diff --git a/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dcce45792f17c6f4a217b759836daa4a81605ce7
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md
@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.5.0
diff --git a/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..6161dcd434f61070bb6c4fb09ef34f3898594544
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "vicuna-v1-3-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "o_proj",
+    "up_proj",
+    "k_proj",
+    "q_proj",
+    "v_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..f4aa9675535a6cb4597b0be9d56c6c871bc02ae0
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df3eab2c5f130a406bd6b6a70ff6f7d3af7060199b9b081057595807e207ab94
+size 319970957
diff --git a/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8a11c13945ca6691233666c289a41f105fe5499a
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json
@@ -0,0 +1,36 @@
+{
+  "_name_or_path": "vicuna-v1-3-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "mm_graph_tower": "hvqvae2",
+  "mm_hidden_size": 308,
+  "mm_projector_type": "hlinear",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "model_type": "llava_graph",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.33.2",
+  "tune_mm_mlp_adapter": false,
+  "use_cache": true,
+  "use_lap_pe": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}
diff --git a/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..76563f694e375d562e9c1b0ec519e8ffddf8d471
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c5c0f66e5988f59e86ec4f21bf39322f67bb3847df4da7ed49f2b5e227a66e0
+size 11335231
diff --git a/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..fe2f25b7b743c1f670fa957825282a64aac06eb4
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/forward_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json
@@ -0,0 +1,58348 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 9720,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.84931506849315e-08,
+      "loss": 0.7129,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.36986301369863e-07,
+      "loss": 0.6816,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0547945205479452e-07,
+      "loss": 0.6914,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 0.6211,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4246575342465755e-07,
+      "loss": 0.6328,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 0.7051,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.794520547945206e-07,
+      "loss": 0.6533,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 0.6846,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.164383561643836e-07,
+      "loss": 0.6797,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.849315068493151e-07,
+      "loss": 0.6484,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.534246575342466e-07,
+      "loss": 0.6699,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 0.6914,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.904109589041097e-07,
+      "loss": 0.6826,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.589041095890411e-07,
+      "loss": 0.625,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0273972602739727e-06,
+      "loss": 0.6719,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 0.6084,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1643835616438357e-06,
+      "loss": 0.7021,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2328767123287673e-06,
+      "loss": 0.7256,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3013698630136986e-06,
+      "loss": 0.709,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 0.7168,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4383561643835616e-06,
+      "loss": 0.6543,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5068493150684932e-06,
+      "loss": 0.7041,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5753424657534248e-06,
+      "loss": 0.6143,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 0.6719,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7123287671232877e-06,
+      "loss": 0.665,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7808219178082193e-06,
+      "loss": 0.5923,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8493150684931507e-06,
+      "loss": 0.6621,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 0.6758,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9863013698630136e-06,
+      "loss": 0.6206,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0547945205479454e-06,
+      "loss": 0.6318,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.123287671232877e-06,
+      "loss": 0.6738,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 0.7402,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.26027397260274e-06,
+      "loss": 0.668,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3287671232876713e-06,
+      "loss": 0.6758,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3972602739726027e-06,
+      "loss": 0.6367,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 0.6494,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.534246575342466e-06,
+      "loss": 0.6025,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 0.7031,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.671232876712329e-06,
+      "loss": 0.6768,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 0.6465,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8082191780821922e-06,
+      "loss": 0.6943,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 0.7012,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.945205479452055e-06,
+      "loss": 0.7002,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 0.6084,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.082191780821918e-06,
+      "loss": 0.585,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1506849315068495e-06,
+      "loss": 0.6592,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2191780821917813e-06,
+      "loss": 0.6416,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 0.5752,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.356164383561644e-06,
+      "loss": 0.6689,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 0.6582,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4931506849315072e-06,
+      "loss": 0.623,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 0.627,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6301369863013704e-06,
+      "loss": 0.6484,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6986301369863014e-06,
+      "loss": 0.6328,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.767123287671233e-06,
+      "loss": 0.5879,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 0.6758,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.904109589041096e-06,
+      "loss": 0.6807,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.972602739726027e-06,
+      "loss": 0.666,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0410958904109595e-06,
+      "loss": 0.6865,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 0.5698,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.178082191780822e-06,
+      "loss": 0.6143,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.246575342465754e-06,
+      "loss": 0.623,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.315068493150685e-06,
+      "loss": 0.6309,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 0.6484,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4520547945205486e-06,
+      "loss": 0.6504,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.52054794520548e-06,
+      "loss": 0.6143,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.589041095890411e-06,
+      "loss": 0.6592,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 0.5605,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.726027397260274e-06,
+      "loss": 0.6113,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7945205479452054e-06,
+      "loss": 0.5986,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.863013698630138e-06,
+      "loss": 0.6162,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 0.583,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 0.6562,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.068493150684932e-06,
+      "loss": 0.5859,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.136986301369864e-06,
+      "loss": 0.5811,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 0.6318,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.273972602739727e-06,
+      "loss": 0.5664,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.342465753424658e-06,
+      "loss": 0.6152,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.41095890410959e-06,
+      "loss": 0.5693,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 0.6372,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.547945205479452e-06,
+      "loss": 0.6182,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6164383561643845e-06,
+      "loss": 0.5518,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.684931506849316e-06,
+      "loss": 0.6367,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 0.6318,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.821917808219179e-06,
+      "loss": 0.5986,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.89041095890411e-06,
+      "loss": 0.5327,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.958904109589042e-06,
+      "loss": 0.5576,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 0.4893,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.095890410958905e-06,
+      "loss": 0.585,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.164383561643836e-06,
+      "loss": 0.5645,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2328767123287685e-06,
+      "loss": 0.6006,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 0.582,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.36986301369863e-06,
+      "loss": 0.5576,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.438356164383563e-06,
+      "loss": 0.6143,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.506849315068494e-06,
+      "loss": 0.5645,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.5488,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.643835616438357e-06,
+      "loss": 0.5049,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.712328767123288e-06,
+      "loss": 0.5752,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.78082191780822e-06,
+      "loss": 0.5684,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 0.521,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.917808219178082e-06,
+      "loss": 0.5469,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9863013698630145e-06,
+      "loss": 0.5879,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.054794520547946e-06,
+      "loss": 0.5435,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 0.5254,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.191780821917809e-06,
+      "loss": 0.5405,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.260273972602741e-06,
+      "loss": 0.5137,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.328767123287672e-06,
+      "loss": 0.5713,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 0.5278,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.465753424657535e-06,
+      "loss": 0.4912,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.534246575342466e-06,
+      "loss": 0.4976,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6027397260273985e-06,
+      "loss": 0.4624,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 0.4717,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.739726027397261e-06,
+      "loss": 0.4927,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 0.4858,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.876712328767124e-06,
+      "loss": 0.4771,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 0.5537,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.013698630136987e-06,
+      "loss": 0.4634,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.082191780821919e-06,
+      "loss": 0.4839,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.150684931506851e-06,
+      "loss": 0.4492,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 0.5,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.287671232876712e-06,
+      "loss": 0.5024,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.356164383561644e-06,
+      "loss": 0.4624,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.424657534246577e-06,
+      "loss": 0.5171,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 0.458,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.56164383561644e-06,
+      "loss": 0.4517,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.63013698630137e-06,
+      "loss": 0.4712,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.698630136986302e-06,
+      "loss": 0.4512,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 0.4619,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.835616438356165e-06,
+      "loss": 0.46,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.904109589041097e-06,
+      "loss": 0.4624,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.972602739726028e-06,
+      "loss": 0.4736,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 0.4521,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.10958904109589e-06,
+      "loss": 0.4644,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.178082191780823e-06,
+      "loss": 0.4429,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.246575342465755e-06,
+      "loss": 0.4209,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 0.4238,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.383561643835618e-06,
+      "loss": 0.4043,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 0.458,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.52054794520548e-06,
+      "loss": 0.4336,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 0.4702,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.657534246575343e-06,
+      "loss": 0.4922,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.726027397260275e-06,
+      "loss": 0.4595,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794520547945206e-06,
+      "loss": 0.4551,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 0.4604,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.931506849315069e-06,
+      "loss": 0.4761,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.4453,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0068493150684933e-05,
+      "loss": 0.4346,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 0.4712,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0205479452054796e-05,
+      "loss": 0.4658,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0273972602739728e-05,
+      "loss": 0.4414,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0342465753424657e-05,
+      "loss": 0.4404,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 0.4248,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0479452054794521e-05,
+      "loss": 0.4775,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0547945205479453e-05,
+      "loss": 0.4497,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0616438356164384e-05,
+      "loss": 0.4521,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0684931506849316e-05,
+      "loss": 0.4531,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0753424657534248e-05,
+      "loss": 0.4873,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.082191780821918e-05,
+      "loss": 0.3945,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.089041095890411e-05,
+      "loss": 0.3955,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 0.4385,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1027397260273974e-05,
+      "loss": 0.4111,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1095890410958904e-05,
+      "loss": 0.4092,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1164383561643837e-05,
+      "loss": 0.4409,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1232876712328769e-05,
+      "loss": 0.4072,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1301369863013701e-05,
+      "loss": 0.4722,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1369863013698632e-05,
+      "loss": 0.3789,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1438356164383562e-05,
+      "loss": 0.3818,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.4346,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1575342465753425e-05,
+      "loss": 0.3657,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1643835616438357e-05,
+      "loss": 0.3975,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.171232876712329e-05,
+      "loss": 0.4272,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.178082191780822e-05,
+      "loss": 0.3853,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1849315068493152e-05,
+      "loss": 0.3931,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1917808219178084e-05,
+      "loss": 0.4302,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1986301369863013e-05,
+      "loss": 0.4644,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 0.4092,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2123287671232878e-05,
+      "loss": 0.4111,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.219178082191781e-05,
+      "loss": 0.4092,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.226027397260274e-05,
+      "loss": 0.3755,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2328767123287673e-05,
+      "loss": 0.4155,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2397260273972605e-05,
+      "loss": 0.4448,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2465753424657537e-05,
+      "loss": 0.3481,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2534246575342466e-05,
+      "loss": 0.3921,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 0.4224,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2671232876712329e-05,
+      "loss": 0.4082,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.273972602739726e-05,
+      "loss": 0.374,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2808219178082193e-05,
+      "loss": 0.3892,
+      "step": 187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2876712328767125e-05,
+      "loss": 0.4312,
+      "step": 188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2945205479452056e-05,
+      "loss": 0.4121,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3013698630136988e-05,
+      "loss": 0.3984,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3082191780821919e-05,
+      "loss": 0.4346,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.375,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3219178082191781e-05,
+      "loss": 0.374,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3287671232876714e-05,
+      "loss": 0.3657,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3356164383561646e-05,
+      "loss": 0.3911,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3424657534246576e-05,
+      "loss": 0.3857,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3493150684931508e-05,
+      "loss": 0.418,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.356164383561644e-05,
+      "loss": 0.4048,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.363013698630137e-05,
+      "loss": 0.3882,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.3892,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3767123287671234e-05,
+      "loss": 0.3582,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3835616438356164e-05,
+      "loss": 0.3965,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3904109589041097e-05,
+      "loss": 0.396,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3972602739726029e-05,
+      "loss": 0.4404,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4041095890410961e-05,
+      "loss": 0.3608,
+      "step": 205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4109589041095892e-05,
+      "loss": 0.3682,
+      "step": 206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4178082191780822e-05,
+      "loss": 0.3926,
+      "step": 207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 0.3672,
+      "step": 208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4315068493150685e-05,
+      "loss": 0.3921,
+      "step": 209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4383561643835617e-05,
+      "loss": 0.3945,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.445205479452055e-05,
+      "loss": 0.4111,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4520547945205482e-05,
+      "loss": 0.4004,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4589041095890412e-05,
+      "loss": 0.3975,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4657534246575344e-05,
+      "loss": 0.3623,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4726027397260275e-05,
+      "loss": 0.377,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 0.395,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4863013698630138e-05,
+      "loss": 0.3755,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.493150684931507e-05,
+      "loss": 0.3755,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.395,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5068493150684933e-05,
+      "loss": 0.3501,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5136986301369865e-05,
+      "loss": 0.3872,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5205479452054797e-05,
+      "loss": 0.3667,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5273972602739728e-05,
+      "loss": 0.3877,
+      "step": 223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 0.4033,
+      "step": 224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.541095890410959e-05,
+      "loss": 0.3828,
+      "step": 225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5479452054794523e-05,
+      "loss": 0.3843,
+      "step": 226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5547945205479453e-05,
+      "loss": 0.3877,
+      "step": 227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.3833,
+      "step": 228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5684931506849318e-05,
+      "loss": 0.3589,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5753424657534248e-05,
+      "loss": 0.2964,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5821917808219182e-05,
+      "loss": 0.3755,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.3701,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5958904109589043e-05,
+      "loss": 0.353,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6027397260273974e-05,
+      "loss": 0.3628,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6095890410958904e-05,
+      "loss": 0.4072,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6164383561643838e-05,
+      "loss": 0.3291,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.623287671232877e-05,
+      "loss": 0.3398,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6301369863013702e-05,
+      "loss": 0.3716,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6369863013698633e-05,
+      "loss": 0.3809,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 0.3496,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6506849315068494e-05,
+      "loss": 0.3887,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6575342465753425e-05,
+      "loss": 0.3345,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.664383561643836e-05,
+      "loss": 0.3843,
+      "step": 243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.671232876712329e-05,
+      "loss": 0.3447,
+      "step": 244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.678082191780822e-05,
+      "loss": 0.3169,
+      "step": 245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6849315068493153e-05,
+      "loss": 0.3491,
+      "step": 246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6917808219178084e-05,
+      "loss": 0.3828,
+      "step": 247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 0.3237,
+      "step": 248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7054794520547945e-05,
+      "loss": 0.2729,
+      "step": 249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.712328767123288e-05,
+      "loss": 0.3843,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.719178082191781e-05,
+      "loss": 0.3354,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.726027397260274e-05,
+      "loss": 0.3203,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7328767123287674e-05,
+      "loss": 0.3608,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7397260273972604e-05,
+      "loss": 0.3828,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7465753424657538e-05,
+      "loss": 0.3696,
+      "step": 255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 0.3525,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.76027397260274e-05,
+      "loss": 0.3501,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.767123287671233e-05,
+      "loss": 0.3608,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.773972602739726e-05,
+      "loss": 0.375,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7808219178082194e-05,
+      "loss": 0.3535,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7876712328767125e-05,
+      "loss": 0.3315,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7945205479452055e-05,
+      "loss": 0.3364,
+      "step": 262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.801369863013699e-05,
+      "loss": 0.353,
+      "step": 263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 0.335,
+      "step": 264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.815068493150685e-05,
+      "loss": 0.3291,
+      "step": 265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.821917808219178e-05,
+      "loss": 0.2983,
+      "step": 266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8287671232876715e-05,
+      "loss": 0.3564,
+      "step": 267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8356164383561645e-05,
+      "loss": 0.3296,
+      "step": 268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8424657534246576e-05,
+      "loss": 0.3071,
+      "step": 269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.849315068493151e-05,
+      "loss": 0.3369,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.856164383561644e-05,
+      "loss": 0.3335,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 0.3247,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.86986301369863e-05,
+      "loss": 0.2993,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8767123287671235e-05,
+      "loss": 0.3496,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8835616438356166e-05,
+      "loss": 0.3242,
+      "step": 275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 0.3452,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.897260273972603e-05,
+      "loss": 0.2966,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904109589041096e-05,
+      "loss": 0.2646,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910958904109589e-05,
+      "loss": 0.3091,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 0.2935,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9246575342465756e-05,
+      "loss": 0.3433,
+      "step": 281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315068493150686e-05,
+      "loss": 0.3477,
+      "step": 282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9383561643835617e-05,
+      "loss": 0.2842,
+      "step": 283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.945205479452055e-05,
+      "loss": 0.3042,
+      "step": 284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.952054794520548e-05,
+      "loss": 0.3164,
+      "step": 285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9589041095890412e-05,
+      "loss": 0.3262,
+      "step": 286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9657534246575346e-05,
+      "loss": 0.3018,
+      "step": 287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 0.3035,
+      "step": 288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9794520547945207e-05,
+      "loss": 0.2942,
+      "step": 289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9863013698630137e-05,
+      "loss": 0.3306,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993150684931507e-05,
+      "loss": 0.3145,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3091,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999444824108e-05,
+      "loss": 0.3132,
+      "step": 293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997779296502e-05,
+      "loss": 0.2849,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995003417356e-05,
+      "loss": 0.2961,
+      "step": 295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999111718698e-05,
+      "loss": 0.2891,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999986120605816e-05,
+      "loss": 0.2822,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998001367441e-05,
+      "loss": 0.3159,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997279639344e-05,
+      "loss": 0.3276,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996446876371e-05,
+      "loss": 0.3127,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999955030786143e-05,
+      "loss": 0.3135,
+      "step": 301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999994448246179e-05,
+      "loss": 0.2935,
+      "step": 302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999932823791816e-05,
+      "loss": 0.3267,
+      "step": 303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999920054777522e-05,
+      "loss": 0.3152,
+      "step": 304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999990617542032e-05,
+      "loss": 0.3301,
+      "step": 305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999989118572176e-05,
+      "loss": 0.2861,
+      "step": 306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999875085683498e-05,
+      "loss": 0.2791,
+      "step": 307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999857875307324e-05,
+      "loss": 0.3008,
+      "step": 308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999839554595152e-05,
+      "loss": 0.29,
+      "step": 309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999820123549014e-05,
+      "loss": 0.2991,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999799582171066e-05,
+      "loss": 0.2778,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999977793046359e-05,
+      "loss": 0.3027,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999755168428986e-05,
+      "loss": 0.2722,
+      "step": 313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999731296069788e-05,
+      "loss": 0.2969,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999706313388645e-05,
+      "loss": 0.2969,
+      "step": 315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999968022038833e-05,
+      "loss": 0.2974,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999965301707174e-05,
+      "loss": 0.29,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99996247034419e-05,
+      "loss": 0.2622,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999595279501944e-05,
+      "loss": 0.3149,
+      "step": 319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999564745255148e-05,
+      "loss": 0.2827,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99995331007049e-05,
+      "loss": 0.2969,
+      "step": 321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999950034585471e-05,
+      "loss": 0.2908,
+      "step": 322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946648070822e-05,
+      "loss": 0.2673,
+      "step": 323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999431505269185e-05,
+      "loss": 0.2903,
+      "step": 324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999395419541494e-05,
+      "loss": 0.2688,
+      "step": 325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999935822352915e-05,
+      "loss": 0.2964,
+      "step": 326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999319917236287e-05,
+      "loss": 0.3135,
+      "step": 327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999280500667154e-05,
+      "loss": 0.2615,
+      "step": 328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999923997382613e-05,
+      "loss": 0.3081,
+      "step": 329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999198336717712e-05,
+      "loss": 0.2693,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999155589346528e-05,
+      "loss": 0.2839,
+      "step": 331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999911173171732e-05,
+      "loss": 0.2563,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999906676383496e-05,
+      "loss": 0.3076,
+      "step": 333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999902068570444e-05,
+      "loss": 0.2786,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998973497330878e-05,
+      "loss": 0.2942,
+      "step": 335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998925198719514e-05,
+      "loss": 0.2888,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998875789875707e-05,
+      "loss": 0.2939,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999882527080495e-05,
+      "loss": 0.2944,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998773641512842e-05,
+      "loss": 0.248,
+      "step": 339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998720902005125e-05,
+      "loss": 0.3091,
+      "step": 340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998667052287647e-05,
+      "loss": 0.2317,
+      "step": 341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998612092366396e-05,
+      "loss": 0.2498,
+      "step": 342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998556022247468e-05,
+      "loss": 0.2456,
+      "step": 343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999849884193709e-05,
+      "loss": 0.293,
+      "step": 344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998440551441618e-05,
+      "loss": 0.2483,
+      "step": 345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998381150767514e-05,
+      "loss": 0.2598,
+      "step": 346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998320639921377e-05,
+      "loss": 0.2661,
+      "step": 347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999825901890993e-05,
+      "loss": 0.312,
+      "step": 348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998196287740006e-05,
+      "loss": 0.2646,
+      "step": 349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998132446418583e-05,
+      "loss": 0.3145,
+      "step": 350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998067494952736e-05,
+      "loss": 0.29,
+      "step": 351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999800143334969e-05,
+      "loss": 0.2302,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997934261616768e-05,
+      "loss": 0.2585,
+      "step": 353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997865979761436e-05,
+      "loss": 0.2639,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997796587791276e-05,
+      "loss": 0.2959,
+      "step": 355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997726085713993e-05,
+      "loss": 0.2563,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999765447353741e-05,
+      "loss": 0.2695,
+      "step": 357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999758175126948e-05,
+      "loss": 0.2324,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999750791891828e-05,
+      "loss": 0.2844,
+      "step": 359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997432976492006e-05,
+      "loss": 0.2773,
+      "step": 360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999735692399898e-05,
+      "loss": 0.2422,
+      "step": 361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997279761447652e-05,
+      "loss": 0.2681,
+      "step": 362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997201488846585e-05,
+      "loss": 0.2617,
+      "step": 363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997122106204466e-05,
+      "loss": 0.2559,
+      "step": 364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997041613530115e-05,
+      "loss": 0.2773,
+      "step": 365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996960010832466e-05,
+      "loss": 0.2703,
+      "step": 366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996877298120583e-05,
+      "loss": 0.2437,
+      "step": 367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996793475403647e-05,
+      "loss": 0.3135,
+      "step": 368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996708542690966e-05,
+      "loss": 0.252,
+      "step": 369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996622499991973e-05,
+      "loss": 0.2695,
+      "step": 370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996535347316217e-05,
+      "loss": 0.2952,
+      "step": 371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996447084673383e-05,
+      "loss": 0.26,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996357712073263e-05,
+      "loss": 0.272,
+      "step": 373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996267229525783e-05,
+      "loss": 0.2319,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999617563704099e-05,
+      "loss": 0.2439,
+      "step": 375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996082934629057e-05,
+      "loss": 0.2478,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995989122300275e-05,
+      "loss": 0.2722,
+      "step": 377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995894200065055e-05,
+      "loss": 0.2825,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995798167933945e-05,
+      "loss": 0.2424,
+      "step": 379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995701025917607e-05,
+      "loss": 0.238,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995602774026826e-05,
+      "loss": 0.2668,
+      "step": 381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995503412272504e-05,
+      "loss": 0.2319,
+      "step": 382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995402940665684e-05,
+      "loss": 0.2627,
+      "step": 383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999530135921752e-05,
+      "loss": 0.2588,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995198667939285e-05,
+      "loss": 0.2627,
+      "step": 385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995094866842386e-05,
+      "loss": 0.2842,
+      "step": 386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994989955938352e-05,
+      "loss": 0.261,
+      "step": 387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994883935238822e-05,
+      "loss": 0.261,
+      "step": 388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994776804755576e-05,
+      "loss": 0.26,
+      "step": 389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999466856450051e-05,
+      "loss": 0.2251,
+      "step": 390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994559214485637e-05,
+      "loss": 0.2224,
+      "step": 391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994448754723098e-05,
+      "loss": 0.2366,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994337185225164e-05,
+      "loss": 0.2507,
+      "step": 393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999422450600422e-05,
+      "loss": 0.2668,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999411071707278e-05,
+      "loss": 0.2371,
+      "step": 395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993995818443473e-05,
+      "loss": 0.2664,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999387981012906e-05,
+      "loss": 0.2419,
+      "step": 397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993762692142423e-05,
+      "loss": 0.2595,
+      "step": 398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993644464496566e-05,
+      "loss": 0.2383,
+      "step": 399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993525127204615e-05,
+      "loss": 0.2432,
+      "step": 400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999340468027982e-05,
+      "loss": 0.2261,
+      "step": 401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999328312373556e-05,
+      "loss": 0.2273,
+      "step": 402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993160457585325e-05,
+      "loss": 0.2532,
+      "step": 403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993036681842737e-05,
+      "loss": 0.2246,
+      "step": 404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992911796521543e-05,
+      "loss": 0.2524,
+      "step": 405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999278580163561e-05,
+      "loss": 0.2524,
+      "step": 406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992658697198922e-05,
+      "loss": 0.2363,
+      "step": 407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992530483225596e-05,
+      "loss": 0.2529,
+      "step": 408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992401159729866e-05,
+      "loss": 0.2395,
+      "step": 409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992270726726096e-05,
+      "loss": 0.2458,
+      "step": 410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992139184228766e-05,
+      "loss": 0.2622,
+      "step": 411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999200653225248e-05,
+      "loss": 0.2678,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999187277081197e-05,
+      "loss": 0.2568,
+      "step": 413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991737899922086e-05,
+      "loss": 0.2246,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991601919597802e-05,
+      "loss": 0.2322,
+      "step": 415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991464829854223e-05,
+      "loss": 0.2439,
+      "step": 416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991326630706564e-05,
+      "loss": 0.2754,
+      "step": 417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991187322170175e-05,
+      "loss": 0.218,
+      "step": 418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999104690426052e-05,
+      "loss": 0.2063,
+      "step": 419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999090537699319e-05,
+      "loss": 0.2422,
+      "step": 420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990762740383908e-05,
+      "loss": 0.2432,
+      "step": 421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990618994448502e-05,
+      "loss": 0.2375,
+      "step": 422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990474139202936e-05,
+      "loss": 0.26,
+      "step": 423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990328174663297e-05,
+      "loss": 0.2649,
+      "step": 424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990181100845786e-05,
+      "loss": 0.2361,
+      "step": 425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999003291776674e-05,
+      "loss": 0.2378,
+      "step": 426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989883625442603e-05,
+      "loss": 0.2786,
+      "step": 427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989733223889964e-05,
+      "loss": 0.2449,
+      "step": 428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989581713125516e-05,
+      "loss": 0.26,
+      "step": 429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998942909316608e-05,
+      "loss": 0.2302,
+      "step": 430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989275364028608e-05,
+      "loss": 0.2529,
+      "step": 431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989120525730165e-05,
+      "loss": 0.249,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998896457828795e-05,
+      "loss": 0.2551,
+      "step": 433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988807521719264e-05,
+      "loss": 0.2239,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988649356041562e-05,
+      "loss": 0.2234,
+      "step": 435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988490081272397e-05,
+      "loss": 0.2563,
+      "step": 436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998832969742946e-05,
+      "loss": 0.2107,
+      "step": 437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998816820453055e-05,
+      "loss": 0.2634,
+      "step": 438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998800560259361e-05,
+      "loss": 0.2175,
+      "step": 439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987841891636687e-05,
+      "loss": 0.2693,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998767707167796e-05,
+      "loss": 0.2476,
+      "step": 441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998751114273573e-05,
+      "loss": 0.2324,
+      "step": 442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998734410482842e-05,
+      "loss": 0.2449,
+      "step": 443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987175957974577e-05,
+      "loss": 0.2485,
+      "step": 444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987006702192875e-05,
+      "loss": 0.2397,
+      "step": 445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986836337502106e-05,
+      "loss": 0.2498,
+      "step": 446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986664863921183e-05,
+      "loss": 0.2307,
+      "step": 447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986492281469147e-05,
+      "loss": 0.2483,
+      "step": 448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986318590165162e-05,
+      "loss": 0.2625,
+      "step": 449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986143790028513e-05,
+      "loss": 0.2324,
+      "step": 450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998596788107861e-05,
+      "loss": 0.2427,
+      "step": 451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985790863334988e-05,
+      "loss": 0.2273,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985612736817295e-05,
+      "loss": 0.2168,
+      "step": 453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985433501545312e-05,
+      "loss": 0.2195,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998525315753894e-05,
+      "loss": 0.1938,
+      "step": 455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998507170481821e-05,
+      "loss": 0.2397,
+      "step": 456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984889143403263e-05,
+      "loss": 0.2505,
+      "step": 457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998470547331437e-05,
+      "loss": 0.2524,
+      "step": 458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984520694571926e-05,
+      "loss": 0.2197,
+      "step": 459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998433480719645e-05,
+      "loss": 0.2378,
+      "step": 460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984147811208576e-05,
+      "loss": 0.2244,
+      "step": 461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983959706629076e-05,
+      "loss": 0.2434,
+      "step": 462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983770493478828e-05,
+      "loss": 0.2107,
+      "step": 463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983580171778846e-05,
+      "loss": 0.2451,
+      "step": 464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998338874155026e-05,
+      "loss": 0.229,
+      "step": 465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998319620281433e-05,
+      "loss": 0.2419,
+      "step": 466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983002555592428e-05,
+      "loss": 0.2097,
+      "step": 467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982807799906057e-05,
+      "loss": 0.24,
+      "step": 468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982611935776847e-05,
+      "loss": 0.2302,
+      "step": 469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982414963226544e-05,
+      "loss": 0.2214,
+      "step": 470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982216882277013e-05,
+      "loss": 0.2288,
+      "step": 471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982017692950254e-05,
+      "loss": 0.2378,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981817395268384e-05,
+      "loss": 0.2515,
+      "step": 473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998161598925364e-05,
+      "loss": 0.1936,
+      "step": 474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981413474928386e-05,
+      "loss": 0.2041,
+      "step": 475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981209852315108e-05,
+      "loss": 0.2319,
+      "step": 476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9981005121436418e-05,
+      "loss": 0.229,
+      "step": 477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980799282315045e-05,
+      "loss": 0.2026,
+      "step": 478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980592334973847e-05,
+      "loss": 0.2354,
+      "step": 479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99803842794358e-05,
+      "loss": 0.2129,
+      "step": 480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980175115724007e-05,
+      "loss": 0.2334,
+      "step": 481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979964843861693e-05,
+      "loss": 0.2402,
+      "step": 482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979753463872203e-05,
+      "loss": 0.2219,
+      "step": 483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997954097577901e-05,
+      "loss": 0.2656,
+      "step": 484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979327379605707e-05,
+      "loss": 0.2732,
+      "step": 485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979112675376014e-05,
+      "loss": 0.249,
+      "step": 486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997889686311376e-05,
+      "loss": 0.2019,
+      "step": 487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978679942842922e-05,
+      "loss": 0.251,
+      "step": 488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978461914587577e-05,
+      "loss": 0.2405,
+      "step": 489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978242778371934e-05,
+      "loss": 0.2297,
+      "step": 490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997802253422033e-05,
+      "loss": 0.2246,
+      "step": 491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977801182157213e-05,
+      "loss": 0.1921,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997757872220717e-05,
+      "loss": 0.259,
+      "step": 493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977355154394885e-05,
+      "loss": 0.2302,
+      "step": 494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99771304787452e-05,
+      "loss": 0.2278,
+      "step": 495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976904695283057e-05,
+      "loss": 0.2344,
+      "step": 496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997667780403352e-05,
+      "loss": 0.199,
+      "step": 497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976449805021788e-05,
+      "loss": 0.1921,
+      "step": 498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976220698273177e-05,
+      "loss": 0.2124,
+      "step": 499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997599048381312e-05,
+      "loss": 0.2158,
+      "step": 500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975759161667182e-05,
+      "loss": 0.2344,
+      "step": 501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997552673186105e-05,
+      "loss": 0.2007,
+      "step": 502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975293194420532e-05,
+      "loss": 0.2034,
+      "step": 503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975058549371553e-05,
+      "loss": 0.2219,
+      "step": 504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974822796740174e-05,
+      "loss": 0.2197,
+      "step": 505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974585936552565e-05,
+      "loss": 0.2114,
+      "step": 506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974347968835036e-05,
+      "loss": 0.2109,
+      "step": 507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974108893614e-05,
+      "loss": 0.21,
+      "step": 508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973868710916004e-05,
+      "loss": 0.2163,
+      "step": 509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997362742076772e-05,
+      "loss": 0.2239,
+      "step": 510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973385023195943e-05,
+      "loss": 0.2183,
+      "step": 511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997314151822758e-05,
+      "loss": 0.2102,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972896905889674e-05,
+      "loss": 0.1814,
+      "step": 513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997265118620938e-05,
+      "loss": 0.22,
+      "step": 514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972404359213987e-05,
+      "loss": 0.2039,
+      "step": 515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972156424930898e-05,
+      "loss": 0.2278,
+      "step": 516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971907383387644e-05,
+      "loss": 0.2258,
+      "step": 517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997165723461188e-05,
+      "loss": 0.2051,
+      "step": 518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971405978631378e-05,
+      "loss": 0.2183,
+      "step": 519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971153615474036e-05,
+      "loss": 0.2026,
+      "step": 520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970900145167877e-05,
+      "loss": 0.1698,
+      "step": 521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997064556774104e-05,
+      "loss": 0.2168,
+      "step": 522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99703898832218e-05,
+      "loss": 0.2166,
+      "step": 523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997013309163854e-05,
+      "loss": 0.2112,
+      "step": 524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969875193019783e-05,
+      "loss": 0.1912,
+      "step": 525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996961618739415e-05,
+      "loss": 0.2024,
+      "step": 526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969356074790412e-05,
+      "loss": 0.2085,
+      "step": 527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969094855237446e-05,
+      "loss": 0.2158,
+      "step": 528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968832528764256e-05,
+      "loss": 0.2097,
+      "step": 529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996856909539997e-05,
+      "loss": 0.2314,
+      "step": 530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996830455517384e-05,
+      "loss": 0.2241,
+      "step": 531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968038908115237e-05,
+      "loss": 0.179,
+      "step": 532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996777215425366e-05,
+      "loss": 0.1958,
+      "step": 533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967504293618725e-05,
+      "loss": 0.2324,
+      "step": 534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967235326240174e-05,
+      "loss": 0.1941,
+      "step": 535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966965252147873e-05,
+      "loss": 0.2109,
+      "step": 536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996669407137181e-05,
+      "loss": 0.2173,
+      "step": 537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966421783942094e-05,
+      "loss": 0.2124,
+      "step": 538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966148389888964e-05,
+      "loss": 0.209,
+      "step": 539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965873889242768e-05,
+      "loss": 0.207,
+      "step": 540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965598282033994e-05,
+      "loss": 0.1924,
+      "step": 541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965321568293233e-05,
+      "loss": 0.2319,
+      "step": 542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965043748051222e-05,
+      "loss": 0.2041,
+      "step": 543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99647648213388e-05,
+      "loss": 0.1926,
+      "step": 544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996448478818694e-05,
+      "loss": 0.219,
+      "step": 545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996420364862674e-05,
+      "loss": 0.1877,
+      "step": 546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963921402689412e-05,
+      "loss": 0.1865,
+      "step": 547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963638050406297e-05,
+      "loss": 0.2034,
+      "step": 548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963353591808853e-05,
+      "loss": 0.1846,
+      "step": 549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963068026928673e-05,
+      "loss": 0.2158,
+      "step": 550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962781355797456e-05,
+      "loss": 0.2163,
+      "step": 551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962493578447038e-05,
+      "loss": 0.1946,
+      "step": 552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962204694909368e-05,
+      "loss": 0.1863,
+      "step": 553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9961914705216526e-05,
+      "loss": 0.2205,
+      "step": 554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961623609400712e-05,
+      "loss": 0.2046,
+      "step": 555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961331407494245e-05,
+      "loss": 0.2019,
+      "step": 556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996103809952957e-05,
+      "loss": 0.1978,
+      "step": 557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996074368553926e-05,
+      "loss": 0.1897,
+      "step": 558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960448165555992e-05,
+      "loss": 0.2383,
+      "step": 559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960151539612593e-05,
+      "loss": 0.1921,
+      "step": 560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959853807741992e-05,
+      "loss": 0.2192,
+      "step": 561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995955496997725e-05,
+      "loss": 0.1924,
+      "step": 562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959255026351548e-05,
+      "loss": 0.2041,
+      "step": 563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995895397689819e-05,
+      "loss": 0.1699,
+      "step": 564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958651821650604e-05,
+      "loss": 0.2048,
+      "step": 565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958348560642336e-05,
+      "loss": 0.2015,
+      "step": 566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958044193907063e-05,
+      "loss": 0.2075,
+      "step": 567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995773872147858e-05,
+      "loss": 0.2034,
+      "step": 568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99574321433908e-05,
+      "loss": 0.2073,
+      "step": 569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957124459677772e-05,
+      "loss": 0.2146,
+      "step": 570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956815670373655e-05,
+      "loss": 0.1921,
+      "step": 571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995650577551274e-05,
+      "loss": 0.1963,
+      "step": 572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956194775129426e-05,
+      "loss": 0.1892,
+      "step": 573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955882669258256e-05,
+      "loss": 0.1975,
+      "step": 574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995556945793388e-05,
+      "loss": 0.2112,
+      "step": 575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955255141191074e-05,
+      "loss": 0.1956,
+      "step": 576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995493971906474e-05,
+      "loss": 0.1882,
+      "step": 577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99546231915899e-05,
+      "loss": 0.1887,
+      "step": 578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954305558801703e-05,
+      "loss": 0.1753,
+      "step": 579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953986820735414e-05,
+      "loss": 0.2036,
+      "step": 580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953666977426428e-05,
+      "loss": 0.2031,
+      "step": 581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995334602891025e-05,
+      "loss": 0.1841,
+      "step": 582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953023975222527e-05,
+      "loss": 0.2039,
+      "step": 583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952700816399014e-05,
+      "loss": 0.2048,
+      "step": 584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995237655247559e-05,
+      "loss": 0.2019,
+      "step": 585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952051183488262e-05,
+      "loss": 0.2051,
+      "step": 586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951724709473163e-05,
+      "loss": 0.1875,
+      "step": 587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951397130466535e-05,
+      "loss": 0.2075,
+      "step": 588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951068446504753e-05,
+      "loss": 0.1995,
+      "step": 589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950738657624318e-05,
+      "loss": 0.2119,
+      "step": 590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950407763861837e-05,
+      "loss": 0.1892,
+      "step": 591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950075765254063e-05,
+      "loss": 0.2153,
+      "step": 592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994974266183785e-05,
+      "loss": 0.1824,
+      "step": 593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949408453650194e-05,
+      "loss": 0.2305,
+      "step": 594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994907314072819e-05,
+      "loss": 0.2002,
+      "step": 595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948736723109082e-05,
+      "loss": 0.2061,
+      "step": 596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994839920083022e-05,
+      "loss": 0.2126,
+      "step": 597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948060573929075e-05,
+      "loss": 0.1821,
+      "step": 598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947720842443255e-05,
+      "loss": 0.1772,
+      "step": 599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947380006410484e-05,
+      "loss": 0.2007,
+      "step": 600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947038065868597e-05,
+      "loss": 0.2112,
+      "step": 601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994669502085557e-05,
+      "loss": 0.2134,
+      "step": 602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946350871409484e-05,
+      "loss": 0.2092,
+      "step": 603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946005617568563e-05,
+      "loss": 0.1863,
+      "step": 604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945659259371133e-05,
+      "loss": 0.2114,
+      "step": 605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945311796855654e-05,
+      "loss": 0.2029,
+      "step": 606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944963230060713e-05,
+      "loss": 0.1973,
+      "step": 607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944613559025005e-05,
+      "loss": 0.2102,
+      "step": 608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994426278378736e-05,
+      "loss": 0.1948,
+      "step": 609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943910904386725e-05,
+      "loss": 0.1978,
+      "step": 610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994355792086217e-05,
+      "loss": 0.1676,
+      "step": 611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994320383325289e-05,
+      "loss": 0.1904,
+      "step": 612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942848641598204e-05,
+      "loss": 0.1951,
+      "step": 613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942492345937545e-05,
+      "loss": 0.1978,
+      "step": 614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942134946310477e-05,
+      "loss": 0.1628,
+      "step": 615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941776442756685e-05,
+      "loss": 0.219,
+      "step": 616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941416835315977e-05,
+      "loss": 0.1987,
+      "step": 617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941056124028277e-05,
+      "loss": 0.2244,
+      "step": 618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940694308933638e-05,
+      "loss": 0.2009,
+      "step": 619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994033139007224e-05,
+      "loss": 0.2004,
+      "step": 620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939967367484372e-05,
+      "loss": 0.1831,
+      "step": 621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939602241210457e-05,
+      "loss": 0.1958,
+      "step": 622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993923601129104e-05,
+      "loss": 0.1833,
+      "step": 623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938868677766778e-05,
+      "loss": 0.1697,
+      "step": 624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993850024067846e-05,
+      "loss": 0.2009,
+      "step": 625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938130700067005e-05,
+      "loss": 0.1748,
+      "step": 626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937760055973433e-05,
+      "loss": 0.2073,
+      "step": 627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99373883084389e-05,
+      "loss": 0.1919,
+      "step": 628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993701545750469e-05,
+      "loss": 0.2097,
+      "step": 629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936641503212195e-05,
+      "loss": 0.2056,
+      "step": 630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936266445602944e-05,
+      "loss": 0.1965,
+      "step": 631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935890284718574e-05,
+      "loss": 0.2227,
+      "step": 632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993551302060086e-05,
+      "loss": 0.2185,
+      "step": 633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993513465329169e-05,
+      "loss": 0.2239,
+      "step": 634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993475518283307e-05,
+      "loss": 0.1926,
+      "step": 635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993437460926714e-05,
+      "loss": 0.207,
+      "step": 636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933992932636154e-05,
+      "loss": 0.1724,
+      "step": 637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993361015298249e-05,
+      "loss": 0.1863,
+      "step": 638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993322627034866e-05,
+      "loss": 0.1816,
+      "step": 639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932841284777277e-05,
+      "loss": 0.196,
+      "step": 640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932455196311093e-05,
+      "loss": 0.2039,
+      "step": 641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993206800499298e-05,
+      "loss": 0.178,
+      "step": 642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931679710865923e-05,
+      "loss": 0.2095,
+      "step": 643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931290313973043e-05,
+      "loss": 0.1963,
+      "step": 644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993089981435757e-05,
+      "loss": 0.1814,
+      "step": 645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930508212062874e-05,
+      "loss": 0.2024,
+      "step": 646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930115507132424e-05,
+      "loss": 0.2041,
+      "step": 647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929721699609828e-05,
+      "loss": 0.1787,
+      "step": 648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929326789538818e-05,
+      "loss": 0.22,
+      "step": 649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992893077696324e-05,
+      "loss": 0.1549,
+      "step": 650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928533661927064e-05,
+      "loss": 0.166,
+      "step": 651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928135444474382e-05,
+      "loss": 0.1689,
+      "step": 652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927736124649413e-05,
+      "loss": 0.1726,
+      "step": 653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927335702496496e-05,
+      "loss": 0.1899,
+      "step": 654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992693417806009e-05,
+      "loss": 0.1946,
+      "step": 655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992653155138478e-05,
+      "loss": 0.1636,
+      "step": 656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926127822515266e-05,
+      "loss": 0.2283,
+      "step": 657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925722991496386e-05,
+      "loss": 0.1953,
+      "step": 658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925317058373086e-05,
+      "loss": 0.2024,
+      "step": 659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924910023190434e-05,
+      "loss": 0.187,
+      "step": 660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924501885993635e-05,
+      "loss": 0.1836,
+      "step": 661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924092646828e-05,
+      "loss": 0.1724,
+      "step": 662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923682305738966e-05,
+      "loss": 0.1505,
+      "step": 663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923270862772104e-05,
+      "loss": 0.1909,
+      "step": 664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992285831797309e-05,
+      "loss": 0.2212,
+      "step": 665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992244467138774e-05,
+      "loss": 0.2109,
+      "step": 666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922029923061973e-05,
+      "loss": 0.2004,
+      "step": 667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921614073041847e-05,
+      "loss": 0.1929,
+      "step": 668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992119712137354e-05,
+      "loss": 0.2026,
+      "step": 669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920779068103336e-05,
+      "loss": 0.2043,
+      "step": 670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920359913277667e-05,
+      "loss": 0.217,
+      "step": 671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919939656943062e-05,
+      "loss": 0.1975,
+      "step": 672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919518299146196e-05,
+      "loss": 0.1788,
+      "step": 673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919095839933846e-05,
+      "loss": 0.1707,
+      "step": 674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918672279352923e-05,
+      "loss": 0.2024,
+      "step": 675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918247617450454e-05,
+      "loss": 0.2041,
+      "step": 676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917821854273597e-05,
+      "loss": 0.2122,
+      "step": 677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991739498986962e-05,
+      "loss": 0.1887,
+      "step": 678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916967024285928e-05,
+      "loss": 0.1968,
+      "step": 679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916537957570035e-05,
+      "loss": 0.1743,
+      "step": 680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916107789769583e-05,
+      "loss": 0.1874,
+      "step": 681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915676520932334e-05,
+      "loss": 0.1643,
+      "step": 682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915244151106177e-05,
+      "loss": 0.1995,
+      "step": 683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991481068033912e-05,
+      "loss": 0.1753,
+      "step": 684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914376108679295e-05,
+      "loss": 0.178,
+      "step": 685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991394043617495e-05,
+      "loss": 0.1829,
+      "step": 686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913503662874462e-05,
+      "loss": 0.2017,
+      "step": 687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991306578882633e-05,
+      "loss": 0.199,
+      "step": 688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912626814079172e-05,
+      "loss": 0.1846,
+      "step": 689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991218673868173e-05,
+      "loss": 0.1833,
+      "step": 690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911745562682866e-05,
+      "loss": 0.1785,
+      "step": 691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911303286131574e-05,
+      "loss": 0.1746,
+      "step": 692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991085990907695e-05,
+      "loss": 0.1931,
+      "step": 693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910415431568233e-05,
+      "loss": 0.1698,
+      "step": 694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990996985365477e-05,
+      "loss": 0.1649,
+      "step": 695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990952317538604e-05,
+      "loss": 0.1909,
+      "step": 696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909075396811643e-05,
+      "loss": 0.1836,
+      "step": 697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990862651798129e-05,
+      "loss": 0.1956,
+      "step": 698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990817653894483e-05,
+      "loss": 0.1738,
+      "step": 699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990772545975222e-05,
+      "loss": 0.1531,
+      "step": 700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990727328045355e-05,
+      "loss": 0.2017,
+      "step": 701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906820001099024e-05,
+      "loss": 0.1699,
+      "step": 702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906365621738975e-05,
+      "loss": 0.1753,
+      "step": 703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905910142423853e-05,
+      "loss": 0.1416,
+      "step": 704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905453563204237e-05,
+      "loss": 0.189,
+      "step": 705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990499588413082e-05,
+      "loss": 0.1929,
+      "step": 706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990453710525442e-05,
+      "loss": 0.1677,
+      "step": 707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904077226625978e-05,
+      "loss": 0.1978,
+      "step": 708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990361624829656e-05,
+      "loss": 0.1688,
+      "step": 709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990315417031734e-05,
+      "loss": 0.1848,
+      "step": 710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990269099273964e-05,
+      "loss": 0.1699,
+      "step": 711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902226715614876e-05,
+      "loss": 0.1689,
+      "step": 712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901761338994606e-05,
+      "loss": 0.1636,
+      "step": 713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901294862930504e-05,
+      "loss": 0.1887,
+      "step": 714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990082728747436e-05,
+      "loss": 0.1914,
+      "step": 715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99003586126781e-05,
+      "loss": 0.1877,
+      "step": 716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899888838593754e-05,
+      "loss": 0.2029,
+      "step": 717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899417965273486e-05,
+      "loss": 0.1611,
+      "step": 718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898945992769584e-05,
+      "loss": 0.1958,
+      "step": 719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989847292113445e-05,
+      "loss": 0.1553,
+      "step": 720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989799875042061e-05,
+      "loss": 0.218,
+      "step": 721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897523480680715e-05,
+      "loss": 0.1609,
+      "step": 722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989704711196754e-05,
+      "loss": 0.207,
+      "step": 723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989656964433397e-05,
+      "loss": 0.1782,
+      "step": 724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989609107783303e-05,
+      "loss": 0.1914,
+      "step": 725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895611412517856e-05,
+      "loss": 0.1833,
+      "step": 726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895130648441706e-05,
+      "loss": 0.1478,
+      "step": 727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989464878565796e-05,
+      "loss": 0.1873,
+      "step": 728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9894165824220125e-05,
+      "loss": 0.1865,
+      "step": 729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893681764181823e-05,
+      "loss": 0.1675,
+      "step": 730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893196605596804e-05,
+      "loss": 0.175,
+      "step": 731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989271034851894e-05,
+      "loss": 0.1704,
+      "step": 732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989222299300222e-05,
+      "loss": 0.1755,
+      "step": 733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891734539100754e-05,
+      "loss": 0.2151,
+      "step": 734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891244986868784e-05,
+      "loss": 0.1748,
+      "step": 735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890754336360666e-05,
+      "loss": 0.2063,
+      "step": 736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890262587630877e-05,
+      "loss": 0.1672,
+      "step": 737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889769740734018e-05,
+      "loss": 0.1797,
+      "step": 738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889275795724815e-05,
+      "loss": 0.1733,
+      "step": 739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888780752658115e-05,
+      "loss": 0.1765,
+      "step": 740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988828461158888e-05,
+      "loss": 0.1682,
+      "step": 741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887787372572203e-05,
+      "loss": 0.1665,
+      "step": 742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887289035663294e-05,
+      "loss": 0.1841,
+      "step": 743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886789600917483e-05,
+      "loss": 0.1694,
+      "step": 744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988628906839023e-05,
+      "loss": 0.1885,
+      "step": 745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988578743813711e-05,
+      "loss": 0.1729,
+      "step": 746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885284710213816e-05,
+      "loss": 0.1697,
+      "step": 747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884780884676177e-05,
+      "loss": 0.1985,
+      "step": 748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884275961580134e-05,
+      "loss": 0.1594,
+      "step": 749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988376994098174e-05,
+      "loss": 0.1821,
+      "step": 750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98832628229372e-05,
+      "loss": 0.1987,
+      "step": 751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882754607502807e-05,
+      "loss": 0.144,
+      "step": 752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882245294734998e-05,
+      "loss": 0.179,
+      "step": 753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881734884690317e-05,
+      "loss": 0.1797,
+      "step": 754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988122337742545e-05,
+      "loss": 0.1882,
+      "step": 755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988071077299718e-05,
+      "loss": 0.1924,
+      "step": 756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988019707146243e-05,
+      "loss": 0.1575,
+      "step": 757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879682272878236e-05,
+      "loss": 0.1672,
+      "step": 758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879166377301768e-05,
+      "loss": 0.2075,
+      "step": 759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878649384790294e-05,
+      "loss": 0.1606,
+      "step": 760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987813129540123e-05,
+      "loss": 0.162,
+      "step": 761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877612109192095e-05,
+      "loss": 0.1699,
+      "step": 762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877091826220543e-05,
+      "loss": 0.168,
+      "step": 763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876570446544335e-05,
+      "loss": 0.1731,
+      "step": 764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876047970221376e-05,
+      "loss": 0.1873,
+      "step": 765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875524397309666e-05,
+      "loss": 0.1714,
+      "step": 766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874999727867347e-05,
+      "loss": 0.176,
+      "step": 767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874473961952672e-05,
+      "loss": 0.1417,
+      "step": 768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873947099624026e-05,
+      "loss": 0.186,
+      "step": 769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873419140939903e-05,
+      "loss": 0.1658,
+      "step": 770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872890085958923e-05,
+      "loss": 0.2009,
+      "step": 771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987235993473984e-05,
+      "loss": 0.167,
+      "step": 772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987182868734151e-05,
+      "loss": 0.1597,
+      "step": 773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987129634382292e-05,
+      "loss": 0.167,
+      "step": 774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870762904243185e-05,
+      "loss": 0.1738,
+      "step": 775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987022836866153e-05,
+      "loss": 0.1777,
+      "step": 776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869692737137312e-05,
+      "loss": 0.1697,
+      "step": 777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869156009730005e-05,
+      "loss": 0.1855,
+      "step": 778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98686181864992e-05,
+      "loss": 0.1846,
+      "step": 779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868079267504615e-05,
+      "loss": 0.1572,
+      "step": 780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867539252806093e-05,
+      "loss": 0.1874,
+      "step": 781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866998142463592e-05,
+      "loss": 0.1659,
+      "step": 782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866455936537195e-05,
+      "loss": 0.1965,
+      "step": 783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865912635087105e-05,
+      "loss": 0.1532,
+      "step": 784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865368238173647e-05,
+      "loss": 0.1829,
+      "step": 785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986482274585727e-05,
+      "loss": 0.188,
+      "step": 786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864276158198544e-05,
+      "loss": 0.1505,
+      "step": 787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863728475258156e-05,
+      "loss": 0.1758,
+      "step": 788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986317969709692e-05,
+      "loss": 0.1794,
+      "step": 789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986262982377577e-05,
+      "loss": 0.1763,
+      "step": 790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862078855355755e-05,
+      "loss": 0.1836,
+      "step": 791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861526791898062e-05,
+      "loss": 0.1609,
+      "step": 792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860973633463986e-05,
+      "loss": 0.1836,
+      "step": 793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860419380114945e-05,
+      "loss": 0.1863,
+      "step": 794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859864031912482e-05,
+      "loss": 0.1694,
+      "step": 795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859307588918258e-05,
+      "loss": 0.186,
+      "step": 796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858750051194062e-05,
+      "loss": 0.1702,
+      "step": 797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858191418801794e-05,
+      "loss": 0.1683,
+      "step": 798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985763169180349e-05,
+      "loss": 0.1897,
+      "step": 799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857070870261292e-05,
+      "loss": 0.1997,
+      "step": 800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856508954237473e-05,
+      "loss": 0.1711,
+      "step": 801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985594594379443e-05,
+      "loss": 0.207,
+      "step": 802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855381838994673e-05,
+      "loss": 0.1729,
+      "step": 803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854816639900837e-05,
+      "loss": 0.1865,
+      "step": 804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854250346575677e-05,
+      "loss": 0.144,
+      "step": 805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853682959082077e-05,
+      "loss": 0.1696,
+      "step": 806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9853114477483036e-05,
+      "loss": 0.155,
+      "step": 807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985254490184167e-05,
+      "loss": 0.1549,
+      "step": 808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851974232221233e-05,
+      "loss": 0.2029,
+      "step": 809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851402468685075e-05,
+      "loss": 0.1653,
+      "step": 810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985082961129669e-05,
+      "loss": 0.2051,
+      "step": 811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850255660119683e-05,
+      "loss": 0.1753,
+      "step": 812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984968061521779e-05,
+      "loss": 0.1692,
+      "step": 813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984910447665485e-05,
+      "loss": 0.1641,
+      "step": 814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848527244494843e-05,
+      "loss": 0.2,
+      "step": 815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847948918801857e-05,
+      "loss": 0.1423,
+      "step": 816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847369499640108e-05,
+      "loss": 0.1543,
+      "step": 817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846788987073933e-05,
+      "loss": 0.1765,
+      "step": 818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984620738116779e-05,
+      "loss": 0.1633,
+      "step": 819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845624681986254e-05,
+      "loss": 0.178,
+      "step": 820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984504088959403e-05,
+      "loss": 0.1738,
+      "step": 821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844456004055935e-05,
+      "loss": 0.1868,
+      "step": 822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843870025436914e-05,
+      "loss": 0.1953,
+      "step": 823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984328295380203e-05,
+      "loss": 0.1887,
+      "step": 824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842694789216473e-05,
+      "loss": 0.2036,
+      "step": 825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842105531745547e-05,
+      "loss": 0.1508,
+      "step": 826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984151518145468e-05,
+      "loss": 0.1682,
+      "step": 827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984092373840942e-05,
+      "loss": 0.1782,
+      "step": 828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840331202675437e-05,
+      "loss": 0.1826,
+      "step": 829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983973757431853e-05,
+      "loss": 0.1653,
+      "step": 830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839142853404606e-05,
+      "loss": 0.1877,
+      "step": 831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838547039999706e-05,
+      "loss": 0.1812,
+      "step": 832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983795013416998e-05,
+      "loss": 0.1792,
+      "step": 833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983735213598171e-05,
+      "loss": 0.1599,
+      "step": 834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836753045501293e-05,
+      "loss": 0.1626,
+      "step": 835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836152862795245e-05,
+      "loss": 0.1731,
+      "step": 836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835551587930217e-05,
+      "loss": 0.1899,
+      "step": 837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834949220972962e-05,
+      "loss": 0.1941,
+      "step": 838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834345761990376e-05,
+      "loss": 0.1609,
+      "step": 839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833741211049448e-05,
+      "loss": 0.1534,
+      "step": 840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833135568217315e-05,
+      "loss": 0.1794,
+      "step": 841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832528833561224e-05,
+      "loss": 0.177,
+      "step": 842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983192100714854e-05,
+      "loss": 0.1853,
+      "step": 843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983131208904676e-05,
+      "loss": 0.1591,
+      "step": 844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983070207932349e-05,
+      "loss": 0.1458,
+      "step": 845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830090978046458e-05,
+      "loss": 0.1833,
+      "step": 846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829478785283527e-05,
+      "loss": 0.1522,
+      "step": 847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982886550110267e-05,
+      "loss": 0.1917,
+      "step": 848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982825112557198e-05,
+      "loss": 0.1741,
+      "step": 849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827635658759673e-05,
+      "loss": 0.1552,
+      "step": 850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982701910073409e-05,
+      "loss": 0.1787,
+      "step": 851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826401451563693e-05,
+      "loss": 0.1736,
+      "step": 852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982578271131706e-05,
+      "loss": 0.1768,
+      "step": 853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825162880062893e-05,
+      "loss": 0.1687,
+      "step": 854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824541957870016e-05,
+      "loss": 0.1714,
+      "step": 855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982391994480737e-05,
+      "loss": 0.1694,
+      "step": 856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823296840944027e-05,
+      "loss": 0.1407,
+      "step": 857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822672646349167e-05,
+      "loss": 0.1685,
+      "step": 858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822047361092096e-05,
+      "loss": 0.189,
+      "step": 859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982142098524225e-05,
+      "loss": 0.1748,
+      "step": 860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820793518869177e-05,
+      "loss": 0.1719,
+      "step": 861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820164962042544e-05,
+      "loss": 0.1594,
+      "step": 862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981953531483215e-05,
+      "loss": 0.1702,
+      "step": 863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818904577307897e-05,
+      "loss": 0.178,
+      "step": 864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818272749539827e-05,
+      "loss": 0.1503,
+      "step": 865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981763983159809e-05,
+      "loss": 0.1943,
+      "step": 866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981700582355297e-05,
+      "loss": 0.1904,
+      "step": 867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981637072547486e-05,
+      "loss": 0.1812,
+      "step": 868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815734537434274e-05,
+      "loss": 0.1628,
+      "step": 869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981509725950186e-05,
+      "loss": 0.1631,
+      "step": 870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981445889174837e-05,
+      "loss": 0.14,
+      "step": 871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813819434244687e-05,
+      "loss": 0.1858,
+      "step": 872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981317888706182e-05,
+      "loss": 0.1772,
+      "step": 873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812537250270882e-05,
+      "loss": 0.1663,
+      "step": 874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811894523943125e-05,
+      "loss": 0.179,
+      "step": 875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981125070814991e-05,
+      "loss": 0.1606,
+      "step": 876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810605802962728e-05,
+      "loss": 0.1567,
+      "step": 877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980995980845318e-05,
+      "loss": 0.1659,
+      "step": 878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809312724692997e-05,
+      "loss": 0.1638,
+      "step": 879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980866455175403e-05,
+      "loss": 0.1809,
+      "step": 880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808015289708248e-05,
+      "loss": 0.1846,
+      "step": 881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807364938627737e-05,
+      "loss": 0.1425,
+      "step": 882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806713498584718e-05,
+      "loss": 0.1599,
+      "step": 883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806060969651514e-05,
+      "loss": 0.1799,
+      "step": 884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805407351900588e-05,
+      "loss": 0.1641,
+      "step": 885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804752645404505e-05,
+      "loss": 0.1687,
+      "step": 886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980409685023597e-05,
+      "loss": 0.1527,
+      "step": 887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803439966467792e-05,
+      "loss": 0.1619,
+      "step": 888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980278199417291e-05,
+      "loss": 0.1748,
+      "step": 889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802122933424387e-05,
+      "loss": 0.1747,
+      "step": 890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801462784295395e-05,
+      "loss": 0.1549,
+      "step": 891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800801546859238e-05,
+      "loss": 0.1562,
+      "step": 892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800139221189332e-05,
+      "loss": 0.1581,
+      "step": 893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979947580735923e-05,
+      "loss": 0.1611,
+      "step": 894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979881130544258e-05,
+      "loss": 0.1758,
+      "step": 895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798145715513168e-05,
+      "loss": 0.1462,
+      "step": 896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797479037644906e-05,
+      "loss": 0.1653,
+      "step": 897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979681127191181e-05,
+      "loss": 0.1542,
+      "step": 898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796142418388035e-05,
+      "loss": 0.1265,
+      "step": 899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795472477147836e-05,
+      "loss": 0.1765,
+      "step": 900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979480144826561e-05,
+      "loss": 0.1658,
+      "step": 901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979412933181586e-05,
+      "loss": 0.2048,
+      "step": 902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979345612787321e-05,
+      "loss": 0.1832,
+      "step": 903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792781836512418e-05,
+      "loss": 0.1545,
+      "step": 904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792106457808348e-05,
+      "loss": 0.1619,
+      "step": 905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791429991835995e-05,
+      "loss": 0.1599,
+      "step": 906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979075243867047e-05,
+      "loss": 0.1677,
+      "step": 907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790073798387003e-05,
+      "loss": 0.1531,
+      "step": 908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789394071060946e-05,
+      "loss": 0.1606,
+      "step": 909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788713256767777e-05,
+      "loss": 0.1569,
+      "step": 910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788031355583085e-05,
+      "loss": 0.1707,
+      "step": 911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787348367582586e-05,
+      "loss": 0.1544,
+      "step": 912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786664292842122e-05,
+      "loss": 0.1892,
+      "step": 913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785979131437646e-05,
+      "loss": 0.1648,
+      "step": 914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978529288344523e-05,
+      "loss": 0.1442,
+      "step": 915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784605548941074e-05,
+      "loss": 0.1301,
+      "step": 916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783917128001503e-05,
+      "loss": 0.1743,
+      "step": 917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783227620702946e-05,
+      "loss": 0.1669,
+      "step": 918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978253702712197e-05,
+      "loss": 0.1462,
+      "step": 919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781845347335253e-05,
+      "loss": 0.1313,
+      "step": 920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781152581419595e-05,
+      "loss": 0.2026,
+      "step": 921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780458729451916e-05,
+      "loss": 0.145,
+      "step": 922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779763791509262e-05,
+      "loss": 0.1558,
+      "step": 923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779067767668794e-05,
+      "loss": 0.1552,
+      "step": 924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778370658007792e-05,
+      "loss": 0.167,
+      "step": 925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977767246260366e-05,
+      "loss": 0.1624,
+      "step": 926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776973181533926e-05,
+      "loss": 0.1415,
+      "step": 927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776272814876235e-05,
+      "loss": 0.1552,
+      "step": 928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775571362708353e-05,
+      "loss": 0.1388,
+      "step": 929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977486882510816e-05,
+      "loss": 0.1755,
+      "step": 930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774165202153665e-05,
+      "loss": 0.1643,
+      "step": 931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773460493922994e-05,
+      "loss": 0.1624,
+      "step": 932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97727547004944e-05,
+      "loss": 0.1567,
+      "step": 933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772047821946242e-05,
+      "loss": 0.1638,
+      "step": 934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977133985835702e-05,
+      "loss": 0.1478,
+      "step": 935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977063080980533e-05,
+      "loss": 0.1655,
+      "step": 936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976992067636991e-05,
+      "loss": 0.1819,
+      "step": 937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976920945812961e-05,
+      "loss": 0.1606,
+      "step": 938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768497155163392e-05,
+      "loss": 0.1643,
+      "step": 939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767783767550358e-05,
+      "loss": 0.1611,
+      "step": 940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767069295369707e-05,
+      "loss": 0.1687,
+      "step": 941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976635373870078e-05,
+      "loss": 0.1777,
+      "step": 942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765637097623028e-05,
+      "loss": 0.1829,
+      "step": 943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976491937221602e-05,
+      "loss": 0.1462,
+      "step": 944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976420056255945e-05,
+      "loss": 0.1724,
+      "step": 945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763480668733132e-05,
+      "loss": 0.198,
+      "step": 946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762759690816996e-05,
+      "loss": 0.1514,
+      "step": 947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762037628891097e-05,
+      "loss": 0.1633,
+      "step": 948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761314483035617e-05,
+      "loss": 0.1621,
+      "step": 949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976059025333084e-05,
+      "loss": 0.1469,
+      "step": 950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759864939857186e-05,
+      "loss": 0.1516,
+      "step": 951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975913854269519e-05,
+      "loss": 0.1724,
+      "step": 952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975841106192551e-05,
+      "loss": 0.1831,
+      "step": 953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757682497628915e-05,
+      "loss": 0.177,
+      "step": 954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975695284988631e-05,
+      "loss": 0.1655,
+      "step": 955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756222118778704e-05,
+      "loss": 0.145,
+      "step": 956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755490304387236e-05,
+      "loss": 0.179,
+      "step": 957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754757406793172e-05,
+      "loss": 0.1582,
+      "step": 958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975402342607787e-05,
+      "loss": 0.1829,
+      "step": 959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975328836232285e-05,
+      "loss": 0.1929,
+      "step": 960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752552215609713e-05,
+      "loss": 0.1516,
+      "step": 961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751814986020203e-05,
+      "loss": 0.1665,
+      "step": 962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975107667363618e-05,
+      "loss": 0.1711,
+      "step": 963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750337278539623e-05,
+      "loss": 0.1602,
+      "step": 964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974959680081263e-05,
+      "loss": 0.1616,
+      "step": 965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748855240537418e-05,
+      "loss": 0.1755,
+      "step": 966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974811259779633e-05,
+      "loss": 0.1733,
+      "step": 967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974736887267182e-05,
+      "loss": 0.1787,
+      "step": 968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974662406524647e-05,
+      "loss": 0.1464,
+      "step": 969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745878175602984e-05,
+      "loss": 0.1768,
+      "step": 970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745131203824177e-05,
+      "loss": 0.1628,
+      "step": 971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974438314999299e-05,
+      "loss": 0.155,
+      "step": 972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743634014192486e-05,
+      "loss": 0.1545,
+      "step": 973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742883796505843e-05,
+      "loss": 0.1555,
+      "step": 974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974213249701636e-05,
+      "loss": 0.1685,
+      "step": 975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974138011580746e-05,
+      "loss": 0.1497,
+      "step": 976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974062665296269e-05,
+      "loss": 0.1201,
+      "step": 977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739872108565697e-05,
+      "loss": 0.1667,
+      "step": 978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973911648270027e-05,
+      "loss": 0.1355,
+      "step": 979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738359775450313e-05,
+      "loss": 0.1281,
+      "step": 980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973760198689984e-05,
+      "loss": 0.1534,
+      "step": 981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736843117132996e-05,
+      "loss": 0.1604,
+      "step": 982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736083166234047e-05,
+      "loss": 0.1638,
+      "step": 983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735322134287364e-05,
+      "loss": 0.1561,
+      "step": 984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734560021377454e-05,
+      "loss": 0.1536,
+      "step": 985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973379682758894e-05,
+      "loss": 0.1572,
+      "step": 986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973303255300656e-05,
+      "loss": 0.1479,
+      "step": 987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732267197715176e-05,
+      "loss": 0.1541,
+      "step": 988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973150076179977e-05,
+      "loss": 0.178,
+      "step": 989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730733245345445e-05,
+      "loss": 0.1702,
+      "step": 990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972996464843742e-05,
+      "loss": 0.1451,
+      "step": 991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972919497116104e-05,
+      "loss": 0.1621,
+      "step": 992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728424213601758e-05,
+      "loss": 0.1562,
+      "step": 993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727652375845164e-05,
+      "loss": 0.1506,
+      "step": 994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726879457976954e-05,
+      "loss": 0.1533,
+      "step": 995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972610546008295e-05,
+      "loss": 0.1606,
+      "step": 996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725330382249094e-05,
+      "loss": 0.1558,
+      "step": 997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972455422456145e-05,
+      "loss": 0.146,
+      "step": 998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723776987106193e-05,
+      "loss": 0.1458,
+      "step": 999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722998669969626e-05,
+      "loss": 0.1763,
+      "step": 1000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722219273238166e-05,
+      "loss": 0.1444,
+      "step": 1001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972143879699836e-05,
+      "loss": 0.1943,
+      "step": 1002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720657241336866e-05,
+      "loss": 0.1575,
+      "step": 1003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971987460634046e-05,
+      "loss": 0.1729,
+      "step": 1004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971909089209605e-05,
+      "loss": 0.1553,
+      "step": 1005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971830609869065e-05,
+      "loss": 0.1375,
+      "step": 1006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97175202262114e-05,
+      "loss": 0.1754,
+      "step": 1007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716733274745566e-05,
+      "loss": 0.1448,
+      "step": 1008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715945244380513e-05,
+      "loss": 0.1431,
+      "step": 1009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971515613520376e-05,
+      "loss": 0.1643,
+      "step": 1010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714365947302905e-05,
+      "loss": 0.1868,
+      "step": 1011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97135746807657e-05,
+      "loss": 0.1321,
+      "step": 1012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971278233568e-05,
+      "loss": 0.142,
+      "step": 1013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711988912133783e-05,
+      "loss": 0.1624,
+      "step": 1014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711194410215148e-05,
+      "loss": 0.1565,
+      "step": 1015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710398830012313e-05,
+      "loss": 0.1677,
+      "step": 1016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970960217161361e-05,
+      "loss": 0.1434,
+      "step": 1017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708804435107504e-05,
+      "loss": 0.1709,
+      "step": 1018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708005620582564e-05,
+      "loss": 0.1509,
+      "step": 1019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707205728127496e-05,
+      "loss": 0.1589,
+      "step": 1020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706404757831104e-05,
+      "loss": 0.1531,
+      "step": 1021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705602709782336e-05,
+      "loss": 0.1252,
+      "step": 1022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970479958407024e-05,
+      "loss": 0.1736,
+      "step": 1023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703995380783993e-05,
+      "loss": 0.1641,
+      "step": 1024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970319010001289e-05,
+      "loss": 0.1484,
+      "step": 1025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702383741846346e-05,
+      "loss": 0.1643,
+      "step": 1026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701576306373896e-05,
+      "loss": 0.1604,
+      "step": 1027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700767793685195e-05,
+      "loss": 0.1462,
+      "step": 1028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969995820387001e-05,
+      "loss": 0.1346,
+      "step": 1029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969914753701824e-05,
+      "loss": 0.158,
+      "step": 1030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969833579321989e-05,
+      "loss": 0.1831,
+      "step": 1031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697522972565103e-05,
+      "loss": 0.1715,
+      "step": 1032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696709075144123e-05,
+      "loss": 0.1672,
+      "step": 1033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695894101047327e-05,
+      "loss": 0.1484,
+      "step": 1034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96950780503652e-05,
+      "loss": 0.1721,
+      "step": 1035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694260923188354e-05,
+      "loss": 0.1351,
+      "step": 1036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969344271960752e-05,
+      "loss": 0.1254,
+      "step": 1037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692623439713547e-05,
+      "loss": 0.1631,
+      "step": 1038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691803083597403e-05,
+      "loss": 0.1492,
+      "step": 1039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969098165135018e-05,
+      "loss": 0.1489,
+      "step": 1040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969015914306308e-05,
+      "loss": 0.1675,
+      "step": 1041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689335558827433e-05,
+      "loss": 0.167,
+      "step": 1042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688510898734687e-05,
+      "loss": 0.1532,
+      "step": 1043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687685162876406e-05,
+      "loss": 0.1594,
+      "step": 1044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686858351344284e-05,
+      "loss": 0.1785,
+      "step": 1045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968603046423011e-05,
+      "loss": 0.1697,
+      "step": 1046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685201501625822e-05,
+      "loss": 0.1453,
+      "step": 1047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968437146362346e-05,
+      "loss": 0.1473,
+      "step": 1048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968354035031519e-05,
+      "loss": 0.1287,
+      "step": 1049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682708161793287e-05,
+      "loss": 0.188,
+      "step": 1050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968187489815016e-05,
+      "loss": 0.1321,
+      "step": 1051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968104055947833e-05,
+      "loss": 0.1455,
+      "step": 1052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968020514587044e-05,
+      "loss": 0.158,
+      "step": 1053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967936865741924e-05,
+      "loss": 0.1624,
+      "step": 1054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678531094217622e-05,
+      "loss": 0.1655,
+      "step": 1055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967769245635858e-05,
+      "loss": 0.1516,
+      "step": 1056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967685274393523e-05,
+      "loss": 0.1333,
+      "step": 1057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676011957040812e-05,
+      "loss": 0.1516,
+      "step": 1058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675170095768685e-05,
+      "loss": 0.1572,
+      "step": 1059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967432716021232e-05,
+      "loss": 0.1528,
+      "step": 1060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673483150465314e-05,
+      "loss": 0.1641,
+      "step": 1061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967263806662139e-05,
+      "loss": 0.1495,
+      "step": 1062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967179190877437e-05,
+      "loss": 0.1377,
+      "step": 1063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670944677018214e-05,
+      "loss": 0.1582,
+      "step": 1064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670096371446992e-05,
+      "loss": 0.168,
+      "step": 1065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96692469921549e-05,
+      "loss": 0.1699,
+      "step": 1066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966839653923624e-05,
+      "loss": 0.1409,
+      "step": 1067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667545012785448e-05,
+      "loss": 0.1184,
+      "step": 1068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966669241289708e-05,
+      "loss": 0.1292,
+      "step": 1069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665838739665793e-05,
+      "loss": 0.1611,
+      "step": 1070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966498399318638e-05,
+      "loss": 0.145,
+      "step": 1071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664128173553748e-05,
+      "loss": 0.1692,
+      "step": 1072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663271280862924e-05,
+      "loss": 0.1621,
+      "step": 1073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966241331520905e-05,
+      "loss": 0.1302,
+      "step": 1074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661554276687394e-05,
+      "loss": 0.1714,
+      "step": 1075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660694165393334e-05,
+      "loss": 0.1536,
+      "step": 1076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659832981422383e-05,
+      "loss": 0.1655,
+      "step": 1077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658970724870153e-05,
+      "loss": 0.1555,
+      "step": 1078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9658107395832387e-05,
+      "loss": 0.1361,
+      "step": 1079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657242994404947e-05,
+      "loss": 0.1614,
+      "step": 1080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656377520683808e-05,
+      "loss": 0.1509,
+      "step": 1081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655510974765074e-05,
+      "loss": 0.1321,
+      "step": 1082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965464335674496e-05,
+      "loss": 0.1389,
+      "step": 1083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653774666719796e-05,
+      "loss": 0.1544,
+      "step": 1084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652904904786046e-05,
+      "loss": 0.1478,
+      "step": 1085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652034071040278e-05,
+      "loss": 0.1716,
+      "step": 1086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651162165579188e-05,
+      "loss": 0.1899,
+      "step": 1087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650289188499587e-05,
+      "loss": 0.1627,
+      "step": 1088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649415139898407e-05,
+      "loss": 0.1865,
+      "step": 1089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96485400198727e-05,
+      "loss": 0.1587,
+      "step": 1090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647663828519633e-05,
+      "loss": 0.1711,
+      "step": 1091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646786565936497e-05,
+      "loss": 0.1538,
+      "step": 1092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645908232220692e-05,
+      "loss": 0.1792,
+      "step": 1093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645028827469748e-05,
+      "loss": 0.1465,
+      "step": 1094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964414835178131e-05,
+      "loss": 0.1495,
+      "step": 1095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964326680525314e-05,
+      "loss": 0.1594,
+      "step": 1096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642384187983126e-05,
+      "loss": 0.1746,
+      "step": 1097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641500500069267e-05,
+      "loss": 0.1559,
+      "step": 1098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964061574160968e-05,
+      "loss": 0.1353,
+      "step": 1099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963972991270261e-05,
+      "loss": 0.1431,
+      "step": 1100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638843013446408e-05,
+      "loss": 0.1572,
+      "step": 1101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637955043939554e-05,
+      "loss": 0.1381,
+      "step": 1102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637066004280646e-05,
+      "loss": 0.1439,
+      "step": 1103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636175894568397e-05,
+      "loss": 0.131,
+      "step": 1104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635284714901646e-05,
+      "loss": 0.1401,
+      "step": 1105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634392465379337e-05,
+      "loss": 0.1562,
+      "step": 1106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963349914610054e-05,
+      "loss": 0.1516,
+      "step": 1107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632604757164456e-05,
+      "loss": 0.1313,
+      "step": 1108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631709298670382e-05,
+      "loss": 0.1332,
+      "step": 1109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630812770717753e-05,
+      "loss": 0.166,
+      "step": 1110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962991517340611e-05,
+      "loss": 0.1505,
+      "step": 1111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629016506835122e-05,
+      "loss": 0.1477,
+      "step": 1112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962811677110457e-05,
+      "loss": 0.1345,
+      "step": 1113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962721596631436e-05,
+      "loss": 0.1558,
+      "step": 1114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626314092564506e-05,
+      "loss": 0.1646,
+      "step": 1115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625411149955156e-05,
+      "loss": 0.178,
+      "step": 1116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962450713858656e-05,
+      "loss": 0.1409,
+      "step": 1117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623602058559103e-05,
+      "loss": 0.1694,
+      "step": 1118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622695909973276e-05,
+      "loss": 0.1537,
+      "step": 1119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621788692929695e-05,
+      "loss": 0.1519,
+      "step": 1120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620880407529092e-05,
+      "loss": 0.1676,
+      "step": 1121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619971053872318e-05,
+      "loss": 0.1266,
+      "step": 1122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619060632060343e-05,
+      "loss": 0.1492,
+      "step": 1123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618149142194262e-05,
+      "loss": 0.1298,
+      "step": 1124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617236584375275e-05,
+      "loss": 0.1428,
+      "step": 1125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616322958704708e-05,
+      "loss": 0.1545,
+      "step": 1126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961540826528401e-05,
+      "loss": 0.1655,
+      "step": 1127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614492504214744e-05,
+      "loss": 0.146,
+      "step": 1128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613575675598588e-05,
+      "loss": 0.1355,
+      "step": 1129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961265777953735e-05,
+      "loss": 0.156,
+      "step": 1130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611738816132936e-05,
+      "loss": 0.1536,
+      "step": 1131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610818785487392e-05,
+      "loss": 0.1527,
+      "step": 1132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609897687702874e-05,
+      "loss": 0.1449,
+      "step": 1133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960897552288165e-05,
+      "loss": 0.1476,
+      "step": 1134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608052291126123e-05,
+      "loss": 0.1589,
+      "step": 1135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607127992538796e-05,
+      "loss": 0.1641,
+      "step": 1136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606202627222298e-05,
+      "loss": 0.1469,
+      "step": 1137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605276195279385e-05,
+      "loss": 0.1672,
+      "step": 1138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604348696812917e-05,
+      "loss": 0.1541,
+      "step": 1139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960342013192588e-05,
+      "loss": 0.1323,
+      "step": 1140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602490500721375e-05,
+      "loss": 0.1412,
+      "step": 1141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960155980330263e-05,
+      "loss": 0.1536,
+      "step": 1142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960062803977298e-05,
+      "loss": 0.1294,
+      "step": 1143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599695210235886e-05,
+      "loss": 0.1263,
+      "step": 1144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959876131479493e-05,
+      "loss": 0.1316,
+      "step": 1145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597826353553794e-05,
+      "loss": 0.1604,
+      "step": 1146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95968903266163e-05,
+      "loss": 0.1345,
+      "step": 1147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959595323408638e-05,
+      "loss": 0.1582,
+      "step": 1148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959501507606808e-05,
+      "loss": 0.1554,
+      "step": 1149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959407585266558e-05,
+      "loss": 0.1371,
+      "step": 1150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593135563983152e-05,
+      "loss": 0.153,
+      "step": 1151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959219421012521e-05,
+      "loss": 0.1398,
+      "step": 1152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591251791196274e-05,
+      "loss": 0.1395,
+      "step": 1153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590308307300988e-05,
+      "loss": 0.1583,
+      "step": 1154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589363758544108e-05,
+      "loss": 0.1709,
+      "step": 1155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958841814503052e-05,
+      "loss": 0.1743,
+      "step": 1156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958747146686521e-05,
+      "loss": 0.1499,
+      "step": 1157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95865237241533e-05,
+      "loss": 0.1416,
+      "step": 1158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958557491700002e-05,
+      "loss": 0.157,
+      "step": 1159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584625045510725e-05,
+      "loss": 0.1694,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583674109790878e-05,
+      "loss": 0.1467,
+      "step": 1161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582722109946067e-05,
+      "loss": 0.1354,
+      "step": 1162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581769046082002e-05,
+      "loss": 0.1394,
+      "step": 1163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580814918304504e-05,
+      "loss": 0.1633,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579859726719513e-05,
+      "loss": 0.1458,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957890347143309e-05,
+      "loss": 0.1497,
+      "step": 1166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577946152551417e-05,
+      "loss": 0.1172,
+      "step": 1167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576987770180788e-05,
+      "loss": 0.1343,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576028324427612e-05,
+      "loss": 0.1484,
+      "step": 1169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575067815398423e-05,
+      "loss": 0.1587,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957410624319988e-05,
+      "loss": 0.1459,
+      "step": 1171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957314360793874e-05,
+      "loss": 0.1455,
+      "step": 1172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572179909721894e-05,
+      "loss": 0.1426,
+      "step": 1173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957121514865635e-05,
+      "loss": 0.1477,
+      "step": 1174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570249324849223e-05,
+      "loss": 0.1631,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569282438407763e-05,
+      "loss": 0.1471,
+      "step": 1176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956831448943932e-05,
+      "loss": 0.1538,
+      "step": 1177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956734547805137e-05,
+      "loss": 0.1428,
+      "step": 1178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956637540435151e-05,
+      "loss": 0.1573,
+      "step": 1179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956540426844746e-05,
+      "loss": 0.1545,
+      "step": 1180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564432070447035e-05,
+      "loss": 0.132,
+      "step": 1181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563458810458195e-05,
+      "loss": 0.1398,
+      "step": 1182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562484488589005e-05,
+      "loss": 0.1642,
+      "step": 1183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561509104947643e-05,
+      "loss": 0.1215,
+      "step": 1184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560532659642413e-05,
+      "loss": 0.1416,
+      "step": 1185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955955515278174e-05,
+      "loss": 0.1334,
+      "step": 1186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558576584474154e-05,
+      "loss": 0.142,
+      "step": 1187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557596954828315e-05,
+      "loss": 0.1414,
+      "step": 1188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556616263953e-05,
+      "loss": 0.123,
+      "step": 1189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955563451195709e-05,
+      "loss": 0.1429,
+      "step": 1190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554651698949603e-05,
+      "loss": 0.1469,
+      "step": 1191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955366782503966e-05,
+      "loss": 0.1526,
+      "step": 1192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552682890336508e-05,
+      "loss": 0.1255,
+      "step": 1193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551696894949513e-05,
+      "loss": 0.155,
+      "step": 1194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955070983898815e-05,
+      "loss": 0.1473,
+      "step": 1195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954972172256202e-05,
+      "loss": 0.1489,
+      "step": 1196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548732545780833e-05,
+      "loss": 0.1592,
+      "step": 1197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954774230875443e-05,
+      "loss": 0.1393,
+      "step": 1198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954675101159276e-05,
+      "loss": 0.1268,
+      "step": 1199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545758654405888e-05,
+      "loss": 0.1389,
+      "step": 1200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544765237304006e-05,
+      "loss": 0.1416,
+      "step": 1201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543770760397413e-05,
+      "loss": 0.1509,
+      "step": 1202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542775223796534e-05,
+      "loss": 0.1392,
+      "step": 1203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541778627611908e-05,
+      "loss": 0.1335,
+      "step": 1204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540780971954193e-05,
+      "loss": 0.1333,
+      "step": 1205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539782256934166e-05,
+      "loss": 0.129,
+      "step": 1206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953878248266271e-05,
+      "loss": 0.1395,
+      "step": 1207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537781649250848e-05,
+      "loss": 0.14,
+      "step": 1208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95367797568097e-05,
+      "loss": 0.1643,
+      "step": 1209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535776805450512e-05,
+      "loss": 0.1519,
+      "step": 1210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953477279528465e-05,
+      "loss": 0.1174,
+      "step": 1211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533767726423586e-05,
+      "loss": 0.1185,
+      "step": 1212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532761598978932e-05,
+      "loss": 0.1638,
+      "step": 1213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531754413062392e-05,
+      "loss": 0.1255,
+      "step": 1214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95307461687858e-05,
+      "loss": 0.1608,
+      "step": 1215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529736866261112e-05,
+      "loss": 0.1512,
+      "step": 1216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528726505600396e-05,
+      "loss": 0.1445,
+      "step": 1217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952771508691583e-05,
+      "loss": 0.1473,
+      "step": 1218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526702610319727e-05,
+      "loss": 0.1326,
+      "step": 1219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525689075924498e-05,
+      "loss": 0.1471,
+      "step": 1220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524674483842687e-05,
+      "loss": 0.1631,
+      "step": 1221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952365883418695e-05,
+      "loss": 0.1571,
+      "step": 1222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522642127070057e-05,
+      "loss": 0.1589,
+      "step": 1223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521624362604896e-05,
+      "loss": 0.1385,
+      "step": 1224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952060554090448e-05,
+      "loss": 0.1711,
+      "step": 1225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519585662081932e-05,
+      "loss": 0.1208,
+      "step": 1226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518564726250496e-05,
+      "loss": 0.1487,
+      "step": 1227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517542733523528e-05,
+      "loss": 0.1677,
+      "step": 1228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516519684014505e-05,
+      "loss": 0.1527,
+      "step": 1229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515495577837026e-05,
+      "loss": 0.1418,
+      "step": 1230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514470415104802e-05,
+      "loss": 0.1575,
+      "step": 1231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951344419593166e-05,
+      "loss": 0.1455,
+      "step": 1232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512416920431544e-05,
+      "loss": 0.1458,
+      "step": 1233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511388588718522e-05,
+      "loss": 0.1351,
+      "step": 1234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510359200906776e-05,
+      "loss": 0.1509,
+      "step": 1235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509328757110598e-05,
+      "loss": 0.1779,
+      "step": 1236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508297257444408e-05,
+      "loss": 0.1438,
+      "step": 1237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950726470202274e-05,
+      "loss": 0.1611,
+      "step": 1238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506231090960244e-05,
+      "loss": 0.1332,
+      "step": 1239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505196424371685e-05,
+      "loss": 0.1624,
+      "step": 1240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504160702371947e-05,
+      "loss": 0.1458,
+      "step": 1241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503123925076028e-05,
+      "loss": 0.1572,
+      "step": 1242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502086092599054e-05,
+      "loss": 0.129,
+      "step": 1243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501047205056262e-05,
+      "loss": 0.1187,
+      "step": 1244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500007262562994e-05,
+      "loss": 0.1378,
+      "step": 1245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498966265234735e-05,
+      "loss": 0.1312,
+      "step": 1246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497924213187057e-05,
+      "loss": 0.1719,
+      "step": 1247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496881106535675e-05,
+      "loss": 0.1675,
+      "step": 1248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495836945396413e-05,
+      "loss": 0.1428,
+      "step": 1249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494791729885198e-05,
+      "loss": 0.1462,
+      "step": 1250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949374546011809e-05,
+      "loss": 0.1453,
+      "step": 1251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949269813621127e-05,
+      "loss": 0.1466,
+      "step": 1252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491649758281017e-05,
+      "loss": 0.1462,
+      "step": 1253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490600326443743e-05,
+      "loss": 0.131,
+      "step": 1254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489549840815974e-05,
+      "loss": 0.155,
+      "step": 1255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488498301514343e-05,
+      "loss": 0.111,
+      "step": 1256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487445708655616e-05,
+      "loss": 0.1492,
+      "step": 1257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486392062356663e-05,
+      "loss": 0.1407,
+      "step": 1258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948533736273448e-05,
+      "loss": 0.1099,
+      "step": 1259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484281609906172e-05,
+      "loss": 0.1733,
+      "step": 1260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483224803988965e-05,
+      "loss": 0.1465,
+      "step": 1261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482166945100204e-05,
+      "loss": 0.1456,
+      "step": 1262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481108033357344e-05,
+      "loss": 0.131,
+      "step": 1263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948004806887797e-05,
+      "loss": 0.1298,
+      "step": 1264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478987051779767e-05,
+      "loss": 0.126,
+      "step": 1265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477924982180548e-05,
+      "loss": 0.1687,
+      "step": 1266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476861860198238e-05,
+      "loss": 0.1716,
+      "step": 1267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475797685950885e-05,
+      "loss": 0.1487,
+      "step": 1268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474732459556647e-05,
+      "loss": 0.1647,
+      "step": 1269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473666181133805e-05,
+      "loss": 0.1536,
+      "step": 1270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947259885080075e-05,
+      "loss": 0.1179,
+      "step": 1271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471530468675995e-05,
+      "loss": 0.1465,
+      "step": 1272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470461034878167e-05,
+      "loss": 0.1667,
+      "step": 1273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469390549526007e-05,
+      "loss": 0.1643,
+      "step": 1274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468319012738383e-05,
+      "loss": 0.1511,
+      "step": 1275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946724642463427e-05,
+      "loss": 0.1395,
+      "step": 1276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466172785332767e-05,
+      "loss": 0.147,
+      "step": 1277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946509809495308e-05,
+      "loss": 0.1057,
+      "step": 1278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946402235361454e-05,
+      "loss": 0.1415,
+      "step": 1279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946294556143659e-05,
+      "loss": 0.1553,
+      "step": 1280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94618677185388e-05,
+      "loss": 0.1344,
+      "step": 1281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460788825040837e-05,
+      "loss": 0.1388,
+      "step": 1282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459708881062504e-05,
+      "loss": 0.1243,
+      "step": 1283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458627886723706e-05,
+      "loss": 0.1398,
+      "step": 1284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945754584214448e-05,
+      "loss": 0.1682,
+      "step": 1285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456462747444965e-05,
+      "loss": 0.1416,
+      "step": 1286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455378602745426e-05,
+      "loss": 0.1453,
+      "step": 1287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945429340816624e-05,
+      "loss": 0.177,
+      "step": 1288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94532071638279e-05,
+      "loss": 0.104,
+      "step": 1289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452119869851014e-05,
+      "loss": 0.1093,
+      "step": 1290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451031526356318e-05,
+      "loss": 0.1475,
+      "step": 1291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449942133464654e-05,
+      "loss": 0.1357,
+      "step": 1292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944885169129698e-05,
+      "loss": 0.1619,
+      "step": 1293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447760199974376e-05,
+      "loss": 0.1533,
+      "step": 1294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944666765961804e-05,
+      "loss": 0.135,
+      "step": 1295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445574070349272e-05,
+      "loss": 0.1516,
+      "step": 1296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444479432289505e-05,
+      "loss": 0.1538,
+      "step": 1297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944338374556028e-05,
+      "loss": 0.1494,
+      "step": 1298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442287010283264e-05,
+      "loss": 0.1422,
+      "step": 1299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441189226580225e-05,
+      "loss": 0.166,
+      "step": 1300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440090394573056e-05,
+      "loss": 0.1348,
+      "step": 1301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943899051438377e-05,
+      "loss": 0.1475,
+      "step": 1302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437889586134493e-05,
+      "loss": 0.1399,
+      "step": 1303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436787609947464e-05,
+      "loss": 0.1512,
+      "step": 1304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435684585945037e-05,
+      "loss": 0.1553,
+      "step": 1305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434580514249698e-05,
+      "loss": 0.1393,
+      "step": 1306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433475394984028e-05,
+      "loss": 0.1498,
+      "step": 1307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432369228270733e-05,
+      "loss": 0.1207,
+      "step": 1308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431262014232645e-05,
+      "loss": 0.1267,
+      "step": 1309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94301537529927e-05,
+      "loss": 0.1672,
+      "step": 1310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942904444467395e-05,
+      "loss": 0.1389,
+      "step": 1311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942793408939957e-05,
+      "loss": 0.125,
+      "step": 1312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426822687292852e-05,
+      "loss": 0.1472,
+      "step": 1313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425710238477197e-05,
+      "loss": 0.1338,
+      "step": 1314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424596743076125e-05,
+      "loss": 0.1635,
+      "step": 1315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423482201213275e-05,
+      "loss": 0.1331,
+      "step": 1316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422366613012404e-05,
+      "loss": 0.1305,
+      "step": 1317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421249978597375e-05,
+      "loss": 0.1467,
+      "step": 1318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420132298092173e-05,
+      "loss": 0.1609,
+      "step": 1319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419013571620907e-05,
+      "loss": 0.1226,
+      "step": 1320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941789379930779e-05,
+      "loss": 0.1443,
+      "step": 1321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416772981277156e-05,
+      "loss": 0.1765,
+      "step": 1322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941565111765346e-05,
+      "loss": 0.155,
+      "step": 1323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414528208561262e-05,
+      "loss": 0.1324,
+      "step": 1324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413404254125246e-05,
+      "loss": 0.1697,
+      "step": 1325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412279254470215e-05,
+      "loss": 0.1389,
+      "step": 1326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411153209721078e-05,
+      "loss": 0.1404,
+      "step": 1327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941002612000287e-05,
+      "loss": 0.147,
+      "step": 1328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408897985440735e-05,
+      "loss": 0.1716,
+      "step": 1329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407768806159935e-05,
+      "loss": 0.1152,
+      "step": 1330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940663858228585e-05,
+      "loss": 0.1537,
+      "step": 1331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405507313943975e-05,
+      "loss": 0.1427,
+      "step": 1332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404375001259916e-05,
+      "loss": 0.1427,
+      "step": 1333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940324164435941e-05,
+      "loss": 0.1445,
+      "step": 1334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402107243368288e-05,
+      "loss": 0.1321,
+      "step": 1335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400971798412514e-05,
+      "loss": 0.166,
+      "step": 1336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399835309618165e-05,
+      "loss": 0.1401,
+      "step": 1337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398697777111427e-05,
+      "loss": 0.1509,
+      "step": 1338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397559201018604e-05,
+      "loss": 0.1364,
+      "step": 1339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939641958146612e-05,
+      "loss": 0.1506,
+      "step": 1340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939527891858052e-05,
+      "loss": 0.1472,
+      "step": 1341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939413721248845e-05,
+      "loss": 0.1484,
+      "step": 1342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392994463316677e-05,
+      "loss": 0.157,
+      "step": 1343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391850671192092e-05,
+      "loss": 0.1162,
+      "step": 1344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390705836241698e-05,
+      "loss": 0.1572,
+      "step": 1345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389559958592607e-05,
+      "loss": 0.1449,
+      "step": 1346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938841303837205e-05,
+      "loss": 0.1414,
+      "step": 1347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938726507570738e-05,
+      "loss": 0.1298,
+      "step": 1348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386116070726063e-05,
+      "loss": 0.1245,
+      "step": 1349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384966023555673e-05,
+      "loss": 0.1396,
+      "step": 1350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938381493432391e-05,
+      "loss": 0.1655,
+      "step": 1351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382662803158585e-05,
+      "loss": 0.1283,
+      "step": 1352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381509630187626e-05,
+      "loss": 0.1161,
+      "step": 1353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938035541553907e-05,
+      "loss": 0.1371,
+      "step": 1354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937920015934108e-05,
+      "loss": 0.1368,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937804386172193e-05,
+      "loss": 0.1254,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937688652281001e-05,
+      "loss": 0.1378,
+      "step": 1357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375728142733825e-05,
+      "loss": 0.1271,
+      "step": 1358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374568721621996e-05,
+      "loss": 0.1443,
+      "step": 1359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373408259603254e-05,
+      "loss": 0.1576,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9372246756806462e-05,
+      "loss": 0.1556,
+      "step": 1361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937108421336058e-05,
+      "loss": 0.1332,
+      "step": 1362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369920629394693e-05,
+      "loss": 0.1475,
+      "step": 1363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368756005038e-05,
+      "loss": 0.1636,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367590340419814e-05,
+      "loss": 0.1338,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366423635669568e-05,
+      "loss": 0.1343,
+      "step": 1366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365255890916802e-05,
+      "loss": 0.1121,
+      "step": 1367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936408710629118e-05,
+      "loss": 0.1362,
+      "step": 1368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936291728192248e-05,
+      "loss": 0.1249,
+      "step": 1369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9361746417940592e-05,
+      "loss": 0.1178,
+      "step": 1370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360574514475518e-05,
+      "loss": 0.1309,
+      "step": 1371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935940157165739e-05,
+      "loss": 0.1445,
+      "step": 1372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935822758961644e-05,
+      "loss": 0.119,
+      "step": 1373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357052568483022e-05,
+      "loss": 0.1345,
+      "step": 1374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355876508387606e-05,
+      "loss": 0.1072,
+      "step": 1375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354699409460775e-05,
+      "loss": 0.1395,
+      "step": 1376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935352127183323e-05,
+      "loss": 0.1411,
+      "step": 1377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352342095635782e-05,
+      "loss": 0.1354,
+      "step": 1378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351161880999363e-05,
+      "loss": 0.1118,
+      "step": 1379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9349980628055023e-05,
+      "loss": 0.1714,
+      "step": 1380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348798336933916e-05,
+      "loss": 0.1317,
+      "step": 1381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934761500776732e-05,
+      "loss": 0.1328,
+      "step": 1382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346430640686625e-05,
+      "loss": 0.1602,
+      "step": 1383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345245235823343e-05,
+      "loss": 0.1472,
+      "step": 1384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344058793309085e-05,
+      "loss": 0.1471,
+      "step": 1385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93428713132756e-05,
+      "loss": 0.1339,
+      "step": 1386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934168279585473e-05,
+      "loss": 0.1334,
+      "step": 1387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340493241178452e-05,
+      "loss": 0.1349,
+      "step": 1388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933930264937884e-05,
+      "loss": 0.1191,
+      "step": 1389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9338111020588092e-05,
+      "loss": 0.1301,
+      "step": 1390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933691835493853e-05,
+      "loss": 0.119,
+      "step": 1391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933572465256257e-05,
+      "loss": 0.1384,
+      "step": 1392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933452991359276e-05,
+      "loss": 0.1396,
+      "step": 1393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933333413816176e-05,
+      "loss": 0.1229,
+      "step": 1394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933213732640234e-05,
+      "loss": 0.1285,
+      "step": 1395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330939478447392e-05,
+      "loss": 0.1456,
+      "step": 1396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329740594429913e-05,
+      "loss": 0.1377,
+      "step": 1397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932854067448303e-05,
+      "loss": 0.1295,
+      "step": 1398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932733971873997e-05,
+      "loss": 0.1331,
+      "step": 1399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932613772733408e-05,
+      "loss": 0.1428,
+      "step": 1400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324934700398827e-05,
+      "loss": 0.1343,
+      "step": 1401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932373063806779e-05,
+      "loss": 0.1567,
+      "step": 1402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932252554047466e-05,
+      "loss": 0.135,
+      "step": 1403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321319407753244e-05,
+      "loss": 0.1241,
+      "step": 1404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320112240037466e-05,
+      "loss": 0.1177,
+      "step": 1405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931890403746137e-05,
+      "loss": 0.1299,
+      "step": 1406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317694800159103e-05,
+      "loss": 0.146,
+      "step": 1407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316484528264932e-05,
+      "loss": 0.1361,
+      "step": 1408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931527322191324e-05,
+      "loss": 0.14,
+      "step": 1409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9314060881238532e-05,
+      "loss": 0.1384,
+      "step": 1410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312847506375413e-05,
+      "loss": 0.1677,
+      "step": 1411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311633097458608e-05,
+      "loss": 0.1296,
+      "step": 1412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931041765462297e-05,
+      "loss": 0.1427,
+      "step": 1413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309201178003443e-05,
+      "loss": 0.1475,
+      "step": 1414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307983667735106e-05,
+      "loss": 0.1252,
+      "step": 1415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930676512395315e-05,
+      "loss": 0.1301,
+      "step": 1416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305545546792863e-05,
+      "loss": 0.149,
+      "step": 1417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304324936389675e-05,
+      "loss": 0.1243,
+      "step": 1418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303103292879104e-05,
+      "loss": 0.1364,
+      "step": 1419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301880616396803e-05,
+      "loss": 0.1481,
+      "step": 1420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300656907078533e-05,
+      "loss": 0.1395,
+      "step": 1421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299432165060166e-05,
+      "loss": 0.1321,
+      "step": 1422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298206390477693e-05,
+      "loss": 0.1401,
+      "step": 1423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929697958346722e-05,
+      "loss": 0.1301,
+      "step": 1424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295751744164955e-05,
+      "loss": 0.1285,
+      "step": 1425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929452287270724e-05,
+      "loss": 0.1461,
+      "step": 1426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293292969230527e-05,
+      "loss": 0.147,
+      "step": 1427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292062033871374e-05,
+      "loss": 0.1219,
+      "step": 1428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9290830066766454e-05,
+      "loss": 0.1401,
+      "step": 1429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289597068052563e-05,
+      "loss": 0.1531,
+      "step": 1430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928836303786661e-05,
+      "loss": 0.1316,
+      "step": 1431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928712797634561e-05,
+      "loss": 0.1071,
+      "step": 1432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285891883626698e-05,
+      "loss": 0.1223,
+      "step": 1433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284654759847127e-05,
+      "loss": 0.1436,
+      "step": 1434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283416605144264e-05,
+      "loss": 0.1259,
+      "step": 1435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282177419655586e-05,
+      "loss": 0.1196,
+      "step": 1436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928093720351868e-05,
+      "loss": 0.1471,
+      "step": 1437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927969595687126e-05,
+      "loss": 0.1343,
+      "step": 1438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278453679851147e-05,
+      "loss": 0.1511,
+      "step": 1439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277210372596278e-05,
+      "loss": 0.1575,
+      "step": 1440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275966035244702e-05,
+      "loss": 0.1301,
+      "step": 1441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274720667934585e-05,
+      "loss": 0.1501,
+      "step": 1442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273474270804206e-05,
+      "loss": 0.1292,
+      "step": 1443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272226843991956e-05,
+      "loss": 0.118,
+      "step": 1444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927097838763635e-05,
+      "loss": 0.1057,
+      "step": 1445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926972890187601e-05,
+      "loss": 0.1295,
+      "step": 1446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268478386849664e-05,
+      "loss": 0.1331,
+      "step": 1447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267226842696174e-05,
+      "loss": 0.1207,
+      "step": 1448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265974269554497e-05,
+      "loss": 0.1288,
+      "step": 1449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264720667563718e-05,
+      "loss": 0.1517,
+      "step": 1450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926346603686303e-05,
+      "loss": 0.1165,
+      "step": 1451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926221037759174e-05,
+      "loss": 0.1339,
+      "step": 1452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926095368988927e-05,
+      "loss": 0.1229,
+      "step": 1453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925969597389516e-05,
+      "loss": 0.1285,
+      "step": 1454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258437229749054e-05,
+      "loss": 0.1394,
+      "step": 1455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925717745759072e-05,
+      "loss": 0.1398,
+      "step": 1456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255916657560042e-05,
+      "loss": 0.1158,
+      "step": 1457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254654829797007e-05,
+      "loss": 0.1465,
+      "step": 1458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925339197444173e-05,
+      "loss": 0.1193,
+      "step": 1459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252128091634418e-05,
+      "loss": 0.1399,
+      "step": 1460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925086318151542e-05,
+      "loss": 0.1315,
+      "step": 1461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924959724422518e-05,
+      "loss": 0.1241,
+      "step": 1462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248330279904262e-05,
+      "loss": 0.1334,
+      "step": 1463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247062288693342e-05,
+      "loss": 0.1414,
+      "step": 1464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245793270733216e-05,
+      "loss": 0.1499,
+      "step": 1465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244523226164785e-05,
+      "loss": 0.1426,
+      "step": 1466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243252155129075e-05,
+      "loss": 0.1206,
+      "step": 1467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924198005776721e-05,
+      "loss": 0.1356,
+      "step": 1468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240706934220447e-05,
+      "loss": 0.1393,
+      "step": 1469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239432784630145e-05,
+      "loss": 0.1442,
+      "step": 1470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238157609137775e-05,
+      "loss": 0.1477,
+      "step": 1471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236881407884928e-05,
+      "loss": 0.118,
+      "step": 1472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235604181013306e-05,
+      "loss": 0.1184,
+      "step": 1473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234325928664736e-05,
+      "loss": 0.1146,
+      "step": 1474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233046650981137e-05,
+      "loss": 0.1471,
+      "step": 1475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231766348104556e-05,
+      "loss": 0.1316,
+      "step": 1476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923048502017716e-05,
+      "loss": 0.1638,
+      "step": 1477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922920266734121e-05,
+      "loss": 0.1482,
+      "step": 1478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92279192897391e-05,
+      "loss": 0.1571,
+      "step": 1479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922663488751333e-05,
+      "loss": 0.1255,
+      "step": 1480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225349460806507e-05,
+      "loss": 0.1234,
+      "step": 1481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224063009761367e-05,
+      "loss": 0.1536,
+      "step": 1482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9222775534520744e-05,
+      "loss": 0.1377,
+      "step": 1483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9221487035227602e-05,
+      "loss": 0.1074,
+      "step": 1484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220197512025e-05,
+      "loss": 0.1237,
+      "step": 1485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218906965056126e-05,
+      "loss": 0.1508,
+      "step": 1486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921761539446427e-05,
+      "loss": 0.1621,
+      "step": 1487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216322800392855e-05,
+      "loss": 0.1211,
+      "step": 1488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215029182985392e-05,
+      "loss": 0.1212,
+      "step": 1489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921373454238552e-05,
+      "loss": 0.1604,
+      "step": 1490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9212438878736997e-05,
+      "loss": 0.1123,
+      "step": 1491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211142192183683e-05,
+      "loss": 0.1638,
+      "step": 1492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920984448286955e-05,
+      "loss": 0.1321,
+      "step": 1493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208545750938693e-05,
+      "loss": 0.1401,
+      "step": 1494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920724599653532e-05,
+      "loss": 0.1187,
+      "step": 1495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920594521980375e-05,
+      "loss": 0.1461,
+      "step": 1496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920464342088841e-05,
+      "loss": 0.1603,
+      "step": 1497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203340599933852e-05,
+      "loss": 0.1389,
+      "step": 1498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202036757084725e-05,
+      "loss": 0.1251,
+      "step": 1499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920073189248581e-05,
+      "loss": 0.1543,
+      "step": 1500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199426006281992e-05,
+      "loss": 0.1278,
+      "step": 1501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198119098618268e-05,
+      "loss": 0.1226,
+      "step": 1502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919681116963975e-05,
+      "loss": 0.1274,
+      "step": 1503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195502219491663e-05,
+      "loss": 0.1226,
+      "step": 1504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194192248319355e-05,
+      "loss": 0.1549,
+      "step": 1505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919288125626827e-05,
+      "loss": 0.1323,
+      "step": 1506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191569243483977e-05,
+      "loss": 0.1367,
+      "step": 1507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190256210112155e-05,
+      "loss": 0.1626,
+      "step": 1508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91889421562986e-05,
+      "loss": 0.1348,
+      "step": 1509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9187627082189212e-05,
+      "loss": 0.1124,
+      "step": 1510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186310987930014e-05,
+      "loss": 0.1255,
+      "step": 1511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918499387366714e-05,
+      "loss": 0.1222,
+      "step": 1512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918367573954684e-05,
+      "loss": 0.132,
+      "step": 1513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9182356585715464e-05,
+      "loss": 0.1521,
+      "step": 1514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918103641231949e-05,
+      "loss": 0.1378,
+      "step": 1515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9179715219505498e-05,
+      "loss": 0.131,
+      "step": 1516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91783930074202e-05,
+      "loss": 0.1255,
+      "step": 1517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917706977621039e-05,
+      "loss": 0.1389,
+      "step": 1518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175745526023005e-05,
+      "loss": 0.1433,
+      "step": 1519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174420257005085e-05,
+      "loss": 0.1357,
+      "step": 1520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173093969303775e-05,
+      "loss": 0.1354,
+      "step": 1521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171766663066345e-05,
+      "loss": 0.1451,
+      "step": 1522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170438338440165e-05,
+      "loss": 0.1414,
+      "step": 1523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169108995572735e-05,
+      "loss": 0.1417,
+      "step": 1524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167778634611653e-05,
+      "loss": 0.1467,
+      "step": 1525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166447255704632e-05,
+      "loss": 0.1411,
+      "step": 1526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165114858999512e-05,
+      "loss": 0.1276,
+      "step": 1527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916378144464423e-05,
+      "loss": 0.144,
+      "step": 1528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162447012786843e-05,
+      "loss": 0.1265,
+      "step": 1529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161111563575518e-05,
+      "loss": 0.1356,
+      "step": 1530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159775097158542e-05,
+      "loss": 0.1445,
+      "step": 1531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158437613684305e-05,
+      "loss": 0.1453,
+      "step": 1532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157099113301316e-05,
+      "loss": 0.1335,
+      "step": 1533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155759596158195e-05,
+      "loss": 0.1334,
+      "step": 1534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154419062403675e-05,
+      "loss": 0.1178,
+      "step": 1535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153077512186605e-05,
+      "loss": 0.1357,
+      "step": 1536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151734945655946e-05,
+      "loss": 0.1323,
+      "step": 1537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915039136296076e-05,
+      "loss": 0.1196,
+      "step": 1538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149046764250244e-05,
+      "loss": 0.1316,
+      "step": 1539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914770114967369e-05,
+      "loss": 0.1234,
+      "step": 1540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914635451938051e-05,
+      "loss": 0.1324,
+      "step": 1541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145006873520227e-05,
+      "loss": 0.1494,
+      "step": 1542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143658212242475e-05,
+      "loss": 0.1409,
+      "step": 1543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142308535697005e-05,
+      "loss": 0.1185,
+      "step": 1544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914095784403368e-05,
+      "loss": 0.1499,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139606137402468e-05,
+      "loss": 0.1318,
+      "step": 1546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138253415953466e-05,
+      "loss": 0.1271,
+      "step": 1547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136899679836863e-05,
+      "loss": 0.1406,
+      "step": 1548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135544929202977e-05,
+      "loss": 0.1589,
+      "step": 1549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134189164202237e-05,
+      "loss": 0.1592,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913283238498517e-05,
+      "loss": 0.1284,
+      "step": 1551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9131474591702438e-05,
+      "loss": 0.1432,
+      "step": 1552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913011578450479e-05,
+      "loss": 0.1432,
+      "step": 1553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128755963543118e-05,
+      "loss": 0.119,
+      "step": 1554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91273951289684e-05,
+      "loss": 0.1381,
+      "step": 1555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126033280931734e-05,
+      "loss": 0.1393,
+      "step": 1556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9124670419584338e-05,
+      "loss": 0.137,
+      "step": 1557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123306545077536e-05,
+      "loss": 0.135,
+      "step": 1558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912194165756277e-05,
+      "loss": 0.1237,
+      "step": 1559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120575757191584e-05,
+      "loss": 0.1477,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119208844115644e-05,
+      "loss": 0.1361,
+      "step": 1561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117840918486727e-05,
+      "loss": 0.1388,
+      "step": 1562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911647198045672e-05,
+      "loss": 0.1346,
+      "step": 1563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911510203017762e-05,
+      "loss": 0.1085,
+      "step": 1564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113731067801543e-05,
+      "loss": 0.1223,
+      "step": 1565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112359093480716e-05,
+      "loss": 0.132,
+      "step": 1566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911098610736747e-05,
+      "loss": 0.1343,
+      "step": 1567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9109612109614263e-05,
+      "loss": 0.1257,
+      "step": 1568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108237100373647e-05,
+      "loss": 0.1387,
+      "step": 1569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106861079798308e-05,
+      "loss": 0.1217,
+      "step": 1570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105484048041024e-05,
+      "loss": 0.136,
+      "step": 1571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104106005254696e-05,
+      "loss": 0.1317,
+      "step": 1572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102726951592338e-05,
+      "loss": 0.1344,
+      "step": 1573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101346887207065e-05,
+      "loss": 0.1376,
+      "step": 1574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099965812252125e-05,
+      "loss": 0.1235,
+      "step": 1575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909858372688086e-05,
+      "loss": 0.1177,
+      "step": 1576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097200631246727e-05,
+      "loss": 0.1161,
+      "step": 1577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9095816525503304e-05,
+      "loss": 0.1377,
+      "step": 1578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094431409804273e-05,
+      "loss": 0.1315,
+      "step": 1579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909304528430343e-05,
+      "loss": 0.1086,
+      "step": 1580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091658149154683e-05,
+      "loss": 0.129,
+      "step": 1581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090270004512053e-05,
+      "loss": 0.1184,
+      "step": 1582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088880850529677e-05,
+      "loss": 0.1367,
+      "step": 1583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087490687361794e-05,
+      "loss": 0.1265,
+      "step": 1584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9086099515162763e-05,
+      "loss": 0.1199,
+      "step": 1585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084707334087056e-05,
+      "loss": 0.1453,
+      "step": 1586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908331414428925e-05,
+      "loss": 0.1415,
+      "step": 1587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908191994592404e-05,
+      "loss": 0.1497,
+      "step": 1588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080524739146232e-05,
+      "loss": 0.1267,
+      "step": 1589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079128524110745e-05,
+      "loss": 0.1427,
+      "step": 1590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90777313009726e-05,
+      "loss": 0.134,
+      "step": 1591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076333069886943e-05,
+      "loss": 0.1257,
+      "step": 1592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074933831009028e-05,
+      "loss": 0.1473,
+      "step": 1593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073533584494218e-05,
+      "loss": 0.1245,
+      "step": 1594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072132330497993e-05,
+      "loss": 0.1283,
+      "step": 1595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070730069175936e-05,
+      "loss": 0.115,
+      "step": 1596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906932680068375e-05,
+      "loss": 0.1205,
+      "step": 1597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906792252517725e-05,
+      "loss": 0.1278,
+      "step": 1598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066517242812353e-05,
+      "loss": 0.1344,
+      "step": 1599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065110953745098e-05,
+      "loss": 0.1346,
+      "step": 1600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063703658131637e-05,
+      "loss": 0.1414,
+      "step": 1601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062295356128225e-05,
+      "loss": 0.1189,
+      "step": 1602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060886047891233e-05,
+      "loss": 0.125,
+      "step": 1603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9059475733577147e-05,
+      "loss": 0.1213,
+      "step": 1604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058064413342555e-05,
+      "loss": 0.129,
+      "step": 1605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905665208734417e-05,
+      "loss": 0.1249,
+      "step": 1606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055238755738805e-05,
+      "loss": 0.1274,
+      "step": 1607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053824418683395e-05,
+      "loss": 0.1118,
+      "step": 1608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052409076334974e-05,
+      "loss": 0.1361,
+      "step": 1609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90509927288507e-05,
+      "loss": 0.1107,
+      "step": 1610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049575376387838e-05,
+      "loss": 0.114,
+      "step": 1611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048157019103758e-05,
+      "loss": 0.1265,
+      "step": 1612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046737657155953e-05,
+      "loss": 0.1255,
+      "step": 1613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904531729070202e-05,
+      "loss": 0.1262,
+      "step": 1614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043895919899677e-05,
+      "loss": 0.1292,
+      "step": 1615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042473544906733e-05,
+      "loss": 0.105,
+      "step": 1616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041050165881126e-05,
+      "loss": 0.1294,
+      "step": 1617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039625782980907e-05,
+      "loss": 0.1244,
+      "step": 1618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903820039636423e-05,
+      "loss": 0.1129,
+      "step": 1619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903677400618936e-05,
+      "loss": 0.1206,
+      "step": 1620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903534661261468e-05,
+      "loss": 0.1145,
+      "step": 1621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903391821579868e-05,
+      "loss": 0.1324,
+      "step": 1622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032488815899958e-05,
+      "loss": 0.1257,
+      "step": 1623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031058413077233e-05,
+      "loss": 0.1345,
+      "step": 1624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902962700748933e-05,
+      "loss": 0.1233,
+      "step": 1625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902819459929518e-05,
+      "loss": 0.1454,
+      "step": 1626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026761188653837e-05,
+      "loss": 0.1411,
+      "step": 1627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902532677572446e-05,
+      "loss": 0.1315,
+      "step": 1628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902389136066631e-05,
+      "loss": 0.1357,
+      "step": 1629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902245494363878e-05,
+      "loss": 0.1194,
+      "step": 1630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021017524801356e-05,
+      "loss": 0.1169,
+      "step": 1631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019579104313645e-05,
+      "loss": 0.1307,
+      "step": 1632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018139682335363e-05,
+      "loss": 0.1146,
+      "step": 1633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016699259026332e-05,
+      "loss": 0.134,
+      "step": 1634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015257834546492e-05,
+      "loss": 0.1246,
+      "step": 1635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013815409055895e-05,
+      "loss": 0.1107,
+      "step": 1636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012371982714698e-05,
+      "loss": 0.1144,
+      "step": 1637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010927555683173e-05,
+      "loss": 0.1244,
+      "step": 1638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009482128121698e-05,
+      "loss": 0.0997,
+      "step": 1639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9008035700190774e-05,
+      "loss": 0.139,
+      "step": 1640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006588272051e-05,
+      "loss": 0.1155,
+      "step": 1641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900513984386309e-05,
+      "loss": 0.1255,
+      "step": 1642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003690415787882e-05,
+      "loss": 0.1157,
+      "step": 1643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002239987986296e-05,
+      "loss": 0.1211,
+      "step": 1644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900078856061939e-05,
+      "loss": 0.1373,
+      "step": 1645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999336133848327e-05,
+      "loss": 0.1125,
+      "step": 1646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997882707834372e-05,
+      "loss": 0.1241,
+      "step": 1647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996428282738906e-05,
+      "loss": 0.1299,
+      "step": 1648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994972858723425e-05,
+      "loss": 0.1115,
+      "step": 1649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899351643594953e-05,
+      "loss": 0.139,
+      "step": 1650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992059014578933e-05,
+      "loss": 0.1486,
+      "step": 1651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899060059477346e-05,
+      "loss": 0.1509,
+      "step": 1652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989141176695054e-05,
+      "loss": 0.1229,
+      "step": 1653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8987680760505753e-05,
+      "loss": 0.1266,
+      "step": 1654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986219346367717e-05,
+      "loss": 0.1262,
+      "step": 1655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984756934443215e-05,
+      "loss": 0.1405,
+      "step": 1656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898329352489463e-05,
+      "loss": 0.1399,
+      "step": 1657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981829117884446e-05,
+      "loss": 0.1328,
+      "step": 1658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8980363713575264e-05,
+      "loss": 0.1323,
+      "step": 1659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89788973121298e-05,
+      "loss": 0.1436,
+      "step": 1660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897742991371087e-05,
+      "loss": 0.1108,
+      "step": 1661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8975961518481412e-05,
+      "loss": 0.1299,
+      "step": 1662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897449212660447e-05,
+      "loss": 0.1234,
+      "step": 1663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973021738243193e-05,
+      "loss": 0.1305,
+      "step": 1664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971550353560852e-05,
+      "loss": 0.1323,
+      "step": 1665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970077972720816e-05,
+      "loss": 0.1487,
+      "step": 1666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968604595886578e-05,
+      "loss": 0.1417,
+      "step": 1667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896713022322173e-05,
+      "loss": 0.1281,
+      "step": 1668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896565485488998e-05,
+      "loss": 0.1233,
+      "step": 1669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964178491055144e-05,
+      "loss": 0.1266,
+      "step": 1670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962701131881153e-05,
+      "loss": 0.132,
+      "step": 1671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961222777532048e-05,
+      "loss": 0.1138,
+      "step": 1672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959743428171972e-05,
+      "loss": 0.131,
+      "step": 1673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958263083965187e-05,
+      "loss": 0.1285,
+      "step": 1674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956781745076068e-05,
+      "loss": 0.1353,
+      "step": 1675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955299411669093e-05,
+      "loss": 0.1158,
+      "step": 1676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895381608390885e-05,
+      "loss": 0.1304,
+      "step": 1677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8952331761960044e-05,
+      "loss": 0.0944,
+      "step": 1678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950846445987486e-05,
+      "loss": 0.1304,
+      "step": 1679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89493601361561e-05,
+      "loss": 0.127,
+      "step": 1680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947872832630916e-05,
+      "loss": 0.1218,
+      "step": 1681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946384535577078e-05,
+      "loss": 0.1257,
+      "step": 1682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944895245159838e-05,
+      "loss": 0.1227,
+      "step": 1683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943404961544565e-05,
+      "loss": 0.137,
+      "step": 1684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941913684896724e-05,
+      "loss": 0.1426,
+      "step": 1685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8940421415381908e-05,
+      "loss": 0.1263,
+      "step": 1686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938928153165802e-05,
+      "loss": 0.1171,
+      "step": 1687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937433898414223e-05,
+      "loss": 0.14,
+      "step": 1688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8935938651293076e-05,
+      "loss": 0.126,
+      "step": 1689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8934442411968387e-05,
+      "loss": 0.1248,
+      "step": 1690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8932945180606296e-05,
+      "loss": 0.1256,
+      "step": 1691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931446957373045e-05,
+      "loss": 0.1201,
+      "step": 1692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892994774243499e-05,
+      "loss": 0.1692,
+      "step": 1693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928447535958598e-05,
+      "loss": 0.1368,
+      "step": 1694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892694633811044e-05,
+      "loss": 0.1244,
+      "step": 1695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925444149057206e-05,
+      "loss": 0.1445,
+      "step": 1696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923940968965694e-05,
+      "loss": 0.1472,
+      "step": 1697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922436798002803e-05,
+      "loss": 0.101,
+      "step": 1698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920931636335553e-05,
+      "loss": 0.1073,
+      "step": 1699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919425484131072e-05,
+      "loss": 0.1233,
+      "step": 1700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917918341556593e-05,
+      "loss": 0.1614,
+      "step": 1701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891641020877946e-05,
+      "loss": 0.121,
+      "step": 1702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8914901085967136e-05,
+      "loss": 0.1409,
+      "step": 1703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913390973287176e-05,
+      "loss": 0.1129,
+      "step": 1704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911879870907266e-05,
+      "loss": 0.1072,
+      "step": 1705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910367778995186e-05,
+      "loss": 0.1168,
+      "step": 1706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890885469771883e-05,
+      "loss": 0.1215,
+      "step": 1707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907340627246204e-05,
+      "loss": 0.1487,
+      "step": 1708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890582556774543e-05,
+      "loss": 0.1345,
+      "step": 1709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904309519384726e-05,
+      "loss": 0.1194,
+      "step": 1710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902792482332425e-05,
+      "loss": 0.113,
+      "step": 1711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890127445675698e-05,
+      "loss": 0.1227,
+      "step": 1712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899755442826936e-05,
+      "loss": 0.1234,
+      "step": 1713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898235440710962e-05,
+      "loss": 0.1473,
+      "step": 1714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889671445057783e-05,
+      "loss": 0.1135,
+      "step": 1715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895192472596425e-05,
+      "loss": 0.1378,
+      "step": 1716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889366950693574e-05,
+      "loss": 0.1288,
+      "step": 1717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892145553764877e-05,
+      "loss": 0.1497,
+      "step": 1718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889062061325305e-05,
+      "loss": 0.1339,
+      "step": 1719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889094685569577e-05,
+      "loss": 0.1195,
+      "step": 1720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888756777088389e-05,
+      "loss": 0.1292,
+      "step": 1721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888603986936554e-05,
+      "loss": 0.1345,
+      "step": 1722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888451098118416e-05,
+      "loss": 0.1284,
+      "step": 1723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882981106509528e-05,
+      "loss": 0.1116,
+      "step": 1724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881450245511502e-05,
+      "loss": 0.1163,
+      "step": 1725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879918398360067e-05,
+      "loss": 0.1353,
+      "step": 1726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878385565225314e-05,
+      "loss": 0.1331,
+      "step": 1727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876851746277434e-05,
+      "loss": 0.1604,
+      "step": 1728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887531694168674e-05,
+      "loss": 0.1249,
+      "step": 1729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8873781151623648e-05,
+      "loss": 0.1105,
+      "step": 1730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887224437625869e-05,
+      "loss": 0.1306,
+      "step": 1731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870706615762492e-05,
+      "loss": 0.1318,
+      "step": 1732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8869167870305806e-05,
+      "loss": 0.1437,
+      "step": 1733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867628140059485e-05,
+      "loss": 0.0989,
+      "step": 1734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866087425194493e-05,
+      "loss": 0.1156,
+      "step": 1735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864545725881908e-05,
+      "loss": 0.14,
+      "step": 1736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8863003042292904e-05,
+      "loss": 0.1353,
+      "step": 1737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861459374598783e-05,
+      "loss": 0.1406,
+      "step": 1738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.885991472297094e-05,
+      "loss": 0.1283,
+      "step": 1739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8858369087580887e-05,
+      "loss": 0.151,
+      "step": 1740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856822468600245e-05,
+      "loss": 0.1265,
+      "step": 1741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885527486620074e-05,
+      "loss": 0.1118,
+      "step": 1742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853726280554215e-05,
+      "loss": 0.1279,
+      "step": 1743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852176711832614e-05,
+      "loss": 0.1101,
+      "step": 1744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850626160207998e-05,
+      "loss": 0.1047,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849074625852527e-05,
+      "loss": 0.1322,
+      "step": 1746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8847522108938482e-05,
+      "loss": 0.1453,
+      "step": 1747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884596860963824e-05,
+      "loss": 0.1167,
+      "step": 1748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844414128124294e-05,
+      "loss": 0.1156,
+      "step": 1749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8842858664569257e-05,
+      "loss": 0.1423,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884130221914583e-05,
+      "loss": 0.1245,
+      "step": 1751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839744792026837e-05,
+      "loss": 0.1136,
+      "step": 1752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838186383385205e-05,
+      "loss": 0.1449,
+      "step": 1753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8836626993393972e-05,
+      "loss": 0.1133,
+      "step": 1754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883506662222629e-05,
+      "loss": 0.1246,
+      "step": 1755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883350527005541e-05,
+      "loss": 0.1053,
+      "step": 1756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8831942937054697e-05,
+      "loss": 0.0878,
+      "step": 1757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883037962339763e-05,
+      "loss": 0.1046,
+      "step": 1758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882881532925779e-05,
+      "loss": 0.1337,
+      "step": 1759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827250054808864e-05,
+      "loss": 0.1322,
+      "step": 1760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8825683800224655e-05,
+      "loss": 0.1394,
+      "step": 1761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824116565679074e-05,
+      "loss": 0.1263,
+      "step": 1762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882254835134614e-05,
+      "loss": 0.1108,
+      "step": 1763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820979157399976e-05,
+      "loss": 0.1299,
+      "step": 1764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881940898401482e-05,
+      "loss": 0.098,
+      "step": 1765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817837831365015e-05,
+      "loss": 0.1069,
+      "step": 1766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816265699625015e-05,
+      "loss": 0.1179,
+      "step": 1767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814692588969387e-05,
+      "loss": 0.1167,
+      "step": 1768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813118499572796e-05,
+      "loss": 0.1172,
+      "step": 1769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881154343161002e-05,
+      "loss": 0.1189,
+      "step": 1770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8809967385255952e-05,
+      "loss": 0.1338,
+      "step": 1771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808390360685586e-05,
+      "loss": 0.1333,
+      "step": 1772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806812358074024e-05,
+      "loss": 0.1156,
+      "step": 1773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805233377596484e-05,
+      "loss": 0.1208,
+      "step": 1774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880365341942829e-05,
+      "loss": 0.1294,
+      "step": 1775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802072483744867e-05,
+      "loss": 0.1351,
+      "step": 1776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880049057072176e-05,
+      "loss": 0.1111,
+      "step": 1777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8798907680534615e-05,
+      "loss": 0.1254,
+      "step": 1778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8797323813359186e-05,
+      "loss": 0.105,
+      "step": 1779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795738969371343e-05,
+      "loss": 0.126,
+      "step": 1780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794153148747055e-05,
+      "loss": 0.1438,
+      "step": 1781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792566351662405e-05,
+      "loss": 0.1047,
+      "step": 1782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790978578293584e-05,
+      "loss": 0.1256,
+      "step": 1783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789389828816894e-05,
+      "loss": 0.1471,
+      "step": 1784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787800103408733e-05,
+      "loss": 0.1127,
+      "step": 1785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786209402245624e-05,
+      "loss": 0.1188,
+      "step": 1786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878461772550419e-05,
+      "loss": 0.1346,
+      "step": 1787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8783025073361162e-05,
+      "loss": 0.1215,
+      "step": 1788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878143144599338e-05,
+      "loss": 0.1013,
+      "step": 1789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8779836843577796e-05,
+      "loss": 0.1145,
+      "step": 1790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877824126629146e-05,
+      "loss": 0.1298,
+      "step": 1791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877664471431154e-05,
+      "loss": 0.1621,
+      "step": 1792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775047187815313e-05,
+      "loss": 0.1219,
+      "step": 1793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773448686980156e-05,
+      "loss": 0.11,
+      "step": 1794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771849211983567e-05,
+      "loss": 0.1222,
+      "step": 1795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770248763003135e-05,
+      "loss": 0.123,
+      "step": 1796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768647340216567e-05,
+      "loss": 0.1284,
+      "step": 1797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767044943801683e-05,
+      "loss": 0.1261,
+      "step": 1798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87654415739364e-05,
+      "loss": 0.1294,
+      "step": 1799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876383723079875e-05,
+      "loss": 0.127,
+      "step": 1800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762231914566877e-05,
+      "loss": 0.1217,
+      "step": 1801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8760625625419014e-05,
+      "loss": 0.123,
+      "step": 1802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8759018363533528e-05,
+      "loss": 0.1271,
+      "step": 1803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875741012908888e-05,
+      "loss": 0.1044,
+      "step": 1804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755800922263633e-05,
+      "loss": 0.1095,
+      "step": 1805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8754190743236476e-05,
+      "loss": 0.113,
+      "step": 1806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875257959218619e-05,
+      "loss": 0.1198,
+      "step": 1807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8750967469291666e-05,
+      "loss": 0.1123,
+      "step": 1808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874935437473191e-05,
+      "loss": 0.1019,
+      "step": 1809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874774030868604e-05,
+      "loss": 0.1263,
+      "step": 1810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746125271333257e-05,
+      "loss": 0.1428,
+      "step": 1811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744509262852902e-05,
+      "loss": 0.1354,
+      "step": 1812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87428922834244e-05,
+      "loss": 0.1406,
+      "step": 1813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87412743332273e-05,
+      "loss": 0.1078,
+      "step": 1814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739655412441243e-05,
+      "loss": 0.1207,
+      "step": 1815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873803552124599e-05,
+      "loss": 0.1024,
+      "step": 1816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873641465982141e-05,
+      "loss": 0.115,
+      "step": 1817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734792828347472e-05,
+      "loss": 0.1461,
+      "step": 1818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733170027004254e-05,
+      "loss": 0.105,
+      "step": 1819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731546255971948e-05,
+      "loss": 0.1084,
+      "step": 1820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872992151543085e-05,
+      "loss": 0.1305,
+      "step": 1821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728295805561355e-05,
+      "loss": 0.1217,
+      "step": 1822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726669126543985e-05,
+      "loss": 0.1292,
+      "step": 1823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725041478559354e-05,
+      "loss": 0.1188,
+      "step": 1824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723412861788187e-05,
+      "loss": 0.1299,
+      "step": 1825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872178327641132e-05,
+      "loss": 0.1101,
+      "step": 1826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720152722609692e-05,
+      "loss": 0.1279,
+      "step": 1827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718521200564352e-05,
+      "loss": 0.1414,
+      "step": 1828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716888710456458e-05,
+      "loss": 0.1067,
+      "step": 1829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715255252467274e-05,
+      "loss": 0.1079,
+      "step": 1830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871362082677817e-05,
+      "loss": 0.1385,
+      "step": 1831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711985433570628e-05,
+      "loss": 0.1017,
+      "step": 1832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8710349073026227e-05,
+      "loss": 0.1246,
+      "step": 1833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708711745326668e-05,
+      "loss": 0.1195,
+      "step": 1834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870707345065375e-05,
+      "loss": 0.1035,
+      "step": 1835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705434189189374e-05,
+      "loss": 0.1195,
+      "step": 1836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870379396111557e-05,
+      "loss": 0.1152,
+      "step": 1837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870215276661445e-05,
+      "loss": 0.1456,
+      "step": 1838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8700510605868246e-05,
+      "loss": 0.1149,
+      "step": 1839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.86988674790593e-05,
+      "loss": 0.1202,
+      "step": 1840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697223386370048e-05,
+      "loss": 0.0968,
+      "step": 1841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695578327983054e-05,
+      "loss": 0.1442,
+      "step": 1842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8693932304080967e-05,
+      "loss": 0.1382,
+      "step": 1843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869228531484656e-05,
+      "loss": 0.1479,
+      "step": 1844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690637360462706e-05,
+      "loss": 0.1174,
+      "step": 1845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868898844111238e-05,
+      "loss": 0.1348,
+      "step": 1846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868733855697868e-05,
+      "loss": 0.1066,
+      "step": 1847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685687708244794e-05,
+      "loss": 0.1342,
+      "step": 1848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684035895094025e-05,
+      "loss": 0.1294,
+      "step": 1849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682383117709783e-05,
+      "loss": 0.1077,
+      "step": 1850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680729376275584e-05,
+      "loss": 0.1332,
+      "step": 1851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8679074670975056e-05,
+      "loss": 0.1348,
+      "step": 1852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677419001991924e-05,
+      "loss": 0.1245,
+      "step": 1853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675762369510027e-05,
+      "loss": 0.1199,
+      "step": 1854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867410477371331e-05,
+      "loss": 0.1519,
+      "step": 1855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8672446214785824e-05,
+      "loss": 0.1105,
+      "step": 1856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670786692911727e-05,
+      "loss": 0.1288,
+      "step": 1857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669126208275286e-05,
+      "loss": 0.1134,
+      "step": 1858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667464761060874e-05,
+      "loss": 0.1144,
+      "step": 1859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665802351452966e-05,
+      "loss": 0.1219,
+      "step": 1860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664138979636152e-05,
+      "loss": 0.1207,
+      "step": 1861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866247464579512e-05,
+      "loss": 0.1214,
+      "step": 1862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8660809350114673e-05,
+      "loss": 0.1212,
+      "step": 1863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865914309277972e-05,
+      "loss": 0.1473,
+      "step": 1864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657475873975267e-05,
+      "loss": 0.1116,
+      "step": 1865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865580769388644e-05,
+      "loss": 0.1208,
+      "step": 1866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654138552698463e-05,
+      "loss": 0.1553,
+      "step": 1867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652468450596673e-05,
+      "loss": 0.1132,
+      "step": 1868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650797387766502e-05,
+      "loss": 0.1306,
+      "step": 1869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864912536439351e-05,
+      "loss": 0.1138,
+      "step": 1870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8647452380663335e-05,
+      "loss": 0.123,
+      "step": 1871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8645778436761748e-05,
+      "loss": 0.119,
+      "step": 1872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644103532874612e-05,
+      "loss": 0.1122,
+      "step": 1873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86424276691879e-05,
+      "loss": 0.1162,
+      "step": 1874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864075084588769e-05,
+      "loss": 0.1356,
+      "step": 1875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8639073063160172e-05,
+      "loss": 0.1188,
+      "step": 1876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863739432119164e-05,
+      "loss": 0.1265,
+      "step": 1877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635714620168488e-05,
+      "loss": 0.1113,
+      "step": 1878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634033960277226e-05,
+      "loss": 0.1191,
+      "step": 1879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863235234170446e-05,
+      "loss": 0.1306,
+      "step": 1880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630669764636922e-05,
+      "loss": 0.1145,
+      "step": 1881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8628986229261426e-05,
+      "loss": 0.13,
+      "step": 1882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627301735764907e-05,
+      "loss": 0.1315,
+      "step": 1883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625616284334405e-05,
+      "loss": 0.1169,
+      "step": 1884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862392987515706e-05,
+      "loss": 0.1343,
+      "step": 1885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622242508420123e-05,
+      "loss": 0.1312,
+      "step": 1886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620554184310954e-05,
+      "loss": 0.1238,
+      "step": 1887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618864903017018e-05,
+      "loss": 0.1218,
+      "step": 1888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8617174664725877e-05,
+      "loss": 0.1281,
+      "step": 1889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861548346962522e-05,
+      "loss": 0.1326,
+      "step": 1890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613791317902815e-05,
+      "loss": 0.1228,
+      "step": 1891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861209820974656e-05,
+      "loss": 0.1128,
+      "step": 1892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610404145344445e-05,
+      "loss": 0.1279,
+      "step": 1893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860870912488457e-05,
+      "loss": 0.1146,
+      "step": 1894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8607013148555148e-05,
+      "loss": 0.1177,
+      "step": 1895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605316216544485e-05,
+      "loss": 0.1156,
+      "step": 1896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603618329041002e-05,
+      "loss": 0.1368,
+      "step": 1897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601919486233227e-05,
+      "loss": 0.1371,
+      "step": 1898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860021968830979e-05,
+      "loss": 0.1038,
+      "step": 1899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8598518935459424e-05,
+      "loss": 0.1172,
+      "step": 1900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859681722787098e-05,
+      "loss": 0.1351,
+      "step": 1901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85951145657334e-05,
+      "loss": 0.12,
+      "step": 1902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593410949235747e-05,
+      "loss": 0.1467,
+      "step": 1903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859170637856718e-05,
+      "loss": 0.1067,
+      "step": 1904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859000085391696e-05,
+      "loss": 0.144,
+      "step": 1905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588294375474466e-05,
+      "loss": 0.1111,
+      "step": 1906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586586943429177e-05,
+      "loss": 0.1077,
+      "step": 1907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584878557970677e-05,
+      "loss": 0.1237,
+      "step": 1908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583169219288658e-05,
+      "loss": 0.1162,
+      "step": 1909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581458927572912e-05,
+      "loss": 0.1129,
+      "step": 1910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857974768301335e-05,
+      "loss": 0.1285,
+      "step": 1911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857803548579997e-05,
+      "loss": 0.1333,
+      "step": 1912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576322336122898e-05,
+      "loss": 0.1117,
+      "step": 1913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574608234172347e-05,
+      "loss": 0.1273,
+      "step": 1914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857289318013864e-05,
+      "loss": 0.1322,
+      "step": 1915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8571177174212214e-05,
+      "loss": 0.1167,
+      "step": 1916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85694602165836e-05,
+      "loss": 0.1245,
+      "step": 1917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856774230744345e-05,
+      "loss": 0.1241,
+      "step": 1918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566023446982503e-05,
+      "loss": 0.1241,
+      "step": 1919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564303635391617e-05,
+      "loss": 0.1146,
+      "step": 1920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562582872861748e-05,
+      "loss": 0.0977,
+      "step": 1921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856086115958397e-05,
+      "loss": 0.1183,
+      "step": 1922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559138495749445e-05,
+      "loss": 0.1116,
+      "step": 1923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557414881549453e-05,
+      "loss": 0.1259,
+      "step": 1924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555690317175375e-05,
+      "loss": 0.132,
+      "step": 1925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85539648028187e-05,
+      "loss": 0.1221,
+      "step": 1926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855223833867102e-05,
+      "loss": 0.1266,
+      "step": 1927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855051092492403e-05,
+      "loss": 0.1125,
+      "step": 1928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548782561769535e-05,
+      "loss": 0.1442,
+      "step": 1929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547053249399448e-05,
+      "loss": 0.1107,
+      "step": 1930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854532298800578e-05,
+      "loss": 0.1382,
+      "step": 1931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543591777780653e-05,
+      "loss": 0.1243,
+      "step": 1932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854185961891629e-05,
+      "loss": 0.1302,
+      "step": 1933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854012651160502e-05,
+      "loss": 0.099,
+      "step": 1934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8538392456039286e-05,
+      "loss": 0.1134,
+      "step": 1935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853665745241162e-05,
+      "loss": 0.1472,
+      "step": 1936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534921500914677e-05,
+      "loss": 0.1128,
+      "step": 1937
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533184601741205e-05,
+      "loss": 0.1093,
+      "step": 1938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8531446755084057e-05,
+      "loss": 0.1307,
+      "step": 1939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8529707961136202e-05,
+      "loss": 0.1121,
+      "step": 1940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527968220090705e-05,
+      "loss": 0.1281,
+      "step": 1941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526227532140734e-05,
+      "loss": 0.1093,
+      "step": 1942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852448589747957e-05,
+      "loss": 0.1206,
+      "step": 1943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522743316300597e-05,
+      "loss": 0.1221,
+      "step": 1944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.84931506849315e-08,
+      "loss": 0.115,
+      "step": 1945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.36986301369863e-07,
+      "loss": 0.115,
+      "step": 1946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0547945205479452e-07,
+      "loss": 0.1182,
+      "step": 1947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 0.1143,
+      "step": 1948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4246575342465755e-07,
+      "loss": 0.1152,
+      "step": 1949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 0.1168,
+      "step": 1950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.794520547945206e-07,
+      "loss": 0.1305,
+      "step": 1951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 0.1327,
+      "step": 1952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.164383561643836e-07,
+      "loss": 0.1124,
+      "step": 1953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.849315068493151e-07,
+      "loss": 0.1301,
+      "step": 1954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.534246575342466e-07,
+      "loss": 0.1156,
+      "step": 1955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 0.1232,
+      "step": 1956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.904109589041097e-07,
+      "loss": 0.1381,
+      "step": 1957
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.589041095890411e-07,
+      "loss": 0.1267,
+      "step": 1958
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0273972602739727e-06,
+      "loss": 0.1133,
+      "step": 1959
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 0.098,
+      "step": 1960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1643835616438357e-06,
+      "loss": 0.1239,
+      "step": 1961
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2328767123287673e-06,
+      "loss": 0.1082,
+      "step": 1962
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3013698630136986e-06,
+      "loss": 0.141,
+      "step": 1963
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 0.1259,
+      "step": 1964
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4383561643835616e-06,
+      "loss": 0.1198,
+      "step": 1965
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5068493150684932e-06,
+      "loss": 0.1075,
+      "step": 1966
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5753424657534248e-06,
+      "loss": 0.1283,
+      "step": 1967
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 0.1165,
+      "step": 1968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7123287671232877e-06,
+      "loss": 0.1033,
+      "step": 1969
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7808219178082193e-06,
+      "loss": 0.1257,
+      "step": 1970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8493150684931507e-06,
+      "loss": 0.1299,
+      "step": 1971
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 0.1174,
+      "step": 1972
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9863013698630136e-06,
+      "loss": 0.1222,
+      "step": 1973
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.0547945205479454e-06,
+      "loss": 0.1057,
+      "step": 1974
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.123287671232877e-06,
+      "loss": 0.0964,
+      "step": 1975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 0.1365,
+      "step": 1976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.26027397260274e-06,
+      "loss": 0.1104,
+      "step": 1977
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3287671232876713e-06,
+      "loss": 0.1052,
+      "step": 1978
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3972602739726027e-06,
+      "loss": 0.1301,
+      "step": 1979
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 0.1316,
+      "step": 1980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.534246575342466e-06,
+      "loss": 0.1134,
+      "step": 1981
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 0.1252,
+      "step": 1982
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.671232876712329e-06,
+      "loss": 0.1243,
+      "step": 1983
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 0.099,
+      "step": 1984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8082191780821922e-06,
+      "loss": 0.1362,
+      "step": 1985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 0.1007,
+      "step": 1986
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.945205479452055e-06,
+      "loss": 0.1217,
+      "step": 1987
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 0.1344,
+      "step": 1988
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.082191780821918e-06,
+      "loss": 0.1288,
+      "step": 1989
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.1506849315068495e-06,
+      "loss": 0.1056,
+      "step": 1990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2191780821917813e-06,
+      "loss": 0.1251,
+      "step": 1991
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 0.0916,
+      "step": 1992
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.356164383561644e-06,
+      "loss": 0.1167,
+      "step": 1993
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 0.1021,
+      "step": 1994
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.4931506849315072e-06,
+      "loss": 0.1074,
+      "step": 1995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 0.0854,
+      "step": 1996
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6301369863013704e-06,
+      "loss": 0.115,
+      "step": 1997
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.6986301369863014e-06,
+      "loss": 0.1212,
+      "step": 1998
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.767123287671233e-06,
+      "loss": 0.1067,
+      "step": 1999
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 0.1061,
+      "step": 2000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.904109589041096e-06,
+      "loss": 0.1245,
+      "step": 2001
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.972602739726027e-06,
+      "loss": 0.0984,
+      "step": 2002
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.0410958904109595e-06,
+      "loss": 0.1261,
+      "step": 2003
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 0.116,
+      "step": 2004
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.178082191780822e-06,
+      "loss": 0.1039,
+      "step": 2005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.246575342465754e-06,
+      "loss": 0.0969,
+      "step": 2006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.315068493150685e-06,
+      "loss": 0.1248,
+      "step": 2007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 0.1049,
+      "step": 2008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4520547945205486e-06,
+      "loss": 0.1345,
+      "step": 2009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.52054794520548e-06,
+      "loss": 0.1076,
+      "step": 2010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.589041095890411e-06,
+      "loss": 0.1251,
+      "step": 2011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 0.1106,
+      "step": 2012
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.726027397260274e-06,
+      "loss": 0.093,
+      "step": 2013
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.7945205479452054e-06,
+      "loss": 0.1044,
+      "step": 2014
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.863013698630138e-06,
+      "loss": 0.1053,
+      "step": 2015
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 0.1055,
+      "step": 2016
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5e-06,
+      "loss": 0.1068,
+      "step": 2017
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.068493150684932e-06,
+      "loss": 0.1307,
+      "step": 2018
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.136986301369864e-06,
+      "loss": 0.129,
+      "step": 2019
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 0.0948,
+      "step": 2020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.273972602739727e-06,
+      "loss": 0.1046,
+      "step": 2021
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.342465753424658e-06,
+      "loss": 0.1091,
+      "step": 2022
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.41095890410959e-06,
+      "loss": 0.1162,
+      "step": 2023
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 0.0941,
+      "step": 2024
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.547945205479452e-06,
+      "loss": 0.1097,
+      "step": 2025
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.6164383561643845e-06,
+      "loss": 0.1073,
+      "step": 2026
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.684931506849316e-06,
+      "loss": 0.1237,
+      "step": 2027
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 0.1008,
+      "step": 2028
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.821917808219179e-06,
+      "loss": 0.111,
+      "step": 2029
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.89041095890411e-06,
+      "loss": 0.1029,
+      "step": 2030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.958904109589042e-06,
+      "loss": 0.1075,
+      "step": 2031
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 0.1077,
+      "step": 2032
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.095890410958905e-06,
+      "loss": 0.12,
+      "step": 2033
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.164383561643836e-06,
+      "loss": 0.1245,
+      "step": 2034
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.2328767123287685e-06,
+      "loss": 0.1195,
+      "step": 2035
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 0.1105,
+      "step": 2036
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.36986301369863e-06,
+      "loss": 0.0945,
+      "step": 2037
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.438356164383563e-06,
+      "loss": 0.0941,
+      "step": 2038
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.506849315068494e-06,
+      "loss": 0.1129,
+      "step": 2039
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.0994,
+      "step": 2040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.643835616438357e-06,
+      "loss": 0.1201,
+      "step": 2041
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.712328767123288e-06,
+      "loss": 0.1105,
+      "step": 2042
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.78082191780822e-06,
+      "loss": 0.1089,
+      "step": 2043
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 0.119,
+      "step": 2044
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.917808219178082e-06,
+      "loss": 0.1124,
+      "step": 2045
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.9863013698630145e-06,
+      "loss": 0.1165,
+      "step": 2046
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.054794520547946e-06,
+      "loss": 0.1301,
+      "step": 2047
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 0.1166,
+      "step": 2048
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.191780821917809e-06,
+      "loss": 0.1219,
+      "step": 2049
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.260273972602741e-06,
+      "loss": 0.104,
+      "step": 2050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.328767123287672e-06,
+      "loss": 0.1071,
+      "step": 2051
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 0.1294,
+      "step": 2052
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.465753424657535e-06,
+      "loss": 0.1158,
+      "step": 2053
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.534246575342466e-06,
+      "loss": 0.1121,
+      "step": 2054
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.6027397260273985e-06,
+      "loss": 0.125,
+      "step": 2055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 0.1238,
+      "step": 2056
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.739726027397261e-06,
+      "loss": 0.0872,
+      "step": 2057
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 0.1191,
+      "step": 2058
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.876712328767124e-06,
+      "loss": 0.0947,
+      "step": 2059
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 0.1106,
+      "step": 2060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.013698630136987e-06,
+      "loss": 0.1244,
+      "step": 2061
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.082191780821919e-06,
+      "loss": 0.1096,
+      "step": 2062
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.150684931506851e-06,
+      "loss": 0.0996,
+      "step": 2063
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 0.1356,
+      "step": 2064
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.287671232876712e-06,
+      "loss": 0.1211,
+      "step": 2065
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.356164383561644e-06,
+      "loss": 0.1112,
+      "step": 2066
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.424657534246577e-06,
+      "loss": 0.141,
+      "step": 2067
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 0.1003,
+      "step": 2068
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.56164383561644e-06,
+      "loss": 0.1157,
+      "step": 2069
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.63013698630137e-06,
+      "loss": 0.092,
+      "step": 2070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.698630136986302e-06,
+      "loss": 0.1412,
+      "step": 2071
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 0.0977,
+      "step": 2072
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.835616438356165e-06,
+      "loss": 0.1013,
+      "step": 2073
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.904109589041097e-06,
+      "loss": 0.1085,
+      "step": 2074
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.972602739726028e-06,
+      "loss": 0.0979,
+      "step": 2075
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 0.1025,
+      "step": 2076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.10958904109589e-06,
+      "loss": 0.0952,
+      "step": 2077
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.178082191780823e-06,
+      "loss": 0.1257,
+      "step": 2078
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.246575342465755e-06,
+      "loss": 0.105,
+      "step": 2079
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 0.1201,
+      "step": 2080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.383561643835618e-06,
+      "loss": 0.1074,
+      "step": 2081
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 0.0885,
+      "step": 2082
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.52054794520548e-06,
+      "loss": 0.1111,
+      "step": 2083
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 0.1027,
+      "step": 2084
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.657534246575343e-06,
+      "loss": 0.1074,
+      "step": 2085
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.726027397260275e-06,
+      "loss": 0.114,
+      "step": 2086
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.794520547945206e-06,
+      "loss": 0.1006,
+      "step": 2087
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 0.1318,
+      "step": 2088
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.931506849315069e-06,
+      "loss": 0.1228,
+      "step": 2089
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1e-05,
+      "loss": 0.0955,
+      "step": 2090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0068493150684933e-05,
+      "loss": 0.0969,
+      "step": 2091
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 0.1411,
+      "step": 2092
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0205479452054796e-05,
+      "loss": 0.1029,
+      "step": 2093
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0273972602739728e-05,
+      "loss": 0.1228,
+      "step": 2094
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0342465753424657e-05,
+      "loss": 0.1115,
+      "step": 2095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 0.1273,
+      "step": 2096
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0479452054794521e-05,
+      "loss": 0.1021,
+      "step": 2097
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0547945205479453e-05,
+      "loss": 0.1036,
+      "step": 2098
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0616438356164384e-05,
+      "loss": 0.1213,
+      "step": 2099
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0684931506849316e-05,
+      "loss": 0.1205,
+      "step": 2100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0753424657534248e-05,
+      "loss": 0.127,
+      "step": 2101
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.082191780821918e-05,
+      "loss": 0.0918,
+      "step": 2102
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.089041095890411e-05,
+      "loss": 0.1353,
+      "step": 2103
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 0.1177,
+      "step": 2104
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1027397260273974e-05,
+      "loss": 0.1198,
+      "step": 2105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1095890410958904e-05,
+      "loss": 0.1169,
+      "step": 2106
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1164383561643837e-05,
+      "loss": 0.0931,
+      "step": 2107
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1232876712328769e-05,
+      "loss": 0.115,
+      "step": 2108
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.1301369863013701e-05,
+      "loss": 0.1174,
+      "step": 2109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1369863013698632e-05,
+      "loss": 0.1241,
+      "step": 2110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1438356164383562e-05,
+      "loss": 0.0857,
+      "step": 2111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.0911,
+      "step": 2112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1575342465753425e-05,
+      "loss": 0.1036,
+      "step": 2113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1643835616438357e-05,
+      "loss": 0.1168,
+      "step": 2114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.171232876712329e-05,
+      "loss": 0.1259,
+      "step": 2115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.178082191780822e-05,
+      "loss": 0.1022,
+      "step": 2116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1849315068493152e-05,
+      "loss": 0.1124,
+      "step": 2117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1917808219178084e-05,
+      "loss": 0.1259,
+      "step": 2118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.1986301369863013e-05,
+      "loss": 0.108,
+      "step": 2119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 0.121,
+      "step": 2120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2123287671232878e-05,
+      "loss": 0.1134,
+      "step": 2121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.219178082191781e-05,
+      "loss": 0.1177,
+      "step": 2122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.226027397260274e-05,
+      "loss": 0.1039,
+      "step": 2123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2328767123287673e-05,
+      "loss": 0.1194,
+      "step": 2124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2397260273972605e-05,
+      "loss": 0.1141,
+      "step": 2125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2465753424657537e-05,
+      "loss": 0.1289,
+      "step": 2126
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2534246575342466e-05,
+      "loss": 0.1007,
+      "step": 2127
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 0.1115,
+      "step": 2128
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2671232876712329e-05,
+      "loss": 0.1141,
+      "step": 2129
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.273972602739726e-05,
+      "loss": 0.0959,
+      "step": 2130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2808219178082193e-05,
+      "loss": 0.1263,
+      "step": 2131
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2876712328767125e-05,
+      "loss": 0.1202,
+      "step": 2132
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.2945205479452056e-05,
+      "loss": 0.1161,
+      "step": 2133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3013698630136988e-05,
+      "loss": 0.1111,
+      "step": 2134
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3082191780821919e-05,
+      "loss": 0.1042,
+      "step": 2135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.1199,
+      "step": 2136
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3219178082191781e-05,
+      "loss": 0.1322,
+      "step": 2137
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3287671232876714e-05,
+      "loss": 0.1083,
+      "step": 2138
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3356164383561646e-05,
+      "loss": 0.0996,
+      "step": 2139
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3424657534246576e-05,
+      "loss": 0.1238,
+      "step": 2140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3493150684931508e-05,
+      "loss": 0.1127,
+      "step": 2141
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.356164383561644e-05,
+      "loss": 0.0988,
+      "step": 2142
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.363013698630137e-05,
+      "loss": 0.1042,
+      "step": 2143
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.127,
+      "step": 2144
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3767123287671234e-05,
+      "loss": 0.0967,
+      "step": 2145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3835616438356164e-05,
+      "loss": 0.1141,
+      "step": 2146
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3904109589041097e-05,
+      "loss": 0.0973,
+      "step": 2147
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3972602739726029e-05,
+      "loss": 0.0928,
+      "step": 2148
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4041095890410961e-05,
+      "loss": 0.1185,
+      "step": 2149
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4109589041095892e-05,
+      "loss": 0.0968,
+      "step": 2150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4178082191780822e-05,
+      "loss": 0.116,
+      "step": 2151
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 0.1116,
+      "step": 2152
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4315068493150685e-05,
+      "loss": 0.1234,
+      "step": 2153
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4383561643835617e-05,
+      "loss": 0.123,
+      "step": 2154
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.445205479452055e-05,
+      "loss": 0.1064,
+      "step": 2155
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520547945205482e-05,
+      "loss": 0.0961,
+      "step": 2156
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4589041095890412e-05,
+      "loss": 0.1317,
+      "step": 2157
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4657534246575344e-05,
+      "loss": 0.1203,
+      "step": 2158
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4726027397260275e-05,
+      "loss": 0.1124,
+      "step": 2159
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 0.0892,
+      "step": 2160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4863013698630138e-05,
+      "loss": 0.1102,
+      "step": 2161
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.493150684931507e-05,
+      "loss": 0.1194,
+      "step": 2162
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1205,
+      "step": 2163
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5068493150684933e-05,
+      "loss": 0.105,
+      "step": 2164
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5136986301369865e-05,
+      "loss": 0.1082,
+      "step": 2165
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5205479452054797e-05,
+      "loss": 0.1271,
+      "step": 2166
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5273972602739728e-05,
+      "loss": 0.0988,
+      "step": 2167
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 0.1201,
+      "step": 2168
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.541095890410959e-05,
+      "loss": 0.1117,
+      "step": 2169
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5479452054794523e-05,
+      "loss": 0.0933,
+      "step": 2170
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5547945205479453e-05,
+      "loss": 0.1276,
+      "step": 2171
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.1039,
+      "step": 2172
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5684931506849318e-05,
+      "loss": 0.1216,
+      "step": 2173
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5753424657534248e-05,
+      "loss": 0.1089,
+      "step": 2174
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5821917808219182e-05,
+      "loss": 0.1223,
+      "step": 2175
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.1139,
+      "step": 2176
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5958904109589043e-05,
+      "loss": 0.12,
+      "step": 2177
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6027397260273974e-05,
+      "loss": 0.1007,
+      "step": 2178
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6095890410958904e-05,
+      "loss": 0.1271,
+      "step": 2179
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6164383561643838e-05,
+      "loss": 0.1187,
+      "step": 2180
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.623287671232877e-05,
+      "loss": 0.1265,
+      "step": 2181
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6301369863013702e-05,
+      "loss": 0.1208,
+      "step": 2182
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6369863013698633e-05,
+      "loss": 0.123,
+      "step": 2183
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 0.123,
+      "step": 2184
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6506849315068494e-05,
+      "loss": 0.105,
+      "step": 2185
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6575342465753425e-05,
+      "loss": 0.1165,
+      "step": 2186
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.664383561643836e-05,
+      "loss": 0.1144,
+      "step": 2187
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.671232876712329e-05,
+      "loss": 0.1284,
+      "step": 2188
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.678082191780822e-05,
+      "loss": 0.1277,
+      "step": 2189
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6849315068493153e-05,
+      "loss": 0.1162,
+      "step": 2190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6917808219178084e-05,
+      "loss": 0.1444,
+      "step": 2191
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 0.106,
+      "step": 2192
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7054794520547945e-05,
+      "loss": 0.1024,
+      "step": 2193
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.712328767123288e-05,
+      "loss": 0.134,
+      "step": 2194
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.719178082191781e-05,
+      "loss": 0.1448,
+      "step": 2195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.726027397260274e-05,
+      "loss": 0.1088,
+      "step": 2196
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7328767123287674e-05,
+      "loss": 0.1272,
+      "step": 2197
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7397260273972604e-05,
+      "loss": 0.1089,
+      "step": 2198
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7465753424657538e-05,
+      "loss": 0.1224,
+      "step": 2199
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 0.1021,
+      "step": 2200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.76027397260274e-05,
+      "loss": 0.1069,
+      "step": 2201
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.767123287671233e-05,
+      "loss": 0.1039,
+      "step": 2202
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.773972602739726e-05,
+      "loss": 0.1171,
+      "step": 2203
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7808219178082194e-05,
+      "loss": 0.1243,
+      "step": 2204
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7876712328767125e-05,
+      "loss": 0.1163,
+      "step": 2205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7945205479452055e-05,
+      "loss": 0.1179,
+      "step": 2206
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801369863013699e-05,
+      "loss": 0.1123,
+      "step": 2207
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 0.1113,
+      "step": 2208
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.815068493150685e-05,
+      "loss": 0.1224,
+      "step": 2209
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.821917808219178e-05,
+      "loss": 0.1261,
+      "step": 2210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8287671232876715e-05,
+      "loss": 0.1134,
+      "step": 2211
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8356164383561645e-05,
+      "loss": 0.14,
+      "step": 2212
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8424657534246576e-05,
+      "loss": 0.104,
+      "step": 2213
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.849315068493151e-05,
+      "loss": 0.1193,
+      "step": 2214
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.856164383561644e-05,
+      "loss": 0.1107,
+      "step": 2215
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 0.1162,
+      "step": 2216
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.86986301369863e-05,
+      "loss": 0.1024,
+      "step": 2217
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8767123287671235e-05,
+      "loss": 0.1326,
+      "step": 2218
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8835616438356166e-05,
+      "loss": 0.1184,
+      "step": 2219
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 0.1323,
+      "step": 2220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.897260273972603e-05,
+      "loss": 0.1366,
+      "step": 2221
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.904109589041096e-05,
+      "loss": 0.1305,
+      "step": 2222
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.910958904109589e-05,
+      "loss": 0.1129,
+      "step": 2223
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 0.1021,
+      "step": 2224
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9246575342465756e-05,
+      "loss": 0.1168,
+      "step": 2225
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9315068493150686e-05,
+      "loss": 0.1084,
+      "step": 2226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9383561643835617e-05,
+      "loss": 0.1189,
+      "step": 2227
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.945205479452055e-05,
+      "loss": 0.1101,
+      "step": 2228
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.952054794520548e-05,
+      "loss": 0.1257,
+      "step": 2229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9589041095890412e-05,
+      "loss": 0.0939,
+      "step": 2230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9657534246575346e-05,
+      "loss": 0.1182,
+      "step": 2231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 0.1067,
+      "step": 2232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9794520547945207e-05,
+      "loss": 0.1023,
+      "step": 2233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9863013698630137e-05,
+      "loss": 0.1046,
+      "step": 2234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.993150684931507e-05,
+      "loss": 0.1086,
+      "step": 2235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2e-05,
+      "loss": 0.1102,
+      "step": 2236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9999999444824108e-05,
+      "loss": 0.1135,
+      "step": 2237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9999997779296502e-05,
+      "loss": 0.1215,
+      "step": 2238
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9999995003417356e-05,
+      "loss": 0.1305,
+      "step": 2239
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.999999111718698e-05,
+      "loss": 0.1306,
+      "step": 2240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9999986120605816e-05,
+      "loss": 0.1079,
+      "step": 2241
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.999998001367441e-05,
+      "loss": 0.1194,
+      "step": 2242
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.999997279639344e-05,
+      "loss": 0.1097,
+      "step": 2243
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.999996446876371e-05,
+      "loss": 0.0983,
+      "step": 2244
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9999955030786143e-05,
+      "loss": 0.1024,
+      "step": 2245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.999994448246179e-05,
+      "loss": 0.1219,
+      "step": 2246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999932823791816e-05,
+      "loss": 0.1221,
+      "step": 2247
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999920054777522e-05,
+      "loss": 0.0986,
+      "step": 2248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.999990617542032e-05,
+      "loss": 0.1003,
+      "step": 2249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.999989118572176e-05,
+      "loss": 0.1243,
+      "step": 2250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999875085683498e-05,
+      "loss": 0.1096,
+      "step": 2251
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999857875307324e-05,
+      "loss": 0.1184,
+      "step": 2252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999839554595152e-05,
+      "loss": 0.1069,
+      "step": 2253
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999820123549014e-05,
+      "loss": 0.108,
+      "step": 2254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999799582171066e-05,
+      "loss": 0.1281,
+      "step": 2255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.999977793046359e-05,
+      "loss": 0.1147,
+      "step": 2256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999755168428986e-05,
+      "loss": 0.1235,
+      "step": 2257
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999731296069788e-05,
+      "loss": 0.1152,
+      "step": 2258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999706313388645e-05,
+      "loss": 0.1064,
+      "step": 2259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.999968022038833e-05,
+      "loss": 0.1268,
+      "step": 2260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.999965301707174e-05,
+      "loss": 0.0947,
+      "step": 2261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.99996247034419e-05,
+      "loss": 0.1097,
+      "step": 2262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999595279501944e-05,
+      "loss": 0.1068,
+      "step": 2263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9999564745255148e-05,
+      "loss": 0.1154,
+      "step": 2264
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.99995331007049e-05,
+      "loss": 0.1127,
+      "step": 2265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999950034585471e-05,
+      "loss": 0.0927,
+      "step": 2266
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999946648070822e-05,
+      "loss": 0.1123,
+      "step": 2267
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9999431505269185e-05,
+      "loss": 0.1257,
+      "step": 2268
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9999395419541494e-05,
+      "loss": 0.1312,
+      "step": 2269
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999935822352915e-05,
+      "loss": 0.0999,
+      "step": 2270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9999319917236287e-05,
+      "loss": 0.1116,
+      "step": 2271
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9999280500667154e-05,
+      "loss": 0.1045,
+      "step": 2272
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999923997382613e-05,
+      "loss": 0.1339,
+      "step": 2273
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9999198336717712e-05,
+      "loss": 0.1106,
+      "step": 2274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9999155589346528e-05,
+      "loss": 0.1139,
+      "step": 2275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999911173171732e-05,
+      "loss": 0.1278,
+      "step": 2276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999906676383496e-05,
+      "loss": 0.1246,
+      "step": 2277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999902068570444e-05,
+      "loss": 0.1158,
+      "step": 2278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9998973497330878e-05,
+      "loss": 0.1003,
+      "step": 2279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9998925198719514e-05,
+      "loss": 0.1182,
+      "step": 2280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9998875789875707e-05,
+      "loss": 0.1033,
+      "step": 2281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.999882527080495e-05,
+      "loss": 0.1122,
+      "step": 2282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9998773641512842e-05,
+      "loss": 0.1122,
+      "step": 2283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9998720902005125e-05,
+      "loss": 0.1372,
+      "step": 2284
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998667052287647e-05,
+      "loss": 0.1041,
+      "step": 2285
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998612092366396e-05,
+      "loss": 0.1208,
+      "step": 2286
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998556022247468e-05,
+      "loss": 0.1027,
+      "step": 2287
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.999849884193709e-05,
+      "loss": 0.1102,
+      "step": 2288
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998440551441618e-05,
+      "loss": 0.0967,
+      "step": 2289
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998381150767514e-05,
+      "loss": 0.1151,
+      "step": 2290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998320639921377e-05,
+      "loss": 0.1259,
+      "step": 2291
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.999825901890993e-05,
+      "loss": 0.12,
+      "step": 2292
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998196287740006e-05,
+      "loss": 0.1006,
+      "step": 2293
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998132446418583e-05,
+      "loss": 0.1083,
+      "step": 2294
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9998067494952736e-05,
+      "loss": 0.1058,
+      "step": 2295
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.999800143334969e-05,
+      "loss": 0.1094,
+      "step": 2296
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9997934261616768e-05,
+      "loss": 0.131,
+      "step": 2297
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9997865979761436e-05,
+      "loss": 0.1079,
+      "step": 2298
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9997796587791276e-05,
+      "loss": 0.1113,
+      "step": 2299
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9997726085713993e-05,
+      "loss": 0.0981,
+      "step": 2300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.999765447353741e-05,
+      "loss": 0.1425,
+      "step": 2301
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.999758175126948e-05,
+      "loss": 0.1195,
+      "step": 2302
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.999750791891828e-05,
+      "loss": 0.1218,
+      "step": 2303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9997432976492006e-05,
+      "loss": 0.1066,
+      "step": 2304
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.999735692399898e-05,
+      "loss": 0.1117,
+      "step": 2305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9997279761447652e-05,
+      "loss": 0.1285,
+      "step": 2306
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9997201488846585e-05,
+      "loss": 0.0983,
+      "step": 2307
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9997122106204466e-05,
+      "loss": 0.1136,
+      "step": 2308
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9997041613530115e-05,
+      "loss": 0.1183,
+      "step": 2309
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996960010832466e-05,
+      "loss": 0.1249,
+      "step": 2310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996877298120583e-05,
+      "loss": 0.1099,
+      "step": 2311
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996793475403647e-05,
+      "loss": 0.1104,
+      "step": 2312
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996708542690966e-05,
+      "loss": 0.1389,
+      "step": 2313
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996622499991973e-05,
+      "loss": 0.0979,
+      "step": 2314
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996535347316217e-05,
+      "loss": 0.1025,
+      "step": 2315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996447084673383e-05,
+      "loss": 0.0995,
+      "step": 2316
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996357712073263e-05,
+      "loss": 0.0848,
+      "step": 2317
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996267229525783e-05,
+      "loss": 0.1089,
+      "step": 2318
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.999617563704099e-05,
+      "loss": 0.1357,
+      "step": 2319
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9996082934629057e-05,
+      "loss": 0.1154,
+      "step": 2320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9995989122300275e-05,
+      "loss": 0.0986,
+      "step": 2321
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9995894200065055e-05,
+      "loss": 0.1348,
+      "step": 2322
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9995798167933945e-05,
+      "loss": 0.1124,
+      "step": 2323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9995701025917607e-05,
+      "loss": 0.1043,
+      "step": 2324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9995602774026826e-05,
+      "loss": 0.0991,
+      "step": 2325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9995503412272504e-05,
+      "loss": 0.0846,
+      "step": 2326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9995402940665684e-05,
+      "loss": 0.1342,
+      "step": 2327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.999530135921752e-05,
+      "loss": 0.1152,
+      "step": 2328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9995198667939285e-05,
+      "loss": 0.1218,
+      "step": 2329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9995094866842386e-05,
+      "loss": 0.1224,
+      "step": 2330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9994989955938352e-05,
+      "loss": 0.113,
+      "step": 2331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9994883935238822e-05,
+      "loss": 0.1079,
+      "step": 2332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9994776804755576e-05,
+      "loss": 0.1263,
+      "step": 2333
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.999466856450051e-05,
+      "loss": 0.118,
+      "step": 2334
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9994559214485637e-05,
+      "loss": 0.1272,
+      "step": 2335
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9994448754723098e-05,
+      "loss": 0.1041,
+      "step": 2336
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9994337185225164e-05,
+      "loss": 0.115,
+      "step": 2337
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.999422450600422e-05,
+      "loss": 0.1074,
+      "step": 2338
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.999411071707278e-05,
+      "loss": 0.1052,
+      "step": 2339
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9993995818443473e-05,
+      "loss": 0.1201,
+      "step": 2340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.999387981012906e-05,
+      "loss": 0.1083,
+      "step": 2341
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9993762692142423e-05,
+      "loss": 0.1025,
+      "step": 2342
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9993644464496566e-05,
+      "loss": 0.1016,
+      "step": 2343
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9993525127204615e-05,
+      "loss": 0.1157,
+      "step": 2344
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.999340468027982e-05,
+      "loss": 0.1221,
+      "step": 2345
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.999328312373556e-05,
+      "loss": 0.125,
+      "step": 2346
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9993160457585325e-05,
+      "loss": 0.1139,
+      "step": 2347
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9993036681842737e-05,
+      "loss": 0.0988,
+      "step": 2348
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9992911796521543e-05,
+      "loss": 0.1088,
+      "step": 2349
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.999278580163561e-05,
+      "loss": 0.1266,
+      "step": 2350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9992658697198922e-05,
+      "loss": 0.1116,
+      "step": 2351
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9992530483225596e-05,
+      "loss": 0.1276,
+      "step": 2352
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9992401159729866e-05,
+      "loss": 0.1127,
+      "step": 2353
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9992270726726096e-05,
+      "loss": 0.1299,
+      "step": 2354
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9992139184228766e-05,
+      "loss": 0.1111,
+      "step": 2355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.999200653225248e-05,
+      "loss": 0.1144,
+      "step": 2356
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.999187277081197e-05,
+      "loss": 0.1095,
+      "step": 2357
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9991737899922086e-05,
+      "loss": 0.1243,
+      "step": 2358
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9991601919597802e-05,
+      "loss": 0.1219,
+      "step": 2359
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9991464829854223e-05,
+      "loss": 0.0991,
+      "step": 2360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9991326630706564e-05,
+      "loss": 0.1125,
+      "step": 2361
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9991187322170175e-05,
+      "loss": 0.0981,
+      "step": 2362
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.999104690426052e-05,
+      "loss": 0.1091,
+      "step": 2363
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.999090537699319e-05,
+      "loss": 0.1157,
+      "step": 2364
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9990762740383908e-05,
+      "loss": 0.0985,
+      "step": 2365
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9990618994448502e-05,
+      "loss": 0.1228,
+      "step": 2366
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9990474139202936e-05,
+      "loss": 0.0956,
+      "step": 2367
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9990328174663297e-05,
+      "loss": 0.1219,
+      "step": 2368
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9990181100845786e-05,
+      "loss": 0.1344,
+      "step": 2369
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.999003291776674e-05,
+      "loss": 0.1069,
+      "step": 2370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9989883625442603e-05,
+      "loss": 0.1412,
+      "step": 2371
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9989733223889964e-05,
+      "loss": 0.1077,
+      "step": 2372
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9989581713125516e-05,
+      "loss": 0.1097,
+      "step": 2373
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998942909316608e-05,
+      "loss": 0.1003,
+      "step": 2374
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9989275364028608e-05,
+      "loss": 0.1096,
+      "step": 2375
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9989120525730165e-05,
+      "loss": 0.0952,
+      "step": 2376
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998896457828795e-05,
+      "loss": 0.1475,
+      "step": 2377
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9988807521719264e-05,
+      "loss": 0.1169,
+      "step": 2378
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9988649356041562e-05,
+      "loss": 0.1124,
+      "step": 2379
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9988490081272397e-05,
+      "loss": 0.1097,
+      "step": 2380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.998832969742946e-05,
+      "loss": 0.105,
+      "step": 2381
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998816820453055e-05,
+      "loss": 0.1174,
+      "step": 2382
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998800560259361e-05,
+      "loss": 0.1066,
+      "step": 2383
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9987841891636687e-05,
+      "loss": 0.1156,
+      "step": 2384
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998767707167796e-05,
+      "loss": 0.1029,
+      "step": 2385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998751114273573e-05,
+      "loss": 0.123,
+      "step": 2386
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998734410482842e-05,
+      "loss": 0.1256,
+      "step": 2387
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9987175957974577e-05,
+      "loss": 0.1252,
+      "step": 2388
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9987006702192875e-05,
+      "loss": 0.1088,
+      "step": 2389
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9986836337502106e-05,
+      "loss": 0.1097,
+      "step": 2390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9986664863921183e-05,
+      "loss": 0.1273,
+      "step": 2391
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9986492281469147e-05,
+      "loss": 0.1191,
+      "step": 2392
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9986318590165162e-05,
+      "loss": 0.1133,
+      "step": 2393
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9986143790028513e-05,
+      "loss": 0.1188,
+      "step": 2394
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998596788107861e-05,
+      "loss": 0.1151,
+      "step": 2395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9985790863334988e-05,
+      "loss": 0.1085,
+      "step": 2396
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9985612736817295e-05,
+      "loss": 0.1066,
+      "step": 2397
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9985433501545312e-05,
+      "loss": 0.1206,
+      "step": 2398
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998525315753894e-05,
+      "loss": 0.099,
+      "step": 2399
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.998507170481821e-05,
+      "loss": 0.0977,
+      "step": 2400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9984889143403263e-05,
+      "loss": 0.1246,
+      "step": 2401
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998470547331437e-05,
+      "loss": 0.1332,
+      "step": 2402
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9984520694571926e-05,
+      "loss": 0.1086,
+      "step": 2403
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998433480719645e-05,
+      "loss": 0.11,
+      "step": 2404
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9984147811208576e-05,
+      "loss": 0.1157,
+      "step": 2405
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9983959706629076e-05,
+      "loss": 0.1073,
+      "step": 2406
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9983770493478828e-05,
+      "loss": 0.1227,
+      "step": 2407
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9983580171778846e-05,
+      "loss": 0.1014,
+      "step": 2408
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998338874155026e-05,
+      "loss": 0.1156,
+      "step": 2409
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998319620281433e-05,
+      "loss": 0.1163,
+      "step": 2410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9983002555592428e-05,
+      "loss": 0.0846,
+      "step": 2411
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9982807799906057e-05,
+      "loss": 0.1331,
+      "step": 2412
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9982611935776847e-05,
+      "loss": 0.0901,
+      "step": 2413
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9982414963226544e-05,
+      "loss": 0.108,
+      "step": 2414
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9982216882277013e-05,
+      "loss": 0.1176,
+      "step": 2415
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9982017692950254e-05,
+      "loss": 0.113,
+      "step": 2416
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9981817395268384e-05,
+      "loss": 0.0984,
+      "step": 2417
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.998161598925364e-05,
+      "loss": 0.1364,
+      "step": 2418
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9981413474928386e-05,
+      "loss": 0.1119,
+      "step": 2419
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9981209852315108e-05,
+      "loss": 0.108,
+      "step": 2420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9981005121436418e-05,
+      "loss": 0.0952,
+      "step": 2421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9980799282315045e-05,
+      "loss": 0.1056,
+      "step": 2422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9980592334973847e-05,
+      "loss": 0.0973,
+      "step": 2423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.99803842794358e-05,
+      "loss": 0.123,
+      "step": 2424
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9980175115724007e-05,
+      "loss": 0.1057,
+      "step": 2425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979964843861693e-05,
+      "loss": 0.0896,
+      "step": 2426
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979753463872203e-05,
+      "loss": 0.1097,
+      "step": 2427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997954097577901e-05,
+      "loss": 0.097,
+      "step": 2428
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979327379605707e-05,
+      "loss": 0.1005,
+      "step": 2429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9979112675376014e-05,
+      "loss": 0.1123,
+      "step": 2430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997889686311376e-05,
+      "loss": 0.1062,
+      "step": 2431
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978679942842922e-05,
+      "loss": 0.1061,
+      "step": 2432
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978461914587577e-05,
+      "loss": 0.1233,
+      "step": 2433
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9978242778371934e-05,
+      "loss": 0.0992,
+      "step": 2434
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997802253422033e-05,
+      "loss": 0.1165,
+      "step": 2435
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9977801182157213e-05,
+      "loss": 0.1191,
+      "step": 2436
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.997757872220717e-05,
+      "loss": 0.1315,
+      "step": 2437
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9977355154394885e-05,
+      "loss": 0.1123,
+      "step": 2438
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.99771304787452e-05,
+      "loss": 0.1031,
+      "step": 2439
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9976904695283057e-05,
+      "loss": 0.1125,
+      "step": 2440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997667780403352e-05,
+      "loss": 0.1123,
+      "step": 2441
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9976449805021788e-05,
+      "loss": 0.1311,
+      "step": 2442
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9976220698273177e-05,
+      "loss": 0.145,
+      "step": 2443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997599048381312e-05,
+      "loss": 0.1112,
+      "step": 2444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9975759161667182e-05,
+      "loss": 0.1012,
+      "step": 2445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997552673186105e-05,
+      "loss": 0.0972,
+      "step": 2446
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9975293194420532e-05,
+      "loss": 0.1156,
+      "step": 2447
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9975058549371553e-05,
+      "loss": 0.1198,
+      "step": 2448
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9974822796740174e-05,
+      "loss": 0.0966,
+      "step": 2449
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9974585936552565e-05,
+      "loss": 0.1211,
+      "step": 2450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9974347968835036e-05,
+      "loss": 0.1267,
+      "step": 2451
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9974108893614e-05,
+      "loss": 0.1085,
+      "step": 2452
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9973868710916004e-05,
+      "loss": 0.1068,
+      "step": 2453
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997362742076772e-05,
+      "loss": 0.1086,
+      "step": 2454
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9973385023195943e-05,
+      "loss": 0.1047,
+      "step": 2455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997314151822758e-05,
+      "loss": 0.1127,
+      "step": 2456
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9972896905889674e-05,
+      "loss": 0.106,
+      "step": 2457
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.997265118620938e-05,
+      "loss": 0.1191,
+      "step": 2458
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9972404359213987e-05,
+      "loss": 0.1129,
+      "step": 2459
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9972156424930898e-05,
+      "loss": 0.088,
+      "step": 2460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9971907383387644e-05,
+      "loss": 0.1132,
+      "step": 2461
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997165723461188e-05,
+      "loss": 0.1041,
+      "step": 2462
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9971405978631378e-05,
+      "loss": 0.1138,
+      "step": 2463
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9971153615474036e-05,
+      "loss": 0.1117,
+      "step": 2464
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9970900145167877e-05,
+      "loss": 0.1068,
+      "step": 2465
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997064556774104e-05,
+      "loss": 0.1378,
+      "step": 2466
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.99703898832218e-05,
+      "loss": 0.0978,
+      "step": 2467
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.997013309163854e-05,
+      "loss": 0.1066,
+      "step": 2468
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9969875193019783e-05,
+      "loss": 0.1233,
+      "step": 2469
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.996961618739415e-05,
+      "loss": 0.1068,
+      "step": 2470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9969356074790412e-05,
+      "loss": 0.1257,
+      "step": 2471
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9969094855237446e-05,
+      "loss": 0.116,
+      "step": 2472
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9968832528764256e-05,
+      "loss": 0.1217,
+      "step": 2473
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.996856909539997e-05,
+      "loss": 0.0988,
+      "step": 2474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.996830455517384e-05,
+      "loss": 0.1084,
+      "step": 2475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9968038908115237e-05,
+      "loss": 0.1028,
+      "step": 2476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.996777215425366e-05,
+      "loss": 0.1204,
+      "step": 2477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9967504293618725e-05,
+      "loss": 0.106,
+      "step": 2478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9967235326240174e-05,
+      "loss": 0.0992,
+      "step": 2479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9966965252147873e-05,
+      "loss": 0.12,
+      "step": 2480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.996669407137181e-05,
+      "loss": 0.1134,
+      "step": 2481
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9966421783942094e-05,
+      "loss": 0.108,
+      "step": 2482
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9966148389888964e-05,
+      "loss": 0.1072,
+      "step": 2483
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9965873889242768e-05,
+      "loss": 0.1155,
+      "step": 2484
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9965598282033994e-05,
+      "loss": 0.1353,
+      "step": 2485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9965321568293233e-05,
+      "loss": 0.0978,
+      "step": 2486
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9965043748051222e-05,
+      "loss": 0.0868,
+      "step": 2487
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.99647648213388e-05,
+      "loss": 0.1044,
+      "step": 2488
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.996448478818694e-05,
+      "loss": 0.0964,
+      "step": 2489
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.996420364862674e-05,
+      "loss": 0.1057,
+      "step": 2490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9963921402689412e-05,
+      "loss": 0.1013,
+      "step": 2491
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9963638050406297e-05,
+      "loss": 0.0937,
+      "step": 2492
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9963353591808853e-05,
+      "loss": 0.0974,
+      "step": 2493
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9963068026928673e-05,
+      "loss": 0.1041,
+      "step": 2494
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9962781355797456e-05,
+      "loss": 0.1201,
+      "step": 2495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9962493578447038e-05,
+      "loss": 0.0986,
+      "step": 2496
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9962204694909368e-05,
+      "loss": 0.1105,
+      "step": 2497
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9961914705216526e-05,
+      "loss": 0.1044,
+      "step": 2498
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9961623609400712e-05,
+      "loss": 0.1163,
+      "step": 2499
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9961331407494245e-05,
+      "loss": 0.1045,
+      "step": 2500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.996103809952957e-05,
+      "loss": 0.1099,
+      "step": 2501
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.996074368553926e-05,
+      "loss": 0.1121,
+      "step": 2502
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9960448165555992e-05,
+      "loss": 0.1217,
+      "step": 2503
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9960151539612593e-05,
+      "loss": 0.1217,
+      "step": 2504
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9959853807741992e-05,
+      "loss": 0.1008,
+      "step": 2505
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.995955496997725e-05,
+      "loss": 0.1028,
+      "step": 2506
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9959255026351548e-05,
+      "loss": 0.1018,
+      "step": 2507
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.995895397689819e-05,
+      "loss": 0.1038,
+      "step": 2508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9958651821650604e-05,
+      "loss": 0.121,
+      "step": 2509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9958348560642336e-05,
+      "loss": 0.1243,
+      "step": 2510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9958044193907063e-05,
+      "loss": 0.1147,
+      "step": 2511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.995773872147858e-05,
+      "loss": 0.1019,
+      "step": 2512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.99574321433908e-05,
+      "loss": 0.1324,
+      "step": 2513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9957124459677772e-05,
+      "loss": 0.1208,
+      "step": 2514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9956815670373655e-05,
+      "loss": 0.1064,
+      "step": 2515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.995650577551274e-05,
+      "loss": 0.1093,
+      "step": 2516
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9956194775129426e-05,
+      "loss": 0.1111,
+      "step": 2517
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9955882669258256e-05,
+      "loss": 0.1042,
+      "step": 2518
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.995556945793388e-05,
+      "loss": 0.1158,
+      "step": 2519
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9955255141191074e-05,
+      "loss": 0.1129,
+      "step": 2520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.995493971906474e-05,
+      "loss": 0.1195,
+      "step": 2521
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.99546231915899e-05,
+      "loss": 0.1184,
+      "step": 2522
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9954305558801703e-05,
+      "loss": 0.0773,
+      "step": 2523
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9953986820735414e-05,
+      "loss": 0.1033,
+      "step": 2524
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9953666977426428e-05,
+      "loss": 0.0936,
+      "step": 2525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.995334602891025e-05,
+      "loss": 0.1229,
+      "step": 2526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9953023975222527e-05,
+      "loss": 0.1353,
+      "step": 2527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9952700816399014e-05,
+      "loss": 0.0879,
+      "step": 2528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.995237655247559e-05,
+      "loss": 0.1029,
+      "step": 2529
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9952051183488262e-05,
+      "loss": 0.1334,
+      "step": 2530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9951724709473163e-05,
+      "loss": 0.1127,
+      "step": 2531
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9951397130466535e-05,
+      "loss": 0.1216,
+      "step": 2532
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9951068446504753e-05,
+      "loss": 0.1156,
+      "step": 2533
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9950738657624318e-05,
+      "loss": 0.097,
+      "step": 2534
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9950407763861837e-05,
+      "loss": 0.1085,
+      "step": 2535
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9950075765254063e-05,
+      "loss": 0.1077,
+      "step": 2536
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994974266183785e-05,
+      "loss": 0.119,
+      "step": 2537
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9949408453650194e-05,
+      "loss": 0.1215,
+      "step": 2538
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994907314072819e-05,
+      "loss": 0.1174,
+      "step": 2539
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9948736723109082e-05,
+      "loss": 0.0901,
+      "step": 2540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994839920083022e-05,
+      "loss": 0.1339,
+      "step": 2541
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9948060573929075e-05,
+      "loss": 0.0952,
+      "step": 2542
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9947720842443255e-05,
+      "loss": 0.1011,
+      "step": 2543
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9947380006410484e-05,
+      "loss": 0.0992,
+      "step": 2544
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9947038065868597e-05,
+      "loss": 0.1184,
+      "step": 2545
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994669502085557e-05,
+      "loss": 0.1116,
+      "step": 2546
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9946350871409484e-05,
+      "loss": 0.1107,
+      "step": 2547
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9946005617568563e-05,
+      "loss": 0.1415,
+      "step": 2548
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9945659259371133e-05,
+      "loss": 0.0999,
+      "step": 2549
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9945311796855654e-05,
+      "loss": 0.1133,
+      "step": 2550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9944963230060713e-05,
+      "loss": 0.1217,
+      "step": 2551
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9944613559025005e-05,
+      "loss": 0.1029,
+      "step": 2552
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994426278378736e-05,
+      "loss": 0.156,
+      "step": 2553
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9943910904386725e-05,
+      "loss": 0.0997,
+      "step": 2554
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994355792086217e-05,
+      "loss": 0.1135,
+      "step": 2555
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.994320383325289e-05,
+      "loss": 0.104,
+      "step": 2556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9942848641598204e-05,
+      "loss": 0.1072,
+      "step": 2557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9942492345937545e-05,
+      "loss": 0.1138,
+      "step": 2558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9942134946310477e-05,
+      "loss": 0.1074,
+      "step": 2559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9941776442756685e-05,
+      "loss": 0.1111,
+      "step": 2560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9941416835315977e-05,
+      "loss": 0.1001,
+      "step": 2561
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9941056124028277e-05,
+      "loss": 0.1281,
+      "step": 2562
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9940694308933638e-05,
+      "loss": 0.1061,
+      "step": 2563
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.994033139007224e-05,
+      "loss": 0.1021,
+      "step": 2564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9939967367484372e-05,
+      "loss": 0.1022,
+      "step": 2565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9939602241210457e-05,
+      "loss": 0.1006,
+      "step": 2566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993923601129104e-05,
+      "loss": 0.1349,
+      "step": 2567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9938868677766778e-05,
+      "loss": 0.1123,
+      "step": 2568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993850024067846e-05,
+      "loss": 0.083,
+      "step": 2569
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9938130700067005e-05,
+      "loss": 0.1055,
+      "step": 2570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9937760055973433e-05,
+      "loss": 0.1011,
+      "step": 2571
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.99373883084389e-05,
+      "loss": 0.0911,
+      "step": 2572
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.993701545750469e-05,
+      "loss": 0.0995,
+      "step": 2573
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9936641503212195e-05,
+      "loss": 0.1179,
+      "step": 2574
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9936266445602944e-05,
+      "loss": 0.1049,
+      "step": 2575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9935890284718574e-05,
+      "loss": 0.103,
+      "step": 2576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993551302060086e-05,
+      "loss": 0.0889,
+      "step": 2577
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993513465329169e-05,
+      "loss": 0.1016,
+      "step": 2578
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993475518283307e-05,
+      "loss": 0.0902,
+      "step": 2579
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993437460926714e-05,
+      "loss": 0.1182,
+      "step": 2580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9933992932636154e-05,
+      "loss": 0.1127,
+      "step": 2581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993361015298249e-05,
+      "loss": 0.0958,
+      "step": 2582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993322627034866e-05,
+      "loss": 0.0984,
+      "step": 2583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9932841284777277e-05,
+      "loss": 0.1021,
+      "step": 2584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9932455196311093e-05,
+      "loss": 0.104,
+      "step": 2585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993206800499298e-05,
+      "loss": 0.1215,
+      "step": 2586
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931679710865923e-05,
+      "loss": 0.1014,
+      "step": 2587
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9931290313973043e-05,
+      "loss": 0.0844,
+      "step": 2588
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.993089981435757e-05,
+      "loss": 0.1233,
+      "step": 2589
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930508212062874e-05,
+      "loss": 0.1096,
+      "step": 2590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9930115507132424e-05,
+      "loss": 0.1052,
+      "step": 2591
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929721699609828e-05,
+      "loss": 0.1128,
+      "step": 2592
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9929326789538818e-05,
+      "loss": 0.1163,
+      "step": 2593
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.992893077696324e-05,
+      "loss": 0.1201,
+      "step": 2594
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9928533661927064e-05,
+      "loss": 0.1121,
+      "step": 2595
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9928135444474382e-05,
+      "loss": 0.1279,
+      "step": 2596
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927736124649413e-05,
+      "loss": 0.1227,
+      "step": 2597
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9927335702496496e-05,
+      "loss": 0.104,
+      "step": 2598
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992693417806009e-05,
+      "loss": 0.1222,
+      "step": 2599
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992653155138478e-05,
+      "loss": 0.1014,
+      "step": 2600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9926127822515266e-05,
+      "loss": 0.0994,
+      "step": 2601
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9925722991496386e-05,
+      "loss": 0.1014,
+      "step": 2602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9925317058373086e-05,
+      "loss": 0.1034,
+      "step": 2603
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9924910023190434e-05,
+      "loss": 0.1108,
+      "step": 2604
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9924501885993635e-05,
+      "loss": 0.1143,
+      "step": 2605
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9924092646828e-05,
+      "loss": 0.1198,
+      "step": 2606
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9923682305738966e-05,
+      "loss": 0.0968,
+      "step": 2607
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9923270862772104e-05,
+      "loss": 0.1078,
+      "step": 2608
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992285831797309e-05,
+      "loss": 0.097,
+      "step": 2609
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992244467138774e-05,
+      "loss": 0.1145,
+      "step": 2610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9922029923061973e-05,
+      "loss": 0.1036,
+      "step": 2611
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9921614073041847e-05,
+      "loss": 0.0979,
+      "step": 2612
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.992119712137354e-05,
+      "loss": 0.1008,
+      "step": 2613
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9920779068103336e-05,
+      "loss": 0.0892,
+      "step": 2614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9920359913277667e-05,
+      "loss": 0.094,
+      "step": 2615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9919939656943062e-05,
+      "loss": 0.13,
+      "step": 2616
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9919518299146196e-05,
+      "loss": 0.1409,
+      "step": 2617
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9919095839933846e-05,
+      "loss": 0.1027,
+      "step": 2618
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9918672279352923e-05,
+      "loss": 0.0986,
+      "step": 2619
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9918247617450454e-05,
+      "loss": 0.1056,
+      "step": 2620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9917821854273597e-05,
+      "loss": 0.1075,
+      "step": 2621
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.991739498986962e-05,
+      "loss": 0.1138,
+      "step": 2622
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9916967024285928e-05,
+      "loss": 0.1013,
+      "step": 2623
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9916537957570035e-05,
+      "loss": 0.1064,
+      "step": 2624
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9916107789769583e-05,
+      "loss": 0.0984,
+      "step": 2625
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9915676520932334e-05,
+      "loss": 0.1206,
+      "step": 2626
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9915244151106177e-05,
+      "loss": 0.1002,
+      "step": 2627
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.991481068033912e-05,
+      "loss": 0.1007,
+      "step": 2628
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9914376108679295e-05,
+      "loss": 0.1304,
+      "step": 2629
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.991394043617495e-05,
+      "loss": 0.0892,
+      "step": 2630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9913503662874462e-05,
+      "loss": 0.0947,
+      "step": 2631
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.991306578882633e-05,
+      "loss": 0.106,
+      "step": 2632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9912626814079172e-05,
+      "loss": 0.1014,
+      "step": 2633
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.991218673868173e-05,
+      "loss": 0.1221,
+      "step": 2634
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9911745562682866e-05,
+      "loss": 0.1178,
+      "step": 2635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9911303286131574e-05,
+      "loss": 0.1084,
+      "step": 2636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.991085990907695e-05,
+      "loss": 0.1016,
+      "step": 2637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9910415431568233e-05,
+      "loss": 0.1116,
+      "step": 2638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990996985365477e-05,
+      "loss": 0.1058,
+      "step": 2639
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990952317538604e-05,
+      "loss": 0.1169,
+      "step": 2640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9909075396811643e-05,
+      "loss": 0.1244,
+      "step": 2641
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990862651798129e-05,
+      "loss": 0.1034,
+      "step": 2642
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990817653894483e-05,
+      "loss": 0.0996,
+      "step": 2643
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990772545975222e-05,
+      "loss": 0.1046,
+      "step": 2644
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990727328045355e-05,
+      "loss": 0.1323,
+      "step": 2645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9906820001099024e-05,
+      "loss": 0.1132,
+      "step": 2646
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9906365621738975e-05,
+      "loss": 0.1062,
+      "step": 2647
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9905910142423853e-05,
+      "loss": 0.1002,
+      "step": 2648
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9905453563204237e-05,
+      "loss": 0.1048,
+      "step": 2649
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990499588413082e-05,
+      "loss": 0.1053,
+      "step": 2650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990453710525442e-05,
+      "loss": 0.1047,
+      "step": 2651
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9904077226625978e-05,
+      "loss": 0.1033,
+      "step": 2652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.990361624829656e-05,
+      "loss": 0.1089,
+      "step": 2653
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.990315417031734e-05,
+      "loss": 0.0884,
+      "step": 2654
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.990269099273964e-05,
+      "loss": 0.1042,
+      "step": 2655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9902226715614876e-05,
+      "loss": 0.109,
+      "step": 2656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9901761338994606e-05,
+      "loss": 0.0875,
+      "step": 2657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9901294862930504e-05,
+      "loss": 0.1284,
+      "step": 2658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.990082728747436e-05,
+      "loss": 0.1331,
+      "step": 2659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.99003586126781e-05,
+      "loss": 0.1014,
+      "step": 2660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9899888838593754e-05,
+      "loss": 0.11,
+      "step": 2661
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9899417965273486e-05,
+      "loss": 0.0977,
+      "step": 2662
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9898945992769584e-05,
+      "loss": 0.0952,
+      "step": 2663
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.989847292113445e-05,
+      "loss": 0.121,
+      "step": 2664
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.989799875042061e-05,
+      "loss": 0.129,
+      "step": 2665
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9897523480680715e-05,
+      "loss": 0.1327,
+      "step": 2666
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.989704711196754e-05,
+      "loss": 0.1077,
+      "step": 2667
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.989656964433397e-05,
+      "loss": 0.1157,
+      "step": 2668
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.989609107783303e-05,
+      "loss": 0.1034,
+      "step": 2669
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9895611412517856e-05,
+      "loss": 0.1145,
+      "step": 2670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9895130648441706e-05,
+      "loss": 0.1082,
+      "step": 2671
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.989464878565796e-05,
+      "loss": 0.1152,
+      "step": 2672
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9894165824220125e-05,
+      "loss": 0.0936,
+      "step": 2673
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9893681764181823e-05,
+      "loss": 0.097,
+      "step": 2674
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9893196605596804e-05,
+      "loss": 0.1193,
+      "step": 2675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.989271034851894e-05,
+      "loss": 0.1179,
+      "step": 2676
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.989222299300222e-05,
+      "loss": 0.1018,
+      "step": 2677
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9891734539100754e-05,
+      "loss": 0.12,
+      "step": 2678
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9891244986868784e-05,
+      "loss": 0.0959,
+      "step": 2679
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9890754336360666e-05,
+      "loss": 0.1235,
+      "step": 2680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9890262587630877e-05,
+      "loss": 0.0999,
+      "step": 2681
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9889769740734018e-05,
+      "loss": 0.1025,
+      "step": 2682
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9889275795724815e-05,
+      "loss": 0.1302,
+      "step": 2683
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9888780752658115e-05,
+      "loss": 0.119,
+      "step": 2684
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.988828461158888e-05,
+      "loss": 0.0985,
+      "step": 2685
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9887787372572203e-05,
+      "loss": 0.1208,
+      "step": 2686
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9887289035663294e-05,
+      "loss": 0.0854,
+      "step": 2687
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9886789600917483e-05,
+      "loss": 0.1024,
+      "step": 2688
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.988628906839023e-05,
+      "loss": 0.1095,
+      "step": 2689
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.988578743813711e-05,
+      "loss": 0.105,
+      "step": 2690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9885284710213816e-05,
+      "loss": 0.1213,
+      "step": 2691
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9884780884676177e-05,
+      "loss": 0.0869,
+      "step": 2692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9884275961580134e-05,
+      "loss": 0.1124,
+      "step": 2693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.988376994098174e-05,
+      "loss": 0.0985,
+      "step": 2694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.98832628229372e-05,
+      "loss": 0.1223,
+      "step": 2695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9882754607502807e-05,
+      "loss": 0.1133,
+      "step": 2696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9882245294734998e-05,
+      "loss": 0.1151,
+      "step": 2697
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9881734884690317e-05,
+      "loss": 0.1084,
+      "step": 2698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.988122337742545e-05,
+      "loss": 0.1226,
+      "step": 2699
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.988071077299718e-05,
+      "loss": 0.0984,
+      "step": 2700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.988019707146243e-05,
+      "loss": 0.0913,
+      "step": 2701
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9879682272878236e-05,
+      "loss": 0.1003,
+      "step": 2702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9879166377301768e-05,
+      "loss": 0.0981,
+      "step": 2703
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9878649384790294e-05,
+      "loss": 0.1312,
+      "step": 2704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.987813129540123e-05,
+      "loss": 0.1107,
+      "step": 2705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9877612109192095e-05,
+      "loss": 0.1019,
+      "step": 2706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9877091826220543e-05,
+      "loss": 0.1099,
+      "step": 2707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9876570446544335e-05,
+      "loss": 0.0962,
+      "step": 2708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9876047970221376e-05,
+      "loss": 0.0933,
+      "step": 2709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9875524397309666e-05,
+      "loss": 0.1068,
+      "step": 2710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9874999727867347e-05,
+      "loss": 0.1266,
+      "step": 2711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9874473961952672e-05,
+      "loss": 0.1105,
+      "step": 2712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9873947099624026e-05,
+      "loss": 0.1115,
+      "step": 2713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9873419140939903e-05,
+      "loss": 0.1051,
+      "step": 2714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9872890085958923e-05,
+      "loss": 0.0846,
+      "step": 2715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.987235993473984e-05,
+      "loss": 0.111,
+      "step": 2716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.987182868734151e-05,
+      "loss": 0.1068,
+      "step": 2717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.987129634382292e-05,
+      "loss": 0.1057,
+      "step": 2718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9870762904243185e-05,
+      "loss": 0.1157,
+      "step": 2719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.987022836866153e-05,
+      "loss": 0.1198,
+      "step": 2720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9869692737137312e-05,
+      "loss": 0.1031,
+      "step": 2721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9869156009730005e-05,
+      "loss": 0.1124,
+      "step": 2722
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.98686181864992e-05,
+      "loss": 0.0928,
+      "step": 2723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9868079267504615e-05,
+      "loss": 0.1199,
+      "step": 2724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9867539252806093e-05,
+      "loss": 0.1187,
+      "step": 2725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9866998142463592e-05,
+      "loss": 0.0879,
+      "step": 2726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9866455936537195e-05,
+      "loss": 0.1251,
+      "step": 2727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9865912635087105e-05,
+      "loss": 0.0999,
+      "step": 2728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9865368238173647e-05,
+      "loss": 0.1118,
+      "step": 2729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.986482274585727e-05,
+      "loss": 0.1119,
+      "step": 2730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9864276158198544e-05,
+      "loss": 0.1102,
+      "step": 2731
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9863728475258156e-05,
+      "loss": 0.1183,
+      "step": 2732
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.986317969709692e-05,
+      "loss": 0.126,
+      "step": 2733
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.986262982377577e-05,
+      "loss": 0.1216,
+      "step": 2734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9862078855355755e-05,
+      "loss": 0.0964,
+      "step": 2735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9861526791898062e-05,
+      "loss": 0.0967,
+      "step": 2736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9860973633463986e-05,
+      "loss": 0.1044,
+      "step": 2737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9860419380114945e-05,
+      "loss": 0.1115,
+      "step": 2738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9859864031912482e-05,
+      "loss": 0.1078,
+      "step": 2739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9859307588918258e-05,
+      "loss": 0.1276,
+      "step": 2740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9858750051194062e-05,
+      "loss": 0.1178,
+      "step": 2741
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9858191418801794e-05,
+      "loss": 0.1078,
+      "step": 2742
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.985763169180349e-05,
+      "loss": 0.1005,
+      "step": 2743
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9857070870261292e-05,
+      "loss": 0.1179,
+      "step": 2744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9856508954237473e-05,
+      "loss": 0.1168,
+      "step": 2745
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.985594594379443e-05,
+      "loss": 0.1178,
+      "step": 2746
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9855381838994673e-05,
+      "loss": 0.1089,
+      "step": 2747
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9854816639900837e-05,
+      "loss": 0.1041,
+      "step": 2748
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9854250346575677e-05,
+      "loss": 0.1067,
+      "step": 2749
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9853682959082077e-05,
+      "loss": 0.0827,
+      "step": 2750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9853114477483036e-05,
+      "loss": 0.1034,
+      "step": 2751
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.985254490184167e-05,
+      "loss": 0.106,
+      "step": 2752
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9851974232221233e-05,
+      "loss": 0.1042,
+      "step": 2753
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9851402468685075e-05,
+      "loss": 0.1237,
+      "step": 2754
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.985082961129669e-05,
+      "loss": 0.1047,
+      "step": 2755
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9850255660119683e-05,
+      "loss": 0.1029,
+      "step": 2756
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.984968061521779e-05,
+      "loss": 0.0974,
+      "step": 2757
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.984910447665485e-05,
+      "loss": 0.1202,
+      "step": 2758
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9848527244494843e-05,
+      "loss": 0.1025,
+      "step": 2759
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9847948918801857e-05,
+      "loss": 0.1072,
+      "step": 2760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9847369499640108e-05,
+      "loss": 0.1008,
+      "step": 2761
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9846788987073933e-05,
+      "loss": 0.1241,
+      "step": 2762
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.984620738116779e-05,
+      "loss": 0.1075,
+      "step": 2763
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9845624681986254e-05,
+      "loss": 0.095,
+      "step": 2764
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.984504088959403e-05,
+      "loss": 0.1133,
+      "step": 2765
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9844456004055935e-05,
+      "loss": 0.0956,
+      "step": 2766
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9843870025436914e-05,
+      "loss": 0.1165,
+      "step": 2767
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.984328295380203e-05,
+      "loss": 0.1118,
+      "step": 2768
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9842694789216473e-05,
+      "loss": 0.0988,
+      "step": 2769
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9842105531745547e-05,
+      "loss": 0.1064,
+      "step": 2770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.984151518145468e-05,
+      "loss": 0.1008,
+      "step": 2771
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.984092373840942e-05,
+      "loss": 0.1006,
+      "step": 2772
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9840331202675437e-05,
+      "loss": 0.1195,
+      "step": 2773
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983973757431853e-05,
+      "loss": 0.0909,
+      "step": 2774
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9839142853404606e-05,
+      "loss": 0.124,
+      "step": 2775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9838547039999706e-05,
+      "loss": 0.0994,
+      "step": 2776
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983795013416998e-05,
+      "loss": 0.0969,
+      "step": 2777
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983735213598171e-05,
+      "loss": 0.0966,
+      "step": 2778
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9836753045501293e-05,
+      "loss": 0.1056,
+      "step": 2779
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9836152862795245e-05,
+      "loss": 0.1016,
+      "step": 2780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9835551587930217e-05,
+      "loss": 0.09,
+      "step": 2781
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9834949220972962e-05,
+      "loss": 0.1287,
+      "step": 2782
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9834345761990376e-05,
+      "loss": 0.106,
+      "step": 2783
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9833741211049448e-05,
+      "loss": 0.1161,
+      "step": 2784
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9833135568217315e-05,
+      "loss": 0.0924,
+      "step": 2785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9832528833561224e-05,
+      "loss": 0.0981,
+      "step": 2786
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983192100714854e-05,
+      "loss": 0.1072,
+      "step": 2787
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983131208904676e-05,
+      "loss": 0.1257,
+      "step": 2788
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.983070207932349e-05,
+      "loss": 0.0917,
+      "step": 2789
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9830090978046458e-05,
+      "loss": 0.0941,
+      "step": 2790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9829478785283527e-05,
+      "loss": 0.1135,
+      "step": 2791
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.982886550110267e-05,
+      "loss": 0.1183,
+      "step": 2792
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.982825112557198e-05,
+      "loss": 0.113,
+      "step": 2793
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9827635658759673e-05,
+      "loss": 0.1006,
+      "step": 2794
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.982701910073409e-05,
+      "loss": 0.0996,
+      "step": 2795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9826401451563693e-05,
+      "loss": 0.1222,
+      "step": 2796
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.982578271131706e-05,
+      "loss": 0.1057,
+      "step": 2797
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9825162880062893e-05,
+      "loss": 0.1116,
+      "step": 2798
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9824541957870016e-05,
+      "loss": 0.1,
+      "step": 2799
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.982391994480737e-05,
+      "loss": 0.1086,
+      "step": 2800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9823296840944027e-05,
+      "loss": 0.1066,
+      "step": 2801
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9822672646349167e-05,
+      "loss": 0.1094,
+      "step": 2802
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9822047361092096e-05,
+      "loss": 0.1254,
+      "step": 2803
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.982142098524225e-05,
+      "loss": 0.1187,
+      "step": 2804
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9820793518869177e-05,
+      "loss": 0.1024,
+      "step": 2805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9820164962042544e-05,
+      "loss": 0.103,
+      "step": 2806
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.981953531483215e-05,
+      "loss": 0.0988,
+      "step": 2807
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9818904577307897e-05,
+      "loss": 0.1122,
+      "step": 2808
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9818272749539827e-05,
+      "loss": 0.1031,
+      "step": 2809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981763983159809e-05,
+      "loss": 0.0892,
+      "step": 2810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981700582355297e-05,
+      "loss": 0.0875,
+      "step": 2811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981637072547486e-05,
+      "loss": 0.0935,
+      "step": 2812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9815734537434274e-05,
+      "loss": 0.0985,
+      "step": 2813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981509725950186e-05,
+      "loss": 0.1018,
+      "step": 2814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981445889174837e-05,
+      "loss": 0.0962,
+      "step": 2815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9813819434244687e-05,
+      "loss": 0.1069,
+      "step": 2816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981317888706182e-05,
+      "loss": 0.0944,
+      "step": 2817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9812537250270882e-05,
+      "loss": 0.0994,
+      "step": 2818
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9811894523943125e-05,
+      "loss": 0.1157,
+      "step": 2819
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.981125070814991e-05,
+      "loss": 0.1061,
+      "step": 2820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9810605802962728e-05,
+      "loss": 0.1018,
+      "step": 2821
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.980995980845318e-05,
+      "loss": 0.0919,
+      "step": 2822
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9809312724692997e-05,
+      "loss": 0.095,
+      "step": 2823
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.980866455175403e-05,
+      "loss": 0.1047,
+      "step": 2824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9808015289708248e-05,
+      "loss": 0.1066,
+      "step": 2825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9807364938627737e-05,
+      "loss": 0.1106,
+      "step": 2826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9806713498584718e-05,
+      "loss": 0.1075,
+      "step": 2827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9806060969651514e-05,
+      "loss": 0.1011,
+      "step": 2828
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9805407351900588e-05,
+      "loss": 0.1141,
+      "step": 2829
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9804752645404505e-05,
+      "loss": 0.1082,
+      "step": 2830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.980409685023597e-05,
+      "loss": 0.135,
+      "step": 2831
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9803439966467792e-05,
+      "loss": 0.1095,
+      "step": 2832
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.980278199417291e-05,
+      "loss": 0.1138,
+      "step": 2833
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9802122933424387e-05,
+      "loss": 0.1047,
+      "step": 2834
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9801462784295395e-05,
+      "loss": 0.1102,
+      "step": 2835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9800801546859238e-05,
+      "loss": 0.0891,
+      "step": 2836
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9800139221189332e-05,
+      "loss": 0.1132,
+      "step": 2837
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.979947580735923e-05,
+      "loss": 0.1071,
+      "step": 2838
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.979881130544258e-05,
+      "loss": 0.1074,
+      "step": 2839
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9798145715513168e-05,
+      "loss": 0.11,
+      "step": 2840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9797479037644906e-05,
+      "loss": 0.1251,
+      "step": 2841
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.979681127191181e-05,
+      "loss": 0.0996,
+      "step": 2842
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9796142418388035e-05,
+      "loss": 0.1323,
+      "step": 2843
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9795472477147836e-05,
+      "loss": 0.1082,
+      "step": 2844
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.979480144826561e-05,
+      "loss": 0.0881,
+      "step": 2845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.979412933181586e-05,
+      "loss": 0.0917,
+      "step": 2846
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.979345612787321e-05,
+      "loss": 0.1091,
+      "step": 2847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9792781836512418e-05,
+      "loss": 0.0963,
+      "step": 2848
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9792106457808348e-05,
+      "loss": 0.1152,
+      "step": 2849
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9791429991835995e-05,
+      "loss": 0.0977,
+      "step": 2850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.979075243867047e-05,
+      "loss": 0.1145,
+      "step": 2851
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9790073798387003e-05,
+      "loss": 0.1158,
+      "step": 2852
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9789394071060946e-05,
+      "loss": 0.1115,
+      "step": 2853
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9788713256767777e-05,
+      "loss": 0.1019,
+      "step": 2854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9788031355583085e-05,
+      "loss": 0.1,
+      "step": 2855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9787348367582586e-05,
+      "loss": 0.1215,
+      "step": 2856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9786664292842122e-05,
+      "loss": 0.1091,
+      "step": 2857
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9785979131437646e-05,
+      "loss": 0.0897,
+      "step": 2858
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.978529288344523e-05,
+      "loss": 0.0895,
+      "step": 2859
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9784605548941074e-05,
+      "loss": 0.1117,
+      "step": 2860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9783917128001503e-05,
+      "loss": 0.1039,
+      "step": 2861
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9783227620702946e-05,
+      "loss": 0.1102,
+      "step": 2862
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.978253702712197e-05,
+      "loss": 0.0923,
+      "step": 2863
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9781845347335253e-05,
+      "loss": 0.1079,
+      "step": 2864
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9781152581419595e-05,
+      "loss": 0.1023,
+      "step": 2865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9780458729451916e-05,
+      "loss": 0.0986,
+      "step": 2866
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9779763791509262e-05,
+      "loss": 0.1156,
+      "step": 2867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9779067767668794e-05,
+      "loss": 0.0986,
+      "step": 2868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9778370658007792e-05,
+      "loss": 0.0841,
+      "step": 2869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.977767246260366e-05,
+      "loss": 0.1301,
+      "step": 2870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9776973181533926e-05,
+      "loss": 0.088,
+      "step": 2871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9776272814876235e-05,
+      "loss": 0.1119,
+      "step": 2872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9775571362708353e-05,
+      "loss": 0.1145,
+      "step": 2873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.977486882510816e-05,
+      "loss": 0.1063,
+      "step": 2874
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9774165202153665e-05,
+      "loss": 0.1316,
+      "step": 2875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9773460493922994e-05,
+      "loss": 0.0829,
+      "step": 2876
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.97727547004944e-05,
+      "loss": 0.1311,
+      "step": 2877
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9772047821946242e-05,
+      "loss": 0.1049,
+      "step": 2878
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.977133985835702e-05,
+      "loss": 0.1108,
+      "step": 2879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.977063080980533e-05,
+      "loss": 0.1146,
+      "step": 2880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.976992067636991e-05,
+      "loss": 0.1155,
+      "step": 2881
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.976920945812961e-05,
+      "loss": 0.1018,
+      "step": 2882
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9768497155163392e-05,
+      "loss": 0.091,
+      "step": 2883
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9767783767550358e-05,
+      "loss": 0.0985,
+      "step": 2884
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9767069295369707e-05,
+      "loss": 0.1169,
+      "step": 2885
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.976635373870078e-05,
+      "loss": 0.0955,
+      "step": 2886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9765637097623028e-05,
+      "loss": 0.0989,
+      "step": 2887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.976491937221602e-05,
+      "loss": 0.1027,
+      "step": 2888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.976420056255945e-05,
+      "loss": 0.0919,
+      "step": 2889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9763480668733132e-05,
+      "loss": 0.1118,
+      "step": 2890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9762759690816996e-05,
+      "loss": 0.1052,
+      "step": 2891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9762037628891097e-05,
+      "loss": 0.0883,
+      "step": 2892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9761314483035617e-05,
+      "loss": 0.1165,
+      "step": 2893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.976059025333084e-05,
+      "loss": 0.0939,
+      "step": 2894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9759864939857186e-05,
+      "loss": 0.1124,
+      "step": 2895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.975913854269519e-05,
+      "loss": 0.0929,
+      "step": 2896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.975841106192551e-05,
+      "loss": 0.0828,
+      "step": 2897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9757682497628915e-05,
+      "loss": 0.1102,
+      "step": 2898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.975695284988631e-05,
+      "loss": 0.0911,
+      "step": 2899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9756222118778704e-05,
+      "loss": 0.1008,
+      "step": 2900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9755490304387236e-05,
+      "loss": 0.1005,
+      "step": 2901
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9754757406793172e-05,
+      "loss": 0.1035,
+      "step": 2902
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.975402342607787e-05,
+      "loss": 0.0994,
+      "step": 2903
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.975328836232285e-05,
+      "loss": 0.1018,
+      "step": 2904
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9752552215609713e-05,
+      "loss": 0.0918,
+      "step": 2905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9751814986020203e-05,
+      "loss": 0.1051,
+      "step": 2906
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.975107667363618e-05,
+      "loss": 0.1023,
+      "step": 2907
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9750337278539623e-05,
+      "loss": 0.1086,
+      "step": 2908
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974959680081263e-05,
+      "loss": 0.1272,
+      "step": 2909
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9748855240537418e-05,
+      "loss": 0.1077,
+      "step": 2910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974811259779633e-05,
+      "loss": 0.1166,
+      "step": 2911
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974736887267182e-05,
+      "loss": 0.1129,
+      "step": 2912
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974662406524647e-05,
+      "loss": 0.1046,
+      "step": 2913
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9745878175602984e-05,
+      "loss": 0.1045,
+      "step": 2914
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9745131203824177e-05,
+      "loss": 0.0974,
+      "step": 2915
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974438314999299e-05,
+      "loss": 0.0792,
+      "step": 2916
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9743634014192486e-05,
+      "loss": 0.1133,
+      "step": 2917
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9742883796505843e-05,
+      "loss": 0.118,
+      "step": 2918
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974213249701636e-05,
+      "loss": 0.1077,
+      "step": 2919
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974138011580746e-05,
+      "loss": 0.1042,
+      "step": 2920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.974062665296269e-05,
+      "loss": 0.1119,
+      "step": 2921
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9739872108565697e-05,
+      "loss": 0.1063,
+      "step": 2922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.973911648270027e-05,
+      "loss": 0.1039,
+      "step": 2923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9738359775450313e-05,
+      "loss": 0.087,
+      "step": 2924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.973760198689984e-05,
+      "loss": 0.0945,
+      "step": 2925
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9736843117132996e-05,
+      "loss": 0.1058,
+      "step": 2926
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9736083166234047e-05,
+      "loss": 0.1,
+      "step": 2927
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9735322134287364e-05,
+      "loss": 0.0827,
+      "step": 2928
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9734560021377454e-05,
+      "loss": 0.1185,
+      "step": 2929
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.973379682758894e-05,
+      "loss": 0.1147,
+      "step": 2930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.973303255300656e-05,
+      "loss": 0.1235,
+      "step": 2931
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9732267197715176e-05,
+      "loss": 0.0916,
+      "step": 2932
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.973150076179977e-05,
+      "loss": 0.0813,
+      "step": 2933
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9730733245345445e-05,
+      "loss": 0.0997,
+      "step": 2934
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.972996464843742e-05,
+      "loss": 0.1167,
+      "step": 2935
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.972919497116104e-05,
+      "loss": 0.097,
+      "step": 2936
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9728424213601758e-05,
+      "loss": 0.1201,
+      "step": 2937
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9727652375845164e-05,
+      "loss": 0.1279,
+      "step": 2938
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9726879457976954e-05,
+      "loss": 0.0988,
+      "step": 2939
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.972610546008295e-05,
+      "loss": 0.0843,
+      "step": 2940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9725330382249094e-05,
+      "loss": 0.0895,
+      "step": 2941
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.972455422456145e-05,
+      "loss": 0.1074,
+      "step": 2942
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9723776987106193e-05,
+      "loss": 0.1156,
+      "step": 2943
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9722998669969626e-05,
+      "loss": 0.09,
+      "step": 2944
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9722219273238166e-05,
+      "loss": 0.0969,
+      "step": 2945
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.972143879699836e-05,
+      "loss": 0.1088,
+      "step": 2946
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9720657241336866e-05,
+      "loss": 0.0916,
+      "step": 2947
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.971987460634046e-05,
+      "loss": 0.0754,
+      "step": 2948
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.971909089209605e-05,
+      "loss": 0.0922,
+      "step": 2949
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.971830609869065e-05,
+      "loss": 0.1033,
+      "step": 2950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.97175202262114e-05,
+      "loss": 0.1091,
+      "step": 2951
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9716733274745566e-05,
+      "loss": 0.1013,
+      "step": 2952
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9715945244380513e-05,
+      "loss": 0.1152,
+      "step": 2953
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.971515613520376e-05,
+      "loss": 0.1113,
+      "step": 2954
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9714365947302905e-05,
+      "loss": 0.0938,
+      "step": 2955
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.97135746807657e-05,
+      "loss": 0.116,
+      "step": 2956
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.971278233568e-05,
+      "loss": 0.1127,
+      "step": 2957
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9711988912133783e-05,
+      "loss": 0.093,
+      "step": 2958
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9711194410215148e-05,
+      "loss": 0.1189,
+      "step": 2959
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9710398830012313e-05,
+      "loss": 0.0999,
+      "step": 2960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.970960217161361e-05,
+      "loss": 0.0913,
+      "step": 2961
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9708804435107504e-05,
+      "loss": 0.1118,
+      "step": 2962
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9708005620582564e-05,
+      "loss": 0.1022,
+      "step": 2963
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9707205728127496e-05,
+      "loss": 0.0953,
+      "step": 2964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9706404757831104e-05,
+      "loss": 0.103,
+      "step": 2965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9705602709782336e-05,
+      "loss": 0.1124,
+      "step": 2966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.970479958407024e-05,
+      "loss": 0.0981,
+      "step": 2967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9703995380783993e-05,
+      "loss": 0.1055,
+      "step": 2968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.970319010001289e-05,
+      "loss": 0.0999,
+      "step": 2969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9702383741846346e-05,
+      "loss": 0.1196,
+      "step": 2970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9701576306373896e-05,
+      "loss": 0.1042,
+      "step": 2971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9700767793685195e-05,
+      "loss": 0.0889,
+      "step": 2972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.969995820387001e-05,
+      "loss": 0.0886,
+      "step": 2973
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.969914753701824e-05,
+      "loss": 0.1135,
+      "step": 2974
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.969833579321989e-05,
+      "loss": 0.1022,
+      "step": 2975
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9697522972565103e-05,
+      "loss": 0.1075,
+      "step": 2976
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9696709075144123e-05,
+      "loss": 0.0944,
+      "step": 2977
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9695894101047327e-05,
+      "loss": 0.0952,
+      "step": 2978
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.96950780503652e-05,
+      "loss": 0.1024,
+      "step": 2979
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9694260923188354e-05,
+      "loss": 0.1038,
+      "step": 2980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.969344271960752e-05,
+      "loss": 0.1128,
+      "step": 2981
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9692623439713547e-05,
+      "loss": 0.0992,
+      "step": 2982
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9691803083597403e-05,
+      "loss": 0.093,
+      "step": 2983
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.969098165135018e-05,
+      "loss": 0.0892,
+      "step": 2984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.969015914306308e-05,
+      "loss": 0.0886,
+      "step": 2985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9689335558827433e-05,
+      "loss": 0.0944,
+      "step": 2986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9688510898734687e-05,
+      "loss": 0.1162,
+      "step": 2987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9687685162876406e-05,
+      "loss": 0.1035,
+      "step": 2988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9686858351344284e-05,
+      "loss": 0.098,
+      "step": 2989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.968603046423011e-05,
+      "loss": 0.0985,
+      "step": 2990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9685201501625822e-05,
+      "loss": 0.1174,
+      "step": 2991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.968437146362346e-05,
+      "loss": 0.1113,
+      "step": 2992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.968354035031519e-05,
+      "loss": 0.0941,
+      "step": 2993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9682708161793287e-05,
+      "loss": 0.0981,
+      "step": 2994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.968187489815016e-05,
+      "loss": 0.0776,
+      "step": 2995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.968104055947833e-05,
+      "loss": 0.1151,
+      "step": 2996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.968020514587044e-05,
+      "loss": 0.0914,
+      "step": 2997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.967936865741924e-05,
+      "loss": 0.1046,
+      "step": 2998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9678531094217622e-05,
+      "loss": 0.1122,
+      "step": 2999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.967769245635858e-05,
+      "loss": 0.1044,
+      "step": 3000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.967685274393523e-05,
+      "loss": 0.1091,
+      "step": 3001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9676011957040812e-05,
+      "loss": 0.1152,
+      "step": 3002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9675170095768685e-05,
+      "loss": 0.099,
+      "step": 3003
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.967432716021232e-05,
+      "loss": 0.1086,
+      "step": 3004
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9673483150465314e-05,
+      "loss": 0.1083,
+      "step": 3005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.967263806662139e-05,
+      "loss": 0.1045,
+      "step": 3006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.967179190877437e-05,
+      "loss": 0.0884,
+      "step": 3007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9670944677018214e-05,
+      "loss": 0.101,
+      "step": 3008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9670096371446992e-05,
+      "loss": 0.1292,
+      "step": 3009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.96692469921549e-05,
+      "loss": 0.1112,
+      "step": 3010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.966839653923624e-05,
+      "loss": 0.1051,
+      "step": 3011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9667545012785448e-05,
+      "loss": 0.1173,
+      "step": 3012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.966669241289708e-05,
+      "loss": 0.0896,
+      "step": 3013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9665838739665793e-05,
+      "loss": 0.093,
+      "step": 3014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.966498399318638e-05,
+      "loss": 0.1145,
+      "step": 3015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9664128173553748e-05,
+      "loss": 0.1106,
+      "step": 3016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9663271280862924e-05,
+      "loss": 0.1033,
+      "step": 3017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.966241331520905e-05,
+      "loss": 0.1139,
+      "step": 3018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9661554276687394e-05,
+      "loss": 0.1068,
+      "step": 3019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9660694165393334e-05,
+      "loss": 0.1097,
+      "step": 3020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9659832981422383e-05,
+      "loss": 0.0929,
+      "step": 3021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9658970724870153e-05,
+      "loss": 0.0876,
+      "step": 3022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9658107395832387e-05,
+      "loss": 0.1155,
+      "step": 3023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9657242994404947e-05,
+      "loss": 0.0955,
+      "step": 3024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9656377520683808e-05,
+      "loss": 0.1082,
+      "step": 3025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9655510974765074e-05,
+      "loss": 0.0918,
+      "step": 3026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.965464335674496e-05,
+      "loss": 0.1075,
+      "step": 3027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9653774666719796e-05,
+      "loss": 0.1093,
+      "step": 3028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9652904904786046e-05,
+      "loss": 0.1118,
+      "step": 3029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9652034071040278e-05,
+      "loss": 0.1158,
+      "step": 3030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9651162165579188e-05,
+      "loss": 0.1074,
+      "step": 3031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9650289188499587e-05,
+      "loss": 0.1094,
+      "step": 3032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9649415139898407e-05,
+      "loss": 0.0999,
+      "step": 3033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.96485400198727e-05,
+      "loss": 0.0956,
+      "step": 3034
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9647663828519633e-05,
+      "loss": 0.1106,
+      "step": 3035
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9646786565936497e-05,
+      "loss": 0.1019,
+      "step": 3036
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9645908232220692e-05,
+      "loss": 0.1056,
+      "step": 3037
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9645028827469748e-05,
+      "loss": 0.1106,
+      "step": 3038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.964414835178131e-05,
+      "loss": 0.0929,
+      "step": 3039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.964326680525314e-05,
+      "loss": 0.0954,
+      "step": 3040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9642384187983126e-05,
+      "loss": 0.1145,
+      "step": 3041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9641500500069267e-05,
+      "loss": 0.108,
+      "step": 3042
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.964061574160968e-05,
+      "loss": 0.1167,
+      "step": 3043
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.963972991270261e-05,
+      "loss": 0.0964,
+      "step": 3044
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9638843013446408e-05,
+      "loss": 0.098,
+      "step": 3045
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9637955043939554e-05,
+      "loss": 0.0959,
+      "step": 3046
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9637066004280646e-05,
+      "loss": 0.0917,
+      "step": 3047
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9636175894568397e-05,
+      "loss": 0.085,
+      "step": 3048
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9635284714901646e-05,
+      "loss": 0.0894,
+      "step": 3049
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9634392465379337e-05,
+      "loss": 0.1005,
+      "step": 3050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.963349914610054e-05,
+      "loss": 0.0966,
+      "step": 3051
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9632604757164456e-05,
+      "loss": 0.1024,
+      "step": 3052
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9631709298670382e-05,
+      "loss": 0.1011,
+      "step": 3053
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9630812770717753e-05,
+      "loss": 0.0844,
+      "step": 3054
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.962991517340611e-05,
+      "loss": 0.1064,
+      "step": 3055
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9629016506835122e-05,
+      "loss": 0.0906,
+      "step": 3056
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.962811677110457e-05,
+      "loss": 0.1078,
+      "step": 3057
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.962721596631436e-05,
+      "loss": 0.0941,
+      "step": 3058
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9626314092564506e-05,
+      "loss": 0.1063,
+      "step": 3059
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9625411149955156e-05,
+      "loss": 0.1082,
+      "step": 3060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.962450713858656e-05,
+      "loss": 0.0929,
+      "step": 3061
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9623602058559103e-05,
+      "loss": 0.1089,
+      "step": 3062
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9622695909973276e-05,
+      "loss": 0.1068,
+      "step": 3063
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9621788692929695e-05,
+      "loss": 0.1071,
+      "step": 3064
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9620880407529092e-05,
+      "loss": 0.0997,
+      "step": 3065
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9619971053872318e-05,
+      "loss": 0.082,
+      "step": 3066
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9619060632060343e-05,
+      "loss": 0.1151,
+      "step": 3067
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9618149142194262e-05,
+      "loss": 0.092,
+      "step": 3068
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9617236584375275e-05,
+      "loss": 0.0945,
+      "step": 3069
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9616322958704708e-05,
+      "loss": 0.1063,
+      "step": 3070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.961540826528401e-05,
+      "loss": 0.0897,
+      "step": 3071
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9614492504214744e-05,
+      "loss": 0.0967,
+      "step": 3072
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9613575675598588e-05,
+      "loss": 0.1005,
+      "step": 3073
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.961265777953735e-05,
+      "loss": 0.1003,
+      "step": 3074
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9611738816132936e-05,
+      "loss": 0.1115,
+      "step": 3075
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9610818785487392e-05,
+      "loss": 0.0765,
+      "step": 3076
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9609897687702874e-05,
+      "loss": 0.1202,
+      "step": 3077
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.960897552288165e-05,
+      "loss": 0.0989,
+      "step": 3078
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9608052291126123e-05,
+      "loss": 0.1129,
+      "step": 3079
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9607127992538796e-05,
+      "loss": 0.0991,
+      "step": 3080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9606202627222298e-05,
+      "loss": 0.1052,
+      "step": 3081
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9605276195279385e-05,
+      "loss": 0.0728,
+      "step": 3082
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9604348696812917e-05,
+      "loss": 0.0939,
+      "step": 3083
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.960342013192588e-05,
+      "loss": 0.0925,
+      "step": 3084
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9602490500721375e-05,
+      "loss": 0.1001,
+      "step": 3085
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.960155980330263e-05,
+      "loss": 0.0991,
+      "step": 3086
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.960062803977298e-05,
+      "loss": 0.1176,
+      "step": 3087
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9599695210235886e-05,
+      "loss": 0.1049,
+      "step": 3088
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.959876131479493e-05,
+      "loss": 0.1094,
+      "step": 3089
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9597826353553794e-05,
+      "loss": 0.0959,
+      "step": 3090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.95968903266163e-05,
+      "loss": 0.0774,
+      "step": 3091
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.959595323408638e-05,
+      "loss": 0.0891,
+      "step": 3092
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.959501507606808e-05,
+      "loss": 0.1096,
+      "step": 3093
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.959407585266558e-05,
+      "loss": 0.1064,
+      "step": 3094
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9593135563983152e-05,
+      "loss": 0.1204,
+      "step": 3095
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.959219421012521e-05,
+      "loss": 0.1049,
+      "step": 3096
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9591251791196274e-05,
+      "loss": 0.1184,
+      "step": 3097
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9590308307300988e-05,
+      "loss": 0.0992,
+      "step": 3098
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9589363758544108e-05,
+      "loss": 0.0853,
+      "step": 3099
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.958841814503052e-05,
+      "loss": 0.1141,
+      "step": 3100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.958747146686521e-05,
+      "loss": 0.0856,
+      "step": 3101
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.95865237241533e-05,
+      "loss": 0.1072,
+      "step": 3102
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.958557491700002e-05,
+      "loss": 0.1129,
+      "step": 3103
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9584625045510725e-05,
+      "loss": 0.0891,
+      "step": 3104
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9583674109790878e-05,
+      "loss": 0.1,
+      "step": 3105
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9582722109946067e-05,
+      "loss": 0.0911,
+      "step": 3106
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9581769046082002e-05,
+      "loss": 0.1018,
+      "step": 3107
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9580814918304504e-05,
+      "loss": 0.0995,
+      "step": 3108
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9579859726719513e-05,
+      "loss": 0.1067,
+      "step": 3109
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.957890347143309e-05,
+      "loss": 0.1176,
+      "step": 3110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9577946152551417e-05,
+      "loss": 0.1063,
+      "step": 3111
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9576987770180788e-05,
+      "loss": 0.0861,
+      "step": 3112
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9576028324427612e-05,
+      "loss": 0.1023,
+      "step": 3113
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9575067815398423e-05,
+      "loss": 0.1011,
+      "step": 3114
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.957410624319988e-05,
+      "loss": 0.0912,
+      "step": 3115
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.957314360793874e-05,
+      "loss": 0.0786,
+      "step": 3116
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9572179909721894e-05,
+      "loss": 0.1069,
+      "step": 3117
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.957121514865635e-05,
+      "loss": 0.0946,
+      "step": 3118
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9570249324849223e-05,
+      "loss": 0.09,
+      "step": 3119
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9569282438407763e-05,
+      "loss": 0.0887,
+      "step": 3120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.956831448943932e-05,
+      "loss": 0.0763,
+      "step": 3121
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.956734547805137e-05,
+      "loss": 0.0704,
+      "step": 3122
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.956637540435151e-05,
+      "loss": 0.1136,
+      "step": 3123
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.956540426844746e-05,
+      "loss": 0.1029,
+      "step": 3124
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9564432070447035e-05,
+      "loss": 0.0918,
+      "step": 3125
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9563458810458195e-05,
+      "loss": 0.1066,
+      "step": 3126
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9562484488589005e-05,
+      "loss": 0.0895,
+      "step": 3127
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9561509104947643e-05,
+      "loss": 0.0869,
+      "step": 3128
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9560532659642413e-05,
+      "loss": 0.0867,
+      "step": 3129
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.955955515278174e-05,
+      "loss": 0.1099,
+      "step": 3130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9558576584474154e-05,
+      "loss": 0.0997,
+      "step": 3131
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9557596954828315e-05,
+      "loss": 0.1245,
+      "step": 3132
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9556616263953e-05,
+      "loss": 0.0963,
+      "step": 3133
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.955563451195709e-05,
+      "loss": 0.0936,
+      "step": 3134
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9554651698949603e-05,
+      "loss": 0.1086,
+      "step": 3135
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.955366782503966e-05,
+      "loss": 0.1104,
+      "step": 3136
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9552682890336508e-05,
+      "loss": 0.0832,
+      "step": 3137
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9551696894949513e-05,
+      "loss": 0.0919,
+      "step": 3138
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.955070983898815e-05,
+      "loss": 0.1135,
+      "step": 3139
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.954972172256202e-05,
+      "loss": 0.1235,
+      "step": 3140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9548732545780833e-05,
+      "loss": 0.1088,
+      "step": 3141
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.954774230875443e-05,
+      "loss": 0.1132,
+      "step": 3142
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.954675101159276e-05,
+      "loss": 0.0804,
+      "step": 3143
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9545758654405888e-05,
+      "loss": 0.1028,
+      "step": 3144
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9544765237304006e-05,
+      "loss": 0.1119,
+      "step": 3145
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9543770760397413e-05,
+      "loss": 0.1105,
+      "step": 3146
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9542775223796534e-05,
+      "loss": 0.098,
+      "step": 3147
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9541778627611908e-05,
+      "loss": 0.0885,
+      "step": 3148
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9540780971954193e-05,
+      "loss": 0.1128,
+      "step": 3149
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9539782256934166e-05,
+      "loss": 0.0892,
+      "step": 3150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.953878248266271e-05,
+      "loss": 0.0951,
+      "step": 3151
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9537781649250848e-05,
+      "loss": 0.1436,
+      "step": 3152
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.95367797568097e-05,
+      "loss": 0.0938,
+      "step": 3153
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9535776805450512e-05,
+      "loss": 0.1208,
+      "step": 3154
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.953477279528465e-05,
+      "loss": 0.0972,
+      "step": 3155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9533767726423586e-05,
+      "loss": 0.0996,
+      "step": 3156
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9532761598978932e-05,
+      "loss": 0.1248,
+      "step": 3157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9531754413062392e-05,
+      "loss": 0.1118,
+      "step": 3158
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.95307461687858e-05,
+      "loss": 0.1066,
+      "step": 3159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9529736866261112e-05,
+      "loss": 0.1105,
+      "step": 3160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9528726505600396e-05,
+      "loss": 0.0891,
+      "step": 3161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.952771508691583e-05,
+      "loss": 0.0994,
+      "step": 3162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9526702610319727e-05,
+      "loss": 0.1083,
+      "step": 3163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9525689075924498e-05,
+      "loss": 0.0853,
+      "step": 3164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9524674483842687e-05,
+      "loss": 0.12,
+      "step": 3165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.952365883418695e-05,
+      "loss": 0.1021,
+      "step": 3166
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9522642127070057e-05,
+      "loss": 0.0861,
+      "step": 3167
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9521624362604896e-05,
+      "loss": 0.1019,
+      "step": 3168
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.952060554090448e-05,
+      "loss": 0.1095,
+      "step": 3169
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9519585662081932e-05,
+      "loss": 0.1091,
+      "step": 3170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9518564726250496e-05,
+      "loss": 0.0854,
+      "step": 3171
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9517542733523528e-05,
+      "loss": 0.1,
+      "step": 3172
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9516519684014505e-05,
+      "loss": 0.0895,
+      "step": 3173
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9515495577837026e-05,
+      "loss": 0.0956,
+      "step": 3174
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9514470415104802e-05,
+      "loss": 0.1067,
+      "step": 3175
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.951344419593166e-05,
+      "loss": 0.075,
+      "step": 3176
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9512416920431544e-05,
+      "loss": 0.1094,
+      "step": 3177
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9511388588718522e-05,
+      "loss": 0.1101,
+      "step": 3178
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9510359200906776e-05,
+      "loss": 0.0929,
+      "step": 3179
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9509328757110598e-05,
+      "loss": 0.0898,
+      "step": 3180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9508297257444408e-05,
+      "loss": 0.1005,
+      "step": 3181
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.950726470202274e-05,
+      "loss": 0.098,
+      "step": 3182
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9506231090960244e-05,
+      "loss": 0.082,
+      "step": 3183
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9505196424371685e-05,
+      "loss": 0.1218,
+      "step": 3184
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9504160702371947e-05,
+      "loss": 0.1119,
+      "step": 3185
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9503123925076028e-05,
+      "loss": 0.1008,
+      "step": 3186
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9502086092599054e-05,
+      "loss": 0.0951,
+      "step": 3187
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9501047205056262e-05,
+      "loss": 0.1086,
+      "step": 3188
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9500007262562994e-05,
+      "loss": 0.1017,
+      "step": 3189
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9498966265234735e-05,
+      "loss": 0.116,
+      "step": 3190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9497924213187057e-05,
+      "loss": 0.1111,
+      "step": 3191
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9496881106535675e-05,
+      "loss": 0.0836,
+      "step": 3192
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9495836945396413e-05,
+      "loss": 0.1024,
+      "step": 3193
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9494791729885198e-05,
+      "loss": 0.077,
+      "step": 3194
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.949374546011809e-05,
+      "loss": 0.1179,
+      "step": 3195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.949269813621127e-05,
+      "loss": 0.1083,
+      "step": 3196
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9491649758281017e-05,
+      "loss": 0.1061,
+      "step": 3197
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9490600326443743e-05,
+      "loss": 0.0948,
+      "step": 3198
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9489549840815974e-05,
+      "loss": 0.115,
+      "step": 3199
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9488498301514343e-05,
+      "loss": 0.11,
+      "step": 3200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9487445708655616e-05,
+      "loss": 0.0996,
+      "step": 3201
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9486392062356663e-05,
+      "loss": 0.094,
+      "step": 3202
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.948533736273448e-05,
+      "loss": 0.0905,
+      "step": 3203
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9484281609906172e-05,
+      "loss": 0.1002,
+      "step": 3204
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9483224803988965e-05,
+      "loss": 0.0867,
+      "step": 3205
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9482166945100204e-05,
+      "loss": 0.0846,
+      "step": 3206
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9481108033357344e-05,
+      "loss": 0.126,
+      "step": 3207
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.948004806887797e-05,
+      "loss": 0.1168,
+      "step": 3208
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9478987051779767e-05,
+      "loss": 0.0999,
+      "step": 3209
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9477924982180548e-05,
+      "loss": 0.0961,
+      "step": 3210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9476861860198238e-05,
+      "loss": 0.0958,
+      "step": 3211
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9475797685950885e-05,
+      "loss": 0.1077,
+      "step": 3212
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9474732459556647e-05,
+      "loss": 0.1226,
+      "step": 3213
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9473666181133805e-05,
+      "loss": 0.0981,
+      "step": 3214
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.947259885080075e-05,
+      "loss": 0.1051,
+      "step": 3215
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9471530468675995e-05,
+      "loss": 0.1005,
+      "step": 3216
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9470461034878167e-05,
+      "loss": 0.0876,
+      "step": 3217
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9469390549526007e-05,
+      "loss": 0.1036,
+      "step": 3218
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9468319012738383e-05,
+      "loss": 0.1067,
+      "step": 3219
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.946724642463427e-05,
+      "loss": 0.1085,
+      "step": 3220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9466172785332767e-05,
+      "loss": 0.0855,
+      "step": 3221
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.946509809495308e-05,
+      "loss": 0.0844,
+      "step": 3222
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.946402235361454e-05,
+      "loss": 0.1177,
+      "step": 3223
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.946294556143659e-05,
+      "loss": 0.0938,
+      "step": 3224
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.94618677185388e-05,
+      "loss": 0.1156,
+      "step": 3225
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9460788825040837e-05,
+      "loss": 0.0874,
+      "step": 3226
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9459708881062504e-05,
+      "loss": 0.0991,
+      "step": 3227
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9458627886723706e-05,
+      "loss": 0.0828,
+      "step": 3228
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.945754584214448e-05,
+      "loss": 0.0974,
+      "step": 3229
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9456462747444965e-05,
+      "loss": 0.0903,
+      "step": 3230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9455378602745426e-05,
+      "loss": 0.1082,
+      "step": 3231
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.945429340816624e-05,
+      "loss": 0.0962,
+      "step": 3232
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.94532071638279e-05,
+      "loss": 0.108,
+      "step": 3233
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9452119869851014e-05,
+      "loss": 0.1049,
+      "step": 3234
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9451031526356318e-05,
+      "loss": 0.1138,
+      "step": 3235
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9449942133464654e-05,
+      "loss": 0.0837,
+      "step": 3236
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944885169129698e-05,
+      "loss": 0.1016,
+      "step": 3237
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9447760199974376e-05,
+      "loss": 0.1058,
+      "step": 3238
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944666765961804e-05,
+      "loss": 0.0936,
+      "step": 3239
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9445574070349272e-05,
+      "loss": 0.0897,
+      "step": 3240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9444479432289505e-05,
+      "loss": 0.1184,
+      "step": 3241
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.944338374556028e-05,
+      "loss": 0.1127,
+      "step": 3242
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9442287010283264e-05,
+      "loss": 0.0922,
+      "step": 3243
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9441189226580225e-05,
+      "loss": 0.0861,
+      "step": 3244
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9440090394573056e-05,
+      "loss": 0.1027,
+      "step": 3245
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.943899051438377e-05,
+      "loss": 0.11,
+      "step": 3246
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9437889586134493e-05,
+      "loss": 0.1039,
+      "step": 3247
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9436787609947464e-05,
+      "loss": 0.111,
+      "step": 3248
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9435684585945037e-05,
+      "loss": 0.0853,
+      "step": 3249
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9434580514249698e-05,
+      "loss": 0.098,
+      "step": 3250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9433475394984028e-05,
+      "loss": 0.11,
+      "step": 3251
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9432369228270733e-05,
+      "loss": 0.1097,
+      "step": 3252
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9431262014232645e-05,
+      "loss": 0.0944,
+      "step": 3253
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.94301537529927e-05,
+      "loss": 0.0994,
+      "step": 3254
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.942904444467395e-05,
+      "loss": 0.1284,
+      "step": 3255
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.942793408939957e-05,
+      "loss": 0.0918,
+      "step": 3256
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9426822687292852e-05,
+      "loss": 0.082,
+      "step": 3257
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9425710238477197e-05,
+      "loss": 0.1049,
+      "step": 3258
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9424596743076125e-05,
+      "loss": 0.1044,
+      "step": 3259
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9423482201213275e-05,
+      "loss": 0.0991,
+      "step": 3260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9422366613012404e-05,
+      "loss": 0.1191,
+      "step": 3261
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9421249978597375e-05,
+      "loss": 0.0843,
+      "step": 3262
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9420132298092173e-05,
+      "loss": 0.0936,
+      "step": 3263
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9419013571620907e-05,
+      "loss": 0.1063,
+      "step": 3264
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.941789379930779e-05,
+      "loss": 0.1041,
+      "step": 3265
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9416772981277156e-05,
+      "loss": 0.098,
+      "step": 3266
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.941565111765346e-05,
+      "loss": 0.0956,
+      "step": 3267
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9414528208561262e-05,
+      "loss": 0.1021,
+      "step": 3268
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9413404254125246e-05,
+      "loss": 0.1097,
+      "step": 3269
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9412279254470215e-05,
+      "loss": 0.0955,
+      "step": 3270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9411153209721078e-05,
+      "loss": 0.0939,
+      "step": 3271
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.941002612000287e-05,
+      "loss": 0.0896,
+      "step": 3272
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9408897985440735e-05,
+      "loss": 0.105,
+      "step": 3273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9407768806159935e-05,
+      "loss": 0.0964,
+      "step": 3274
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.940663858228585e-05,
+      "loss": 0.0997,
+      "step": 3275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9405507313943975e-05,
+      "loss": 0.0917,
+      "step": 3276
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9404375001259916e-05,
+      "loss": 0.1183,
+      "step": 3277
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.940324164435941e-05,
+      "loss": 0.087,
+      "step": 3278
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9402107243368288e-05,
+      "loss": 0.093,
+      "step": 3279
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9400971798412514e-05,
+      "loss": 0.0975,
+      "step": 3280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9399835309618165e-05,
+      "loss": 0.1075,
+      "step": 3281
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9398697777111427e-05,
+      "loss": 0.0814,
+      "step": 3282
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9397559201018604e-05,
+      "loss": 0.0928,
+      "step": 3283
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.939641958146612e-05,
+      "loss": 0.0922,
+      "step": 3284
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.939527891858052e-05,
+      "loss": 0.108,
+      "step": 3285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.939413721248845e-05,
+      "loss": 0.1016,
+      "step": 3286
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9392994463316677e-05,
+      "loss": 0.09,
+      "step": 3287
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9391850671192092e-05,
+      "loss": 0.0896,
+      "step": 3288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9390705836241698e-05,
+      "loss": 0.1008,
+      "step": 3289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9389559958592607e-05,
+      "loss": 0.0936,
+      "step": 3290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.938841303837205e-05,
+      "loss": 0.1143,
+      "step": 3291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.938726507570738e-05,
+      "loss": 0.0859,
+      "step": 3292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9386116070726063e-05,
+      "loss": 0.0978,
+      "step": 3293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9384966023555673e-05,
+      "loss": 0.1172,
+      "step": 3294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.938381493432391e-05,
+      "loss": 0.1161,
+      "step": 3295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9382662803158585e-05,
+      "loss": 0.0996,
+      "step": 3296
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9381509630187626e-05,
+      "loss": 0.0917,
+      "step": 3297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.938035541553907e-05,
+      "loss": 0.0946,
+      "step": 3298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.937920015934108e-05,
+      "loss": 0.0955,
+      "step": 3299
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.937804386172193e-05,
+      "loss": 0.1112,
+      "step": 3300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.937688652281001e-05,
+      "loss": 0.1005,
+      "step": 3301
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9375728142733825e-05,
+      "loss": 0.0757,
+      "step": 3302
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9374568721621996e-05,
+      "loss": 0.095,
+      "step": 3303
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9373408259603254e-05,
+      "loss": 0.0975,
+      "step": 3304
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9372246756806462e-05,
+      "loss": 0.1023,
+      "step": 3305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.937108421336058e-05,
+      "loss": 0.1034,
+      "step": 3306
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9369920629394693e-05,
+      "loss": 0.1083,
+      "step": 3307
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9368756005038e-05,
+      "loss": 0.1099,
+      "step": 3308
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9367590340419814e-05,
+      "loss": 0.0746,
+      "step": 3309
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9366423635669568e-05,
+      "loss": 0.1061,
+      "step": 3310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9365255890916802e-05,
+      "loss": 0.1041,
+      "step": 3311
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.936408710629118e-05,
+      "loss": 0.0886,
+      "step": 3312
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.936291728192248e-05,
+      "loss": 0.0925,
+      "step": 3313
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9361746417940592e-05,
+      "loss": 0.1019,
+      "step": 3314
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9360574514475518e-05,
+      "loss": 0.0992,
+      "step": 3315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.935940157165739e-05,
+      "loss": 0.0878,
+      "step": 3316
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.935822758961644e-05,
+      "loss": 0.1062,
+      "step": 3317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9357052568483022e-05,
+      "loss": 0.0884,
+      "step": 3318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9355876508387606e-05,
+      "loss": 0.085,
+      "step": 3319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9354699409460775e-05,
+      "loss": 0.0862,
+      "step": 3320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.935352127183323e-05,
+      "loss": 0.0973,
+      "step": 3321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9352342095635782e-05,
+      "loss": 0.1005,
+      "step": 3322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9351161880999363e-05,
+      "loss": 0.0907,
+      "step": 3323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9349980628055023e-05,
+      "loss": 0.1031,
+      "step": 3324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9348798336933916e-05,
+      "loss": 0.1008,
+      "step": 3325
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.934761500776732e-05,
+      "loss": 0.1106,
+      "step": 3326
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9346430640686625e-05,
+      "loss": 0.1052,
+      "step": 3327
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9345245235823343e-05,
+      "loss": 0.1078,
+      "step": 3328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9344058793309085e-05,
+      "loss": 0.0988,
+      "step": 3329
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.93428713132756e-05,
+      "loss": 0.0947,
+      "step": 3330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.934168279585473e-05,
+      "loss": 0.0817,
+      "step": 3331
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9340493241178452e-05,
+      "loss": 0.1021,
+      "step": 3332
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.933930264937884e-05,
+      "loss": 0.0842,
+      "step": 3333
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9338111020588092e-05,
+      "loss": 0.1031,
+      "step": 3334
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.933691835493853e-05,
+      "loss": 0.0948,
+      "step": 3335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.933572465256257e-05,
+      "loss": 0.0979,
+      "step": 3336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.933452991359276e-05,
+      "loss": 0.1073,
+      "step": 3337
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.933333413816176e-05,
+      "loss": 0.1221,
+      "step": 3338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.933213732640234e-05,
+      "loss": 0.0828,
+      "step": 3339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9330939478447392e-05,
+      "loss": 0.1119,
+      "step": 3340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9329740594429913e-05,
+      "loss": 0.1112,
+      "step": 3341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.932854067448303e-05,
+      "loss": 0.0934,
+      "step": 3342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.932733971873997e-05,
+      "loss": 0.0941,
+      "step": 3343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.932613772733408e-05,
+      "loss": 0.0999,
+      "step": 3344
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9324934700398827e-05,
+      "loss": 0.1013,
+      "step": 3345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.932373063806779e-05,
+      "loss": 0.1011,
+      "step": 3346
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.932252554047466e-05,
+      "loss": 0.1012,
+      "step": 3347
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9321319407753244e-05,
+      "loss": 0.0906,
+      "step": 3348
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9320112240037466e-05,
+      "loss": 0.09,
+      "step": 3349
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.931890403746137e-05,
+      "loss": 0.0939,
+      "step": 3350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9317694800159103e-05,
+      "loss": 0.0902,
+      "step": 3351
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9316484528264932e-05,
+      "loss": 0.0979,
+      "step": 3352
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.931527322191324e-05,
+      "loss": 0.0957,
+      "step": 3353
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9314060881238532e-05,
+      "loss": 0.0822,
+      "step": 3354
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9312847506375413e-05,
+      "loss": 0.1055,
+      "step": 3355
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9311633097458608e-05,
+      "loss": 0.0946,
+      "step": 3356
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.931041765462297e-05,
+      "loss": 0.1071,
+      "step": 3357
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9309201178003443e-05,
+      "loss": 0.0817,
+      "step": 3358
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9307983667735106e-05,
+      "loss": 0.072,
+      "step": 3359
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.930676512395315e-05,
+      "loss": 0.1047,
+      "step": 3360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9305545546792863e-05,
+      "loss": 0.0923,
+      "step": 3361
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9304324936389675e-05,
+      "loss": 0.0685,
+      "step": 3362
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9303103292879104e-05,
+      "loss": 0.0979,
+      "step": 3363
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9301880616396803e-05,
+      "loss": 0.1194,
+      "step": 3364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9300656907078533e-05,
+      "loss": 0.1052,
+      "step": 3365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9299432165060166e-05,
+      "loss": 0.0956,
+      "step": 3366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9298206390477693e-05,
+      "loss": 0.0929,
+      "step": 3367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.929697958346722e-05,
+      "loss": 0.1121,
+      "step": 3368
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9295751744164955e-05,
+      "loss": 0.0944,
+      "step": 3369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.929452287270724e-05,
+      "loss": 0.0891,
+      "step": 3370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9293292969230527e-05,
+      "loss": 0.0824,
+      "step": 3371
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9292062033871374e-05,
+      "loss": 0.0874,
+      "step": 3372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9290830066766454e-05,
+      "loss": 0.0933,
+      "step": 3373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9289597068052563e-05,
+      "loss": 0.0965,
+      "step": 3374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.928836303786661e-05,
+      "loss": 0.0994,
+      "step": 3375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.928712797634561e-05,
+      "loss": 0.0917,
+      "step": 3376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9285891883626698e-05,
+      "loss": 0.0979,
+      "step": 3377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9284654759847127e-05,
+      "loss": 0.1254,
+      "step": 3378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9283416605144264e-05,
+      "loss": 0.0795,
+      "step": 3379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9282177419655586e-05,
+      "loss": 0.1093,
+      "step": 3380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.928093720351868e-05,
+      "loss": 0.0939,
+      "step": 3381
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.927969595687126e-05,
+      "loss": 0.0834,
+      "step": 3382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9278453679851147e-05,
+      "loss": 0.0869,
+      "step": 3383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9277210372596278e-05,
+      "loss": 0.109,
+      "step": 3384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9275966035244702e-05,
+      "loss": 0.1045,
+      "step": 3385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9274720667934585e-05,
+      "loss": 0.1025,
+      "step": 3386
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9273474270804206e-05,
+      "loss": 0.0984,
+      "step": 3387
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9272226843991956e-05,
+      "loss": 0.0777,
+      "step": 3388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.927097838763635e-05,
+      "loss": 0.0931,
+      "step": 3389
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.926972890187601e-05,
+      "loss": 0.0864,
+      "step": 3390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9268478386849664e-05,
+      "loss": 0.0995,
+      "step": 3391
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9267226842696174e-05,
+      "loss": 0.1156,
+      "step": 3392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9265974269554497e-05,
+      "loss": 0.0879,
+      "step": 3393
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9264720667563718e-05,
+      "loss": 0.1008,
+      "step": 3394
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.926346603686303e-05,
+      "loss": 0.1107,
+      "step": 3395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.926221037759174e-05,
+      "loss": 0.0919,
+      "step": 3396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.926095368988927e-05,
+      "loss": 0.1066,
+      "step": 3397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.925969597389516e-05,
+      "loss": 0.1068,
+      "step": 3398
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9258437229749054e-05,
+      "loss": 0.0956,
+      "step": 3399
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.925717745759072e-05,
+      "loss": 0.0768,
+      "step": 3400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9255916657560042e-05,
+      "loss": 0.101,
+      "step": 3401
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9254654829797007e-05,
+      "loss": 0.0911,
+      "step": 3402
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.925339197444173e-05,
+      "loss": 0.1041,
+      "step": 3403
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9252128091634418e-05,
+      "loss": 0.0961,
+      "step": 3404
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.925086318151542e-05,
+      "loss": 0.0973,
+      "step": 3405
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.924959724422518e-05,
+      "loss": 0.1003,
+      "step": 3406
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9248330279904262e-05,
+      "loss": 0.0958,
+      "step": 3407
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9247062288693342e-05,
+      "loss": 0.0883,
+      "step": 3408
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9245793270733216e-05,
+      "loss": 0.0834,
+      "step": 3409
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9244523226164785e-05,
+      "loss": 0.0947,
+      "step": 3410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9243252155129075e-05,
+      "loss": 0.0865,
+      "step": 3411
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.924198005776721e-05,
+      "loss": 0.0995,
+      "step": 3412
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9240706934220447e-05,
+      "loss": 0.0986,
+      "step": 3413
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9239432784630145e-05,
+      "loss": 0.0853,
+      "step": 3414
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9238157609137775e-05,
+      "loss": 0.1193,
+      "step": 3415
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9236881407884928e-05,
+      "loss": 0.0994,
+      "step": 3416
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9235604181013306e-05,
+      "loss": 0.088,
+      "step": 3417
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9234325928664736e-05,
+      "loss": 0.0902,
+      "step": 3418
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9233046650981137e-05,
+      "loss": 0.1096,
+      "step": 3419
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9231766348104556e-05,
+      "loss": 0.0892,
+      "step": 3420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.923048502017716e-05,
+      "loss": 0.1096,
+      "step": 3421
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.922920266734121e-05,
+      "loss": 0.1152,
+      "step": 3422
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.92279192897391e-05,
+      "loss": 0.1146,
+      "step": 3423
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.922663488751333e-05,
+      "loss": 0.0907,
+      "step": 3424
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9225349460806507e-05,
+      "loss": 0.0872,
+      "step": 3425
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9224063009761367e-05,
+      "loss": 0.1254,
+      "step": 3426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9222775534520744e-05,
+      "loss": 0.0962,
+      "step": 3427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9221487035227602e-05,
+      "loss": 0.0803,
+      "step": 3428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9220197512025e-05,
+      "loss": 0.094,
+      "step": 3429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9218906965056126e-05,
+      "loss": 0.0975,
+      "step": 3430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.921761539446427e-05,
+      "loss": 0.0948,
+      "step": 3431
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9216322800392855e-05,
+      "loss": 0.1169,
+      "step": 3432
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9215029182985392e-05,
+      "loss": 0.115,
+      "step": 3433
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.921373454238552e-05,
+      "loss": 0.0842,
+      "step": 3434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9212438878736997e-05,
+      "loss": 0.105,
+      "step": 3435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9211142192183683e-05,
+      "loss": 0.0895,
+      "step": 3436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.920984448286955e-05,
+      "loss": 0.0953,
+      "step": 3437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9208545750938693e-05,
+      "loss": 0.0996,
+      "step": 3438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.920724599653532e-05,
+      "loss": 0.085,
+      "step": 3439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.920594521980375e-05,
+      "loss": 0.0923,
+      "step": 3440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.920464342088841e-05,
+      "loss": 0.0979,
+      "step": 3441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9203340599933852e-05,
+      "loss": 0.1232,
+      "step": 3442
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9202036757084725e-05,
+      "loss": 0.1122,
+      "step": 3443
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.920073189248581e-05,
+      "loss": 0.0999,
+      "step": 3444
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9199426006281992e-05,
+      "loss": 0.0963,
+      "step": 3445
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9198119098618268e-05,
+      "loss": 0.0934,
+      "step": 3446
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.919681116963975e-05,
+      "loss": 0.1334,
+      "step": 3447
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9195502219491663e-05,
+      "loss": 0.0973,
+      "step": 3448
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9194192248319355e-05,
+      "loss": 0.0858,
+      "step": 3449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.919288125626827e-05,
+      "loss": 0.0933,
+      "step": 3450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9191569243483977e-05,
+      "loss": 0.0994,
+      "step": 3451
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9190256210112155e-05,
+      "loss": 0.0848,
+      "step": 3452
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.91889421562986e-05,
+      "loss": 0.0875,
+      "step": 3453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9187627082189212e-05,
+      "loss": 0.0988,
+      "step": 3454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9186310987930014e-05,
+      "loss": 0.0844,
+      "step": 3455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.918499387366714e-05,
+      "loss": 0.0922,
+      "step": 3456
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.918367573954684e-05,
+      "loss": 0.1028,
+      "step": 3457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9182356585715464e-05,
+      "loss": 0.098,
+      "step": 3458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.918103641231949e-05,
+      "loss": 0.0957,
+      "step": 3459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9179715219505498e-05,
+      "loss": 0.0934,
+      "step": 3460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.91783930074202e-05,
+      "loss": 0.0782,
+      "step": 3461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.917706977621039e-05,
+      "loss": 0.0994,
+      "step": 3462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9175745526023005e-05,
+      "loss": 0.0925,
+      "step": 3463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9174420257005085e-05,
+      "loss": 0.0679,
+      "step": 3464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9173093969303775e-05,
+      "loss": 0.0825,
+      "step": 3465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9171766663066345e-05,
+      "loss": 0.0878,
+      "step": 3466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9170438338440165e-05,
+      "loss": 0.0977,
+      "step": 3467
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9169108995572735e-05,
+      "loss": 0.0991,
+      "step": 3468
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9167778634611653e-05,
+      "loss": 0.0864,
+      "step": 3469
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9166447255704632e-05,
+      "loss": 0.0945,
+      "step": 3470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9165114858999512e-05,
+      "loss": 0.1051,
+      "step": 3471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.916378144464423e-05,
+      "loss": 0.1051,
+      "step": 3472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9162447012786843e-05,
+      "loss": 0.1008,
+      "step": 3473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9161111563575518e-05,
+      "loss": 0.08,
+      "step": 3474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9159775097158542e-05,
+      "loss": 0.0966,
+      "step": 3475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9158437613684305e-05,
+      "loss": 0.0961,
+      "step": 3476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9157099113301316e-05,
+      "loss": 0.095,
+      "step": 3477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9155759596158195e-05,
+      "loss": 0.1,
+      "step": 3478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9154419062403675e-05,
+      "loss": 0.1066,
+      "step": 3479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9153077512186605e-05,
+      "loss": 0.0961,
+      "step": 3480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9151734945655946e-05,
+      "loss": 0.098,
+      "step": 3481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.915039136296076e-05,
+      "loss": 0.0876,
+      "step": 3482
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9149046764250244e-05,
+      "loss": 0.0947,
+      "step": 3483
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.914770114967369e-05,
+      "loss": 0.0681,
+      "step": 3484
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.914635451938051e-05,
+      "loss": 0.0827,
+      "step": 3485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9145006873520227e-05,
+      "loss": 0.1014,
+      "step": 3486
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9143658212242475e-05,
+      "loss": 0.0993,
+      "step": 3487
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9142308535697005e-05,
+      "loss": 0.1072,
+      "step": 3488
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.914095784403368e-05,
+      "loss": 0.104,
+      "step": 3489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9139606137402468e-05,
+      "loss": 0.0959,
+      "step": 3490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9138253415953466e-05,
+      "loss": 0.095,
+      "step": 3491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9136899679836863e-05,
+      "loss": 0.0837,
+      "step": 3492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9135544929202977e-05,
+      "loss": 0.1064,
+      "step": 3493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9134189164202237e-05,
+      "loss": 0.0972,
+      "step": 3494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.913283238498517e-05,
+      "loss": 0.0997,
+      "step": 3495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9131474591702438e-05,
+      "loss": 0.0837,
+      "step": 3496
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.913011578450479e-05,
+      "loss": 0.0973,
+      "step": 3497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9128755963543118e-05,
+      "loss": 0.0964,
+      "step": 3498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.91273951289684e-05,
+      "loss": 0.0957,
+      "step": 3499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9126033280931734e-05,
+      "loss": 0.089,
+      "step": 3500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9124670419584338e-05,
+      "loss": 0.0924,
+      "step": 3501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9123306545077536e-05,
+      "loss": 0.0907,
+      "step": 3502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.912194165756277e-05,
+      "loss": 0.0867,
+      "step": 3503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9120575757191584e-05,
+      "loss": 0.1123,
+      "step": 3504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9119208844115644e-05,
+      "loss": 0.0972,
+      "step": 3505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9117840918486727e-05,
+      "loss": 0.0883,
+      "step": 3506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.911647198045672e-05,
+      "loss": 0.0924,
+      "step": 3507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.911510203017762e-05,
+      "loss": 0.1012,
+      "step": 3508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9113731067801543e-05,
+      "loss": 0.1083,
+      "step": 3509
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9112359093480716e-05,
+      "loss": 0.0975,
+      "step": 3510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.911098610736747e-05,
+      "loss": 0.0991,
+      "step": 3511
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9109612109614263e-05,
+      "loss": 0.094,
+      "step": 3512
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9108237100373647e-05,
+      "loss": 0.1,
+      "step": 3513
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9106861079798308e-05,
+      "loss": 0.0764,
+      "step": 3514
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9105484048041024e-05,
+      "loss": 0.0682,
+      "step": 3515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9104106005254696e-05,
+      "loss": 0.0866,
+      "step": 3516
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9102726951592338e-05,
+      "loss": 0.1008,
+      "step": 3517
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9101346887207065e-05,
+      "loss": 0.1121,
+      "step": 3518
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9099965812252125e-05,
+      "loss": 0.0936,
+      "step": 3519
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.909858372688086e-05,
+      "loss": 0.1168,
+      "step": 3520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9097200631246727e-05,
+      "loss": 0.0728,
+      "step": 3521
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9095816525503304e-05,
+      "loss": 0.0942,
+      "step": 3522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9094431409804273e-05,
+      "loss": 0.0999,
+      "step": 3523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.909304528430343e-05,
+      "loss": 0.1082,
+      "step": 3524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9091658149154683e-05,
+      "loss": 0.0795,
+      "step": 3525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9090270004512053e-05,
+      "loss": 0.1143,
+      "step": 3526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9088880850529677e-05,
+      "loss": 0.0815,
+      "step": 3527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9087490687361794e-05,
+      "loss": 0.0962,
+      "step": 3528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9086099515162763e-05,
+      "loss": 0.0883,
+      "step": 3529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9084707334087056e-05,
+      "loss": 0.1039,
+      "step": 3530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.908331414428925e-05,
+      "loss": 0.0992,
+      "step": 3531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.908191994592404e-05,
+      "loss": 0.0798,
+      "step": 3532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9080524739146232e-05,
+      "loss": 0.0833,
+      "step": 3533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9079128524110745e-05,
+      "loss": 0.0803,
+      "step": 3534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.90777313009726e-05,
+      "loss": 0.0791,
+      "step": 3535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9076333069886943e-05,
+      "loss": 0.0961,
+      "step": 3536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9074933831009028e-05,
+      "loss": 0.0983,
+      "step": 3537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9073533584494218e-05,
+      "loss": 0.1018,
+      "step": 3538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9072132330497993e-05,
+      "loss": 0.0755,
+      "step": 3539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9070730069175936e-05,
+      "loss": 0.0742,
+      "step": 3540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.906932680068375e-05,
+      "loss": 0.0886,
+      "step": 3541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.906792252517725e-05,
+      "loss": 0.1111,
+      "step": 3542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9066517242812353e-05,
+      "loss": 0.103,
+      "step": 3543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9065110953745098e-05,
+      "loss": 0.0977,
+      "step": 3544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9063703658131637e-05,
+      "loss": 0.0844,
+      "step": 3545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9062295356128225e-05,
+      "loss": 0.1091,
+      "step": 3546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9060886047891233e-05,
+      "loss": 0.1033,
+      "step": 3547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9059475733577147e-05,
+      "loss": 0.12,
+      "step": 3548
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9058064413342555e-05,
+      "loss": 0.0951,
+      "step": 3549
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.905665208734417e-05,
+      "loss": 0.098,
+      "step": 3550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9055238755738805e-05,
+      "loss": 0.1115,
+      "step": 3551
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9053824418683395e-05,
+      "loss": 0.0967,
+      "step": 3552
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9052409076334974e-05,
+      "loss": 0.111,
+      "step": 3553
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.90509927288507e-05,
+      "loss": 0.0992,
+      "step": 3554
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9049575376387838e-05,
+      "loss": 0.1117,
+      "step": 3555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9048157019103758e-05,
+      "loss": 0.0906,
+      "step": 3556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9046737657155953e-05,
+      "loss": 0.0923,
+      "step": 3557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.904531729070202e-05,
+      "loss": 0.0892,
+      "step": 3558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9043895919899677e-05,
+      "loss": 0.0857,
+      "step": 3559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9042473544906733e-05,
+      "loss": 0.1056,
+      "step": 3560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9041050165881126e-05,
+      "loss": 0.0795,
+      "step": 3561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9039625782980907e-05,
+      "loss": 0.0833,
+      "step": 3562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.903820039636423e-05,
+      "loss": 0.0861,
+      "step": 3563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.903677400618936e-05,
+      "loss": 0.1064,
+      "step": 3564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.903534661261468e-05,
+      "loss": 0.0814,
+      "step": 3565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.903391821579868e-05,
+      "loss": 0.1201,
+      "step": 3566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9032488815899958e-05,
+      "loss": 0.0826,
+      "step": 3567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9031058413077233e-05,
+      "loss": 0.1024,
+      "step": 3568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902962700748933e-05,
+      "loss": 0.0994,
+      "step": 3569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902819459929518e-05,
+      "loss": 0.1028,
+      "step": 3570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9026761188653837e-05,
+      "loss": 0.0775,
+      "step": 3571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902532677572446e-05,
+      "loss": 0.1023,
+      "step": 3572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902389136066631e-05,
+      "loss": 0.1042,
+      "step": 3573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.902245494363878e-05,
+      "loss": 0.1056,
+      "step": 3574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9021017524801356e-05,
+      "loss": 0.1003,
+      "step": 3575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9019579104313645e-05,
+      "loss": 0.0712,
+      "step": 3576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9018139682335363e-05,
+      "loss": 0.097,
+      "step": 3577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9016699259026332e-05,
+      "loss": 0.1019,
+      "step": 3578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9015257834546492e-05,
+      "loss": 0.1149,
+      "step": 3579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9013815409055895e-05,
+      "loss": 0.1011,
+      "step": 3580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9012371982714698e-05,
+      "loss": 0.1167,
+      "step": 3581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9010927555683173e-05,
+      "loss": 0.0969,
+      "step": 3582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9009482128121698e-05,
+      "loss": 0.1035,
+      "step": 3583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9008035700190774e-05,
+      "loss": 0.1044,
+      "step": 3584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9006588272051e-05,
+      "loss": 0.1016,
+      "step": 3585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.900513984386309e-05,
+      "loss": 0.0942,
+      "step": 3586
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9003690415787882e-05,
+      "loss": 0.0957,
+      "step": 3587
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9002239987986296e-05,
+      "loss": 0.077,
+      "step": 3588
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.900078856061939e-05,
+      "loss": 0.1021,
+      "step": 3589
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8999336133848327e-05,
+      "loss": 0.0792,
+      "step": 3590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8997882707834372e-05,
+      "loss": 0.0912,
+      "step": 3591
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8996428282738906e-05,
+      "loss": 0.1014,
+      "step": 3592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8994972858723425e-05,
+      "loss": 0.0781,
+      "step": 3593
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.899351643594953e-05,
+      "loss": 0.1085,
+      "step": 3594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8992059014578933e-05,
+      "loss": 0.0903,
+      "step": 3595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.899060059477346e-05,
+      "loss": 0.0918,
+      "step": 3596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8989141176695054e-05,
+      "loss": 0.0851,
+      "step": 3597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8987680760505753e-05,
+      "loss": 0.0924,
+      "step": 3598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8986219346367717e-05,
+      "loss": 0.0942,
+      "step": 3599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8984756934443215e-05,
+      "loss": 0.106,
+      "step": 3600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.898329352489463e-05,
+      "loss": 0.105,
+      "step": 3601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8981829117884446e-05,
+      "loss": 0.0865,
+      "step": 3602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8980363713575264e-05,
+      "loss": 0.103,
+      "step": 3603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.89788973121298e-05,
+      "loss": 0.0902,
+      "step": 3604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.897742991371087e-05,
+      "loss": 0.0958,
+      "step": 3605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8975961518481412e-05,
+      "loss": 0.0974,
+      "step": 3606
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.897449212660447e-05,
+      "loss": 0.1228,
+      "step": 3607
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8973021738243193e-05,
+      "loss": 0.0911,
+      "step": 3608
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8971550353560852e-05,
+      "loss": 0.1136,
+      "step": 3609
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8970077972720816e-05,
+      "loss": 0.1067,
+      "step": 3610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8968604595886578e-05,
+      "loss": 0.1093,
+      "step": 3611
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.896713022322173e-05,
+      "loss": 0.0858,
+      "step": 3612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.896565485488998e-05,
+      "loss": 0.0996,
+      "step": 3613
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8964178491055144e-05,
+      "loss": 0.1068,
+      "step": 3614
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8962701131881153e-05,
+      "loss": 0.0984,
+      "step": 3615
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8961222777532048e-05,
+      "loss": 0.0918,
+      "step": 3616
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8959743428171972e-05,
+      "loss": 0.0978,
+      "step": 3617
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8958263083965187e-05,
+      "loss": 0.0858,
+      "step": 3618
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8956781745076068e-05,
+      "loss": 0.0995,
+      "step": 3619
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8955299411669093e-05,
+      "loss": 0.1017,
+      "step": 3620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.895381608390885e-05,
+      "loss": 0.0929,
+      "step": 3621
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8952331761960044e-05,
+      "loss": 0.0795,
+      "step": 3622
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8950846445987486e-05,
+      "loss": 0.1187,
+      "step": 3623
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.89493601361561e-05,
+      "loss": 0.0867,
+      "step": 3624
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8947872832630916e-05,
+      "loss": 0.0963,
+      "step": 3625
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8946384535577078e-05,
+      "loss": 0.1066,
+      "step": 3626
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8944895245159838e-05,
+      "loss": 0.0875,
+      "step": 3627
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8943404961544565e-05,
+      "loss": 0.0886,
+      "step": 3628
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8941913684896724e-05,
+      "loss": 0.0861,
+      "step": 3629
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8940421415381908e-05,
+      "loss": 0.1183,
+      "step": 3630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8938928153165802e-05,
+      "loss": 0.11,
+      "step": 3631
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8937433898414223e-05,
+      "loss": 0.0868,
+      "step": 3632
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8935938651293076e-05,
+      "loss": 0.0889,
+      "step": 3633
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8934442411968387e-05,
+      "loss": 0.105,
+      "step": 3634
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8932945180606296e-05,
+      "loss": 0.1014,
+      "step": 3635
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8931446957373045e-05,
+      "loss": 0.0939,
+      "step": 3636
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.892994774243499e-05,
+      "loss": 0.1027,
+      "step": 3637
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8928447535958598e-05,
+      "loss": 0.1002,
+      "step": 3638
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.892694633811044e-05,
+      "loss": 0.0897,
+      "step": 3639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8925444149057206e-05,
+      "loss": 0.0888,
+      "step": 3640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8923940968965694e-05,
+      "loss": 0.1082,
+      "step": 3641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8922436798002803e-05,
+      "loss": 0.0903,
+      "step": 3642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8920931636335553e-05,
+      "loss": 0.0908,
+      "step": 3643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8919425484131072e-05,
+      "loss": 0.067,
+      "step": 3644
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8917918341556593e-05,
+      "loss": 0.1117,
+      "step": 3645
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.891641020877946e-05,
+      "loss": 0.1041,
+      "step": 3646
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8914901085967136e-05,
+      "loss": 0.092,
+      "step": 3647
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8913390973287176e-05,
+      "loss": 0.0865,
+      "step": 3648
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8911879870907266e-05,
+      "loss": 0.0833,
+      "step": 3649
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8910367778995186e-05,
+      "loss": 0.1068,
+      "step": 3650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.890885469771883e-05,
+      "loss": 0.1206,
+      "step": 3651
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8907340627246204e-05,
+      "loss": 0.1165,
+      "step": 3652
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.890582556774543e-05,
+      "loss": 0.0787,
+      "step": 3653
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8904309519384726e-05,
+      "loss": 0.0911,
+      "step": 3654
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8902792482332425e-05,
+      "loss": 0.0822,
+      "step": 3655
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.890127445675698e-05,
+      "loss": 0.0857,
+      "step": 3656
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8899755442826936e-05,
+      "loss": 0.0958,
+      "step": 3657
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8898235440710962e-05,
+      "loss": 0.0929,
+      "step": 3658
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.889671445057783e-05,
+      "loss": 0.1044,
+      "step": 3659
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8895192472596425e-05,
+      "loss": 0.0935,
+      "step": 3660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.889366950693574e-05,
+      "loss": 0.0981,
+      "step": 3661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8892145553764877e-05,
+      "loss": 0.1149,
+      "step": 3662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.889062061325305e-05,
+      "loss": 0.0997,
+      "step": 3663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8889094685569577e-05,
+      "loss": 0.094,
+      "step": 3664
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.888756777088389e-05,
+      "loss": 0.0876,
+      "step": 3665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.888603986936554e-05,
+      "loss": 0.1049,
+      "step": 3666
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.888451098118416e-05,
+      "loss": 0.0973,
+      "step": 3667
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8882981106509528e-05,
+      "loss": 0.1071,
+      "step": 3668
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8881450245511502e-05,
+      "loss": 0.0916,
+      "step": 3669
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8879918398360067e-05,
+      "loss": 0.0757,
+      "step": 3670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8878385565225314e-05,
+      "loss": 0.0903,
+      "step": 3671
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8876851746277434e-05,
+      "loss": 0.1113,
+      "step": 3672
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.887531694168674e-05,
+      "loss": 0.0865,
+      "step": 3673
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8873781151623648e-05,
+      "loss": 0.1019,
+      "step": 3674
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.887224437625869e-05,
+      "loss": 0.0911,
+      "step": 3675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8870706615762492e-05,
+      "loss": 0.1039,
+      "step": 3676
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8869167870305806e-05,
+      "loss": 0.1118,
+      "step": 3677
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8867628140059485e-05,
+      "loss": 0.0966,
+      "step": 3678
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8866087425194493e-05,
+      "loss": 0.0884,
+      "step": 3679
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8864545725881908e-05,
+      "loss": 0.0846,
+      "step": 3680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8863003042292904e-05,
+      "loss": 0.083,
+      "step": 3681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8861459374598783e-05,
+      "loss": 0.1019,
+      "step": 3682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.885991472297094e-05,
+      "loss": 0.0818,
+      "step": 3683
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8858369087580887e-05,
+      "loss": 0.1016,
+      "step": 3684
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8856822468600245e-05,
+      "loss": 0.0901,
+      "step": 3685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.885527486620074e-05,
+      "loss": 0.0876,
+      "step": 3686
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8853726280554215e-05,
+      "loss": 0.0986,
+      "step": 3687
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8852176711832614e-05,
+      "loss": 0.085,
+      "step": 3688
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8850626160207998e-05,
+      "loss": 0.0883,
+      "step": 3689
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8849074625852527e-05,
+      "loss": 0.0816,
+      "step": 3690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8847522108938482e-05,
+      "loss": 0.0931,
+      "step": 3691
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.884596860963824e-05,
+      "loss": 0.0861,
+      "step": 3692
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8844414128124294e-05,
+      "loss": 0.1071,
+      "step": 3693
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8842858664569257e-05,
+      "loss": 0.099,
+      "step": 3694
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.884130221914583e-05,
+      "loss": 0.0852,
+      "step": 3695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8839744792026837e-05,
+      "loss": 0.1125,
+      "step": 3696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8838186383385205e-05,
+      "loss": 0.0977,
+      "step": 3697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8836626993393972e-05,
+      "loss": 0.0905,
+      "step": 3698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.883506662222629e-05,
+      "loss": 0.0912,
+      "step": 3699
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.883350527005541e-05,
+      "loss": 0.0979,
+      "step": 3700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8831942937054697e-05,
+      "loss": 0.1066,
+      "step": 3701
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.883037962339763e-05,
+      "loss": 0.1082,
+      "step": 3702
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.882881532925779e-05,
+      "loss": 0.0829,
+      "step": 3703
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8827250054808864e-05,
+      "loss": 0.094,
+      "step": 3704
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8825683800224655e-05,
+      "loss": 0.0776,
+      "step": 3705
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8824116565679074e-05,
+      "loss": 0.1007,
+      "step": 3706
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.882254835134614e-05,
+      "loss": 0.0963,
+      "step": 3707
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8820979157399976e-05,
+      "loss": 0.0918,
+      "step": 3708
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.881940898401482e-05,
+      "loss": 0.0887,
+      "step": 3709
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8817837831365015e-05,
+      "loss": 0.1093,
+      "step": 3710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8816265699625015e-05,
+      "loss": 0.0992,
+      "step": 3711
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8814692588969387e-05,
+      "loss": 0.0918,
+      "step": 3712
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8813118499572796e-05,
+      "loss": 0.1002,
+      "step": 3713
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.881154343161002e-05,
+      "loss": 0.0948,
+      "step": 3714
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8809967385255952e-05,
+      "loss": 0.0578,
+      "step": 3715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8808390360685586e-05,
+      "loss": 0.0913,
+      "step": 3716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8806812358074024e-05,
+      "loss": 0.111,
+      "step": 3717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8805233377596484e-05,
+      "loss": 0.0702,
+      "step": 3718
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.880365341942829e-05,
+      "loss": 0.0913,
+      "step": 3719
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8802072483744867e-05,
+      "loss": 0.1067,
+      "step": 3720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.880049057072176e-05,
+      "loss": 0.1,
+      "step": 3721
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8798907680534615e-05,
+      "loss": 0.0823,
+      "step": 3722
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8797323813359186e-05,
+      "loss": 0.1168,
+      "step": 3723
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8795738969371343e-05,
+      "loss": 0.0859,
+      "step": 3724
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8794153148747055e-05,
+      "loss": 0.0841,
+      "step": 3725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8792566351662405e-05,
+      "loss": 0.1073,
+      "step": 3726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8790978578293584e-05,
+      "loss": 0.0848,
+      "step": 3727
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8789389828816894e-05,
+      "loss": 0.1058,
+      "step": 3728
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8787800103408733e-05,
+      "loss": 0.0973,
+      "step": 3729
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8786209402245624e-05,
+      "loss": 0.092,
+      "step": 3730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.878461772550419e-05,
+      "loss": 0.0747,
+      "step": 3731
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8783025073361162e-05,
+      "loss": 0.1152,
+      "step": 3732
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.878143144599338e-05,
+      "loss": 0.0756,
+      "step": 3733
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8779836843577796e-05,
+      "loss": 0.0831,
+      "step": 3734
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.877824126629146e-05,
+      "loss": 0.1039,
+      "step": 3735
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.877664471431154e-05,
+      "loss": 0.1089,
+      "step": 3736
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8775047187815313e-05,
+      "loss": 0.0812,
+      "step": 3737
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8773448686980156e-05,
+      "loss": 0.0835,
+      "step": 3738
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8771849211983567e-05,
+      "loss": 0.0948,
+      "step": 3739
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8770248763003135e-05,
+      "loss": 0.0906,
+      "step": 3740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8768647340216567e-05,
+      "loss": 0.0892,
+      "step": 3741
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8767044943801683e-05,
+      "loss": 0.0935,
+      "step": 3742
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.87654415739364e-05,
+      "loss": 0.0835,
+      "step": 3743
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.876383723079875e-05,
+      "loss": 0.104,
+      "step": 3744
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8762231914566877e-05,
+      "loss": 0.0822,
+      "step": 3745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8760625625419014e-05,
+      "loss": 0.0843,
+      "step": 3746
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8759018363533528e-05,
+      "loss": 0.0992,
+      "step": 3747
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.875741012908888e-05,
+      "loss": 0.1107,
+      "step": 3748
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8755800922263633e-05,
+      "loss": 0.0997,
+      "step": 3749
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8754190743236476e-05,
+      "loss": 0.0796,
+      "step": 3750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.875257959218619e-05,
+      "loss": 0.1118,
+      "step": 3751
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8750967469291666e-05,
+      "loss": 0.0894,
+      "step": 3752
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.874935437473191e-05,
+      "loss": 0.1061,
+      "step": 3753
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.874774030868604e-05,
+      "loss": 0.0898,
+      "step": 3754
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8746125271333257e-05,
+      "loss": 0.1301,
+      "step": 3755
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8744509262852902e-05,
+      "loss": 0.1027,
+      "step": 3756
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.87428922834244e-05,
+      "loss": 0.0851,
+      "step": 3757
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.87412743332273e-05,
+      "loss": 0.0831,
+      "step": 3758
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8739655412441243e-05,
+      "loss": 0.0947,
+      "step": 3759
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.873803552124599e-05,
+      "loss": 0.0863,
+      "step": 3760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.873641465982141e-05,
+      "loss": 0.0918,
+      "step": 3761
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8734792828347472e-05,
+      "loss": 0.0914,
+      "step": 3762
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8733170027004254e-05,
+      "loss": 0.0996,
+      "step": 3763
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8731546255971948e-05,
+      "loss": 0.0922,
+      "step": 3764
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.872992151543085e-05,
+      "loss": 0.0977,
+      "step": 3765
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8728295805561355e-05,
+      "loss": 0.1016,
+      "step": 3766
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8726669126543985e-05,
+      "loss": 0.1002,
+      "step": 3767
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8725041478559354e-05,
+      "loss": 0.1084,
+      "step": 3768
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8723412861788187e-05,
+      "loss": 0.0956,
+      "step": 3769
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.872178327641132e-05,
+      "loss": 0.1029,
+      "step": 3770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8720152722609692e-05,
+      "loss": 0.0864,
+      "step": 3771
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8718521200564352e-05,
+      "loss": 0.093,
+      "step": 3772
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8716888710456458e-05,
+      "loss": 0.1099,
+      "step": 3773
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8715255252467274e-05,
+      "loss": 0.0959,
+      "step": 3774
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.871362082677817e-05,
+      "loss": 0.1086,
+      "step": 3775
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8711985433570628e-05,
+      "loss": 0.1024,
+      "step": 3776
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8710349073026227e-05,
+      "loss": 0.1133,
+      "step": 3777
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8708711745326668e-05,
+      "loss": 0.1022,
+      "step": 3778
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.870707345065375e-05,
+      "loss": 0.0905,
+      "step": 3779
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8705434189189374e-05,
+      "loss": 0.0746,
+      "step": 3780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.870379396111557e-05,
+      "loss": 0.0742,
+      "step": 3781
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.870215276661445e-05,
+      "loss": 0.1172,
+      "step": 3782
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8700510605868246e-05,
+      "loss": 0.0923,
+      "step": 3783
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.86988674790593e-05,
+      "loss": 0.103,
+      "step": 3784
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8697223386370048e-05,
+      "loss": 0.0797,
+      "step": 3785
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8695578327983054e-05,
+      "loss": 0.0885,
+      "step": 3786
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8693932304080967e-05,
+      "loss": 0.0984,
+      "step": 3787
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.869228531484656e-05,
+      "loss": 0.0955,
+      "step": 3788
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8690637360462706e-05,
+      "loss": 0.0986,
+      "step": 3789
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.868898844111238e-05,
+      "loss": 0.0895,
+      "step": 3790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.868733855697868e-05,
+      "loss": 0.0873,
+      "step": 3791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8685687708244794e-05,
+      "loss": 0.0968,
+      "step": 3792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8684035895094025e-05,
+      "loss": 0.087,
+      "step": 3793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8682383117709783e-05,
+      "loss": 0.0892,
+      "step": 3794
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8680729376275584e-05,
+      "loss": 0.0928,
+      "step": 3795
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8679074670975056e-05,
+      "loss": 0.076,
+      "step": 3796
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8677419001991924e-05,
+      "loss": 0.0901,
+      "step": 3797
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8675762369510027e-05,
+      "loss": 0.0847,
+      "step": 3798
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.867410477371331e-05,
+      "loss": 0.0913,
+      "step": 3799
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8672446214785824e-05,
+      "loss": 0.1008,
+      "step": 3800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8670786692911727e-05,
+      "loss": 0.0914,
+      "step": 3801
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8669126208275286e-05,
+      "loss": 0.0862,
+      "step": 3802
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8667464761060874e-05,
+      "loss": 0.1127,
+      "step": 3803
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8665802351452966e-05,
+      "loss": 0.0894,
+      "step": 3804
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8664138979636152e-05,
+      "loss": 0.1006,
+      "step": 3805
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.866247464579512e-05,
+      "loss": 0.1002,
+      "step": 3806
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8660809350114673e-05,
+      "loss": 0.1047,
+      "step": 3807
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.865914309277972e-05,
+      "loss": 0.0992,
+      "step": 3808
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8657475873975267e-05,
+      "loss": 0.093,
+      "step": 3809
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.865580769388644e-05,
+      "loss": 0.097,
+      "step": 3810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8654138552698463e-05,
+      "loss": 0.0834,
+      "step": 3811
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8652468450596673e-05,
+      "loss": 0.1064,
+      "step": 3812
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8650797387766502e-05,
+      "loss": 0.0868,
+      "step": 3813
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.864912536439351e-05,
+      "loss": 0.1024,
+      "step": 3814
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8647452380663335e-05,
+      "loss": 0.1005,
+      "step": 3815
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8645778436761748e-05,
+      "loss": 0.0872,
+      "step": 3816
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8644103532874612e-05,
+      "loss": 0.0889,
+      "step": 3817
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.86424276691879e-05,
+      "loss": 0.094,
+      "step": 3818
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.864075084588769e-05,
+      "loss": 0.0848,
+      "step": 3819
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8639073063160172e-05,
+      "loss": 0.0804,
+      "step": 3820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.863739432119164e-05,
+      "loss": 0.0986,
+      "step": 3821
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8635714620168488e-05,
+      "loss": 0.0847,
+      "step": 3822
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8634033960277226e-05,
+      "loss": 0.097,
+      "step": 3823
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.863235234170446e-05,
+      "loss": 0.0925,
+      "step": 3824
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8630669764636922e-05,
+      "loss": 0.1083,
+      "step": 3825
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8628986229261426e-05,
+      "loss": 0.099,
+      "step": 3826
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8627301735764907e-05,
+      "loss": 0.0798,
+      "step": 3827
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8625616284334405e-05,
+      "loss": 0.0945,
+      "step": 3828
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.862392987515706e-05,
+      "loss": 0.0858,
+      "step": 3829
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8622242508420123e-05,
+      "loss": 0.0948,
+      "step": 3830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8620554184310954e-05,
+      "loss": 0.1005,
+      "step": 3831
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8618864903017018e-05,
+      "loss": 0.0831,
+      "step": 3832
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8617174664725877e-05,
+      "loss": 0.0935,
+      "step": 3833
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.861548346962522e-05,
+      "loss": 0.0988,
+      "step": 3834
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8613791317902815e-05,
+      "loss": 0.079,
+      "step": 3835
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.861209820974656e-05,
+      "loss": 0.0856,
+      "step": 3836
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8610404145344445e-05,
+      "loss": 0.1057,
+      "step": 3837
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.860870912488457e-05,
+      "loss": 0.088,
+      "step": 3838
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8607013148555148e-05,
+      "loss": 0.0944,
+      "step": 3839
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8605316216544485e-05,
+      "loss": 0.0908,
+      "step": 3840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8603618329041002e-05,
+      "loss": 0.0983,
+      "step": 3841
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8601919486233227e-05,
+      "loss": 0.0929,
+      "step": 3842
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.860021968830979e-05,
+      "loss": 0.0933,
+      "step": 3843
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8598518935459424e-05,
+      "loss": 0.0969,
+      "step": 3844
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.859681722787098e-05,
+      "loss": 0.1063,
+      "step": 3845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.85951145657334e-05,
+      "loss": 0.0938,
+      "step": 3846
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8593410949235747e-05,
+      "loss": 0.0881,
+      "step": 3847
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.859170637856718e-05,
+      "loss": 0.0925,
+      "step": 3848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.859000085391696e-05,
+      "loss": 0.0986,
+      "step": 3849
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8588294375474466e-05,
+      "loss": 0.0811,
+      "step": 3850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8586586943429177e-05,
+      "loss": 0.1006,
+      "step": 3851
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8584878557970677e-05,
+      "loss": 0.0889,
+      "step": 3852
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8583169219288658e-05,
+      "loss": 0.1001,
+      "step": 3853
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8581458927572912e-05,
+      "loss": 0.1104,
+      "step": 3854
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.857974768301335e-05,
+      "loss": 0.0704,
+      "step": 3855
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.857803548579997e-05,
+      "loss": 0.093,
+      "step": 3856
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8576322336122898e-05,
+      "loss": 0.0881,
+      "step": 3857
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8574608234172347e-05,
+      "loss": 0.0881,
+      "step": 3858
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.857289318013864e-05,
+      "loss": 0.0978,
+      "step": 3859
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8571177174212214e-05,
+      "loss": 0.0903,
+      "step": 3860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.85694602165836e-05,
+      "loss": 0.1071,
+      "step": 3861
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.856774230744345e-05,
+      "loss": 0.1058,
+      "step": 3862
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8566023446982503e-05,
+      "loss": 0.0781,
+      "step": 3863
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8564303635391617e-05,
+      "loss": 0.0826,
+      "step": 3864
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8562582872861748e-05,
+      "loss": 0.0968,
+      "step": 3865
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.856086115958397e-05,
+      "loss": 0.109,
+      "step": 3866
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8559138495749445e-05,
+      "loss": 0.0945,
+      "step": 3867
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8557414881549453e-05,
+      "loss": 0.0863,
+      "step": 3868
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8555690317175375e-05,
+      "loss": 0.0697,
+      "step": 3869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.85539648028187e-05,
+      "loss": 0.11,
+      "step": 3870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.855223833867102e-05,
+      "loss": 0.0948,
+      "step": 3871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.855051092492403e-05,
+      "loss": 0.0985,
+      "step": 3872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8548782561769535e-05,
+      "loss": 0.0718,
+      "step": 3873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8547053249399448e-05,
+      "loss": 0.0997,
+      "step": 3874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.854532298800578e-05,
+      "loss": 0.1047,
+      "step": 3875
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8543591777780653e-05,
+      "loss": 0.0779,
+      "step": 3876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.854185961891629e-05,
+      "loss": 0.0818,
+      "step": 3877
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.854012651160502e-05,
+      "loss": 0.0967,
+      "step": 3878
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8538392456039286e-05,
+      "loss": 0.0826,
+      "step": 3879
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.853665745241162e-05,
+      "loss": 0.0844,
+      "step": 3880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8534921500914677e-05,
+      "loss": 0.0852,
+      "step": 3881
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8533184601741205e-05,
+      "loss": 0.0707,
+      "step": 3882
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8531446755084057e-05,
+      "loss": 0.0938,
+      "step": 3883
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8529707961136202e-05,
+      "loss": 0.0903,
+      "step": 3884
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8527968220090705e-05,
+      "loss": 0.1024,
+      "step": 3885
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8526227532140734e-05,
+      "loss": 0.1116,
+      "step": 3886
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.852448589747957e-05,
+      "loss": 0.0814,
+      "step": 3887
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8522743316300597e-05,
+      "loss": 0.0741,
+      "step": 3888
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8520999788797303e-05,
+      "loss": 0.0997,
+      "step": 3889
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8519255315163278e-05,
+      "loss": 0.0796,
+      "step": 3890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8517509895592223e-05,
+      "loss": 0.0873,
+      "step": 3891
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8515763530277938e-05,
+      "loss": 0.079,
+      "step": 3892
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8514016219414334e-05,
+      "loss": 0.0977,
+      "step": 3893
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.851226796319542e-05,
+      "loss": 0.0936,
+      "step": 3894
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8510518761815318e-05,
+      "loss": 0.0999,
+      "step": 3895
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8508768615468248e-05,
+      "loss": 0.0857,
+      "step": 3896
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.850701752434854e-05,
+      "loss": 0.0874,
+      "step": 3897
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8505265488650625e-05,
+      "loss": 0.0996,
+      "step": 3898
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.850351250856904e-05,
+      "loss": 0.0868,
+      "step": 3899
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8501758584298436e-05,
+      "loss": 0.0679,
+      "step": 3900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8500003716033546e-05,
+      "loss": 0.0666,
+      "step": 3901
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8498247903969232e-05,
+      "loss": 0.0844,
+      "step": 3902
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8496491148300446e-05,
+      "loss": 0.0848,
+      "step": 3903
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8494733449222254e-05,
+      "loss": 0.0809,
+      "step": 3904
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8492974806929816e-05,
+      "loss": 0.082,
+      "step": 3905
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8491215221618413e-05,
+      "loss": 0.0833,
+      "step": 3906
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.848945469348341e-05,
+      "loss": 0.0952,
+      "step": 3907
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8487693222720297e-05,
+      "loss": 0.0807,
+      "step": 3908
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8485930809524655e-05,
+      "loss": 0.0906,
+      "step": 3909
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.848416745409217e-05,
+      "loss": 0.0924,
+      "step": 3910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8482403156618644e-05,
+      "loss": 0.0802,
+      "step": 3911
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.848063791729997e-05,
+      "loss": 0.0857,
+      "step": 3912
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8478871736332156e-05,
+      "loss": 0.1002,
+      "step": 3913
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.847710461391131e-05,
+      "loss": 0.0807,
+      "step": 3914
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.847533655023364e-05,
+      "loss": 0.0779,
+      "step": 3915
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8473567545495464e-05,
+      "loss": 0.0749,
+      "step": 3916
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8471797599893213e-05,
+      "loss": 0.1035,
+      "step": 3917
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.84700267136234e-05,
+      "loss": 0.0809,
+      "step": 3918
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8468254886882668e-05,
+      "loss": 0.1075,
+      "step": 3919
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.846648211986774e-05,
+      "loss": 0.0916,
+      "step": 3920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8464708412775464e-05,
+      "loss": 0.1011,
+      "step": 3921
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.846293376580278e-05,
+      "loss": 0.088,
+      "step": 3922
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.846115817914674e-05,
+      "loss": 0.0846,
+      "step": 3923
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8459381653004495e-05,
+      "loss": 0.0861,
+      "step": 3924
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.84576041875733e-05,
+      "loss": 0.0795,
+      "step": 3925
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8455825783050517e-05,
+      "loss": 0.0804,
+      "step": 3926
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8454046439633612e-05,
+      "loss": 0.0884,
+      "step": 3927
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8452266157520156e-05,
+      "loss": 0.1021,
+      "step": 3928
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.845048493690782e-05,
+      "loss": 0.1096,
+      "step": 3929
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8448702777994386e-05,
+      "loss": 0.0879,
+      "step": 3930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.844691968097773e-05,
+      "loss": 0.0941,
+      "step": 3931
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8445135646055844e-05,
+      "loss": 0.0942,
+      "step": 3932
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.844335067342682e-05,
+      "loss": 0.0966,
+      "step": 3933
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8441564763288847e-05,
+      "loss": 0.1118,
+      "step": 3934
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.843977791584023e-05,
+      "loss": 0.0739,
+      "step": 3935
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8437990131279364e-05,
+      "loss": 0.0812,
+      "step": 3936
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8436201409804763e-05,
+      "loss": 0.0876,
+      "step": 3937
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8434411751615032e-05,
+      "loss": 0.0887,
+      "step": 3938
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8432621156908894e-05,
+      "loss": 0.0984,
+      "step": 3939
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8430829625885166e-05,
+      "loss": 0.0669,
+      "step": 3940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.842903715874276e-05,
+      "loss": 0.1006,
+      "step": 3941
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8427243755680718e-05,
+      "loss": 0.1152,
+      "step": 3942
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8425449416898164e-05,
+      "loss": 0.0822,
+      "step": 3943
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8423654142594333e-05,
+      "loss": 0.108,
+      "step": 3944
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.842185793296856e-05,
+      "loss": 0.0997,
+      "step": 3945
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8420060788220296e-05,
+      "loss": 0.0947,
+      "step": 3946
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8418262708549083e-05,
+      "loss": 0.0836,
+      "step": 3947
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8416463694154564e-05,
+      "loss": 0.0741,
+      "step": 3948
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8414663745236504e-05,
+      "loss": 0.0759,
+      "step": 3949
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8412862861994755e-05,
+      "loss": 0.0938,
+      "step": 3950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8411061044629282e-05,
+      "loss": 0.0794,
+      "step": 3951
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8409258293340146e-05,
+      "loss": 0.0933,
+      "step": 3952
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8407454608327516e-05,
+      "loss": 0.075,
+      "step": 3953
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8405649989791666e-05,
+      "loss": 0.0948,
+      "step": 3954
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8403844437932973e-05,
+      "loss": 0.0977,
+      "step": 3955
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8402037952951916e-05,
+      "loss": 0.0947,
+      "step": 3956
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.840023053504908e-05,
+      "loss": 0.1005,
+      "step": 3957
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8398422184425144e-05,
+      "loss": 0.0679,
+      "step": 3958
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8396612901280907e-05,
+      "loss": 0.0865,
+      "step": 3959
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8394802685817262e-05,
+      "loss": 0.1046,
+      "step": 3960
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8392991538235208e-05,
+      "loss": 0.0824,
+      "step": 3961
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.839117945873584e-05,
+      "loss": 0.077,
+      "step": 3962
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.838936644752037e-05,
+      "loss": 0.085,
+      "step": 3963
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8387552504790097e-05,
+      "loss": 0.0792,
+      "step": 3964
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.838573763074644e-05,
+      "loss": 0.0841,
+      "step": 3965
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.838392182559091e-05,
+      "loss": 0.0749,
+      "step": 3966
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8382105089525126e-05,
+      "loss": 0.0809,
+      "step": 3967
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8380287422750812e-05,
+      "loss": 0.0808,
+      "step": 3968
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8378468825469792e-05,
+      "loss": 0.0831,
+      "step": 3969
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.837664929788399e-05,
+      "loss": 0.0979,
+      "step": 3970
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8374828840195445e-05,
+      "loss": 0.0773,
+      "step": 3971
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8373007452606283e-05,
+      "loss": 0.0889,
+      "step": 3972
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.837118513531875e-05,
+      "loss": 0.0899,
+      "step": 3973
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8369361888535187e-05,
+      "loss": 0.1041,
+      "step": 3974
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.836753771245803e-05,
+      "loss": 0.0859,
+      "step": 3975
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8365712607289835e-05,
+      "loss": 0.084,
+      "step": 3976
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8363886573233254e-05,
+      "loss": 0.0929,
+      "step": 3977
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.836205961049103e-05,
+      "loss": 0.0881,
+      "step": 3978
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8360231719266036e-05,
+      "loss": 0.0908,
+      "step": 3979
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8358402899761218e-05,
+      "loss": 0.108,
+      "step": 3980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8356573152179646e-05,
+      "loss": 0.0959,
+      "step": 3981
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.835474247672449e-05,
+      "loss": 0.0879,
+      "step": 3982
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8352910873599006e-05,
+      "loss": 0.0841,
+      "step": 3983
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8351078343006583e-05,
+      "loss": 0.0861,
+      "step": 3984
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8349244885150684e-05,
+      "loss": 0.1018,
+      "step": 3985
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.834741050023489e-05,
+      "loss": 0.0945,
+      "step": 3986
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.834557518846289e-05,
+      "loss": 0.0889,
+      "step": 3987
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.834373895003846e-05,
+      "loss": 0.0809,
+      "step": 3988
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.834190178516549e-05,
+      "loss": 0.0911,
+      "step": 3989
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8340063694047968e-05,
+      "loss": 0.0863,
+      "step": 3990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8338224676889987e-05,
+      "loss": 0.1111,
+      "step": 3991
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8336384733895748e-05,
+      "loss": 0.0874,
+      "step": 3992
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.833454386526954e-05,
+      "loss": 0.0854,
+      "step": 3993
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8332702071215773e-05,
+      "loss": 0.0905,
+      "step": 3994
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8330859351938946e-05,
+      "loss": 0.0887,
+      "step": 3995
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8329015707643665e-05,
+      "loss": 0.0952,
+      "step": 3996
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8327171138534644e-05,
+      "loss": 0.0867,
+      "step": 3997
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8325325644816686e-05,
+      "loss": 0.088,
+      "step": 3998
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.832347922669472e-05,
+      "loss": 0.0839,
+      "step": 3999
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8321631884373753e-05,
+      "loss": 0.0759,
+      "step": 4000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8319783618058902e-05,
+      "loss": 0.0885,
+      "step": 4001
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8317934427955403e-05,
+      "loss": 0.0814,
+      "step": 4002
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8316084314268568e-05,
+      "loss": 0.0829,
+      "step": 4003
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.831423327720383e-05,
+      "loss": 0.0853,
+      "step": 4004
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8312381316966718e-05,
+      "loss": 0.0878,
+      "step": 4005
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.831052843376287e-05,
+      "loss": 0.1052,
+      "step": 4006
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8308674627798014e-05,
+      "loss": 0.0895,
+      "step": 4007
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.830681989927799e-05,
+      "loss": 0.0857,
+      "step": 4008
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.830496424840874e-05,
+      "loss": 0.0847,
+      "step": 4009
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8303107675396307e-05,
+      "loss": 0.0894,
+      "step": 4010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.830125018044683e-05,
+      "loss": 0.0868,
+      "step": 4011
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.829939176376656e-05,
+      "loss": 0.0991,
+      "step": 4012
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.829753242556185e-05,
+      "loss": 0.0865,
+      "step": 4013
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.829567216603915e-05,
+      "loss": 0.0692,
+      "step": 4014
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8293810985405013e-05,
+      "loss": 0.0886,
+      "step": 4015
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8291948883866095e-05,
+      "loss": 0.0797,
+      "step": 4016
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8290085861629153e-05,
+      "loss": 0.0886,
+      "step": 4017
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8288221918901053e-05,
+      "loss": 0.0997,
+      "step": 4018
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8286357055888756e-05,
+      "loss": 0.1006,
+      "step": 4019
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8284491272799327e-05,
+      "loss": 0.0836,
+      "step": 4020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8282624569839934e-05,
+      "loss": 0.1085,
+      "step": 4021
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8280756947217845e-05,
+      "loss": 0.0894,
+      "step": 4022
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8278888405140435e-05,
+      "loss": 0.0898,
+      "step": 4023
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8277018943815173e-05,
+      "loss": 0.0945,
+      "step": 4024
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8275148563449642e-05,
+      "loss": 0.0794,
+      "step": 4025
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8273277264251515e-05,
+      "loss": 0.0928,
+      "step": 4026
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8271405046428572e-05,
+      "loss": 0.0817,
+      "step": 4027
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.82695319101887e-05,
+      "loss": 0.0826,
+      "step": 4028
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8267657855739874e-05,
+      "loss": 0.0775,
+      "step": 4029
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8265782883290187e-05,
+      "loss": 0.094,
+      "step": 4030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8263906993047823e-05,
+      "loss": 0.0874,
+      "step": 4031
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.826203018522108e-05,
+      "loss": 0.0969,
+      "step": 4032
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8260152460018336e-05,
+      "loss": 0.0786,
+      "step": 4033
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8258273817648097e-05,
+      "loss": 0.0887,
+      "step": 4034
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8256394258318954e-05,
+      "loss": 0.0746,
+      "step": 4035
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.82545137822396e-05,
+      "loss": 0.0883,
+      "step": 4036
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.825263238961884e-05,
+      "loss": 0.073,
+      "step": 4037
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8250750080665575e-05,
+      "loss": 0.0731,
+      "step": 4038
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.82488668555888e-05,
+      "loss": 0.1036,
+      "step": 4039
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.824698271459763e-05,
+      "loss": 0.0728,
+      "step": 4040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8245097657901262e-05,
+      "loss": 0.075,
+      "step": 4041
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8243211685709002e-05,
+      "loss": 0.095,
+      "step": 4042
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.824132479823027e-05,
+      "loss": 0.0836,
+      "step": 4043
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.823943699567457e-05,
+      "loss": 0.0863,
+      "step": 4044
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8237548278251517e-05,
+      "loss": 0.1002,
+      "step": 4045
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.823565864617082e-05,
+      "loss": 0.0783,
+      "step": 4046
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.82337680996423e-05,
+      "loss": 0.0746,
+      "step": 4047
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8231876638875873e-05,
+      "loss": 0.0956,
+      "step": 4048
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8229984264081555e-05,
+      "loss": 0.0823,
+      "step": 4049
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.822809097546947e-05,
+      "loss": 0.0839,
+      "step": 4050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8226196773249836e-05,
+      "loss": 0.0956,
+      "step": 4051
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.822430165763298e-05,
+      "loss": 0.0952,
+      "step": 4052
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8222405628829322e-05,
+      "loss": 0.0798,
+      "step": 4053
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.822050868704939e-05,
+      "loss": 0.079,
+      "step": 4054
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8218610832503814e-05,
+      "loss": 0.095,
+      "step": 4055
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8216712065403318e-05,
+      "loss": 0.0895,
+      "step": 4056
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8214812385958738e-05,
+      "loss": 0.0844,
+      "step": 4057
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8212911794380996e-05,
+      "loss": 0.115,
+      "step": 4058
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8211010290881132e-05,
+      "loss": 0.0931,
+      "step": 4059
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8209107875670278e-05,
+      "loss": 0.0751,
+      "step": 4060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.820720454895967e-05,
+      "loss": 0.0834,
+      "step": 4061
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8205300310960642e-05,
+      "loss": 0.1029,
+      "step": 4062
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.820339516188463e-05,
+      "loss": 0.0994,
+      "step": 4063
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8201489101943178e-05,
+      "loss": 0.062,
+      "step": 4064
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.819958213134792e-05,
+      "loss": 0.0991,
+      "step": 4065
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.81976742503106e-05,
+      "loss": 0.0826,
+      "step": 4066
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.819576545904306e-05,
+      "loss": 0.0813,
+      "step": 4067
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8193855757757242e-05,
+      "loss": 0.0817,
+      "step": 4068
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8191945146665192e-05,
+      "loss": 0.0901,
+      "step": 4069
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.819003362597905e-05,
+      "loss": 0.083,
+      "step": 4070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8188121195911067e-05,
+      "loss": 0.0943,
+      "step": 4071
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.818620785667359e-05,
+      "loss": 0.092,
+      "step": 4072
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8184293608479066e-05,
+      "loss": 0.0894,
+      "step": 4073
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.818237845154004e-05,
+      "loss": 0.0774,
+      "step": 4074
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8180462386069163e-05,
+      "loss": 0.0974,
+      "step": 4075
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8178545412279194e-05,
+      "loss": 0.0966,
+      "step": 4076
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8176627530382977e-05,
+      "loss": 0.0919,
+      "step": 4077
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.817470874059346e-05,
+      "loss": 0.0813,
+      "step": 4078
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8172789043123708e-05,
+      "loss": 0.0906,
+      "step": 4079
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8170868438186862e-05,
+      "loss": 0.0803,
+      "step": 4080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8168946925996192e-05,
+      "loss": 0.0894,
+      "step": 4081
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.816702450676504e-05,
+      "loss": 0.104,
+      "step": 4082
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8165101180706865e-05,
+      "loss": 0.0677,
+      "step": 4083
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.816317694803523e-05,
+      "loss": 0.0972,
+      "step": 4084
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8161251808963784e-05,
+      "loss": 0.0821,
+      "step": 4085
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8159325763706294e-05,
+      "loss": 0.0942,
+      "step": 4086
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8157398812476613e-05,
+      "loss": 0.0975,
+      "step": 4087
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.81554709554887e-05,
+      "loss": 0.0815,
+      "step": 4088
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.815354219295662e-05,
+      "loss": 0.0803,
+      "step": 4089
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8151612525094525e-05,
+      "loss": 0.0842,
+      "step": 4090
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8149681952116686e-05,
+      "loss": 0.0946,
+      "step": 4091
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8147750474237456e-05,
+      "loss": 0.0833,
+      "step": 4092
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8145818091671304e-05,
+      "loss": 0.0974,
+      "step": 4093
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.814388480463279e-05,
+      "loss": 0.0837,
+      "step": 4094
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.814195061333657e-05,
+      "loss": 0.1097,
+      "step": 4095
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.814001551799742e-05,
+      "loss": 0.0734,
+      "step": 4096
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8138079518830192e-05,
+      "loss": 0.0841,
+      "step": 4097
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8136142616049857e-05,
+      "loss": 0.0781,
+      "step": 4098
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8134204809871475e-05,
+      "loss": 0.0823,
+      "step": 4099
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8132266100510217e-05,
+      "loss": 0.0847,
+      "step": 4100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.813032648818134e-05,
+      "loss": 0.0796,
+      "step": 4101
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8128385973100215e-05,
+      "loss": 0.101,
+      "step": 4102
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8126444555482306e-05,
+      "loss": 0.0853,
+      "step": 4103
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8124502235543174e-05,
+      "loss": 0.0844,
+      "step": 4104
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8122559013498496e-05,
+      "loss": 0.0912,
+      "step": 4105
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8120614889564026e-05,
+      "loss": 0.0935,
+      "step": 4106
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8118669863955637e-05,
+      "loss": 0.0839,
+      "step": 4107
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8116723936889294e-05,
+      "loss": 0.083,
+      "step": 4108
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.811477710858106e-05,
+      "loss": 0.0837,
+      "step": 4109
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8112829379247106e-05,
+      "loss": 0.0803,
+      "step": 4110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8110880749103696e-05,
+      "loss": 0.0858,
+      "step": 4111
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8108931218367198e-05,
+      "loss": 0.0791,
+      "step": 4112
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8106980787254075e-05,
+      "loss": 0.0781,
+      "step": 4113
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.81050294559809e-05,
+      "loss": 0.0775,
+      "step": 4114
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8103077224764333e-05,
+      "loss": 0.095,
+      "step": 4115
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8101124093821144e-05,
+      "loss": 0.09,
+      "step": 4116
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8099170063368197e-05,
+      "loss": 0.082,
+      "step": 4117
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8097215133622464e-05,
+      "loss": 0.0883,
+      "step": 4118
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8095259304801002e-05,
+      "loss": 0.0955,
+      "step": 4119
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.809330257712098e-05,
+      "loss": 0.1051,
+      "step": 4120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8091344950799665e-05,
+      "loss": 0.0861,
+      "step": 4121
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8089386426054423e-05,
+      "loss": 0.0864,
+      "step": 4122
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.808742700310272e-05,
+      "loss": 0.0924,
+      "step": 4123
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8085466682162116e-05,
+      "loss": 0.1024,
+      "step": 4124
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.808350546345028e-05,
+      "loss": 0.0821,
+      "step": 4125
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8081543347184972e-05,
+      "loss": 0.0611,
+      "step": 4126
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8079580333584062e-05,
+      "loss": 0.0906,
+      "step": 4127
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.807761642286551e-05,
+      "loss": 0.1005,
+      "step": 4128
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8075651615247382e-05,
+      "loss": 0.1104,
+      "step": 4129
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8073685910947833e-05,
+      "loss": 0.0683,
+      "step": 4130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8071719310185133e-05,
+      "loss": 0.0941,
+      "step": 4131
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.806975181317764e-05,
+      "loss": 0.0983,
+      "step": 4132
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.806778342014382e-05,
+      "loss": 0.0839,
+      "step": 4133
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.806581413130223e-05,
+      "loss": 0.089,
+      "step": 4134
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8063843946871527e-05,
+      "loss": 0.0876,
+      "step": 4135
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.806187286707048e-05,
+      "loss": 0.0747,
+      "step": 4136
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8059900892117942e-05,
+      "loss": 0.0967,
+      "step": 4137
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8057928022232872e-05,
+      "loss": 0.0862,
+      "step": 4138
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.805595425763433e-05,
+      "loss": 0.0763,
+      "step": 4139
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8053979598541473e-05,
+      "loss": 0.0773,
+      "step": 4140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.805200404517355e-05,
+      "loss": 0.0778,
+      "step": 4141
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.805002759774993e-05,
+      "loss": 0.0835,
+      "step": 4142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8048050256490058e-05,
+      "loss": 0.0879,
+      "step": 4143
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8046072021613496e-05,
+      "loss": 0.079,
+      "step": 4144
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.804409289333989e-05,
+      "loss": 0.0839,
+      "step": 4145
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8042112871888994e-05,
+      "loss": 0.0859,
+      "step": 4146
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.804013195748067e-05,
+      "loss": 0.0808,
+      "step": 4147
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8038150150334856e-05,
+      "loss": 0.0745,
+      "step": 4148
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.803616745067161e-05,
+      "loss": 0.0837,
+      "step": 4149
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.803418385871108e-05,
+      "loss": 0.0873,
+      "step": 4150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8032199374673512e-05,
+      "loss": 0.0754,
+      "step": 4151
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.803021399877926e-05,
+      "loss": 0.077,
+      "step": 4152
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8028227731248763e-05,
+      "loss": 0.0792,
+      "step": 4153
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8026240572302567e-05,
+      "loss": 0.1056,
+      "step": 4154
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8024252522161326e-05,
+      "loss": 0.0817,
+      "step": 4155
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8022263581045775e-05,
+      "loss": 0.0838,
+      "step": 4156
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8020273749176756e-05,
+      "loss": 0.0941,
+      "step": 4157
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8018283026775213e-05,
+      "loss": 0.0825,
+      "step": 4158
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8016291414062184e-05,
+      "loss": 0.0759,
+      "step": 4159
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8014298911258813e-05,
+      "loss": 0.0762,
+      "step": 4160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8012305518586334e-05,
+      "loss": 0.0906,
+      "step": 4161
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8010311236266084e-05,
+      "loss": 0.095,
+      "step": 4162
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8008316064519498e-05,
+      "loss": 0.0793,
+      "step": 4163
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.800632000356811e-05,
+      "loss": 0.084,
+      "step": 4164
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.800432305363356e-05,
+      "loss": 0.0746,
+      "step": 4165
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.800232521493757e-05,
+      "loss": 0.0687,
+      "step": 4166
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.800032648770197e-05,
+      "loss": 0.0922,
+      "step": 4167
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.79983268721487e-05,
+      "loss": 0.0886,
+      "step": 4168
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.7996326368499776e-05,
+      "loss": 0.0886,
+      "step": 4169
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7994324976977333e-05,
+      "loss": 0.0822,
+      "step": 4170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.799232269780359e-05,
+      "loss": 0.0977,
+      "step": 4171
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.799031953120087e-05,
+      "loss": 0.0946,
+      "step": 4172
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7988315477391604e-05,
+      "loss": 0.0792,
+      "step": 4173
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7986310536598304e-05,
+      "loss": 0.1182,
+      "step": 4174
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.798430470904359e-05,
+      "loss": 0.0981,
+      "step": 4175
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.798229799495018e-05,
+      "loss": 0.0901,
+      "step": 4176
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.798029039454089e-05,
+      "loss": 0.0865,
+      "step": 4177
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7978281908038633e-05,
+      "loss": 0.0737,
+      "step": 4178
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7976272535666424e-05,
+      "loss": 0.0901,
+      "step": 4179
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7974262277647376e-05,
+      "loss": 0.0869,
+      "step": 4180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7972251134204693e-05,
+      "loss": 0.0904,
+      "step": 4181
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.797023910556168e-05,
+      "loss": 0.0946,
+      "step": 4182
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.796822619194176e-05,
+      "loss": 0.0739,
+      "step": 4183
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7966212393568418e-05,
+      "loss": 0.0848,
+      "step": 4184
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7964197710665262e-05,
+      "loss": 0.0969,
+      "step": 4185
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7962182143455997e-05,
+      "loss": 0.0955,
+      "step": 4186
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.796016569216442e-05,
+      "loss": 0.0952,
+      "step": 4187
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.7958148357014424e-05,
+      "loss": 0.0917,
+      "step": 4188
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.795613013823001e-05,
+      "loss": 0.0742,
+      "step": 4189
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.795411103603527e-05,
+      "loss": 0.0983,
+      "step": 4190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7952091050654392e-05,
+      "loss": 0.0906,
+      "step": 4191
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.795007018231167e-05,
+      "loss": 0.0793,
+      "step": 4192
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.794804843123149e-05,
+      "loss": 0.0666,
+      "step": 4193
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7946025797638332e-05,
+      "loss": 0.0939,
+      "step": 4194
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7944002281756784e-05,
+      "loss": 0.0769,
+      "step": 4195
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.794197788381153e-05,
+      "loss": 0.0907,
+      "step": 4196
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.793995260402735e-05,
+      "loss": 0.0895,
+      "step": 4197
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7937926442629116e-05,
+      "loss": 0.0854,
+      "step": 4198
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.79358993998418e-05,
+      "loss": 0.0674,
+      "step": 4199
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7933871475890484e-05,
+      "loss": 0.0629,
+      "step": 4200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.793184267100034e-05,
+      "loss": 0.0883,
+      "step": 4201
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7929812985396625e-05,
+      "loss": 0.1119,
+      "step": 4202
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7927782419304716e-05,
+      "loss": 0.0711,
+      "step": 4203
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7925750972950074e-05,
+      "loss": 0.0963,
+      "step": 4204
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.792371864655826e-05,
+      "loss": 0.0825,
+      "step": 4205
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7921685440354933e-05,
+      "loss": 0.0989,
+      "step": 4206
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.7919651354565857e-05,
+      "loss": 0.0977,
+      "step": 4207
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.791761638941688e-05,
+      "loss": 0.0988,
+      "step": 4208
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7915580545133956e-05,
+      "loss": 0.085,
+      "step": 4209
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7913543821943135e-05,
+      "loss": 0.1012,
+      "step": 4210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7911506220070563e-05,
+      "loss": 0.0853,
+      "step": 4211
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7909467739742494e-05,
+      "loss": 0.0938,
+      "step": 4212
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7907428381185264e-05,
+      "loss": 0.0811,
+      "step": 4213
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7905388144625316e-05,
+      "loss": 0.0903,
+      "step": 4214
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7903347030289184e-05,
+      "loss": 0.097,
+      "step": 4215
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.790130503840351e-05,
+      "loss": 0.078,
+      "step": 4216
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7899262169195024e-05,
+      "loss": 0.0895,
+      "step": 4217
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.789721842289055e-05,
+      "loss": 0.0946,
+      "step": 4218
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7895173799717027e-05,
+      "loss": 0.0881,
+      "step": 4219
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.789312829990147e-05,
+      "loss": 0.078,
+      "step": 4220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.789108192367101e-05,
+      "loss": 0.0848,
+      "step": 4221
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7889034671252866e-05,
+      "loss": 0.1,
+      "step": 4222
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7886986542874348e-05,
+      "loss": 0.0911,
+      "step": 4223
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7884937538762872e-05,
+      "loss": 0.0762,
+      "step": 4224
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7882887659145955e-05,
+      "loss": 0.0753,
+      "step": 4225
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.78808369042512e-05,
+      "loss": 0.0874,
+      "step": 4226
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.787878527430632e-05,
+      "loss": 0.0961,
+      "step": 4227
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.7876732769539108e-05,
+      "loss": 0.0918,
+      "step": 4228
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7874679390177476e-05,
+      "loss": 0.0972,
+      "step": 4229
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7872625136449412e-05,
+      "loss": 0.0907,
+      "step": 4230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.787057000858301e-05,
+      "loss": 0.0811,
+      "step": 4231
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.786851400680647e-05,
+      "loss": 0.0684,
+      "step": 4232
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7866457131348073e-05,
+      "loss": 0.087,
+      "step": 4233
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7864399382436208e-05,
+      "loss": 0.0813,
+      "step": 4234
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7862340760299358e-05,
+      "loss": 0.0839,
+      "step": 4235
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7860281265166097e-05,
+      "loss": 0.1007,
+      "step": 4236
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7858220897265107e-05,
+      "loss": 0.0734,
+      "step": 4237
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.785615965682516e-05,
+      "loss": 0.0837,
+      "step": 4238
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.785409754407513e-05,
+      "loss": 0.0818,
+      "step": 4239
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7852034559243977e-05,
+      "loss": 0.0913,
+      "step": 4240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.784997070256077e-05,
+      "loss": 0.0754,
+      "step": 4241
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7847905974254663e-05,
+      "loss": 0.0711,
+      "step": 4242
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.784584037455492e-05,
+      "loss": 0.0681,
+      "step": 4243
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7843773903690896e-05,
+      "loss": 0.0968,
+      "step": 4244
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7841706561892037e-05,
+      "loss": 0.1046,
+      "step": 4245
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7839638349387892e-05,
+      "loss": 0.058,
+      "step": 4246
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.7837569266408107e-05,
+      "loss": 0.0696,
+      "step": 4247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7835499313182426e-05,
+      "loss": 0.093,
+      "step": 4248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.783342848994068e-05,
+      "loss": 0.0811,
+      "step": 4249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7831356796912805e-05,
+      "loss": 0.0737,
+      "step": 4250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.782928423432883e-05,
+      "loss": 0.0767,
+      "step": 4251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7827210802418894e-05,
+      "loss": 0.1039,
+      "step": 4252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7825136501413206e-05,
+      "loss": 0.0816,
+      "step": 4253
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7823061331542094e-05,
+      "loss": 0.0707,
+      "step": 4254
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7820985293035974e-05,
+      "loss": 0.0811,
+      "step": 4255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7818908386125358e-05,
+      "loss": 0.1057,
+      "step": 4256
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7816830611040855e-05,
+      "loss": 0.1001,
+      "step": 4257
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7814751968013176e-05,
+      "loss": 0.0818,
+      "step": 4258
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7812672457273118e-05,
+      "loss": 0.0829,
+      "step": 4259
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7810592079051586e-05,
+      "loss": 0.097,
+      "step": 4260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7808510833579565e-05,
+      "loss": 0.104,
+      "step": 4261
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7806428721088156e-05,
+      "loss": 0.0787,
+      "step": 4262
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7804345741808543e-05,
+      "loss": 0.088,
+      "step": 4263
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.780226189597201e-05,
+      "loss": 0.0914,
+      "step": 4264
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7800177183809937e-05,
+      "loss": 0.0908,
+      "step": 4265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.77980916055538e-05,
+      "loss": 0.0964,
+      "step": 4266
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.7796005161435173e-05,
+      "loss": 0.078,
+      "step": 4267
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7793917851685725e-05,
+      "loss": 0.0934,
+      "step": 4268
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7791829676537223e-05,
+      "loss": 0.0807,
+      "step": 4269
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.778974063622152e-05,
+      "loss": 0.0828,
+      "step": 4270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7787650730970578e-05,
+      "loss": 0.0724,
+      "step": 4271
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7785559961016452e-05,
+      "loss": 0.0925,
+      "step": 4272
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.778346832659129e-05,
+      "loss": 0.0912,
+      "step": 4273
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7781375827927334e-05,
+      "loss": 0.0753,
+      "step": 4274
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7779282465256928e-05,
+      "loss": 0.0901,
+      "step": 4275
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7777188238812503e-05,
+      "loss": 0.1003,
+      "step": 4276
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7775093148826602e-05,
+      "loss": 0.0802,
+      "step": 4277
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7772997195531846e-05,
+      "loss": 0.1063,
+      "step": 4278
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.777090037916096e-05,
+      "loss": 0.0912,
+      "step": 4279
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7768802699946767e-05,
+      "loss": 0.0986,
+      "step": 4280
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7766704158122184e-05,
+      "loss": 0.0867,
+      "step": 4281
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.776460475392022e-05,
+      "loss": 0.0895,
+      "step": 4282
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7762504487573986e-05,
+      "loss": 0.071,
+      "step": 4283
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.776040335931668e-05,
+      "loss": 0.0875,
+      "step": 4284
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7758301369381604e-05,
+      "loss": 0.1029,
+      "step": 4285
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.7756198518002158e-05,
+      "loss": 0.093,
+      "step": 4286
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7754094805411825e-05,
+      "loss": 0.076,
+      "step": 4287
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7751990231844195e-05,
+      "loss": 0.0767,
+      "step": 4288
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.774988479753295e-05,
+      "loss": 0.074,
+      "step": 4289
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.774777850271186e-05,
+      "loss": 0.0754,
+      "step": 4290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.774567134761481e-05,
+      "loss": 0.0879,
+      "step": 4291
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.774356333247576e-05,
+      "loss": 0.0823,
+      "step": 4292
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7741454457528774e-05,
+      "loss": 0.0988,
+      "step": 4293
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7739344723008017e-05,
+      "loss": 0.076,
+      "step": 4294
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7737234129147737e-05,
+      "loss": 0.082,
+      "step": 4295
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7735122676182288e-05,
+      "loss": 0.0792,
+      "step": 4296
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7733010364346115e-05,
+      "loss": 0.0847,
+      "step": 4297
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7730897193873758e-05,
+      "loss": 0.1166,
+      "step": 4298
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7728783164999855e-05,
+      "loss": 0.0743,
+      "step": 4299
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7726668277959137e-05,
+      "loss": 0.0749,
+      "step": 4300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.772455253298643e-05,
+      "loss": 0.0939,
+      "step": 4301
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7722435930316656e-05,
+      "loss": 0.088,
+      "step": 4302
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.772031847018483e-05,
+      "loss": 0.0811,
+      "step": 4303
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.771820015282607e-05,
+      "loss": 0.0841,
+      "step": 4304
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.7716080978475584e-05,
+      "loss": 0.0847,
+      "step": 4305
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7713960947368666e-05,
+      "loss": 0.102,
+      "step": 4306
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7711840059740724e-05,
+      "loss": 0.083,
+      "step": 4307
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7709718315827246e-05,
+      "loss": 0.0927,
+      "step": 4308
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7707595715863823e-05,
+      "loss": 0.0712,
+      "step": 4309
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7705472260086134e-05,
+      "loss": 0.0985,
+      "step": 4310
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7703347948729965e-05,
+      "loss": 0.0803,
+      "step": 4311
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7701222782031177e-05,
+      "loss": 0.0799,
+      "step": 4312
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7699096760225752e-05,
+      "loss": 0.0724,
+      "step": 4313
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7696969883549742e-05,
+      "loss": 0.078,
+      "step": 4314
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7694842152239312e-05,
+      "loss": 0.0929,
+      "step": 4315
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7692713566530712e-05,
+      "loss": 0.0902,
+      "step": 4316
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7690584126660292e-05,
+      "loss": 0.0768,
+      "step": 4317
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.768845383286449e-05,
+      "loss": 0.0812,
+      "step": 4318
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.768632268537985e-05,
+      "loss": 0.0905,
+      "step": 4319
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7684190684443003e-05,
+      "loss": 0.084,
+      "step": 4320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7682057830290674e-05,
+      "loss": 0.0792,
+      "step": 4321
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.767992412315968e-05,
+      "loss": 0.0918,
+      "step": 4322
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7677789563286948e-05,
+      "loss": 0.0773,
+      "step": 4323
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7675654150909485e-05,
+      "loss": 0.0779,
+      "step": 4324
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.7673517886264392e-05,
+      "loss": 0.0817,
+      "step": 4325
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7671380769588878e-05,
+      "loss": 0.0834,
+      "step": 4326
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.766924280112023e-05,
+      "loss": 0.09,
+      "step": 4327
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7667103981095844e-05,
+      "loss": 0.0903,
+      "step": 4328
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7664964309753202e-05,
+      "loss": 0.0853,
+      "step": 4329
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7662823787329877e-05,
+      "loss": 0.0811,
+      "step": 4330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.766068241406355e-05,
+      "loss": 0.066,
+      "step": 4331
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7658540190191992e-05,
+      "loss": 0.0691,
+      "step": 4332
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7656397115953055e-05,
+      "loss": 0.0664,
+      "step": 4333
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.76542531915847e-05,
+      "loss": 0.0929,
+      "step": 4334
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7652108417324976e-05,
+      "loss": 0.0934,
+      "step": 4335
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7649962793412036e-05,
+      "loss": 0.0891,
+      "step": 4336
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7647816320084113e-05,
+      "loss": 0.071,
+      "step": 4337
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7645668997579544e-05,
+      "loss": 0.0851,
+      "step": 4338
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7643520826136752e-05,
+      "loss": 0.0961,
+      "step": 4339
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7641371805994266e-05,
+      "loss": 0.0953,
+      "step": 4340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.76392219373907e-05,
+      "loss": 0.1008,
+      "step": 4341
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7637071220564765e-05,
+      "loss": 0.1002,
+      "step": 4342
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7634919655755267e-05,
+      "loss": 0.0762,
+      "step": 4343
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.7632767243201104e-05,
+      "loss": 0.101,
+      "step": 4344
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.763061398314127e-05,
+      "loss": 0.0929,
+      "step": 4345
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7628459875814856e-05,
+      "loss": 0.0883,
+      "step": 4346
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7626304921461036e-05,
+      "loss": 0.0779,
+      "step": 4347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7624149120319092e-05,
+      "loss": 0.08,
+      "step": 4348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7621992472628396e-05,
+      "loss": 0.0762,
+      "step": 4349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7619834978628406e-05,
+      "loss": 0.0973,
+      "step": 4350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.761767663855868e-05,
+      "loss": 0.0967,
+      "step": 4351
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7615517452658873e-05,
+      "loss": 0.0836,
+      "step": 4352
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7613357421168728e-05,
+      "loss": 0.0842,
+      "step": 4353
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.761119654432809e-05,
+      "loss": 0.0768,
+      "step": 4354
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7609034822376882e-05,
+      "loss": 0.0796,
+      "step": 4355
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.760687225555514e-05,
+      "loss": 0.0905,
+      "step": 4356
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.760470884410298e-05,
+      "loss": 0.0956,
+      "step": 4357
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.760254458826062e-05,
+      "loss": 0.1034,
+      "step": 4358
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7600379488268366e-05,
+      "loss": 0.0641,
+      "step": 4359
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7598213544366624e-05,
+      "loss": 0.0993,
+      "step": 4360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7596046756795886e-05,
+      "loss": 0.0895,
+      "step": 4361
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.7593879125796746e-05,
+      "loss": 0.0742,
+      "step": 4362
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.759171065160988e-05,
+      "loss": 0.0829,
+      "step": 4363
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.758954133447607e-05,
+      "loss": 0.0826,
+      "step": 4364
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.758737117463619e-05,
+      "loss": 0.0855,
+      "step": 4365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7585200172331197e-05,
+      "loss": 0.0713,
+      "step": 4366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.758302832780215e-05,
+      "loss": 0.0823,
+      "step": 4367
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7580855641290203e-05,
+      "loss": 0.0723,
+      "step": 4368
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7578682113036603e-05,
+      "loss": 0.0776,
+      "step": 4369
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7576507743282683e-05,
+      "loss": 0.0975,
+      "step": 4370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7574332532269873e-05,
+      "loss": 0.0815,
+      "step": 4371
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7572156480239702e-05,
+      "loss": 0.092,
+      "step": 4372
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.756997958743379e-05,
+      "loss": 0.0775,
+      "step": 4373
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7567801854093845e-05,
+      "loss": 0.0674,
+      "step": 4374
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.756562328046167e-05,
+      "loss": 0.0894,
+      "step": 4375
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.756344386677917e-05,
+      "loss": 0.0849,
+      "step": 4376
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.756126361328833e-05,
+      "loss": 0.0883,
+      "step": 4377
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.755908252023124e-05,
+      "loss": 0.1108,
+      "step": 4378
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7556900587850076e-05,
+      "loss": 0.0757,
+      "step": 4379
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.755471781638711e-05,
+      "loss": 0.0717,
+      "step": 4380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7552534206084703e-05,
+      "loss": 0.0761,
+      "step": 4381
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.755034975718531e-05,
+      "loss": 0.0782,
+      "step": 4382
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7548164469931495e-05,
+      "loss": 0.0759,
+      "step": 4383
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.754597834456589e-05,
+      "loss": 0.0807,
+      "step": 4384
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7543791381331236e-05,
+      "loss": 0.0884,
+      "step": 4385
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7541603580470364e-05,
+      "loss": 0.0896,
+      "step": 4386
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.753941494222619e-05,
+      "loss": 0.0781,
+      "step": 4387
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.753722546684174e-05,
+      "loss": 0.0779,
+      "step": 4388
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7535035154560114e-05,
+      "loss": 0.0815,
+      "step": 4389
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7532844005624522e-05,
+      "loss": 0.084,
+      "step": 4390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7530652020278247e-05,
+      "loss": 0.087,
+      "step": 4391
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7528459198764688e-05,
+      "loss": 0.0867,
+      "step": 4392
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.752626554132732e-05,
+      "loss": 0.0695,
+      "step": 4393
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7524071048209715e-05,
+      "loss": 0.1011,
+      "step": 4394
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7521875719655544e-05,
+      "loss": 0.0948,
+      "step": 4395
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.751967955590856e-05,
+      "loss": 0.1002,
+      "step": 4396
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.751748255721262e-05,
+      "loss": 0.0719,
+      "step": 4397
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7515284723811664e-05,
+      "loss": 0.0792,
+      "step": 4398
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.751308605594973e-05,
+      "loss": 0.0897,
+      "step": 4399
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.751088655387094e-05,
+      "loss": 0.0802,
+      "step": 4400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7508686217819532e-05,
+      "loss": 0.0802,
+      "step": 4401
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.750648504803981e-05,
+      "loss": 0.0811,
+      "step": 4402
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7504283044776186e-05,
+      "loss": 0.0885,
+      "step": 4403
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7502080208273152e-05,
+      "loss": 0.0806,
+      "step": 4404
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7499876538775312e-05,
+      "loss": 0.0847,
+      "step": 4405
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.749767203652734e-05,
+      "loss": 0.0846,
+      "step": 4406
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.749546670177402e-05,
+      "loss": 0.0901,
+      "step": 4407
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.749326053476022e-05,
+      "loss": 0.082,
+      "step": 4408
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7491053535730898e-05,
+      "loss": 0.0766,
+      "step": 4409
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7488845704931113e-05,
+      "loss": 0.0865,
+      "step": 4410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7486637042606016e-05,
+      "loss": 0.0712,
+      "step": 4411
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7484427549000836e-05,
+      "loss": 0.0833,
+      "step": 4412
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.748221722436091e-05,
+      "loss": 0.0896,
+      "step": 4413
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.748000606893166e-05,
+      "loss": 0.0963,
+      "step": 4414
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.747779408295861e-05,
+      "loss": 0.0822,
+      "step": 4415
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7475581266687362e-05,
+      "loss": 0.0778,
+      "step": 4416
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.747336762036361e-05,
+      "loss": 0.0718,
+      "step": 4417
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7471153144233158e-05,
+      "loss": 0.0852,
+      "step": 4418
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7468937838541885e-05,
+      "loss": 0.1038,
+      "step": 4419
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7466721703535765e-05,
+      "loss": 0.0934,
+      "step": 4420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7464504739460874e-05,
+      "loss": 0.0966,
+      "step": 4421
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7462286946563366e-05,
+      "loss": 0.0911,
+      "step": 4422
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7460068325089502e-05,
+      "loss": 0.1,
+      "step": 4423
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7457848875285618e-05,
+      "loss": 0.0998,
+      "step": 4424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.745562859739816e-05,
+      "loss": 0.0946,
+      "step": 4425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.745340749167365e-05,
+      "loss": 0.0912,
+      "step": 4426
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7451185558358714e-05,
+      "loss": 0.0886,
+      "step": 4427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.744896279770006e-05,
+      "loss": 0.0995,
+      "step": 4428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7446739209944496e-05,
+      "loss": 0.0972,
+      "step": 4429
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7444514795338917e-05,
+      "loss": 0.0864,
+      "step": 4430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7442289554130307e-05,
+      "loss": 0.0884,
+      "step": 4431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7440063486565755e-05,
+      "loss": 0.0762,
+      "step": 4432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.743783659289243e-05,
+      "loss": 0.0938,
+      "step": 4433
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.743560887335759e-05,
+      "loss": 0.0959,
+      "step": 4434
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7433380328208594e-05,
+      "loss": 0.0837,
+      "step": 4435
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7431150957692896e-05,
+      "loss": 0.0712,
+      "step": 4436
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7428920762058022e-05,
+      "loss": 0.0813,
+      "step": 4437
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.742668974155161e-05,
+      "loss": 0.0702,
+      "step": 4438
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7424457896421376e-05,
+      "loss": 0.087,
+      "step": 4439
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.7422225226915138e-05,
+      "loss": 0.0904,
+      "step": 4440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.74199917332808e-05,
+      "loss": 0.0962,
+      "step": 4441
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.741775741576636e-05,
+      "loss": 0.085,
+      "step": 4442
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7415522274619902e-05,
+      "loss": 0.095,
+      "step": 4443
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.741328631008961e-05,
+      "loss": 0.0983,
+      "step": 4444
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.741104952242375e-05,
+      "loss": 0.0807,
+      "step": 4445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7408811911870685e-05,
+      "loss": 0.0819,
+      "step": 4446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.740657347867887e-05,
+      "loss": 0.0804,
+      "step": 4447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7404334223096852e-05,
+      "loss": 0.0764,
+      "step": 4448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.740209414537326e-05,
+      "loss": 0.0874,
+      "step": 4449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.739985324575683e-05,
+      "loss": 0.0806,
+      "step": 4450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7397611524496375e-05,
+      "loss": 0.0895,
+      "step": 4451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7395368981840804e-05,
+      "loss": 0.1061,
+      "step": 4452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7393125618039124e-05,
+      "loss": 0.0773,
+      "step": 4453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7390881433340424e-05,
+      "loss": 0.0872,
+      "step": 4454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7388636427993886e-05,
+      "loss": 0.0826,
+      "step": 4455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7386390602248787e-05,
+      "loss": 0.094,
+      "step": 4456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.738414395635449e-05,
+      "loss": 0.1085,
+      "step": 4457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7381896490560456e-05,
+      "loss": 0.0875,
+      "step": 4458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.737964820511623e-05,
+      "loss": 0.0969,
+      "step": 4459
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.737739910027145e-05,
+      "loss": 0.086,
+      "step": 4460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7375149176275847e-05,
+      "loss": 0.0992,
+      "step": 4461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7372898433379243e-05,
+      "loss": 0.1115,
+      "step": 4462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7370646871831546e-05,
+      "loss": 0.0723,
+      "step": 4463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7368394491882757e-05,
+      "loss": 0.0928,
+      "step": 4464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7366141293782978e-05,
+      "loss": 0.1105,
+      "step": 4465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7363887277782388e-05,
+      "loss": 0.0787,
+      "step": 4466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.736163244413126e-05,
+      "loss": 0.09,
+      "step": 4467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7359376793079963e-05,
+      "loss": 0.0771,
+      "step": 4468
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7357120324878957e-05,
+      "loss": 0.0885,
+      "step": 4469
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.735486303977878e-05,
+      "loss": 0.0996,
+      "step": 4470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7352604938030074e-05,
+      "loss": 0.0892,
+      "step": 4471
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.735034601988357e-05,
+      "loss": 0.0977,
+      "step": 4472
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.734808628559009e-05,
+      "loss": 0.0902,
+      "step": 4473
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7345825735400538e-05,
+      "loss": 0.0887,
+      "step": 4474
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.734356436956592e-05,
+      "loss": 0.0717,
+      "step": 4475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.734130218833732e-05,
+      "loss": 0.0668,
+      "step": 4476
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7339039191965924e-05,
+      "loss": 0.079,
+      "step": 4477
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7336775380703005e-05,
+      "loss": 0.0919,
+      "step": 4478
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.733451075479992e-05,
+      "loss": 0.0746,
+      "step": 4479
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7332245314508137e-05,
+      "loss": 0.0774,
+      "step": 4480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7329979060079184e-05,
+      "loss": 0.0811,
+      "step": 4481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7327711991764698e-05,
+      "loss": 0.0952,
+      "step": 4482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7325444109816408e-05,
+      "loss": 0.0778,
+      "step": 4483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7323175414486125e-05,
+      "loss": 0.0735,
+      "step": 4484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7320905906025752e-05,
+      "loss": 0.0837,
+      "step": 4485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7318635584687294e-05,
+      "loss": 0.0769,
+      "step": 4486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7316364450722827e-05,
+      "loss": 0.0904,
+      "step": 4487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.731409250438453e-05,
+      "loss": 0.0905,
+      "step": 4488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7311819745924672e-05,
+      "loss": 0.0909,
+      "step": 4489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7309546175595602e-05,
+      "loss": 0.0968,
+      "step": 4490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.730727179364977e-05,
+      "loss": 0.0697,
+      "step": 4491
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7304996600339718e-05,
+      "loss": 0.0879,
+      "step": 4492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.730272059591806e-05,
+      "loss": 0.1001,
+      "step": 4493
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7300443780637527e-05,
+      "loss": 0.0864,
+      "step": 4494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7298166154750914e-05,
+      "loss": 0.0867,
+      "step": 4495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7295887718511123e-05,
+      "loss": 0.0868,
+      "step": 4496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.729360847217114e-05,
+      "loss": 0.0823,
+      "step": 4497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7291328415984038e-05,
+      "loss": 0.0927,
+      "step": 4498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.728904755020299e-05,
+      "loss": 0.0787,
+      "step": 4499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7286765875081243e-05,
+      "loss": 0.0883,
+      "step": 4500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7284483390872156e-05,
+      "loss": 0.0859,
+      "step": 4501
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7282200097829153e-05,
+      "loss": 0.0934,
+      "step": 4502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7279915996205768e-05,
+      "loss": 0.0969,
+      "step": 4503
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.727763108625561e-05,
+      "loss": 0.0928,
+      "step": 4504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7275345368232392e-05,
+      "loss": 0.0717,
+      "step": 4505
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7273058842389906e-05,
+      "loss": 0.0839,
+      "step": 4506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7270771508982035e-05,
+      "loss": 0.0622,
+      "step": 4507
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.726848336826275e-05,
+      "loss": 0.0764,
+      "step": 4508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7266194420486123e-05,
+      "loss": 0.0802,
+      "step": 4509
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7263904665906302e-05,
+      "loss": 0.0953,
+      "step": 4510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7261614104777534e-05,
+      "loss": 0.087,
+      "step": 4511
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.725932273735415e-05,
+      "loss": 0.1057,
+      "step": 4512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7257030563890575e-05,
+      "loss": 0.0847,
+      "step": 4513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.725473758464131e-05,
+      "loss": 0.1118,
+      "step": 4514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7252443799860973e-05,
+      "loss": 0.1024,
+      "step": 4515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7250149209804247e-05,
+      "loss": 0.0945,
+      "step": 4516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.724785381472591e-05,
+      "loss": 0.085,
+      "step": 4517
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7245557614880836e-05,
+      "loss": 0.0842,
+      "step": 4518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7243260610523976e-05,
+      "loss": 0.068,
+      "step": 4519
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7240962801910387e-05,
+      "loss": 0.0856,
+      "step": 4520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7238664189295204e-05,
+      "loss": 0.0776,
+      "step": 4521
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7236364772933653e-05,
+      "loss": 0.0934,
+      "step": 4522
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.723406455308105e-05,
+      "loss": 0.0743,
+      "step": 4523
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7231763529992796e-05,
+      "loss": 0.0771,
+      "step": 4524
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7229461703924398e-05,
+      "loss": 0.0911,
+      "step": 4525
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7227159075131424e-05,
+      "loss": 0.0839,
+      "step": 4526
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7224855643869563e-05,
+      "loss": 0.0791,
+      "step": 4527
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7222551410394564e-05,
+      "loss": 0.08,
+      "step": 4528
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7220246374962283e-05,
+      "loss": 0.1094,
+      "step": 4529
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7217940537828658e-05,
+      "loss": 0.0892,
+      "step": 4530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.721563389924972e-05,
+      "loss": 0.0802,
+      "step": 4531
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.721332645948159e-05,
+      "loss": 0.0773,
+      "step": 4532
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.721101821878047e-05,
+      "loss": 0.074,
+      "step": 4533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.720870917740266e-05,
+      "loss": 0.1014,
+      "step": 4534
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7206399335604536e-05,
+      "loss": 0.0907,
+      "step": 4535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.720408869364258e-05,
+      "loss": 0.0905,
+      "step": 4536
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7201777251773356e-05,
+      "loss": 0.1007,
+      "step": 4537
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7199465010253508e-05,
+      "loss": 0.089,
+      "step": 4538
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7197151969339783e-05,
+      "loss": 0.0798,
+      "step": 4539
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7194838129289006e-05,
+      "loss": 0.0775,
+      "step": 4540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7192523490358095e-05,
+      "loss": 0.0688,
+      "step": 4541
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7190208052804056e-05,
+      "loss": 0.0823,
+      "step": 4542
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.718789181688399e-05,
+      "loss": 0.0808,
+      "step": 4543
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.718557478285507e-05,
+      "loss": 0.0955,
+      "step": 4544
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7183256950974578e-05,
+      "loss": 0.0941,
+      "step": 4545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7180938321499867e-05,
+      "loss": 0.0853,
+      "step": 4546
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7178618894688394e-05,
+      "loss": 0.0751,
+      "step": 4547
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.717629867079769e-05,
+      "loss": 0.0924,
+      "step": 4548
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7173977650085385e-05,
+      "loss": 0.0917,
+      "step": 4549
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7171655832809194e-05,
+      "loss": 0.0795,
+      "step": 4550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.716933321922692e-05,
+      "loss": 0.0914,
+      "step": 4551
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7167009809596457e-05,
+      "loss": 0.0723,
+      "step": 4552
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.716468560417578e-05,
+      "loss": 0.0935,
+      "step": 4553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7162360603222964e-05,
+      "loss": 0.1025,
+      "step": 4554
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.716003480699616e-05,
+      "loss": 0.0931,
+      "step": 4555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7157708215753615e-05,
+      "loss": 0.0829,
+      "step": 4556
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.715538082975367e-05,
+      "loss": 0.0916,
+      "step": 4557
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7153052649254734e-05,
+      "loss": 0.0886,
+      "step": 4558
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7150723674515322e-05,
+      "loss": 0.0858,
+      "step": 4559
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7148393905794036e-05,
+      "loss": 0.0958,
+      "step": 4560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7146063343349556e-05,
+      "loss": 0.076,
+      "step": 4561
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7143731987440664e-05,
+      "loss": 0.106,
+      "step": 4562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7141399838326213e-05,
+      "loss": 0.085,
+      "step": 4563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7139066896265162e-05,
+      "loss": 0.0795,
+      "step": 4564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7136733161516547e-05,
+      "loss": 0.0856,
+      "step": 4565
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7134398634339492e-05,
+      "loss": 0.0988,
+      "step": 4566
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7132063314993213e-05,
+      "loss": 0.0748,
+      "step": 4567
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7129727203737018e-05,
+      "loss": 0.097,
+      "step": 4568
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7127390300830288e-05,
+      "loss": 0.0765,
+      "step": 4569
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.712505260653251e-05,
+      "loss": 0.0837,
+      "step": 4570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7122714121103244e-05,
+      "loss": 0.0948,
+      "step": 4571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.712037484480215e-05,
+      "loss": 0.0792,
+      "step": 4572
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7118034777888957e-05,
+      "loss": 0.0862,
+      "step": 4573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7115693920623516e-05,
+      "loss": 0.0896,
+      "step": 4574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7113352273265727e-05,
+      "loss": 0.0869,
+      "step": 4575
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7111009836075604e-05,
+      "loss": 0.0861,
+      "step": 4576
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7108666609313233e-05,
+      "loss": 0.0861,
+      "step": 4577
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7106322593238802e-05,
+      "loss": 0.0759,
+      "step": 4578
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7103977788112578e-05,
+      "loss": 0.0718,
+      "step": 4579
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.710163219419491e-05,
+      "loss": 0.0839,
+      "step": 4580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.709928581174625e-05,
+      "loss": 0.0931,
+      "step": 4581
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7096938641027124e-05,
+      "loss": 0.0699,
+      "step": 4582
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.709459068229815e-05,
+      "loss": 0.0874,
+      "step": 4583
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7092241935820036e-05,
+      "loss": 0.0785,
+      "step": 4584
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7089892401853577e-05,
+      "loss": 0.0905,
+      "step": 4585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7087542080659654e-05,
+      "loss": 0.0979,
+      "step": 4586
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7085190972499228e-05,
+      "loss": 0.0975,
+      "step": 4587
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7082839077633363e-05,
+      "loss": 0.0975,
+      "step": 4588
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.70804863963232e-05,
+      "loss": 0.0824,
+      "step": 4589
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.707813292882997e-05,
+      "loss": 0.0934,
+      "step": 4590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.707577867541499e-05,
+      "loss": 0.0881,
+      "step": 4591
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.707342363633966e-05,
+      "loss": 0.0854,
+      "step": 4592
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.0808,
+      "step": 4593
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7068711202254024e-05,
+      "loss": 0.0887,
+      "step": 4594
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7066353807766957e-05,
+      "loss": 0.0891,
+      "step": 4595
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.706399562866604e-05,
+      "loss": 0.0903,
+      "step": 4596
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.706163666521311e-05,
+      "loss": 0.1018,
+      "step": 4597
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7059276917670097e-05,
+      "loss": 0.063,
+      "step": 4598
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.705691638629901e-05,
+      "loss": 0.0809,
+      "step": 4599
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7054555071361954e-05,
+      "loss": 0.0829,
+      "step": 4600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7052192973121124e-05,
+      "loss": 0.074,
+      "step": 4601
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7049830091838788e-05,
+      "loss": 0.0868,
+      "step": 4602
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7047466427777313e-05,
+      "loss": 0.0674,
+      "step": 4603
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7045101981199144e-05,
+      "loss": 0.0901,
+      "step": 4604
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7042736752366828e-05,
+      "loss": 0.1047,
+      "step": 4605
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7040370741542978e-05,
+      "loss": 0.0898,
+      "step": 4606
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.703800394899031e-05,
+      "loss": 0.0881,
+      "step": 4607
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7035636374971618e-05,
+      "loss": 0.0875,
+      "step": 4608
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.703326801974979e-05,
+      "loss": 0.0968,
+      "step": 4609
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7030898883587794e-05,
+      "loss": 0.0737,
+      "step": 4610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7028528966748686e-05,
+      "loss": 0.0871,
+      "step": 4611
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7026158269495612e-05,
+      "loss": 0.0867,
+      "step": 4612
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7023786792091805e-05,
+      "loss": 0.0791,
+      "step": 4613
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.702141453480058e-05,
+      "loss": 0.0813,
+      "step": 4614
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.701904149788534e-05,
+      "loss": 0.1016,
+      "step": 4615
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.701666768160958e-05,
+      "loss": 0.0875,
+      "step": 4616
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.701429308623687e-05,
+      "loss": 0.1163,
+      "step": 4617
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.701191771203088e-05,
+      "loss": 0.074,
+      "step": 4618
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.700954155925536e-05,
+      "loss": 0.0906,
+      "step": 4619
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.700716462817414e-05,
+      "loss": 0.1086,
+      "step": 4620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.700478691905115e-05,
+      "loss": 0.0849,
+      "step": 4621
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7002408432150396e-05,
+      "loss": 0.0745,
+      "step": 4622
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7000029167735972e-05,
+      "loss": 0.0826,
+      "step": 4623
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6997649126072064e-05,
+      "loss": 0.0812,
+      "step": 4624
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.699526830742294e-05,
+      "loss": 0.085,
+      "step": 4625
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.699288671205295e-05,
+      "loss": 0.0782,
+      "step": 4626
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6990504340226545e-05,
+      "loss": 0.0868,
+      "step": 4627
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6988121192208237e-05,
+      "loss": 0.0863,
+      "step": 4628
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6985737268262653e-05,
+      "loss": 0.0944,
+      "step": 4629
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6983352568654488e-05,
+      "loss": 0.0944,
+      "step": 4630
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.698096709364852e-05,
+      "loss": 0.0853,
+      "step": 4631
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6978580843509635e-05,
+      "loss": 0.095,
+      "step": 4632
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6976193818502776e-05,
+      "loss": 0.0802,
+      "step": 4633
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6973806018893e-05,
+      "loss": 0.0845,
+      "step": 4634
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6971417444945423e-05,
+      "loss": 0.1094,
+      "step": 4635
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.6969028096925275e-05,
+      "loss": 0.0963,
+      "step": 4636
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6966637975097846e-05,
+      "loss": 0.0854,
+      "step": 4637
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.696424707972853e-05,
+      "loss": 0.0685,
+      "step": 4638
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.69618554110828e-05,
+      "loss": 0.0995,
+      "step": 4639
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6959462969426215e-05,
+      "loss": 0.0941,
+      "step": 4640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6957069755024416e-05,
+      "loss": 0.0933,
+      "step": 4641
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6954675768143138e-05,
+      "loss": 0.0671,
+      "step": 4642
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.69522810090482e-05,
+      "loss": 0.0883,
+      "step": 4643
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6949885478005497e-05,
+      "loss": 0.0933,
+      "step": 4644
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6947489175281027e-05,
+      "loss": 0.0815,
+      "step": 4645
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.694509210114086e-05,
+      "loss": 0.0919,
+      "step": 4646
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.694269425585115e-05,
+      "loss": 0.068,
+      "step": 4647
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6940295639678147e-05,
+      "loss": 0.0851,
+      "step": 4648
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6937896252888183e-05,
+      "loss": 0.0659,
+      "step": 4649
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.693549609574767e-05,
+      "loss": 0.0962,
+      "step": 4650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6933095168523116e-05,
+      "loss": 0.0795,
+      "step": 4651
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.69306934714811e-05,
+      "loss": 0.0933,
+      "step": 4652
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6928291004888306e-05,
+      "loss": 0.0944,
+      "step": 4653
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6925887769011483e-05,
+      "loss": 0.0801,
+      "step": 4654
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6923483764117477e-05,
+      "loss": 0.0729,
+      "step": 4655
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6921078990473216e-05,
+      "loss": 0.0856,
+      "step": 4656
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.691867344834572e-05,
+      "loss": 0.0861,
+      "step": 4657
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6916267138002086e-05,
+      "loss": 0.0831,
+      "step": 4658
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.69138600597095e-05,
+      "loss": 0.0806,
+      "step": 4659
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6911452213735223e-05,
+      "loss": 0.0904,
+      "step": 4660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.690904360034662e-05,
+      "loss": 0.0934,
+      "step": 4661
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6906634219811136e-05,
+      "loss": 0.0852,
+      "step": 4662
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6904224072396288e-05,
+      "loss": 0.0803,
+      "step": 4663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6901813158369686e-05,
+      "loss": 0.1002,
+      "step": 4664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6899401477999034e-05,
+      "loss": 0.0824,
+      "step": 4665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.689698903155211e-05,
+      "loss": 0.1,
+      "step": 4666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.689457581929678e-05,
+      "loss": 0.0873,
+      "step": 4667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6892161841501e-05,
+      "loss": 0.0909,
+      "step": 4668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6889747098432795e-05,
+      "loss": 0.0846,
+      "step": 4669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.68873315903603e-05,
+      "loss": 0.0875,
+      "step": 4670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.688491531755171e-05,
+      "loss": 0.0731,
+      "step": 4671
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6882498280275322e-05,
+      "loss": 0.0813,
+      "step": 4672
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6880080478799512e-05,
+      "loss": 0.0833,
+      "step": 4673
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.687766191339274e-05,
+      "loss": 0.0721,
+      "step": 4674
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6875242584323553e-05,
+      "loss": 0.0693,
+      "step": 4675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6872822491860583e-05,
+      "loss": 0.0833,
+      "step": 4676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.687040163627254e-05,
+      "loss": 0.0861,
+      "step": 4677
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6867980017828228e-05,
+      "loss": 0.0834,
+      "step": 4678
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6865557636796533e-05,
+      "loss": 0.0717,
+      "step": 4679
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.686313449344642e-05,
+      "loss": 0.0804,
+      "step": 4680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6860710588046945e-05,
+      "loss": 0.079,
+      "step": 4681
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6858285920867254e-05,
+      "loss": 0.0861,
+      "step": 4682
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.685586049217656e-05,
+      "loss": 0.0829,
+      "step": 4683
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6853434302244175e-05,
+      "loss": 0.079,
+      "step": 4684
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6851007351339493e-05,
+      "loss": 0.0958,
+      "step": 4685
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6848579639731987e-05,
+      "loss": 0.0869,
+      "step": 4686
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.684615116769122e-05,
+      "loss": 0.0912,
+      "step": 4687
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.684372193548684e-05,
+      "loss": 0.0863,
+      "step": 4688
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6841291943388576e-05,
+      "loss": 0.0767,
+      "step": 4689
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.683886119166624e-05,
+      "loss": 0.0855,
+      "step": 4690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.683642968058974e-05,
+      "loss": 0.0794,
+      "step": 4691
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6833997410429046e-05,
+      "loss": 0.0744,
+      "step": 4692
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6831564381454235e-05,
+      "loss": 0.0813,
+      "step": 4693
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6829130593935454e-05,
+      "loss": 0.0757,
+      "step": 4694
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6826696048142946e-05,
+      "loss": 0.0866,
+      "step": 4695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.682426074434702e-05,
+      "loss": 0.0912,
+      "step": 4696
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.682182468281809e-05,
+      "loss": 0.0623,
+      "step": 4697
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.681938786382664e-05,
+      "loss": 0.0773,
+      "step": 4698
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6816950287643243e-05,
+      "loss": 0.0886,
+      "step": 4699
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.681451195453856e-05,
+      "loss": 0.0952,
+      "step": 4700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6812072864783324e-05,
+      "loss": 0.079,
+      "step": 4701
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6809633018648365e-05,
+      "loss": 0.0738,
+      "step": 4702
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.680719241640459e-05,
+      "loss": 0.0802,
+      "step": 4703
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6804751058322994e-05,
+      "loss": 0.0806,
+      "step": 4704
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.680230894467465e-05,
+      "loss": 0.0917,
+      "step": 4705
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6799866075730724e-05,
+      "loss": 0.0648,
+      "step": 4706
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6797422451762454e-05,
+      "loss": 0.0775,
+      "step": 4707
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6794978073041176e-05,
+      "loss": 0.079,
+      "step": 4708
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.679253293983829e-05,
+      "loss": 0.0798,
+      "step": 4709
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6790087052425303e-05,
+      "loss": 0.087,
+      "step": 4710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.678764041107379e-05,
+      "loss": 0.0831,
+      "step": 4711
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6785193016055415e-05,
+      "loss": 0.0956,
+      "step": 4712
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6782744867641924e-05,
+      "loss": 0.0867,
+      "step": 4713
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.6780295966105148e-05,
+      "loss": 0.0792,
+      "step": 4714
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6777846311717005e-05,
+      "loss": 0.08,
+      "step": 4715
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6775395904749486e-05,
+      "loss": 0.0864,
+      "step": 4716
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6772944745474676e-05,
+      "loss": 0.0841,
+      "step": 4717
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.677049283416474e-05,
+      "loss": 0.0958,
+      "step": 4718
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6768040171091926e-05,
+      "loss": 0.0743,
+      "step": 4719
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.676558675652857e-05,
+      "loss": 0.0773,
+      "step": 4720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6763132590747076e-05,
+      "loss": 0.084,
+      "step": 4721
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6760677674019953e-05,
+      "loss": 0.0931,
+      "step": 4722
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.675822200661978e-05,
+      "loss": 0.0663,
+      "step": 4723
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6755765588819226e-05,
+      "loss": 0.0714,
+      "step": 4724
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6753308420891034e-05,
+      "loss": 0.0895,
+      "step": 4725
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.675085050310804e-05,
+      "loss": 0.0749,
+      "step": 4726
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6748391835743153e-05,
+      "loss": 0.0782,
+      "step": 4727
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.674593241906938e-05,
+      "loss": 0.0973,
+      "step": 4728
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.67434722533598e-05,
+      "loss": 0.0903,
+      "step": 4729
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6741011338887573e-05,
+      "loss": 0.0879,
+      "step": 4730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6738549675925956e-05,
+      "loss": 0.0702,
+      "step": 4731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6736087264748267e-05,
+      "loss": 0.0788,
+      "step": 4732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.6733624105627937e-05,
+      "loss": 0.0711,
+      "step": 4733
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.673116019883845e-05,
+      "loss": 0.0927,
+      "step": 4734
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6728695544653397e-05,
+      "loss": 0.0721,
+      "step": 4735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6726230143346433e-05,
+      "loss": 0.0775,
+      "step": 4736
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6723763995191308e-05,
+      "loss": 0.0834,
+      "step": 4737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6721297100461845e-05,
+      "loss": 0.077,
+      "step": 4738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6718829459431964e-05,
+      "loss": 0.0933,
+      "step": 4739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6716361072375657e-05,
+      "loss": 0.0837,
+      "step": 4740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6713891939567002e-05,
+      "loss": 0.0903,
+      "step": 4741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.671142206128016e-05,
+      "loss": 0.0697,
+      "step": 4742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6708951437789373e-05,
+      "loss": 0.0895,
+      "step": 4743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6706480069368968e-05,
+      "loss": 0.0868,
+      "step": 4744
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6704007956293354e-05,
+      "loss": 0.0858,
+      "step": 4745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6701535098837024e-05,
+      "loss": 0.0905,
+      "step": 4746
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6699061497274548e-05,
+      "loss": 0.0948,
+      "step": 4747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6696587151880586e-05,
+      "loss": 0.0663,
+      "step": 4748
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.669411206292988e-05,
+      "loss": 0.0902,
+      "step": 4749
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6691636230697246e-05,
+      "loss": 0.095,
+      "step": 4750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.668915965545759e-05,
+      "loss": 0.0752,
+      "step": 4751
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6686682337485897e-05,
+      "loss": 0.0707,
+      "step": 4752
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6684204277057246e-05,
+      "loss": 0.0798,
+      "step": 4753
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.668172547444678e-05,
+      "loss": 0.0684,
+      "step": 4754
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6679245929929735e-05,
+      "loss": 0.0993,
+      "step": 4755
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.667676564378143e-05,
+      "loss": 0.0906,
+      "step": 4756
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6674284616277263e-05,
+      "loss": 0.0923,
+      "step": 4757
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.667180284769271e-05,
+      "loss": 0.0886,
+      "step": 4758
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.666932033830334e-05,
+      "loss": 0.1018,
+      "step": 4759
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.66668370883848e-05,
+      "loss": 0.078,
+      "step": 4760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6664353098212817e-05,
+      "loss": 0.0829,
+      "step": 4761
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.66618683680632e-05,
+      "loss": 0.0726,
+      "step": 4762
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6659382898211843e-05,
+      "loss": 0.0908,
+      "step": 4763
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6656896688934717e-05,
+      "loss": 0.0831,
+      "step": 4764
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6654409740507884e-05,
+      "loss": 0.0817,
+      "step": 4765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6651922053207478e-05,
+      "loss": 0.0901,
+      "step": 4766
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6649433627309725e-05,
+      "loss": 0.0903,
+      "step": 4767
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6646944463090922e-05,
+      "loss": 0.1135,
+      "step": 4768
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6644454560827457e-05,
+      "loss": 0.0651,
+      "step": 4769
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6641963920795795e-05,
+      "loss": 0.077,
+      "step": 4770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6639472543272488e-05,
+      "loss": 0.0974,
+      "step": 4771
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6636980428534163e-05,
+      "loss": 0.0803,
+      "step": 4772
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6634487576857534e-05,
+      "loss": 0.0731,
+      "step": 4773
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6631993988519396e-05,
+      "loss": 0.0728,
+      "step": 4774
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6629499663796622e-05,
+      "loss": 0.0889,
+      "step": 4775
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6627004602966176e-05,
+      "loss": 0.0861,
+      "step": 4776
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6624508806305088e-05,
+      "loss": 0.0741,
+      "step": 4777
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6622012274090487e-05,
+      "loss": 0.0945,
+      "step": 4778
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6619515006599573e-05,
+      "loss": 0.0835,
+      "step": 4779
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.661701700410963e-05,
+      "loss": 0.0939,
+      "step": 4780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.661451826689803e-05,
+      "loss": 0.0889,
+      "step": 4781
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6612018795242214e-05,
+      "loss": 0.0775,
+      "step": 4782
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6609518589419708e-05,
+      "loss": 0.0887,
+      "step": 4783
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6607017649708133e-05,
+      "loss": 0.0731,
+      "step": 4784
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6604515976385176e-05,
+      "loss": 0.0732,
+      "step": 4785
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.660201356972861e-05,
+      "loss": 0.0927,
+      "step": 4786
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.659951043001629e-05,
+      "loss": 0.0654,
+      "step": 4787
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6597006557526156e-05,
+      "loss": 0.0839,
+      "step": 4788
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6594501952536225e-05,
+      "loss": 0.0748,
+      "step": 4789
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6591996615324593e-05,
+      "loss": 0.0814,
+      "step": 4790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6589490546169443e-05,
+      "loss": 0.0817,
+      "step": 4791
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6586983745349033e-05,
+      "loss": 0.0994,
+      "step": 4792
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.658447621314171e-05,
+      "loss": 0.0811,
+      "step": 4793
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6581967949825902e-05,
+      "loss": 0.0897,
+      "step": 4794
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6579458955680106e-05,
+      "loss": 0.0774,
+      "step": 4795
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6576949230982918e-05,
+      "loss": 0.0856,
+      "step": 4796
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6574438776012998e-05,
+      "loss": 0.0966,
+      "step": 4797
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6571927591049094e-05,
+      "loss": 0.0869,
+      "step": 4798
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6569415676370044e-05,
+      "loss": 0.0796,
+      "step": 4799
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6566903032254754e-05,
+      "loss": 0.082,
+      "step": 4800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.656438965898221e-05,
+      "loss": 0.0923,
+      "step": 4801
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6561875556831497e-05,
+      "loss": 0.0675,
+      "step": 4802
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6559360726081762e-05,
+      "loss": 0.0784,
+      "step": 4803
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6556845167012238e-05,
+      "loss": 0.0817,
+      "step": 4804
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6554328879902245e-05,
+      "loss": 0.0917,
+      "step": 4805
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6551811865031174e-05,
+      "loss": 0.0911,
+      "step": 4806
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6549294122678507e-05,
+      "loss": 0.0829,
+      "step": 4807
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6546775653123803e-05,
+      "loss": 0.0802,
+      "step": 4808
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6544256456646693e-05,
+      "loss": 0.0806,
+      "step": 4809
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6541736533526903e-05,
+      "loss": 0.0562,
+      "step": 4810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6539215884044235e-05,
+      "loss": 0.0791,
+      "step": 4811
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6536694508478565e-05,
+      "loss": 0.0845,
+      "step": 4812
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6534172407109857e-05,
+      "loss": 0.0817,
+      "step": 4813
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.653164958021815e-05,
+      "loss": 0.0577,
+      "step": 4814
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6529126028083566e-05,
+      "loss": 0.0948,
+      "step": 4815
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6526601750986314e-05,
+      "loss": 0.0808,
+      "step": 4816
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6524076749206674e-05,
+      "loss": 0.0748,
+      "step": 4817
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6521551023025006e-05,
+      "loss": 0.0734,
+      "step": 4818
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.651902457272176e-05,
+      "loss": 0.0945,
+      "step": 4819
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.651649739857746e-05,
+      "loss": 0.1024,
+      "step": 4820
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6513969500872713e-05,
+      "loss": 0.0704,
+      "step": 4821
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.65114408798882e-05,
+      "loss": 0.0583,
+      "step": 4822
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.650891153590469e-05,
+      "loss": 0.089,
+      "step": 4823
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6506381469203025e-05,
+      "loss": 0.0883,
+      "step": 4824
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6503850680064135e-05,
+      "loss": 0.0757,
+      "step": 4825
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.650131916876903e-05,
+      "loss": 0.0801,
+      "step": 4826
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.649878693559879e-05,
+      "loss": 0.0852,
+      "step": 4827
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6496253980834586e-05,
+      "loss": 0.0996,
+      "step": 4828
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6493720304757666e-05,
+      "loss": 0.0826,
+      "step": 4829
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.649118590764935e-05,
+      "loss": 0.0721,
+      "step": 4830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6488650789791054e-05,
+      "loss": 0.0752,
+      "step": 4831
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.648611495146426e-05,
+      "loss": 0.0845,
+      "step": 4832
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.648357839295054e-05,
+      "loss": 0.0776,
+      "step": 4833
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6481041114531535e-05,
+      "loss": 0.0696,
+      "step": 4834
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6478503116488975e-05,
+      "loss": 0.0715,
+      "step": 4835
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.647596439910467e-05,
+      "loss": 0.0745,
+      "step": 4836
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6473424962660503e-05,
+      "loss": 0.0919,
+      "step": 4837
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.647088480743844e-05,
+      "loss": 0.0798,
+      "step": 4838
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6468343933720532e-05,
+      "loss": 0.0767,
+      "step": 4839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6465802341788903e-05,
+      "loss": 0.0824,
+      "step": 4840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.646326003192576e-05,
+      "loss": 0.0807,
+      "step": 4841
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6460717004413383e-05,
+      "loss": 0.0922,
+      "step": 4842
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6458173259534148e-05,
+      "loss": 0.0892,
+      "step": 4843
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6455628797570494e-05,
+      "loss": 0.0881,
+      "step": 4844
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6453083618804944e-05,
+      "loss": 0.0807,
+      "step": 4845
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.645053772352011e-05,
+      "loss": 0.0914,
+      "step": 4846
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6447991111998665e-05,
+      "loss": 0.0807,
+      "step": 4847
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.644544378452338e-05,
+      "loss": 0.0797,
+      "step": 4848
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6442895741377098e-05,
+      "loss": 0.085,
+      "step": 4849
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6440346982842735e-05,
+      "loss": 0.1138,
+      "step": 4850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6437797509203305e-05,
+      "loss": 0.0791,
+      "step": 4851
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6435247320741873e-05,
+      "loss": 0.0847,
+      "step": 4852
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6432696417741615e-05,
+      "loss": 0.071,
+      "step": 4853
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.643014480048576e-05,
+      "loss": 0.1068,
+      "step": 4854
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6427592469257635e-05,
+      "loss": 0.0692,
+      "step": 4855
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6425039424340633e-05,
+      "loss": 0.0726,
+      "step": 4856
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6422485666018235e-05,
+      "loss": 0.085,
+      "step": 4857
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6419931194573998e-05,
+      "loss": 0.0652,
+      "step": 4858
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6417376010291556e-05,
+      "loss": 0.0779,
+      "step": 4859
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6414820113454624e-05,
+      "loss": 0.0717,
+      "step": 4860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6412263504347002e-05,
+      "loss": 0.0824,
+      "step": 4861
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6409706183252555e-05,
+      "loss": 0.0748,
+      "step": 4862
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6407148150455242e-05,
+      "loss": 0.0768,
+      "step": 4863
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6404589406239094e-05,
+      "loss": 0.0928,
+      "step": 4864
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.640202995088822e-05,
+      "loss": 0.0803,
+      "step": 4865
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.639946978468681e-05,
+      "loss": 0.0903,
+      "step": 4866
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.639690890791913e-05,
+      "loss": 0.09,
+      "step": 4867
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.639434732086953e-05,
+      "loss": 0.0715,
+      "step": 4868
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6391785023822436e-05,
+      "loss": 0.0873,
+      "step": 4869
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.638922201706236e-05,
+      "loss": 0.0731,
+      "step": 4870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.638665830087387e-05,
+      "loss": 0.0782,
+      "step": 4871
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6384093875541642e-05,
+      "loss": 0.072,
+      "step": 4872
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6381528741350414e-05,
+      "loss": 0.0762,
+      "step": 4873
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6378962898585005e-05,
+      "loss": 0.0857,
+      "step": 4874
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6376396347530314e-05,
+      "loss": 0.088,
+      "step": 4875
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.637382908847132e-05,
+      "loss": 0.0739,
+      "step": 4876
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6371261121693075e-05,
+      "loss": 0.0917,
+      "step": 4877
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6368692447480716e-05,
+      "loss": 0.0969,
+      "step": 4878
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6366123066119458e-05,
+      "loss": 0.0624,
+      "step": 4879
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.636355297789459e-05,
+      "loss": 0.0841,
+      "step": 4880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6360982183091486e-05,
+      "loss": 0.0972,
+      "step": 4881
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.635841068199559e-05,
+      "loss": 0.0758,
+      "step": 4882
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6355838474892435e-05,
+      "loss": 0.097,
+      "step": 4883
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.635326556206762e-05,
+      "loss": 0.0917,
+      "step": 4884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.635069194380683e-05,
+      "loss": 0.1008,
+      "step": 4885
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.634811762039583e-05,
+      "loss": 0.0851,
+      "step": 4886
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.634554259212046e-05,
+      "loss": 0.077,
+      "step": 4887
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6342966859266637e-05,
+      "loss": 0.0983,
+      "step": 4888
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.634039042212036e-05,
+      "loss": 0.0936,
+      "step": 4889
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6337813280967703e-05,
+      "loss": 0.0883,
+      "step": 4890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.633523543609482e-05,
+      "loss": 0.0997,
+      "step": 4891
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6332656887787937e-05,
+      "loss": 0.1089,
+      "step": 4892
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.633007763633337e-05,
+      "loss": 0.0823,
+      "step": 4893
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6327497682017506e-05,
+      "loss": 0.0952,
+      "step": 4894
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6324917025126816e-05,
+      "loss": 0.0947,
+      "step": 4895
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.632233566594783e-05,
+      "loss": 0.0779,
+      "step": 4896
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.631975360476718e-05,
+      "loss": 0.0748,
+      "step": 4897
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.631717084187156e-05,
+      "loss": 0.0875,
+      "step": 4898
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6314587377547754e-05,
+      "loss": 0.0758,
+      "step": 4899
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.631200321208261e-05,
+      "loss": 0.0785,
+      "step": 4900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6309418345763067e-05,
+      "loss": 0.1007,
+      "step": 4901
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6306832778876135e-05,
+      "loss": 0.0809,
+      "step": 4902
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.63042465117089e-05,
+      "loss": 0.086,
+      "step": 4903
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6301659544548528e-05,
+      "loss": 0.0952,
+      "step": 4904
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.629907187768227e-05,
+      "loss": 0.0919,
+      "step": 4905
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6296483511397442e-05,
+      "loss": 0.0945,
+      "step": 4906
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6293894445981448e-05,
+      "loss": 0.0662,
+      "step": 4907
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.629130468172176e-05,
+      "loss": 0.078,
+      "step": 4908
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6288714218905938e-05,
+      "loss": 0.0826,
+      "step": 4909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.628612305782161e-05,
+      "loss": 0.0979,
+      "step": 4910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.628353119875649e-05,
+      "loss": 0.0698,
+      "step": 4911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6280938641998366e-05,
+      "loss": 0.0994,
+      "step": 4912
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.62783453878351e-05,
+      "loss": 0.0963,
+      "step": 4913
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6275751436554632e-05,
+      "loss": 0.1021,
+      "step": 4914
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6273156788444988e-05,
+      "loss": 0.0693,
+      "step": 4915
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6270561443794262e-05,
+      "loss": 0.09,
+      "step": 4916
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.626796540289063e-05,
+      "loss": 0.071,
+      "step": 4917
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6265368666022343e-05,
+      "loss": 0.0922,
+      "step": 4918
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.626277123347773e-05,
+      "loss": 0.0967,
+      "step": 4919
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6260173105545198e-05,
+      "loss": 0.0787,
+      "step": 4920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6257574282513227e-05,
+      "loss": 0.0962,
+      "step": 4921
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6254974764670382e-05,
+      "loss": 0.09,
+      "step": 4922
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.62523745523053e-05,
+      "loss": 0.0802,
+      "step": 4923
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6249773645706698e-05,
+      "loss": 0.0889,
+      "step": 4924
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.6247172045163366e-05,
+      "loss": 0.0927,
+      "step": 4925
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.624456975096417e-05,
+      "loss": 0.0776,
+      "step": 4926
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.624196676339806e-05,
+      "loss": 0.0933,
+      "step": 4927
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.623936308275406e-05,
+      "loss": 0.0779,
+      "step": 4928
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6236758709321265e-05,
+      "loss": 0.0842,
+      "step": 4929
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.623415364338886e-05,
+      "loss": 0.0878,
+      "step": 4930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6231547885246094e-05,
+      "loss": 0.0776,
+      "step": 4931
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6228941435182294e-05,
+      "loss": 0.0797,
+      "step": 4932
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6226334293486878e-05,
+      "loss": 0.0776,
+      "step": 4933
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6223726460449322e-05,
+      "loss": 0.0696,
+      "step": 4934
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.622111793635919e-05,
+      "loss": 0.0712,
+      "step": 4935
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.621850872150612e-05,
+      "loss": 0.0892,
+      "step": 4936
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6215898816179826e-05,
+      "loss": 0.0835,
+      "step": 4937
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6213288220670097e-05,
+      "loss": 0.0771,
+      "step": 4938
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6210676935266806e-05,
+      "loss": 0.0837,
+      "step": 4939
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6208064960259897e-05,
+      "loss": 0.0984,
+      "step": 4940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.620545229593939e-05,
+      "loss": 0.0934,
+      "step": 4941
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6202838942595378e-05,
+      "loss": 0.0801,
+      "step": 4942
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.620022490051804e-05,
+      "loss": 0.0823,
+      "step": 4943
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6197610169997625e-05,
+      "loss": 0.0701,
+      "step": 4944
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6194994751324462e-05,
+      "loss": 0.0649,
+      "step": 4945
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6192378644788955e-05,
+      "loss": 0.0759,
+      "step": 4946
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6189761850681578e-05,
+      "loss": 0.0939,
+      "step": 4947
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6187144369292894e-05,
+      "loss": 0.0746,
+      "step": 4948
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6184526200913533e-05,
+      "loss": 0.0815,
+      "step": 4949
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6181907345834202e-05,
+      "loss": 0.0759,
+      "step": 4950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6179287804345692e-05,
+      "loss": 0.0717,
+      "step": 4951
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6176667576738852e-05,
+      "loss": 0.0792,
+      "step": 4952
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6174046663304634e-05,
+      "loss": 0.093,
+      "step": 4953
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.617142506433404e-05,
+      "loss": 0.0912,
+      "step": 4954
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.616880278011817e-05,
+      "loss": 0.0837,
+      "step": 4955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6166179810948182e-05,
+      "loss": 0.0859,
+      "step": 4956
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6163556157115317e-05,
+      "loss": 0.084,
+      "step": 4957
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6160931818910902e-05,
+      "loss": 0.0665,
+      "step": 4958
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.615830679662632e-05,
+      "loss": 0.0813,
+      "step": 4959
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.615568109055305e-05,
+      "loss": 0.0734,
+      "step": 4960
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6153054700982628e-05,
+      "loss": 0.0734,
+      "step": 4961
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6150427628206686e-05,
+      "loss": 0.0748,
+      "step": 4962
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6147799872516915e-05,
+      "loss": 0.0916,
+      "step": 4963
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.614517143420509e-05,
+      "loss": 0.0823,
+      "step": 4964
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.614254231356306e-05,
+      "loss": 0.079,
+      "step": 4965
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6139912510882752e-05,
+      "loss": 0.0875,
+      "step": 4966
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6137282026456162e-05,
+      "loss": 0.0883,
+      "step": 4967
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.613465086057537e-05,
+      "loss": 0.069,
+      "step": 4968
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6132019013532526e-05,
+      "loss": 0.0853,
+      "step": 4969
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.612938648561986e-05,
+      "loss": 0.0726,
+      "step": 4970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6126753277129672e-05,
+      "loss": 0.0753,
+      "step": 4971
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6124119388354343e-05,
+      "loss": 0.091,
+      "step": 4972
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6121484819586326e-05,
+      "loss": 0.0729,
+      "step": 4973
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6118849571118154e-05,
+      "loss": 0.0745,
+      "step": 4974
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.611621364324243e-05,
+      "loss": 0.0801,
+      "step": 4975
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.611357703625183e-05,
+      "loss": 0.0764,
+      "step": 4976
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6110939750439118e-05,
+      "loss": 0.0873,
+      "step": 4977
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.610830178609712e-05,
+      "loss": 0.0839,
+      "step": 4978
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6105663143518748e-05,
+      "loss": 0.0742,
+      "step": 4979
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6103023822996982e-05,
+      "loss": 0.077,
+      "step": 4980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6100383824824874e-05,
+      "loss": 0.0863,
+      "step": 4981
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6097743149295565e-05,
+      "loss": 0.0737,
+      "step": 4982
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6095101796702257e-05,
+      "loss": 0.0923,
+      "step": 4983
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6092459767338236e-05,
+      "loss": 0.0823,
+      "step": 4984
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.608981706149686e-05,
+      "loss": 0.079,
+      "step": 4985
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6087173679471565e-05,
+      "loss": 0.078,
+      "step": 4986
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.608452962155585e-05,
+      "loss": 0.0944,
+      "step": 4987
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6081884888043307e-05,
+      "loss": 0.0823,
+      "step": 4988
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6079239479227596e-05,
+      "loss": 0.0963,
+      "step": 4989
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.607659339540244e-05,
+      "loss": 0.1083,
+      "step": 4990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.607394663686166e-05,
+      "loss": 0.0709,
+      "step": 4991
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6071299203899127e-05,
+      "loss": 0.0693,
+      "step": 4992
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.606865109680881e-05,
+      "loss": 0.0989,
+      "step": 4993
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6066002315884733e-05,
+      "loss": 0.0711,
+      "step": 4994
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.606335286142101e-05,
+      "loss": 0.1123,
+      "step": 4995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6060702733711823e-05,
+      "loss": 0.0889,
+      "step": 4996
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6058051933051427e-05,
+      "loss": 0.0918,
+      "step": 4997
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6055400459734158e-05,
+      "loss": 0.0749,
+      "step": 4998
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.605274831405442e-05,
+      "loss": 0.0892,
+      "step": 4999
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.605009549630669e-05,
+      "loss": 0.0699,
+      "step": 5000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6047442006785533e-05,
+      "loss": 0.066,
+      "step": 5001
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6044787845785576e-05,
+      "loss": 0.0688,
+      "step": 5002
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6042133013601523e-05,
+      "loss": 0.0883,
+      "step": 5003
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6039477510528155e-05,
+      "loss": 0.0883,
+      "step": 5004
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6036821336860324e-05,
+      "loss": 0.0738,
+      "step": 5005
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.603416449289296e-05,
+      "loss": 0.0924,
+      "step": 5006
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6031506978921066e-05,
+      "loss": 0.0697,
+      "step": 5007
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6028848795239725e-05,
+      "loss": 0.0786,
+      "step": 5008
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.602618994214408e-05,
+      "loss": 0.0734,
+      "step": 5009
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6023530419929362e-05,
+      "loss": 0.0929,
+      "step": 5010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6020870228890873e-05,
+      "loss": 0.0885,
+      "step": 5011
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6018209369323983e-05,
+      "loss": 0.09,
+      "step": 5012
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6015547841524144e-05,
+      "loss": 0.0736,
+      "step": 5013
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6012885645786877e-05,
+      "loss": 0.0873,
+      "step": 5014
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6010222782407784e-05,
+      "loss": 0.0879,
+      "step": 5015
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6007559251682532e-05,
+      "loss": 0.0472,
+      "step": 5016
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.600489505390687e-05,
+      "loss": 0.0811,
+      "step": 5017
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6002230189376614e-05,
+      "loss": 0.0865,
+      "step": 5018
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.599956465838766e-05,
+      "loss": 0.0922,
+      "step": 5019
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5996898461235976e-05,
+      "loss": 0.0674,
+      "step": 5020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5994231598217607e-05,
+      "loss": 0.0794,
+      "step": 5021
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.599156406962866e-05,
+      "loss": 0.0965,
+      "step": 5022
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.598889587576533e-05,
+      "loss": 0.0754,
+      "step": 5023
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.598622701692388e-05,
+      "loss": 0.0666,
+      "step": 5024
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5983557493400645e-05,
+      "loss": 0.0803,
+      "step": 5025
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.598088730549204e-05,
+      "loss": 0.0851,
+      "step": 5026
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5978216453494544e-05,
+      "loss": 0.0822,
+      "step": 5027
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5975544937704724e-05,
+      "loss": 0.0833,
+      "step": 5028
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5972872758419207e-05,
+      "loss": 0.1071,
+      "step": 5029
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5970199915934696e-05,
+      "loss": 0.0857,
+      "step": 5030
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.596752641054798e-05,
+      "loss": 0.0638,
+      "step": 5031
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.59648522425559e-05,
+      "loss": 0.0733,
+      "step": 5032
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5962177412255392e-05,
+      "loss": 0.0693,
+      "step": 5033
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.595950191994345e-05,
+      "loss": 0.0689,
+      "step": 5034
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.595682576591715e-05,
+      "loss": 0.0751,
+      "step": 5035
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5954148950473642e-05,
+      "loss": 0.0858,
+      "step": 5036
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5951471473910145e-05,
+      "loss": 0.0821,
+      "step": 5037
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5948793336523953e-05,
+      "loss": 0.0824,
+      "step": 5038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.594611453861243e-05,
+      "loss": 0.0892,
+      "step": 5039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5943435080473025e-05,
+      "loss": 0.073,
+      "step": 5040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5940754962403242e-05,
+      "loss": 0.0714,
+      "step": 5041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5938074184700675e-05,
+      "loss": 0.0765,
+      "step": 5042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5935392747662987e-05,
+      "loss": 0.0735,
+      "step": 5043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5932710651587902e-05,
+      "loss": 0.082,
+      "step": 5044
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5930027896773235e-05,
+      "loss": 0.0917,
+      "step": 5045
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5927344483516865e-05,
+      "loss": 0.0803,
+      "step": 5046
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5924660412116746e-05,
+      "loss": 0.1011,
+      "step": 5047
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.59219756828709e-05,
+      "loss": 0.1032,
+      "step": 5048
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.591929029607743e-05,
+      "loss": 0.0938,
+      "step": 5049
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5916604252034508e-05,
+      "loss": 0.0839,
+      "step": 5050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5913917551040377e-05,
+      "loss": 0.0717,
+      "step": 5051
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.591123019339336e-05,
+      "loss": 0.0722,
+      "step": 5052
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5908542179391844e-05,
+      "loss": 0.0707,
+      "step": 5053
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5905853509334295e-05,
+      "loss": 0.0754,
+      "step": 5054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5903164183519248e-05,
+      "loss": 0.0914,
+      "step": 5055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5900474202245315e-05,
+      "loss": 0.1021,
+      "step": 5056
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.589778356581118e-05,
+      "loss": 0.0679,
+      "step": 5057
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5895092274515597e-05,
+      "loss": 0.1006,
+      "step": 5058
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.589240032865739e-05,
+      "loss": 0.067,
+      "step": 5059
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.588970772853546e-05,
+      "loss": 0.0839,
+      "step": 5060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.588701447444879e-05,
+      "loss": 0.0804,
+      "step": 5061
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.588432056669641e-05,
+      "loss": 0.0928,
+      "step": 5062
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5881626005577456e-05,
+      "loss": 0.0958,
+      "step": 5063
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5878930791391106e-05,
+      "loss": 0.0704,
+      "step": 5064
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5876234924436633e-05,
+      "loss": 0.0857,
+      "step": 5065
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5873538405013368e-05,
+      "loss": 0.0908,
+      "step": 5066
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5870841233420718e-05,
+      "loss": 0.0807,
+      "step": 5067
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5868143409958162e-05,
+      "loss": 0.0637,
+      "step": 5068
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5865444934925263e-05,
+      "loss": 0.062,
+      "step": 5069
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.586274580862164e-05,
+      "loss": 0.0886,
+      "step": 5070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5860046031346988e-05,
+      "loss": 0.0847,
+      "step": 5071
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5857345603401082e-05,
+      "loss": 0.073,
+      "step": 5072
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5854644525083767e-05,
+      "loss": 0.0693,
+      "step": 5073
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.585194279669495e-05,
+      "loss": 0.0795,
+      "step": 5074
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5849240418534622e-05,
+      "loss": 0.0907,
+      "step": 5075
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5846537390902845e-05,
+      "loss": 0.0739,
+      "step": 5076
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5843833714099748e-05,
+      "loss": 0.0851,
+      "step": 5077
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.584112938842553e-05,
+      "loss": 0.0765,
+      "step": 5078
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5838424414180473e-05,
+      "loss": 0.0737,
+      "step": 5079
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.583571879166492e-05,
+      "loss": 0.0793,
+      "step": 5080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.5833012521179293e-05,
+      "loss": 0.0996,
+      "step": 5081
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.583030560302408e-05,
+      "loss": 0.0906,
+      "step": 5082
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.582759803749984e-05,
+      "loss": 0.0823,
+      "step": 5083
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5824889824907225e-05,
+      "loss": 0.1115,
+      "step": 5084
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5822180965546925e-05,
+      "loss": 0.0809,
+      "step": 5085
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5819471459719723e-05,
+      "loss": 0.0996,
+      "step": 5086
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5816761307726474e-05,
+      "loss": 0.0865,
+      "step": 5087
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5814050509868093e-05,
+      "loss": 0.098,
+      "step": 5088
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.581133906644558e-05,
+      "loss": 0.0864,
+      "step": 5089
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5808626977759998e-05,
+      "loss": 0.0787,
+      "step": 5090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5805914244112485e-05,
+      "loss": 0.0972,
+      "step": 5091
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5803200865804253e-05,
+      "loss": 0.0865,
+      "step": 5092
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5800486843136576e-05,
+      "loss": 0.0857,
+      "step": 5093
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.579777217641081e-05,
+      "loss": 0.0875,
+      "step": 5094
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5795056865928376e-05,
+      "loss": 0.0927,
+      "step": 5095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.579234091199077e-05,
+      "loss": 0.094,
+      "step": 5096
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.578962431489956e-05,
+      "loss": 0.0714,
+      "step": 5097
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5786907074956384e-05,
+      "loss": 0.0928,
+      "step": 5098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5784189192462952e-05,
+      "loss": 0.0867,
+      "step": 5099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.578147066772104e-05,
+      "loss": 0.0972,
+      "step": 5100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5778751501032502e-05,
+      "loss": 0.0867,
+      "step": 5101
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.577603169269926e-05,
+      "loss": 0.0912,
+      "step": 5102
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.5773311243023314e-05,
+      "loss": 0.0721,
+      "step": 5103
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5770590152306723e-05,
+      "loss": 0.0824,
+      "step": 5104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5767868420851628e-05,
+      "loss": 0.0747,
+      "step": 5105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5765146048960234e-05,
+      "loss": 0.0841,
+      "step": 5106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.576242303693482e-05,
+      "loss": 0.1128,
+      "step": 5107
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5759699385077744e-05,
+      "loss": 0.0684,
+      "step": 5108
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5756975093691415e-05,
+      "loss": 0.076,
+      "step": 5109
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.575425016307833e-05,
+      "loss": 0.0875,
+      "step": 5110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5751524593541057e-05,
+      "loss": 0.0805,
+      "step": 5111
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5748798385382226e-05,
+      "loss": 0.0792,
+      "step": 5112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5746071538904542e-05,
+      "loss": 0.0917,
+      "step": 5113
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5743344054410777e-05,
+      "loss": 0.0811,
+      "step": 5114
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5740615932203788e-05,
+      "loss": 0.0723,
+      "step": 5115
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.573788717258648e-05,
+      "loss": 0.095,
+      "step": 5116
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.573515777586185e-05,
+      "loss": 0.0925,
+      "step": 5117
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5732427742332955e-05,
+      "loss": 0.0737,
+      "step": 5118
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5729697072302927e-05,
+      "loss": 0.0772,
+      "step": 5119
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.5726965766074962e-05,
+      "loss": 0.087,
+      "step": 5120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.572423382395233e-05,
+      "loss": 0.0707,
+      "step": 5121
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.572150124623838e-05,
+      "loss": 0.0679,
+      "step": 5122
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5718768033236518e-05,
+      "loss": 0.0781,
+      "step": 5123
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.571603418525023e-05,
+      "loss": 0.0831,
+      "step": 5124
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5713299702583067e-05,
+      "loss": 0.0984,
+      "step": 5125
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5710564585538653e-05,
+      "loss": 0.0737,
+      "step": 5126
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5707828834420683e-05,
+      "loss": 0.0819,
+      "step": 5127
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5705092449532922e-05,
+      "loss": 0.0729,
+      "step": 5128
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5702355431179202e-05,
+      "loss": 0.0748,
+      "step": 5129
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5699617779663438e-05,
+      "loss": 0.085,
+      "step": 5130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.569687949528959e-05,
+      "loss": 0.0873,
+      "step": 5131
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5694140578361722e-05,
+      "loss": 0.0906,
+      "step": 5132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5691401029183934e-05,
+      "loss": 0.0624,
+      "step": 5133
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.568866084806042e-05,
+      "loss": 0.084,
+      "step": 5134
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5685920035295436e-05,
+      "loss": 0.0886,
+      "step": 5135
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5683178591193306e-05,
+      "loss": 0.0674,
+      "step": 5136
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.568043651605843e-05,
+      "loss": 0.087,
+      "step": 5137
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5677693810195272e-05,
+      "loss": 0.075,
+      "step": 5138
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5674950473908373e-05,
+      "loss": 0.0959,
+      "step": 5139
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.5672206507502337e-05,
+      "loss": 0.0728,
+      "step": 5140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.566946191128184e-05,
+      "loss": 0.0825,
+      "step": 5141
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.566671668555163e-05,
+      "loss": 0.0601,
+      "step": 5142
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5663970830616523e-05,
+      "loss": 0.0898,
+      "step": 5143
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.566122434678141e-05,
+      "loss": 0.0795,
+      "step": 5144
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5658477234351234e-05,
+      "loss": 0.0884,
+      "step": 5145
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5655729493631038e-05,
+      "loss": 0.0771,
+      "step": 5146
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5652981124925907e-05,
+      "loss": 0.0742,
+      "step": 5147
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.565023212854101e-05,
+      "loss": 0.0991,
+      "step": 5148
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.564748250478158e-05,
+      "loss": 0.0775,
+      "step": 5149
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.564473225395293e-05,
+      "loss": 0.0747,
+      "step": 5150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5641981376360423e-05,
+      "loss": 0.0764,
+      "step": 5151
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5639229872309512e-05,
+      "loss": 0.0734,
+      "step": 5152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5636477742105706e-05,
+      "loss": 0.0769,
+      "step": 5153
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.563372498605459e-05,
+      "loss": 0.0728,
+      "step": 5154
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5630971604461817e-05,
+      "loss": 0.0797,
+      "step": 5155
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5628217597633112e-05,
+      "loss": 0.0714,
+      "step": 5156
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.562546296587426e-05,
+      "loss": 0.0968,
+      "step": 5157
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5622707709491125e-05,
+      "loss": 0.0774,
+      "step": 5158
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5619951828789635e-05,
+      "loss": 0.085,
+      "step": 5159
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.56171953240758e-05,
+      "loss": 0.0815,
+      "step": 5160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5614438195655678e-05,
+      "loss": 0.076,
+      "step": 5161
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.561168044383541e-05,
+      "loss": 0.0812,
+      "step": 5162
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5608922068921203e-05,
+      "loss": 0.0692,
+      "step": 5163
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.560616307121934e-05,
+      "loss": 0.0704,
+      "step": 5164
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5603403451036156e-05,
+      "loss": 0.0851,
+      "step": 5165
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5600643208678075e-05,
+      "loss": 0.0697,
+      "step": 5166
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5597882344451575e-05,
+      "loss": 0.0868,
+      "step": 5167
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5595120858663215e-05,
+      "loss": 0.0851,
+      "step": 5168
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.559235875161961e-05,
+      "loss": 0.0707,
+      "step": 5169
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.558959602362746e-05,
+      "loss": 0.0925,
+      "step": 5170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5586832674993514e-05,
+      "loss": 0.074,
+      "step": 5171
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5584068706024612e-05,
+      "loss": 0.1108,
+      "step": 5172
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.558130411702764e-05,
+      "loss": 0.0721,
+      "step": 5173
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5578538908309578e-05,
+      "loss": 0.0782,
+      "step": 5174
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.557577308017745e-05,
+      "loss": 0.0804,
+      "step": 5175
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5573006632938364e-05,
+      "loss": 0.0775,
+      "step": 5176
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.55702395668995e-05,
+      "loss": 0.0897,
+      "step": 5177
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5567471882368086e-05,
+      "loss": 0.0815,
+      "step": 5178
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5564703579651444e-05,
+      "loss": 0.0767,
+      "step": 5179
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5561934659056947e-05,
+      "loss": 0.0746,
+      "step": 5180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5559165120892048e-05,
+      "loss": 0.0721,
+      "step": 5181
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5556394965464256e-05,
+      "loss": 0.0811,
+      "step": 5182
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.555362419308116e-05,
+      "loss": 0.0728,
+      "step": 5183
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5550852804050412e-05,
+      "loss": 0.0894,
+      "step": 5184
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5548080798679732e-05,
+      "loss": 0.0822,
+      "step": 5185
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5545308177276915e-05,
+      "loss": 0.0697,
+      "step": 5186
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5542534940149816e-05,
+      "loss": 0.0703,
+      "step": 5187
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5539761087606364e-05,
+      "loss": 0.0791,
+      "step": 5188
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5536986619954553e-05,
+      "loss": 0.089,
+      "step": 5189
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5534211537502444e-05,
+      "loss": 0.0807,
+      "step": 5190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.553143584055817e-05,
+      "loss": 0.0672,
+      "step": 5191
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5528659529429935e-05,
+      "loss": 0.0905,
+      "step": 5192
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5525882604426005e-05,
+      "loss": 0.0814,
+      "step": 5193
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5523105065854712e-05,
+      "loss": 0.0962,
+      "step": 5194
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5520326914024467e-05,
+      "loss": 0.0672,
+      "step": 5195
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.551754814924374e-05,
+      "loss": 0.0802,
+      "step": 5196
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.551476877182107e-05,
+      "loss": 0.077,
+      "step": 5197
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5511988782065067e-05,
+      "loss": 0.0807,
+      "step": 5198
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5509208180284406e-05,
+      "loss": 0.0868,
+      "step": 5199
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.5506426966787836e-05,
+      "loss": 0.079,
+      "step": 5200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5503645141884166e-05,
+      "loss": 0.0769,
+      "step": 5201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5500862705882278e-05,
+      "loss": 0.0879,
+      "step": 5202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5498079659091118e-05,
+      "loss": 0.0834,
+      "step": 5203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5495296001819704e-05,
+      "loss": 0.0803,
+      "step": 5204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.549251173437712e-05,
+      "loss": 0.0854,
+      "step": 5205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5489726857072517e-05,
+      "loss": 0.0773,
+      "step": 5206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5486941370215114e-05,
+      "loss": 0.0783,
+      "step": 5207
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5484155274114197e-05,
+      "loss": 0.0859,
+      "step": 5208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5481368569079126e-05,
+      "loss": 0.0961,
+      "step": 5209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.547858125541932e-05,
+      "loss": 0.1021,
+      "step": 5210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5475793333444263e-05,
+      "loss": 0.0651,
+      "step": 5211
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.547300480346352e-05,
+      "loss": 0.0718,
+      "step": 5212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5470215665786715e-05,
+      "loss": 0.0741,
+      "step": 5213
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.546742592072354e-05,
+      "loss": 0.0799,
+      "step": 5214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.546463556858375e-05,
+      "loss": 0.0772,
+      "step": 5215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.546184460967718e-05,
+      "loss": 0.0789,
+      "step": 5216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5459053044313722e-05,
+      "loss": 0.0889,
+      "step": 5217
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.5456260872803332e-05,
+      "loss": 0.0818,
+      "step": 5218
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.545346809545605e-05,
+      "loss": 0.09,
+      "step": 5219
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.545067471258196e-05,
+      "loss": 0.0721,
+      "step": 5220
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5447880724491234e-05,
+      "loss": 0.0885,
+      "step": 5221
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5445086131494103e-05,
+      "loss": 0.0747,
+      "step": 5222
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5442290933900864e-05,
+      "loss": 0.0809,
+      "step": 5223
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.543949513202188e-05,
+      "loss": 0.0887,
+      "step": 5224
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5436698726167585e-05,
+      "loss": 0.0895,
+      "step": 5225
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.543390171664848e-05,
+      "loss": 0.0846,
+      "step": 5226
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.543110410377513e-05,
+      "loss": 0.0908,
+      "step": 5227
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5428305887858166e-05,
+      "loss": 0.0815,
+      "step": 5228
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.542550706920829e-05,
+      "loss": 0.0886,
+      "step": 5229
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5422707648136275e-05,
+      "loss": 0.067,
+      "step": 5230
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5419907624952947e-05,
+      "loss": 0.0875,
+      "step": 5231
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.541710699996921e-05,
+      "loss": 0.0898,
+      "step": 5232
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5414305773496033e-05,
+      "loss": 0.0667,
+      "step": 5233
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.541150394584445e-05,
+      "loss": 0.0762,
+      "step": 5234
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5408701517325565e-05,
+      "loss": 0.0754,
+      "step": 5235
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.540589848825054e-05,
+      "loss": 0.0813,
+      "step": 5236
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.540309485893061e-05,
+      "loss": 0.067,
+      "step": 5237
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.5400290629677085e-05,
+      "loss": 0.0836,
+      "step": 5238
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.539748580080133e-05,
+      "loss": 0.0875,
+      "step": 5239
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5394680372614773e-05,
+      "loss": 0.0709,
+      "step": 5240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5391874345428923e-05,
+      "loss": 0.082,
+      "step": 5241
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.538906771955534e-05,
+      "loss": 0.0721,
+      "step": 5242
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5386260495305666e-05,
+      "loss": 0.0822,
+      "step": 5243
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5383452672991598e-05,
+      "loss": 0.0737,
+      "step": 5244
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5380644252924902e-05,
+      "loss": 0.0768,
+      "step": 5245
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5377835235417418e-05,
+      "loss": 0.0815,
+      "step": 5246
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5375025620781034e-05,
+      "loss": 0.0911,
+      "step": 5247
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5372215409327726e-05,
+      "loss": 0.087,
+      "step": 5248
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5369404601369523e-05,
+      "loss": 0.0717,
+      "step": 5249
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5366593197218524e-05,
+      "loss": 0.069,
+      "step": 5250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5363781197186893e-05,
+      "loss": 0.0748,
+      "step": 5251
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5360968601586858e-05,
+      "loss": 0.0721,
+      "step": 5252
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5358155410730723e-05,
+      "loss": 0.0787,
+      "step": 5253
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5355341624930846e-05,
+      "loss": 0.0848,
+      "step": 5254
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5352527244499658e-05,
+      "loss": 0.0743,
+      "step": 5255
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5349712269749656e-05,
+      "loss": 0.0812,
+      "step": 5256
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.5346896700993398e-05,
+      "loss": 0.1182,
+      "step": 5257
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.534408053854351e-05,
+      "loss": 0.0823,
+      "step": 5258
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.534126378271269e-05,
+      "loss": 0.078,
+      "step": 5259
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5338446433813695e-05,
+      "loss": 0.0801,
+      "step": 5260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5335628492159345e-05,
+      "loss": 0.074,
+      "step": 5261
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.533280995806254e-05,
+      "loss": 0.0637,
+      "step": 5262
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5329990831836227e-05,
+      "loss": 0.0862,
+      "step": 5263
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5327171113793437e-05,
+      "loss": 0.0708,
+      "step": 5264
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5324350804247248e-05,
+      "loss": 0.0747,
+      "step": 5265
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5321529903510822e-05,
+      "loss": 0.0747,
+      "step": 5266
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5318708411897375e-05,
+      "loss": 0.0897,
+      "step": 5267
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.531588632972019e-05,
+      "loss": 0.0746,
+      "step": 5268
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5313063657292623e-05,
+      "loss": 0.0833,
+      "step": 5269
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5310240394928086e-05,
+      "loss": 0.1,
+      "step": 5270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5307416542940058e-05,
+      "loss": 0.0929,
+      "step": 5271
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5304592101642087e-05,
+      "loss": 0.0771,
+      "step": 5272
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5301767071347794e-05,
+      "loss": 0.0803,
+      "step": 5273
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5298941452370846e-05,
+      "loss": 0.0813,
+      "step": 5274
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.529611524502499e-05,
+      "loss": 0.0833,
+      "step": 5275
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5293288449624033e-05,
+      "loss": 0.0878,
+      "step": 5276
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.5290461066481853e-05,
+      "loss": 0.075,
+      "step": 5277
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5287633095912383e-05,
+      "loss": 0.0817,
+      "step": 5278
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.528480453822963e-05,
+      "loss": 0.0571,
+      "step": 5279
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.528197539374767e-05,
+      "loss": 0.089,
+      "step": 5280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5279145662780622e-05,
+      "loss": 0.0734,
+      "step": 5281
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5276315345642703e-05,
+      "loss": 0.0806,
+      "step": 5282
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5273484442648165e-05,
+      "loss": 0.0759,
+      "step": 5283
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5270652954111344e-05,
+      "loss": 0.0736,
+      "step": 5284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5267820880346632e-05,
+      "loss": 0.0645,
+      "step": 5285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.526498822166849e-05,
+      "loss": 0.0977,
+      "step": 5286
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5262154978391442e-05,
+      "loss": 0.0806,
+      "step": 5287
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5259321150830082e-05,
+      "loss": 0.0677,
+      "step": 5288
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5256486739299061e-05,
+      "loss": 0.0835,
+      "step": 5289
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5253651744113096e-05,
+      "loss": 0.0775,
+      "step": 5290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5250816165586978e-05,
+      "loss": 0.0715,
+      "step": 5291
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5247980004035547e-05,
+      "loss": 0.087,
+      "step": 5292
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5245143259773723e-05,
+      "loss": 0.092,
+      "step": 5293
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.524230593311648e-05,
+      "loss": 0.0696,
+      "step": 5294
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5239468024378868e-05,
+      "loss": 0.0736,
+      "step": 5295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.523662953387599e-05,
+      "loss": 0.0721,
+      "step": 5296
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.523379046192302e-05,
+      "loss": 0.0693,
+      "step": 5297
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5230950808835187e-05,
+      "loss": 0.0898,
+      "step": 5298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5228110574927806e-05,
+      "loss": 0.0724,
+      "step": 5299
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5225269760516233e-05,
+      "loss": 0.059,
+      "step": 5300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5222428365915902e-05,
+      "loss": 0.0667,
+      "step": 5301
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.521958639144231e-05,
+      "loss": 0.0833,
+      "step": 5302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5216743837411008e-05,
+      "loss": 0.0773,
+      "step": 5303
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5213900704137627e-05,
+      "loss": 0.0876,
+      "step": 5304
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5211056991937856e-05,
+      "loss": 0.0721,
+      "step": 5305
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5208212701127437e-05,
+      "loss": 0.0683,
+      "step": 5306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5205367832022197e-05,
+      "loss": 0.0629,
+      "step": 5307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5202522384938012e-05,
+      "loss": 0.0654,
+      "step": 5308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5199676360190827e-05,
+      "loss": 0.0626,
+      "step": 5309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5196829758096652e-05,
+      "loss": 0.0746,
+      "step": 5310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5193982578971556e-05,
+      "loss": 0.0767,
+      "step": 5311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.519113482313168e-05,
+      "loss": 0.0851,
+      "step": 5312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5188286490893227e-05,
+      "loss": 0.0978,
+      "step": 5313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5185437582572458e-05,
+      "loss": 0.0927,
+      "step": 5314
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.51825880984857e-05,
+      "loss": 0.0743,
+      "step": 5315
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5179738038949354e-05,
+      "loss": 0.0791,
+      "step": 5316
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5176887404279872e-05,
+      "loss": 0.0735,
+      "step": 5317
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.517403619479377e-05,
+      "loss": 0.0902,
+      "step": 5318
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.517118441080764e-05,
+      "loss": 0.0817,
+      "step": 5319
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.516833205263813e-05,
+      "loss": 0.0599,
+      "step": 5320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5165479120601947e-05,
+      "loss": 0.0825,
+      "step": 5321
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.516262561501587e-05,
+      "loss": 0.0807,
+      "step": 5322
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.515977153619674e-05,
+      "loss": 0.1024,
+      "step": 5323
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5156916884461457e-05,
+      "loss": 0.0897,
+      "step": 5324
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5154061660126989e-05,
+      "loss": 0.0974,
+      "step": 5325
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5151205863510367e-05,
+      "loss": 0.0785,
+      "step": 5326
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5148349494928684e-05,
+      "loss": 0.0817,
+      "step": 5327
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.51454925546991e-05,
+      "loss": 0.0706,
+      "step": 5328
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.514263504313883e-05,
+      "loss": 0.0789,
+      "step": 5329
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5139776960565162e-05,
+      "loss": 0.0854,
+      "step": 5330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5136918307295445e-05,
+      "loss": 0.0766,
+      "step": 5331
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5134059083647086e-05,
+      "loss": 0.0657,
+      "step": 5332
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5131199289937566e-05,
+      "loss": 0.0924,
+      "step": 5333
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5128338926484419e-05,
+      "loss": 0.0676,
+      "step": 5334
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5125477993605243e-05,
+      "loss": 0.0577,
+      "step": 5335
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5122616491617706e-05,
+      "loss": 0.0847,
+      "step": 5336
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5119754420839534e-05,
+      "loss": 0.0734,
+      "step": 5337
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5116891781588518e-05,
+      "loss": 0.0858,
+      "step": 5338
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5114028574182511e-05,
+      "loss": 0.0668,
+      "step": 5339
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5111164798939433e-05,
+      "loss": 0.0797,
+      "step": 5340
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5108300456177257e-05,
+      "loss": 0.0601,
+      "step": 5341
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5105435546214031e-05,
+      "loss": 0.0734,
+      "step": 5342
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5102570069367861e-05,
+      "loss": 0.0892,
+      "step": 5343
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5099704025956911e-05,
+      "loss": 0.0842,
+      "step": 5344
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5096837416299421e-05,
+      "loss": 0.0822,
+      "step": 5345
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.509397024071368e-05,
+      "loss": 0.0751,
+      "step": 5346
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5091102499518042e-05,
+      "loss": 0.0742,
+      "step": 5347
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5088234193030933e-05,
+      "loss": 0.0768,
+      "step": 5348
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5085365321570833e-05,
+      "loss": 0.0702,
+      "step": 5349
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5082495885456293e-05,
+      "loss": 0.0719,
+      "step": 5350
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5079625885005912e-05,
+      "loss": 0.0602,
+      "step": 5351
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5076755320538368e-05,
+      "loss": 0.0724,
+      "step": 5352
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5073884192372393e-05,
+      "loss": 0.0828,
+      "step": 5353
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5071012500826783e-05,
+      "loss": 0.0911,
+      "step": 5354
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5068140246220396e-05,
+      "loss": 0.0864,
+      "step": 5355
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5065267428872152e-05,
+      "loss": 0.0762,
+      "step": 5356
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5062394049101038e-05,
+      "loss": 0.0828,
+      "step": 5357
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5059520107226097e-05,
+      "loss": 0.0851,
+      "step": 5358
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5056645603566444e-05,
+      "loss": 0.0718,
+      "step": 5359
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5053770538441246e-05,
+      "loss": 0.0738,
+      "step": 5360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.505089491216973e-05,
+      "loss": 0.0872,
+      "step": 5361
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5048018725071202e-05,
+      "loss": 0.0981,
+      "step": 5362
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5045141977465016e-05,
+      "loss": 0.0873,
+      "step": 5363
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.504226466967059e-05,
+      "loss": 0.0956,
+      "step": 5364
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.503938680200741e-05,
+      "loss": 0.0797,
+      "step": 5365
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5036508374795018e-05,
+      "loss": 0.0731,
+      "step": 5366
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.503362938835302e-05,
+      "loss": 0.074,
+      "step": 5367
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.503074984300109e-05,
+      "loss": 0.0727,
+      "step": 5368
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5027869739058951e-05,
+      "loss": 0.0731,
+      "step": 5369
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5024989076846401e-05,
+      "loss": 0.0756,
+      "step": 5370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5022107856683294e-05,
+      "loss": 0.0806,
+      "step": 5371
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5019226078889543e-05,
+      "loss": 0.075,
+      "step": 5372
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5016343743785135e-05,
+      "loss": 0.0859,
+      "step": 5373
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5013460851690106e-05,
+      "loss": 0.0881,
+      "step": 5374
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5010577402924556e-05,
+      "loss": 0.0741,
+      "step": 5375
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5007693397808652e-05,
+      "loss": 0.0831,
+      "step": 5376
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.500480883666262e-05,
+      "loss": 0.083,
+      "step": 5377
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5001923719806746e-05,
+      "loss": 0.0808,
+      "step": 5378
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.499903804756138e-05,
+      "loss": 0.0566,
+      "step": 5379
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4996151820246936e-05,
+      "loss": 0.0643,
+      "step": 5380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4993265038183885e-05,
+      "loss": 0.0779,
+      "step": 5381
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.499037770169276e-05,
+      "loss": 0.0649,
+      "step": 5382
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4987489811094156e-05,
+      "loss": 0.0749,
+      "step": 5383
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4984601366708735e-05,
+      "loss": 0.0956,
+      "step": 5384
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4981712368857213e-05,
+      "loss": 0.0771,
+      "step": 5385
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.497882281786037e-05,
+      "loss": 0.0745,
+      "step": 5386
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4975932714039048e-05,
+      "loss": 0.0789,
+      "step": 5387
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4973042057714154e-05,
+      "loss": 0.0762,
+      "step": 5388
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4970150849206643e-05,
+      "loss": 0.0713,
+      "step": 5389
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4967259088837552e-05,
+      "loss": 0.0955,
+      "step": 5390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4964366776927963e-05,
+      "loss": 0.0836,
+      "step": 5391
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.496147391379902e-05,
+      "loss": 0.075,
+      "step": 5392
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4958580499771943e-05,
+      "loss": 0.0626,
+      "step": 5393
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.4955686535167994e-05,
+      "loss": 0.0754,
+      "step": 5394
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.495279202030851e-05,
+      "loss": 0.0854,
+      "step": 5395
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4949896955514877e-05,
+      "loss": 0.0739,
+      "step": 5396
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4947001341108557e-05,
+      "loss": 0.0714,
+      "step": 5397
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4944105177411061e-05,
+      "loss": 0.0729,
+      "step": 5398
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4941208464743968e-05,
+      "loss": 0.0718,
+      "step": 5399
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4938311203428912e-05,
+      "loss": 0.0573,
+      "step": 5400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.493541339378759e-05,
+      "loss": 0.083,
+      "step": 5401
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4932515036141763e-05,
+      "loss": 0.0736,
+      "step": 5402
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4929616130813252e-05,
+      "loss": 0.0691,
+      "step": 5403
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4926716678123933e-05,
+      "loss": 0.0924,
+      "step": 5404
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4923816678395753e-05,
+      "loss": 0.0925,
+      "step": 5405
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.492091613195071e-05,
+      "loss": 0.0801,
+      "step": 5406
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4918015039110867e-05,
+      "loss": 0.0948,
+      "step": 5407
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4915113400198349e-05,
+      "loss": 0.0601,
+      "step": 5408
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4912211215535337e-05,
+      "loss": 0.084,
+      "step": 5409
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.490930848544408e-05,
+      "loss": 0.0797,
+      "step": 5410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4906405210246881e-05,
+      "loss": 0.0964,
+      "step": 5411
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4903501390266104e-05,
+      "loss": 0.069,
+      "step": 5412
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4900597025824177e-05,
+      "loss": 0.087,
+      "step": 5413
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4897692117243588e-05,
+      "loss": 0.1115,
+      "step": 5414
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4894786664846879e-05,
+      "loss": 0.0903,
+      "step": 5415
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4891880668956661e-05,
+      "loss": 0.0704,
+      "step": 5416
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4888974129895606e-05,
+      "loss": 0.0939,
+      "step": 5417
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4886067047986432e-05,
+      "loss": 0.0778,
+      "step": 5418
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4883159423551937e-05,
+      "loss": 0.0811,
+      "step": 5419
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4880251256914964e-05,
+      "loss": 0.0706,
+      "step": 5420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4877342548398421e-05,
+      "loss": 0.067,
+      "step": 5421
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4874433298325282e-05,
+      "loss": 0.0872,
+      "step": 5422
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.487152350701857e-05,
+      "loss": 0.0682,
+      "step": 5423
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4868613174801382e-05,
+      "loss": 0.0902,
+      "step": 5424
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4865702301996863e-05,
+      "loss": 0.0731,
+      "step": 5425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4862790888928219e-05,
+      "loss": 0.0745,
+      "step": 5426
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4859878935918724e-05,
+      "loss": 0.0557,
+      "step": 5427
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4856966443291705e-05,
+      "loss": 0.075,
+      "step": 5428
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4854053411370552e-05,
+      "loss": 0.0935,
+      "step": 5429
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4851139840478714e-05,
+      "loss": 0.0622,
+      "step": 5430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.48482257309397e-05,
+      "loss": 0.0763,
+      "step": 5431
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4845311083077076e-05,
+      "loss": 0.0696,
+      "step": 5432
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4842395897214477e-05,
+      "loss": 0.089,
+      "step": 5433
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4839480173675584e-05,
+      "loss": 0.0752,
+      "step": 5434
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4836563912784149e-05,
+      "loss": 0.0873,
+      "step": 5435
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4833647114863978e-05,
+      "loss": 0.0834,
+      "step": 5436
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4830729780238937e-05,
+      "loss": 0.0731,
+      "step": 5437
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4827811909232956e-05,
+      "loss": 0.0675,
+      "step": 5438
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4824893502170019e-05,
+      "loss": 0.093,
+      "step": 5439
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4821974559374172e-05,
+      "loss": 0.0818,
+      "step": 5440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4819055081169523e-05,
+      "loss": 0.1041,
+      "step": 5441
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4816135067880232e-05,
+      "loss": 0.0831,
+      "step": 5442
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4813214519830527e-05,
+      "loss": 0.0789,
+      "step": 5443
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.481029343734469e-05,
+      "loss": 0.0643,
+      "step": 5444
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4807371820747066e-05,
+      "loss": 0.07,
+      "step": 5445
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4804449670362053e-05,
+      "loss": 0.0909,
+      "step": 5446
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4801526986514117e-05,
+      "loss": 0.0887,
+      "step": 5447
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4798603769527775e-05,
+      "loss": 0.0825,
+      "step": 5448
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4795680019727607e-05,
+      "loss": 0.0845,
+      "step": 5449
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4792755737438258e-05,
+      "loss": 0.0818,
+      "step": 5450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.478983092298442e-05,
+      "loss": 0.0889,
+      "step": 5451
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4786905576690852e-05,
+      "loss": 0.0594,
+      "step": 5452
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4783979698882372e-05,
+      "loss": 0.0841,
+      "step": 5453
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4781053289883851e-05,
+      "loss": 0.0878,
+      "step": 5454
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4778126350020227e-05,
+      "loss": 0.0704,
+      "step": 5455
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4775198879616495e-05,
+      "loss": 0.0652,
+      "step": 5456
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4772270878997701e-05,
+      "loss": 0.0641,
+      "step": 5457
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4769342348488964e-05,
+      "loss": 0.0941,
+      "step": 5458
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4766413288415447e-05,
+      "loss": 0.0724,
+      "step": 5459
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.476348369910238e-05,
+      "loss": 0.0962,
+      "step": 5460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4760553580875057e-05,
+      "loss": 0.067,
+      "step": 5461
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4757622934058816e-05,
+      "loss": 0.0883,
+      "step": 5462
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4754691758979065e-05,
+      "loss": 0.0615,
+      "step": 5463
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4751760055961267e-05,
+      "loss": 0.0795,
+      "step": 5464
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4748827825330947e-05,
+      "loss": 0.0779,
+      "step": 5465
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.474589506741368e-05,
+      "loss": 0.0918,
+      "step": 5466
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.474296178253511e-05,
+      "loss": 0.0716,
+      "step": 5467
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4740027971020936e-05,
+      "loss": 0.0786,
+      "step": 5468
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4737093633196909e-05,
+      "loss": 0.0817,
+      "step": 5469
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.473415876938885e-05,
+      "loss": 0.075,
+      "step": 5470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4731223379922625e-05,
+      "loss": 0.0822,
+      "step": 5471
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4728287465124169e-05,
+      "loss": 0.0643,
+      "step": 5472
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4725351025319474e-05,
+      "loss": 0.0591,
+      "step": 5473
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4722414060834587e-05,
+      "loss": 0.0707,
+      "step": 5474
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.471947657199561e-05,
+      "loss": 0.0649,
+      "step": 5475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4716538559128715e-05,
+      "loss": 0.0817,
+      "step": 5476
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4713600022560117e-05,
+      "loss": 0.0714,
+      "step": 5477
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4710660962616104e-05,
+      "loss": 0.0878,
+      "step": 5478
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4707721379623012e-05,
+      "loss": 0.088,
+      "step": 5479
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4704781273907234e-05,
+      "loss": 0.0846,
+      "step": 5480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4701840645795234e-05,
+      "loss": 0.0781,
+      "step": 5481
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4698899495613519e-05,
+      "loss": 0.0563,
+      "step": 5482
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4695957823688659e-05,
+      "loss": 0.0763,
+      "step": 5483
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4693015630347287e-05,
+      "loss": 0.0699,
+      "step": 5484
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4690072915916091e-05,
+      "loss": 0.0946,
+      "step": 5485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4687129680721807e-05,
+      "loss": 0.0787,
+      "step": 5486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.468418592509125e-05,
+      "loss": 0.0637,
+      "step": 5487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4681241649351275e-05,
+      "loss": 0.0697,
+      "step": 5488
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4678296853828795e-05,
+      "loss": 0.0845,
+      "step": 5489
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4675351538850793e-05,
+      "loss": 0.0611,
+      "step": 5490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.46724057047443e-05,
+      "loss": 0.076,
+      "step": 5491
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4669459351836404e-05,
+      "loss": 0.0739,
+      "step": 5492
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4666512480454261e-05,
+      "loss": 0.0893,
+      "step": 5493
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.466356509092507e-05,
+      "loss": 0.0807,
+      "step": 5494
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4660617183576102e-05,
+      "loss": 0.0749,
+      "step": 5495
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4657668758734673e-05,
+      "loss": 0.0892,
+      "step": 5496
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4654719816728164e-05,
+      "loss": 0.0764,
+      "step": 5497
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.465177035788401e-05,
+      "loss": 0.0896,
+      "step": 5498
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4648820382529707e-05,
+      "loss": 0.0863,
+      "step": 5499
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4645869890992804e-05,
+      "loss": 0.0795,
+      "step": 5500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4642918883600909e-05,
+      "loss": 0.0659,
+      "step": 5501
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4639967360681691e-05,
+      "loss": 0.0773,
+      "step": 5502
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.463701532256287e-05,
+      "loss": 0.0687,
+      "step": 5503
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4634062769572225e-05,
+      "loss": 0.0688,
+      "step": 5504
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4631109702037597e-05,
+      "loss": 0.0765,
+      "step": 5505
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4628156120286879e-05,
+      "loss": 0.0848,
+      "step": 5506
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4625202024648019e-05,
+      "loss": 0.092,
+      "step": 5507
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4622247415449028e-05,
+      "loss": 0.0828,
+      "step": 5508
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4619292293017972e-05,
+      "loss": 0.0828,
+      "step": 5509
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4616336657682977e-05,
+      "loss": 0.0741,
+      "step": 5510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4613380509772215e-05,
+      "loss": 0.0735,
+      "step": 5511
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4610423849613926e-05,
+      "loss": 0.1011,
+      "step": 5512
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4607466677536405e-05,
+      "loss": 0.0815,
+      "step": 5513
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4604508993868e-05,
+      "loss": 0.0764,
+      "step": 5514
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4601550798937121e-05,
+      "loss": 0.0635,
+      "step": 5515
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4598592093072228e-05,
+      "loss": 0.0609,
+      "step": 5516
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4595632876601841e-05,
+      "loss": 0.0695,
+      "step": 5517
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4592673149854541e-05,
+      "loss": 0.0688,
+      "step": 5518
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.458971291315896e-05,
+      "loss": 0.0894,
+      "step": 5519
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4586752166843786e-05,
+      "loss": 0.0779,
+      "step": 5520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.458379091123777e-05,
+      "loss": 0.0673,
+      "step": 5521
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4580829146669713e-05,
+      "loss": 0.071,
+      "step": 5522
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4577866873468473e-05,
+      "loss": 0.0864,
+      "step": 5523
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4574904091962973e-05,
+      "loss": 0.0892,
+      "step": 5524
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4571940802482183e-05,
+      "loss": 0.0727,
+      "step": 5525
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4568977005355128e-05,
+      "loss": 0.0767,
+      "step": 5526
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4566012700910896e-05,
+      "loss": 0.0741,
+      "step": 5527
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4563047889478633e-05,
+      "loss": 0.0721,
+      "step": 5528
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4560082571387537e-05,
+      "loss": 0.0671,
+      "step": 5529
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4557116746966857e-05,
+      "loss": 0.0748,
+      "step": 5530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4554150416545908e-05,
+      "loss": 0.0607,
+      "step": 5531
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4551183580454056e-05,
+      "loss": 0.0918,
+      "step": 5532
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4548216239020727e-05,
+      "loss": 0.0577,
+      "step": 5533
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4545248392575392e-05,
+      "loss": 0.0951,
+      "step": 5534
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4542280041447599e-05,
+      "loss": 0.0696,
+      "step": 5535
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4539311185966929e-05,
+      "loss": 0.0839,
+      "step": 5536
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4536341826463032e-05,
+      "loss": 0.0858,
+      "step": 5537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4533371963265613e-05,
+      "loss": 0.0731,
+      "step": 5538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4530401596704432e-05,
+      "loss": 0.0652,
+      "step": 5539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4527430727109299e-05,
+      "loss": 0.0673,
+      "step": 5540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4524459354810091e-05,
+      "loss": 0.0815,
+      "step": 5541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4521487480136734e-05,
+      "loss": 0.0878,
+      "step": 5542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4518515103419207e-05,
+      "loss": 0.0861,
+      "step": 5543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4515542224987552e-05,
+      "loss": 0.0736,
+      "step": 5544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4512568845171861e-05,
+      "loss": 0.0927,
+      "step": 5545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4509594964302283e-05,
+      "loss": 0.0688,
+      "step": 5546
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4506620582709026e-05,
+      "loss": 0.0887,
+      "step": 5547
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.450364570072235e-05,
+      "loss": 0.0842,
+      "step": 5548
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4500670318672571e-05,
+      "loss": 0.0756,
+      "step": 5549
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4497694436890062e-05,
+      "loss": 0.0758,
+      "step": 5550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.449471805570525e-05,
+      "loss": 0.0767,
+      "step": 5551
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4491741175448617e-05,
+      "loss": 0.0897,
+      "step": 5552
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4488763796450704e-05,
+      "loss": 0.0782,
+      "step": 5553
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4485785919042101e-05,
+      "loss": 0.0808,
+      "step": 5554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.448280754355346e-05,
+      "loss": 0.0834,
+      "step": 5555
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4479828670315486e-05,
+      "loss": 0.0858,
+      "step": 5556
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4476849299658938e-05,
+      "loss": 0.0789,
+      "step": 5557
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4473869431914628e-05,
+      "loss": 0.0851,
+      "step": 5558
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4470889067413432e-05,
+      "loss": 0.0687,
+      "step": 5559
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.446790820648627e-05,
+      "loss": 0.0699,
+      "step": 5560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4464926849464122e-05,
+      "loss": 0.0944,
+      "step": 5561
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.446194499667803e-05,
+      "loss": 0.0774,
+      "step": 5562
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4458962648459079e-05,
+      "loss": 0.0616,
+      "step": 5563
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4455979805138416e-05,
+      "loss": 0.0603,
+      "step": 5564
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4452996467047243e-05,
+      "loss": 0.0739,
+      "step": 5565
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.445001263451681e-05,
+      "loss": 0.0828,
+      "step": 5566
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4447028307878435e-05,
+      "loss": 0.0933,
+      "step": 5567
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4444043487463479e-05,
+      "loss": 0.0801,
+      "step": 5568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.444105817360336e-05,
+      "loss": 0.0761,
+      "step": 5569
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4438072366629561e-05,
+      "loss": 0.0824,
+      "step": 5570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4435086066873602e-05,
+      "loss": 0.0869,
+      "step": 5571
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4432099274667073e-05,
+      "loss": 0.0626,
+      "step": 5572
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.442911199034161e-05,
+      "loss": 0.0799,
+      "step": 5573
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.442612421422891e-05,
+      "loss": 0.0887,
+      "step": 5574
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.442313594666072e-05,
+      "loss": 0.0746,
+      "step": 5575
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.442014718796884e-05,
+      "loss": 0.0725,
+      "step": 5576
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4417157938485128e-05,
+      "loss": 0.082,
+      "step": 5577
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4414168198541503e-05,
+      "loss": 0.072,
+      "step": 5578
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4411177968469922e-05,
+      "loss": 0.0858,
+      "step": 5579
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.440818724860241e-05,
+      "loss": 0.0997,
+      "step": 5580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.440519603927104e-05,
+      "loss": 0.0733,
+      "step": 5581
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4402204340807947e-05,
+      "loss": 0.0726,
+      "step": 5582
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4399212153545305e-05,
+      "loss": 0.0778,
+      "step": 5583
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4396219477815363e-05,
+      "loss": 0.0748,
+      "step": 5584
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4393226313950406e-05,
+      "loss": 0.0767,
+      "step": 5585
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4390232662282784e-05,
+      "loss": 0.0645,
+      "step": 5586
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4387238523144894e-05,
+      "loss": 0.0924,
+      "step": 5587
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4384243896869194e-05,
+      "loss": 0.0629,
+      "step": 5588
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4381248783788194e-05,
+      "loss": 0.0906,
+      "step": 5589
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4378253184234452e-05,
+      "loss": 0.0862,
+      "step": 5590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4375257098540588e-05,
+      "loss": 0.0666,
+      "step": 5591
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4372260527039275e-05,
+      "loss": 0.0669,
+      "step": 5592
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4369263470063235e-05,
+      "loss": 0.0726,
+      "step": 5593
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4366265927945244e-05,
+      "loss": 0.0848,
+      "step": 5594
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4363267901018141e-05,
+      "loss": 0.0673,
+      "step": 5595
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4360269389614809e-05,
+      "loss": 0.0666,
+      "step": 5596
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4357270394068186e-05,
+      "loss": 0.0652,
+      "step": 5597
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4354270914711273e-05,
+      "loss": 0.0843,
+      "step": 5598
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.435127095187711e-05,
+      "loss": 0.0656,
+      "step": 5599
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4348270505898802e-05,
+      "loss": 0.0748,
+      "step": 5600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4345269577109503e-05,
+      "loss": 0.0648,
+      "step": 5601
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4342268165842426e-05,
+      "loss": 0.064,
+      "step": 5602
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4339266272430828e-05,
+      "loss": 0.0659,
+      "step": 5603
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4336263897208025e-05,
+      "loss": 0.0961,
+      "step": 5604
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.433326104050739e-05,
+      "loss": 0.0787,
+      "step": 5605
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.433025770266234e-05,
+      "loss": 0.0804,
+      "step": 5606
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4327253884006358e-05,
+      "loss": 0.0669,
+      "step": 5607
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4324249584872969e-05,
+      "loss": 0.0743,
+      "step": 5608
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4321244805595758e-05,
+      "loss": 0.0814,
+      "step": 5609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.431823954650836e-05,
+      "loss": 0.0742,
+      "step": 5610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4315233807944462e-05,
+      "loss": 0.0844,
+      "step": 5611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.431222759023781e-05,
+      "loss": 0.0691,
+      "step": 5612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4309220893722203e-05,
+      "loss": 0.0836,
+      "step": 5613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.430621371873148e-05,
+      "loss": 0.0892,
+      "step": 5614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4303206065599554e-05,
+      "loss": 0.0706,
+      "step": 5615
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4300197934660378e-05,
+      "loss": 0.0698,
+      "step": 5616
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4297189326247956e-05,
+      "loss": 0.064,
+      "step": 5617
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4294180240696351e-05,
+      "loss": 0.0875,
+      "step": 5618
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4291170678339679e-05,
+      "loss": 0.076,
+      "step": 5619
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4288160639512105e-05,
+      "loss": 0.0659,
+      "step": 5620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4285150124547853e-05,
+      "loss": 0.0715,
+      "step": 5621
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4282139133781192e-05,
+      "loss": 0.0556,
+      "step": 5622
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4279127667546448e-05,
+      "loss": 0.0665,
+      "step": 5623
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4276115726178003e-05,
+      "loss": 0.071,
+      "step": 5624
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4273103310010288e-05,
+      "loss": 0.0871,
+      "step": 5625
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4270090419377783e-05,
+      "loss": 0.0819,
+      "step": 5626
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4267077054615026e-05,
+      "loss": 0.0723,
+      "step": 5627
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4264063216056609e-05,
+      "loss": 0.0827,
+      "step": 5628
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4261048904037178e-05,
+      "loss": 0.0681,
+      "step": 5629
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.425803411889142e-05,
+      "loss": 0.0702,
+      "step": 5630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4255018860954084e-05,
+      "loss": 0.0728,
+      "step": 5631
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4252003130559971e-05,
+      "loss": 0.084,
+      "step": 5632
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4248986928043936e-05,
+      "loss": 0.0778,
+      "step": 5633
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4245970253740874e-05,
+      "loss": 0.0687,
+      "step": 5634
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4242953107985756e-05,
+      "loss": 0.0751,
+      "step": 5635
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4239935491113579e-05,
+      "loss": 0.0845,
+      "step": 5636
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4236917403459412e-05,
+      "loss": 0.0738,
+      "step": 5637
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4233898845358365e-05,
+      "loss": 0.0764,
+      "step": 5638
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4230879817145607e-05,
+      "loss": 0.0747,
+      "step": 5639
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4227860319156353e-05,
+      "loss": 0.0621,
+      "step": 5640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4224840351725875e-05,
+      "loss": 0.0851,
+      "step": 5641
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4221819915189496e-05,
+      "loss": 0.0839,
+      "step": 5642
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4218799009882593e-05,
+      "loss": 0.0695,
+      "step": 5643
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4215777636140589e-05,
+      "loss": 0.0836,
+      "step": 5644
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4212755794298962e-05,
+      "loss": 0.0837,
+      "step": 5645
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.420973348469325e-05,
+      "loss": 0.0825,
+      "step": 5646
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.4206710707659025e-05,
+      "loss": 0.0772,
+      "step": 5647
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.420368746353193e-05,
+      "loss": 0.0912,
+      "step": 5648
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4200663752647649e-05,
+      "loss": 0.0632,
+      "step": 5649
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.419763957534192e-05,
+      "loss": 0.0641,
+      "step": 5650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.419461493195053e-05,
+      "loss": 0.0806,
+      "step": 5651
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4191589822809326e-05,
+      "loss": 0.0662,
+      "step": 5652
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4188564248254201e-05,
+      "loss": 0.0725,
+      "step": 5653
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4185538208621097e-05,
+      "loss": 0.0717,
+      "step": 5654
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4182511704246012e-05,
+      "loss": 0.0852,
+      "step": 5655
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4179484735464995e-05,
+      "loss": 0.0675,
+      "step": 5656
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4176457302614145e-05,
+      "loss": 0.0676,
+      "step": 5657
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4173429406029616e-05,
+      "loss": 0.0599,
+      "step": 5658
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4170401046047608e-05,
+      "loss": 0.0912,
+      "step": 5659
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4167372223004379e-05,
+      "loss": 0.0723,
+      "step": 5660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4164342937236231e-05,
+      "loss": 0.0612,
+      "step": 5661
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4161313189079523e-05,
+      "loss": 0.0817,
+      "step": 5662
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4158282978870665e-05,
+      "loss": 0.0592,
+      "step": 5663
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4155252306946117e-05,
+      "loss": 0.0938,
+      "step": 5664
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4152221173642385e-05,
+      "loss": 0.0614,
+      "step": 5665
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.414918957929604e-05,
+      "loss": 0.0859,
+      "step": 5666
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4146157524243686e-05,
+      "loss": 0.0751,
+      "step": 5667
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4143125008821996e-05,
+      "loss": 0.0581,
+      "step": 5668
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.414009203336768e-05,
+      "loss": 0.071,
+      "step": 5669
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4137058598217511e-05,
+      "loss": 0.0959,
+      "step": 5670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4134024703708301e-05,
+      "loss": 0.0961,
+      "step": 5671
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4130990350176924e-05,
+      "loss": 0.0857,
+      "step": 5672
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4127955537960296e-05,
+      "loss": 0.0768,
+      "step": 5673
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4124920267395388e-05,
+      "loss": 0.0787,
+      "step": 5674
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4121884538819224e-05,
+      "loss": 0.0924,
+      "step": 5675
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.411884835256888e-05,
+      "loss": 0.0619,
+      "step": 5676
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4115811708981469e-05,
+      "loss": 0.0847,
+      "step": 5677
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4112774608394174e-05,
+      "loss": 0.0763,
+      "step": 5678
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4109737051144221e-05,
+      "loss": 0.0663,
+      "step": 5679
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4106699037568878e-05,
+      "loss": 0.0737,
+      "step": 5680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4103660568005479e-05,
+      "loss": 0.0688,
+      "step": 5681
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4100621642791395e-05,
+      "loss": 0.0834,
+      "step": 5682
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4097582262264058e-05,
+      "loss": 0.0771,
+      "step": 5683
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4094542426760947e-05,
+      "loss": 0.0839,
+      "step": 5684
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4091502136619583e-05,
+      "loss": 0.1001,
+      "step": 5685
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4088461392177555e-05,
+      "loss": 0.0817,
+      "step": 5686
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4085420193772485e-05,
+      "loss": 0.0816,
+      "step": 5687
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4082378541742057e-05,
+      "loss": 0.0735,
+      "step": 5688
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4079336436424e-05,
+      "loss": 0.0645,
+      "step": 5689
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4076293878156097e-05,
+      "loss": 0.0726,
+      "step": 5690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4073250867276174e-05,
+      "loss": 0.0804,
+      "step": 5691
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4070207404122118e-05,
+      "loss": 0.0884,
+      "step": 5692
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4067163489031858e-05,
+      "loss": 0.0847,
+      "step": 5693
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4064119122343374e-05,
+      "loss": 0.0763,
+      "step": 5694
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.40610743043947e-05,
+      "loss": 0.0682,
+      "step": 5695
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4058029035523917e-05,
+      "loss": 0.0725,
+      "step": 5696
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4054983316069155e-05,
+      "loss": 0.0808,
+      "step": 5697
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4051937146368602e-05,
+      "loss": 0.0618,
+      "step": 5698
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4048890526760484e-05,
+      "loss": 0.0722,
+      "step": 5699
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4045843457583086e-05,
+      "loss": 0.0818,
+      "step": 5700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4042795939174738e-05,
+      "loss": 0.0941,
+      "step": 5701
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4039747971873822e-05,
+      "loss": 0.0813,
+      "step": 5702
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.403669955601877e-05,
+      "loss": 0.0789,
+      "step": 5703
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4033650691948068e-05,
+      "loss": 0.0741,
+      "step": 5704
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.403060138000024e-05,
+      "loss": 0.0564,
+      "step": 5705
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.402755162051387e-05,
+      "loss": 0.0737,
+      "step": 5706
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.402450141382759e-05,
+      "loss": 0.0876,
+      "step": 5707
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4021450760280076e-05,
+      "loss": 0.0835,
+      "step": 5708
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.401839966021006e-05,
+      "loss": 0.053,
+      "step": 5709
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4015348113956324e-05,
+      "loss": 0.0664,
+      "step": 5710
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.401229612185769e-05,
+      "loss": 0.0763,
+      "step": 5711
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.400924368425305e-05,
+      "loss": 0.0579,
+      "step": 5712
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4006190801481317e-05,
+      "loss": 0.0848,
+      "step": 5713
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4003137473881476e-05,
+      "loss": 0.0712,
+      "step": 5714
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4000083701792554e-05,
+      "loss": 0.0756,
+      "step": 5715
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3997029485553622e-05,
+      "loss": 0.0679,
+      "step": 5716
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3993974825503812e-05,
+      "loss": 0.08,
+      "step": 5717
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3990919721982294e-05,
+      "loss": 0.0864,
+      "step": 5718
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3987864175328294e-05,
+      "loss": 0.0634,
+      "step": 5719
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3984808185881086e-05,
+      "loss": 0.0964,
+      "step": 5720
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3981751753979993e-05,
+      "loss": 0.0726,
+      "step": 5721
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3978694879964381e-05,
+      "loss": 0.0656,
+      "step": 5722
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3975637564173676e-05,
+      "loss": 0.0836,
+      "step": 5723
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3972579806947347e-05,
+      "loss": 0.0659,
+      "step": 5724
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3969521608624909e-05,
+      "loss": 0.0952,
+      "step": 5725
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3966462969545933e-05,
+      "loss": 0.1018,
+      "step": 5726
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3963403890050036e-05,
+      "loss": 0.0768,
+      "step": 5727
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.396034437047688e-05,
+      "loss": 0.0791,
+      "step": 5728
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3957284411166183e-05,
+      "loss": 0.069,
+      "step": 5729
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3954224012457707e-05,
+      "loss": 0.0826,
+      "step": 5730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.395116317469126e-05,
+      "loss": 0.0793,
+      "step": 5731
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.394810189820671e-05,
+      "loss": 0.0748,
+      "step": 5732
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3945040183343958e-05,
+      "loss": 0.0745,
+      "step": 5733
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3941978030442968e-05,
+      "loss": 0.0911,
+      "step": 5734
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3938915439843748e-05,
+      "loss": 0.0785,
+      "step": 5735
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3935852411886347e-05,
+      "loss": 0.0646,
+      "step": 5736
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3932788946910877e-05,
+      "loss": 0.0678,
+      "step": 5737
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.392972504525748e-05,
+      "loss": 0.0723,
+      "step": 5738
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3926660707266364e-05,
+      "loss": 0.0884,
+      "step": 5739
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.392359593327778e-05,
+      "loss": 0.0663,
+      "step": 5740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.392053072363202e-05,
+      "loss": 0.084,
+      "step": 5741
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3917465078669433e-05,
+      "loss": 0.0598,
+      "step": 5742
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3914398998730414e-05,
+      "loss": 0.0723,
+      "step": 5743
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3911332484155406e-05,
+      "loss": 0.0702,
+      "step": 5744
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3908265535284895e-05,
+      "loss": 0.0784,
+      "step": 5745
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3905198152459427e-05,
+      "loss": 0.0865,
+      "step": 5746
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3902130336019585e-05,
+      "loss": 0.0701,
+      "step": 5747
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3899062086306008e-05,
+      "loss": 0.0878,
+      "step": 5748
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3895993403659377e-05,
+      "loss": 0.0884,
+      "step": 5749
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3892924288420424e-05,
+      "loss": 0.0909,
+      "step": 5750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3889854740929927e-05,
+      "loss": 0.0704,
+      "step": 5751
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.388678476152872e-05,
+      "loss": 0.0874,
+      "step": 5752
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.388371435055767e-05,
+      "loss": 0.0672,
+      "step": 5753
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3880643508357708e-05,
+      "loss": 0.0767,
+      "step": 5754
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3877572235269802e-05,
+      "loss": 0.0732,
+      "step": 5755
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3874500531634975e-05,
+      "loss": 0.0823,
+      "step": 5756
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3871428397794286e-05,
+      "loss": 0.08,
+      "step": 5757
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.386835583408886e-05,
+      "loss": 0.0863,
+      "step": 5758
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.386528284085985e-05,
+      "loss": 0.0696,
+      "step": 5759
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3862209418448476e-05,
+      "loss": 0.0841,
+      "step": 5760
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3859135567195989e-05,
+      "loss": 0.0732,
+      "step": 5761
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3856061287443696e-05,
+      "loss": 0.0868,
+      "step": 5762
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3852986579532954e-05,
+      "loss": 0.0878,
+      "step": 5763
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.384991144380516e-05,
+      "loss": 0.071,
+      "step": 5764
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3846835880601761e-05,
+      "loss": 0.0847,
+      "step": 5765
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3843759890264256e-05,
+      "loss": 0.0739,
+      "step": 5766
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3840683473134186e-05,
+      "loss": 0.0684,
+      "step": 5767
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3837606629553143e-05,
+      "loss": 0.0732,
+      "step": 5768
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3834529359862763e-05,
+      "loss": 0.0789,
+      "step": 5769
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3831451664404737e-05,
+      "loss": 0.0728,
+      "step": 5770
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3828373543520787e-05,
+      "loss": 0.0721,
+      "step": 5771
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3825294997552703e-05,
+      "loss": 0.0853,
+      "step": 5772
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3822216026842306e-05,
+      "loss": 0.0686,
+      "step": 5773
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3819136631731471e-05,
+      "loss": 0.0781,
+      "step": 5774
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.381605681256212e-05,
+      "loss": 0.0748,
+      "step": 5775
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.381297656967622e-05,
+      "loss": 0.0664,
+      "step": 5776
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3809895903415788e-05,
+      "loss": 0.0637,
+      "step": 5777
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3806814814122888e-05,
+      "loss": 0.0941,
+      "step": 5778
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3803733302139624e-05,
+      "loss": 0.0857,
+      "step": 5779
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3800651367808158e-05,
+      "loss": 0.0617,
+      "step": 5780
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3797569011470688e-05,
+      "loss": 0.0528,
+      "step": 5781
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3794486233469466e-05,
+      "loss": 0.0794,
+      "step": 5782
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3791403034146788e-05,
+      "loss": 0.0629,
+      "step": 5783
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3788319413845004e-05,
+      "loss": 0.0902,
+      "step": 5784
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3785235372906494e-05,
+      "loss": 0.084,
+      "step": 5785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3782150911673702e-05,
+      "loss": 0.0907,
+      "step": 5786
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3779066030489107e-05,
+      "loss": 0.0821,
+      "step": 5787
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3775980729695247e-05,
+      "loss": 0.0758,
+      "step": 5788
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3772895009634693e-05,
+      "loss": 0.0667,
+      "step": 5789
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3769808870650066e-05,
+      "loss": 0.0732,
+      "step": 5790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3766722313084042e-05,
+      "loss": 0.0936,
+      "step": 5791
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3763635337279335e-05,
+      "loss": 0.0652,
+      "step": 5792
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3760547943578706e-05,
+      "loss": 0.078,
+      "step": 5793
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3757460132324968e-05,
+      "loss": 0.084,
+      "step": 5794
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3754371903860976e-05,
+      "loss": 0.0714,
+      "step": 5795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.375128325852963e-05,
+      "loss": 0.0856,
+      "step": 5796
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3748194196673876e-05,
+      "loss": 0.0948,
+      "step": 5797
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3745104718636718e-05,
+      "loss": 0.0908,
+      "step": 5798
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3742014824761183e-05,
+      "loss": 0.0785,
+      "step": 5799
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.373892451539037e-05,
+      "loss": 0.0621,
+      "step": 5800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3735833790867407e-05,
+      "loss": 0.0722,
+      "step": 5801
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3732742651535471e-05,
+      "loss": 0.0682,
+      "step": 5802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3729651097737793e-05,
+      "loss": 0.0612,
+      "step": 5803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3726559129817641e-05,
+      "loss": 0.0757,
+      "step": 5804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3723466748118329e-05,
+      "loss": 0.0862,
+      "step": 5805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3720373952983226e-05,
+      "loss": 0.0859,
+      "step": 5806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3717280744755738e-05,
+      "loss": 0.0764,
+      "step": 5807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3714187123779316e-05,
+      "loss": 0.0786,
+      "step": 5808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3711093090397471e-05,
+      "loss": 0.0784,
+      "step": 5809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.370799864495374e-05,
+      "loss": 0.0812,
+      "step": 5810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3704903787791722e-05,
+      "loss": 0.0723,
+      "step": 5811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3701808519255051e-05,
+      "loss": 0.0875,
+      "step": 5812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3698712839687409e-05,
+      "loss": 0.0782,
+      "step": 5813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.369561674943253e-05,
+      "loss": 0.0738,
+      "step": 5814
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.369252024883419e-05,
+      "loss": 0.0806,
+      "step": 5815
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3689423338236201e-05,
+      "loss": 0.0916,
+      "step": 5816
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3686326017982436e-05,
+      "loss": 0.0741,
+      "step": 5817
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3683228288416807e-05,
+      "loss": 0.0718,
+      "step": 5818
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3680130149883266e-05,
+      "loss": 0.0793,
+      "step": 5819
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.367703160272582e-05,
+      "loss": 0.0721,
+      "step": 5820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3673932647288515e-05,
+      "loss": 0.0732,
+      "step": 5821
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3670833283915444e-05,
+      "loss": 0.0929,
+      "step": 5822
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3667733512950742e-05,
+      "loss": 0.0754,
+      "step": 5823
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3664633334738601e-05,
+      "loss": 0.0834,
+      "step": 5824
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3661532749623242e-05,
+      "loss": 0.0549,
+      "step": 5825
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3658431757948941e-05,
+      "loss": 0.0721,
+      "step": 5826
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3655330360060025e-05,
+      "loss": 0.0781,
+      "step": 5827
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3652228556300844e-05,
+      "loss": 0.0822,
+      "step": 5828
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3649126347015816e-05,
+      "loss": 0.0701,
+      "step": 5829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3646023732549396e-05,
+      "loss": 0.0745,
+      "step": 5830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.364292071324608e-05,
+      "loss": 0.0723,
+      "step": 5831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3639817289450412e-05,
+      "loss": 0.0862,
+      "step": 5832
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3636713461506985e-05,
+      "loss": 0.074,
+      "step": 5833
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3633609229760429e-05,
+      "loss": 0.0781,
+      "step": 5834
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.363050459455543e-05,
+      "loss": 0.0759,
+      "step": 5835
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.36273995562367e-05,
+      "loss": 0.0766,
+      "step": 5836
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.362429411514902e-05,
+      "loss": 0.0648,
+      "step": 5837
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3621188271637193e-05,
+      "loss": 0.081,
+      "step": 5838
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3618082026046084e-05,
+      "loss": 0.0671,
+      "step": 5839
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3614975378720592e-05,
+      "loss": 0.0703,
+      "step": 5840
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3611868330005663e-05,
+      "loss": 0.0744,
+      "step": 5841
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3608760880246293e-05,
+      "loss": 0.085,
+      "step": 5842
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3605653029787513e-05,
+      "loss": 0.0765,
+      "step": 5843
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3602544778974407e-05,
+      "loss": 0.0665,
+      "step": 5844
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3599436128152101e-05,
+      "loss": 0.0787,
+      "step": 5845
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3596327077665765e-05,
+      "loss": 0.0622,
+      "step": 5846
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.359321762786061e-05,
+      "loss": 0.0631,
+      "step": 5847
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3590107779081894e-05,
+      "loss": 0.064,
+      "step": 5848
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3586997531674924e-05,
+      "loss": 0.0744,
+      "step": 5849
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.358388688598504e-05,
+      "loss": 0.0736,
+      "step": 5850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3580775842357641e-05,
+      "loss": 0.0913,
+      "step": 5851
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3577664401138158e-05,
+      "loss": 0.0865,
+      "step": 5852
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3574552562672068e-05,
+      "loss": 0.0688,
+      "step": 5853
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.35714403273049e-05,
+      "loss": 0.0696,
+      "step": 5854
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3568327695382219e-05,
+      "loss": 0.0806,
+      "step": 5855
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3565214667249633e-05,
+      "loss": 0.0889,
+      "step": 5856
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3562101243252805e-05,
+      "loss": 0.0727,
+      "step": 5857
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3558987423737434e-05,
+      "loss": 0.0835,
+      "step": 5858
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3555873209049252e-05,
+      "loss": 0.0662,
+      "step": 5859
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3552758599534062e-05,
+      "loss": 0.0715,
+      "step": 5860
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3549643595537685e-05,
+      "loss": 0.0918,
+      "step": 5861
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3546528197405998e-05,
+      "loss": 0.08,
+      "step": 5862
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3543412405484924e-05,
+      "loss": 0.0846,
+      "step": 5863
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.354029622012042e-05,
+      "loss": 0.0724,
+      "step": 5864
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3537179641658495e-05,
+      "loss": 0.0675,
+      "step": 5865
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3534062670445198e-05,
+      "loss": 0.0773,
+      "step": 5866
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3530945306826624e-05,
+      "loss": 0.0742,
+      "step": 5867
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3527827551148906e-05,
+      "loss": 0.063,
+      "step": 5868
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3524709403758231e-05,
+      "loss": 0.0832,
+      "step": 5869
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3521590865000817e-05,
+      "loss": 0.0776,
+      "step": 5870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3518471935222936e-05,
+      "loss": 0.0824,
+      "step": 5871
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3515352614770895e-05,
+      "loss": 0.0958,
+      "step": 5872
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3512232903991051e-05,
+      "loss": 0.0702,
+      "step": 5873
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3509112803229803e-05,
+      "loss": 0.0637,
+      "step": 5874
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3505992312833586e-05,
+      "loss": 0.0787,
+      "step": 5875
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.350287143314889e-05,
+      "loss": 0.0787,
+      "step": 5876
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.349975016452224e-05,
+      "loss": 0.0626,
+      "step": 5877
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.349662850730021e-05,
+      "loss": 0.0648,
+      "step": 5878
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3493506461829407e-05,
+      "loss": 0.0845,
+      "step": 5879
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3490384028456492e-05,
+      "loss": 0.0822,
+      "step": 5880
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3487261207528167e-05,
+      "loss": 0.0723,
+      "step": 5881
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3484137999391169e-05,
+      "loss": 0.0624,
+      "step": 5882
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.348101440439229e-05,
+      "loss": 0.0632,
+      "step": 5883
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3477890422878357e-05,
+      "loss": 0.0654,
+      "step": 5884
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3474766055196238e-05,
+      "loss": 0.0749,
+      "step": 5885
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3471641301692855e-05,
+      "loss": 0.0767,
+      "step": 5886
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.346851616271516e-05,
+      "loss": 0.0595,
+      "step": 5887
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3465390638610154e-05,
+      "loss": 0.0789,
+      "step": 5888
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3462264729724882e-05,
+      "loss": 0.0798,
+      "step": 5889
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.345913843640643e-05,
+      "loss": 0.0898,
+      "step": 5890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3456011759001926e-05,
+      "loss": 0.0764,
+      "step": 5891
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3452884697858538e-05,
+      "loss": 0.0726,
+      "step": 5892
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3449757253323484e-05,
+      "loss": 0.0692,
+      "step": 5893
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3446629425744018e-05,
+      "loss": 0.0695,
+      "step": 5894
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.344350121546744e-05,
+      "loss": 0.0894,
+      "step": 5895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3440372622841091e-05,
+      "loss": 0.0817,
+      "step": 5896
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3437243648212358e-05,
+      "loss": 0.0659,
+      "step": 5897
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.343411429192866e-05,
+      "loss": 0.0743,
+      "step": 5898
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3430984554337472e-05,
+      "loss": 0.0784,
+      "step": 5899
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3427854435786303e-05,
+      "loss": 0.0729,
+      "step": 5900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3424723936622706e-05,
+      "loss": 0.0623,
+      "step": 5901
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3421593057194273e-05,
+      "loss": 0.0716,
+      "step": 5902
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.341846179784865e-05,
+      "loss": 0.0896,
+      "step": 5903
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.341533015893351e-05,
+      "loss": 0.073,
+      "step": 5904
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3412198140796577e-05,
+      "loss": 0.0623,
+      "step": 5905
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3409065743785619e-05,
+      "loss": 0.0797,
+      "step": 5906
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3405932968248436e-05,
+      "loss": 0.0781,
+      "step": 5907
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3402799814532881e-05,
+      "loss": 0.0845,
+      "step": 5908
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3399666282986843e-05,
+      "loss": 0.071,
+      "step": 5909
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3396532373958248e-05,
+      "loss": 0.0647,
+      "step": 5910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3393398087795083e-05,
+      "loss": 0.0718,
+      "step": 5911
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3390263424845353e-05,
+      "loss": 0.074,
+      "step": 5912
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3387128385457117e-05,
+      "loss": 0.0697,
+      "step": 5913
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3383992969978481e-05,
+      "loss": 0.0856,
+      "step": 5914
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3380857178757584e-05,
+      "loss": 0.0684,
+      "step": 5915
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3377721012142603e-05,
+      "loss": 0.0709,
+      "step": 5916
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3374584470481772e-05,
+      "loss": 0.0693,
+      "step": 5917
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.337144755412335e-05,
+      "loss": 0.0812,
+      "step": 5918
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3368310263415652e-05,
+      "loss": 0.067,
+      "step": 5919
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3365172598707023e-05,
+      "loss": 0.0714,
+      "step": 5920
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3362034560345854e-05,
+      "loss": 0.0823,
+      "step": 5921
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3358896148680578e-05,
+      "loss": 0.0587,
+      "step": 5922
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3355757364059671e-05,
+      "loss": 0.0745,
+      "step": 5923
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3352618206831647e-05,
+      "loss": 0.0679,
+      "step": 5924
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3349478677345063e-05,
+      "loss": 0.0692,
+      "step": 5925
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3346338775948518e-05,
+      "loss": 0.0725,
+      "step": 5926
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3343198502990649e-05,
+      "loss": 0.0812,
+      "step": 5927
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3340057858820141e-05,
+      "loss": 0.0837,
+      "step": 5928
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3336916843785712e-05,
+      "loss": 0.0604,
+      "step": 5929
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3333775458236126e-05,
+      "loss": 0.0803,
+      "step": 5930
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.333063370252019e-05,
+      "loss": 0.0719,
+      "step": 5931
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3327491576986749e-05,
+      "loss": 0.0967,
+      "step": 5932
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3324349081984684e-05,
+      "loss": 0.0732,
+      "step": 5933
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.332120621786293e-05,
+      "loss": 0.0793,
+      "step": 5934
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.331806298497045e-05,
+      "loss": 0.0869,
+      "step": 5935
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3314919383656256e-05,
+      "loss": 0.0704,
+      "step": 5936
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3311775414269396e-05,
+      "loss": 0.0535,
+      "step": 5937
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3308631077158967e-05,
+      "loss": 0.0721,
+      "step": 5938
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3305486372674092e-05,
+      "loss": 0.0822,
+      "step": 5939
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3302341301163952e-05,
+      "loss": 0.0696,
+      "step": 5940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3299195862977756e-05,
+      "loss": 0.0609,
+      "step": 5941
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3296050058464761e-05,
+      "loss": 0.0607,
+      "step": 5942
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.329290388797426e-05,
+      "loss": 0.0703,
+      "step": 5943
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3289757351855587e-05,
+      "loss": 0.0773,
+      "step": 5944
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3286610450458123e-05,
+      "loss": 0.0583,
+      "step": 5945
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3283463184131283e-05,
+      "loss": 0.0704,
+      "step": 5946
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3280315553224521e-05,
+      "loss": 0.0592,
+      "step": 5947
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3277167558087338e-05,
+      "loss": 0.066,
+      "step": 5948
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3274019199069272e-05,
+      "loss": 0.0801,
+      "step": 5949
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.32708704765199e-05,
+      "loss": 0.075,
+      "step": 5950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3267721390788844e-05,
+      "loss": 0.0698,
+      "step": 5951
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.326457194222576e-05,
+      "loss": 0.0783,
+      "step": 5952
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3261422131180348e-05,
+      "loss": 0.0677,
+      "step": 5953
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.325827195800235e-05,
+      "loss": 0.0656,
+      "step": 5954
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3255121423041544e-05,
+      "loss": 0.084,
+      "step": 5955
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.325197052664775e-05,
+      "loss": 0.0675,
+      "step": 5956
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3248819269170831e-05,
+      "loss": 0.0678,
+      "step": 5957
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3245667650960686e-05,
+      "loss": 0.076,
+      "step": 5958
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3242515672367254e-05,
+      "loss": 0.066,
+      "step": 5959
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3239363333740517e-05,
+      "loss": 0.0811,
+      "step": 5960
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3236210635430493e-05,
+      "loss": 0.0711,
+      "step": 5961
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3233057577787248e-05,
+      "loss": 0.085,
+      "step": 5962
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3229904161160877e-05,
+      "loss": 0.0734,
+      "step": 5963
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3226750385901521e-05,
+      "loss": 0.0821,
+      "step": 5964
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3223596252359366e-05,
+      "loss": 0.0662,
+      "step": 5965
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3220441760884621e-05,
+      "loss": 0.0591,
+      "step": 5966
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3217286911827554e-05,
+      "loss": 0.0742,
+      "step": 5967
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3214131705538463e-05,
+      "loss": 0.077,
+      "step": 5968
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3210976142367684e-05,
+      "loss": 0.0637,
+      "step": 5969
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3207820222665594e-05,
+      "loss": 0.0742,
+      "step": 5970
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3204663946782618e-05,
+      "loss": 0.0729,
+      "step": 5971
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.320150731506921e-05,
+      "loss": 0.0645,
+      "step": 5972
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3198350327875866e-05,
+      "loss": 0.0706,
+      "step": 5973
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3195192985553128e-05,
+      "loss": 0.0759,
+      "step": 5974
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3192035288451562e-05,
+      "loss": 0.0685,
+      "step": 5975
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3188877236921789e-05,
+      "loss": 0.0589,
+      "step": 5976
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.3185718831314466e-05,
+      "loss": 0.0698,
+      "step": 5977
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3182560071980284e-05,
+      "loss": 0.0773,
+      "step": 5978
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.317940095926998e-05,
+      "loss": 0.0681,
+      "step": 5979
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3176241493534321e-05,
+      "loss": 0.0653,
+      "step": 5980
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3173081675124122e-05,
+      "loss": 0.0655,
+      "step": 5981
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3169921504390236e-05,
+      "loss": 0.075,
+      "step": 5982
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.316676098168355e-05,
+      "loss": 0.0633,
+      "step": 5983
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3163600107354992e-05,
+      "loss": 0.0687,
+      "step": 5984
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3160438881755536e-05,
+      "loss": 0.0775,
+      "step": 5985
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3157277305236185e-05,
+      "loss": 0.0798,
+      "step": 5986
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3154115378147984e-05,
+      "loss": 0.0848,
+      "step": 5987
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3150953100842023e-05,
+      "loss": 0.0713,
+      "step": 5988
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3147790473669422e-05,
+      "loss": 0.0688,
+      "step": 5989
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3144627496981344e-05,
+      "loss": 0.061,
+      "step": 5990
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3141464171128993e-05,
+      "loss": 0.0767,
+      "step": 5991
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3138300496463606e-05,
+      "loss": 0.0783,
+      "step": 5992
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3135136473336467e-05,
+      "loss": 0.0732,
+      "step": 5993
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3131972102098889e-05,
+      "loss": 0.0691,
+      "step": 5994
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.312880738310223e-05,
+      "loss": 0.0591,
+      "step": 5995
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3125642316697889e-05,
+      "loss": 0.0835,
+      "step": 5996
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.3122476903237296e-05,
+      "loss": 0.0715,
+      "step": 5997
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3119311143071922e-05,
+      "loss": 0.0718,
+      "step": 5998
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3116145036553283e-05,
+      "loss": 0.077,
+      "step": 5999
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3112978584032923e-05,
+      "loss": 0.0584,
+      "step": 6000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3109811785862428e-05,
+      "loss": 0.0771,
+      "step": 6001
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.310664464239343e-05,
+      "loss": 0.0621,
+      "step": 6002
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3103477153977593e-05,
+      "loss": 0.0693,
+      "step": 6003
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3100309320966613e-05,
+      "loss": 0.0646,
+      "step": 6004
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3097141143712237e-05,
+      "loss": 0.0715,
+      "step": 6005
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3093972622566245e-05,
+      "loss": 0.0832,
+      "step": 6006
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3090803757880448e-05,
+      "loss": 0.0725,
+      "step": 6007
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3087634550006707e-05,
+      "loss": 0.0632,
+      "step": 6008
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3084464999296916e-05,
+      "loss": 0.0819,
+      "step": 6009
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3081295106102999e-05,
+      "loss": 0.0725,
+      "step": 6010
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3078124870776934e-05,
+      "loss": 0.0627,
+      "step": 6011
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3074954293670726e-05,
+      "loss": 0.0837,
+      "step": 6012
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3071783375136417e-05,
+      "loss": 0.0684,
+      "step": 6013
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3068612115526099e-05,
+      "loss": 0.0727,
+      "step": 6014
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3065440515191883e-05,
+      "loss": 0.074,
+      "step": 6015
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.3062268574485937e-05,
+      "loss": 0.0789,
+      "step": 6016
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3059096293760454e-05,
+      "loss": 0.0817,
+      "step": 6017
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3055923673367667e-05,
+      "loss": 0.0787,
+      "step": 6018
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3052750713659852e-05,
+      "loss": 0.0692,
+      "step": 6019
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3049577414989318e-05,
+      "loss": 0.0581,
+      "step": 6020
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3046403777708409e-05,
+      "loss": 0.0584,
+      "step": 6021
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3043229802169519e-05,
+      "loss": 0.0731,
+      "step": 6022
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3040055488725064e-05,
+      "loss": 0.071,
+      "step": 6023
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3036880837727503e-05,
+      "loss": 0.0792,
+      "step": 6024
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3033705849529342e-05,
+      "loss": 0.0648,
+      "step": 6025
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.303053052448311e-05,
+      "loss": 0.0813,
+      "step": 6026
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3027354862941379e-05,
+      "loss": 0.0951,
+      "step": 6027
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3024178865256764e-05,
+      "loss": 0.0693,
+      "step": 6028
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3021002531781909e-05,
+      "loss": 0.0781,
+      "step": 6029
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.30178258628695e-05,
+      "loss": 0.0663,
+      "step": 6030
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.301464885887226e-05,
+      "loss": 0.0653,
+      "step": 6031
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3011471520142946e-05,
+      "loss": 0.0759,
+      "step": 6032
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3008293847034355e-05,
+      "loss": 0.0695,
+      "step": 6033
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.300511583989932e-05,
+      "loss": 0.0784,
+      "step": 6034
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.3001937499090716e-05,
+      "loss": 0.0927,
+      "step": 6035
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2998758824961442e-05,
+      "loss": 0.0692,
+      "step": 6036
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.299557981786445e-05,
+      "loss": 0.0579,
+      "step": 6037
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2992400478152717e-05,
+      "loss": 0.0759,
+      "step": 6038
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2989220806179268e-05,
+      "loss": 0.0853,
+      "step": 6039
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2986040802297153e-05,
+      "loss": 0.0601,
+      "step": 6040
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2982860466859464e-05,
+      "loss": 0.0762,
+      "step": 6041
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2979679800219333e-05,
+      "loss": 0.0811,
+      "step": 6042
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2976498802729927e-05,
+      "loss": 0.0819,
+      "step": 6043
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.297331747474444e-05,
+      "loss": 0.065,
+      "step": 6044
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2970135816616123e-05,
+      "loss": 0.0811,
+      "step": 6045
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2966953828698243e-05,
+      "loss": 0.061,
+      "step": 6046
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2963771511344119e-05,
+      "loss": 0.065,
+      "step": 6047
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2960588864907096e-05,
+      "loss": 0.0758,
+      "step": 6048
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.295740588974056e-05,
+      "loss": 0.0692,
+      "step": 6049
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2954222586197935e-05,
+      "loss": 0.069,
+      "step": 6050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.295103895463268e-05,
+      "loss": 0.0641,
+      "step": 6051
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2947854995398286e-05,
+      "loss": 0.0762,
+      "step": 6052
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.294467070884829e-05,
+      "loss": 0.0621,
+      "step": 6053
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2941486095336258e-05,
+      "loss": 0.059,
+      "step": 6054
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.293830115521579e-05,
+      "loss": 0.0739,
+      "step": 6055
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2935115888840532e-05,
+      "loss": 0.0716,
+      "step": 6056
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2931930296564159e-05,
+      "loss": 0.066,
+      "step": 6057
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2928744378740382e-05,
+      "loss": 0.0664,
+      "step": 6058
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2925558135722951e-05,
+      "loss": 0.0627,
+      "step": 6059
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2922371567865652e-05,
+      "loss": 0.065,
+      "step": 6060
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2919184675522305e-05,
+      "loss": 0.0727,
+      "step": 6061
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2915997459046768e-05,
+      "loss": 0.0846,
+      "step": 6062
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2912809918792934e-05,
+      "loss": 0.0717,
+      "step": 6063
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.290962205511473e-05,
+      "loss": 0.0782,
+      "step": 6064
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2906433868366123e-05,
+      "loss": 0.0813,
+      "step": 6065
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2903245358901118e-05,
+      "loss": 0.0758,
+      "step": 6066
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2900056527073743e-05,
+      "loss": 0.0836,
+      "step": 6067
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2896867373238075e-05,
+      "loss": 0.0696,
+      "step": 6068
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2893677897748222e-05,
+      "loss": 0.066,
+      "step": 6069
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.289048810095833e-05,
+      "loss": 0.0629,
+      "step": 6070
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2887297983222577e-05,
+      "loss": 0.0763,
+      "step": 6071
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2884107544895175e-05,
+      "loss": 0.0633,
+      "step": 6072
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2880916786330381e-05,
+      "loss": 0.0558,
+      "step": 6073
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2877725707882478e-05,
+      "loss": 0.0596,
+      "step": 6074
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2874534309905789e-05,
+      "loss": 0.0782,
+      "step": 6075
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.287134259275467e-05,
+      "loss": 0.0692,
+      "step": 6076
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2868150556783514e-05,
+      "loss": 0.0671,
+      "step": 6077
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2864958202346748e-05,
+      "loss": 0.066,
+      "step": 6078
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2861765529798843e-05,
+      "loss": 0.064,
+      "step": 6079
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2858572539494289e-05,
+      "loss": 0.0808,
+      "step": 6080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2855379231787623e-05,
+      "loss": 0.0617,
+      "step": 6081
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2852185607033418e-05,
+      "loss": 0.0695,
+      "step": 6082
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2848991665586276e-05,
+      "loss": 0.0812,
+      "step": 6083
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2845797407800834e-05,
+      "loss": 0.067,
+      "step": 6084
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2842602834031773e-05,
+      "loss": 0.0583,
+      "step": 6085
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.28394079446338e-05,
+      "loss": 0.0897,
+      "step": 6086
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2836212739961657e-05,
+      "loss": 0.0599,
+      "step": 6087
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.283301722037013e-05,
+      "loss": 0.064,
+      "step": 6088
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.282982138621403e-05,
+      "loss": 0.0753,
+      "step": 6089
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2826625237848207e-05,
+      "loss": 0.0725,
+      "step": 6090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2823428775627551e-05,
+      "loss": 0.0634,
+      "step": 6091
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2820231999906974e-05,
+      "loss": 0.0671,
+      "step": 6092
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2817034911041436e-05,
+      "loss": 0.0748,
+      "step": 6093
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2813837509385926e-05,
+      "loss": 0.0728,
+      "step": 6094
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2810639795295466e-05,
+      "loss": 0.0839,
+      "step": 6095
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2807441769125116e-05,
+      "loss": 0.0786,
+      "step": 6096
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2804243431229968e-05,
+      "loss": 0.0635,
+      "step": 6097
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.280104478196515e-05,
+      "loss": 0.085,
+      "step": 6098
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2797845821685828e-05,
+      "loss": 0.0617,
+      "step": 6099
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2794646550747196e-05,
+      "loss": 0.0571,
+      "step": 6100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2791446969504486e-05,
+      "loss": 0.0627,
+      "step": 6101
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2788247078312966e-05,
+      "loss": 0.078,
+      "step": 6102
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2785046877527934e-05,
+      "loss": 0.0642,
+      "step": 6103
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2781846367504725e-05,
+      "loss": 0.0834,
+      "step": 6104
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.277864554859871e-05,
+      "loss": 0.0562,
+      "step": 6105
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.277544442116529e-05,
+      "loss": 0.0579,
+      "step": 6106
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2772242985559906e-05,
+      "loss": 0.0706,
+      "step": 6107
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2769041242138027e-05,
+      "loss": 0.0623,
+      "step": 6108
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2765839191255163e-05,
+      "loss": 0.0702,
+      "step": 6109
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.276263683326685e-05,
+      "loss": 0.0551,
+      "step": 6110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2759434168528663e-05,
+      "loss": 0.0861,
+      "step": 6111
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2756231197396214e-05,
+      "loss": 0.0657,
+      "step": 6112
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.275302792022514e-05,
+      "loss": 0.0885,
+      "step": 6113
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2749824337371124e-05,
+      "loss": 0.083,
+      "step": 6114
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.274662044918987e-05,
+      "loss": 0.08,
+      "step": 6115
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2743416256037126e-05,
+      "loss": 0.0747,
+      "step": 6116
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.274021175826867e-05,
+      "loss": 0.0613,
+      "step": 6117
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2737006956240312e-05,
+      "loss": 0.0731,
+      "step": 6118
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2733801850307898e-05,
+      "loss": 0.069,
+      "step": 6119
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2730596440827312e-05,
+      "loss": 0.0686,
+      "step": 6120
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.272739072815446e-05,
+      "loss": 0.0814,
+      "step": 6121
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2724184712645296e-05,
+      "loss": 0.0663,
+      "step": 6122
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2720978394655793e-05,
+      "loss": 0.0778,
+      "step": 6123
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2717771774541973e-05,
+      "loss": 0.0563,
+      "step": 6124
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.271456485265988e-05,
+      "loss": 0.0719,
+      "step": 6125
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2711357629365592e-05,
+      "loss": 0.0595,
+      "step": 6126
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2708150105015229e-05,
+      "loss": 0.0771,
+      "step": 6127
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2704942279964936e-05,
+      "loss": 0.0603,
+      "step": 6128
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2701734154570894e-05,
+      "loss": 0.0764,
+      "step": 6129
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2698525729189321e-05,
+      "loss": 0.0773,
+      "step": 6130
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2695317004176464e-05,
+      "loss": 0.0752,
+      "step": 6131
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2692107979888599e-05,
+      "loss": 0.073,
+      "step": 6132
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2688898656682048e-05,
+      "loss": 0.0612,
+      "step": 6133
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2685689034913156e-05,
+      "loss": 0.0848,
+      "step": 6134
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.26824791149383e-05,
+      "loss": 0.0731,
+      "step": 6135
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.26792688971139e-05,
+      "loss": 0.0653,
+      "step": 6136
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2676058381796402e-05,
+      "loss": 0.0634,
+      "step": 6137
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2672847569342284e-05,
+      "loss": 0.0659,
+      "step": 6138
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.266963646010806e-05,
+      "loss": 0.0636,
+      "step": 6139
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2666425054450276e-05,
+      "loss": 0.0676,
+      "step": 6140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2663213352725507e-05,
+      "loss": 0.0732,
+      "step": 6141
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2660001355290374e-05,
+      "loss": 0.0847,
+      "step": 6142
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2656789062501515e-05,
+      "loss": 0.0617,
+      "step": 6143
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.265357647471561e-05,
+      "loss": 0.0733,
+      "step": 6144
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2650363592289368e-05,
+      "loss": 0.0664,
+      "step": 6145
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2647150415579532e-05,
+      "loss": 0.084,
+      "step": 6146
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2643936944942878e-05,
+      "loss": 0.0874,
+      "step": 6147
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2640723180736216e-05,
+      "loss": 0.0923,
+      "step": 6148
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2637509123316383e-05,
+      "loss": 0.0715,
+      "step": 6149
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2634294773040256e-05,
+      "loss": 0.0643,
+      "step": 6150
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2631080130264742e-05,
+      "loss": 0.0748,
+      "step": 6151
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.2627865195346771e-05,
+      "loss": 0.0842,
+      "step": 6152
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2624649968643323e-05,
+      "loss": 0.0609,
+      "step": 6153
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.26214344505114e-05,
+      "loss": 0.0751,
+      "step": 6154
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2618218641308033e-05,
+      "loss": 0.0554,
+      "step": 6155
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2615002541390293e-05,
+      "loss": 0.07,
+      "step": 6156
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2611786151115282e-05,
+      "loss": 0.0648,
+      "step": 6157
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2608569470840128e-05,
+      "loss": 0.0642,
+      "step": 6158
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2605352500921998e-05,
+      "loss": 0.0836,
+      "step": 6159
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2602135241718094e-05,
+      "loss": 0.0811,
+      "step": 6160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2598917693585635e-05,
+      "loss": 0.0654,
+      "step": 6161
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2595699856881888e-05,
+      "loss": 0.0676,
+      "step": 6162
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2592481731964146e-05,
+      "loss": 0.0691,
+      "step": 6163
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2589263319189731e-05,
+      "loss": 0.0782,
+      "step": 6164
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2586044618916002e-05,
+      "loss": 0.0659,
+      "step": 6165
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2582825631500351e-05,
+      "loss": 0.0656,
+      "step": 6166
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2579606357300193e-05,
+      "loss": 0.0712,
+      "step": 6167
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2576386796672985e-05,
+      "loss": 0.0649,
+      "step": 6168
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.257316694997621e-05,
+      "loss": 0.0657,
+      "step": 6169
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2569946817567384e-05,
+      "loss": 0.058,
+      "step": 6170
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2566726399804056e-05,
+      "loss": 0.078,
+      "step": 6171
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.2563505697043805e-05,
+      "loss": 0.0701,
+      "step": 6172
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.256028470964424e-05,
+      "loss": 0.0584,
+      "step": 6173
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2557063437963011e-05,
+      "loss": 0.0723,
+      "step": 6174
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2553841882357783e-05,
+      "loss": 0.0569,
+      "step": 6175
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.255062004318627e-05,
+      "loss": 0.066,
+      "step": 6176
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2547397920806203e-05,
+      "loss": 0.0825,
+      "step": 6177
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2544175515575356e-05,
+      "loss": 0.0815,
+      "step": 6178
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.254095282785153e-05,
+      "loss": 0.0684,
+      "step": 6179
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2537729857992549e-05,
+      "loss": 0.0736,
+      "step": 6180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2534506606356282e-05,
+      "loss": 0.0726,
+      "step": 6181
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2531283073300627e-05,
+      "loss": 0.0696,
+      "step": 6182
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2528059259183499e-05,
+      "loss": 0.072,
+      "step": 6183
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2524835164362866e-05,
+      "loss": 0.0714,
+      "step": 6184
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2521610789196707e-05,
+      "loss": 0.0566,
+      "step": 6185
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2518386134043046e-05,
+      "loss": 0.0728,
+      "step": 6186
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.251516119925993e-05,
+      "loss": 0.0746,
+      "step": 6187
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2511935985205446e-05,
+      "loss": 0.0686,
+      "step": 6188
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2508710492237698e-05,
+      "loss": 0.0689,
+      "step": 6189
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2505484720714838e-05,
+      "loss": 0.0941,
+      "step": 6190
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2502258670995036e-05,
+      "loss": 0.0746,
+      "step": 6191
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2499032343436492e-05,
+      "loss": 0.0641,
+      "step": 6192
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.249580573839745e-05,
+      "loss": 0.0687,
+      "step": 6193
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2492578856236173e-05,
+      "loss": 0.0722,
+      "step": 6194
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2489351697310957e-05,
+      "loss": 0.0806,
+      "step": 6195
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2486124261980134e-05,
+      "loss": 0.0697,
+      "step": 6196
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2482896550602057e-05,
+      "loss": 0.0717,
+      "step": 6197
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2479668563535124e-05,
+      "loss": 0.0691,
+      "step": 6198
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2476440301137749e-05,
+      "loss": 0.0804,
+      "step": 6199
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.247321176376838e-05,
+      "loss": 0.074,
+      "step": 6200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2469982951785507e-05,
+      "loss": 0.0764,
+      "step": 6201
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2466753865547636e-05,
+      "loss": 0.0653,
+      "step": 6202
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2463524505413308e-05,
+      "loss": 0.0727,
+      "step": 6203
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2460294871741102e-05,
+      "loss": 0.0929,
+      "step": 6204
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2457064964889614e-05,
+      "loss": 0.0741,
+      "step": 6205
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2453834785217478e-05,
+      "loss": 0.0685,
+      "step": 6206
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.245060433308336e-05,
+      "loss": 0.0913,
+      "step": 6207
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2447373608845956e-05,
+      "loss": 0.0739,
+      "step": 6208
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2444142612863985e-05,
+      "loss": 0.0698,
+      "step": 6209
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2440911345496204e-05,
+      "loss": 0.0617,
+      "step": 6210
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.2437679807101399e-05,
+      "loss": 0.0684,
+      "step": 6211
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2434447998038379e-05,
+      "loss": 0.0649,
+      "step": 6212
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2431215918665995e-05,
+      "loss": 0.0721,
+      "step": 6213
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2427983569343118e-05,
+      "loss": 0.0978,
+      "step": 6214
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.242475095042865e-05,
+      "loss": 0.0755,
+      "step": 6215
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2421518062281532e-05,
+      "loss": 0.0779,
+      "step": 6216
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2418284905260722e-05,
+      "loss": 0.0689,
+      "step": 6217
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2415051479725218e-05,
+      "loss": 0.072,
+      "step": 6218
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.241181778603404e-05,
+      "loss": 0.0721,
+      "step": 6219
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2408583824546247e-05,
+      "loss": 0.0658,
+      "step": 6220
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2405349595620921e-05,
+      "loss": 0.0734,
+      "step": 6221
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2402115099617174e-05,
+      "loss": 0.0695,
+      "step": 6222
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2398880336894146e-05,
+      "loss": 0.0763,
+      "step": 6223
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2395645307811014e-05,
+      "loss": 0.0681,
+      "step": 6224
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2392410012726981e-05,
+      "loss": 0.0663,
+      "step": 6225
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.238917445200127e-05,
+      "loss": 0.0662,
+      "step": 6226
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2385938625993154e-05,
+      "loss": 0.0627,
+      "step": 6227
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2382702535061917e-05,
+      "loss": 0.0604,
+      "step": 6228
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2379466179566878e-05,
+      "loss": 0.0543,
+      "step": 6229
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.2376229559867388e-05,
+      "loss": 0.0735,
+      "step": 6230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2372992676322826e-05,
+      "loss": 0.0847,
+      "step": 6231
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2369755529292598e-05,
+      "loss": 0.0655,
+      "step": 6232
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2366518119136143e-05,
+      "loss": 0.0778,
+      "step": 6233
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2363280446212929e-05,
+      "loss": 0.0656,
+      "step": 6234
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2360042510882449e-05,
+      "loss": 0.0689,
+      "step": 6235
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2356804313504227e-05,
+      "loss": 0.0839,
+      "step": 6236
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2353565854437818e-05,
+      "loss": 0.0623,
+      "step": 6237
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2350327134042806e-05,
+      "loss": 0.0763,
+      "step": 6238
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2347088152678803e-05,
+      "loss": 0.0741,
+      "step": 6239
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2343848910705449e-05,
+      "loss": 0.0726,
+      "step": 6240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2340609408482411e-05,
+      "loss": 0.0876,
+      "step": 6241
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2337369646369392e-05,
+      "loss": 0.0698,
+      "step": 6242
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2334129624726118e-05,
+      "loss": 0.0793,
+      "step": 6243
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2330889343912347e-05,
+      "loss": 0.0673,
+      "step": 6244
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2327648804287862e-05,
+      "loss": 0.0959,
+      "step": 6245
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2324408006212475e-05,
+      "loss": 0.0609,
+      "step": 6246
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2321166950046033e-05,
+      "loss": 0.0733,
+      "step": 6247
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2317925636148404e-05,
+      "loss": 0.0713,
+      "step": 6248
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.231468406487949e-05,
+      "loss": 0.0669,
+      "step": 6249
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2311442236599218e-05,
+      "loss": 0.0545,
+      "step": 6250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2308200151667545e-05,
+      "loss": 0.0758,
+      "step": 6251
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2304957810444455e-05,
+      "loss": 0.0793,
+      "step": 6252
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.230171521328997e-05,
+      "loss": 0.0918,
+      "step": 6253
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.229847236056412e-05,
+      "loss": 0.0767,
+      "step": 6254
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2295229252626985e-05,
+      "loss": 0.0622,
+      "step": 6255
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.229198588983866e-05,
+      "loss": 0.0732,
+      "step": 6256
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2288742272559273e-05,
+      "loss": 0.0663,
+      "step": 6257
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2285498401148984e-05,
+      "loss": 0.0587,
+      "step": 6258
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.228225427596797e-05,
+      "loss": 0.0715,
+      "step": 6259
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2279009897376444e-05,
+      "loss": 0.0809,
+      "step": 6260
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2275765265734649e-05,
+      "loss": 0.0713,
+      "step": 6261
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2272520381402856e-05,
+      "loss": 0.0706,
+      "step": 6262
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2269275244741352e-05,
+      "loss": 0.0795,
+      "step": 6263
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.226602985611047e-05,
+      "loss": 0.0709,
+      "step": 6264
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2262784215870562e-05,
+      "loss": 0.0711,
+      "step": 6265
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2259538324382001e-05,
+      "loss": 0.0792,
+      "step": 6266
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2256292182005201e-05,
+      "loss": 0.0661,
+      "step": 6267
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2253045789100597e-05,
+      "loss": 0.0631,
+      "step": 6268
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.2249799146028651e-05,
+      "loss": 0.0641,
+      "step": 6269
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.224655225314986e-05,
+      "loss": 0.0655,
+      "step": 6270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2243305110824738e-05,
+      "loss": 0.0796,
+      "step": 6271
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.224005771941383e-05,
+      "loss": 0.0724,
+      "step": 6272
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2236810079277717e-05,
+      "loss": 0.087,
+      "step": 6273
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2233562190776996e-05,
+      "loss": 0.072,
+      "step": 6274
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2230314054272302e-05,
+      "loss": 0.0748,
+      "step": 6275
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2227065670124288e-05,
+      "loss": 0.0614,
+      "step": 6276
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2223817038693639e-05,
+      "loss": 0.075,
+      "step": 6277
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2220568160341072e-05,
+      "loss": 0.0793,
+      "step": 6278
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2217319035427323e-05,
+      "loss": 0.0566,
+      "step": 6279
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2214069664313159e-05,
+      "loss": 0.0693,
+      "step": 6280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2210820047359377e-05,
+      "loss": 0.064,
+      "step": 6281
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2207570184926797e-05,
+      "loss": 0.0683,
+      "step": 6282
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2204320077376266e-05,
+      "loss": 0.0685,
+      "step": 6283
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2201069725068666e-05,
+      "loss": 0.0707,
+      "step": 6284
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2197819128364894e-05,
+      "loss": 0.063,
+      "step": 6285
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2194568287625886e-05,
+      "loss": 0.0754,
+      "step": 6286
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.2191317203212595e-05,
+      "loss": 0.0831,
+      "step": 6287
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.218806587548601e-05,
+      "loss": 0.0847,
+      "step": 6288
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2184814304807138e-05,
+      "loss": 0.0731,
+      "step": 6289
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2181562491537022e-05,
+      "loss": 0.0754,
+      "step": 6290
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2178310436036726e-05,
+      "loss": 0.0609,
+      "step": 6291
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2175058138667343e-05,
+      "loss": 0.0742,
+      "step": 6292
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2171805599789993e-05,
+      "loss": 0.0729,
+      "step": 6293
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.216855281976582e-05,
+      "loss": 0.0601,
+      "step": 6294
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2165299798955998e-05,
+      "loss": 0.0726,
+      "step": 6295
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2162046537721728e-05,
+      "loss": 0.0701,
+      "step": 6296
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2158793036424235e-05,
+      "loss": 0.0646,
+      "step": 6297
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2155539295424774e-05,
+      "loss": 0.0656,
+      "step": 6298
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2152285315084623e-05,
+      "loss": 0.0768,
+      "step": 6299
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2149031095765087e-05,
+      "loss": 0.0822,
+      "step": 6300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2145776637827503e-05,
+      "loss": 0.082,
+      "step": 6301
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2142521941633226e-05,
+      "loss": 0.0632,
+      "step": 6302
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2139267007543642e-05,
+      "loss": 0.0586,
+      "step": 6303
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2136011835920167e-05,
+      "loss": 0.0859,
+      "step": 6304
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2132756427124237e-05,
+      "loss": 0.0723,
+      "step": 6305
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2129500781517317e-05,
+      "loss": 0.0827,
+      "step": 6306
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2126244899460896e-05,
+      "loss": 0.0587,
+      "step": 6307
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.2122988781316499e-05,
+      "loss": 0.0699,
+      "step": 6308
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.211973242744566e-05,
+      "loss": 0.0692,
+      "step": 6309
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2116475838209956e-05,
+      "loss": 0.0776,
+      "step": 6310
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2113219013970978e-05,
+      "loss": 0.0743,
+      "step": 6311
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2109961955090354e-05,
+      "loss": 0.0725,
+      "step": 6312
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2106704661929727e-05,
+      "loss": 0.0701,
+      "step": 6313
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.210344713485077e-05,
+      "loss": 0.0757,
+      "step": 6314
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2100189374215188e-05,
+      "loss": 0.0669,
+      "step": 6315
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2096931380384705e-05,
+      "loss": 0.0582,
+      "step": 6316
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2093673153721071e-05,
+      "loss": 0.0759,
+      "step": 6317
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2090414694586065e-05,
+      "loss": 0.0908,
+      "step": 6318
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2087156003341494e-05,
+      "loss": 0.0702,
+      "step": 6319
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2083897080349181e-05,
+      "loss": 0.0709,
+      "step": 6320
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2080637925970985e-05,
+      "loss": 0.0677,
+      "step": 6321
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2077378540568789e-05,
+      "loss": 0.0649,
+      "step": 6322
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2074118924504493e-05,
+      "loss": 0.0495,
+      "step": 6323
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2070859078140034e-05,
+      "loss": 0.0569,
+      "step": 6324
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2067599001837369e-05,
+      "loss": 0.075,
+      "step": 6325
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.206433869595848e-05,
+      "loss": 0.0717,
+      "step": 6326
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2061078160865379e-05,
+      "loss": 0.0734,
+      "step": 6327
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2057817396920091e-05,
+      "loss": 0.0839,
+      "step": 6328
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2054556404484688e-05,
+      "loss": 0.0648,
+      "step": 6329
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2051295183921245e-05,
+      "loss": 0.0694,
+      "step": 6330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2048033735591878e-05,
+      "loss": 0.0953,
+      "step": 6331
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2044772059858718e-05,
+      "loss": 0.0869,
+      "step": 6332
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2041510157083931e-05,
+      "loss": 0.0796,
+      "step": 6333
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2038248027629695e-05,
+      "loss": 0.0784,
+      "step": 6334
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2034985671858233e-05,
+      "loss": 0.0589,
+      "step": 6335
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2031723090131771e-05,
+      "loss": 0.0632,
+      "step": 6336
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.202846028281257e-05,
+      "loss": 0.0483,
+      "step": 6337
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2025197250262926e-05,
+      "loss": 0.0734,
+      "step": 6338
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2021933992845142e-05,
+      "loss": 0.0663,
+      "step": 6339
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2018670510921558e-05,
+      "loss": 0.0552,
+      "step": 6340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2015406804854535e-05,
+      "loss": 0.0626,
+      "step": 6341
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2012142875006456e-05,
+      "loss": 0.0641,
+      "step": 6342
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2008878721739735e-05,
+      "loss": 0.0847,
+      "step": 6343
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2005614345416808e-05,
+      "loss": 0.0673,
+      "step": 6344
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.2002349746400135e-05,
+      "loss": 0.0698,
+      "step": 6345
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.19990849250522e-05,
+      "loss": 0.0743,
+      "step": 6346
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.1995819881735516e-05,
+      "loss": 0.0563,
+      "step": 6347
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1992554616812613e-05,
+      "loss": 0.0764,
+      "step": 6348
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1989289130646056e-05,
+      "loss": 0.0785,
+      "step": 6349
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1986023423598425e-05,
+      "loss": 0.0604,
+      "step": 6350
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1982757496032329e-05,
+      "loss": 0.0663,
+      "step": 6351
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1979491348310402e-05,
+      "loss": 0.0775,
+      "step": 6352
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.19762249807953e-05,
+      "loss": 0.0902,
+      "step": 6353
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1972958393849706e-05,
+      "loss": 0.0726,
+      "step": 6354
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1969691587836322e-05,
+      "loss": 0.0511,
+      "step": 6355
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1966424563117886e-05,
+      "loss": 0.0812,
+      "step": 6356
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1963157320057145e-05,
+      "loss": 0.0769,
+      "step": 6357
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1959889859016885e-05,
+      "loss": 0.0659,
+      "step": 6358
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1956622180359905e-05,
+      "loss": 0.0665,
+      "step": 6359
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1953354284449028e-05,
+      "loss": 0.069,
+      "step": 6360
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1950086171647115e-05,
+      "loss": 0.0649,
+      "step": 6361
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1946817842317036e-05,
+      "loss": 0.0861,
+      "step": 6362
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1943549296821686e-05,
+      "loss": 0.0811,
+      "step": 6363
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1940280535523998e-05,
+      "loss": 0.0616,
+      "step": 6364
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1937011558786915e-05,
+      "loss": 0.0627,
+      "step": 6365
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.1933742366973406e-05,
+      "loss": 0.0728,
+      "step": 6366
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1930472960446473e-05,
+      "loss": 0.0665,
+      "step": 6367
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1927203339569128e-05,
+      "loss": 0.0734,
+      "step": 6368
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1923933504704417e-05,
+      "loss": 0.0873,
+      "step": 6369
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.192066345621541e-05,
+      "loss": 0.0668,
+      "step": 6370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1917393194465188e-05,
+      "loss": 0.0586,
+      "step": 6371
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1914122719816874e-05,
+      "loss": 0.0754,
+      "step": 6372
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1910852032633605e-05,
+      "loss": 0.0731,
+      "step": 6373
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1907581133278536e-05,
+      "loss": 0.0753,
+      "step": 6374
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1904310022114857e-05,
+      "loss": 0.083,
+      "step": 6375
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1901038699505779e-05,
+      "loss": 0.0718,
+      "step": 6376
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1897767165814527e-05,
+      "loss": 0.0697,
+      "step": 6377
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1894495421404364e-05,
+      "loss": 0.0706,
+      "step": 6378
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1891223466638557e-05,
+      "loss": 0.0813,
+      "step": 6379
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1887951301880421e-05,
+      "loss": 0.0797,
+      "step": 6380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1884678927493276e-05,
+      "loss": 0.0767,
+      "step": 6381
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1881406343840468e-05,
+      "loss": 0.071,
+      "step": 6382
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.187813355128537e-05,
+      "loss": 0.0647,
+      "step": 6383
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1874860550191385e-05,
+      "loss": 0.0624,
+      "step": 6384
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1871587340921918e-05,
+      "loss": 0.0641,
+      "step": 6385
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.1868313923840423e-05,
+      "loss": 0.0743,
+      "step": 6386
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1865040299310356e-05,
+      "loss": 0.0728,
+      "step": 6387
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1861766467695204e-05,
+      "loss": 0.0498,
+      "step": 6388
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1858492429358483e-05,
+      "loss": 0.0645,
+      "step": 6389
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1855218184663725e-05,
+      "loss": 0.0775,
+      "step": 6390
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1851943733974484e-05,
+      "loss": 0.0657,
+      "step": 6391
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1848669077654342e-05,
+      "loss": 0.0668,
+      "step": 6392
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.18453942160669e-05,
+      "loss": 0.0796,
+      "step": 6393
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1842119149575781e-05,
+      "loss": 0.0714,
+      "step": 6394
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1838843878544635e-05,
+      "loss": 0.1007,
+      "step": 6395
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1835568403337131e-05,
+      "loss": 0.0887,
+      "step": 6396
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.183229272431696e-05,
+      "loss": 0.0468,
+      "step": 6397
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1829016841847845e-05,
+      "loss": 0.061,
+      "step": 6398
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1825740756293515e-05,
+      "loss": 0.0778,
+      "step": 6399
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1822464468017735e-05,
+      "loss": 0.0745,
+      "step": 6400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1819187977384292e-05,
+      "loss": 0.0606,
+      "step": 6401
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1815911284756983e-05,
+      "loss": 0.0609,
+      "step": 6402
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1812634390499645e-05,
+      "loss": 0.073,
+      "step": 6403
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.1809357294976122e-05,
+      "loss": 0.0689,
+      "step": 6404
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.180607999855029e-05,
+      "loss": 0.0715,
+      "step": 6405
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1802802501586044e-05,
+      "loss": 0.0695,
+      "step": 6406
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1799524804447302e-05,
+      "loss": 0.0572,
+      "step": 6407
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1796246907498001e-05,
+      "loss": 0.054,
+      "step": 6408
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1792968811102106e-05,
+      "loss": 0.0759,
+      "step": 6409
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.17896905156236e-05,
+      "loss": 0.0638,
+      "step": 6410
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1786412021426485e-05,
+      "loss": 0.0705,
+      "step": 6411
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1783133328874797e-05,
+      "loss": 0.0679,
+      "step": 6412
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1779854438332584e-05,
+      "loss": 0.0783,
+      "step": 6413
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.177657535016391e-05,
+      "loss": 0.0526,
+      "step": 6414
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.177329606473288e-05,
+      "loss": 0.0802,
+      "step": 6415
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1770016582403605e-05,
+      "loss": 0.0748,
+      "step": 6416
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1766736903540222e-05,
+      "loss": 0.0636,
+      "step": 6417
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1763457028506893e-05,
+      "loss": 0.0747,
+      "step": 6418
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1760176957667799e-05,
+      "loss": 0.0624,
+      "step": 6419
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1756896691387141e-05,
+      "loss": 0.0654,
+      "step": 6420
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.175361623002915e-05,
+      "loss": 0.0754,
+      "step": 6421
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1750335573958066e-05,
+      "loss": 0.049,
+      "step": 6422
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1747054723538158e-05,
+      "loss": 0.0747,
+      "step": 6423
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.1743773679133721e-05,
+      "loss": 0.0883,
+      "step": 6424
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1740492441109059e-05,
+      "loss": 0.0608,
+      "step": 6425
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1737211009828513e-05,
+      "loss": 0.072,
+      "step": 6426
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.173392938565643e-05,
+      "loss": 0.0775,
+      "step": 6427
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.173064756895719e-05,
+      "loss": 0.0747,
+      "step": 6428
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.172736556009519e-05,
+      "loss": 0.0709,
+      "step": 6429
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1724083359434845e-05,
+      "loss": 0.0697,
+      "step": 6430
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1720800967340597e-05,
+      "loss": 0.0763,
+      "step": 6431
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.171751838417691e-05,
+      "loss": 0.052,
+      "step": 6432
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1714235610308261e-05,
+      "loss": 0.0721,
+      "step": 6433
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1710952646099157e-05,
+      "loss": 0.0917,
+      "step": 6434
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.170766949191412e-05,
+      "loss": 0.075,
+      "step": 6435
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.17043861481177e-05,
+      "loss": 0.0675,
+      "step": 6436
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1701102615074453e-05,
+      "loss": 0.0808,
+      "step": 6437
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.169781889314898e-05,
+      "loss": 0.0743,
+      "step": 6438
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1694534982705883e-05,
+      "loss": 0.0681,
+      "step": 6439
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1691250884109791e-05,
+      "loss": 0.0773,
+      "step": 6440
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1687966597725358e-05,
+      "loss": 0.0801,
+      "step": 6441
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.168468212391725e-05,
+      "loss": 0.0682,
+      "step": 6442
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.1681397463050164e-05,
+      "loss": 0.0629,
+      "step": 6443
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.167811261548881e-05,
+      "loss": 0.0667,
+      "step": 6444
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1674827581597925e-05,
+      "loss": 0.0645,
+      "step": 6445
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.167154236174226e-05,
+      "loss": 0.0784,
+      "step": 6446
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1668256956286595e-05,
+      "loss": 0.0774,
+      "step": 6447
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1664971365595714e-05,
+      "loss": 0.0693,
+      "step": 6448
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1661685590034446e-05,
+      "loss": 0.0725,
+      "step": 6449
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1658399629967621e-05,
+      "loss": 0.0683,
+      "step": 6450
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1655113485760097e-05,
+      "loss": 0.077,
+      "step": 6451
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1651827157776754e-05,
+      "loss": 0.0845,
+      "step": 6452
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1648540646382487e-05,
+      "loss": 0.0732,
+      "step": 6453
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1645253951942215e-05,
+      "loss": 0.0617,
+      "step": 6454
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.164196707482088e-05,
+      "loss": 0.078,
+      "step": 6455
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1638680015383434e-05,
+      "loss": 0.0765,
+      "step": 6456
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1635392773994865e-05,
+      "loss": 0.067,
+      "step": 6457
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1632105351020166e-05,
+      "loss": 0.0729,
+      "step": 6458
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1628817746824356e-05,
+      "loss": 0.067,
+      "step": 6459
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1625529961772481e-05,
+      "loss": 0.0678,
+      "step": 6460
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1622241996229597e-05,
+      "loss": 0.0775,
+      "step": 6461
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.1618953850560783e-05,
+      "loss": 0.0721,
+      "step": 6462
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.161566552513114e-05,
+      "loss": 0.0604,
+      "step": 6463
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.161237702030579e-05,
+      "loss": 0.075,
+      "step": 6464
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1609088336449865e-05,
+      "loss": 0.0809,
+      "step": 6465
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1605799473928534e-05,
+      "loss": 0.0698,
+      "step": 6466
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1602510433106972e-05,
+      "loss": 0.0785,
+      "step": 6467
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1599221214350376e-05,
+      "loss": 0.0815,
+      "step": 6468
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1595931818023971e-05,
+      "loss": 0.0642,
+      "step": 6469
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.159264224449299e-05,
+      "loss": 0.0631,
+      "step": 6470
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1589352494122694e-05,
+      "loss": 0.0837,
+      "step": 6471
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1586062567278361e-05,
+      "loss": 0.068,
+      "step": 6472
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1582772464325292e-05,
+      "loss": 0.0693,
+      "step": 6473
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1579482185628794e-05,
+      "loss": 0.0724,
+      "step": 6474
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1576191731554214e-05,
+      "loss": 0.065,
+      "step": 6475
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1572901102466902e-05,
+      "loss": 0.077,
+      "step": 6476
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1569610298732236e-05,
+      "loss": 0.0692,
+      "step": 6477
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1566319320715614e-05,
+      "loss": 0.063,
+      "step": 6478
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1563028168782444e-05,
+      "loss": 0.0687,
+      "step": 6479
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1559736843298164e-05,
+      "loss": 0.0796,
+      "step": 6480
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1556445344628226e-05,
+      "loss": 0.059,
+      "step": 6481
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1553153673138103e-05,
+      "loss": 0.0692,
+      "step": 6482
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.1549861829193282e-05,
+      "loss": 0.0752,
+      "step": 6483
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.154656981315928e-05,
+      "loss": 0.0668,
+      "step": 6484
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.154327762540162e-05,
+      "loss": 0.0618,
+      "step": 6485
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1539985266285858e-05,
+      "loss": 0.0674,
+      "step": 6486
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1536692736177555e-05,
+      "loss": 0.0798,
+      "step": 6487
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.15334000354423e-05,
+      "loss": 0.0594,
+      "step": 6488
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1530107164445701e-05,
+      "loss": 0.0682,
+      "step": 6489
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.152681412355338e-05,
+      "loss": 0.0533,
+      "step": 6490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1523520913130979e-05,
+      "loss": 0.0863,
+      "step": 6491
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1520227533544164e-05,
+      "loss": 0.0742,
+      "step": 6492
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1516933985158612e-05,
+      "loss": 0.0679,
+      "step": 6493
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1513640268340025e-05,
+      "loss": 0.0635,
+      "step": 6494
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1510346383454122e-05,
+      "loss": 0.0922,
+      "step": 6495
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.150705233086664e-05,
+      "loss": 0.0815,
+      "step": 6496
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.150375811094333e-05,
+      "loss": 0.0803,
+      "step": 6497
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1500463724049975e-05,
+      "loss": 0.0688,
+      "step": 6498
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.149716917055236e-05,
+      "loss": 0.0751,
+      "step": 6499
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1493874450816302e-05,
+      "loss": 0.0706,
+      "step": 6500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1490579565207629e-05,
+      "loss": 0.0846,
+      "step": 6501
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1487284514092186e-05,
+      "loss": 0.0823,
+      "step": 6502
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1483989297835843e-05,
+      "loss": 0.0718,
+      "step": 6503
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1480693916804484e-05,
+      "loss": 0.0679,
+      "step": 6504
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.147739837136401e-05,
+      "loss": 0.0638,
+      "step": 6505
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1474102661880347e-05,
+      "loss": 0.0922,
+      "step": 6506
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1470806788719433e-05,
+      "loss": 0.0781,
+      "step": 6507
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1467510752247222e-05,
+      "loss": 0.0566,
+      "step": 6508
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1464214552829695e-05,
+      "loss": 0.086,
+      "step": 6509
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1460918190832847e-05,
+      "loss": 0.09,
+      "step": 6510
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1457621666622683e-05,
+      "loss": 0.0804,
+      "step": 6511
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1454324980565239e-05,
+      "loss": 0.0786,
+      "step": 6512
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1451028133026563e-05,
+      "loss": 0.0719,
+      "step": 6513
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1447731124372721e-05,
+      "loss": 0.0693,
+      "step": 6514
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1444433954969794e-05,
+      "loss": 0.073,
+      "step": 6515
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1441136625183885e-05,
+      "loss": 0.0664,
+      "step": 6516
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1437839135381112e-05,
+      "loss": 0.0624,
+      "step": 6517
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1434541485927619e-05,
+      "loss": 0.0689,
+      "step": 6518
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1431243677189552e-05,
+      "loss": 0.0736,
+      "step": 6519
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1427945709533092e-05,
+      "loss": 0.0659,
+      "step": 6520
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1424647583324424e-05,
+      "loss": 0.0565,
+      "step": 6521
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1421349298929758e-05,
+      "loss": 0.0557,
+      "step": 6522
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.141805085671532e-05,
+      "loss": 0.0805,
+      "step": 6523
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1414752257047352e-05,
+      "loss": 0.079,
+      "step": 6524
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1411453500292115e-05,
+      "loss": 0.0752,
+      "step": 6525
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1408154586815887e-05,
+      "loss": 0.0866,
+      "step": 6526
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1404855516984965e-05,
+      "loss": 0.0706,
+      "step": 6527
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1401556291165659e-05,
+      "loss": 0.0804,
+      "step": 6528
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.13982569097243e-05,
+      "loss": 0.0807,
+      "step": 6529
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1394957373027236e-05,
+      "loss": 0.0585,
+      "step": 6530
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1391657681440834e-05,
+      "loss": 0.0607,
+      "step": 6531
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1388357835331472e-05,
+      "loss": 0.0735,
+      "step": 6532
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1385057835065549e-05,
+      "loss": 0.0658,
+      "step": 6533
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1381757681009484e-05,
+      "loss": 0.0534,
+      "step": 6534
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.137845737352971e-05,
+      "loss": 0.063,
+      "step": 6535
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1375156912992674e-05,
+      "loss": 0.0685,
+      "step": 6536
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1371856299764845e-05,
+      "loss": 0.0673,
+      "step": 6537
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1368555534212711e-05,
+      "loss": 0.069,
+      "step": 6538
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1365254616702763e-05,
+      "loss": 0.0645,
+      "step": 6539
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1361953547601532e-05,
+      "loss": 0.0728,
+      "step": 6540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1358652327275544e-05,
+      "loss": 0.0845,
+      "step": 6541
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.135535095609135e-05,
+      "loss": 0.077,
+      "step": 6542
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1352049434415524e-05,
+      "loss": 0.0829,
+      "step": 6543
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1348747762614647e-05,
+      "loss": 0.0796,
+      "step": 6544
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1345445941055322e-05,
+      "loss": 0.0646,
+      "step": 6545
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1342143970104167e-05,
+      "loss": 0.0594,
+      "step": 6546
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.133884185012782e-05,
+      "loss": 0.0966,
+      "step": 6547
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1335539581492925e-05,
+      "loss": 0.0709,
+      "step": 6548
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1332237164566158e-05,
+      "loss": 0.061,
+      "step": 6549
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1328934599714201e-05,
+      "loss": 0.0895,
+      "step": 6550
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.132563188730375e-05,
+      "loss": 0.0582,
+      "step": 6551
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1322329027701529e-05,
+      "loss": 0.0735,
+      "step": 6552
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1319026021274267e-05,
+      "loss": 0.0633,
+      "step": 6553
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1315722868388719e-05,
+      "loss": 0.066,
+      "step": 6554
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1312419569411645e-05,
+      "loss": 0.0754,
+      "step": 6555
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1309116124709832e-05,
+      "loss": 0.066,
+      "step": 6556
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1305812534650075e-05,
+      "loss": 0.0695,
+      "step": 6557
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1302508799599193e-05,
+      "loss": 0.0962,
+      "step": 6558
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1299204919924011e-05,
+      "loss": 0.0795,
+      "step": 6559
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.1295900895991384e-05,
+      "loss": 0.0601,
+      "step": 6560
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1292596728168168e-05,
+      "loss": 0.0772,
+      "step": 6561
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1289292416821243e-05,
+      "loss": 0.0583,
+      "step": 6562
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1285987962317505e-05,
+      "loss": 0.075,
+      "step": 6563
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1282683365023866e-05,
+      "loss": 0.0726,
+      "step": 6564
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.127937862530725e-05,
+      "loss": 0.0615,
+      "step": 6565
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.12760737435346e-05,
+      "loss": 0.093,
+      "step": 6566
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1272768720072875e-05,
+      "loss": 0.0826,
+      "step": 6567
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1269463555289049e-05,
+      "loss": 0.078,
+      "step": 6568
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.126615824955011e-05,
+      "loss": 0.0685,
+      "step": 6569
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1262852803223065e-05,
+      "loss": 0.0703,
+      "step": 6570
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1259547216674932e-05,
+      "loss": 0.0616,
+      "step": 6571
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1256241490272753e-05,
+      "loss": 0.0648,
+      "step": 6572
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1252935624383574e-05,
+      "loss": 0.0809,
+      "step": 6573
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1249629619374465e-05,
+      "loss": 0.0651,
+      "step": 6574
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1246323475612509e-05,
+      "loss": 0.0692,
+      "step": 6575
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1243017193464802e-05,
+      "loss": 0.0634,
+      "step": 6576
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1239710773298462e-05,
+      "loss": 0.066,
+      "step": 6577
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1236404215480617e-05,
+      "loss": 0.0862,
+      "step": 6578
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1233097520378404e-05,
+      "loss": 0.0746,
+      "step": 6579
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.1229790688358995e-05,
+      "loss": 0.0811,
+      "step": 6580
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1226483719789556e-05,
+      "loss": 0.0716,
+      "step": 6581
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1223176615037274e-05,
+      "loss": 0.0856,
+      "step": 6582
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1219869374469364e-05,
+      "loss": 0.084,
+      "step": 6583
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1216561998453038e-05,
+      "loss": 0.0667,
+      "step": 6584
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1213254487355533e-05,
+      "loss": 0.0546,
+      "step": 6585
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1209946841544101e-05,
+      "loss": 0.0836,
+      "step": 6586
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1206639061386006e-05,
+      "loss": 0.0609,
+      "step": 6587
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.120333114724853e-05,
+      "loss": 0.0815,
+      "step": 6588
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1200023099498962e-05,
+      "loss": 0.0723,
+      "step": 6589
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1196714918504616e-05,
+      "loss": 0.0841,
+      "step": 6590
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1193406604632818e-05,
+      "loss": 0.0703,
+      "step": 6591
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1190098158250904e-05,
+      "loss": 0.0745,
+      "step": 6592
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1186789579726226e-05,
+      "loss": 0.0572,
+      "step": 6593
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.118348086942616e-05,
+      "loss": 0.0684,
+      "step": 6594
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1180172027718084e-05,
+      "loss": 0.0602,
+      "step": 6595
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1176863054969396e-05,
+      "loss": 0.065,
+      "step": 6596
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1173553951547508e-05,
+      "loss": 0.0673,
+      "step": 6597
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1170244717819849e-05,
+      "loss": 0.0501,
+      "step": 6598
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1166935354153857e-05,
+      "loss": 0.0641,
+      "step": 6599
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1163625860916996e-05,
+      "loss": 0.0659,
+      "step": 6600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1160316238476727e-05,
+      "loss": 0.0842,
+      "step": 6601
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1157006487200534e-05,
+      "loss": 0.0866,
+      "step": 6602
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1153696607455924e-05,
+      "loss": 0.0653,
+      "step": 6603
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1150386599610406e-05,
+      "loss": 0.0605,
+      "step": 6604
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1147076464031503e-05,
+      "loss": 0.0717,
+      "step": 6605
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1143766201086765e-05,
+      "loss": 0.056,
+      "step": 6606
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.114045581114374e-05,
+      "loss": 0.0674,
+      "step": 6607
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1137145294570002e-05,
+      "loss": 0.0594,
+      "step": 6608
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1133834651733135e-05,
+      "loss": 0.0678,
+      "step": 6609
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1130523883000732e-05,
+      "loss": 0.0684,
+      "step": 6610
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.112721298874041e-05,
+      "loss": 0.0637,
+      "step": 6611
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1123901969319795e-05,
+      "loss": 0.0743,
+      "step": 6612
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1120590825106521e-05,
+      "loss": 0.0666,
+      "step": 6613
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1117279556468248e-05,
+      "loss": 0.0795,
+      "step": 6614
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1113968163772641e-05,
+      "loss": 0.0689,
+      "step": 6615
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1110656647387376e-05,
+      "loss": 0.0652,
+      "step": 6616
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1107345007680157e-05,
+      "loss": 0.0864,
+      "step": 6617
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1104033245018686e-05,
+      "loss": 0.0684,
+      "step": 6618
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1100721359770686e-05,
+      "loss": 0.075,
+      "step": 6619
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1097409352303896e-05,
+      "loss": 0.0797,
+      "step": 6620
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1094097222986063e-05,
+      "loss": 0.0734,
+      "step": 6621
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1090784972184947e-05,
+      "loss": 0.0658,
+      "step": 6622
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.108747260026833e-05,
+      "loss": 0.0598,
+      "step": 6623
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1084160107603998e-05,
+      "loss": 0.0566,
+      "step": 6624
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1080847494559754e-05,
+      "loss": 0.0696,
+      "step": 6625
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.107753476150342e-05,
+      "loss": 0.0709,
+      "step": 6626
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1074221908802819e-05,
+      "loss": 0.064,
+      "step": 6627
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1070908936825795e-05,
+      "loss": 0.0698,
+      "step": 6628
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.106759584594021e-05,
+      "loss": 0.077,
+      "step": 6629
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1064282636513927e-05,
+      "loss": 0.083,
+      "step": 6630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1060969308914835e-05,
+      "loss": 0.0808,
+      "step": 6631
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1057655863510826e-05,
+      "loss": 0.0829,
+      "step": 6632
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1054342300669806e-05,
+      "loss": 0.0629,
+      "step": 6633
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1051028620759704e-05,
+      "loss": 0.0618,
+      "step": 6634
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1047714824148454e-05,
+      "loss": 0.0481,
+      "step": 6635
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1044400911203998e-05,
+      "loss": 0.0775,
+      "step": 6636
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1041086882294306e-05,
+      "loss": 0.075,
+      "step": 6637
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1037772737787342e-05,
+      "loss": 0.0713,
+      "step": 6638
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1034458478051098e-05,
+      "loss": 0.0629,
+      "step": 6639
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1031144103453572e-05,
+      "loss": 0.067,
+      "step": 6640
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1027829614362779e-05,
+      "loss": 0.0649,
+      "step": 6641
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1024515011146739e-05,
+      "loss": 0.0652,
+      "step": 6642
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1021200294173493e-05,
+      "loss": 0.0729,
+      "step": 6643
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1017885463811094e-05,
+      "loss": 0.0617,
+      "step": 6644
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1014570520427594e-05,
+      "loss": 0.0624,
+      "step": 6645
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1011255464391081e-05,
+      "loss": 0.0714,
+      "step": 6646
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1007940296069638e-05,
+      "loss": 0.0724,
+      "step": 6647
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1004625015831362e-05,
+      "loss": 0.0753,
+      "step": 6648
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1001309624044372e-05,
+      "loss": 0.0688,
+      "step": 6649
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0997994121076784e-05,
+      "loss": 0.0785,
+      "step": 6650
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0994678507296743e-05,
+      "loss": 0.0795,
+      "step": 6651
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0991362783072399e-05,
+      "loss": 0.0779,
+      "step": 6652
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0988046948771909e-05,
+      "loss": 0.0703,
+      "step": 6653
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0984731004763452e-05,
+      "loss": 0.0786,
+      "step": 6654
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0981414951415212e-05,
+      "loss": 0.0699,
+      "step": 6655
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0978098789095386e-05,
+      "loss": 0.075,
+      "step": 6656
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.097478251817219e-05,
+      "loss": 0.0912,
+      "step": 6657
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0971466139013841e-05,
+      "loss": 0.0717,
+      "step": 6658
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0968149651988578e-05,
+      "loss": 0.078,
+      "step": 6659
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0964833057464645e-05,
+      "loss": 0.0728,
+      "step": 6660
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.09615163558103e-05,
+      "loss": 0.0695,
+      "step": 6661
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0958199547393817e-05,
+      "loss": 0.0828,
+      "step": 6662
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0954882632583477e-05,
+      "loss": 0.064,
+      "step": 6663
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0951565611747572e-05,
+      "loss": 0.082,
+      "step": 6664
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.094824848525441e-05,
+      "loss": 0.0796,
+      "step": 6665
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0944931253472313e-05,
+      "loss": 0.0858,
+      "step": 6666
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.09416139167696e-05,
+      "loss": 0.077,
+      "step": 6667
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.093829647551462e-05,
+      "loss": 0.0778,
+      "step": 6668
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0934978930075726e-05,
+      "loss": 0.0651,
+      "step": 6669
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0931661280821276e-05,
+      "loss": 0.0738,
+      "step": 6670
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0928343528119651e-05,
+      "loss": 0.0806,
+      "step": 6671
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.092502567233924e-05,
+      "loss": 0.0677,
+      "step": 6672
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0921707713848434e-05,
+      "loss": 0.0732,
+      "step": 6673
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0918389653015651e-05,
+      "loss": 0.0712,
+      "step": 6674
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.091507149020931e-05,
+      "loss": 0.0682,
+      "step": 6675
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0911753225797841e-05,
+      "loss": 0.0483,
+      "step": 6676
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.090843486014969e-05,
+      "loss": 0.0686,
+      "step": 6677
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0905116393633314e-05,
+      "loss": 0.0645,
+      "step": 6678
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0901797826617176e-05,
+      "loss": 0.0695,
+      "step": 6679
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0898479159469756e-05,
+      "loss": 0.058,
+      "step": 6680
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0895160392559543e-05,
+      "loss": 0.0642,
+      "step": 6681
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0891841526255038e-05,
+      "loss": 0.0657,
+      "step": 6682
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.088852256092475e-05,
+      "loss": 0.0689,
+      "step": 6683
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0885203496937198e-05,
+      "loss": 0.0862,
+      "step": 6684
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0881884334660921e-05,
+      "loss": 0.0638,
+      "step": 6685
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0878565074464459e-05,
+      "loss": 0.0667,
+      "step": 6686
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0875245716716364e-05,
+      "loss": 0.0699,
+      "step": 6687
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.087192626178521e-05,
+      "loss": 0.072,
+      "step": 6688
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0868606710039569e-05,
+      "loss": 0.079,
+      "step": 6689
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0865287061848022e-05,
+      "loss": 0.0717,
+      "step": 6690
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0861967317579176e-05,
+      "loss": 0.0754,
+      "step": 6691
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0858647477601633e-05,
+      "loss": 0.0617,
+      "step": 6692
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0855327542284015e-05,
+      "loss": 0.0712,
+      "step": 6693
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0852007511994953e-05,
+      "loss": 0.0657,
+      "step": 6694
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0848687387103082e-05,
+      "loss": 0.0758,
+      "step": 6695
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0845367167977056e-05,
+      "loss": 0.074,
+      "step": 6696
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.084204685498554e-05,
+      "loss": 0.0727,
+      "step": 6697
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.08387264484972e-05,
+      "loss": 0.0846,
+      "step": 6698
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0835405948880716e-05,
+      "loss": 0.0532,
+      "step": 6699
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0832085356504787e-05,
+      "loss": 0.0756,
+      "step": 6700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.082876467173811e-05,
+      "loss": 0.044,
+      "step": 6701
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0825443894949398e-05,
+      "loss": 0.0823,
+      "step": 6702
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.082212302650738e-05,
+      "loss": 0.0808,
+      "step": 6703
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0818802066780785e-05,
+      "loss": 0.0593,
+      "step": 6704
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0815481016138356e-05,
+      "loss": 0.0751,
+      "step": 6705
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0812159874948847e-05,
+      "loss": 0.0637,
+      "step": 6706
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0808838643581022e-05,
+      "loss": 0.0723,
+      "step": 6707
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0805517322403654e-05,
+      "loss": 0.0739,
+      "step": 6708
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0802195911785527e-05,
+      "loss": 0.0791,
+      "step": 6709
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.079887441209543e-05,
+      "loss": 0.0668,
+      "step": 6710
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0795552823702176e-05,
+      "loss": 0.0796,
+      "step": 6711
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.079223114697457e-05,
+      "loss": 0.0877,
+      "step": 6712
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0788909382281437e-05,
+      "loss": 0.0677,
+      "step": 6713
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0785587529991612e-05,
+      "loss": 0.0687,
+      "step": 6714
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0782265590473934e-05,
+      "loss": 0.0768,
+      "step": 6715
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.077894356409726e-05,
+      "loss": 0.0668,
+      "step": 6716
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0775621451230444e-05,
+      "loss": 0.0769,
+      "step": 6717
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0772299252242364e-05,
+      "loss": 0.058,
+      "step": 6718
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.07689769675019e-05,
+      "loss": 0.0741,
+      "step": 6719
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0765654597377941e-05,
+      "loss": 0.0565,
+      "step": 6720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0762332142239384e-05,
+      "loss": 0.0625,
+      "step": 6721
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0759009602455146e-05,
+      "loss": 0.0767,
+      "step": 6722
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0755686978394142e-05,
+      "loss": 0.1029,
+      "step": 6723
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0752364270425294e-05,
+      "loss": 0.0656,
+      "step": 6724
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0749041478917552e-05,
+      "loss": 0.0789,
+      "step": 6725
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0745718604239852e-05,
+      "loss": 0.0967,
+      "step": 6726
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0742395646761155e-05,
+      "loss": 0.0587,
+      "step": 6727
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0739072606850425e-05,
+      "loss": 0.0722,
+      "step": 6728
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0735749484876638e-05,
+      "loss": 0.0666,
+      "step": 6729
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0732426281208772e-05,
+      "loss": 0.0656,
+      "step": 6730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0729102996215828e-05,
+      "loss": 0.0773,
+      "step": 6731
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.07257796302668e-05,
+      "loss": 0.0648,
+      "step": 6732
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.07224561837307e-05,
+      "loss": 0.0681,
+      "step": 6733
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0719132656976551e-05,
+      "loss": 0.0693,
+      "step": 6734
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.0715809050373381e-05,
+      "loss": 0.0763,
+      "step": 6735
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0712485364290224e-05,
+      "loss": 0.0598,
+      "step": 6736
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0709161599096128e-05,
+      "loss": 0.0629,
+      "step": 6737
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0705837755160147e-05,
+      "loss": 0.0667,
+      "step": 6738
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0702513832851344e-05,
+      "loss": 0.0773,
+      "step": 6739
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0699189832538795e-05,
+      "loss": 0.0848,
+      "step": 6740
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0695865754591576e-05,
+      "loss": 0.0723,
+      "step": 6741
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.069254159937878e-05,
+      "loss": 0.0718,
+      "step": 6742
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0689217367269507e-05,
+      "loss": 0.073,
+      "step": 6743
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0685893058632855e-05,
+      "loss": 0.0818,
+      "step": 6744
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.068256867383795e-05,
+      "loss": 0.0784,
+      "step": 6745
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0679244213253907e-05,
+      "loss": 0.058,
+      "step": 6746
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0675919677249864e-05,
+      "loss": 0.0671,
+      "step": 6747
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0672595066194959e-05,
+      "loss": 0.0486,
+      "step": 6748
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0669270380458342e-05,
+      "loss": 0.0475,
+      "step": 6749
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0665945620409166e-05,
+      "loss": 0.0687,
+      "step": 6750
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0662620786416601e-05,
+      "loss": 0.0817,
+      "step": 6751
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0659295878849822e-05,
+      "loss": 0.0656,
+      "step": 6752
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0655970898078003e-05,
+      "loss": 0.0708,
+      "step": 6753
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0652645844470342e-05,
+      "loss": 0.0674,
+      "step": 6754
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.0649320718396032e-05,
+      "loss": 0.0617,
+      "step": 6755
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0645995520224279e-05,
+      "loss": 0.0544,
+      "step": 6756
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0642670250324301e-05,
+      "loss": 0.0601,
+      "step": 6757
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0639344909065312e-05,
+      "loss": 0.079,
+      "step": 6758
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0636019496816552e-05,
+      "loss": 0.0616,
+      "step": 6759
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0632694013947254e-05,
+      "loss": 0.058,
+      "step": 6760
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0629368460826662e-05,
+      "loss": 0.0795,
+      "step": 6761
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0626042837824032e-05,
+      "loss": 0.063,
+      "step": 6762
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0622717145308623e-05,
+      "loss": 0.0717,
+      "step": 6763
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0619391383649701e-05,
+      "loss": 0.0656,
+      "step": 6764
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0616065553216553e-05,
+      "loss": 0.0743,
+      "step": 6765
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0612739654378455e-05,
+      "loss": 0.0677,
+      "step": 6766
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0609413687504698e-05,
+      "loss": 0.084,
+      "step": 6767
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0606087652964585e-05,
+      "loss": 0.0684,
+      "step": 6768
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0602761551127421e-05,
+      "loss": 0.0533,
+      "step": 6769
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.059943538236252e-05,
+      "loss": 0.0775,
+      "step": 6770
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0596109147039207e-05,
+      "loss": 0.0739,
+      "step": 6771
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0592782845526808e-05,
+      "loss": 0.0667,
+      "step": 6772
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0589456478194658e-05,
+      "loss": 0.069,
+      "step": 6773
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.0586130045412103e-05,
+      "loss": 0.0707,
+      "step": 6774
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0582803547548497e-05,
+      "loss": 0.083,
+      "step": 6775
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0579476984973191e-05,
+      "loss": 0.0663,
+      "step": 6776
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0576150358055559e-05,
+      "loss": 0.0654,
+      "step": 6777
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0572823667164968e-05,
+      "loss": 0.067,
+      "step": 6778
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0569496912670798e-05,
+      "loss": 0.071,
+      "step": 6779
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.056617009494244e-05,
+      "loss": 0.0613,
+      "step": 6780
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.056284321434928e-05,
+      "loss": 0.0668,
+      "step": 6781
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0559516271260727e-05,
+      "loss": 0.0711,
+      "step": 6782
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0556189266046187e-05,
+      "loss": 0.075,
+      "step": 6783
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0552862199075067e-05,
+      "loss": 0.0694,
+      "step": 6784
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0549535070716799e-05,
+      "loss": 0.0715,
+      "step": 6785
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0546207881340804e-05,
+      "loss": 0.0615,
+      "step": 6786
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.054288063131652e-05,
+      "loss": 0.0906,
+      "step": 6787
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0539553321013388e-05,
+      "loss": 0.0662,
+      "step": 6788
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0536225950800856e-05,
+      "loss": 0.0715,
+      "step": 6789
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0532898521048382e-05,
+      "loss": 0.071,
+      "step": 6790
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0529571032125426e-05,
+      "loss": 0.0682,
+      "step": 6791
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0526243484401453e-05,
+      "loss": 0.0652,
+      "step": 6792
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0522915878245945e-05,
+      "loss": 0.0733,
+      "step": 6793
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0519588214028377e-05,
+      "loss": 0.0571,
+      "step": 6794
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0516260492118238e-05,
+      "loss": 0.0741,
+      "step": 6795
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0512932712885027e-05,
+      "loss": 0.0538,
+      "step": 6796
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0509604876698238e-05,
+      "loss": 0.0704,
+      "step": 6797
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0506276983927382e-05,
+      "loss": 0.0831,
+      "step": 6798
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0502949034941969e-05,
+      "loss": 0.0701,
+      "step": 6799
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0499621030111523e-05,
+      "loss": 0.0828,
+      "step": 6800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0496292969805563e-05,
+      "loss": 0.0657,
+      "step": 6801
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.049296485439363e-05,
+      "loss": 0.0774,
+      "step": 6802
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0489636684245255e-05,
+      "loss": 0.0536,
+      "step": 6803
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.048630845972998e-05,
+      "loss": 0.0771,
+      "step": 6804
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.048298018121736e-05,
+      "loss": 0.0784,
+      "step": 6805
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0479651849076953e-05,
+      "loss": 0.0776,
+      "step": 6806
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0476323463678313e-05,
+      "loss": 0.0687,
+      "step": 6807
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0472995025391015e-05,
+      "loss": 0.0845,
+      "step": 6808
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.046966653458463e-05,
+      "loss": 0.0751,
+      "step": 6809
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0466337991628737e-05,
+      "loss": 0.0522,
+      "step": 6810
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0463009396892923e-05,
+      "loss": 0.0622,
+      "step": 6811
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0459680750746775e-05,
+      "loss": 0.0579,
+      "step": 6812
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0456352053559895e-05,
+      "loss": 0.0593,
+      "step": 6813
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0453023305701886e-05,
+      "loss": 0.0629,
+      "step": 6814
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.044969450754235e-05,
+      "loss": 0.0783,
+      "step": 6815
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0446365659450906e-05,
+      "loss": 0.0685,
+      "step": 6816
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.044303676179717e-05,
+      "loss": 0.0787,
+      "step": 6817
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0439707814950768e-05,
+      "loss": 0.0723,
+      "step": 6818
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0436378819281333e-05,
+      "loss": 0.0684,
+      "step": 6819
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0433049775158497e-05,
+      "loss": 0.0772,
+      "step": 6820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.04297206829519e-05,
+      "loss": 0.0709,
+      "step": 6821
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0426391543031194e-05,
+      "loss": 0.063,
+      "step": 6822
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0423062355766025e-05,
+      "loss": 0.077,
+      "step": 6823
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.041973312152605e-05,
+      "loss": 0.0884,
+      "step": 6824
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0416403840680938e-05,
+      "loss": 0.0752,
+      "step": 6825
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0413074513600348e-05,
+      "loss": 0.0561,
+      "step": 6826
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0409745140653955e-05,
+      "loss": 0.064,
+      "step": 6827
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.040641572221144e-05,
+      "loss": 0.0701,
+      "step": 6828
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0403086258642482e-05,
+      "loss": 0.0673,
+      "step": 6829
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0399756750316767e-05,
+      "loss": 0.0675,
+      "step": 6830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0396427197603992e-05,
+      "loss": 0.0847,
+      "step": 6831
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0393097600873854e-05,
+      "loss": 0.0634,
+      "step": 6832
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0389767960496051e-05,
+      "loss": 0.0792,
+      "step": 6833
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0386438276840296e-05,
+      "loss": 0.0655,
+      "step": 6834
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0383108550276295e-05,
+      "loss": 0.0796,
+      "step": 6835
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.037977878117377e-05,
+      "loss": 0.0773,
+      "step": 6836
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0376448969902441e-05,
+      "loss": 0.0665,
+      "step": 6837
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0373119116832031e-05,
+      "loss": 0.062,
+      "step": 6838
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0369789222332276e-05,
+      "loss": 0.0868,
+      "step": 6839
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0366459286772907e-05,
+      "loss": 0.0723,
+      "step": 6840
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0363129310523666e-05,
+      "loss": 0.0632,
+      "step": 6841
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0359799293954298e-05,
+      "loss": 0.0711,
+      "step": 6842
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0356469237434552e-05,
+      "loss": 0.0684,
+      "step": 6843
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0353139141334179e-05,
+      "loss": 0.0701,
+      "step": 6844
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.034980900602294e-05,
+      "loss": 0.0806,
+      "step": 6845
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0346478831870596e-05,
+      "loss": 0.0663,
+      "step": 6846
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.034314861924691e-05,
+      "loss": 0.0693,
+      "step": 6847
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.033981836852166e-05,
+      "loss": 0.09,
+      "step": 6848
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0336488080064614e-05,
+      "loss": 0.0676,
+      "step": 6849
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0333157754245557e-05,
+      "loss": 0.0696,
+      "step": 6850
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0329827391434268e-05,
+      "loss": 0.0643,
+      "step": 6851
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0326496992000535e-05,
+      "loss": 0.0709,
+      "step": 6852
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.032316655631415e-05,
+      "loss": 0.0529,
+      "step": 6853
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0319836084744912e-05,
+      "loss": 0.0913,
+      "step": 6854
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0316505577662611e-05,
+      "loss": 0.0674,
+      "step": 6855
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0313175035437065e-05,
+      "loss": 0.0628,
+      "step": 6856
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0309844458438068e-05,
+      "loss": 0.0708,
+      "step": 6857
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0306513847035435e-05,
+      "loss": 0.0762,
+      "step": 6858
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0303183201598985e-05,
+      "loss": 0.0699,
+      "step": 6859
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0299852522498535e-05,
+      "loss": 0.07,
+      "step": 6860
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0296521810103906e-05,
+      "loss": 0.0576,
+      "step": 6861
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0293191064784924e-05,
+      "loss": 0.065,
+      "step": 6862
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.028986028691142e-05,
+      "loss": 0.0707,
+      "step": 6863
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0286529476853228e-05,
+      "loss": 0.0657,
+      "step": 6864
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0283198634980185e-05,
+      "loss": 0.0837,
+      "step": 6865
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0279867761662127e-05,
+      "loss": 0.0663,
+      "step": 6866
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.027653685726891e-05,
+      "loss": 0.0658,
+      "step": 6867
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0273205922170369e-05,
+      "loss": 0.076,
+      "step": 6868
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0269874956736359e-05,
+      "loss": 0.0687,
+      "step": 6869
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0266543961336737e-05,
+      "loss": 0.074,
+      "step": 6870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0263212936341358e-05,
+      "loss": 0.0701,
+      "step": 6871
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0259881882120082e-05,
+      "loss": 0.0627,
+      "step": 6872
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0256550799042779e-05,
+      "loss": 0.0734,
+      "step": 6873
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.025321968747931e-05,
+      "loss": 0.0865,
+      "step": 6874
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0249888547799547e-05,
+      "loss": 0.0719,
+      "step": 6875
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0246557380373366e-05,
+      "loss": 0.0732,
+      "step": 6876
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0243226185570643e-05,
+      "loss": 0.0698,
+      "step": 6877
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0239894963761254e-05,
+      "loss": 0.0672,
+      "step": 6878
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0236563715315089e-05,
+      "loss": 0.0655,
+      "step": 6879
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0233232440602027e-05,
+      "loss": 0.0595,
+      "step": 6880
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.022990113999196e-05,
+      "loss": 0.0764,
+      "step": 6881
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.022656981385478e-05,
+      "loss": 0.0764,
+      "step": 6882
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.022323846256038e-05,
+      "loss": 0.0684,
+      "step": 6883
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0219907086478655e-05,
+      "loss": 0.0612,
+      "step": 6884
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.021657568597951e-05,
+      "loss": 0.0758,
+      "step": 6885
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0213244261432844e-05,
+      "loss": 0.062,
+      "step": 6886
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0209912813208565e-05,
+      "loss": 0.0738,
+      "step": 6887
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.020658134167658e-05,
+      "loss": 0.0677,
+      "step": 6888
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0203249847206796e-05,
+      "loss": 0.0728,
+      "step": 6889
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.019991833016913e-05,
+      "loss": 0.072,
+      "step": 6890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0196586790933498e-05,
+      "loss": 0.0754,
+      "step": 6891
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0193255229869813e-05,
+      "loss": 0.0677,
+      "step": 6892
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0189923647348003e-05,
+      "loss": 0.0685,
+      "step": 6893
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0186592043737988e-05,
+      "loss": 0.0556,
+      "step": 6894
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0183260419409689e-05,
+      "loss": 0.055,
+      "step": 6895
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.017992877473304e-05,
+      "loss": 0.0806,
+      "step": 6896
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0176597110077964e-05,
+      "loss": 0.0672,
+      "step": 6897
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0173265425814397e-05,
+      "loss": 0.0695,
+      "step": 6898
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0169933722312273e-05,
+      "loss": 0.0693,
+      "step": 6899
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0166601999941528e-05,
+      "loss": 0.0818,
+      "step": 6900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.01632702590721e-05,
+      "loss": 0.0651,
+      "step": 6901
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0159938500073928e-05,
+      "loss": 0.0647,
+      "step": 6902
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0156606723316962e-05,
+      "loss": 0.0634,
+      "step": 6903
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0153274929171134e-05,
+      "loss": 0.0665,
+      "step": 6904
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.01499431180064e-05,
+      "loss": 0.0566,
+      "step": 6905
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0146611290192706e-05,
+      "loss": 0.0656,
+      "step": 6906
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0143279446099999e-05,
+      "loss": 0.0582,
+      "step": 6907
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0139947586098235e-05,
+      "loss": 0.0668,
+      "step": 6908
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0136615710557362e-05,
+      "loss": 0.0659,
+      "step": 6909
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0133283819847344e-05,
+      "loss": 0.0737,
+      "step": 6910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0129951914338132e-05,
+      "loss": 0.0792,
+      "step": 6911
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0126619994399684e-05,
+      "loss": 0.0771,
+      "step": 6912
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0123288060401966e-05,
+      "loss": 0.0668,
+      "step": 6913
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0119956112714934e-05,
+      "loss": 0.0669,
+      "step": 6914
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0116624151708554e-05,
+      "loss": 0.0693,
+      "step": 6915
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0113292177752791e-05,
+      "loss": 0.0797,
+      "step": 6916
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0109960191217612e-05,
+      "loss": 0.0646,
+      "step": 6917
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0106628192472983e-05,
+      "loss": 0.0739,
+      "step": 6918
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0103296181888873e-05,
+      "loss": 0.0672,
+      "step": 6919
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0099964159835249e-05,
+      "loss": 0.0861,
+      "step": 6920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0096632126682093e-05,
+      "loss": 0.0788,
+      "step": 6921
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0093300082799368e-05,
+      "loss": 0.0634,
+      "step": 6922
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0089968028557054e-05,
+      "loss": 0.0612,
+      "step": 6923
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0086635964325121e-05,
+      "loss": 0.0546,
+      "step": 6924
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.008330389047355e-05,
+      "loss": 0.0633,
+      "step": 6925
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0079971807372318e-05,
+      "loss": 0.0767,
+      "step": 6926
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0076639715391399e-05,
+      "loss": 0.0654,
+      "step": 6927
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0073307614900778e-05,
+      "loss": 0.0532,
+      "step": 6928
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.006997550627043e-05,
+      "loss": 0.0733,
+      "step": 6929
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0066643389870343e-05,
+      "loss": 0.0582,
+      "step": 6930
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0063311266070495e-05,
+      "loss": 0.0684,
+      "step": 6931
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0059979135240868e-05,
+      "loss": 0.0643,
+      "step": 6932
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0056646997751448e-05,
+      "loss": 0.0914,
+      "step": 6933
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0053314853972216e-05,
+      "loss": 0.0654,
+      "step": 6934
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.004998270427316e-05,
+      "loss": 0.0535,
+      "step": 6935
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0046650549024267e-05,
+      "loss": 0.0699,
+      "step": 6936
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0043318388595523e-05,
+      "loss": 0.0729,
+      "step": 6937
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.003998622335691e-05,
+      "loss": 0.0723,
+      "step": 6938
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0036654053678425e-05,
+      "loss": 0.0773,
+      "step": 6939
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0033321879930045e-05,
+      "loss": 0.0617,
+      "step": 6940
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0029989702481768e-05,
+      "loss": 0.0723,
+      "step": 6941
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.002665752170358e-05,
+      "loss": 0.0837,
+      "step": 6942
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0023325337965466e-05,
+      "loss": 0.0632,
+      "step": 6943
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0019993151637419e-05,
+      "loss": 0.0587,
+      "step": 6944
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0016660963089433e-05,
+      "loss": 0.0645,
+      "step": 6945
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0013328772691489e-05,
+      "loss": 0.062,
+      "step": 6946
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0009996580813585e-05,
+      "loss": 0.0605,
+      "step": 6947
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0006664387825709e-05,
+      "loss": 0.0633,
+      "step": 6948
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.000333219409785e-05,
+      "loss": 0.0657,
+      "step": 6949
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1e-05,
+      "loss": 0.0594,
+      "step": 6950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.996667805902154e-06,
+      "loss": 0.0735,
+      "step": 6951
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.993335612174295e-06,
+      "loss": 0.0582,
+      "step": 6952
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.99000341918642e-06,
+      "loss": 0.072,
+      "step": 6953
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.986671227308514e-06,
+      "loss": 0.0861,
+      "step": 6954
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.983339036910573e-06,
+      "loss": 0.0792,
+      "step": 6955
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.980006848362583e-06,
+      "loss": 0.079,
+      "step": 6956
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.976674662034537e-06,
+      "loss": 0.0634,
+      "step": 6957
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.973342478296423e-06,
+      "loss": 0.0587,
+      "step": 6958
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.970010297518237e-06,
+      "loss": 0.0665,
+      "step": 6959
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.966678120069957e-06,
+      "loss": 0.0745,
+      "step": 6960
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.963345946321582e-06,
+      "loss": 0.0699,
+      "step": 6961
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.960013776643091e-06,
+      "loss": 0.0616,
+      "step": 6962
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.956681611404482e-06,
+      "loss": 0.0723,
+      "step": 6963
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.953349450975736e-06,
+      "loss": 0.0762,
+      "step": 6964
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.950017295726845e-06,
+      "loss": 0.0635,
+      "step": 6965
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.946685146027787e-06,
+      "loss": 0.0536,
+      "step": 6966
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.94335300224856e-06,
+      "loss": 0.0938,
+      "step": 6967
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.940020864759137e-06,
+      "loss": 0.0671,
+      "step": 6968
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.93668873392951e-06,
+      "loss": 0.0687,
+      "step": 6969
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.93335661012966e-06,
+      "loss": 0.0584,
+      "step": 6970
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.930024493729573e-06,
+      "loss": 0.0655,
+      "step": 6971
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.926692385099224e-06,
+      "loss": 0.0569,
+      "step": 6972
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.923360284608606e-06,
+      "loss": 0.07,
+      "step": 6973
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.920028192627686e-06,
+      "loss": 0.0613,
+      "step": 6974
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.916696109526454e-06,
+      "loss": 0.0731,
+      "step": 6975
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.913364035674882e-06,
+      "loss": 0.0679,
+      "step": 6976
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.91003197144295e-06,
+      "loss": 0.0804,
+      "step": 6977
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.906699917200633e-06,
+      "loss": 0.0636,
+      "step": 6978
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.903367873317913e-06,
+      "loss": 0.0607,
+      "step": 6979
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.900035840164753e-06,
+      "loss": 0.0764,
+      "step": 6980
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.896703818111134e-06,
+      "loss": 0.0706,
+      "step": 6981
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.893371807527022e-06,
+      "loss": 0.0778,
+      "step": 6982
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.890039808782393e-06,
+      "loss": 0.0804,
+      "step": 6983
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.88670782224721e-06,
+      "loss": 0.0649,
+      "step": 6984
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.88337584829145e-06,
+      "loss": 0.058,
+      "step": 6985
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.880043887285065e-06,
+      "loss": 0.0756,
+      "step": 6986
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.876711939598037e-06,
+      "loss": 0.0773,
+      "step": 6987
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.873380005600316e-06,
+      "loss": 0.0719,
+      "step": 6988
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.87004808566187e-06,
+      "loss": 0.0828,
+      "step": 6989
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.866716180152656e-06,
+      "loss": 0.0587,
+      "step": 6990
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.86338428944264e-06,
+      "loss": 0.072,
+      "step": 6991
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.860052413901766e-06,
+      "loss": 0.0663,
+      "step": 6992
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.856720553900004e-06,
+      "loss": 0.078,
+      "step": 6993
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.853388709807296e-06,
+      "loss": 0.0668,
+      "step": 6994
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.850056881993602e-06,
+      "loss": 0.0601,
+      "step": 6995
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.846725070828866e-06,
+      "loss": 0.073,
+      "step": 6996
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.843393276683042e-06,
+      "loss": 0.0699,
+      "step": 6997
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.84006149992607e-06,
+      "loss": 0.0634,
+      "step": 6998
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.836729740927904e-06,
+      "loss": 0.0561,
+      "step": 6999
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.833398000058473e-06,
+      "loss": 0.08,
+      "step": 7000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.83006627768773e-06,
+      "loss": 0.0817,
+      "step": 7001
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.826734574185605e-06,
+      "loss": 0.067,
+      "step": 7002
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.823402889922039e-06,
+      "loss": 0.0714,
+      "step": 7003
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.820071225266962e-06,
+      "loss": 0.0642,
+      "step": 7004
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.816739580590315e-06,
+      "loss": 0.0604,
+      "step": 7005
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.813407956262016e-06,
+      "loss": 0.0679,
+      "step": 7006
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.810076352651999e-06,
+      "loss": 0.0581,
+      "step": 7007
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.806744770130185e-06,
+      "loss": 0.0745,
+      "step": 7008
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.803413209066504e-06,
+      "loss": 0.084,
+      "step": 7009
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.80008166983087e-06,
+      "loss": 0.0731,
+      "step": 7010
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.796750152793208e-06,
+      "loss": 0.0612,
+      "step": 7011
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.793418658323422e-06,
+      "loss": 0.0739,
+      "step": 7012
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.790087186791438e-06,
+      "loss": 0.0804,
+      "step": 7013
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.786755738567156e-06,
+      "loss": 0.0595,
+      "step": 7014
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.783424314020492e-06,
+      "loss": 0.0746,
+      "step": 7015
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.780092913521345e-06,
+      "loss": 0.0587,
+      "step": 7016
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.776761537439623e-06,
+      "loss": 0.0699,
+      "step": 7017
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.773430186145225e-06,
+      "loss": 0.0756,
+      "step": 7018
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.770098860008043e-06,
+      "loss": 0.0712,
+      "step": 7019
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.766767559397978e-06,
+      "loss": 0.0609,
+      "step": 7020
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.763436284684915e-06,
+      "loss": 0.063,
+      "step": 7021
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.76010503623875e-06,
+      "loss": 0.0704,
+      "step": 7022
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.75677381442936e-06,
+      "loss": 0.0709,
+      "step": 7023
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.753442619626638e-06,
+      "loss": 0.0712,
+      "step": 7024
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.750111452200455e-06,
+      "loss": 0.059,
+      "step": 7025
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.746780312520695e-06,
+      "loss": 0.059,
+      "step": 7026
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.743449200957225e-06,
+      "loss": 0.0624,
+      "step": 7027
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.74011811787992e-06,
+      "loss": 0.0673,
+      "step": 7028
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.736787063658645e-06,
+      "loss": 0.0728,
+      "step": 7029
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.733456038663268e-06,
+      "loss": 0.0445,
+      "step": 7030
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.730125043263645e-06,
+      "loss": 0.059,
+      "step": 7031
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.726794077829636e-06,
+      "loss": 0.066,
+      "step": 7032
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.723463142731094e-06,
+      "loss": 0.066,
+      "step": 7033
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.720132238337874e-06,
+      "loss": 0.0657,
+      "step": 7034
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.716801365019819e-06,
+      "loss": 0.076,
+      "step": 7035
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.713470523146777e-06,
+      "loss": 0.0731,
+      "step": 7036
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.710139713088585e-06,
+      "loss": 0.0721,
+      "step": 7037
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.706808935215081e-06,
+      "loss": 0.0674,
+      "step": 7038
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.703478189896099e-06,
+      "loss": 0.0719,
+      "step": 7039
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.700147477501469e-06,
+      "loss": 0.072,
+      "step": 7040
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.696816798401017e-06,
+      "loss": 0.075,
+      "step": 7041
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.69348615296457e-06,
+      "loss": 0.0526,
+      "step": 7042
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.690155541561936e-06,
+      "loss": 0.0715,
+      "step": 7043
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.686824964562942e-06,
+      "loss": 0.0601,
+      "step": 7044
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.68349442233739e-06,
+      "loss": 0.0714,
+      "step": 7045
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.680163915255095e-06,
+      "loss": 0.0776,
+      "step": 7046
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.676833443685852e-06,
+      "loss": 0.0728,
+      "step": 7047
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.673503007999469e-06,
+      "loss": 0.0779,
+      "step": 7048
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.670172608565735e-06,
+      "loss": 0.0676,
+      "step": 7049
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.666842245754448e-06,
+      "loss": 0.0642,
+      "step": 7050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.663511919935387e-06,
+      "loss": 0.0679,
+      "step": 7051
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.660181631478343e-06,
+      "loss": 0.0834,
+      "step": 7052
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.65685138075309e-06,
+      "loss": 0.0594,
+      "step": 7053
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.653521168129407e-06,
+      "loss": 0.0815,
+      "step": 7054
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.65019099397706e-06,
+      "loss": 0.0602,
+      "step": 7055
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.646860858665825e-06,
+      "loss": 0.0686,
+      "step": 7056
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.64353076256545e-06,
+      "loss": 0.0835,
+      "step": 7057
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.640200706045705e-06,
+      "loss": 0.0885,
+      "step": 7058
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.636870689476334e-06,
+      "loss": 0.0641,
+      "step": 7059
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.633540713227095e-06,
+      "loss": 0.0815,
+      "step": 7060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.630210777667726e-06,
+      "loss": 0.0612,
+      "step": 7061
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.626880883167972e-06,
+      "loss": 0.0709,
+      "step": 7062
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.62355103009756e-06,
+      "loss": 0.0662,
+      "step": 7063
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.620221218826233e-06,
+      "loss": 0.0658,
+      "step": 7064
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.616891449723705e-06,
+      "loss": 0.0706,
+      "step": 7065
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.613561723159707e-06,
+      "loss": 0.0661,
+      "step": 7066
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.610232039503949e-06,
+      "loss": 0.0717,
+      "step": 7067
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.606902399126148e-06,
+      "loss": 0.0668,
+      "step": 7068
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.603572802396007e-06,
+      "loss": 0.0557,
+      "step": 7069
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.600243249683235e-06,
+      "loss": 0.0736,
+      "step": 7070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.59691374135752e-06,
+      "loss": 0.0735,
+      "step": 7071
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.593584277788563e-06,
+      "loss": 0.0706,
+      "step": 7072
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.590254859346046e-06,
+      "loss": 0.0806,
+      "step": 7073
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.586925486399656e-06,
+      "loss": 0.0613,
+      "step": 7074
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.583596159319064e-06,
+      "loss": 0.0598,
+      "step": 7075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.580266878473952e-06,
+      "loss": 0.0806,
+      "step": 7076
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.576937644233977e-06,
+      "loss": 0.0686,
+      "step": 7077
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.573608456968811e-06,
+      "loss": 0.0748,
+      "step": 7078
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.5702793170481e-06,
+      "loss": 0.0583,
+      "step": 7079
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.566950224841506e-06,
+      "loss": 0.0768,
+      "step": 7080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.563621180718669e-06,
+      "loss": 0.0786,
+      "step": 7081
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.560292185049235e-06,
+      "loss": 0.0745,
+      "step": 7082
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.55696323820283e-06,
+      "loss": 0.0783,
+      "step": 7083
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.553634340549098e-06,
+      "loss": 0.0674,
+      "step": 7084
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.550305492457655e-06,
+      "loss": 0.0702,
+      "step": 7085
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.546976694298117e-06,
+      "loss": 0.0712,
+      "step": 7086
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.54364794644011e-06,
+      "loss": 0.0757,
+      "step": 7087
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.540319249253229e-06,
+      "loss": 0.0692,
+      "step": 7088
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.536990603107084e-06,
+      "loss": 0.0561,
+      "step": 7089
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.533662008371267e-06,
+      "loss": 0.0814,
+      "step": 7090
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.530333465415375e-06,
+      "loss": 0.058,
+      "step": 7091
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.527004974608987e-06,
+      "loss": 0.058,
+      "step": 7092
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.523676536321692e-06,
+      "loss": 0.0722,
+      "step": 7093
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.52034815092305e-06,
+      "loss": 0.084,
+      "step": 7094
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.517019818782644e-06,
+      "loss": 0.0719,
+      "step": 7095
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.513691540270023e-06,
+      "loss": 0.0719,
+      "step": 7096
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.510363315754752e-06,
+      "loss": 0.063,
+      "step": 7097
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.507035145606374e-06,
+      "loss": 0.0697,
+      "step": 7098
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.503707030194439e-06,
+      "loss": 0.067,
+      "step": 7099
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.500378969888479e-06,
+      "loss": 0.0698,
+      "step": 7100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.497050965058036e-06,
+      "loss": 0.0821,
+      "step": 7101
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.493723016072623e-06,
+      "loss": 0.0594,
+      "step": 7102
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.490395123301767e-06,
+      "loss": 0.0613,
+      "step": 7103
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.487067287114978e-06,
+      "loss": 0.0728,
+      "step": 7104
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.483739507881764e-06,
+      "loss": 0.0719,
+      "step": 7105
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.480411785971626e-06,
+      "loss": 0.0703,
+      "step": 7106
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.477084121754062e-06,
+      "loss": 0.0783,
+      "step": 7107
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.47375651559855e-06,
+      "loss": 0.0565,
+      "step": 7108
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.470428967874579e-06,
+      "loss": 0.0817,
+      "step": 7109
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.467101478951621e-06,
+      "loss": 0.0692,
+      "step": 7110
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.463774049199148e-06,
+      "loss": 0.0717,
+      "step": 7111
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.460446678986616e-06,
+      "loss": 0.05,
+      "step": 7112
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.457119368683486e-06,
+      "loss": 0.0635,
+      "step": 7113
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.453792118659198e-06,
+      "loss": 0.0635,
+      "step": 7114
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.450464929283208e-06,
+      "loss": 0.0694,
+      "step": 7115
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.447137800924937e-06,
+      "loss": 0.0635,
+      "step": 7116
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.443810733953818e-06,
+      "loss": 0.0587,
+      "step": 7117
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.440483728739274e-06,
+      "loss": 0.0735,
+      "step": 7118
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.437156785650722e-06,
+      "loss": 0.0728,
+      "step": 7119
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.433829905057562e-06,
+      "loss": 0.0807,
+      "step": 7120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.430503087329204e-06,
+      "loss": 0.0739,
+      "step": 7121
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.427176332835033e-06,
+      "loss": 0.0645,
+      "step": 7122
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.423849641944443e-06,
+      "loss": 0.0676,
+      "step": 7123
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.420523015026809e-06,
+      "loss": 0.0676,
+      "step": 7124
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.417196452451506e-06,
+      "loss": 0.0602,
+      "step": 7125
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.413869954587897e-06,
+      "loss": 0.0759,
+      "step": 7126
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.410543521805345e-06,
+      "loss": 0.0583,
+      "step": 7127
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.407217154473195e-06,
+      "loss": 0.0722,
+      "step": 7128
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.403890852960796e-06,
+      "loss": 0.0701,
+      "step": 7129
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.400564617637481e-06,
+      "loss": 0.0664,
+      "step": 7130
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.39723844887258e-06,
+      "loss": 0.078,
+      "step": 7131
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.393912347035417e-06,
+      "loss": 0.0523,
+      "step": 7132
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.390586312495306e-06,
+      "loss": 0.0707,
+      "step": 7133
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.387260345621548e-06,
+      "loss": 0.072,
+      "step": 7134
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.38393444678345e-06,
+      "loss": 0.0705,
+      "step": 7135
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.380608616350297e-06,
+      "loss": 0.067,
+      "step": 7136
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.37728285469138e-06,
+      "loss": 0.066,
+      "step": 7137
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.37395716217597e-06,
+      "loss": 0.0624,
+      "step": 7138
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.370631539173343e-06,
+      "loss": 0.0643,
+      "step": 7139
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.367305986052748e-06,
+      "loss": 0.0569,
+      "step": 7140
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.36398050318345e-06,
+      "loss": 0.0751,
+      "step": 7141
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.360655090934688e-06,
+      "loss": 0.0549,
+      "step": 7142
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.357329749675704e-06,
+      "loss": 0.0712,
+      "step": 7143
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.354004479775722e-06,
+      "loss": 0.0726,
+      "step": 7144
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.350679281603972e-06,
+      "loss": 0.0562,
+      "step": 7145
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.34735415552966e-06,
+      "loss": 0.0796,
+      "step": 7146
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.344029101921999e-06,
+      "loss": 0.0643,
+      "step": 7147
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.340704121150182e-06,
+      "loss": 0.0668,
+      "step": 7148
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.3373792135834e-06,
+      "loss": 0.0731,
+      "step": 7149
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.334054379590834e-06,
+      "loss": 0.0627,
+      "step": 7150
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.330729619541661e-06,
+      "loss": 0.0767,
+      "step": 7151
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.327404933805041e-06,
+      "loss": 0.0584,
+      "step": 7152
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.324080322750139e-06,
+      "loss": 0.0734,
+      "step": 7153
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.320755786746096e-06,
+      "loss": 0.0675,
+      "step": 7154
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.317431326162054e-06,
+      "loss": 0.0662,
+      "step": 7155
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.314106941367148e-06,
+      "loss": 0.0693,
+      "step": 7156
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.310782632730498e-06,
+      "loss": 0.064,
+      "step": 7157
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.307458400621224e-06,
+      "loss": 0.0687,
+      "step": 7158
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.304134245408427e-06,
+      "loss": 0.0613,
+      "step": 7159
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.300810167461209e-06,
+      "loss": 0.0696,
+      "step": 7160
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.297486167148657e-06,
+      "loss": 0.0808,
+      "step": 7161
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.294162244839857e-06,
+      "loss": 0.071,
+      "step": 7162
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.290838400903874e-06,
+      "loss": 0.0507,
+      "step": 7163
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.28751463570978e-06,
+      "loss": 0.0556,
+      "step": 7164
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.284190949626624e-06,
+      "loss": 0.0645,
+      "step": 7165
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.280867343023452e-06,
+      "loss": 0.0712,
+      "step": 7166
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.277543816269302e-06,
+      "loss": 0.0737,
+      "step": 7167
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.274220369733205e-06,
+      "loss": 0.066,
+      "step": 7168
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.270897003784175e-06,
+      "loss": 0.0732,
+      "step": 7169
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.267573718791233e-06,
+      "loss": 0.0751,
+      "step": 7170
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.264250515123365e-06,
+      "loss": 0.0674,
+      "step": 7171
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.260927393149579e-06,
+      "loss": 0.0741,
+      "step": 7172
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.257604353238848e-06,
+      "loss": 0.0535,
+      "step": 7173
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.254281395760151e-06,
+      "loss": 0.063,
+      "step": 7174
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.250958521082452e-06,
+      "loss": 0.0772,
+      "step": 7175
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.247635729574707e-06,
+      "loss": 0.0746,
+      "step": 7176
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.244313021605862e-06,
+      "loss": 0.0712,
+      "step": 7177
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.240990397544859e-06,
+      "loss": 0.0602,
+      "step": 7178
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.237667857760617e-06,
+      "loss": 0.0949,
+      "step": 7179
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.234345402622065e-06,
+      "loss": 0.0619,
+      "step": 7180
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.231023032498103e-06,
+      "loss": 0.0802,
+      "step": 7181
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.227700747757638e-06,
+      "loss": 0.0732,
+      "step": 7182
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.224378548769557e-06,
+      "loss": 0.059,
+      "step": 7183
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.221056435902747e-06,
+      "loss": 0.0702,
+      "step": 7184
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.217734409526067e-06,
+      "loss": 0.0578,
+      "step": 7185
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.214412470008392e-06,
+      "loss": 0.0621,
+      "step": 7186
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.211090617718563e-06,
+      "loss": 0.0703,
+      "step": 7187
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.207768853025432e-06,
+      "loss": 0.0576,
+      "step": 7188
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.204447176297826e-06,
+      "loss": 0.0496,
+      "step": 7189
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.201125587904572e-06,
+      "loss": 0.0538,
+      "step": 7190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.197804088214475e-06,
+      "loss": 0.0663,
+      "step": 7191
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.19448267759635e-06,
+      "loss": 0.0764,
+      "step": 7192
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.19116135641898e-06,
+      "loss": 0.0662,
+      "step": 7193
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.187840125051155e-06,
+      "loss": 0.0727,
+      "step": 7194
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.184518983861646e-06,
+      "loss": 0.062,
+      "step": 7195
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.181197933219218e-06,
+      "loss": 0.0729,
+      "step": 7196
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.177876973492618e-06,
+      "loss": 0.0638,
+      "step": 7197
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.174556105050603e-06,
+      "loss": 0.0649,
+      "step": 7198
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.171235328261893e-06,
+      "loss": 0.0658,
+      "step": 7199
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.167914643495217e-06,
+      "loss": 0.0616,
+      "step": 7200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.164594051119286e-06,
+      "loss": 0.0704,
+      "step": 7201
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.161273551502804e-06,
+      "loss": 0.0501,
+      "step": 7202
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.15795314501446e-06,
+      "loss": 0.0645,
+      "step": 7203
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.154632832022945e-06,
+      "loss": 0.0751,
+      "step": 7204
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.15131261289692e-06,
+      "loss": 0.0796,
+      "step": 7205
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.14799248800505e-06,
+      "loss": 0.0812,
+      "step": 7206
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.144672457715985e-06,
+      "loss": 0.0739,
+      "step": 7207
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.141352522398369e-06,
+      "loss": 0.0891,
+      "step": 7208
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.138032682420824e-06,
+      "loss": 0.0673,
+      "step": 7209
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.13471293815198e-06,
+      "loss": 0.0667,
+      "step": 7210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.131393289960435e-06,
+      "loss": 0.067,
+      "step": 7211
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.128073738214792e-06,
+      "loss": 0.0603,
+      "step": 7212
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.124754283283634e-06,
+      "loss": 0.0617,
+      "step": 7213
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.121434925535544e-06,
+      "loss": 0.0703,
+      "step": 7214
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.11811566533908e-06,
+      "loss": 0.072,
+      "step": 7215
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.114796503062805e-06,
+      "loss": 0.0713,
+      "step": 7216
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.111477439075252e-06,
+      "loss": 0.0673,
+      "step": 7217
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.108158473744965e-06,
+      "loss": 0.0635,
+      "step": 7218
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.104839607440457e-06,
+      "loss": 0.0759,
+      "step": 7219
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.101520840530245e-06,
+      "loss": 0.0573,
+      "step": 7220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.098202173382829e-06,
+      "loss": 0.076,
+      "step": 7221
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.094883606366689e-06,
+      "loss": 0.0869,
+      "step": 7222
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.091565139850315e-06,
+      "loss": 0.0707,
+      "step": 7223
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.088246774202162e-06,
+      "loss": 0.0824,
+      "step": 7224
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.084928509790696e-06,
+      "loss": 0.0732,
+      "step": 7225
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.08161034698435e-06,
+      "loss": 0.0613,
+      "step": 7226
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.07829228615157e-06,
+      "loss": 0.0549,
+      "step": 7227
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.074974327660763e-06,
+      "loss": 0.0676,
+      "step": 7228
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.071656471880352e-06,
+      "loss": 0.0582,
+      "step": 7229
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.068338719178726e-06,
+      "loss": 0.0757,
+      "step": 7230
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.06502106992428e-06,
+      "loss": 0.0847,
+      "step": 7231
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.061703524485383e-06,
+      "loss": 0.0611,
+      "step": 7232
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.058386083230404e-06,
+      "loss": 0.0815,
+      "step": 7233
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.05506874652769e-06,
+      "loss": 0.071,
+      "step": 7234
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.051751514745594e-06,
+      "loss": 0.048,
+      "step": 7235
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.048434388252431e-06,
+      "loss": 0.0644,
+      "step": 7236
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.045117367416527e-06,
+      "loss": 0.0778,
+      "step": 7237
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.041800452606186e-06,
+      "loss": 0.0722,
+      "step": 7238
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.038483644189704e-06,
+      "loss": 0.0654,
+      "step": 7239
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.035166942535358e-06,
+      "loss": 0.0778,
+      "step": 7240
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.031850348011429e-06,
+      "loss": 0.0589,
+      "step": 7241
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.02853386098616e-06,
+      "loss": 0.0637,
+      "step": 7242
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.025217481827815e-06,
+      "loss": 0.0682,
+      "step": 7243
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.021901210904616e-06,
+      "loss": 0.0727,
+      "step": 7244
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.018585048584793e-06,
+      "loss": 0.0729,
+      "step": 7245
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.015268995236551e-06,
+      "loss": 0.0804,
+      "step": 7246
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.011953051228096e-06,
+      "loss": 0.0639,
+      "step": 7247
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.008637216927604e-06,
+      "loss": 0.0648,
+      "step": 7248
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.00532149270326e-06,
+      "loss": 0.0791,
+      "step": 7249
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.00200587892322e-06,
+      "loss": 0.0742,
+      "step": 7250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.998690375955635e-06,
+      "loss": 0.07,
+      "step": 7251
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.99537498416864e-06,
+      "loss": 0.0802,
+      "step": 7252
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.992059703930366e-06,
+      "loss": 0.0644,
+      "step": 7253
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.988744535608919e-06,
+      "loss": 0.0911,
+      "step": 7254
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.985429479572407e-06,
+      "loss": 0.0734,
+      "step": 7255
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.982114536188911e-06,
+      "loss": 0.0749,
+      "step": 7256
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.978799705826509e-06,
+      "loss": 0.0753,
+      "step": 7257
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.975484988853263e-06,
+      "loss": 0.0724,
+      "step": 7258
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.972170385637224e-06,
+      "loss": 0.0723,
+      "step": 7259
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.96885589654643e-06,
+      "loss": 0.069,
+      "step": 7260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.965541521948907e-06,
+      "loss": 0.0469,
+      "step": 7261
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.96222726221266e-06,
+      "loss": 0.0686,
+      "step": 7262
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.9589131177057e-06,
+      "loss": 0.088,
+      "step": 7263
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.955599088796002e-06,
+      "loss": 0.065,
+      "step": 7264
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.952285175851548e-06,
+      "loss": 0.072,
+      "step": 7265
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.948971379240294e-06,
+      "loss": 0.077,
+      "step": 7266
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.945657699330195e-06,
+      "loss": 0.0627,
+      "step": 7267
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.942344136489174e-06,
+      "loss": 0.0814,
+      "step": 7268
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.939030691085168e-06,
+      "loss": 0.0668,
+      "step": 7269
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.935717363486074e-06,
+      "loss": 0.0669,
+      "step": 7270
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.932404154059793e-06,
+      "loss": 0.0826,
+      "step": 7271
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.929091063174205e-06,
+      "loss": 0.0754,
+      "step": 7272
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.925778091197183e-06,
+      "loss": 0.0706,
+      "step": 7273
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.92246523849658e-06,
+      "loss": 0.0605,
+      "step": 7274
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.919152505440248e-06,
+      "loss": 0.0745,
+      "step": 7275
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.915839892396004e-06,
+      "loss": 0.066,
+      "step": 7276
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.912527399731674e-06,
+      "loss": 0.068,
+      "step": 7277
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.909215027815053e-06,
+      "loss": 0.0876,
+      "step": 7278
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.90590277701394e-06,
+      "loss": 0.0726,
+      "step": 7279
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 8.902590647696104e-06,
+      "loss": 0.0539,
+      "step": 7280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.899278640229316e-06,
+      "loss": 0.08,
+      "step": 7281
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.895966754981316e-06,
+      "loss": 0.0782,
+      "step": 7282
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.892654992319846e-06,
+      "loss": 0.0715,
+      "step": 7283
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.889343352612624e-06,
+      "loss": 0.0732,
+      "step": 7284
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.886031836227362e-06,
+      "loss": 0.072,
+      "step": 7285
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.882720443531752e-06,
+      "loss": 0.0632,
+      "step": 7286
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.87940917489348e-06,
+      "loss": 0.0644,
+      "step": 7287
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.87609803068021e-06,
+      "loss": 0.0692,
+      "step": 7288
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.872787011259592e-06,
+      "loss": 0.0679,
+      "step": 7289
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.86947611699927e-06,
+      "loss": 0.0617,
+      "step": 7290
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.866165348266869e-06,
+      "loss": 0.0747,
+      "step": 7291
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.862854705430003e-06,
+      "loss": 0.0683,
+      "step": 7292
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.859544188856264e-06,
+      "loss": 0.0668,
+      "step": 7293
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.856233798913241e-06,
+      "loss": 0.0555,
+      "step": 7294
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.852923535968499e-06,
+      "loss": 0.0585,
+      "step": 7295
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.849613400389599e-06,
+      "loss": 0.0607,
+      "step": 7296
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.846303392544077e-06,
+      "loss": 0.0758,
+      "step": 7297
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.84299351279947e-06,
+      "loss": 0.0605,
+      "step": 7298
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.839683761523277e-06,
+      "loss": 0.0645,
+      "step": 7299
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.836374139083011e-06,
+      "loss": 0.0626,
+      "step": 7300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.833064645846144e-06,
+      "loss": 0.0532,
+      "step": 7301
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.829755282180155e-06,
+      "loss": 0.0758,
+      "step": 7302
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.826446048452494e-06,
+      "loss": 0.0702,
+      "step": 7303
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.823136945030607e-06,
+      "loss": 0.0682,
+      "step": 7304
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.819827972281917e-06,
+      "loss": 0.0769,
+      "step": 7305
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.816519130573844e-06,
+      "loss": 0.0771,
+      "step": 7306
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.813210420273775e-06,
+      "loss": 0.0711,
+      "step": 7307
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.809901841749101e-06,
+      "loss": 0.0651,
+      "step": 7308
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.806593395367186e-06,
+      "loss": 0.0607,
+      "step": 7309
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.803285081495388e-06,
+      "loss": 0.0646,
+      "step": 7310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.799976900501041e-06,
+      "loss": 0.0698,
+      "step": 7311
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.796668852751476e-06,
+      "loss": 0.0696,
+      "step": 7312
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.793360938613997e-06,
+      "loss": 0.0691,
+      "step": 7313
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.790053158455904e-06,
+      "loss": 0.0746,
+      "step": 7314
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.78674551264447e-06,
+      "loss": 0.0812,
+      "step": 7315
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.783438001546967e-06,
+      "loss": 0.0848,
+      "step": 7316
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.78013062553064e-06,
+      "loss": 0.0765,
+      "step": 7317
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.77682338496273e-06,
+      "loss": 0.0706,
+      "step": 7318
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 8.773516280210448e-06,
+      "loss": 0.0929,
+      "step": 7319
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.77020931164101e-06,
+      "loss": 0.0782,
+      "step": 7320
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.766902479621596e-06,
+      "loss": 0.0641,
+      "step": 7321
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.763595784519388e-06,
+      "loss": 0.0622,
+      "step": 7322
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.760289226701538e-06,
+      "loss": 0.0634,
+      "step": 7323
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.756982806535201e-06,
+      "loss": 0.0618,
+      "step": 7324
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.753676524387491e-06,
+      "loss": 0.0756,
+      "step": 7325
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.750370380625539e-06,
+      "loss": 0.058,
+      "step": 7326
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.747064375616428e-06,
+      "loss": 0.0631,
+      "step": 7327
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.74375850972725e-06,
+      "loss": 0.0598,
+      "step": 7328
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.740452783325069e-06,
+      "loss": 0.0715,
+      "step": 7329
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.737147196776938e-06,
+      "loss": 0.0844,
+      "step": 7330
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.73384175044989e-06,
+      "loss": 0.0614,
+      "step": 7331
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.730536444710955e-06,
+      "loss": 0.0635,
+      "step": 7332
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.727231279927128e-06,
+      "loss": 0.0671,
+      "step": 7333
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.723926256465402e-06,
+      "loss": 0.0603,
+      "step": 7334
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.720621374692752e-06,
+      "loss": 0.0755,
+      "step": 7335
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.717316634976137e-06,
+      "loss": 0.0651,
+      "step": 7336
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.714012037682496e-06,
+      "loss": 0.0828,
+      "step": 7337
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.71070758317876e-06,
+      "loss": 0.0645,
+      "step": 7338
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.707403271831835e-06,
+      "loss": 0.0639,
+      "step": 7339
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.70409910400862e-06,
+      "loss": 0.0561,
+      "step": 7340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.700795080075989e-06,
+      "loss": 0.0677,
+      "step": 7341
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.697491200400809e-06,
+      "loss": 0.0598,
+      "step": 7342
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.694187465349925e-06,
+      "loss": 0.0701,
+      "step": 7343
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.690883875290173e-06,
+      "loss": 0.0637,
+      "step": 7344
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.687580430588355e-06,
+      "loss": 0.0776,
+      "step": 7345
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.684277131611284e-06,
+      "loss": 0.0459,
+      "step": 7346
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.680973978725733e-06,
+      "loss": 0.067,
+      "step": 7347
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.677670972298472e-06,
+      "loss": 0.0751,
+      "step": 7348
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.67436811269625e-06,
+      "loss": 0.0861,
+      "step": 7349
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.671065400285802e-06,
+      "loss": 0.0641,
+      "step": 7350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.667762835433842e-06,
+      "loss": 0.0773,
+      "step": 7351
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.664460418507077e-06,
+      "loss": 0.0714,
+      "step": 7352
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.661158149872183e-06,
+      "loss": 0.0759,
+      "step": 7353
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.657856029895835e-06,
+      "loss": 0.0609,
+      "step": 7354
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.654554058944683e-06,
+      "loss": 0.0665,
+      "step": 7355
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.651252237385354e-06,
+      "loss": 0.0544,
+      "step": 7356
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.647950565584483e-06,
+      "loss": 0.0688,
+      "step": 7357
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.644649043908652e-06,
+      "loss": 0.066,
+      "step": 7358
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.641347672724463e-06,
+      "loss": 0.0669,
+      "step": 7359
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.638046452398473e-06,
+      "loss": 0.0667,
+      "step": 7360
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.634745383297239e-06,
+      "loss": 0.0619,
+      "step": 7361
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.631444465787292e-06,
+      "loss": 0.0927,
+      "step": 7362
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.628143700235159e-06,
+      "loss": 0.0707,
+      "step": 7363
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.62484308700733e-06,
+      "loss": 0.0686,
+      "step": 7364
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.621542626470295e-06,
+      "loss": 0.0612,
+      "step": 7365
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.618242318990517e-06,
+      "loss": 0.0726,
+      "step": 7366
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.614942164934453e-06,
+      "loss": 0.061,
+      "step": 7367
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.61164216466853e-06,
+      "loss": 0.0738,
+      "step": 7368
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.608342318559171e-06,
+      "loss": 0.0681,
+      "step": 7369
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.605042626972766e-06,
+      "loss": 0.077,
+      "step": 7370
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.601743090275705e-06,
+      "loss": 0.0663,
+      "step": 7371
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.598443708834345e-06,
+      "loss": 0.0641,
+      "step": 7372
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.595144483015038e-06,
+      "loss": 0.0662,
+      "step": 7373
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.591845413184114e-06,
+      "loss": 0.0723,
+      "step": 7374
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.58854649970789e-06,
+      "loss": 0.054,
+      "step": 7375
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.58524774295265e-06,
+      "loss": 0.0716,
+      "step": 7376
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 8.581949143284685e-06,
+      "loss": 0.0638,
+      "step": 7377
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.578650701070244e-06,
+      "loss": 0.0508,
+      "step": 7378
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.57535241667558e-06,
+      "loss": 0.0713,
+      "step": 7379
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.572054290466911e-06,
+      "loss": 0.0681,
+      "step": 7380
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.56875632281045e-06,
+      "loss": 0.0691,
+      "step": 7381
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.565458514072385e-06,
+      "loss": 0.0606,
+      "step": 7382
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.56216086461889e-06,
+      "loss": 0.0724,
+      "step": 7383
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.55886337481612e-06,
+      "loss": 0.0732,
+      "step": 7384
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.555566045030211e-06,
+      "loss": 0.0592,
+      "step": 7385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.552268875627284e-06,
+      "loss": 0.0703,
+      "step": 7386
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.548971866973439e-06,
+      "loss": 0.0673,
+      "step": 7387
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.545675019434761e-06,
+      "loss": 0.064,
+      "step": 7388
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.54237833337732e-06,
+      "loss": 0.0539,
+      "step": 7389
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.539081809167157e-06,
+      "loss": 0.0664,
+      "step": 7390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.535785447170307e-06,
+      "loss": 0.0531,
+      "step": 7391
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.532489247752778e-06,
+      "loss": 0.0804,
+      "step": 7392
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.529193211280572e-06,
+      "loss": 0.0644,
+      "step": 7393
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.525897338119655e-06,
+      "loss": 0.0604,
+      "step": 7394
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.522601628635993e-06,
+      "loss": 0.0568,
+      "step": 7395
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.519306083195518e-06,
+      "loss": 0.0613,
+      "step": 7396
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.516010702164162e-06,
+      "loss": 0.0647,
+      "step": 7397
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.512715485907816e-06,
+      "loss": 0.0662,
+      "step": 7398
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.509420434792374e-06,
+      "loss": 0.0681,
+      "step": 7399
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.506125549183698e-06,
+      "loss": 0.0841,
+      "step": 7400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.502830829447642e-06,
+      "loss": 0.0681,
+      "step": 7401
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.499536275950025e-06,
+      "loss": 0.068,
+      "step": 7402
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.496241889056671e-06,
+      "loss": 0.076,
+      "step": 7403
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.492947669133362e-06,
+      "loss": 0.0742,
+      "step": 7404
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.489653616545881e-06,
+      "loss": 0.062,
+      "step": 7405
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.486359731659977e-06,
+      "loss": 0.0781,
+      "step": 7406
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.48306601484139e-06,
+      "loss": 0.0476,
+      "step": 7407
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.479772466455838e-06,
+      "loss": 0.0623,
+      "step": 7408
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.476479086869024e-06,
+      "loss": 0.0767,
+      "step": 7409
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.473185876446623e-06,
+      "loss": 0.0651,
+      "step": 7410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.469892835554302e-06,
+      "loss": 0.0909,
+      "step": 7411
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.4665999645577e-06,
+      "loss": 0.0758,
+      "step": 7412
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.463307263822447e-06,
+      "loss": 0.0663,
+      "step": 7413
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.460014733714144e-06,
+      "loss": 0.0653,
+      "step": 7414
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.456722374598381e-06,
+      "loss": 0.0567,
+      "step": 7415
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.453430186840721e-06,
+      "loss": 0.0699,
+      "step": 7416
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.45013817080672e-06,
+      "loss": 0.0856,
+      "step": 7417
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.446846326861899e-06,
+      "loss": 0.0677,
+      "step": 7418
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.443554655371776e-06,
+      "loss": 0.0532,
+      "step": 7419
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.440263156701835e-06,
+      "loss": 0.0645,
+      "step": 7420
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.43697183121756e-06,
+      "loss": 0.0687,
+      "step": 7421
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.43368067928439e-06,
+      "loss": 0.0671,
+      "step": 7422
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.430389701267765e-06,
+      "loss": 0.0609,
+      "step": 7423
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.427098897533101e-06,
+      "loss": 0.0684,
+      "step": 7424
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.42380826844579e-06,
+      "loss": 0.0654,
+      "step": 7425
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.420517814371211e-06,
+      "loss": 0.0486,
+      "step": 7426
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.417227535674712e-06,
+      "loss": 0.0614,
+      "step": 7427
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.413937432721644e-06,
+      "loss": 0.0779,
+      "step": 7428
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.410647505877308e-06,
+      "loss": 0.072,
+      "step": 7429
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.407357755507013e-06,
+      "loss": 0.0715,
+      "step": 7430
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.404068181976032e-06,
+      "loss": 0.057,
+      "step": 7431
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.400778785649626e-06,
+      "loss": 0.0812,
+      "step": 7432
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.39748956689303e-06,
+      "loss": 0.0662,
+      "step": 7433
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.39420052607147e-06,
+      "loss": 0.0692,
+      "step": 7434
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 8.390911663550137e-06,
+      "loss": 0.0567,
+      "step": 7435
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.387622979694215e-06,
+      "loss": 0.0607,
+      "step": 7436
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.38433447486886e-06,
+      "loss": 0.0818,
+      "step": 7437
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.38104614943922e-06,
+      "loss": 0.0638,
+      "step": 7438
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.377758003770404e-06,
+      "loss": 0.0684,
+      "step": 7439
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.374470038227522e-06,
+      "loss": 0.0676,
+      "step": 7440
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.371182253175645e-06,
+      "loss": 0.066,
+      "step": 7441
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.36789464897984e-06,
+      "loss": 0.0601,
+      "step": 7442
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.36460722600514e-06,
+      "loss": 0.0798,
+      "step": 7443
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.361319984616568e-06,
+      "loss": 0.0782,
+      "step": 7444
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.358032925179124e-06,
+      "loss": 0.0846,
+      "step": 7445
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.35474604805779e-06,
+      "loss": 0.0652,
+      "step": 7446
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.351459353617515e-06,
+      "loss": 0.065,
+      "step": 7447
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.348172842223251e-06,
+      "loss": 0.0632,
+      "step": 7448
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.344886514239906e-06,
+      "loss": 0.0583,
+      "step": 7449
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.341600370032384e-06,
+      "loss": 0.067,
+      "step": 7450
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.338314409965557e-06,
+      "loss": 0.0628,
+      "step": 7451
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.335028634404292e-06,
+      "loss": 0.0765,
+      "step": 7452
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.33174304371341e-06,
+      "loss": 0.0764,
+      "step": 7453
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.328457638257743e-06,
+      "loss": 0.0702,
+      "step": 7454
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.325172418402076e-06,
+      "loss": 0.0776,
+      "step": 7455
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.321887384511191e-06,
+      "loss": 0.0687,
+      "step": 7456
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.318602536949837e-06,
+      "loss": 0.0574,
+      "step": 7457
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.315317876082752e-06,
+      "loss": 0.0703,
+      "step": 7458
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.312033402274644e-06,
+      "loss": 0.0798,
+      "step": 7459
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.308749115890212e-06,
+      "loss": 0.0593,
+      "step": 7460
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.30546501729412e-06,
+      "loss": 0.0667,
+      "step": 7461
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.302181106851022e-06,
+      "loss": 0.0679,
+      "step": 7462
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.298897384925547e-06,
+      "loss": 0.0664,
+      "step": 7463
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.295613851882305e-06,
+      "loss": 0.0602,
+      "step": 7464
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.29233050808588e-06,
+      "loss": 0.0759,
+      "step": 7465
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.289047353900847e-06,
+      "loss": 0.0643,
+      "step": 7466
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.28576438969174e-06,
+      "loss": 0.0643,
+      "step": 7467
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.282481615823092e-06,
+      "loss": 0.0526,
+      "step": 7468
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.279199032659403e-06,
+      "loss": 0.0717,
+      "step": 7469
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.275916640565157e-06,
+      "loss": 0.0799,
+      "step": 7470
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.272634439904812e-06,
+      "loss": 0.0662,
+      "step": 7471
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.269352431042813e-06,
+      "loss": 0.0657,
+      "step": 7472
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.266070614343569e-06,
+      "loss": 0.0599,
+      "step": 7473
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.26278899017149e-06,
+      "loss": 0.0535,
+      "step": 7474
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.259507558890941e-06,
+      "loss": 0.0806,
+      "step": 7475
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.256226320866282e-06,
+      "loss": 0.0661,
+      "step": 7476
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.252945276461842e-06,
+      "loss": 0.0581,
+      "step": 7477
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.249664426041936e-06,
+      "loss": 0.0743,
+      "step": 7478
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.246383769970852e-06,
+      "loss": 0.0718,
+      "step": 7479
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.24310330861286e-06,
+      "loss": 0.0522,
+      "step": 7480
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.239823042332203e-06,
+      "loss": 0.0663,
+      "step": 7481
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.23654297149311e-06,
+      "loss": 0.0683,
+      "step": 7482
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.23326309645978e-06,
+      "loss": 0.0757,
+      "step": 7483
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.229983417596399e-06,
+      "loss": 0.0789,
+      "step": 7484
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.226703935267121e-06,
+      "loss": 0.0706,
+      "step": 7485
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.223424649836093e-06,
+      "loss": 0.0685,
+      "step": 7486
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.220145561667421e-06,
+      "loss": 0.0604,
+      "step": 7487
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.216866671125205e-06,
+      "loss": 0.0652,
+      "step": 7488
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.213587978573517e-06,
+      "loss": 0.0721,
+      "step": 7489
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.210309484376404e-06,
+      "loss": 0.0837,
+      "step": 7490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.2070311888979e-06,
+      "loss": 0.0626,
+      "step": 7491
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.203753092502002e-06,
+      "loss": 0.0817,
+      "step": 7492
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.200475195552703e-06,
+      "loss": 0.0493,
+      "step": 7493
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.197197498413958e-06,
+      "loss": 0.0608,
+      "step": 7494
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.193920001449713e-06,
+      "loss": 0.0595,
+      "step": 7495
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.19064270502388e-06,
+      "loss": 0.0636,
+      "step": 7496
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.18736560950036e-06,
+      "loss": 0.0673,
+      "step": 7497
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.184088715243019e-06,
+      "loss": 0.0778,
+      "step": 7498
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.180812022615714e-06,
+      "loss": 0.0722,
+      "step": 7499
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.177535531982266e-06,
+      "loss": 0.0551,
+      "step": 7500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.174259243706488e-06,
+      "loss": 0.069,
+      "step": 7501
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.17098315815216e-06,
+      "loss": 0.0677,
+      "step": 7502
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.167707275683043e-06,
+      "loss": 0.0694,
+      "step": 7503
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.164431596662872e-06,
+      "loss": 0.0687,
+      "step": 7504
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.16115612145537e-06,
+      "loss": 0.0604,
+      "step": 7505
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.157880850424222e-06,
+      "loss": 0.0646,
+      "step": 7506
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.154605783933104e-06,
+      "loss": 0.0828,
+      "step": 7507
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.15133092234566e-06,
+      "loss": 0.0605,
+      "step": 7508
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.148056266025517e-06,
+      "loss": 0.059,
+      "step": 7509
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.144781815336276e-06,
+      "loss": 0.0657,
+      "step": 7510
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.141507570641522e-06,
+      "loss": 0.0822,
+      "step": 7511
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.1382335323048e-06,
+      "loss": 0.0646,
+      "step": 7512
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.13495970068965e-06,
+      "loss": 0.0617,
+      "step": 7513
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.131686076159582e-06,
+      "loss": 0.0776,
+      "step": 7514
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.128412659078084e-06,
+      "loss": 0.0702,
+      "step": 7515
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.125139449808618e-06,
+      "loss": 0.0681,
+      "step": 7516
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.121866448714633e-06,
+      "loss": 0.0588,
+      "step": 7517
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.118593656159536e-06,
+      "loss": 0.0731,
+      "step": 7518
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.11532107250673e-06,
+      "loss": 0.0536,
+      "step": 7519
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.11204869811958e-06,
+      "loss": 0.0572,
+      "step": 7520
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.108776533361445e-06,
+      "loss": 0.0589,
+      "step": 7521
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.10550457859564e-06,
+      "loss": 0.0801,
+      "step": 7522
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.102232834185478e-06,
+      "loss": 0.0619,
+      "step": 7523
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.098961300494221e-06,
+      "loss": 0.0505,
+      "step": 7524
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.095689977885144e-06,
+      "loss": 0.058,
+      "step": 7525
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.092418866721466e-06,
+      "loss": 0.0462,
+      "step": 7526
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.0891479673664e-06,
+      "loss": 0.0712,
+      "step": 7527
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.085877280183126e-06,
+      "loss": 0.0673,
+      "step": 7528
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.082606805534817e-06,
+      "loss": 0.0623,
+      "step": 7529
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.079336543784592e-06,
+      "loss": 0.0724,
+      "step": 7530
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.076066495295586e-06,
+      "loss": 0.0602,
+      "step": 7531
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.072796660430873e-06,
+      "loss": 0.0742,
+      "step": 7532
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.06952703955353e-06,
+      "loss": 0.0677,
+      "step": 7533
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.066257633026594e-06,
+      "loss": 0.0895,
+      "step": 7534
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.062988441213087e-06,
+      "loss": 0.0636,
+      "step": 7535
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.059719464476e-06,
+      "loss": 0.0851,
+      "step": 7536
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.056450703178316e-06,
+      "loss": 0.0544,
+      "step": 7537
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.053182157682968e-06,
+      "loss": 0.0635,
+      "step": 7538
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.049913828352889e-06,
+      "loss": 0.0663,
+      "step": 7539
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.046645715550972e-06,
+      "loss": 0.0802,
+      "step": 7540
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.043377819640099e-06,
+      "loss": 0.0714,
+      "step": 7541
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.040110140983115e-06,
+      "loss": 0.0707,
+      "step": 7542
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.036842679942856e-06,
+      "loss": 0.0585,
+      "step": 7543
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.033575436882115e-06,
+      "loss": 0.0561,
+      "step": 7544
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.03030841216368e-06,
+      "loss": 0.0802,
+      "step": 7545
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.027041606150296e-06,
+      "loss": 0.0552,
+      "step": 7546
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.023775019204703e-06,
+      "loss": 0.0554,
+      "step": 7547
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.0205086516896e-06,
+      "loss": 0.0519,
+      "step": 7548
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.017242503967675e-06,
+      "loss": 0.0712,
+      "step": 7549
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.013976576401575e-06,
+      "loss": 0.0733,
+      "step": 7550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.010710869353948e-06,
+      "loss": 0.0668,
+      "step": 7551
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.007445383187387e-06,
+      "loss": 0.0862,
+      "step": 7552
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.004180118264487e-06,
+      "loss": 0.0746,
+      "step": 7553
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.000915074947802e-06,
+      "loss": 0.0747,
+      "step": 7554
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.997650253599868e-06,
+      "loss": 0.0748,
+      "step": 7555
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.994385654583192e-06,
+      "loss": 0.0737,
+      "step": 7556
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.991121278260267e-06,
+      "loss": 0.0636,
+      "step": 7557
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.98785712499355e-06,
+      "loss": 0.0625,
+      "step": 7558
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.984593195145468e-06,
+      "loss": 0.0579,
+      "step": 7559
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.981329489078447e-06,
+      "loss": 0.0636,
+      "step": 7560
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.97806600715486e-06,
+      "loss": 0.0692,
+      "step": 7561
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.974802749737079e-06,
+      "loss": 0.0626,
+      "step": 7562
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.971539717187431e-06,
+      "loss": 0.0703,
+      "step": 7563
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.968276909868234e-06,
+      "loss": 0.0634,
+      "step": 7564
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.96501432814177e-06,
+      "loss": 0.0786,
+      "step": 7565
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.961751972370306e-06,
+      "loss": 0.0604,
+      "step": 7566
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.958489842916072e-06,
+      "loss": 0.0765,
+      "step": 7567
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.955227940141287e-06,
+      "loss": 0.0782,
+      "step": 7568
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.951966264408126e-06,
+      "loss": 0.0649,
+      "step": 7569
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.94870481607876e-06,
+      "loss": 0.0645,
+      "step": 7570
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 7.945443595515317e-06,
+      "loss": 0.0731,
+      "step": 7571
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.94218260307991e-06,
+      "loss": 0.0638,
+      "step": 7572
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.938921839134626e-06,
+      "loss": 0.0651,
+      "step": 7573
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.935661304041524e-06,
+      "loss": 0.066,
+      "step": 7574
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.932400998162635e-06,
+      "loss": 0.0768,
+      "step": 7575
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.92914092185997e-06,
+      "loss": 0.0664,
+      "step": 7576
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.92588107549551e-06,
+      "loss": 0.0623,
+      "step": 7577
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.922621459431216e-06,
+      "loss": 0.0577,
+      "step": 7578
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.919362074029018e-06,
+      "loss": 0.0763,
+      "step": 7579
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.916102919650826e-06,
+      "loss": 0.074,
+      "step": 7580
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.91284399665851e-06,
+      "loss": 0.0656,
+      "step": 7581
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.90958530541394e-06,
+      "loss": 0.0701,
+      "step": 7582
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.906326846278934e-06,
+      "loss": 0.0804,
+      "step": 7583
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.9030686196153e-06,
+      "loss": 0.0724,
+      "step": 7584
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.899810625784815e-06,
+      "loss": 0.0691,
+      "step": 7585
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.896552865149232e-06,
+      "loss": 0.085,
+      "step": 7586
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.893295338070278e-06,
+      "loss": 0.0748,
+      "step": 7587
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.890038044909651e-06,
+      "loss": 0.0671,
+      "step": 7588
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.886780986029023e-06,
+      "loss": 0.0533,
+      "step": 7589
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.883524161790048e-06,
+      "loss": 0.0544,
+      "step": 7590
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 7.880267572554341e-06,
+      "loss": 0.0704,
+      "step": 7591
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.877011218683503e-06,
+      "loss": 0.0594,
+      "step": 7592
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.873755100539102e-06,
+      "loss": 0.0844,
+      "step": 7593
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.870499218482687e-06,
+      "loss": 0.0787,
+      "step": 7594
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.867243572875766e-06,
+      "loss": 0.072,
+      "step": 7595
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.863988164079837e-06,
+      "loss": 0.0854,
+      "step": 7596
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.86073299245636e-06,
+      "loss": 0.0604,
+      "step": 7597
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.857478058366778e-06,
+      "loss": 0.0598,
+      "step": 7598
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.854223362172499e-06,
+      "loss": 0.0627,
+      "step": 7599
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.850968904234918e-06,
+      "loss": 0.0781,
+      "step": 7600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.847714684915379e-06,
+      "loss": 0.0815,
+      "step": 7601
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.84446070457523e-06,
+      "loss": 0.0828,
+      "step": 7602
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.841206963575767e-06,
+      "loss": 0.0724,
+      "step": 7603
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.837953462278273e-06,
+      "loss": 0.067,
+      "step": 7604
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.834700201044002e-06,
+      "loss": 0.0633,
+      "step": 7605
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.831447180234182e-06,
+      "loss": 0.0649,
+      "step": 7606
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.828194400210007e-06,
+      "loss": 0.0605,
+      "step": 7607
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.82494186133266e-06,
+      "loss": 0.0662,
+      "step": 7608
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.821689563963276e-06,
+      "loss": 0.0658,
+      "step": 7609
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 7.818437508462981e-06,
+      "loss": 0.0822,
+      "step": 7610
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.815185695192862e-06,
+      "loss": 0.0532,
+      "step": 7611
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.811934124513994e-06,
+      "loss": 0.07,
+      "step": 7612
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.808682796787406e-06,
+      "loss": 0.0648,
+      "step": 7613
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.805431712374119e-06,
+      "loss": 0.062,
+      "step": 7614
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.802180871635107e-06,
+      "loss": 0.0593,
+      "step": 7615
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.798930274931338e-06,
+      "loss": 0.0612,
+      "step": 7616
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.795679922623734e-06,
+      "loss": 0.0711,
+      "step": 7617
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.792429815073206e-06,
+      "loss": 0.0596,
+      "step": 7618
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.789179952640625e-06,
+      "loss": 0.0715,
+      "step": 7619
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.785930335686845e-06,
+      "loss": 0.0626,
+      "step": 7620
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.782680964572676e-06,
+      "loss": 0.0609,
+      "step": 7621
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.779431839658931e-06,
+      "loss": 0.0693,
+      "step": 7622
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.776182961306361e-06,
+      "loss": 0.0562,
+      "step": 7623
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.772934329875714e-06,
+      "loss": 0.0773,
+      "step": 7624
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.769685945727703e-06,
+      "loss": 0.0809,
+      "step": 7625
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.766437809223008e-06,
+      "loss": 0.071,
+      "step": 7626
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.76318992072229e-06,
+      "loss": 0.0669,
+      "step": 7627
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.759942280586174e-06,
+      "loss": 0.0511,
+      "step": 7628
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.756694889175269e-06,
+      "loss": 0.0683,
+      "step": 7629
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.753447746850145e-06,
+      "loss": 0.0576,
+      "step": 7630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.750200853971354e-06,
+      "loss": 0.0654,
+      "step": 7631
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.746954210899404e-06,
+      "loss": 0.0634,
+      "step": 7632
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.743707817994804e-06,
+      "loss": 0.084,
+      "step": 7633
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.740461675618004e-06,
+      "loss": 0.0794,
+      "step": 7634
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.737215784129443e-06,
+      "loss": 0.088,
+      "step": 7635
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.733970143889531e-06,
+      "loss": 0.055,
+      "step": 7636
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.73072475525865e-06,
+      "loss": 0.0748,
+      "step": 7637
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.727479618597148e-06,
+      "loss": 0.0617,
+      "step": 7638
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.724234734265355e-06,
+      "loss": 0.0646,
+      "step": 7639
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.72099010262356e-06,
+      "loss": 0.0615,
+      "step": 7640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.717745724032036e-06,
+      "loss": 0.0612,
+      "step": 7641
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.714501598851021e-06,
+      "loss": 0.0607,
+      "step": 7642
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.711257727440729e-06,
+      "loss": 0.0832,
+      "step": 7643
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.708014110161342e-06,
+      "loss": 0.0737,
+      "step": 7644
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.70477074737302e-06,
+      "loss": 0.0566,
+      "step": 7645
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.701527639435883e-06,
+      "loss": 0.0665,
+      "step": 7646
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.698284786710036e-06,
+      "loss": 0.0601,
+      "step": 7647
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.695042189555547e-06,
+      "loss": 0.0751,
+      "step": 7648
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 7.691799848332459e-06,
+      "loss": 0.0685,
+      "step": 7649
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.688557763400785e-06,
+      "loss": 0.0688,
+      "step": 7650
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.685315935120517e-06,
+      "loss": 0.0575,
+      "step": 7651
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.682074363851598e-06,
+      "loss": 0.0541,
+      "step": 7652
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.678833049953972e-06,
+      "loss": 0.0632,
+      "step": 7653
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.675591993787528e-06,
+      "loss": 0.0539,
+      "step": 7654
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.672351195712142e-06,
+      "loss": 0.0707,
+      "step": 7655
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.669110656087655e-06,
+      "loss": 0.0707,
+      "step": 7656
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.665870375273885e-06,
+      "loss": 0.0475,
+      "step": 7657
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.662630353630606e-06,
+      "loss": 0.0613,
+      "step": 7658
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.65939059151759e-06,
+      "loss": 0.0676,
+      "step": 7659
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.656151089294553e-06,
+      "loss": 0.072,
+      "step": 7660
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.652911847321199e-06,
+      "loss": 0.0573,
+      "step": 7661
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.649672865957194e-06,
+      "loss": 0.077,
+      "step": 7662
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.646434145562183e-06,
+      "loss": 0.0678,
+      "step": 7663
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.643195686495773e-06,
+      "loss": 0.0566,
+      "step": 7664
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.639957489117555e-06,
+      "loss": 0.0718,
+      "step": 7665
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.636719553787073e-06,
+      "loss": 0.0722,
+      "step": 7666
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.633481880863859e-06,
+      "loss": 0.0707,
+      "step": 7667
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.630244470707404e-06,
+      "loss": 0.0784,
+      "step": 7668
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.627007323677177e-06,
+      "loss": 0.0471,
+      "step": 7669
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.623770440132613e-06,
+      "loss": 0.063,
+      "step": 7670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.620533820433126e-06,
+      "loss": 0.0663,
+      "step": 7671
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.617297464938086e-06,
+      "loss": 0.0621,
+      "step": 7672
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.614061374006848e-06,
+      "loss": 0.063,
+      "step": 7673
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.610825547998728e-06,
+      "loss": 0.0721,
+      "step": 7674
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.607589987273022e-06,
+      "loss": 0.0704,
+      "step": 7675
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.604354692188986e-06,
+      "loss": 0.0709,
+      "step": 7676
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.601119663105857e-06,
+      "loss": 0.0511,
+      "step": 7677
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.597884900382827e-06,
+      "loss": 0.0771,
+      "step": 7678
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.594650404379082e-06,
+      "loss": 0.0607,
+      "step": 7679
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.591416175453753e-06,
+      "loss": 0.0651,
+      "step": 7680
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.58818221396596e-06,
+      "loss": 0.0601,
+      "step": 7681
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.584948520274784e-06,
+      "loss": 0.071,
+      "step": 7682
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.581715094739279e-06,
+      "loss": 0.0684,
+      "step": 7683
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.578481937718469e-06,
+      "loss": 0.0672,
+      "step": 7684
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.575249049571352e-06,
+      "loss": 0.0779,
+      "step": 7685
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.572016430656884e-06,
+      "loss": 0.0628,
+      "step": 7686
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.5687840813340076e-06,
+      "loss": 0.0649,
+      "step": 7687
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.56555200196162e-06,
+      "loss": 0.0763,
+      "step": 7688
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.562320192898604e-06,
+      "loss": 0.0579,
+      "step": 7689
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.559088654503796e-06,
+      "loss": 0.0578,
+      "step": 7690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.555857387136018e-06,
+      "loss": 0.0814,
+      "step": 7691
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.552626391154047e-06,
+      "loss": 0.062,
+      "step": 7692
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.5493956669166415e-06,
+      "loss": 0.0546,
+      "step": 7693
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.5461652147825255e-06,
+      "loss": 0.065,
+      "step": 7694
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.54293503511039e-06,
+      "loss": 0.0569,
+      "step": 7695
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.539705128258904e-06,
+      "loss": 0.0675,
+      "step": 7696
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.536475494586695e-06,
+      "loss": 0.0662,
+      "step": 7697
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.533246134452368e-06,
+      "loss": 0.0676,
+      "step": 7698
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.530017048214495e-06,
+      "loss": 0.0639,
+      "step": 7699
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.526788236231622e-06,
+      "loss": 0.0632,
+      "step": 7700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.523559698862254e-06,
+      "loss": 0.0758,
+      "step": 7701
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.520331436464881e-06,
+      "loss": 0.0721,
+      "step": 7702
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.517103449397944e-06,
+      "loss": 0.0575,
+      "step": 7703
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.5138757380198714e-06,
+      "loss": 0.0679,
+      "step": 7704
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.510648302689045e-06,
+      "loss": 0.0597,
+      "step": 7705
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.507421143763831e-06,
+      "loss": 0.076,
+      "step": 7706
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.504194261602553e-06,
+      "loss": 0.0779,
+      "step": 7707
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.500967656563513e-06,
+      "loss": 0.0702,
+      "step": 7708
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.497741329004968e-06,
+      "loss": 0.0756,
+      "step": 7709
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.494515279285166e-06,
+      "loss": 0.0554,
+      "step": 7710
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.4912895077623025e-06,
+      "loss": 0.0611,
+      "step": 7711
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.488064014794558e-06,
+      "loss": 0.0673,
+      "step": 7712
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.4848388007400705e-06,
+      "loss": 0.0657,
+      "step": 7713
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.481613865956958e-06,
+      "loss": 0.0647,
+      "step": 7714
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.478389210803296e-06,
+      "loss": 0.0726,
+      "step": 7715
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.475164835637141e-06,
+      "loss": 0.0753,
+      "step": 7716
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.471940740816504e-06,
+      "loss": 0.059,
+      "step": 7717
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.46871692669938e-06,
+      "loss": 0.077,
+      "step": 7718
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.465493393643719e-06,
+      "loss": 0.0685,
+      "step": 7719
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.462270142007455e-06,
+      "loss": 0.0704,
+      "step": 7720
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.459047172148474e-06,
+      "loss": 0.0656,
+      "step": 7721
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.455824484424647e-06,
+      "loss": 0.0691,
+      "step": 7722
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.4526020791937995e-06,
+      "loss": 0.0656,
+      "step": 7723
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.4493799568137335e-06,
+      "loss": 0.0593,
+      "step": 7724
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.446158117642218e-06,
+      "loss": 0.0726,
+      "step": 7725
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.442936562036993e-06,
+      "loss": 0.0585,
+      "step": 7726
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.439715290355759e-06,
+      "loss": 0.0623,
+      "step": 7727
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.436494302956198e-06,
+      "loss": 0.067,
+      "step": 7728
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.433273600195943e-06,
+      "loss": 0.0803,
+      "step": 7729
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.430053182432617e-06,
+      "loss": 0.0715,
+      "step": 7730
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.426833050023791e-06,
+      "loss": 0.0626,
+      "step": 7731
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.423613203327016e-06,
+      "loss": 0.0551,
+      "step": 7732
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.4203936426998065e-06,
+      "loss": 0.0823,
+      "step": 7733
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.417174368499653e-06,
+      "loss": 0.0813,
+      "step": 7734
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.413955381083997e-06,
+      "loss": 0.0599,
+      "step": 7735
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.410736680810272e-06,
+      "loss": 0.0791,
+      "step": 7736
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.407518268035857e-06,
+      "loss": 0.0653,
+      "step": 7737
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.404300143118115e-06,
+      "loss": 0.0597,
+      "step": 7738
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.401082306414367e-06,
+      "loss": 0.0659,
+      "step": 7739
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.39786475828191e-06,
+      "loss": 0.0603,
+      "step": 7740
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.394647499078001e-06,
+      "loss": 0.0751,
+      "step": 7741
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.391430529159875e-06,
+      "loss": 0.0762,
+      "step": 7742
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.38821384888472e-06,
+      "loss": 0.0765,
+      "step": 7743
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.384997458609708e-06,
+      "loss": 0.0708,
+      "step": 7744
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.381781358691968e-06,
+      "loss": 0.0507,
+      "step": 7745
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.378565549488604e-06,
+      "loss": 0.0607,
+      "step": 7746
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.375350031356677e-06,
+      "loss": 0.0677,
+      "step": 7747
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.372134804653232e-06,
+      "loss": 0.0814,
+      "step": 7748
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.3689198697352626e-06,
+      "loss": 0.0522,
+      "step": 7749
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.365705226959747e-06,
+      "loss": 0.056,
+      "step": 7750
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.362490876683618e-06,
+      "loss": 0.077,
+      "step": 7751
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.359276819263786e-06,
+      "loss": 0.0745,
+      "step": 7752
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.356063055057122e-06,
+      "loss": 0.0709,
+      "step": 7753
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.352849584420472e-06,
+      "loss": 0.0679,
+      "step": 7754
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.349636407710632e-06,
+      "loss": 0.0674,
+      "step": 7755
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.346423525284393e-06,
+      "loss": 0.0616,
+      "step": 7756
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.343210937498486e-06,
+      "loss": 0.0563,
+      "step": 7757
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.339998644709627e-06,
+      "loss": 0.0613,
+      "step": 7758
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.3367866472744964e-06,
+      "loss": 0.0638,
+      "step": 7759
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.333574945549727e-06,
+      "loss": 0.0693,
+      "step": 7760
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.330363539891946e-06,
+      "loss": 0.0746,
+      "step": 7761
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.327152430657721e-06,
+      "loss": 0.0637,
+      "step": 7762
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.3239416182036024e-06,
+      "loss": 0.0719,
+      "step": 7763
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.320731102886102e-06,
+      "loss": 0.0717,
+      "step": 7764
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.317520885061704e-06,
+      "loss": 0.0544,
+      "step": 7765
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.314310965086848e-06,
+      "loss": 0.0534,
+      "step": 7766
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.311101343317956e-06,
+      "loss": 0.0757,
+      "step": 7767
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.3078920201114045e-06,
+      "loss": 0.0685,
+      "step": 7768
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.304682995823542e-06,
+      "loss": 0.0592,
+      "step": 7769
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.301474270810681e-06,
+      "loss": 0.0649,
+      "step": 7770
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.298265845429109e-06,
+      "loss": 0.0624,
+      "step": 7771
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.295057720035066e-06,
+      "loss": 0.0681,
+      "step": 7772
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.291849894984775e-06,
+      "loss": 0.0618,
+      "step": 7773
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.288642370634411e-06,
+      "loss": 0.0732,
+      "step": 7774
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.2854351473401255e-06,
+      "loss": 0.0846,
+      "step": 7775
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.28222822545803e-06,
+      "loss": 0.0539,
+      "step": 7776
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.27902160534421e-06,
+      "loss": 0.0479,
+      "step": 7777
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.275815287354707e-06,
+      "loss": 0.064,
+      "step": 7778
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.2726092718455455e-06,
+      "loss": 0.067,
+      "step": 7779
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.269403559172691e-06,
+      "loss": 0.0656,
+      "step": 7780
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.2661981496921055e-06,
+      "loss": 0.0563,
+      "step": 7781
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.262993043759692e-06,
+      "loss": 0.0704,
+      "step": 7782
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.259788241731336e-06,
+      "loss": 0.0531,
+      "step": 7783
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.2565837439628775e-06,
+      "loss": 0.0585,
+      "step": 7784
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.253379550810136e-06,
+      "loss": 0.0587,
+      "step": 7785
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.250175662628879e-06,
+      "loss": 0.0682,
+      "step": 7786
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.2469720797748635e-06,
+      "loss": 0.0628,
+      "step": 7787
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.24376880260379e-06,
+      "loss": 0.0642,
+      "step": 7788
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.24056583147134e-06,
+      "loss": 0.0611,
+      "step": 7789
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.237363166733153e-06,
+      "loss": 0.0643,
+      "step": 7790
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.23416080874484e-06,
+      "loss": 0.0618,
+      "step": 7791
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.230958757861972e-06,
+      "loss": 0.0524,
+      "step": 7792
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.227757014440098e-06,
+      "loss": 0.0781,
+      "step": 7793
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.224555578834711e-06,
+      "loss": 0.0701,
+      "step": 7794
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.221354451401294e-06,
+      "loss": 0.0638,
+      "step": 7795
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.218153632495277e-06,
+      "loss": 0.0646,
+      "step": 7796
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.21495312247207e-06,
+      "loss": 0.0656,
+      "step": 7797
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.211752921687036e-06,
+      "loss": 0.0656,
+      "step": 7798
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.208553030495518e-06,
+      "loss": 0.0625,
+      "step": 7799
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.205353449252807e-06,
+      "loss": 0.0696,
+      "step": 7800
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.202154178314175e-06,
+      "loss": 0.0617,
+      "step": 7801
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.198955218034851e-06,
+      "loss": 0.0657,
+      "step": 7802
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.195756568770036e-06,
+      "loss": 0.0554,
+      "step": 7803
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.192558230874887e-06,
+      "loss": 0.0767,
+      "step": 7804
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.1893602047045385e-06,
+      "loss": 0.0597,
+      "step": 7805
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.186162490614075e-06,
+      "loss": 0.056,
+      "step": 7806
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.182965088958567e-06,
+      "loss": 0.0638,
+      "step": 7807
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.179768000093027e-06,
+      "loss": 0.0698,
+      "step": 7808
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.176571224372453e-06,
+      "loss": 0.0772,
+      "step": 7809
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.173374762151792e-06,
+      "loss": 0.0739,
+      "step": 7810
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.170178613785972e-06,
+      "loss": 0.0599,
+      "step": 7811
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.166982779629871e-06,
+      "loss": 0.0699,
+      "step": 7812
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.163787260038346e-06,
+      "loss": 0.053,
+      "step": 7813
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.160592055366202e-06,
+      "loss": 0.0735,
+      "step": 7814
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.15739716596823e-06,
+      "loss": 0.0639,
+      "step": 7815
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.154202592199166e-06,
+      "loss": 0.047,
+      "step": 7816
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.151008334413727e-06,
+      "loss": 0.0544,
+      "step": 7817
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.147814392966581e-06,
+      "loss": 0.058,
+      "step": 7818
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.144620768212378e-06,
+      "loss": 0.0687,
+      "step": 7819
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.141427460505712e-06,
+      "loss": 0.068,
+      "step": 7820
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.13823447020116e-06,
+      "loss": 0.0674,
+      "step": 7821
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.135041797653251e-06,
+      "loss": 0.0701,
+      "step": 7822
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.131849443216487e-06,
+      "loss": 0.0734,
+      "step": 7823
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.1286574072453315e-06,
+      "loss": 0.0634,
+      "step": 7824
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.125465690094215e-06,
+      "loss": 0.0651,
+      "step": 7825
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.122274292117526e-06,
+      "loss": 0.0576,
+      "step": 7826
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.119083213669621e-06,
+      "loss": 0.0591,
+      "step": 7827
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.115892455104827e-06,
+      "loss": 0.0708,
+      "step": 7828
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.1127020167774265e-06,
+      "loss": 0.056,
+      "step": 7829
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.109511899041674e-06,
+      "loss": 0.063,
+      "step": 7830
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.1063221022517805e-06,
+      "loss": 0.0726,
+      "step": 7831
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.103132626761929e-06,
+      "loss": 0.0673,
+      "step": 7832
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.099943472926261e-06,
+      "loss": 0.0775,
+      "step": 7833
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.096754641098887e-06,
+      "loss": 0.0667,
+      "step": 7834
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.093566131633878e-06,
+      "loss": 0.0616,
+      "step": 7835
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.090377944885275e-06,
+      "loss": 0.0765,
+      "step": 7836
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.0871900812070695e-06,
+      "loss": 0.062,
+      "step": 7837
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.084002540953236e-06,
+      "loss": 0.0679,
+      "step": 7838
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.080815324477699e-06,
+      "loss": 0.0588,
+      "step": 7839
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.077628432134352e-06,
+      "loss": 0.0689,
+      "step": 7840
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.0744418642770516e-06,
+      "loss": 0.057,
+      "step": 7841
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.071255621259622e-06,
+      "loss": 0.0676,
+      "step": 7842
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.068069703435845e-06,
+      "loss": 0.0662,
+      "step": 7843
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.064884111159474e-06,
+      "loss": 0.0612,
+      "step": 7844
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.061698844784213e-06,
+      "loss": 0.0684,
+      "step": 7845
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.0585139046637485e-06,
+      "loss": 0.0706,
+      "step": 7846
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.0553292911517134e-06,
+      "loss": 0.0601,
+      "step": 7847
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.052145004601717e-06,
+      "loss": 0.0474,
+      "step": 7848
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.048961045367324e-06,
+      "loss": 0.0637,
+      "step": 7849
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.04577741380207e-06,
+      "loss": 0.0562,
+      "step": 7850
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.042594110259444e-06,
+      "loss": 0.0623,
+      "step": 7851
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.0394111350929095e-06,
+      "loss": 0.0697,
+      "step": 7852
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.036228488655885e-06,
+      "loss": 0.0717,
+      "step": 7853
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.033046171301761e-06,
+      "loss": 0.0496,
+      "step": 7854
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.02986418338388e-06,
+      "loss": 0.0679,
+      "step": 7855
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.026682525255563e-06,
+      "loss": 0.074,
+      "step": 7856
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.023501197270077e-06,
+      "loss": 0.0602,
+      "step": 7857
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.02032019978067e-06,
+      "loss": 0.0696,
+      "step": 7858
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.017139533140538e-06,
+      "loss": 0.0535,
+      "step": 7859
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.013959197702851e-06,
+      "loss": 0.0679,
+      "step": 7860
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.0107791938207335e-06,
+      "loss": 0.0597,
+      "step": 7861
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.007599521847286e-06,
+      "loss": 0.0569,
+      "step": 7862
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.0044201821355515e-06,
+      "loss": 0.0585,
+      "step": 7863
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.001241175038562e-06,
+      "loss": 0.0711,
+      "step": 7864
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.998062500909288e-06,
+      "loss": 0.0704,
+      "step": 7865
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.9948841601006824e-06,
+      "loss": 0.0545,
+      "step": 7866
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.991706152965647e-06,
+      "loss": 0.0625,
+      "step": 7867
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.988528479857057e-06,
+      "loss": 0.0627,
+      "step": 7868
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.985351141127741e-06,
+      "loss": 0.0699,
+      "step": 7869
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.982174137130502e-06,
+      "loss": 0.0519,
+      "step": 7870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.978997468218092e-06,
+      "loss": 0.071,
+      "step": 7871
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.975821134743238e-06,
+      "loss": 0.0482,
+      "step": 7872
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.972645137058621e-06,
+      "loss": 0.0702,
+      "step": 7873
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.969469475516892e-06,
+      "loss": 0.0579,
+      "step": 7874
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.966294150470658e-06,
+      "loss": 0.0479,
+      "step": 7875
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.963119162272498e-06,
+      "loss": 0.0615,
+      "step": 7876
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.959944511274939e-06,
+      "loss": 0.0721,
+      "step": 7877
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.956770197830485e-06,
+      "loss": 0.0698,
+      "step": 7878
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.95359622229159e-06,
+      "loss": 0.0573,
+      "step": 7879
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.950422585010686e-06,
+      "loss": 0.0654,
+      "step": 7880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.947249286340149e-06,
+      "loss": 0.0599,
+      "step": 7881
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 6.944076326632336e-06,
+      "loss": 0.0625,
+      "step": 7882
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.940903706239548e-06,
+      "loss": 0.0639,
+      "step": 7883
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.937731425514066e-06,
+      "loss": 0.0581,
+      "step": 7884
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.934559484808117e-06,
+      "loss": 0.0591,
+      "step": 7885
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.931387884473905e-06,
+      "loss": 0.066,
+      "step": 7886
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.928216624863582e-06,
+      "loss": 0.0609,
+      "step": 7887
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.925045706329277e-06,
+      "loss": 0.0721,
+      "step": 7888
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.921875129223067e-06,
+      "loss": 0.054,
+      "step": 7889
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.918704893897004e-06,
+      "loss": 0.0789,
+      "step": 7890
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.915535000703088e-06,
+      "loss": 0.077,
+      "step": 7891
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.912365449993296e-06,
+      "loss": 0.0637,
+      "step": 7892
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.9091962421195566e-06,
+      "loss": 0.0773,
+      "step": 7893
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.906027377433758e-06,
+      "loss": 0.071,
+      "step": 7894
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.902858856287766e-06,
+      "loss": 0.0599,
+      "step": 7895
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.899690679033389e-06,
+      "loss": 0.0704,
+      "step": 7896
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.896522846022413e-06,
+      "loss": 0.0781,
+      "step": 7897
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.893355357606573e-06,
+      "loss": 0.0524,
+      "step": 7898
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.890188214137575e-06,
+      "loss": 0.0787,
+      "step": 7899
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.887021415967081e-06,
+      "loss": 0.0594,
+      "step": 7900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.883854963446723e-06,
+      "loss": 0.0765,
+      "step": 7901
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.880688856928081e-06,
+      "loss": 0.069,
+      "step": 7902
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.877523096762708e-06,
+      "loss": 0.0705,
+      "step": 7903
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.8743576833021135e-06,
+      "loss": 0.0679,
+      "step": 7904
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.871192616897772e-06,
+      "loss": 0.0656,
+      "step": 7905
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.868027897901113e-06,
+      "loss": 0.0628,
+      "step": 7906
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.864863526663539e-06,
+      "loss": 0.0481,
+      "step": 7907
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.861699503536397e-06,
+      "loss": 0.0726,
+      "step": 7908
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.858535828871012e-06,
+      "loss": 0.058,
+      "step": 7909
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.8553725030186605e-06,
+      "loss": 0.0581,
+      "step": 7910
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.852209526330583e-06,
+      "loss": 0.0717,
+      "step": 7911
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.849046899157981e-06,
+      "loss": 0.0545,
+      "step": 7912
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.845884621852021e-06,
+      "loss": 0.0575,
+      "step": 7913
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.842722694763817e-06,
+      "loss": 0.0706,
+      "step": 7914
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.8395611182444684e-06,
+      "loss": 0.0636,
+      "step": 7915
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.83639989264501e-06,
+      "loss": 0.0576,
+      "step": 7916
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.833239018316456e-06,
+      "loss": 0.0558,
+      "step": 7917
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.830078495609768e-06,
+      "loss": 0.0633,
+      "step": 7918
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.826918324875882e-06,
+      "loss": 0.0562,
+      "step": 7919
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.823758506465682e-06,
+      "loss": 0.0706,
+      "step": 7920
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.8205990407300275e-06,
+      "loss": 0.0612,
+      "step": 7921
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.817439928019719e-06,
+      "loss": 0.054,
+      "step": 7922
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.814281168685538e-06,
+      "loss": 0.0527,
+      "step": 7923
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.811122763078213e-06,
+      "loss": 0.0679,
+      "step": 7924
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.807964711548442e-06,
+      "loss": 0.0633,
+      "step": 7925
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.804807014446877e-06,
+      "loss": 0.0542,
+      "step": 7926
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.801649672124137e-06,
+      "loss": 0.0719,
+      "step": 7927
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.7984926849307905e-06,
+      "loss": 0.0623,
+      "step": 7928
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.795336053217383e-06,
+      "loss": 0.0544,
+      "step": 7929
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.792179777334404e-06,
+      "loss": 0.0529,
+      "step": 7930
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.789023857632319e-06,
+      "loss": 0.0648,
+      "step": 7931
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.785868294461538e-06,
+      "loss": 0.0653,
+      "step": 7932
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.782713088172449e-06,
+      "loss": 0.0591,
+      "step": 7933
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.779558239115378e-06,
+      "loss": 0.0578,
+      "step": 7934
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.776403747640638e-06,
+      "loss": 0.0597,
+      "step": 7935
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.773249614098479e-06,
+      "loss": 0.0607,
+      "step": 7936
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.770095838839126e-06,
+      "loss": 0.0518,
+      "step": 7937
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.766942422212755e-06,
+      "loss": 0.0576,
+      "step": 7938
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.7637893645695105e-06,
+      "loss": 0.0627,
+      "step": 7939
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.760636666259485e-06,
+      "loss": 0.0585,
+      "step": 7940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 6.75748432763275e-06,
+      "loss": 0.0658,
+      "step": 7941
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.754332349039316e-06,
+      "loss": 0.0483,
+      "step": 7942
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.7511807308291724e-06,
+      "loss": 0.0558,
+      "step": 7943
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.7480294733522515e-06,
+      "loss": 0.0603,
+      "step": 7944
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.744878576958458e-06,
+      "loss": 0.0814,
+      "step": 7945
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.74172804199765e-06,
+      "loss": 0.0598,
+      "step": 7946
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.738577868819656e-06,
+      "loss": 0.0598,
+      "step": 7947
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.735428057774243e-06,
+      "loss": 0.0637,
+      "step": 7948
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.732278609211159e-06,
+      "loss": 0.0775,
+      "step": 7949
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.7291295234801e-06,
+      "loss": 0.0568,
+      "step": 7950
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.725980800930729e-06,
+      "loss": 0.0654,
+      "step": 7951
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.7228324419126616e-06,
+      "loss": 0.0683,
+      "step": 7952
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.7196844467754815e-06,
+      "loss": 0.0774,
+      "step": 7953
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.716536815868717e-06,
+      "loss": 0.0731,
+      "step": 7954
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.7133895495418775e-06,
+      "loss": 0.0602,
+      "step": 7955
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.710242648144413e-06,
+      "loss": 0.0713,
+      "step": 7956
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.707096112025741e-06,
+      "loss": 0.0541,
+      "step": 7957
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.703949941535239e-06,
+      "loss": 0.0665,
+      "step": 7958
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.700804137022247e-06,
+      "loss": 0.0635,
+      "step": 7959
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.697658698836047e-06,
+      "loss": 0.0598,
+      "step": 7960
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.694513627325909e-06,
+      "loss": 0.0709,
+      "step": 7961
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.691368922841037e-06,
+      "loss": 0.076,
+      "step": 7962
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.688224585730605e-06,
+      "loss": 0.06,
+      "step": 7963
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.6850806163437486e-06,
+      "loss": 0.0486,
+      "step": 7964
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.681937015029552e-06,
+      "loss": 0.0597,
+      "step": 7965
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.678793782137075e-06,
+      "loss": 0.072,
+      "step": 7966
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.675650918015318e-06,
+      "loss": 0.069,
+      "step": 7967
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.672508423013255e-06,
+      "loss": 0.0514,
+      "step": 7968
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.669366297479811e-06,
+      "loss": 0.0598,
+      "step": 7969
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.666224541763876e-06,
+      "loss": 0.0677,
+      "step": 7970
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.663083156214289e-06,
+      "loss": 0.0549,
+      "step": 7971
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.659942141179865e-06,
+      "loss": 0.0801,
+      "step": 7972
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.656801497009353e-06,
+      "loss": 0.0565,
+      "step": 7973
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.653661224051486e-06,
+      "loss": 0.0558,
+      "step": 7974
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.65052132265494e-06,
+      "loss": 0.0527,
+      "step": 7975
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.647381793168357e-06,
+      "loss": 0.0651,
+      "step": 7976
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.6442426359403315e-06,
+      "loss": 0.0627,
+      "step": 7977
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.641103851319426e-06,
+      "loss": 0.0568,
+      "step": 7978
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.63796543965415e-06,
+      "loss": 0.0558,
+      "step": 7979
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.634827401292981e-06,
+      "loss": 0.0624,
+      "step": 7980
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.63168973658435e-06,
+      "loss": 0.0671,
+      "step": 7981
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.62855244587665e-06,
+      "loss": 0.0669,
+      "step": 7982
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.625415529518228e-06,
+      "loss": 0.0812,
+      "step": 7983
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.6222789878574e-06,
+      "loss": 0.0668,
+      "step": 7984
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.619142821242419e-06,
+      "loss": 0.0615,
+      "step": 7985
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.616007030021523e-06,
+      "loss": 0.0742,
+      "step": 7986
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.612871614542884e-06,
+      "loss": 0.068,
+      "step": 7987
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.609736575154653e-06,
+      "loss": 0.0554,
+      "step": 7988
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.6066019122049216e-06,
+      "loss": 0.0638,
+      "step": 7989
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.603467626041755e-06,
+      "loss": 0.0609,
+      "step": 7990
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.600333717013161e-06,
+      "loss": 0.0469,
+      "step": 7991
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.5972001854671245e-06,
+      "loss": 0.0523,
+      "step": 7992
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.594067031751565e-06,
+      "loss": 0.0624,
+      "step": 7993
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.590934256214385e-06,
+      "loss": 0.0538,
+      "step": 7994
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.587801859203422e-06,
+      "loss": 0.0537,
+      "step": 7995
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.584669841066491e-06,
+      "loss": 0.0697,
+      "step": 7996
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.581538202151351e-06,
+      "loss": 0.0669,
+      "step": 7997
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.578406942805729e-06,
+      "loss": 0.051,
+      "step": 7998
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.575276063377299e-06,
+      "loss": 0.0541,
+      "step": 7999
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.572145564213701e-06,
+      "loss": 0.0645,
+      "step": 8000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.56901544566253e-06,
+      "loss": 0.071,
+      "step": 8001
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5658857080713425e-06,
+      "loss": 0.0643,
+      "step": 8002
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.562756351787645e-06,
+      "loss": 0.0571,
+      "step": 8003
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.559627377158911e-06,
+      "loss": 0.0572,
+      "step": 8004
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5564987845325614e-06,
+      "loss": 0.0613,
+      "step": 8005
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5533705742559835e-06,
+      "loss": 0.0514,
+      "step": 8006
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.550242746676519e-06,
+      "loss": 0.0601,
+      "step": 8007
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.547115302141465e-06,
+      "loss": 0.0623,
+      "step": 8008
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.543988240998076e-06,
+      "loss": 0.0574,
+      "step": 8009
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5408615635935735e-06,
+      "loss": 0.0533,
+      "step": 8010
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5377352702751175e-06,
+      "loss": 0.0722,
+      "step": 8011
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.534609361389847e-06,
+      "loss": 0.0694,
+      "step": 8012
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5314838372848425e-06,
+      "loss": 0.0627,
+      "step": 8013
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.528358698307148e-06,
+      "loss": 0.0508,
+      "step": 8014
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.5252339448037615e-06,
+      "loss": 0.0554,
+      "step": 8015
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.522109577121646e-06,
+      "loss": 0.0519,
+      "step": 8016
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.51898559560771e-06,
+      "loss": 0.05,
+      "step": 8017
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.515862000608834e-06,
+      "loss": 0.0576,
+      "step": 8018
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.512738792471837e-06,
+      "loss": 0.063,
+      "step": 8019
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.5096159715435105e-06,
+      "loss": 0.0709,
+      "step": 8020
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.506493538170595e-06,
+      "loss": 0.0516,
+      "step": 8021
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.503371492699794e-06,
+      "loss": 0.0526,
+      "step": 8022
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.5002498354777584e-06,
+      "loss": 0.0508,
+      "step": 8023
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.497128566851112e-06,
+      "loss": 0.061,
+      "step": 8024
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.4940076871664145e-06,
+      "loss": 0.0697,
+      "step": 8025
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.4908871967702e-06,
+      "loss": 0.066,
+      "step": 8026
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.487767096008949e-06,
+      "loss": 0.0579,
+      "step": 8027
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.4846473852291054e-06,
+      "loss": 0.0649,
+      "step": 8028
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.481528064777069e-06,
+      "loss": 0.0424,
+      "step": 8029
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.4784091349991855e-06,
+      "loss": 0.0559,
+      "step": 8030
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.475290596241772e-06,
+      "loss": 0.0599,
+      "step": 8031
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.472172448851095e-06,
+      "loss": 0.0566,
+      "step": 8032
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.46905469317338e-06,
+      "loss": 0.0696,
+      "step": 8033
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.465937329554804e-06,
+      "loss": 0.0599,
+      "step": 8034
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.462820358341511e-06,
+      "loss": 0.0625,
+      "step": 8035
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.459703779879585e-06,
+      "loss": 0.0665,
+      "step": 8036
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.456587594515081e-06,
+      "loss": 0.0634,
+      "step": 8037
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.453471802594005e-06,
+      "loss": 0.0743,
+      "step": 8038
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.450356404462319e-06,
+      "loss": 0.0886,
+      "step": 8039
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.447241400465942e-06,
+      "loss": 0.0629,
+      "step": 8040
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.444126790950751e-06,
+      "loss": 0.0627,
+      "step": 8041
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.441012576262571e-06,
+      "loss": 0.0598,
+      "step": 8042
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.437898756747198e-06,
+      "loss": 0.0508,
+      "step": 8043
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.434785332750368e-06,
+      "loss": 0.0658,
+      "step": 8044
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.431672304617786e-06,
+      "loss": 0.067,
+      "step": 8045
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.428559672695102e-06,
+      "loss": 0.066,
+      "step": 8046
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.425447437327936e-06,
+      "loss": 0.0614,
+      "step": 8047
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.422335598861846e-06,
+      "loss": 0.0653,
+      "step": 8048
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.419224157642365e-06,
+      "loss": 0.0587,
+      "step": 8049
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.416113114014964e-06,
+      "loss": 0.0684,
+      "step": 8050
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.413002468325082e-06,
+      "loss": 0.0584,
+      "step": 8051
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.409892220918108e-06,
+      "loss": 0.069,
+      "step": 8052
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.406782372139395e-06,
+      "loss": 0.0576,
+      "step": 8053
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.4036729223342375e-06,
+      "loss": 0.0583,
+      "step": 8054
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.400563871847902e-06,
+      "loss": 0.0643,
+      "step": 8055
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.397455221025595e-06,
+      "loss": 0.0623,
+      "step": 8056
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.394346970212491e-06,
+      "loss": 0.0472,
+      "step": 8057
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.39123911975371e-06,
+      "loss": 0.0667,
+      "step": 8058
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.38813166999434e-06,
+      "loss": 0.0679,
+      "step": 8059
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.385024621279411e-06,
+      "loss": 0.0663,
+      "step": 8060
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.3819179739539214e-06,
+      "loss": 0.0744,
+      "step": 8061
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.378811728362808e-06,
+      "loss": 0.0708,
+      "step": 8062
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.375705884850985e-06,
+      "loss": 0.0603,
+      "step": 8063
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.3726004437633e-06,
+      "loss": 0.0595,
+      "step": 8064
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.3694954054445745e-06,
+      "loss": 0.0775,
+      "step": 8065
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.36639077023957e-06,
+      "loss": 0.0616,
+      "step": 8066
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.363286538493019e-06,
+      "loss": 0.063,
+      "step": 8067
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.360182710549587e-06,
+      "loss": 0.0642,
+      "step": 8068
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.357079286753925e-06,
+      "loss": 0.0518,
+      "step": 8069
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.353976267450608e-06,
+      "loss": 0.0637,
+      "step": 8070
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.350873652984186e-06,
+      "loss": 0.0634,
+      "step": 8071
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.347771443699157e-06,
+      "loss": 0.0527,
+      "step": 8072
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.34466963993998e-06,
+      "loss": 0.0601,
+      "step": 8073
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.341568242051057e-06,
+      "loss": 0.0756,
+      "step": 8074
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.338467250376761e-06,
+      "loss": 0.0675,
+      "step": 8075
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.335366665261401e-06,
+      "loss": 0.061,
+      "step": 8076
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.332266487049259e-06,
+      "loss": 0.0618,
+      "step": 8077
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.329166716084558e-06,
+      "loss": 0.0633,
+      "step": 8078
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.326067352711487e-06,
+      "loss": 0.0756,
+      "step": 8079
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.322968397274181e-06,
+      "loss": 0.0635,
+      "step": 8080
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.319869850116738e-06,
+      "loss": 0.0667,
+      "step": 8081
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.316771711583196e-06,
+      "loss": 0.0532,
+      "step": 8082
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.313673982017567e-06,
+      "loss": 0.0609,
+      "step": 8083
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.310576661763801e-06,
+      "loss": 0.0694,
+      "step": 8084
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.307479751165814e-06,
+      "loss": 0.065,
+      "step": 8085
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.30438325056747e-06,
+      "loss": 0.0719,
+      "step": 8086
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.301287160312593e-06,
+      "loss": 0.0624,
+      "step": 8087
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.298191480744951e-06,
+      "loss": 0.0773,
+      "step": 8088
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.295096212208281e-06,
+      "loss": 0.0661,
+      "step": 8089
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.29200135504626e-06,
+      "loss": 0.0698,
+      "step": 8090
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.288906909602531e-06,
+      "loss": 0.0609,
+      "step": 8091
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.285812876220682e-06,
+      "loss": 0.067,
+      "step": 8092
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.282719255244265e-06,
+      "loss": 0.0656,
+      "step": 8093
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.279626047016776e-06,
+      "loss": 0.0514,
+      "step": 8094
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.276533251881674e-06,
+      "loss": 0.0712,
+      "step": 8095
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.273440870182365e-06,
+      "loss": 0.0627,
+      "step": 8096
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.270348902262209e-06,
+      "loss": 0.0583,
+      "step": 8097
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.267257348464532e-06,
+      "loss": 0.0687,
+      "step": 8098
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.264166209132596e-06,
+      "loss": 0.0642,
+      "step": 8099
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.261075484609635e-06,
+      "loss": 0.063,
+      "step": 8100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.2579851752388186e-06,
+      "loss": 0.0682,
+      "step": 8101
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.254895281363289e-06,
+      "loss": 0.0664,
+      "step": 8102
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.2518058033261255e-06,
+      "loss": 0.0647,
+      "step": 8103
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.248716741470376e-06,
+      "loss": 0.0538,
+      "step": 8104
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.245628096139027e-06,
+      "loss": 0.0587,
+      "step": 8105
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.242539867675036e-06,
+      "loss": 0.0526,
+      "step": 8106
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.2394520564212954e-06,
+      "loss": 0.0551,
+      "step": 8107
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.236364662720668e-06,
+      "loss": 0.0618,
+      "step": 8108
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.233277686915961e-06,
+      "loss": 0.053,
+      "step": 8109
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.230191129349938e-06,
+      "loss": 0.0747,
+      "step": 8110
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.2271049903653115e-06,
+      "loss": 0.0607,
+      "step": 8111
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.224019270304757e-06,
+      "loss": 0.0663,
+      "step": 8112
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.220933969510894e-06,
+      "loss": 0.0627,
+      "step": 8113
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.217849088326302e-06,
+      "loss": 0.0521,
+      "step": 8114
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.214764627093509e-06,
+      "loss": 0.0743,
+      "step": 8115
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.211680586155003e-06,
+      "loss": 0.0635,
+      "step": 8116
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.208596965853213e-06,
+      "loss": 0.0534,
+      "step": 8117
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.20551376653054e-06,
+      "loss": 0.0795,
+      "step": 8118
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.202430988529315e-06,
+      "loss": 0.0581,
+      "step": 8119
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.199348632191848e-06,
+      "loss": 0.0716,
+      "step": 8120
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.196266697860379e-06,
+      "loss": 0.0604,
+      "step": 8121
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.193185185877117e-06,
+      "loss": 0.0731,
+      "step": 8122
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.190104096584213e-06,
+      "loss": 0.0642,
+      "step": 8123
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.187023430323782e-06,
+      "loss": 0.078,
+      "step": 8124
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.1839431874378816e-06,
+      "loss": 0.0603,
+      "step": 8125
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.1808633682685345e-06,
+      "loss": 0.0667,
+      "step": 8126
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.177783973157699e-06,
+      "loss": 0.0439,
+      "step": 8127
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.174705002447301e-06,
+      "loss": 0.0648,
+      "step": 8128
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.171626456479214e-06,
+      "loss": 0.0703,
+      "step": 8129
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.168548335595268e-06,
+      "loss": 0.079,
+      "step": 8130
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.165470640137237e-06,
+      "loss": 0.0657,
+      "step": 8131
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.162393370446861e-06,
+      "loss": 0.0682,
+      "step": 8132
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.159316526865816e-06,
+      "loss": 0.0687,
+      "step": 8133
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.156240109735747e-06,
+      "loss": 0.0651,
+      "step": 8134
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.153164119398241e-06,
+      "loss": 0.061,
+      "step": 8135
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.150088556194843e-06,
+      "loss": 0.082,
+      "step": 8136
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.147013420467047e-06,
+      "loss": 0.054,
+      "step": 8137
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.143938712556305e-06,
+      "loss": 0.058,
+      "step": 8138
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.14086443280401e-06,
+      "loss": 0.0606,
+      "step": 8139
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.1377905815515255e-06,
+      "loss": 0.0605,
+      "step": 8140
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.134717159140149e-06,
+      "loss": 0.0601,
+      "step": 8141
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.131644165911144e-06,
+      "loss": 0.0522,
+      "step": 8142
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.128571602205713e-06,
+      "loss": 0.0525,
+      "step": 8143
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.125499468365028e-06,
+      "loss": 0.0572,
+      "step": 8144
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.122427764730198e-06,
+      "loss": 0.0592,
+      "step": 8145
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.119356491642294e-06,
+      "loss": 0.0596,
+      "step": 8146
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.116285649442331e-06,
+      "loss": 0.0612,
+      "step": 8147
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.113215238471284e-06,
+      "loss": 0.061,
+      "step": 8148
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.1101452590700735e-06,
+      "loss": 0.0673,
+      "step": 8149
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.10707571157958e-06,
+      "loss": 0.0674,
+      "step": 8150
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.104006596340625e-06,
+      "loss": 0.0785,
+      "step": 8151
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.100937913693996e-06,
+      "loss": 0.0695,
+      "step": 8152
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.0978696639804155e-06,
+      "loss": 0.0568,
+      "step": 8153
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.094801847540576e-06,
+      "loss": 0.0663,
+      "step": 8154
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.091734464715105e-06,
+      "loss": 0.0585,
+      "step": 8155
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.0886675158445976e-06,
+      "loss": 0.0629,
+      "step": 8156
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.085601001269586e-06,
+      "loss": 0.0596,
+      "step": 8157
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.08253492133057e-06,
+      "loss": 0.0689,
+      "step": 8158
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.07946927636798e-06,
+      "loss": 0.0523,
+      "step": 8159
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.076404066722224e-06,
+      "loss": 0.0751,
+      "step": 8160
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.073339292733635e-06,
+      "loss": 0.0772,
+      "step": 8161
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.070274954742521e-06,
+      "loss": 0.0558,
+      "step": 8162
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.06721105308913e-06,
+      "loss": 0.0709,
+      "step": 8163
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.064147588113656e-06,
+      "loss": 0.0566,
+      "step": 8164
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.061084560156257e-06,
+      "loss": 0.0502,
+      "step": 8165
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.058021969557033e-06,
+      "loss": 0.0529,
+      "step": 8166
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.054959816656045e-06,
+      "loss": 0.0626,
+      "step": 8167
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.051898101793294e-06,
+      "loss": 0.0591,
+      "step": 8168
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.048836825308745e-06,
+      "loss": 0.0674,
+      "step": 8169
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.045775987542298e-06,
+      "loss": 0.0644,
+      "step": 8170
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.042715588833822e-06,
+      "loss": 0.0778,
+      "step": 8171
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.039655629523122e-06,
+      "loss": 0.0571,
+      "step": 8172
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.036596109949968e-06,
+      "loss": 0.049,
+      "step": 8173
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.033537030454069e-06,
+      "loss": 0.0578,
+      "step": 8174
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.030478391375094e-06,
+      "loss": 0.0718,
+      "step": 8175
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.027420193052656e-06,
+      "loss": 0.0478,
+      "step": 8176
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.024362435826328e-06,
+      "loss": 0.0615,
+      "step": 8177
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.021305120035621e-06,
+      "loss": 0.056,
+      "step": 8178
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.018248246020012e-06,
+      "loss": 0.0699,
+      "step": 8179
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.0151918141189156e-06,
+      "loss": 0.0654,
+      "step": 8180
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.012135824671707e-06,
+      "loss": 0.0694,
+      "step": 8181
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.009080278017707e-06,
+      "loss": 0.0585,
+      "step": 8182
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.0060251744961926e-06,
+      "loss": 0.0593,
+      "step": 8183
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.002970514446382e-06,
+      "loss": 0.0706,
+      "step": 8184
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.999916298207452e-06,
+      "loss": 0.049,
+      "step": 8185
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.996862526118528e-06,
+      "loss": 0.0596,
+      "step": 8186
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.993809198518687e-06,
+      "loss": 0.0603,
+      "step": 8187
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.990756315746955e-06,
+      "loss": 0.062,
+      "step": 8188
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.987703878142313e-06,
+      "loss": 0.0648,
+      "step": 8189
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.98465188604368e-06,
+      "loss": 0.059,
+      "step": 8190
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.981600339789945e-06,
+      "loss": 0.0533,
+      "step": 8191
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.978549239719929e-06,
+      "loss": 0.0682,
+      "step": 8192
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.975498586172416e-06,
+      "loss": 0.0526,
+      "step": 8193
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.972448379486133e-06,
+      "loss": 0.078,
+      "step": 8194
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.969398619999764e-06,
+      "loss": 0.0572,
+      "step": 8195
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.966349308051933e-06,
+      "loss": 0.0668,
+      "step": 8196
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.963300443981231e-06,
+      "loss": 0.0553,
+      "step": 8197
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.960252028126179e-06,
+      "loss": 0.0563,
+      "step": 8198
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.957204060825265e-06,
+      "loss": 0.0709,
+      "step": 8199
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.954156542416915e-06,
+      "loss": 0.0445,
+      "step": 8200
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.951109473239518e-06,
+      "loss": 0.0541,
+      "step": 8201
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.948062853631399e-06,
+      "loss": 0.0526,
+      "step": 8202
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.945016683930846e-06,
+      "loss": 0.0592,
+      "step": 8203
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.941970964476085e-06,
+      "loss": 0.0583,
+      "step": 8204
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.938925695605303e-06,
+      "loss": 0.0712,
+      "step": 8205
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.935880877656628e-06,
+      "loss": 0.0698,
+      "step": 8206
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.932836510968145e-06,
+      "loss": 0.0732,
+      "step": 8207
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.929792595877881e-06,
+      "loss": 0.0644,
+      "step": 8208
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.926749132723828e-06,
+      "loss": 0.0648,
+      "step": 8209
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.923706121843905e-06,
+      "loss": 0.0562,
+      "step": 8210
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.920663563576001e-06,
+      "loss": 0.0655,
+      "step": 8211
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.917621458257944e-06,
+      "loss": 0.0695,
+      "step": 8212
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.914579806227516e-06,
+      "loss": 0.0499,
+      "step": 8213
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.911538607822447e-06,
+      "loss": 0.0584,
+      "step": 8214
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.90849786338042e-06,
+      "loss": 0.0509,
+      "step": 8215
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.905457573239055e-06,
+      "loss": 0.0693,
+      "step": 8216
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.902417737735944e-06,
+      "loss": 0.07,
+      "step": 8217
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.899378357208606e-06,
+      "loss": 0.0655,
+      "step": 8218
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8963394319945244e-06,
+      "loss": 0.0684,
+      "step": 8219
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8933009624311235e-06,
+      "loss": 0.0674,
+      "step": 8220
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.890262948855783e-06,
+      "loss": 0.068,
+      "step": 8221
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8872253916058255e-06,
+      "loss": 0.0688,
+      "step": 8222
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.884188291018533e-06,
+      "loss": 0.0703,
+      "step": 8223
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.881151647431125e-06,
+      "loss": 0.0559,
+      "step": 8224
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.878115461180778e-06,
+      "loss": 0.069,
+      "step": 8225
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8750797326046134e-06,
+      "loss": 0.0605,
+      "step": 8226
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.872044462039708e-06,
+      "loss": 0.0626,
+      "step": 8227
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.869009649823079e-06,
+      "loss": 0.0665,
+      "step": 8228
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.865975296291702e-06,
+      "loss": 0.0643,
+      "step": 8229
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.862941401782494e-06,
+      "loss": 0.066,
+      "step": 8230
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.859907966632321e-06,
+      "loss": 0.0543,
+      "step": 8231
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.856874991178008e-06,
+      "loss": 0.0585,
+      "step": 8232
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.853842475756315e-06,
+      "loss": 0.0859,
+      "step": 8233
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8508104207039674e-06,
+      "loss": 0.062,
+      "step": 8234
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8477788263576175e-06,
+      "loss": 0.0631,
+      "step": 8235
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.844747693053889e-06,
+      "loss": 0.0485,
+      "step": 8236
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.841717021129337e-06,
+      "loss": 0.0707,
+      "step": 8237
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.83868681092048e-06,
+      "loss": 0.0546,
+      "step": 8238
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.835657062763773e-06,
+      "loss": 0.0572,
+      "step": 8239
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.832627776995625e-06,
+      "loss": 0.0694,
+      "step": 8240
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.829598953952394e-06,
+      "loss": 0.0566,
+      "step": 8241
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.826570593970389e-06,
+      "loss": 0.0709,
+      "step": 8242
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.823542697385858e-06,
+      "loss": 0.0626,
+      "step": 8243
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.820515264535011e-06,
+      "loss": 0.0769,
+      "step": 8244
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8174882957539925e-06,
+      "loss": 0.0694,
+      "step": 8245
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.814461791378907e-06,
+      "loss": 0.0597,
+      "step": 8246
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.811435751745802e-06,
+      "loss": 0.0511,
+      "step": 8247
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.808410177190677e-06,
+      "loss": 0.0519,
+      "step": 8248
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.805385068049472e-06,
+      "loss": 0.0722,
+      "step": 8249
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.802360424658086e-06,
+      "loss": 0.0699,
+      "step": 8250
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.799336247352356e-06,
+      "loss": 0.0542,
+      "step": 8251
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.796312536468074e-06,
+      "loss": 0.0655,
+      "step": 8252
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.793289292340976e-06,
+      "loss": 0.0505,
+      "step": 8253
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.790266515306757e-06,
+      "loss": 0.0664,
+      "step": 8254
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7872442057010405e-06,
+      "loss": 0.0638,
+      "step": 8255
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.784222363859418e-06,
+      "loss": 0.064,
+      "step": 8256
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.781200990117408e-06,
+      "loss": 0.0588,
+      "step": 8257
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7781800848105065e-06,
+      "loss": 0.0598,
+      "step": 8258
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7751596482741265e-06,
+      "loss": 0.0709,
+      "step": 8259
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.772139680843652e-06,
+      "loss": 0.0689,
+      "step": 8260
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.769120182854397e-06,
+      "loss": 0.0621,
+      "step": 8261
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.766101154641637e-06,
+      "loss": 0.056,
+      "step": 8262
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.763082596540588e-06,
+      "loss": 0.0616,
+      "step": 8263
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.760064508886423e-06,
+      "loss": 0.0521,
+      "step": 8264
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.757046892014247e-06,
+      "loss": 0.0752,
+      "step": 8265
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7540297462591264e-06,
+      "loss": 0.0743,
+      "step": 8266
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.751013071956068e-06,
+      "loss": 0.049,
+      "step": 8267
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.747996869440029e-06,
+      "loss": 0.0634,
+      "step": 8268
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.7449811390459155e-06,
+      "loss": 0.0594,
+      "step": 8269
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.741965881108583e-06,
+      "loss": 0.0651,
+      "step": 8270
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.738951095962824e-06,
+      "loss": 0.0834,
+      "step": 8271
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.735936783943392e-06,
+      "loss": 0.0616,
+      "step": 8272
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.732922945384975e-06,
+      "loss": 0.0574,
+      "step": 8273
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.729909580622219e-06,
+      "loss": 0.0729,
+      "step": 8274
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.7268966899897134e-06,
+      "loss": 0.0679,
+      "step": 8275
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.723884273822e-06,
+      "loss": 0.0644,
+      "step": 8276
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.720872332453553e-06,
+      "loss": 0.0547,
+      "step": 8277
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.717860866218814e-06,
+      "loss": 0.0604,
+      "step": 8278
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.714849875452151e-06,
+      "loss": 0.0578,
+      "step": 8279
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.711839360487897e-06,
+      "loss": 0.0651,
+      "step": 8280
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.708829321660321e-06,
+      "loss": 0.062,
+      "step": 8281
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.705819759303653e-06,
+      "loss": 0.0778,
+      "step": 8282
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.702810673752046e-06,
+      "loss": 0.0587,
+      "step": 8283
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.699802065339626e-06,
+      "loss": 0.0618,
+      "step": 8284
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.696793934400446e-06,
+      "loss": 0.0586,
+      "step": 8285
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.693786281268519e-06,
+      "loss": 0.0726,
+      "step": 8286
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.6907791062778e-06,
+      "loss": 0.0717,
+      "step": 8287
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.687772409762193e-06,
+      "loss": 0.0524,
+      "step": 8288
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.6847661920555395e-06,
+      "loss": 0.0626,
+      "step": 8289
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.681760453491647e-06,
+      "loss": 0.0676,
+      "step": 8290
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.678755194404246e-06,
+      "loss": 0.0632,
+      "step": 8291
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.675750415127032e-06,
+      "loss": 0.0599,
+      "step": 8292
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.672746115993643e-06,
+      "loss": 0.0665,
+      "step": 8293
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.669742297337662e-06,
+      "loss": 0.0696,
+      "step": 8294
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.666738959492612e-06,
+      "loss": 0.0587,
+      "step": 8295
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.663736102791979e-06,
+      "loss": 0.0688,
+      "step": 8296
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.660733727569176e-06,
+      "loss": 0.0685,
+      "step": 8297
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.657731834157576e-06,
+      "loss": 0.0652,
+      "step": 8298
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.6547304228905e-06,
+      "loss": 0.0692,
+      "step": 8299
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.651729494101201e-06,
+      "loss": 0.0725,
+      "step": 8300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.648729048122896e-06,
+      "loss": 0.0626,
+      "step": 8301
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.6457290852887336e-06,
+      "loss": 0.0696,
+      "step": 8302
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.642729605931816e-06,
+      "loss": 0.0489,
+      "step": 8303
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.6397306103851944e-06,
+      "loss": 0.0527,
+      "step": 8304
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.6367320989818644e-06,
+      "loss": 0.0701,
+      "step": 8305
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.63373407205476e-06,
+      "loss": 0.063,
+      "step": 8306
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.6307365299367735e-06,
+      "loss": 0.0717,
+      "step": 8307
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.627739472960727e-06,
+      "loss": 0.0583,
+      "step": 8308
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.624742901459415e-06,
+      "loss": 0.0535,
+      "step": 8309
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.62174681576555e-06,
+      "loss": 0.0613,
+      "step": 8310
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.618751216211812e-06,
+      "loss": 0.0618,
+      "step": 8311
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.615756103130809e-06,
+      "loss": 0.0583,
+      "step": 8312
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.612761476855112e-06,
+      "loss": 0.0562,
+      "step": 8313
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.609767337717218e-06,
+      "loss": 0.0729,
+      "step": 8314
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.606773686049598e-06,
+      "loss": 0.0705,
+      "step": 8315
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.6037805221846395e-06,
+      "loss": 0.0505,
+      "step": 8316
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.600787846454698e-06,
+      "loss": 0.0818,
+      "step": 8317
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.5977956591920576e-06,
+      "loss": 0.0605,
+      "step": 8318
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.594803960728962e-06,
+      "loss": 0.0641,
+      "step": 8319
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.591812751397592e-06,
+      "loss": 0.05,
+      "step": 8320
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.588822031530083e-06,
+      "loss": 0.076,
+      "step": 8321
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.585831801458501e-06,
+      "loss": 0.0712,
+      "step": 8322
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.582842061514875e-06,
+      "loss": 0.0598,
+      "step": 8323
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.579852812031165e-06,
+      "loss": 0.0714,
+      "step": 8324
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.576864053339284e-06,
+      "loss": 0.062,
+      "step": 8325
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.573875785771091e-06,
+      "loss": 0.0757,
+      "step": 8326
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.570888009658393e-06,
+      "loss": 0.0505,
+      "step": 8327
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.56790072533293e-06,
+      "loss": 0.0853,
+      "step": 8328
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.564913933126404e-06,
+      "loss": 0.0676,
+      "step": 8329
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.561927633370443e-06,
+      "loss": 0.0618,
+      "step": 8330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.55894182639664e-06,
+      "loss": 0.0673,
+      "step": 8331
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.555956512536522e-06,
+      "loss": 0.0615,
+      "step": 8332
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.552971692121568e-06,
+      "loss": 0.0739,
+      "step": 8333
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.54998736548319e-06,
+      "loss": 0.0572,
+      "step": 8334
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.547003532952763e-06,
+      "loss": 0.0595,
+      "step": 8335
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.544020194861587e-06,
+      "loss": 0.068,
+      "step": 8336
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.541037351540923e-06,
+      "loss": 0.0535,
+      "step": 8337
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.538055003321969e-06,
+      "loss": 0.0641,
+      "step": 8338
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.535073150535879e-06,
+      "loss": 0.0681,
+      "step": 8339
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.5320917935137315e-06,
+      "loss": 0.0752,
+      "step": 8340
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.529110932586573e-06,
+      "loss": 0.0487,
+      "step": 8341
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.5261305680853736e-06,
+      "loss": 0.0663,
+      "step": 8342
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.523150700341065e-06,
+      "loss": 0.0556,
+      "step": 8343
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.520171329684514e-06,
+      "loss": 0.0623,
+      "step": 8344
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.517192456446543e-06,
+      "loss": 0.0582,
+      "step": 8345
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.5142140809579e-06,
+      "loss": 0.049,
+      "step": 8346
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.511236203549302e-06,
+      "loss": 0.0569,
+      "step": 8347
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.508258824551387e-06,
+      "loss": 0.0595,
+      "step": 8348
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.505281944294752e-06,
+      "loss": 0.063,
+      "step": 8349
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.502305563109939e-06,
+      "loss": 0.0701,
+      "step": 8350
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.499329681327432e-06,
+      "loss": 0.0692,
+      "step": 8351
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.496354299277652e-06,
+      "loss": 0.0636,
+      "step": 8352
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.493379417290978e-06,
+      "loss": 0.0527,
+      "step": 8353
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.490405035697717e-06,
+      "loss": 0.0579,
+      "step": 8354
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.487431154828142e-06,
+      "loss": 0.061,
+      "step": 8355
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.484457775012451e-06,
+      "loss": 0.0546,
+      "step": 8356
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.481484896580798e-06,
+      "loss": 0.0554,
+      "step": 8357
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.478512519863269e-06,
+      "loss": 0.0592,
+      "step": 8358
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.475540645189914e-06,
+      "loss": 0.0743,
+      "step": 8359
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.4725692728906994e-06,
+      "loss": 0.068,
+      "step": 8360
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.469598403295572e-06,
+      "loss": 0.0618,
+      "step": 8361
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.4666280367343885e-06,
+      "loss": 0.0622,
+      "step": 8362
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.463658173536972e-06,
+      "loss": 0.0543,
+      "step": 8363
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.460688814033075e-06,
+      "loss": 0.0727,
+      "step": 8364
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.457719958552404e-06,
+      "loss": 0.0642,
+      "step": 8365
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.4547516074246085e-06,
+      "loss": 0.0537,
+      "step": 8366
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.451783760979276e-06,
+      "loss": 0.0785,
+      "step": 8367
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.448816419545948e-06,
+      "loss": 0.0635,
+      "step": 8368
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.445849583454095e-06,
+      "loss": 0.0646,
+      "step": 8369
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.4428832530331445e-06,
+      "loss": 0.0609,
+      "step": 8370
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.4399174286124644e-06,
+      "loss": 0.0604,
+      "step": 8371
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.43695211052137e-06,
+      "loss": 0.0685,
+      "step": 8372
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.433987299089105e-06,
+      "loss": 0.0873,
+      "step": 8373
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.431022994644879e-06,
+      "loss": 0.0638,
+      "step": 8374
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.428059197517823e-06,
+      "loss": 0.0746,
+      "step": 8375
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.4250959080370305e-06,
+      "loss": 0.0643,
+      "step": 8376
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.422133126531527e-06,
+      "loss": 0.0588,
+      "step": 8377
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.419170853330291e-06,
+      "loss": 0.0507,
+      "step": 8378
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.416209088762233e-06,
+      "loss": 0.0529,
+      "step": 8379
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.413247833156219e-06,
+      "loss": 0.0596,
+      "step": 8380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.410287086841044e-06,
+      "loss": 0.0615,
+      "step": 8381
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.4073268501454615e-06,
+      "loss": 0.0453,
+      "step": 8382
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.404367123398159e-06,
+      "loss": 0.0732,
+      "step": 8383
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.401407906927776e-06,
+      "loss": 0.0555,
+      "step": 8384
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.398449201062881e-06,
+      "loss": 0.0699,
+      "step": 8385
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.395491006132003e-06,
+      "loss": 0.0641,
+      "step": 8386
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.392533322463598e-06,
+      "loss": 0.0538,
+      "step": 8387
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.389576150386075e-06,
+      "loss": 0.0703,
+      "step": 8388
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.386619490227787e-06,
+      "loss": 0.0532,
+      "step": 8389
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.383663342317029e-06,
+      "loss": 0.0665,
+      "step": 8390
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.3807077069820295e-06,
+      "loss": 0.0585,
+      "step": 8391
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.377752584550977e-06,
+      "loss": 0.0673,
+      "step": 8392
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.374797975351986e-06,
+      "loss": 0.0588,
+      "step": 8393
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.371843879713126e-06,
+      "loss": 0.0517,
+      "step": 8394
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.368890297962403e-06,
+      "loss": 0.0749,
+      "step": 8395
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.365937230427778e-06,
+      "loss": 0.0486,
+      "step": 8396
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.362984677437133e-06,
+      "loss": 0.0477,
+      "step": 8397
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.360032639318313e-06,
+      "loss": 0.0753,
+      "step": 8398
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.357081116399094e-06,
+      "loss": 0.053,
+      "step": 8399
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.354130109007198e-06,
+      "loss": 0.0609,
+      "step": 8400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.351179617470294e-06,
+      "loss": 0.0675,
+      "step": 8401
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.348229642115993e-06,
+      "loss": 0.0644,
+      "step": 8402
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.345280183271838e-06,
+      "loss": 0.0712,
+      "step": 8403
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.342331241265332e-06,
+      "loss": 0.0734,
+      "step": 8404
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.339382816423897e-06,
+      "loss": 0.0709,
+      "step": 8405
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.33643490907493e-06,
+      "loss": 0.0597,
+      "step": 8406
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.33348751954574e-06,
+      "loss": 0.0646,
+      "step": 8407
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.330540648163598e-06,
+      "loss": 0.0647,
+      "step": 8408
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.327594295255705e-06,
+      "loss": 0.0689,
+      "step": 8409
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.3246484611492135e-06,
+      "loss": 0.0602,
+      "step": 8410
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.321703146171204e-06,
+      "loss": 0.0578,
+      "step": 8411
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.3187583506487295e-06,
+      "loss": 0.0616,
+      "step": 8412
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.31581407490875e-06,
+      "loss": 0.0625,
+      "step": 8413
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.3128703192781936e-06,
+      "loss": 0.0533,
+      "step": 8414
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.309927084083912e-06,
+      "loss": 0.0656,
+      "step": 8415
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.306984369652712e-06,
+      "loss": 0.0571,
+      "step": 8416
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.304042176311339e-06,
+      "loss": 0.0742,
+      "step": 8417
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.301100504386485e-06,
+      "loss": 0.0607,
+      "step": 8418
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.298159354204767e-06,
+      "loss": 0.0505,
+      "step": 8419
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.295218726092768e-06,
+      "loss": 0.072,
+      "step": 8420
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.292278620376992e-06,
+      "loss": 0.0754,
+      "step": 8421
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.289339037383896e-06,
+      "loss": 0.062,
+      "step": 8422
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.286399977439882e-06,
+      "loss": 0.0523,
+      "step": 8423
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.283461440871288e-06,
+      "loss": 0.0695,
+      "step": 8424
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.28052342800439e-06,
+      "loss": 0.0594,
+      "step": 8425
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.277585939165417e-06,
+      "loss": 0.0555,
+      "step": 8426
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.2746489746805286e-06,
+      "loss": 0.0632,
+      "step": 8427
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.271712534875831e-06,
+      "loss": 0.0576,
+      "step": 8428
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.268776620077376e-06,
+      "loss": 0.0607,
+      "step": 8429
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.265841230611155e-06,
+      "loss": 0.0604,
+      "step": 8430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.262906366803092e-06,
+      "loss": 0.0585,
+      "step": 8431
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.259972028979068e-06,
+      "loss": 0.0726,
+      "step": 8432
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.257038217464891e-06,
+      "loss": 0.0625,
+      "step": 8433
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.254104932586321e-06,
+      "loss": 0.0592,
+      "step": 8434
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.251172174669058e-06,
+      "loss": 0.0529,
+      "step": 8435
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.248239944038736e-06,
+      "loss": 0.0671,
+      "step": 8436
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.245308241020941e-06,
+      "loss": 0.0554,
+      "step": 8437
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.242377065941188e-06,
+      "loss": 0.0713,
+      "step": 8438
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.239446419124948e-06,
+      "loss": 0.0752,
+      "step": 8439
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.2365163008976195e-06,
+      "loss": 0.0717,
+      "step": 8440
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.233586711584558e-06,
+      "loss": 0.0872,
+      "step": 8441
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.23065765151104e-06,
+      "loss": 0.0583,
+      "step": 8442
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.227729121002303e-06,
+      "loss": 0.0507,
+      "step": 8443
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.224801120383509e-06,
+      "loss": 0.0674,
+      "step": 8444
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.221873649979775e-06,
+      "loss": 0.0588,
+      "step": 8445
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.218946710116151e-06,
+      "loss": 0.0521,
+      "step": 8446
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.216020301117634e-06,
+      "loss": 0.066,
+      "step": 8447
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.213094423309151e-06,
+      "loss": 0.062,
+      "step": 8448
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.2101690770155855e-06,
+      "loss": 0.0677,
+      "step": 8449
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.207244262561746e-06,
+      "loss": 0.0569,
+      "step": 8450
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.204319980272394e-06,
+      "loss": 0.0716,
+      "step": 8451
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.201396230472227e-06,
+      "loss": 0.0612,
+      "step": 8452
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.198473013485888e-06,
+      "loss": 0.0523,
+      "step": 8453
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.195550329637951e-06,
+      "loss": 0.06,
+      "step": 8454
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.192628179252941e-06,
+      "loss": 0.0644,
+      "step": 8455
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.189706562655314e-06,
+      "loss": 0.0715,
+      "step": 8456
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.186785480169476e-06,
+      "loss": 0.0583,
+      "step": 8457
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.183864932119769e-06,
+      "loss": 0.0673,
+      "step": 8458
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.180944918830481e-06,
+      "loss": 0.0538,
+      "step": 8459
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.1780254406258305e-06,
+      "loss": 0.0572,
+      "step": 8460
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.175106497829987e-06,
+      "loss": 0.069,
+      "step": 8461
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.172188090767044e-06,
+      "loss": 0.0655,
+      "step": 8462
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.169270219761065e-06,
+      "loss": 0.067,
+      "step": 8463
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.1663528851360245e-06,
+      "loss": 0.051,
+      "step": 8464
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.163436087215856e-06,
+      "loss": 0.0607,
+      "step": 8465
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.160519826324419e-06,
+      "loss": 0.0546,
+      "step": 8466
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.157604102785525e-06,
+      "loss": 0.0893,
+      "step": 8467
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.154688916922922e-06,
+      "loss": 0.0436,
+      "step": 8468
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.151774269060302e-06,
+      "loss": 0.0505,
+      "step": 8469
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.148860159521287e-06,
+      "loss": 0.0614,
+      "step": 8470
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.1459465886294515e-06,
+      "loss": 0.0668,
+      "step": 8471
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.1430335567082965e-06,
+      "loss": 0.0607,
+      "step": 8472
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.140121064081277e-06,
+      "loss": 0.0711,
+      "step": 8473
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.13720911107178e-06,
+      "loss": 0.0528,
+      "step": 8474
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.134297698003141e-06,
+      "loss": 0.0581,
+      "step": 8475
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.131386825198619e-06,
+      "loss": 0.0699,
+      "step": 8476
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.128476492981432e-06,
+      "loss": 0.0599,
+      "step": 8477
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.12556670167472e-06,
+      "loss": 0.0667,
+      "step": 8478
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.122657451601581e-06,
+      "loss": 0.0515,
+      "step": 8479
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.119748743085038e-06,
+      "loss": 0.0515,
+      "step": 8480
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.116840576448067e-06,
+      "loss": 0.0605,
+      "step": 8481
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.113932952013569e-06,
+      "loss": 0.0566,
+      "step": 8482
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.1110258701044e-06,
+      "loss": 0.0649,
+      "step": 8483
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.1081193310433395e-06,
+      "loss": 0.0639,
+      "step": 8484
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.105213335153123e-06,
+      "loss": 0.0652,
+      "step": 8485
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.102307882756415e-06,
+      "loss": 0.0611,
+      "step": 8486
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.099402974175825e-06,
+      "loss": 0.0692,
+      "step": 8487
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.096498609733898e-06,
+      "loss": 0.0616,
+      "step": 8488
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.0935947897531235e-06,
+      "loss": 0.0607,
+      "step": 8489
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.090691514555922e-06,
+      "loss": 0.0643,
+      "step": 8490
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.0877887844646626e-06,
+      "loss": 0.0592,
+      "step": 8491
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.08488659980165e-06,
+      "loss": 0.0571,
+      "step": 8492
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.081984960889135e-06,
+      "loss": 0.0555,
+      "step": 8493
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.079083868049292e-06,
+      "loss": 0.0765,
+      "step": 8494
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.076183321604251e-06,
+      "loss": 0.0741,
+      "step": 8495
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.073283321876068e-06,
+      "loss": 0.0668,
+      "step": 8496
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.07038386918675e-06,
+      "loss": 0.0508,
+      "step": 8497
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.067484963858237e-06,
+      "loss": 0.0781,
+      "step": 8498
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.0645866062124135e-06,
+      "loss": 0.0538,
+      "step": 8499
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.061688796571095e-06,
+      "loss": 0.0646,
+      "step": 8500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.058791535256037e-06,
+      "loss": 0.0552,
+      "step": 8501
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.0558948225889405e-06,
+      "loss": 0.0609,
+      "step": 8502
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.0529986588914435e-06,
+      "loss": 0.0595,
+      "step": 8503
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.050103044485126e-06,
+      "loss": 0.053,
+      "step": 8504
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.047207979691496e-06,
+      "loss": 0.0676,
+      "step": 8505
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.0443134648320115e-06,
+      "loss": 0.0637,
+      "step": 8506
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.041419500228062e-06,
+      "loss": 0.0791,
+      "step": 8507
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.038526086200981e-06,
+      "loss": 0.059,
+      "step": 8508
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.035633223072041e-06,
+      "loss": 0.0618,
+      "step": 8509
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.032740911162452e-06,
+      "loss": 0.0757,
+      "step": 8510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.02984915079336e-06,
+      "loss": 0.0674,
+      "step": 8511
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.0269579422858546e-06,
+      "loss": 0.0667,
+      "step": 8512
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.024067285960953e-06,
+      "loss": 0.0495,
+      "step": 8513
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.021177182139634e-06,
+      "loss": 0.0573,
+      "step": 8514
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.01828763114279e-06,
+      "loss": 0.0535,
+      "step": 8515
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.01539863329127e-06,
+      "loss": 0.0562,
+      "step": 8516
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.012510188905847e-06,
+      "loss": 0.0695,
+      "step": 8517
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.009622298307247e-06,
+      "loss": 0.0636,
+      "step": 8518
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.006734961816118e-06,
+      "loss": 0.0757,
+      "step": 8519
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.003848179753068e-06,
+      "loss": 0.0549,
+      "step": 8520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.000961952438621e-06,
+      "loss": 0.0798,
+      "step": 8521
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.998076280193259e-06,
+      "loss": 0.0499,
+      "step": 8522
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.995191163337385e-06,
+      "loss": 0.0601,
+      "step": 8523
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.992306602191351e-06,
+      "loss": 0.0594,
+      "step": 8524
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.989422597075446e-06,
+      "loss": 0.0754,
+      "step": 8525
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.9865391483098994e-06,
+      "loss": 0.0587,
+      "step": 8526
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.983656256214866e-06,
+      "loss": 0.0607,
+      "step": 8527
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.98077392111046e-06,
+      "loss": 0.0682,
+      "step": 8528
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.97789214331671e-06,
+      "loss": 0.0703,
+      "step": 8529
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.9750109231536015e-06,
+      "loss": 0.0596,
+      "step": 8530
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.9721302609410505e-06,
+      "loss": 0.0732,
+      "step": 8531
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.9692501569989145e-06,
+      "loss": 0.0638,
+      "step": 8532
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.966370611646981e-06,
+      "loss": 0.0935,
+      "step": 8533
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.963491625204987e-06,
+      "loss": 0.0615,
+      "step": 8534
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.960613197992593e-06,
+      "loss": 0.0679,
+      "step": 8535
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.95773533032941e-06,
+      "loss": 0.0522,
+      "step": 8536
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.954858022534984e-06,
+      "loss": 0.0488,
+      "step": 8537
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.9519812749288e-06,
+      "loss": 0.0551,
+      "step": 8538
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.9491050878302694e-06,
+      "loss": 0.0704,
+      "step": 8539
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.946229461558759e-06,
+      "loss": 0.0744,
+      "step": 8540
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.943354396433557e-06,
+      "loss": 0.0573,
+      "step": 8541
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.940479892773902e-06,
+      "loss": 0.0571,
+      "step": 8542
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.937605950898962e-06,
+      "loss": 0.0602,
+      "step": 8543
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.93473257112785e-06,
+      "loss": 0.0638,
+      "step": 8544
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.931859753779608e-06,
+      "loss": 0.0795,
+      "step": 8545
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.928987499173222e-06,
+      "loss": 0.0685,
+      "step": 8546
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.92611580762761e-06,
+      "loss": 0.073,
+      "step": 8547
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.923244679461634e-06,
+      "loss": 0.0574,
+      "step": 8548
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.9203741149940874e-06,
+      "loss": 0.0625,
+      "step": 8549
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.917504114543711e-06,
+      "loss": 0.0576,
+      "step": 8550
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.914634678429166e-06,
+      "loss": 0.0573,
+      "step": 8551
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.911765806969071e-06,
+      "loss": 0.0606,
+      "step": 8552
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.9088975004819604e-06,
+      "loss": 0.0726,
+      "step": 8553
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.906029759286324e-06,
+      "loss": 0.0743,
+      "step": 8554
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.9031625837005795e-06,
+      "loss": 0.0556,
+      "step": 8555
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.90029597404309e-06,
+      "loss": 0.0591,
+      "step": 8556
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.897429930632141e-06,
+      "loss": 0.0818,
+      "step": 8557
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.894564453785972e-06,
+      "loss": 0.0649,
+      "step": 8558
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.891699543822743e-06,
+      "loss": 0.0584,
+      "step": 8559
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.888835201060571e-06,
+      "loss": 0.0472,
+      "step": 8560
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.88597142581749e-06,
+      "loss": 0.0542,
+      "step": 8561
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.883108218411485e-06,
+      "loss": 0.0654,
+      "step": 8562
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.8802455791604684e-06,
+      "loss": 0.0591,
+      "step": 8563
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.8773835083823e-06,
+      "loss": 0.0641,
+      "step": 8564
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.874522006394757e-06,
+      "loss": 0.0547,
+      "step": 8565
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.871661073515585e-06,
+      "loss": 0.0561,
+      "step": 8566
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.868800710062438e-06,
+      "loss": 0.0596,
+      "step": 8567
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.8659409163529155e-06,
+      "loss": 0.0616,
+      "step": 8568
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.863081692704561e-06,
+      "loss": 0.0616,
+      "step": 8569
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.8602230394348395e-06,
+      "loss": 0.0547,
+      "step": 8570
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.857364956861175e-06,
+      "loss": 0.072,
+      "step": 8571
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.854507445300905e-06,
+      "loss": 0.0635,
+      "step": 8572
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.851650505071321e-06,
+      "loss": 0.0643,
+      "step": 8573
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.848794136489637e-06,
+      "loss": 0.0479,
+      "step": 8574
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.845938339873013e-06,
+      "loss": 0.0624,
+      "step": 8575
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.843083115538544e-06,
+      "loss": 0.0598,
+      "step": 8576
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.840228463803264e-06,
+      "loss": 0.0643,
+      "step": 8577
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.837374384984132e-06,
+      "loss": 0.0815,
+      "step": 8578
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.834520879398058e-06,
+      "loss": 0.065,
+      "step": 8579
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.831667947361874e-06,
+      "loss": 0.0767,
+      "step": 8580
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.828815589192362e-06,
+      "loss": 0.0655,
+      "step": 8581
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.8259638052062304e-06,
+      "loss": 0.0599,
+      "step": 8582
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.823112595720134e-06,
+      "loss": 0.0657,
+      "step": 8583
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.820261961050648e-06,
+      "loss": 0.0547,
+      "step": 8584
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.817411901514303e-06,
+      "loss": 0.0656,
+      "step": 8585
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.814562417427547e-06,
+      "loss": 0.0701,
+      "step": 8586
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.811713509106776e-06,
+      "loss": 0.0621,
+      "step": 8587
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.80886517686832e-06,
+      "loss": 0.0588,
+      "step": 8588
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.806017421028447e-06,
+      "loss": 0.0715,
+      "step": 8589
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.803170241903352e-06,
+      "loss": 0.0718,
+      "step": 8590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.800323639809179e-06,
+      "loss": 0.0773,
+      "step": 8591
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.797477615061992e-06,
+      "loss": 0.0532,
+      "step": 8592
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.794632167977805e-06,
+      "loss": 0.0605,
+      "step": 8593
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.791787298872563e-06,
+      "loss": 0.0621,
+      "step": 8594
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.78894300806215e-06,
+      "loss": 0.0563,
+      "step": 8595
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.786099295862374e-06,
+      "loss": 0.0699,
+      "step": 8596
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.783256162588996e-06,
+      "loss": 0.0554,
+      "step": 8597
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.780413608557694e-06,
+      "loss": 0.0544,
+      "step": 8598
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.777571634084099e-06,
+      "loss": 0.0716,
+      "step": 8599
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.774730239483767e-06,
+      "loss": 0.0712,
+      "step": 8600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.771889425072197e-06,
+      "loss": 0.0647,
+      "step": 8601
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.769049191164813e-06,
+      "loss": 0.056,
+      "step": 8602
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.766209538076987e-06,
+      "loss": 0.0646,
+      "step": 8603
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7633704661240135e-06,
+      "loss": 0.0635,
+      "step": 8604
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7605319756211335e-06,
+      "loss": 0.0669,
+      "step": 8605
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.757694066883518e-06,
+      "loss": 0.0567,
+      "step": 8606
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.754856740226282e-06,
+      "loss": 0.0616,
+      "step": 8607
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.752019995964456e-06,
+      "loss": 0.0581,
+      "step": 8608
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.749183834413028e-06,
+      "loss": 0.0561,
+      "step": 8609
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.746348255886902e-06,
+      "loss": 0.0635,
+      "step": 8610
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.743513260700942e-06,
+      "loss": 0.0551,
+      "step": 8611
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.740678849169917e-06,
+      "loss": 0.0683,
+      "step": 8612
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.737845021608558e-06,
+      "loss": 0.059,
+      "step": 8613
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.735011778331511e-06,
+      "loss": 0.0776,
+      "step": 8614
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.732179119653373e-06,
+      "loss": 0.0422,
+      "step": 8615
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.729347045888657e-06,
+      "loss": 0.0664,
+      "step": 8616
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.726515557351837e-06,
+      "loss": 0.0625,
+      "step": 8617
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.723684654357299e-06,
+      "loss": 0.0712,
+      "step": 8618
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.72085433721938e-06,
+      "loss": 0.058,
+      "step": 8619
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.718024606252335e-06,
+      "loss": 0.0554,
+      "step": 8620
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.7151954617703686e-06,
+      "loss": 0.0538,
+      "step": 8621
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7123669040876164e-06,
+      "loss": 0.0547,
+      "step": 8622
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.70953893351815e-06,
+      "loss": 0.0648,
+      "step": 8623
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.7067115503759665e-06,
+      "loss": 0.0726,
+      "step": 8624
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.703884754975013e-06,
+      "loss": 0.0607,
+      "step": 8625
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.701058547629156e-06,
+      "loss": 0.0619,
+      "step": 8626
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.698232928652206e-06,
+      "loss": 0.0643,
+      "step": 8627
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.695407898357909e-06,
+      "loss": 0.0531,
+      "step": 8628
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.692583457059944e-06,
+      "loss": 0.0521,
+      "step": 8629
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.689759605071916e-06,
+      "loss": 0.0562,
+      "step": 8630
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.68693634270738e-06,
+      "loss": 0.0545,
+      "step": 8631
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.684113670279811e-06,
+      "loss": 0.0554,
+      "step": 8632
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.6812915881026265e-06,
+      "loss": 0.0718,
+      "step": 8633
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.678470096489182e-06,
+      "loss": 0.051,
+      "step": 8634
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.6756491957527535e-06,
+      "loss": 0.0611,
+      "step": 8635
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.6728288862065705e-06,
+      "loss": 0.0731,
+      "step": 8636
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.670009168163777e-06,
+      "loss": 0.067,
+      "step": 8637
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.667190041937464e-06,
+      "loss": 0.0596,
+      "step": 8638
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.664371507840656e-06,
+      "loss": 0.0702,
+      "step": 8639
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.661553566186311e-06,
+      "loss": 0.0659,
+      "step": 8640
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.658736217287313e-06,
+      "loss": 0.0582,
+      "step": 8641
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.655919461456495e-06,
+      "loss": 0.0596,
+      "step": 8642
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.6531032990066075e-06,
+      "loss": 0.073,
+      "step": 8643
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.650287730250347e-06,
+      "loss": 0.0651,
+      "step": 8644
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.647472755500342e-06,
+      "loss": 0.0734,
+      "step": 8645
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.644658375069157e-06,
+      "loss": 0.0535,
+      "step": 8646
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.6418445892692794e-06,
+      "loss": 0.0678,
+      "step": 8647
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.639031398413146e-06,
+      "loss": 0.0627,
+      "step": 8648
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.636218802813113e-06,
+      "loss": 0.0568,
+      "step": 8649
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.6334068027814805e-06,
+      "loss": 0.0652,
+      "step": 8650
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.630595398630479e-06,
+      "loss": 0.0632,
+      "step": 8651
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.627784590672278e-06,
+      "loss": 0.0698,
+      "step": 8652
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.624974379218968e-06,
+      "loss": 0.0497,
+      "step": 8653
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.622164764582589e-06,
+      "loss": 0.057,
+      "step": 8654
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.6193557470751e-06,
+      "loss": 0.0556,
+      "step": 8655
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.616547327008405e-06,
+      "loss": 0.0579,
+      "step": 8656
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.6137395046943355e-06,
+      "loss": 0.0598,
+      "step": 8657
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.610932280444665e-06,
+      "loss": 0.0609,
+      "step": 8658
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.608125654571083e-06,
+      "loss": 0.0724,
+      "step": 8659
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.605319627385233e-06,
+      "loss": 0.0692,
+      "step": 8660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.602514199198677e-06,
+      "loss": 0.0681,
+      "step": 8661
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.599709370322917e-06,
+      "loss": 0.0544,
+      "step": 8662
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.59690514106939e-06,
+      "loss": 0.0706,
+      "step": 8663
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.594101511749466e-06,
+      "loss": 0.0705,
+      "step": 8664
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.591298482674441e-06,
+      "loss": 0.0676,
+      "step": 8665
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.5884960541555556e-06,
+      "loss": 0.0547,
+      "step": 8666
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.585694226503967e-06,
+      "loss": 0.0472,
+      "step": 8667
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.582893000030793e-06,
+      "loss": 0.0461,
+      "step": 8668
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.580092375047055e-06,
+      "loss": 0.0705,
+      "step": 8669
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.577292351863731e-06,
+      "loss": 0.0577,
+      "step": 8670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.574492930791711e-06,
+      "loss": 0.061,
+      "step": 8671
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.571694112141835e-06,
+      "loss": 0.0723,
+      "step": 8672
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.568895896224872e-06,
+      "loss": 0.0504,
+      "step": 8673
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.566098283351523e-06,
+      "loss": 0.0579,
+      "step": 8674
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.5633012738324156e-06,
+      "loss": 0.0607,
+      "step": 8675
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.560504867978124e-06,
+      "loss": 0.0496,
+      "step": 8676
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.5577090660991385e-06,
+      "loss": 0.0629,
+      "step": 8677
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.5549138685058965e-06,
+      "loss": 0.0703,
+      "step": 8678
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.552119275508764e-06,
+      "loss": 0.0785,
+      "step": 8679
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.549325287418042e-06,
+      "loss": 0.047,
+      "step": 8680
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.546531904543954e-06,
+      "loss": 0.0538,
+      "step": 8681
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.54373912719667e-06,
+      "loss": 0.0798,
+      "step": 8682
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.5409469556862815e-06,
+      "loss": 0.0557,
+      "step": 8683
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.538155390322819e-06,
+      "loss": 0.0587,
+      "step": 8684
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.535364431416247e-06,
+      "loss": 0.0709,
+      "step": 8685
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.532574079276462e-06,
+      "loss": 0.0629,
+      "step": 8686
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.529784334213284e-06,
+      "loss": 0.0563,
+      "step": 8687
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.526995196536481e-06,
+      "loss": 0.0659,
+      "step": 8688
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.524206666555738e-06,
+      "loss": 0.0544,
+      "step": 8689
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.521418744580683e-06,
+      "loss": 0.0673,
+      "step": 8690
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.518631430920873e-06,
+      "loss": 0.0545,
+      "step": 8691
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.515844725885803e-06,
+      "loss": 0.0661,
+      "step": 8692
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.513058629784887e-06,
+      "loss": 0.075,
+      "step": 8693
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.510273142927487e-06,
+      "loss": 0.0677,
+      "step": 8694
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.507488265622882e-06,
+      "loss": 0.058,
+      "step": 8695
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.5047039981802975e-06,
+      "loss": 0.0496,
+      "step": 8696
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.501920340908883e-06,
+      "loss": 0.0582,
+      "step": 8697
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.4991372941177255e-06,
+      "loss": 0.0586,
+      "step": 8698
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.496354858115835e-06,
+      "loss": 0.0662,
+      "step": 8699
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.493573033212169e-06,
+      "loss": 0.0694,
+      "step": 8700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.490791819715597e-06,
+      "loss": 0.0778,
+      "step": 8701
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.488011217934936e-06,
+      "loss": 0.0676,
+      "step": 8702
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.485231228178936e-06,
+      "loss": 0.0575,
+      "step": 8703
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.482451850756264e-06,
+      "loss": 0.0582,
+      "step": 8704
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.479673085975539e-06,
+      "loss": 0.0653,
+      "step": 8705
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.4768949341452915e-06,
+      "loss": 0.0567,
+      "step": 8706
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.474117395573999e-06,
+      "loss": 0.076,
+      "step": 8707
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.471340470570067e-06,
+      "loss": 0.0615,
+      "step": 8708
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.468564159441833e-06,
+      "loss": 0.0582,
+      "step": 8709
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.46578846249756e-06,
+      "loss": 0.0665,
+      "step": 8710
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.4630133800454545e-06,
+      "loss": 0.0682,
+      "step": 8711
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.4602389123936406e-06,
+      "loss": 0.0501,
+      "step": 8712
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.457465059850185e-06,
+      "loss": 0.0485,
+      "step": 8713
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.454691822723086e-06,
+      "loss": 0.0569,
+      "step": 8714
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.451919201320271e-06,
+      "loss": 0.0582,
+      "step": 8715
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.449147195949592e-06,
+      "loss": 0.0654,
+      "step": 8716
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.446375806918847e-06,
+      "loss": 0.0757,
+      "step": 8717
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.443605034535746e-06,
+      "loss": 0.0717,
+      "step": 8718
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.440834879107957e-06,
+      "loss": 0.0522,
+      "step": 8719
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.438065340943054e-06,
+      "loss": 0.061,
+      "step": 8720
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.435296420348561e-06,
+      "loss": 0.0666,
+      "step": 8721
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.432528117631916e-06,
+      "loss": 0.0562,
+      "step": 8722
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.429760433100508e-06,
+      "loss": 0.0559,
+      "step": 8723
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.426993367061635e-06,
+      "loss": 0.0524,
+      "step": 8724
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.424226919822554e-06,
+      "loss": 0.0604,
+      "step": 8725
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.421461091690425e-06,
+      "loss": 0.063,
+      "step": 8726
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.418695882972364e-06,
+      "loss": 0.0621,
+      "step": 8727
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.415931293975394e-06,
+      "loss": 0.0711,
+      "step": 8728
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.413167325006488e-06,
+      "loss": 0.056,
+      "step": 8729
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.410403976372544e-06,
+      "loss": 0.0602,
+      "step": 8730
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.407641248380392e-06,
+      "loss": 0.0695,
+      "step": 8731
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.404879141336788e-06,
+      "loss": 0.0511,
+      "step": 8732
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.402117655548429e-06,
+      "loss": 0.0578,
+      "step": 8733
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.3993567913219295e-06,
+      "loss": 0.0709,
+      "step": 8734
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.3965965489638466e-06,
+      "loss": 0.0578,
+      "step": 8735
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.393836928780664e-06,
+      "loss": 0.0611,
+      "step": 8736
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.3910779310788e-06,
+      "loss": 0.0611,
+      "step": 8737
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.388319556164593e-06,
+      "loss": 0.0678,
+      "step": 8738
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.385561804344328e-06,
+      "loss": 0.0565,
+      "step": 8739
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.3828046759242035e-06,
+      "loss": 0.0491,
+      "step": 8740
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.380048171210364e-06,
+      "loss": 0.0573,
+      "step": 8741
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.377292290508876e-06,
+      "loss": 0.0695,
+      "step": 8742
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.374537034125744e-06,
+      "loss": 0.0652,
+      "step": 8743
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.371782402366892e-06,
+      "loss": 0.0739,
+      "step": 8744
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.369028395538186e-06,
+      "loss": 0.0718,
+      "step": 8745
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.3662750139454115e-06,
+      "loss": 0.0495,
+      "step": 8746
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.363522257894295e-06,
+      "loss": 0.0517,
+      "step": 8747
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.360770127690488e-06,
+      "loss": 0.0627,
+      "step": 8748
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.358018623639578e-06,
+      "loss": 0.0436,
+      "step": 8749
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.355267746047072e-06,
+      "loss": 0.0566,
+      "step": 8750
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.352517495218421e-06,
+      "loss": 0.0703,
+      "step": 8751
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.349767871458993e-06,
+      "loss": 0.0573,
+      "step": 8752
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.347018875074095e-06,
+      "loss": 0.0712,
+      "step": 8753
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.344270506368964e-06,
+      "loss": 0.0641,
+      "step": 8754
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.341522765648768e-06,
+      "loss": 0.0648,
+      "step": 8755
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.338775653218595e-06,
+      "loss": 0.0645,
+      "step": 8756
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.336029169383481e-06,
+      "loss": 0.0599,
+      "step": 8757
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.333283314448373e-06,
+      "loss": 0.0546,
+      "step": 8758
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.330538088718162e-06,
+      "loss": 0.0768,
+      "step": 8759
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.327793492497664e-06,
+      "loss": 0.0694,
+      "step": 8760
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.325049526091629e-06,
+      "loss": 0.0633,
+      "step": 8761
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.322306189804728e-06,
+      "loss": 0.0732,
+      "step": 8762
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.319563483941574e-06,
+      "loss": 0.0637,
+      "step": 8763
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.316821408806694e-06,
+      "loss": 0.0701,
+      "step": 8764
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.314079964704568e-06,
+      "loss": 0.066,
+      "step": 8765
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.311339151939582e-06,
+      "loss": 0.0734,
+      "step": 8766
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.3085989708160705e-06,
+      "loss": 0.059,
+      "step": 8767
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.305859421638282e-06,
+      "loss": 0.0596,
+      "step": 8768
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.3031205047104076e-06,
+      "loss": 0.049,
+      "step": 8769
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.300382220336567e-06,
+      "loss": 0.0701,
+      "step": 8770
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.297644568820797e-06,
+      "loss": 0.072,
+      "step": 8771
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.294907550467083e-06,
+      "loss": 0.0765,
+      "step": 8772
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.292171165579319e-06,
+      "loss": 0.0617,
+      "step": 8773
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.2894354144613525e-06,
+      "loss": 0.0489,
+      "step": 8774
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.286700297416935e-06,
+      "loss": 0.0573,
+      "step": 8775
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.283965814749774e-06,
+      "loss": 0.0626,
+      "step": 8776
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.281231966763484e-06,
+      "loss": 0.0476,
+      "step": 8777
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.278498753761624e-06,
+      "loss": 0.0608,
+      "step": 8778
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.275766176047672e-06,
+      "loss": 0.0529,
+      "step": 8779
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.273034233925041e-06,
+      "loss": 0.0477,
+      "step": 8780
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.270302927697076e-06,
+      "loss": 0.0582,
+      "step": 8781
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.267572257667048e-06,
+      "loss": 0.0604,
+      "step": 8782
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.264842224138151e-06,
+      "loss": 0.0648,
+      "step": 8783
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.262112827413525e-06,
+      "loss": 0.0568,
+      "step": 8784
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.259384067796219e-06,
+      "loss": 0.0634,
+      "step": 8785
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.256655945589225e-06,
+      "loss": 0.0731,
+      "step": 8786
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.253928461095461e-06,
+      "loss": 0.0625,
+      "step": 8787
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2512016146177794e-06,
+      "loss": 0.0648,
+      "step": 8788
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.248475406458945e-06,
+      "loss": 0.0717,
+      "step": 8789
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.245749836921673e-06,
+      "loss": 0.0686,
+      "step": 8790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2430249063085896e-06,
+      "loss": 0.055,
+      "step": 8791
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.24030061492226e-06,
+      "loss": 0.0609,
+      "step": 8792
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.237576963065179e-06,
+      "loss": 0.0651,
+      "step": 8793
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.23485395103977e-06,
+      "loss": 0.0692,
+      "step": 8794
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.232131579148376e-06,
+      "loss": 0.0711,
+      "step": 8795
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.2294098476932824e-06,
+      "loss": 0.0588,
+      "step": 8796
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.22668875697669e-06,
+      "loss": 0.0649,
+      "step": 8797
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.223968307300741e-06,
+      "loss": 0.0593,
+      "step": 8798
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2212484989675e-06,
+      "loss": 0.0655,
+      "step": 8799
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.218529332278966e-06,
+      "loss": 0.0613,
+      "step": 8800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.215810807537052e-06,
+      "loss": 0.0662,
+      "step": 8801
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.213092925043619e-06,
+      "loss": 0.0573,
+      "step": 8802
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.210375685100442e-06,
+      "loss": 0.0626,
+      "step": 8803
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.20765908800923e-06,
+      "loss": 0.0698,
+      "step": 8804
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.204943134071625e-06,
+      "loss": 0.0722,
+      "step": 8805
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.2022278235891944e-06,
+      "loss": 0.0723,
+      "step": 8806
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.1995131568634265e-06,
+      "loss": 0.0614,
+      "step": 8807
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.1967991341957525e-06,
+      "loss": 0.0565,
+      "step": 8808
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.1940857558875155e-06,
+      "loss": 0.0571,
+      "step": 8809
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.191373022240002e-06,
+      "loss": 0.0554,
+      "step": 8810
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.188660933554419e-06,
+      "loss": 0.0598,
+      "step": 8811
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.18594949013191e-06,
+      "loss": 0.0584,
+      "step": 8812
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.183238692273529e-06,
+      "loss": 0.074,
+      "step": 8813
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.1805285402802805e-06,
+      "loss": 0.0561,
+      "step": 8814
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.177819034453076e-06,
+      "loss": 0.0696,
+      "step": 8815
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.175110175092778e-06,
+      "loss": 0.0609,
+      "step": 8816
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.172401962500156e-06,
+      "loss": 0.0691,
+      "step": 8817
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.169694396975924e-06,
+      "loss": 0.0864,
+      "step": 8818
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.1669874788207095e-06,
+      "loss": 0.0574,
+      "step": 8819
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.164281208335083e-06,
+      "loss": 0.0585,
+      "step": 8820
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.161575585819526e-06,
+      "loss": 0.0524,
+      "step": 8821
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.158870611574471e-06,
+      "loss": 0.0616,
+      "step": 8822
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.1561662859002526e-06,
+      "loss": 0.0601,
+      "step": 8823
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.1534626090971576e-06,
+      "loss": 0.067,
+      "step": 8824
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.150759581465378e-06,
+      "loss": 0.0435,
+      "step": 8825
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.1480572033050505e-06,
+      "loss": 0.055,
+      "step": 8826
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.145355474916234e-06,
+      "loss": 0.0528,
+      "step": 8827
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.1426543965989195e-06,
+      "loss": 0.0542,
+      "step": 8828
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.139953968653013e-06,
+      "loss": 0.0607,
+      "step": 8829
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.137254191378366e-06,
+      "loss": 0.0574,
+      "step": 8830
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.13455506507474e-06,
+      "loss": 0.0648,
+      "step": 8831
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.131856590041837e-06,
+      "loss": 0.062,
+      "step": 8832
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.129158766579284e-06,
+      "loss": 0.0647,
+      "step": 8833
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.126461594986636e-06,
+      "loss": 0.0626,
+      "step": 8834
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.123765075563367e-06,
+      "loss": 0.0643,
+      "step": 8835
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.121069208608894e-06,
+      "loss": 0.059,
+      "step": 8836
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1183739944225455e-06,
+      "loss": 0.0588,
+      "step": 8837
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.115679433303587e-06,
+      "loss": 0.0622,
+      "step": 8838
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.112985525551216e-06,
+      "loss": 0.0535,
+      "step": 8839
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.110292271464541e-06,
+      "loss": 0.0679,
+      "step": 8840
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.107599671342617e-06,
+      "loss": 0.0633,
+      "step": 8841
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.104907725484409e-06,
+      "loss": 0.0607,
+      "step": 8842
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.102216434188823e-06,
+      "loss": 0.0613,
+      "step": 8843
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.099525797754686e-06,
+      "loss": 0.0608,
+      "step": 8844
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.096835816480755e-06,
+      "loss": 0.0452,
+      "step": 8845
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.094146490665708e-06,
+      "loss": 0.0729,
+      "step": 8846
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.091457820608162e-06,
+      "loss": 0.0634,
+      "step": 8847
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.088769806606644e-06,
+      "loss": 0.0588,
+      "step": 8848
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.086082448959624e-06,
+      "loss": 0.0759,
+      "step": 8849
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.083395747965494e-06,
+      "loss": 0.0802,
+      "step": 8850
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.080709703922574e-06,
+      "loss": 0.0609,
+      "step": 8851
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.078024317129102e-06,
+      "loss": 0.0711,
+      "step": 8852
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.075339587883259e-06,
+      "loss": 0.0624,
+      "step": 8853
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.072655516483137e-06,
+      "loss": 0.0702,
+      "step": 8854
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.069972103226766e-06,
+      "loss": 0.0627,
+      "step": 8855
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.067289348412099e-06,
+      "loss": 0.0693,
+      "step": 8856
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.06460725233702e-06,
+      "loss": 0.0473,
+      "step": 8857
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.061925815299327e-06,
+      "loss": 0.0492,
+      "step": 8858
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.059245037596763e-06,
+      "loss": 0.0497,
+      "step": 8859
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0565649195269804e-06,
+      "loss": 0.0649,
+      "step": 8860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0538854613875726e-06,
+      "loss": 0.0655,
+      "step": 8861
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.051206663476049e-06,
+      "loss": 0.0699,
+      "step": 8862
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.048528526089859e-06,
+      "loss": 0.0641,
+      "step": 8863
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.04585104952636e-06,
+      "loss": 0.0751,
+      "step": 8864
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.043174234082854e-06,
+      "loss": 0.0521,
+      "step": 8865
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.040498080056555e-06,
+      "loss": 0.0566,
+      "step": 8866
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0378225877446135e-06,
+      "loss": 0.0701,
+      "step": 8867
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.035147757444102e-06,
+      "loss": 0.0621,
+      "step": 8868
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0324735894520265e-06,
+      "loss": 0.0539,
+      "step": 8869
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.029800084065304e-06,
+      "loss": 0.0572,
+      "step": 8870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.027127241580797e-06,
+      "loss": 0.072,
+      "step": 8871
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.024455062295274e-06,
+      "loss": 0.0684,
+      "step": 8872
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.021783546505455e-06,
+      "loss": 0.0508,
+      "step": 8873
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0191126945079604e-06,
+      "loss": 0.0639,
+      "step": 8874
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.016442506599357e-06,
+      "loss": 0.0616,
+      "step": 8875
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.013772983076123e-06,
+      "loss": 0.0529,
+      "step": 8876
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.011104124234672e-06,
+      "loss": 0.0806,
+      "step": 8877
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.00843593037134e-06,
+      "loss": 0.0569,
+      "step": 8878
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.005768401782396e-06,
+      "loss": 0.0722,
+      "step": 8879
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.003101538764023e-06,
+      "loss": 0.0511,
+      "step": 8880
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.000435341612341e-06,
+      "loss": 0.0627,
+      "step": 8881
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.997769810623386e-06,
+      "loss": 0.0538,
+      "step": 8882
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.995104946093131e-06,
+      "loss": 0.0611,
+      "step": 8883
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.992440748317466e-06,
+      "loss": 0.0665,
+      "step": 8884
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.989777217592218e-06,
+      "loss": 0.0646,
+      "step": 8885
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.987114354213123e-06,
+      "loss": 0.063,
+      "step": 8886
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.98445215847586e-06,
+      "loss": 0.0672,
+      "step": 8887
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.98179063067602e-06,
+      "loss": 0.0554,
+      "step": 8888
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.97912977110913e-06,
+      "loss": 0.0632,
+      "step": 8889
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.976469580070638e-06,
+      "loss": 0.0768,
+      "step": 8890
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.973810057855922e-06,
+      "loss": 0.0588,
+      "step": 8891
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.971151204760277e-06,
+      "loss": 0.0578,
+      "step": 8892
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.968493021078935e-06,
+      "loss": 0.0662,
+      "step": 8893
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.965835507107042e-06,
+      "loss": 0.0573,
+      "step": 8894
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.963178663139678e-06,
+      "loss": 0.0577,
+      "step": 8895
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.960522489471847e-06,
+      "loss": 0.0575,
+      "step": 8896
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.95786698639848e-06,
+      "loss": 0.0684,
+      "step": 8897
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.955212154214425e-06,
+      "loss": 0.0577,
+      "step": 8898
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.952557993214468e-06,
+      "loss": 0.058,
+      "step": 8899
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.94990450369331e-06,
+      "loss": 0.0466,
+      "step": 8900
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.947251685945582e-06,
+      "loss": 0.0532,
+      "step": 8901
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.944599540265842e-06,
+      "loss": 0.0569,
+      "step": 8902
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.941948066948574e-06,
+      "loss": 0.0478,
+      "step": 8903
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.9392972662881815e-06,
+      "loss": 0.0547,
+      "step": 8904
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.936647138578993e-06,
+      "loss": 0.0555,
+      "step": 8905
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.93399768411527e-06,
+      "loss": 0.0627,
+      "step": 8906
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.931348903191193e-06,
+      "loss": 0.0604,
+      "step": 8907
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.9287007961008775e-06,
+      "loss": 0.0731,
+      "step": 8908
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.926053363138346e-06,
+      "loss": 0.069,
+      "step": 8909
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.923406604597565e-06,
+      "loss": 0.0642,
+      "step": 8910
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.920760520772411e-06,
+      "loss": 0.0745,
+      "step": 8911
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.918115111956695e-06,
+      "loss": 0.0623,
+      "step": 8912
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.915470378444151e-06,
+      "loss": 0.0636,
+      "step": 8913
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.912826320528441e-06,
+      "loss": 0.0604,
+      "step": 8914
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.910182938503141e-06,
+      "loss": 0.068,
+      "step": 8915
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.907540232661767e-06,
+      "loss": 0.0673,
+      "step": 8916
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.904898203297746e-06,
+      "loss": 0.0627,
+      "step": 8917
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.902256850704438e-06,
+      "loss": 0.0707,
+      "step": 8918
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.899616175175126e-06,
+      "loss": 0.0504,
+      "step": 8919
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8969761770030235e-06,
+      "loss": 0.0588,
+      "step": 8920
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8943368564812545e-06,
+      "loss": 0.0466,
+      "step": 8921
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8916982139028835e-06,
+      "loss": 0.0582,
+      "step": 8922
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.889060249560883e-06,
+      "loss": 0.0527,
+      "step": 8923
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.886422963748173e-06,
+      "loss": 0.0504,
+      "step": 8924
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.883786356757575e-06,
+      "loss": 0.0535,
+      "step": 8925
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.8811504288818515e-06,
+      "loss": 0.0615,
+      "step": 8926
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.878515180413676e-06,
+      "loss": 0.058,
+      "step": 8927
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.875880611645662e-06,
+      "loss": 0.064,
+      "step": 8928
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.87324672287033e-06,
+      "loss": 0.0571,
+      "step": 8929
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.870613514380144e-06,
+      "loss": 0.0701,
+      "step": 8930
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.867980986467475e-06,
+      "loss": 0.0613,
+      "step": 8931
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.865349139424634e-06,
+      "loss": 0.0678,
+      "step": 8932
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.862717973543841e-06,
+      "loss": 0.0653,
+      "step": 8933
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.860087489117251e-06,
+      "loss": 0.0616,
+      "step": 8934
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.857457686436939e-06,
+      "loss": 0.0632,
+      "step": 8935
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8548285657949126e-06,
+      "loss": 0.0641,
+      "step": 8936
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.852200127483087e-06,
+      "loss": 0.0624,
+      "step": 8937
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.849572371793318e-06,
+      "loss": 0.0562,
+      "step": 8938
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.846945299017373e-06,
+      "loss": 0.0527,
+      "step": 8939
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8443189094469535e-06,
+      "loss": 0.062,
+      "step": 8940
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.84169320337368e-06,
+      "loss": 0.0633,
+      "step": 8941
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.839068181089102e-06,
+      "loss": 0.0527,
+      "step": 8942
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8364438428846825e-06,
+      "loss": 0.0696,
+      "step": 8943
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.833820189051822e-06,
+      "loss": 0.0582,
+      "step": 8944
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.831197219881833e-06,
+      "loss": 0.0588,
+      "step": 8945
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.82857493566596e-06,
+      "loss": 0.064,
+      "step": 8946
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.825953336695366e-06,
+      "loss": 0.058,
+      "step": 8947
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.823332423261148e-06,
+      "loss": 0.0681,
+      "step": 8948
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.820712195654312e-06,
+      "loss": 0.0515,
+      "step": 8949
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8180926541658005e-06,
+      "loss": 0.079,
+      "step": 8950
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.815473799086469e-06,
+      "loss": 0.0598,
+      "step": 8951
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.8128556307071062e-06,
+      "loss": 0.0664,
+      "step": 8952
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8102381493184206e-06,
+      "loss": 0.0601,
+      "step": 8953
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8076213552110487e-06,
+      "loss": 0.0617,
+      "step": 8954
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8050052486755384e-06,
+      "loss": 0.0638,
+      "step": 8955
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.8023898300023776e-06,
+      "loss": 0.0601,
+      "step": 8956
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7997750994819626e-06,
+      "loss": 0.0554,
+      "step": 8957
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7971610574046237e-06,
+      "loss": 0.0529,
+      "step": 8958
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7945477040606125e-06,
+      "loss": 0.0654,
+      "step": 8959
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7919350397401054e-06,
+      "loss": 0.0477,
+      "step": 8960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7893230647331935e-06,
+      "loss": 0.0574,
+      "step": 8961
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7867117793299047e-06,
+      "loss": 0.0508,
+      "step": 8962
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.784101183820178e-06,
+      "loss": 0.0561,
+      "step": 8963
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.781491278493883e-06,
+      "loss": 0.0507,
+      "step": 8964
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7788820636408107e-06,
+      "loss": 0.0553,
+      "step": 8965
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.776273539550681e-06,
+      "loss": 0.0616,
+      "step": 8966
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7736657065131244e-06,
+      "loss": 0.062,
+      "step": 8967
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7710585648177076e-06,
+      "loss": 0.071,
+      "step": 8968
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7684521147539065e-06,
+      "loss": 0.0682,
+      "step": 8969
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7658463566111423e-06,
+      "loss": 0.0465,
+      "step": 8970
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.7632412906787384e-06,
+      "loss": 0.0599,
+      "step": 8971
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7606369172459445e-06,
+      "loss": 0.0655,
+      "step": 8972
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.758033236601946e-06,
+      "loss": 0.0618,
+      "step": 8973
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.755430249035832e-06,
+      "loss": 0.0637,
+      "step": 8974
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7528279548366397e-06,
+      "loss": 0.0678,
+      "step": 8975
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.750226354293305e-06,
+      "loss": 0.0599,
+      "step": 8976
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7476254476947025e-06,
+      "loss": 0.0638,
+      "step": 8977
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7450252353296202e-06,
+      "loss": 0.069,
+      "step": 8978
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7424257174867784e-06,
+      "loss": 0.0551,
+      "step": 8979
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7398268944548043e-06,
+      "loss": 0.0505,
+      "step": 8980
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.737228766522274e-06,
+      "loss": 0.0679,
+      "step": 8981
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7346313339776597e-06,
+      "loss": 0.0569,
+      "step": 8982
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.732034597109374e-06,
+      "loss": 0.0557,
+      "step": 8983
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.72943855620574e-06,
+      "loss": 0.0626,
+      "step": 8984
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7268432115550135e-06,
+      "loss": 0.0658,
+      "step": 8985
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7242485634453686e-06,
+      "loss": 0.0568,
+      "step": 8986
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7216546121649054e-06,
+      "loss": 0.0737,
+      "step": 8987
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7190613580016376e-06,
+      "loss": 0.0524,
+      "step": 8988
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7164688012435136e-06,
+      "loss": 0.0537,
+      "step": 8989
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.7138769421783925e-06,
+      "loss": 0.0596,
+      "step": 8990
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.711285781094065e-06,
+      "loss": 0.0595,
+      "step": 8991
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.7086953182782413e-06,
+      "loss": 0.0729,
+      "step": 8992
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.7061055540185576e-06,
+      "loss": 0.0597,
+      "step": 8993
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.70351648860256e-06,
+      "loss": 0.0511,
+      "step": 8994
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.700928122317735e-06,
+      "loss": 0.0598,
+      "step": 8995
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6983404554514746e-06,
+      "loss": 0.0555,
+      "step": 8996
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.695753488291105e-06,
+      "loss": 0.0678,
+      "step": 8997
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.693167221123869e-06,
+      "loss": 0.0633,
+      "step": 8998
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6905816542369376e-06,
+      "loss": 0.0761,
+      "step": 8999
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.687996787917393e-06,
+      "loss": 0.0566,
+      "step": 9000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6854126224522522e-06,
+      "loss": 0.0627,
+      "step": 9001
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6828291581284426e-06,
+      "loss": 0.0682,
+      "step": 9002
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6802463952328237e-06,
+      "loss": 0.0592,
+      "step": 9003
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6776643340521713e-06,
+      "loss": 0.0556,
+      "step": 9004
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6750829748731885e-06,
+      "loss": 0.0609,
+      "step": 9005
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6725023179824924e-06,
+      "loss": 0.0465,
+      "step": 9006
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6699223636666316e-06,
+      "loss": 0.0557,
+      "step": 9007
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.667343112212065e-06,
+      "loss": 0.0472,
+      "step": 9008
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.6647645639051842e-06,
+      "loss": 0.0462,
+      "step": 9009
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.662186719032299e-06,
+      "loss": 0.0723,
+      "step": 9010
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6596095778796424e-06,
+      "loss": 0.0616,
+      "step": 9011
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6570331407333635e-06,
+      "loss": 0.0677,
+      "step": 9012
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6544574078795434e-06,
+      "loss": 0.059,
+      "step": 9013
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6518823796041724e-06,
+      "loss": 0.0552,
+      "step": 9014
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6493080561931713e-06,
+      "loss": 0.0588,
+      "step": 9015
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.646734437932381e-06,
+      "loss": 0.0525,
+      "step": 9016
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6441615251075692e-06,
+      "loss": 0.051,
+      "step": 9017
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6415893180044105e-06,
+      "loss": 0.0678,
+      "step": 9018
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.639017816908518e-06,
+      "loss": 0.0734,
+      "step": 9019
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.636447022105407e-06,
+      "loss": 0.0691,
+      "step": 9020
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6338769338805434e-06,
+      "loss": 0.0583,
+      "step": 9021
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6313075525192843e-06,
+      "loss": 0.0696,
+      "step": 9022
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.628738878306929e-06,
+      "loss": 0.0494,
+      "step": 9023
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.626170911528684e-06,
+      "loss": 0.0602,
+      "step": 9024
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6236036524696904e-06,
+      "loss": 0.0678,
+      "step": 9025
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.621037101414995e-06,
+      "loss": 0.065,
+      "step": 9026
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6184712586495873e-06,
+      "loss": 0.0452,
+      "step": 9027
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.615906124458358e-06,
+      "loss": 0.0701,
+      "step": 9028
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.6133416991261316e-06,
+      "loss": 0.0688,
+      "step": 9029
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6107779829376445e-06,
+      "loss": 0.0538,
+      "step": 9030
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6082149761775632e-06,
+      "loss": 0.0513,
+      "step": 9031
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.605652679130469e-06,
+      "loss": 0.0689,
+      "step": 9032
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6030910920808727e-06,
+      "loss": 0.0566,
+      "step": 9033
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.600530215313194e-06,
+      "loss": 0.0694,
+      "step": 9034
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5979700491117853e-06,
+      "loss": 0.0588,
+      "step": 9035
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5954105937609084e-06,
+      "loss": 0.0664,
+      "step": 9036
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5928518495447583e-06,
+      "loss": 0.0571,
+      "step": 9037
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.590293816747448e-06,
+      "loss": 0.0679,
+      "step": 9038
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5877364956530013e-06,
+      "loss": 0.0698,
+      "step": 9039
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.585179886545379e-06,
+      "loss": 0.066,
+      "step": 9040
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.582623989708448e-06,
+      "loss": 0.0524,
+      "step": 9041
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5800688054260047e-06,
+      "loss": 0.0532,
+      "step": 9042
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5775143339817652e-06,
+      "loss": 0.0558,
+      "step": 9043
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5749605756593697e-06,
+      "loss": 0.0495,
+      "step": 9044
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5724075307423667e-06,
+      "loss": 0.0514,
+      "step": 9045
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5698551995142427e-06,
+      "loss": 0.0732,
+      "step": 9046
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.567303582258389e-06,
+      "loss": 0.0538,
+      "step": 9047
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.564752679258128e-06,
+      "loss": 0.0716,
+      "step": 9048
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.562202490796699e-06,
+      "loss": 0.0751,
+      "step": 9049
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5596530171572675e-06,
+      "loss": 0.057,
+      "step": 9050
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5571042586229053e-06,
+      "loss": 0.0651,
+      "step": 9051
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5545562154766245e-06,
+      "loss": 0.0621,
+      "step": 9052
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5520088880013393e-06,
+      "loss": 0.054,
+      "step": 9053
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5494622764798946e-06,
+      "loss": 0.069,
+      "step": 9054
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.546916381195056e-06,
+      "loss": 0.0599,
+      "step": 9055
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5443712024295095e-06,
+      "loss": 0.0539,
+      "step": 9056
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5418267404658534e-06,
+      "loss": 0.0591,
+      "step": 9057
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5392829955866194e-06,
+      "loss": 0.0604,
+      "step": 9058
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5367399680742455e-06,
+      "loss": 0.06,
+      "step": 9059
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5341976582111003e-06,
+      "loss": 0.046,
+      "step": 9060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5316560662794697e-06,
+      "loss": 0.0547,
+      "step": 9061
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.529115192561563e-06,
+      "loss": 0.0628,
+      "step": 9062
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.526575037339501e-06,
+      "loss": 0.0561,
+      "step": 9063
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5240356008953367e-06,
+      "loss": 0.0696,
+      "step": 9064
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5214968835110284e-06,
+      "loss": 0.0577,
+      "step": 9065
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5189588854684684e-06,
+      "loss": 0.0517,
+      "step": 9066
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.516421607049464e-06,
+      "loss": 0.049,
+      "step": 9067
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5138850485357436e-06,
+      "loss": 0.0645,
+      "step": 9068
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.5113492102089496e-06,
+      "loss": 0.0664,
+      "step": 9069
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.5088140923506543e-06,
+      "loss": 0.0682,
+      "step": 9070
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.5062796952423407e-06,
+      "loss": 0.0743,
+      "step": 9071
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.503746019165416e-06,
+      "loss": 0.0541,
+      "step": 9072
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.501213064401211e-06,
+      "loss": 0.0614,
+      "step": 9073
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.498680831230974e-06,
+      "loss": 0.0768,
+      "step": 9074
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4961493199358653e-06,
+      "loss": 0.0505,
+      "step": 9075
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4936185307969796e-06,
+      "loss": 0.0579,
+      "step": 9076
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4910884640953115e-06,
+      "loss": 0.0669,
+      "step": 9077
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.488559120111803e-06,
+      "loss": 0.05,
+      "step": 9078
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4860304991272887e-06,
+      "loss": 0.0542,
+      "step": 9079
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4835026014225413e-06,
+      "loss": 0.0703,
+      "step": 9080
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4809754272782403e-06,
+      "loss": 0.0759,
+      "step": 9081
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4784489769749953e-06,
+      "loss": 0.0531,
+      "step": 9082
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4759232507933284e-06,
+      "loss": 0.0617,
+      "step": 9083
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4733982490136884e-06,
+      "loss": 0.0522,
+      "step": 9084
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4708739719164352e-06,
+      "loss": 0.058,
+      "step": 9085
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.468350419781855e-06,
+      "loss": 0.057,
+      "step": 9086
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4658275928901474e-06,
+      "loss": 0.0642,
+      "step": 9087
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.463305491521437e-06,
+      "loss": 0.0574,
+      "step": 9088
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4607841159557653e-06,
+      "loss": 0.0674,
+      "step": 9089
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4582634664730974e-06,
+      "loss": 0.06,
+      "step": 9090
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.455743543353307e-06,
+      "loss": 0.0526,
+      "step": 9091
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.453224346876203e-06,
+      "loss": 0.0811,
+      "step": 9092
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.450705877321495e-06,
+      "loss": 0.0515,
+      "step": 9093
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.448188134968827e-06,
+      "loss": 0.073,
+      "step": 9094
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.445671120097758e-06,
+      "loss": 0.0563,
+      "step": 9095
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.443154832987765e-06,
+      "loss": 0.0588,
+      "step": 9096
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4406392739182402e-06,
+      "loss": 0.075,
+      "step": 9097
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4381244431685066e-06,
+      "loss": 0.0613,
+      "step": 9098
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4356103410177897e-06,
+      "loss": 0.0596,
+      "step": 9099
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4330969677452496e-06,
+      "loss": 0.0556,
+      "step": 9100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4305843236299564e-06,
+      "loss": 0.0602,
+      "step": 9101
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4280724089509064e-06,
+      "loss": 0.0512,
+      "step": 9102
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4255612239870038e-06,
+      "loss": 0.0529,
+      "step": 9103
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4230507690170854e-06,
+      "loss": 0.0695,
+      "step": 9104
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4205410443198938e-06,
+      "loss": 0.0789,
+      "step": 9105
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.418032050174098e-06,
+      "loss": 0.0778,
+      "step": 9106
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.415523786858291e-06,
+      "loss": 0.0561,
+      "step": 9107
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4130162546509695e-06,
+      "loss": 0.0656,
+      "step": 9108
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4105094538305638e-06,
+      "loss": 0.0707,
+      "step": 9109
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4080033846754124e-06,
+      "loss": 0.066,
+      "step": 9110
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.405498047463779e-06,
+      "loss": 0.0684,
+      "step": 9111
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4029934424738455e-06,
+      "loss": 0.0648,
+      "step": 9112
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4004895699837128e-06,
+      "loss": 0.0579,
+      "step": 9113
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3979864302713928e-06,
+      "loss": 0.0568,
+      "step": 9114
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.395484023614829e-06,
+      "loss": 0.0555,
+      "step": 9115
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3929823502918703e-06,
+      "loss": 0.0651,
+      "step": 9116
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.390481410580293e-06,
+      "loss": 0.0588,
+      "step": 9117
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.38798120475779e-06,
+      "loss": 0.056,
+      "step": 9118
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3854817331019753e-06,
+      "loss": 0.0653,
+      "step": 9119
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3829829958903704e-06,
+      "loss": 0.061,
+      "step": 9120
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.380484993400431e-06,
+      "loss": 0.0554,
+      "step": 9121
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3779877259095172e-06,
+      "loss": 0.0493,
+      "step": 9122
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3754911936949154e-06,
+      "loss": 0.0512,
+      "step": 9123
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3729953970338282e-06,
+      "loss": 0.0469,
+      "step": 9124
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.3705003362033816e-06,
+      "loss": 0.0596,
+      "step": 9125
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.368006011480608e-06,
+      "loss": 0.0499,
+      "step": 9126
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.365512423142472e-06,
+      "loss": 0.0583,
+      "step": 9127
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3630195714658374e-06,
+      "loss": 0.0574,
+      "step": 9128
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.360527456727516e-06,
+      "loss": 0.068,
+      "step": 9129
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3580360792042065e-06,
+      "loss": 0.067,
+      "step": 9130
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3555454391725475e-06,
+      "loss": 0.0536,
+      "step": 9131
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3530555369090813e-06,
+      "loss": 0.0675,
+      "step": 9132
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.350566372690278e-06,
+      "loss": 0.0561,
+      "step": 9133
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3480779467925217e-06,
+      "loss": 0.0545,
+      "step": 9134
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3455902594921186e-06,
+      "loss": 0.0685,
+      "step": 9135
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3431033110652834e-06,
+      "loss": 0.0657,
+      "step": 9136
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3406171017881607e-06,
+      "loss": 0.0605,
+      "step": 9137
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.338131631936802e-06,
+      "loss": 0.08,
+      "step": 9138
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.335646901787184e-06,
+      "loss": 0.0644,
+      "step": 9139
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3331629116151987e-06,
+      "loss": 0.0613,
+      "step": 9140
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.330679661696661e-06,
+      "loss": 0.0769,
+      "step": 9141
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.328197152307292e-06,
+      "loss": 0.0723,
+      "step": 9142
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.325715383722744e-06,
+      "loss": 0.0525,
+      "step": 9143
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3232343562185742e-06,
+      "loss": 0.0714,
+      "step": 9144
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3207540700702665e-06,
+      "loss": 0.0735,
+      "step": 9145
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.3182745255532212e-06,
+      "loss": 0.055,
+      "step": 9146
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.315795722942756e-06,
+      "loss": 0.0579,
+      "step": 9147
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.3133176625141016e-06,
+      "loss": 0.0736,
+      "step": 9148
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.310840344542414e-06,
+      "loss": 0.0493,
+      "step": 9149
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.308363769302758e-06,
+      "loss": 0.0597,
+      "step": 9150
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.3058879370701226e-06,
+      "loss": 0.0623,
+      "step": 9151
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.3034128481194126e-06,
+      "loss": 0.0691,
+      "step": 9152
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.300938502725454e-06,
+      "loss": 0.0564,
+      "step": 9153
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2984649011629775e-06,
+      "loss": 0.0643,
+      "step": 9154
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2959920437066486e-06,
+      "loss": 0.0443,
+      "step": 9155
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2935199306310327e-06,
+      "loss": 0.0541,
+      "step": 9156
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.291048562210628e-06,
+      "loss": 0.0726,
+      "step": 9157
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2885779387198403e-06,
+      "loss": 0.0569,
+      "step": 9158
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2861080604329996e-06,
+      "loss": 0.0728,
+      "step": 9159
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2836389276243442e-06,
+      "loss": 0.0566,
+      "step": 9160
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2811705405680392e-06,
+      "loss": 0.0593,
+      "step": 9161
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2787028995381574e-06,
+      "loss": 0.0643,
+      "step": 9162
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.276236004808696e-06,
+      "loss": 0.0756,
+      "step": 9163
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.273769856653568e-06,
+      "loss": 0.0726,
+      "step": 9164
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.2713044553466055e-06,
+      "loss": 0.0598,
+      "step": 9165
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2688398011615485e-06,
+      "loss": 0.0546,
+      "step": 9166
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2663758943720658e-06,
+      "loss": 0.0607,
+      "step": 9167
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.263912735251733e-06,
+      "loss": 0.0723,
+      "step": 9168
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.261450324074048e-06,
+      "loss": 0.0782,
+      "step": 9169
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.258988661112428e-06,
+      "loss": 0.061,
+      "step": 9170
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.256527746640206e-06,
+      "loss": 0.061,
+      "step": 9171
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2540675809306233e-06,
+      "loss": 0.0545,
+      "step": 9172
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2516081642568508e-06,
+      "loss": 0.0533,
+      "step": 9173
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.249149496891966e-06,
+      "loss": 0.0561,
+      "step": 9174
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2466915791089693e-06,
+      "loss": 0.0519,
+      "step": 9175
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.244234411180779e-06,
+      "loss": 0.0609,
+      "step": 9176
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2417779933802205e-06,
+      "loss": 0.0717,
+      "step": 9177
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.239322325980051e-06,
+      "loss": 0.076,
+      "step": 9178
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.236867409252924e-06,
+      "loss": 0.0548,
+      "step": 9179
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.234413243471436e-06,
+      "loss": 0.0614,
+      "step": 9180
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2319598289080758e-06,
+      "loss": 0.0735,
+      "step": 9181
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.229507165835264e-06,
+      "loss": 0.0527,
+      "step": 9182
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2270552545253265e-06,
+      "loss": 0.0656,
+      "step": 9183
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.22460409525052e-06,
+      "loss": 0.073,
+      "step": 9184
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.2221536882829975e-06,
+      "loss": 0.0646,
+      "step": 9185
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.219704033894854e-06,
+      "loss": 0.0715,
+      "step": 9186
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.217255132358078e-06,
+      "loss": 0.0751,
+      "step": 9187
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.2148069839445893e-06,
+      "loss": 0.0581,
+      "step": 9188
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.2123595889262126e-06,
+      "loss": 0.0635,
+      "step": 9189
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.209912947574699e-06,
+      "loss": 0.0533,
+      "step": 9190
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.207467060161711e-06,
+      "loss": 0.0585,
+      "step": 9191
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.2050219269588302e-06,
+      "loss": 0.0565,
+      "step": 9192
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.2025775482375478e-06,
+      "loss": 0.0474,
+      "step": 9193
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.200133924269281e-06,
+      "loss": 0.0699,
+      "step": 9194
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1976910553253514e-06,
+      "loss": 0.0729,
+      "step": 9195
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1952489416770083e-06,
+      "loss": 0.0516,
+      "step": 9196
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.192807583595411e-06,
+      "loss": 0.0553,
+      "step": 9197
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1903669813516393e-06,
+      "loss": 0.0576,
+      "step": 9198
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1879271352166785e-06,
+      "loss": 0.0751,
+      "step": 9199
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1854880454614466e-06,
+      "loss": 0.0677,
+      "step": 9200
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1830497123567594e-06,
+      "loss": 0.0632,
+      "step": 9201
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1806121361733623e-06,
+      "loss": 0.0598,
+      "step": 9202
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.178175317181912e-06,
+      "loss": 0.0703,
+      "step": 9203
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.1757392556529833e-06,
+      "loss": 0.0522,
+      "step": 9204
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1733039518570574e-06,
+      "loss": 0.0593,
+      "step": 9205
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1708694060645483e-06,
+      "loss": 0.0577,
+      "step": 9206
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1684356185457677e-06,
+      "loss": 0.0521,
+      "step": 9207
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1660025895709547e-06,
+      "loss": 0.0534,
+      "step": 9208
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1635703194102618e-06,
+      "loss": 0.0687,
+      "step": 9209
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.161138808333759e-06,
+      "loss": 0.0542,
+      "step": 9210
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1587080566114247e-06,
+      "loss": 0.0595,
+      "step": 9211
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.156278064513163e-06,
+      "loss": 0.0647,
+      "step": 9212
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.153848832308781e-06,
+      "loss": 0.0608,
+      "step": 9213
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.151420360268015e-06,
+      "loss": 0.0603,
+      "step": 9214
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1489926486605094e-06,
+      "loss": 0.0724,
+      "step": 9215
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.146565697755828e-06,
+      "loss": 0.0579,
+      "step": 9216
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1441395078234418e-06,
+      "loss": 0.0728,
+      "step": 9217
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1417140791327507e-06,
+      "loss": 0.0673,
+      "step": 9218
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1392894119530548e-06,
+      "loss": 0.0562,
+      "step": 9219
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1368655065535814e-06,
+      "loss": 0.0554,
+      "step": 9220
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1344423632034692e-06,
+      "loss": 0.0585,
+      "step": 9221
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1320199821717746e-06,
+      "loss": 0.05,
+      "step": 9222
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.1295983637274617e-06,
+      "loss": 0.0688,
+      "step": 9223
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.127177508139423e-06,
+      "loss": 0.0656,
+      "step": 9224
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.124757415676446e-06,
+      "loss": 0.0756,
+      "step": 9225
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1223380866072617e-06,
+      "loss": 0.0651,
+      "step": 9226
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1199195212004884e-06,
+      "loss": 0.0699,
+      "step": 9227
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1175017197246814e-06,
+      "loss": 0.062,
+      "step": 9228
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1150846824482928e-06,
+      "loss": 0.0494,
+      "step": 9229
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1126684096397076e-06,
+      "loss": 0.0521,
+      "step": 9230
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1102529015672044e-06,
+      "loss": 0.0498,
+      "step": 9231
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1078381584990047e-06,
+      "loss": 0.0714,
+      "step": 9232
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1054241807032195e-06,
+      "loss": 0.0681,
+      "step": 9233
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1030109684478917e-06,
+      "loss": 0.0658,
+      "step": 9234
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1005985220009673e-06,
+      "loss": 0.0684,
+      "step": 9235
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.098186841630314e-06,
+      "loss": 0.0669,
+      "step": 9236
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.0957759276037148e-06,
+      "loss": 0.0616,
+      "step": 9237
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.0933657801888683e-06,
+      "loss": 0.0509,
+      "step": 9238
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.0909563996533797e-06,
+      "loss": 0.0651,
+      "step": 9239
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.0885477862647805e-06,
+      "loss": 0.0759,
+      "step": 9240
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.086139940290507e-06,
+      "loss": 0.0433,
+      "step": 9241
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.083732861997917e-06,
+      "loss": 0.0656,
+      "step": 9242
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.0813265516542825e-06,
+      "loss": 0.0513,
+      "step": 9243
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.078921009526785e-06,
+      "loss": 0.0587,
+      "step": 9244
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0765162358825285e-06,
+      "loss": 0.059,
+      "step": 9245
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0741122309885218e-06,
+      "loss": 0.0593,
+      "step": 9246
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0717089951116985e-06,
+      "loss": 0.066,
+      "step": 9247
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0693065285188995e-06,
+      "loss": 0.0562,
+      "step": 9248
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.066904831476889e-06,
+      "loss": 0.0526,
+      "step": 9249
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0645039042523318e-06,
+      "loss": 0.0557,
+      "step": 9250
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.062103747111823e-06,
+      "loss": 0.0502,
+      "step": 9251
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0597043603218567e-06,
+      "loss": 0.0492,
+      "step": 9252
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.057305744148854e-06,
+      "loss": 0.0493,
+      "step": 9253
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.054907898859144e-06,
+      "loss": 0.0629,
+      "step": 9254
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0525108247189763e-06,
+      "loss": 0.0588,
+      "step": 9255
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.050114521994503e-06,
+      "loss": 0.0689,
+      "step": 9256
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0477189909518047e-06,
+      "loss": 0.0612,
+      "step": 9257
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0453242318568643e-06,
+      "loss": 0.0633,
+      "step": 9258
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0429302449755873e-06,
+      "loss": 0.0558,
+      "step": 9259
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.040537030573788e-06,
+      "loss": 0.0673,
+      "step": 9260
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0381445889172047e-06,
+      "loss": 0.0494,
+      "step": 9261
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.035752920271472e-06,
+      "loss": 0.0629,
+      "step": 9262
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.033362024902159e-06,
+      "loss": 0.0535,
+      "step": 9263
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.030971903074731e-06,
+      "loss": 0.0709,
+      "step": 9264
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.028582555054579e-06,
+      "loss": 0.0667,
+      "step": 9265
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.026193981107004e-06,
+      "loss": 0.0673,
+      "step": 9266
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0238061814972274e-06,
+      "loss": 0.0564,
+      "step": 9267
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0214191564903694e-06,
+      "loss": 0.0588,
+      "step": 9268
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0190329063514835e-06,
+      "loss": 0.0474,
+      "step": 9269
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0166474313455186e-06,
+      "loss": 0.0688,
+      "step": 9270
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0142627317373495e-06,
+      "loss": 0.059,
+      "step": 9271
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0118788077917626e-06,
+      "loss": 0.0554,
+      "step": 9272
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0094956597734603e-06,
+      "loss": 0.0477,
+      "step": 9273
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0071132879470497e-06,
+      "loss": 0.0703,
+      "step": 9274
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.004731692577064e-06,
+      "loss": 0.0743,
+      "step": 9275
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.0023508739279384e-06,
+      "loss": 0.0756,
+      "step": 9276
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9999708322640296e-06,
+      "loss": 0.0662,
+      "step": 9277
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.997591567849607e-06,
+      "loss": 0.0585,
+      "step": 9278
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9952130809488545e-06,
+      "loss": 0.0649,
+      "step": 9279
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.992835371825862e-06,
+      "loss": 0.0638,
+      "step": 9280
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9904584407446457e-06,
+      "loss": 0.0646,
+      "step": 9281
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9880822879691184e-06,
+      "loss": 0.0595,
+      "step": 9282
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.985706913763131e-06,
+      "loss": 0.0541,
+      "step": 9283
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9833323183904216e-06,
+      "loss": 0.0546,
+      "step": 9284
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9809585021146615e-06,
+      "loss": 0.0717,
+      "step": 9285
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.978585465199422e-06,
+      "loss": 0.0543,
+      "step": 9286
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9762132079081952e-06,
+      "loss": 0.0474,
+      "step": 9287
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9738417305043867e-06,
+      "loss": 0.0487,
+      "step": 9288
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9714710332513165e-06,
+      "loss": 0.0701,
+      "step": 9289
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9691011164122085e-06,
+      "loss": 0.0704,
+      "step": 9290
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.966731980250214e-06,
+      "loss": 0.0676,
+      "step": 9291
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9643636250283837e-06,
+      "loss": 0.0528,
+      "step": 9292
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9619960510096925e-06,
+      "loss": 0.0612,
+      "step": 9293
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9596292584570218e-06,
+      "loss": 0.0579,
+      "step": 9294
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.957263247633175e-06,
+      "loss": 0.0554,
+      "step": 9295
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9548980188008545e-06,
+      "loss": 0.0468,
+      "step": 9296
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9525335722226912e-06,
+      "loss": 0.0483,
+      "step": 9297
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9501699081612143e-06,
+      "loss": 0.0401,
+      "step": 9298
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9478070268788774e-06,
+      "loss": 0.0505,
+      "step": 9299
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.945444928638044e-06,
+      "loss": 0.0626,
+      "step": 9300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.9430836137009934e-06,
+      "loss": 0.0759,
+      "step": 9301
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.940723082329906e-06,
+      "loss": 0.0695,
+      "step": 9302
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9383633347868935e-06,
+      "loss": 0.0677,
+      "step": 9303
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9360043713339616e-06,
+      "loss": 0.0587,
+      "step": 9304
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9336461922330438e-06,
+      "loss": 0.0718,
+      "step": 9305
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.931288797745979e-06,
+      "loss": 0.0625,
+      "step": 9306
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.0631,
+      "step": 9307
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.926576363660346e-06,
+      "loss": 0.0731,
+      "step": 9308
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.924221324585017e-06,
+      "loss": 0.0471,
+      "step": 9309
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.921867071170034e-06,
+      "loss": 0.0677,
+      "step": 9310
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9195136036768e-06,
+      "loss": 0.0592,
+      "step": 9311
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9171609223666396e-06,
+      "loss": 0.0666,
+      "step": 9312
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9148090275007733e-06,
+      "loss": 0.0492,
+      "step": 9313
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9124579193403522e-06,
+      "loss": 0.0532,
+      "step": 9314
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9101075981464267e-06,
+      "loss": 0.0709,
+      "step": 9315
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.907758064179965e-06,
+      "loss": 0.0511,
+      "step": 9316
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9054093177018515e-06,
+      "loss": 0.0556,
+      "step": 9317
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9030613589728805e-06,
+      "loss": 0.0558,
+      "step": 9318
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.9007141882537535e-06,
+      "loss": 0.0591,
+      "step": 9319
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.8983678058050934e-06,
+      "loss": 0.0707,
+      "step": 9320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.896022211887427e-06,
+      "loss": 0.0525,
+      "step": 9321
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8936774067611996e-06,
+      "loss": 0.0684,
+      "step": 9322
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.891333390686767e-06,
+      "loss": 0.0658,
+      "step": 9323
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.888990163924401e-06,
+      "loss": 0.0696,
+      "step": 9324
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8866477267342752e-06,
+      "loss": 0.0514,
+      "step": 9325
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.88430607937649e-06,
+      "loss": 0.0551,
+      "step": 9326
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.881965222111043e-06,
+      "loss": 0.0602,
+      "step": 9327
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.879625155197856e-06,
+      "loss": 0.0652,
+      "step": 9328
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.877285878896758e-06,
+      "loss": 0.0612,
+      "step": 9329
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8749473934674953e-06,
+      "loss": 0.0769,
+      "step": 9330
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.872609699169714e-06,
+      "loss": 0.0739,
+      "step": 9331
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8702727962629883e-06,
+      "loss": 0.0577,
+      "step": 9332
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8679366850067857e-06,
+      "loss": 0.0634,
+      "step": 9333
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8656013656605107e-06,
+      "loss": 0.0676,
+      "step": 9334
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8632668384834563e-06,
+      "loss": 0.0629,
+      "step": 9335
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.860933103734842e-06,
+      "loss": 0.0814,
+      "step": 9336
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.858600161673789e-06,
+      "loss": 0.0505,
+      "step": 9337
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.85626801255934e-06,
+      "loss": 0.0583,
+      "step": 9338
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8539366566504445e-06,
+      "loss": 0.0547,
+      "step": 9339
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.8516060942059677e-06,
+      "loss": 0.0665,
+      "step": 9340
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.849276325484679e-06,
+      "loss": 0.0585,
+      "step": 9341
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8469473507452707e-06,
+      "loss": 0.0501,
+      "step": 9342
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8446191702463343e-06,
+      "loss": 0.0555,
+      "step": 9343
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.842291784246384e-06,
+      "loss": 0.0621,
+      "step": 9344
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8399651930038385e-06,
+      "loss": 0.0517,
+      "step": 9345
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.837639396777038e-06,
+      "loss": 0.054,
+      "step": 9346
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.835314395824219e-06,
+      "loss": 0.0642,
+      "step": 9347
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.832990190403546e-06,
+      "loss": 0.0562,
+      "step": 9348
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8306667807730802e-06,
+      "loss": 0.0565,
+      "step": 9349
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8283441671908064e-06,
+      "loss": 0.067,
+      "step": 9350
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8260223499146154e-06,
+      "loss": 0.066,
+      "step": 9351
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.823701329202313e-06,
+      "loss": 0.078,
+      "step": 9352
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.821381105311609e-06,
+      "loss": 0.0654,
+      "step": 9353
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.819061678500137e-06,
+      "loss": 0.0707,
+      "step": 9354
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8167430490254265e-06,
+      "loss": 0.0643,
+      "step": 9355
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8144252171449316e-06,
+      "loss": 0.0395,
+      "step": 9356
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8121081831160125e-06,
+      "loss": 0.053,
+      "step": 9357
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8097919471959457e-06,
+      "loss": 0.0573,
+      "step": 9358
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8074765096419066e-06,
+      "loss": 0.0704,
+      "step": 9359
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.8051618707109984e-06,
+      "loss": 0.0468,
+      "step": 9360
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.8028480306602203e-06,
+      "loss": 0.0667,
+      "step": 9361
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.800534989746493e-06,
+      "loss": 0.076,
+      "step": 9362
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7982227482266454e-06,
+      "loss": 0.0599,
+      "step": 9363
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7959113063574216e-06,
+      "loss": 0.0583,
+      "step": 9364
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7936006643954662e-06,
+      "loss": 0.0592,
+      "step": 9365
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.791290822597347e-06,
+      "loss": 0.0647,
+      "step": 9366
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.788981781219533e-06,
+      "loss": 0.0744,
+      "step": 9367
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7866735405184118e-06,
+      "loss": 0.0595,
+      "step": 9368
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7843661007502787e-06,
+      "loss": 0.0577,
+      "step": 9369
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.782059462171344e-06,
+      "loss": 0.0634,
+      "step": 9370
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7797536250377187e-06,
+      "loss": 0.0784,
+      "step": 9371
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7774485896054404e-06,
+      "loss": 0.0561,
+      "step": 9372
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7751443561304413e-06,
+      "loss": 0.0557,
+      "step": 9373
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.772840924868575e-06,
+      "loss": 0.0693,
+      "step": 9374
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.770538296075608e-06,
+      "loss": 0.0568,
+      "step": 9375
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7682364700072053e-06,
+      "loss": 0.0568,
+      "step": 9376
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.765935446918957e-06,
+      "loss": 0.0621,
+      "step": 9377
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7636352270663524e-06,
+      "loss": 0.0574,
+      "step": 9378
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.7613358107048007e-06,
+      "loss": 0.0504,
+      "step": 9379
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.759037198089616e-06,
+      "loss": 0.0544,
+      "step": 9380
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7567393894760273e-06,
+      "loss": 0.0686,
+      "step": 9381
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7544423851191705e-06,
+      "loss": 0.0637,
+      "step": 9382
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.752146185274095e-06,
+      "loss": 0.0481,
+      "step": 9383
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7498507901957537e-06,
+      "loss": 0.0533,
+      "step": 9384
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.747556200139028e-06,
+      "loss": 0.0679,
+      "step": 9385
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7452624153586883e-06,
+      "loss": 0.0588,
+      "step": 9386
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7429694361094315e-06,
+      "loss": 0.0629,
+      "step": 9387
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7406772626458535e-06,
+      "loss": 0.0626,
+      "step": 9388
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7383858952224708e-06,
+      "loss": 0.0618,
+      "step": 9389
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.736095334093698e-06,
+      "loss": 0.0687,
+      "step": 9390
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.733805579513881e-06,
+      "loss": 0.0686,
+      "step": 9391
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7315166317372523e-06,
+      "loss": 0.0646,
+      "step": 9392
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7292284910179713e-06,
+      "loss": 0.0673,
+      "step": 9393
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7269411576100977e-06,
+      "loss": 0.0664,
+      "step": 9394
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.724654631767608e-06,
+      "loss": 0.0544,
+      "step": 9395
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7223689137443876e-06,
+      "loss": 0.048,
+      "step": 9396
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7200840037942345e-06,
+      "loss": 0.0746,
+      "step": 9397
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7177999021708477e-06,
+      "loss": 0.0706,
+      "step": 9398
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.7155166091278496e-06,
+      "loss": 0.0637,
+      "step": 9399
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.7132341249187587e-06,
+      "loss": 0.0577,
+      "step": 9400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.710952449797014e-06,
+      "loss": 0.0605,
+      "step": 9401
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.708671584015964e-06,
+      "loss": 0.0449,
+      "step": 9402
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.7063915278288657e-06,
+      "loss": 0.0626,
+      "step": 9403
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.704112281488881e-06,
+      "loss": 0.0611,
+      "step": 9404
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.701833845249091e-06,
+      "loss": 0.0704,
+      "step": 9405
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.699556219362478e-06,
+      "loss": 0.0604,
+      "step": 9406
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6972794040819405e-06,
+      "loss": 0.0524,
+      "step": 9407
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6950033996602844e-06,
+      "loss": 0.0731,
+      "step": 9408
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6927282063502313e-06,
+      "loss": 0.0538,
+      "step": 9409
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6904538244044e-06,
+      "loss": 0.046,
+      "step": 9410
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.688180254075333e-06,
+      "loss": 0.0635,
+      "step": 9411
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6859074956154717e-06,
+      "loss": 0.0615,
+      "step": 9412
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.683635549277174e-06,
+      "loss": 0.0661,
+      "step": 9413
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6813644153127073e-06,
+      "loss": 0.0455,
+      "step": 9414
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6790940939742484e-06,
+      "loss": 0.0656,
+      "step": 9415
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6768245855138784e-06,
+      "loss": 0.0662,
+      "step": 9416
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.674555890183598e-06,
+      "loss": 0.067,
+      "step": 9417
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6722880082353065e-06,
+      "loss": 0.0641,
+      "step": 9418
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.670020939920821e-06,
+      "loss": 0.0634,
+      "step": 9419
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.667754685491867e-06,
+      "loss": 0.0607,
+      "step": 9420
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.665489245200079e-06,
+      "loss": 0.0559,
+      "step": 9421
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.663224619296998e-06,
+      "loss": 0.0634,
+      "step": 9422
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6609608080340797e-06,
+      "loss": 0.0569,
+      "step": 9423
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6586978116626837e-06,
+      "loss": 0.0672,
+      "step": 9424
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6564356304340844e-06,
+      "loss": 0.0736,
+      "step": 9425
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.654174264599462e-06,
+      "loss": 0.053,
+      "step": 9426
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.651913714409912e-06,
+      "loss": 0.0722,
+      "step": 9427
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.649653980116429e-06,
+      "loss": 0.0537,
+      "step": 9428
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6473950619699286e-06,
+      "loss": 0.0688,
+      "step": 9429
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6451369602212206e-06,
+      "loss": 0.0698,
+      "step": 9430
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.642879675121047e-06,
+      "loss": 0.0531,
+      "step": 9431
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6406232069200365e-06,
+      "loss": 0.0508,
+      "step": 9432
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6383675558687417e-06,
+      "loss": 0.0648,
+      "step": 9433
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.636112722217614e-06,
+      "loss": 0.0614,
+      "step": 9434
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.6338587062170253e-06,
+      "loss": 0.0535,
+      "step": 9435
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.631605508117241e-06,
+      "loss": 0.0407,
+      "step": 9436
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.629353128168457e-06,
+      "loss": 0.0677,
+      "step": 9437
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.62710156662076e-06,
+      "loss": 0.0605,
+      "step": 9438
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6248508237241556e-06,
+      "loss": 0.0599,
+      "step": 9439
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6226008997285513e-06,
+      "loss": 0.0758,
+      "step": 9440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.620351794883771e-06,
+      "loss": 0.0508,
+      "step": 9441
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6181035094395456e-06,
+      "loss": 0.0668,
+      "step": 9442
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6158560436455095e-06,
+      "loss": 0.0647,
+      "step": 9443
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6136093977512157e-06,
+      "loss": 0.0541,
+      "step": 9444
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.611363572006116e-06,
+      "loss": 0.0561,
+      "step": 9445
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6091185666595777e-06,
+      "loss": 0.0646,
+      "step": 9446
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6068743819608756e-06,
+      "loss": 0.0549,
+      "step": 9447
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.604631018159197e-06,
+      "loss": 0.0614,
+      "step": 9448
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6023884755036277e-06,
+      "loss": 0.0668,
+      "step": 9449
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.6001467542431746e-06,
+      "loss": 0.046,
+      "step": 9450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.597905854626742e-06,
+      "loss": 0.0809,
+      "step": 9451
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.595665776903152e-06,
+      "loss": 0.0673,
+      "step": 9452
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5934265213211305e-06,
+      "loss": 0.0713,
+      "step": 9453
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5911880881293185e-06,
+      "loss": 0.0571,
+      "step": 9454
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.588950477576253e-06,
+      "loss": 0.0596,
+      "step": 9455
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.586713689910395e-06,
+      "loss": 0.0627,
+      "step": 9456
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5844777253801e-06,
+      "loss": 0.0751,
+      "step": 9457
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5822425842336417e-06,
+      "loss": 0.0558,
+      "step": 9458
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5800082667192004e-06,
+      "loss": 0.053,
+      "step": 9459
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5777747730848646e-06,
+      "loss": 0.0637,
+      "step": 9460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5755421035786265e-06,
+      "loss": 0.0538,
+      "step": 9461
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.573310258448397e-06,
+      "loss": 0.0477,
+      "step": 9462
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5710792379419826e-06,
+      "loss": 0.054,
+      "step": 9463
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5688490423071076e-06,
+      "loss": 0.0688,
+      "step": 9464
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5666196717914036e-06,
+      "loss": 0.0472,
+      "step": 9465
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.564391126642413e-06,
+      "loss": 0.0648,
+      "step": 9466
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.562163407107573e-06,
+      "loss": 0.0563,
+      "step": 9467
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.559936513434248e-06,
+      "loss": 0.0765,
+      "step": 9468
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5577104458696942e-06,
+      "loss": 0.05,
+      "step": 9469
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.555485204661087e-06,
+      "loss": 0.0632,
+      "step": 9470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.553260790055506e-06,
+      "loss": 0.0518,
+      "step": 9471
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.551037202299944e-06,
+      "loss": 0.0652,
+      "step": 9472
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5488144416412885e-06,
+      "loss": 0.0537,
+      "step": 9473
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.546592508326353e-06,
+      "loss": 0.0712,
+      "step": 9474
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5443714026018427e-06,
+      "loss": 0.0684,
+      "step": 9475
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5421511247143826e-06,
+      "loss": 0.0623,
+      "step": 9476
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5399316749105e-06,
+      "loss": 0.0593,
+      "step": 9477
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.537713053436636e-06,
+      "loss": 0.0529,
+      "step": 9478
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.535495260539129e-06,
+      "loss": 0.0559,
+      "step": 9479
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.533278296464239e-06,
+      "loss": 0.0717,
+      "step": 9480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.53106216145812e-06,
+      "loss": 0.0501,
+      "step": 9481
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5288468557668457e-06,
+      "loss": 0.051,
+      "step": 9482
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.52663237963639e-06,
+      "loss": 0.0617,
+      "step": 9483
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5244187333126437e-06,
+      "loss": 0.0549,
+      "step": 9484
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5222059170413915e-06,
+      "loss": 0.0544,
+      "step": 9485
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5199939310683396e-06,
+      "loss": 0.0754,
+      "step": 9486
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5177827756390895e-06,
+      "loss": 0.0717,
+      "step": 9487
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5155724509991663e-06,
+      "loss": 0.0676,
+      "step": 9488
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5133629573939878e-06,
+      "loss": 0.0519,
+      "step": 9489
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5111542950688893e-06,
+      "loss": 0.0654,
+      "step": 9490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5089464642691043e-06,
+      "loss": 0.0586,
+      "step": 9491
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5067394652397836e-06,
+      "loss": 0.0549,
+      "step": 9492
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5045332982259805e-06,
+      "loss": 0.06,
+      "step": 9493
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.502327963472662e-06,
+      "loss": 0.0706,
+      "step": 9494
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5001234612246893e-06,
+      "loss": 0.0608,
+      "step": 9495
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.4979197917268484e-06,
+      "loss": 0.0627,
+      "step": 9496
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.495716955223817e-06,
+      "loss": 0.0733,
+      "step": 9497
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.49351495196019e-06,
+      "loss": 0.0474,
+      "step": 9498
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4913137821804667e-06,
+      "loss": 0.0516,
+      "step": 9499
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.489113446129059e-06,
+      "loss": 0.0674,
+      "step": 9500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4869139440502744e-06,
+      "loss": 0.0522,
+      "step": 9501
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4847152761883408e-06,
+      "loss": 0.0533,
+      "step": 9502
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.482517442787383e-06,
+      "loss": 0.0584,
+      "step": 9503
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4803204440914396e-06,
+      "loss": 0.0569,
+      "step": 9504
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.478124280344456e-06,
+      "loss": 0.0499,
+      "step": 9505
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4759289517902864e-06,
+      "loss": 0.0601,
+      "step": 9506
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4737344586726818e-06,
+      "loss": 0.0401,
+      "step": 9507
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.471540801235316e-06,
+      "loss": 0.0608,
+      "step": 9508
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4693479797217545e-06,
+      "loss": 0.0606,
+      "step": 9509
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4671559943754818e-06,
+      "loss": 0.0502,
+      "step": 9510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4649648454398887e-06,
+      "loss": 0.0737,
+      "step": 9511
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4627745331582632e-06,
+      "loss": 0.064,
+      "step": 9512
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4605850577738134e-06,
+      "loss": 0.0555,
+      "step": 9513
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4583964195296407e-06,
+      "loss": 0.0674,
+      "step": 9514
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.456208618668766e-06,
+      "loss": 0.0668,
+      "step": 9515
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.454021655434111e-06,
+      "loss": 0.054,
+      "step": 9516
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4518355300685083e-06,
+      "loss": 0.0613,
+      "step": 9517
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4496502428146896e-06,
+      "loss": 0.0604,
+      "step": 9518
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4474657939153044e-06,
+      "loss": 0.0508,
+      "step": 9519
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.445282183612897e-06,
+      "loss": 0.0685,
+      "step": 9520
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4430994121499273e-06,
+      "loss": 0.0592,
+      "step": 9521
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.440917479768762e-06,
+      "loss": 0.0643,
+      "step": 9522
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.438736386711673e-06,
+      "loss": 0.0448,
+      "step": 9523
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.436556133220833e-06,
+      "loss": 0.0681,
+      "step": 9524
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4343767195383338e-06,
+      "loss": 0.0564,
+      "step": 9525
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4321981459061594e-06,
+      "loss": 0.0621,
+      "step": 9526
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.430020412566213e-06,
+      "loss": 0.0486,
+      "step": 9527
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.427843519760298e-06,
+      "loss": 0.0594,
+      "step": 9528
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4256674677301294e-06,
+      "loss": 0.0576,
+      "step": 9529
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.42349225671732e-06,
+      "loss": 0.0402,
+      "step": 9530
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.421317886963401e-06,
+      "loss": 0.0615,
+      "step": 9531
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4191443587097975e-06,
+      "loss": 0.0624,
+      "step": 9532
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.416971672197851e-06,
+      "loss": 0.0665,
+      "step": 9533
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.4147998276688046e-06,
+      "loss": 0.0682,
+      "step": 9534
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.412628825363814e-06,
+      "loss": 0.0579,
+      "step": 9535
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4104586655239305e-06,
+      "loss": 0.0809,
+      "step": 9536
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4082893483901237e-06,
+      "loss": 0.0698,
+      "step": 9537
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.406120874203256e-06,
+      "loss": 0.0565,
+      "step": 9538
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4039532432041156e-06,
+      "loss": 0.0659,
+      "step": 9539
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.401786455633377e-06,
+      "loss": 0.0557,
+      "step": 9540
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3996205117316364e-06,
+      "loss": 0.0722,
+      "step": 9541
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3974554117393823e-06,
+      "loss": 0.0573,
+      "step": 9542
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3952911558970214e-06,
+      "loss": 0.0496,
+      "step": 9543
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.393127744444862e-06,
+      "loss": 0.0501,
+      "step": 9544
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.39096517762312e-06,
+      "loss": 0.0642,
+      "step": 9545
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3888034556719142e-06,
+      "loss": 0.0636,
+      "step": 9546
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.386642578831274e-06,
+      "loss": 0.0732,
+      "step": 9547
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3844825473411282e-06,
+      "loss": 0.0565,
+      "step": 9548
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.38232336144132e-06,
+      "loss": 0.0662,
+      "step": 9549
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3801650213715944e-06,
+      "loss": 0.0563,
+      "step": 9550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3780075273716063e-06,
+      "loss": 0.0707,
+      "step": 9551
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3758508796809067e-06,
+      "loss": 0.0698,
+      "step": 9552
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3736950785389658e-06,
+      "loss": 0.0637,
+      "step": 9553
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.371540124185148e-06,
+      "loss": 0.0549,
+      "step": 9554
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.369386016858731e-06,
+      "loss": 0.0569,
+      "step": 9555
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3672327567988963e-06,
+      "loss": 0.0544,
+      "step": 9556
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3650803442447356e-06,
+      "loss": 0.0555,
+      "step": 9557
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.362928779435236e-06,
+      "loss": 0.0689,
+      "step": 9558
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3607780626093036e-06,
+      "loss": 0.0491,
+      "step": 9559
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3586281940057367e-06,
+      "loss": 0.0503,
+      "step": 9560
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3564791738632485e-06,
+      "loss": 0.0619,
+      "step": 9561
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.354331002420458e-06,
+      "loss": 0.0484,
+      "step": 9562
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.352183679915888e-06,
+      "loss": 0.0583,
+      "step": 9563
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.350037206587964e-06,
+      "loss": 0.0535,
+      "step": 9564
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3478915826750236e-06,
+      "loss": 0.0693,
+      "step": 9565
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.345746808415302e-06,
+      "loss": 0.0694,
+      "step": 9566
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.343602884046947e-06,
+      "loss": 0.0663,
+      "step": 9567
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.34145980980801e-06,
+      "loss": 0.068,
+      "step": 9568
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3393175859364503e-06,
+      "loss": 0.0599,
+      "step": 9569
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3371762126701237e-06,
+      "loss": 0.0488,
+      "step": 9570
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.335035690246804e-06,
+      "loss": 0.0514,
+      "step": 9571
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3328960189041593e-06,
+      "loss": 0.0732,
+      "step": 9572
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3307571988797705e-06,
+      "loss": 0.0448,
+      "step": 9573
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.3286192304111232e-06,
+      "loss": 0.0729,
+      "step": 9574
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3264821137356086e-06,
+      "loss": 0.0613,
+      "step": 9575
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.324345849090517e-06,
+      "loss": 0.0565,
+      "step": 9576
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3222104367130536e-06,
+      "loss": 0.0717,
+      "step": 9577
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3200758768403196e-06,
+      "loss": 0.0608,
+      "step": 9578
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3179421697093285e-06,
+      "loss": 0.0558,
+      "step": 9579
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3158093155570003e-06,
+      "loss": 0.0841,
+      "step": 9580
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3136773146201506e-06,
+      "loss": 0.0531,
+      "step": 9581
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3115461671355122e-06,
+      "loss": 0.0536,
+      "step": 9582
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.309415873339712e-06,
+      "loss": 0.053,
+      "step": 9583
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3072864334692903e-06,
+      "loss": 0.049,
+      "step": 9584
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3051578477606907e-06,
+      "loss": 0.0615,
+      "step": 9585
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.303030116450262e-06,
+      "loss": 0.0556,
+      "step": 9586
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3009032397742528e-06,
+      "loss": 0.0504,
+      "step": 9587
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.2987772179688263e-06,
+      "loss": 0.0533,
+      "step": 9588
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.2966520512700385e-06,
+      "loss": 0.048,
+      "step": 9589
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.294527739913868e-06,
+      "loss": 0.0628,
+      "step": 9590
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.2924042841361793e-06,
+      "loss": 0.0504,
+      "step": 9591
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.290281684172757e-06,
+      "loss": 0.0529,
+      "step": 9592
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.288159940259278e-06,
+      "loss": 0.0575,
+      "step": 9593
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.286039052631337e-06,
+      "loss": 0.0662,
+      "step": 9594
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2839190215244177e-06,
+      "loss": 0.0549,
+      "step": 9595
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.281799847173931e-06,
+      "loss": 0.0588,
+      "step": 9596
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.27968152981517e-06,
+      "loss": 0.069,
+      "step": 9597
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.277564069683349e-06,
+      "loss": 0.0721,
+      "step": 9598
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.275447467013574e-06,
+      "loss": 0.0535,
+      "step": 9599
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2733317220408647e-06,
+      "loss": 0.0716,
+      "step": 9600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.271216835000145e-06,
+      "loss": 0.0579,
+      "step": 9601
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2691028061262433e-06,
+      "loss": 0.0663,
+      "step": 9602
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2669896356538856e-06,
+      "loss": 0.0507,
+      "step": 9603
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2648773238177147e-06,
+      "loss": 0.0596,
+      "step": 9604
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.262765870852265e-06,
+      "loss": 0.0661,
+      "step": 9605
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2606552769919855e-06,
+      "loss": 0.0563,
+      "step": 9606
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2585455424712255e-06,
+      "loss": 0.0665,
+      "step": 9607
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2564366675242433e-06,
+      "loss": 0.0647,
+      "step": 9608
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.254328652385193e-06,
+      "loss": 0.0569,
+      "step": 9609
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2522214972881417e-06,
+      "loss": 0.0637,
+      "step": 9610
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2501152024670545e-06,
+      "loss": 0.0607,
+      "step": 9611
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2480097681558077e-06,
+      "loss": 0.0621,
+      "step": 9612
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.245905194588176e-06,
+      "loss": 0.0573,
+      "step": 9613
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.243801481997845e-06,
+      "loss": 0.059,
+      "step": 9614
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2416986306183963e-06,
+      "loss": 0.0477,
+      "step": 9615
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2395966406833237e-06,
+      "loss": 0.074,
+      "step": 9616
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.237495512426019e-06,
+      "loss": 0.0578,
+      "step": 9617
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.235395246079782e-06,
+      "loss": 0.0505,
+      "step": 9618
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2332958418778174e-06,
+      "loss": 0.0684,
+      "step": 9619
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.231197300053235e-06,
+      "loss": 0.0615,
+      "step": 9620
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2290996208390423e-06,
+      "loss": 0.0618,
+      "step": 9621
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.227002804468159e-06,
+      "loss": 0.0537,
+      "step": 9622
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.224906851173403e-06,
+      "loss": 0.0763,
+      "step": 9623
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2228117611874987e-06,
+      "loss": 0.0743,
+      "step": 9624
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2207175347430754e-06,
+      "loss": 0.0612,
+      "step": 9625
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2186241720726698e-06,
+      "loss": 0.0546,
+      "step": 9626
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2165316734087128e-06,
+      "loss": 0.0773,
+      "step": 9627
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2144400389835506e-06,
+      "loss": 0.0665,
+      "step": 9628
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2123492690294237e-06,
+      "loss": 0.0748,
+      "step": 9629
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.210259363778482e-06,
+      "loss": 0.0585,
+      "step": 9630
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2081703234627793e-06,
+      "loss": 0.0578,
+      "step": 9631
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.206082148314276e-06,
+      "loss": 0.0668,
+      "step": 9632
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.2039948385648267e-06,
+      "loss": 0.0689,
+      "step": 9633
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.2019083944462028e-06,
+      "loss": 0.0519,
+      "step": 9634
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1998228161900635e-06,
+      "loss": 0.0551,
+      "step": 9635
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.197738104027992e-06,
+      "loss": 0.0551,
+      "step": 9636
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1956542581914585e-06,
+      "loss": 0.0636,
+      "step": 9637
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.193571278911847e-06,
+      "loss": 0.0677,
+      "step": 9638
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1914891664204362e-06,
+      "loss": 0.0591,
+      "step": 9639
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.18940792094842e-06,
+      "loss": 0.0645,
+      "step": 9640
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1873275427268804e-06,
+      "loss": 0.0596,
+      "step": 9641
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1852480319868253e-06,
+      "loss": 0.063,
+      "step": 9642
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.183169388959144e-06,
+      "loss": 0.0607,
+      "step": 9643
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.181091613874645e-06,
+      "loss": 0.0537,
+      "step": 9644
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.179014706964031e-06,
+      "loss": 0.0608,
+      "step": 9645
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1769386684579064e-06,
+      "loss": 0.0697,
+      "step": 9646
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1748634985867965e-06,
+      "loss": 0.061,
+      "step": 9647
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1727891975811098e-06,
+      "loss": 0.059,
+      "step": 9648
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.170715765671171e-06,
+      "loss": 0.0565,
+      "step": 9649
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1686432030871995e-06,
+      "loss": 0.0601,
+      "step": 9650
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.1665715100593244e-06,
+      "loss": 0.0688,
+      "step": 9651
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1645006868175765e-06,
+      "loss": 0.0535,
+      "step": 9652
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.162430733591895e-06,
+      "loss": 0.0571,
+      "step": 9653
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1603616506121093e-06,
+      "loss": 0.0515,
+      "step": 9654
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1582934381079678e-06,
+      "loss": 0.0504,
+      "step": 9655
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1562260963091086e-06,
+      "loss": 0.0567,
+      "step": 9656
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1541596254450815e-06,
+      "loss": 0.0558,
+      "step": 9657
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.152094025745338e-06,
+      "loss": 0.0695,
+      "step": 9658
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1500292974392357e-06,
+      "loss": 0.07,
+      "step": 9659
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1479654407560256e-06,
+      "loss": 0.0724,
+      "step": 9660
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.145902455924874e-06,
+      "loss": 0.0621,
+      "step": 9661
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.14384034317484e-06,
+      "loss": 0.0742,
+      "step": 9662
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.141779102734893e-06,
+      "loss": 0.0684,
+      "step": 9663
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.139718734833903e-06,
+      "loss": 0.067,
+      "step": 9664
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.137659239700647e-06,
+      "loss": 0.0563,
+      "step": 9665
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1356006175637944e-06,
+      "loss": 0.06,
+      "step": 9666
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1335428686519312e-06,
+      "loss": 0.0645,
+      "step": 9667
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.131485993193534e-06,
+      "loss": 0.064,
+      "step": 9668
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1294299914169905e-06,
+      "loss": 0.0814,
+      "step": 9669
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.127374863550591e-06,
+      "loss": 0.0685,
+      "step": 9670
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1253206098225286e-06,
+      "loss": 0.06,
+      "step": 9671
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1232672304608914e-06,
+      "loss": 0.0472,
+      "step": 9672
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1212147256936845e-06,
+      "loss": 0.0571,
+      "step": 9673
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.119163095748801e-06,
+      "loss": 0.0493,
+      "step": 9674
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1171123408540463e-06,
+      "loss": 0.0541,
+      "step": 9675
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1150624612371273e-06,
+      "loss": 0.0424,
+      "step": 9676
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1130134571256556e-06,
+      "loss": 0.0586,
+      "step": 9677
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1109653287471375e-06,
+      "loss": 0.0615,
+      "step": 9678
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.108918076328992e-06,
+      "loss": 0.0612,
+      "step": 9679
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1068717000985294e-06,
+      "loss": 0.0522,
+      "step": 9680
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1048262002829754e-06,
+      "loss": 0.061,
+      "step": 9681
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1027815771094494e-06,
+      "loss": 0.0446,
+      "step": 9682
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.100737830804982e-06,
+      "loss": 0.0613,
+      "step": 9683
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0986949615964926e-06,
+      "loss": 0.0654,
+      "step": 9684
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0966529697108196e-06,
+      "loss": 0.068,
+      "step": 9685
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.094611855374685e-06,
+      "loss": 0.0557,
+      "step": 9686
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0925716188147384e-06,
+      "loss": 0.0498,
+      "step": 9687
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.090532260257507e-06,
+      "loss": 0.0779,
+      "step": 9688
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0884937799294382e-06,
+      "loss": 0.0682,
+      "step": 9689
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0864561780568693e-06,
+      "loss": 0.0665,
+      "step": 9690
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.08441945486605e-06,
+      "loss": 0.0681,
+      "step": 9691
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.082383610583122e-06,
+      "loss": 0.0715,
+      "step": 9692
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.080348645434146e-06,
+      "loss": 0.0598,
+      "step": 9693
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.078314559645066e-06,
+      "loss": 0.0741,
+      "step": 9694
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0762813534417424e-06,
+      "loss": 0.0649,
+      "step": 9695
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0742490270499284e-06,
+      "loss": 0.0651,
+      "step": 9696
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.072217580695285e-06,
+      "loss": 0.0572,
+      "step": 9697
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0701870146033744e-06,
+      "loss": 0.0595,
+      "step": 9698
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0681573289996646e-06,
+      "loss": 0.0513,
+      "step": 9699
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.066128524109515e-06,
+      "loss": 0.0609,
+      "step": 9700
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.064100600158202e-06,
+      "loss": 0.0536,
+      "step": 9701
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0620735573708893e-06,
+      "loss": 0.0561,
+      "step": 9702
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.060047395972653e-06,
+      "loss": 0.0528,
+      "step": 9703
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0580221161884693e-06,
+      "loss": 0.0629,
+      "step": 9704
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.055997718243217e-06,
+      "loss": 0.0696,
+      "step": 9705
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0539742023616703e-06,
+      "loss": 0.066,
+      "step": 9706
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0519515687685165e-06,
+      "loss": 0.0567,
+      "step": 9707
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.049929817688333e-06,
+      "loss": 0.0746,
+      "step": 9708
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.047908949345608e-06,
+      "loss": 0.055,
+      "step": 9709
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0458889639647308e-06,
+      "loss": 0.0603,
+      "step": 9710
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0438698617699914e-06,
+      "loss": 0.0546,
+      "step": 9711
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0418516429855796e-06,
+      "loss": 0.0519,
+      "step": 9712
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0398343078355853e-06,
+      "loss": 0.0711,
+      "step": 9713
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0378178565440067e-06,
+      "loss": 0.0534,
+      "step": 9714
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0358022893347396e-06,
+      "loss": 0.0578,
+      "step": 9715
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0337876064315888e-06,
+      "loss": 0.0677,
+      "step": 9716
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0317738080582463e-06,
+      "loss": 0.0654,
+      "step": 9717
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0297608944383208e-06,
+      "loss": 0.0518,
+      "step": 9718
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0277488657953125e-06,
+      "loss": 0.0606,
+      "step": 9719
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0257377223526285e-06,
+      "loss": 0.0506,
+      "step": 9720
+    },
+    {
+      "epoch": 5.0,
+      "step": 9720,
+      "total_flos": 1.5587807120689463e+19,
+      "train_loss": 0.06375732421875,
+      "train_runtime": 32855.1146,
+      "train_samples_per_second": 18.929,
+      "train_steps_per_second": 0.296
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 9720,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 1.5587807120689463e+19,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/README.md b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dcce45792f17c6f4a217b759836daa4a81605ce7
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/README.md
@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.5.0
diff --git a/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/adapter_config.json b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b74a32cb782f414c5b24805dce59d60db770a4c9
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "vicuna-v1-3-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "k_proj",
+    "gate_proj",
+    "q_proj",
+    "o_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/adapter_model.bin b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..f3c6b6483d4e347842a2ce618658b312967133c3
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd133c19a6a2a4b6b5b403f19c54a54146895e78912b17bea72704409c68fafa
+size 319970957
diff --git a/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/config.json b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8a11c13945ca6691233666c289a41f105fe5499a
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/config.json
@@ -0,0 +1,36 @@
+{
+  "_name_or_path": "vicuna-v1-3-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "mm_graph_tower": "hvqvae2",
+  "mm_hidden_size": 308,
+  "mm_projector_type": "hlinear",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "model_type": "llava_graph",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.33.2",
+  "tune_mm_mlp_adapter": false,
+  "use_cache": true,
+  "use_lap_pe": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}
diff --git a/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/non_lora_trainables.bin b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..11ca60789a1248bb82c65b46e31fdc65283fb4ac
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd9fee870b58c4519fcfef92bc257da4d3fbce206c3fd051151c930347e4fb36
+size 11335231
diff --git a/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/trainer_state.json b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..b90ef576a12953fbf33aa0f70477da064516f60e
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/property_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep/trainer_state.json
@@ -0,0 +1,168838 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 28135,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.366863905325444e-08,
+      "loss": 4.125,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733727810650888e-08,
+      "loss": 4.1523,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.100591715976332e-08,
+      "loss": 4.1406,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.467455621301776e-08,
+      "loss": 4.1172,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.183431952662722e-07,
+      "loss": 4.1016,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4201183431952663e-07,
+      "loss": 4.0781,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6568047337278109e-07,
+      "loss": 4.1328,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8934911242603552e-07,
+      "loss": 4.0156,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1301775147928995e-07,
+      "loss": 4.1328,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.366863905325444e-07,
+      "loss": 4.1484,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6035502958579883e-07,
+      "loss": 4.1641,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8402366863905326e-07,
+      "loss": 4.1055,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 4.1562,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3136094674556217e-07,
+      "loss": 4.1641,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.550295857988166e-07,
+      "loss": 4.1406,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7869822485207103e-07,
+      "loss": 4.1406,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0236686390532546e-07,
+      "loss": 4.1406,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.260355029585799e-07,
+      "loss": 4.1719,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.497041420118344e-07,
+      "loss": 4.1602,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733727810650888e-07,
+      "loss": 4.1875,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.970414201183432e-07,
+      "loss": 4.1953,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.207100591715977e-07,
+      "loss": 4.125,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.44378698224852e-07,
+      "loss": 4.1641,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.680473372781065e-07,
+      "loss": 4.0547,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.91715976331361e-07,
+      "loss": 4.1094,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 4.1484,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.390532544378699e-07,
+      "loss": 4.1094,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.627218934911243e-07,
+      "loss": 4.25,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.863905325443787e-07,
+      "loss": 4.1875,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.100591715976332e-07,
+      "loss": 4.0742,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.337278106508876e-07,
+      "loss": 4.0859,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.573964497041421e-07,
+      "loss": 4.1875,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.810650887573964e-07,
+      "loss": 4.1406,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.047337278106509e-07,
+      "loss": 4.082,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.284023668639055e-07,
+      "loss": 4.0977,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.520710059171598e-07,
+      "loss": 4.0938,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.757396449704144e-07,
+      "loss": 4.1328,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.994082840236687e-07,
+      "loss": 4.1406,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 4.0391,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.467455621301776e-07,
+      "loss": 4.1875,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.70414201183432e-07,
+      "loss": 4.2188,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.940828402366864e-07,
+      "loss": 4.1719,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.017751479289941e-06,
+      "loss": 4.1641,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0414201183431953e-06,
+      "loss": 4.0703,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06508875739645e-06,
+      "loss": 4.1484,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.088757396449704e-06,
+      "loss": 4.0547,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1124260355029587e-06,
+      "loss": 4.1797,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.136094674556213e-06,
+      "loss": 4.0781,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1597633136094676e-06,
+      "loss": 4.125,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.183431952662722e-06,
+      "loss": 4.1484,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2071005917159764e-06,
+      "loss": 4.1719,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 4.1484,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2544378698224854e-06,
+      "loss": 4.0625,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2781065088757397e-06,
+      "loss": 4.0273,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.301775147928994e-06,
+      "loss": 4.0859,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3254437869822487e-06,
+      "loss": 4.0547,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3491124260355033e-06,
+      "loss": 4.1797,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3727810650887574e-06,
+      "loss": 4.1406,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3964497041420118e-06,
+      "loss": 4.0625,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4201183431952664e-06,
+      "loss": 4.043,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.443786982248521e-06,
+      "loss": 4.0703,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4674556213017752e-06,
+      "loss": 3.9922,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4911242603550298e-06,
+      "loss": 4.0,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5147928994082841e-06,
+      "loss": 4.125,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 4.0117,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5621301775147929e-06,
+      "loss": 3.9883,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5857988165680475e-06,
+      "loss": 4.0117,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6094674556213018e-06,
+      "loss": 3.9414,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6331360946745564e-06,
+      "loss": 4.1172,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.656804733727811e-06,
+      "loss": 4.0547,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6804733727810652e-06,
+      "loss": 4.0352,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7041420118343196e-06,
+      "loss": 4.0508,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7278106508875742e-06,
+      "loss": 4.0273,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7514792899408287e-06,
+      "loss": 3.9922,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.775147928994083e-06,
+      "loss": 3.9961,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7988165680473375e-06,
+      "loss": 3.8867,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8224852071005919e-06,
+      "loss": 3.9961,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 4.0352,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8698224852071006e-06,
+      "loss": 4.043,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8934911242603552e-06,
+      "loss": 4.0352,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9171597633136096e-06,
+      "loss": 3.9609,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.940828402366864e-06,
+      "loss": 3.9688,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9644970414201183e-06,
+      "loss": 4.0391,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9881656804733727e-06,
+      "loss": 3.9688,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0118343195266275e-06,
+      "loss": 3.9531,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.035502958579882e-06,
+      "loss": 3.9219,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0591715976331363e-06,
+      "loss": 3.9414,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0828402366863907e-06,
+      "loss": 4.0117,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.106508875739645e-06,
+      "loss": 3.8984,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1301775147929e-06,
+      "loss": 3.8398,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 3.9492,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.177514792899408e-06,
+      "loss": 3.7461,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.201183431952663e-06,
+      "loss": 3.7969,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2248520710059173e-06,
+      "loss": 3.8281,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2485207100591717e-06,
+      "loss": 3.8594,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.272189349112426e-06,
+      "loss": 3.7734,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2958579881656805e-06,
+      "loss": 3.8203,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3195266272189353e-06,
+      "loss": 3.9062,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3431952662721896e-06,
+      "loss": 3.7148,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.366863905325444e-06,
+      "loss": 3.7617,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3905325443786984e-06,
+      "loss": 3.7891,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4142011834319528e-06,
+      "loss": 3.7148,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4378698224852076e-06,
+      "loss": 3.6836,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 3.6523,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.485207100591716e-06,
+      "loss": 3.6406,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5088757396449707e-06,
+      "loss": 3.6328,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.532544378698225e-06,
+      "loss": 3.5469,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5562130177514795e-06,
+      "loss": 3.6758,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5798816568047343e-06,
+      "loss": 3.5742,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.603550295857988e-06,
+      "loss": 3.5625,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6272189349112426e-06,
+      "loss": 3.4453,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6508875739644974e-06,
+      "loss": 3.3555,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6745562130177518e-06,
+      "loss": 3.4648,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6982248520710066e-06,
+      "loss": 3.3125,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7218934911242605e-06,
+      "loss": 3.4023,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.745562130177515e-06,
+      "loss": 3.418,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 3.3438,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7928994082840236e-06,
+      "loss": 3.1797,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.816568047337278e-06,
+      "loss": 3.1797,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.840236686390533e-06,
+      "loss": 3.1914,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.863905325443787e-06,
+      "loss": 3.1211,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.887573964497042e-06,
+      "loss": 3.0352,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.911242603550296e-06,
+      "loss": 2.918,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9349112426035503e-06,
+      "loss": 2.9961,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.958579881656805e-06,
+      "loss": 3.0039,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9822485207100595e-06,
+      "loss": 2.8555,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0059171597633143e-06,
+      "loss": 2.8203,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0295857988165683e-06,
+      "loss": 2.6992,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0532544378698226e-06,
+      "loss": 2.7148,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 2.7109,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1005917159763314e-06,
+      "loss": 2.6133,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1242603550295858e-06,
+      "loss": 2.5859,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1479289940828406e-06,
+      "loss": 2.5234,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.171597633136095e-06,
+      "loss": 2.4883,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1952662721893497e-06,
+      "loss": 2.4219,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2189349112426037e-06,
+      "loss": 2.3555,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.242603550295858e-06,
+      "loss": 2.3945,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.266272189349113e-06,
+      "loss": 2.2617,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2899408284023672e-06,
+      "loss": 2.2852,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.313609467455622e-06,
+      "loss": 2.2148,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.337278106508876e-06,
+      "loss": 2.1953,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3609467455621304e-06,
+      "loss": 2.1035,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 2.125,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.408284023668639e-06,
+      "loss": 2.1406,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4319526627218935e-06,
+      "loss": 2.0859,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4556213017751483e-06,
+      "loss": 2.0176,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4792899408284027e-06,
+      "loss": 1.9922,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5029585798816575e-06,
+      "loss": 1.9668,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5266272189349114e-06,
+      "loss": 1.8965,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.550295857988166e-06,
+      "loss": 1.9062,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5739644970414206e-06,
+      "loss": 1.916,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.597633136094675e-06,
+      "loss": 1.8398,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.621301775147929e-06,
+      "loss": 1.8398,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6449704142011837e-06,
+      "loss": 1.7969,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.668639053254438e-06,
+      "loss": 1.791,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 1.7734,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.715976331360947e-06,
+      "loss": 1.7637,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7396449704142013e-06,
+      "loss": 1.7402,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.763313609467456e-06,
+      "loss": 1.7168,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7869822485207104e-06,
+      "loss": 1.7129,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8106508875739652e-06,
+      "loss": 1.7207,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.834319526627219e-06,
+      "loss": 1.6387,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8579881656804736e-06,
+      "loss": 1.623,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.881656804733728e-06,
+      "loss": 1.627,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.905325443786982e-06,
+      "loss": 1.6094,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.928994082840237e-06,
+      "loss": 1.5781,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.952662721893492e-06,
+      "loss": 1.5859,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9763313609467454e-06,
+      "loss": 1.5938,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.5586,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.023668639053255e-06,
+      "loss": 1.5215,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.047337278106509e-06,
+      "loss": 1.5195,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.071005917159764e-06,
+      "loss": 1.5273,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.094674556213018e-06,
+      "loss": 1.5234,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1183431952662725e-06,
+      "loss": 1.5098,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.142011834319527e-06,
+      "loss": 1.4727,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.165680473372781e-06,
+      "loss": 1.5,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.189349112426036e-06,
+      "loss": 1.457,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.21301775147929e-06,
+      "loss": 1.4551,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2366863905325444e-06,
+      "loss": 1.4492,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2603550295858e-06,
+      "loss": 1.4355,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.284023668639053e-06,
+      "loss": 1.4238,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 1.3887,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.331360946745563e-06,
+      "loss": 1.4199,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.355029585798816e-06,
+      "loss": 1.3984,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3786982248520715e-06,
+      "loss": 1.3945,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.402366863905326e-06,
+      "loss": 1.3926,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42603550295858e-06,
+      "loss": 1.3789,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.449704142011835e-06,
+      "loss": 1.3594,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.473372781065089e-06,
+      "loss": 1.334,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.497041420118343e-06,
+      "loss": 1.332,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520710059171598e-06,
+      "loss": 1.3574,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.544378698224852e-06,
+      "loss": 1.3477,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.568047337278107e-06,
+      "loss": 1.3281,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.591715976331361e-06,
+      "loss": 1.3008,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.332,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6390532544378705e-06,
+      "loss": 1.293,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.662721893491124e-06,
+      "loss": 1.2695,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.686390532544379e-06,
+      "loss": 1.2852,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.710059171597634e-06,
+      "loss": 1.291,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.733727810650888e-06,
+      "loss": 1.2656,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.757396449704142e-06,
+      "loss": 1.2832,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781065088757397e-06,
+      "loss": 1.2656,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.804733727810651e-06,
+      "loss": 1.2559,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8284023668639055e-06,
+      "loss": 1.2734,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.85207100591716e-06,
+      "loss": 1.2422,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.875739644970415e-06,
+      "loss": 1.2598,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.899408284023669e-06,
+      "loss": 1.2148,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 1.2305,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.946745562130178e-06,
+      "loss": 1.207,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.970414201183432e-06,
+      "loss": 1.2207,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.994082840236687e-06,
+      "loss": 1.2266,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.017751479289941e-06,
+      "loss": 1.2051,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.041420118343196e-06,
+      "loss": 1.207,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.06508875739645e-06,
+      "loss": 1.2305,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.088757396449705e-06,
+      "loss": 1.2109,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.112426035502959e-06,
+      "loss": 1.209,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.136094674556213e-06,
+      "loss": 1.2031,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.1597633136094685e-06,
+      "loss": 1.2109,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.183431952662722e-06,
+      "loss": 1.2285,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.207100591715976e-06,
+      "loss": 1.1934,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 1.1777,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.254437869822485e-06,
+      "loss": 1.1738,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2781065088757395e-06,
+      "loss": 1.1934,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.301775147928995e-06,
+      "loss": 1.1934,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.325443786982249e-06,
+      "loss": 1.168,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.3491124260355035e-06,
+      "loss": 1.1777,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.372781065088758e-06,
+      "loss": 1.1641,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.396449704142013e-06,
+      "loss": 1.2109,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.420118343195267e-06,
+      "loss": 1.166,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.443786982248521e-06,
+      "loss": 1.1523,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.467455621301776e-06,
+      "loss": 1.1719,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.49112426035503e-06,
+      "loss": 1.1875,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.514792899408284e-06,
+      "loss": 1.1484,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 1.1543,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.562130177514793e-06,
+      "loss": 1.168,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.585798816568047e-06,
+      "loss": 1.1309,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6094674556213025e-06,
+      "loss": 1.1504,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.633136094674556e-06,
+      "loss": 1.1621,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.656804733727811e-06,
+      "loss": 1.1602,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.680473372781066e-06,
+      "loss": 1.1758,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.704142011834321e-06,
+      "loss": 1.1621,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.727810650887574e-06,
+      "loss": 1.1523,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.751479289940829e-06,
+      "loss": 1.1504,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.775147928994084e-06,
+      "loss": 1.1426,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7988165680473375e-06,
+      "loss": 1.1348,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.822485207100592e-06,
+      "loss": 1.1484,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 1.1445,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.869822485207101e-06,
+      "loss": 1.1504,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.893491124260355e-06,
+      "loss": 1.125,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.91715976331361e-06,
+      "loss": 1.1133,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.940828402366864e-06,
+      "loss": 1.1465,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.964497041420119e-06,
+      "loss": 1.1133,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.988165680473373e-06,
+      "loss": 1.1348,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.011834319526629e-06,
+      "loss": 1.1133,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.035502958579882e-06,
+      "loss": 1.1328,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0591715976331365e-06,
+      "loss": 1.0918,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.082840236686392e-06,
+      "loss": 1.0723,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.106508875739645e-06,
+      "loss": 1.1055,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.1301775147929e-06,
+      "loss": 1.1191,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.0859,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.177514792899408e-06,
+      "loss": 1.1152,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.201183431952663e-06,
+      "loss": 1.1309,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.224852071005918e-06,
+      "loss": 1.0918,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2485207100591715e-06,
+      "loss": 1.0801,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.272189349112427e-06,
+      "loss": 1.082,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.295857988165681e-06,
+      "loss": 1.1152,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.319526627218936e-06,
+      "loss": 1.125,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.34319526627219e-06,
+      "loss": 1.0742,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.366863905325444e-06,
+      "loss": 1.082,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3905325443786995e-06,
+      "loss": 1.1191,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.414201183431953e-06,
+      "loss": 1.0859,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.437869822485207e-06,
+      "loss": 1.0879,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 1.0918,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.485207100591716e-06,
+      "loss": 1.1035,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5088757396449705e-06,
+      "loss": 1.0977,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.532544378698226e-06,
+      "loss": 1.0625,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.556213017751479e-06,
+      "loss": 1.0801,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5798816568047345e-06,
+      "loss": 1.0645,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.603550295857989e-06,
+      "loss": 1.043,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.627218934911244e-06,
+      "loss": 1.0469,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.650887573964498e-06,
+      "loss": 1.0547,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.674556213017752e-06,
+      "loss": 1.0996,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.698224852071007e-06,
+      "loss": 1.0586,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.721893491124261e-06,
+      "loss": 1.0762,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.745562130177515e-06,
+      "loss": 1.0859,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 1.0605,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.792899408284024e-06,
+      "loss": 1.0254,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.816568047337278e-06,
+      "loss": 1.0361,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.8402366863905335e-06,
+      "loss": 1.0762,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.863905325443787e-06,
+      "loss": 1.0547,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.887573964497042e-06,
+      "loss": 1.043,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.911242603550297e-06,
+      "loss": 1.0527,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.93491124260355e-06,
+      "loss": 1.0176,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.958579881656805e-06,
+      "loss": 1.0332,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.98224852071006e-06,
+      "loss": 1.0273,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.005917159763315e-06,
+      "loss": 1.0508,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.0295857988165685e-06,
+      "loss": 1.0127,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.053254437869823e-06,
+      "loss": 1.0088,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 1.0234,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.100591715976332e-06,
+      "loss": 1.0527,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.124260355029586e-06,
+      "loss": 1.043,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.147928994082841e-06,
+      "loss": 1.0176,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.171597633136095e-06,
+      "loss": 1.043,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.19526627218935e-06,
+      "loss": 1.0547,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.218934911242604e-06,
+      "loss": 0.9619,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.242603550295858e-06,
+      "loss": 1.0312,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.266272189349113e-06,
+      "loss": 0.9766,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2899408284023675e-06,
+      "loss": 1.0039,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.313609467455623e-06,
+      "loss": 0.999,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.337278106508876e-06,
+      "loss": 0.9883,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.360946745562131e-06,
+      "loss": 1.0059,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 1.002,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.408284023668639e-06,
+      "loss": 0.9873,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.431952662721894e-06,
+      "loss": 1.0029,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.455621301775149e-06,
+      "loss": 1.0303,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.4792899408284025e-06,
+      "loss": 0.9961,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.502958579881658e-06,
+      "loss": 0.9932,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.526627218934912e-06,
+      "loss": 0.9883,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.550295857988166e-06,
+      "loss": 0.9795,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.573964497041421e-06,
+      "loss": 0.9854,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.597633136094675e-06,
+      "loss": 1.0186,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6213017751479305e-06,
+      "loss": 1.0078,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.644970414201183e-06,
+      "loss": 0.9961,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.668639053254438e-06,
+      "loss": 1.0059,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.9736,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.715976331360947e-06,
+      "loss": 0.9912,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.739644970414202e-06,
+      "loss": 0.9922,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.763313609467456e-06,
+      "loss": 0.9658,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.786982248520711e-06,
+      "loss": 0.9678,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.810650887573965e-06,
+      "loss": 0.9531,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.83431952662722e-06,
+      "loss": 0.9531,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.857988165680473e-06,
+      "loss": 0.9873,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.881656804733729e-06,
+      "loss": 0.959,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.905325443786984e-06,
+      "loss": 0.9658,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.928994082840237e-06,
+      "loss": 0.9707,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.952662721893491e-06,
+      "loss": 0.9883,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.976331360946746e-06,
+      "loss": 0.9844,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9746,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.023668639053255e-06,
+      "loss": 0.96,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.04733727810651e-06,
+      "loss": 0.9609,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.071005917159764e-06,
+      "loss": 0.9785,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.094674556213019e-06,
+      "loss": 0.9541,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.118343195266272e-06,
+      "loss": 0.96,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.142011834319528e-06,
+      "loss": 0.9727,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.165680473372781e-06,
+      "loss": 0.9756,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.189349112426036e-06,
+      "loss": 0.958,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.213017751479292e-06,
+      "loss": 0.9824,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.236686390532545e-06,
+      "loss": 0.9385,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.260355029585799e-06,
+      "loss": 1.0088,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.284023668639054e-06,
+      "loss": 0.9619,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 0.9717,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.331360946745563e-06,
+      "loss": 0.9463,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.355029585798818e-06,
+      "loss": 0.9512,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.378698224852071e-06,
+      "loss": 0.9639,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.402366863905327e-06,
+      "loss": 0.9463,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.42603550295858e-06,
+      "loss": 0.9434,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.449704142011835e-06,
+      "loss": 0.9629,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.473372781065089e-06,
+      "loss": 0.9912,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.497041420118344e-06,
+      "loss": 0.958,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.5207100591716e-06,
+      "loss": 0.9512,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.544378698224853e-06,
+      "loss": 0.9365,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.568047337278106e-06,
+      "loss": 0.9375,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.591715976331362e-06,
+      "loss": 0.9639,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.9434,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.63905325443787e-06,
+      "loss": 0.9453,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.662721893491126e-06,
+      "loss": 0.9688,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.686390532544379e-06,
+      "loss": 0.9287,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.710059171597633e-06,
+      "loss": 0.9209,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.733727810650888e-06,
+      "loss": 0.9219,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.757396449704143e-06,
+      "loss": 0.9629,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.781065088757397e-06,
+      "loss": 0.9541,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.804733727810652e-06,
+      "loss": 0.958,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.828402366863907e-06,
+      "loss": 0.9395,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.85207100591716e-06,
+      "loss": 0.9199,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.875739644970414e-06,
+      "loss": 0.9531,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.89940828402367e-06,
+      "loss": 0.9336,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 0.9473,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.946745562130178e-06,
+      "loss": 0.9326,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.970414201183433e-06,
+      "loss": 0.9346,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.994082840236687e-06,
+      "loss": 0.9512,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.01775147928994e-06,
+      "loss": 0.9492,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.041420118343196e-06,
+      "loss": 0.9355,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.06508875739645e-06,
+      "loss": 0.9473,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.088757396449704e-06,
+      "loss": 0.9072,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.11242603550296e-06,
+      "loss": 0.9502,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.136094674556215e-06,
+      "loss": 0.9229,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.159763313609468e-06,
+      "loss": 0.9023,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.183431952662722e-06,
+      "loss": 0.9365,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.207100591715977e-06,
+      "loss": 0.9023,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9033,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.254437869822486e-06,
+      "loss": 0.9277,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.278106508875741e-06,
+      "loss": 0.9189,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.301775147928995e-06,
+      "loss": 0.9082,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.325443786982248e-06,
+      "loss": 0.9297,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.349112426035503e-06,
+      "loss": 0.9043,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.372781065088759e-06,
+      "loss": 0.9473,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.396449704142012e-06,
+      "loss": 0.9258,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.420118343195267e-06,
+      "loss": 0.9268,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.443786982248523e-06,
+      "loss": 0.9062,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.467455621301776e-06,
+      "loss": 0.9062,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.49112426035503e-06,
+      "loss": 0.8984,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.514792899408285e-06,
+      "loss": 0.915,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 0.9248,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.562130177514794e-06,
+      "loss": 0.9062,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.585798816568049e-06,
+      "loss": 0.9453,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.609467455621302e-06,
+      "loss": 0.8867,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.633136094674556e-06,
+      "loss": 0.9072,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.656804733727811e-06,
+      "loss": 0.9229,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.680473372781066e-06,
+      "loss": 0.917,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.70414201183432e-06,
+      "loss": 0.9043,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.727810650887575e-06,
+      "loss": 0.8975,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.75147928994083e-06,
+      "loss": 0.9355,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.775147928994084e-06,
+      "loss": 0.9092,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.798816568047337e-06,
+      "loss": 0.9473,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.822485207100593e-06,
+      "loss": 0.9023,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.9238,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869822485207101e-06,
+      "loss": 0.9053,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893491124260357e-06,
+      "loss": 0.875,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91715976331361e-06,
+      "loss": 0.9092,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.940828402366864e-06,
+      "loss": 0.9121,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964497041420119e-06,
+      "loss": 0.9229,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988165680473374e-06,
+      "loss": 0.9258,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0011834319526628e-05,
+      "loss": 0.9316,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0035502958579883e-05,
+      "loss": 0.9551,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0059171597633138e-05,
+      "loss": 0.9189,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0082840236686392e-05,
+      "loss": 0.8955,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0106508875739647e-05,
+      "loss": 0.915,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.01301775147929e-05,
+      "loss": 0.9277,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 0.8955,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.017751479289941e-05,
+      "loss": 0.9023,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0201183431952664e-05,
+      "loss": 0.9102,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0224852071005918e-05,
+      "loss": 0.9346,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0248520710059173e-05,
+      "loss": 0.9102,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0272189349112427e-05,
+      "loss": 0.8799,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.029585798816568e-05,
+      "loss": 0.9014,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0319526627218937e-05,
+      "loss": 0.9062,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.034319526627219e-05,
+      "loss": 0.8965,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0366863905325444e-05,
+      "loss": 0.9092,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.03905325443787e-05,
+      "loss": 0.8848,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0414201183431953e-05,
+      "loss": 0.9092,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0437869822485206e-05,
+      "loss": 0.9121,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.8975,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0485207100591717e-05,
+      "loss": 0.916,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.050887573964497e-05,
+      "loss": 0.9072,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0532544378698226e-05,
+      "loss": 0.8828,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0556213017751479e-05,
+      "loss": 0.9258,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0579881656804734e-05,
+      "loss": 0.9189,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.060355029585799e-05,
+      "loss": 0.9102,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0627218934911243e-05,
+      "loss": 0.8838,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0650887573964498e-05,
+      "loss": 0.915,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0674556213017754e-05,
+      "loss": 0.9092,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0698224852071007e-05,
+      "loss": 0.8906,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0721893491124262e-05,
+      "loss": 0.8633,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0745562130177516e-05,
+      "loss": 0.8867,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9121,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0792899408284026e-05,
+      "loss": 0.877,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.081656804733728e-05,
+      "loss": 0.8721,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0840236686390533e-05,
+      "loss": 0.8984,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0863905325443789e-05,
+      "loss": 0.8818,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0887573964497042e-05,
+      "loss": 0.9287,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0911242603550296e-05,
+      "loss": 0.8848,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0934911242603553e-05,
+      "loss": 0.8799,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0958579881656806e-05,
+      "loss": 0.9062,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.098224852071006e-05,
+      "loss": 0.8945,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1005917159763315e-05,
+      "loss": 0.917,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1029585798816568e-05,
+      "loss": 0.9121,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1053254437869822e-05,
+      "loss": 0.8887,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.9023,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1100591715976332e-05,
+      "loss": 0.9131,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1124260355029586e-05,
+      "loss": 0.9072,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1147928994082841e-05,
+      "loss": 0.8818,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1171597633136095e-05,
+      "loss": 0.8936,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.119526627218935e-05,
+      "loss": 0.8857,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1218934911242605e-05,
+      "loss": 0.8975,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1242603550295859e-05,
+      "loss": 0.8965,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1266272189349112e-05,
+      "loss": 0.8936,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1289940828402369e-05,
+      "loss": 0.8896,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1313609467455623e-05,
+      "loss": 0.8965,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1337278106508878e-05,
+      "loss": 0.873,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1360946745562131e-05,
+      "loss": 0.8867,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 0.8896,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1408284023668642e-05,
+      "loss": 0.8672,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1431952662721895e-05,
+      "loss": 0.915,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1455621301775149e-05,
+      "loss": 0.8789,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1479289940828404e-05,
+      "loss": 0.8984,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1502958579881658e-05,
+      "loss": 0.8838,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1526627218934911e-05,
+      "loss": 0.8916,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1550295857988168e-05,
+      "loss": 0.8887,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1573964497041422e-05,
+      "loss": 0.8779,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1597633136094675e-05,
+      "loss": 0.916,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.162130177514793e-05,
+      "loss": 0.8633,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1644970414201184e-05,
+      "loss": 0.9307,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1668639053254437e-05,
+      "loss": 0.8691,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.8896,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1715976331360948e-05,
+      "loss": 0.8809,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1739644970414201e-05,
+      "loss": 0.8682,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1763313609467457e-05,
+      "loss": 0.8965,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.178698224852071e-05,
+      "loss": 0.9072,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1810650887573965e-05,
+      "loss": 0.9082,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.183431952662722e-05,
+      "loss": 0.8887,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1857988165680474e-05,
+      "loss": 0.8652,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1881656804733728e-05,
+      "loss": 0.8594,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1905325443786983e-05,
+      "loss": 0.8682,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1928994082840238e-05,
+      "loss": 0.8711,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1952662721893493e-05,
+      "loss": 0.8896,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1976331360946747e-05,
+      "loss": 0.8594,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8955,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2023668639053257e-05,
+      "loss": 0.8545,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.204733727810651e-05,
+      "loss": 0.8545,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2071005917159764e-05,
+      "loss": 0.8662,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.209467455621302e-05,
+      "loss": 0.9053,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2118343195266273e-05,
+      "loss": 0.8584,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2142011834319527e-05,
+      "loss": 0.876,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2165680473372783e-05,
+      "loss": 0.8955,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2189349112426037e-05,
+      "loss": 0.8789,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.221301775147929e-05,
+      "loss": 0.8623,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2236686390532546e-05,
+      "loss": 0.9092,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.22603550295858e-05,
+      "loss": 0.8477,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2284023668639053e-05,
+      "loss": 0.877,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.8857,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2331360946745563e-05,
+      "loss": 0.877,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2355029585798817e-05,
+      "loss": 0.8896,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2378698224852072e-05,
+      "loss": 0.8652,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2402366863905326e-05,
+      "loss": 0.8506,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.242603550295858e-05,
+      "loss": 0.8555,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2449704142011836e-05,
+      "loss": 0.8711,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.247337278106509e-05,
+      "loss": 0.8867,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2497041420118343e-05,
+      "loss": 0.9062,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2520710059171598e-05,
+      "loss": 0.8809,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2544378698224854e-05,
+      "loss": 0.877,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2568047337278107e-05,
+      "loss": 0.8662,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2591715976331362e-05,
+      "loss": 0.8877,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 0.8564,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2639053254437873e-05,
+      "loss": 0.875,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2662721893491126e-05,
+      "loss": 0.8926,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.268639053254438e-05,
+      "loss": 0.8945,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2710059171597635e-05,
+      "loss": 0.8701,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2733727810650889e-05,
+      "loss": 0.8887,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2757396449704142e-05,
+      "loss": 0.8691,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2781065088757399e-05,
+      "loss": 0.8877,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2804733727810652e-05,
+      "loss": 0.8535,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2828402366863906e-05,
+      "loss": 0.8926,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2852071005917161e-05,
+      "loss": 0.9062,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2875739644970415e-05,
+      "loss": 0.8721,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2899408284023668e-05,
+      "loss": 0.874,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.8789,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2946745562130179e-05,
+      "loss": 0.8818,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2970414201183432e-05,
+      "loss": 0.8896,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2994082840236688e-05,
+      "loss": 0.8682,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3017751479289941e-05,
+      "loss": 0.8398,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3041420118343196e-05,
+      "loss": 0.8838,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3065088757396451e-05,
+      "loss": 0.8799,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3088757396449705e-05,
+      "loss": 0.8789,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3112426035502959e-05,
+      "loss": 0.8936,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3136094674556214e-05,
+      "loss": 0.8662,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3159763313609469e-05,
+      "loss": 0.8574,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3183431952662723e-05,
+      "loss": 0.8438,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3207100591715978e-05,
+      "loss": 0.8467,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 0.8779,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3254437869822488e-05,
+      "loss": 0.8799,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3278106508875742e-05,
+      "loss": 0.8516,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3301775147928995e-05,
+      "loss": 0.8799,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.332544378698225e-05,
+      "loss": 0.8369,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3349112426035504e-05,
+      "loss": 0.8613,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3372781065088758e-05,
+      "loss": 0.8994,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3396449704142014e-05,
+      "loss": 0.8564,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3420118343195268e-05,
+      "loss": 0.8955,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3443786982248522e-05,
+      "loss": 0.8848,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3467455621301777e-05,
+      "loss": 0.8682,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.349112426035503e-05,
+      "loss": 0.8252,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3514792899408284e-05,
+      "loss": 0.8828,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.8662,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3562130177514794e-05,
+      "loss": 0.8564,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3585798816568048e-05,
+      "loss": 0.8613,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3609467455621303e-05,
+      "loss": 0.8682,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3633136094674557e-05,
+      "loss": 0.8623,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.365680473372781e-05,
+      "loss": 0.8652,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3680473372781067e-05,
+      "loss": 0.8652,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.370414201183432e-05,
+      "loss": 0.8486,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3727810650887574e-05,
+      "loss": 0.8701,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.375147928994083e-05,
+      "loss": 0.8965,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3775147928994084e-05,
+      "loss": 0.8672,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3798816568047338e-05,
+      "loss": 0.8623,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3822485207100593e-05,
+      "loss": 0.8672,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.9189,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.38698224852071e-05,
+      "loss": 0.8525,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3893491124260357e-05,
+      "loss": 0.8398,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.391715976331361e-05,
+      "loss": 0.874,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3940828402366866e-05,
+      "loss": 0.8789,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.396449704142012e-05,
+      "loss": 0.8594,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3988165680473373e-05,
+      "loss": 0.8682,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.401183431952663e-05,
+      "loss": 0.8945,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4035502958579883e-05,
+      "loss": 0.8809,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4059171597633137e-05,
+      "loss": 0.8525,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4082840236686392e-05,
+      "loss": 0.8486,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4106508875739646e-05,
+      "loss": 0.8389,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.41301775147929e-05,
+      "loss": 0.8701,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.8643,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.417751479289941e-05,
+      "loss": 0.874,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4201183431952663e-05,
+      "loss": 0.8545,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4224852071005918e-05,
+      "loss": 0.8701,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4248520710059172e-05,
+      "loss": 0.8594,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4272189349112426e-05,
+      "loss": 0.8604,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4295857988165682e-05,
+      "loss": 0.8613,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4319526627218936e-05,
+      "loss": 0.8574,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.434319526627219e-05,
+      "loss": 0.8809,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4366863905325445e-05,
+      "loss": 0.8467,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.43905325443787e-05,
+      "loss": 0.8652,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4414201183431953e-05,
+      "loss": 0.8604,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4437869822485209e-05,
+      "loss": 0.8584,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 0.8604,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4485207100591716e-05,
+      "loss": 0.8672,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4508875739644973e-05,
+      "loss": 0.877,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4532544378698226e-05,
+      "loss": 0.8584,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4556213017751481e-05,
+      "loss": 0.832,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4579881656804735e-05,
+      "loss": 0.8486,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4603550295857989e-05,
+      "loss": 0.8389,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4627218934911245e-05,
+      "loss": 0.8574,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4650887573964499e-05,
+      "loss": 0.8662,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4674556213017752e-05,
+      "loss": 0.8477,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4698224852071008e-05,
+      "loss": 0.8613,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4721893491124261e-05,
+      "loss": 0.874,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4745562130177515e-05,
+      "loss": 0.8564,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.8525,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4792899408284025e-05,
+      "loss": 0.8584,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4816568047337279e-05,
+      "loss": 0.8594,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4840236686390534e-05,
+      "loss": 0.8701,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4863905325443788e-05,
+      "loss": 0.8682,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4887573964497041e-05,
+      "loss": 0.8604,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4911242603550298e-05,
+      "loss": 0.8799,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4934911242603551e-05,
+      "loss": 0.873,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4958579881656805e-05,
+      "loss": 0.8545,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.498224852071006e-05,
+      "loss": 0.873,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5005917159763315e-05,
+      "loss": 0.8418,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5029585798816569e-05,
+      "loss": 0.8672,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5053254437869824e-05,
+      "loss": 0.8535,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 0.8643,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5100591715976331e-05,
+      "loss": 0.8359,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5124260355029588e-05,
+      "loss": 0.8721,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5147928994082842e-05,
+      "loss": 0.8555,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5171597633136097e-05,
+      "loss": 0.8516,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.519526627218935e-05,
+      "loss": 0.8535,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5218934911242604e-05,
+      "loss": 0.8379,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5242603550295861e-05,
+      "loss": 0.8574,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5266272189349113e-05,
+      "loss": 0.8613,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5289940828402366e-05,
+      "loss": 0.8574,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5313609467455623e-05,
+      "loss": 0.8535,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5337278106508877e-05,
+      "loss": 0.8418,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.536094674556213e-05,
+      "loss": 0.8613,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.8838,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.540828402366864e-05,
+      "loss": 0.8408,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5431952662721894e-05,
+      "loss": 0.8828,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.545562130177515e-05,
+      "loss": 0.8643,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5479289940828405e-05,
+      "loss": 0.8848,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5502958579881658e-05,
+      "loss": 0.8164,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5526627218934912e-05,
+      "loss": 0.8691,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5550295857988165e-05,
+      "loss": 0.8613,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5573964497041422e-05,
+      "loss": 0.8604,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5597633136094676e-05,
+      "loss": 0.8613,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.562130177514793e-05,
+      "loss": 0.876,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5644970414201183e-05,
+      "loss": 0.8643,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.566863905325444e-05,
+      "loss": 0.8564,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 0.8477,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5715976331360947e-05,
+      "loss": 0.8652,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5739644970414204e-05,
+      "loss": 0.8633,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5763313609467457e-05,
+      "loss": 0.834,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.578698224852071e-05,
+      "loss": 0.8379,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5810650887573968e-05,
+      "loss": 0.8594,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.583431952662722e-05,
+      "loss": 0.8789,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5857988165680475e-05,
+      "loss": 0.8887,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5881656804733728e-05,
+      "loss": 0.8711,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5905325443786982e-05,
+      "loss": 0.8691,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.592899408284024e-05,
+      "loss": 0.8555,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5952662721893492e-05,
+      "loss": 0.8652,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5976331360946746e-05,
+      "loss": 0.8496,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.875,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6023668639053256e-05,
+      "loss": 0.873,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.604733727810651e-05,
+      "loss": 0.8496,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6071005917159767e-05,
+      "loss": 0.8545,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.609467455621302e-05,
+      "loss": 0.832,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6118343195266274e-05,
+      "loss": 0.8682,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6142011834319527e-05,
+      "loss": 0.8545,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.616568047337278e-05,
+      "loss": 0.835,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6189349112426038e-05,
+      "loss": 0.8506,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.621301775147929e-05,
+      "loss": 0.8652,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6236686390532545e-05,
+      "loss": 0.8506,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6260355029585798e-05,
+      "loss": 0.8428,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6284023668639055e-05,
+      "loss": 0.8418,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 0.8203,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6331360946745562e-05,
+      "loss": 0.8564,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.635502958579882e-05,
+      "loss": 0.8604,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6378698224852073e-05,
+      "loss": 0.8848,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6402366863905326e-05,
+      "loss": 0.8613,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6426035502958583e-05,
+      "loss": 0.8496,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6449704142011837e-05,
+      "loss": 0.834,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.647337278106509e-05,
+      "loss": 0.8867,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6497041420118344e-05,
+      "loss": 0.8652,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6520710059171597e-05,
+      "loss": 0.8506,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6544378698224854e-05,
+      "loss": 0.8506,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6568047337278108e-05,
+      "loss": 0.877,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.659171597633136e-05,
+      "loss": 0.8604,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.8408,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.663905325443787e-05,
+      "loss": 0.8652,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6662721893491125e-05,
+      "loss": 0.8574,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6686390532544382e-05,
+      "loss": 0.8789,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6710059171597636e-05,
+      "loss": 0.8643,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.673372781065089e-05,
+      "loss": 0.8555,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6757396449704143e-05,
+      "loss": 0.877,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6781065088757396e-05,
+      "loss": 0.8789,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6804733727810653e-05,
+      "loss": 0.8389,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6828402366863907e-05,
+      "loss": 0.8711,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.685207100591716e-05,
+      "loss": 0.8662,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6875739644970414e-05,
+      "loss": 0.8457,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.689940828402367e-05,
+      "loss": 0.8643,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.8623,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6946745562130178e-05,
+      "loss": 0.8477,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6970414201183435e-05,
+      "loss": 0.8477,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6994082840236688e-05,
+      "loss": 0.8652,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7017751479289942e-05,
+      "loss": 0.8896,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.70414201183432e-05,
+      "loss": 0.8408,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7065088757396452e-05,
+      "loss": 0.8457,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7088757396449706e-05,
+      "loss": 0.8799,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.711242603550296e-05,
+      "loss": 0.874,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7136094674556213e-05,
+      "loss": 0.8359,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.715976331360947e-05,
+      "loss": 0.8369,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7183431952662723e-05,
+      "loss": 0.8711,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7207100591715977e-05,
+      "loss": 0.8809,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.876,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7254437869822487e-05,
+      "loss": 0.834,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.727810650887574e-05,
+      "loss": 0.8691,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7301775147928998e-05,
+      "loss": 0.8525,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.732544378698225e-05,
+      "loss": 0.8633,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7349112426035505e-05,
+      "loss": 0.8643,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7372781065088758e-05,
+      "loss": 0.8447,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7396449704142012e-05,
+      "loss": 0.8545,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7420118343195265e-05,
+      "loss": 0.9062,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7443786982248522e-05,
+      "loss": 0.8652,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7467455621301776e-05,
+      "loss": 0.8633,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.749112426035503e-05,
+      "loss": 0.833,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7514792899408286e-05,
+      "loss": 0.8379,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 0.8604,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7562130177514793e-05,
+      "loss": 0.8672,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.758579881656805e-05,
+      "loss": 0.8369,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7609467455621304e-05,
+      "loss": 0.833,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7633136094674557e-05,
+      "loss": 0.8418,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7656804733727814e-05,
+      "loss": 0.877,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7680473372781068e-05,
+      "loss": 0.8438,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.770414201183432e-05,
+      "loss": 0.8721,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7727810650887575e-05,
+      "loss": 0.8486,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7751479289940828e-05,
+      "loss": 0.8525,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7775147928994085e-05,
+      "loss": 0.8193,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.779881656804734e-05,
+      "loss": 0.8193,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7822485207100592e-05,
+      "loss": 0.8408,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.8672,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7869822485207103e-05,
+      "loss": 0.833,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7893491124260356e-05,
+      "loss": 0.8779,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7917159763313613e-05,
+      "loss": 0.8506,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7940828402366867e-05,
+      "loss": 0.8428,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.796449704142012e-05,
+      "loss": 0.8535,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7988165680473374e-05,
+      "loss": 0.8652,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8011834319526627e-05,
+      "loss": 0.8291,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.803550295857988e-05,
+      "loss": 0.8301,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8059171597633138e-05,
+      "loss": 0.8633,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.808284023668639e-05,
+      "loss": 0.8506,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8106508875739645e-05,
+      "loss": 0.8809,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.81301775147929e-05,
+      "loss": 0.8662,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 0.832,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.817751479289941e-05,
+      "loss": 0.8633,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8201183431952666e-05,
+      "loss": 0.8555,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.822485207100592e-05,
+      "loss": 0.8594,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8248520710059173e-05,
+      "loss": 0.8564,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.827218934911243e-05,
+      "loss": 0.8477,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8295857988165683e-05,
+      "loss": 0.8535,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8319526627218937e-05,
+      "loss": 0.8545,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.834319526627219e-05,
+      "loss": 0.833,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8366863905325444e-05,
+      "loss": 0.8574,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.83905325443787e-05,
+      "loss": 0.8711,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8414201183431954e-05,
+      "loss": 0.8662,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8437869822485208e-05,
+      "loss": 0.8613,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.8379,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8485207100591718e-05,
+      "loss": 0.8867,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.850887573964497e-05,
+      "loss": 0.8721,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.853254437869823e-05,
+      "loss": 0.8555,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8556213017751482e-05,
+      "loss": 0.8643,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8579881656804736e-05,
+      "loss": 0.8535,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.860355029585799e-05,
+      "loss": 0.8564,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8627218934911243e-05,
+      "loss": 0.8545,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8650887573964496e-05,
+      "loss": 0.8359,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8674556213017753e-05,
+      "loss": 0.8633,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8698224852071007e-05,
+      "loss": 0.8359,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.872189349112426e-05,
+      "loss": 0.8584,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8745562130177517e-05,
+      "loss": 0.8584,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 0.8799,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8792899408284024e-05,
+      "loss": 0.8799,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.881656804733728e-05,
+      "loss": 0.8633,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8840236686390535e-05,
+      "loss": 0.8564,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8863905325443788e-05,
+      "loss": 0.8613,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8887573964497045e-05,
+      "loss": 0.8428,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.89112426035503e-05,
+      "loss": 0.8506,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8934911242603552e-05,
+      "loss": 0.8506,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8958579881656806e-05,
+      "loss": 0.8271,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898224852071006e-05,
+      "loss": 0.8633,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9005917159763313e-05,
+      "loss": 0.8486,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902958579881657e-05,
+      "loss": 0.8701,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9053254437869823e-05,
+      "loss": 0.8545,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.8262,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9100591715976334e-05,
+      "loss": 0.8545,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9124260355029587e-05,
+      "loss": 0.8691,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9147928994082844e-05,
+      "loss": 0.8477,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9171597633136098e-05,
+      "loss": 0.8408,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.919526627218935e-05,
+      "loss": 0.8193,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9218934911242605e-05,
+      "loss": 0.8535,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9242603550295858e-05,
+      "loss": 0.8447,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9266272189349112e-05,
+      "loss": 0.8438,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.928994082840237e-05,
+      "loss": 0.834,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313609467455622e-05,
+      "loss": 0.8398,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9337278106508876e-05,
+      "loss": 0.8447,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9360946745562133e-05,
+      "loss": 0.8291,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 0.8643,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.940828402366864e-05,
+      "loss": 0.8311,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9431952662721897e-05,
+      "loss": 0.8506,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.945562130177515e-05,
+      "loss": 0.8057,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9479289940828404e-05,
+      "loss": 0.8643,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.950295857988166e-05,
+      "loss": 0.8428,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9526627218934914e-05,
+      "loss": 0.8203,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9550295857988168e-05,
+      "loss": 0.8633,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.957396449704142e-05,
+      "loss": 0.8545,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9597633136094675e-05,
+      "loss": 0.8555,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9621301775147928e-05,
+      "loss": 0.8691,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9644970414201185e-05,
+      "loss": 0.8271,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.966863905325444e-05,
+      "loss": 0.8291,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 0.874,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.971597633136095e-05,
+      "loss": 0.832,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9739644970414203e-05,
+      "loss": 0.8604,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.976331360946746e-05,
+      "loss": 0.8496,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9786982248520713e-05,
+      "loss": 0.8301,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9810650887573967e-05,
+      "loss": 0.8711,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.983431952662722e-05,
+      "loss": 0.8467,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9857988165680474e-05,
+      "loss": 0.8281,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9881656804733727e-05,
+      "loss": 0.8477,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9905325443786984e-05,
+      "loss": 0.8428,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9928994082840238e-05,
+      "loss": 0.874,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995266272189349e-05,
+      "loss": 0.8545,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976331360946748e-05,
+      "loss": 0.8652,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.8643,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999933738285e-05,
+      "loss": 0.8672,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999973495314e-05,
+      "loss": 0.8438,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999403644565e-05,
+      "loss": 0.8262,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999998939812565e-05,
+      "loss": 0.8467,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999834345715e-05,
+      "loss": 0.8516,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997614578324e-05,
+      "loss": 0.8457,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999996753176097e-05,
+      "loss": 0.8408,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995759250482e-05,
+      "loss": 0.832,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999994632801488e-05,
+      "loss": 0.8672,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999993373829137e-05,
+      "loss": 0.8467,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999991982333444e-05,
+      "loss": 0.835,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999045831442e-05,
+      "loss": 0.8369,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999988801772096e-05,
+      "loss": 0.834,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999987012706486e-05,
+      "loss": 0.8711,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998509111762e-05,
+      "loss": 0.8359,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999983037005516e-05,
+      "loss": 0.8779,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999980850370203e-05,
+      "loss": 0.835,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999978531211717e-05,
+      "loss": 0.8584,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997607953008e-05,
+      "loss": 0.8779,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999973495325333e-05,
+      "loss": 0.8438,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999970778597503e-05,
+      "loss": 0.8564,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999967929346628e-05,
+      "loss": 0.8486,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999964947572746e-05,
+      "loss": 0.8291,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999961833275897e-05,
+      "loss": 0.8555,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999958586456122e-05,
+      "loss": 0.8447,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999955207113462e-05,
+      "loss": 0.8525,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999951695247968e-05,
+      "loss": 0.8369,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999994805085968e-05,
+      "loss": 0.8516,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999994427394865e-05,
+      "loss": 0.8594,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999940364514924e-05,
+      "loss": 0.8262,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999993632255856e-05,
+      "loss": 0.8281,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999932148079607e-05,
+      "loss": 0.8525,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999927841078118e-05,
+      "loss": 0.8525,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999923401554157e-05,
+      "loss": 0.8594,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999918829507782e-05,
+      "loss": 0.833,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999914124939047e-05,
+      "loss": 0.8438,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999990928784802e-05,
+      "loss": 0.8604,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999904318234763e-05,
+      "loss": 0.8486,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999989921609934e-05,
+      "loss": 0.8223,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999893981441825e-05,
+      "loss": 0.8535,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999888614262278e-05,
+      "loss": 0.8145,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999988311456078e-05,
+      "loss": 0.8184,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999877482337398e-05,
+      "loss": 0.8164,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999871717592207e-05,
+      "loss": 0.8447,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999865820325282e-05,
+      "loss": 0.8877,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999985979053671e-05,
+      "loss": 0.8477,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999985362822656e-05,
+      "loss": 0.8691,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999984733339492e-05,
+      "loss": 0.8223,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999984090604187e-05,
+      "loss": 0.8672,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999834346167496e-05,
+      "loss": 0.8564,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999827653771886e-05,
+      "loss": 0.8428,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999982082885513e-05,
+      "loss": 0.8672,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999981387141731e-05,
+      "loss": 0.8516,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999806781458534e-05,
+      "loss": 0.8975,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999799558978884e-05,
+      "loss": 0.833,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999792203978455e-05,
+      "loss": 0.8545,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999784716457354e-05,
+      "loss": 0.8672,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999777096415668e-05,
+      "loss": 0.8525,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999976934385351e-05,
+      "loss": 0.8467,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999761458770975e-05,
+      "loss": 0.8359,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999753441168165e-05,
+      "loss": 0.8232,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999745291045197e-05,
+      "loss": 0.8438,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999973700840217e-05,
+      "loss": 0.8447,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999728593239197e-05,
+      "loss": 0.8457,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999972004555639e-05,
+      "loss": 0.8447,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999711365353856e-05,
+      "loss": 0.8271,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999970255263172e-05,
+      "loss": 0.8145,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999693607390096e-05,
+      "loss": 0.8389,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99996845296291e-05,
+      "loss": 0.8145,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999967531934885e-05,
+      "loss": 0.8564,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999966597654947e-05,
+      "loss": 0.8555,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999965650123109e-05,
+      "loss": 0.8447,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999646893393825e-05,
+      "loss": 0.8467,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999963715303781e-05,
+      "loss": 0.8359,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999627280163172e-05,
+      "loss": 0.8398,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999961727477004e-05,
+      "loss": 0.8223,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999607136858552e-05,
+      "loss": 0.8516,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999596866428836e-05,
+      "loss": 0.8477,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999586463481032e-05,
+      "loss": 0.8662,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999575928015276e-05,
+      "loss": 0.8535,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999565260031707e-05,
+      "loss": 0.8428,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999955445953047e-05,
+      "loss": 0.8604,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999543526511704e-05,
+      "loss": 0.8447,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999532460975555e-05,
+      "loss": 0.8623,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999521262922172e-05,
+      "loss": 0.8525,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999509932351698e-05,
+      "loss": 0.8428,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999949846926429e-05,
+      "loss": 0.8535,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999486873660098e-05,
+      "loss": 0.8467,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999475145539273e-05,
+      "loss": 0.8525,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946328490197e-05,
+      "loss": 0.8486,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999451291748352e-05,
+      "loss": 0.8311,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999943916607857e-05,
+      "loss": 0.8379,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999942690789279e-05,
+      "loss": 0.8223,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999414517191177e-05,
+      "loss": 0.8301,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999401993973885e-05,
+      "loss": 0.834,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999938933824109e-05,
+      "loss": 0.8359,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999376549992955e-05,
+      "loss": 0.832,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999936362922965e-05,
+      "loss": 0.8418,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999350575951345e-05,
+      "loss": 0.8555,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999337390158217e-05,
+      "loss": 0.8369,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999932407185044e-05,
+      "loss": 0.8271,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999310621028186e-05,
+      "loss": 0.8623,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999297037691633e-05,
+      "loss": 0.8486,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999928332184097e-05,
+      "loss": 0.8252,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999269473476373e-05,
+      "loss": 0.8525,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999925549259802e-05,
+      "loss": 0.8545,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999241379206105e-05,
+      "loss": 0.8203,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999227133300814e-05,
+      "loss": 0.8311,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999921275488233e-05,
+      "loss": 0.832,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999919824395085e-05,
+      "loss": 0.8623,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999183600506563e-05,
+      "loss": 0.835,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999916882454966e-05,
+      "loss": 0.835,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999153916080344e-05,
+      "loss": 0.8301,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999138875098808e-05,
+      "loss": 0.8398,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999123701605253e-05,
+      "loss": 0.8213,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999108395599875e-05,
+      "loss": 0.8496,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999092957082884e-05,
+      "loss": 0.8428,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999077386054483e-05,
+      "loss": 0.8369,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999061682514876e-05,
+      "loss": 0.8262,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999045846464272e-05,
+      "loss": 0.8691,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999902987790288e-05,
+      "loss": 0.8662,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999013776830916e-05,
+      "loss": 0.8545,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998997543248587e-05,
+      "loss": 0.8584,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999898117715611e-05,
+      "loss": 0.8428,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998964678553707e-05,
+      "loss": 0.8789,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998948047441587e-05,
+      "loss": 0.8564,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998931283819982e-05,
+      "loss": 0.832,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998914387689105e-05,
+      "loss": 0.8467,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998897359049187e-05,
+      "loss": 0.874,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998880197900447e-05,
+      "loss": 0.8691,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998862904243113e-05,
+      "loss": 0.8496,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999884547807742e-05,
+      "loss": 0.8379,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999882791940359e-05,
+      "loss": 0.8535,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998810228221867e-05,
+      "loss": 0.8184,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999879240453248e-05,
+      "loss": 0.8379,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998774448335662e-05,
+      "loss": 0.8652,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999875635963165e-05,
+      "loss": 0.8418,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998738138420694e-05,
+      "loss": 0.8457,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998719784703026e-05,
+      "loss": 0.8398,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998701298478895e-05,
+      "loss": 0.8291,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998682679748537e-05,
+      "loss": 0.8311,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999866392851221e-05,
+      "loss": 0.8369,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998645044770155e-05,
+      "loss": 0.832,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998626028522625e-05,
+      "loss": 0.8711,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998606879769875e-05,
+      "loss": 0.8408,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998587598512152e-05,
+      "loss": 0.8682,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998568184749715e-05,
+      "loss": 0.8311,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999854863848282e-05,
+      "loss": 0.8545,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998528959711726e-05,
+      "loss": 0.8457,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99985091484367e-05,
+      "loss": 0.8887,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998489204657995e-05,
+      "loss": 0.8438,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999846912837588e-05,
+      "loss": 0.8438,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998448919590626e-05,
+      "loss": 0.8242,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998428578302493e-05,
+      "loss": 0.8506,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999840810451175e-05,
+      "loss": 0.8438,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998387498218677e-05,
+      "loss": 0.8418,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999836675942354e-05,
+      "loss": 0.8242,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998345888126613e-05,
+      "loss": 0.8643,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998324884328174e-05,
+      "loss": 0.8301,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999830374802851e-05,
+      "loss": 0.8477,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998282479227886e-05,
+      "loss": 0.8398,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998261077926595e-05,
+      "loss": 0.8555,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998239544124917e-05,
+      "loss": 0.8457,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998217877823135e-05,
+      "loss": 0.8535,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999819607902154e-05,
+      "loss": 0.8252,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998174147720418e-05,
+      "loss": 0.8467,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998152083920063e-05,
+      "loss": 0.8896,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998129887620767e-05,
+      "loss": 0.833,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998107558822818e-05,
+      "loss": 0.8672,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999808509752652e-05,
+      "loss": 0.8662,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999806250373217e-05,
+      "loss": 0.8301,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999803977744006e-05,
+      "loss": 0.8281,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998016918650495e-05,
+      "loss": 0.8389,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999799392736378e-05,
+      "loss": 0.8564,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997970803580223e-05,
+      "loss": 0.8584,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999794754730012e-05,
+      "loss": 0.8115,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997924158523788e-05,
+      "loss": 0.8486,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997900637251533e-05,
+      "loss": 0.8398,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997876983483665e-05,
+      "loss": 0.833,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997853197220504e-05,
+      "loss": 0.8477,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997829278462362e-05,
+      "loss": 0.8193,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999780522720955e-05,
+      "loss": 0.8252,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997781043462397e-05,
+      "loss": 0.8672,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997756727221218e-05,
+      "loss": 0.8125,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997732278486338e-05,
+      "loss": 0.8652,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997707697258077e-05,
+      "loss": 0.834,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997682983536763e-05,
+      "loss": 0.8281,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999765813732272e-05,
+      "loss": 0.8076,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997633158616284e-05,
+      "loss": 0.835,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997608047417777e-05,
+      "loss": 0.8486,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997582803727545e-05,
+      "loss": 0.8662,
+      "step": 1036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999755742754591e-05,
+      "loss": 0.8301,
+      "step": 1037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997531918873214e-05,
+      "loss": 0.8154,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997506277709793e-05,
+      "loss": 0.834,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999748050405599e-05,
+      "loss": 0.8389,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997454597912142e-05,
+      "loss": 0.8457,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99974285592786e-05,
+      "loss": 0.8506,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99974023881557e-05,
+      "loss": 0.8506,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997376084543792e-05,
+      "loss": 0.8477,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997349648443227e-05,
+      "loss": 0.834,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997323079854348e-05,
+      "loss": 0.8359,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999729637877752e-05,
+      "loss": 0.8438,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997269545213086e-05,
+      "loss": 0.8428,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997242579161407e-05,
+      "loss": 0.8379,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999721548062284e-05,
+      "loss": 0.8545,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999718824959774e-05,
+      "loss": 0.8174,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999716088608647e-05,
+      "loss": 0.8389,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997133390089395e-05,
+      "loss": 0.8535,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997105761606875e-05,
+      "loss": 0.8428,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999707800063928e-05,
+      "loss": 0.833,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999705010718698e-05,
+      "loss": 0.8418,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997022081250336e-05,
+      "loss": 0.8301,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996993922829727e-05,
+      "loss": 0.8281,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996965631925525e-05,
+      "loss": 0.8242,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99969372085381e-05,
+      "loss": 0.8262,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996908652667838e-05,
+      "loss": 0.8311,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999687996431511e-05,
+      "loss": 0.8389,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996851143480294e-05,
+      "loss": 0.8262,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999682219016378e-05,
+      "loss": 0.8398,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996793104365947e-05,
+      "loss": 0.8604,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999676388608718e-05,
+      "loss": 0.8506,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999673453532787e-05,
+      "loss": 0.8369,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996705052088398e-05,
+      "loss": 0.8486,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996675436369165e-05,
+      "loss": 0.8301,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996645688170557e-05,
+      "loss": 0.8457,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999661580749297e-05,
+      "loss": 0.8535,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99965857943368e-05,
+      "loss": 0.8291,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996555648702444e-05,
+      "loss": 0.8408,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99965253705903e-05,
+      "loss": 0.833,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996494960000774e-05,
+      "loss": 0.8486,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996464416934268e-05,
+      "loss": 0.8525,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999643374139118e-05,
+      "loss": 0.8457,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996402933371923e-05,
+      "loss": 0.875,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996371992876908e-05,
+      "loss": 0.877,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996340919906536e-05,
+      "loss": 0.8379,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996309714461227e-05,
+      "loss": 0.8076,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999627837654139e-05,
+      "loss": 0.8281,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996246906147437e-05,
+      "loss": 0.8477,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996215303279792e-05,
+      "loss": 0.8408,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996183567938874e-05,
+      "loss": 0.8193,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99961517001251e-05,
+      "loss": 0.8057,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996119699838894e-05,
+      "loss": 0.8477,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999608756708068e-05,
+      "loss": 0.8643,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996055301850877e-05,
+      "loss": 0.8086,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996022904149927e-05,
+      "loss": 0.8252,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995990373978245e-05,
+      "loss": 0.876,
+      "step": 1091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995957711336272e-05,
+      "loss": 0.8496,
+      "step": 1092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995924916224435e-05,
+      "loss": 0.8428,
+      "step": 1093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995891988643173e-05,
+      "loss": 0.8252,
+      "step": 1094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995858928592917e-05,
+      "loss": 0.8447,
+      "step": 1095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999582573607411e-05,
+      "loss": 0.8389,
+      "step": 1096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999579241108719e-05,
+      "loss": 0.8223,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995758953632597e-05,
+      "loss": 0.8242,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995725363710775e-05,
+      "loss": 0.8291,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995691641322174e-05,
+      "loss": 0.8213,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995657786467237e-05,
+      "loss": 0.833,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995623799146412e-05,
+      "loss": 0.8447,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995589679360148e-05,
+      "loss": 0.8594,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995555427108902e-05,
+      "loss": 0.8369,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995521042393124e-05,
+      "loss": 0.8369,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999548652521327e-05,
+      "loss": 0.8535,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995451875569797e-05,
+      "loss": 0.8262,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995417093463172e-05,
+      "loss": 0.8184,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995382178893842e-05,
+      "loss": 0.8516,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999534713186228e-05,
+      "loss": 0.8555,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999531195236895e-05,
+      "loss": 0.8359,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995276640414312e-05,
+      "loss": 0.8369,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995241195998843e-05,
+      "loss": 0.8408,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995205619123003e-05,
+      "loss": 0.8252,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999516990978727e-05,
+      "loss": 0.8594,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995134067992118e-05,
+      "loss": 0.8428,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995098093738015e-05,
+      "loss": 0.8379,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999506198702544e-05,
+      "loss": 0.8164,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999502574785488e-05,
+      "loss": 0.8193,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994989376226804e-05,
+      "loss": 0.8428,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99949528721417e-05,
+      "loss": 0.8223,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999491623560005e-05,
+      "loss": 0.8291,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994879466602343e-05,
+      "loss": 0.8271,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999484256514906e-05,
+      "loss": 0.8682,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994805531240695e-05,
+      "loss": 0.8301,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994768364877734e-05,
+      "loss": 0.8506,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994731066060677e-05,
+      "loss": 0.8125,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994693634790012e-05,
+      "loss": 0.8467,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994656071066237e-05,
+      "loss": 0.8076,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994618374889853e-05,
+      "loss": 0.8379,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999458054626135e-05,
+      "loss": 0.8545,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999454258518124e-05,
+      "loss": 0.8311,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994504491650022e-05,
+      "loss": 0.8506,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99944662656682e-05,
+      "loss": 0.8447,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999442790723628e-05,
+      "loss": 0.8418,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994389416354772e-05,
+      "loss": 0.8418,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994350793024188e-05,
+      "loss": 0.8428,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994312037245034e-05,
+      "loss": 0.8242,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994273149017828e-05,
+      "loss": 0.8408,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994234128343085e-05,
+      "loss": 0.8174,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999419497522132e-05,
+      "loss": 0.8594,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994155689653056e-05,
+      "loss": 0.8486,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994116271638814e-05,
+      "loss": 0.8213,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994076721179108e-05,
+      "loss": 0.8525,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994037038274468e-05,
+      "loss": 0.8291,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999399722292542e-05,
+      "loss": 0.8555,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999395727513249e-05,
+      "loss": 0.8271,
+      "step": 1147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999391719489621e-05,
+      "loss": 0.8174,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993876982217115e-05,
+      "loss": 0.8682,
+      "step": 1149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993836637095726e-05,
+      "loss": 0.8545,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993796159532585e-05,
+      "loss": 0.8428,
+      "step": 1151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999375554952823e-05,
+      "loss": 0.8145,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993714807083197e-05,
+      "loss": 0.8398,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999367393219802e-05,
+      "loss": 0.834,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999363292487325e-05,
+      "loss": 0.8379,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999359178510943e-05,
+      "loss": 0.8164,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993550512907097e-05,
+      "loss": 0.8555,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993509108266805e-05,
+      "loss": 0.8389,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993467571189106e-05,
+      "loss": 0.8359,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999342590167454e-05,
+      "loss": 0.8408,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993384099723662e-05,
+      "loss": 0.8271,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993342165337032e-05,
+      "loss": 0.8242,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993300098515204e-05,
+      "loss": 0.8418,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999325789925873e-05,
+      "loss": 0.8477,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993215567568173e-05,
+      "loss": 0.8428,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99931731034441e-05,
+      "loss": 0.8398,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993130506887062e-05,
+      "loss": 0.8447,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993087777897628e-05,
+      "loss": 0.8584,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993044916476368e-05,
+      "loss": 0.8213,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993001922623847e-05,
+      "loss": 0.8232,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992958796340636e-05,
+      "loss": 0.8643,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992915537627303e-05,
+      "loss": 0.8223,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992872146484424e-05,
+      "loss": 0.8438,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992828622912576e-05,
+      "loss": 0.8184,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992784966912337e-05,
+      "loss": 0.8125,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992741178484277e-05,
+      "loss": 0.8477,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992697257628982e-05,
+      "loss": 0.8525,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992653204347038e-05,
+      "loss": 0.8242,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999260901863902e-05,
+      "loss": 0.8447,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992564700505518e-05,
+      "loss": 0.835,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992520249947124e-05,
+      "loss": 0.834,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999247566696442e-05,
+      "loss": 0.8564,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992430951557996e-05,
+      "loss": 0.8262,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992386103728455e-05,
+      "loss": 0.8271,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992341123476377e-05,
+      "loss": 0.8564,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999229601080237e-05,
+      "loss": 0.834,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992250765707025e-05,
+      "loss": 0.8281,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992205388190943e-05,
+      "loss": 0.8398,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999215987825473e-05,
+      "loss": 0.8525,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992114235898984e-05,
+      "loss": 0.8105,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992068461124307e-05,
+      "loss": 0.8174,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992022553931312e-05,
+      "loss": 0.8643,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991976514320607e-05,
+      "loss": 0.835,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99919303422928e-05,
+      "loss": 0.8516,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99918840378485e-05,
+      "loss": 0.8896,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991837600988326e-05,
+      "loss": 0.8213,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991791031712893e-05,
+      "loss": 0.8125,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991744330022814e-05,
+      "loss": 0.8447,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999169749591871e-05,
+      "loss": 0.8359,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99916505294012e-05,
+      "loss": 0.8477,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999160343047091e-05,
+      "loss": 0.8242,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991556199128467e-05,
+      "loss": 0.8379,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991508835374486e-05,
+      "loss": 0.8262,
+      "step": 1203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991461339209604e-05,
+      "loss": 0.8506,
+      "step": 1204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991413710634448e-05,
+      "loss": 0.8223,
+      "step": 1205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991365949649648e-05,
+      "loss": 0.8506,
+      "step": 1206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991318056255838e-05,
+      "loss": 0.8291,
+      "step": 1207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999127003045365e-05,
+      "loss": 0.8506,
+      "step": 1208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991221872243728e-05,
+      "loss": 0.8242,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99911735816267e-05,
+      "loss": 0.8477,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991125158603214e-05,
+      "loss": 0.8379,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991076603173907e-05,
+      "loss": 0.8301,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999102791533942e-05,
+      "loss": 0.8477,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990979095100412e-05,
+      "loss": 0.835,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990930142457515e-05,
+      "loss": 0.832,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999088105741138e-05,
+      "loss": 0.8359,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990831839962668e-05,
+      "loss": 0.8135,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999078249011202e-05,
+      "loss": 0.8086,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999073300786009e-05,
+      "loss": 0.7969,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990683393207546e-05,
+      "loss": 0.8271,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999063364615503e-05,
+      "loss": 0.8105,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999058376670321e-05,
+      "loss": 0.8584,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999053375485275e-05,
+      "loss": 0.8496,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990483610604305e-05,
+      "loss": 0.8486,
+      "step": 1224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999043333395854e-05,
+      "loss": 0.8408,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990382924916128e-05,
+      "loss": 0.8496,
+      "step": 1226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999033238347773e-05,
+      "loss": 0.8438,
+      "step": 1227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999028170964402e-05,
+      "loss": 0.8311,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999023090341567e-05,
+      "loss": 0.8389,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999017996479335e-05,
+      "loss": 0.8389,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990128893777735e-05,
+      "loss": 0.8672,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999007769036951e-05,
+      "loss": 0.8369,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990026354569338e-05,
+      "loss": 0.832,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989974886377915e-05,
+      "loss": 0.8145,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989923285795916e-05,
+      "loss": 0.873,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989871552824022e-05,
+      "loss": 0.8135,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989819687462923e-05,
+      "loss": 0.8672,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989767689713307e-05,
+      "loss": 0.8398,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998971555957586e-05,
+      "loss": 0.8506,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989663297051275e-05,
+      "loss": 0.8398,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989610902140243e-05,
+      "loss": 0.8359,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998955837484346e-05,
+      "loss": 0.8213,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989505715161623e-05,
+      "loss": 0.8389,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989452923095426e-05,
+      "loss": 0.8291,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998939999864557e-05,
+      "loss": 0.8574,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989346941812754e-05,
+      "loss": 0.8398,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998929375259769e-05,
+      "loss": 0.8691,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989240431001075e-05,
+      "loss": 0.833,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989186977023617e-05,
+      "loss": 0.8418,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998913339066603e-05,
+      "loss": 0.8232,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998907967192901e-05,
+      "loss": 0.8037,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989025820813286e-05,
+      "loss": 0.8291,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998897183731956e-05,
+      "loss": 0.8145,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988917721448553e-05,
+      "loss": 0.8369,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998886347320098e-05,
+      "loss": 0.8369,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988809092577562e-05,
+      "loss": 0.833,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988754579579016e-05,
+      "loss": 0.8164,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998869993420607e-05,
+      "loss": 0.8193,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998864515645944e-05,
+      "loss": 0.8389,
+      "step": 1259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988590246339857e-05,
+      "loss": 0.8271,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988535203848046e-05,
+      "loss": 0.832,
+      "step": 1261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988480028984745e-05,
+      "loss": 0.8516,
+      "step": 1262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988424721750675e-05,
+      "loss": 0.8525,
+      "step": 1263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988369282146575e-05,
+      "loss": 0.8574,
+      "step": 1264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988313710173174e-05,
+      "loss": 0.8428,
+      "step": 1265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988258005831213e-05,
+      "loss": 0.8418,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988202169121432e-05,
+      "loss": 0.8525,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988146200044566e-05,
+      "loss": 0.8213,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998809009860136e-05,
+      "loss": 0.8525,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988033864792554e-05,
+      "loss": 0.8262,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987977498618897e-05,
+      "loss": 0.8594,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987921000081135e-05,
+      "loss": 0.8311,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998786436918002e-05,
+      "loss": 0.8701,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987807605916296e-05,
+      "loss": 0.8301,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987750710290715e-05,
+      "loss": 0.8623,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987693682304036e-05,
+      "loss": 0.833,
+      "step": 1276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998763652195701e-05,
+      "loss": 0.8154,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99875792292504e-05,
+      "loss": 0.8359,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987521804184964e-05,
+      "loss": 0.833,
+      "step": 1279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998746424676146e-05,
+      "loss": 0.8379,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987406556980654e-05,
+      "loss": 0.8418,
+      "step": 1281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987348734843307e-05,
+      "loss": 0.8271,
+      "step": 1282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987290780350187e-05,
+      "loss": 0.8828,
+      "step": 1283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987232693502064e-05,
+      "loss": 0.8408,
+      "step": 1284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987174474299703e-05,
+      "loss": 0.8311,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998711612274388e-05,
+      "loss": 0.8242,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998705763883537e-05,
+      "loss": 0.8359,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998699902257494e-05,
+      "loss": 0.8369,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986940273963374e-05,
+      "loss": 0.8506,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998688139300145e-05,
+      "loss": 0.8389,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986822379689942e-05,
+      "loss": 0.8311,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998676323402964e-05,
+      "loss": 0.8223,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998670395602132e-05,
+      "loss": 0.8506,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998664454566578e-05,
+      "loss": 0.8447,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986585002963794e-05,
+      "loss": 0.8564,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998652532791616e-05,
+      "loss": 0.8574,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998646552052366e-05,
+      "loss": 0.8164,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986405580787096e-05,
+      "loss": 0.8428,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986345508707258e-05,
+      "loss": 0.8262,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998628530428494e-05,
+      "loss": 0.8428,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998622496752094e-05,
+      "loss": 0.8359,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986164498416067e-05,
+      "loss": 0.8232,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986103896971106e-05,
+      "loss": 0.873,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986043163186876e-05,
+      "loss": 0.8262,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985982297064174e-05,
+      "loss": 0.835,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985921298603802e-05,
+      "loss": 0.8281,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998586016780658e-05,
+      "loss": 0.834,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998579890467331e-05,
+      "loss": 0.8291,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998573750920481e-05,
+      "loss": 0.8535,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985675981401886e-05,
+      "loss": 0.8057,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985614321265355e-05,
+      "loss": 0.8232,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998555252879604e-05,
+      "loss": 0.8516,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985490603994756e-05,
+      "loss": 0.8262,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985428546862323e-05,
+      "loss": 0.8398,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985366357399566e-05,
+      "loss": 0.834,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99853040356073e-05,
+      "loss": 0.8477,
+      "step": 1316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985241581486365e-05,
+      "loss": 0.8242,
+      "step": 1317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998517899503758e-05,
+      "loss": 0.8193,
+      "step": 1318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985116276261778e-05,
+      "loss": 0.8643,
+      "step": 1319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985053425159786e-05,
+      "loss": 0.8359,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998499044173244e-05,
+      "loss": 0.8311,
+      "step": 1321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984927325980575e-05,
+      "loss": 0.8125,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984864077905026e-05,
+      "loss": 0.8174,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984800697506633e-05,
+      "loss": 0.8525,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998473718478623e-05,
+      "loss": 0.8252,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998467353974467e-05,
+      "loss": 0.8643,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984609762382785e-05,
+      "loss": 0.8086,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998454585270143e-05,
+      "loss": 0.8467,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998448181070144e-05,
+      "loss": 0.8408,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984417636383674e-05,
+      "loss": 0.8389,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984353329748977e-05,
+      "loss": 0.8613,
+      "step": 1331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984288890798206e-05,
+      "loss": 0.8438,
+      "step": 1332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984224319532212e-05,
+      "loss": 0.8135,
+      "step": 1333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998415961595185e-05,
+      "loss": 0.8359,
+      "step": 1334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998409478005798e-05,
+      "loss": 0.8408,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984029811851457e-05,
+      "loss": 0.875,
+      "step": 1336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983964711333148e-05,
+      "loss": 0.8682,
+      "step": 1337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983899478503912e-05,
+      "loss": 0.8076,
+      "step": 1338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998383411336461e-05,
+      "loss": 0.8418,
+      "step": 1339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983768615916117e-05,
+      "loss": 0.8213,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983702986159293e-05,
+      "loss": 0.8223,
+      "step": 1341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998363722409501e-05,
+      "loss": 0.8379,
+      "step": 1342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983571329724146e-05,
+      "loss": 0.8467,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983505303047565e-05,
+      "loss": 0.832,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983439144066144e-05,
+      "loss": 0.8682,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983372852780763e-05,
+      "loss": 0.8145,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983306429192298e-05,
+      "loss": 0.8164,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983239873301633e-05,
+      "loss": 0.8359,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983173185109643e-05,
+      "loss": 0.8301,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983106364617216e-05,
+      "loss": 0.8096,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983039411825242e-05,
+      "loss": 0.8604,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982972326734603e-05,
+      "loss": 0.8232,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982905109346185e-05,
+      "loss": 0.8398,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982837759660888e-05,
+      "loss": 0.8369,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982770277679595e-05,
+      "loss": 0.833,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982702663403204e-05,
+      "loss": 0.8262,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982634916832616e-05,
+      "loss": 0.8193,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998256703796872e-05,
+      "loss": 0.8174,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982499026812423e-05,
+      "loss": 0.8047,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998243088336462e-05,
+      "loss": 0.8193,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982362607626216e-05,
+      "loss": 0.8359,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982294199598124e-05,
+      "loss": 0.8467,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982225659281235e-05,
+      "loss": 0.835,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982156986676472e-05,
+      "loss": 0.8428,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982088181784736e-05,
+      "loss": 0.834,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998201924460694e-05,
+      "loss": 0.8477,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981950175144007e-05,
+      "loss": 0.832,
+      "step": 1367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998188097339684e-05,
+      "loss": 0.8115,
+      "step": 1368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998181163936636e-05,
+      "loss": 0.8301,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981742173053487e-05,
+      "loss": 0.8184,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998167257445914e-05,
+      "loss": 0.8467,
+      "step": 1371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981602843584243e-05,
+      "loss": 0.8291,
+      "step": 1372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981532980429726e-05,
+      "loss": 0.8008,
+      "step": 1373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981462984996503e-05,
+      "loss": 0.8301,
+      "step": 1374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981392857285505e-05,
+      "loss": 0.8379,
+      "step": 1375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981322597297668e-05,
+      "loss": 0.8174,
+      "step": 1376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981252205033917e-05,
+      "loss": 0.835,
+      "step": 1377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981181680495186e-05,
+      "loss": 0.8496,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998111102368241e-05,
+      "loss": 0.8232,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998104023459653e-05,
+      "loss": 0.8359,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980969313238472e-05,
+      "loss": 0.8281,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980898259609186e-05,
+      "loss": 0.8477,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998082707370961e-05,
+      "loss": 0.8223,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980755755540694e-05,
+      "loss": 0.8594,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998068430510337e-05,
+      "loss": 0.8428,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99806127223986e-05,
+      "loss": 0.8193,
+      "step": 1386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980541007427322e-05,
+      "loss": 0.835,
+      "step": 1387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980469160190486e-05,
+      "loss": 0.8496,
+      "step": 1388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980397180689053e-05,
+      "loss": 0.8174,
+      "step": 1389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998032506892397e-05,
+      "loss": 0.8262,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980252824896193e-05,
+      "loss": 0.8779,
+      "step": 1391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998018044860668e-05,
+      "loss": 0.8359,
+      "step": 1392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980107940056394e-05,
+      "loss": 0.833,
+      "step": 1393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998003529924629e-05,
+      "loss": 0.8281,
+      "step": 1394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979962526177332e-05,
+      "loss": 0.8242,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979889620850486e-05,
+      "loss": 0.8604,
+      "step": 1396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979816583266723e-05,
+      "loss": 0.8438,
+      "step": 1397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979743413427002e-05,
+      "loss": 0.8213,
+      "step": 1398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979670111332295e-05,
+      "loss": 0.8398,
+      "step": 1399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997959667698358e-05,
+      "loss": 0.8281,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979523110381817e-05,
+      "loss": 0.8203,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979449411527997e-05,
+      "loss": 0.832,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979375580423085e-05,
+      "loss": 0.8389,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979301617068062e-05,
+      "loss": 0.8291,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979227521463912e-05,
+      "loss": 0.8564,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979153293611616e-05,
+      "loss": 0.832,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997907893351215e-05,
+      "loss": 0.8574,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997900444116651e-05,
+      "loss": 0.8359,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978929816575677e-05,
+      "loss": 0.8164,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978855059740644e-05,
+      "loss": 0.8271,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997878017066239e-05,
+      "loss": 0.8203,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978705149341927e-05,
+      "loss": 0.835,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978629995780235e-05,
+      "loss": 0.7998,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978554709978317e-05,
+      "loss": 0.8457,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978479291937164e-05,
+      "loss": 0.8369,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978403741657783e-05,
+      "loss": 0.832,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978328059141167e-05,
+      "loss": 0.8047,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978252244388325e-05,
+      "loss": 0.8135,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997817629740026e-05,
+      "loss": 0.8398,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997810021817798e-05,
+      "loss": 0.8408,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978024006722487e-05,
+      "loss": 0.8291,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977947663034798e-05,
+      "loss": 0.8311,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977871187115925e-05,
+      "loss": 0.8232,
+      "step": 1423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977794578966875e-05,
+      "loss": 0.8447,
+      "step": 1424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997771783858867e-05,
+      "loss": 0.835,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977640965982322e-05,
+      "loss": 0.8467,
+      "step": 1426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977563961148853e-05,
+      "loss": 0.8262,
+      "step": 1427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977486824089283e-05,
+      "loss": 0.8135,
+      "step": 1428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977409554804633e-05,
+      "loss": 0.8223,
+      "step": 1429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977332153295928e-05,
+      "loss": 0.8506,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977254619564194e-05,
+      "loss": 0.8301,
+      "step": 1431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977176953610455e-05,
+      "loss": 0.8506,
+      "step": 1432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977099155435743e-05,
+      "loss": 0.8486,
+      "step": 1433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977021225041092e-05,
+      "loss": 0.8438,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976943162427532e-05,
+      "loss": 0.8311,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976864967596093e-05,
+      "loss": 0.8613,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997678664054782e-05,
+      "loss": 0.8486,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997670818128375e-05,
+      "loss": 0.8301,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997662958980491e-05,
+      "loss": 0.8242,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976550866112355e-05,
+      "loss": 0.834,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976472010207127e-05,
+      "loss": 0.8291,
+      "step": 1441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997639302209027e-05,
+      "loss": 0.8311,
+      "step": 1442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976313901762822e-05,
+      "loss": 0.8408,
+      "step": 1443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997623464922584e-05,
+      "loss": 0.8145,
+      "step": 1444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976155264480377e-05,
+      "loss": 0.8203,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976075747527476e-05,
+      "loss": 0.8408,
+      "step": 1446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99759960983682e-05,
+      "loss": 0.8428,
+      "step": 1447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99759163170036e-05,
+      "loss": 0.8193,
+      "step": 1448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975836403434733e-05,
+      "loss": 0.833,
+      "step": 1449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975756357662656e-05,
+      "loss": 0.8184,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975676179688435e-05,
+      "loss": 0.8662,
+      "step": 1451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975595869513126e-05,
+      "loss": 0.8408,
+      "step": 1452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99755154271378e-05,
+      "loss": 0.8467,
+      "step": 1453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997543485256352e-05,
+      "loss": 0.8613,
+      "step": 1454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975354145791354e-05,
+      "loss": 0.8525,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975273306822372e-05,
+      "loss": 0.8037,
+      "step": 1456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975192335657645e-05,
+      "loss": 0.8311,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975111232298245e-05,
+      "loss": 0.8154,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997502999674525e-05,
+      "loss": 0.8604,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997494862899973e-05,
+      "loss": 0.8379,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974867129062773e-05,
+      "loss": 0.835,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997478549693545e-05,
+      "loss": 0.835,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974703732618846e-05,
+      "loss": 0.8779,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974621836114046e-05,
+      "loss": 0.8125,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974539807422136e-05,
+      "loss": 0.834,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974457646544198e-05,
+      "loss": 0.8564,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974375353481328e-05,
+      "loss": 0.8164,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974292928234612e-05,
+      "loss": 0.8516,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974210370805142e-05,
+      "loss": 0.8408,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974127681194013e-05,
+      "loss": 0.832,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974044859402324e-05,
+      "loss": 0.8154,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997396190543117e-05,
+      "loss": 0.8291,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973878819281646e-05,
+      "loss": 0.8555,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973795600954858e-05,
+      "loss": 0.8301,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997371225045191e-05,
+      "loss": 0.8252,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973628767773904e-05,
+      "loss": 0.8428,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973545152921943e-05,
+      "loss": 0.8525,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973461405897143e-05,
+      "loss": 0.832,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997337752670061e-05,
+      "loss": 0.8125,
+      "step": 1479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973293515333454e-05,
+      "loss": 0.8369,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973209371796793e-05,
+      "loss": 0.8154,
+      "step": 1481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973125096091733e-05,
+      "loss": 0.8252,
+      "step": 1482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973040688219402e-05,
+      "loss": 0.8213,
+      "step": 1483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997295614818091e-05,
+      "loss": 0.8359,
+      "step": 1484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997287147597738e-05,
+      "loss": 0.8369,
+      "step": 1485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997278667160994e-05,
+      "loss": 0.8721,
+      "step": 1486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972701735079704e-05,
+      "loss": 0.8281,
+      "step": 1487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972616666387802e-05,
+      "loss": 0.7949,
+      "step": 1488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972531465535363e-05,
+      "loss": 0.8477,
+      "step": 1489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972446132523517e-05,
+      "loss": 0.8438,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997236066735339e-05,
+      "loss": 0.8154,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972275070026115e-05,
+      "loss": 0.8184,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972189340542832e-05,
+      "loss": 0.8223,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972103478904673e-05,
+      "loss": 0.8057,
+      "step": 1494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972017485112775e-05,
+      "loss": 0.8135,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997193135916828e-05,
+      "loss": 0.8545,
+      "step": 1496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997184510107233e-05,
+      "loss": 0.8467,
+      "step": 1497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971758710826064e-05,
+      "loss": 0.8076,
+      "step": 1498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997167218843063e-05,
+      "loss": 0.834,
+      "step": 1499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971585533887173e-05,
+      "loss": 0.8398,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971498747196848e-05,
+      "loss": 0.8535,
+      "step": 1501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971411828360794e-05,
+      "loss": 0.8066,
+      "step": 1502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997132477738017e-05,
+      "loss": 0.8135,
+      "step": 1503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997123759425613e-05,
+      "loss": 0.8145,
+      "step": 1504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997115027898983e-05,
+      "loss": 0.8457,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997106283158242e-05,
+      "loss": 0.8105,
+      "step": 1506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970975252035064e-05,
+      "loss": 0.8486,
+      "step": 1507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997088754034892e-05,
+      "loss": 0.8105,
+      "step": 1508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970799696525156e-05,
+      "loss": 0.833,
+      "step": 1509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970711720564933e-05,
+      "loss": 0.8262,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970623612469417e-05,
+      "loss": 0.832,
+      "step": 1511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970535372239773e-05,
+      "loss": 0.8193,
+      "step": 1512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997044699987717e-05,
+      "loss": 0.8398,
+      "step": 1513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997035849538279e-05,
+      "loss": 0.8252,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997026985875779e-05,
+      "loss": 0.832,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970181090003355e-05,
+      "loss": 0.8115,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997009218912066e-05,
+      "loss": 0.8125,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997000315611088e-05,
+      "loss": 0.8252,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969913990975198e-05,
+      "loss": 0.8281,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969824693714792e-05,
+      "loss": 0.8076,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996973526433085e-05,
+      "loss": 0.8613,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969645702824557e-05,
+      "loss": 0.833,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969556009197096e-05,
+      "loss": 0.8379,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969466183449657e-05,
+      "loss": 0.8232,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969376225583427e-05,
+      "loss": 0.8203,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969286135599608e-05,
+      "loss": 0.8223,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969195913499387e-05,
+      "loss": 0.8086,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969105559283958e-05,
+      "loss": 0.8477,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996901507295452e-05,
+      "loss": 0.793,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968924454512276e-05,
+      "loss": 0.8203,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968833703958425e-05,
+      "loss": 0.8125,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968742821294166e-05,
+      "loss": 0.834,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968651806520707e-05,
+      "loss": 0.8291,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968560659639252e-05,
+      "loss": 0.8281,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968469380651015e-05,
+      "loss": 0.8516,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968377969557195e-05,
+      "loss": 0.8135,
+      "step": 1536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968286426359013e-05,
+      "loss": 0.8252,
+      "step": 1537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968194751057677e-05,
+      "loss": 0.8047,
+      "step": 1538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99681029436544e-05,
+      "loss": 0.8203,
+      "step": 1539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996801100415041e-05,
+      "loss": 0.8164,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967918932546912e-05,
+      "loss": 0.8203,
+      "step": 1541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967826728845132e-05,
+      "loss": 0.8408,
+      "step": 1542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996773439304629e-05,
+      "loss": 0.8213,
+      "step": 1543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967641925151615e-05,
+      "loss": 0.8066,
+      "step": 1544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967549325162327e-05,
+      "loss": 0.8252,
+      "step": 1545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967456593079654e-05,
+      "loss": 0.8037,
+      "step": 1546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967363728904824e-05,
+      "loss": 0.834,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967270732639073e-05,
+      "loss": 0.8184,
+      "step": 1548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996717760428363e-05,
+      "loss": 0.833,
+      "step": 1549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967084343839723e-05,
+      "loss": 0.8301,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99669909513086e-05,
+      "loss": 0.833,
+      "step": 1551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996689742669149e-05,
+      "loss": 0.8516,
+      "step": 1552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966803769989637e-05,
+      "loss": 0.833,
+      "step": 1553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996670998120428e-05,
+      "loss": 0.8203,
+      "step": 1554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966616060336657e-05,
+      "loss": 0.8262,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966522007388024e-05,
+      "loss": 0.8486,
+      "step": 1556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966427822359616e-05,
+      "loss": 0.8135,
+      "step": 1557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996633350525269e-05,
+      "loss": 0.8418,
+      "step": 1558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966239056068487e-05,
+      "loss": 0.8408,
+      "step": 1559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966144474808268e-05,
+      "loss": 0.8184,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966049761473284e-05,
+      "loss": 0.8223,
+      "step": 1561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965954916064783e-05,
+      "loss": 0.8145,
+      "step": 1562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965859938584035e-05,
+      "loss": 0.8105,
+      "step": 1563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965764829032287e-05,
+      "loss": 0.8516,
+      "step": 1564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965669587410803e-05,
+      "loss": 0.8164,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965574213720846e-05,
+      "loss": 0.8359,
+      "step": 1566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965478707963677e-05,
+      "loss": 0.8301,
+      "step": 1567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965383070140566e-05,
+      "loss": 0.7988,
+      "step": 1568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996528730025278e-05,
+      "loss": 0.8311,
+      "step": 1569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996519139830159e-05,
+      "loss": 0.8174,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996509536428826e-05,
+      "loss": 0.7998,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964999198214065e-05,
+      "loss": 0.8291,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996490290008028e-05,
+      "loss": 0.834,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964806469888186e-05,
+      "loss": 0.8105,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964709907639056e-05,
+      "loss": 0.8369,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996461321333417e-05,
+      "loss": 0.8125,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996451638697481e-05,
+      "loss": 0.833,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996441942856226e-05,
+      "loss": 0.8271,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964322338097803e-05,
+      "loss": 0.8252,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964225115582727e-05,
+      "loss": 0.8115,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996412776101832e-05,
+      "loss": 0.833,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964030274405877e-05,
+      "loss": 0.8262,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996393265574668e-05,
+      "loss": 0.8428,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996383490504203e-05,
+      "loss": 0.835,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996373702229322e-05,
+      "loss": 0.8301,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996363900750155e-05,
+      "loss": 0.7988,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963540860668313e-05,
+      "loss": 0.8037,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963442581794816e-05,
+      "loss": 0.8135,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963344170882357e-05,
+      "loss": 0.8408,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963245627932245e-05,
+      "loss": 0.8135,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996314695294578e-05,
+      "loss": 0.8311,
+      "step": 1591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963048145924274e-05,
+      "loss": 0.8242,
+      "step": 1592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962949206869033e-05,
+      "loss": 0.832,
+      "step": 1593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996285013578137e-05,
+      "loss": 0.8213,
+      "step": 1594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99627509326626e-05,
+      "loss": 0.8516,
+      "step": 1595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962651597514033e-05,
+      "loss": 0.8164,
+      "step": 1596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996255213033699e-05,
+      "loss": 0.8428,
+      "step": 1597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962452531132785e-05,
+      "loss": 0.873,
+      "step": 1598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962352799902746e-05,
+      "loss": 0.8223,
+      "step": 1599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996225293664818e-05,
+      "loss": 0.8145,
+      "step": 1600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962152941370427e-05,
+      "loss": 0.8164,
+      "step": 1601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99620528140708e-05,
+      "loss": 0.834,
+      "step": 1602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9961952554750634e-05,
+      "loss": 0.8242,
+      "step": 1603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961852163411254e-05,
+      "loss": 0.8438,
+      "step": 1604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961751640053987e-05,
+      "loss": 0.8213,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996165098468017e-05,
+      "loss": 0.832,
+      "step": 1606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961550197291136e-05,
+      "loss": 0.8125,
+      "step": 1607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961449277888222e-05,
+      "loss": 0.8232,
+      "step": 1608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996134822647276e-05,
+      "loss": 0.8086,
+      "step": 1609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961247043046096e-05,
+      "loss": 0.8447,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961145727609566e-05,
+      "loss": 0.8408,
+      "step": 1611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961044280164514e-05,
+      "loss": 0.8281,
+      "step": 1612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960942700712287e-05,
+      "loss": 0.8203,
+      "step": 1613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960840989254228e-05,
+      "loss": 0.8428,
+      "step": 1614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960739145791687e-05,
+      "loss": 0.8057,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996063717032601e-05,
+      "loss": 0.8145,
+      "step": 1616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996053506285855e-05,
+      "loss": 0.8252,
+      "step": 1617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960432823390662e-05,
+      "loss": 0.8301,
+      "step": 1618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960330451923704e-05,
+      "loss": 0.8379,
+      "step": 1619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960227948459027e-05,
+      "loss": 0.8232,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996012531299799e-05,
+      "loss": 0.8291,
+      "step": 1621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960022545541953e-05,
+      "loss": 0.8555,
+      "step": 1622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995991964609228e-05,
+      "loss": 0.8418,
+      "step": 1623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959816614650334e-05,
+      "loss": 0.833,
+      "step": 1624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995971345121748e-05,
+      "loss": 0.8203,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959610155795087e-05,
+      "loss": 0.8389,
+      "step": 1626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959506728384523e-05,
+      "loss": 0.8428,
+      "step": 1627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959403168987158e-05,
+      "loss": 0.8555,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995929947760436e-05,
+      "loss": 0.834,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959195654237515e-05,
+      "loss": 0.8701,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959091698887985e-05,
+      "loss": 0.8457,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958987611557155e-05,
+      "loss": 0.8594,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958883392246406e-05,
+      "loss": 0.8574,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958779040957115e-05,
+      "loss": 0.8164,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958674557690668e-05,
+      "loss": 0.8262,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958569942448445e-05,
+      "loss": 0.8438,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958465195231834e-05,
+      "loss": 0.8271,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958360316042232e-05,
+      "loss": 0.8496,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958255304881017e-05,
+      "loss": 0.8086,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995815016174959e-05,
+      "loss": 0.8223,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958044886649333e-05,
+      "loss": 0.8271,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995793947958165e-05,
+      "loss": 0.8359,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957833940547937e-05,
+      "loss": 0.8379,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995772826954959e-05,
+      "loss": 0.8398,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995762246658801e-05,
+      "loss": 0.8379,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957516531664604e-05,
+      "loss": 0.8047,
+      "step": 1646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957410464780768e-05,
+      "loss": 0.8193,
+      "step": 1647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995730426593791e-05,
+      "loss": 0.8037,
+      "step": 1648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957197935137443e-05,
+      "loss": 0.8359,
+      "step": 1649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957091472380772e-05,
+      "loss": 0.8164,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956984877669304e-05,
+      "loss": 0.8184,
+      "step": 1651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956878151004458e-05,
+      "loss": 0.8369,
+      "step": 1652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956771292387644e-05,
+      "loss": 0.8223,
+      "step": 1653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956664301820282e-05,
+      "loss": 0.834,
+      "step": 1654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956557179303787e-05,
+      "loss": 0.8164,
+      "step": 1655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995644992483958e-05,
+      "loss": 0.8252,
+      "step": 1656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956342538429082e-05,
+      "loss": 0.8047,
+      "step": 1657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956235020073715e-05,
+      "loss": 0.8486,
+      "step": 1658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956127369774904e-05,
+      "loss": 0.8047,
+      "step": 1659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9956019587534075e-05,
+      "loss": 0.8369,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955911673352663e-05,
+      "loss": 0.8213,
+      "step": 1661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995580362723209e-05,
+      "loss": 0.8105,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955695449173792e-05,
+      "loss": 0.8164,
+      "step": 1663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99555871391792e-05,
+      "loss": 0.8262,
+      "step": 1664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955478697249752e-05,
+      "loss": 0.8203,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955370123386883e-05,
+      "loss": 0.8203,
+      "step": 1666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955261417592032e-05,
+      "loss": 0.7988,
+      "step": 1667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995515257986664e-05,
+      "loss": 0.8115,
+      "step": 1668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995504361021215e-05,
+      "loss": 0.7988,
+      "step": 1669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954934508630007e-05,
+      "loss": 0.7939,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954825275121654e-05,
+      "loss": 0.8564,
+      "step": 1671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995471590968854e-05,
+      "loss": 0.8174,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954606412332115e-05,
+      "loss": 0.8291,
+      "step": 1673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995449678305383e-05,
+      "loss": 0.8193,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995438702185514e-05,
+      "loss": 0.793,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954277128737495e-05,
+      "loss": 0.8164,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995416710370235e-05,
+      "loss": 0.833,
+      "step": 1677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995405694675117e-05,
+      "loss": 0.8242,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995394665788541e-05,
+      "loss": 0.7998,
+      "step": 1679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953836237106535e-05,
+      "loss": 0.8174,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953725684416005e-05,
+      "loss": 0.8076,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995361499981529e-05,
+      "loss": 0.8232,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953504183305847e-05,
+      "loss": 0.8174,
+      "step": 1683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953393234889153e-05,
+      "loss": 0.8281,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995328215456668e-05,
+      "loss": 0.8496,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953170942339894e-05,
+      "loss": 0.8174,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953059598210268e-05,
+      "loss": 0.835,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995294812217928e-05,
+      "loss": 0.8545,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952836514248412e-05,
+      "loss": 0.8105,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952724774419135e-05,
+      "loss": 0.8271,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952612902692938e-05,
+      "loss": 0.834,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99525008990713e-05,
+      "loss": 0.8242,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99523887635557e-05,
+      "loss": 0.8213,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952276496147628e-05,
+      "loss": 0.8145,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952164096848578e-05,
+      "loss": 0.8271,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995205156566003e-05,
+      "loss": 0.8086,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995193890258348e-05,
+      "loss": 0.8691,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951826107620424e-05,
+      "loss": 0.8574,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951713180772348e-05,
+      "loss": 0.8232,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951600122040758e-05,
+      "loss": 0.8164,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951486931427145e-05,
+      "loss": 0.8281,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951373608933012e-05,
+      "loss": 0.832,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951260154559864e-05,
+      "loss": 0.8096,
+      "step": 1703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951146568309202e-05,
+      "loss": 0.8125,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951032850182524e-05,
+      "loss": 0.8115,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995091900018135e-05,
+      "loss": 0.8193,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995080501830718e-05,
+      "loss": 0.8525,
+      "step": 1707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995069090456153e-05,
+      "loss": 0.8223,
+      "step": 1708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950576658945903e-05,
+      "loss": 0.8252,
+      "step": 1709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950462281461826e-05,
+      "loss": 0.8584,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950347772110806e-05,
+      "loss": 0.8301,
+      "step": 1711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995023313089436e-05,
+      "loss": 0.8174,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995011835781401e-05,
+      "loss": 0.875,
+      "step": 1713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995000345287128e-05,
+      "loss": 0.8467,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.994988841606769e-05,
+      "loss": 0.8018,
+      "step": 1715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.994977324740476e-05,
+      "loss": 0.8311,
+      "step": 1716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949657946884026e-05,
+      "loss": 0.8066,
+      "step": 1717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949542514507007e-05,
+      "loss": 0.8271,
+      "step": 1718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949426950275235e-05,
+      "loss": 0.8311,
+      "step": 1719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949311254190243e-05,
+      "loss": 0.8096,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949195426253567e-05,
+      "loss": 0.8174,
+      "step": 1721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949079466466737e-05,
+      "loss": 0.8096,
+      "step": 1722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994896337483129e-05,
+      "loss": 0.8291,
+      "step": 1723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994884715134877e-05,
+      "loss": 0.8164,
+      "step": 1724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994873079602071e-05,
+      "loss": 0.8027,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948614308848656e-05,
+      "loss": 0.8262,
+      "step": 1726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994849768983415e-05,
+      "loss": 0.8164,
+      "step": 1727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948380938978742e-05,
+      "loss": 0.8096,
+      "step": 1728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948264056283976e-05,
+      "loss": 0.8477,
+      "step": 1729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948147041751394e-05,
+      "loss": 0.8252,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948029895382562e-05,
+      "loss": 0.834,
+      "step": 1731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994791261717902e-05,
+      "loss": 0.8398,
+      "step": 1732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947795207142326e-05,
+      "loss": 0.8408,
+      "step": 1733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947677665274035e-05,
+      "loss": 0.8398,
+      "step": 1734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947559991575708e-05,
+      "loss": 0.8164,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947442186048903e-05,
+      "loss": 0.8184,
+      "step": 1736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947324248695175e-05,
+      "loss": 0.8115,
+      "step": 1737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947206179516098e-05,
+      "loss": 0.8008,
+      "step": 1738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994708797851323e-05,
+      "loss": 0.8223,
+      "step": 1739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946969645688136e-05,
+      "loss": 0.8086,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994685118104239e-05,
+      "loss": 0.8408,
+      "step": 1741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946732584577557e-05,
+      "loss": 0.8379,
+      "step": 1742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946613856295207e-05,
+      "loss": 0.8193,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946494996196923e-05,
+      "loss": 0.832,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946376004284273e-05,
+      "loss": 0.8389,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994625688055883e-05,
+      "loss": 0.8232,
+      "step": 1746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946137625022183e-05,
+      "loss": 0.8291,
+      "step": 1747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99460182376759e-05,
+      "loss": 0.8438,
+      "step": 1748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945898718521575e-05,
+      "loss": 0.8184,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994577906756079e-05,
+      "loss": 0.8213,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945659284795122e-05,
+      "loss": 0.8018,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945539370226166e-05,
+      "loss": 0.8203,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994541932385551e-05,
+      "loss": 0.8008,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945299145684745e-05,
+      "loss": 0.8242,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994517883571546e-05,
+      "loss": 0.834,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945058393949252e-05,
+      "loss": 0.8291,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994493782038772e-05,
+      "loss": 0.8398,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944817115032457e-05,
+      "loss": 0.8262,
+      "step": 1758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944696277885067e-05,
+      "loss": 0.8457,
+      "step": 1759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944575308947147e-05,
+      "loss": 0.8252,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.99444542082203e-05,
+      "loss": 0.8047,
+      "step": 1761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994433297570614e-05,
+      "loss": 0.8535,
+      "step": 1762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944211611406262e-05,
+      "loss": 0.8096,
+      "step": 1763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994409011532228e-05,
+      "loss": 0.8477,
+      "step": 1764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943968487455803e-05,
+      "loss": 0.8018,
+      "step": 1765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994384672780844e-05,
+      "loss": 0.8213,
+      "step": 1766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943724836381815e-05,
+      "loss": 0.8438,
+      "step": 1767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994360281317753e-05,
+      "loss": 0.8027,
+      "step": 1768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943480658197213e-05,
+      "loss": 0.834,
+      "step": 1769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943358371442475e-05,
+      "loss": 0.8018,
+      "step": 1770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994323595291494e-05,
+      "loss": 0.8164,
+      "step": 1771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943113402616228e-05,
+      "loss": 0.8418,
+      "step": 1772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942990720547968e-05,
+      "loss": 0.8457,
+      "step": 1773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942867906711783e-05,
+      "loss": 0.8037,
+      "step": 1774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942744961109296e-05,
+      "loss": 0.8232,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942621883742148e-05,
+      "loss": 0.8311,
+      "step": 1776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942498674611953e-05,
+      "loss": 0.8477,
+      "step": 1777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942375333720362e-05,
+      "loss": 0.8232,
+      "step": 1778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942251861068994e-05,
+      "loss": 0.8086,
+      "step": 1779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942128256659497e-05,
+      "loss": 0.8174,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942004520493507e-05,
+      "loss": 0.8213,
+      "step": 1781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994188065257266e-05,
+      "loss": 0.833,
+      "step": 1782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941756652898595e-05,
+      "loss": 0.7998,
+      "step": 1783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994163252147296e-05,
+      "loss": 0.8184,
+      "step": 1784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99415082582974e-05,
+      "loss": 0.8242,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994138386337356e-05,
+      "loss": 0.8242,
+      "step": 1786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994125933670309e-05,
+      "loss": 0.7871,
+      "step": 1787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994113467828764e-05,
+      "loss": 0.8525,
+      "step": 1788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941009888128864e-05,
+      "loss": 0.8193,
+      "step": 1789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994088496622841e-05,
+      "loss": 0.834,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994075991258794e-05,
+      "loss": 0.8213,
+      "step": 1791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940634727209105e-05,
+      "loss": 0.8232,
+      "step": 1792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994050941009357e-05,
+      "loss": 0.8164,
+      "step": 1793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994038396124299e-05,
+      "loss": 0.8408,
+      "step": 1794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940258380659034e-05,
+      "loss": 0.8281,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940132668343358e-05,
+      "loss": 0.8213,
+      "step": 1796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940006824297636e-05,
+      "loss": 0.8477,
+      "step": 1797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939880848523533e-05,
+      "loss": 0.8252,
+      "step": 1798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939754741022718e-05,
+      "loss": 0.8086,
+      "step": 1799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939628501796858e-05,
+      "loss": 0.8057,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939502130847634e-05,
+      "loss": 0.8008,
+      "step": 1801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939375628176716e-05,
+      "loss": 0.8066,
+      "step": 1802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993924899378578e-05,
+      "loss": 0.8496,
+      "step": 1803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939122227676507e-05,
+      "loss": 0.8184,
+      "step": 1804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938995329850574e-05,
+      "loss": 0.8115,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938868300309666e-05,
+      "loss": 0.8535,
+      "step": 1806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938741139055462e-05,
+      "loss": 0.8311,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938613846089652e-05,
+      "loss": 0.834,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993848642141392e-05,
+      "loss": 0.8066,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938358865029955e-05,
+      "loss": 0.8379,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938231176939446e-05,
+      "loss": 0.8174,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993810335714409e-05,
+      "loss": 0.8057,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937975405645575e-05,
+      "loss": 0.8232,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99378473224456e-05,
+      "loss": 0.8535,
+      "step": 1814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993771910754586e-05,
+      "loss": 0.8252,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937590760948064e-05,
+      "loss": 0.8174,
+      "step": 1816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937462282653898e-05,
+      "loss": 0.8174,
+      "step": 1817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937333672665072e-05,
+      "loss": 0.8115,
+      "step": 1818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937204930983294e-05,
+      "loss": 0.8057,
+      "step": 1819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937076057610262e-05,
+      "loss": 0.835,
+      "step": 1820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936947052547692e-05,
+      "loss": 0.8555,
+      "step": 1821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936817915797286e-05,
+      "loss": 0.8281,
+      "step": 1822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936688647360758e-05,
+      "loss": 0.8262,
+      "step": 1823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936559247239823e-05,
+      "loss": 0.835,
+      "step": 1824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936429715436198e-05,
+      "loss": 0.8281,
+      "step": 1825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993630005195159e-05,
+      "loss": 0.8145,
+      "step": 1826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993617025678773e-05,
+      "loss": 0.8486,
+      "step": 1827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936040329946333e-05,
+      "loss": 0.835,
+      "step": 1828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935910271429117e-05,
+      "loss": 0.8242,
+      "step": 1829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935780081237807e-05,
+      "loss": 0.8232,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993564975937413e-05,
+      "loss": 0.8145,
+      "step": 1831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935519305839814e-05,
+      "loss": 0.8037,
+      "step": 1832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935388720636583e-05,
+      "loss": 0.8301,
+      "step": 1833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935258003766177e-05,
+      "loss": 0.7861,
+      "step": 1834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935127155230316e-05,
+      "loss": 0.8291,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934996175030746e-05,
+      "loss": 0.834,
+      "step": 1836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934865063169195e-05,
+      "loss": 0.8076,
+      "step": 1837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934733819647404e-05,
+      "loss": 0.7979,
+      "step": 1838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934602444467108e-05,
+      "loss": 0.8076,
+      "step": 1839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993447093763005e-05,
+      "loss": 0.8184,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934339299137983e-05,
+      "loss": 0.8242,
+      "step": 1841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934207528992634e-05,
+      "loss": 0.8145,
+      "step": 1842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993407562719576e-05,
+      "loss": 0.8262,
+      "step": 1843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933943593749107e-05,
+      "loss": 0.7979,
+      "step": 1844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933811428654422e-05,
+      "loss": 0.8398,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993367913191346e-05,
+      "loss": 0.8076,
+      "step": 1846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933546703527975e-05,
+      "loss": 0.8789,
+      "step": 1847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993341414349972e-05,
+      "loss": 0.8281,
+      "step": 1848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933281451830453e-05,
+      "loss": 0.8115,
+      "step": 1849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993314862852193e-05,
+      "loss": 0.8213,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933015673575912e-05,
+      "loss": 0.8096,
+      "step": 1851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993288258699416e-05,
+      "loss": 0.8447,
+      "step": 1852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993274936877844e-05,
+      "loss": 0.8535,
+      "step": 1853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993261601893052e-05,
+      "loss": 0.8135,
+      "step": 1854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993248253745216e-05,
+      "loss": 0.8311,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932348924345134e-05,
+      "loss": 0.8184,
+      "step": 1856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932215179611212e-05,
+      "loss": 0.8096,
+      "step": 1857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932081303252167e-05,
+      "loss": 0.8389,
+      "step": 1858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993194729526977e-05,
+      "loss": 0.8262,
+      "step": 1859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.99318131556658e-05,
+      "loss": 0.8232,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931678884442033e-05,
+      "loss": 0.8154,
+      "step": 1861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931544481600248e-05,
+      "loss": 0.7988,
+      "step": 1862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993140994714223e-05,
+      "loss": 0.8389,
+      "step": 1863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993127528106976e-05,
+      "loss": 0.8154,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931140483384622e-05,
+      "loss": 0.8301,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931005554088598e-05,
+      "loss": 0.8477,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930870493183484e-05,
+      "loss": 0.8418,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930735300671066e-05,
+      "loss": 0.8105,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930599976553134e-05,
+      "loss": 0.8174,
+      "step": 1869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930464520831482e-05,
+      "loss": 0.8008,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930328933507907e-05,
+      "loss": 0.8203,
+      "step": 1871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930193214584208e-05,
+      "loss": 0.8262,
+      "step": 1872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993005736406218e-05,
+      "loss": 0.8047,
+      "step": 1873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929921381943623e-05,
+      "loss": 0.7959,
+      "step": 1874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992978526823034e-05,
+      "loss": 0.8184,
+      "step": 1875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929649022924137e-05,
+      "loss": 0.8174,
+      "step": 1876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929512646026814e-05,
+      "loss": 0.8252,
+      "step": 1877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929376137540184e-05,
+      "loss": 0.8086,
+      "step": 1878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929239497466054e-05,
+      "loss": 0.8252,
+      "step": 1879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929102725806233e-05,
+      "loss": 0.8037,
+      "step": 1880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928965822562537e-05,
+      "loss": 0.8047,
+      "step": 1881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992882878773678e-05,
+      "loss": 0.8564,
+      "step": 1882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928691621330773e-05,
+      "loss": 0.8359,
+      "step": 1883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992855432334634e-05,
+      "loss": 0.8047,
+      "step": 1884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928416893785298e-05,
+      "loss": 0.7969,
+      "step": 1885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928279332649465e-05,
+      "loss": 0.8193,
+      "step": 1886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992814163994067e-05,
+      "loss": 0.8262,
+      "step": 1887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928003815660735e-05,
+      "loss": 0.8213,
+      "step": 1888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927865859811483e-05,
+      "loss": 0.8389,
+      "step": 1889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927727772394748e-05,
+      "loss": 0.7979,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992758955341236e-05,
+      "loss": 0.8203,
+      "step": 1891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927451202866144e-05,
+      "loss": 0.8164,
+      "step": 1892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992731272075794e-05,
+      "loss": 0.8223,
+      "step": 1893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927174107089583e-05,
+      "loss": 0.834,
+      "step": 1894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927035361862907e-05,
+      "loss": 0.8301,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992689648507975e-05,
+      "loss": 0.8389,
+      "step": 1896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926757476741953e-05,
+      "loss": 0.8193,
+      "step": 1897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992661833685136e-05,
+      "loss": 0.8584,
+      "step": 1898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926479065409818e-05,
+      "loss": 0.8242,
+      "step": 1899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926339662419165e-05,
+      "loss": 0.832,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926200127881256e-05,
+      "loss": 0.8281,
+      "step": 1901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992606046179793e-05,
+      "loss": 0.8242,
+      "step": 1902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992592066417105e-05,
+      "loss": 0.8496,
+      "step": 1903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925780735002462e-05,
+      "loss": 0.8008,
+      "step": 1904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992564067429402e-05,
+      "loss": 0.8242,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925500482047583e-05,
+      "loss": 0.7871,
+      "step": 1906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925360158265007e-05,
+      "loss": 0.8252,
+      "step": 1907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925219702948153e-05,
+      "loss": 0.8184,
+      "step": 1908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992507911609888e-05,
+      "loss": 0.8457,
+      "step": 1909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924938397719052e-05,
+      "loss": 0.8203,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924797547810537e-05,
+      "loss": 0.8145,
+      "step": 1911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924656566375196e-05,
+      "loss": 0.8174,
+      "step": 1912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924515453414903e-05,
+      "loss": 0.834,
+      "step": 1913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924374208931523e-05,
+      "loss": 0.8223,
+      "step": 1914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992423283292693e-05,
+      "loss": 0.7861,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924091325403e-05,
+      "loss": 0.8301,
+      "step": 1916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923949686361603e-05,
+      "loss": 0.8184,
+      "step": 1917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923807915804623e-05,
+      "loss": 0.8184,
+      "step": 1918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923666013733932e-05,
+      "loss": 0.8232,
+      "step": 1919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923523980151415e-05,
+      "loss": 0.8184,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923381815058952e-05,
+      "loss": 0.8193,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923239518458427e-05,
+      "loss": 0.835,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923097090351728e-05,
+      "loss": 0.8242,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922954530740738e-05,
+      "loss": 0.833,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922811839627354e-05,
+      "loss": 0.8076,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992266901701346e-05,
+      "loss": 0.8301,
+      "step": 1926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992252606290095e-05,
+      "loss": 0.8359,
+      "step": 1927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922382977291722e-05,
+      "loss": 0.832,
+      "step": 1928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922239760187666e-05,
+      "loss": 0.8037,
+      "step": 1929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922096411590686e-05,
+      "loss": 0.793,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992195293150268e-05,
+      "loss": 0.8291,
+      "step": 1931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921809319925548e-05,
+      "loss": 0.8311,
+      "step": 1932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921665576861194e-05,
+      "loss": 0.8184,
+      "step": 1933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921521702311523e-05,
+      "loss": 0.8203,
+      "step": 1934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992137769627844e-05,
+      "loss": 0.8252,
+      "step": 1935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921233558763856e-05,
+      "loss": 0.8174,
+      "step": 1936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921089289769678e-05,
+      "loss": 0.793,
+      "step": 1937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920944889297823e-05,
+      "loss": 0.8291,
+      "step": 1938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920800357350204e-05,
+      "loss": 0.8506,
+      "step": 1939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992065569392873e-05,
+      "loss": 0.8115,
+      "step": 1940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920510899035323e-05,
+      "loss": 0.8174,
+      "step": 1941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920365972671902e-05,
+      "loss": 0.8398,
+      "step": 1942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.992022091484039e-05,
+      "loss": 0.8301,
+      "step": 1943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99200757255427e-05,
+      "loss": 0.8232,
+      "step": 1944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919930404780766e-05,
+      "loss": 0.8311,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919784952556506e-05,
+      "loss": 0.8242,
+      "step": 1946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919639368871856e-05,
+      "loss": 0.8154,
+      "step": 1947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919493653728738e-05,
+      "loss": 0.8369,
+      "step": 1948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991934780712909e-05,
+      "loss": 0.8057,
+      "step": 1949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919201829074836e-05,
+      "loss": 0.8242,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919055719567917e-05,
+      "loss": 0.8096,
+      "step": 1951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991890947861027e-05,
+      "loss": 0.8027,
+      "step": 1952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918763106203825e-05,
+      "loss": 0.7988,
+      "step": 1953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918616602350533e-05,
+      "loss": 0.8154,
+      "step": 1954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918469967052326e-05,
+      "loss": 0.8418,
+      "step": 1955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918323200311157e-05,
+      "loss": 0.8164,
+      "step": 1956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991817630212896e-05,
+      "loss": 0.8252,
+      "step": 1957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918029272507688e-05,
+      "loss": 0.8096,
+      "step": 1958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991788211144929e-05,
+      "loss": 0.8213,
+      "step": 1959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917734818955713e-05,
+      "loss": 0.8379,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991758739502891e-05,
+      "loss": 0.8076,
+      "step": 1961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917439839670835e-05,
+      "loss": 0.8252,
+      "step": 1962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917292152883448e-05,
+      "loss": 0.8311,
+      "step": 1963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917144334668697e-05,
+      "loss": 0.8047,
+      "step": 1964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916996385028552e-05,
+      "loss": 0.8369,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916848303964963e-05,
+      "loss": 0.8057,
+      "step": 1966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916700091479896e-05,
+      "loss": 0.8076,
+      "step": 1967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991655174757532e-05,
+      "loss": 0.8486,
+      "step": 1968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916403272253197e-05,
+      "loss": 0.8496,
+      "step": 1969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991625466551549e-05,
+      "loss": 0.8037,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991610592736418e-05,
+      "loss": 0.793,
+      "step": 1971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915957057801226e-05,
+      "loss": 0.8164,
+      "step": 1972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915808056828608e-05,
+      "loss": 0.8232,
+      "step": 1973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915658924448298e-05,
+      "loss": 0.7891,
+      "step": 1974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915509660662274e-05,
+      "loss": 0.8076,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915360265472516e-05,
+      "loss": 0.8252,
+      "step": 1976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915210738880998e-05,
+      "loss": 0.8262,
+      "step": 1977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915061080889703e-05,
+      "loss": 0.8379,
+      "step": 1978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914911291500617e-05,
+      "loss": 0.8115,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914761370715724e-05,
+      "loss": 0.8379,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914611318537013e-05,
+      "loss": 0.8447,
+      "step": 1981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991446113496647e-05,
+      "loss": 0.8232,
+      "step": 1982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914310820006085e-05,
+      "loss": 0.8086,
+      "step": 1983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914160373657848e-05,
+      "loss": 0.8135,
+      "step": 1984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991400979592376e-05,
+      "loss": 0.8115,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991385908680581e-05,
+      "loss": 0.7939,
+      "step": 1986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913708246305998e-05,
+      "loss": 0.8281,
+      "step": 1987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991355727442632e-05,
+      "loss": 0.8105,
+      "step": 1988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991340617116878e-05,
+      "loss": 0.7939,
+      "step": 1989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991325493653538e-05,
+      "loss": 0.8271,
+      "step": 1990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913103570528124e-05,
+      "loss": 0.8496,
+      "step": 1991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912952073149016e-05,
+      "loss": 0.8232,
+      "step": 1992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912800444400064e-05,
+      "loss": 0.8184,
+      "step": 1993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912648684283282e-05,
+      "loss": 0.8281,
+      "step": 1994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991249679280068e-05,
+      "loss": 0.8428,
+      "step": 1995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912344769954263e-05,
+      "loss": 0.8047,
+      "step": 1996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912192615746054e-05,
+      "loss": 0.8418,
+      "step": 1997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9912040330178065e-05,
+      "loss": 0.8037,
+      "step": 1998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911887913252315e-05,
+      "loss": 0.8047,
+      "step": 1999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911735364970827e-05,
+      "loss": 0.8105,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911582685335623e-05,
+      "loss": 0.8125,
+      "step": 2001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911429874348723e-05,
+      "loss": 0.7939,
+      "step": 2002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991127693201215e-05,
+      "loss": 0.8057,
+      "step": 2003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911123858327938e-05,
+      "loss": 0.8008,
+      "step": 2004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991097065329811e-05,
+      "loss": 0.8242,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910817316924695e-05,
+      "loss": 0.7988,
+      "step": 2006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910663849209728e-05,
+      "loss": 0.8076,
+      "step": 2007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910510250155242e-05,
+      "loss": 0.8311,
+      "step": 2008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910356519763277e-05,
+      "loss": 0.8154,
+      "step": 2009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910202658035865e-05,
+      "loss": 0.8242,
+      "step": 2010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910048664975045e-05,
+      "loss": 0.7949,
+      "step": 2011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909894540582862e-05,
+      "loss": 0.8174,
+      "step": 2012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909740284861354e-05,
+      "loss": 0.7969,
+      "step": 2013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909585897812565e-05,
+      "loss": 0.8096,
+      "step": 2014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909431379438544e-05,
+      "loss": 0.8232,
+      "step": 2015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990927672974134e-05,
+      "loss": 0.8311,
+      "step": 2016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909121948723e-05,
+      "loss": 0.832,
+      "step": 2017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908967036385573e-05,
+      "loss": 0.8145,
+      "step": 2018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908811992731113e-05,
+      "loss": 0.7998,
+      "step": 2019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990865681776168e-05,
+      "loss": 0.8135,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908501511479324e-05,
+      "loss": 0.8125,
+      "step": 2021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908346073886108e-05,
+      "loss": 0.8213,
+      "step": 2022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990819050498409e-05,
+      "loss": 0.8584,
+      "step": 2023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908034804775327e-05,
+      "loss": 0.8154,
+      "step": 2024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990787897326189e-05,
+      "loss": 0.8193,
+      "step": 2025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990772301044584e-05,
+      "loss": 0.8105,
+      "step": 2026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990756691632924e-05,
+      "loss": 0.8115,
+      "step": 2027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990741069091417e-05,
+      "loss": 0.7998,
+      "step": 2028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990725433420269e-05,
+      "loss": 0.8027,
+      "step": 2029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907097846196878e-05,
+      "loss": 0.8174,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906941226898805e-05,
+      "loss": 0.8398,
+      "step": 2031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906784476310547e-05,
+      "loss": 0.8027,
+      "step": 2032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990662759443418e-05,
+      "loss": 0.8164,
+      "step": 2033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906470581271784e-05,
+      "loss": 0.7988,
+      "step": 2034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990631343682544e-05,
+      "loss": 0.8301,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906156161097232e-05,
+      "loss": 0.8027,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905998754089242e-05,
+      "loss": 0.7979,
+      "step": 2037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990584121580356e-05,
+      "loss": 0.8018,
+      "step": 2038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905683546242266e-05,
+      "loss": 0.8232,
+      "step": 2039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990552574540746e-05,
+      "loss": 0.8213,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905367813301223e-05,
+      "loss": 0.8057,
+      "step": 2041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905209749925656e-05,
+      "loss": 0.8262,
+      "step": 2042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905051555282846e-05,
+      "loss": 0.8447,
+      "step": 2043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904893229374896e-05,
+      "loss": 0.8213,
+      "step": 2044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99047347722039e-05,
+      "loss": 0.8105,
+      "step": 2045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904576183771963e-05,
+      "loss": 0.8223,
+      "step": 2046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990441746408118e-05,
+      "loss": 0.8311,
+      "step": 2047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904258613133662e-05,
+      "loss": 0.8223,
+      "step": 2048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904099630931507e-05,
+      "loss": 0.8145,
+      "step": 2049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903940517476826e-05,
+      "loss": 0.7939,
+      "step": 2050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903781272771726e-05,
+      "loss": 0.8193,
+      "step": 2051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903621896818316e-05,
+      "loss": 0.8252,
+      "step": 2052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990346238961871e-05,
+      "loss": 0.8086,
+      "step": 2053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903302751175023e-05,
+      "loss": 0.8096,
+      "step": 2054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903142981489372e-05,
+      "loss": 0.8213,
+      "step": 2055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902983080563868e-05,
+      "loss": 0.8125,
+      "step": 2056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902823048400636e-05,
+      "loss": 0.8213,
+      "step": 2057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902662885001797e-05,
+      "loss": 0.8213,
+      "step": 2058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902502590369466e-05,
+      "loss": 0.8018,
+      "step": 2059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902342164505774e-05,
+      "loss": 0.8193,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902181607412844e-05,
+      "loss": 0.8018,
+      "step": 2061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990202091909281e-05,
+      "loss": 0.8086,
+      "step": 2062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901860099547795e-05,
+      "loss": 0.7969,
+      "step": 2063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901699148779928e-05,
+      "loss": 0.79,
+      "step": 2064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990153806679135e-05,
+      "loss": 0.8398,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901376853584188e-05,
+      "loss": 0.8223,
+      "step": 2066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901215509160584e-05,
+      "loss": 0.8047,
+      "step": 2067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901054033522676e-05,
+      "loss": 0.8174,
+      "step": 2068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99008924266726e-05,
+      "loss": 0.8008,
+      "step": 2069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99007306886125e-05,
+      "loss": 0.8164,
+      "step": 2070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990056881934452e-05,
+      "loss": 0.8193,
+      "step": 2071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900406818870802e-05,
+      "loss": 0.8135,
+      "step": 2072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900244687193497e-05,
+      "loss": 0.8203,
+      "step": 2073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990008242431475e-05,
+      "loss": 0.8486,
+      "step": 2074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989992003023672e-05,
+      "loss": 0.8174,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899757504961545e-05,
+      "loss": 0.8428,
+      "step": 2076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989959484849139e-05,
+      "loss": 0.7988,
+      "step": 2077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899432060828406e-05,
+      "loss": 0.8193,
+      "step": 2078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989926914197475e-05,
+      "loss": 0.8174,
+      "step": 2079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899106091932582e-05,
+      "loss": 0.8037,
+      "step": 2080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898942910704067e-05,
+      "loss": 0.8262,
+      "step": 2081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898779598291357e-05,
+      "loss": 0.7939,
+      "step": 2082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989861615469663e-05,
+      "loss": 0.791,
+      "step": 2083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898452579922037e-05,
+      "loss": 0.8447,
+      "step": 2084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898288873969756e-05,
+      "loss": 0.8008,
+      "step": 2085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989812503684196e-05,
+      "loss": 0.8252,
+      "step": 2086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897961068540807e-05,
+      "loss": 0.8047,
+      "step": 2087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989779696906848e-05,
+      "loss": 0.834,
+      "step": 2088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897632738427153e-05,
+      "loss": 0.8262,
+      "step": 2089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897468376618997e-05,
+      "loss": 0.8027,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897303883646195e-05,
+      "loss": 0.7979,
+      "step": 2091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897139259510925e-05,
+      "loss": 0.8457,
+      "step": 2092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896974504215368e-05,
+      "loss": 0.8174,
+      "step": 2093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896809617761708e-05,
+      "loss": 0.8008,
+      "step": 2094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896644600152136e-05,
+      "loss": 0.8242,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896479451388825e-05,
+      "loss": 0.8145,
+      "step": 2096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989631417147398e-05,
+      "loss": 0.8096,
+      "step": 2097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989614876040978e-05,
+      "loss": 0.8105,
+      "step": 2098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989598321819842e-05,
+      "loss": 0.7881,
+      "step": 2099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895817544842095e-05,
+      "loss": 0.8057,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895651740342997e-05,
+      "loss": 0.8105,
+      "step": 2101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989548580470333e-05,
+      "loss": 0.8311,
+      "step": 2102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989531973792529e-05,
+      "loss": 0.8057,
+      "step": 2103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895153540011077e-05,
+      "loss": 0.8174,
+      "step": 2104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989498721096289e-05,
+      "loss": 0.8271,
+      "step": 2105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989482075078294e-05,
+      "loss": 0.8027,
+      "step": 2106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989465415947343e-05,
+      "loss": 0.8252,
+      "step": 2107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894487437036565e-05,
+      "loss": 0.8262,
+      "step": 2108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894320583474558e-05,
+      "loss": 0.8057,
+      "step": 2109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894153598789618e-05,
+      "loss": 0.8135,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989398648298396e-05,
+      "loss": 0.8203,
+      "step": 2111
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893819236059796e-05,
+      "loss": 0.8193,
+      "step": 2112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893651858019346e-05,
+      "loss": 0.8047,
+      "step": 2113
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989348434886483e-05,
+      "loss": 0.8164,
+      "step": 2114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989331670859846e-05,
+      "loss": 0.8232,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893148937222463e-05,
+      "loss": 0.8281,
+      "step": 2116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989298103473906e-05,
+      "loss": 0.8359,
+      "step": 2117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892813001150477e-05,
+      "loss": 0.8086,
+      "step": 2118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892644836458944e-05,
+      "loss": 0.832,
+      "step": 2119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892476540666684e-05,
+      "loss": 0.8291,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989230811377593e-05,
+      "loss": 0.8115,
+      "step": 2121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892139555788917e-05,
+      "loss": 0.8008,
+      "step": 2122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891970866707872e-05,
+      "loss": 0.8047,
+      "step": 2123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989180204653504e-05,
+      "loss": 0.8125,
+      "step": 2124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891633095272644e-05,
+      "loss": 0.8047,
+      "step": 2125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989146401292294e-05,
+      "loss": 0.7988,
+      "step": 2126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891294799488155e-05,
+      "loss": 0.8008,
+      "step": 2127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989112545497054e-05,
+      "loss": 0.8125,
+      "step": 2128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890955979372335e-05,
+      "loss": 0.8027,
+      "step": 2129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989078637269579e-05,
+      "loss": 0.832,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890616634943144e-05,
+      "loss": 0.7998,
+      "step": 2131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890446766116653e-05,
+      "loss": 0.7959,
+      "step": 2132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989027676621857e-05,
+      "loss": 0.8066,
+      "step": 2133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890106635251144e-05,
+      "loss": 0.8467,
+      "step": 2134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889936373216634e-05,
+      "loss": 0.7979,
+      "step": 2135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889765980117288e-05,
+      "loss": 0.8223,
+      "step": 2136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889595455955372e-05,
+      "loss": 0.8184,
+      "step": 2137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889424800733142e-05,
+      "loss": 0.8096,
+      "step": 2138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889254014452863e-05,
+      "loss": 0.8135,
+      "step": 2139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889083097116796e-05,
+      "loss": 0.8066,
+      "step": 2140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888912048727205e-05,
+      "loss": 0.8125,
+      "step": 2141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888740869286357e-05,
+      "loss": 0.8135,
+      "step": 2142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888569558796522e-05,
+      "loss": 0.8291,
+      "step": 2143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988839811725997e-05,
+      "loss": 0.791,
+      "step": 2144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988822654467897e-05,
+      "loss": 0.7939,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888054841055803e-05,
+      "loss": 0.7969,
+      "step": 2146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988788300639274e-05,
+      "loss": 0.8174,
+      "step": 2147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988771104069206e-05,
+      "loss": 0.8311,
+      "step": 2148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988753894395603e-05,
+      "loss": 0.8018,
+      "step": 2149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988736671618695e-05,
+      "loss": 0.8105,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988719435738709e-05,
+      "loss": 0.8252,
+      "step": 2151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887021867558736e-05,
+      "loss": 0.835,
+      "step": 2152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988684924670418e-05,
+      "loss": 0.8135,
+      "step": 2153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886676494825702e-05,
+      "loss": 0.8281,
+      "step": 2154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886503611925588e-05,
+      "loss": 0.8174,
+      "step": 2155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886330598006143e-05,
+      "loss": 0.8232,
+      "step": 2156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988615745306965e-05,
+      "loss": 0.8057,
+      "step": 2157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988598417711841e-05,
+      "loss": 0.8359,
+      "step": 2158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885810770154708e-05,
+      "loss": 0.8086,
+      "step": 2159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988563723218085e-05,
+      "loss": 0.792,
+      "step": 2160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885463563199142e-05,
+      "loss": 0.8047,
+      "step": 2161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885289763211872e-05,
+      "loss": 0.8408,
+      "step": 2162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988511583222135e-05,
+      "loss": 0.8545,
+      "step": 2163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884941770229884e-05,
+      "loss": 0.834,
+      "step": 2164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884767577239773e-05,
+      "loss": 0.8115,
+      "step": 2165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884593253253334e-05,
+      "loss": 0.8311,
+      "step": 2166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988441879827287e-05,
+      "loss": 0.8281,
+      "step": 2167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98842442123007e-05,
+      "loss": 0.8291,
+      "step": 2168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988406949533913e-05,
+      "loss": 0.7988,
+      "step": 2169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988389464739048e-05,
+      "loss": 0.8145,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883719668457066e-05,
+      "loss": 0.7764,
+      "step": 2171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988354455854121e-05,
+      "loss": 0.8125,
+      "step": 2172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883369317645224e-05,
+      "loss": 0.8047,
+      "step": 2173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883193945771443e-05,
+      "loss": 0.793,
+      "step": 2174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988301844292218e-05,
+      "loss": 0.8076,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882842809099766e-05,
+      "loss": 0.8408,
+      "step": 2176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882667044306527e-05,
+      "loss": 0.8125,
+      "step": 2177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988249114854479e-05,
+      "loss": 0.8086,
+      "step": 2178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882315121816892e-05,
+      "loss": 0.8213,
+      "step": 2179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882138964125164e-05,
+      "loss": 0.8145,
+      "step": 2180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881962675471936e-05,
+      "loss": 0.8135,
+      "step": 2181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988178625585955e-05,
+      "loss": 0.7891,
+      "step": 2182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988160970529034e-05,
+      "loss": 0.8174,
+      "step": 2183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881433023766646e-05,
+      "loss": 0.8145,
+      "step": 2184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881256211290812e-05,
+      "loss": 0.8076,
+      "step": 2185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881079267865182e-05,
+      "loss": 0.7979,
+      "step": 2186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880902193492097e-05,
+      "loss": 0.8125,
+      "step": 2187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880724988173905e-05,
+      "loss": 0.7861,
+      "step": 2188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880547651912954e-05,
+      "loss": 0.8203,
+      "step": 2189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880370184711592e-05,
+      "loss": 0.8027,
+      "step": 2190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880192586572177e-05,
+      "loss": 0.7998,
+      "step": 2191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988001485749706e-05,
+      "loss": 0.8174,
+      "step": 2192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987983699748859e-05,
+      "loss": 0.8242,
+      "step": 2193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987965900654914e-05,
+      "loss": 0.8145,
+      "step": 2194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987948088468105e-05,
+      "loss": 0.793,
+      "step": 2195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987930263188669e-05,
+      "loss": 0.8193,
+      "step": 2196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879124248168426e-05,
+      "loss": 0.8086,
+      "step": 2197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987894573352861e-05,
+      "loss": 0.8018,
+      "step": 2198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878767087969623e-05,
+      "loss": 0.8027,
+      "step": 2199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878588311493815e-05,
+      "loss": 0.8174,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987840940410357e-05,
+      "loss": 0.8359,
+      "step": 2201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878230365801256e-05,
+      "loss": 0.8213,
+      "step": 2202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987805119658924e-05,
+      "loss": 0.8154,
+      "step": 2203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98778718964699e-05,
+      "loss": 0.8076,
+      "step": 2204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987769246544561e-05,
+      "loss": 0.8096,
+      "step": 2205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877512903518756e-05,
+      "loss": 0.8115,
+      "step": 2206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877333210691704e-05,
+      "loss": 0.8125,
+      "step": 2207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877153386966844e-05,
+      "loss": 0.8027,
+      "step": 2208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987697343234656e-05,
+      "loss": 0.835,
+      "step": 2209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876793346833232e-05,
+      "loss": 0.833,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987661313042925e-05,
+      "loss": 0.8262,
+      "step": 2211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876432783137006e-05,
+      "loss": 0.8486,
+      "step": 2212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987625230495888e-05,
+      "loss": 0.8057,
+      "step": 2213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876071695897274e-05,
+      "loss": 0.8281,
+      "step": 2214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875890955954574e-05,
+      "loss": 0.8125,
+      "step": 2215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875710085133175e-05,
+      "loss": 0.8535,
+      "step": 2216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875529083435482e-05,
+      "loss": 0.7891,
+      "step": 2217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875347950863884e-05,
+      "loss": 0.8301,
+      "step": 2218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875166687420785e-05,
+      "loss": 0.8115,
+      "step": 2219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874985293108593e-05,
+      "loss": 0.7988,
+      "step": 2220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874803767929706e-05,
+      "loss": 0.833,
+      "step": 2221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874622111886528e-05,
+      "loss": 0.8301,
+      "step": 2222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874440324981473e-05,
+      "loss": 0.8262,
+      "step": 2223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874258407216943e-05,
+      "loss": 0.8027,
+      "step": 2224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874076358595353e-05,
+      "loss": 0.8223,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873894179119113e-05,
+      "loss": 0.8184,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987371186879064e-05,
+      "loss": 0.8008,
+      "step": 2227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987352942761235e-05,
+      "loss": 0.8135,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873346855586663e-05,
+      "loss": 0.79,
+      "step": 2229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873164152715988e-05,
+      "loss": 0.8242,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872981319002758e-05,
+      "loss": 0.8174,
+      "step": 2231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987279835444939e-05,
+      "loss": 0.8252,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987261525905831e-05,
+      "loss": 0.8047,
+      "step": 2233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872432032831943e-05,
+      "loss": 0.8135,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987224867577272e-05,
+      "loss": 0.8037,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987206518788307e-05,
+      "loss": 0.8076,
+      "step": 2236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987188156916543e-05,
+      "loss": 0.8232,
+      "step": 2237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987169781962222e-05,
+      "loss": 0.8125,
+      "step": 2238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871513939255882e-05,
+      "loss": 0.8232,
+      "step": 2239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871329928068857e-05,
+      "loss": 0.8193,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987114578606358e-05,
+      "loss": 0.8486,
+      "step": 2241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987096151324249e-05,
+      "loss": 0.7998,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870777109608035e-05,
+      "loss": 0.8115,
+      "step": 2243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987059257516265e-05,
+      "loss": 0.8154,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870407909908787e-05,
+      "loss": 0.791,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870223113848886e-05,
+      "loss": 0.835,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870038186985408e-05,
+      "loss": 0.8145,
+      "step": 2247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986985312932079e-05,
+      "loss": 0.7969,
+      "step": 2248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98696679408575e-05,
+      "loss": 0.7949,
+      "step": 2249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986948262159798e-05,
+      "loss": 0.8271,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869297171544685e-05,
+      "loss": 0.8018,
+      "step": 2251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869111590700082e-05,
+      "loss": 0.8115,
+      "step": 2252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868925879066623e-05,
+      "loss": 0.7988,
+      "step": 2253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986874003664677e-05,
+      "loss": 0.8281,
+      "step": 2254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986855406344299e-05,
+      "loss": 0.8057,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868367959457744e-05,
+      "loss": 0.8193,
+      "step": 2256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98681817246935e-05,
+      "loss": 0.8184,
+      "step": 2257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867995359152723e-05,
+      "loss": 0.7969,
+      "step": 2258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867808862837886e-05,
+      "loss": 0.7998,
+      "step": 2259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986762223575146e-05,
+      "loss": 0.8047,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986743547789592e-05,
+      "loss": 0.8311,
+      "step": 2261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867248589273737e-05,
+      "loss": 0.8223,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986706156988739e-05,
+      "loss": 0.79,
+      "step": 2263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866874419739357e-05,
+      "loss": 0.8027,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866687138832117e-05,
+      "loss": 0.7949,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986649972716815e-05,
+      "loss": 0.8135,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866312184749947e-05,
+      "loss": 0.8066,
+      "step": 2267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986612451157999e-05,
+      "loss": 0.7979,
+      "step": 2268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986593670766076e-05,
+      "loss": 0.8076,
+      "step": 2269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986574877299476e-05,
+      "loss": 0.8105,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865560707584465e-05,
+      "loss": 0.8047,
+      "step": 2271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865372511432373e-05,
+      "loss": 0.8096,
+      "step": 2272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986518418454098e-05,
+      "loss": 0.8193,
+      "step": 2273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864995726912783e-05,
+      "loss": 0.8193,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864807138550276e-05,
+      "loss": 0.8232,
+      "step": 2275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986461841945596e-05,
+      "loss": 0.8262,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864429569632335e-05,
+      "loss": 0.8154,
+      "step": 2277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864240589081902e-05,
+      "loss": 0.8008,
+      "step": 2278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986405147780717e-05,
+      "loss": 0.8008,
+      "step": 2279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863862235810643e-05,
+      "loss": 0.835,
+      "step": 2280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863672863094827e-05,
+      "loss": 0.8213,
+      "step": 2281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863483359662235e-05,
+      "loss": 0.7998,
+      "step": 2282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863293725515372e-05,
+      "loss": 0.8057,
+      "step": 2283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863103960656764e-05,
+      "loss": 0.8066,
+      "step": 2284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862914065088913e-05,
+      "loss": 0.8203,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862724038814337e-05,
+      "loss": 0.8086,
+      "step": 2286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986253388183556e-05,
+      "loss": 0.8145,
+      "step": 2287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862343594155102e-05,
+      "loss": 0.791,
+      "step": 2288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986215317577548e-05,
+      "loss": 0.8115,
+      "step": 2289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861962626699217e-05,
+      "loss": 0.8096,
+      "step": 2290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861771946928846e-05,
+      "loss": 0.8057,
+      "step": 2291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861581136466885e-05,
+      "loss": 0.8066,
+      "step": 2292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986139019531587e-05,
+      "loss": 0.8125,
+      "step": 2293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861199123478325e-05,
+      "loss": 0.792,
+      "step": 2294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861007920956786e-05,
+      "loss": 0.8037,
+      "step": 2295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860816587753785e-05,
+      "loss": 0.8291,
+      "step": 2296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860625123871863e-05,
+      "loss": 0.8418,
+      "step": 2297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860433529313546e-05,
+      "loss": 0.8184,
+      "step": 2298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860241804081383e-05,
+      "loss": 0.7959,
+      "step": 2299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860049948177914e-05,
+      "loss": 0.8076,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859857961605678e-05,
+      "loss": 0.8105,
+      "step": 2301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985966584436722e-05,
+      "loss": 0.8301,
+      "step": 2302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859473596465085e-05,
+      "loss": 0.8066,
+      "step": 2303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859281217901827e-05,
+      "loss": 0.79,
+      "step": 2304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859088708679986e-05,
+      "loss": 0.8115,
+      "step": 2305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985889606880212e-05,
+      "loss": 0.7979,
+      "step": 2306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858703298270782e-05,
+      "loss": 0.8232,
+      "step": 2307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858510397088524e-05,
+      "loss": 0.8096,
+      "step": 2308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858317365257902e-05,
+      "loss": 0.8076,
+      "step": 2309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858124202781474e-05,
+      "loss": 0.7988,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98579309096618e-05,
+      "loss": 0.8203,
+      "step": 2311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857737485901445e-05,
+      "loss": 0.7764,
+      "step": 2312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857543931502974e-05,
+      "loss": 0.8018,
+      "step": 2313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857350246468943e-05,
+      "loss": 0.8301,
+      "step": 2314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857156430801922e-05,
+      "loss": 0.8115,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856962484504485e-05,
+      "loss": 0.7998,
+      "step": 2316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856768407579198e-05,
+      "loss": 0.791,
+      "step": 2317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985657420002863e-05,
+      "loss": 0.8145,
+      "step": 2318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985637986185536e-05,
+      "loss": 0.8066,
+      "step": 2319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856185393061963e-05,
+      "loss": 0.7988,
+      "step": 2320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855990793651017e-05,
+      "loss": 0.8232,
+      "step": 2321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855796063625096e-05,
+      "loss": 0.8174,
+      "step": 2322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855601202986783e-05,
+      "loss": 0.8047,
+      "step": 2323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855406211738663e-05,
+      "loss": 0.8125,
+      "step": 2324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855211089883314e-05,
+      "loss": 0.8242,
+      "step": 2325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855015837423324e-05,
+      "loss": 0.8037,
+      "step": 2326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854820454361287e-05,
+      "loss": 0.8027,
+      "step": 2327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854624940699785e-05,
+      "loss": 0.7979,
+      "step": 2328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854429296441413e-05,
+      "loss": 0.7822,
+      "step": 2329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985423352158876e-05,
+      "loss": 0.8076,
+      "step": 2330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854037616144425e-05,
+      "loss": 0.7988,
+      "step": 2331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853841580111e-05,
+      "loss": 0.8008,
+      "step": 2332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985364541349109e-05,
+      "loss": 0.8037,
+      "step": 2333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853449116287282e-05,
+      "loss": 0.8193,
+      "step": 2334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985325268850219e-05,
+      "loss": 0.8145,
+      "step": 2335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9853056130138407e-05,
+      "loss": 0.8037,
+      "step": 2336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852859441198546e-05,
+      "loss": 0.8174,
+      "step": 2337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852662621685212e-05,
+      "loss": 0.8193,
+      "step": 2338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985246567160101e-05,
+      "loss": 0.8145,
+      "step": 2339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985226859094855e-05,
+      "loss": 0.8096,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985207137973045e-05,
+      "loss": 0.7939,
+      "step": 2341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851874037949317e-05,
+      "loss": 0.7998,
+      "step": 2342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851676565607768e-05,
+      "loss": 0.8174,
+      "step": 2343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985147896270842e-05,
+      "loss": 0.8164,
+      "step": 2344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985128122925389e-05,
+      "loss": 0.8018,
+      "step": 2345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851083365246804e-05,
+      "loss": 0.8135,
+      "step": 2346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985088537068978e-05,
+      "loss": 0.8418,
+      "step": 2347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985068724558544e-05,
+      "loss": 0.8057,
+      "step": 2348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850488989936414e-05,
+      "loss": 0.7969,
+      "step": 2349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850290603745327e-05,
+      "loss": 0.8086,
+      "step": 2350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985009208701481e-05,
+      "loss": 0.7939,
+      "step": 2351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984989343974749e-05,
+      "loss": 0.8008,
+      "step": 2352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849694661946004e-05,
+      "loss": 0.8174,
+      "step": 2353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849495753612985e-05,
+      "loss": 0.8506,
+      "step": 2354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849296714751065e-05,
+      "loss": 0.7979,
+      "step": 2355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849097545362886e-05,
+      "loss": 0.7988,
+      "step": 2356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848898245451087e-05,
+      "loss": 0.8086,
+      "step": 2357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984869881501831e-05,
+      "loss": 0.8311,
+      "step": 2358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848499254067196e-05,
+      "loss": 0.8076,
+      "step": 2359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848299562600386e-05,
+      "loss": 0.8164,
+      "step": 2360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984809974062053e-05,
+      "loss": 0.8184,
+      "step": 2361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847899788130284e-05,
+      "loss": 0.8291,
+      "step": 2362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847699705132286e-05,
+      "loss": 0.791,
+      "step": 2363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847499491629195e-05,
+      "loss": 0.8164,
+      "step": 2364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847299147623656e-05,
+      "loss": 0.8086,
+      "step": 2365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847098673118333e-05,
+      "loss": 0.8115,
+      "step": 2366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846898068115878e-05,
+      "loss": 0.8428,
+      "step": 2367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984669733261895e-05,
+      "loss": 0.8047,
+      "step": 2368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846496466630213e-05,
+      "loss": 0.8037,
+      "step": 2369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846295470152325e-05,
+      "loss": 0.8262,
+      "step": 2370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846094343187948e-05,
+      "loss": 0.8027,
+      "step": 2371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984589308573975e-05,
+      "loss": 0.8115,
+      "step": 2372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845691697810398e-05,
+      "loss": 0.8408,
+      "step": 2373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845490179402562e-05,
+      "loss": 0.792,
+      "step": 2374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845288530518912e-05,
+      "loss": 0.8154,
+      "step": 2375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984508675116212e-05,
+      "loss": 0.8164,
+      "step": 2376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844884841334858e-05,
+      "loss": 0.8066,
+      "step": 2377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844682801039804e-05,
+      "loss": 0.8242,
+      "step": 2378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844480630279633e-05,
+      "loss": 0.8203,
+      "step": 2379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984427832905703e-05,
+      "loss": 0.8203,
+      "step": 2380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984407589737467e-05,
+      "loss": 0.7988,
+      "step": 2381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843873335235238e-05,
+      "loss": 0.832,
+      "step": 2382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843670642641423e-05,
+      "loss": 0.8135,
+      "step": 2383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98434678195959e-05,
+      "loss": 0.8223,
+      "step": 2384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843264866101365e-05,
+      "loss": 0.8281,
+      "step": 2385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984306178216051e-05,
+      "loss": 0.8105,
+      "step": 2386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842858567776018e-05,
+      "loss": 0.7852,
+      "step": 2387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842655222950587e-05,
+      "loss": 0.8154,
+      "step": 2388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842451747686914e-05,
+      "loss": 0.8193,
+      "step": 2389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984224814198769e-05,
+      "loss": 0.7998,
+      "step": 2390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842044405855618e-05,
+      "loss": 0.832,
+      "step": 2391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841840539293395e-05,
+      "loss": 0.8252,
+      "step": 2392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841636542303725e-05,
+      "loss": 0.7959,
+      "step": 2393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841432414889308e-05,
+      "loss": 0.8076,
+      "step": 2394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841228157052853e-05,
+      "loss": 0.8115,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841023768797062e-05,
+      "loss": 0.8184,
+      "step": 2396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840819250124652e-05,
+      "loss": 0.7881,
+      "step": 2397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840614601038326e-05,
+      "loss": 0.8174,
+      "step": 2398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840409821540798e-05,
+      "loss": 0.8037,
+      "step": 2399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840204911634783e-05,
+      "loss": 0.8096,
+      "step": 2400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839999871322993e-05,
+      "loss": 0.8086,
+      "step": 2401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839794700608147e-05,
+      "loss": 0.8418,
+      "step": 2402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839589399492968e-05,
+      "loss": 0.8174,
+      "step": 2403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839383967980174e-05,
+      "loss": 0.8066,
+      "step": 2404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839178406072482e-05,
+      "loss": 0.792,
+      "step": 2405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983897271377263e-05,
+      "loss": 0.7979,
+      "step": 2406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838766891083327e-05,
+      "loss": 0.8262,
+      "step": 2407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838560938007312e-05,
+      "loss": 0.8037,
+      "step": 2408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838354854547312e-05,
+      "loss": 0.8154,
+      "step": 2409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838148640706055e-05,
+      "loss": 0.7988,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983794229648628e-05,
+      "loss": 0.8066,
+      "step": 2411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837735821890713e-05,
+      "loss": 0.7871,
+      "step": 2412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.98375292169221e-05,
+      "loss": 0.834,
+      "step": 2413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837322481583165e-05,
+      "loss": 0.8379,
+      "step": 2414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837115615876664e-05,
+      "loss": 0.8174,
+      "step": 2415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983690861980533e-05,
+      "loss": 0.8076,
+      "step": 2416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836701493371903e-05,
+      "loss": 0.8076,
+      "step": 2417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836494236579137e-05,
+      "loss": 0.8057,
+      "step": 2418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983628684942977e-05,
+      "loss": 0.8164,
+      "step": 2419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836079331926558e-05,
+      "loss": 0.8105,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835871684072246e-05,
+      "loss": 0.8154,
+      "step": 2421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835663905869586e-05,
+      "loss": 0.8076,
+      "step": 2422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835455997321333e-05,
+      "loss": 0.7822,
+      "step": 2423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835247958430243e-05,
+      "loss": 0.8281,
+      "step": 2424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835039789199068e-05,
+      "loss": 0.7998,
+      "step": 2425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834831489630573e-05,
+      "loss": 0.8027,
+      "step": 2426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983462305972752e-05,
+      "loss": 0.8027,
+      "step": 2427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834414499492662e-05,
+      "loss": 0.8164,
+      "step": 2428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834205808928768e-05,
+      "loss": 0.8242,
+      "step": 2429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833996988038606e-05,
+      "loss": 0.8301,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983378803682494e-05,
+      "loss": 0.8096,
+      "step": 2431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983357895529054e-05,
+      "loss": 0.8223,
+      "step": 2432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983336974343818e-05,
+      "loss": 0.8184,
+      "step": 2433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833160401270626e-05,
+      "loss": 0.79,
+      "step": 2434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983295092879066e-05,
+      "loss": 0.8057,
+      "step": 2435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983274132600105e-05,
+      "loss": 0.8223,
+      "step": 2436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983253159290458e-05,
+      "loss": 0.8408,
+      "step": 2437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832321729504023e-05,
+      "loss": 0.8291,
+      "step": 2438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983211173580217e-05,
+      "loss": 0.8076,
+      "step": 2439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831901611801793e-05,
+      "loss": 0.7949,
+      "step": 2440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831691357505686e-05,
+      "loss": 0.7783,
+      "step": 2441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983148097291663e-05,
+      "loss": 0.8076,
+      "step": 2442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831270458037412e-05,
+      "loss": 0.8271,
+      "step": 2443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831059812870826e-05,
+      "loss": 0.8115,
+      "step": 2444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830849037419656e-05,
+      "loss": 0.8242,
+      "step": 2445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830638131686708e-05,
+      "loss": 0.8174,
+      "step": 2446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830427095674766e-05,
+      "loss": 0.8242,
+      "step": 2447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830215929386637e-05,
+      "loss": 0.8066,
+      "step": 2448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830004632825107e-05,
+      "loss": 0.8262,
+      "step": 2449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829793205992984e-05,
+      "loss": 0.8076,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829581648893064e-05,
+      "loss": 0.7959,
+      "step": 2451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982936996152816e-05,
+      "loss": 0.8057,
+      "step": 2452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829158143901074e-05,
+      "loss": 0.8037,
+      "step": 2453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828946196014603e-05,
+      "loss": 0.832,
+      "step": 2454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828734117871572e-05,
+      "loss": 0.8086,
+      "step": 2455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982852190947478e-05,
+      "loss": 0.8027,
+      "step": 2456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828309570827042e-05,
+      "loss": 0.835,
+      "step": 2457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828097101931175e-05,
+      "loss": 0.7959,
+      "step": 2458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982788450278999e-05,
+      "loss": 0.7803,
+      "step": 2459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827671773406306e-05,
+      "loss": 0.8174,
+      "step": 2460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827458913782943e-05,
+      "loss": 0.8135,
+      "step": 2461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827245923922728e-05,
+      "loss": 0.7881,
+      "step": 2462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827032803828473e-05,
+      "loss": 0.792,
+      "step": 2463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826819553503007e-05,
+      "loss": 0.8203,
+      "step": 2464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826606172949156e-05,
+      "loss": 0.8066,
+      "step": 2465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826392662169747e-05,
+      "loss": 0.8037,
+      "step": 2466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826179021167613e-05,
+      "loss": 0.8115,
+      "step": 2467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982596524994558e-05,
+      "loss": 0.8125,
+      "step": 2468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825751348506483e-05,
+      "loss": 0.8125,
+      "step": 2469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982553731685316e-05,
+      "loss": 0.8203,
+      "step": 2470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825323154988446e-05,
+      "loss": 0.8154,
+      "step": 2471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825108862915174e-05,
+      "loss": 0.7998,
+      "step": 2472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824894440636186e-05,
+      "loss": 0.7969,
+      "step": 2473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982467988815433e-05,
+      "loss": 0.8213,
+      "step": 2474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982446520547244e-05,
+      "loss": 0.7988,
+      "step": 2475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982425039259337e-05,
+      "loss": 0.8203,
+      "step": 2476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982403544951996e-05,
+      "loss": 0.8252,
+      "step": 2477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982382037625506e-05,
+      "loss": 0.8271,
+      "step": 2478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823605172801523e-05,
+      "loss": 0.8037,
+      "step": 2479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823389839162196e-05,
+      "loss": 0.8154,
+      "step": 2480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823174375339942e-05,
+      "loss": 0.8076,
+      "step": 2481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822958781337605e-05,
+      "loss": 0.8008,
+      "step": 2482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822743057158048e-05,
+      "loss": 0.8203,
+      "step": 2483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982252720280413e-05,
+      "loss": 0.7988,
+      "step": 2484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982231121827871e-05,
+      "loss": 0.8262,
+      "step": 2485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982209510358465e-05,
+      "loss": 0.8086,
+      "step": 2486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821878858724818e-05,
+      "loss": 0.8057,
+      "step": 2487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821662483702074e-05,
+      "loss": 0.793,
+      "step": 2488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982144597851929e-05,
+      "loss": 0.8105,
+      "step": 2489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982122934317933e-05,
+      "loss": 0.8242,
+      "step": 2490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821012577685074e-05,
+      "loss": 0.8105,
+      "step": 2491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820795682039384e-05,
+      "loss": 0.7793,
+      "step": 2492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820578656245143e-05,
+      "loss": 0.8037,
+      "step": 2493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982036150030522e-05,
+      "loss": 0.8135,
+      "step": 2494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98201442142225e-05,
+      "loss": 0.8271,
+      "step": 2495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819926797999854e-05,
+      "loss": 0.8135,
+      "step": 2496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981970925164017e-05,
+      "loss": 0.8115,
+      "step": 2497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819491575146333e-05,
+      "loss": 0.8096,
+      "step": 2498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819273768521217e-05,
+      "loss": 0.8174,
+      "step": 2499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981905583176772e-05,
+      "loss": 0.7891,
+      "step": 2500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818837764888727e-05,
+      "loss": 0.8359,
+      "step": 2501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818619567887123e-05,
+      "loss": 0.7812,
+      "step": 2502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818401240765804e-05,
+      "loss": 0.8135,
+      "step": 2503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818182783527665e-05,
+      "loss": 0.7988,
+      "step": 2504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817964196175596e-05,
+      "loss": 0.8193,
+      "step": 2505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817745478712496e-05,
+      "loss": 0.8008,
+      "step": 2506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817526631141263e-05,
+      "loss": 0.8086,
+      "step": 2507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98173076534648e-05,
+      "loss": 0.8018,
+      "step": 2508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817088545686e-05,
+      "loss": 0.7998,
+      "step": 2509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816869307807784e-05,
+      "loss": 0.792,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981664993983304e-05,
+      "loss": 0.7988,
+      "step": 2511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816430441764685e-05,
+      "loss": 0.8174,
+      "step": 2512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9816210813605625e-05,
+      "loss": 0.8076,
+      "step": 2513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815991055358773e-05,
+      "loss": 0.8242,
+      "step": 2514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815771167027035e-05,
+      "loss": 0.8018,
+      "step": 2515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815551148613326e-05,
+      "loss": 0.7979,
+      "step": 2516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981533100012057e-05,
+      "loss": 0.8174,
+      "step": 2517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981511072155168e-05,
+      "loss": 0.8105,
+      "step": 2518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814890312909575e-05,
+      "loss": 0.792,
+      "step": 2519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981466977419717e-05,
+      "loss": 0.8262,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814449105417396e-05,
+      "loss": 0.8057,
+      "step": 2521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814228306573174e-05,
+      "loss": 0.7949,
+      "step": 2522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981400737766743e-05,
+      "loss": 0.8164,
+      "step": 2523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813786318703094e-05,
+      "loss": 0.7988,
+      "step": 2524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813565129683093e-05,
+      "loss": 0.8105,
+      "step": 2525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813343810610358e-05,
+      "loss": 0.7852,
+      "step": 2526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813122361487826e-05,
+      "loss": 0.8066,
+      "step": 2527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812900782318425e-05,
+      "loss": 0.7979,
+      "step": 2528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98126790731051e-05,
+      "loss": 0.8223,
+      "step": 2529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981245723385078e-05,
+      "loss": 0.8223,
+      "step": 2530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812235264558414e-05,
+      "loss": 0.793,
+      "step": 2531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812013165230932e-05,
+      "loss": 0.8027,
+      "step": 2532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981179093587129e-05,
+      "loss": 0.8203,
+      "step": 2533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811568576482426e-05,
+      "loss": 0.7842,
+      "step": 2534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811346087067286e-05,
+      "loss": 0.8154,
+      "step": 2535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981112346762882e-05,
+      "loss": 0.7852,
+      "step": 2536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810900718169987e-05,
+      "loss": 0.8145,
+      "step": 2537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810677838693725e-05,
+      "loss": 0.8213,
+      "step": 2538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810454829202992e-05,
+      "loss": 0.7881,
+      "step": 2539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981023168970075e-05,
+      "loss": 0.8379,
+      "step": 2540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810008420189945e-05,
+      "loss": 0.8018,
+      "step": 2541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809785020673544e-05,
+      "loss": 0.8271,
+      "step": 2542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980956149115451e-05,
+      "loss": 0.8076,
+      "step": 2543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809337831635794e-05,
+      "loss": 0.8359,
+      "step": 2544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809114042120368e-05,
+      "loss": 0.792,
+      "step": 2545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808890122611196e-05,
+      "loss": 0.7979,
+      "step": 2546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808666073111248e-05,
+      "loss": 0.8164,
+      "step": 2547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808441893623494e-05,
+      "loss": 0.8164,
+      "step": 2548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808217584150896e-05,
+      "loss": 0.8408,
+      "step": 2549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807993144696438e-05,
+      "loss": 0.8232,
+      "step": 2550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807768575263086e-05,
+      "loss": 0.8262,
+      "step": 2551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980754387585382e-05,
+      "loss": 0.8242,
+      "step": 2552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807319046471617e-05,
+      "loss": 0.8027,
+      "step": 2553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807094087119455e-05,
+      "loss": 0.8203,
+      "step": 2554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980686899780032e-05,
+      "loss": 0.7939,
+      "step": 2555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980664377851719e-05,
+      "loss": 0.8027,
+      "step": 2556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980641842927305e-05,
+      "loss": 0.8008,
+      "step": 2557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806192950070885e-05,
+      "loss": 0.8223,
+      "step": 2558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980596734091369e-05,
+      "loss": 0.8223,
+      "step": 2559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980574160180445e-05,
+      "loss": 0.8057,
+      "step": 2560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805515732746156e-05,
+      "loss": 0.7949,
+      "step": 2561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805289733741808e-05,
+      "loss": 0.8164,
+      "step": 2562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805063604794386e-05,
+      "loss": 0.8125,
+      "step": 2563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804837345906902e-05,
+      "loss": 0.8125,
+      "step": 2564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980461095708235e-05,
+      "loss": 0.7852,
+      "step": 2565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804384438323723e-05,
+      "loss": 0.8008,
+      "step": 2566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804157789634036e-05,
+      "loss": 0.79,
+      "step": 2567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980393101101628e-05,
+      "loss": 0.8486,
+      "step": 2568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803704102473466e-05,
+      "loss": 0.8232,
+      "step": 2569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803477064008604e-05,
+      "loss": 0.8301,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803249895624698e-05,
+      "loss": 0.8047,
+      "step": 2571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980302259732476e-05,
+      "loss": 0.8125,
+      "step": 2572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802795169111804e-05,
+      "loss": 0.835,
+      "step": 2573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980256761098884e-05,
+      "loss": 0.8105,
+      "step": 2574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980233992295889e-05,
+      "loss": 0.8203,
+      "step": 2575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980211210502496e-05,
+      "loss": 0.8555,
+      "step": 2576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801884157190084e-05,
+      "loss": 0.7852,
+      "step": 2577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980165607945727e-05,
+      "loss": 0.8096,
+      "step": 2578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801427871829553e-05,
+      "loss": 0.8066,
+      "step": 2579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801199534309945e-05,
+      "loss": 0.7881,
+      "step": 2580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800971066901477e-05,
+      "loss": 0.8066,
+      "step": 2581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980074246960718e-05,
+      "loss": 0.8223,
+      "step": 2582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800513742430077e-05,
+      "loss": 0.7988,
+      "step": 2583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800284885373203e-05,
+      "loss": 0.7959,
+      "step": 2584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980005589843959e-05,
+      "loss": 0.8125,
+      "step": 2585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799826781632277e-05,
+      "loss": 0.8438,
+      "step": 2586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799597534954294e-05,
+      "loss": 0.8145,
+      "step": 2587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979936815840868e-05,
+      "loss": 0.8037,
+      "step": 2588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979913865199848e-05,
+      "loss": 0.8086,
+      "step": 2589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798909015726728e-05,
+      "loss": 0.7939,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798679249596472e-05,
+      "loss": 0.8164,
+      "step": 2591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798449353610757e-05,
+      "loss": 0.7959,
+      "step": 2592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979821932777263e-05,
+      "loss": 0.7959,
+      "step": 2593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797989172085134e-05,
+      "loss": 0.8047,
+      "step": 2594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797758886551328e-05,
+      "loss": 0.8057,
+      "step": 2595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797528471174255e-05,
+      "loss": 0.8076,
+      "step": 2596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797297925956973e-05,
+      "loss": 0.8193,
+      "step": 2597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797067250902537e-05,
+      "loss": 0.8477,
+      "step": 2598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796836446014006e-05,
+      "loss": 0.8193,
+      "step": 2599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979660551129443e-05,
+      "loss": 0.7949,
+      "step": 2600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796374446746882e-05,
+      "loss": 0.8145,
+      "step": 2601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796143252374418e-05,
+      "loss": 0.7988,
+      "step": 2602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795911928180098e-05,
+      "loss": 0.8076,
+      "step": 2603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795680474166994e-05,
+      "loss": 0.8193,
+      "step": 2604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795448890338172e-05,
+      "loss": 0.7998,
+      "step": 2605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795217176696697e-05,
+      "loss": 0.7959,
+      "step": 2606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794985333245643e-05,
+      "loss": 0.8125,
+      "step": 2607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979475335998808e-05,
+      "loss": 0.8027,
+      "step": 2608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794521256927088e-05,
+      "loss": 0.8057,
+      "step": 2609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794289024065735e-05,
+      "loss": 0.8164,
+      "step": 2610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794056661407107e-05,
+      "loss": 0.7773,
+      "step": 2611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979382416895428e-05,
+      "loss": 0.8096,
+      "step": 2612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979359154671033e-05,
+      "loss": 0.791,
+      "step": 2613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793358794678347e-05,
+      "loss": 0.8066,
+      "step": 2614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793125912861407e-05,
+      "loss": 0.7959,
+      "step": 2615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9792892901262605e-05,
+      "loss": 0.8057,
+      "step": 2616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979265975988503e-05,
+      "loss": 0.8125,
+      "step": 2617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979242648873176e-05,
+      "loss": 0.8027,
+      "step": 2618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792193087805896e-05,
+      "loss": 0.8037,
+      "step": 2619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791959557110535e-05,
+      "loss": 0.7998,
+      "step": 2620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979172589664876e-05,
+      "loss": 0.8018,
+      "step": 2621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791492106423672e-05,
+      "loss": 0.8369,
+      "step": 2622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791258186438372e-05,
+      "loss": 0.8115,
+      "step": 2623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979102413669596e-05,
+      "loss": 0.8438,
+      "step": 2624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790789957199533e-05,
+      "loss": 0.792,
+      "step": 2625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97905556479522e-05,
+      "loss": 0.8486,
+      "step": 2626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979032120895706e-05,
+      "loss": 0.7998,
+      "step": 2627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790086640217226e-05,
+      "loss": 0.7998,
+      "step": 2628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789851941735806e-05,
+      "loss": 0.8125,
+      "step": 2629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789617113515907e-05,
+      "loss": 0.8154,
+      "step": 2630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978938215556064e-05,
+      "loss": 0.8105,
+      "step": 2631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789147067873125e-05,
+      "loss": 0.8037,
+      "step": 2632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978891185045647e-05,
+      "loss": 0.7949,
+      "step": 2633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97886765033138e-05,
+      "loss": 0.8105,
+      "step": 2634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788441026448225e-05,
+      "loss": 0.7998,
+      "step": 2635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788205419862877e-05,
+      "loss": 0.8105,
+      "step": 2636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787969683560867e-05,
+      "loss": 0.8096,
+      "step": 2637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787733817545325e-05,
+      "loss": 0.7959,
+      "step": 2638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787497821819373e-05,
+      "loss": 0.8076,
+      "step": 2639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787261696386145e-05,
+      "loss": 0.8242,
+      "step": 2640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787025441248763e-05,
+      "loss": 0.7969,
+      "step": 2641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786789056410366e-05,
+      "loss": 0.8027,
+      "step": 2642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786552541874075e-05,
+      "loss": 0.8164,
+      "step": 2643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786315897643035e-05,
+      "loss": 0.793,
+      "step": 2644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978607912372038e-05,
+      "loss": 0.7949,
+      "step": 2645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785842220109244e-05,
+      "loss": 0.8193,
+      "step": 2646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785605186812767e-05,
+      "loss": 0.7998,
+      "step": 2647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785368023834095e-05,
+      "loss": 0.8262,
+      "step": 2648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785130731176364e-05,
+      "loss": 0.792,
+      "step": 2649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978489330884273e-05,
+      "loss": 0.7764,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784655756836323e-05,
+      "loss": 0.8311,
+      "step": 2651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978441807516031e-05,
+      "loss": 0.793,
+      "step": 2652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978418026381782e-05,
+      "loss": 0.8096,
+      "step": 2653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783942322812022e-05,
+      "loss": 0.7773,
+      "step": 2654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978370425214606e-05,
+      "loss": 0.792,
+      "step": 2655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783466051823093e-05,
+      "loss": 0.8066,
+      "step": 2656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978322772184628e-05,
+      "loss": 0.8301,
+      "step": 2657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978298926221877e-05,
+      "loss": 0.8018,
+      "step": 2658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782750672943732e-05,
+      "loss": 0.8076,
+      "step": 2659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782511954024324e-05,
+      "loss": 0.7959,
+      "step": 2660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978227310546371e-05,
+      "loss": 0.8115,
+      "step": 2661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782034127265058e-05,
+      "loss": 0.7998,
+      "step": 2662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978179501943153e-05,
+      "loss": 0.7979,
+      "step": 2663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97815557819663e-05,
+      "loss": 0.8164,
+      "step": 2664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781316414872537e-05,
+      "loss": 0.8213,
+      "step": 2665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978107691815341e-05,
+      "loss": 0.7783,
+      "step": 2666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780837291812093e-05,
+      "loss": 0.8193,
+      "step": 2667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780597535851765e-05,
+      "loss": 0.8008,
+      "step": 2668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780357650275603e-05,
+      "loss": 0.7822,
+      "step": 2669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780117635086785e-05,
+      "loss": 0.8115,
+      "step": 2670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.977987749028849e-05,
+      "loss": 0.7832,
+      "step": 2671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779637215883905e-05,
+      "loss": 0.8135,
+      "step": 2672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977939681187621e-05,
+      "loss": 0.8145,
+      "step": 2673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779156278268594e-05,
+      "loss": 0.793,
+      "step": 2674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778915615064242e-05,
+      "loss": 0.8086,
+      "step": 2675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778674822266344e-05,
+      "loss": 0.8008,
+      "step": 2676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778433899878093e-05,
+      "loss": 0.7852,
+      "step": 2677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977819284790268e-05,
+      "loss": 0.7812,
+      "step": 2678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777951666343295e-05,
+      "loss": 0.7979,
+      "step": 2679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9777710355203147e-05,
+      "loss": 0.7998,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977746891448542e-05,
+      "loss": 0.8213,
+      "step": 2681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977722734419332e-05,
+      "loss": 0.7891,
+      "step": 2682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776985644330048e-05,
+      "loss": 0.793,
+      "step": 2683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776743814898812e-05,
+      "loss": 0.7959,
+      "step": 2684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776501855902804e-05,
+      "loss": 0.8428,
+      "step": 2685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776259767345243e-05,
+      "loss": 0.8096,
+      "step": 2686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776017549229334e-05,
+      "loss": 0.8174,
+      "step": 2687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775775201558283e-05,
+      "loss": 0.8076,
+      "step": 2688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775532724335303e-05,
+      "loss": 0.8066,
+      "step": 2689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775290117563612e-05,
+      "loss": 0.8008,
+      "step": 2690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977504738124642e-05,
+      "loss": 0.7998,
+      "step": 2691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774804515386944e-05,
+      "loss": 0.8018,
+      "step": 2692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977456151998841e-05,
+      "loss": 0.7979,
+      "step": 2693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774318395054026e-05,
+      "loss": 0.8232,
+      "step": 2694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774075140587024e-05,
+      "loss": 0.8105,
+      "step": 2695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773831756590623e-05,
+      "loss": 0.8145,
+      "step": 2696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773588243068048e-05,
+      "loss": 0.7998,
+      "step": 2697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773344600022527e-05,
+      "loss": 0.8105,
+      "step": 2698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773100827457293e-05,
+      "loss": 0.8271,
+      "step": 2699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977285692537557e-05,
+      "loss": 0.793,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772612893780595e-05,
+      "loss": 0.7871,
+      "step": 2701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772368732675598e-05,
+      "loss": 0.793,
+      "step": 2702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772124442063817e-05,
+      "loss": 0.8086,
+      "step": 2703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771880021948492e-05,
+      "loss": 0.8164,
+      "step": 2704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977163547233286e-05,
+      "loss": 0.8027,
+      "step": 2705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977139079322016e-05,
+      "loss": 0.8047,
+      "step": 2706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771145984613633e-05,
+      "loss": 0.7764,
+      "step": 2707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977090104651653e-05,
+      "loss": 0.7861,
+      "step": 2708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770655978932088e-05,
+      "loss": 0.8076,
+      "step": 2709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770410781863568e-05,
+      "loss": 0.7959,
+      "step": 2710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770165455314206e-05,
+      "loss": 0.8525,
+      "step": 2711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976991999928726e-05,
+      "loss": 0.8242,
+      "step": 2712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976967441378598e-05,
+      "loss": 0.8359,
+      "step": 2713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769428698813624e-05,
+      "loss": 0.8105,
+      "step": 2714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9769182854373444e-05,
+      "loss": 0.8379,
+      "step": 2715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768936880468704e-05,
+      "loss": 0.8301,
+      "step": 2716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768690777102658e-05,
+      "loss": 0.7998,
+      "step": 2717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768444544278566e-05,
+      "loss": 0.8174,
+      "step": 2718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97681981819997e-05,
+      "loss": 0.8154,
+      "step": 2719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767951690269314e-05,
+      "loss": 0.8018,
+      "step": 2720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767705069090683e-05,
+      "loss": 0.8135,
+      "step": 2721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767458318467073e-05,
+      "loss": 0.8145,
+      "step": 2722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767211438401756e-05,
+      "loss": 0.8135,
+      "step": 2723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766964428897996e-05,
+      "loss": 0.8281,
+      "step": 2724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766717289959075e-05,
+      "loss": 0.8076,
+      "step": 2725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766470021588263e-05,
+      "loss": 0.8135,
+      "step": 2726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976622262378884e-05,
+      "loss": 0.8184,
+      "step": 2727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9765975096564082e-05,
+      "loss": 0.8154,
+      "step": 2728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976572743991727e-05,
+      "loss": 0.8008,
+      "step": 2729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976547965385169e-05,
+      "loss": 0.8047,
+      "step": 2730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976523173837062e-05,
+      "loss": 0.793,
+      "step": 2731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764983693477348e-05,
+      "loss": 0.8145,
+      "step": 2732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976473551917516e-05,
+      "loss": 0.8252,
+      "step": 2733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764487215467345e-05,
+      "loss": 0.8115,
+      "step": 2734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764238782357198e-05,
+      "loss": 0.7939,
+      "step": 2735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763990219848006e-05,
+      "loss": 0.7852,
+      "step": 2736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763741527943067e-05,
+      "loss": 0.7715,
+      "step": 2737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763492706645672e-05,
+      "loss": 0.8184,
+      "step": 2738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976324375595912e-05,
+      "loss": 0.791,
+      "step": 2739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762994675886715e-05,
+      "loss": 0.8135,
+      "step": 2740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976274546643175e-05,
+      "loss": 0.7881,
+      "step": 2741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762496127597536e-05,
+      "loss": 0.7969,
+      "step": 2742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976224665938737e-05,
+      "loss": 0.7979,
+      "step": 2743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761997061804562e-05,
+      "loss": 0.7988,
+      "step": 2744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976174733485242e-05,
+      "loss": 0.7959,
+      "step": 2745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976149747853425e-05,
+      "loss": 0.7998,
+      "step": 2746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761247492853366e-05,
+      "loss": 0.8154,
+      "step": 2747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760997377813084e-05,
+      "loss": 0.7939,
+      "step": 2748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976074713341671e-05,
+      "loss": 0.7998,
+      "step": 2749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760496759667568e-05,
+      "loss": 0.79,
+      "step": 2750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760246256568973e-05,
+      "loss": 0.7871,
+      "step": 2751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759995624124247e-05,
+      "loss": 0.793,
+      "step": 2752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975974486233671e-05,
+      "loss": 0.79,
+      "step": 2753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975949397120968e-05,
+      "loss": 0.7959,
+      "step": 2754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759242950746487e-05,
+      "loss": 0.7998,
+      "step": 2755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975899180095046e-05,
+      "loss": 0.7959,
+      "step": 2756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758740521824924e-05,
+      "loss": 0.7793,
+      "step": 2757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975848911337321e-05,
+      "loss": 0.7988,
+      "step": 2758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975823757559865e-05,
+      "loss": 0.8184,
+      "step": 2759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757985908504576e-05,
+      "loss": 0.8018,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757734112094323e-05,
+      "loss": 0.8086,
+      "step": 2761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757482186371232e-05,
+      "loss": 0.8223,
+      "step": 2762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757230131338635e-05,
+      "loss": 0.8105,
+      "step": 2763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975697794699988e-05,
+      "loss": 0.8145,
+      "step": 2764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97567256333583e-05,
+      "loss": 0.8096,
+      "step": 2765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756473190417248e-05,
+      "loss": 0.7969,
+      "step": 2766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975622061818006e-05,
+      "loss": 0.8008,
+      "step": 2767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755967916650092e-05,
+      "loss": 0.7939,
+      "step": 2768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755715085830688e-05,
+      "loss": 0.8008,
+      "step": 2769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755462125725202e-05,
+      "loss": 0.8232,
+      "step": 2770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755209036336983e-05,
+      "loss": 0.8115,
+      "step": 2771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754955817669385e-05,
+      "loss": 0.8086,
+      "step": 2772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754702469725764e-05,
+      "loss": 0.7764,
+      "step": 2773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975444899250948e-05,
+      "loss": 0.7969,
+      "step": 2774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975419538602389e-05,
+      "loss": 0.8164,
+      "step": 2775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753941650272357e-05,
+      "loss": 0.792,
+      "step": 2776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975368778525824e-05,
+      "loss": 0.8291,
+      "step": 2777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753433790984906e-05,
+      "loss": 0.8271,
+      "step": 2778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975317966745572e-05,
+      "loss": 0.8115,
+      "step": 2779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975292541467405e-05,
+      "loss": 0.7969,
+      "step": 2780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752671032643266e-05,
+      "loss": 0.7939,
+      "step": 2781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752416521366737e-05,
+      "loss": 0.7871,
+      "step": 2782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975216188084784e-05,
+      "loss": 0.8193,
+      "step": 2783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751907111089945e-05,
+      "loss": 0.8281,
+      "step": 2784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975165221209643e-05,
+      "loss": 0.793,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9751397183870677e-05,
+      "loss": 0.7979,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975114202641606e-05,
+      "loss": 0.7861,
+      "step": 2787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975088673973596e-05,
+      "loss": 0.8105,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750631323833766e-05,
+      "loss": 0.8086,
+      "step": 2789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750375778712857e-05,
+      "loss": 0.8271,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750120104376625e-05,
+      "loss": 0.8027,
+      "step": 2791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974986430082845e-05,
+      "loss": 0.7949,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749608368071733e-05,
+      "loss": 0.8047,
+      "step": 2793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974935230610986e-05,
+      "loss": 0.8018,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974909611494622e-05,
+      "loss": 0.8066,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748839794584216e-05,
+      "loss": 0.8037,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748583345027242e-05,
+      "loss": 0.8184,
+      "step": 2797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748326766278693e-05,
+      "loss": 0.7959,
+      "step": 2798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748070058341975e-05,
+      "loss": 0.8105,
+      "step": 2799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747813221220486e-05,
+      "loss": 0.7754,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747556254917632e-05,
+      "loss": 0.7891,
+      "step": 2801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747299159436816e-05,
+      "loss": 0.8242,
+      "step": 2802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974704193478145e-05,
+      "loss": 0.8193,
+      "step": 2803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746784580954936e-05,
+      "loss": 0.8037,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746527097960687e-05,
+      "loss": 0.8154,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974626948580212e-05,
+      "loss": 0.8037,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974601174448264e-05,
+      "loss": 0.8164,
+      "step": 2807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974575387400567e-05,
+      "loss": 0.791,
+      "step": 2808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745495874374628e-05,
+      "loss": 0.7812,
+      "step": 2809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974523774559293e-05,
+      "loss": 0.793,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744979487663994e-05,
+      "loss": 0.8242,
+      "step": 2811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744721100591246e-05,
+      "loss": 0.8125,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974446258437811e-05,
+      "loss": 0.8174,
+      "step": 2813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974420393902801e-05,
+      "loss": 0.7803,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974394516454438e-05,
+      "loss": 0.7783,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743686260930645e-05,
+      "loss": 0.8037,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974342722819023e-05,
+      "loss": 0.8076,
+      "step": 2817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974316806632658e-05,
+      "loss": 0.8125,
+      "step": 2818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974290877534312e-05,
+      "loss": 0.793,
+      "step": 2819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974264935524329e-05,
+      "loss": 0.8037,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742389806030528e-05,
+      "loss": 0.7861,
+      "step": 2821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974213012770827e-05,
+      "loss": 0.7988,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741870320279962e-05,
+      "loss": 0.79,
+      "step": 2823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741610383749047e-05,
+      "loss": 0.7949,
+      "step": 2824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741350318118965e-05,
+      "loss": 0.8018,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741090123393167e-05,
+      "loss": 0.8018,
+      "step": 2826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740829799575102e-05,
+      "loss": 0.7822,
+      "step": 2827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740569346668214e-05,
+      "loss": 0.8008,
+      "step": 2828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740308764675964e-05,
+      "loss": 0.8223,
+      "step": 2829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740048053601795e-05,
+      "loss": 0.832,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739787213449167e-05,
+      "loss": 0.8193,
+      "step": 2831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739526244221538e-05,
+      "loss": 0.8057,
+      "step": 2832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973926514592236e-05,
+      "loss": 0.8018,
+      "step": 2833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739003918555105e-05,
+      "loss": 0.8174,
+      "step": 2834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738742562123227e-05,
+      "loss": 0.8057,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738481076630185e-05,
+      "loss": 0.8008,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738219462079453e-05,
+      "loss": 0.8174,
+      "step": 2837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737957718474497e-05,
+      "loss": 0.8008,
+      "step": 2838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737695845818782e-05,
+      "loss": 0.8223,
+      "step": 2839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737433844115778e-05,
+      "loss": 0.8115,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737171713368958e-05,
+      "loss": 0.8125,
+      "step": 2841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736909453581798e-05,
+      "loss": 0.792,
+      "step": 2842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736647064757775e-05,
+      "loss": 0.791,
+      "step": 2843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736384546900364e-05,
+      "loss": 0.7803,
+      "step": 2844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736121900013042e-05,
+      "loss": 0.8613,
+      "step": 2845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973585912409929e-05,
+      "loss": 0.8047,
+      "step": 2846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735596219162593e-05,
+      "loss": 0.8125,
+      "step": 2847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735333185206435e-05,
+      "loss": 0.8369,
+      "step": 2848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735070022234298e-05,
+      "loss": 0.8105,
+      "step": 2849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734806730249675e-05,
+      "loss": 0.7881,
+      "step": 2850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734543309256052e-05,
+      "loss": 0.8066,
+      "step": 2851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973427975925692e-05,
+      "loss": 0.79,
+      "step": 2852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973401608025577e-05,
+      "loss": 0.7832,
+      "step": 2853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733752272256104e-05,
+      "loss": 0.8281,
+      "step": 2854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733488335261408e-05,
+      "loss": 0.8125,
+      "step": 2855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733224269275184e-05,
+      "loss": 0.7959,
+      "step": 2856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732960074300937e-05,
+      "loss": 0.791,
+      "step": 2857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732695750342156e-05,
+      "loss": 0.8076,
+      "step": 2858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732431297402354e-05,
+      "loss": 0.7891,
+      "step": 2859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973216671548503e-05,
+      "loss": 0.7969,
+      "step": 2860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97319020045937e-05,
+      "loss": 0.7871,
+      "step": 2861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973163716473186e-05,
+      "loss": 0.8008,
+      "step": 2862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731372195903026e-05,
+      "loss": 0.7803,
+      "step": 2863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731107098110705e-05,
+      "loss": 0.792,
+      "step": 2864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730841871358415e-05,
+      "loss": 0.7969,
+      "step": 2865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973057651564967e-05,
+      "loss": 0.8086,
+      "step": 2866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730311030987982e-05,
+      "loss": 0.7793,
+      "step": 2867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730045417376876e-05,
+      "loss": 0.832,
+      "step": 2868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729779674819875e-05,
+      "loss": 0.7949,
+      "step": 2869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729513803320484e-05,
+      "loss": 0.7764,
+      "step": 2870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729247802882242e-05,
+      "loss": 0.8057,
+      "step": 2871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972898167350867e-05,
+      "loss": 0.8037,
+      "step": 2872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728715415203297e-05,
+      "loss": 0.8086,
+      "step": 2873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728449027969644e-05,
+      "loss": 0.7871,
+      "step": 2874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972818251181125e-05,
+      "loss": 0.834,
+      "step": 2875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727915866731636e-05,
+      "loss": 0.8135,
+      "step": 2876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727649092734348e-05,
+      "loss": 0.8203,
+      "step": 2877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727382189822914e-05,
+      "loss": 0.8193,
+      "step": 2878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727115158000875e-05,
+      "loss": 0.8105,
+      "step": 2879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726847997271765e-05,
+      "loss": 0.8076,
+      "step": 2880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726580707639128e-05,
+      "loss": 0.8066,
+      "step": 2881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726313289106507e-05,
+      "loss": 0.7949,
+      "step": 2882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726045741677445e-05,
+      "loss": 0.7979,
+      "step": 2883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725778065355487e-05,
+      "loss": 0.7959,
+      "step": 2884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725510260144178e-05,
+      "loss": 0.7852,
+      "step": 2885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725242326047068e-05,
+      "loss": 0.8027,
+      "step": 2886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724974263067716e-05,
+      "loss": 0.7969,
+      "step": 2887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972470607120966e-05,
+      "loss": 0.791,
+      "step": 2888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724437750476468e-05,
+      "loss": 0.834,
+      "step": 2889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972416930087169e-05,
+      "loss": 0.8145,
+      "step": 2890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972390072239888e-05,
+      "loss": 0.7998,
+      "step": 2891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97236320150616e-05,
+      "loss": 0.7959,
+      "step": 2892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972336317886341e-05,
+      "loss": 0.7939,
+      "step": 2893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723094213807878e-05,
+      "loss": 0.8086,
+      "step": 2894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722825119898566e-05,
+      "loss": 0.7803,
+      "step": 2895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972255589713904e-05,
+      "loss": 0.7764,
+      "step": 2896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722286545532858e-05,
+      "loss": 0.8008,
+      "step": 2897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9722017065083604e-05,
+      "loss": 0.7979,
+      "step": 2898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721747455794842e-05,
+      "loss": 0.791,
+      "step": 2899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972147771767015e-05,
+      "loss": 0.8174,
+      "step": 2900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721207850713098e-05,
+      "loss": 0.7988,
+      "step": 2901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972093785492726e-05,
+      "loss": 0.8018,
+      "step": 2902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720667730316224e-05,
+      "loss": 0.7871,
+      "step": 2903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720397476883558e-05,
+      "loss": 0.832,
+      "step": 2904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972012709463285e-05,
+      "loss": 0.8145,
+      "step": 2905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719856583567686e-05,
+      "loss": 0.8105,
+      "step": 2906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971958594369164e-05,
+      "loss": 0.7842,
+      "step": 2907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971931517500831e-05,
+      "loss": 0.7959,
+      "step": 2908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971904427752128e-05,
+      "loss": 0.8057,
+      "step": 2909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718773251234137e-05,
+      "loss": 0.8135,
+      "step": 2910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718502096150477e-05,
+      "loss": 0.8125,
+      "step": 2911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718230812273892e-05,
+      "loss": 0.8057,
+      "step": 2912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971795939960798e-05,
+      "loss": 0.7803,
+      "step": 2913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971768785815633e-05,
+      "loss": 0.8047,
+      "step": 2914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717416187922553e-05,
+      "loss": 0.8027,
+      "step": 2915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717144388910233e-05,
+      "loss": 0.7842,
+      "step": 2916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716872461122986e-05,
+      "loss": 0.7725,
+      "step": 2917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971660040456441e-05,
+      "loss": 0.8047,
+      "step": 2918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971632821923811e-05,
+      "loss": 0.8047,
+      "step": 2919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716055905147694e-05,
+      "loss": 0.7842,
+      "step": 2920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971578346229677e-05,
+      "loss": 0.8105,
+      "step": 2921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715510890688953e-05,
+      "loss": 0.8213,
+      "step": 2922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971523819032785e-05,
+      "loss": 0.8076,
+      "step": 2923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714965361217076e-05,
+      "loss": 0.8027,
+      "step": 2924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714692403360248e-05,
+      "loss": 0.7803,
+      "step": 2925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714419316760982e-05,
+      "loss": 0.8066,
+      "step": 2926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714146101422896e-05,
+      "loss": 0.7979,
+      "step": 2927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971387275734961e-05,
+      "loss": 0.792,
+      "step": 2928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713599284544755e-05,
+      "loss": 0.8135,
+      "step": 2929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713325683011947e-05,
+      "loss": 0.79,
+      "step": 2930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971305195275481e-05,
+      "loss": 0.8076,
+      "step": 2931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712778093776978e-05,
+      "loss": 0.8223,
+      "step": 2932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712504106082077e-05,
+      "loss": 0.7822,
+      "step": 2933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971222998967374e-05,
+      "loss": 0.7861,
+      "step": 2934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711955744555595e-05,
+      "loss": 0.791,
+      "step": 2935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711681370731282e-05,
+      "loss": 0.8076,
+      "step": 2936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711406868204436e-05,
+      "loss": 0.8086,
+      "step": 2937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711132236978693e-05,
+      "loss": 0.7891,
+      "step": 2938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971085747705769e-05,
+      "loss": 0.7881,
+      "step": 2939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710582588445072e-05,
+      "loss": 0.8125,
+      "step": 2940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710307571144483e-05,
+      "loss": 0.7979,
+      "step": 2941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710032425159565e-05,
+      "loss": 0.7969,
+      "step": 2942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709757150493965e-05,
+      "loss": 0.8135,
+      "step": 2943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970948174715133e-05,
+      "loss": 0.8086,
+      "step": 2944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970920621513531e-05,
+      "loss": 0.7891,
+      "step": 2945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708930554449562e-05,
+      "loss": 0.8203,
+      "step": 2946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970865476509773e-05,
+      "loss": 0.7861,
+      "step": 2947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708378847083473e-05,
+      "loss": 0.7803,
+      "step": 2948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970810280041045e-05,
+      "loss": 0.7891,
+      "step": 2949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707826625082318e-05,
+      "loss": 0.8057,
+      "step": 2950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707550321102734e-05,
+      "loss": 0.8037,
+      "step": 2951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707273888475362e-05,
+      "loss": 0.7891,
+      "step": 2952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9706997327203865e-05,
+      "loss": 0.7979,
+      "step": 2953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970672063729191e-05,
+      "loss": 0.7998,
+      "step": 2954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970644381874316e-05,
+      "loss": 0.7861,
+      "step": 2955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706166871561288e-05,
+      "loss": 0.7803,
+      "step": 2956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705889795749958e-05,
+      "loss": 0.7969,
+      "step": 2957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970561259131285e-05,
+      "loss": 0.7969,
+      "step": 2958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705335258253632e-05,
+      "loss": 0.8115,
+      "step": 2959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970505779657598e-05,
+      "loss": 0.8193,
+      "step": 2960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704780206283568e-05,
+      "loss": 0.8027,
+      "step": 2961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970450248738008e-05,
+      "loss": 0.7803,
+      "step": 2962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97042246398692e-05,
+      "loss": 0.793,
+      "step": 2963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97039466637546e-05,
+      "loss": 0.8154,
+      "step": 2964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703668559039968e-05,
+      "loss": 0.79,
+      "step": 2965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703390325728994e-05,
+      "loss": 0.8223,
+      "step": 2966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703111963825358e-05,
+      "loss": 0.7822,
+      "step": 2967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702833473332755e-05,
+      "loss": 0.8066,
+      "step": 2968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970255485425487e-05,
+      "loss": 0.8027,
+      "step": 2969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97022761065954e-05,
+      "loss": 0.7998,
+      "step": 2970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701997230358037e-05,
+      "loss": 0.8057,
+      "step": 2971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701718225546477e-05,
+      "loss": 0.7861,
+      "step": 2972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970143909216442e-05,
+      "loss": 0.8057,
+      "step": 2973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970115983021556e-05,
+      "loss": 0.7861,
+      "step": 2974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700880439703604e-05,
+      "loss": 0.791,
+      "step": 2975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970060092063225e-05,
+      "loss": 0.7803,
+      "step": 2976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.97003212730052e-05,
+      "loss": 0.8184,
+      "step": 2977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700041496826165e-05,
+      "loss": 0.8027,
+      "step": 2978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699761592098853e-05,
+      "loss": 0.7949,
+      "step": 2979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699481558826973e-05,
+      "loss": 0.791,
+      "step": 2980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699201397014234e-05,
+      "loss": 0.8066,
+      "step": 2981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698921106664348e-05,
+      "loss": 0.8018,
+      "step": 2982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698640687781033e-05,
+      "loss": 0.7861,
+      "step": 2983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698360140368e-05,
+      "loss": 0.8047,
+      "step": 2984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698079464428974e-05,
+      "loss": 0.8281,
+      "step": 2985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969779865996767e-05,
+      "loss": 0.7939,
+      "step": 2986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697517726987808e-05,
+      "loss": 0.8057,
+      "step": 2987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697236665493116e-05,
+      "loss": 0.8086,
+      "step": 2988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969695547548731e-05,
+      "loss": 0.7803,
+      "step": 2989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969667415697413e-05,
+      "loss": 0.7881,
+      "step": 2990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696392709957292e-05,
+      "loss": 0.7949,
+      "step": 2991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696111134440534e-05,
+      "loss": 0.8066,
+      "step": 2992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695829430427576e-05,
+      "loss": 0.8047,
+      "step": 2993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695547597922167e-05,
+      "loss": 0.8164,
+      "step": 2994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695265636928033e-05,
+      "loss": 0.7773,
+      "step": 2995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694983547448907e-05,
+      "loss": 0.8408,
+      "step": 2996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694701329488538e-05,
+      "loss": 0.79,
+      "step": 2997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694418983050655e-05,
+      "loss": 0.7871,
+      "step": 2998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694136508139006e-05,
+      "loss": 0.7578,
+      "step": 2999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969385390475733e-05,
+      "loss": 0.8125,
+      "step": 3000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693571172909384e-05,
+      "loss": 0.8018,
+      "step": 3001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96932883125989e-05,
+      "loss": 0.792,
+      "step": 3002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693005323829633e-05,
+      "loss": 0.79,
+      "step": 3003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692722206605334e-05,
+      "loss": 0.8096,
+      "step": 3004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692438960929756e-05,
+      "loss": 0.8115,
+      "step": 3005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692155586806646e-05,
+      "loss": 0.7998,
+      "step": 3006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691872084239768e-05,
+      "loss": 0.793,
+      "step": 3007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691588453232876e-05,
+      "loss": 0.7959,
+      "step": 3008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691304693789725e-05,
+      "loss": 0.8135,
+      "step": 3009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691020805914078e-05,
+      "loss": 0.8037,
+      "step": 3010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96907367896097e-05,
+      "loss": 0.7949,
+      "step": 3011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969045264488035e-05,
+      "loss": 0.7969,
+      "step": 3012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96901683717298e-05,
+      "loss": 0.8027,
+      "step": 3013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968988397016181e-05,
+      "loss": 0.7988,
+      "step": 3014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689599440180155e-05,
+      "loss": 0.8096,
+      "step": 3015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689314781788598e-05,
+      "loss": 0.7979,
+      "step": 3016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968902999499092e-05,
+      "loss": 0.7744,
+      "step": 3017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688745079790893e-05,
+      "loss": 0.7871,
+      "step": 3018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688460036192288e-05,
+      "loss": 0.7988,
+      "step": 3019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688174864198884e-05,
+      "loss": 0.7998,
+      "step": 3020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687889563814466e-05,
+      "loss": 0.7832,
+      "step": 3021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968760413504281e-05,
+      "loss": 0.7754,
+      "step": 3022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96873185778877e-05,
+      "loss": 0.7812,
+      "step": 3023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968703289235292e-05,
+      "loss": 0.8047,
+      "step": 3024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686747078442252e-05,
+      "loss": 0.7871,
+      "step": 3025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968646113615949e-05,
+      "loss": 0.8047,
+      "step": 3026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686175065508423e-05,
+      "loss": 0.8037,
+      "step": 3027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968588886649284e-05,
+      "loss": 0.8213,
+      "step": 3028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685602539116527e-05,
+      "loss": 0.7969,
+      "step": 3029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685316083383292e-05,
+      "loss": 0.7793,
+      "step": 3030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685029499296925e-05,
+      "loss": 0.7871,
+      "step": 3031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968474278686122e-05,
+      "loss": 0.8037,
+      "step": 3032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968445594607998e-05,
+      "loss": 0.8213,
+      "step": 3033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968416897695701e-05,
+      "loss": 0.8242,
+      "step": 3034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683881879496106e-05,
+      "loss": 0.8086,
+      "step": 3035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683594653701078e-05,
+      "loss": 0.8203,
+      "step": 3036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683307299575727e-05,
+      "loss": 0.8066,
+      "step": 3037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683019817123868e-05,
+      "loss": 0.8242,
+      "step": 3038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682732206349306e-05,
+      "loss": 0.7969,
+      "step": 3039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682444467255856e-05,
+      "loss": 0.8359,
+      "step": 3040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682156599847326e-05,
+      "loss": 0.7969,
+      "step": 3041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681868604127535e-05,
+      "loss": 0.8154,
+      "step": 3042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.96815804801003e-05,
+      "loss": 0.8154,
+      "step": 3043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968129222776944e-05,
+      "loss": 0.833,
+      "step": 3044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681003847138767e-05,
+      "loss": 0.8115,
+      "step": 3045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680715338212115e-05,
+      "loss": 0.8096,
+      "step": 3046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680426700993296e-05,
+      "loss": 0.7773,
+      "step": 3047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680137935486144e-05,
+      "loss": 0.8184,
+      "step": 3048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967984904169448e-05,
+      "loss": 0.8223,
+      "step": 3049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679560019622136e-05,
+      "loss": 0.8115,
+      "step": 3050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967927086927294e-05,
+      "loss": 0.8164,
+      "step": 3051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967898159065072e-05,
+      "loss": 0.8037,
+      "step": 3052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678692183759326e-05,
+      "loss": 0.8047,
+      "step": 3053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678402648602574e-05,
+      "loss": 0.7949,
+      "step": 3054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967811298518431e-05,
+      "loss": 0.8037,
+      "step": 3055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967782319350837e-05,
+      "loss": 0.8105,
+      "step": 3056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677533273578597e-05,
+      "loss": 0.7949,
+      "step": 3057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967724322539883e-05,
+      "loss": 0.8154,
+      "step": 3058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676953048972917e-05,
+      "loss": 0.7949,
+      "step": 3059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676662744304704e-05,
+      "loss": 0.8145,
+      "step": 3060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967637231139803e-05,
+      "loss": 0.8096,
+      "step": 3061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676081750256754e-05,
+      "loss": 0.8135,
+      "step": 3062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967579106088472e-05,
+      "loss": 0.8164,
+      "step": 3063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967550024328578e-05,
+      "loss": 0.7803,
+      "step": 3064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675209297463794e-05,
+      "loss": 0.7988,
+      "step": 3065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9674918223422615e-05,
+      "loss": 0.8096,
+      "step": 3066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674627021166097e-05,
+      "loss": 0.8008,
+      "step": 3067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96743356906981e-05,
+      "loss": 0.7939,
+      "step": 3068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674044232022492e-05,
+      "loss": 0.7861,
+      "step": 3069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673752645143125e-05,
+      "loss": 0.7969,
+      "step": 3070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673460930063868e-05,
+      "loss": 0.7949,
+      "step": 3071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673169086788587e-05,
+      "loss": 0.7861,
+      "step": 3072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9672877115321154e-05,
+      "loss": 0.7852,
+      "step": 3073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967258501566543e-05,
+      "loss": 0.8076,
+      "step": 3074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9672292787825293e-05,
+      "loss": 0.7861,
+      "step": 3075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967200043180461e-05,
+      "loss": 0.7979,
+      "step": 3076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967170794760726e-05,
+      "loss": 0.7842,
+      "step": 3077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671415335237118e-05,
+      "loss": 0.8145,
+      "step": 3078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967112259469806e-05,
+      "loss": 0.8096,
+      "step": 3079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670829725993966e-05,
+      "loss": 0.7939,
+      "step": 3080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670536729128718e-05,
+      "loss": 0.7783,
+      "step": 3081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670243604106196e-05,
+      "loss": 0.7998,
+      "step": 3082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966995035093029e-05,
+      "loss": 0.7939,
+      "step": 3083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966965696960488e-05,
+      "loss": 0.792,
+      "step": 3084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9669363460133864e-05,
+      "loss": 0.8027,
+      "step": 3085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966906982252112e-05,
+      "loss": 0.8027,
+      "step": 3086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668776056770545e-05,
+      "loss": 0.7979,
+      "step": 3087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668482162886032e-05,
+      "loss": 0.7891,
+      "step": 3088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966818814087148e-05,
+      "loss": 0.7979,
+      "step": 3089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667893990730775e-05,
+      "loss": 0.79,
+      "step": 3090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667599712467823e-05,
+      "loss": 0.8076,
+      "step": 3091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667305306086524e-05,
+      "loss": 0.8057,
+      "step": 3092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667010771590773e-05,
+      "loss": 0.8105,
+      "step": 3093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966671610898448e-05,
+      "loss": 0.7949,
+      "step": 3094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966642131827155e-05,
+      "loss": 0.7969,
+      "step": 3095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9666126399455882e-05,
+      "loss": 0.7881,
+      "step": 3096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665831352541396e-05,
+      "loss": 0.8164,
+      "step": 3097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665536177531992e-05,
+      "loss": 0.8047,
+      "step": 3098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966524087443159e-05,
+      "loss": 0.7891,
+      "step": 3099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664945443244093e-05,
+      "loss": 0.8184,
+      "step": 3100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664649883973425e-05,
+      "loss": 0.8027,
+      "step": 3101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96643541966235e-05,
+      "loss": 0.8096,
+      "step": 3102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664058381198235e-05,
+      "loss": 0.791,
+      "step": 3103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663762437701557e-05,
+      "loss": 0.8018,
+      "step": 3104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663466366137377e-05,
+      "loss": 0.7832,
+      "step": 3105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663170166509628e-05,
+      "loss": 0.7969,
+      "step": 3106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966287383882223e-05,
+      "loss": 0.7852,
+      "step": 3107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966257738307911e-05,
+      "loss": 0.7832,
+      "step": 3108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662280799284204e-05,
+      "loss": 0.792,
+      "step": 3109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661984087441436e-05,
+      "loss": 0.8145,
+      "step": 3110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661687247554736e-05,
+      "loss": 0.793,
+      "step": 3111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966139027962804e-05,
+      "loss": 0.7803,
+      "step": 3112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661093183665286e-05,
+      "loss": 0.7959,
+      "step": 3113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660795959670412e-05,
+      "loss": 0.7969,
+      "step": 3114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660498607647352e-05,
+      "loss": 0.8057,
+      "step": 3115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966020112760005e-05,
+      "loss": 0.8242,
+      "step": 3116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965990351953245e-05,
+      "loss": 0.7988,
+      "step": 3117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965960578344849e-05,
+      "loss": 0.8076,
+      "step": 3118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965930791935212e-05,
+      "loss": 0.8164,
+      "step": 3119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659009927247286e-05,
+      "loss": 0.8242,
+      "step": 3120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965871180713794e-05,
+      "loss": 0.8057,
+      "step": 3121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965841355902803e-05,
+      "loss": 0.8125,
+      "step": 3122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9658115182921507e-05,
+      "loss": 0.7812,
+      "step": 3123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965781667882233e-05,
+      "loss": 0.792,
+      "step": 3124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965751804673445e-05,
+      "loss": 0.7852,
+      "step": 3125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657219286661825e-05,
+      "loss": 0.7812,
+      "step": 3126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965692039860842e-05,
+      "loss": 0.7939,
+      "step": 3127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965662138257819e-05,
+      "loss": 0.7949,
+      "step": 3128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656322238575097e-05,
+      "loss": 0.8057,
+      "step": 3129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656022966603113e-05,
+      "loss": 0.7812,
+      "step": 3130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655723566666195e-05,
+      "loss": 0.8213,
+      "step": 3131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655424038768316e-05,
+      "loss": 0.7744,
+      "step": 3132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655124382913444e-05,
+      "loss": 0.8311,
+      "step": 3133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654824599105552e-05,
+      "loss": 0.8223,
+      "step": 3134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654524687348607e-05,
+      "loss": 0.79,
+      "step": 3135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965422464764659e-05,
+      "loss": 0.792,
+      "step": 3136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653924480003475e-05,
+      "loss": 0.8057,
+      "step": 3137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965362418442324e-05,
+      "loss": 0.7832,
+      "step": 3138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653323760909866e-05,
+      "loss": 0.7949,
+      "step": 3139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653023209467334e-05,
+      "loss": 0.8057,
+      "step": 3140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652722530099618e-05,
+      "loss": 0.8008,
+      "step": 3141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652421722810716e-05,
+      "loss": 0.791,
+      "step": 3142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965212078760461e-05,
+      "loss": 0.8145,
+      "step": 3143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651819724485283e-05,
+      "loss": 0.8115,
+      "step": 3144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965151853345673e-05,
+      "loss": 0.7881,
+      "step": 3145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651217214522945e-05,
+      "loss": 0.8174,
+      "step": 3146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965091576768791e-05,
+      "loss": 0.7988,
+      "step": 3147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650614192955636e-05,
+      "loss": 0.8008,
+      "step": 3148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650312490330107e-05,
+      "loss": 0.7891,
+      "step": 3149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650010659815322e-05,
+      "loss": 0.8066,
+      "step": 3150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964970870141529e-05,
+      "loss": 0.7637,
+      "step": 3151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649406615134003e-05,
+      "loss": 0.8115,
+      "step": 3152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964910440097547e-05,
+      "loss": 0.8213,
+      "step": 3153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964880205894369e-05,
+      "loss": 0.79,
+      "step": 3154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648499589042677e-05,
+      "loss": 0.7959,
+      "step": 3155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964819699127644e-05,
+      "loss": 0.8164,
+      "step": 3156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964789426564898e-05,
+      "loss": 0.792,
+      "step": 3157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647591412164315e-05,
+      "loss": 0.7715,
+      "step": 3158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647288430826457e-05,
+      "loss": 0.8164,
+      "step": 3159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646985321639426e-05,
+      "loss": 0.8213,
+      "step": 3160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964668208460723e-05,
+      "loss": 0.7676,
+      "step": 3161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96463787197339e-05,
+      "loss": 0.7949,
+      "step": 3162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964607522702344e-05,
+      "loss": 0.832,
+      "step": 3163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645771606479888e-05,
+      "loss": 0.7871,
+      "step": 3164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645467858107256e-05,
+      "loss": 0.8037,
+      "step": 3165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645163981909573e-05,
+      "loss": 0.7998,
+      "step": 3166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964485997789087e-05,
+      "loss": 0.7939,
+      "step": 3167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644555846055175e-05,
+      "loss": 0.8096,
+      "step": 3168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644251586406514e-05,
+      "loss": 0.8096,
+      "step": 3169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964394719894892e-05,
+      "loss": 0.8066,
+      "step": 3170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643642683686426e-05,
+      "loss": 0.8057,
+      "step": 3171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643338040623073e-05,
+      "loss": 0.8223,
+      "step": 3172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643033269762896e-05,
+      "loss": 0.7861,
+      "step": 3173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964272837110993e-05,
+      "loss": 0.8066,
+      "step": 3174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964242334466822e-05,
+      "loss": 0.7969,
+      "step": 3175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642118190441804e-05,
+      "loss": 0.7871,
+      "step": 3176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641812908434733e-05,
+      "loss": 0.7959,
+      "step": 3177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641507498651045e-05,
+      "loss": 0.7803,
+      "step": 3178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641201961094794e-05,
+      "loss": 0.8076,
+      "step": 3179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640896295770023e-05,
+      "loss": 0.8184,
+      "step": 3180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964059050268079e-05,
+      "loss": 0.7812,
+      "step": 3181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640284581831138e-05,
+      "loss": 0.8096,
+      "step": 3182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639978533225128e-05,
+      "loss": 0.8037,
+      "step": 3183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639672356866817e-05,
+      "loss": 0.8115,
+      "step": 3184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639366052760257e-05,
+      "loss": 0.7969,
+      "step": 3185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963905962090951e-05,
+      "loss": 0.8037,
+      "step": 3186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963875306131864e-05,
+      "loss": 0.8125,
+      "step": 3187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638446373991705e-05,
+      "loss": 0.7988,
+      "step": 3188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963813955893277e-05,
+      "loss": 0.7988,
+      "step": 3189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637832616145906e-05,
+      "loss": 0.7842,
+      "step": 3190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963752554563517e-05,
+      "loss": 0.8408,
+      "step": 3191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637218347404642e-05,
+      "loss": 0.7822,
+      "step": 3192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963691102145839e-05,
+      "loss": 0.79,
+      "step": 3193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636603567800485e-05,
+      "loss": 0.7949,
+      "step": 3194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636295986435004e-05,
+      "loss": 0.8135,
+      "step": 3195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635988277366018e-05,
+      "loss": 0.7891,
+      "step": 3196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635680440597612e-05,
+      "loss": 0.7998,
+      "step": 3197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963537247613386e-05,
+      "loss": 0.8184,
+      "step": 3198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635064383978845e-05,
+      "loss": 0.793,
+      "step": 3199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634756164136652e-05,
+      "loss": 0.8105,
+      "step": 3200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634447816611364e-05,
+      "loss": 0.7803,
+      "step": 3201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634139341407067e-05,
+      "loss": 0.7959,
+      "step": 3202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633830738527848e-05,
+      "loss": 0.7783,
+      "step": 3203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633522007977798e-05,
+      "loss": 0.8291,
+      "step": 3204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963321314976101e-05,
+      "loss": 0.7939,
+      "step": 3205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632904163881574e-05,
+      "loss": 0.8135,
+      "step": 3206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632595050343588e-05,
+      "loss": 0.7969,
+      "step": 3207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632285809151144e-05,
+      "loss": 0.8008,
+      "step": 3208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631976440308345e-05,
+      "loss": 0.7842,
+      "step": 3209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631666943819286e-05,
+      "loss": 0.8135,
+      "step": 3210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631357319688075e-05,
+      "loss": 0.7969,
+      "step": 3211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963104756791881e-05,
+      "loss": 0.8066,
+      "step": 3212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630737688515596e-05,
+      "loss": 0.8096,
+      "step": 3213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630427681482545e-05,
+      "loss": 0.8242,
+      "step": 3214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963011754682376e-05,
+      "loss": 0.7783,
+      "step": 3215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629807284543353e-05,
+      "loss": 0.7988,
+      "step": 3216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962949689464543e-05,
+      "loss": 0.7764,
+      "step": 3217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962918637713411e-05,
+      "loss": 0.7832,
+      "step": 3218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628875732013516e-05,
+      "loss": 0.8027,
+      "step": 3219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962856495928775e-05,
+      "loss": 0.8096,
+      "step": 3220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628254058960937e-05,
+      "loss": 0.7871,
+      "step": 3221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.96279430310372e-05,
+      "loss": 0.8164,
+      "step": 3222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627631875520657e-05,
+      "loss": 0.8018,
+      "step": 3223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627320592415432e-05,
+      "loss": 0.7842,
+      "step": 3224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962700918172565e-05,
+      "loss": 0.7715,
+      "step": 3225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626697643455437e-05,
+      "loss": 0.7861,
+      "step": 3226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626385977608925e-05,
+      "loss": 0.7861,
+      "step": 3227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626074184190246e-05,
+      "loss": 0.8057,
+      "step": 3228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625762263203522e-05,
+      "loss": 0.7627,
+      "step": 3229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625450214652898e-05,
+      "loss": 0.793,
+      "step": 3230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625138038542503e-05,
+      "loss": 0.7959,
+      "step": 3231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624825734876475e-05,
+      "loss": 0.7891,
+      "step": 3232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624513303658954e-05,
+      "loss": 0.7949,
+      "step": 3233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962420074489408e-05,
+      "loss": 0.8008,
+      "step": 3234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9623888058585993e-05,
+      "loss": 0.8027,
+      "step": 3235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623575244738838e-05,
+      "loss": 0.791,
+      "step": 3236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623262303356768e-05,
+      "loss": 0.8125,
+      "step": 3237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962294923444392e-05,
+      "loss": 0.792,
+      "step": 3238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622636038004443e-05,
+      "loss": 0.7832,
+      "step": 3239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622322714042493e-05,
+      "loss": 0.7793,
+      "step": 3240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622009262562223e-05,
+      "loss": 0.7783,
+      "step": 3241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962169568356778e-05,
+      "loss": 0.7939,
+      "step": 3242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621381977063326e-05,
+      "loss": 0.7715,
+      "step": 3243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621068143053016e-05,
+      "loss": 0.8154,
+      "step": 3244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962075418154101e-05,
+      "loss": 0.7979,
+      "step": 3245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620440092531467e-05,
+      "loss": 0.7871,
+      "step": 3246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962012587602855e-05,
+      "loss": 0.7812,
+      "step": 3247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619811532036425e-05,
+      "loss": 0.7881,
+      "step": 3248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619497060559257e-05,
+      "loss": 0.791,
+      "step": 3249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961918246160121e-05,
+      "loss": 0.791,
+      "step": 3250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618867735166457e-05,
+      "loss": 0.7979,
+      "step": 3251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961855288125917e-05,
+      "loss": 0.79,
+      "step": 3252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618237899883516e-05,
+      "loss": 0.7988,
+      "step": 3253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617922791043674e-05,
+      "loss": 0.79,
+      "step": 3254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617607554743818e-05,
+      "loss": 0.7764,
+      "step": 3255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617292190988126e-05,
+      "loss": 0.7969,
+      "step": 3256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616976699780782e-05,
+      "loss": 0.8115,
+      "step": 3257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616661081125958e-05,
+      "loss": 0.7861,
+      "step": 3258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961634533502784e-05,
+      "loss": 0.7871,
+      "step": 3259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616029461490617e-05,
+      "loss": 0.7939,
+      "step": 3260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961571346051847e-05,
+      "loss": 0.7852,
+      "step": 3261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9615397332115588e-05,
+      "loss": 0.792,
+      "step": 3262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961508107628616e-05,
+      "loss": 0.7861,
+      "step": 3263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961476469303438e-05,
+      "loss": 0.7803,
+      "step": 3264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961444818236444e-05,
+      "loss": 0.791,
+      "step": 3265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961413154428053e-05,
+      "loss": 0.7871,
+      "step": 3266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961381477878685e-05,
+      "loss": 0.7881,
+      "step": 3267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613497885887602e-05,
+      "loss": 0.7793,
+      "step": 3268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613180865586973e-05,
+      "loss": 0.8008,
+      "step": 3269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612863717889177e-05,
+      "loss": 0.7861,
+      "step": 3270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961254644279841e-05,
+      "loss": 0.7949,
+      "step": 3271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961222904031888e-05,
+      "loss": 0.8096,
+      "step": 3272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611911510454795e-05,
+      "loss": 0.7705,
+      "step": 3273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611593853210356e-05,
+      "loss": 0.7959,
+      "step": 3274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611276068589778e-05,
+      "loss": 0.8135,
+      "step": 3275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961095815659727e-05,
+      "loss": 0.8135,
+      "step": 3276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961064011723705e-05,
+      "loss": 0.8184,
+      "step": 3277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610321950513324e-05,
+      "loss": 0.7832,
+      "step": 3278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610003656430316e-05,
+      "loss": 0.7754,
+      "step": 3279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960968523499224e-05,
+      "loss": 0.7803,
+      "step": 3280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609366686203318e-05,
+      "loss": 0.7793,
+      "step": 3281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609048010067775e-05,
+      "loss": 0.8213,
+      "step": 3282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608729206589828e-05,
+      "loss": 0.8057,
+      "step": 3283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.96084102757737e-05,
+      "loss": 0.791,
+      "step": 3284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960809121762363e-05,
+      "loss": 0.8076,
+      "step": 3285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960777203214383e-05,
+      "loss": 0.7764,
+      "step": 3286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960745271933854e-05,
+      "loss": 0.7988,
+      "step": 3287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960713327921199e-05,
+      "loss": 0.7881,
+      "step": 3288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606813711768415e-05,
+      "loss": 0.7725,
+      "step": 3289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606494017012047e-05,
+      "loss": 0.7842,
+      "step": 3290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960617419494712e-05,
+      "loss": 0.7939,
+      "step": 3291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605854245577883e-05,
+      "loss": 0.792,
+      "step": 3292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960553416890856e-05,
+      "loss": 0.8359,
+      "step": 3293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605213964943412e-05,
+      "loss": 0.7783,
+      "step": 3294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604893633686663e-05,
+      "loss": 0.8125,
+      "step": 3295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604573175142573e-05,
+      "loss": 0.8262,
+      "step": 3296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960425258931538e-05,
+      "loss": 0.748,
+      "step": 3297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603931876209338e-05,
+      "loss": 0.7832,
+      "step": 3298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9603611035828694e-05,
+      "loss": 0.8203,
+      "step": 3299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.96032900681777e-05,
+      "loss": 0.7783,
+      "step": 3300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602968973260614e-05,
+      "loss": 0.8154,
+      "step": 3301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602647751081685e-05,
+      "loss": 0.7666,
+      "step": 3302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960232640164517e-05,
+      "loss": 0.7842,
+      "step": 3303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602004924955334e-05,
+      "loss": 0.7783,
+      "step": 3304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601683321016433e-05,
+      "loss": 0.7998,
+      "step": 3305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960136158983273e-05,
+      "loss": 0.7988,
+      "step": 3306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601039731408485e-05,
+      "loss": 0.8037,
+      "step": 3307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960071774574797e-05,
+      "loss": 0.79,
+      "step": 3308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600395632855448e-05,
+      "loss": 0.8047,
+      "step": 3309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960007339273519e-05,
+      "loss": 0.793,
+      "step": 3310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599751025391464e-05,
+      "loss": 0.7822,
+      "step": 3311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599428530828542e-05,
+      "loss": 0.7559,
+      "step": 3312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599105909050698e-05,
+      "loss": 0.7627,
+      "step": 3313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959878316006221e-05,
+      "loss": 0.7812,
+      "step": 3314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959846028386735e-05,
+      "loss": 0.793,
+      "step": 3315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598137280470406e-05,
+      "loss": 0.8145,
+      "step": 3316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959781414987565e-05,
+      "loss": 0.792,
+      "step": 3317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597490892087364e-05,
+      "loss": 0.7988,
+      "step": 3318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959716750710984e-05,
+      "loss": 0.7871,
+      "step": 3319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959684399494736e-05,
+      "loss": 0.8262,
+      "step": 3320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596520355604204e-05,
+      "loss": 0.793,
+      "step": 3321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596196589084672e-05,
+      "loss": 0.7764,
+      "step": 3322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595872695393047e-05,
+      "loss": 0.7861,
+      "step": 3323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595548674533624e-05,
+      "loss": 0.791,
+      "step": 3324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95952245265107e-05,
+      "loss": 0.7969,
+      "step": 3325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594900251328564e-05,
+      "loss": 0.8213,
+      "step": 3326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959457584899152e-05,
+      "loss": 0.8008,
+      "step": 3327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594251319503862e-05,
+      "loss": 0.7842,
+      "step": 3328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593926662869892e-05,
+      "loss": 0.7949,
+      "step": 3329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593601879093914e-05,
+      "loss": 0.8008,
+      "step": 3330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593276968180236e-05,
+      "loss": 0.7852,
+      "step": 3331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592951930133154e-05,
+      "loss": 0.8105,
+      "step": 3332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592626764956987e-05,
+      "loss": 0.7822,
+      "step": 3333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959230147265603e-05,
+      "loss": 0.835,
+      "step": 3334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959197605323461e-05,
+      "loss": 0.7842,
+      "step": 3335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591650506697027e-05,
+      "loss": 0.8271,
+      "step": 3336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591324833047602e-05,
+      "loss": 0.7988,
+      "step": 3337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590999032290648e-05,
+      "loss": 0.7881,
+      "step": 3338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590673104430483e-05,
+      "loss": 0.7842,
+      "step": 3339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590347049471427e-05,
+      "loss": 0.8047,
+      "step": 3340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95900208674178e-05,
+      "loss": 0.7979,
+      "step": 3341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589694558273927e-05,
+      "loss": 0.7881,
+      "step": 3342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958936812204413e-05,
+      "loss": 0.7666,
+      "step": 3343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589041558732734e-05,
+      "loss": 0.7881,
+      "step": 3344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588714868344072e-05,
+      "loss": 0.7705,
+      "step": 3345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958838805088247e-05,
+      "loss": 0.8232,
+      "step": 3346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588061106352257e-05,
+      "loss": 0.7744,
+      "step": 3347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958773403475777e-05,
+      "loss": 0.7939,
+      "step": 3348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9587406836103337e-05,
+      "loss": 0.7959,
+      "step": 3349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9587079510393302e-05,
+      "loss": 0.7852,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586752057631995e-05,
+      "loss": 0.7861,
+      "step": 3351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586424477823766e-05,
+      "loss": 0.793,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9586096770972947e-05,
+      "loss": 0.8076,
+      "step": 3353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585768937083883e-05,
+      "loss": 0.792,
+      "step": 3354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958544097616092e-05,
+      "loss": 0.7988,
+      "step": 3355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585112888208404e-05,
+      "loss": 0.7764,
+      "step": 3356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584784673230683e-05,
+      "loss": 0.8008,
+      "step": 3357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584456331232107e-05,
+      "loss": 0.7803,
+      "step": 3358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584127862217028e-05,
+      "loss": 0.8213,
+      "step": 3359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583799266189795e-05,
+      "loss": 0.8154,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583470543154762e-05,
+      "loss": 0.7793,
+      "step": 3361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583141693116294e-05,
+      "loss": 0.7852,
+      "step": 3362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958281271607874e-05,
+      "loss": 0.7871,
+      "step": 3363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582483612046466e-05,
+      "loss": 0.7959,
+      "step": 3364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958215438102383e-05,
+      "loss": 0.7998,
+      "step": 3365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581825023015194e-05,
+      "loss": 0.8115,
+      "step": 3366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581495538024925e-05,
+      "loss": 0.8076,
+      "step": 3367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581165926057393e-05,
+      "loss": 0.8252,
+      "step": 3368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580836187116956e-05,
+      "loss": 0.7949,
+      "step": 3369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580506321207995e-05,
+      "loss": 0.7949,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580176328334873e-05,
+      "loss": 0.7881,
+      "step": 3371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957984620850197e-05,
+      "loss": 0.792,
+      "step": 3372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957951596171365e-05,
+      "loss": 0.7832,
+      "step": 3373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579185587974306e-05,
+      "loss": 0.8174,
+      "step": 3374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9578855087288304e-05,
+      "loss": 0.8057,
+      "step": 3375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9578524459660028e-05,
+      "loss": 0.79,
+      "step": 3376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9578193705093857e-05,
+      "loss": 0.7871,
+      "step": 3377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577862823594177e-05,
+      "loss": 0.7891,
+      "step": 3378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577531815165374e-05,
+      "loss": 0.792,
+      "step": 3379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957720067981183e-05,
+      "loss": 0.791,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576869417537936e-05,
+      "loss": 0.7988,
+      "step": 3381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957653802834808e-05,
+      "loss": 0.8115,
+      "step": 3382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576206512246662e-05,
+      "loss": 0.8242,
+      "step": 3383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575874869238064e-05,
+      "loss": 0.7861,
+      "step": 3384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575543099326685e-05,
+      "loss": 0.8184,
+      "step": 3385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575211202516926e-05,
+      "loss": 0.8223,
+      "step": 3386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574879178813182e-05,
+      "loss": 0.8135,
+      "step": 3387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957454702821985e-05,
+      "loss": 0.8096,
+      "step": 3388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957421475074134e-05,
+      "loss": 0.7959,
+      "step": 3389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573882346382045e-05,
+      "loss": 0.791,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957354981514638e-05,
+      "loss": 0.8105,
+      "step": 3391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573217157038744e-05,
+      "loss": 0.7939,
+      "step": 3392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572884372063552e-05,
+      "loss": 0.8086,
+      "step": 3393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957255146022521e-05,
+      "loss": 0.8105,
+      "step": 3394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957221842152813e-05,
+      "loss": 0.7793,
+      "step": 3395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571885255976728e-05,
+      "loss": 0.7949,
+      "step": 3396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571551963575418e-05,
+      "loss": 0.7793,
+      "step": 3397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571218544328613e-05,
+      "loss": 0.7881,
+      "step": 3398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957088499824074e-05,
+      "loss": 0.7764,
+      "step": 3399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570551325316215e-05,
+      "loss": 0.8174,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570217525559455e-05,
+      "loss": 0.8047,
+      "step": 3401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.956988359897489e-05,
+      "loss": 0.7852,
+      "step": 3402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569549545566942e-05,
+      "loss": 0.7842,
+      "step": 3403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569215365340045e-05,
+      "loss": 0.7852,
+      "step": 3404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568881058298618e-05,
+      "loss": 0.8066,
+      "step": 3405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568546624447095e-05,
+      "loss": 0.7881,
+      "step": 3406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956821206378991e-05,
+      "loss": 0.8008,
+      "step": 3407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567877376331494e-05,
+      "loss": 0.7979,
+      "step": 3408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956754256207628e-05,
+      "loss": 0.8125,
+      "step": 3409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567207621028714e-05,
+      "loss": 0.8301,
+      "step": 3410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566872553193227e-05,
+      "loss": 0.8438,
+      "step": 3411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566537358574265e-05,
+      "loss": 0.8047,
+      "step": 3412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9566202037176264e-05,
+      "loss": 0.7988,
+      "step": 3413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956586658900367e-05,
+      "loss": 0.8203,
+      "step": 3414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956553101406093e-05,
+      "loss": 0.7998,
+      "step": 3415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565195312352493e-05,
+      "loss": 0.7861,
+      "step": 3416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95648594838828e-05,
+      "loss": 0.8145,
+      "step": 3417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956452352865631e-05,
+      "loss": 0.7842,
+      "step": 3418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564187446677474e-05,
+      "loss": 0.8037,
+      "step": 3419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956385123795074e-05,
+      "loss": 0.8125,
+      "step": 3420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956351490248057e-05,
+      "loss": 0.8027,
+      "step": 3421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563178440271415e-05,
+      "loss": 0.7881,
+      "step": 3422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956284185132774e-05,
+      "loss": 0.7754,
+      "step": 3423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562505135654005e-05,
+      "loss": 0.79,
+      "step": 3424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562168293254668e-05,
+      "loss": 0.79,
+      "step": 3425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561831324134194e-05,
+      "loss": 0.7842,
+      "step": 3426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561494228297052e-05,
+      "loss": 0.7617,
+      "step": 3427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561157005747705e-05,
+      "loss": 0.7803,
+      "step": 3428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956081965649063e-05,
+      "loss": 0.7842,
+      "step": 3429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560482180530284e-05,
+      "loss": 0.7998,
+      "step": 3430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560144577871152e-05,
+      "loss": 0.7979,
+      "step": 3431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559806848517703e-05,
+      "loss": 0.8135,
+      "step": 3432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955946899247441e-05,
+      "loss": 0.7715,
+      "step": 3433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559131009745756e-05,
+      "loss": 0.8184,
+      "step": 3434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558792900336217e-05,
+      "loss": 0.8213,
+      "step": 3435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558454664250274e-05,
+      "loss": 0.8037,
+      "step": 3436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955811630149241e-05,
+      "loss": 0.792,
+      "step": 3437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955777781206711e-05,
+      "loss": 0.8018,
+      "step": 3438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557439195978857e-05,
+      "loss": 0.835,
+      "step": 3439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557100453232137e-05,
+      "loss": 0.8145,
+      "step": 3440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556761583831444e-05,
+      "loss": 0.8242,
+      "step": 3441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955642258778127e-05,
+      "loss": 0.8047,
+      "step": 3442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95560834650861e-05,
+      "loss": 0.8252,
+      "step": 3443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555744215750435e-05,
+      "loss": 0.8418,
+      "step": 3444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555404839778767e-05,
+      "loss": 0.8018,
+      "step": 3445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.95550653371756e-05,
+      "loss": 0.791,
+      "step": 3446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955472570794542e-05,
+      "loss": 0.7969,
+      "step": 3447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955438595209274e-05,
+      "loss": 0.8359,
+      "step": 3448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554046069622058e-05,
+      "loss": 0.7988,
+      "step": 3449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955370606053788e-05,
+      "loss": 0.793,
+      "step": 3450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955336592484471e-05,
+      "loss": 0.791,
+      "step": 3451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553025662547056e-05,
+      "loss": 0.7881,
+      "step": 3452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955268527364943e-05,
+      "loss": 0.8076,
+      "step": 3453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552344758156338e-05,
+      "loss": 0.7949,
+      "step": 3454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552004116072295e-05,
+      "loss": 0.8086,
+      "step": 3455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551663347401812e-05,
+      "loss": 0.7969,
+      "step": 3456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551322452149413e-05,
+      "loss": 0.8105,
+      "step": 3457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955098143031961e-05,
+      "loss": 0.7861,
+      "step": 3458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550640281916924e-05,
+      "loss": 0.7812,
+      "step": 3459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550299006945876e-05,
+      "loss": 0.7666,
+      "step": 3460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549957605410988e-05,
+      "loss": 0.7754,
+      "step": 3461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549616077316785e-05,
+      "loss": 0.7861,
+      "step": 3462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954927442266779e-05,
+      "loss": 0.8154,
+      "step": 3463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548932641468533e-05,
+      "loss": 0.79,
+      "step": 3464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548590733723548e-05,
+      "loss": 0.7822,
+      "step": 3465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548248699437356e-05,
+      "loss": 0.7773,
+      "step": 3466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95479065386145e-05,
+      "loss": 0.8213,
+      "step": 3467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547564251259507e-05,
+      "loss": 0.8281,
+      "step": 3468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954722183737692e-05,
+      "loss": 0.7959,
+      "step": 3469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954687929697127e-05,
+      "loss": 0.7881,
+      "step": 3470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95465366300471e-05,
+      "loss": 0.7773,
+      "step": 3471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954619383660895e-05,
+      "loss": 0.8145,
+      "step": 3472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545850916661365e-05,
+      "loss": 0.7852,
+      "step": 3473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545507870208884e-05,
+      "loss": 0.8193,
+      "step": 3474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954516469725606e-05,
+      "loss": 0.7852,
+      "step": 3475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544821397807435e-05,
+      "loss": 0.7959,
+      "step": 3476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544477971867565e-05,
+      "loss": 0.7754,
+      "step": 3477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544134419440996e-05,
+      "loss": 0.7822,
+      "step": 3478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543790740532283e-05,
+      "loss": 0.8125,
+      "step": 3479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954344693514598e-05,
+      "loss": 0.8232,
+      "step": 3480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543103003286645e-05,
+      "loss": 0.79,
+      "step": 3481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954275894495883e-05,
+      "loss": 0.8018,
+      "step": 3482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95424147601671e-05,
+      "loss": 0.8027,
+      "step": 3483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542070448916012e-05,
+      "loss": 0.8008,
+      "step": 3484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541726011210136e-05,
+      "loss": 0.7979,
+      "step": 3485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541381447054032e-05,
+      "loss": 0.7988,
+      "step": 3486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541036756452264e-05,
+      "loss": 0.7656,
+      "step": 3487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540691939409405e-05,
+      "loss": 0.8047,
+      "step": 3488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540346995930022e-05,
+      "loss": 0.7832,
+      "step": 3489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540001926018687e-05,
+      "loss": 0.8008,
+      "step": 3490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539656729679973e-05,
+      "loss": 0.7812,
+      "step": 3491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953931140691845e-05,
+      "loss": 0.8066,
+      "step": 3492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95389659577387e-05,
+      "loss": 0.8018,
+      "step": 3493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538620382145305e-05,
+      "loss": 0.7998,
+      "step": 3494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538274680142836e-05,
+      "loss": 0.8037,
+      "step": 3495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537928851735875e-05,
+      "loss": 0.7998,
+      "step": 3496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953758289692901e-05,
+      "loss": 0.8018,
+      "step": 3497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537236815726825e-05,
+      "loss": 0.7979,
+      "step": 3498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536890608133903e-05,
+      "loss": 0.791,
+      "step": 3499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536544274154833e-05,
+      "loss": 0.8105,
+      "step": 3500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536197813794205e-05,
+      "loss": 0.7949,
+      "step": 3501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535851227056617e-05,
+      "loss": 0.8008,
+      "step": 3502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953550451394665e-05,
+      "loss": 0.7949,
+      "step": 3503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535157674468905e-05,
+      "loss": 0.8086,
+      "step": 3504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953481070862798e-05,
+      "loss": 0.7988,
+      "step": 3505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9534463616428474e-05,
+      "loss": 0.7764,
+      "step": 3506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953411639787498e-05,
+      "loss": 0.7764,
+      "step": 3507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533769052972103e-05,
+      "loss": 0.7881,
+      "step": 3508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533421581724453e-05,
+      "loss": 0.8066,
+      "step": 3509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533073984136625e-05,
+      "loss": 0.7998,
+      "step": 3510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532726260213225e-05,
+      "loss": 0.7891,
+      "step": 3511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953237840995887e-05,
+      "loss": 0.7842,
+      "step": 3512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532030433378164e-05,
+      "loss": 0.8066,
+      "step": 3513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531682330475717e-05,
+      "loss": 0.8066,
+      "step": 3514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531334101256148e-05,
+      "loss": 0.7822,
+      "step": 3515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953098574572407e-05,
+      "loss": 0.8027,
+      "step": 3516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9530637263884092e-05,
+      "loss": 0.7822,
+      "step": 3517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9530288655740845e-05,
+      "loss": 0.8027,
+      "step": 3518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529939921298937e-05,
+      "loss": 0.8115,
+      "step": 3519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529591060562997e-05,
+      "loss": 0.8418,
+      "step": 3520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529242073537646e-05,
+      "loss": 0.8066,
+      "step": 3521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528892960227506e-05,
+      "loss": 0.7764,
+      "step": 3522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528543720637212e-05,
+      "loss": 0.8154,
+      "step": 3523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528194354771382e-05,
+      "loss": 0.8018,
+      "step": 3524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527844862634654e-05,
+      "loss": 0.7783,
+      "step": 3525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527495244231652e-05,
+      "loss": 0.7891,
+      "step": 3526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952714549956702e-05,
+      "loss": 0.7773,
+      "step": 3527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952679562864538e-05,
+      "loss": 0.793,
+      "step": 3528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526445631471376e-05,
+      "loss": 0.7979,
+      "step": 3529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952609550804965e-05,
+      "loss": 0.8066,
+      "step": 3530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525745258384834e-05,
+      "loss": 0.7939,
+      "step": 3531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525394882481572e-05,
+      "loss": 0.793,
+      "step": 3532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952504438034451e-05,
+      "loss": 0.8184,
+      "step": 3533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524693751978287e-05,
+      "loss": 0.8193,
+      "step": 3534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524342997387557e-05,
+      "loss": 0.7832,
+      "step": 3535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523992116576968e-05,
+      "loss": 0.8154,
+      "step": 3536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952364110955116e-05,
+      "loss": 0.7988,
+      "step": 3537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.95232899763148e-05,
+      "loss": 0.7959,
+      "step": 3538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522938716872525e-05,
+      "loss": 0.7744,
+      "step": 3539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522587331229005e-05,
+      "loss": 0.7891,
+      "step": 3540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522235819388885e-05,
+      "loss": 0.7822,
+      "step": 3541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521884181356828e-05,
+      "loss": 0.7852,
+      "step": 3542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521532417137496e-05,
+      "loss": 0.7686,
+      "step": 3543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952118052673555e-05,
+      "loss": 0.7988,
+      "step": 3544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520828510155653e-05,
+      "loss": 0.7822,
+      "step": 3545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520476367402464e-05,
+      "loss": 0.8027,
+      "step": 3546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520124098480657e-05,
+      "loss": 0.7939,
+      "step": 3547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519771703394902e-05,
+      "loss": 0.792,
+      "step": 3548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951941918214986e-05,
+      "loss": 0.8174,
+      "step": 3549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519066534750214e-05,
+      "loss": 0.8086,
+      "step": 3550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951871376120063e-05,
+      "loss": 0.7881,
+      "step": 3551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951836086150578e-05,
+      "loss": 0.8057,
+      "step": 3552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951800783567035e-05,
+      "loss": 0.8018,
+      "step": 3553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517654683699013e-05,
+      "loss": 0.8047,
+      "step": 3554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517301405596448e-05,
+      "loss": 0.7832,
+      "step": 3555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516948001367342e-05,
+      "loss": 0.7988,
+      "step": 3556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516594471016376e-05,
+      "loss": 0.792,
+      "step": 3557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516240814548228e-05,
+      "loss": 0.7734,
+      "step": 3558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515887031967598e-05,
+      "loss": 0.8125,
+      "step": 3559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951553312327916e-05,
+      "loss": 0.7695,
+      "step": 3560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515179088487616e-05,
+      "loss": 0.8047,
+      "step": 3561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514824927597657e-05,
+      "loss": 0.7764,
+      "step": 3562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514470640613968e-05,
+      "loss": 0.7656,
+      "step": 3563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951411622754125e-05,
+      "loss": 0.8047,
+      "step": 3564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513761688384197e-05,
+      "loss": 0.7998,
+      "step": 3565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951340702314751e-05,
+      "loss": 0.7861,
+      "step": 3566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951305223183589e-05,
+      "loss": 0.793,
+      "step": 3567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512697314454038e-05,
+      "loss": 0.7666,
+      "step": 3568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512342271006654e-05,
+      "loss": 0.7822,
+      "step": 3569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511987101498447e-05,
+      "loss": 0.7949,
+      "step": 3570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511631805934122e-05,
+      "loss": 0.7969,
+      "step": 3571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511276384318386e-05,
+      "loss": 0.7988,
+      "step": 3572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9510920836655955e-05,
+      "loss": 0.7959,
+      "step": 3573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.7803,
+      "step": 3574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510209363209845e-05,
+      "loss": 0.7861,
+      "step": 3575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509853437435596e-05,
+      "loss": 0.8154,
+      "step": 3576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509497385633503e-05,
+      "loss": 0.7969,
+      "step": 3577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950914120780829e-05,
+      "loss": 0.7969,
+      "step": 3578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950878490396467e-05,
+      "loss": 0.8115,
+      "step": 3579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508428474107374e-05,
+      "loss": 0.7969,
+      "step": 3580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950807191824112e-05,
+      "loss": 0.7998,
+      "step": 3581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507715236370633e-05,
+      "loss": 0.79,
+      "step": 3582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950735842850064e-05,
+      "loss": 0.8027,
+      "step": 3583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9507001494635875e-05,
+      "loss": 0.7852,
+      "step": 3584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506644434781057e-05,
+      "loss": 0.7852,
+      "step": 3585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950628724894093e-05,
+      "loss": 0.7949,
+      "step": 3586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505929937120216e-05,
+      "loss": 0.8066,
+      "step": 3587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950557249932366e-05,
+      "loss": 0.7773,
+      "step": 3588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505214935555997e-05,
+      "loss": 0.792,
+      "step": 3589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950485724582196e-05,
+      "loss": 0.7969,
+      "step": 3590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504499430126295e-05,
+      "loss": 0.7832,
+      "step": 3591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504141488473744e-05,
+      "loss": 0.8027,
+      "step": 3592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503783420869043e-05,
+      "loss": 0.792,
+      "step": 3593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503425227316947e-05,
+      "loss": 0.8105,
+      "step": 3594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503066907822198e-05,
+      "loss": 0.7725,
+      "step": 3595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502708462389544e-05,
+      "loss": 0.7646,
+      "step": 3596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502349891023738e-05,
+      "loss": 0.8086,
+      "step": 3597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950199119372953e-05,
+      "loss": 0.7939,
+      "step": 3598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501632370511673e-05,
+      "loss": 0.79,
+      "step": 3599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950127342137492e-05,
+      "loss": 0.7754,
+      "step": 3600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500914346324037e-05,
+      "loss": 0.8076,
+      "step": 3601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950055514536377e-05,
+      "loss": 0.7949,
+      "step": 3602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950019581849889e-05,
+      "loss": 0.8066,
+      "step": 3603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499836365734154e-05,
+      "loss": 0.7949,
+      "step": 3604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949947678707433e-05,
+      "loss": 0.7891,
+      "step": 3605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499117082524175e-05,
+      "loss": 0.792,
+      "step": 3606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498757252088463e-05,
+      "loss": 0.7812,
+      "step": 3607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498397295771957e-05,
+      "loss": 0.8027,
+      "step": 3608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498037213579435e-05,
+      "loss": 0.7852,
+      "step": 3609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949767700551566e-05,
+      "loss": 0.7646,
+      "step": 3610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497316671585415e-05,
+      "loss": 0.7949,
+      "step": 3611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496956211793466e-05,
+      "loss": 0.8027,
+      "step": 3612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496595626144595e-05,
+      "loss": 0.8086,
+      "step": 3613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949623491464358e-05,
+      "loss": 0.7871,
+      "step": 3614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495874077295202e-05,
+      "loss": 0.8203,
+      "step": 3615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495513114104237e-05,
+      "loss": 0.8027,
+      "step": 3616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949515202507548e-05,
+      "loss": 0.791,
+      "step": 3617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494790810213707e-05,
+      "loss": 0.8057,
+      "step": 3618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949442946952371e-05,
+      "loss": 0.8135,
+      "step": 3619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949406800301027e-05,
+      "loss": 0.7871,
+      "step": 3620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949370641067819e-05,
+      "loss": 0.7725,
+      "step": 3621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493344692532253e-05,
+      "loss": 0.7998,
+      "step": 3622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949298284857725e-05,
+      "loss": 0.792,
+      "step": 3623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492620878817982e-05,
+      "loss": 0.8086,
+      "step": 3624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949225878325925e-05,
+      "loss": 0.8096,
+      "step": 3625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949189656190584e-05,
+      "loss": 0.8184,
+      "step": 3626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491534214762562e-05,
+      "loss": 0.7803,
+      "step": 3627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491171741834215e-05,
+      "loss": 0.8008,
+      "step": 3628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.94908091431256e-05,
+      "loss": 0.8115,
+      "step": 3629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.949044641864153e-05,
+      "loss": 0.8213,
+      "step": 3630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490083568386807e-05,
+      "loss": 0.791,
+      "step": 3631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489720592366236e-05,
+      "loss": 0.7988,
+      "step": 3632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489357490584636e-05,
+      "loss": 0.7832,
+      "step": 3633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948899426304681e-05,
+      "loss": 0.7939,
+      "step": 3634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948863090975758e-05,
+      "loss": 0.7803,
+      "step": 3635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488267430721753e-05,
+      "loss": 0.791,
+      "step": 3636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487903825944153e-05,
+      "loss": 0.7969,
+      "step": 3637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948754009542959e-05,
+      "loss": 0.8008,
+      "step": 3638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487176239182893e-05,
+      "loss": 0.7979,
+      "step": 3639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486812257208884e-05,
+      "loss": 0.7881,
+      "step": 3640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948644814951238e-05,
+      "loss": 0.7725,
+      "step": 3641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486083916098208e-05,
+      "loss": 0.8018,
+      "step": 3642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94857195569712e-05,
+      "loss": 0.8174,
+      "step": 3643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948535507213618e-05,
+      "loss": 0.7744,
+      "step": 3644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948499046159798e-05,
+      "loss": 0.7676,
+      "step": 3645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948462572536143e-05,
+      "loss": 0.7705,
+      "step": 3646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484260863431363e-05,
+      "loss": 0.791,
+      "step": 3647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948389587581262e-05,
+      "loss": 0.7881,
+      "step": 3648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483530762510036e-05,
+      "loss": 0.7852,
+      "step": 3649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483165523528443e-05,
+      "loss": 0.7803,
+      "step": 3650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482800158872687e-05,
+      "loss": 0.7959,
+      "step": 3651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948243466854761e-05,
+      "loss": 0.7598,
+      "step": 3652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482069052558055e-05,
+      "loss": 0.8047,
+      "step": 3653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481703310908866e-05,
+      "loss": 0.8105,
+      "step": 3654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481337443604894e-05,
+      "loss": 0.8018,
+      "step": 3655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948097145065098e-05,
+      "loss": 0.8105,
+      "step": 3656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948060533205198e-05,
+      "loss": 0.7637,
+      "step": 3657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480239087812746e-05,
+      "loss": 0.7822,
+      "step": 3658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947987271793813e-05,
+      "loss": 0.79,
+      "step": 3659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947950622243299e-05,
+      "loss": 0.7549,
+      "step": 3660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479139601302175e-05,
+      "loss": 0.8076,
+      "step": 3661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478772854550548e-05,
+      "loss": 0.7988,
+      "step": 3662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478405982182977e-05,
+      "loss": 0.7744,
+      "step": 3663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947803898420431e-05,
+      "loss": 0.8135,
+      "step": 3664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477671860619426e-05,
+      "loss": 0.7773,
+      "step": 3665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477304611433177e-05,
+      "loss": 0.7754,
+      "step": 3666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476937236650436e-05,
+      "loss": 0.7764,
+      "step": 3667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947656973627607e-05,
+      "loss": 0.7803,
+      "step": 3668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947620211031495e-05,
+      "loss": 0.7686,
+      "step": 3669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475834358771945e-05,
+      "loss": 0.793,
+      "step": 3670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475466481651933e-05,
+      "loss": 0.8057,
+      "step": 3671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947509847895979e-05,
+      "loss": 0.793,
+      "step": 3672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947473035070039e-05,
+      "loss": 0.8076,
+      "step": 3673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474362096878608e-05,
+      "loss": 0.8076,
+      "step": 3674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947399371749933e-05,
+      "loss": 0.7842,
+      "step": 3675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947362521256744e-05,
+      "loss": 0.7969,
+      "step": 3676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473256582087813e-05,
+      "loss": 0.8125,
+      "step": 3677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947288782606534e-05,
+      "loss": 0.793,
+      "step": 3678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9472518944504904e-05,
+      "loss": 0.8359,
+      "step": 3679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94721499374114e-05,
+      "loss": 0.8135,
+      "step": 3680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471780804789714e-05,
+      "loss": 0.7822,
+      "step": 3681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471411546644735e-05,
+      "loss": 0.791,
+      "step": 3682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471042162981365e-05,
+      "loss": 0.8018,
+      "step": 3683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947067265380449e-05,
+      "loss": 0.7734,
+      "step": 3684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947030301911901e-05,
+      "loss": 0.7842,
+      "step": 3685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946993325892983e-05,
+      "loss": 0.8008,
+      "step": 3686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946956337324184e-05,
+      "loss": 0.8008,
+      "step": 3687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946919336205995e-05,
+      "loss": 0.7861,
+      "step": 3688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468823225389055e-05,
+      "loss": 0.7881,
+      "step": 3689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946845296323407e-05,
+      "loss": 0.8135,
+      "step": 3690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468082575599892e-05,
+      "loss": 0.7832,
+      "step": 3691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946771206249144e-05,
+      "loss": 0.8018,
+      "step": 3692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467341423913618e-05,
+      "loss": 0.7949,
+      "step": 3693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466970659871337e-05,
+      "loss": 0.7979,
+      "step": 3694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946659977036951e-05,
+      "loss": 0.7939,
+      "step": 3695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466228755413056e-05,
+      "loss": 0.7891,
+      "step": 3696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465857615006888e-05,
+      "loss": 0.7803,
+      "step": 3697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465486349155933e-05,
+      "loss": 0.7988,
+      "step": 3698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465114957865097e-05,
+      "loss": 0.8047,
+      "step": 3699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946474344113931e-05,
+      "loss": 0.8027,
+      "step": 3700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464371798983496e-05,
+      "loss": 0.8008,
+      "step": 3701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946400003140258e-05,
+      "loss": 0.7852,
+      "step": 3702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463628138401488e-05,
+      "loss": 0.7861,
+      "step": 3703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946325611998515e-05,
+      "loss": 0.7861,
+      "step": 3704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946288397615849e-05,
+      "loss": 0.7822,
+      "step": 3705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946251170692645e-05,
+      "loss": 0.7881,
+      "step": 3706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9462139312293954e-05,
+      "loss": 0.791,
+      "step": 3707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461766792265942e-05,
+      "loss": 0.7969,
+      "step": 3708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461394146847347e-05,
+      "loss": 0.7969,
+      "step": 3709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946102137604311e-05,
+      "loss": 0.8008,
+      "step": 3710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460648479858174e-05,
+      "loss": 0.7754,
+      "step": 3711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946027545829747e-05,
+      "loss": 0.7734,
+      "step": 3712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945990231136596e-05,
+      "loss": 0.7744,
+      "step": 3713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459529039068573e-05,
+      "loss": 0.8271,
+      "step": 3714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945915564141026e-05,
+      "loss": 0.7832,
+      "step": 3715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945878211839597e-05,
+      "loss": 0.7539,
+      "step": 3716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458408470030656e-05,
+      "loss": 0.7803,
+      "step": 3717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458034696319266e-05,
+      "loss": 0.7793,
+      "step": 3718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457660797266755e-05,
+      "loss": 0.8037,
+      "step": 3719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945728677287808e-05,
+      "loss": 0.7764,
+      "step": 3720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456912623158185e-05,
+      "loss": 0.7773,
+      "step": 3721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945653834811205e-05,
+      "loss": 0.791,
+      "step": 3722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456163947744616e-05,
+      "loss": 0.7979,
+      "step": 3723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455789422060855e-05,
+      "loss": 0.8408,
+      "step": 3724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945541477106573e-05,
+      "loss": 0.7734,
+      "step": 3725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94550399947642e-05,
+      "loss": 0.7695,
+      "step": 3726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9454665093161237e-05,
+      "loss": 0.7559,
+      "step": 3727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9454290066261806e-05,
+      "loss": 0.7969,
+      "step": 3728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945391491407088e-05,
+      "loss": 0.8076,
+      "step": 3729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945353963659343e-05,
+      "loss": 0.8086,
+      "step": 3730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9453164233834426e-05,
+      "loss": 0.7939,
+      "step": 3731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452788705798848e-05,
+      "loss": 0.8096,
+      "step": 3732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452413052491668e-05,
+      "loss": 0.8018,
+      "step": 3733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945203727391787e-05,
+      "loss": 0.8076,
+      "step": 3734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451661370082426e-05,
+      "loss": 0.7686,
+      "step": 3735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945128534099032e-05,
+      "loss": 0.791,
+      "step": 3736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450909186646544e-05,
+      "loss": 0.7734,
+      "step": 3737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450532907056074e-05,
+      "loss": 0.8047,
+      "step": 3738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94501565022239e-05,
+      "loss": 0.7803,
+      "step": 3739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449779972155005e-05,
+      "loss": 0.79,
+      "step": 3740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449403316854384e-05,
+      "loss": 0.8086,
+      "step": 3741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944902653632703e-05,
+      "loss": 0.7812,
+      "step": 3742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9448649630577933e-05,
+      "loss": 0.8271,
+      "step": 3743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9448272599612087e-05,
+      "loss": 0.7842,
+      "step": 3744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447895443434488e-05,
+      "loss": 0.8008,
+      "step": 3745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944751816205014e-05,
+      "loss": 0.793,
+      "step": 3746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944714075546404e-05,
+      "loss": 0.8008,
+      "step": 3747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446763223681185e-05,
+      "loss": 0.7676,
+      "step": 3748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446385566706583e-05,
+      "loss": 0.7871,
+      "step": 3749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944600778454524e-05,
+      "loss": 0.7793,
+      "step": 3750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944562987720216e-05,
+      "loss": 0.7793,
+      "step": 3751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445251844682347e-05,
+      "loss": 0.8145,
+      "step": 3752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944487368699082e-05,
+      "loss": 0.7979,
+      "step": 3753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944449540413258e-05,
+      "loss": 0.7803,
+      "step": 3754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444116996112653e-05,
+      "loss": 0.8008,
+      "step": 3755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944373846293604e-05,
+      "loss": 0.79,
+      "step": 3756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944335980460777e-05,
+      "loss": 0.8008,
+      "step": 3757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442981021132853e-05,
+      "loss": 0.8213,
+      "step": 3758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442602112516307e-05,
+      "loss": 0.8037,
+      "step": 3759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442223078763162e-05,
+      "loss": 0.7988,
+      "step": 3760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441843919878436e-05,
+      "loss": 0.8027,
+      "step": 3761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441464635867154e-05,
+      "loss": 0.8018,
+      "step": 3762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944108522673434e-05,
+      "loss": 0.8154,
+      "step": 3763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440705692485025e-05,
+      "loss": 0.8154,
+      "step": 3764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440326033124242e-05,
+      "loss": 0.7871,
+      "step": 3765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439946248657017e-05,
+      "loss": 0.8027,
+      "step": 3766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439566339088386e-05,
+      "loss": 0.7812,
+      "step": 3767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439186304423377e-05,
+      "loss": 0.8105,
+      "step": 3768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9438806144667038e-05,
+      "loss": 0.8027,
+      "step": 3769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9438425859824397e-05,
+      "loss": 0.8047,
+      "step": 3770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94380454499005e-05,
+      "loss": 0.7803,
+      "step": 3771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437664914900385e-05,
+      "loss": 0.7754,
+      "step": 3772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437284254829097e-05,
+      "loss": 0.8115,
+      "step": 3773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436903469691674e-05,
+      "loss": 0.7881,
+      "step": 3774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436522559493173e-05,
+      "loss": 0.8018,
+      "step": 3775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436141524238636e-05,
+      "loss": 0.8125,
+      "step": 3776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435760363933115e-05,
+      "loss": 0.7705,
+      "step": 3777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943537907858165e-05,
+      "loss": 0.7783,
+      "step": 3778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434997668189314e-05,
+      "loss": 0.7832,
+      "step": 3779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434616132761148e-05,
+      "loss": 0.7998,
+      "step": 3780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943423447230221e-05,
+      "loss": 0.8008,
+      "step": 3781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433852686817562e-05,
+      "loss": 0.7822,
+      "step": 3782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433470776312256e-05,
+      "loss": 0.7939,
+      "step": 3783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433088740791362e-05,
+      "loss": 0.7861,
+      "step": 3784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943270658025994e-05,
+      "loss": 0.7734,
+      "step": 3785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943232429472305e-05,
+      "loss": 0.8076,
+      "step": 3786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431941884185762e-05,
+      "loss": 0.7939,
+      "step": 3787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431559348653144e-05,
+      "loss": 0.7959,
+      "step": 3788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431176688130263e-05,
+      "loss": 0.8037,
+      "step": 3789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430793902622196e-05,
+      "loss": 0.7842,
+      "step": 3790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430410992134008e-05,
+      "loss": 0.7861,
+      "step": 3791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943002795667078e-05,
+      "loss": 0.792,
+      "step": 3792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429644796237582e-05,
+      "loss": 0.7881,
+      "step": 3793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94292615108395e-05,
+      "loss": 0.7812,
+      "step": 3794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428878100481604e-05,
+      "loss": 0.8008,
+      "step": 3795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428494565168984e-05,
+      "loss": 0.79,
+      "step": 3796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428110904906712e-05,
+      "loss": 0.8066,
+      "step": 3797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9427727119699885e-05,
+      "loss": 0.7852,
+      "step": 3798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9427343209553583e-05,
+      "loss": 0.8066,
+      "step": 3799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942695917447289e-05,
+      "loss": 0.8008,
+      "step": 3800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426575014462897e-05,
+      "loss": 0.79,
+      "step": 3801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94261907295287e-05,
+      "loss": 0.7617,
+      "step": 3802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942580631967539e-05,
+      "loss": 0.7646,
+      "step": 3803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425421784908057e-05,
+      "loss": 0.7715,
+      "step": 3804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425037125231804e-05,
+      "loss": 0.7949,
+      "step": 3805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942465234065172e-05,
+      "loss": 0.7773,
+      "step": 3806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424267431172912e-05,
+      "loss": 0.7793,
+      "step": 3807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423882396800478e-05,
+      "loss": 0.7793,
+      "step": 3808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423497237539517e-05,
+      "loss": 0.7773,
+      "step": 3809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423111953395143e-05,
+      "loss": 0.792,
+      "step": 3810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942272654437245e-05,
+      "loss": 0.7852,
+      "step": 3811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942234101047655e-05,
+      "loss": 0.8066,
+      "step": 3812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421955351712557e-05,
+      "loss": 0.7764,
+      "step": 3813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942156956808558e-05,
+      "loss": 0.8076,
+      "step": 3814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421183659600727e-05,
+      "loss": 0.79,
+      "step": 3815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420797626263114e-05,
+      "loss": 0.8018,
+      "step": 3816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942041146807786e-05,
+      "loss": 0.7939,
+      "step": 3817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942002518505008e-05,
+      "loss": 0.7881,
+      "step": 3818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419638777184893e-05,
+      "loss": 0.7695,
+      "step": 3819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419252244487418e-05,
+      "loss": 0.7744,
+      "step": 3820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9418865586962786e-05,
+      "loss": 0.8135,
+      "step": 3821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941847880461611e-05,
+      "loss": 0.8428,
+      "step": 3822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941809189745252e-05,
+      "loss": 0.7822,
+      "step": 3823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417704865477147e-05,
+      "loss": 0.791,
+      "step": 3824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417317708695117e-05,
+      "loss": 0.791,
+      "step": 3825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941693042711156e-05,
+      "loss": 0.7891,
+      "step": 3826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941654302073161e-05,
+      "loss": 0.8018,
+      "step": 3827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94161554895604e-05,
+      "loss": 0.7891,
+      "step": 3828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9415767833603065e-05,
+      "loss": 0.8174,
+      "step": 3829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9415380052864745e-05,
+      "loss": 0.7949,
+      "step": 3830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414992147350578e-05,
+      "loss": 0.7812,
+      "step": 3831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414604117065703e-05,
+      "loss": 0.8115,
+      "step": 3832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414215962015263e-05,
+      "loss": 0.7871,
+      "step": 3833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413827682204402e-05,
+      "loss": 0.7744,
+      "step": 3834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413439277638267e-05,
+      "loss": 0.791,
+      "step": 3835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413050748322003e-05,
+      "loss": 0.7842,
+      "step": 3836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412662094260765e-05,
+      "loss": 0.7783,
+      "step": 3837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941227331545969e-05,
+      "loss": 0.8057,
+      "step": 3838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411884411923946e-05,
+      "loss": 0.8066,
+      "step": 3839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941149538365868e-05,
+      "loss": 0.7559,
+      "step": 3840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411106230669045e-05,
+      "loss": 0.8135,
+      "step": 3841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94107169529602e-05,
+      "loss": 0.7998,
+      "step": 3842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410327550537304e-05,
+      "loss": 0.7695,
+      "step": 3843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940993802340552e-05,
+      "loss": 0.7793,
+      "step": 3844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409548371570005e-05,
+      "loss": 0.7969,
+      "step": 3845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409158595035928e-05,
+      "loss": 0.7666,
+      "step": 3846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408768693808453e-05,
+      "loss": 0.7695,
+      "step": 3847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408378667892745e-05,
+      "loss": 0.7871,
+      "step": 3848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407988517293973e-05,
+      "loss": 0.8066,
+      "step": 3849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407598242017315e-05,
+      "loss": 0.8027,
+      "step": 3850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940720784206793e-05,
+      "loss": 0.7832,
+      "step": 3851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406817317451e-05,
+      "loss": 0.7881,
+      "step": 3852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94064266681717e-05,
+      "loss": 0.7832,
+      "step": 3853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406035894235206e-05,
+      "loss": 0.7852,
+      "step": 3854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405644995646694e-05,
+      "loss": 0.7842,
+      "step": 3855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405253972411347e-05,
+      "loss": 0.7881,
+      "step": 3856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940486282453435e-05,
+      "loss": 0.7812,
+      "step": 3857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940447155202088e-05,
+      "loss": 0.7832,
+      "step": 3858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404080154876127e-05,
+      "loss": 0.8018,
+      "step": 3859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403688633105275e-05,
+      "loss": 0.8037,
+      "step": 3860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940329698671352e-05,
+      "loss": 0.7891,
+      "step": 3861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402905215706038e-05,
+      "loss": 0.7891,
+      "step": 3862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402513320088033e-05,
+      "loss": 0.7773,
+      "step": 3863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402121299864696e-05,
+      "loss": 0.8047,
+      "step": 3864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940172915504122e-05,
+      "loss": 0.7891,
+      "step": 3865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.94013368856228e-05,
+      "loss": 0.8027,
+      "step": 3866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400944491614642e-05,
+      "loss": 0.7832,
+      "step": 3867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940055197302194e-05,
+      "loss": 0.7773,
+      "step": 3868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400159329849895e-05,
+      "loss": 0.7852,
+      "step": 3869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399766562103716e-05,
+      "loss": 0.7803,
+      "step": 3870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399373669788605e-05,
+      "loss": 0.7695,
+      "step": 3871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398980652909764e-05,
+      "loss": 0.792,
+      "step": 3872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398587511472407e-05,
+      "loss": 0.7812,
+      "step": 3873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398194245481746e-05,
+      "loss": 0.7871,
+      "step": 3874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397800854942987e-05,
+      "loss": 0.7627,
+      "step": 3875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397407339861346e-05,
+      "loss": 0.793,
+      "step": 3876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939701370024204e-05,
+      "loss": 0.8008,
+      "step": 3877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939661993609028e-05,
+      "loss": 0.7695,
+      "step": 3878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396226047411294e-05,
+      "loss": 0.7832,
+      "step": 3879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9395832034210294e-05,
+      "loss": 0.7793,
+      "step": 3880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.93954378964925e-05,
+      "loss": 0.7588,
+      "step": 3881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939504363426314e-05,
+      "loss": 0.8057,
+      "step": 3882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939464924752744e-05,
+      "loss": 0.8018,
+      "step": 3883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394254736290622e-05,
+      "loss": 0.7715,
+      "step": 3884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393860100557917e-05,
+      "loss": 0.8008,
+      "step": 3885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393465340334553e-05,
+      "loss": 0.7734,
+      "step": 3886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393070455625765e-05,
+      "loss": 0.834,
+      "step": 3887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392675446436782e-05,
+      "loss": 0.7773,
+      "step": 3888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939228031277284e-05,
+      "loss": 0.7852,
+      "step": 3889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391885054639177e-05,
+      "loss": 0.8057,
+      "step": 3890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939148967204103e-05,
+      "loss": 0.8047,
+      "step": 3891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391094164983643e-05,
+      "loss": 0.7812,
+      "step": 3892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390698533472248e-05,
+      "loss": 0.7734,
+      "step": 3893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390302777512097e-05,
+      "loss": 0.7822,
+      "step": 3894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938990689710843e-05,
+      "loss": 0.8262,
+      "step": 3895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938951089226649e-05,
+      "loss": 0.7822,
+      "step": 3896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389114762991533e-05,
+      "loss": 0.7988,
+      "step": 3897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388718509288807e-05,
+      "loss": 0.792,
+      "step": 3898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938832213116356e-05,
+      "loss": 0.7842,
+      "step": 3899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387925628621046e-05,
+      "loss": 0.8027,
+      "step": 3900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938752900166652e-05,
+      "loss": 0.7812,
+      "step": 3901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387132250305236e-05,
+      "loss": 0.7959,
+      "step": 3902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386735374542456e-05,
+      "loss": 0.791,
+      "step": 3903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386338374383438e-05,
+      "loss": 0.7861,
+      "step": 3904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938594124983344e-05,
+      "loss": 0.7871,
+      "step": 3905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385544000897733e-05,
+      "loss": 0.7939,
+      "step": 3906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385146627581572e-05,
+      "loss": 0.7734,
+      "step": 3907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384749129890227e-05,
+      "loss": 0.7803,
+      "step": 3908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384351507828967e-05,
+      "loss": 0.7852,
+      "step": 3909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938395376140306e-05,
+      "loss": 0.7754,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938355589061778e-05,
+      "loss": 0.8145,
+      "step": 3911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938315789547839e-05,
+      "loss": 0.8037,
+      "step": 3912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382759775990178e-05,
+      "loss": 0.8203,
+      "step": 3913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382361532158413e-05,
+      "loss": 0.7988,
+      "step": 3914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938196316398837e-05,
+      "loss": 0.791,
+      "step": 3915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381564671485334e-05,
+      "loss": 0.8047,
+      "step": 3916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381166054654584e-05,
+      "loss": 0.7705,
+      "step": 3917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380767313501397e-05,
+      "loss": 0.8164,
+      "step": 3918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380368448031068e-05,
+      "loss": 0.7842,
+      "step": 3919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379969458248876e-05,
+      "loss": 0.7861,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937957034416011e-05,
+      "loss": 0.7686,
+      "step": 3921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379171105770058e-05,
+      "loss": 0.791,
+      "step": 3922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937877174308401e-05,
+      "loss": 0.7734,
+      "step": 3923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9378372256107263e-05,
+      "loss": 0.7969,
+      "step": 3924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937797264484511e-05,
+      "loss": 0.7705,
+      "step": 3925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937757290930284e-05,
+      "loss": 0.7969,
+      "step": 3926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937717304948576e-05,
+      "loss": 0.791,
+      "step": 3927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376773065399163e-05,
+      "loss": 0.7686,
+      "step": 3928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376372957048354e-05,
+      "loss": 0.7871,
+      "step": 3929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937597272443863e-05,
+      "loss": 0.8174,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.93755723675753e-05,
+      "loss": 0.8057,
+      "step": 3931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375171886463664e-05,
+      "loss": 0.7979,
+      "step": 3932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374771281109036e-05,
+      "loss": 0.8057,
+      "step": 3933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374370551516718e-05,
+      "loss": 0.7891,
+      "step": 3934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937396969769203e-05,
+      "loss": 0.792,
+      "step": 3935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373568719640276e-05,
+      "loss": 0.8125,
+      "step": 3936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937316761736677e-05,
+      "loss": 0.7891,
+      "step": 3937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937276639087683e-05,
+      "loss": 0.79,
+      "step": 3938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937236504017578e-05,
+      "loss": 0.7822,
+      "step": 3939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371963565268927e-05,
+      "loss": 0.8018,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371561966161598e-05,
+      "loss": 0.7871,
+      "step": 3941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9371160242859114e-05,
+      "loss": 0.8018,
+      "step": 3942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370758395366797e-05,
+      "loss": 0.7734,
+      "step": 3943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370356423689976e-05,
+      "loss": 0.7949,
+      "step": 3944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369954327833972e-05,
+      "loss": 0.792,
+      "step": 3945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369552107804123e-05,
+      "loss": 0.8018,
+      "step": 3946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936914976360575e-05,
+      "loss": 0.8008,
+      "step": 3947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936874729524419e-05,
+      "loss": 0.7783,
+      "step": 3948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368344702724777e-05,
+      "loss": 0.7783,
+      "step": 3949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367941986052847e-05,
+      "loss": 0.8223,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936753914523373e-05,
+      "loss": 0.8027,
+      "step": 3951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367136180272773e-05,
+      "loss": 0.8047,
+      "step": 3952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936673309117531e-05,
+      "loss": 0.7754,
+      "step": 3953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936632987794669e-05,
+      "loss": 0.79,
+      "step": 3954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365926540592248e-05,
+      "loss": 0.7812,
+      "step": 3955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365523079117335e-05,
+      "loss": 0.8047,
+      "step": 3956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365119493527295e-05,
+      "loss": 0.8076,
+      "step": 3957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936471578382748e-05,
+      "loss": 0.8105,
+      "step": 3958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364311950023236e-05,
+      "loss": 0.79,
+      "step": 3959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363907992119918e-05,
+      "loss": 0.7607,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363503910122875e-05,
+      "loss": 0.791,
+      "step": 3961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363099704037463e-05,
+      "loss": 0.792,
+      "step": 3962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362695373869047e-05,
+      "loss": 0.833,
+      "step": 3963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362290919622975e-05,
+      "loss": 0.7871,
+      "step": 3964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936188634130461e-05,
+      "loss": 0.7764,
+      "step": 3965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936148163891932e-05,
+      "loss": 0.8076,
+      "step": 3966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9361076812472456e-05,
+      "loss": 0.7861,
+      "step": 3967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360671861969394e-05,
+      "loss": 0.7617,
+      "step": 3968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360266787415494e-05,
+      "loss": 0.7744,
+      "step": 3969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9359861588816126e-05,
+      "loss": 0.7891,
+      "step": 3970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935945626617666e-05,
+      "loss": 0.7793,
+      "step": 3971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935905081950247e-05,
+      "loss": 0.7881,
+      "step": 3972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358645248798928e-05,
+      "loss": 0.793,
+      "step": 3973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358239554071404e-05,
+      "loss": 0.7881,
+      "step": 3974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935783373532528e-05,
+      "loss": 0.8027,
+      "step": 3975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935742779256593e-05,
+      "loss": 0.7568,
+      "step": 3976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357021725798736e-05,
+      "loss": 0.7695,
+      "step": 3977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935661553502908e-05,
+      "loss": 0.7627,
+      "step": 3978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356209220262343e-05,
+      "loss": 0.8223,
+      "step": 3979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935580278150391e-05,
+      "loss": 0.7773,
+      "step": 3980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935539621875917e-05,
+      "loss": 0.8184,
+      "step": 3981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354989532033507e-05,
+      "loss": 0.7705,
+      "step": 3982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354582721332313e-05,
+      "loss": 0.7881,
+      "step": 3983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935417578666098e-05,
+      "loss": 0.8174,
+      "step": 3984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353768728024896e-05,
+      "loss": 0.7822,
+      "step": 3985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935336154542946e-05,
+      "loss": 0.7939,
+      "step": 3986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935295423888007e-05,
+      "loss": 0.793,
+      "step": 3987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935254680838212e-05,
+      "loss": 0.7734,
+      "step": 3988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352139253941007e-05,
+      "loss": 0.79,
+      "step": 3989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351731575562138e-05,
+      "loss": 0.793,
+      "step": 3990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935132377325091e-05,
+      "loss": 0.7959,
+      "step": 3991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350915847012733e-05,
+      "loss": 0.7793,
+      "step": 3992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350507796853007e-05,
+      "loss": 0.7598,
+      "step": 3993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350099622777146e-05,
+      "loss": 0.793,
+      "step": 3994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9349691324790556e-05,
+      "loss": 0.7891,
+      "step": 3995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9349282902898648e-05,
+      "loss": 0.7871,
+      "step": 3996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348874357106833e-05,
+      "loss": 0.7949,
+      "step": 3997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934846568742053e-05,
+      "loss": 0.7949,
+      "step": 3998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934805689384515e-05,
+      "loss": 0.8125,
+      "step": 3999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934764797638611e-05,
+      "loss": 0.7939,
+      "step": 4000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9347238935048833e-05,
+      "loss": 0.7852,
+      "step": 4001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346829769838734e-05,
+      "loss": 0.8203,
+      "step": 4002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346420480761243e-05,
+      "loss": 0.8232,
+      "step": 4003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934601106782178e-05,
+      "loss": 0.8086,
+      "step": 4004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345601531025767e-05,
+      "loss": 0.7881,
+      "step": 4005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934519187037864e-05,
+      "loss": 0.7695,
+      "step": 4006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344782085885826e-05,
+      "loss": 0.7715,
+      "step": 4007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344372177552747e-05,
+      "loss": 0.8037,
+      "step": 4008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343962145384846e-05,
+      "loss": 0.8076,
+      "step": 4009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343551989387546e-05,
+      "loss": 0.7793,
+      "step": 4010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343141709566293e-05,
+      "loss": 0.8271,
+      "step": 4011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934273130592652e-05,
+      "loss": 0.8008,
+      "step": 4012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934232077847366e-05,
+      "loss": 0.8193,
+      "step": 4013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9341910127213167e-05,
+      "loss": 0.7871,
+      "step": 4014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934149935215047e-05,
+      "loss": 0.7949,
+      "step": 4015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934108845329102e-05,
+      "loss": 0.7822,
+      "step": 4016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340677430640258e-05,
+      "loss": 0.7588,
+      "step": 4017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340266284203635e-05,
+      "loss": 0.8086,
+      "step": 4018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93398550139866e-05,
+      "loss": 0.7939,
+      "step": 4019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93394436199946e-05,
+      "loss": 0.7998,
+      "step": 4020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9339032102233085e-05,
+      "loss": 0.7793,
+      "step": 4021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338620460707515e-05,
+      "loss": 0.7822,
+      "step": 4022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338208695423336e-05,
+      "loss": 0.8008,
+      "step": 4023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337796806386017e-05,
+      "loss": 0.7822,
+      "step": 4024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337384793601006e-05,
+      "loss": 0.7939,
+      "step": 4025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9336972657073772e-05,
+      "loss": 0.7812,
+      "step": 4026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933656039680977e-05,
+      "loss": 0.7979,
+      "step": 4027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9336148012814464e-05,
+      "loss": 0.7822,
+      "step": 4028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933573550509332e-05,
+      "loss": 0.791,
+      "step": 4029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933532287365181e-05,
+      "loss": 0.8027,
+      "step": 4030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334910118495394e-05,
+      "loss": 0.7773,
+      "step": 4031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9334497239629545e-05,
+      "loss": 0.7949,
+      "step": 4032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933408423705974e-05,
+      "loss": 0.7939,
+      "step": 4033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333671110791444e-05,
+      "loss": 0.7852,
+      "step": 4034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333257860830134e-05,
+      "loss": 0.8008,
+      "step": 4035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933284448718129e-05,
+      "loss": 0.8096,
+      "step": 4036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332430989850387e-05,
+      "loss": 0.7871,
+      "step": 4037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332017368842906e-05,
+      "loss": 0.7842,
+      "step": 4038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933160362416433e-05,
+      "loss": 0.8086,
+      "step": 4039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933118975582014e-05,
+      "loss": 0.7852,
+      "step": 4040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330775763815815e-05,
+      "loss": 0.7939,
+      "step": 4041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330361648156856e-05,
+      "loss": 0.7988,
+      "step": 4042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932994740884874e-05,
+      "loss": 0.7441,
+      "step": 4043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329533045896957e-05,
+      "loss": 0.7939,
+      "step": 4044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329118559307002e-05,
+      "loss": 0.7871,
+      "step": 4045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328703949084366e-05,
+      "loss": 0.7949,
+      "step": 4046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328289215234544e-05,
+      "loss": 0.8135,
+      "step": 4047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932787435776303e-05,
+      "loss": 0.8086,
+      "step": 4048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327459376675327e-05,
+      "loss": 0.7773,
+      "step": 4049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932704427197693e-05,
+      "loss": 0.8008,
+      "step": 4050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326629043673342e-05,
+      "loss": 0.7969,
+      "step": 4051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326213691770067e-05,
+      "loss": 0.8008,
+      "step": 4052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325798216272606e-05,
+      "loss": 0.8057,
+      "step": 4053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325382617186465e-05,
+      "loss": 0.7891,
+      "step": 4054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324966894517153e-05,
+      "loss": 0.8105,
+      "step": 4055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324551048270182e-05,
+      "loss": 0.7754,
+      "step": 4056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932413507845106e-05,
+      "loss": 0.7637,
+      "step": 4057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93237189850653e-05,
+      "loss": 0.7783,
+      "step": 4058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932330276811842e-05,
+      "loss": 0.8105,
+      "step": 4059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932288642761593e-05,
+      "loss": 0.7676,
+      "step": 4060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322469963563348e-05,
+      "loss": 0.7871,
+      "step": 4061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322053375966193e-05,
+      "loss": 0.7861,
+      "step": 4062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321636664829987e-05,
+      "loss": 0.8057,
+      "step": 4063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321219830160257e-05,
+      "loss": 0.7783,
+      "step": 4064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932080287196252e-05,
+      "loss": 0.7676,
+      "step": 4065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932038579024231e-05,
+      "loss": 0.791,
+      "step": 4066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319968585005145e-05,
+      "loss": 0.7627,
+      "step": 4067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319551256256557e-05,
+      "loss": 0.8096,
+      "step": 4068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931913380400208e-05,
+      "loss": 0.7852,
+      "step": 4069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931871622824724e-05,
+      "loss": 0.7705,
+      "step": 4070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931829852899758e-05,
+      "loss": 0.7617,
+      "step": 4071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317880706258628e-05,
+      "loss": 0.7793,
+      "step": 4072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931746276003592e-05,
+      "loss": 0.7617,
+      "step": 4073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317044690335e-05,
+      "loss": 0.7871,
+      "step": 4074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931662649716141e-05,
+      "loss": 0.8291,
+      "step": 4075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316208180520683e-05,
+      "loss": 0.8164,
+      "step": 4076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315789740418373e-05,
+      "loss": 0.7676,
+      "step": 4077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315371176860017e-05,
+      "loss": 0.7881,
+      "step": 4078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9314952489851167e-05,
+      "loss": 0.7822,
+      "step": 4079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9314533679397368e-05,
+      "loss": 0.7871,
+      "step": 4080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931411474550417e-05,
+      "loss": 0.8047,
+      "step": 4081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313695688177133e-05,
+      "loss": 0.7793,
+      "step": 4082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313276507421804e-05,
+      "loss": 0.8271,
+      "step": 4083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312857203243735e-05,
+      "loss": 0.7988,
+      "step": 4084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312437775648486e-05,
+      "loss": 0.7783,
+      "step": 4085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931201822464162e-05,
+      "loss": 0.8096,
+      "step": 4086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931159855022869e-05,
+      "loss": 0.7861,
+      "step": 4087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931117875241526e-05,
+      "loss": 0.7666,
+      "step": 4088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310758831206896e-05,
+      "loss": 0.7812,
+      "step": 4089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931033878660916e-05,
+      "loss": 0.7949,
+      "step": 4090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930991861862762e-05,
+      "loss": 0.7832,
+      "step": 4091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309498327267842e-05,
+      "loss": 0.792,
+      "step": 4092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309077912535396e-05,
+      "loss": 0.7783,
+      "step": 4093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308657374435857e-05,
+      "loss": 0.8037,
+      "step": 4094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93082367129748e-05,
+      "loss": 0.7881,
+      "step": 4095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307815928157788e-05,
+      "loss": 0.7754,
+      "step": 4096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930739501999041e-05,
+      "loss": 0.7549,
+      "step": 4097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306973988478237e-05,
+      "loss": 0.7959,
+      "step": 4098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306552833626853e-05,
+      "loss": 0.8018,
+      "step": 4099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306131555441834e-05,
+      "loss": 0.7656,
+      "step": 4100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305710153928768e-05,
+      "loss": 0.7705,
+      "step": 4101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305288629093234e-05,
+      "loss": 0.7461,
+      "step": 4102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930486698094083e-05,
+      "loss": 0.7715,
+      "step": 4103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304445209477127e-05,
+      "loss": 0.8018,
+      "step": 4104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304023314707725e-05,
+      "loss": 0.7793,
+      "step": 4105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303601296638218e-05,
+      "loss": 0.7725,
+      "step": 4106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930317915527419e-05,
+      "loss": 0.79,
+      "step": 4107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930275689062124e-05,
+      "loss": 0.8066,
+      "step": 4108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9302334502684963e-05,
+      "loss": 0.7764,
+      "step": 4109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301911991470956e-05,
+      "loss": 0.7676,
+      "step": 4110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930148935698482e-05,
+      "loss": 0.7725,
+      "step": 4111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301066599232155e-05,
+      "loss": 0.7734,
+      "step": 4112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300643718218562e-05,
+      "loss": 0.7803,
+      "step": 4113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300220713949648e-05,
+      "loss": 0.7773,
+      "step": 4114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299797586431023e-05,
+      "loss": 0.7764,
+      "step": 4115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929937433566828e-05,
+      "loss": 0.7988,
+      "step": 4116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929895096166704e-05,
+      "loss": 0.7822,
+      "step": 4117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298527464432914e-05,
+      "loss": 0.7979,
+      "step": 4118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298103843971508e-05,
+      "loss": 0.8027,
+      "step": 4119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929768010028844e-05,
+      "loss": 0.7578,
+      "step": 4120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9297256233389322e-05,
+      "loss": 0.7705,
+      "step": 4121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296832243279774e-05,
+      "loss": 0.7734,
+      "step": 4122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296408129965417e-05,
+      "loss": 0.7725,
+      "step": 4123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929598389345187e-05,
+      "loss": 0.7832,
+      "step": 4124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295559533744754e-05,
+      "loss": 0.792,
+      "step": 4125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295135050849692e-05,
+      "loss": 0.7812,
+      "step": 4126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929471044477231e-05,
+      "loss": 0.7705,
+      "step": 4127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9294285715518236e-05,
+      "loss": 0.8037,
+      "step": 4128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293860863093097e-05,
+      "loss": 0.8164,
+      "step": 4129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293435887502524e-05,
+      "loss": 0.7715,
+      "step": 4130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293010788752155e-05,
+      "loss": 0.7861,
+      "step": 4131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929258556684761e-05,
+      "loss": 0.7861,
+      "step": 4132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929216022179454e-05,
+      "loss": 0.8184,
+      "step": 4133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929173475359857e-05,
+      "loss": 0.7783,
+      "step": 4134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929130916226534e-05,
+      "loss": 0.7871,
+      "step": 4135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9290883447800494e-05,
+      "loss": 0.7988,
+      "step": 4136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.929045761020967e-05,
+      "loss": 0.8262,
+      "step": 4137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.929003164949852e-05,
+      "loss": 0.7744,
+      "step": 4138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289605565672678e-05,
+      "loss": 0.7852,
+      "step": 4139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289179358737797e-05,
+      "loss": 0.791,
+      "step": 4140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928875302869952e-05,
+      "loss": 0.8145,
+      "step": 4141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9288326575563503e-05,
+      "loss": 0.8105,
+      "step": 4142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287899999335393e-05,
+      "loss": 0.7852,
+      "step": 4143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287473300020846e-05,
+      "loss": 0.8037,
+      "step": 4144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287046477625514e-05,
+      "loss": 0.8018,
+      "step": 4145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286619532155057e-05,
+      "loss": 0.7861,
+      "step": 4146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286192463615132e-05,
+      "loss": 0.791,
+      "step": 4147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285765272011395e-05,
+      "loss": 0.79,
+      "step": 4148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285337957349513e-05,
+      "loss": 0.793,
+      "step": 4149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284910519635143e-05,
+      "loss": 0.7871,
+      "step": 4150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284482958873957e-05,
+      "loss": 0.8008,
+      "step": 4151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928405527507161e-05,
+      "loss": 0.7705,
+      "step": 4152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928362746823378e-05,
+      "loss": 0.7764,
+      "step": 4153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283199538366134e-05,
+      "loss": 0.7969,
+      "step": 4154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928277148547434e-05,
+      "loss": 0.7852,
+      "step": 4155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282343309564072e-05,
+      "loss": 0.7852,
+      "step": 4156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281915010641007e-05,
+      "loss": 0.8018,
+      "step": 4157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9281486588710816e-05,
+      "loss": 0.7891,
+      "step": 4158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928105804377918e-05,
+      "loss": 0.7949,
+      "step": 4159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280629375851784e-05,
+      "loss": 0.7822,
+      "step": 4160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280200584934298e-05,
+      "loss": 0.7529,
+      "step": 4161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279771671032408e-05,
+      "loss": 0.7881,
+      "step": 4162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92793426341518e-05,
+      "loss": 0.8027,
+      "step": 4163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927891347429816e-05,
+      "loss": 0.8223,
+      "step": 4164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278484191477173e-05,
+      "loss": 0.7969,
+      "step": 4165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927805478569453e-05,
+      "loss": 0.8115,
+      "step": 4166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927762525695592e-05,
+      "loss": 0.8037,
+      "step": 4167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277195605267042e-05,
+      "loss": 0.7998,
+      "step": 4168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276765830633575e-05,
+      "loss": 0.8027,
+      "step": 4169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927633593306123e-05,
+      "loss": 0.8174,
+      "step": 4170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275905912555696e-05,
+      "loss": 0.7803,
+      "step": 4171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275475769122674e-05,
+      "loss": 0.7998,
+      "step": 4172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275045502767865e-05,
+      "loss": 0.7852,
+      "step": 4173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274615113496965e-05,
+      "loss": 0.7773,
+      "step": 4174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274184601315688e-05,
+      "loss": 0.7979,
+      "step": 4175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273753966229734e-05,
+      "loss": 0.7803,
+      "step": 4176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927332320824481e-05,
+      "loss": 0.7627,
+      "step": 4177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272892327366617e-05,
+      "loss": 0.7871,
+      "step": 4178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272461323600883e-05,
+      "loss": 0.7969,
+      "step": 4179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272030196953304e-05,
+      "loss": 0.8164,
+      "step": 4180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92715989474296e-05,
+      "loss": 0.7939,
+      "step": 4181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271167575035485e-05,
+      "loss": 0.7832,
+      "step": 4182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270736079776677e-05,
+      "loss": 0.7822,
+      "step": 4183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270304461658887e-05,
+      "loss": 0.7832,
+      "step": 4184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269872720687846e-05,
+      "loss": 0.7793,
+      "step": 4185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926944085686927e-05,
+      "loss": 0.7832,
+      "step": 4186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269008870208883e-05,
+      "loss": 0.793,
+      "step": 4187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926857676071241e-05,
+      "loss": 0.7715,
+      "step": 4188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268144528385576e-05,
+      "loss": 0.8018,
+      "step": 4189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926771217323411e-05,
+      "loss": 0.7715,
+      "step": 4190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267279695263742e-05,
+      "loss": 0.7891,
+      "step": 4191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9266847094480205e-05,
+      "loss": 0.791,
+      "step": 4192
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9266414370889226e-05,
+      "loss": 0.7637,
+      "step": 4193
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265981524496546e-05,
+      "loss": 0.7959,
+      "step": 4194
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.92655485553079e-05,
+      "loss": 0.7695,
+      "step": 4195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265115463329025e-05,
+      "loss": 0.7783,
+      "step": 4196
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264682248565657e-05,
+      "loss": 0.8027,
+      "step": 4197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264248911023548e-05,
+      "loss": 0.7734,
+      "step": 4198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263815450708425e-05,
+      "loss": 0.8076,
+      "step": 4199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263381867626047e-05,
+      "loss": 0.7969,
+      "step": 4200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262948161782147e-05,
+      "loss": 0.751,
+      "step": 4201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262514333182482e-05,
+      "loss": 0.7578,
+      "step": 4202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262080381832798e-05,
+      "loss": 0.7793,
+      "step": 4203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261646307738848e-05,
+      "loss": 0.8018,
+      "step": 4204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926121211090638e-05,
+      "loss": 0.7891,
+      "step": 4205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9260777791341155e-05,
+      "loss": 0.793,
+      "step": 4206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9260343349048922e-05,
+      "loss": 0.7861,
+      "step": 4207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925990878403544e-05,
+      "loss": 0.7666,
+      "step": 4208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9259474096306474e-05,
+      "loss": 0.7695,
+      "step": 4209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9259039285867775e-05,
+      "loss": 0.7852,
+      "step": 4210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925860435272511e-05,
+      "loss": 0.8018,
+      "step": 4211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258169296884245e-05,
+      "loss": 0.7998,
+      "step": 4212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925773411835094e-05,
+      "loss": 0.8135,
+      "step": 4213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925729881713097e-05,
+      "loss": 0.792,
+      "step": 4214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9256863393230096e-05,
+      "loss": 0.793,
+      "step": 4215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925642784665409e-05,
+      "loss": 0.7617,
+      "step": 4216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255992177408728e-05,
+      "loss": 0.7783,
+      "step": 4217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255556385499784e-05,
+      "loss": 0.8086,
+      "step": 4218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255120470933025e-05,
+      "loss": 0.7529,
+      "step": 4219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254684433714235e-05,
+      "loss": 0.793,
+      "step": 4220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254248273849195e-05,
+      "loss": 0.7881,
+      "step": 4221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253811991343677e-05,
+      "loss": 0.7979,
+      "step": 4222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925337558620347e-05,
+      "loss": 0.7832,
+      "step": 4223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925293905843435e-05,
+      "loss": 0.8057,
+      "step": 4224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252502408042108e-05,
+      "loss": 0.7881,
+      "step": 4225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925206563503253e-05,
+      "loss": 0.8174,
+      "step": 4226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251628739411405e-05,
+      "loss": 0.7979,
+      "step": 4227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251191721184517e-05,
+      "loss": 0.7715,
+      "step": 4228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925075458035766e-05,
+      "loss": 0.7764,
+      "step": 4229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250317316936633e-05,
+      "loss": 0.8057,
+      "step": 4230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249879930927223e-05,
+      "loss": 0.7949,
+      "step": 4231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249442422335233e-05,
+      "loss": 0.7852,
+      "step": 4232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249004791166458e-05,
+      "loss": 0.7695,
+      "step": 4233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248567037426697e-05,
+      "loss": 0.7949,
+      "step": 4234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924812916112175e-05,
+      "loss": 0.7969,
+      "step": 4235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247691162257422e-05,
+      "loss": 0.7783,
+      "step": 4236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247253040839517e-05,
+      "loss": 0.7949,
+      "step": 4237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246814796873843e-05,
+      "loss": 0.7656,
+      "step": 4238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246376430366206e-05,
+      "loss": 0.7607,
+      "step": 4239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924593794132241e-05,
+      "loss": 0.7842,
+      "step": 4240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924549932974828e-05,
+      "loss": 0.8047,
+      "step": 4241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245060595649615e-05,
+      "loss": 0.8076,
+      "step": 4242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244621739032238e-05,
+      "loss": 0.8105,
+      "step": 4243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244182759901956e-05,
+      "loss": 0.8193,
+      "step": 4244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243743658264592e-05,
+      "loss": 0.7715,
+      "step": 4245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243304434125968e-05,
+      "loss": 0.7637,
+      "step": 4246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242865087491903e-05,
+      "loss": 0.8037,
+      "step": 4247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242425618368213e-05,
+      "loss": 0.7988,
+      "step": 4248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924198602676073e-05,
+      "loss": 0.7803,
+      "step": 4249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241546312675278e-05,
+      "loss": 0.7783,
+      "step": 4250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924110647611768e-05,
+      "loss": 0.8145,
+      "step": 4251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240666517093767e-05,
+      "loss": 0.8037,
+      "step": 4252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240226435609374e-05,
+      "loss": 0.7705,
+      "step": 4253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239786231670325e-05,
+      "loss": 0.8125,
+      "step": 4254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923934590528246e-05,
+      "loss": 0.7686,
+      "step": 4255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238905456451612e-05,
+      "loss": 0.7842,
+      "step": 4256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923846488518362e-05,
+      "loss": 0.7998,
+      "step": 4257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923802419148432e-05,
+      "loss": 0.7979,
+      "step": 4258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923758337535955e-05,
+      "loss": 0.791,
+      "step": 4259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923714243681516e-05,
+      "loss": 0.7754,
+      "step": 4260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236701375856987e-05,
+      "loss": 0.793,
+      "step": 4261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236260192490878e-05,
+      "loss": 0.7832,
+      "step": 4262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235818886722674e-05,
+      "loss": 0.8203,
+      "step": 4263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235377458558236e-05,
+      "loss": 0.7598,
+      "step": 4264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92349359080034e-05,
+      "loss": 0.79,
+      "step": 4265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923449423506403e-05,
+      "loss": 0.7861,
+      "step": 4266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923405243974597e-05,
+      "loss": 0.7764,
+      "step": 4267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233610522055078e-05,
+      "loss": 0.7773,
+      "step": 4268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233168481997214e-05,
+      "loss": 0.7793,
+      "step": 4269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923272631957823e-05,
+      "loss": 0.7803,
+      "step": 4270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232284034803986e-05,
+      "loss": 0.7764,
+      "step": 4271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923184162768035e-05,
+      "loss": 0.8008,
+      "step": 4272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923139909821318e-05,
+      "loss": 0.7822,
+      "step": 4273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230956446408338e-05,
+      "loss": 0.7793,
+      "step": 4274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230513672271697e-05,
+      "loss": 0.7832,
+      "step": 4275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923007077580912e-05,
+      "loss": 0.792,
+      "step": 4276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229627757026478e-05,
+      "loss": 0.7715,
+      "step": 4277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922918461592964e-05,
+      "loss": 0.7861,
+      "step": 4278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922874135252448e-05,
+      "loss": 0.7949,
+      "step": 4279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9228297966816872e-05,
+      "loss": 0.7861,
+      "step": 4280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227854458812696e-05,
+      "loss": 0.8037,
+      "step": 4281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227410828517822e-05,
+      "loss": 0.7734,
+      "step": 4282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226967075938137e-05,
+      "loss": 0.7979,
+      "step": 4283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226523201079515e-05,
+      "loss": 0.7861,
+      "step": 4284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226079203947842e-05,
+      "loss": 0.7979,
+      "step": 4285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225635084549003e-05,
+      "loss": 0.7783,
+      "step": 4286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922519084288888e-05,
+      "loss": 0.7881,
+      "step": 4287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224746478973362e-05,
+      "loss": 0.8115,
+      "step": 4288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224301992808335e-05,
+      "loss": 0.8086,
+      "step": 4289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92238573843997e-05,
+      "loss": 0.7686,
+      "step": 4290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223412653753336e-05,
+      "loss": 0.7861,
+      "step": 4291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9222967800875142e-05,
+      "loss": 0.8037,
+      "step": 4292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922252282577102e-05,
+      "loss": 0.7627,
+      "step": 4293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922207772844685e-05,
+      "loss": 0.7832,
+      "step": 4294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922163250890855e-05,
+      "loss": 0.7998,
+      "step": 4295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922118716716201e-05,
+      "loss": 0.7725,
+      "step": 4296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220741703213132e-05,
+      "loss": 0.8027,
+      "step": 4297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220296117067816e-05,
+      "loss": 0.7871,
+      "step": 4298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9219850408731976e-05,
+      "loss": 0.7939,
+      "step": 4299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9219404578211516e-05,
+      "loss": 0.791,
+      "step": 4300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921895862551234e-05,
+      "loss": 0.7451,
+      "step": 4301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921851255064036e-05,
+      "loss": 0.7891,
+      "step": 4302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218066353601492e-05,
+      "loss": 0.79,
+      "step": 4303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9217620034401642e-05,
+      "loss": 0.7803,
+      "step": 4304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921717359304673e-05,
+      "loss": 0.7822,
+      "step": 4305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921672702954267e-05,
+      "loss": 0.791,
+      "step": 4306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216280343895376e-05,
+      "loss": 0.7891,
+      "step": 4307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215833536110778e-05,
+      "loss": 0.7822,
+      "step": 4308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921538660619479e-05,
+      "loss": 0.7969,
+      "step": 4309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214939554153336e-05,
+      "loss": 0.7686,
+      "step": 4310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214492379992338e-05,
+      "loss": 0.7607,
+      "step": 4311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214045083717727e-05,
+      "loss": 0.8086,
+      "step": 4312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921359766533543e-05,
+      "loss": 0.7744,
+      "step": 4313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213150124851373e-05,
+      "loss": 0.7871,
+      "step": 4314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921270246227149e-05,
+      "loss": 0.7861,
+      "step": 4315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921225467760171e-05,
+      "loss": 0.7783,
+      "step": 4316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921180677084797e-05,
+      "loss": 0.8066,
+      "step": 4317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211358742016205e-05,
+      "loss": 0.793,
+      "step": 4318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210910591112357e-05,
+      "loss": 0.7842,
+      "step": 4319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210462318142357e-05,
+      "loss": 0.7744,
+      "step": 4320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210013923112152e-05,
+      "loss": 0.7979,
+      "step": 4321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920956540602768e-05,
+      "loss": 0.7822,
+      "step": 4322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209116766894888e-05,
+      "loss": 0.79,
+      "step": 4323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208668005719722e-05,
+      "loss": 0.7861,
+      "step": 4324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208219122508122e-05,
+      "loss": 0.7842,
+      "step": 4325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207770117266047e-05,
+      "loss": 0.7959,
+      "step": 4326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920732098999944e-05,
+      "loss": 0.7969,
+      "step": 4327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9206871740714257e-05,
+      "loss": 0.8193,
+      "step": 4328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9206422369416452e-05,
+      "loss": 0.7744,
+      "step": 4329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205972876111973e-05,
+      "loss": 0.8057,
+      "step": 4330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205523260806785e-05,
+      "loss": 0.7754,
+      "step": 4331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205073523506846e-05,
+      "loss": 0.7959,
+      "step": 4332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920462366421811e-05,
+      "loss": 0.7891,
+      "step": 4333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204173682946542e-05,
+      "loss": 0.7812,
+      "step": 4334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203723579698107e-05,
+      "loss": 0.791,
+      "step": 4335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920327335447877e-05,
+      "loss": 0.7949,
+      "step": 4336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202823007294497e-05,
+      "loss": 0.7637,
+      "step": 4337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202372538151254e-05,
+      "loss": 0.7646,
+      "step": 4338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920192194705501e-05,
+      "loss": 0.7656,
+      "step": 4339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920147123401174e-05,
+      "loss": 0.7861,
+      "step": 4340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201020399027416e-05,
+      "loss": 0.792,
+      "step": 4341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200569442108016e-05,
+      "loss": 0.7783,
+      "step": 4342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200118363259507e-05,
+      "loss": 0.7764,
+      "step": 4343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919966716248787e-05,
+      "loss": 0.7842,
+      "step": 4344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919921583979909e-05,
+      "loss": 0.7773,
+      "step": 4345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198764395199147e-05,
+      "loss": 0.7871,
+      "step": 4346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919831282869402e-05,
+      "loss": 0.7871,
+      "step": 4347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9197861140289694e-05,
+      "loss": 0.7568,
+      "step": 4348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9197409329992156e-05,
+      "loss": 0.7568,
+      "step": 4349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196957397807396e-05,
+      "loss": 0.7861,
+      "step": 4350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196505343741395e-05,
+      "loss": 0.7881,
+      "step": 4351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919605316780015e-05,
+      "loss": 0.8105,
+      "step": 4352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195600869989657e-05,
+      "loss": 0.7979,
+      "step": 4353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.91951484503159e-05,
+      "loss": 0.7822,
+      "step": 4354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194695908784885e-05,
+      "loss": 0.7871,
+      "step": 4355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194243245402603e-05,
+      "loss": 0.7969,
+      "step": 4356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193790460175052e-05,
+      "loss": 0.7715,
+      "step": 4357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9193337553108238e-05,
+      "loss": 0.7979,
+      "step": 4358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919288452420816e-05,
+      "loss": 0.7773,
+      "step": 4359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919243137348082e-05,
+      "loss": 0.7998,
+      "step": 4360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191978100932224e-05,
+      "loss": 0.7822,
+      "step": 4361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.919152470656838e-05,
+      "loss": 0.8018,
+      "step": 4362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191071190395297e-05,
+      "loss": 0.7988,
+      "step": 4363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190617552418983e-05,
+      "loss": 0.7705,
+      "step": 4364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190163792645453e-05,
+      "loss": 0.7568,
+      "step": 4365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9189709911080715e-05,
+      "loss": 0.7783,
+      "step": 4366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9189255907730793e-05,
+      "loss": 0.8086,
+      "step": 4367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188801782601697e-05,
+      "loss": 0.7686,
+      "step": 4368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188347535699444e-05,
+      "loss": 0.7676,
+      "step": 4369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9187893167030056e-05,
+      "loss": 0.7803,
+      "step": 4370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9187438676599557e-05,
+      "loss": 0.7793,
+      "step": 4371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186984064413966e-05,
+      "loss": 0.7852,
+      "step": 4372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186529330479313e-05,
+      "loss": 0.8066,
+      "step": 4373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186074474801616e-05,
+      "loss": 0.7646,
+      "step": 4374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9185619497386908e-05,
+      "loss": 0.7969,
+      "step": 4375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9185164398241223e-05,
+      "loss": 0.7959,
+      "step": 4376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9184709177370584e-05,
+      "loss": 0.7832,
+      "step": 4377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918425383478103e-05,
+      "loss": 0.7939,
+      "step": 4378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183798370478586e-05,
+      "loss": 0.792,
+      "step": 4379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91833427844693e-05,
+      "loss": 0.7793,
+      "step": 4380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91828870767592e-05,
+      "loss": 0.7832,
+      "step": 4381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918243124735433e-05,
+      "loss": 0.7812,
+      "step": 4382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181975296260732e-05,
+      "loss": 0.7812,
+      "step": 4383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181519223484447e-05,
+      "loss": 0.7539,
+      "step": 4384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181063029031516e-05,
+      "loss": 0.7939,
+      "step": 4385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9180606712907988e-05,
+      "loss": 0.7744,
+      "step": 4386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918015027511991e-05,
+      "loss": 0.7686,
+      "step": 4387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917969371567333e-05,
+      "loss": 0.7744,
+      "step": 4388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9179237034574295e-05,
+      "loss": 0.7959,
+      "step": 4389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9178780231828864e-05,
+      "loss": 0.8027,
+      "step": 4390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917832330744309e-05,
+      "loss": 0.8096,
+      "step": 4391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917786626142302e-05,
+      "loss": 0.7676,
+      "step": 4392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9177409093774718e-05,
+      "loss": 0.8018,
+      "step": 4393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176951804504242e-05,
+      "loss": 0.7842,
+      "step": 4394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176494393617653e-05,
+      "loss": 0.8018,
+      "step": 4395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176036861121012e-05,
+      "loss": 0.7891,
+      "step": 4396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175579207020377e-05,
+      "loss": 0.8086,
+      "step": 4397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917512143132182e-05,
+      "loss": 0.8018,
+      "step": 4398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174663534031407e-05,
+      "loss": 0.79,
+      "step": 4399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174205515155203e-05,
+      "loss": 0.7705,
+      "step": 4400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173747374699282e-05,
+      "loss": 0.8086,
+      "step": 4401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173289112669705e-05,
+      "loss": 0.8037,
+      "step": 4402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9172830729072556e-05,
+      "loss": 0.7842,
+      "step": 4403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917237222391391e-05,
+      "loss": 0.8027,
+      "step": 4404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171913597199838e-05,
+      "loss": 0.793,
+      "step": 4405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917145484893642e-05,
+      "loss": 0.8281,
+      "step": 4406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917099597912973e-05,
+      "loss": 0.8066,
+      "step": 4407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917053698778586e-05,
+      "loss": 0.7842,
+      "step": 4408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170077874910884e-05,
+      "loss": 0.8145,
+      "step": 4409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169618640510888e-05,
+      "loss": 0.7959,
+      "step": 4410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916915928459196e-05,
+      "loss": 0.7842,
+      "step": 4411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9168699807160188e-05,
+      "loss": 0.792,
+      "step": 4412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9168240208221657e-05,
+      "loss": 0.7744,
+      "step": 4413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167780487782464e-05,
+      "loss": 0.7979,
+      "step": 4414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167320645848694e-05,
+      "loss": 0.7881,
+      "step": 4415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166860682426448e-05,
+      "loss": 0.7842,
+      "step": 4416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916640059752182e-05,
+      "loss": 0.7832,
+      "step": 4417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165940391140897e-05,
+      "loss": 0.7607,
+      "step": 4418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165480063289794e-05,
+      "loss": 0.7744,
+      "step": 4419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165019613974602e-05,
+      "loss": 0.7773,
+      "step": 4420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164559043201425e-05,
+      "loss": 0.792,
+      "step": 4421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164098350976368e-05,
+      "loss": 0.7529,
+      "step": 4422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9163637537305533e-05,
+      "loss": 0.7793,
+      "step": 4423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9163176602195025e-05,
+      "loss": 0.7793,
+      "step": 4424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162715545650964e-05,
+      "loss": 0.7939,
+      "step": 4425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162254367679444e-05,
+      "loss": 0.7783,
+      "step": 4426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916179306828659e-05,
+      "loss": 0.7822,
+      "step": 4427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916133164747851e-05,
+      "loss": 0.8076,
+      "step": 4428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916087010526132e-05,
+      "loss": 0.7725,
+      "step": 4429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916040844164113e-05,
+      "loss": 0.7969,
+      "step": 4430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915994665662407e-05,
+      "loss": 0.7939,
+      "step": 4431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915948475021625e-05,
+      "loss": 0.8076,
+      "step": 4432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159022722423793e-05,
+      "loss": 0.7998,
+      "step": 4433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915856057325283e-05,
+      "loss": 0.7725,
+      "step": 4434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158098302709477e-05,
+      "loss": 0.8203,
+      "step": 4435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157635910799858e-05,
+      "loss": 0.7803,
+      "step": 4436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915717339753011e-05,
+      "loss": 0.7959,
+      "step": 4437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156710762906353e-05,
+      "loss": 0.7803,
+      "step": 4438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915624800693473e-05,
+      "loss": 0.7725,
+      "step": 4439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915578512962136e-05,
+      "loss": 0.7891,
+      "step": 4440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155322130972384e-05,
+      "loss": 0.7959,
+      "step": 4441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915485901099394e-05,
+      "loss": 0.8008,
+      "step": 4442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915439576969216e-05,
+      "loss": 0.7969,
+      "step": 4443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915393240707319e-05,
+      "loss": 0.7617,
+      "step": 4444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915346892314316e-05,
+      "loss": 0.7939,
+      "step": 4445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153005317908224e-05,
+      "loss": 0.751,
+      "step": 4446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915254159137452e-05,
+      "loss": 0.7744,
+      "step": 4447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915207774354819e-05,
+      "loss": 0.8096,
+      "step": 4448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151613774435387e-05,
+      "loss": 0.7939,
+      "step": 4449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151149684042257e-05,
+      "loss": 0.7803,
+      "step": 4450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9150685472374952e-05,
+      "loss": 0.7852,
+      "step": 4451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9150221139439623e-05,
+      "loss": 0.7861,
+      "step": 4452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149756685242425e-05,
+      "loss": 0.7793,
+      "step": 4453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149292109789508e-05,
+      "loss": 0.7627,
+      "step": 4454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9148827413087034e-05,
+      "loss": 0.7793,
+      "step": 4455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914836259514116e-05,
+      "loss": 0.8076,
+      "step": 4456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147897655958044e-05,
+      "loss": 0.8076,
+      "step": 4457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147432595543847e-05,
+      "loss": 0.7881,
+      "step": 4458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146967413904738e-05,
+      "loss": 0.7988,
+      "step": 4459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146502111046876e-05,
+      "loss": 0.8203,
+      "step": 4460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146036686976427e-05,
+      "loss": 0.7725,
+      "step": 4461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145571141699565e-05,
+      "loss": 0.7803,
+      "step": 4462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145105475222456e-05,
+      "loss": 0.7715,
+      "step": 4463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914463968755127e-05,
+      "loss": 0.7646,
+      "step": 4464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914417377869218e-05,
+      "loss": 0.7871,
+      "step": 4465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914370774865136e-05,
+      "loss": 0.7617,
+      "step": 4466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143241597434986e-05,
+      "loss": 0.7812,
+      "step": 4467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914277532504924e-05,
+      "loss": 0.7871,
+      "step": 4468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91423089315003e-05,
+      "loss": 0.7725,
+      "step": 4469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141842416794337e-05,
+      "loss": 0.7715,
+      "step": 4470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141375780937544e-05,
+      "loss": 0.7637,
+      "step": 4471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9140909023936104e-05,
+      "loss": 0.7803,
+      "step": 4472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9140442145796202e-05,
+      "loss": 0.7832,
+      "step": 4473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139975146524023e-05,
+      "loss": 0.7734,
+      "step": 4474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139508026125756e-05,
+      "loss": 0.7637,
+      "step": 4475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913904078460759e-05,
+      "loss": 0.7764,
+      "step": 4476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913857342197572e-05,
+      "loss": 0.8213,
+      "step": 4477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138105938236338e-05,
+      "loss": 0.8174,
+      "step": 4478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9137638333395646e-05,
+      "loss": 0.7686,
+      "step": 4479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913717060745983e-05,
+      "loss": 0.8027,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136702760435092e-05,
+      "loss": 0.7617,
+      "step": 4481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136234792327637e-05,
+      "loss": 0.7939,
+      "step": 4482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135766703143663e-05,
+      "loss": 0.7686,
+      "step": 4483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913529849288937e-05,
+      "loss": 0.7803,
+      "step": 4484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134830161570973e-05,
+      "loss": 0.7773,
+      "step": 4485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134361709194667e-05,
+      "loss": 0.7822,
+      "step": 4486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9133893135766667e-05,
+      "loss": 0.7812,
+      "step": 4487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913342444129318e-05,
+      "loss": 0.792,
+      "step": 4488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132955625780415e-05,
+      "loss": 0.8164,
+      "step": 4489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132486689234594e-05,
+      "loss": 0.7725,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132017631661917e-05,
+      "loss": 0.79,
+      "step": 4491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9131548453068616e-05,
+      "loss": 0.7812,
+      "step": 4492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91310791534609e-05,
+      "loss": 0.7783,
+      "step": 4493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9130609732844985e-05,
+      "loss": 0.7695,
+      "step": 4494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91301401912271e-05,
+      "loss": 0.7842,
+      "step": 4495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912967052861346e-05,
+      "loss": 0.7822,
+      "step": 4496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91292007450103e-05,
+      "loss": 0.7773,
+      "step": 4497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128730840423835e-05,
+      "loss": 0.7881,
+      "step": 4498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128260814860296e-05,
+      "loss": 0.7803,
+      "step": 4499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912779066832591e-05,
+      "loss": 0.7793,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127320400826914e-05,
+      "loss": 0.7764,
+      "step": 4501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126850012369535e-05,
+      "loss": 0.7598,
+      "step": 4502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126379502960005e-05,
+      "loss": 0.751,
+      "step": 4503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125908872604565e-05,
+      "loss": 0.7861,
+      "step": 4504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912543812130945e-05,
+      "loss": 0.7773,
+      "step": 4505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912496724908089e-05,
+      "loss": 0.79,
+      "step": 4506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912449625592514e-05,
+      "loss": 0.7832,
+      "step": 4507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912402514184843e-05,
+      "loss": 0.8213,
+      "step": 4508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912355390685701e-05,
+      "loss": 0.7881,
+      "step": 4509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912308255095712e-05,
+      "loss": 0.7861,
+      "step": 4510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912261107415501e-05,
+      "loss": 0.7725,
+      "step": 4511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9122139476456932e-05,
+      "loss": 0.7939,
+      "step": 4512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9121667757869126e-05,
+      "loss": 0.7705,
+      "step": 4513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912119591839785e-05,
+      "loss": 0.7715,
+      "step": 4514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120723958049355e-05,
+      "loss": 0.7803,
+      "step": 4515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120251876829896e-05,
+      "loss": 0.7803,
+      "step": 4516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911977967474573e-05,
+      "loss": 0.8047,
+      "step": 4517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119307351803114e-05,
+      "loss": 0.8125,
+      "step": 4518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911883490800831e-05,
+      "loss": 0.7939,
+      "step": 4519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9118362343367573e-05,
+      "loss": 0.7754,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911788965788717e-05,
+      "loss": 0.7861,
+      "step": 4521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117416851573364e-05,
+      "loss": 0.792,
+      "step": 4522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116943924432423e-05,
+      "loss": 0.7666,
+      "step": 4523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911647087647061e-05,
+      "loss": 0.792,
+      "step": 4524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115997707694198e-05,
+      "loss": 0.8037,
+      "step": 4525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911552441810945e-05,
+      "loss": 0.8184,
+      "step": 4526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115051007722652e-05,
+      "loss": 0.7773,
+      "step": 4527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114577476540067e-05,
+      "loss": 0.8125,
+      "step": 4528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114103824567975e-05,
+      "loss": 0.7539,
+      "step": 4529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911363005181265e-05,
+      "loss": 0.7871,
+      "step": 4530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113156158280373e-05,
+      "loss": 0.7979,
+      "step": 4531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112682143977422e-05,
+      "loss": 0.7637,
+      "step": 4532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911220800891008e-05,
+      "loss": 0.79,
+      "step": 4533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911173375308463e-05,
+      "loss": 0.751,
+      "step": 4534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911125937650736e-05,
+      "loss": 0.8047,
+      "step": 4535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911078487918455e-05,
+      "loss": 0.7881,
+      "step": 4536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110310261122497e-05,
+      "loss": 0.7637,
+      "step": 4537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910983552232748e-05,
+      "loss": 0.8027,
+      "step": 4538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9109360662805803e-05,
+      "loss": 0.7969,
+      "step": 4539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910888568256375e-05,
+      "loss": 0.7734,
+      "step": 4540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108410581607613e-05,
+      "loss": 0.8135,
+      "step": 4541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91079353599437e-05,
+      "loss": 0.7803,
+      "step": 4542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107460017578295e-05,
+      "loss": 0.7783,
+      "step": 4543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910698455451771e-05,
+      "loss": 0.7969,
+      "step": 4544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910650897076824e-05,
+      "loss": 0.7891,
+      "step": 4545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106033266336188e-05,
+      "loss": 0.8018,
+      "step": 4546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910555744122786e-05,
+      "loss": 0.7852,
+      "step": 4547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105081495449553e-05,
+      "loss": 0.8008,
+      "step": 4548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910460542900759e-05,
+      "loss": 0.79,
+      "step": 4549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104129241908264e-05,
+      "loss": 0.8018,
+      "step": 4550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103652934157896e-05,
+      "loss": 0.8008,
+      "step": 4551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103176505762795e-05,
+      "loss": 0.7695,
+      "step": 4552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102699956729277e-05,
+      "loss": 0.7832,
+      "step": 4553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910222328706365e-05,
+      "loss": 0.7734,
+      "step": 4554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101746496772243e-05,
+      "loss": 0.7686,
+      "step": 4555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101269585861364e-05,
+      "loss": 0.7754,
+      "step": 4556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100792554337338e-05,
+      "loss": 0.7695,
+      "step": 4557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9100315402206485e-05,
+      "loss": 0.8096,
+      "step": 4558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909983812947513e-05,
+      "loss": 0.7969,
+      "step": 4559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099360736149598e-05,
+      "loss": 0.7988,
+      "step": 4560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098883222236212e-05,
+      "loss": 0.7646,
+      "step": 4561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098405587741305e-05,
+      "loss": 0.7568,
+      "step": 4562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097927832671208e-05,
+      "loss": 0.7617,
+      "step": 4563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909744995703225e-05,
+      "loss": 0.7793,
+      "step": 4564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096971960830757e-05,
+      "loss": 0.7861,
+      "step": 4565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909649384407307e-05,
+      "loss": 0.7734,
+      "step": 4566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909601560676553e-05,
+      "loss": 0.7861,
+      "step": 4567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9095537248914464e-05,
+      "loss": 0.7803,
+      "step": 4568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909505877052622e-05,
+      "loss": 0.7861,
+      "step": 4569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094580171607137e-05,
+      "loss": 0.7764,
+      "step": 4570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094101452163555e-05,
+      "loss": 0.7812,
+      "step": 4571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909362261220182e-05,
+      "loss": 0.7656,
+      "step": 4572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9093143651728277e-05,
+      "loss": 0.7861,
+      "step": 4573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9092664570749275e-05,
+      "loss": 0.8105,
+      "step": 4574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909218536927116e-05,
+      "loss": 0.7773,
+      "step": 4575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091706047300283e-05,
+      "loss": 0.7881,
+      "step": 4576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091226604843004e-05,
+      "loss": 0.7803,
+      "step": 4577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090747041905663e-05,
+      "loss": 0.7881,
+      "step": 4578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909026735849463e-05,
+      "loss": 0.7783,
+      "step": 4579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9089787554616248e-05,
+      "loss": 0.7871,
+      "step": 4580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908930763027688e-05,
+      "loss": 0.7832,
+      "step": 4581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088827585482895e-05,
+      "loss": 0.7803,
+      "step": 4582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908834742024064e-05,
+      "loss": 0.7695,
+      "step": 4583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087867134556493e-05,
+      "loss": 0.7861,
+      "step": 4584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087386728436813e-05,
+      "loss": 0.7744,
+      "step": 4585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908690620188796e-05,
+      "loss": 0.7676,
+      "step": 4586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908642555491631e-05,
+      "loss": 0.7998,
+      "step": 4587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085944787528232e-05,
+      "loss": 0.8242,
+      "step": 4588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085463899730093e-05,
+      "loss": 0.7754,
+      "step": 4589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908498289152827e-05,
+      "loss": 0.7852,
+      "step": 4590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084501762929134e-05,
+      "loss": 0.7803,
+      "step": 4591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084020513939064e-05,
+      "loss": 0.7949,
+      "step": 4592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908353914456444e-05,
+      "loss": 0.8018,
+      "step": 4593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9083057654811633e-05,
+      "loss": 0.8057,
+      "step": 4594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082576044687033e-05,
+      "loss": 0.7666,
+      "step": 4595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082094314197014e-05,
+      "loss": 0.7773,
+      "step": 4596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081612463347967e-05,
+      "loss": 0.7959,
+      "step": 4597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081130492146273e-05,
+      "loss": 0.7861,
+      "step": 4598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080648400598326e-05,
+      "loss": 0.7881,
+      "step": 4599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080166188710505e-05,
+      "loss": 0.8018,
+      "step": 4600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907968385648921e-05,
+      "loss": 0.7891,
+      "step": 4601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079201403940825e-05,
+      "loss": 0.7744,
+      "step": 4602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078718831071746e-05,
+      "loss": 0.7773,
+      "step": 4603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078236137888376e-05,
+      "loss": 0.7705,
+      "step": 4604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9077753324397102e-05,
+      "loss": 0.7812,
+      "step": 4605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9077270390604323e-05,
+      "loss": 0.7832,
+      "step": 4606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076787336516443e-05,
+      "loss": 0.7568,
+      "step": 4607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076304162139863e-05,
+      "loss": 0.7686,
+      "step": 4608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9075820867480986e-05,
+      "loss": 0.8115,
+      "step": 4609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9075337452546215e-05,
+      "loss": 0.7656,
+      "step": 4610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907485391734196e-05,
+      "loss": 0.7793,
+      "step": 4611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074370261874628e-05,
+      "loss": 0.792,
+      "step": 4612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073886486150622e-05,
+      "loss": 0.7959,
+      "step": 4613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907340259017636e-05,
+      "loss": 0.7744,
+      "step": 4614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072918573958254e-05,
+      "loss": 0.7969,
+      "step": 4615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907243443750272e-05,
+      "loss": 0.7852,
+      "step": 4616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9071950180816167e-05,
+      "loss": 0.7715,
+      "step": 4617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907146580390502e-05,
+      "loss": 0.7715,
+      "step": 4618
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070981306775695e-05,
+      "loss": 0.7666,
+      "step": 4619
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907049668943461e-05,
+      "loss": 0.7725,
+      "step": 4620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907001195188819e-05,
+      "loss": 0.7773,
+      "step": 4621
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069527094142862e-05,
+      "loss": 0.7959,
+      "step": 4622
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069042116205048e-05,
+      "loss": 0.7656,
+      "step": 4623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068557018081173e-05,
+      "loss": 0.792,
+      "step": 4624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068071799777672e-05,
+      "loss": 0.7764,
+      "step": 4625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906758646130097e-05,
+      "loss": 0.793,
+      "step": 4626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90671010026575e-05,
+      "loss": 0.7871,
+      "step": 4627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066615423853693e-05,
+      "loss": 0.7725,
+      "step": 4628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066129724895993e-05,
+      "loss": 0.7773,
+      "step": 4629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065643905790827e-05,
+      "loss": 0.7812,
+      "step": 4630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065157966544635e-05,
+      "loss": 0.8096,
+      "step": 4631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9064671907163865e-05,
+      "loss": 0.7871,
+      "step": 4632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9064185727654948e-05,
+      "loss": 0.8057,
+      "step": 4633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063699428024328e-05,
+      "loss": 0.7812,
+      "step": 4634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906321300827846e-05,
+      "loss": 0.791,
+      "step": 4635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906272646842378e-05,
+      "loss": 0.8164,
+      "step": 4636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062239808466736e-05,
+      "loss": 0.7891,
+      "step": 4637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906175302841378e-05,
+      "loss": 0.8018,
+      "step": 4638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061266128271366e-05,
+      "loss": 0.7939,
+      "step": 4639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060779108045945e-05,
+      "loss": 0.7725,
+      "step": 4640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060291967743968e-05,
+      "loss": 0.7705,
+      "step": 4641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9059804707371893e-05,
+      "loss": 0.7861,
+      "step": 4642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9059317326936173e-05,
+      "loss": 0.7852,
+      "step": 4643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058829826443274e-05,
+      "loss": 0.7734,
+      "step": 4644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058342205899657e-05,
+      "loss": 0.7822,
+      "step": 4645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9057854465311773e-05,
+      "loss": 0.7852,
+      "step": 4646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90573666046861e-05,
+      "loss": 0.7754,
+      "step": 4647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905687862402909e-05,
+      "loss": 0.7881,
+      "step": 4648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056390523347223e-05,
+      "loss": 0.8008,
+      "step": 4649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905590230264696e-05,
+      "loss": 0.792,
+      "step": 4650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905541396193477e-05,
+      "loss": 0.7881,
+      "step": 4651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9054925501217125e-05,
+      "loss": 0.7744,
+      "step": 4652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90544369205005e-05,
+      "loss": 0.7715,
+      "step": 4653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905394821979137e-05,
+      "loss": 0.79,
+      "step": 4654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053459399096213e-05,
+      "loss": 0.7949,
+      "step": 4655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052970458421505e-05,
+      "loss": 0.7705,
+      "step": 4656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905248139777372e-05,
+      "loss": 0.8057,
+      "step": 4657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051992217159354e-05,
+      "loss": 0.7822,
+      "step": 4658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051502916584877e-05,
+      "loss": 0.7715,
+      "step": 4659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051013496056775e-05,
+      "loss": 0.791,
+      "step": 4660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905052395558154e-05,
+      "loss": 0.7891,
+      "step": 4661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050034295165654e-05,
+      "loss": 0.7861,
+      "step": 4662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904954451481561e-05,
+      "loss": 0.8018,
+      "step": 4663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049054614537888e-05,
+      "loss": 0.793,
+      "step": 4664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048564594338998e-05,
+      "loss": 0.7783,
+      "step": 4665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904807445422542e-05,
+      "loss": 0.7891,
+      "step": 4666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047584194203655e-05,
+      "loss": 0.7578,
+      "step": 4667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90470938142802e-05,
+      "loss": 0.7725,
+      "step": 4668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046603314461554e-05,
+      "loss": 0.7861,
+      "step": 4669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046112694754214e-05,
+      "loss": 0.7783,
+      "step": 4670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045621955164686e-05,
+      "loss": 0.7715,
+      "step": 4671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045131095699473e-05,
+      "loss": 0.7725,
+      "step": 4672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044640116365075e-05,
+      "loss": 0.792,
+      "step": 4673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044149017168006e-05,
+      "loss": 0.8018,
+      "step": 4674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043657798114766e-05,
+      "loss": 0.7832,
+      "step": 4675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043166459211873e-05,
+      "loss": 0.7539,
+      "step": 4676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042675000465832e-05,
+      "loss": 0.7812,
+      "step": 4677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042183421883163e-05,
+      "loss": 0.7725,
+      "step": 4678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041691723470373e-05,
+      "loss": 0.7656,
+      "step": 4679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904119990523398e-05,
+      "loss": 0.7627,
+      "step": 4680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040707967180506e-05,
+      "loss": 0.7871,
+      "step": 4681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040215909316468e-05,
+      "loss": 0.8125,
+      "step": 4682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039723731648384e-05,
+      "loss": 0.7646,
+      "step": 4683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903923143418278e-05,
+      "loss": 0.8018,
+      "step": 4684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903873901692618e-05,
+      "loss": 0.7842,
+      "step": 4685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9038246479885106e-05,
+      "loss": 0.7822,
+      "step": 4686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903775382306609e-05,
+      "loss": 0.792,
+      "step": 4687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9037261046475656e-05,
+      "loss": 0.7549,
+      "step": 4688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903676815012034e-05,
+      "loss": 0.7891,
+      "step": 4689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903627513400667e-05,
+      "loss": 0.7734,
+      "step": 4690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035781998141182e-05,
+      "loss": 0.7666,
+      "step": 4691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035288742530414e-05,
+      "loss": 0.7812,
+      "step": 4692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903479536718089e-05,
+      "loss": 0.7676,
+      "step": 4693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034301872099164e-05,
+      "loss": 0.7988,
+      "step": 4694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903380825729177e-05,
+      "loss": 0.7852,
+      "step": 4695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9033314522765246e-05,
+      "loss": 0.7578,
+      "step": 4696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032820668526137e-05,
+      "loss": 0.7949,
+      "step": 4697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032326694580988e-05,
+      "loss": 0.7871,
+      "step": 4698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031832600936348e-05,
+      "loss": 0.7773,
+      "step": 4699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031338387598762e-05,
+      "loss": 0.7744,
+      "step": 4700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.903084405457478e-05,
+      "loss": 0.7744,
+      "step": 4701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030349601870957e-05,
+      "loss": 0.7754,
+      "step": 4702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902985502949384e-05,
+      "loss": 0.8018,
+      "step": 4703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029360337449986e-05,
+      "loss": 0.7998,
+      "step": 4704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028865525745948e-05,
+      "loss": 0.7676,
+      "step": 4705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028370594388285e-05,
+      "loss": 0.7812,
+      "step": 4706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027875543383558e-05,
+      "loss": 0.7725,
+      "step": 4707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9027380372738328e-05,
+      "loss": 0.7949,
+      "step": 4708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026885082459153e-05,
+      "loss": 0.79,
+      "step": 4709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.90263896725526e-05,
+      "loss": 0.7744,
+      "step": 4710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9025894143025234e-05,
+      "loss": 0.7764,
+      "step": 4711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902539849388362e-05,
+      "loss": 0.7988,
+      "step": 4712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9024902725134325e-05,
+      "loss": 0.8066,
+      "step": 4713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902440683678393e-05,
+      "loss": 0.7803,
+      "step": 4714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902391082883899e-05,
+      "loss": 0.7754,
+      "step": 4715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9023414701306092e-05,
+      "loss": 0.7871,
+      "step": 4716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022918454191805e-05,
+      "loss": 0.7617,
+      "step": 4717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022422087502708e-05,
+      "loss": 0.7725,
+      "step": 4718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021925601245375e-05,
+      "loss": 0.7725,
+      "step": 4719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021428995426388e-05,
+      "loss": 0.8047,
+      "step": 4720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902093227005233e-05,
+      "loss": 0.8008,
+      "step": 4721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9020435425129777e-05,
+      "loss": 0.7881,
+      "step": 4722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901993846066532e-05,
+      "loss": 0.7881,
+      "step": 4723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019441376665545e-05,
+      "loss": 0.7793,
+      "step": 4724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018944173137037e-05,
+      "loss": 0.793,
+      "step": 4725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018446850086386e-05,
+      "loss": 0.7803,
+      "step": 4726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9017949407520182e-05,
+      "loss": 0.7852,
+      "step": 4727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9017451845445016e-05,
+      "loss": 0.7949,
+      "step": 4728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016954163867487e-05,
+      "loss": 0.7891,
+      "step": 4729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016456362794186e-05,
+      "loss": 0.7812,
+      "step": 4730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901595844223171e-05,
+      "loss": 0.8232,
+      "step": 4731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015460402186658e-05,
+      "loss": 0.7715,
+      "step": 4732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901496224266563e-05,
+      "loss": 0.7734,
+      "step": 4733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9014463963675233e-05,
+      "loss": 0.7637,
+      "step": 4734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013965565222062e-05,
+      "loss": 0.7695,
+      "step": 4735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901346704731273e-05,
+      "loss": 0.7861,
+      "step": 4736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012968409953834e-05,
+      "loss": 0.7734,
+      "step": 4737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012469653151995e-05,
+      "loss": 0.7568,
+      "step": 4738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901197077691381e-05,
+      "loss": 0.7744,
+      "step": 4739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9011471781245897e-05,
+      "loss": 0.7734,
+      "step": 4740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010972666154867e-05,
+      "loss": 0.7461,
+      "step": 4741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010473431647332e-05,
+      "loss": 0.7607,
+      "step": 4742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009974077729915e-05,
+      "loss": 0.7568,
+      "step": 4743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009474604409227e-05,
+      "loss": 0.7666,
+      "step": 4744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9008975011691893e-05,
+      "loss": 0.7969,
+      "step": 4745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900847529958453e-05,
+      "loss": 0.8184,
+      "step": 4746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007975468093758e-05,
+      "loss": 0.7852,
+      "step": 4747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007475517226207e-05,
+      "loss": 0.793,
+      "step": 4748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006975446988496e-05,
+      "loss": 0.7764,
+      "step": 4749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006475257387258e-05,
+      "loss": 0.7539,
+      "step": 4750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9005974948429116e-05,
+      "loss": 0.7705,
+      "step": 4751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9005474520120705e-05,
+      "loss": 0.8105,
+      "step": 4752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900497397246866e-05,
+      "loss": 0.8008,
+      "step": 4753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9004473305479604e-05,
+      "loss": 0.7754,
+      "step": 4754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003972519160178e-05,
+      "loss": 0.7725,
+      "step": 4755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900347161351702e-05,
+      "loss": 0.7725,
+      "step": 4756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002970588556766e-05,
+      "loss": 0.7646,
+      "step": 4757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002469444286055e-05,
+      "loss": 0.7539,
+      "step": 4758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900196818071153e-05,
+      "loss": 0.7842,
+      "step": 4759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9001466797839833e-05,
+      "loss": 0.7861,
+      "step": 4760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900096529567761e-05,
+      "loss": 0.7734,
+      "step": 4761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000463674231505e-05,
+      "loss": 0.749,
+      "step": 4762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999961933508165e-05,
+      "loss": 0.8037,
+      "step": 4763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999460073514248e-05,
+      "loss": 0.7871,
+      "step": 4764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899895809425639e-05,
+      "loss": 0.7959,
+      "step": 4765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8998455995741255e-05,
+      "loss": 0.7646,
+      "step": 4766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899795377797549e-05,
+      "loss": 0.7598,
+      "step": 4767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899745144096576e-05,
+      "loss": 0.79,
+      "step": 4768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899694898471871e-05,
+      "loss": 0.7842,
+      "step": 4769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996446409241006e-05,
+      "loss": 0.7842,
+      "step": 4770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8995943714539308e-05,
+      "loss": 0.7637,
+      "step": 4771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8995440900620274e-05,
+      "loss": 0.7852,
+      "step": 4772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899493796749057e-05,
+      "loss": 0.8281,
+      "step": 4773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994434915156865e-05,
+      "loss": 0.7773,
+      "step": 4774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8993931743625823e-05,
+      "loss": 0.7988,
+      "step": 4775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8993428452904106e-05,
+      "loss": 0.7852,
+      "step": 4776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992925042998388e-05,
+      "loss": 0.7959,
+      "step": 4777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992421513915345e-05,
+      "loss": 0.7705,
+      "step": 4778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991917865661648e-05,
+      "loss": 0.7705,
+      "step": 4779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991414098243964e-05,
+      "loss": 0.7861,
+      "step": 4780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8990910211668982e-05,
+      "loss": 0.7578,
+      "step": 4781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8990406205943362e-05,
+      "loss": 0.7861,
+      "step": 4782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989902081073804e-05,
+      "loss": 0.7764,
+      "step": 4783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989397837066974e-05,
+      "loss": 0.7627,
+      "step": 4784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898889347392956e-05,
+      "loss": 0.7939,
+      "step": 4785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8988388991668244e-05,
+      "loss": 0.7988,
+      "step": 4786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898788439028971e-05,
+      "loss": 0.7637,
+      "step": 4787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898737966980065e-05,
+      "loss": 0.7822,
+      "step": 4788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986874830207754e-05,
+      "loss": 0.7695,
+      "step": 4789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986369871517702e-05,
+      "loss": 0.7822,
+      "step": 4790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8985864793737196e-05,
+      "loss": 0.7715,
+      "step": 4791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8985359596872928e-05,
+      "loss": 0.7744,
+      "step": 4792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984854280931588e-05,
+      "loss": 0.7939,
+      "step": 4793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984348845919877e-05,
+      "loss": 0.7529,
+      "step": 4794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983843291844494e-05,
+      "loss": 0.7832,
+      "step": 4795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898333761871213e-05,
+      "loss": 0.8154,
+      "step": 4796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89828318265295e-05,
+      "loss": 0.8271,
+      "step": 4797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982325915303296e-05,
+      "loss": 0.7627,
+      "step": 4798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981819885040227e-05,
+      "loss": 0.7764,
+      "step": 4799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981313735747e-05,
+      "loss": 0.7881,
+      "step": 4800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898080746743032e-05,
+      "loss": 0.7617,
+      "step": 4801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8980301080096897e-05,
+      "loss": 0.7764,
+      "step": 4802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897979457375344e-05,
+      "loss": 0.7969,
+      "step": 4803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897928794840667e-05,
+      "loss": 0.7773,
+      "step": 4804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897878120406329e-05,
+      "loss": 0.7715,
+      "step": 4805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8978274340730024e-05,
+      "loss": 0.7656,
+      "step": 4806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977767358413585e-05,
+      "loss": 0.7812,
+      "step": 4807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897726025712069e-05,
+      "loss": 0.7832,
+      "step": 4808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897675303685806e-05,
+      "loss": 0.7988,
+      "step": 4809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897624569763242e-05,
+      "loss": 0.7891,
+      "step": 4810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897573823945049e-05,
+      "loss": 0.7598,
+      "step": 4811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8975230662318998e-05,
+      "loss": 0.7783,
+      "step": 4812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897472296624467e-05,
+      "loss": 0.7441,
+      "step": 4813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974215151234233e-05,
+      "loss": 0.7637,
+      "step": 4814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973707217294414e-05,
+      "loss": 0.7637,
+      "step": 4815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973199164431952e-05,
+      "loss": 0.8027,
+      "step": 4816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972690992653572e-05,
+      "loss": 0.7783,
+      "step": 4817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972182701966014e-05,
+      "loss": 0.7852,
+      "step": 4818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897167429237601e-05,
+      "loss": 0.7646,
+      "step": 4819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89711657638903e-05,
+      "loss": 0.792,
+      "step": 4820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970657116515623e-05,
+      "loss": 0.7451,
+      "step": 4821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897014835025872e-05,
+      "loss": 0.7656,
+      "step": 4822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896963946512633e-05,
+      "loss": 0.7715,
+      "step": 4823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969130461125204e-05,
+      "loss": 0.7754,
+      "step": 4824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896862133826208e-05,
+      "loss": 0.7783,
+      "step": 4825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968112096543706e-05,
+      "loss": 0.7568,
+      "step": 4826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967602735976838e-05,
+      "loss": 0.7617,
+      "step": 4827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896709325656822e-05,
+      "loss": 0.7783,
+      "step": 4828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8966583658324604e-05,
+      "loss": 0.7988,
+      "step": 4829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8966073941252743e-05,
+      "loss": 0.7705,
+      "step": 4830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8965564105359394e-05,
+      "loss": 0.8135,
+      "step": 4831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896505415065131e-05,
+      "loss": 0.791,
+      "step": 4832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964544077135257e-05,
+      "loss": 0.7812,
+      "step": 4833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964033884817986e-05,
+      "loss": 0.7861,
+      "step": 4834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963523573706262e-05,
+      "loss": 0.7842,
+      "step": 4835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896301314380685e-05,
+      "loss": 0.7812,
+      "step": 4836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962502595126513e-05,
+      "loss": 0.7891,
+      "step": 4837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961991927672013e-05,
+      "loss": 0.7627,
+      "step": 4838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896148114145012e-05,
+      "loss": 0.79,
+      "step": 4839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8960970236467606e-05,
+      "loss": 0.7949,
+      "step": 4840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8960459212731237e-05,
+      "loss": 0.7715,
+      "step": 4841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959948070247786e-05,
+      "loss": 0.7871,
+      "step": 4842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959436809024033e-05,
+      "loss": 0.7734,
+      "step": 4843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958925429066746e-05,
+      "loss": 0.7715,
+      "step": 4844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958413930382704e-05,
+      "loss": 0.7783,
+      "step": 4845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895790231297869e-05,
+      "loss": 0.7549,
+      "step": 4846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957390576861474e-05,
+      "loss": 0.7715,
+      "step": 4847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895687872203785e-05,
+      "loss": 0.793,
+      "step": 4848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956366748514593e-05,
+      "loss": 0.7705,
+      "step": 4849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955854656298492e-05,
+      "loss": 0.7637,
+      "step": 4850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955342445396327e-05,
+      "loss": 0.8027,
+      "step": 4851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8954830115814895e-05,
+      "loss": 0.792,
+      "step": 4852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895431766756098e-05,
+      "loss": 0.7969,
+      "step": 4853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8953805100641377e-05,
+      "loss": 0.7949,
+      "step": 4854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8953292415062873e-05,
+      "loss": 0.8096,
+      "step": 4855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8952779610832265e-05,
+      "loss": 0.7773,
+      "step": 4856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895226668795635e-05,
+      "loss": 0.7842,
+      "step": 4857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8951753646441924e-05,
+      "loss": 0.7734,
+      "step": 4858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8951240486295786e-05,
+      "loss": 0.7998,
+      "step": 4859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895072720752474e-05,
+      "loss": 0.7578,
+      "step": 4860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950213810135585e-05,
+      "loss": 0.7812,
+      "step": 4861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949700294135124e-05,
+      "loss": 0.7617,
+      "step": 4862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949186659530166e-05,
+      "loss": 0.7871,
+      "step": 4863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894867290632751e-05,
+      "loss": 0.7695,
+      "step": 4864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948159034533974e-05,
+      "loss": 0.7725,
+      "step": 4865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.894764504415636e-05,
+      "loss": 0.7832,
+      "step": 4866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947130935201487e-05,
+      "loss": 0.7686,
+      "step": 4867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946616707676166e-05,
+      "loss": 0.7666,
+      "step": 4868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946102361587205e-05,
+      "loss": 0.791,
+      "step": 4869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945587896941428e-05,
+      "loss": 0.7793,
+      "step": 4870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8945073313745648e-05,
+      "loss": 0.7773,
+      "step": 4871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944558612006688e-05,
+      "loss": 0.7725,
+      "step": 4872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894404379173137e-05,
+      "loss": 0.7646,
+      "step": 4873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943528852926513e-05,
+      "loss": 0.7783,
+      "step": 4874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943013795598942e-05,
+      "loss": 0.7979,
+      "step": 4875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8942498619755483e-05,
+      "loss": 0.7725,
+      "step": 4876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941983325402963e-05,
+      "loss": 0.7627,
+      "step": 4877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941467912548213e-05,
+      "loss": 0.7773,
+      "step": 4878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894095238119806e-05,
+      "loss": 0.8018,
+      "step": 4879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894043673135934e-05,
+      "loss": 0.7822,
+      "step": 4880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8939920963038883e-05,
+      "loss": 0.79,
+      "step": 4881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8939405076243523e-05,
+      "loss": 0.8047,
+      "step": 4882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938889070980103e-05,
+      "loss": 0.792,
+      "step": 4883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938372947255456e-05,
+      "loss": 0.7832,
+      "step": 4884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937856705076422e-05,
+      "loss": 0.7988,
+      "step": 4885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937340344449846e-05,
+      "loss": 0.7852,
+      "step": 4886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893682386538257e-05,
+      "loss": 0.7617,
+      "step": 4887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893630726788143e-05,
+      "loss": 0.7607,
+      "step": 4888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8935790551953287e-05,
+      "loss": 0.7783,
+      "step": 4889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893527371760498e-05,
+      "loss": 0.7744,
+      "step": 4890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8934756764843355e-05,
+      "loss": 0.7744,
+      "step": 4891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893423969367527e-05,
+      "loss": 0.7686,
+      "step": 4892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933722504107574e-05,
+      "loss": 0.7734,
+      "step": 4893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933205196147122e-05,
+      "loss": 0.7676,
+      "step": 4894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8932687769800768e-05,
+      "loss": 0.7881,
+      "step": 4895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8932170225075373e-05,
+      "loss": 0.7852,
+      "step": 4896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931652561977787e-05,
+      "loss": 0.791,
+      "step": 4897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931134780514883e-05,
+      "loss": 0.7979,
+      "step": 4898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8930616880693513e-05,
+      "loss": 0.793,
+      "step": 4899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8930098862520543e-05,
+      "loss": 0.792,
+      "step": 4900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892958072600284e-05,
+      "loss": 0.7793,
+      "step": 4901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8929062471147265e-05,
+      "loss": 0.7744,
+      "step": 4902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928544097960696e-05,
+      "loss": 0.7773,
+      "step": 4903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892802560644999e-05,
+      "loss": 0.7852,
+      "step": 4904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892750699662203e-05,
+      "loss": 0.7666,
+      "step": 4905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892698826848368e-05,
+      "loss": 0.7695,
+      "step": 4906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892646942204182e-05,
+      "loss": 0.7988,
+      "step": 4907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925950457303325e-05,
+      "loss": 0.7764,
+      "step": 4908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892543137427507e-05,
+      "loss": 0.7939,
+      "step": 4909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8924912172963933e-05,
+      "loss": 0.7627,
+      "step": 4910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89243928533768e-05,
+      "loss": 0.7725,
+      "step": 4911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892387341552055e-05,
+      "loss": 0.7988,
+      "step": 4912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923353859402068e-05,
+      "loss": 0.7949,
+      "step": 4913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922834185028236e-05,
+      "loss": 0.7822,
+      "step": 4914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922314392405945e-05,
+      "loss": 0.7822,
+      "step": 4915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892179448154208e-05,
+      "loss": 0.791,
+      "step": 4916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921274452443533e-05,
+      "loss": 0.7979,
+      "step": 4917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920754305117196e-05,
+      "loss": 0.7861,
+      "step": 4918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920234039569963e-05,
+      "loss": 0.7852,
+      "step": 4919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919713655808727e-05,
+      "loss": 0.7783,
+      "step": 4920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891919315384038e-05,
+      "loss": 0.7734,
+      "step": 4921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891867253367183e-05,
+      "loss": 0.7832,
+      "step": 4922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.891815179530997e-05,
+      "loss": 0.7627,
+      "step": 4923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.89176309387617e-05,
+      "loss": 0.8037,
+      "step": 4924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917109964033927e-05,
+      "loss": 0.79,
+      "step": 4925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891658887113355e-05,
+      "loss": 0.7822,
+      "step": 4926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891606766006748e-05,
+      "loss": 0.7646,
+      "step": 4927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915546330842617e-05,
+      "loss": 0.7627,
+      "step": 4928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891502488346588e-05,
+      "loss": 0.7744,
+      "step": 4929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8914503317944172e-05,
+      "loss": 0.7812,
+      "step": 4930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913981634284404e-05,
+      "loss": 0.7773,
+      "step": 4931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913459832493493e-05,
+      "loss": 0.8008,
+      "step": 4932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912937912578357e-05,
+      "loss": 0.7861,
+      "step": 4933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891241587454591e-05,
+      "loss": 0.7861,
+      "step": 4934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911893718403064e-05,
+      "loss": 0.7432,
+      "step": 4935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911371444156746e-05,
+      "loss": 0.7666,
+      "step": 4936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910849051813875e-05,
+      "loss": 0.7832,
+      "step": 4937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910326541381378e-05,
+      "loss": 0.7686,
+      "step": 4938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8909803912866172e-05,
+      "loss": 0.7607,
+      "step": 4939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890928116627519e-05,
+      "loss": 0.7754,
+      "step": 4940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908758301615354e-05,
+      "loss": 0.7783,
+      "step": 4941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908235318893595e-05,
+      "loss": 0.7705,
+      "step": 4942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907712218116847e-05,
+      "loss": 0.7666,
+      "step": 4943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890718899929204e-05,
+      "loss": 0.7627,
+      "step": 4944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906665662426105e-05,
+      "loss": 0.791,
+      "step": 4945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906142207525983e-05,
+      "loss": 0.7861,
+      "step": 4946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905618634598605e-05,
+      "loss": 0.7637,
+      "step": 4947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905094943650916e-05,
+      "loss": 0.7881,
+      "step": 4948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904571134689852e-05,
+      "loss": 0.7656,
+      "step": 4949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890404720772235e-05,
+      "loss": 0.7402,
+      "step": 4950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8903523162755367e-05,
+      "loss": 0.7852,
+      "step": 4951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902998999795834e-05,
+      "loss": 0.8086,
+      "step": 4952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902474718850706e-05,
+      "loss": 0.7705,
+      "step": 4953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901950319926925e-05,
+      "loss": 0.7822,
+      "step": 4954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890142580303145e-05,
+      "loss": 0.7939,
+      "step": 4955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890090116817122e-05,
+      "loss": 0.7988,
+      "step": 4956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890037641535319e-05,
+      "loss": 0.7959,
+      "step": 4957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899851544584324e-05,
+      "loss": 0.7773,
+      "step": 4958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899326555871567e-05,
+      "loss": 0.7793,
+      "step": 4959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898801449221883e-05,
+      "loss": 0.7666,
+      "step": 4960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898276224642224e-05,
+      "loss": 0.7607,
+      "step": 4961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889775088213956e-05,
+      "loss": 0.7666,
+      "step": 4962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8897225421720846e-05,
+      "loss": 0.7959,
+      "step": 4963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8896699843393046e-05,
+      "loss": 0.7852,
+      "step": 4964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8896174147163127e-05,
+      "loss": 0.8086,
+      "step": 4965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895648333038058e-05,
+      "loss": 0.7646,
+      "step": 4966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895122401024803e-05,
+      "loss": 0.7949,
+      "step": 4967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8894596351130335e-05,
+      "loss": 0.8018,
+      "step": 4968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889407018336162e-05,
+      "loss": 0.7666,
+      "step": 4969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8893543897725635e-05,
+      "loss": 0.7666,
+      "step": 4970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889301749422936e-05,
+      "loss": 0.7744,
+      "step": 4971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892490972879758e-05,
+      "loss": 0.7676,
+      "step": 4972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8891964333683818e-05,
+      "loss": 0.7979,
+      "step": 4973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8891437576648512e-05,
+      "loss": 0.7881,
+      "step": 4974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890910701780827e-05,
+      "loss": 0.7432,
+      "step": 4975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889038370908774e-05,
+      "loss": 0.7959,
+      "step": 4976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889856598576242e-05,
+      "loss": 0.7988,
+      "step": 4977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.888932937025331e-05,
+      "loss": 0.7842,
+      "step": 4978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8888802024125934e-05,
+      "loss": 0.7832,
+      "step": 4979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8888274560201102e-05,
+      "loss": 0.7783,
+      "step": 4980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8887746978485808e-05,
+      "loss": 0.7969,
+      "step": 4981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888721927898704e-05,
+      "loss": 0.7793,
+      "step": 4982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8886691461711793e-05,
+      "loss": 0.791,
+      "step": 4983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888616352666706e-05,
+      "loss": 0.7695,
+      "step": 4984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888563547385984e-05,
+      "loss": 0.7803,
+      "step": 4985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885107303297127e-05,
+      "loss": 0.7627,
+      "step": 4986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8884579014985925e-05,
+      "loss": 0.7656,
+      "step": 4987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888405060893323e-05,
+      "loss": 0.7686,
+      "step": 4988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8883522085146052e-05,
+      "loss": 0.7871,
+      "step": 4989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882993443631388e-05,
+      "loss": 0.7617,
+      "step": 4990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882464684396247e-05,
+      "loss": 0.7842,
+      "step": 4991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881935807447634e-05,
+      "loss": 0.7988,
+      "step": 4992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881406812792564e-05,
+      "loss": 0.7842,
+      "step": 4993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880877700438037e-05,
+      "loss": 0.7734,
+      "step": 4994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880348470391078e-05,
+      "loss": 0.7705,
+      "step": 4995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879819122658686e-05,
+      "loss": 0.7793,
+      "step": 4996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879289657247887e-05,
+      "loss": 0.7969,
+      "step": 4997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878760074165697e-05,
+      "loss": 0.8115,
+      "step": 4998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878230373419127e-05,
+      "loss": 0.7441,
+      "step": 4999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8877700555015205e-05,
+      "loss": 0.7783,
+      "step": 5000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8877170618960948e-05,
+      "loss": 0.8027,
+      "step": 5001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887664056526338e-05,
+      "loss": 0.7705,
+      "step": 5002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876110393929522e-05,
+      "loss": 0.7969,
+      "step": 5003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8875580104966404e-05,
+      "loss": 0.7598,
+      "step": 5004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8875049698381052e-05,
+      "loss": 0.7949,
+      "step": 5005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88745191741805e-05,
+      "loss": 0.7725,
+      "step": 5006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887398853237177e-05,
+      "loss": 0.7803,
+      "step": 5007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8873457772961902e-05,
+      "loss": 0.7783,
+      "step": 5008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8872926895957927e-05,
+      "loss": 0.7764,
+      "step": 5009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8872395901366877e-05,
+      "loss": 0.7764,
+      "step": 5010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871864789195792e-05,
+      "loss": 0.7725,
+      "step": 5011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871333559451706e-05,
+      "loss": 0.8057,
+      "step": 5012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870802212141673e-05,
+      "loss": 0.7393,
+      "step": 5013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870270747272716e-05,
+      "loss": 0.7803,
+      "step": 5014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886973916485189e-05,
+      "loss": 0.7637,
+      "step": 5015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886920746488624e-05,
+      "loss": 0.7461,
+      "step": 5016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868675647382805e-05,
+      "loss": 0.7822,
+      "step": 5017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8868143712348638e-05,
+      "loss": 0.7764,
+      "step": 5018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867611659790787e-05,
+      "loss": 0.752,
+      "step": 5019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867079489716303e-05,
+      "loss": 0.7822,
+      "step": 5020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886654720213224e-05,
+      "loss": 0.7822,
+      "step": 5021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866014797045647e-05,
+      "loss": 0.7715,
+      "step": 5022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886548227446359e-05,
+      "loss": 0.793,
+      "step": 5023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864949634393115e-05,
+      "loss": 0.7754,
+      "step": 5024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864416876841286e-05,
+      "loss": 0.7812,
+      "step": 5025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886388400181516e-05,
+      "loss": 0.7725,
+      "step": 5026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8863351009321803e-05,
+      "loss": 0.791,
+      "step": 5027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886281789936828e-05,
+      "loss": 0.7666,
+      "step": 5028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886228467196165e-05,
+      "loss": 0.8037,
+      "step": 5029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886175132710898e-05,
+      "loss": 0.7666,
+      "step": 5030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861217864817344e-05,
+      "loss": 0.7607,
+      "step": 5031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8860684285093807e-05,
+      "loss": 0.7949,
+      "step": 5032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886015058794544e-05,
+      "loss": 0.7549,
+      "step": 5033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8859616773379318e-05,
+      "loss": 0.7803,
+      "step": 5034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8859082841402514e-05,
+      "loss": 0.7998,
+      "step": 5035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8858548792022105e-05,
+      "loss": 0.791,
+      "step": 5036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8858014625245166e-05,
+      "loss": 0.7861,
+      "step": 5037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8857480341078774e-05,
+      "loss": 0.7754,
+      "step": 5038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856945939530018e-05,
+      "loss": 0.8193,
+      "step": 5039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856411420605975e-05,
+      "loss": 0.7988,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8855876784313723e-05,
+      "loss": 0.7617,
+      "step": 5041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885534203066036e-05,
+      "loss": 0.7607,
+      "step": 5042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8854807159652958e-05,
+      "loss": 0.75,
+      "step": 5043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8854272171298617e-05,
+      "loss": 0.7754,
+      "step": 5044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853737065604426e-05,
+      "loss": 0.7861,
+      "step": 5045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853201842577466e-05,
+      "loss": 0.7725,
+      "step": 5046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852666502224842e-05,
+      "loss": 0.7744,
+      "step": 5047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852131044553642e-05,
+      "loss": 0.7695,
+      "step": 5048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8851595469570963e-05,
+      "loss": 0.7637,
+      "step": 5049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8851059777283905e-05,
+      "loss": 0.79,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850523967699565e-05,
+      "loss": 0.7822,
+      "step": 5051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884998804082504e-05,
+      "loss": 0.7959,
+      "step": 5052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849451996667442e-05,
+      "loss": 0.8232,
+      "step": 5053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8848915835233866e-05,
+      "loss": 0.7676,
+      "step": 5054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884837955653142e-05,
+      "loss": 0.8008,
+      "step": 5055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8847843160567215e-05,
+      "loss": 0.7705,
+      "step": 5056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884730664734835e-05,
+      "loss": 0.7539,
+      "step": 5057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8846770016881948e-05,
+      "loss": 0.7871,
+      "step": 5058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884623326917511e-05,
+      "loss": 0.792,
+      "step": 5059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845696404234955e-05,
+      "loss": 0.7715,
+      "step": 5060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845159422068594e-05,
+      "loss": 0.7832,
+      "step": 5061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844622322683147e-05,
+      "loss": 0.7734,
+      "step": 5062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844085106085727e-05,
+      "loss": 0.791,
+      "step": 5063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843547772283457e-05,
+      "loss": 0.7979,
+      "step": 5064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843010321283456e-05,
+      "loss": 0.7744,
+      "step": 5065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884247275309285e-05,
+      "loss": 0.7842,
+      "step": 5066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884193506771876e-05,
+      "loss": 0.7676,
+      "step": 5067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884139726516831e-05,
+      "loss": 0.793,
+      "step": 5068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884085934544863e-05,
+      "loss": 0.7715,
+      "step": 5069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840321308566853e-05,
+      "loss": 0.7559,
+      "step": 5070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839783154530098e-05,
+      "loss": 0.7988,
+      "step": 5071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839244883345505e-05,
+      "loss": 0.7666,
+      "step": 5072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838706495020206e-05,
+      "loss": 0.7529,
+      "step": 5073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838167989561333e-05,
+      "loss": 0.7676,
+      "step": 5074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883762936697603e-05,
+      "loss": 0.7656,
+      "step": 5075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8837090627271424e-05,
+      "loss": 0.7578,
+      "step": 5076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883655177045466e-05,
+      "loss": 0.8096,
+      "step": 5077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8836012796532887e-05,
+      "loss": 0.7812,
+      "step": 5078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8835473705513233e-05,
+      "loss": 0.7646,
+      "step": 5079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8834934497402854e-05,
+      "loss": 0.7812,
+      "step": 5080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8834395172208886e-05,
+      "loss": 0.7891,
+      "step": 5081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8833855729938485e-05,
+      "loss": 0.793,
+      "step": 5082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8833316170598794e-05,
+      "loss": 0.7598,
+      "step": 5083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883277649419697e-05,
+      "loss": 0.7861,
+      "step": 5084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8832236700740156e-05,
+      "loss": 0.7734,
+      "step": 5085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8831696790235512e-05,
+      "loss": 0.7969,
+      "step": 5086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8831156762690194e-05,
+      "loss": 0.7822,
+      "step": 5087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8830616618111352e-05,
+      "loss": 0.7725,
+      "step": 5088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883007635650615e-05,
+      "loss": 0.7705,
+      "step": 5089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8829535977881745e-05,
+      "loss": 0.7734,
+      "step": 5090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88289954822453e-05,
+      "loss": 0.7852,
+      "step": 5091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8828454869603975e-05,
+      "loss": 0.7852,
+      "step": 5092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882791413996494e-05,
+      "loss": 0.7812,
+      "step": 5093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827373293335355e-05,
+      "loss": 0.7949,
+      "step": 5094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882683232972239e-05,
+      "loss": 0.7627,
+      "step": 5095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8826291249133215e-05,
+      "loss": 0.7549,
+      "step": 5096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8825750051574998e-05,
+      "loss": 0.751,
+      "step": 5097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8825208737054914e-05,
+      "loss": 0.793,
+      "step": 5098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824667305580132e-05,
+      "loss": 0.7852,
+      "step": 5099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824125757157835e-05,
+      "loss": 0.7812,
+      "step": 5100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823584091795193e-05,
+      "loss": 0.791,
+      "step": 5101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823042309499385e-05,
+      "loss": 0.7725,
+      "step": 5102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8822500410277594e-05,
+      "loss": 0.7578,
+      "step": 5103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821958394137e-05,
+      "loss": 0.7588,
+      "step": 5104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821416261084787e-05,
+      "loss": 0.7822,
+      "step": 5105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820874011128136e-05,
+      "loss": 0.8008,
+      "step": 5106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820331644274237e-05,
+      "loss": 0.7793,
+      "step": 5107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819789160530276e-05,
+      "loss": 0.7451,
+      "step": 5108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819246559903442e-05,
+      "loss": 0.7881,
+      "step": 5109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8818703842400925e-05,
+      "loss": 0.7695,
+      "step": 5110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8818161008029922e-05,
+      "loss": 0.8027,
+      "step": 5111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817618056797618e-05,
+      "loss": 0.7744,
+      "step": 5112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817074988711217e-05,
+      "loss": 0.7793,
+      "step": 5113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816531803777912e-05,
+      "loss": 0.791,
+      "step": 5114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88159885020049e-05,
+      "loss": 0.7734,
+      "step": 5115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8815445083399385e-05,
+      "loss": 0.793,
+      "step": 5116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814901547968566e-05,
+      "loss": 0.7666,
+      "step": 5117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881435789571965e-05,
+      "loss": 0.7695,
+      "step": 5118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813814126659833e-05,
+      "loss": 0.7764,
+      "step": 5119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813270240796328e-05,
+      "loss": 0.792,
+      "step": 5120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812726238136342e-05,
+      "loss": 0.7549,
+      "step": 5121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812182118687083e-05,
+      "loss": 0.7734,
+      "step": 5122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8811637882455766e-05,
+      "loss": 0.7891,
+      "step": 5123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8811093529449593e-05,
+      "loss": 0.7598,
+      "step": 5124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881054905967579e-05,
+      "loss": 0.791,
+      "step": 5125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810004473141565e-05,
+      "loss": 0.7725,
+      "step": 5126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8809459769854135e-05,
+      "loss": 0.7734,
+      "step": 5127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808914949820728e-05,
+      "loss": 0.7744,
+      "step": 5128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880837001304855e-05,
+      "loss": 0.7686,
+      "step": 5129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8807824959544837e-05,
+      "loss": 0.7881,
+      "step": 5130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88072797893168e-05,
+      "loss": 0.7793,
+      "step": 5131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880673450237167e-05,
+      "loss": 0.7939,
+      "step": 5132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806189098716672e-05,
+      "loss": 0.7832,
+      "step": 5133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805643578359036e-05,
+      "loss": 0.7617,
+      "step": 5134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805097941305985e-05,
+      "loss": 0.7715,
+      "step": 5135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8804552187564758e-05,
+      "loss": 0.7793,
+      "step": 5136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880400631714258e-05,
+      "loss": 0.7822,
+      "step": 5137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8803460330046694e-05,
+      "loss": 0.75,
+      "step": 5138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802914226284324e-05,
+      "loss": 0.7822,
+      "step": 5139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802368005862722e-05,
+      "loss": 0.7363,
+      "step": 5140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880182166878911e-05,
+      "loss": 0.7822,
+      "step": 5141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8801275215070746e-05,
+      "loss": 0.7715,
+      "step": 5142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800728644714856e-05,
+      "loss": 0.751,
+      "step": 5143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800181957728693e-05,
+      "loss": 0.7656,
+      "step": 5144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799635154119496e-05,
+      "loss": 0.793,
+      "step": 5145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799088233894516e-05,
+      "loss": 0.7803,
+      "step": 5146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8798541197060997e-05,
+      "loss": 0.7822,
+      "step": 5147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8797994043626194e-05,
+      "loss": 0.8008,
+      "step": 5148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8797446773597352e-05,
+      "loss": 0.7656,
+      "step": 5149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796899386981727e-05,
+      "loss": 0.7939,
+      "step": 5150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796351883786572e-05,
+      "loss": 0.7842,
+      "step": 5151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879580426401914e-05,
+      "loss": 0.7979,
+      "step": 5152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87952565276867e-05,
+      "loss": 0.7656,
+      "step": 5153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794708674796495e-05,
+      "loss": 0.7803,
+      "step": 5154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794160705355795e-05,
+      "loss": 0.7842,
+      "step": 5155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879361261937186e-05,
+      "loss": 0.7773,
+      "step": 5156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879306441685195e-05,
+      "loss": 0.7725,
+      "step": 5157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792516097803334e-05,
+      "loss": 0.7705,
+      "step": 5158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791967662233278e-05,
+      "loss": 0.7812,
+      "step": 5159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791419110149052e-05,
+      "loss": 0.7764,
+      "step": 5160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879087044155792e-05,
+      "loss": 0.7627,
+      "step": 5161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790321656467156e-05,
+      "loss": 0.7627,
+      "step": 5162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789772754884034e-05,
+      "loss": 0.7686,
+      "step": 5163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789223736815823e-05,
+      "loss": 0.7979,
+      "step": 5164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8788674602269808e-05,
+      "loss": 0.7646,
+      "step": 5165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878812535125326e-05,
+      "loss": 0.7686,
+      "step": 5166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787575983773458e-05,
+      "loss": 0.8096,
+      "step": 5167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787026499837685e-05,
+      "loss": 0.7754,
+      "step": 5168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878647689945322e-05,
+      "loss": 0.7852,
+      "step": 5169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8785927182627347e-05,
+      "loss": 0.7676,
+      "step": 5170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8785377349367354e-05,
+      "loss": 0.7764,
+      "step": 5171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784827399680523e-05,
+      "loss": 0.7588,
+      "step": 5172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784277333574144e-05,
+      "loss": 0.7646,
+      "step": 5173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878372715105551e-05,
+      "loss": 0.7861,
+      "step": 5174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8783176852131907e-05,
+      "loss": 0.7998,
+      "step": 5175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782626436810632e-05,
+      "loss": 0.7832,
+      "step": 5176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782075905098976e-05,
+      "loss": 0.7578,
+      "step": 5177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8781525257004238e-05,
+      "loss": 0.7832,
+      "step": 5178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780974492533715e-05,
+      "loss": 0.7832,
+      "step": 5179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780423611694703e-05,
+      "loss": 0.7637,
+      "step": 5180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87798726144945e-05,
+      "loss": 0.7959,
+      "step": 5181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8779321500940417e-05,
+      "loss": 0.7773,
+      "step": 5182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778770271039754e-05,
+      "loss": 0.751,
+      "step": 5183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8778218924799813e-05,
+      "loss": 0.7852,
+      "step": 5184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87776674622279e-05,
+      "loss": 0.7705,
+      "step": 5185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777115883331326e-05,
+      "loss": 0.79,
+      "step": 5186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8776564188117402e-05,
+      "loss": 0.7725,
+      "step": 5187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8776012376593438e-05,
+      "loss": 0.7969,
+      "step": 5188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775460448766743e-05,
+      "loss": 0.7607,
+      "step": 5189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8774908404644638e-05,
+      "loss": 0.791,
+      "step": 5190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8774356244234436e-05,
+      "loss": 0.7939,
+      "step": 5191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877380396754345e-05,
+      "loss": 0.7754,
+      "step": 5192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773251574579004e-05,
+      "loss": 0.793,
+      "step": 5193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8772699065348417e-05,
+      "loss": 0.7598,
+      "step": 5194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8772146439859017e-05,
+      "loss": 0.7471,
+      "step": 5195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771593698118117e-05,
+      "loss": 0.7871,
+      "step": 5196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771040840133047e-05,
+      "loss": 0.7754,
+      "step": 5197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770487865911137e-05,
+      "loss": 0.7686,
+      "step": 5198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769934775459708e-05,
+      "loss": 0.79,
+      "step": 5199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769381568786095e-05,
+      "loss": 0.7764,
+      "step": 5200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768828245897628e-05,
+      "loss": 0.8008,
+      "step": 5201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876827480680164e-05,
+      "loss": 0.793,
+      "step": 5202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767721251505465e-05,
+      "loss": 0.8066,
+      "step": 5203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876716758001644e-05,
+      "loss": 0.7666,
+      "step": 5204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8766613792341903e-05,
+      "loss": 0.7744,
+      "step": 5205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876605988848919e-05,
+      "loss": 0.7715,
+      "step": 5206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876550586846564e-05,
+      "loss": 0.793,
+      "step": 5207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87649517322786e-05,
+      "loss": 0.8027,
+      "step": 5208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764397479935415e-05,
+      "loss": 0.7783,
+      "step": 5209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8763843111443426e-05,
+      "loss": 0.7969,
+      "step": 5210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876328862680998e-05,
+      "loss": 0.7734,
+      "step": 5211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762734026042423e-05,
+      "loss": 0.7686,
+      "step": 5212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876217930914811e-05,
+      "loss": 0.7725,
+      "step": 5213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761624476134386e-05,
+      "loss": 0.7881,
+      "step": 5214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761069527008613e-05,
+      "loss": 0.7646,
+      "step": 5215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876051446177814e-05,
+      "loss": 0.7783,
+      "step": 5216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875995928045032e-05,
+      "loss": 0.791,
+      "step": 5217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8759403983032513e-05,
+      "loss": 0.7734,
+      "step": 5218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875884856953208e-05,
+      "loss": 0.7812,
+      "step": 5219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875829303995638e-05,
+      "loss": 0.7773,
+      "step": 5220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8757737394312776e-05,
+      "loss": 0.7686,
+      "step": 5221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875718163260863e-05,
+      "loss": 0.7646,
+      "step": 5222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8756625754851305e-05,
+      "loss": 0.7773,
+      "step": 5223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8756069761048173e-05,
+      "loss": 0.7705,
+      "step": 5224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87555136512066e-05,
+      "loss": 0.7646,
+      "step": 5225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8754957425333954e-05,
+      "loss": 0.7773,
+      "step": 5226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875440108343761e-05,
+      "loss": 0.7822,
+      "step": 5227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8753844625524938e-05,
+      "loss": 0.7646,
+      "step": 5228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875328805160331e-05,
+      "loss": 0.793,
+      "step": 5229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752731361680106e-05,
+      "loss": 0.7949,
+      "step": 5230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752174555762703e-05,
+      "loss": 0.8008,
+      "step": 5231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875161763385848e-05,
+      "loss": 0.7539,
+      "step": 5232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8751060595974816e-05,
+      "loss": 0.7812,
+      "step": 5233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8750503442119097e-05,
+      "loss": 0.7881,
+      "step": 5234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87499461722987e-05,
+      "loss": 0.7812,
+      "step": 5235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8749388786521013e-05,
+      "loss": 0.7793,
+      "step": 5236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8748831284793426e-05,
+      "loss": 0.8311,
+      "step": 5237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874827366712332e-05,
+      "loss": 0.7852,
+      "step": 5238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8747715933518092e-05,
+      "loss": 0.7598,
+      "step": 5239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874715808398513e-05,
+      "loss": 0.7803,
+      "step": 5240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746600118531827e-05,
+      "loss": 0.7822,
+      "step": 5241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746042037165578e-05,
+      "loss": 0.7725,
+      "step": 5242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874548383989378e-05,
+      "loss": 0.7607,
+      "step": 5243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744925526723826e-05,
+      "loss": 0.7842,
+      "step": 5244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744367097663122e-05,
+      "loss": 0.8037,
+      "step": 5245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874380855271906e-05,
+      "loss": 0.7734,
+      "step": 5246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874324989189905e-05,
+      "loss": 0.7939,
+      "step": 5247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874269111521049e-05,
+      "loss": 0.7568,
+      "step": 5248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8742132222660786e-05,
+      "loss": 0.7822,
+      "step": 5249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874157321425735e-05,
+      "loss": 0.7764,
+      "step": 5250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874101409000758e-05,
+      "loss": 0.7676,
+      "step": 5251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8740454849918896e-05,
+      "loss": 0.8057,
+      "step": 5252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739895493998705e-05,
+      "loss": 0.7695,
+      "step": 5253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873933602225442e-05,
+      "loss": 0.7949,
+      "step": 5254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738776434693448e-05,
+      "loss": 0.7627,
+      "step": 5255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873821673132322e-05,
+      "loss": 0.7803,
+      "step": 5256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737656912151138e-05,
+      "loss": 0.7832,
+      "step": 5257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737096977184635e-05,
+      "loss": 0.7832,
+      "step": 5258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873653692643112e-05,
+      "loss": 0.7539,
+      "step": 5259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873597675989802e-05,
+      "loss": 0.7744,
+      "step": 5260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8735416477592762e-05,
+      "loss": 0.7764,
+      "step": 5261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734856079522762e-05,
+      "loss": 0.793,
+      "step": 5262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734295565695454e-05,
+      "loss": 0.7793,
+      "step": 5263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873373493611826e-05,
+      "loss": 0.7578,
+      "step": 5264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873317419079862e-05,
+      "loss": 0.749,
+      "step": 5265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732613329743956e-05,
+      "loss": 0.7773,
+      "step": 5266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732052352961704e-05,
+      "loss": 0.8037,
+      "step": 5267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731491260459295e-05,
+      "loss": 0.7725,
+      "step": 5268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8730930052244168e-05,
+      "loss": 0.7529,
+      "step": 5269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8730368728323762e-05,
+      "loss": 0.7627,
+      "step": 5270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872980728870551e-05,
+      "loss": 0.7725,
+      "step": 5271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872924573339686e-05,
+      "loss": 0.7734,
+      "step": 5272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728684062405246e-05,
+      "loss": 0.7646,
+      "step": 5273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728122275738116e-05,
+      "loss": 0.7559,
+      "step": 5274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872756037340292e-05,
+      "loss": 0.793,
+      "step": 5275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872699835540709e-05,
+      "loss": 0.7686,
+      "step": 5276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872643622175809e-05,
+      "loss": 0.7988,
+      "step": 5277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725873972463357e-05,
+      "loss": 0.7686,
+      "step": 5278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725311607530355e-05,
+      "loss": 0.7881,
+      "step": 5279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8724749126966523e-05,
+      "loss": 0.79,
+      "step": 5280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8724186530779323e-05,
+      "loss": 0.7676,
+      "step": 5281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723623818976212e-05,
+      "loss": 0.7744,
+      "step": 5282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872306099156464e-05,
+      "loss": 0.7686,
+      "step": 5283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872249804855207e-05,
+      "loss": 0.8135,
+      "step": 5284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721934989945966e-05,
+      "loss": 0.7559,
+      "step": 5285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721371815753783e-05,
+      "loss": 0.7705,
+      "step": 5286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720808525982992e-05,
+      "loss": 0.7969,
+      "step": 5287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720245120641053e-05,
+      "loss": 0.7773,
+      "step": 5288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719681599735428e-05,
+      "loss": 0.7705,
+      "step": 5289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719117963273598e-05,
+      "loss": 0.7734,
+      "step": 5290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718554211263015e-05,
+      "loss": 0.752,
+      "step": 5291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8717990343711166e-05,
+      "loss": 0.7627,
+      "step": 5292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8717426360625513e-05,
+      "loss": 0.7607,
+      "step": 5293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871686226201354e-05,
+      "loss": 0.7588,
+      "step": 5294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716298047882715e-05,
+      "loss": 0.7627,
+      "step": 5295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715733718240515e-05,
+      "loss": 0.7764,
+      "step": 5296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715169273094422e-05,
+      "loss": 0.7822,
+      "step": 5297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8714604712451917e-05,
+      "loss": 0.7852,
+      "step": 5298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8714040036320475e-05,
+      "loss": 0.7852,
+      "step": 5299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713475244707586e-05,
+      "loss": 0.7822,
+      "step": 5300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712910337620735e-05,
+      "loss": 0.7734,
+      "step": 5301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712345315067403e-05,
+      "loss": 0.7686,
+      "step": 5302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711780177055085e-05,
+      "loss": 0.7793,
+      "step": 5303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711214923591264e-05,
+      "loss": 0.7334,
+      "step": 5304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8710649554683433e-05,
+      "loss": 0.7666,
+      "step": 5305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8710084070339085e-05,
+      "loss": 0.7715,
+      "step": 5306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8709518470565715e-05,
+      "loss": 0.7529,
+      "step": 5307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708952755370818e-05,
+      "loss": 0.7822,
+      "step": 5308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708386924761885e-05,
+      "loss": 0.7715,
+      "step": 5309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707820978746424e-05,
+      "loss": 0.7598,
+      "step": 5310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707254917331932e-05,
+      "loss": 0.7822,
+      "step": 5311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8706688740525905e-05,
+      "loss": 0.7832,
+      "step": 5312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870612244833586e-05,
+      "loss": 0.7891,
+      "step": 5313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705556040769287e-05,
+      "loss": 0.7637,
+      "step": 5314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8704989517833695e-05,
+      "loss": 0.7773,
+      "step": 5315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.87044228795366e-05,
+      "loss": 0.7754,
+      "step": 5316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.87038561258855e-05,
+      "loss": 0.7734,
+      "step": 5317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8703289256887916e-05,
+      "loss": 0.7725,
+      "step": 5318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702722272551357e-05,
+      "loss": 0.7539,
+      "step": 5319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702155172883333e-05,
+      "loss": 0.7783,
+      "step": 5320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8701587957891364e-05,
+      "loss": 0.7891,
+      "step": 5321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870102062758296e-05,
+      "loss": 0.7881,
+      "step": 5322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870045318196565e-05,
+      "loss": 0.7539,
+      "step": 5323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869988562104695e-05,
+      "loss": 0.7871,
+      "step": 5324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869931794483438e-05,
+      "loss": 0.7627,
+      "step": 5325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8698750153335462e-05,
+      "loss": 0.7754,
+      "step": 5326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869818224655772e-05,
+      "loss": 0.7764,
+      "step": 5327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869761422450868e-05,
+      "loss": 0.8047,
+      "step": 5328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697046087195875e-05,
+      "loss": 0.7754,
+      "step": 5329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869647783462683e-05,
+      "loss": 0.7949,
+      "step": 5330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695909466809076e-05,
+      "loss": 0.7861,
+      "step": 5331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695340983750146e-05,
+      "loss": 0.793,
+      "step": 5332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694772385457575e-05,
+      "loss": 0.7832,
+      "step": 5333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694203671938894e-05,
+      "loss": 0.8105,
+      "step": 5334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869363484320164e-05,
+      "loss": 0.7998,
+      "step": 5335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869306589925336e-05,
+      "loss": 0.7803,
+      "step": 5336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8692496840101585e-05,
+      "loss": 0.7686,
+      "step": 5337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8691927665753856e-05,
+      "loss": 0.7656,
+      "step": 5338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869135837621772e-05,
+      "loss": 0.7754,
+      "step": 5339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690788971500723e-05,
+      "loss": 0.7607,
+      "step": 5340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690219451610403e-05,
+      "loss": 0.7607,
+      "step": 5341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868964981655432e-05,
+      "loss": 0.7861,
+      "step": 5342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868908006634001e-05,
+      "loss": 0.7793,
+      "step": 5343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8688510200975026e-05,
+      "loss": 0.7812,
+      "step": 5344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868794022046693e-05,
+      "loss": 0.7725,
+      "step": 5345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8687370124823266e-05,
+      "loss": 0.7744,
+      "step": 5346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8686799914051593e-05,
+      "loss": 0.7676,
+      "step": 5347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8686229588159465e-05,
+      "loss": 0.7891,
+      "step": 5348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685659147154442e-05,
+      "loss": 0.7715,
+      "step": 5349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685088591044082e-05,
+      "loss": 0.7617,
+      "step": 5350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684517919835948e-05,
+      "loss": 0.7715,
+      "step": 5351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8683947133537605e-05,
+      "loss": 0.7666,
+      "step": 5352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868337623215661e-05,
+      "loss": 0.7773,
+      "step": 5353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682805215700535e-05,
+      "loss": 0.7764,
+      "step": 5354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682234084176947e-05,
+      "loss": 0.7891,
+      "step": 5355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868166283759341e-05,
+      "loss": 0.7891,
+      "step": 5356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.86810914759575e-05,
+      "loss": 0.7676,
+      "step": 5357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868051999927679e-05,
+      "loss": 0.7549,
+      "step": 5358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8679948407558846e-05,
+      "loss": 0.7695,
+      "step": 5359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867937670081125e-05,
+      "loss": 0.7832,
+      "step": 5360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678804879041573e-05,
+      "loss": 0.7734,
+      "step": 5361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678232942257395e-05,
+      "loss": 0.8096,
+      "step": 5362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677660890466296e-05,
+      "loss": 0.792,
+      "step": 5363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677088723675858e-05,
+      "loss": 0.7686,
+      "step": 5364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8676516441893663e-05,
+      "loss": 0.7578,
+      "step": 5365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675944045127292e-05,
+      "loss": 0.8154,
+      "step": 5366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675371533384336e-05,
+      "loss": 0.7754,
+      "step": 5367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867479890667238e-05,
+      "loss": 0.7559,
+      "step": 5368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867422616499901e-05,
+      "loss": 0.7803,
+      "step": 5369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8673653308371816e-05,
+      "loss": 0.8174,
+      "step": 5370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8673080336798394e-05,
+      "loss": 0.7725,
+      "step": 5371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8672507250286334e-05,
+      "loss": 0.7656,
+      "step": 5372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8671934048843233e-05,
+      "loss": 0.7607,
+      "step": 5373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8671360732476684e-05,
+      "loss": 0.7861,
+      "step": 5374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.867078730119429e-05,
+      "loss": 0.793,
+      "step": 5375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670213755003646e-05,
+      "loss": 0.7959,
+      "step": 5376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669640093912353e-05,
+      "loss": 0.7783,
+      "step": 5377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669066317928014e-05,
+      "loss": 0.7705,
+      "step": 5378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8668492427058234e-05,
+      "loss": 0.7861,
+      "step": 5379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667918421310617e-05,
+      "loss": 0.7891,
+      "step": 5380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866734430069277e-05,
+      "loss": 0.792,
+      "step": 5381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666770065212304e-05,
+      "loss": 0.7803,
+      "step": 5382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8666195714876828e-05,
+      "loss": 0.7627,
+      "step": 5383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866562124969395e-05,
+      "loss": 0.7773,
+      "step": 5384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665046669671283e-05,
+      "loss": 0.7598,
+      "step": 5385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664471974816444e-05,
+      "loss": 0.7646,
+      "step": 5386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8663897165137053e-05,
+      "loss": 0.8037,
+      "step": 5387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8663322240640724e-05,
+      "loss": 0.7842,
+      "step": 5388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866274720133507e-05,
+      "loss": 0.7822,
+      "step": 5389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8662172047227724e-05,
+      "loss": 0.7471,
+      "step": 5390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86615967783263e-05,
+      "loss": 0.7734,
+      "step": 5391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8661021394638417e-05,
+      "loss": 0.7803,
+      "step": 5392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8660445896171712e-05,
+      "loss": 0.7773,
+      "step": 5393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659870282933804e-05,
+      "loss": 0.7715,
+      "step": 5394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659294554932324e-05,
+      "loss": 0.7539,
+      "step": 5395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86587187121749e-05,
+      "loss": 0.8047,
+      "step": 5396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8658142754669163e-05,
+      "loss": 0.7969,
+      "step": 5397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865756668242275e-05,
+      "loss": 0.7588,
+      "step": 5398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865699049544329e-05,
+      "loss": 0.7646,
+      "step": 5399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656414193738422e-05,
+      "loss": 0.7803,
+      "step": 5400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8655837777315783e-05,
+      "loss": 0.7686,
+      "step": 5401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8655261246183012e-05,
+      "loss": 0.7764,
+      "step": 5402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654684600347746e-05,
+      "loss": 0.7676,
+      "step": 5403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865410783981763e-05,
+      "loss": 0.792,
+      "step": 5404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865353096460031e-05,
+      "loss": 0.7549,
+      "step": 5405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865295397470343e-05,
+      "loss": 0.7744,
+      "step": 5406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652376870134628e-05,
+      "loss": 0.7871,
+      "step": 5407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8651799650901557e-05,
+      "loss": 0.7725,
+      "step": 5408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8651222317011874e-05,
+      "loss": 0.7598,
+      "step": 5409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650644868473222e-05,
+      "loss": 0.7988,
+      "step": 5410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650067305293252e-05,
+      "loss": 0.7354,
+      "step": 5411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864948962747963e-05,
+      "loss": 0.7578,
+      "step": 5412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648911835039998e-05,
+      "loss": 0.7822,
+      "step": 5413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648333927982015e-05,
+      "loss": 0.8105,
+      "step": 5414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864775590631335e-05,
+      "loss": 0.7539,
+      "step": 5415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8647177770041647e-05,
+      "loss": 0.7783,
+      "step": 5416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646599519174584e-05,
+      "loss": 0.7871,
+      "step": 5417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646021153719815e-05,
+      "loss": 0.7725,
+      "step": 5418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864544267368501e-05,
+      "loss": 0.7666,
+      "step": 5419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644864079077827e-05,
+      "loss": 0.75,
+      "step": 5420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644285369905938e-05,
+      "loss": 0.7959,
+      "step": 5421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8643706546177018e-05,
+      "loss": 0.7734,
+      "step": 5422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864312760789873e-05,
+      "loss": 0.7588,
+      "step": 5423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8642548555078747e-05,
+      "loss": 0.7783,
+      "step": 5424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8641969387724745e-05,
+      "loss": 0.7812,
+      "step": 5425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86413901058444e-05,
+      "loss": 0.7959,
+      "step": 5426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640810709445384e-05,
+      "loss": 0.7861,
+      "step": 5427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640231198535385e-05,
+      "loss": 0.7695,
+      "step": 5428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639651573122075e-05,
+      "loss": 0.7754,
+      "step": 5429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639071833213137e-05,
+      "loss": 0.7656,
+      "step": 5430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8638491978816254e-05,
+      "loss": 0.7773,
+      "step": 5431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863791200993911e-05,
+      "loss": 0.7617,
+      "step": 5432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8637331926589395e-05,
+      "loss": 0.7783,
+      "step": 5433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863675172877479e-05,
+      "loss": 0.7783,
+      "step": 5434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636171416502987e-05,
+      "loss": 0.7734,
+      "step": 5435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863559098978168e-05,
+      "loss": 0.79,
+      "step": 5436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635010448618557e-05,
+      "loss": 0.7832,
+      "step": 5437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863442979302131e-05,
+      "loss": 0.7646,
+      "step": 5438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8633849022997637e-05,
+      "loss": 0.75,
+      "step": 5439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8633268138555233e-05,
+      "loss": 0.7783,
+      "step": 5440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86326871397018e-05,
+      "loss": 0.7764,
+      "step": 5441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632106026445033e-05,
+      "loss": 0.7773,
+      "step": 5442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8631524798792634e-05,
+      "loss": 0.7725,
+      "step": 5443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630943456752307e-05,
+      "loss": 0.7744,
+      "step": 5444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630362000331755e-05,
+      "loss": 0.7744,
+      "step": 5445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8629780429538682e-05,
+      "loss": 0.7861,
+      "step": 5446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86291987443808e-05,
+      "loss": 0.7861,
+      "step": 5447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8628616944865815e-05,
+      "loss": 0.7568,
+      "step": 5448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8628035031001434e-05,
+      "loss": 0.7852,
+      "step": 5449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627453002795374e-05,
+      "loss": 0.7871,
+      "step": 5450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8626870860255346e-05,
+      "loss": 0.7715,
+      "step": 5451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8626288603389065e-05,
+      "loss": 0.7881,
+      "step": 5452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625706232204247e-05,
+      "loss": 0.7637,
+      "step": 5453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625123746708612e-05,
+      "loss": 0.7939,
+      "step": 5454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624541146909872e-05,
+      "loss": 0.7715,
+      "step": 5455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623958432815755e-05,
+      "loss": 0.752,
+      "step": 5456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623375604433987e-05,
+      "loss": 0.7754,
+      "step": 5457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622792661772278e-05,
+      "loss": 0.7715,
+      "step": 5458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622209604838368e-05,
+      "loss": 0.7754,
+      "step": 5459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8621626433639973e-05,
+      "loss": 0.7861,
+      "step": 5460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8621043148184827e-05,
+      "loss": 0.7686,
+      "step": 5461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862045974848066e-05,
+      "loss": 0.7666,
+      "step": 5462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8619876234535202e-05,
+      "loss": 0.7617,
+      "step": 5463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8619292606356184e-05,
+      "loss": 0.7832,
+      "step": 5464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618708863951343e-05,
+      "loss": 0.791,
+      "step": 5465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618125007328416e-05,
+      "loss": 0.7734,
+      "step": 5466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8617541036495138e-05,
+      "loss": 0.7686,
+      "step": 5467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861695695145925e-05,
+      "loss": 0.8008,
+      "step": 5468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861637275222849e-05,
+      "loss": 0.7744,
+      "step": 5469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86157884388106e-05,
+      "loss": 0.7539,
+      "step": 5470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8615204011213326e-05,
+      "loss": 0.75,
+      "step": 5471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8614619469444414e-05,
+      "loss": 0.7969,
+      "step": 5472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8614034813511606e-05,
+      "loss": 0.7676,
+      "step": 5473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613450043422653e-05,
+      "loss": 0.7891,
+      "step": 5474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8612865159185304e-05,
+      "loss": 0.7852,
+      "step": 5475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861228016080731e-05,
+      "loss": 0.8193,
+      "step": 5476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8611695048296426e-05,
+      "loss": 0.7812,
+      "step": 5477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8611109821660404e-05,
+      "loss": 0.7754,
+      "step": 5478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610524480906994e-05,
+      "loss": 0.7734,
+      "step": 5479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609939026043965e-05,
+      "loss": 0.7773,
+      "step": 5480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609353457079063e-05,
+      "loss": 0.7803,
+      "step": 5481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608767774020063e-05,
+      "loss": 0.7949,
+      "step": 5482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860818197687471e-05,
+      "loss": 0.7764,
+      "step": 5483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8607596065650783e-05,
+      "loss": 0.7783,
+      "step": 5484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860701004035603e-05,
+      "loss": 0.7695,
+      "step": 5485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860642390099823e-05,
+      "loss": 0.792,
+      "step": 5486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860583764758515e-05,
+      "loss": 0.8135,
+      "step": 5487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605251280124555e-05,
+      "loss": 0.7764,
+      "step": 5488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604664798624218e-05,
+      "loss": 0.791,
+      "step": 5489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604078203091907e-05,
+      "loss": 0.7666,
+      "step": 5490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603491493535403e-05,
+      "loss": 0.7549,
+      "step": 5491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860290466996247e-05,
+      "loss": 0.7637,
+      "step": 5492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8602317732380898e-05,
+      "loss": 0.7686,
+      "step": 5493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601730680798456e-05,
+      "loss": 0.7939,
+      "step": 5494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860114351522293e-05,
+      "loss": 0.7617,
+      "step": 5495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86005562356621e-05,
+      "loss": 0.7578,
+      "step": 5496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8599968842123745e-05,
+      "loss": 0.7773,
+      "step": 5497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8599381334615648e-05,
+      "loss": 0.7744,
+      "step": 5498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8598793713145603e-05,
+      "loss": 0.7842,
+      "step": 5499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859820597772139e-05,
+      "loss": 0.7861,
+      "step": 5500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8597618128350806e-05,
+      "loss": 0.7617,
+      "step": 5501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859703016504163e-05,
+      "loss": 0.7783,
+      "step": 5502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8596442087801662e-05,
+      "loss": 0.7715,
+      "step": 5503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8595853896638694e-05,
+      "loss": 0.8066,
+      "step": 5504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859526559156052e-05,
+      "loss": 0.7812,
+      "step": 5505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8594677172574937e-05,
+      "loss": 0.7705,
+      "step": 5506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8594088639689744e-05,
+      "loss": 0.7617,
+      "step": 5507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593499992912737e-05,
+      "loss": 0.791,
+      "step": 5508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8592911232251717e-05,
+      "loss": 0.7598,
+      "step": 5509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859232235771449e-05,
+      "loss": 0.7705,
+      "step": 5510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8591733369308863e-05,
+      "loss": 0.7773,
+      "step": 5511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859114426704263e-05,
+      "loss": 0.8066,
+      "step": 5512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8590555050923612e-05,
+      "loss": 0.7578,
+      "step": 5513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589965720959605e-05,
+      "loss": 0.7559,
+      "step": 5514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589376277158426e-05,
+      "loss": 0.7451,
+      "step": 5515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588786719527884e-05,
+      "loss": 0.7646,
+      "step": 5516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588197048075796e-05,
+      "loss": 0.7891,
+      "step": 5517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858760726280997e-05,
+      "loss": 0.7686,
+      "step": 5518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8587017363738228e-05,
+      "loss": 0.7451,
+      "step": 5519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586427350868385e-05,
+      "loss": 0.7344,
+      "step": 5520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8585837224208258e-05,
+      "loss": 0.7686,
+      "step": 5521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858524698376567e-05,
+      "loss": 0.7773,
+      "step": 5522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584656629548444e-05,
+      "loss": 0.7744,
+      "step": 5523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584066161564404e-05,
+      "loss": 0.7773,
+      "step": 5524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858347557982137e-05,
+      "loss": 0.7598,
+      "step": 5525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8582884884327172e-05,
+      "loss": 0.793,
+      "step": 5526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858229407508964e-05,
+      "loss": 0.7598,
+      "step": 5527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85817031521166e-05,
+      "loss": 0.7656,
+      "step": 5528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581112115415886e-05,
+      "loss": 0.793,
+      "step": 5529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858052096499533e-05,
+      "loss": 0.7627,
+      "step": 5530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579929700862763e-05,
+      "loss": 0.7969,
+      "step": 5531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579338323026027e-05,
+      "loss": 0.7695,
+      "step": 5532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578746831492954e-05,
+      "loss": 0.7822,
+      "step": 5533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578155226271382e-05,
+      "loss": 0.7881,
+      "step": 5534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577563507369153e-05,
+      "loss": 0.7803,
+      "step": 5535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576971674794113e-05,
+      "loss": 0.7734,
+      "step": 5536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85763797285541e-05,
+      "loss": 0.8047,
+      "step": 5537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8575787668656953e-05,
+      "loss": 0.7773,
+      "step": 5538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8575195495110532e-05,
+      "loss": 0.7646,
+      "step": 5539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574603207922673e-05,
+      "loss": 0.7666,
+      "step": 5540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574010807101232e-05,
+      "loss": 0.79,
+      "step": 5541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8573418292654057e-05,
+      "loss": 0.7744,
+      "step": 5542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8572825664589004e-05,
+      "loss": 0.7617,
+      "step": 5543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857223292291392e-05,
+      "loss": 0.748,
+      "step": 5544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857164006763666e-05,
+      "loss": 0.7676,
+      "step": 5545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857104709876509e-05,
+      "loss": 0.7891,
+      "step": 5546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857045401630706e-05,
+      "loss": 0.752,
+      "step": 5547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569860820270434e-05,
+      "loss": 0.7598,
+      "step": 5548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856926751066307e-05,
+      "loss": 0.7969,
+      "step": 5549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856867408749283e-05,
+      "loss": 0.8037,
+      "step": 5550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8568080550767583e-05,
+      "loss": 0.7539,
+      "step": 5551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8567486900495197e-05,
+      "loss": 0.7412,
+      "step": 5552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566893136683526e-05,
+      "loss": 0.7881,
+      "step": 5553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856629925934045e-05,
+      "loss": 0.7852,
+      "step": 5554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8565705268473836e-05,
+      "loss": 0.7998,
+      "step": 5555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856511116409156e-05,
+      "loss": 0.7803,
+      "step": 5556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856451694620149e-05,
+      "loss": 0.7803,
+      "step": 5557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85639226148115e-05,
+      "loss": 0.7832,
+      "step": 5558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856332816992947e-05,
+      "loss": 0.7812,
+      "step": 5559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562733611563275e-05,
+      "loss": 0.7793,
+      "step": 5560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85621389397208e-05,
+      "loss": 0.7715,
+      "step": 5561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8561544154409917e-05,
+      "loss": 0.7861,
+      "step": 5562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560949255638512e-05,
+      "loss": 0.7842,
+      "step": 5563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560354243414474e-05,
+      "loss": 0.791,
+      "step": 5564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855975911774568e-05,
+      "loss": 0.7666,
+      "step": 5565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559163878640025e-05,
+      "loss": 0.793,
+      "step": 5566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855856852610539e-05,
+      "loss": 0.7568,
+      "step": 5567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557973060149667e-05,
+      "loss": 0.8164,
+      "step": 5568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557377480780746e-05,
+      "loss": 0.7803,
+      "step": 5569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556781788006528e-05,
+      "loss": 0.8066,
+      "step": 5570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85561859818349e-05,
+      "loss": 0.7568,
+      "step": 5571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555590062273756e-05,
+      "loss": 0.7695,
+      "step": 5572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554994029330997e-05,
+      "loss": 0.7588,
+      "step": 5573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554397883014522e-05,
+      "loss": 0.7646,
+      "step": 5574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855380162333223e-05,
+      "loss": 0.7725,
+      "step": 5575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8553205250292023e-05,
+      "loss": 0.7549,
+      "step": 5576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8552608763901808e-05,
+      "loss": 0.8066,
+      "step": 5577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8552012164169487e-05,
+      "loss": 0.7705,
+      "step": 5578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551415451102962e-05,
+      "loss": 0.7598,
+      "step": 5579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8550818624710144e-05,
+      "loss": 0.7939,
+      "step": 5580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8550221684998944e-05,
+      "loss": 0.7412,
+      "step": 5581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549624631977277e-05,
+      "loss": 0.7559,
+      "step": 5582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549027465653048e-05,
+      "loss": 0.7559,
+      "step": 5583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854843018603417e-05,
+      "loss": 0.7588,
+      "step": 5584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547832793128563e-05,
+      "loss": 0.7607,
+      "step": 5585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547235286944144e-05,
+      "loss": 0.7832,
+      "step": 5586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854663766748883e-05,
+      "loss": 0.7578,
+      "step": 5587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8546039934770543e-05,
+      "loss": 0.7744,
+      "step": 5588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8545442088797197e-05,
+      "loss": 0.7832,
+      "step": 5589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854484412957672e-05,
+      "loss": 0.7754,
+      "step": 5590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8544246057117038e-05,
+      "loss": 0.7646,
+      "step": 5591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543647871426075e-05,
+      "loss": 0.7578,
+      "step": 5592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543049572511757e-05,
+      "loss": 0.7666,
+      "step": 5593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8542451160382016e-05,
+      "loss": 0.7695,
+      "step": 5594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854185263504478e-05,
+      "loss": 0.7695,
+      "step": 5595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854125399650798e-05,
+      "loss": 0.8037,
+      "step": 5596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8540655244779556e-05,
+      "loss": 0.752,
+      "step": 5597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8540056379867435e-05,
+      "loss": 0.7812,
+      "step": 5598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8539457401779555e-05,
+      "loss": 0.7559,
+      "step": 5599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8538858310523855e-05,
+      "loss": 0.7461,
+      "step": 5600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8538259106108274e-05,
+      "loss": 0.8047,
+      "step": 5601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537659788540753e-05,
+      "loss": 0.7832,
+      "step": 5602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537060357829238e-05,
+      "loss": 0.7363,
+      "step": 5603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8536460813981666e-05,
+      "loss": 0.7812,
+      "step": 5604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535861157005986e-05,
+      "loss": 0.7471,
+      "step": 5605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535261386910148e-05,
+      "loss": 0.7607,
+      "step": 5606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534661503702095e-05,
+      "loss": 0.7686,
+      "step": 5607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534061507389777e-05,
+      "loss": 0.7803,
+      "step": 5608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853346139798115e-05,
+      "loss": 0.7939,
+      "step": 5609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8532861175484163e-05,
+      "loss": 0.7783,
+      "step": 5610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853226083990677e-05,
+      "loss": 0.7617,
+      "step": 5611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853166039125693e-05,
+      "loss": 0.7705,
+      "step": 5612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85310598295426e-05,
+      "loss": 0.8018,
+      "step": 5613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530459154771733e-05,
+      "loss": 0.7783,
+      "step": 5614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8529858366952296e-05,
+      "loss": 0.7783,
+      "step": 5615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852925746609225e-05,
+      "loss": 0.7695,
+      "step": 5616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8528656452199558e-05,
+      "loss": 0.7539,
+      "step": 5617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852805532528218e-05,
+      "loss": 0.791,
+      "step": 5618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852745408534809e-05,
+      "loss": 0.7656,
+      "step": 5619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852685273240525e-05,
+      "loss": 0.7725,
+      "step": 5620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526251266461632e-05,
+      "loss": 0.7832,
+      "step": 5621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8525649687525207e-05,
+      "loss": 0.7646,
+      "step": 5622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8525047995603947e-05,
+      "loss": 0.7529,
+      "step": 5623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8524446190705823e-05,
+      "loss": 0.791,
+      "step": 5624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852384427283881e-05,
+      "loss": 0.7764,
+      "step": 5625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8523242242010893e-05,
+      "loss": 0.8096,
+      "step": 5626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522640098230042e-05,
+      "loss": 0.7988,
+      "step": 5627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522037841504242e-05,
+      "loss": 0.7568,
+      "step": 5628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521435471841473e-05,
+      "loss": 0.7764,
+      "step": 5629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8520832989249713e-05,
+      "loss": 0.7783,
+      "step": 5630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852023039373695e-05,
+      "loss": 0.752,
+      "step": 5631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8519627685311175e-05,
+      "loss": 0.75,
+      "step": 5632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8519024863980365e-05,
+      "loss": 0.7637,
+      "step": 5633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8518421929752514e-05,
+      "loss": 0.7852,
+      "step": 5634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8517818882635616e-05,
+      "loss": 0.7871,
+      "step": 5635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851721572263766e-05,
+      "loss": 0.7949,
+      "step": 5636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8516612449766632e-05,
+      "loss": 0.7686,
+      "step": 5637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8516009064030536e-05,
+      "loss": 0.7764,
+      "step": 5638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851540556543737e-05,
+      "loss": 0.7725,
+      "step": 5639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851480195399512e-05,
+      "loss": 0.7705,
+      "step": 5640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8514198229711796e-05,
+      "loss": 0.7734,
+      "step": 5641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513594392595396e-05,
+      "loss": 0.7734,
+      "step": 5642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851299044265392e-05,
+      "loss": 0.7578,
+      "step": 5643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512386379895372e-05,
+      "loss": 0.7539,
+      "step": 5644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851178220432776e-05,
+      "loss": 0.749,
+      "step": 5645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8511177915959088e-05,
+      "loss": 0.8154,
+      "step": 5646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851057351479737e-05,
+      "loss": 0.7871,
+      "step": 5647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509969000850607e-05,
+      "loss": 0.7549,
+      "step": 5648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509364374126813e-05,
+      "loss": 0.7803,
+      "step": 5649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508759634634007e-05,
+      "loss": 0.7539,
+      "step": 5650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508154782380194e-05,
+      "loss": 0.7744,
+      "step": 5651
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8507549817373397e-05,
+      "loss": 0.7949,
+      "step": 5652
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850694473962163e-05,
+      "loss": 0.793,
+      "step": 5653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850633954913291e-05,
+      "loss": 0.7676,
+      "step": 5654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8505734245915258e-05,
+      "loss": 0.793,
+      "step": 5655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.85051288299767e-05,
+      "loss": 0.7715,
+      "step": 5656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8504523301325254e-05,
+      "loss": 0.79,
+      "step": 5657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503917659968947e-05,
+      "loss": 0.7529,
+      "step": 5658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503311905915808e-05,
+      "loss": 0.7764,
+      "step": 5659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8502706039173856e-05,
+      "loss": 0.7793,
+      "step": 5660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850210005975113e-05,
+      "loss": 0.7686,
+      "step": 5661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501493967655656e-05,
+      "loss": 0.7734,
+      "step": 5662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500887762895466e-05,
+      "loss": 0.7588,
+      "step": 5663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500281445478592e-05,
+      "loss": 0.7686,
+      "step": 5664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499675015413072e-05,
+      "loss": 0.7734,
+      "step": 5665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8499068472706943e-05,
+      "loss": 0.7764,
+      "step": 5666
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849846181736824e-05,
+      "loss": 0.7539,
+      "step": 5667
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497855049405007e-05,
+      "loss": 0.749,
+      "step": 5668
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497248168825282e-05,
+      "loss": 0.7666,
+      "step": 5669
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8496641175637107e-05,
+      "loss": 0.7803,
+      "step": 5670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849603406984853e-05,
+      "loss": 0.7529,
+      "step": 5671
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8495426851467592e-05,
+      "loss": 0.7783,
+      "step": 5672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849481952050234e-05,
+      "loss": 0.7754,
+      "step": 5673
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494212076960827e-05,
+      "loss": 0.7686,
+      "step": 5674
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84936045208511e-05,
+      "loss": 0.7812,
+      "step": 5675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849299685218121e-05,
+      "loss": 0.7686,
+      "step": 5676
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492389070959212e-05,
+      "loss": 0.7725,
+      "step": 5677
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849178117719316e-05,
+      "loss": 0.7754,
+      "step": 5678
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849117317089111e-05,
+      "loss": 0.7949,
+      "step": 5679
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490565052061118e-05,
+      "loss": 0.7549,
+      "step": 5680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8489956820711246e-05,
+      "loss": 0.7725,
+      "step": 5681
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848934847684955e-05,
+      "loss": 0.7607,
+      "step": 5682
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8488740020484097e-05,
+      "loss": 0.7461,
+      "step": 5683
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848813145162295e-05,
+      "loss": 0.7627,
+      "step": 5684
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848752277027417e-05,
+      "loss": 0.7539,
+      "step": 5685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8486913976445825e-05,
+      "loss": 0.7686,
+      "step": 5686
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848630507014598e-05,
+      "loss": 0.7578,
+      "step": 5687
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848569605138271e-05,
+      "loss": 0.7637,
+      "step": 5688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8485086920164087e-05,
+      "loss": 0.7695,
+      "step": 5689
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8484477676498177e-05,
+      "loss": 0.7744,
+      "step": 5690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8483868320393062e-05,
+      "loss": 0.7783,
+      "step": 5691
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8483258851856808e-05,
+      "loss": 0.7754,
+      "step": 5692
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84826492708975e-05,
+      "loss": 0.7764,
+      "step": 5693
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8482039577523208e-05,
+      "loss": 0.7832,
+      "step": 5694
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848142977174202e-05,
+      "loss": 0.7793,
+      "step": 5695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8480819853562012e-05,
+      "loss": 0.7676,
+      "step": 5696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848020982299127e-05,
+      "loss": 0.7568,
+      "step": 5697
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847959968003788e-05,
+      "loss": 0.7773,
+      "step": 5698
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8478989424709923e-05,
+      "loss": 0.7881,
+      "step": 5699
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8478379057015487e-05,
+      "loss": 0.7754,
+      "step": 5700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8477768576962662e-05,
+      "loss": 0.7578,
+      "step": 5701
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847715798455954e-05,
+      "loss": 0.7588,
+      "step": 5702
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847654727981421e-05,
+      "loss": 0.7695,
+      "step": 5703
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847593646273477e-05,
+      "loss": 0.7744,
+      "step": 5704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847532553332931e-05,
+      "loss": 0.7959,
+      "step": 5705
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8474714491605925e-05,
+      "loss": 0.7734,
+      "step": 5706
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8474103337572714e-05,
+      "loss": 0.7646,
+      "step": 5707
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847349207123778e-05,
+      "loss": 0.792,
+      "step": 5708
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8472880692609225e-05,
+      "loss": 0.75,
+      "step": 5709
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847226920169514e-05,
+      "loss": 0.7793,
+      "step": 5710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8471657598503642e-05,
+      "loss": 0.752,
+      "step": 5711
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8471045883042825e-05,
+      "loss": 0.7451,
+      "step": 5712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8470434055320804e-05,
+      "loss": 0.7734,
+      "step": 5713
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469822115345685e-05,
+      "loss": 0.7715,
+      "step": 5714
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469210063125574e-05,
+      "loss": 0.7812,
+      "step": 5715
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468597898668587e-05,
+      "loss": 0.7715,
+      "step": 5716
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8467985621982835e-05,
+      "loss": 0.7676,
+      "step": 5717
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8467373233076428e-05,
+      "loss": 0.7705,
+      "step": 5718
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846676073195749e-05,
+      "loss": 0.7559,
+      "step": 5719
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8466148118634127e-05,
+      "loss": 0.7578,
+      "step": 5720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846553539311447e-05,
+      "loss": 0.7695,
+      "step": 5721
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846492255540663e-05,
+      "loss": 0.7715,
+      "step": 5722
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846430960551873e-05,
+      "loss": 0.7812,
+      "step": 5723
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8463696543458897e-05,
+      "loss": 0.7549,
+      "step": 5724
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8463083369235255e-05,
+      "loss": 0.7861,
+      "step": 5725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8462470082855922e-05,
+      "loss": 0.7676,
+      "step": 5726
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461856684329034e-05,
+      "loss": 0.79,
+      "step": 5727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461243173662716e-05,
+      "loss": 0.7705,
+      "step": 5728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8460629550865102e-05,
+      "loss": 0.7715,
+      "step": 5729
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846001581594432e-05,
+      "loss": 0.79,
+      "step": 5730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459401968908507e-05,
+      "loss": 0.7949,
+      "step": 5731
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8458788009765796e-05,
+      "loss": 0.75,
+      "step": 5732
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8458173938524322e-05,
+      "loss": 0.7705,
+      "step": 5733
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8457559755192226e-05,
+      "loss": 0.7666,
+      "step": 5734
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8456945459777645e-05,
+      "loss": 0.7607,
+      "step": 5735
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845633105228872e-05,
+      "loss": 0.7686,
+      "step": 5736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845571653273359e-05,
+      "loss": 0.7852,
+      "step": 5737
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845510190112041e-05,
+      "loss": 0.7725,
+      "step": 5738
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454487157457315e-05,
+      "loss": 0.748,
+      "step": 5739
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8453872301752455e-05,
+      "loss": 0.7754,
+      "step": 5740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845325733401398e-05,
+      "loss": 0.7627,
+      "step": 5741
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8452642254250033e-05,
+      "loss": 0.7754,
+      "step": 5742
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845202706246877e-05,
+      "loss": 0.7656,
+      "step": 5743
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8451411758678352e-05,
+      "loss": 0.7734,
+      "step": 5744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8450796342886915e-05,
+      "loss": 0.7695,
+      "step": 5745
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845018081510263e-05,
+      "loss": 0.7627,
+      "step": 5746
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844956517533365e-05,
+      "loss": 0.7773,
+      "step": 5747
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844894942358813e-05,
+      "loss": 0.7891,
+      "step": 5748
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8448333559874234e-05,
+      "loss": 0.7979,
+      "step": 5749
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844771758420012e-05,
+      "loss": 0.7852,
+      "step": 5750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8447101496573957e-05,
+      "loss": 0.7578,
+      "step": 5751
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8446485297003906e-05,
+      "loss": 0.7832,
+      "step": 5752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445868985498133e-05,
+      "loss": 0.7529,
+      "step": 5753
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445252562064802e-05,
+      "loss": 0.7529,
+      "step": 5754
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844463602671209e-05,
+      "loss": 0.751,
+      "step": 5755
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8444019379448162e-05,
+      "loss": 0.7432,
+      "step": 5756
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8443402620281193e-05,
+      "loss": 0.7695,
+      "step": 5757
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8442785749219352e-05,
+      "loss": 0.8271,
+      "step": 5758
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844216876627082e-05,
+      "loss": 0.7852,
+      "step": 5759
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844155167144377e-05,
+      "loss": 0.7891,
+      "step": 5760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844093446474638e-05,
+      "loss": 0.791,
+      "step": 5761
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8440317146186825e-05,
+      "loss": 0.79,
+      "step": 5762
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439699715773296e-05,
+      "loss": 0.7939,
+      "step": 5763
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843908217351397e-05,
+      "loss": 0.7832,
+      "step": 5764
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8438464519417027e-05,
+      "loss": 0.7852,
+      "step": 5765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843784675349066e-05,
+      "loss": 0.7783,
+      "step": 5766
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8437228875743048e-05,
+      "loss": 0.7695,
+      "step": 5767
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843661088618239e-05,
+      "loss": 0.8096,
+      "step": 5768
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8435992784816865e-05,
+      "loss": 0.7705,
+      "step": 5769
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843537457165467e-05,
+      "loss": 0.7822,
+      "step": 5770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434756246703995e-05,
+      "loss": 0.791,
+      "step": 5771
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434137809973038e-05,
+      "loss": 0.7891,
+      "step": 5772
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8433519261469988e-05,
+      "loss": 0.7656,
+      "step": 5773
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843290060120305e-05,
+      "loss": 0.7822,
+      "step": 5774
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843228182918042e-05,
+      "loss": 0.7578,
+      "step": 5775
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8431662945410296e-05,
+      "loss": 0.7695,
+      "step": 5776
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843104394990088e-05,
+      "loss": 0.7422,
+      "step": 5777
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8430424842660378e-05,
+      "loss": 0.7627,
+      "step": 5778
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842980562369699e-05,
+      "loss": 0.75,
+      "step": 5779
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842918629301893e-05,
+      "loss": 0.7705,
+      "step": 5780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8428566850634398e-05,
+      "loss": 0.7725,
+      "step": 5781
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8427947296551604e-05,
+      "loss": 0.7686,
+      "step": 5782
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8427327630778764e-05,
+      "loss": 0.7783,
+      "step": 5783
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842670785332408e-05,
+      "loss": 0.7812,
+      "step": 5784
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842608796419578e-05,
+      "loss": 0.7568,
+      "step": 5785
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425467963402065e-05,
+      "loss": 0.7773,
+      "step": 5786
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842484785095116e-05,
+      "loss": 0.7686,
+      "step": 5787
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8424227626851276e-05,
+      "loss": 0.7686,
+      "step": 5788
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8423607291110638e-05,
+      "loss": 0.7686,
+      "step": 5789
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842298684373747e-05,
+      "loss": 0.7666,
+      "step": 5790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8422366284739987e-05,
+      "loss": 0.7861,
+      "step": 5791
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421745614126415e-05,
+      "loss": 0.791,
+      "step": 5792
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421124831904977e-05,
+      "loss": 0.792,
+      "step": 5793
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8420503938083906e-05,
+      "loss": 0.7832,
+      "step": 5794
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8419882932671428e-05,
+      "loss": 0.7646,
+      "step": 5795
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8419261815675772e-05,
+      "loss": 0.7588,
+      "step": 5796
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418640587105166e-05,
+      "loss": 0.7773,
+      "step": 5797
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841801924696785e-05,
+      "loss": 0.7773,
+      "step": 5798
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841739779527205e-05,
+      "loss": 0.7559,
+      "step": 5799
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8416776232026008e-05,
+      "loss": 0.7559,
+      "step": 5800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8416154557237957e-05,
+      "loss": 0.8018,
+      "step": 5801
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8415532770916142e-05,
+      "loss": 0.7656,
+      "step": 5802
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414910873068795e-05,
+      "loss": 0.8037,
+      "step": 5803
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414288863704163e-05,
+      "loss": 0.7744,
+      "step": 5804
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413666742830486e-05,
+      "loss": 0.7764,
+      "step": 5805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841304451045601e-05,
+      "loss": 0.7451,
+      "step": 5806
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8412422166588982e-05,
+      "loss": 0.7793,
+      "step": 5807
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411799711237648e-05,
+      "loss": 0.8008,
+      "step": 5808
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841117714441026e-05,
+      "loss": 0.7715,
+      "step": 5809
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8410554466115062e-05,
+      "loss": 0.7588,
+      "step": 5810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840993167636031e-05,
+      "loss": 0.7734,
+      "step": 5811
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840930877515426e-05,
+      "loss": 0.7588,
+      "step": 5812
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408685762505162e-05,
+      "loss": 0.7773,
+      "step": 5813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408062638421275e-05,
+      "loss": 0.7656,
+      "step": 5814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407439402910858e-05,
+      "loss": 0.7656,
+      "step": 5815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840681605598217e-05,
+      "loss": 0.7803,
+      "step": 5816
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8406192597643467e-05,
+      "loss": 0.7754,
+      "step": 5817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8405569027903018e-05,
+      "loss": 0.7588,
+      "step": 5818
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8404945346769082e-05,
+      "loss": 0.7598,
+      "step": 5819
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8404321554249927e-05,
+      "loss": 0.751,
+      "step": 5820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8403697650353818e-05,
+      "loss": 0.7627,
+      "step": 5821
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8403073635089023e-05,
+      "loss": 0.7549,
+      "step": 5822
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840244950846381e-05,
+      "loss": 0.7969,
+      "step": 5823
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.840182527048646e-05,
+      "loss": 0.7764,
+      "step": 5824
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8401200921165233e-05,
+      "loss": 0.7422,
+      "step": 5825
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400576460508407e-05,
+      "loss": 0.7646,
+      "step": 5826
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8399951888524263e-05,
+      "loss": 0.7891,
+      "step": 5827
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8399327205221073e-05,
+      "loss": 0.7705,
+      "step": 5828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398702410607115e-05,
+      "loss": 0.7676,
+      "step": 5829
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839807750469067e-05,
+      "loss": 0.7988,
+      "step": 5830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839745248748002e-05,
+      "loss": 0.7627,
+      "step": 5831
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839682735898345e-05,
+      "loss": 0.7744,
+      "step": 5832
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396202119209242e-05,
+      "loss": 0.7842,
+      "step": 5833
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839557676816568e-05,
+      "loss": 0.7637,
+      "step": 5834
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8394951305861055e-05,
+      "loss": 0.7764,
+      "step": 5835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8394325732303655e-05,
+      "loss": 0.7441,
+      "step": 5836
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393700047501767e-05,
+      "loss": 0.7627,
+      "step": 5837
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393074251463686e-05,
+      "loss": 0.7627,
+      "step": 5838
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392448344197708e-05,
+      "loss": 0.7676,
+      "step": 5839
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8391822325712123e-05,
+      "loss": 0.7529,
+      "step": 5840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839119619601523e-05,
+      "loss": 0.7861,
+      "step": 5841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390569955115322e-05,
+      "loss": 0.7803,
+      "step": 5842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.83899436030207e-05,
+      "loss": 0.7598,
+      "step": 5843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838931713973967e-05,
+      "loss": 0.7715,
+      "step": 5844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838869056528053e-05,
+      "loss": 0.7627,
+      "step": 5845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8388063879651583e-05,
+      "loss": 0.7793,
+      "step": 5846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8387437082861134e-05,
+      "loss": 0.7695,
+      "step": 5847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8386810174917493e-05,
+      "loss": 0.7559,
+      "step": 5848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8386183155828963e-05,
+      "loss": 0.7686,
+      "step": 5849
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385556025603856e-05,
+      "loss": 0.7578,
+      "step": 5850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8384928784250486e-05,
+      "loss": 0.7715,
+      "step": 5851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8384301431777158e-05,
+      "loss": 0.7754,
+      "step": 5852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383673968192192e-05,
+      "loss": 0.7793,
+      "step": 5853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383046393503903e-05,
+      "loss": 0.7852,
+      "step": 5854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382418707720603e-05,
+      "loss": 0.7842,
+      "step": 5855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8381790910850615e-05,
+      "loss": 0.7744,
+      "step": 5856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8381163002902262e-05,
+      "loss": 0.7734,
+      "step": 5857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8380534983883857e-05,
+      "loss": 0.7549,
+      "step": 5858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8379906853803725e-05,
+      "loss": 0.7637,
+      "step": 5859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837927861267019e-05,
+      "loss": 0.75,
+      "step": 5860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378650260491584e-05,
+      "loss": 0.7529,
+      "step": 5861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378021797276226e-05,
+      "loss": 0.7627,
+      "step": 5862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837739322303245e-05,
+      "loss": 0.7803,
+      "step": 5863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8376764537768585e-05,
+      "loss": 0.7832,
+      "step": 5864
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837613574149296e-05,
+      "loss": 0.7803,
+      "step": 5865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837550683421391e-05,
+      "loss": 0.7666,
+      "step": 5866
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837487781593977e-05,
+      "loss": 0.7979,
+      "step": 5867
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374248686678876e-05,
+      "loss": 0.7637,
+      "step": 5868
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8373619446439562e-05,
+      "loss": 0.7559,
+      "step": 5869
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837299009523017e-05,
+      "loss": 0.7568,
+      "step": 5870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837236063305904e-05,
+      "loss": 0.7832,
+      "step": 5871
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8371731059934514e-05,
+      "loss": 0.7686,
+      "step": 5872
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8371101375864935e-05,
+      "loss": 0.7822,
+      "step": 5873
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8370471580858648e-05,
+      "loss": 0.7695,
+      "step": 5874
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369841674923998e-05,
+      "loss": 0.7695,
+      "step": 5875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369211658069336e-05,
+      "loss": 0.7666,
+      "step": 5876
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8368581530303005e-05,
+      "loss": 0.7637,
+      "step": 5877
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8367951291633363e-05,
+      "loss": 0.79,
+      "step": 5878
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8367320942068757e-05,
+      "loss": 0.7529,
+      "step": 5879
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8366690481617544e-05,
+      "loss": 0.7529,
+      "step": 5880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8366059910288075e-05,
+      "loss": 0.8115,
+      "step": 5881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8365429228088715e-05,
+      "loss": 0.8086,
+      "step": 5882
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836479843502781e-05,
+      "loss": 0.7607,
+      "step": 5883
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8364167531113726e-05,
+      "loss": 0.7725,
+      "step": 5884
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836353651635483e-05,
+      "loss": 0.8076,
+      "step": 5885
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836290539075947e-05,
+      "loss": 0.7861,
+      "step": 5886
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8362274154336024e-05,
+      "loss": 0.7734,
+      "step": 5887
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8361642807092845e-05,
+      "loss": 0.7656,
+      "step": 5888
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8361011349038313e-05,
+      "loss": 0.7578,
+      "step": 5889
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8360379780180784e-05,
+      "loss": 0.7773,
+      "step": 5890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8359748100528633e-05,
+      "loss": 0.7715,
+      "step": 5891
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8359116310090233e-05,
+      "loss": 0.7686,
+      "step": 5892
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8358484408873954e-05,
+      "loss": 0.752,
+      "step": 5893
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8357852396888175e-05,
+      "loss": 0.7715,
+      "step": 5894
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8357220274141264e-05,
+      "loss": 0.7549,
+      "step": 5895
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8356588040641603e-05,
+      "loss": 0.7988,
+      "step": 5896
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835595569639757e-05,
+      "loss": 0.7676,
+      "step": 5897
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355323241417543e-05,
+      "loss": 0.7676,
+      "step": 5898
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8354690675709907e-05,
+      "loss": 0.7607,
+      "step": 5899
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835405799928304e-05,
+      "loss": 0.7754,
+      "step": 5900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8353425212145332e-05,
+      "loss": 0.7607,
+      "step": 5901
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8352792314305165e-05,
+      "loss": 0.7842,
+      "step": 5902
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835215930577093e-05,
+      "loss": 0.7832,
+      "step": 5903
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8351526186551014e-05,
+      "loss": 0.7656,
+      "step": 5904
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350892956653802e-05,
+      "loss": 0.7529,
+      "step": 5905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350259616087697e-05,
+      "loss": 0.7568,
+      "step": 5906
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8349626164861084e-05,
+      "loss": 0.7695,
+      "step": 5907
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348992602982357e-05,
+      "loss": 0.7695,
+      "step": 5908
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348358930459916e-05,
+      "loss": 0.7646,
+      "step": 5909
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834772514730216e-05,
+      "loss": 0.7754,
+      "step": 5910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8347091253517484e-05,
+      "loss": 0.7764,
+      "step": 5911
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8346457249114287e-05,
+      "loss": 0.7627,
+      "step": 5912
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8345823134100978e-05,
+      "loss": 0.7686,
+      "step": 5913
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8345188908485956e-05,
+      "loss": 0.7432,
+      "step": 5914
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834455457227763e-05,
+      "loss": 0.8066,
+      "step": 5915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8343920125484398e-05,
+      "loss": 0.748,
+      "step": 5916
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8343285568114674e-05,
+      "loss": 0.7686,
+      "step": 5917
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8342650900176868e-05,
+      "loss": 0.7744,
+      "step": 5918
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834201612167939e-05,
+      "loss": 0.75,
+      "step": 5919
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834138123263065e-05,
+      "loss": 0.8018,
+      "step": 5920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8340746233039063e-05,
+      "loss": 0.7705,
+      "step": 5921
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8340111122913047e-05,
+      "loss": 0.7578,
+      "step": 5922
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833947590226101e-05,
+      "loss": 0.749,
+      "step": 5923
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338840571091383e-05,
+      "loss": 0.7744,
+      "step": 5924
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338205129412577e-05,
+      "loss": 0.7822,
+      "step": 5925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337569577233015e-05,
+      "loss": 0.7783,
+      "step": 5926
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8336933914561122e-05,
+      "loss": 0.7695,
+      "step": 5927
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8336298141405317e-05,
+      "loss": 0.7793,
+      "step": 5928
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833566225777403e-05,
+      "loss": 0.8096,
+      "step": 5929
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8335026263675682e-05,
+      "loss": 0.748,
+      "step": 5930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833439015911871e-05,
+      "loss": 0.7588,
+      "step": 5931
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8333753944111538e-05,
+      "loss": 0.7539,
+      "step": 5932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.83331176186626e-05,
+      "loss": 0.7803,
+      "step": 5933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332481182780327e-05,
+      "loss": 0.7998,
+      "step": 5934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8331844636473154e-05,
+      "loss": 0.752,
+      "step": 5935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8331207979749516e-05,
+      "loss": 0.7451,
+      "step": 5936
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.833057121261785e-05,
+      "loss": 0.7617,
+      "step": 5937
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329934335086595e-05,
+      "loss": 0.7832,
+      "step": 5938
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329297347164195e-05,
+      "loss": 0.7949,
+      "step": 5939
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8328660248859086e-05,
+      "loss": 0.7666,
+      "step": 5940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8328023040179714e-05,
+      "loss": 0.7676,
+      "step": 5941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832738572113452e-05,
+      "loss": 0.7812,
+      "step": 5942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8326748291731954e-05,
+      "loss": 0.7588,
+      "step": 5943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8326110751980463e-05,
+      "loss": 0.7656,
+      "step": 5944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8325473101888494e-05,
+      "loss": 0.7617,
+      "step": 5945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83248353414645e-05,
+      "loss": 0.7666,
+      "step": 5946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8324197470716932e-05,
+      "loss": 0.752,
+      "step": 5947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832355948965424e-05,
+      "loss": 0.7656,
+      "step": 5948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8322921398284885e-05,
+      "loss": 0.7471,
+      "step": 5949
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8322283196617312e-05,
+      "loss": 0.748,
+      "step": 5950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321644884659992e-05,
+      "loss": 0.7549,
+      "step": 5951
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321006462421377e-05,
+      "loss": 0.7686,
+      "step": 5952
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832036792990993e-05,
+      "loss": 0.7803,
+      "step": 5953
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319729287134106e-05,
+      "loss": 0.7598,
+      "step": 5954
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319090534102383e-05,
+      "loss": 0.7559,
+      "step": 5955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831845167082321e-05,
+      "loss": 0.7793,
+      "step": 5956
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317812697305063e-05,
+      "loss": 0.7402,
+      "step": 5957
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831717361355641e-05,
+      "loss": 0.7793,
+      "step": 5958
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316534419585715e-05,
+      "loss": 0.7344,
+      "step": 5959
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8315895115401458e-05,
+      "loss": 0.7852,
+      "step": 5960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8315255701012098e-05,
+      "loss": 0.7715,
+      "step": 5961
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831461617642612e-05,
+      "loss": 0.7783,
+      "step": 5962
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831397654165199e-05,
+      "loss": 0.7432,
+      "step": 5963
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8313336796698195e-05,
+      "loss": 0.7773,
+      "step": 5964
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8312696941573206e-05,
+      "loss": 0.752,
+      "step": 5965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8312056976285503e-05,
+      "loss": 0.7617,
+      "step": 5966
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831141690084357e-05,
+      "loss": 0.7969,
+      "step": 5967
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831077671525589e-05,
+      "loss": 0.7754,
+      "step": 5968
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8310136419530944e-05,
+      "loss": 0.7637,
+      "step": 5969
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8309496013677214e-05,
+      "loss": 0.7568,
+      "step": 5970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8308855497703194e-05,
+      "loss": 0.7832,
+      "step": 5971
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830821487161737e-05,
+      "loss": 0.7764,
+      "step": 5972
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8307574135428235e-05,
+      "loss": 0.75,
+      "step": 5973
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830693328914427e-05,
+      "loss": 0.7607,
+      "step": 5974
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830629233277398e-05,
+      "loss": 0.7363,
+      "step": 5975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830565126632585e-05,
+      "loss": 0.7549,
+      "step": 5976
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830501008980838e-05,
+      "loss": 0.7705,
+      "step": 5977
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8304368803230064e-05,
+      "loss": 0.7861,
+      "step": 5978
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830372740659941e-05,
+      "loss": 0.7588,
+      "step": 5979
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8303085899924903e-05,
+      "loss": 0.7979,
+      "step": 5980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8302444283215054e-05,
+      "loss": 0.7637,
+      "step": 5981
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8301802556478363e-05,
+      "loss": 0.7734,
+      "step": 5982
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830116071972334e-05,
+      "loss": 0.751,
+      "step": 5983
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830051877295848e-05,
+      "loss": 0.7568,
+      "step": 5984
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8299876716192305e-05,
+      "loss": 0.7627,
+      "step": 5985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829923454943331e-05,
+      "loss": 0.7588,
+      "step": 5986
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8298592272690012e-05,
+      "loss": 0.7715,
+      "step": 5987
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829794988597092e-05,
+      "loss": 0.7568,
+      "step": 5988
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829730738928455e-05,
+      "loss": 0.7734,
+      "step": 5989
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296664782639414e-05,
+      "loss": 0.7627,
+      "step": 5990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829602206604403e-05,
+      "loss": 0.7676,
+      "step": 5991
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8295379239506918e-05,
+      "loss": 0.7764,
+      "step": 5992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829473630303659e-05,
+      "loss": 0.7559,
+      "step": 5993
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8294093256641566e-05,
+      "loss": 0.7598,
+      "step": 5994
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8293450100330375e-05,
+      "loss": 0.791,
+      "step": 5995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829280683411154e-05,
+      "loss": 0.7627,
+      "step": 5996
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829216345799358e-05,
+      "loss": 0.7803,
+      "step": 5997
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8291519971985026e-05,
+      "loss": 0.7715,
+      "step": 5998
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8290876376094405e-05,
+      "loss": 0.7676,
+      "step": 5999
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829023267033024e-05,
+      "loss": 0.7773,
+      "step": 6000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828958885470107e-05,
+      "loss": 0.7393,
+      "step": 6001
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288944929215425e-05,
+      "loss": 0.7637,
+      "step": 6002
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288300893881837e-05,
+      "loss": 0.7451,
+      "step": 6003
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828765674870884e-05,
+      "loss": 0.8037,
+      "step": 6004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828701249370497e-05,
+      "loss": 0.7842,
+      "step": 6005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8286368128878767e-05,
+      "loss": 0.7451,
+      "step": 6006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8285723654238775e-05,
+      "loss": 0.7559,
+      "step": 6007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8285079069793524e-05,
+      "loss": 0.7891,
+      "step": 6008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8284434375551563e-05,
+      "loss": 0.8086,
+      "step": 6009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8283789571521437e-05,
+      "loss": 0.7676,
+      "step": 6010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8283144657711686e-05,
+      "loss": 0.8037,
+      "step": 6011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828249963413086e-05,
+      "loss": 0.7646,
+      "step": 6012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8281854500787508e-05,
+      "loss": 0.7666,
+      "step": 6013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8281209257690176e-05,
+      "loss": 0.7734,
+      "step": 6014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8280563904847416e-05,
+      "loss": 0.7432,
+      "step": 6015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279918442267784e-05,
+      "loss": 0.7656,
+      "step": 6016
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827927286995983e-05,
+      "loss": 0.7578,
+      "step": 6017
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8278627187932113e-05,
+      "loss": 0.7656,
+      "step": 6018
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277981396193188e-05,
+      "loss": 0.7705,
+      "step": 6019
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827733549475161e-05,
+      "loss": 0.7734,
+      "step": 6020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8276689483615942e-05,
+      "loss": 0.7773,
+      "step": 6021
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8276043362794745e-05,
+      "loss": 0.7578,
+      "step": 6022
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827539713229658e-05,
+      "loss": 0.7607,
+      "step": 6023
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8274750792130013e-05,
+      "loss": 0.7842,
+      "step": 6024
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827410434230361e-05,
+      "loss": 0.751,
+      "step": 6025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827345778282594e-05,
+      "loss": 0.7432,
+      "step": 6026
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8272811113705563e-05,
+      "loss": 0.7773,
+      "step": 6027
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8272164334951053e-05,
+      "loss": 0.7832,
+      "step": 6028
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8271517446570985e-05,
+      "loss": 0.792,
+      "step": 6029
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827087044857393e-05,
+      "loss": 0.7773,
+      "step": 6030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827022334096846e-05,
+      "loss": 0.7969,
+      "step": 6031
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826957612376315e-05,
+      "loss": 0.7871,
+      "step": 6032
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8268928796966585e-05,
+      "loss": 0.7559,
+      "step": 6033
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8268281360587334e-05,
+      "loss": 0.7734,
+      "step": 6034
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8267633814633982e-05,
+      "loss": 0.7744,
+      "step": 6035
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826698615911511e-05,
+      "loss": 0.7725,
+      "step": 6036
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82663383940393e-05,
+      "loss": 0.7578,
+      "step": 6037
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8265690519415135e-05,
+      "loss": 0.7852,
+      "step": 6038
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8265042535251206e-05,
+      "loss": 0.7451,
+      "step": 6039
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8264394441556095e-05,
+      "loss": 0.7793,
+      "step": 6040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263746238338393e-05,
+      "loss": 0.7832,
+      "step": 6041
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826309792560669e-05,
+      "loss": 0.7676,
+      "step": 6042
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8262449503369577e-05,
+      "loss": 0.7676,
+      "step": 6043
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826180097163565e-05,
+      "loss": 0.7598,
+      "step": 6044
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8261152330413495e-05,
+      "loss": 0.7871,
+      "step": 6045
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8260503579711718e-05,
+      "loss": 0.75,
+      "step": 6046
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8259854719538915e-05,
+      "loss": 0.7637,
+      "step": 6047
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8259205749903683e-05,
+      "loss": 0.7646,
+      "step": 6048
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.825855667081462e-05,
+      "loss": 0.7783,
+      "step": 6049
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825790748228033e-05,
+      "loss": 0.791,
+      "step": 6050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8257258184309417e-05,
+      "loss": 0.792,
+      "step": 6051
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8256608776910484e-05,
+      "loss": 0.7822,
+      "step": 6052
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8255959260092136e-05,
+      "loss": 0.7656,
+      "step": 6053
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8255309633862987e-05,
+      "loss": 0.7695,
+      "step": 6054
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8254659898231638e-05,
+      "loss": 0.7529,
+      "step": 6055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8254010053206706e-05,
+      "loss": 0.7832,
+      "step": 6056
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82533600987968e-05,
+      "loss": 0.7803,
+      "step": 6057
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825271003501053e-05,
+      "loss": 0.7627,
+      "step": 6058
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825205986185652e-05,
+      "loss": 0.7646,
+      "step": 6059
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8251409579343374e-05,
+      "loss": 0.792,
+      "step": 6060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250759187479724e-05,
+      "loss": 0.749,
+      "step": 6061
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250108686274176e-05,
+      "loss": 0.7793,
+      "step": 6062
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8249458075735358e-05,
+      "loss": 0.7793,
+      "step": 6063
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8248807355871894e-05,
+      "loss": 0.7539,
+      "step": 6064
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8248156526692404e-05,
+      "loss": 0.7773,
+      "step": 6065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8247505588205508e-05,
+      "loss": 0.7549,
+      "step": 6066
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246854540419838e-05,
+      "loss": 0.79,
+      "step": 6067
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246203383344025e-05,
+      "loss": 0.7637,
+      "step": 6068
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8245552116986694e-05,
+      "loss": 0.793,
+      "step": 6069
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8244900741356475e-05,
+      "loss": 0.7783,
+      "step": 6070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8244249256462004e-05,
+      "loss": 0.7695,
+      "step": 6071
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8243597662311915e-05,
+      "loss": 0.7637,
+      "step": 6072
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242945958914837e-05,
+      "loss": 0.7803,
+      "step": 6073
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824229414627941e-05,
+      "loss": 0.7695,
+      "step": 6074
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8241642224414274e-05,
+      "loss": 0.7568,
+      "step": 6075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8240990193328067e-05,
+      "loss": 0.751,
+      "step": 6076
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824033805302943e-05,
+      "loss": 0.8125,
+      "step": 6077
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8239685803527e-05,
+      "loss": 0.7598,
+      "step": 6078
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823903344482943e-05,
+      "loss": 0.7715,
+      "step": 6079
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8238380976945365e-05,
+      "loss": 0.7744,
+      "step": 6080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8237728399883444e-05,
+      "loss": 0.7725,
+      "step": 6081
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8237075713652323e-05,
+      "loss": 0.752,
+      "step": 6082
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8236422918260643e-05,
+      "loss": 0.7656,
+      "step": 6083
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8235770013717062e-05,
+      "loss": 0.7441,
+      "step": 6084
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823511700003023e-05,
+      "loss": 0.7803,
+      "step": 6085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234463877208803e-05,
+      "loss": 0.7773,
+      "step": 6086
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8233810645261434e-05,
+      "loss": 0.7803,
+      "step": 6087
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823315730419678e-05,
+      "loss": 0.7881,
+      "step": 6088
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8232503854023504e-05,
+      "loss": 0.7744,
+      "step": 6089
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8231850294750258e-05,
+      "loss": 0.7529,
+      "step": 6090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8231196626385704e-05,
+      "loss": 0.7666,
+      "step": 6091
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8230542848938513e-05,
+      "loss": 0.7529,
+      "step": 6092
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822988896241734e-05,
+      "loss": 0.7568,
+      "step": 6093
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8229234966830858e-05,
+      "loss": 0.7783,
+      "step": 6094
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8228580862187728e-05,
+      "loss": 0.7285,
+      "step": 6095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822792664849662e-05,
+      "loss": 0.7793,
+      "step": 6096
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8227272325766203e-05,
+      "loss": 0.7471,
+      "step": 6097
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8226617894005155e-05,
+      "loss": 0.7363,
+      "step": 6098
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822596335322214e-05,
+      "loss": 0.79,
+      "step": 6099
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8225308703425835e-05,
+      "loss": 0.751,
+      "step": 6100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8224653944624915e-05,
+      "loss": 0.7686,
+      "step": 6101
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8223999076828064e-05,
+      "loss": 0.7852,
+      "step": 6102
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822334410004395e-05,
+      "loss": 0.7637,
+      "step": 6103
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8222689014281257e-05,
+      "loss": 0.7441,
+      "step": 6104
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8222033819548673e-05,
+      "loss": 0.7529,
+      "step": 6105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8221378515854874e-05,
+      "loss": 0.7598,
+      "step": 6106
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.822072310320854e-05,
+      "loss": 0.7637,
+      "step": 6107
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8220067581618367e-05,
+      "loss": 0.7754,
+      "step": 6108
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821941195109304e-05,
+      "loss": 0.7734,
+      "step": 6109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218756211641237e-05,
+      "loss": 0.7822,
+      "step": 6110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218100363271663e-05,
+      "loss": 0.7764,
+      "step": 6111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8217444405993e-05,
+      "loss": 0.7803,
+      "step": 6112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8216788339813944e-05,
+      "loss": 0.7871,
+      "step": 6113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821613216474319e-05,
+      "loss": 0.7646,
+      "step": 6114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8215475880789433e-05,
+      "loss": 0.7939,
+      "step": 6115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821481948796137e-05,
+      "loss": 0.7471,
+      "step": 6116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82141629862677e-05,
+      "loss": 0.7695,
+      "step": 6117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821350637571712e-05,
+      "loss": 0.7744,
+      "step": 6118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821284965631834e-05,
+      "loss": 0.7578,
+      "step": 6119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8212192828080056e-05,
+      "loss": 0.7656,
+      "step": 6120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821153589101097e-05,
+      "loss": 0.7773,
+      "step": 6121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82108788451198e-05,
+      "loss": 0.7607,
+      "step": 6122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8210221690415242e-05,
+      "loss": 0.7432,
+      "step": 6123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820956442690601e-05,
+      "loss": 0.7676,
+      "step": 6124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820890705460081e-05,
+      "loss": 0.7559,
+      "step": 6125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820824957350836e-05,
+      "loss": 0.7861,
+      "step": 6126
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820759198363737e-05,
+      "loss": 0.8066,
+      "step": 6127
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820693428499655e-05,
+      "loss": 0.7607,
+      "step": 6128
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8206276477594625e-05,
+      "loss": 0.751,
+      "step": 6129
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205618561440305e-05,
+      "loss": 0.748,
+      "step": 6130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8204960536542314e-05,
+      "loss": 0.7666,
+      "step": 6131
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8204302402909368e-05,
+      "loss": 0.7578,
+      "step": 6132
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8203644160550193e-05,
+      "loss": 0.7686,
+      "step": 6133
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820298580947351e-05,
+      "loss": 0.751,
+      "step": 6134
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8202327349688046e-05,
+      "loss": 0.7754,
+      "step": 6135
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820166878120252e-05,
+      "loss": 0.7949,
+      "step": 6136
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201010104025666e-05,
+      "loss": 0.7754,
+      "step": 6137
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8200351318166214e-05,
+      "loss": 0.7432,
+      "step": 6138
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8199692423632893e-05,
+      "loss": 0.7383,
+      "step": 6139
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8199033420434432e-05,
+      "loss": 0.7793,
+      "step": 6140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8198374308579565e-05,
+      "loss": 0.7998,
+      "step": 6141
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819771508807703e-05,
+      "loss": 0.7422,
+      "step": 6142
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8197055758935564e-05,
+      "loss": 0.7656,
+      "step": 6143
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196396321163895e-05,
+      "loss": 0.7725,
+      "step": 6144
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8195736774770775e-05,
+      "loss": 0.75,
+      "step": 6145
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8195077119764938e-05,
+      "loss": 0.7549,
+      "step": 6146
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8194417356155124e-05,
+      "loss": 0.7686,
+      "step": 6147
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193757483950083e-05,
+      "loss": 0.7852,
+      "step": 6148
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193097503158556e-05,
+      "loss": 0.7637,
+      "step": 6149
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8192437413789286e-05,
+      "loss": 0.749,
+      "step": 6150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191777215851027e-05,
+      "loss": 0.7676,
+      "step": 6151
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191116909352522e-05,
+      "loss": 0.7578,
+      "step": 6152
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8190456494302523e-05,
+      "loss": 0.75,
+      "step": 6153
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8189795970709786e-05,
+      "loss": 0.7461,
+      "step": 6154
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8189135338583065e-05,
+      "loss": 0.7773,
+      "step": 6155
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818847459793111e-05,
+      "loss": 0.7812,
+      "step": 6156
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818781374876268e-05,
+      "loss": 0.7773,
+      "step": 6157
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8187152791086533e-05,
+      "loss": 0.7803,
+      "step": 6158
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8186491724911425e-05,
+      "loss": 0.7646,
+      "step": 6159
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8185830550246125e-05,
+      "loss": 0.7588,
+      "step": 6160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8185169267099384e-05,
+      "loss": 0.7568,
+      "step": 6161
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818450787547997e-05,
+      "loss": 0.7764,
+      "step": 6162
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8183846375396653e-05,
+      "loss": 0.7744,
+      "step": 6163
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8183184766858194e-05,
+      "loss": 0.7725,
+      "step": 6164
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818252304987336e-05,
+      "loss": 0.7578,
+      "step": 6165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8181861224450927e-05,
+      "loss": 0.7441,
+      "step": 6166
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818119929059966e-05,
+      "loss": 0.7832,
+      "step": 6167
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818053724832833e-05,
+      "loss": 0.7754,
+      "step": 6168
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817987509764571e-05,
+      "loss": 0.7676,
+      "step": 6169
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8179212838560585e-05,
+      "loss": 0.7715,
+      "step": 6170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8178550471081723e-05,
+      "loss": 0.7549,
+      "step": 6171
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81778879952179e-05,
+      "loss": 0.7627,
+      "step": 6172
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81772254109779e-05,
+      "loss": 0.7607,
+      "step": 6173
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8176562718370507e-05,
+      "loss": 0.7666,
+      "step": 6174
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8175899917404492e-05,
+      "loss": 0.7793,
+      "step": 6175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8175237008088653e-05,
+      "loss": 0.793,
+      "step": 6176
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817457399043176e-05,
+      "loss": 0.7705,
+      "step": 6177
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8173910864442615e-05,
+      "loss": 0.7676,
+      "step": 6178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8173247630129994e-05,
+      "loss": 0.7783,
+      "step": 6179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817258428750269e-05,
+      "loss": 0.7715,
+      "step": 6180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8171920836569494e-05,
+      "loss": 0.7793,
+      "step": 6181
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81712572773392e-05,
+      "loss": 0.7578,
+      "step": 6182
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81705936098206e-05,
+      "loss": 0.7734,
+      "step": 6183
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816992983402249e-05,
+      "loss": 0.7686,
+      "step": 6184
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8169265949953664e-05,
+      "loss": 0.7686,
+      "step": 6185
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168601957622924e-05,
+      "loss": 0.7461,
+      "step": 6186
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816793785703907e-05,
+      "loss": 0.7539,
+      "step": 6187
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8167273648210897e-05,
+      "loss": 0.749,
+      "step": 6188
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8166609331147213e-05,
+      "loss": 0.748,
+      "step": 6189
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816594490585682e-05,
+      "loss": 0.7607,
+      "step": 6190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8165280372348517e-05,
+      "loss": 0.7539,
+      "step": 6191
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816461573063112e-05,
+      "loss": 0.7646,
+      "step": 6192
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8163950980713436e-05,
+      "loss": 0.7598,
+      "step": 6193
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816328612260427e-05,
+      "loss": 0.749,
+      "step": 6194
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8162621156312435e-05,
+      "loss": 0.7812,
+      "step": 6195
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8161956081846743e-05,
+      "loss": 0.7695,
+      "step": 6196
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816129089921601e-05,
+      "loss": 0.7646,
+      "step": 6197
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8160625608429047e-05,
+      "loss": 0.7998,
+      "step": 6198
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8159960209494673e-05,
+      "loss": 0.7363,
+      "step": 6199
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8159294702421707e-05,
+      "loss": 0.7715,
+      "step": 6200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8158629087218968e-05,
+      "loss": 0.7539,
+      "step": 6201
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815796336389527e-05,
+      "loss": 0.75,
+      "step": 6202
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815729753245945e-05,
+      "loss": 0.7764,
+      "step": 6203
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815663159292032e-05,
+      "loss": 0.7715,
+      "step": 6204
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815596554528671e-05,
+      "loss": 0.7705,
+      "step": 6205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8155299389567447e-05,
+      "loss": 0.7324,
+      "step": 6206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8154633125771356e-05,
+      "loss": 0.7871,
+      "step": 6207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8153966753907268e-05,
+      "loss": 0.7979,
+      "step": 6208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8153300273984016e-05,
+      "loss": 0.7646,
+      "step": 6209
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815263368601043e-05,
+      "loss": 0.7549,
+      "step": 6210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8151966989995343e-05,
+      "loss": 0.7314,
+      "step": 6211
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8151300185947595e-05,
+      "loss": 0.7832,
+      "step": 6212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815063327387602e-05,
+      "loss": 0.7383,
+      "step": 6213
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.814996625378945e-05,
+      "loss": 0.7812,
+      "step": 6214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8149299125696735e-05,
+      "loss": 0.7764,
+      "step": 6215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148631889606713e-05,
+      "loss": 0.7832,
+      "step": 6216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.814796454552822e-05,
+      "loss": 0.7646,
+      "step": 6217
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814729709347011e-05,
+      "loss": 0.7451,
+      "step": 6218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8146629533441217e-05,
+      "loss": 0.791,
+      "step": 6219
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.81459618654504e-05,
+      "loss": 0.7588,
+      "step": 6220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8145294089506494e-05,
+      "loss": 0.7832,
+      "step": 6221
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814462620561836e-05,
+      "loss": 0.7725,
+      "step": 6222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8143958213794842e-05,
+      "loss": 0.7773,
+      "step": 6223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.81432901140448e-05,
+      "loss": 0.7695,
+      "step": 6224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8142621906377077e-05,
+      "loss": 0.7725,
+      "step": 6225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8141953590800536e-05,
+      "loss": 0.7598,
+      "step": 6226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8141285167324035e-05,
+      "loss": 0.7627,
+      "step": 6227
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8140616635956424e-05,
+      "loss": 0.7549,
+      "step": 6228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813994799670657e-05,
+      "loss": 0.7637,
+      "step": 6229
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8139279249583334e-05,
+      "loss": 0.7744,
+      "step": 6230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8138610394595574e-05,
+      "loss": 0.7871,
+      "step": 6231
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8137941431752155e-05,
+      "loss": 0.7881,
+      "step": 6232
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8137272361061943e-05,
+      "loss": 0.7734,
+      "step": 6233
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136603182533808e-05,
+      "loss": 0.7852,
+      "step": 6234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8135933896176614e-05,
+      "loss": 0.7373,
+      "step": 6235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8135264501999233e-05,
+      "loss": 0.7588,
+      "step": 6236
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8134595000010534e-05,
+      "loss": 0.7373,
+      "step": 6237
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8133925390219393e-05,
+      "loss": 0.7686,
+      "step": 6238
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813325567263468e-05,
+      "loss": 0.7842,
+      "step": 6239
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132585847265267e-05,
+      "loss": 0.7432,
+      "step": 6240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813191591412004e-05,
+      "loss": 0.7725,
+      "step": 6241
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813124587320787e-05,
+      "loss": 0.7598,
+      "step": 6242
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813057572453764e-05,
+      "loss": 0.7754,
+      "step": 6243
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812990546811823e-05,
+      "loss": 0.7598,
+      "step": 6244
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812923510395853e-05,
+      "loss": 0.7812,
+      "step": 6245
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128564632067406e-05,
+      "loss": 0.7686,
+      "step": 6246
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8127894052453762e-05,
+      "loss": 0.7803,
+      "step": 6247
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8127223365126472e-05,
+      "loss": 0.7676,
+      "step": 6248
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812655257009443e-05,
+      "loss": 0.7656,
+      "step": 6249
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8125881667366526e-05,
+      "loss": 0.7627,
+      "step": 6250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8125210656951647e-05,
+      "loss": 0.7803,
+      "step": 6251
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812453953885869e-05,
+      "loss": 0.7754,
+      "step": 6252
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8123868313096547e-05,
+      "loss": 0.7666,
+      "step": 6253
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812319697967411e-05,
+      "loss": 0.7656,
+      "step": 6254
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8122525538600282e-05,
+      "loss": 0.7832,
+      "step": 6255
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812185398988396e-05,
+      "loss": 0.7871,
+      "step": 6256
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812118233353404e-05,
+      "loss": 0.7588,
+      "step": 6257
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8120510569559422e-05,
+      "loss": 0.7607,
+      "step": 6258
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8119838697969016e-05,
+      "loss": 0.7588,
+      "step": 6259
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8119166718771716e-05,
+      "loss": 0.7627,
+      "step": 6260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118494631976435e-05,
+      "loss": 0.7744,
+      "step": 6261
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8117822437592076e-05,
+      "loss": 0.7822,
+      "step": 6262
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811715013562755e-05,
+      "loss": 0.7695,
+      "step": 6263
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8116477726091762e-05,
+      "loss": 0.8203,
+      "step": 6264
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811580520899363e-05,
+      "loss": 0.7705,
+      "step": 6265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811513258434206e-05,
+      "loss": 0.751,
+      "step": 6266
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8114459852145965e-05,
+      "loss": 0.7598,
+      "step": 6267
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113787012414268e-05,
+      "loss": 0.7832,
+      "step": 6268
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113114065155878e-05,
+      "loss": 0.7812,
+      "step": 6269
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112441010379716e-05,
+      "loss": 0.7441,
+      "step": 6270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8111767848094704e-05,
+      "loss": 0.7539,
+      "step": 6271
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811109457830976e-05,
+      "loss": 0.75,
+      "step": 6272
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110421201033806e-05,
+      "loss": 0.749,
+      "step": 6273
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8109747716275772e-05,
+      "loss": 0.7715,
+      "step": 6274
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810907412404457e-05,
+      "loss": 0.7832,
+      "step": 6275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108400424349142e-05,
+      "loss": 0.792,
+      "step": 6276
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8107726617198408e-05,
+      "loss": 0.7617,
+      "step": 6277
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8107052702601294e-05,
+      "loss": 0.7637,
+      "step": 6278
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810637868056674e-05,
+      "loss": 0.7637,
+      "step": 6279
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810570455110367e-05,
+      "loss": 0.7607,
+      "step": 6280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8105030314221026e-05,
+      "loss": 0.7656,
+      "step": 6281
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8104355969927737e-05,
+      "loss": 0.7734,
+      "step": 6282
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810368151823274e-05,
+      "loss": 0.8184,
+      "step": 6283
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103006959144977e-05,
+      "loss": 0.7549,
+      "step": 6284
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810233229267338e-05,
+      "loss": 0.7559,
+      "step": 6285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.81016575188269e-05,
+      "loss": 0.7598,
+      "step": 6286
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8100982637614472e-05,
+      "loss": 0.7666,
+      "step": 6287
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8100307649045043e-05,
+      "loss": 0.752,
+      "step": 6288
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8099632553127557e-05,
+      "loss": 0.7676,
+      "step": 6289
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809895734987096e-05,
+      "loss": 0.7842,
+      "step": 6290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8098282039284202e-05,
+      "loss": 0.7725,
+      "step": 6291
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8097606621376232e-05,
+      "loss": 0.7783,
+      "step": 6292
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8096931096155998e-05,
+      "loss": 0.79,
+      "step": 6293
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8096255463632454e-05,
+      "loss": 0.7607,
+      "step": 6294
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809557972381456e-05,
+      "loss": 0.7764,
+      "step": 6295
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8094903876711257e-05,
+      "loss": 0.7607,
+      "step": 6296
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8094227922331514e-05,
+      "loss": 0.7588,
+      "step": 6297
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8093551860684288e-05,
+      "loss": 0.7715,
+      "step": 6298
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809287569177853e-05,
+      "loss": 0.7539,
+      "step": 6299
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8092199415623207e-05,
+      "loss": 0.7607,
+      "step": 6300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091523032227284e-05,
+      "loss": 0.7812,
+      "step": 6301
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090846541599716e-05,
+      "loss": 0.7754,
+      "step": 6302
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.7764,
+      "step": 6303
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8089493238685527e-05,
+      "loss": 0.7422,
+      "step": 6304
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808881642641684e-05,
+      "loss": 0.7656,
+      "step": 6305
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088139506952377e-05,
+      "loss": 0.8037,
+      "step": 6306
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8087462480301114e-05,
+      "loss": 0.7852,
+      "step": 6307
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8086785346472027e-05,
+      "loss": 0.7529,
+      "step": 6308
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8086108105474082e-05,
+      "loss": 0.7744,
+      "step": 6309
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8085430757316258e-05,
+      "loss": 0.7822,
+      "step": 6310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8084753302007532e-05,
+      "loss": 0.7627,
+      "step": 6311
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808407573955688e-05,
+      "loss": 0.793,
+      "step": 6312
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083398069973285e-05,
+      "loss": 0.7559,
+      "step": 6313
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808272029326572e-05,
+      "loss": 0.751,
+      "step": 6314
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8082042409443176e-05,
+      "loss": 0.7715,
+      "step": 6315
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808136441851463e-05,
+      "loss": 0.7559,
+      "step": 6316
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080686320489067e-05,
+      "loss": 0.7842,
+      "step": 6317
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080008115375477e-05,
+      "loss": 0.7822,
+      "step": 6318
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807932980318285e-05,
+      "loss": 0.7686,
+      "step": 6319
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8078651383920173e-05,
+      "loss": 0.7549,
+      "step": 6320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807797285759643e-05,
+      "loss": 0.751,
+      "step": 6321
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807729422422062e-05,
+      "loss": 0.748,
+      "step": 6322
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807661548380174e-05,
+      "loss": 0.7803,
+      "step": 6323
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807593663634877e-05,
+      "loss": 0.7725,
+      "step": 6324
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8075257681870723e-05,
+      "loss": 0.7852,
+      "step": 6325
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807457862037659e-05,
+      "loss": 0.7646,
+      "step": 6326
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807389945187537e-05,
+      "loss": 0.7744,
+      "step": 6327
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8073220176376064e-05,
+      "loss": 0.7627,
+      "step": 6328
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807254079388767e-05,
+      "loss": 0.7627,
+      "step": 6329
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8071861304419198e-05,
+      "loss": 0.7734,
+      "step": 6330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8071181707979653e-05,
+      "loss": 0.7734,
+      "step": 6331
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.807050200457803e-05,
+      "loss": 0.7705,
+      "step": 6332
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806982219422335e-05,
+      "loss": 0.7705,
+      "step": 6333
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806914227692462e-05,
+      "loss": 0.7559,
+      "step": 6334
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068462252690844e-05,
+      "loss": 0.7666,
+      "step": 6335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8067782121531038e-05,
+      "loss": 0.7607,
+      "step": 6336
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8067101883454213e-05,
+      "loss": 0.7725,
+      "step": 6337
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8066421538469386e-05,
+      "loss": 0.7861,
+      "step": 6338
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8065741086585572e-05,
+      "loss": 0.79,
+      "step": 6339
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8065060527811794e-05,
+      "loss": 0.7666,
+      "step": 6340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806437986215706e-05,
+      "loss": 0.7686,
+      "step": 6341
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80636990896304e-05,
+      "loss": 0.7441,
+      "step": 6342
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8063018210240834e-05,
+      "loss": 0.7588,
+      "step": 6343
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8062337223997378e-05,
+      "loss": 0.7617,
+      "step": 6344
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8061656130909067e-05,
+      "loss": 0.7646,
+      "step": 6345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806097493098492e-05,
+      "loss": 0.7676,
+      "step": 6346
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806029362423397e-05,
+      "loss": 0.7842,
+      "step": 6347
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8059612210665238e-05,
+      "loss": 0.748,
+      "step": 6348
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8058930690287765e-05,
+      "loss": 0.7578,
+      "step": 6349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805824906311057e-05,
+      "loss": 0.7812,
+      "step": 6350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8057567329142696e-05,
+      "loss": 0.7607,
+      "step": 6351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056885488393175e-05,
+      "loss": 0.7959,
+      "step": 6352
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056203540871045e-05,
+      "loss": 0.7646,
+      "step": 6353
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8055521486585338e-05,
+      "loss": 0.7539,
+      "step": 6354
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8054839325545094e-05,
+      "loss": 0.7559,
+      "step": 6355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805415705775936e-05,
+      "loss": 0.7773,
+      "step": 6356
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805347468323717e-05,
+      "loss": 0.7832,
+      "step": 6357
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805279220198757e-05,
+      "loss": 0.7607,
+      "step": 6358
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8052109614019605e-05,
+      "loss": 0.7744,
+      "step": 6359
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805142691934232e-05,
+      "loss": 0.7803,
+      "step": 6360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805074411796476e-05,
+      "loss": 0.7588,
+      "step": 6361
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805006120989598e-05,
+      "loss": 0.7842,
+      "step": 6362
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8049378195145024e-05,
+      "loss": 0.7676,
+      "step": 6363
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8048695073720945e-05,
+      "loss": 0.7627,
+      "step": 6364
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80480118456328e-05,
+      "loss": 0.8008,
+      "step": 6365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8047328510889636e-05,
+      "loss": 0.7607,
+      "step": 6366
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8046645069500518e-05,
+      "loss": 0.7754,
+      "step": 6367
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8045961521474492e-05,
+      "loss": 0.7607,
+      "step": 6368
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8045277866820625e-05,
+      "loss": 0.7812,
+      "step": 6369
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044594105547976e-05,
+      "loss": 0.7852,
+      "step": 6370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8043910237665606e-05,
+      "loss": 0.7588,
+      "step": 6371
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8043226263182578e-05,
+      "loss": 0.7881,
+      "step": 6372
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804254218210795e-05,
+      "loss": 0.7627,
+      "step": 6373
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8041857994450795e-05,
+      "loss": 0.7871,
+      "step": 6374
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804117370022018e-05,
+      "loss": 0.7881,
+      "step": 6375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8040489299425172e-05,
+      "loss": 0.7451,
+      "step": 6376
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803980479207484e-05,
+      "loss": 0.7568,
+      "step": 6377
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8039120178178255e-05,
+      "loss": 0.7607,
+      "step": 6378
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803843545774449e-05,
+      "loss": 0.751,
+      "step": 6379
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8037750630782622e-05,
+      "loss": 0.7617,
+      "step": 6380
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803706569730172e-05,
+      "loss": 0.8066,
+      "step": 6381
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8036380657310873e-05,
+      "loss": 0.7607,
+      "step": 6382
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8035695510819144e-05,
+      "loss": 0.7764,
+      "step": 6383
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8035010257835623e-05,
+      "loss": 0.751,
+      "step": 6384
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803432489836939e-05,
+      "loss": 0.7754,
+      "step": 6385
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8033639432429527e-05,
+      "loss": 0.7598,
+      "step": 6386
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8032953860025114e-05,
+      "loss": 0.7539,
+      "step": 6387
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803226818116524e-05,
+      "loss": 0.7676,
+      "step": 6388
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8031582395858994e-05,
+      "loss": 0.7568,
+      "step": 6389
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803089650411546e-05,
+      "loss": 0.7734,
+      "step": 6390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803021050594373e-05,
+      "loss": 0.7568,
+      "step": 6391
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8029524401352897e-05,
+      "loss": 0.7402,
+      "step": 6392
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8028838190352047e-05,
+      "loss": 0.752,
+      "step": 6393
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8028151872950282e-05,
+      "loss": 0.7705,
+      "step": 6394
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802746544915669e-05,
+      "loss": 0.7715,
+      "step": 6395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802677891898037e-05,
+      "loss": 0.7822,
+      "step": 6396
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8026092282430426e-05,
+      "loss": 0.7471,
+      "step": 6397
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802540553951595e-05,
+      "loss": 0.7705,
+      "step": 6398
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024718690246047e-05,
+      "loss": 0.7783,
+      "step": 6399
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024031734629815e-05,
+      "loss": 0.7422,
+      "step": 6400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8023344672676363e-05,
+      "loss": 0.7734,
+      "step": 6401
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8022657504394793e-05,
+      "loss": 0.7764,
+      "step": 6402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8021970229794214e-05,
+      "loss": 0.7559,
+      "step": 6403
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802128284888373e-05,
+      "loss": 0.7432,
+      "step": 6404
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020595361672458e-05,
+      "loss": 0.7695,
+      "step": 6405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.80199077681695e-05,
+      "loss": 0.7637,
+      "step": 6406
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019220068383975e-05,
+      "loss": 0.7393,
+      "step": 6407
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801853226232499e-05,
+      "loss": 0.7861,
+      "step": 6408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8017844350001663e-05,
+      "loss": 0.7646,
+      "step": 6409
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8017156331423115e-05,
+      "loss": 0.75,
+      "step": 6410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801646820659846e-05,
+      "loss": 0.7725,
+      "step": 6411
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8015779975536814e-05,
+      "loss": 0.7793,
+      "step": 6412
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8015091638247303e-05,
+      "loss": 0.7578,
+      "step": 6413
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8014403194739044e-05,
+      "loss": 0.749,
+      "step": 6414
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8013714645021168e-05,
+      "loss": 0.7949,
+      "step": 6415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8013025989102792e-05,
+      "loss": 0.7637,
+      "step": 6416
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012337226993047e-05,
+      "loss": 0.7812,
+      "step": 6417
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8011648358701062e-05,
+      "loss": 0.7549,
+      "step": 6418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801095938423596e-05,
+      "loss": 0.7754,
+      "step": 6419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801027030360688e-05,
+      "loss": 0.7715,
+      "step": 6420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8009581116822942e-05,
+      "loss": 0.7627,
+      "step": 6421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800889182389329e-05,
+      "loss": 0.7539,
+      "step": 6422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8008202424827054e-05,
+      "loss": 0.7734,
+      "step": 6423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800751291963337e-05,
+      "loss": 0.7627,
+      "step": 6424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8006823308321383e-05,
+      "loss": 0.793,
+      "step": 6425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800613359090022e-05,
+      "loss": 0.7783,
+      "step": 6426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800544376737903e-05,
+      "loss": 0.7676,
+      "step": 6427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800475383776695e-05,
+      "loss": 0.7852,
+      "step": 6428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8004063802073128e-05,
+      "loss": 0.7461,
+      "step": 6429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8003373660306703e-05,
+      "loss": 0.7744,
+      "step": 6430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8002683412476822e-05,
+      "loss": 0.7705,
+      "step": 6431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8001993058592636e-05,
+      "loss": 0.7578,
+      "step": 6432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8001302598663292e-05,
+      "loss": 0.7607,
+      "step": 6433
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8000612032697943e-05,
+      "loss": 0.7793,
+      "step": 6434
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7999921360705736e-05,
+      "loss": 0.751,
+      "step": 6435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7999230582695822e-05,
+      "loss": 0.792,
+      "step": 6436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7998539698677363e-05,
+      "loss": 0.7354,
+      "step": 6437
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7997848708659507e-05,
+      "loss": 0.7646,
+      "step": 6438
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7997157612651423e-05,
+      "loss": 0.7861,
+      "step": 6439
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7996466410662257e-05,
+      "loss": 0.7832,
+      "step": 6440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7995775102701172e-05,
+      "loss": 0.7852,
+      "step": 6441
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7995083688777333e-05,
+      "loss": 0.7568,
+      "step": 6442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7994392168899904e-05,
+      "loss": 0.7373,
+      "step": 6443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7993700543078044e-05,
+      "loss": 0.7568,
+      "step": 6444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799300881132092e-05,
+      "loss": 0.7617,
+      "step": 6445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7992316973637702e-05,
+      "loss": 0.7441,
+      "step": 6446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7991625030037558e-05,
+      "loss": 0.7822,
+      "step": 6447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990932980529655e-05,
+      "loss": 0.7686,
+      "step": 6448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990240825123165e-05,
+      "loss": 0.7441,
+      "step": 6449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7989548563827264e-05,
+      "loss": 0.7979,
+      "step": 6450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988856196651122e-05,
+      "loss": 0.7695,
+      "step": 6451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988163723603916e-05,
+      "loss": 0.7715,
+      "step": 6452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798747114469482e-05,
+      "loss": 0.7695,
+      "step": 6453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798677845993302e-05,
+      "loss": 0.7529,
+      "step": 6454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798608566932769e-05,
+      "loss": 0.7725,
+      "step": 6455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7985392772888013e-05,
+      "loss": 0.7598,
+      "step": 6456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7984699770623167e-05,
+      "loss": 0.7656,
+      "step": 6457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798400666254234e-05,
+      "loss": 0.7812,
+      "step": 6458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798331344865472e-05,
+      "loss": 0.7822,
+      "step": 6459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798262012896949e-05,
+      "loss": 0.7744,
+      "step": 6460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981926703495838e-05,
+      "loss": 0.7842,
+      "step": 6461
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7981233172242954e-05,
+      "loss": 0.7412,
+      "step": 6462
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7980539535220026e-05,
+      "loss": 0.7529,
+      "step": 6463
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7979845792436256e-05,
+      "loss": 0.7627,
+      "step": 6464
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7979151943900826e-05,
+      "loss": 0.7773,
+      "step": 6465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7978457989622938e-05,
+      "loss": 0.7676,
+      "step": 6466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7977763929611786e-05,
+      "loss": 0.7783,
+      "step": 6467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7977069763876568e-05,
+      "loss": 0.7676,
+      "step": 6468
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976375492426484e-05,
+      "loss": 0.7686,
+      "step": 6469
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7975681115270734e-05,
+      "loss": 0.75,
+      "step": 6470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7974986632418524e-05,
+      "loss": 0.7695,
+      "step": 6471
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7974292043879054e-05,
+      "loss": 0.748,
+      "step": 6472
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7973597349661528e-05,
+      "loss": 0.7344,
+      "step": 6473
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797290254977515e-05,
+      "loss": 0.7773,
+      "step": 6474
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797220764422914e-05,
+      "loss": 0.7617,
+      "step": 6475
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797151263303269e-05,
+      "loss": 0.7744,
+      "step": 6476
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797081751619502e-05,
+      "loss": 0.7178,
+      "step": 6477
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7970122293725344e-05,
+      "loss": 0.7539,
+      "step": 6478
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7969426965632873e-05,
+      "loss": 0.7588,
+      "step": 6479
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796873153192682e-05,
+      "loss": 0.7646,
+      "step": 6480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79680359926164e-05,
+      "loss": 0.7715,
+      "step": 6481
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7967340347710832e-05,
+      "loss": 0.7441,
+      "step": 6482
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796664459721934e-05,
+      "loss": 0.7734,
+      "step": 6483
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7965948741151137e-05,
+      "loss": 0.7754,
+      "step": 6484
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796525277951545e-05,
+      "loss": 0.7656,
+      "step": 6485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7964556712321495e-05,
+      "loss": 0.7559,
+      "step": 6486
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7963860539578503e-05,
+      "loss": 0.7959,
+      "step": 6487
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79631642612957e-05,
+      "loss": 0.7871,
+      "step": 6488
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7962467877482314e-05,
+      "loss": 0.7617,
+      "step": 6489
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7961771388147565e-05,
+      "loss": 0.7637,
+      "step": 6490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7961074793300693e-05,
+      "loss": 0.7812,
+      "step": 6491
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7960378092950924e-05,
+      "loss": 0.7549,
+      "step": 6492
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795968128710749e-05,
+      "loss": 0.7852,
+      "step": 6493
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7958984375779633e-05,
+      "loss": 0.7734,
+      "step": 6494
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7958287358976583e-05,
+      "loss": 0.7656,
+      "step": 6495
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7957590236707577e-05,
+      "loss": 0.7666,
+      "step": 6496
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7956893008981854e-05,
+      "loss": 0.7617,
+      "step": 6497
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7956195675808655e-05,
+      "loss": 0.7588,
+      "step": 6498
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795549823719722e-05,
+      "loss": 0.7539,
+      "step": 6499
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795480069315679e-05,
+      "loss": 0.7881,
+      "step": 6500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7954103043696617e-05,
+      "loss": 0.7559,
+      "step": 6501
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7953405288825936e-05,
+      "loss": 0.7686,
+      "step": 6502
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7952707428554e-05,
+      "loss": 0.7666,
+      "step": 6503
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7952009462890056e-05,
+      "loss": 0.7598,
+      "step": 6504
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7951311391843352e-05,
+      "loss": 0.7588,
+      "step": 6505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7950613215423145e-05,
+      "loss": 0.7656,
+      "step": 6506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7949914933638682e-05,
+      "loss": 0.7754,
+      "step": 6507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7949216546499215e-05,
+      "loss": 0.7578,
+      "step": 6508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7948518054014006e-05,
+      "loss": 0.7754,
+      "step": 6509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7947819456192304e-05,
+      "loss": 0.7559,
+      "step": 6510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7947120753043377e-05,
+      "loss": 0.751,
+      "step": 6511
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7946421944576474e-05,
+      "loss": 0.7549,
+      "step": 6512
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7945723030800862e-05,
+      "loss": 0.7529,
+      "step": 6513
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79450240117258e-05,
+      "loss": 0.7881,
+      "step": 6514
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7944324887360554e-05,
+      "loss": 0.7441,
+      "step": 6515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794362565771439e-05,
+      "loss": 0.7705,
+      "step": 6516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7942926322796567e-05,
+      "loss": 0.7871,
+      "step": 6517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7942226882616365e-05,
+      "loss": 0.7744,
+      "step": 6518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7941527337183045e-05,
+      "loss": 0.7637,
+      "step": 6519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794082768650588e-05,
+      "loss": 0.7754,
+      "step": 6520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794012793059414e-05,
+      "loss": 0.7559,
+      "step": 6521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7939428069457098e-05,
+      "loss": 0.7686,
+      "step": 6522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938728103104033e-05,
+      "loss": 0.7471,
+      "step": 6523
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7938028031544222e-05,
+      "loss": 0.79,
+      "step": 6524
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7937327854786933e-05,
+      "loss": 0.7637,
+      "step": 6525
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936627572841458e-05,
+      "loss": 0.7627,
+      "step": 6526
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7935927185717065e-05,
+      "loss": 0.7588,
+      "step": 6527
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7935226693423043e-05,
+      "loss": 0.7324,
+      "step": 6528
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7934526095968677e-05,
+      "loss": 0.7549,
+      "step": 6529
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7933825393363242e-05,
+      "loss": 0.7852,
+      "step": 6530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7933124585616034e-05,
+      "loss": 0.7871,
+      "step": 6531
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7932423672736336e-05,
+      "loss": 0.7627,
+      "step": 6532
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793172265473344e-05,
+      "loss": 0.7734,
+      "step": 6533
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793102153161663e-05,
+      "loss": 0.7695,
+      "step": 6534
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7930320303395202e-05,
+      "loss": 0.7627,
+      "step": 6535
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792961897007845e-05,
+      "loss": 0.7686,
+      "step": 6536
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928917531675663e-05,
+      "loss": 0.749,
+      "step": 6537
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792821598819614e-05,
+      "loss": 0.7842,
+      "step": 6538
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792751433964918e-05,
+      "loss": 0.7715,
+      "step": 6539
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7926812586044077e-05,
+      "loss": 0.7764,
+      "step": 6540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7926110727390135e-05,
+      "loss": 0.7607,
+      "step": 6541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7925408763696653e-05,
+      "loss": 0.7617,
+      "step": 6542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7924706694972936e-05,
+      "loss": 0.7539,
+      "step": 6543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7924004521228284e-05,
+      "loss": 0.7695,
+      "step": 6544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7923302242472008e-05,
+      "loss": 0.7646,
+      "step": 6545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7922599858713407e-05,
+      "loss": 0.7637,
+      "step": 6546
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7921897369961796e-05,
+      "loss": 0.7568,
+      "step": 6547
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7921194776226482e-05,
+      "loss": 0.7559,
+      "step": 6548
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7920492077516776e-05,
+      "loss": 0.7656,
+      "step": 6549
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919789273841995e-05,
+      "loss": 0.7461,
+      "step": 6550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919086365211443e-05,
+      "loss": 0.7422,
+      "step": 6551
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7918383351634447e-05,
+      "loss": 0.7832,
+      "step": 6552
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7917680233120314e-05,
+      "loss": 0.7539,
+      "step": 6553
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7916977009678363e-05,
+      "loss": 0.7852,
+      "step": 6554
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791627368131792e-05,
+      "loss": 0.7764,
+      "step": 6555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7915570248048296e-05,
+      "loss": 0.7686,
+      "step": 6556
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7914866709878825e-05,
+      "loss": 0.7812,
+      "step": 6557
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791416306681882e-05,
+      "loss": 0.8027,
+      "step": 6558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791345931887761e-05,
+      "loss": 0.7559,
+      "step": 6559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7912755466064523e-05,
+      "loss": 0.7422,
+      "step": 6560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7912051508388887e-05,
+      "loss": 0.7686,
+      "step": 6561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911347445860028e-05,
+      "loss": 0.7451,
+      "step": 6562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7910643278487277e-05,
+      "loss": 0.7891,
+      "step": 6563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7909939006279966e-05,
+      "loss": 0.7676,
+      "step": 6564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7909234629247428e-05,
+      "loss": 0.7686,
+      "step": 6565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7908530147398998e-05,
+      "loss": 0.7568,
+      "step": 6566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907825560744017e-05,
+      "loss": 0.7617,
+      "step": 6567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907120869291816e-05,
+      "loss": 0.7627,
+      "step": 6568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7906416073051732e-05,
+      "loss": 0.7646,
+      "step": 6569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7905711172033114e-05,
+      "loss": 0.7754,
+      "step": 6570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7905006166245295e-05,
+      "loss": 0.7646,
+      "step": 6571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790430105569762e-05,
+      "loss": 0.7598,
+      "step": 6572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790359584039944e-05,
+      "loss": 0.7754,
+      "step": 6573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7902890520360095e-05,
+      "loss": 0.7549,
+      "step": 6574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790218509558893e-05,
+      "loss": 0.748,
+      "step": 6575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7901479566095297e-05,
+      "loss": 0.7588,
+      "step": 6576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7900773931888548e-05,
+      "loss": 0.748,
+      "step": 6577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7900068192978026e-05,
+      "loss": 0.7695,
+      "step": 6578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899362349373094e-05,
+      "loss": 0.7686,
+      "step": 6579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78986564010831e-05,
+      "loss": 0.749,
+      "step": 6580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78979503481174e-05,
+      "loss": 0.7549,
+      "step": 6581
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789724419048535e-05,
+      "loss": 0.8008,
+      "step": 6582
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7896537928196313e-05,
+      "loss": 0.7529,
+      "step": 6583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895831561259644e-05,
+      "loss": 0.7607,
+      "step": 6584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895125089684705e-05,
+      "loss": 0.7676,
+      "step": 6585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789441851348086e-05,
+      "loss": 0.7617,
+      "step": 6586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789371183265747e-05,
+      "loss": 0.751,
+      "step": 6587
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7893005047223906e-05,
+      "loss": 0.7744,
+      "step": 6588
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7892298157189528e-05,
+      "loss": 0.7568,
+      "step": 6589
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891591162563704e-05,
+      "loss": 0.751,
+      "step": 6590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789088406335581e-05,
+      "loss": 0.7559,
+      "step": 6591
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789017685957521e-05,
+      "loss": 0.7578,
+      "step": 6592
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7889469551231275e-05,
+      "loss": 0.7754,
+      "step": 6593
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788876213833339e-05,
+      "loss": 0.7637,
+      "step": 6594
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7888054620890915e-05,
+      "loss": 0.7607,
+      "step": 6595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7887346998913236e-05,
+      "loss": 0.7627,
+      "step": 6596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788663927240973e-05,
+      "loss": 0.7588,
+      "step": 6597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788593144138977e-05,
+      "loss": 0.7705,
+      "step": 6598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885223505862742e-05,
+      "loss": 0.7588,
+      "step": 6599
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884515465838027e-05,
+      "loss": 0.7832,
+      "step": 6600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7883807321325004e-05,
+      "loss": 0.7617,
+      "step": 6601
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7883099072333062e-05,
+      "loss": 0.7666,
+      "step": 6602
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788239071887159e-05,
+      "loss": 0.7734,
+      "step": 6603
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7881682260949966e-05,
+      "loss": 0.7695,
+      "step": 6604
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788097369857759e-05,
+      "loss": 0.7891,
+      "step": 6605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7880265031763837e-05,
+      "loss": 0.7822,
+      "step": 6606
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7879556260518112e-05,
+      "loss": 0.7734,
+      "step": 6607
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7878847384849803e-05,
+      "loss": 0.7861,
+      "step": 6608
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787813840476831e-05,
+      "loss": 0.7539,
+      "step": 6609
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7877429320283017e-05,
+      "loss": 0.7715,
+      "step": 6610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787672013140333e-05,
+      "loss": 0.7529,
+      "step": 6611
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7876010838138644e-05,
+      "loss": 0.7666,
+      "step": 6612
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7875301440498357e-05,
+      "loss": 0.7734,
+      "step": 6613
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7874591938491877e-05,
+      "loss": 0.749,
+      "step": 6614
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7873882332128597e-05,
+      "loss": 0.7676,
+      "step": 6615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7873172621417933e-05,
+      "loss": 0.7803,
+      "step": 6616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787246280636928e-05,
+      "loss": 0.7822,
+      "step": 6617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7871752886992048e-05,
+      "loss": 0.7852,
+      "step": 6618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7871042863295645e-05,
+      "loss": 0.7686,
+      "step": 6619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7870332735289482e-05,
+      "loss": 0.7852,
+      "step": 6620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786962250298297e-05,
+      "loss": 0.7832,
+      "step": 6621
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7868912166385515e-05,
+      "loss": 0.7617,
+      "step": 6622
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786820172550654e-05,
+      "loss": 0.7607,
+      "step": 6623
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7867491180355453e-05,
+      "loss": 0.7715,
+      "step": 6624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7866780530941676e-05,
+      "loss": 0.7656,
+      "step": 6625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786606977727462e-05,
+      "loss": 0.7686,
+      "step": 6626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786535891936371e-05,
+      "loss": 0.752,
+      "step": 6627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864647957218368e-05,
+      "loss": 0.7676,
+      "step": 6628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786393689084801e-05,
+      "loss": 0.7646,
+      "step": 6629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786322572026206e-05,
+      "loss": 0.7617,
+      "step": 6630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786251444546994e-05,
+      "loss": 0.749,
+      "step": 6631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7861803066481086e-05,
+      "loss": 0.7666,
+      "step": 6632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786109158330492e-05,
+      "loss": 0.7578,
+      "step": 6633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786037999595087e-05,
+      "loss": 0.7744,
+      "step": 6634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7859668304428365e-05,
+      "loss": 0.7646,
+      "step": 6635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7858956508746842e-05,
+      "loss": 0.7812,
+      "step": 6636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7858244608915724e-05,
+      "loss": 0.7695,
+      "step": 6637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7857532604944457e-05,
+      "loss": 0.7598,
+      "step": 6638
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856820496842472e-05,
+      "loss": 0.7656,
+      "step": 6639
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856108284619203e-05,
+      "loss": 0.7656,
+      "step": 6640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785539596828409e-05,
+      "loss": 0.7793,
+      "step": 6641
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7854683547846576e-05,
+      "loss": 0.7871,
+      "step": 6642
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7853971023316096e-05,
+      "loss": 0.7314,
+      "step": 6643
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7853258394702103e-05,
+      "loss": 0.7725,
+      "step": 6644
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785254566201403e-05,
+      "loss": 0.7744,
+      "step": 6645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785183282526133e-05,
+      "loss": 0.7607,
+      "step": 6646
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7851119884453444e-05,
+      "loss": 0.7695,
+      "step": 6647
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7850406839599824e-05,
+      "loss": 0.7598,
+      "step": 6648
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7849693690709924e-05,
+      "loss": 0.7568,
+      "step": 6649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7848980437793183e-05,
+      "loss": 0.7627,
+      "step": 6650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7848267080859062e-05,
+      "loss": 0.752,
+      "step": 6651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7847553619917014e-05,
+      "loss": 0.7666,
+      "step": 6652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7846840054976486e-05,
+      "loss": 0.7764,
+      "step": 6653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7846126386046948e-05,
+      "loss": 0.7812,
+      "step": 6654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7845412613137846e-05,
+      "loss": 0.7812,
+      "step": 6655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7844698736258642e-05,
+      "loss": 0.7695,
+      "step": 6656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843984755418802e-05,
+      "loss": 0.7686,
+      "step": 6657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843270670627782e-05,
+      "loss": 0.7773,
+      "step": 6658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7842556481895052e-05,
+      "loss": 0.7695,
+      "step": 6659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7841842189230068e-05,
+      "loss": 0.7549,
+      "step": 6660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.78411277926423e-05,
+      "loss": 0.7832,
+      "step": 6661
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784041329214122e-05,
+      "loss": 0.7627,
+      "step": 6662
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839698687736288e-05,
+      "loss": 0.7773,
+      "step": 6663
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783898397943698e-05,
+      "loss": 0.7568,
+      "step": 6664
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7838269167252768e-05,
+      "loss": 0.752,
+      "step": 6665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783755425119312e-05,
+      "loss": 0.7842,
+      "step": 6666
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7836839231267515e-05,
+      "loss": 0.7637,
+      "step": 6667
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7836124107485427e-05,
+      "loss": 0.7695,
+      "step": 6668
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7835408879856332e-05,
+      "loss": 0.7559,
+      "step": 6669
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783469354838971e-05,
+      "loss": 0.7656,
+      "step": 6670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7833978113095042e-05,
+      "loss": 0.7627,
+      "step": 6671
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7833262573981807e-05,
+      "loss": 0.7754,
+      "step": 6672
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832546931059488e-05,
+      "loss": 0.7646,
+      "step": 6673
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783183118433757e-05,
+      "loss": 0.7656,
+      "step": 6674
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7831115333825534e-05,
+      "loss": 0.7656,
+      "step": 6675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830399379532873e-05,
+      "loss": 0.7822,
+      "step": 6676
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7829683321469075e-05,
+      "loss": 0.7539,
+      "step": 6677
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782896715964362e-05,
+      "loss": 0.7686,
+      "step": 6678
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7828250894066007e-05,
+      "loss": 0.7666,
+      "step": 6679
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782753452474573e-05,
+      "loss": 0.7822,
+      "step": 6680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7826818051692282e-05,
+      "loss": 0.7646,
+      "step": 6681
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7826101474915146e-05,
+      "loss": 0.749,
+      "step": 6682
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7825384794423836e-05,
+      "loss": 0.7666,
+      "step": 6683
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824668010227836e-05,
+      "loss": 0.7383,
+      "step": 6684
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7823951122336654e-05,
+      "loss": 0.7861,
+      "step": 6685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7823234130759785e-05,
+      "loss": 0.7559,
+      "step": 6686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782251703550673e-05,
+      "loss": 0.7744,
+      "step": 6687
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7821799836586998e-05,
+      "loss": 0.79,
+      "step": 6688
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782108253401009e-05,
+      "loss": 0.7598,
+      "step": 6689
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7820365127785512e-05,
+      "loss": 0.749,
+      "step": 6690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7819647617922773e-05,
+      "loss": 0.748,
+      "step": 6691
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781893000443138e-05,
+      "loss": 0.748,
+      "step": 6692
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781821228732084e-05,
+      "loss": 0.752,
+      "step": 6693
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781749446660067e-05,
+      "loss": 0.7471,
+      "step": 6694
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781677654228038e-05,
+      "loss": 0.7832,
+      "step": 6695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7816058514369484e-05,
+      "loss": 0.7559,
+      "step": 6696
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78153403828775e-05,
+      "loss": 0.791,
+      "step": 6697
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814622147813945e-05,
+      "loss": 0.7461,
+      "step": 6698
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7813903809188332e-05,
+      "loss": 0.7461,
+      "step": 6699
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7813185367010184e-05,
+      "loss": 0.7598,
+      "step": 6700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812466821289024e-05,
+      "loss": 0.7686,
+      "step": 6701
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7811748172034373e-05,
+      "loss": 0.7461,
+      "step": 6702
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781102941925575e-05,
+      "loss": 0.7852,
+      "step": 6703
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7810310562962693e-05,
+      "loss": 0.751,
+      "step": 6704
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809591603164713e-05,
+      "loss": 0.7441,
+      "step": 6705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7808872539871348e-05,
+      "loss": 0.791,
+      "step": 6706
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7808153373092123e-05,
+      "loss": 0.7832,
+      "step": 6707
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780743410283657e-05,
+      "loss": 0.7598,
+      "step": 6708
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7806714729114223e-05,
+      "loss": 0.7705,
+      "step": 6709
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7805995251934614e-05,
+      "loss": 0.7588,
+      "step": 6710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780527567130727e-05,
+      "loss": 0.7607,
+      "step": 6711
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804555987241744e-05,
+      "loss": 0.7754,
+      "step": 6712
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7803836199747556e-05,
+      "loss": 0.7539,
+      "step": 6713
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780311630883426e-05,
+      "loss": 0.7773,
+      "step": 6714
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780239631451138e-05,
+      "loss": 0.8018,
+      "step": 6715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780167621678847e-05,
+      "loss": 0.7666,
+      "step": 6716
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7800956015675074e-05,
+      "loss": 0.7617,
+      "step": 6717
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7800235711180726e-05,
+      "loss": 0.7773,
+      "step": 6718
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779951530331498e-05,
+      "loss": 0.7393,
+      "step": 6719
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779879479208738e-05,
+      "loss": 0.7568,
+      "step": 6720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7798074177507477e-05,
+      "loss": 0.7607,
+      "step": 6721
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7797353459584815e-05,
+      "loss": 0.8027,
+      "step": 6722
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7796632638328955e-05,
+      "loss": 0.7695,
+      "step": 6723
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779591171374944e-05,
+      "loss": 0.7559,
+      "step": 6724
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7795190685855826e-05,
+      "loss": 0.7646,
+      "step": 6725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779446955465767e-05,
+      "loss": 0.7539,
+      "step": 6726
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7793748320164532e-05,
+      "loss": 0.7705,
+      "step": 6727
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7793026982385965e-05,
+      "loss": 0.7588,
+      "step": 6728
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779230554133153e-05,
+      "loss": 0.7578,
+      "step": 6729
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791583997010787e-05,
+      "loss": 0.7578,
+      "step": 6730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77908623494333e-05,
+      "loss": 0.7549,
+      "step": 6731
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7790140598608633e-05,
+      "loss": 0.7656,
+      "step": 6732
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7789418744546347e-05,
+      "loss": 0.7471,
+      "step": 6733
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778869678725601e-05,
+      "loss": 0.7471,
+      "step": 6734
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778797472674719e-05,
+      "loss": 0.7686,
+      "step": 6735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7787252563029458e-05,
+      "loss": 0.75,
+      "step": 6736
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7786530296112383e-05,
+      "loss": 0.7754,
+      "step": 6737
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7785807926005536e-05,
+      "loss": 0.7559,
+      "step": 6738
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7785085452718493e-05,
+      "loss": 0.7705,
+      "step": 6739
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7784362876260823e-05,
+      "loss": 0.7549,
+      "step": 6740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7783640196642103e-05,
+      "loss": 0.7773,
+      "step": 6741
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7782917413871918e-05,
+      "loss": 0.7637,
+      "step": 6742
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7782194527959835e-05,
+      "loss": 0.7383,
+      "step": 6743
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778147153891544e-05,
+      "loss": 0.7734,
+      "step": 6744
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7780748446748316e-05,
+      "loss": 0.7559,
+      "step": 6745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7780025251468044e-05,
+      "loss": 0.7881,
+      "step": 6746
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779301953084204e-05,
+      "loss": 0.7744,
+      "step": 6747
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7778578551606384e-05,
+      "loss": 0.7588,
+      "step": 6748
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777855047044174e-05,
+      "loss": 0.7676,
+      "step": 6749
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777131439407162e-05,
+      "loss": 0.7715,
+      "step": 6750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7776407728704935e-05,
+      "loss": 0.7607,
+      "step": 6751
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777568391494708e-05,
+      "loss": 0.7637,
+      "step": 6752
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774959998143194e-05,
+      "loss": 0.7393,
+      "step": 6753
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774235978302873e-05,
+      "loss": 0.7568,
+      "step": 6754
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777351185543571e-05,
+      "loss": 0.7578,
+      "step": 6755
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7772787629551297e-05,
+      "loss": 0.7559,
+      "step": 6756
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7772063300659235e-05,
+      "loss": 0.7529,
+      "step": 6757
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7771338868769125e-05,
+      "loss": 0.7422,
+      "step": 6758
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7770614333890563e-05,
+      "loss": 0.793,
+      "step": 6759
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7769889696033154e-05,
+      "loss": 0.7451,
+      "step": 6760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77691649552065e-05,
+      "loss": 0.7686,
+      "step": 6761
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7768440111420208e-05,
+      "loss": 0.8008,
+      "step": 6762
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776771516468388e-05,
+      "loss": 0.7451,
+      "step": 6763
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766990115007122e-05,
+      "loss": 0.7686,
+      "step": 6764
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776626496239955e-05,
+      "loss": 0.7373,
+      "step": 6765
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7765539706870768e-05,
+      "loss": 0.751,
+      "step": 6766
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776481434843039e-05,
+      "loss": 0.7656,
+      "step": 6767
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7764088887088026e-05,
+      "loss": 0.7373,
+      "step": 6768
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776336332285329e-05,
+      "loss": 0.7646,
+      "step": 6769
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762637655735804e-05,
+      "loss": 0.7695,
+      "step": 6770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776191188574518e-05,
+      "loss": 0.7656,
+      "step": 6771
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776118601289103e-05,
+      "loss": 0.7568,
+      "step": 6772
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7760460037182982e-05,
+      "loss": 0.75,
+      "step": 6773
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7759733958630655e-05,
+      "loss": 0.7588,
+      "step": 6774
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7759007777243673e-05,
+      "loss": 0.7539,
+      "step": 6775
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775828149303165e-05,
+      "loss": 0.7793,
+      "step": 6776
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7757555106004226e-05,
+      "loss": 0.7754,
+      "step": 6777
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7756828616171016e-05,
+      "loss": 0.7705,
+      "step": 6778
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7756102023541654e-05,
+      "loss": 0.7646,
+      "step": 6779
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7755375328125762e-05,
+      "loss": 0.7861,
+      "step": 6780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775464852993298e-05,
+      "loss": 0.7812,
+      "step": 6781
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775392162897293e-05,
+      "loss": 0.7666,
+      "step": 6782
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775319462525525e-05,
+      "loss": 0.7812,
+      "step": 6783
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7752467518789577e-05,
+      "loss": 0.7568,
+      "step": 6784
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751740309585542e-05,
+      "loss": 0.7803,
+      "step": 6785
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751012997652782e-05,
+      "loss": 0.7812,
+      "step": 6786
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775028558300094e-05,
+      "loss": 0.7793,
+      "step": 6787
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7749558065639655e-05,
+      "loss": 0.793,
+      "step": 6788
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7748830445578565e-05,
+      "loss": 0.7764,
+      "step": 6789
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7748102722827315e-05,
+      "loss": 0.7695,
+      "step": 6790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7747374897395552e-05,
+      "loss": 0.7773,
+      "step": 6791
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7746646969292917e-05,
+      "loss": 0.7422,
+      "step": 6792
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745918938529058e-05,
+      "loss": 0.7676,
+      "step": 6793
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774519080511362e-05,
+      "loss": 0.75,
+      "step": 6794
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7744462569056255e-05,
+      "loss": 0.7646,
+      "step": 6795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774373423036662e-05,
+      "loss": 0.7363,
+      "step": 6796
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7743005789054355e-05,
+      "loss": 0.7822,
+      "step": 6797
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7742277245129124e-05,
+      "loss": 0.7637,
+      "step": 6798
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7741548598600577e-05,
+      "loss": 0.7773,
+      "step": 6799
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774081984947837e-05,
+      "loss": 0.7783,
+      "step": 6800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7740090997772163e-05,
+      "loss": 0.7617,
+      "step": 6801
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7739362043491616e-05,
+      "loss": 0.7568,
+      "step": 6802
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7738632986646383e-05,
+      "loss": 0.7607,
+      "step": 6803
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7737903827246136e-05,
+      "loss": 0.751,
+      "step": 6804
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7737174565300525e-05,
+      "loss": 0.7568,
+      "step": 6805
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7736445200819225e-05,
+      "loss": 0.7451,
+      "step": 6806
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735715733811898e-05,
+      "loss": 0.7803,
+      "step": 6807
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7734986164288214e-05,
+      "loss": 0.7559,
+      "step": 6808
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7734256492257836e-05,
+      "loss": 0.752,
+      "step": 6809
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7733526717730437e-05,
+      "loss": 0.7451,
+      "step": 6810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773279684071569e-05,
+      "loss": 0.7646,
+      "step": 6811
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7732066861223263e-05,
+      "loss": 0.7559,
+      "step": 6812
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7731336779262833e-05,
+      "loss": 0.7676,
+      "step": 6813
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7730606594844072e-05,
+      "loss": 0.7461,
+      "step": 6814
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7729876307976662e-05,
+      "loss": 0.7578,
+      "step": 6815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772914591867028e-05,
+      "loss": 0.7666,
+      "step": 6816
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7728415426934603e-05,
+      "loss": 0.7764,
+      "step": 6817
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7727684832779313e-05,
+      "loss": 0.7607,
+      "step": 6818
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772695413621409e-05,
+      "loss": 0.7656,
+      "step": 6819
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726223337248624e-05,
+      "loss": 0.7646,
+      "step": 6820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7725492435892587e-05,
+      "loss": 0.7549,
+      "step": 6821
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7724761432155678e-05,
+      "loss": 0.7734,
+      "step": 6822
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772403032604758e-05,
+      "loss": 0.7422,
+      "step": 6823
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772329911757798e-05,
+      "loss": 0.749,
+      "step": 6824
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7722567806756566e-05,
+      "loss": 0.7969,
+      "step": 6825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7721836393593038e-05,
+      "loss": 0.7861,
+      "step": 6826
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7721104878097084e-05,
+      "loss": 0.7744,
+      "step": 6827
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.77203732602784e-05,
+      "loss": 0.75,
+      "step": 6828
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7719641540146677e-05,
+      "loss": 0.7471,
+      "step": 6829
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718909717711615e-05,
+      "loss": 0.7666,
+      "step": 6830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718177792982914e-05,
+      "loss": 0.7832,
+      "step": 6831
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7717445765970272e-05,
+      "loss": 0.7852,
+      "step": 6832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771671363668339e-05,
+      "loss": 0.7656,
+      "step": 6833
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7715981405131968e-05,
+      "loss": 0.7754,
+      "step": 6834
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7715249071325717e-05,
+      "loss": 0.7666,
+      "step": 6835
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7714516635274336e-05,
+      "loss": 0.749,
+      "step": 6836
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771378409698753e-05,
+      "loss": 0.7959,
+      "step": 6837
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7713051456475018e-05,
+      "loss": 0.7861,
+      "step": 6838
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771231871374649e-05,
+      "loss": 0.8018,
+      "step": 6839
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7711585868811676e-05,
+      "loss": 0.7529,
+      "step": 6840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771085292168028e-05,
+      "loss": 0.7539,
+      "step": 6841
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771011987236201e-05,
+      "loss": 0.7529,
+      "step": 6842
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770938672086659e-05,
+      "loss": 0.7578,
+      "step": 6843
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7708653467203726e-05,
+      "loss": 0.7529,
+      "step": 6844
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7707920111383147e-05,
+      "loss": 0.7832,
+      "step": 6845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7707186653414562e-05,
+      "loss": 0.7871,
+      "step": 6846
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7706453093307695e-05,
+      "loss": 0.7988,
+      "step": 6847
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7705719431072266e-05,
+      "loss": 0.7441,
+      "step": 6848
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7704985666718003e-05,
+      "loss": 0.7979,
+      "step": 6849
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7704251800254623e-05,
+      "loss": 0.7617,
+      "step": 6850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770351783169185e-05,
+      "loss": 0.7783,
+      "step": 6851
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770278376103942e-05,
+      "loss": 0.7861,
+      "step": 6852
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7702049588307054e-05,
+      "loss": 0.7617,
+      "step": 6853
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7701315313504485e-05,
+      "loss": 0.791,
+      "step": 6854
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7700580936641444e-05,
+      "loss": 0.7539,
+      "step": 6855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7699846457727662e-05,
+      "loss": 0.7783,
+      "step": 6856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769911187677287e-05,
+      "loss": 0.8018,
+      "step": 6857
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7698377193786803e-05,
+      "loss": 0.7744,
+      "step": 6858
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7697642408779205e-05,
+      "loss": 0.7764,
+      "step": 6859
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696907521759806e-05,
+      "loss": 0.7432,
+      "step": 6860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696172532738345e-05,
+      "loss": 0.7393,
+      "step": 6861
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7695437441724566e-05,
+      "loss": 0.7617,
+      "step": 6862
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769470224872821e-05,
+      "loss": 0.7568,
+      "step": 6863
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7693966953759018e-05,
+      "loss": 0.7529,
+      "step": 6864
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7693231556826734e-05,
+      "loss": 0.7871,
+      "step": 6865
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769249605794111e-05,
+      "loss": 0.7432,
+      "step": 6866
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7691760457111884e-05,
+      "loss": 0.7861,
+      "step": 6867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7691024754348813e-05,
+      "loss": 0.7666,
+      "step": 6868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7690288949661637e-05,
+      "loss": 0.7852,
+      "step": 6869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7689553043060118e-05,
+      "loss": 0.7578,
+      "step": 6870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7688817034554e-05,
+      "loss": 0.7666,
+      "step": 6871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768808092415304e-05,
+      "loss": 0.7676,
+      "step": 6872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7687344711866993e-05,
+      "loss": 0.7549,
+      "step": 6873
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768660839770562e-05,
+      "loss": 0.7451,
+      "step": 6874
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7685871981678674e-05,
+      "loss": 0.7842,
+      "step": 6875
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768513546379591e-05,
+      "loss": 0.7559,
+      "step": 6876
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7684398844067096e-05,
+      "loss": 0.7725,
+      "step": 6877
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7683662122501992e-05,
+      "loss": 0.7725,
+      "step": 6878
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768292529911036e-05,
+      "loss": 0.7529,
+      "step": 6879
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7682188373901965e-05,
+      "loss": 0.7559,
+      "step": 6880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7681451346886574e-05,
+      "loss": 0.7539,
+      "step": 6881
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7680714218073955e-05,
+      "loss": 0.7842,
+      "step": 6882
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679976987473874e-05,
+      "loss": 0.8008,
+      "step": 6883
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679239655096103e-05,
+      "loss": 0.7715,
+      "step": 6884
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7678502220950414e-05,
+      "loss": 0.7764,
+      "step": 6885
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7677764685046573e-05,
+      "loss": 0.7705,
+      "step": 6886
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7677027047394367e-05,
+      "loss": 0.7559,
+      "step": 6887
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676289308003557e-05,
+      "loss": 0.7842,
+      "step": 6888
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767555146688393e-05,
+      "loss": 0.7451,
+      "step": 6889
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7674813524045262e-05,
+      "loss": 0.752,
+      "step": 6890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767407547949733e-05,
+      "loss": 0.7471,
+      "step": 6891
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7673337333249918e-05,
+      "loss": 0.7861,
+      "step": 6892
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7672599085312804e-05,
+      "loss": 0.7676,
+      "step": 6893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7671860735695775e-05,
+      "loss": 0.7559,
+      "step": 6894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7671122284408616e-05,
+      "loss": 0.8047,
+      "step": 6895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.767038373146111e-05,
+      "loss": 0.7812,
+      "step": 6896
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7669645076863047e-05,
+      "loss": 0.7432,
+      "step": 6897
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7668906320624216e-05,
+      "loss": 0.7646,
+      "step": 6898
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7668167462754405e-05,
+      "loss": 0.7568,
+      "step": 6899
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766742850326341e-05,
+      "loss": 0.7451,
+      "step": 6900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766668944216102e-05,
+      "loss": 0.7812,
+      "step": 6901
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766595027945703e-05,
+      "loss": 0.7295,
+      "step": 6902
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7665211015161234e-05,
+      "loss": 0.7617,
+      "step": 6903
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664471649283433e-05,
+      "loss": 0.7598,
+      "step": 6904
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7663732181833423e-05,
+      "loss": 0.7588,
+      "step": 6905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662992612821007e-05,
+      "loss": 0.7666,
+      "step": 6906
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662252942255982e-05,
+      "loss": 0.7451,
+      "step": 6907
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766151317014815e-05,
+      "loss": 0.7832,
+      "step": 6908
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7660773296507317e-05,
+      "loss": 0.7461,
+      "step": 6909
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7660033321343288e-05,
+      "loss": 0.7363,
+      "step": 6910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7659293244665866e-05,
+      "loss": 0.7363,
+      "step": 6911
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7658553066484865e-05,
+      "loss": 0.7842,
+      "step": 6912
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7657812786810086e-05,
+      "loss": 0.7578,
+      "step": 6913
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765707240565135e-05,
+      "loss": 0.7617,
+      "step": 6914
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765633192301846e-05,
+      "loss": 0.7676,
+      "step": 6915
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765559133892123e-05,
+      "loss": 0.7793,
+      "step": 6916
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7654850653369478e-05,
+      "loss": 0.748,
+      "step": 6917
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765410986637302e-05,
+      "loss": 0.7861,
+      "step": 6918
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653368977941668e-05,
+      "loss": 0.7686,
+      "step": 6919
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652627988085246e-05,
+      "loss": 0.7236,
+      "step": 6920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765188689681357e-05,
+      "loss": 0.7881,
+      "step": 6921
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7651145704136464e-05,
+      "loss": 0.7695,
+      "step": 6922
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765040441006375e-05,
+      "loss": 0.7471,
+      "step": 6923
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764966301460525e-05,
+      "loss": 0.7812,
+      "step": 6924
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764892151777079e-05,
+      "loss": 0.749,
+      "step": 6925
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76481799195702e-05,
+      "loss": 0.7764,
+      "step": 6926
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7647438220013305e-05,
+      "loss": 0.7891,
+      "step": 6927
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764669641910993e-05,
+      "loss": 0.7598,
+      "step": 6928
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645954516869914e-05,
+      "loss": 0.7598,
+      "step": 6929
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645212513303085e-05,
+      "loss": 0.7559,
+      "step": 6930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7644470408419275e-05,
+      "loss": 0.7441,
+      "step": 6931
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764372820222832e-05,
+      "loss": 0.7734,
+      "step": 6932
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642985894740055e-05,
+      "loss": 0.7725,
+      "step": 6933
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642243485964316e-05,
+      "loss": 0.7822,
+      "step": 6934
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7641500975910945e-05,
+      "loss": 0.7656,
+      "step": 6935
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764075836458978e-05,
+      "loss": 0.7744,
+      "step": 6936
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7640015652010667e-05,
+      "loss": 0.75,
+      "step": 6937
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7639272838183444e-05,
+      "loss": 0.752,
+      "step": 6938
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7638529923117953e-05,
+      "loss": 0.7627,
+      "step": 6939
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763778690682404e-05,
+      "loss": 0.7695,
+      "step": 6940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763704378931156e-05,
+      "loss": 0.7734,
+      "step": 6941
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763630057059035e-05,
+      "loss": 0.752,
+      "step": 6942
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7635557250670266e-05,
+      "loss": 0.7598,
+      "step": 6943
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763481382956116e-05,
+      "loss": 0.7422,
+      "step": 6944
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763407030727288e-05,
+      "loss": 0.7598,
+      "step": 6945
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7633326683815276e-05,
+      "loss": 0.7383,
+      "step": 6946
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763258295919821e-05,
+      "loss": 0.7402,
+      "step": 6947
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7631839133431534e-05,
+      "loss": 0.7939,
+      "step": 6948
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763109520652511e-05,
+      "loss": 0.7559,
+      "step": 6949
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.763035117848879e-05,
+      "loss": 0.7422,
+      "step": 6950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762960704933244e-05,
+      "loss": 0.7744,
+      "step": 6951
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762886281906592e-05,
+      "loss": 0.791,
+      "step": 6952
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7628118487699088e-05,
+      "loss": 0.7666,
+      "step": 6953
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7627374055241814e-05,
+      "loss": 0.7666,
+      "step": 6954
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762662952170396e-05,
+      "loss": 0.7627,
+      "step": 6955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76258848870954e-05,
+      "loss": 0.7695,
+      "step": 6956
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7625140151425992e-05,
+      "loss": 0.7773,
+      "step": 6957
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7624395314705608e-05,
+      "loss": 0.7607,
+      "step": 6958
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7623650376944127e-05,
+      "loss": 0.7529,
+      "step": 6959
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762290533815141e-05,
+      "loss": 0.7832,
+      "step": 6960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622160198337338e-05,
+      "loss": 0.752,
+      "step": 6961
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762141495751178e-05,
+      "loss": 0.7598,
+      "step": 6962
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7620669615684618e-05,
+      "loss": 0.7598,
+      "step": 6963
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761992417286573e-05,
+      "loss": 0.7832,
+      "step": 6964
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619178629064982e-05,
+      "loss": 0.752,
+      "step": 6965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7618432984292273e-05,
+      "loss": 0.7393,
+      "step": 6966
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7617687238557475e-05,
+      "loss": 0.7559,
+      "step": 6967
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761694139187047e-05,
+      "loss": 0.7676,
+      "step": 6968
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7616195444241143e-05,
+      "loss": 0.791,
+      "step": 6969
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761544939567938e-05,
+      "loss": 0.7793,
+      "step": 6970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7614703246195067e-05,
+      "loss": 0.7637,
+      "step": 6971
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613956995798097e-05,
+      "loss": 0.7656,
+      "step": 6972
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613210644498354e-05,
+      "loss": 0.7637,
+      "step": 6973
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761246419230573e-05,
+      "loss": 0.7432,
+      "step": 6974
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761171763923012e-05,
+      "loss": 0.7646,
+      "step": 6975
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7610970985281415e-05,
+      "loss": 0.7559,
+      "step": 6976
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761022423046951e-05,
+      "loss": 0.7812,
+      "step": 6977
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609477374804305e-05,
+      "loss": 0.7656,
+      "step": 6978
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760873041829569e-05,
+      "loss": 0.7588,
+      "step": 6979
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760798336095357e-05,
+      "loss": 0.7744,
+      "step": 6980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607236202787844e-05,
+      "loss": 0.7852,
+      "step": 6981
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7606488943808412e-05,
+      "loss": 0.7227,
+      "step": 6982
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760574158402518e-05,
+      "loss": 0.7744,
+      "step": 6983
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760499412344805e-05,
+      "loss": 0.7656,
+      "step": 6984
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760424656208693e-05,
+      "loss": 0.75,
+      "step": 6985
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760349889995172e-05,
+      "loss": 0.7773,
+      "step": 6986
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760275113705234e-05,
+      "loss": 0.7539,
+      "step": 6987
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7602003273398688e-05,
+      "loss": 0.7852,
+      "step": 6988
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760125530900068e-05,
+      "loss": 0.7578,
+      "step": 6989
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7600507243868227e-05,
+      "loss": 0.7549,
+      "step": 6990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7599759078011248e-05,
+      "loss": 0.7676,
+      "step": 6991
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759901081143965e-05,
+      "loss": 0.7666,
+      "step": 6992
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598262444163356e-05,
+      "loss": 0.7588,
+      "step": 6993
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7597513976192276e-05,
+      "loss": 0.7686,
+      "step": 6994
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7596765407536338e-05,
+      "loss": 0.7842,
+      "step": 6995
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7596016738205456e-05,
+      "loss": 0.7646,
+      "step": 6996
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7595267968209552e-05,
+      "loss": 0.7695,
+      "step": 6997
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7594519097558552e-05,
+      "loss": 0.7871,
+      "step": 6998
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759377012626238e-05,
+      "loss": 0.7617,
+      "step": 6999
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7593021054330956e-05,
+      "loss": 0.7578,
+      "step": 7000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7592271881774215e-05,
+      "loss": 0.7998,
+      "step": 7001
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759152260860208e-05,
+      "loss": 0.7637,
+      "step": 7002
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7590773234824486e-05,
+      "loss": 0.7695,
+      "step": 7003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7590023760451356e-05,
+      "loss": 0.7402,
+      "step": 7004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.758927418549263e-05,
+      "loss": 0.7549,
+      "step": 7005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758852450995823e-05,
+      "loss": 0.7549,
+      "step": 7006
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587774733858104e-05,
+      "loss": 0.7617,
+      "step": 7007
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7587024857202186e-05,
+      "loss": 0.7793,
+      "step": 7008
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7586274880000408e-05,
+      "loss": 0.7812,
+      "step": 7009
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758552480226271e-05,
+      "loss": 0.7451,
+      "step": 7010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758477462399904e-05,
+      "loss": 0.7598,
+      "step": 7011
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7584024345219332e-05,
+      "loss": 0.7812,
+      "step": 7012
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583273965933528e-05,
+      "loss": 0.7559,
+      "step": 7013
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582523486151575e-05,
+      "loss": 0.7676,
+      "step": 7014
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7581772905883423e-05,
+      "loss": 0.75,
+      "step": 7015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758102222513901e-05,
+      "loss": 0.7461,
+      "step": 7016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758027144392829e-05,
+      "loss": 0.7783,
+      "step": 7017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757952056226121e-05,
+      "loss": 0.7676,
+      "step": 7018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578769580147726e-05,
+      "loss": 0.7695,
+      "step": 7019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578018497597785e-05,
+      "loss": 0.7734,
+      "step": 7020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757726731462134e-05,
+      "loss": 0.7656,
+      "step": 7021
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7576516031228354e-05,
+      "loss": 0.75,
+      "step": 7022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757576464742877e-05,
+      "loss": 0.7617,
+      "step": 7023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757501316323256e-05,
+      "loss": 0.7568,
+      "step": 7024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7574261578649672e-05,
+      "loss": 0.7832,
+      "step": 7025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757350989369007e-05,
+      "loss": 0.7422,
+      "step": 7026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572758108363717e-05,
+      "loss": 0.7646,
+      "step": 7027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572006222680578e-05,
+      "loss": 0.7422,
+      "step": 7028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7571254236650607e-05,
+      "loss": 0.7725,
+      "step": 7029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7570502150283783e-05,
+      "loss": 0.7949,
+      "step": 7030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756974996359006e-05,
+      "loss": 0.7871,
+      "step": 7031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756899767657942e-05,
+      "loss": 0.7588,
+      "step": 7032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756824528926182e-05,
+      "loss": 0.7666,
+      "step": 7033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756749280164724e-05,
+      "loss": 0.7588,
+      "step": 7034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756674021374565e-05,
+      "loss": 0.7842,
+      "step": 7035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7565987525567015e-05,
+      "loss": 0.7725,
+      "step": 7036
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756523473712132e-05,
+      "loss": 0.748,
+      "step": 7037
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756448184841854e-05,
+      "loss": 0.7402,
+      "step": 7038
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7563728859468647e-05,
+      "loss": 0.7617,
+      "step": 7039
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7562975770281628e-05,
+      "loss": 0.7529,
+      "step": 7040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7562222580867456e-05,
+      "loss": 0.7471,
+      "step": 7041
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7561469291236114e-05,
+      "loss": 0.7773,
+      "step": 7042
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756071590139759e-05,
+      "loss": 0.7646,
+      "step": 7043
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755996241136186e-05,
+      "loss": 0.7568,
+      "step": 7044
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7559208821138914e-05,
+      "loss": 0.7676,
+      "step": 7045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755845513073874e-05,
+      "loss": 0.7432,
+      "step": 7046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7557701340171327e-05,
+      "loss": 0.7646,
+      "step": 7047
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7556947449446663e-05,
+      "loss": 0.7695,
+      "step": 7048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7556193458574734e-05,
+      "loss": 0.7705,
+      "step": 7049
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7555439367565537e-05,
+      "loss": 0.7568,
+      "step": 7050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7554685176429064e-05,
+      "loss": 0.7529,
+      "step": 7051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755393088517531e-05,
+      "loss": 0.748,
+      "step": 7052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7553176493814273e-05,
+      "loss": 0.748,
+      "step": 7053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7552422002355953e-05,
+      "loss": 0.7559,
+      "step": 7054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755166741081034e-05,
+      "loss": 0.7529,
+      "step": 7055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755091271918744e-05,
+      "loss": 0.7812,
+      "step": 7056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755015792749725e-05,
+      "loss": 0.75,
+      "step": 7057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754940303574978e-05,
+      "loss": 0.7529,
+      "step": 7058
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754864804395503e-05,
+      "loss": 0.7715,
+      "step": 7059
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7547892952123007e-05,
+      "loss": 0.7461,
+      "step": 7060
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7547137760263713e-05,
+      "loss": 0.7568,
+      "step": 7061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7546382468387162e-05,
+      "loss": 0.7559,
+      "step": 7062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754562707650336e-05,
+      "loss": 0.7549,
+      "step": 7063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7544871584622316e-05,
+      "loss": 0.7676,
+      "step": 7064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7544115992754047e-05,
+      "loss": 0.7588,
+      "step": 7065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7543360300908566e-05,
+      "loss": 0.7607,
+      "step": 7066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7542604509095883e-05,
+      "loss": 0.791,
+      "step": 7067
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7541848617326014e-05,
+      "loss": 0.749,
+      "step": 7068
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7541092625608986e-05,
+      "loss": 0.75,
+      "step": 7069
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75403365339548e-05,
+      "loss": 0.7539,
+      "step": 7070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7539580342373498e-05,
+      "loss": 0.7725,
+      "step": 7071
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7538824050875084e-05,
+      "loss": 0.7646,
+      "step": 7072
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753806765946959e-05,
+      "loss": 0.7705,
+      "step": 7073
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7537311168167034e-05,
+      "loss": 0.7598,
+      "step": 7074
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7536554576977442e-05,
+      "loss": 0.748,
+      "step": 7075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535797885910842e-05,
+      "loss": 0.7715,
+      "step": 7076
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753504109497727e-05,
+      "loss": 0.79,
+      "step": 7077
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7534284204186737e-05,
+      "loss": 0.7422,
+      "step": 7078
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7533527213549293e-05,
+      "loss": 0.7529,
+      "step": 7079
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532770123074955e-05,
+      "loss": 0.7373,
+      "step": 7080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532012932773766e-05,
+      "loss": 0.7627,
+      "step": 7081
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7531255642655755e-05,
+      "loss": 0.7637,
+      "step": 7082
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753049825273096e-05,
+      "loss": 0.7637,
+      "step": 7083
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752974076300942e-05,
+      "loss": 0.751,
+      "step": 7084
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7528983173501168e-05,
+      "loss": 0.7607,
+      "step": 7085
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7528225484216248e-05,
+      "loss": 0.7734,
+      "step": 7086
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75274676951647e-05,
+      "loss": 0.8154,
+      "step": 7087
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752670980635657e-05,
+      "loss": 0.7783,
+      "step": 7088
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7525951817801895e-05,
+      "loss": 0.7773,
+      "step": 7089
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7525193729510728e-05,
+      "loss": 0.75,
+      "step": 7090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7524435541493107e-05,
+      "loss": 0.7432,
+      "step": 7091
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7523677253759087e-05,
+      "loss": 0.7695,
+      "step": 7092
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7522918866318712e-05,
+      "loss": 0.7949,
+      "step": 7093
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7522160379182038e-05,
+      "loss": 0.7441,
+      "step": 7094
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752140179235911e-05,
+      "loss": 0.7598,
+      "step": 7095
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7520643105859983e-05,
+      "loss": 0.7764,
+      "step": 7096
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519884319694715e-05,
+      "loss": 0.749,
+      "step": 7097
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7519125433873357e-05,
+      "loss": 0.7617,
+      "step": 7098
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751836644840597e-05,
+      "loss": 0.7676,
+      "step": 7099
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7517607363302615e-05,
+      "loss": 0.7627,
+      "step": 7100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751684817857334e-05,
+      "loss": 0.7705,
+      "step": 7101
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7516088894228217e-05,
+      "loss": 0.7617,
+      "step": 7102
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7515329510277307e-05,
+      "loss": 0.7666,
+      "step": 7103
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7514570026730667e-05,
+      "loss": 0.7598,
+      "step": 7104
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7513810443598367e-05,
+      "loss": 0.7588,
+      "step": 7105
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7513050760890476e-05,
+      "loss": 0.7656,
+      "step": 7106
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7512290978617053e-05,
+      "loss": 0.7695,
+      "step": 7107
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7511531096788176e-05,
+      "loss": 0.751,
+      "step": 7108
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751077111541391e-05,
+      "loss": 0.7637,
+      "step": 7109
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7510011034504324e-05,
+      "loss": 0.7598,
+      "step": 7110
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7509250854069496e-05,
+      "loss": 0.752,
+      "step": 7111
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75084905741195e-05,
+      "loss": 0.7607,
+      "step": 7112
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750773019466441e-05,
+      "loss": 0.7432,
+      "step": 7113
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7506969715714304e-05,
+      "loss": 0.7773,
+      "step": 7114
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750620913727926e-05,
+      "loss": 0.7598,
+      "step": 7115
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7505448459369357e-05,
+      "loss": 0.7627,
+      "step": 7116
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7504687681994672e-05,
+      "loss": 0.7871,
+      "step": 7117
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7503926805165293e-05,
+      "loss": 0.7812,
+      "step": 7118
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.75031658288913e-05,
+      "loss": 0.75,
+      "step": 7119
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7502404753182775e-05,
+      "loss": 0.7676,
+      "step": 7120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.750164357804981e-05,
+      "loss": 0.791,
+      "step": 7121
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500882303502494e-05,
+      "loss": 0.7324,
+      "step": 7122
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500120929550907e-05,
+      "loss": 0.7461,
+      "step": 7123
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7499359456205147e-05,
+      "loss": 0.7822,
+      "step": 7124
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.74985978834753e-05,
+      "loss": 0.7549,
+      "step": 7125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749783621137146e-05,
+      "loss": 0.7334,
+      "step": 7126
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749707443990372e-05,
+      "loss": 0.7705,
+      "step": 7127
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7496312569082182e-05,
+      "loss": 0.7549,
+      "step": 7128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749555059891693e-05,
+      "loss": 0.7207,
+      "step": 7129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7494788529418073e-05,
+      "loss": 0.7559,
+      "step": 7130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749402636059571e-05,
+      "loss": 0.7471,
+      "step": 7131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7493264092459933e-05,
+      "loss": 0.7441,
+      "step": 7132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749250172502085e-05,
+      "loss": 0.7451,
+      "step": 7133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7491739258288563e-05,
+      "loss": 0.751,
+      "step": 7134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7490976692273177e-05,
+      "loss": 0.7744,
+      "step": 7135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7490214026984795e-05,
+      "loss": 0.7627,
+      "step": 7136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748945126243353e-05,
+      "loss": 0.7666,
+      "step": 7137
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7488688398629485e-05,
+      "loss": 0.7725,
+      "step": 7138
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748792543558277e-05,
+      "loss": 0.7891,
+      "step": 7139
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.74871623733035e-05,
+      "loss": 0.7646,
+      "step": 7140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748639921180178e-05,
+      "loss": 0.7656,
+      "step": 7141
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7485635951087733e-05,
+      "loss": 0.75,
+      "step": 7142
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748487259117147e-05,
+      "loss": 0.7764,
+      "step": 7143
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7484109132063104e-05,
+      "loss": 0.752,
+      "step": 7144
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7483345573772755e-05,
+      "loss": 0.7656,
+      "step": 7145
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7482581916310544e-05,
+      "loss": 0.7793,
+      "step": 7146
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748181815968659e-05,
+      "loss": 0.7744,
+      "step": 7147
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7481054303911012e-05,
+      "loss": 0.7373,
+      "step": 7148
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748029034899394e-05,
+      "loss": 0.7568,
+      "step": 7149
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7479526294945487e-05,
+      "loss": 0.7578,
+      "step": 7150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747876214177579e-05,
+      "loss": 0.7676,
+      "step": 7151
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7477997889494966e-05,
+      "loss": 0.7725,
+      "step": 7152
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7477233538113153e-05,
+      "loss": 0.7607,
+      "step": 7153
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747646908764047e-05,
+      "loss": 0.7695,
+      "step": 7154
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475704538087055e-05,
+      "loss": 0.7549,
+      "step": 7155
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747493988946304e-05,
+      "loss": 0.7617,
+      "step": 7156
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7474175141778554e-05,
+      "loss": 0.7393,
+      "step": 7157
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7473410295043732e-05,
+      "loss": 0.7539,
+      "step": 7158
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747264534926872e-05,
+      "loss": 0.7549,
+      "step": 7159
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7471880304463638e-05,
+      "loss": 0.7783,
+      "step": 7160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747111516063864e-05,
+      "loss": 0.7568,
+      "step": 7161
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7470349917803856e-05,
+      "loss": 0.7578,
+      "step": 7162
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7469584575969433e-05,
+      "loss": 0.7998,
+      "step": 7163
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468819135145513e-05,
+      "loss": 0.7812,
+      "step": 7164
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468053595342238e-05,
+      "loss": 0.7568,
+      "step": 7165
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7467287956569753e-05,
+      "loss": 0.7764,
+      "step": 7166
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7466522218838205e-05,
+      "loss": 0.7803,
+      "step": 7167
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7465756382157746e-05,
+      "loss": 0.7676,
+      "step": 7168
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746499044653852e-05,
+      "loss": 0.7676,
+      "step": 7169
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464224411990676e-05,
+      "loss": 0.752,
+      "step": 7170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746345827852437e-05,
+      "loss": 0.7637,
+      "step": 7171
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462692046149753e-05,
+      "loss": 0.7559,
+      "step": 7172
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7461925714876984e-05,
+      "loss": 0.7861,
+      "step": 7173
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7461159284716213e-05,
+      "loss": 0.7607,
+      "step": 7174
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7460392755677592e-05,
+      "loss": 0.7705,
+      "step": 7175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745962612777129e-05,
+      "loss": 0.7432,
+      "step": 7176
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7458859401007465e-05,
+      "loss": 0.7734,
+      "step": 7177
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7458092575396274e-05,
+      "loss": 0.7393,
+      "step": 7178
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7457325650947884e-05,
+      "loss": 0.7578,
+      "step": 7179
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745655862767245e-05,
+      "loss": 0.7803,
+      "step": 7180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455791505580148e-05,
+      "loss": 0.7793,
+      "step": 7181
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455024284681133e-05,
+      "loss": 0.7559,
+      "step": 7182
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745425696498558e-05,
+      "loss": 0.7734,
+      "step": 7183
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7453489546503653e-05,
+      "loss": 0.7451,
+      "step": 7184
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7452722029245527e-05,
+      "loss": 0.7539,
+      "step": 7185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7451954413221373e-05,
+      "loss": 0.7803,
+      "step": 7186
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745118669844136e-05,
+      "loss": 0.7637,
+      "step": 7187
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7450418884915663e-05,
+      "loss": 0.7871,
+      "step": 7188
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744965097265446e-05,
+      "loss": 0.7373,
+      "step": 7189
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7448882961667922e-05,
+      "loss": 0.7363,
+      "step": 7190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7448114851966235e-05,
+      "loss": 0.7646,
+      "step": 7191
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7447346643559574e-05,
+      "loss": 0.7803,
+      "step": 7192
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7446578336458115e-05,
+      "loss": 0.7764,
+      "step": 7193
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744580993067205e-05,
+      "loss": 0.748,
+      "step": 7194
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744504142621155e-05,
+      "loss": 0.7549,
+      "step": 7195
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7444272823086813e-05,
+      "loss": 0.7686,
+      "step": 7196
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443504121308014e-05,
+      "loss": 0.749,
+      "step": 7197
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7442735320885344e-05,
+      "loss": 0.7529,
+      "step": 7198
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7441966421828995e-05,
+      "loss": 0.7891,
+      "step": 7199
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7441197424149146e-05,
+      "loss": 0.7822,
+      "step": 7200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7440428327856e-05,
+      "loss": 0.7666,
+      "step": 7201
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7439659132959744e-05,
+      "loss": 0.7773,
+      "step": 7202
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7438889839470573e-05,
+      "loss": 0.7441,
+      "step": 7203
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743812044739868e-05,
+      "loss": 0.7461,
+      "step": 7204
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743735095675426e-05,
+      "loss": 0.7578,
+      "step": 7205
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7436581367547514e-05,
+      "loss": 0.7607,
+      "step": 7206
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7435811679788643e-05,
+      "loss": 0.7695,
+      "step": 7207
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743504189348784e-05,
+      "loss": 0.7617,
+      "step": 7208
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743427200865531e-05,
+      "loss": 0.752,
+      "step": 7209
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743350202530126e-05,
+      "loss": 0.7676,
+      "step": 7210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432731943435886e-05,
+      "loss": 0.7656,
+      "step": 7211
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431961763069402e-05,
+      "loss": 0.7363,
+      "step": 7212
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431191484212007e-05,
+      "loss": 0.7676,
+      "step": 7213
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7430421106873914e-05,
+      "loss": 0.7383,
+      "step": 7214
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742965063106533e-05,
+      "loss": 0.7412,
+      "step": 7215
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7428880056796467e-05,
+      "loss": 0.7598,
+      "step": 7216
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7428109384077533e-05,
+      "loss": 0.7852,
+      "step": 7217
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742733861291875e-05,
+      "loss": 0.7646,
+      "step": 7218
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742656774333032e-05,
+      "loss": 0.7725,
+      "step": 7219
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742579677532247e-05,
+      "loss": 0.7529,
+      "step": 7220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7425025708905413e-05,
+      "loss": 0.7627,
+      "step": 7221
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7424254544089363e-05,
+      "loss": 0.7871,
+      "step": 7222
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7423483280884545e-05,
+      "loss": 0.7568,
+      "step": 7223
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7422711919301184e-05,
+      "loss": 0.752,
+      "step": 7224
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7421940459349494e-05,
+      "loss": 0.7656,
+      "step": 7225
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74211689010397e-05,
+      "loss": 0.7607,
+      "step": 7226
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742039724438203e-05,
+      "loss": 0.7588,
+      "step": 7227
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7419625489386713e-05,
+      "loss": 0.7793,
+      "step": 7228
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741885363606397e-05,
+      "loss": 0.7764,
+      "step": 7229
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741808168442403e-05,
+      "loss": 0.7715,
+      "step": 7230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7417309634477132e-05,
+      "loss": 0.751,
+      "step": 7231
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7416537486233496e-05,
+      "loss": 0.7598,
+      "step": 7232
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7415765239703363e-05,
+      "loss": 0.7627,
+      "step": 7233
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7414992894896963e-05,
+      "loss": 0.751,
+      "step": 7234
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741422045182453e-05,
+      "loss": 0.7578,
+      "step": 7235
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7413447910496305e-05,
+      "loss": 0.7832,
+      "step": 7236
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7412675270922528e-05,
+      "loss": 0.7705,
+      "step": 7237
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741190253311343e-05,
+      "loss": 0.7539,
+      "step": 7238
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7411129697079258e-05,
+      "loss": 0.7549,
+      "step": 7239
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741035676283025e-05,
+      "loss": 0.7363,
+      "step": 7240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409583730376653e-05,
+      "loss": 0.7695,
+      "step": 7241
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740881059972871e-05,
+      "loss": 0.7744,
+      "step": 7242
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7408037370896664e-05,
+      "loss": 0.7656,
+      "step": 7243
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7407264043890767e-05,
+      "loss": 0.748,
+      "step": 7244
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7406490618721267e-05,
+      "loss": 0.7412,
+      "step": 7245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7405717095398405e-05,
+      "loss": 0.7197,
+      "step": 7246
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404943473932445e-05,
+      "loss": 0.7363,
+      "step": 7247
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404169754333628e-05,
+      "loss": 0.7871,
+      "step": 7248
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7403395936612214e-05,
+      "loss": 0.7529,
+      "step": 7249
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740262202077846e-05,
+      "loss": 0.7617,
+      "step": 7250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7401848006842615e-05,
+      "loss": 0.7842,
+      "step": 7251
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7401073894814938e-05,
+      "loss": 0.7715,
+      "step": 7252
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7400299684705694e-05,
+      "loss": 0.7764,
+      "step": 7253
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739952537652514e-05,
+      "loss": 0.7637,
+      "step": 7254
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7398750970283532e-05,
+      "loss": 0.7871,
+      "step": 7255
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7397976465991137e-05,
+      "loss": 0.7705,
+      "step": 7256
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7397201863658225e-05,
+      "loss": 0.752,
+      "step": 7257
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7396427163295052e-05,
+      "loss": 0.7627,
+      "step": 7258
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7395652364911886e-05,
+      "loss": 0.7773,
+      "step": 7259
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7394877468518998e-05,
+      "loss": 0.7432,
+      "step": 7260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7394102474126654e-05,
+      "loss": 0.7988,
+      "step": 7261
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739332738174513e-05,
+      "loss": 0.7578,
+      "step": 7262
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7392552191384694e-05,
+      "loss": 0.7471,
+      "step": 7263
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7391776903055618e-05,
+      "loss": 0.75,
+      "step": 7264
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739100151676818e-05,
+      "loss": 0.7471,
+      "step": 7265
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739022603253265e-05,
+      "loss": 0.7695,
+      "step": 7266
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389450450359314e-05,
+      "loss": 0.748,
+      "step": 7267
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7388674770258437e-05,
+      "loss": 0.7666,
+      "step": 7268
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738789899224031e-05,
+      "loss": 0.8018,
+      "step": 7269
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7387123116315213e-05,
+      "loss": 0.7637,
+      "step": 7270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7386347142493424e-05,
+      "loss": 0.75,
+      "step": 7271
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7385571070785228e-05,
+      "loss": 0.7744,
+      "step": 7272
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7384794901200908e-05,
+      "loss": 0.7734,
+      "step": 7273
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7384018633750753e-05,
+      "loss": 0.7627,
+      "step": 7274
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7383242268445047e-05,
+      "loss": 0.7432,
+      "step": 7275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7382465805294083e-05,
+      "loss": 0.7832,
+      "step": 7276
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7381689244308148e-05,
+      "loss": 0.7695,
+      "step": 7277
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7380912585497533e-05,
+      "loss": 0.7773,
+      "step": 7278
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7380135828872533e-05,
+      "loss": 0.7568,
+      "step": 7279
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737935897444344e-05,
+      "loss": 0.7451,
+      "step": 7280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7378582022220548e-05,
+      "loss": 0.7666,
+      "step": 7281
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7377804972214157e-05,
+      "loss": 0.7734,
+      "step": 7282
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737702782443456e-05,
+      "loss": 0.8037,
+      "step": 7283
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737625057889206e-05,
+      "loss": 0.7568,
+      "step": 7284
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737547323559696e-05,
+      "loss": 0.75,
+      "step": 7285
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7374695794559552e-05,
+      "loss": 0.7432,
+      "step": 7286
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7373918255790142e-05,
+      "loss": 0.7402,
+      "step": 7287
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7373140619299046e-05,
+      "loss": 0.75,
+      "step": 7288
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737236288509655e-05,
+      "loss": 0.792,
+      "step": 7289
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7371585053192982e-05,
+      "loss": 0.75,
+      "step": 7290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737080712359863e-05,
+      "loss": 0.748,
+      "step": 7291
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737002909632382e-05,
+      "loss": 0.7627,
+      "step": 7292
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7369250971378847e-05,
+      "loss": 0.7686,
+      "step": 7293
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368472748774037e-05,
+      "loss": 0.7812,
+      "step": 7294
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7367694428519697e-05,
+      "loss": 0.7432,
+      "step": 7295
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7366916010626142e-05,
+      "loss": 0.752,
+      "step": 7296
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7366137495103684e-05,
+      "loss": 0.7637,
+      "step": 7297
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7365358881962648e-05,
+      "loss": 0.7939,
+      "step": 7298
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7364580171213346e-05,
+      "loss": 0.7881,
+      "step": 7299
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363801362866103e-05,
+      "loss": 0.7412,
+      "step": 7300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363022456931235e-05,
+      "loss": 0.7715,
+      "step": 7301
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7362243453419067e-05,
+      "loss": 0.7471,
+      "step": 7302
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7361464352339924e-05,
+      "loss": 0.7656,
+      "step": 7303
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7360685153704127e-05,
+      "loss": 0.793,
+      "step": 7304
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7359905857522007e-05,
+      "loss": 0.75,
+      "step": 7305
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7359126463803887e-05,
+      "loss": 0.751,
+      "step": 7306
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358346972560095e-05,
+      "loss": 0.7598,
+      "step": 7307
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735756738380097e-05,
+      "loss": 0.7891,
+      "step": 7308
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7356787697536834e-05,
+      "loss": 0.7471,
+      "step": 7309
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735600791377802e-05,
+      "loss": 0.7686,
+      "step": 7310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7355228032534867e-05,
+      "loss": 0.7461,
+      "step": 7311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735444805381771e-05,
+      "loss": 0.7637,
+      "step": 7312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735366797763688e-05,
+      "loss": 0.7588,
+      "step": 7313
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7352887804002717e-05,
+      "loss": 0.7852,
+      "step": 7314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735210753292557e-05,
+      "loss": 0.7979,
+      "step": 7315
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735132716441576e-05,
+      "loss": 0.7666,
+      "step": 7316
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7350546698483648e-05,
+      "loss": 0.8057,
+      "step": 7317
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734976613513956e-05,
+      "loss": 0.7676,
+      "step": 7318
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734898547439386e-05,
+      "loss": 0.7783,
+      "step": 7319
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7348204716256876e-05,
+      "loss": 0.7461,
+      "step": 7320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7347423860738962e-05,
+      "loss": 0.7686,
+      "step": 7321
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7346642907850468e-05,
+      "loss": 0.7725,
+      "step": 7322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734586185760174e-05,
+      "loss": 0.751,
+      "step": 7323
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7345080710003125e-05,
+      "loss": 0.7539,
+      "step": 7324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7344299465064985e-05,
+      "loss": 0.7441,
+      "step": 7325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7343518122797663e-05,
+      "loss": 0.7451,
+      "step": 7326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7342736683211525e-05,
+      "loss": 0.7383,
+      "step": 7327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7341955146316916e-05,
+      "loss": 0.7588,
+      "step": 7328
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73411735121242e-05,
+      "loss": 0.7637,
+      "step": 7329
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7340391780643738e-05,
+      "loss": 0.7422,
+      "step": 7330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733960995188588e-05,
+      "loss": 0.7627,
+      "step": 7331
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733882802586099e-05,
+      "loss": 0.7529,
+      "step": 7332
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7338046002579437e-05,
+      "loss": 0.7871,
+      "step": 7333
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7337263882051576e-05,
+      "loss": 0.748,
+      "step": 7334
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7336481664287778e-05,
+      "loss": 0.7598,
+      "step": 7335
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335699349298408e-05,
+      "loss": 0.7422,
+      "step": 7336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7334916937093833e-05,
+      "loss": 0.752,
+      "step": 7337
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733413442768442e-05,
+      "loss": 0.7871,
+      "step": 7338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733335182108054e-05,
+      "loss": 0.7588,
+      "step": 7339
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7332569117292568e-05,
+      "loss": 0.7324,
+      "step": 7340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7331786316330875e-05,
+      "loss": 0.7705,
+      "step": 7341
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733100341820583e-05,
+      "loss": 0.7764,
+      "step": 7342
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733022042292781e-05,
+      "loss": 0.7715,
+      "step": 7343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7329437330507196e-05,
+      "loss": 0.7617,
+      "step": 7344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7328654140954363e-05,
+      "loss": 0.7822,
+      "step": 7345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7327870854279692e-05,
+      "loss": 0.7344,
+      "step": 7346
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732708747049356e-05,
+      "loss": 0.7559,
+      "step": 7347
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732630398960635e-05,
+      "loss": 0.7529,
+      "step": 7348
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7325520411628448e-05,
+      "loss": 0.7783,
+      "step": 7349
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7324736736570234e-05,
+      "loss": 0.7676,
+      "step": 7350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7323952964442096e-05,
+      "loss": 0.7686,
+      "step": 7351
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7323169095254417e-05,
+      "loss": 0.7666,
+      "step": 7352
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732238512901759e-05,
+      "loss": 0.7529,
+      "step": 7353
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7321601065742e-05,
+      "loss": 0.7803,
+      "step": 7354
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7320816905438043e-05,
+      "loss": 0.7695,
+      "step": 7355
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7320032648116107e-05,
+      "loss": 0.7764,
+      "step": 7356
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7319248293786588e-05,
+      "loss": 0.7559,
+      "step": 7357
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7318463842459876e-05,
+      "loss": 0.752,
+      "step": 7358
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731767929414637e-05,
+      "loss": 0.7578,
+      "step": 7359
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731689464885647e-05,
+      "loss": 0.7539,
+      "step": 7360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731610990660057e-05,
+      "loss": 0.7607,
+      "step": 7361
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731532506738907e-05,
+      "loss": 0.7715,
+      "step": 7362
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731454013123237e-05,
+      "loss": 0.7803,
+      "step": 7363
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7313755098140877e-05,
+      "loss": 0.7568,
+      "step": 7364
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7312969968124988e-05,
+      "loss": 0.7705,
+      "step": 7365
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7312184741195116e-05,
+      "loss": 0.7461,
+      "step": 7366
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731139941736166e-05,
+      "loss": 0.7539,
+      "step": 7367
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7310613996635028e-05,
+      "loss": 0.75,
+      "step": 7368
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309828479025634e-05,
+      "loss": 0.7871,
+      "step": 7369
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730904286454388e-05,
+      "loss": 0.7393,
+      "step": 7370
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7308257153200187e-05,
+      "loss": 0.7549,
+      "step": 7371
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7307471345004956e-05,
+      "loss": 0.7451,
+      "step": 7372
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7306685439968612e-05,
+      "loss": 0.7803,
+      "step": 7373
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7305899438101565e-05,
+      "loss": 0.75,
+      "step": 7374
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730511333941423e-05,
+      "loss": 0.7754,
+      "step": 7375
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7304327143917027e-05,
+      "loss": 0.752,
+      "step": 7376
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7303540851620373e-05,
+      "loss": 0.7607,
+      "step": 7377
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730275446253469e-05,
+      "loss": 0.793,
+      "step": 7378
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.73019679766704e-05,
+      "loss": 0.7627,
+      "step": 7379
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7301181394037922e-05,
+      "loss": 0.752,
+      "step": 7380
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7300394714647683e-05,
+      "loss": 0.751,
+      "step": 7381
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729960793851011e-05,
+      "loss": 0.7529,
+      "step": 7382
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7298821065635626e-05,
+      "loss": 0.7656,
+      "step": 7383
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729803409603466e-05,
+      "loss": 0.7549,
+      "step": 7384
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7297247029717642e-05,
+      "loss": 0.7637,
+      "step": 7385
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7296459866695002e-05,
+      "loss": 0.75,
+      "step": 7386
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729567260697717e-05,
+      "loss": 0.7412,
+      "step": 7387
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7294885250574584e-05,
+      "loss": 0.7637,
+      "step": 7388
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7294097797497674e-05,
+      "loss": 0.7617,
+      "step": 7389
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7293310247756876e-05,
+      "loss": 0.7637,
+      "step": 7390
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7292522601362633e-05,
+      "loss": 0.7764,
+      "step": 7391
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7291734858325372e-05,
+      "loss": 0.7549,
+      "step": 7392
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729094701865554e-05,
+      "loss": 0.7432,
+      "step": 7393
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7290159082363573e-05,
+      "loss": 0.7734,
+      "step": 7394
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7289371049459922e-05,
+      "loss": 0.7461,
+      "step": 7395
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728858291995502e-05,
+      "loss": 0.7539,
+      "step": 7396
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728779469385932e-05,
+      "loss": 0.7666,
+      "step": 7397
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728700637118326e-05,
+      "loss": 0.7559,
+      "step": 7398
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728621795193729e-05,
+      "loss": 0.7764,
+      "step": 7399
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7285429436131862e-05,
+      "loss": 0.7705,
+      "step": 7400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7284640823777424e-05,
+      "loss": 0.7607,
+      "step": 7401
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728385211488442e-05,
+      "loss": 0.7666,
+      "step": 7402
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7283063309463314e-05,
+      "loss": 0.752,
+      "step": 7403
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728227440752455e-05,
+      "loss": 0.7627,
+      "step": 7404
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728148540907859e-05,
+      "loss": 0.749,
+      "step": 7405
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7280696314135885e-05,
+      "loss": 0.7539,
+      "step": 7406
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279907122706894e-05,
+      "loss": 0.7236,
+      "step": 7407
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279117834802075e-05,
+      "loss": 0.7607,
+      "step": 7408
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727832845043189e-05,
+      "loss": 0.7451,
+      "step": 7409
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7277538969606796e-05,
+      "loss": 0.7656,
+      "step": 7410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727674939233726e-05,
+      "loss": 0.7773,
+      "step": 7411
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7275959718633743e-05,
+      "loss": 0.7725,
+      "step": 7412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727516994850671e-05,
+      "loss": 0.7275,
+      "step": 7413
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727438008196663e-05,
+      "loss": 0.7559,
+      "step": 7414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7273590119023968e-05,
+      "loss": 0.7637,
+      "step": 7415
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7272800059689194e-05,
+      "loss": 0.7695,
+      "step": 7416
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727200990397278e-05,
+      "loss": 0.7861,
+      "step": 7417
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7271219651885193e-05,
+      "loss": 0.752,
+      "step": 7418
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7270429303436908e-05,
+      "loss": 0.7549,
+      "step": 7419
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7269638858638395e-05,
+      "loss": 0.7783,
+      "step": 7420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7268848317500137e-05,
+      "loss": 0.7578,
+      "step": 7421
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7268057680032606e-05,
+      "loss": 0.752,
+      "step": 7422
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7267266946246282e-05,
+      "loss": 0.7764,
+      "step": 7423
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7266476116151642e-05,
+      "loss": 0.7861,
+      "step": 7424
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7265685189759164e-05,
+      "loss": 0.7852,
+      "step": 7425
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264894167079334e-05,
+      "loss": 0.7627,
+      "step": 7426
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264103048122636e-05,
+      "loss": 0.7627,
+      "step": 7427
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7263311832899547e-05,
+      "loss": 0.7607,
+      "step": 7428
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726252052142056e-05,
+      "loss": 0.7627,
+      "step": 7429
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726172911369616e-05,
+      "loss": 0.791,
+      "step": 7430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726093760973683e-05,
+      "loss": 0.752,
+      "step": 7431
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7260146009553065e-05,
+      "loss": 0.7441,
+      "step": 7432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7259354313155352e-05,
+      "loss": 0.7617,
+      "step": 7433
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7258562520554188e-05,
+      "loss": 0.7402,
+      "step": 7434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725777063176006e-05,
+      "loss": 0.7578,
+      "step": 7435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7256978646783466e-05,
+      "loss": 0.7627,
+      "step": 7436
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7256186565634898e-05,
+      "loss": 0.7764,
+      "step": 7437
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7255394388324858e-05,
+      "loss": 0.7461,
+      "step": 7438
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7254602114863842e-05,
+      "loss": 0.7559,
+      "step": 7439
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253809745262347e-05,
+      "loss": 0.7422,
+      "step": 7440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253017279530877e-05,
+      "loss": 0.7568,
+      "step": 7441
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7252224717679933e-05,
+      "loss": 0.7275,
+      "step": 7442
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725143205972002e-05,
+      "loss": 0.7656,
+      "step": 7443
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7250639305661638e-05,
+      "loss": 0.7549,
+      "step": 7444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72498464555153e-05,
+      "loss": 0.7402,
+      "step": 7445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72490535092915e-05,
+      "loss": 0.7363,
+      "step": 7446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7248260467000763e-05,
+      "loss": 0.7598,
+      "step": 7447
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7247467328653588e-05,
+      "loss": 0.7637,
+      "step": 7448
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724667409426049e-05,
+      "loss": 0.7461,
+      "step": 7449
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7245880763831983e-05,
+      "loss": 0.749,
+      "step": 7450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7245087337378572e-05,
+      "loss": 0.7549,
+      "step": 7451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724429381491078e-05,
+      "loss": 0.7646,
+      "step": 7452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724350019643912e-05,
+      "loss": 0.7822,
+      "step": 7453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7242706481974107e-05,
+      "loss": 0.7578,
+      "step": 7454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7241912671526265e-05,
+      "loss": 0.8008,
+      "step": 7455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7241118765106114e-05,
+      "loss": 0.7588,
+      "step": 7456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7240324762724164e-05,
+      "loss": 0.7793,
+      "step": 7457
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723953066439095e-05,
+      "loss": 0.7734,
+      "step": 7458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238736470116997e-05,
+      "loss": 0.7324,
+      "step": 7459
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7237942179912818e-05,
+      "loss": 0.7295,
+      "step": 7460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7237147793788947e-05,
+      "loss": 0.752,
+      "step": 7461
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7236353311755905e-05,
+      "loss": 0.7158,
+      "step": 7462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7235558733824232e-05,
+      "loss": 0.7529,
+      "step": 7463
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723476406000445e-05,
+      "loss": 0.7412,
+      "step": 7464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723396929030709e-05,
+      "loss": 0.7617,
+      "step": 7465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7233174424742688e-05,
+      "loss": 0.7627,
+      "step": 7466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7232379463321774e-05,
+      "loss": 0.7656,
+      "step": 7467
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723158440605489e-05,
+      "loss": 0.7646,
+      "step": 7468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7230789252952564e-05,
+      "loss": 0.7803,
+      "step": 7469
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7229994004025337e-05,
+      "loss": 0.7783,
+      "step": 7470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722919865928375e-05,
+      "loss": 0.749,
+      "step": 7471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722840321873834e-05,
+      "loss": 0.7529,
+      "step": 7472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227607682399653e-05,
+      "loss": 0.751,
+      "step": 7473
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7226812050278224e-05,
+      "loss": 0.7676,
+      "step": 7474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7226016322384603e-05,
+      "loss": 0.7539,
+      "step": 7475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7225220498729337e-05,
+      "loss": 0.7607,
+      "step": 7476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7224424579322966e-05,
+      "loss": 0.7461,
+      "step": 7477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7223628564176044e-05,
+      "loss": 0.7734,
+      "step": 7478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222832453299116e-05,
+      "loss": 0.7373,
+      "step": 7479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222036246702734e-05,
+      "loss": 0.7676,
+      "step": 7480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722123994439745e-05,
+      "loss": 0.7812,
+      "step": 7481
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220443546393815e-05,
+      "loss": 0.7725,
+      "step": 7482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7219647052702385e-05,
+      "loss": 0.7734,
+      "step": 7483
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7218850463333712e-05,
+      "loss": 0.7373,
+      "step": 7484
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7218053778298357e-05,
+      "loss": 0.7607,
+      "step": 7485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721725699760688e-05,
+      "loss": 0.7432,
+      "step": 7486
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7216460121269833e-05,
+      "loss": 0.7646,
+      "step": 7487
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7215663149297777e-05,
+      "loss": 0.7695,
+      "step": 7488
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721486608170128e-05,
+      "loss": 0.7559,
+      "step": 7489
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7214068918490904e-05,
+      "loss": 0.7539,
+      "step": 7490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721327165967721e-05,
+      "loss": 0.7568,
+      "step": 7491
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7212474305270763e-05,
+      "loss": 0.7686,
+      "step": 7492
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7211676855282128e-05,
+      "loss": 0.7549,
+      "step": 7493
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721087930972188e-05,
+      "loss": 0.7461,
+      "step": 7494
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7210081668600586e-05,
+      "loss": 0.7617,
+      "step": 7495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7209283931928815e-05,
+      "loss": 0.7354,
+      "step": 7496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7208486099717138e-05,
+      "loss": 0.7422,
+      "step": 7497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720768817197613e-05,
+      "loss": 0.7461,
+      "step": 7498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7206890148716365e-05,
+      "loss": 0.7324,
+      "step": 7499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720609202994842e-05,
+      "loss": 0.7686,
+      "step": 7500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720529381568287e-05,
+      "loss": 0.7656,
+      "step": 7501
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720449550593029e-05,
+      "loss": 0.7578,
+      "step": 7502
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7203697100701268e-05,
+      "loss": 0.749,
+      "step": 7503
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7202898600006378e-05,
+      "loss": 0.7432,
+      "step": 7504
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7202100003856205e-05,
+      "loss": 0.748,
+      "step": 7505
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7201301312261325e-05,
+      "loss": 0.7676,
+      "step": 7506
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7200502525232335e-05,
+      "loss": 0.7578,
+      "step": 7507
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719970364277981e-05,
+      "loss": 0.7363,
+      "step": 7508
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7198904664914348e-05,
+      "loss": 0.7471,
+      "step": 7509
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7198105591646528e-05,
+      "loss": 0.7852,
+      "step": 7510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719730642298694e-05,
+      "loss": 0.7686,
+      "step": 7511
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719650715894618e-05,
+      "loss": 0.7832,
+      "step": 7512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7195707799534837e-05,
+      "loss": 0.7764,
+      "step": 7513
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194908344763503e-05,
+      "loss": 0.7686,
+      "step": 7514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194108794642777e-05,
+      "loss": 0.7607,
+      "step": 7515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719330914918325e-05,
+      "loss": 0.7529,
+      "step": 7516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7192509408395522e-05,
+      "loss": 0.7725,
+      "step": 7517
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7191709572290192e-05,
+      "loss": 0.751,
+      "step": 7518
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7190909640877864e-05,
+      "loss": 0.8047,
+      "step": 7519
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7190109614169124e-05,
+      "loss": 0.7607,
+      "step": 7520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718930949217459e-05,
+      "loss": 0.749,
+      "step": 7521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7188509274904856e-05,
+      "loss": 0.8096,
+      "step": 7522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7187708962370534e-05,
+      "loss": 0.7695,
+      "step": 7523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7186908554582223e-05,
+      "loss": 0.7646,
+      "step": 7524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7186108051550533e-05,
+      "loss": 0.7686,
+      "step": 7525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718530745328608e-05,
+      "loss": 0.7588,
+      "step": 7526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184506759799458e-05,
+      "loss": 0.7744,
+      "step": 7527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718370597110129e-05,
+      "loss": 0.7588,
+      "step": 7528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7182905087202186e-05,
+      "loss": 0.7568,
+      "step": 7529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7182104108112758e-05,
+      "loss": 0.7578,
+      "step": 7530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7181303033843624e-05,
+      "loss": 0.748,
+      "step": 7531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180501864405395e-05,
+      "loss": 0.7949,
+      "step": 7532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717970059980869e-05,
+      "loss": 0.7461,
+      "step": 7533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178899240064133e-05,
+      "loss": 0.7441,
+      "step": 7534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178097785182336e-05,
+      "loss": 0.7627,
+      "step": 7535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717729623517393e-05,
+      "loss": 0.7441,
+      "step": 7536
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7176494590049525e-05,
+      "loss": 0.7578,
+      "step": 7537
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717569284981975e-05,
+      "loss": 0.7734,
+      "step": 7538
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7174891014495233e-05,
+      "loss": 0.7549,
+      "step": 7539
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71740890840866e-05,
+      "loss": 0.7432,
+      "step": 7540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7173287058604473e-05,
+      "loss": 0.7695,
+      "step": 7541
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7172484938059487e-05,
+      "loss": 0.7617,
+      "step": 7542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717168272246227e-05,
+      "loss": 0.7539,
+      "step": 7543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717088041182345e-05,
+      "loss": 0.7705,
+      "step": 7544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717007800615366e-05,
+      "loss": 0.7607,
+      "step": 7545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716927550546354e-05,
+      "loss": 0.7471,
+      "step": 7546
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716847290976372e-05,
+      "loss": 0.75,
+      "step": 7547
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7167670219064834e-05,
+      "loss": 0.7695,
+      "step": 7548
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166867433377526e-05,
+      "loss": 0.7529,
+      "step": 7549
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166064552712427e-05,
+      "loss": 0.7734,
+      "step": 7550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7165261577080182e-05,
+      "loss": 0.7861,
+      "step": 7551
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164458506491435e-05,
+      "loss": 0.748,
+      "step": 7552
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7163655340956823e-05,
+      "loss": 0.7627,
+      "step": 7553
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7162852080486995e-05,
+      "loss": 0.7646,
+      "step": 7554
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716204872509259e-05,
+      "loss": 0.7617,
+      "step": 7555
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7161245274784258e-05,
+      "loss": 0.7578,
+      "step": 7556
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7160441729572647e-05,
+      "loss": 0.7441,
+      "step": 7557
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7159638089468405e-05,
+      "loss": 0.7803,
+      "step": 7558
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7158834354482183e-05,
+      "loss": 0.792,
+      "step": 7559
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7158030524624628e-05,
+      "loss": 0.7559,
+      "step": 7560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71572265999064e-05,
+      "loss": 0.7773,
+      "step": 7561
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7156422580338143e-05,
+      "loss": 0.7627,
+      "step": 7562
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7155618465930525e-05,
+      "loss": 0.7715,
+      "step": 7563
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715481425669419e-05,
+      "loss": 0.7471,
+      "step": 7564
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7154009952639806e-05,
+      "loss": 0.7539,
+      "step": 7565
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7153205553778022e-05,
+      "loss": 0.7793,
+      "step": 7566
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7152401060119507e-05,
+      "loss": 0.7656,
+      "step": 7567
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7151596471674917e-05,
+      "loss": 0.7666,
+      "step": 7568
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7150791788454917e-05,
+      "loss": 0.7529,
+      "step": 7569
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7149987010470172e-05,
+      "loss": 0.7568,
+      "step": 7570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714918213773134e-05,
+      "loss": 0.752,
+      "step": 7571
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148377170249096e-05,
+      "loss": 0.752,
+      "step": 7572
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7147572108034104e-05,
+      "loss": 0.7451,
+      "step": 7573
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146766951097037e-05,
+      "loss": 0.7617,
+      "step": 7574
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714596169944856e-05,
+      "loss": 0.7656,
+      "step": 7575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7145156353099344e-05,
+      "loss": 0.7578,
+      "step": 7576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7144350912060064e-05,
+      "loss": 0.7695,
+      "step": 7577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7143545376341396e-05,
+      "loss": 0.7656,
+      "step": 7578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7142739745954013e-05,
+      "loss": 0.79,
+      "step": 7579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714193402090859e-05,
+      "loss": 0.7539,
+      "step": 7580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141128201215808e-05,
+      "loss": 0.75,
+      "step": 7581
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7140322286886345e-05,
+      "loss": 0.7549,
+      "step": 7582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713951627793088e-05,
+      "loss": 0.7588,
+      "step": 7583
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7138710174360092e-05,
+      "loss": 0.7324,
+      "step": 7584
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7137903976184674e-05,
+      "loss": 0.749,
+      "step": 7585
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7137097683415297e-05,
+      "loss": 0.7695,
+      "step": 7586
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7136291296062656e-05,
+      "loss": 0.7676,
+      "step": 7587
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7135484814137432e-05,
+      "loss": 0.7363,
+      "step": 7588
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7134678237650313e-05,
+      "loss": 0.7549,
+      "step": 7589
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713387156661199e-05,
+      "loss": 0.7637,
+      "step": 7590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7133064801033155e-05,
+      "loss": 0.7686,
+      "step": 7591
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7132257940924494e-05,
+      "loss": 0.7598,
+      "step": 7592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7131450986296707e-05,
+      "loss": 0.7686,
+      "step": 7593
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7130643937160484e-05,
+      "loss": 0.7773,
+      "step": 7594
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7129836793526518e-05,
+      "loss": 0.7686,
+      "step": 7595
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7129029555405508e-05,
+      "loss": 0.7793,
+      "step": 7596
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7128222222808154e-05,
+      "loss": 0.751,
+      "step": 7597
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7127414795745153e-05,
+      "loss": 0.7637,
+      "step": 7598
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.71266072742272e-05,
+      "loss": 0.7725,
+      "step": 7599
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712579965826501e-05,
+      "loss": 0.7324,
+      "step": 7600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712499194786927e-05,
+      "loss": 0.7529,
+      "step": 7601
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7124184143050696e-05,
+      "loss": 0.7666,
+      "step": 7602
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7123376243819986e-05,
+      "loss": 0.7598,
+      "step": 7603
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712256825018785e-05,
+      "loss": 0.7627,
+      "step": 7604
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7121760162164996e-05,
+      "loss": 0.7314,
+      "step": 7605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7120951979762136e-05,
+      "loss": 0.7764,
+      "step": 7606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7120143702989973e-05,
+      "loss": 0.7666,
+      "step": 7607
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711933533185922e-05,
+      "loss": 0.7695,
+      "step": 7608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.71185268663806e-05,
+      "loss": 0.75,
+      "step": 7609
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711771830656481e-05,
+      "loss": 0.7646,
+      "step": 7610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7116909652422577e-05,
+      "loss": 0.7529,
+      "step": 7611
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7116100903964617e-05,
+      "loss": 0.7559,
+      "step": 7612
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7115292061201644e-05,
+      "loss": 0.7461,
+      "step": 7613
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711448312414438e-05,
+      "loss": 0.7529,
+      "step": 7614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7113674092803546e-05,
+      "loss": 0.7393,
+      "step": 7615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7112864967189858e-05,
+      "loss": 0.7559,
+      "step": 7616
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7112055747314044e-05,
+      "loss": 0.7373,
+      "step": 7617
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7111246433186824e-05,
+      "loss": 0.749,
+      "step": 7618
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711043702481893e-05,
+      "loss": 0.791,
+      "step": 7619
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710962752222108e-05,
+      "loss": 0.7354,
+      "step": 7620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710881792540401e-05,
+      "loss": 0.752,
+      "step": 7621
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7108008234378446e-05,
+      "loss": 0.7256,
+      "step": 7622
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7107198449155116e-05,
+      "loss": 0.7324,
+      "step": 7623
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7106388569744755e-05,
+      "loss": 0.79,
+      "step": 7624
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710557859615809e-05,
+      "loss": 0.7637,
+      "step": 7625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7104768528405863e-05,
+      "loss": 0.7588,
+      "step": 7626
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103958366498802e-05,
+      "loss": 0.7441,
+      "step": 7627
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710314811044765e-05,
+      "loss": 0.7734,
+      "step": 7628
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7102337760263138e-05,
+      "loss": 0.7578,
+      "step": 7629
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710152731595601e-05,
+      "loss": 0.7715,
+      "step": 7630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7100716777537005e-05,
+      "loss": 0.7383,
+      "step": 7631
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7099906145016864e-05,
+      "loss": 0.7451,
+      "step": 7632
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709909541840633e-05,
+      "loss": 0.7588,
+      "step": 7633
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709828459771615e-05,
+      "loss": 0.7432,
+      "step": 7634
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7097473682957068e-05,
+      "loss": 0.7451,
+      "step": 7635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7096662674139822e-05,
+      "loss": 0.7744,
+      "step": 7636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709585157127517e-05,
+      "loss": 0.751,
+      "step": 7637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7095040374373862e-05,
+      "loss": 0.7754,
+      "step": 7638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709422908344664e-05,
+      "loss": 0.7588,
+      "step": 7639
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709341769850426e-05,
+      "loss": 0.7598,
+      "step": 7640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092606219557475e-05,
+      "loss": 0.7764,
+      "step": 7641
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7091794646617035e-05,
+      "loss": 0.7656,
+      "step": 7642
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7090982979693703e-05,
+      "loss": 0.751,
+      "step": 7643
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709017121879823e-05,
+      "loss": 0.7725,
+      "step": 7644
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7089359363941374e-05,
+      "loss": 0.7627,
+      "step": 7645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70885474151339e-05,
+      "loss": 0.7686,
+      "step": 7646
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7087735372386554e-05,
+      "loss": 0.7471,
+      "step": 7647
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7086923235710113e-05,
+      "loss": 0.7715,
+      "step": 7648
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7086111005115333e-05,
+      "loss": 0.7402,
+      "step": 7649
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7085298680612977e-05,
+      "loss": 0.7422,
+      "step": 7650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7084486262213808e-05,
+      "loss": 0.7734,
+      "step": 7651
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70836737499286e-05,
+      "loss": 0.7832,
+      "step": 7652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7082861143768114e-05,
+      "loss": 0.7549,
+      "step": 7653
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7082048443743124e-05,
+      "loss": 0.7803,
+      "step": 7654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7081235649864396e-05,
+      "loss": 0.7715,
+      "step": 7655
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7080422762142706e-05,
+      "loss": 0.7646,
+      "step": 7656
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707960978058882e-05,
+      "loss": 0.7686,
+      "step": 7657
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7078796705213515e-05,
+      "loss": 0.7549,
+      "step": 7658
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707798353602757e-05,
+      "loss": 0.7607,
+      "step": 7659
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7077170273041756e-05,
+      "loss": 0.75,
+      "step": 7660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7076356916266857e-05,
+      "loss": 0.7344,
+      "step": 7661
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7075543465713643e-05,
+      "loss": 0.7871,
+      "step": 7662
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70747299213929e-05,
+      "loss": 0.7334,
+      "step": 7663
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7073916283315407e-05,
+      "loss": 0.7529,
+      "step": 7664
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707310255149195e-05,
+      "loss": 0.7393,
+      "step": 7665
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707228872593331e-05,
+      "loss": 0.7549,
+      "step": 7666
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7071474806650275e-05,
+      "loss": 0.7646,
+      "step": 7667
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7070660793653625e-05,
+      "loss": 0.7471,
+      "step": 7668
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7069846686954155e-05,
+      "loss": 0.7695,
+      "step": 7669
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706903248656265e-05,
+      "loss": 0.7783,
+      "step": 7670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7068218192489897e-05,
+      "loss": 0.7568,
+      "step": 7671
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7067403804746698e-05,
+      "loss": 0.7627,
+      "step": 7672
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7066589323343833e-05,
+      "loss": 0.8047,
+      "step": 7673
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70657747482921e-05,
+      "loss": 0.7607,
+      "step": 7674
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7064960079602297e-05,
+      "loss": 0.7656,
+      "step": 7675
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7064145317285218e-05,
+      "loss": 0.7705,
+      "step": 7676
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7063330461351665e-05,
+      "loss": 0.748,
+      "step": 7677
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706251551181243e-05,
+      "loss": 0.7549,
+      "step": 7678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7061700468678315e-05,
+      "loss": 0.791,
+      "step": 7679
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7060885331960124e-05,
+      "loss": 0.7461,
+      "step": 7680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7060070101668654e-05,
+      "loss": 0.7451,
+      "step": 7681
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7059254777814714e-05,
+      "loss": 0.7578,
+      "step": 7682
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705843936040911e-05,
+      "loss": 0.7529,
+      "step": 7683
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7057623849462638e-05,
+      "loss": 0.7451,
+      "step": 7684
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7056808244986117e-05,
+      "loss": 0.7627,
+      "step": 7685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7055992546990356e-05,
+      "loss": 0.7734,
+      "step": 7686
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7055176755486155e-05,
+      "loss": 0.7754,
+      "step": 7687
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705436087048433e-05,
+      "loss": 0.7754,
+      "step": 7688
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7053544891995698e-05,
+      "loss": 0.7559,
+      "step": 7689
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052728820031065e-05,
+      "loss": 0.748,
+      "step": 7690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705191265460125e-05,
+      "loss": 0.7705,
+      "step": 7691
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705109639571707e-05,
+      "loss": 0.7627,
+      "step": 7692
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7050280043389338e-05,
+      "loss": 0.7539,
+      "step": 7693
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7049463597628875e-05,
+      "loss": 0.7334,
+      "step": 7694
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7048647058446507e-05,
+      "loss": 0.7676,
+      "step": 7695
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7047830425853046e-05,
+      "loss": 0.7705,
+      "step": 7696
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7047013699859313e-05,
+      "loss": 0.7646,
+      "step": 7697
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704619688047614e-05,
+      "loss": 0.7783,
+      "step": 7698
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704537996771435e-05,
+      "loss": 0.7637,
+      "step": 7699
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7044562961584764e-05,
+      "loss": 0.8008,
+      "step": 7700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7043745862098215e-05,
+      "loss": 0.7656,
+      "step": 7701
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7042928669265525e-05,
+      "loss": 0.7578,
+      "step": 7702
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704211138309753e-05,
+      "loss": 0.749,
+      "step": 7703
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7041294003605056e-05,
+      "loss": 0.7549,
+      "step": 7704
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704047653079894e-05,
+      "loss": 0.7764,
+      "step": 7705
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7039658964690012e-05,
+      "loss": 0.7676,
+      "step": 7706
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7038841305289108e-05,
+      "loss": 0.7529,
+      "step": 7707
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703802355260706e-05,
+      "loss": 0.7588,
+      "step": 7708
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7037205706654713e-05,
+      "loss": 0.7939,
+      "step": 7709
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70363877674429e-05,
+      "loss": 0.748,
+      "step": 7710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703556973498246e-05,
+      "loss": 0.7617,
+      "step": 7711
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7034751609284237e-05,
+      "loss": 0.7617,
+      "step": 7712
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703393339035907e-05,
+      "loss": 0.7637,
+      "step": 7713
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7033115078217806e-05,
+      "loss": 0.7666,
+      "step": 7714
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7032296672871286e-05,
+      "loss": 0.7559,
+      "step": 7715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7031478174330356e-05,
+      "loss": 0.7627,
+      "step": 7716
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030659582605867e-05,
+      "loss": 0.7373,
+      "step": 7717
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7029840897708662e-05,
+      "loss": 0.7539,
+      "step": 7718
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7029022119649593e-05,
+      "loss": 0.7803,
+      "step": 7719
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7028203248439512e-05,
+      "loss": 0.7637,
+      "step": 7720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702738428408927e-05,
+      "loss": 0.752,
+      "step": 7721
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7026565226609718e-05,
+      "loss": 0.7744,
+      "step": 7722
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7025746076011715e-05,
+      "loss": 0.7646,
+      "step": 7723
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7024926832306112e-05,
+      "loss": 0.7354,
+      "step": 7724
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702410749550377e-05,
+      "loss": 0.7627,
+      "step": 7725
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7023288065615543e-05,
+      "loss": 0.7666,
+      "step": 7726
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7022468542652292e-05,
+      "loss": 0.7539,
+      "step": 7727
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7021648926624877e-05,
+      "loss": 0.7539,
+      "step": 7728
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7020829217544164e-05,
+      "loss": 0.7578,
+      "step": 7729
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7020009415421008e-05,
+      "loss": 0.7451,
+      "step": 7730
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7019189520266282e-05,
+      "loss": 0.7441,
+      "step": 7731
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701836953209085e-05,
+      "loss": 0.7461,
+      "step": 7732
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701754945090557e-05,
+      "loss": 0.75,
+      "step": 7733
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701672927672132e-05,
+      "loss": 0.7705,
+      "step": 7734
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015909009548967e-05,
+      "loss": 0.7607,
+      "step": 7735
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015088649399377e-05,
+      "loss": 0.7598,
+      "step": 7736
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7014268196283426e-05,
+      "loss": 0.7715,
+      "step": 7737
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7013447650211988e-05,
+      "loss": 0.7607,
+      "step": 7738
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701262701119593e-05,
+      "loss": 0.749,
+      "step": 7739
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7011806279246138e-05,
+      "loss": 0.7354,
+      "step": 7740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701098545437348e-05,
+      "loss": 0.7461,
+      "step": 7741
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7010164536588836e-05,
+      "loss": 0.7451,
+      "step": 7742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7009343525903086e-05,
+      "loss": 0.7656,
+      "step": 7743
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7008522422327112e-05,
+      "loss": 0.7607,
+      "step": 7744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7007701225871794e-05,
+      "loss": 0.7822,
+      "step": 7745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700687993654801e-05,
+      "loss": 0.7598,
+      "step": 7746
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7006058554366657e-05,
+      "loss": 0.7686,
+      "step": 7747
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005237079338604e-05,
+      "loss": 0.7812,
+      "step": 7748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7004415511474747e-05,
+      "loss": 0.7793,
+      "step": 7749
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7003593850785973e-05,
+      "loss": 0.748,
+      "step": 7750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700277209728317e-05,
+      "loss": 0.7422,
+      "step": 7751
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7001950250977225e-05,
+      "loss": 0.7627,
+      "step": 7752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7001128311879034e-05,
+      "loss": 0.7686,
+      "step": 7753
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700030627999949e-05,
+      "loss": 0.7637,
+      "step": 7754
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6999484155349483e-05,
+      "loss": 0.7715,
+      "step": 7755
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699866193793991e-05,
+      "loss": 0.7627,
+      "step": 7756
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6997839627781672e-05,
+      "loss": 0.7461,
+      "step": 7757
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699701722488566e-05,
+      "loss": 0.7734,
+      "step": 7758
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6996194729262766e-05,
+      "loss": 0.7666,
+      "step": 7759
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6995372140923907e-05,
+      "loss": 0.7559,
+      "step": 7760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6994549459879973e-05,
+      "loss": 0.748,
+      "step": 7761
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699372668614187e-05,
+      "loss": 0.7617,
+      "step": 7762
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69929038197205e-05,
+      "loss": 0.7471,
+      "step": 7763
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699208086062677e-05,
+      "loss": 0.7637,
+      "step": 7764
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6991257808871586e-05,
+      "loss": 0.7734,
+      "step": 7765
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699043466446585e-05,
+      "loss": 0.749,
+      "step": 7766
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698961142742048e-05,
+      "loss": 0.7627,
+      "step": 7767
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6988788097746375e-05,
+      "loss": 0.7451,
+      "step": 7768
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6987964675454452e-05,
+      "loss": 0.7637,
+      "step": 7769
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6987141160555628e-05,
+      "loss": 0.7998,
+      "step": 7770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6986317553060812e-05,
+      "loss": 0.7686,
+      "step": 7771
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6985493852980914e-05,
+      "loss": 0.7617,
+      "step": 7772
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6984670060326857e-05,
+      "loss": 0.7646,
+      "step": 7773
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698384617510955e-05,
+      "loss": 0.7686,
+      "step": 7774
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6983022197339923e-05,
+      "loss": 0.7432,
+      "step": 7775
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698219812702889e-05,
+      "loss": 0.7539,
+      "step": 7776
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6981373964187368e-05,
+      "loss": 0.7412,
+      "step": 7777
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6980549708826283e-05,
+      "loss": 0.7559,
+      "step": 7778
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697972536095656e-05,
+      "loss": 0.7695,
+      "step": 7779
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697890092058912e-05,
+      "loss": 0.7266,
+      "step": 7780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697807638773489e-05,
+      "loss": 0.751,
+      "step": 7781
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6977251762404795e-05,
+      "loss": 0.7754,
+      "step": 7782
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6976427044609766e-05,
+      "loss": 0.7471,
+      "step": 7783
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697560223436073e-05,
+      "loss": 0.7549,
+      "step": 7784
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697477733166862e-05,
+      "loss": 0.7598,
+      "step": 7785
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6973952336544373e-05,
+      "loss": 0.7373,
+      "step": 7786
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697312724899891e-05,
+      "loss": 0.7568,
+      "step": 7787
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6972302069043176e-05,
+      "loss": 0.7373,
+      "step": 7788
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69714767966881e-05,
+      "loss": 0.7354,
+      "step": 7789
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697065143194462e-05,
+      "loss": 0.7686,
+      "step": 7790
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969825974823676e-05,
+      "loss": 0.7139,
+      "step": 7791
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969000425336205e-05,
+      "loss": 0.7432,
+      "step": 7792
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6968174783493154e-05,
+      "loss": 0.7725,
+      "step": 7793
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6967349049305456e-05,
+      "loss": 0.7656,
+      "step": 7794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696652322278406e-05,
+      "loss": 0.7861,
+      "step": 7795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6965697303939902e-05,
+      "loss": 0.7705,
+      "step": 7796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6964871292783933e-05,
+      "loss": 0.748,
+      "step": 7797
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6964045189327102e-05,
+      "loss": 0.7617,
+      "step": 7798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6963218993580353e-05,
+      "loss": 0.7686,
+      "step": 7799
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6962392705554644e-05,
+      "loss": 0.7773,
+      "step": 7800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6961566325260907e-05,
+      "loss": 0.7627,
+      "step": 7801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696073985271011e-05,
+      "loss": 0.7559,
+      "step": 7802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69599132879132e-05,
+      "loss": 0.7617,
+      "step": 7803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6959086630881127e-05,
+      "loss": 0.7744,
+      "step": 7804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695825988162485e-05,
+      "loss": 0.7529,
+      "step": 7805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6957433040155326e-05,
+      "loss": 0.7676,
+      "step": 7806
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6956606106483513e-05,
+      "loss": 0.7715,
+      "step": 7807
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6955779080620366e-05,
+      "loss": 0.7568,
+      "step": 7808
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6954951962576853e-05,
+      "loss": 0.7695,
+      "step": 7809
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6954124752363923e-05,
+      "loss": 0.7754,
+      "step": 7810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695329744999255e-05,
+      "loss": 0.7441,
+      "step": 7811
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695247005547369e-05,
+      "loss": 0.7734,
+      "step": 7812
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6951642568818315e-05,
+      "loss": 0.7471,
+      "step": 7813
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6950814990037383e-05,
+      "loss": 0.7686,
+      "step": 7814
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694998731914187e-05,
+      "loss": 0.752,
+      "step": 7815
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6949159556142736e-05,
+      "loss": 0.7695,
+      "step": 7816
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6948331701050956e-05,
+      "loss": 0.7539,
+      "step": 7817
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6947503753877495e-05,
+      "loss": 0.7539,
+      "step": 7818
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6946675714633336e-05,
+      "loss": 0.7705,
+      "step": 7819
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6945847583329444e-05,
+      "loss": 0.7568,
+      "step": 7820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69450193599768e-05,
+      "loss": 0.7637,
+      "step": 7821
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6944191044586373e-05,
+      "loss": 0.7461,
+      "step": 7822
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6943362637169144e-05,
+      "loss": 0.7637,
+      "step": 7823
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6942534137736087e-05,
+      "loss": 0.7451,
+      "step": 7824
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694170554629819e-05,
+      "loss": 0.7617,
+      "step": 7825
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694087686286643e-05,
+      "loss": 0.7578,
+      "step": 7826
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6940048087451783e-05,
+      "loss": 0.7686,
+      "step": 7827
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693921922006524e-05,
+      "loss": 0.7725,
+      "step": 7828
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693839026071778e-05,
+      "loss": 0.7627,
+      "step": 7829
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6937561209420394e-05,
+      "loss": 0.7725,
+      "step": 7830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6936732066184065e-05,
+      "loss": 0.7568,
+      "step": 7831
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6935902831019783e-05,
+      "loss": 0.749,
+      "step": 7832
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6935073503938535e-05,
+      "loss": 0.7451,
+      "step": 7833
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693424408495131e-05,
+      "loss": 0.7598,
+      "step": 7834
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693341457406911e-05,
+      "loss": 0.7598,
+      "step": 7835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6932584971302913e-05,
+      "loss": 0.7734,
+      "step": 7836
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6931755276663725e-05,
+      "loss": 0.75,
+      "step": 7837
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693092549016254e-05,
+      "loss": 0.751,
+      "step": 7838
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6930095611810348e-05,
+      "loss": 0.7588,
+      "step": 7839
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6929265641618152e-05,
+      "loss": 0.7471,
+      "step": 7840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692843557959695e-05,
+      "loss": 0.748,
+      "step": 7841
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692760542575774e-05,
+      "loss": 0.7715,
+      "step": 7842
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6926775180111528e-05,
+      "loss": 0.79,
+      "step": 7843
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6925944842669312e-05,
+      "loss": 0.7598,
+      "step": 7844
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69251144134421e-05,
+      "loss": 0.7812,
+      "step": 7845
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6924283892440896e-05,
+      "loss": 0.752,
+      "step": 7846
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6923453279676703e-05,
+      "loss": 0.7324,
+      "step": 7847
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6922622575160535e-05,
+      "loss": 0.7725,
+      "step": 7848
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6921791778903394e-05,
+      "loss": 0.7559,
+      "step": 7849
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.692096089091629e-05,
+      "loss": 0.7842,
+      "step": 7850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6920129911210242e-05,
+      "loss": 0.75,
+      "step": 7851
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6919298839796257e-05,
+      "loss": 0.7812,
+      "step": 7852
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6918467676685345e-05,
+      "loss": 0.7588,
+      "step": 7853
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691763642188853e-05,
+      "loss": 0.7402,
+      "step": 7854
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6916805075416826e-05,
+      "loss": 0.7607,
+      "step": 7855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691597363728124e-05,
+      "loss": 0.7695,
+      "step": 7856
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6915142107492803e-05,
+      "loss": 0.7402,
+      "step": 7857
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6914310486062527e-05,
+      "loss": 0.7588,
+      "step": 7858
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6913478773001434e-05,
+      "loss": 0.7666,
+      "step": 7859
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691264696832055e-05,
+      "loss": 0.7451,
+      "step": 7860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6911815072030896e-05,
+      "loss": 0.7559,
+      "step": 7861
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.69109830841435e-05,
+      "loss": 0.7373,
+      "step": 7862
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691015100466938e-05,
+      "loss": 0.751,
+      "step": 7863
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690931883361957e-05,
+      "loss": 0.7422,
+      "step": 7864
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6908486571005094e-05,
+      "loss": 0.7549,
+      "step": 7865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6907654216836984e-05,
+      "loss": 0.7598,
+      "step": 7866
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6906821771126272e-05,
+      "loss": 0.7686,
+      "step": 7867
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6905989233883985e-05,
+      "loss": 0.7588,
+      "step": 7868
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690515660512116e-05,
+      "loss": 0.7461,
+      "step": 7869
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6904323884848824e-05,
+      "loss": 0.7646,
+      "step": 7870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6903491073078025e-05,
+      "loss": 0.7627,
+      "step": 7871
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690265816981979e-05,
+      "loss": 0.7363,
+      "step": 7872
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6901825175085166e-05,
+      "loss": 0.7539,
+      "step": 7873
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6900992088885182e-05,
+      "loss": 0.7646,
+      "step": 7874
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6900158911230882e-05,
+      "loss": 0.7695,
+      "step": 7875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689932564213331e-05,
+      "loss": 0.7568,
+      "step": 7876
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6898492281603505e-05,
+      "loss": 0.7568,
+      "step": 7877
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6897658829652513e-05,
+      "loss": 0.7539,
+      "step": 7878
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689682528629138e-05,
+      "loss": 0.7607,
+      "step": 7879
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689599165153115e-05,
+      "loss": 0.7568,
+      "step": 7880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6895157925382878e-05,
+      "loss": 0.7617,
+      "step": 7881
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68943241078576e-05,
+      "loss": 0.749,
+      "step": 7882
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6893490198966375e-05,
+      "loss": 0.7412,
+      "step": 7883
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6892656198720253e-05,
+      "loss": 0.749,
+      "step": 7884
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6891822107130284e-05,
+      "loss": 0.7627,
+      "step": 7885
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689098792420752e-05,
+      "loss": 0.752,
+      "step": 7886
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6890153649963032e-05,
+      "loss": 0.749,
+      "step": 7887
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688931928440785e-05,
+      "loss": 0.7705,
+      "step": 7888
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6888484827553052e-05,
+      "loss": 0.8154,
+      "step": 7889
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6887650279409685e-05,
+      "loss": 0.752,
+      "step": 7890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6886815639988815e-05,
+      "loss": 0.7607,
+      "step": 7891
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68859809093015e-05,
+      "loss": 0.7832,
+      "step": 7892
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6885146087358805e-05,
+      "loss": 0.7754,
+      "step": 7893
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6884311174171787e-05,
+      "loss": 0.7676,
+      "step": 7894
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6883476169751518e-05,
+      "loss": 0.7539,
+      "step": 7895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688264107410906e-05,
+      "loss": 0.7715,
+      "step": 7896
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688180588725548e-05,
+      "loss": 0.7715,
+      "step": 7897
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880970609201848e-05,
+      "loss": 0.751,
+      "step": 7898
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688013523995923e-05,
+      "loss": 0.7461,
+      "step": 7899
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68792997795387e-05,
+      "loss": 0.748,
+      "step": 7900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687846422795133e-05,
+      "loss": 0.7588,
+      "step": 7901
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687762858520819e-05,
+      "loss": 0.749,
+      "step": 7902
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6876792851320356e-05,
+      "loss": 0.7383,
+      "step": 7903
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6875957026298904e-05,
+      "loss": 0.7988,
+      "step": 7904
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6875121110154905e-05,
+      "loss": 0.7666,
+      "step": 7905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6874285102899447e-05,
+      "loss": 0.7617,
+      "step": 7906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6873449004543604e-05,
+      "loss": 0.79,
+      "step": 7907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687261281509845e-05,
+      "loss": 0.7773,
+      "step": 7908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687177653457508e-05,
+      "loss": 0.7686,
+      "step": 7909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6870940162984566e-05,
+      "loss": 0.7441,
+      "step": 7910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6870103700337996e-05,
+      "loss": 0.7715,
+      "step": 7911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686926714664645e-05,
+      "loss": 0.7695,
+      "step": 7912
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6868430501921023e-05,
+      "loss": 0.7607,
+      "step": 7913
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6867593766172795e-05,
+      "loss": 0.7764,
+      "step": 7914
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686675693941286e-05,
+      "loss": 0.7451,
+      "step": 7915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865920021652305e-05,
+      "loss": 0.7812,
+      "step": 7916
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865083012902222e-05,
+      "loss": 0.7568,
+      "step": 7917
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68642459131737e-05,
+      "loss": 0.7354,
+      "step": 7918
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6863408722477837e-05,
+      "loss": 0.7686,
+      "step": 7919
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686257144082573e-05,
+      "loss": 0.7451,
+      "step": 7920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6861734068228468e-05,
+      "loss": 0.8008,
+      "step": 7921
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686089660469715e-05,
+      "loss": 0.7764,
+      "step": 7922
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686005905024288e-05,
+      "loss": 0.7549,
+      "step": 7923
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685922140487675e-05,
+      "loss": 0.7471,
+      "step": 7924
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6858383668609863e-05,
+      "loss": 0.7314,
+      "step": 7925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6857545841453323e-05,
+      "loss": 0.7666,
+      "step": 7926
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6856707923418233e-05,
+      "loss": 0.7783,
+      "step": 7927
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6855869914515697e-05,
+      "loss": 0.7695,
+      "step": 7928
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685503181475682e-05,
+      "loss": 0.7559,
+      "step": 7929
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6854193624152707e-05,
+      "loss": 0.7432,
+      "step": 7930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685335534271447e-05,
+      "loss": 0.7646,
+      "step": 7931
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6852516970453216e-05,
+      "loss": 0.75,
+      "step": 7932
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6851678507380054e-05,
+      "loss": 0.7607,
+      "step": 7933
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6850839953506098e-05,
+      "loss": 0.7617,
+      "step": 7934
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685000130884246e-05,
+      "loss": 0.749,
+      "step": 7935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6849162573400255e-05,
+      "loss": 0.7637,
+      "step": 7936
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6848323747190594e-05,
+      "loss": 0.7471,
+      "step": 7937
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68474848302246e-05,
+      "loss": 0.7773,
+      "step": 7938
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6846645822513385e-05,
+      "loss": 0.7539,
+      "step": 7939
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6845806724068073e-05,
+      "loss": 0.7432,
+      "step": 7940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6844967534899777e-05,
+      "loss": 0.7461,
+      "step": 7941
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6844128255019625e-05,
+      "loss": 0.7637,
+      "step": 7942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6843288884438734e-05,
+      "loss": 0.7568,
+      "step": 7943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684244942316823e-05,
+      "loss": 0.7402,
+      "step": 7944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684160987121924e-05,
+      "loss": 0.7529,
+      "step": 7945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684077022860289e-05,
+      "loss": 0.7588,
+      "step": 7946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6839930495330302e-05,
+      "loss": 0.7705,
+      "step": 7947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683909067141261e-05,
+      "loss": 0.7451,
+      "step": 7948
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683825075686094e-05,
+      "loss": 0.751,
+      "step": 7949
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6837410751686426e-05,
+      "loss": 0.7627,
+      "step": 7950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836570655900195e-05,
+      "loss": 0.748,
+      "step": 7951
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6835730469513386e-05,
+      "loss": 0.7568,
+      "step": 7952
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683489019253713e-05,
+      "loss": 0.7637,
+      "step": 7953
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6834049824982565e-05,
+      "loss": 0.7559,
+      "step": 7954
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6833209366860826e-05,
+      "loss": 0.7441,
+      "step": 7955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683236881818305e-05,
+      "loss": 0.7363,
+      "step": 7956
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683152817896038e-05,
+      "loss": 0.7627,
+      "step": 7957
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6830687449203953e-05,
+      "loss": 0.748,
+      "step": 7958
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682984662892491e-05,
+      "loss": 0.7451,
+      "step": 7959
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68290057181344e-05,
+      "loss": 0.7461,
+      "step": 7960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682816471684356e-05,
+      "loss": 0.752,
+      "step": 7961
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6827323625063535e-05,
+      "loss": 0.7373,
+      "step": 7962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6826482442805477e-05,
+      "loss": 0.7656,
+      "step": 7963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6825641170080532e-05,
+      "loss": 0.7451,
+      "step": 7964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6824799806899848e-05,
+      "loss": 0.7568,
+      "step": 7965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6823958353274574e-05,
+      "loss": 0.7588,
+      "step": 7966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6823116809215862e-05,
+      "loss": 0.7656,
+      "step": 7967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6822275174734865e-05,
+      "loss": 0.7588,
+      "step": 7968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6821433449842737e-05,
+      "loss": 0.7646,
+      "step": 7969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6820591634550628e-05,
+      "loss": 0.7764,
+      "step": 7970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6819749728869705e-05,
+      "loss": 0.7578,
+      "step": 7971
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681890773281111e-05,
+      "loss": 0.7578,
+      "step": 7972
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6818065646386016e-05,
+      "loss": 0.7363,
+      "step": 7973
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6817223469605574e-05,
+      "loss": 0.7383,
+      "step": 7974
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6816381202480948e-05,
+      "loss": 0.7686,
+      "step": 7975
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.68155388450233e-05,
+      "loss": 0.7471,
+      "step": 7976
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6814696397243788e-05,
+      "loss": 0.7627,
+      "step": 7977
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6813853859153582e-05,
+      "loss": 0.7256,
+      "step": 7978
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6813011230763847e-05,
+      "loss": 0.749,
+      "step": 7979
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681216851208575e-05,
+      "loss": 0.751,
+      "step": 7980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681132570313046e-05,
+      "loss": 0.7471,
+      "step": 7981
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681048280390914e-05,
+      "loss": 0.7363,
+      "step": 7982
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680963981443297e-05,
+      "loss": 0.7646,
+      "step": 7983
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6808796734713112e-05,
+      "loss": 0.7705,
+      "step": 7984
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6807953564760743e-05,
+      "loss": 0.748,
+      "step": 7985
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680711030458704e-05,
+      "loss": 0.7461,
+      "step": 7986
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6806266954203173e-05,
+      "loss": 0.7686,
+      "step": 7987
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680542351362032e-05,
+      "loss": 0.7715,
+      "step": 7988
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6804579982849657e-05,
+      "loss": 0.7549,
+      "step": 7989
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680373636190237e-05,
+      "loss": 0.7539,
+      "step": 7990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680289265078963e-05,
+      "loss": 0.7383,
+      "step": 7991
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6802048849522625e-05,
+      "loss": 0.7373,
+      "step": 7992
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6801204958112535e-05,
+      "loss": 0.7598,
+      "step": 7993
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680036097657054e-05,
+      "loss": 0.7637,
+      "step": 7994
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679951690490783e-05,
+      "loss": 0.7695,
+      "step": 7995
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6798672743135583e-05,
+      "loss": 0.7529,
+      "step": 7996
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797828491264998e-05,
+      "loss": 0.7939,
+      "step": 7997
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6796984149307255e-05,
+      "loss": 0.7578,
+      "step": 7998
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6796139717273544e-05,
+      "loss": 0.7793,
+      "step": 7999
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679529519517506e-05,
+      "loss": 0.7373,
+      "step": 8000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6794450583022993e-05,
+      "loss": 0.7686,
+      "step": 8001
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6793605880828534e-05,
+      "loss": 0.7783,
+      "step": 8002
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6792761088602875e-05,
+      "loss": 0.7471,
+      "step": 8003
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679191620635722e-05,
+      "loss": 0.7676,
+      "step": 8004
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6791071234102756e-05,
+      "loss": 0.7666,
+      "step": 8005
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6790226171850688e-05,
+      "loss": 0.7471,
+      "step": 8006
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6789381019612215e-05,
+      "loss": 0.752,
+      "step": 8007
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678853577739853e-05,
+      "loss": 0.7695,
+      "step": 8008
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678769044522084e-05,
+      "loss": 0.7539,
+      "step": 8009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678684502309035e-05,
+      "loss": 0.7432,
+      "step": 8010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6785999511018256e-05,
+      "loss": 0.792,
+      "step": 8011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678515390901577e-05,
+      "loss": 0.7998,
+      "step": 8012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6784308217094097e-05,
+      "loss": 0.7627,
+      "step": 8013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6783462435264443e-05,
+      "loss": 0.7412,
+      "step": 8014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6782616563538018e-05,
+      "loss": 0.7754,
+      "step": 8015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6781770601926028e-05,
+      "loss": 0.75,
+      "step": 8016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6780924550439688e-05,
+      "loss": 0.7637,
+      "step": 8017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6780078409090207e-05,
+      "loss": 0.7344,
+      "step": 8018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6779232177888802e-05,
+      "loss": 0.7549,
+      "step": 8019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6778385856846685e-05,
+      "loss": 0.7588,
+      "step": 8020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6777539445975074e-05,
+      "loss": 0.7637,
+      "step": 8021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677669294528518e-05,
+      "loss": 0.7598,
+      "step": 8022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6775846354788233e-05,
+      "loss": 0.7617,
+      "step": 8023
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677499967449544e-05,
+      "loss": 0.791,
+      "step": 8024
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6774152904418025e-05,
+      "loss": 0.7705,
+      "step": 8025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677330604456722e-05,
+      "loss": 0.752,
+      "step": 8026
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6772459094954227e-05,
+      "loss": 0.751,
+      "step": 8027
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6771612055590284e-05,
+      "loss": 0.751,
+      "step": 8028
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677076492648662e-05,
+      "loss": 0.7617,
+      "step": 8029
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676991770765445e-05,
+      "loss": 0.75,
+      "step": 8030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676907039910501e-05,
+      "loss": 0.7539,
+      "step": 8031
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6768223000849526e-05,
+      "loss": 0.7559,
+      "step": 8032
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676737551289923e-05,
+      "loss": 0.7217,
+      "step": 8033
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6766527935265347e-05,
+      "loss": 0.7725,
+      "step": 8034
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676568026795912e-05,
+      "loss": 0.7744,
+      "step": 8035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6764832510991772e-05,
+      "loss": 0.7832,
+      "step": 8036
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676398466437454e-05,
+      "loss": 0.7578,
+      "step": 8037
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6763136728118667e-05,
+      "loss": 0.7588,
+      "step": 8038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6762288702235383e-05,
+      "loss": 0.7842,
+      "step": 8039
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6761440586735926e-05,
+      "loss": 0.7695,
+      "step": 8040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6760592381631543e-05,
+      "loss": 0.7568,
+      "step": 8041
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6759744086933463e-05,
+      "loss": 0.7432,
+      "step": 8042
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675889570265294e-05,
+      "loss": 0.7715,
+      "step": 8043
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675804722880121e-05,
+      "loss": 0.7578,
+      "step": 8044
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6757198665389517e-05,
+      "loss": 0.7256,
+      "step": 8045
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6756350012429115e-05,
+      "loss": 0.7607,
+      "step": 8046
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6755501269931235e-05,
+      "loss": 0.7461,
+      "step": 8047
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675465243790714e-05,
+      "loss": 0.751,
+      "step": 8048
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6753803516368073e-05,
+      "loss": 0.7705,
+      "step": 8049
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6752954505325282e-05,
+      "loss": 0.7686,
+      "step": 8050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675210540479002e-05,
+      "loss": 0.7383,
+      "step": 8051
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675125621477354e-05,
+      "loss": 0.7588,
+      "step": 8052
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6750406935287097e-05,
+      "loss": 0.7432,
+      "step": 8053
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674955756634194e-05,
+      "loss": 0.7461,
+      "step": 8054
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6748708107949338e-05,
+      "loss": 0.75,
+      "step": 8055
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6747858560120534e-05,
+      "loss": 0.7773,
+      "step": 8056
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6747008922866797e-05,
+      "loss": 0.7744,
+      "step": 8057
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674615919619938e-05,
+      "loss": 0.7793,
+      "step": 8058
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6745309380129547e-05,
+      "loss": 0.7393,
+      "step": 8059
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6744459474668556e-05,
+      "loss": 0.7627,
+      "step": 8060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6743609479827676e-05,
+      "loss": 0.7715,
+      "step": 8061
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6742759395618172e-05,
+      "loss": 0.7539,
+      "step": 8062
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741909222051304e-05,
+      "loss": 0.7773,
+      "step": 8063
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674105895913834e-05,
+      "loss": 0.751,
+      "step": 8064
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674020860689055e-05,
+      "loss": 0.7393,
+      "step": 8065
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6739358165319204e-05,
+      "loss": 0.7646,
+      "step": 8066
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6738507634435567e-05,
+      "loss": 0.7295,
+      "step": 8067
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673765701425092e-05,
+      "loss": 0.7695,
+      "step": 8068
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736806304776524e-05,
+      "loss": 0.7568,
+      "step": 8069
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6735955506023663e-05,
+      "loss": 0.7676,
+      "step": 8070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6735104618003606e-05,
+      "loss": 0.7354,
+      "step": 8071
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6734253640727636e-05,
+      "loss": 0.7598,
+      "step": 8072
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673340257420702e-05,
+      "loss": 0.7549,
+      "step": 8073
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6732551418453045e-05,
+      "loss": 0.7549,
+      "step": 8074
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6731700173476987e-05,
+      "loss": 0.7568,
+      "step": 8075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.673084883929013e-05,
+      "loss": 0.7578,
+      "step": 8076
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672999741590375e-05,
+      "loss": 0.7598,
+      "step": 8077
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672914590332914e-05,
+      "loss": 0.7578,
+      "step": 8078
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6728294301577577e-05,
+      "loss": 0.7744,
+      "step": 8079
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672744261066035e-05,
+      "loss": 0.7783,
+      "step": 8080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6726590830588744e-05,
+      "loss": 0.7461,
+      "step": 8081
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672573896137405e-05,
+      "loss": 0.7529,
+      "step": 8082
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672488700302755e-05,
+      "loss": 0.7617,
+      "step": 8083
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724034955560542e-05,
+      "loss": 0.7637,
+      "step": 8084
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6723182818984315e-05,
+      "loss": 0.7598,
+      "step": 8085
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6722330593310167e-05,
+      "loss": 0.7783,
+      "step": 8086
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672147827854938e-05,
+      "loss": 0.7803,
+      "step": 8087
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6720625874713258e-05,
+      "loss": 0.7539,
+      "step": 8088
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.67197733818131e-05,
+      "loss": 0.7578,
+      "step": 8089
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6718920799860193e-05,
+      "loss": 0.7568,
+      "step": 8090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6718068128865845e-05,
+      "loss": 0.7607,
+      "step": 8091
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671721536884135e-05,
+      "loss": 0.7578,
+      "step": 8092
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6716362519798015e-05,
+      "loss": 0.75,
+      "step": 8093
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6715509581747142e-05,
+      "loss": 0.7559,
+      "step": 8094
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6714656554700024e-05,
+      "loss": 0.7656,
+      "step": 8095
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6713803438667976e-05,
+      "loss": 0.7637,
+      "step": 8096
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6712950233662303e-05,
+      "loss": 0.7627,
+      "step": 8097
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6712096939694307e-05,
+      "loss": 0.7695,
+      "step": 8098
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.67112435567753e-05,
+      "loss": 0.7607,
+      "step": 8099
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671039008491659e-05,
+      "loss": 0.7412,
+      "step": 8100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670953652412949e-05,
+      "loss": 0.7295,
+      "step": 8101
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6708682874425307e-05,
+      "loss": 0.7607,
+      "step": 8102
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6707829135815355e-05,
+      "loss": 0.7539,
+      "step": 8103
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6706975308310953e-05,
+      "loss": 0.7695,
+      "step": 8104
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6706121391923412e-05,
+      "loss": 0.7471,
+      "step": 8105
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6705267386664043e-05,
+      "loss": 0.7812,
+      "step": 8106
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6704413292544178e-05,
+      "loss": 0.7549,
+      "step": 8107
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670355910957512e-05,
+      "loss": 0.7588,
+      "step": 8108
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.67027048377682e-05,
+      "loss": 0.7578,
+      "step": 8109
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6701850477134737e-05,
+      "loss": 0.748,
+      "step": 8110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670099602768605e-05,
+      "loss": 0.7393,
+      "step": 8111
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670014148943346e-05,
+      "loss": 0.7529,
+      "step": 8112
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.66992868623883e-05,
+      "loss": 0.7549,
+      "step": 8113
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669843214656189e-05,
+      "loss": 0.749,
+      "step": 8114
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669757734196556e-05,
+      "loss": 0.7637,
+      "step": 8115
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6696722448610637e-05,
+      "loss": 0.7471,
+      "step": 8116
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695867466508448e-05,
+      "loss": 0.7578,
+      "step": 8117
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695012395670328e-05,
+      "loss": 0.7451,
+      "step": 8118
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.66941572361076e-05,
+      "loss": 0.7803,
+      "step": 8119
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6693301987831613e-05,
+      "loss": 0.7617,
+      "step": 8120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669244665085368e-05,
+      "loss": 0.7617,
+      "step": 8121
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6691591225185156e-05,
+      "loss": 0.7705,
+      "step": 8122
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6690735710837366e-05,
+      "loss": 0.751,
+      "step": 8123
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668988010782165e-05,
+      "loss": 0.7783,
+      "step": 8124
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6689024416149345e-05,
+      "loss": 0.7578,
+      "step": 8125
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6688168635831794e-05,
+      "loss": 0.7441,
+      "step": 8126
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668731276688034e-05,
+      "loss": 0.7598,
+      "step": 8127
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668645680930632e-05,
+      "loss": 0.7568,
+      "step": 8128
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6685600763121078e-05,
+      "loss": 0.7275,
+      "step": 8129
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6684744628335962e-05,
+      "loss": 0.7432,
+      "step": 8130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6683888404962317e-05,
+      "loss": 0.7734,
+      "step": 8131
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.668303209301149e-05,
+      "loss": 0.7646,
+      "step": 8132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6682175692494828e-05,
+      "loss": 0.7402,
+      "step": 8133
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6681319203423678e-05,
+      "loss": 0.7588,
+      "step": 8134
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6680462625809392e-05,
+      "loss": 0.7432,
+      "step": 8135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6679605959663324e-05,
+      "loss": 0.7461,
+      "step": 8136
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6678749204996828e-05,
+      "loss": 0.7695,
+      "step": 8137
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6677892361821257e-05,
+      "loss": 0.7402,
+      "step": 8138
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6677035430147958e-05,
+      "loss": 0.7412,
+      "step": 8139
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.66761784099883e-05,
+      "loss": 0.7451,
+      "step": 8140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667532130135363e-05,
+      "loss": 0.7412,
+      "step": 8141
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6674464104255315e-05,
+      "loss": 0.7881,
+      "step": 8142
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6673606818704712e-05,
+      "loss": 0.7461,
+      "step": 8143
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667274944471318e-05,
+      "loss": 0.7559,
+      "step": 8144
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667189198229208e-05,
+      "loss": 0.7695,
+      "step": 8145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667103443145278e-05,
+      "loss": 0.7637,
+      "step": 8146
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6670176792206642e-05,
+      "loss": 0.7676,
+      "step": 8147
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6669319064565034e-05,
+      "loss": 0.7842,
+      "step": 8148
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6668461248539316e-05,
+      "loss": 0.7715,
+      "step": 8149
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6667603344140865e-05,
+      "loss": 0.7578,
+      "step": 8150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6666745351381048e-05,
+      "loss": 0.7627,
+      "step": 8151
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666588727027123e-05,
+      "loss": 0.7646,
+      "step": 8152
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6665029100822786e-05,
+      "loss": 0.7656,
+      "step": 8153
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666417084304709e-05,
+      "loss": 0.7559,
+      "step": 8154
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6663312496955517e-05,
+      "loss": 0.7461,
+      "step": 8155
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666245406255944e-05,
+      "loss": 0.7578,
+      "step": 8156
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6661595539870235e-05,
+      "loss": 0.7744,
+      "step": 8157
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6660736928899278e-05,
+      "loss": 0.7373,
+      "step": 8158
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665987822965795e-05,
+      "loss": 0.7578,
+      "step": 8159
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6659019442157632e-05,
+      "loss": 0.7578,
+      "step": 8160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6658160566409702e-05,
+      "loss": 0.752,
+      "step": 8161
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6657301602425543e-05,
+      "loss": 0.7754,
+      "step": 8162
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6656442550216536e-05,
+      "loss": 0.7432,
+      "step": 8163
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6655583409794073e-05,
+      "loss": 0.7441,
+      "step": 8164
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665472418116953e-05,
+      "loss": 0.748,
+      "step": 8165
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.66538648643543e-05,
+      "loss": 0.7637,
+      "step": 8166
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6653005459359772e-05,
+      "loss": 0.7568,
+      "step": 8167
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665214596619733e-05,
+      "loss": 0.7656,
+      "step": 8168
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6651286384878367e-05,
+      "loss": 0.7598,
+      "step": 8169
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6650426715414277e-05,
+      "loss": 0.7637,
+      "step": 8170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649566957816445e-05,
+      "loss": 0.7275,
+      "step": 8171
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6648707112096274e-05,
+      "loss": 0.7773,
+      "step": 8172
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6647847178265153e-05,
+      "loss": 0.7646,
+      "step": 8173
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664698715633448e-05,
+      "loss": 0.749,
+      "step": 8174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6646127046315653e-05,
+      "loss": 0.751,
+      "step": 8175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664526684822007e-05,
+      "loss": 0.7686,
+      "step": 8176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6644406562059126e-05,
+      "loss": 0.7383,
+      "step": 8177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664354618784423e-05,
+      "loss": 0.7656,
+      "step": 8178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6642685725586778e-05,
+      "loss": 0.7559,
+      "step": 8179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6641825175298177e-05,
+      "loss": 0.7676,
+      "step": 8180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664096453698983e-05,
+      "loss": 0.7373,
+      "step": 8181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6640103810673138e-05,
+      "loss": 0.7588,
+      "step": 8182
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6639242996359516e-05,
+      "loss": 0.7275,
+      "step": 8183
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6638382094060363e-05,
+      "loss": 0.7441,
+      "step": 8184
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6637521103787095e-05,
+      "loss": 0.7754,
+      "step": 8185
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6636660025551118e-05,
+      "loss": 0.8008,
+      "step": 8186
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.663579885936385e-05,
+      "loss": 0.7324,
+      "step": 8187
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6634937605236693e-05,
+      "loss": 0.7607,
+      "step": 8188
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6634076263181067e-05,
+      "loss": 0.7451,
+      "step": 8189
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6633214833208384e-05,
+      "loss": 0.7598,
+      "step": 8190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6632353315330063e-05,
+      "loss": 0.7471,
+      "step": 8191
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.663149170955752e-05,
+      "loss": 0.7529,
+      "step": 8192
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6630630015902175e-05,
+      "loss": 0.7549,
+      "step": 8193
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6629768234375447e-05,
+      "loss": 0.7588,
+      "step": 8194
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662890636498875e-05,
+      "loss": 0.7656,
+      "step": 8195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6628044407753513e-05,
+      "loss": 0.749,
+      "step": 8196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6627182362681158e-05,
+      "loss": 0.7432,
+      "step": 8197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662632022978311e-05,
+      "loss": 0.7793,
+      "step": 8198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6625458009070788e-05,
+      "loss": 0.7822,
+      "step": 8199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6624595700555624e-05,
+      "loss": 0.7559,
+      "step": 8200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6623733304249046e-05,
+      "loss": 0.7637,
+      "step": 8201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622870820162478e-05,
+      "loss": 0.7852,
+      "step": 8202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622008248307356e-05,
+      "loss": 0.7715,
+      "step": 8203
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6621145588695107e-05,
+      "loss": 0.7568,
+      "step": 8204
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6620282841337164e-05,
+      "loss": 0.7617,
+      "step": 8205
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6619420006244964e-05,
+      "loss": 0.7803,
+      "step": 8206
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6618557083429933e-05,
+      "loss": 0.7539,
+      "step": 8207
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6617694072903518e-05,
+      "loss": 0.7764,
+      "step": 8208
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661683097467715e-05,
+      "loss": 0.7617,
+      "step": 8209
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6615967788762262e-05,
+      "loss": 0.791,
+      "step": 8210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66151045151703e-05,
+      "loss": 0.7656,
+      "step": 8211
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6614241153912703e-05,
+      "loss": 0.7617,
+      "step": 8212
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6613377705000914e-05,
+      "loss": 0.7666,
+      "step": 8213
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6612514168446375e-05,
+      "loss": 0.7539,
+      "step": 8214
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6611650544260528e-05,
+      "loss": 0.7422,
+      "step": 8215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6610786832454814e-05,
+      "loss": 0.7432,
+      "step": 8216
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660992303304069e-05,
+      "loss": 0.7539,
+      "step": 8217
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66090591460296e-05,
+      "loss": 0.7539,
+      "step": 8218
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6608195171432985e-05,
+      "loss": 0.7383,
+      "step": 8219
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66073311092623e-05,
+      "loss": 0.7529,
+      "step": 8220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6606466959528997e-05,
+      "loss": 0.7588,
+      "step": 8221
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6605602722244528e-05,
+      "loss": 0.752,
+      "step": 8222
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660473839742034e-05,
+      "loss": 0.7402,
+      "step": 8223
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66038739850679e-05,
+      "loss": 0.7715,
+      "step": 8224
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6603009485198648e-05,
+      "loss": 0.7617,
+      "step": 8225
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6602144897824053e-05,
+      "loss": 0.7549,
+      "step": 8226
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6601280222955567e-05,
+      "loss": 0.7432,
+      "step": 8227
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660041546060465e-05,
+      "loss": 0.7607,
+      "step": 8228
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659955061078276e-05,
+      "loss": 0.7578,
+      "step": 8229
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659868567350136e-05,
+      "loss": 0.7402,
+      "step": 8230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659782064877192e-05,
+      "loss": 0.7803,
+      "step": 8231
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6596955536605893e-05,
+      "loss": 0.7461,
+      "step": 8232
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6596090337014743e-05,
+      "loss": 0.79,
+      "step": 8233
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6595225050009945e-05,
+      "loss": 0.751,
+      "step": 8234
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659435967560296e-05,
+      "loss": 0.7393,
+      "step": 8235
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6593494213805258e-05,
+      "loss": 0.7461,
+      "step": 8236
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659262866462831e-05,
+      "loss": 0.7559,
+      "step": 8237
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6591763028083582e-05,
+      "loss": 0.748,
+      "step": 8238
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6590897304182548e-05,
+      "loss": 0.7471,
+      "step": 8239
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659003149293668e-05,
+      "loss": 0.7393,
+      "step": 8240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6589165594357454e-05,
+      "loss": 0.7344,
+      "step": 8241
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6588299608456345e-05,
+      "loss": 0.7578,
+      "step": 8242
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658743353524483e-05,
+      "loss": 0.749,
+      "step": 8243
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658656737473438e-05,
+      "loss": 0.7412,
+      "step": 8244
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6585701126936485e-05,
+      "loss": 0.7412,
+      "step": 8245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6584834791862616e-05,
+      "loss": 0.7559,
+      "step": 8246
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6583968369524254e-05,
+      "loss": 0.7676,
+      "step": 8247
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658310185993289e-05,
+      "loss": 0.7656,
+      "step": 8248
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6582235263099998e-05,
+      "loss": 0.7676,
+      "step": 8249
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6581368579037062e-05,
+      "loss": 0.793,
+      "step": 8250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6580501807755576e-05,
+      "loss": 0.7227,
+      "step": 8251
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6579634949267018e-05,
+      "loss": 0.75,
+      "step": 8252
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6578768003582882e-05,
+      "loss": 0.7461,
+      "step": 8253
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6577900970714654e-05,
+      "loss": 0.7656,
+      "step": 8254
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6577033850673824e-05,
+      "loss": 0.7627,
+      "step": 8255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6576166643471885e-05,
+      "loss": 0.7461,
+      "step": 8256
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657529934912033e-05,
+      "loss": 0.751,
+      "step": 8257
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6574431967630652e-05,
+      "loss": 0.7451,
+      "step": 8258
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6573564499014345e-05,
+      "loss": 0.7959,
+      "step": 8259
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572696943282904e-05,
+      "loss": 0.7646,
+      "step": 8260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6571829300447827e-05,
+      "loss": 0.7559,
+      "step": 8261
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570961570520615e-05,
+      "loss": 0.7695,
+      "step": 8262
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570093753512765e-05,
+      "loss": 0.7617,
+      "step": 8263
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6569225849435775e-05,
+      "loss": 0.7686,
+      "step": 8264
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6568357858301153e-05,
+      "loss": 0.7715,
+      "step": 8265
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6567489780120397e-05,
+      "loss": 0.7607,
+      "step": 8266
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656662161490501e-05,
+      "loss": 0.7383,
+      "step": 8267
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6565753362666502e-05,
+      "loss": 0.7539,
+      "step": 8268
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6564885023416382e-05,
+      "loss": 0.7715,
+      "step": 8269
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656401659716615e-05,
+      "loss": 0.7725,
+      "step": 8270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6563148083927313e-05,
+      "loss": 0.7461,
+      "step": 8271
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656227948371139e-05,
+      "loss": 0.7441,
+      "step": 8272
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6561410796529886e-05,
+      "loss": 0.7734,
+      "step": 8273
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656054202239431e-05,
+      "loss": 0.7598,
+      "step": 8274
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6559673161316188e-05,
+      "loss": 0.7285,
+      "step": 8275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6558804213307022e-05,
+      "loss": 0.7637,
+      "step": 8276
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6557935178378334e-05,
+      "loss": 0.7441,
+      "step": 8277
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6557066056541633e-05,
+      "loss": 0.7773,
+      "step": 8278
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6556196847808448e-05,
+      "loss": 0.7363,
+      "step": 8279
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6555327552190286e-05,
+      "loss": 0.7568,
+      "step": 8280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655445816969868e-05,
+      "loss": 0.749,
+      "step": 8281
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6553588700345144e-05,
+      "loss": 0.75,
+      "step": 8282
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65527191441412e-05,
+      "loss": 0.7568,
+      "step": 8283
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6551849501098373e-05,
+      "loss": 0.749,
+      "step": 8284
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550979771228188e-05,
+      "loss": 0.7773,
+      "step": 8285
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550109954542174e-05,
+      "loss": 0.7373,
+      "step": 8286
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654924005105185e-05,
+      "loss": 0.7432,
+      "step": 8287
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6548370060768753e-05,
+      "loss": 0.7529,
+      "step": 8288
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6547499983704408e-05,
+      "loss": 0.7373,
+      "step": 8289
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6546629819870347e-05,
+      "loss": 0.7598,
+      "step": 8290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65457595692781e-05,
+      "loss": 0.751,
+      "step": 8291
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65448892319392e-05,
+      "loss": 0.7568,
+      "step": 8292
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6544018807865184e-05,
+      "loss": 0.791,
+      "step": 8293
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6543148297067582e-05,
+      "loss": 0.7725,
+      "step": 8294
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6542277699557934e-05,
+      "loss": 0.7373,
+      "step": 8295
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654140701534778e-05,
+      "loss": 0.7803,
+      "step": 8296
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6540536244448654e-05,
+      "loss": 0.7539,
+      "step": 8297
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.653966538687209e-05,
+      "loss": 0.748,
+      "step": 8298
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6538794442629645e-05,
+      "loss": 0.7812,
+      "step": 8299
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653792341173285e-05,
+      "loss": 0.7471,
+      "step": 8300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653705229419325e-05,
+      "loss": 0.7783,
+      "step": 8301
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653618109002239e-05,
+      "loss": 0.7578,
+      "step": 8302
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6535309799231817e-05,
+      "loss": 0.7451,
+      "step": 8303
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653443842183307e-05,
+      "loss": 0.7783,
+      "step": 8304
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653356695783771e-05,
+      "loss": 0.7588,
+      "step": 8305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6532695407257274e-05,
+      "loss": 0.7686,
+      "step": 8306
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6531823770103314e-05,
+      "loss": 0.7412,
+      "step": 8307
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653095204638739e-05,
+      "loss": 0.7656,
+      "step": 8308
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6530080236121044e-05,
+      "loss": 0.7988,
+      "step": 8309
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6529208339315835e-05,
+      "loss": 0.7666,
+      "step": 8310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6528336355983313e-05,
+      "loss": 0.75,
+      "step": 8311
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6527464286135043e-05,
+      "loss": 0.7529,
+      "step": 8312
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652659212978257e-05,
+      "loss": 0.7539,
+      "step": 8313
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6525719886937464e-05,
+      "loss": 0.7637,
+      "step": 8314
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652484755761128e-05,
+      "loss": 0.7607,
+      "step": 8315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652397514181557e-05,
+      "loss": 0.7422,
+      "step": 8316
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6523102639561907e-05,
+      "loss": 0.7793,
+      "step": 8317
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6522230050861853e-05,
+      "loss": 0.7373,
+      "step": 8318
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652135737572696e-05,
+      "loss": 0.7725,
+      "step": 8319
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6520484614168806e-05,
+      "loss": 0.7764,
+      "step": 8320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651961176619895e-05,
+      "loss": 0.7979,
+      "step": 8321
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6518738831828962e-05,
+      "loss": 0.7344,
+      "step": 8322
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651786581107041e-05,
+      "loss": 0.7598,
+      "step": 8323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6516992703934865e-05,
+      "loss": 0.7627,
+      "step": 8324
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6516119510433893e-05,
+      "loss": 0.7451,
+      "step": 8325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6515246230579075e-05,
+      "loss": 0.7803,
+      "step": 8326
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6514372864381973e-05,
+      "loss": 0.7881,
+      "step": 8327
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6513499411854166e-05,
+      "loss": 0.7656,
+      "step": 8328
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6512625873007233e-05,
+      "loss": 0.7441,
+      "step": 8329
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6511752247852742e-05,
+      "loss": 0.7578,
+      "step": 8330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651087853640228e-05,
+      "loss": 0.749,
+      "step": 8331
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651000473866742e-05,
+      "loss": 0.7461,
+      "step": 8332
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6509130854659747e-05,
+      "loss": 0.7695,
+      "step": 8333
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650825688439083e-05,
+      "loss": 0.7715,
+      "step": 8334
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6507382827872264e-05,
+      "loss": 0.7598,
+      "step": 8335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6506508685115627e-05,
+      "loss": 0.7617,
+      "step": 8336
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6505634456132504e-05,
+      "loss": 0.7686,
+      "step": 8337
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650476014093448e-05,
+      "loss": 0.7441,
+      "step": 8338
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6503885739533142e-05,
+      "loss": 0.7471,
+      "step": 8339
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6503011251940078e-05,
+      "loss": 0.7402,
+      "step": 8340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650213667816688e-05,
+      "loss": 0.7617,
+      "step": 8341
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6501262018225137e-05,
+      "loss": 0.752,
+      "step": 8342
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6500387272126434e-05,
+      "loss": 0.7705,
+      "step": 8343
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649951243988237e-05,
+      "loss": 0.7676,
+      "step": 8344
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6498637521504538e-05,
+      "loss": 0.7598,
+      "step": 8345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6497762517004526e-05,
+      "loss": 0.7383,
+      "step": 8346
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6496887426393945e-05,
+      "loss": 0.7412,
+      "step": 8347
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6496012249684373e-05,
+      "loss": 0.7451,
+      "step": 8348
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649513698688742e-05,
+      "loss": 0.7354,
+      "step": 8349
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649426163801468e-05,
+      "loss": 0.7471,
+      "step": 8350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6493386203077764e-05,
+      "loss": 0.7324,
+      "step": 8351
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649251068208826e-05,
+      "loss": 0.75,
+      "step": 8352
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6491635075057777e-05,
+      "loss": 0.7705,
+      "step": 8353
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649075938199792e-05,
+      "loss": 0.7285,
+      "step": 8354
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6489883602920288e-05,
+      "loss": 0.7734,
+      "step": 8355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6489007737836495e-05,
+      "loss": 0.7295,
+      "step": 8356
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6488131786758147e-05,
+      "loss": 0.7715,
+      "step": 8357
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6487255749696848e-05,
+      "loss": 0.7354,
+      "step": 8358
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6486379626664205e-05,
+      "loss": 0.749,
+      "step": 8359
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6485503417671836e-05,
+      "loss": 0.7656,
+      "step": 8360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6484627122731352e-05,
+      "loss": 0.7725,
+      "step": 8361
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6483750741854362e-05,
+      "loss": 0.75,
+      "step": 8362
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6482874275052482e-05,
+      "loss": 0.7383,
+      "step": 8363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648199772233733e-05,
+      "loss": 0.7578,
+      "step": 8364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6481121083720518e-05,
+      "loss": 0.7646,
+      "step": 8365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6480244359213665e-05,
+      "loss": 0.75,
+      "step": 8366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647936754882839e-05,
+      "loss": 0.7402,
+      "step": 8367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6478490652576315e-05,
+      "loss": 0.752,
+      "step": 8368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6477613670469053e-05,
+      "loss": 0.7559,
+      "step": 8369
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6476736602518238e-05,
+      "loss": 0.7568,
+      "step": 8370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6475859448735484e-05,
+      "loss": 0.7646,
+      "step": 8371
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6474982209132416e-05,
+      "loss": 0.7383,
+      "step": 8372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6474104883720665e-05,
+      "loss": 0.7578,
+      "step": 8373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647322747251185e-05,
+      "loss": 0.7461,
+      "step": 8374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647234997551761e-05,
+      "loss": 0.749,
+      "step": 8375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647147239274956e-05,
+      "loss": 0.7607,
+      "step": 8376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647059472421934e-05,
+      "loss": 0.7676,
+      "step": 8377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646971696993858e-05,
+      "loss": 0.7373,
+      "step": 8378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646883912991891e-05,
+      "loss": 0.7637,
+      "step": 8379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646796120417196e-05,
+      "loss": 0.7783,
+      "step": 8380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6467083192709374e-05,
+      "loss": 0.7627,
+      "step": 8381
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646620509554278e-05,
+      "loss": 0.7617,
+      "step": 8382
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6465326912683816e-05,
+      "loss": 0.7559,
+      "step": 8383
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6464448644144124e-05,
+      "loss": 0.752,
+      "step": 8384
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6463570289935336e-05,
+      "loss": 0.7627,
+      "step": 8385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6462691850069096e-05,
+      "loss": 0.7236,
+      "step": 8386
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646181332455705e-05,
+      "loss": 0.7412,
+      "step": 8387
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6460934713410833e-05,
+      "loss": 0.7568,
+      "step": 8388
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6460056016642095e-05,
+      "loss": 0.7637,
+      "step": 8389
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6459177234262473e-05,
+      "loss": 0.7959,
+      "step": 8390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6458298366283622e-05,
+      "loss": 0.7539,
+      "step": 8391
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645741941271718e-05,
+      "loss": 0.7578,
+      "step": 8392
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6456540373574808e-05,
+      "loss": 0.7705,
+      "step": 8393
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645566124886814e-05,
+      "loss": 0.7363,
+      "step": 8394
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6454782038608837e-05,
+      "loss": 0.7607,
+      "step": 8395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645390274280855e-05,
+      "loss": 0.7568,
+      "step": 8396
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6453023361478923e-05,
+      "loss": 0.7305,
+      "step": 8397
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6452143894631617e-05,
+      "loss": 0.7686,
+      "step": 8398
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6451264342278287e-05,
+      "loss": 0.7695,
+      "step": 8399
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6450384704430587e-05,
+      "loss": 0.7441,
+      "step": 8400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644950498110018e-05,
+      "loss": 0.749,
+      "step": 8401
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6448625172298714e-05,
+      "loss": 0.7471,
+      "step": 8402
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6447745278037855e-05,
+      "loss": 0.7363,
+      "step": 8403
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6446865298329267e-05,
+      "loss": 0.7686,
+      "step": 8404
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6445985233184606e-05,
+      "loss": 0.7705,
+      "step": 8405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6445105082615535e-05,
+      "loss": 0.7598,
+      "step": 8406
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644422484663372e-05,
+      "loss": 0.7314,
+      "step": 8407
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6443344525250825e-05,
+      "loss": 0.7412,
+      "step": 8408
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6442464118478522e-05,
+      "loss": 0.7607,
+      "step": 8409
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644158362632847e-05,
+      "loss": 0.7529,
+      "step": 8410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644070304881234e-05,
+      "loss": 0.7568,
+      "step": 8411
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6439822385941802e-05,
+      "loss": 0.7314,
+      "step": 8412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6438941637728533e-05,
+      "loss": 0.7354,
+      "step": 8413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6438060804184195e-05,
+      "loss": 0.7363,
+      "step": 8414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6437179885320468e-05,
+      "loss": 0.7607,
+      "step": 8415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6436298881149024e-05,
+      "loss": 0.8096,
+      "step": 8416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6435417791681535e-05,
+      "loss": 0.7568,
+      "step": 8417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6434536616929686e-05,
+      "loss": 0.7422,
+      "step": 8418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6433655356905147e-05,
+      "loss": 0.7637,
+      "step": 8419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.64327740116196e-05,
+      "loss": 0.7529,
+      "step": 8420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6431892581084723e-05,
+      "loss": 0.7656,
+      "step": 8421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6431011065312202e-05,
+      "loss": 0.7432,
+      "step": 8422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643012946431371e-05,
+      "loss": 0.7529,
+      "step": 8423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642924777810094e-05,
+      "loss": 0.7598,
+      "step": 8424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642836600668557e-05,
+      "loss": 0.7461,
+      "step": 8425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6427484150079288e-05,
+      "loss": 0.7324,
+      "step": 8426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642660220829378e-05,
+      "loss": 0.7559,
+      "step": 8427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6425720181340734e-05,
+      "loss": 0.7461,
+      "step": 8428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642483806923184e-05,
+      "loss": 0.7363,
+      "step": 8429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6423955871978787e-05,
+      "loss": 0.7598,
+      "step": 8430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6423073589593266e-05,
+      "loss": 0.75,
+      "step": 8431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642219122208697e-05,
+      "loss": 0.7412,
+      "step": 8432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642130876947159e-05,
+      "loss": 0.7637,
+      "step": 8433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6420426231758823e-05,
+      "loss": 0.7441,
+      "step": 8434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641954360896037e-05,
+      "loss": 0.75,
+      "step": 8435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6418660901087916e-05,
+      "loss": 0.7363,
+      "step": 8436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6417778108153165e-05,
+      "loss": 0.7295,
+      "step": 8437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641689523016782e-05,
+      "loss": 0.7402,
+      "step": 8438
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6416012267143573e-05,
+      "loss": 0.7617,
+      "step": 8439
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641512921909213e-05,
+      "loss": 0.7529,
+      "step": 8440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6414246086025193e-05,
+      "loss": 0.7451,
+      "step": 8441
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641336286795447e-05,
+      "loss": 0.7617,
+      "step": 8442
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6412479564891658e-05,
+      "loss": 0.7568,
+      "step": 8443
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6411596176848465e-05,
+      "loss": 0.748,
+      "step": 8444
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.64107127038366e-05,
+      "loss": 0.7617,
+      "step": 8445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6409829145867774e-05,
+      "loss": 0.7676,
+      "step": 8446
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6408945502953688e-05,
+      "loss": 0.749,
+      "step": 8447
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640806177510606e-05,
+      "loss": 0.7451,
+      "step": 8448
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6407177962336596e-05,
+      "loss": 0.7744,
+      "step": 8449
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640629406465701e-05,
+      "loss": 0.7695,
+      "step": 8450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640541008207902e-05,
+      "loss": 0.75,
+      "step": 8451
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6404526014614333e-05,
+      "loss": 0.752,
+      "step": 8452
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6403641862274674e-05,
+      "loss": 0.75,
+      "step": 8453
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6402757625071752e-05,
+      "loss": 0.7539,
+      "step": 8454
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6401873303017288e-05,
+      "loss": 0.7754,
+      "step": 8455
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6400988896123008e-05,
+      "loss": 0.7529,
+      "step": 8456
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640010440440062e-05,
+      "loss": 0.7676,
+      "step": 8457
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6399219827861857e-05,
+      "loss": 0.7383,
+      "step": 8458
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639833516651843e-05,
+      "loss": 0.7334,
+      "step": 8459
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6397450420382076e-05,
+      "loss": 0.7568,
+      "step": 8460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6396565589464515e-05,
+      "loss": 0.7598,
+      "step": 8461
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639568067377747e-05,
+      "loss": 0.7617,
+      "step": 8462
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6394795673332664e-05,
+      "loss": 0.752,
+      "step": 8463
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6393910588141836e-05,
+      "loss": 0.7666,
+      "step": 8464
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639302541821671e-05,
+      "loss": 0.7568,
+      "step": 8465
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6392140163569016e-05,
+      "loss": 0.7656,
+      "step": 8466
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639125482421049e-05,
+      "loss": 0.7578,
+      "step": 8467
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6390369400152862e-05,
+      "loss": 0.7939,
+      "step": 8468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638948389140786e-05,
+      "loss": 0.75,
+      "step": 8469
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638859829798723e-05,
+      "loss": 0.7617,
+      "step": 8470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6387712619902704e-05,
+      "loss": 0.7344,
+      "step": 8471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638682685716601e-05,
+      "loss": 0.7725,
+      "step": 8472
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6385941009788903e-05,
+      "loss": 0.751,
+      "step": 8473
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638505507778311e-05,
+      "loss": 0.7676,
+      "step": 8474
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6384169061160376e-05,
+      "loss": 0.7578,
+      "step": 8475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6383282959932448e-05,
+      "loss": 0.7559,
+      "step": 8476
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638239677411106e-05,
+      "loss": 0.7451,
+      "step": 8477
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638151050370796e-05,
+      "loss": 0.7305,
+      "step": 8478
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6380624148734892e-05,
+      "loss": 0.7451,
+      "step": 8479
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6379737709203603e-05,
+      "loss": 0.751,
+      "step": 8480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637885118512584e-05,
+      "loss": 0.7324,
+      "step": 8481
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6377964576513354e-05,
+      "loss": 0.7344,
+      "step": 8482
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637707788337789e-05,
+      "loss": 0.7275,
+      "step": 8483
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6376191105731206e-05,
+      "loss": 0.7598,
+      "step": 8484
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6375304243585046e-05,
+      "loss": 0.7998,
+      "step": 8485
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6374417296951168e-05,
+      "loss": 0.749,
+      "step": 8486
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637353026584132e-05,
+      "loss": 0.7812,
+      "step": 8487
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6372643150267267e-05,
+      "loss": 0.7549,
+      "step": 8488
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637175595024076e-05,
+      "loss": 0.7383,
+      "step": 8489
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6370868665773553e-05,
+      "loss": 0.749,
+      "step": 8490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636998129687741e-05,
+      "loss": 0.7852,
+      "step": 8491
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6369093843564086e-05,
+      "loss": 0.7764,
+      "step": 8492
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636820630584535e-05,
+      "loss": 0.7441,
+      "step": 8493
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6367318683732955e-05,
+      "loss": 0.7705,
+      "step": 8494
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6366430977238667e-05,
+      "loss": 0.7715,
+      "step": 8495
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6365543186374254e-05,
+      "loss": 0.7461,
+      "step": 8496
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6364655311151474e-05,
+      "loss": 0.7568,
+      "step": 8497
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6363767351582103e-05,
+      "loss": 0.7598,
+      "step": 8498
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63628793076779e-05,
+      "loss": 0.7461,
+      "step": 8499
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6361991179450637e-05,
+      "loss": 0.7686,
+      "step": 8500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6361102966912084e-05,
+      "loss": 0.7725,
+      "step": 8501
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6360214670074013e-05,
+      "loss": 0.751,
+      "step": 8502
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6359326288948192e-05,
+      "loss": 0.752,
+      "step": 8503
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6358437823546398e-05,
+      "loss": 0.7773,
+      "step": 8504
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6357549273880404e-05,
+      "loss": 0.7461,
+      "step": 8505
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6356660639961983e-05,
+      "loss": 0.7881,
+      "step": 8506
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6355771921802918e-05,
+      "loss": 0.7334,
+      "step": 8507
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6354883119414978e-05,
+      "loss": 0.7402,
+      "step": 8508
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635399423280995e-05,
+      "loss": 0.7471,
+      "step": 8509
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6353105261999608e-05,
+      "loss": 0.7549,
+      "step": 8510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6352216206995733e-05,
+      "loss": 0.7705,
+      "step": 8511
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635132706781011e-05,
+      "loss": 0.7656,
+      "step": 8512
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6350437844454524e-05,
+      "loss": 0.7773,
+      "step": 8513
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6349548536940753e-05,
+      "loss": 0.751,
+      "step": 8514
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6348659145280587e-05,
+      "loss": 0.7666,
+      "step": 8515
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634776966948581e-05,
+      "loss": 0.7588,
+      "step": 8516
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634688010956821e-05,
+      "loss": 0.7676,
+      "step": 8517
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6345990465539582e-05,
+      "loss": 0.7461,
+      "step": 8518
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634510073741171e-05,
+      "loss": 0.7568,
+      "step": 8519
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634421092519638e-05,
+      "loss": 0.7725,
+      "step": 8520
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634332102890539e-05,
+      "loss": 0.7539,
+      "step": 8521
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634243104855054e-05,
+      "loss": 0.7627,
+      "step": 8522
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634154098414361e-05,
+      "loss": 0.7588,
+      "step": 8523
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6340650835696405e-05,
+      "loss": 0.7607,
+      "step": 8524
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633976060322072e-05,
+      "loss": 0.7334,
+      "step": 8525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633887028672835e-05,
+      "loss": 0.7461,
+      "step": 8526
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6337979886231095e-05,
+      "loss": 0.7441,
+      "step": 8527
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633708940174076e-05,
+      "loss": 0.7588,
+      "step": 8528
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633619883326914e-05,
+      "loss": 0.7715,
+      "step": 8529
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6335308180828037e-05,
+      "loss": 0.75,
+      "step": 8530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6334417444429257e-05,
+      "loss": 0.7295,
+      "step": 8531
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6333526624084602e-05,
+      "loss": 0.751,
+      "step": 8532
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6332635719805877e-05,
+      "loss": 0.7637,
+      "step": 8533
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6331744731604896e-05,
+      "loss": 0.752,
+      "step": 8534
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6330853659493457e-05,
+      "loss": 0.7588,
+      "step": 8535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632996250348337e-05,
+      "loss": 0.7764,
+      "step": 8536
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6329071263586455e-05,
+      "loss": 0.7588,
+      "step": 8537
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632817993981451e-05,
+      "loss": 0.7402,
+      "step": 8538
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6327288532179355e-05,
+      "loss": 0.7617,
+      "step": 8539
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63263970406928e-05,
+      "loss": 0.7607,
+      "step": 8540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632550546536666e-05,
+      "loss": 0.7637,
+      "step": 8541
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632461380621275e-05,
+      "loss": 0.7646,
+      "step": 8542
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632372206324289e-05,
+      "loss": 0.7568,
+      "step": 8543
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6322830236468896e-05,
+      "loss": 0.7461,
+      "step": 8544
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6321938325902582e-05,
+      "loss": 0.7656,
+      "step": 8545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6321046331555774e-05,
+      "loss": 0.7588,
+      "step": 8546
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632015425344029e-05,
+      "loss": 0.7324,
+      "step": 8547
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6319262091567955e-05,
+      "loss": 0.7656,
+      "step": 8548
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631836984595059e-05,
+      "loss": 0.7314,
+      "step": 8549
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6317477516600015e-05,
+      "loss": 0.7637,
+      "step": 8550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631658510352806e-05,
+      "loss": 0.7607,
+      "step": 8551
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6315692606746558e-05,
+      "loss": 0.7402,
+      "step": 8552
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6314800026267327e-05,
+      "loss": 0.7705,
+      "step": 8553
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6313907362102196e-05,
+      "loss": 0.7734,
+      "step": 8554
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6313014614263004e-05,
+      "loss": 0.752,
+      "step": 8555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631212178276157e-05,
+      "loss": 0.7568,
+      "step": 8556
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6311228867609736e-05,
+      "loss": 0.7383,
+      "step": 8557
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631033586881933e-05,
+      "loss": 0.7529,
+      "step": 8558
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6309442786402185e-05,
+      "loss": 0.7627,
+      "step": 8559
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630854962037014e-05,
+      "loss": 0.7715,
+      "step": 8560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6307656370735035e-05,
+      "loss": 0.7432,
+      "step": 8561
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6306763037508696e-05,
+      "loss": 0.7441,
+      "step": 8562
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6305869620702973e-05,
+      "loss": 0.7568,
+      "step": 8563
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6304976120329702e-05,
+      "loss": 0.7461,
+      "step": 8564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6304082536400723e-05,
+      "loss": 0.748,
+      "step": 8565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630318886892788e-05,
+      "loss": 0.7246,
+      "step": 8566
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6302295117923013e-05,
+      "loss": 0.7422,
+      "step": 8567
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630140128339797e-05,
+      "loss": 0.7559,
+      "step": 8568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6300507365364592e-05,
+      "loss": 0.7754,
+      "step": 8569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6299613363834733e-05,
+      "loss": 0.7568,
+      "step": 8570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6298719278820233e-05,
+      "loss": 0.7559,
+      "step": 8571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6297825110332943e-05,
+      "loss": 0.7568,
+      "step": 8572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6296930858384712e-05,
+      "loss": 0.7686,
+      "step": 8573
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6296036522987394e-05,
+      "loss": 0.751,
+      "step": 8574
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6295142104152842e-05,
+      "loss": 0.7334,
+      "step": 8575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6294247601892904e-05,
+      "loss": 0.7832,
+      "step": 8576
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6293353016219438e-05,
+      "loss": 0.7656,
+      "step": 8577
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6292458347144296e-05,
+      "loss": 0.7363,
+      "step": 8578
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629156359467934e-05,
+      "loss": 0.75,
+      "step": 8579
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629066875883642e-05,
+      "loss": 0.7412,
+      "step": 8580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.62897738396274e-05,
+      "loss": 0.7432,
+      "step": 8581
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628887883706414e-05,
+      "loss": 0.7275,
+      "step": 8582
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6287983751158498e-05,
+      "loss": 0.7686,
+      "step": 8583
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628708858192234e-05,
+      "loss": 0.7568,
+      "step": 8584
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6286193329367522e-05,
+      "loss": 0.7617,
+      "step": 8585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6285297993505916e-05,
+      "loss": 0.7617,
+      "step": 8586
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6284402574349384e-05,
+      "loss": 0.7422,
+      "step": 8587
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628350707190979e-05,
+      "loss": 0.7539,
+      "step": 8588
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6282611486199008e-05,
+      "loss": 0.7695,
+      "step": 8589
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.62817158172289e-05,
+      "loss": 0.7578,
+      "step": 8590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6280820065011337e-05,
+      "loss": 0.7539,
+      "step": 8591
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6279924229558192e-05,
+      "loss": 0.7559,
+      "step": 8592
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6279028310881337e-05,
+      "loss": 0.7549,
+      "step": 8593
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6278132308992645e-05,
+      "loss": 0.7441,
+      "step": 8594
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6277236223903987e-05,
+      "loss": 0.75,
+      "step": 8595
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6276340055627244e-05,
+      "loss": 0.7412,
+      "step": 8596
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6275443804174284e-05,
+      "loss": 0.7695,
+      "step": 8597
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627454746955699e-05,
+      "loss": 0.7305,
+      "step": 8598
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627365105178724e-05,
+      "loss": 0.7471,
+      "step": 8599
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6272754550876915e-05,
+      "loss": 0.7695,
+      "step": 8600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6271857966837895e-05,
+      "loss": 0.752,
+      "step": 8601
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270961299682056e-05,
+      "loss": 0.7666,
+      "step": 8602
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627006454942129e-05,
+      "loss": 0.7588,
+      "step": 8603
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6269167716067474e-05,
+      "loss": 0.752,
+      "step": 8604
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.62682707996325e-05,
+      "loss": 0.7422,
+      "step": 8605
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626737380012825e-05,
+      "loss": 0.7197,
+      "step": 8606
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626647671756661e-05,
+      "loss": 0.7607,
+      "step": 8607
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6265579551959464e-05,
+      "loss": 0.7588,
+      "step": 8608
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6264682303318715e-05,
+      "loss": 0.7373,
+      "step": 8609
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6263784971656245e-05,
+      "loss": 0.7705,
+      "step": 8610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6262887556983947e-05,
+      "loss": 0.7441,
+      "step": 8611
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6261990059313713e-05,
+      "loss": 0.7676,
+      "step": 8612
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6261092478657435e-05,
+      "loss": 0.7617,
+      "step": 8613
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626019481502702e-05,
+      "loss": 0.752,
+      "step": 8614
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6259297068434344e-05,
+      "loss": 0.7646,
+      "step": 8615
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6258399238891322e-05,
+      "loss": 0.7539,
+      "step": 8616
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6257501326409843e-05,
+      "loss": 0.7412,
+      "step": 8617
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6256603331001808e-05,
+      "loss": 0.7412,
+      "step": 8618
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625570525267912e-05,
+      "loss": 0.7539,
+      "step": 8619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6254807091453678e-05,
+      "loss": 0.7549,
+      "step": 8620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6253908847337393e-05,
+      "loss": 0.7568,
+      "step": 8621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625301052034215e-05,
+      "loss": 0.7725,
+      "step": 8622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6252112110479875e-05,
+      "loss": 0.7822,
+      "step": 8623
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625121361776246e-05,
+      "loss": 0.7764,
+      "step": 8624
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625031504220182e-05,
+      "loss": 0.7471,
+      "step": 8625
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6249416383809857e-05,
+      "loss": 0.7422,
+      "step": 8626
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6248517642598488e-05,
+      "loss": 0.7725,
+      "step": 8627
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6247618818579612e-05,
+      "loss": 0.752,
+      "step": 8628
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6246719911765153e-05,
+      "loss": 0.7754,
+      "step": 8629
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6245820922167017e-05,
+      "loss": 0.79,
+      "step": 8630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6244921849797116e-05,
+      "loss": 0.7539,
+      "step": 8631
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624402269466737e-05,
+      "loss": 0.7578,
+      "step": 8632
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6243123456789693e-05,
+      "loss": 0.7773,
+      "step": 8633
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6242224136176e-05,
+      "loss": 0.7295,
+      "step": 8634
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624132473283821e-05,
+      "loss": 0.7295,
+      "step": 8635
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6240425246788246e-05,
+      "loss": 0.7695,
+      "step": 8636
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.623952567803802e-05,
+      "loss": 0.7695,
+      "step": 8637
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6238626026599466e-05,
+      "loss": 0.7607,
+      "step": 8638
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623772629248449e-05,
+      "loss": 0.7578,
+      "step": 8639
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623682647570503e-05,
+      "loss": 0.7383,
+      "step": 8640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6235926576273003e-05,
+      "loss": 0.7656,
+      "step": 8641
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6235026594200342e-05,
+      "loss": 0.7754,
+      "step": 8642
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6234126529498963e-05,
+      "loss": 0.749,
+      "step": 8643
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233226382180802e-05,
+      "loss": 0.7451,
+      "step": 8644
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6232326152257786e-05,
+      "loss": 0.7637,
+      "step": 8645
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6231425839741844e-05,
+      "loss": 0.791,
+      "step": 8646
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623052544464491e-05,
+      "loss": 0.7637,
+      "step": 8647
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622962496697891e-05,
+      "loss": 0.7783,
+      "step": 8648
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622872440675579e-05,
+      "loss": 0.7588,
+      "step": 8649
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622782376398747e-05,
+      "loss": 0.7676,
+      "step": 8650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6226923038685894e-05,
+      "loss": 0.7734,
+      "step": 8651
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6226022230862998e-05,
+      "loss": 0.7363,
+      "step": 8652
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6225121340530717e-05,
+      "loss": 0.75,
+      "step": 8653
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6224220367700995e-05,
+      "loss": 0.7402,
+      "step": 8654
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6223319312385768e-05,
+      "loss": 0.7852,
+      "step": 8655
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6222418174596975e-05,
+      "loss": 0.7354,
+      "step": 8656
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6221516954346564e-05,
+      "loss": 0.7725,
+      "step": 8657
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622061565164647e-05,
+      "loss": 0.7715,
+      "step": 8658
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621971426650865e-05,
+      "loss": 0.7539,
+      "step": 8659
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6218812798945038e-05,
+      "loss": 0.7637,
+      "step": 8660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6217911248967587e-05,
+      "loss": 0.7402,
+      "step": 8661
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621700961658824e-05,
+      "loss": 0.751,
+      "step": 8662
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6216107901818948e-05,
+      "loss": 0.7783,
+      "step": 8663
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6215206104671663e-05,
+      "loss": 0.75,
+      "step": 8664
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621430422515833e-05,
+      "loss": 0.75,
+      "step": 8665
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6213402263290907e-05,
+      "loss": 0.7305,
+      "step": 8666
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6212500219081343e-05,
+      "loss": 0.7559,
+      "step": 8667
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.62115980925416e-05,
+      "loss": 0.7646,
+      "step": 8668
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6210695883683623e-05,
+      "loss": 0.752,
+      "step": 8669
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6209793592519373e-05,
+      "loss": 0.7471,
+      "step": 8670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6208891219060805e-05,
+      "loss": 0.7559,
+      "step": 8671
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6207988763319887e-05,
+      "loss": 0.749,
+      "step": 8672
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6207086225308563e-05,
+      "loss": 0.7637,
+      "step": 8673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6206183605038806e-05,
+      "loss": 0.7744,
+      "step": 8674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6205280902522574e-05,
+      "loss": 0.752,
+      "step": 8675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6204378117771834e-05,
+      "loss": 0.7705,
+      "step": 8676
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620347525079854e-05,
+      "loss": 0.7607,
+      "step": 8677
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6202572301614666e-05,
+      "loss": 0.7598,
+      "step": 8678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6201669270232173e-05,
+      "loss": 0.7607,
+      "step": 8679
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6200766156663032e-05,
+      "loss": 0.7451,
+      "step": 8680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6199862960919214e-05,
+      "loss": 0.7744,
+      "step": 8681
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6198959683012677e-05,
+      "loss": 0.7432,
+      "step": 8682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6198056322955404e-05,
+      "loss": 0.7676,
+      "step": 8683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619715288075936e-05,
+      "loss": 0.752,
+      "step": 8684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619624935643652e-05,
+      "loss": 0.7861,
+      "step": 8685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6195345749998856e-05,
+      "loss": 0.7646,
+      "step": 8686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6194442061458343e-05,
+      "loss": 0.7275,
+      "step": 8687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619353829082696e-05,
+      "loss": 0.7471,
+      "step": 8688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619263443811668e-05,
+      "loss": 0.7354,
+      "step": 8689
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6191730503339488e-05,
+      "loss": 0.7305,
+      "step": 8690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6190826486507355e-05,
+      "loss": 0.7529,
+      "step": 8691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6189922387632268e-05,
+      "loss": 0.7529,
+      "step": 8692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.61890182067262e-05,
+      "loss": 0.7207,
+      "step": 8693
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6188113943801143e-05,
+      "loss": 0.7549,
+      "step": 8694
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6187209598869073e-05,
+      "loss": 0.75,
+      "step": 8695
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6186305171941977e-05,
+      "loss": 0.7578,
+      "step": 8696
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6185400663031847e-05,
+      "loss": 0.748,
+      "step": 8697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6184496072150663e-05,
+      "loss": 0.7734,
+      "step": 8698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6183591399310414e-05,
+      "loss": 0.7578,
+      "step": 8699
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6182686644523092e-05,
+      "loss": 0.748,
+      "step": 8700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618178180780068e-05,
+      "loss": 0.7373,
+      "step": 8701
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6180876889155178e-05,
+      "loss": 0.7637,
+      "step": 8702
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6179971888598577e-05,
+      "loss": 0.749,
+      "step": 8703
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617906680614286e-05,
+      "loss": 0.7705,
+      "step": 8704
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6178161641800036e-05,
+      "loss": 0.7607,
+      "step": 8705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6177256395582087e-05,
+      "loss": 0.752,
+      "step": 8706
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6176351067501024e-05,
+      "loss": 0.751,
+      "step": 8707
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6175445657568832e-05,
+      "loss": 0.7637,
+      "step": 8708
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6174540165797516e-05,
+      "loss": 0.7656,
+      "step": 8709
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6173634592199077e-05,
+      "loss": 0.7725,
+      "step": 8710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6172728936785512e-05,
+      "loss": 0.748,
+      "step": 8711
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6171823199568822e-05,
+      "loss": 0.793,
+      "step": 8712
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6170917380561015e-05,
+      "loss": 0.75,
+      "step": 8713
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6170011479774097e-05,
+      "loss": 0.7383,
+      "step": 8714
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6169105497220064e-05,
+      "loss": 0.7471,
+      "step": 8715
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616819943291093e-05,
+      "loss": 0.7646,
+      "step": 8716
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6167293286858703e-05,
+      "loss": 0.7432,
+      "step": 8717
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6166387059075387e-05,
+      "loss": 0.75,
+      "step": 8718
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616548074957299e-05,
+      "loss": 0.7734,
+      "step": 8719
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616457435836353e-05,
+      "loss": 0.7715,
+      "step": 8720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163667885459012e-05,
+      "loss": 0.7637,
+      "step": 8721
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6162761330871456e-05,
+      "loss": 0.7812,
+      "step": 8722
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616185469461287e-05,
+      "loss": 0.7715,
+      "step": 8723
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160947976695275e-05,
+      "loss": 0.7539,
+      "step": 8724
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160041177130676e-05,
+      "loss": 0.7734,
+      "step": 8725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61591342959311e-05,
+      "loss": 0.7578,
+      "step": 8726
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6158227333108566e-05,
+      "loss": 0.7412,
+      "step": 8727
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6157320288675086e-05,
+      "loss": 0.7559,
+      "step": 8728
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6156413162642685e-05,
+      "loss": 0.7666,
+      "step": 8729
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6155505955023385e-05,
+      "loss": 0.7432,
+      "step": 8730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615459866582921e-05,
+      "loss": 0.7373,
+      "step": 8731
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6153691295072177e-05,
+      "loss": 0.7529,
+      "step": 8732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6152783842764317e-05,
+      "loss": 0.791,
+      "step": 8733
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6151876308917654e-05,
+      "loss": 0.7568,
+      "step": 8734
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6150968693544215e-05,
+      "loss": 0.7383,
+      "step": 8735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615006099665603e-05,
+      "loss": 0.7852,
+      "step": 8736
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6149153218265123e-05,
+      "loss": 0.7676,
+      "step": 8737
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6148245358383528e-05,
+      "loss": 0.749,
+      "step": 8738
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6147337417023277e-05,
+      "loss": 0.7559,
+      "step": 8739
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61464293941964e-05,
+      "loss": 0.7471,
+      "step": 8740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6145521289914933e-05,
+      "loss": 0.7578,
+      "step": 8741
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6144613104190904e-05,
+      "loss": 0.7432,
+      "step": 8742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614370483703636e-05,
+      "loss": 0.7871,
+      "step": 8743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6142796488463324e-05,
+      "loss": 0.7422,
+      "step": 8744
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6141888058483846e-05,
+      "loss": 0.7803,
+      "step": 8745
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614097954710996e-05,
+      "loss": 0.7598,
+      "step": 8746
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6140070954353705e-05,
+      "loss": 0.75,
+      "step": 8747
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.613916228022712e-05,
+      "loss": 0.7871,
+      "step": 8748
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6138253524742253e-05,
+      "loss": 0.7344,
+      "step": 8749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6137344687911143e-05,
+      "loss": 0.7568,
+      "step": 8750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6136435769745835e-05,
+      "loss": 0.75,
+      "step": 8751
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613552677025837e-05,
+      "loss": 0.7607,
+      "step": 8752
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6134617689460807e-05,
+      "loss": 0.7646,
+      "step": 8753
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6133708527365176e-05,
+      "loss": 0.7441,
+      "step": 8754
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6132799283983543e-05,
+      "loss": 0.7441,
+      "step": 8755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6131889959327944e-05,
+      "loss": 0.75,
+      "step": 8756
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6130980553410437e-05,
+      "loss": 0.7646,
+      "step": 8757
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613007106624307e-05,
+      "loss": 0.7676,
+      "step": 8758
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61291614978379e-05,
+      "loss": 0.7637,
+      "step": 8759
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6128251848206974e-05,
+      "loss": 0.7607,
+      "step": 8760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6127342117362353e-05,
+      "loss": 0.7471,
+      "step": 8761
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61264323053161e-05,
+      "loss": 0.7695,
+      "step": 8762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6125522412080252e-05,
+      "loss": 0.7852,
+      "step": 8763
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6124612437666885e-05,
+      "loss": 0.7578,
+      "step": 8764
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612370238208805e-05,
+      "loss": 0.7451,
+      "step": 8765
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612279224535581e-05,
+      "loss": 0.7529,
+      "step": 8766
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6121882027482224e-05,
+      "loss": 0.7373,
+      "step": 8767
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612097172847936e-05,
+      "loss": 0.7773,
+      "step": 8768
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612006134835928e-05,
+      "loss": 0.7695,
+      "step": 8769
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611915088713404e-05,
+      "loss": 0.7656,
+      "step": 8770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611824034481572e-05,
+      "loss": 0.75,
+      "step": 8771
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6117329721416374e-05,
+      "loss": 0.749,
+      "step": 8772
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6116419016948075e-05,
+      "loss": 0.7412,
+      "step": 8773
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6115508231422893e-05,
+      "loss": 0.7432,
+      "step": 8774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61145973648529e-05,
+      "loss": 0.7783,
+      "step": 8775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6113686417250164e-05,
+      "loss": 0.7441,
+      "step": 8776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6112775388626754e-05,
+      "loss": 0.7549,
+      "step": 8777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611186427899475e-05,
+      "loss": 0.7646,
+      "step": 8778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611095308836622e-05,
+      "loss": 0.7559,
+      "step": 8779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6110041816753246e-05,
+      "loss": 0.7695,
+      "step": 8780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61091304641679e-05,
+      "loss": 0.7598,
+      "step": 8781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6108219030622262e-05,
+      "loss": 0.7637,
+      "step": 8782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6107307516128403e-05,
+      "loss": 0.748,
+      "step": 8783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6106395920698415e-05,
+      "loss": 0.7432,
+      "step": 8784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610548424434437e-05,
+      "loss": 0.7539,
+      "step": 8785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6104572487078353e-05,
+      "loss": 0.7598,
+      "step": 8786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610366064891245e-05,
+      "loss": 0.7539,
+      "step": 8787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6102748729858738e-05,
+      "loss": 0.7734,
+      "step": 8788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101836729929305e-05,
+      "loss": 0.7539,
+      "step": 8789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6100924649136238e-05,
+      "loss": 0.7461,
+      "step": 8790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6100012487491623e-05,
+      "loss": 0.7412,
+      "step": 8791
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6099100245007553e-05,
+      "loss": 0.749,
+      "step": 8792
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6098187921696112e-05,
+      "loss": 0.7578,
+      "step": 8793
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609727551756939e-05,
+      "loss": 0.7412,
+      "step": 8794
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6096363032639482e-05,
+      "loss": 0.7637,
+      "step": 8795
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6095450466918478e-05,
+      "loss": 0.7578,
+      "step": 8796
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6094537820418472e-05,
+      "loss": 0.7686,
+      "step": 8797
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609362509315156e-05,
+      "loss": 0.7607,
+      "step": 8798
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609271228512984e-05,
+      "loss": 0.7598,
+      "step": 8799
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6091799396365403e-05,
+      "loss": 0.7334,
+      "step": 8800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6090886426870352e-05,
+      "loss": 0.7559,
+      "step": 8801
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089973376656778e-05,
+      "loss": 0.7451,
+      "step": 8802
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089060245736793e-05,
+      "loss": 0.7363,
+      "step": 8803
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6088147034122486e-05,
+      "loss": 0.7246,
+      "step": 8804
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6087233741825974e-05,
+      "loss": 0.7676,
+      "step": 8805
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6086320368859345e-05,
+      "loss": 0.7354,
+      "step": 8806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6085406915234712e-05,
+      "loss": 0.752,
+      "step": 8807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608449338096418e-05,
+      "loss": 0.7559,
+      "step": 8808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608357976605985e-05,
+      "loss": 0.7861,
+      "step": 8809
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608266607053383e-05,
+      "loss": 0.793,
+      "step": 8810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608175229439824e-05,
+      "loss": 0.7637,
+      "step": 8811
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6080838437665177e-05,
+      "loss": 0.75,
+      "step": 8812
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6079924500346757e-05,
+      "loss": 0.7568,
+      "step": 8813
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607901048245509e-05,
+      "loss": 0.7422,
+      "step": 8814
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6078096384002292e-05,
+      "loss": 0.7656,
+      "step": 8815
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6077182205000476e-05,
+      "loss": 0.7539,
+      "step": 8816
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607626794546175e-05,
+      "loss": 0.7549,
+      "step": 8817
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6075353605398242e-05,
+      "loss": 0.7637,
+      "step": 8818
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607443918482206e-05,
+      "loss": 0.748,
+      "step": 8819
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6073524683745327e-05,
+      "loss": 0.7705,
+      "step": 8820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6072610102180163e-05,
+      "loss": 0.7783,
+      "step": 8821
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607169544013868e-05,
+      "loss": 0.7354,
+      "step": 8822
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607078069763301e-05,
+      "loss": 0.7695,
+      "step": 8823
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606986587467527e-05,
+      "loss": 0.7871,
+      "step": 8824
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606895097127758e-05,
+      "loss": 0.7578,
+      "step": 8825
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6068035987452072e-05,
+      "loss": 0.7354,
+      "step": 8826
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606712092321087e-05,
+      "loss": 0.7559,
+      "step": 8827
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6066205778566098e-05,
+      "loss": 0.7451,
+      "step": 8828
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6065290553529888e-05,
+      "loss": 0.748,
+      "step": 8829
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6064375248114365e-05,
+      "loss": 0.7422,
+      "step": 8830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6063459862331657e-05,
+      "loss": 0.7617,
+      "step": 8831
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.60625443961939e-05,
+      "loss": 0.7871,
+      "step": 8832
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6061628849713226e-05,
+      "loss": 0.7686,
+      "step": 8833
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6060713222901764e-05,
+      "loss": 0.7637,
+      "step": 8834
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605979751577165e-05,
+      "loss": 0.7471,
+      "step": 8835
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605888172833502e-05,
+      "loss": 0.7705,
+      "step": 8836
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6057965860604013e-05,
+      "loss": 0.7705,
+      "step": 8837
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605704991259076e-05,
+      "loss": 0.7578,
+      "step": 8838
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6056133884307405e-05,
+      "loss": 0.7461,
+      "step": 8839
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6055217775766087e-05,
+      "loss": 0.7266,
+      "step": 8840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605430158697894e-05,
+      "loss": 0.7441,
+      "step": 8841
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6053385317958113e-05,
+      "loss": 0.7549,
+      "step": 8842
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6052468968715747e-05,
+      "loss": 0.7471,
+      "step": 8843
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6051552539263986e-05,
+      "loss": 0.7461,
+      "step": 8844
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6050636029614972e-05,
+      "loss": 0.75,
+      "step": 8845
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6049719439780857e-05,
+      "loss": 0.7393,
+      "step": 8846
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604880276977378e-05,
+      "loss": 0.752,
+      "step": 8847
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6047886019605893e-05,
+      "loss": 0.7451,
+      "step": 8848
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6046969189289345e-05,
+      "loss": 0.7588,
+      "step": 8849
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6046052278836288e-05,
+      "loss": 0.7578,
+      "step": 8850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604513528825887e-05,
+      "loss": 0.7637,
+      "step": 8851
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6044218217569246e-05,
+      "loss": 0.7461,
+      "step": 8852
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6043301066779565e-05,
+      "loss": 0.7432,
+      "step": 8853
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604238383590199e-05,
+      "loss": 0.7803,
+      "step": 8854
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6041466524948665e-05,
+      "loss": 0.7539,
+      "step": 8855
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6040549133931756e-05,
+      "loss": 0.7666,
+      "step": 8856
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6039631662863413e-05,
+      "loss": 0.7539,
+      "step": 8857
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6038714111755804e-05,
+      "loss": 0.7617,
+      "step": 8858
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6037796480621077e-05,
+      "loss": 0.7324,
+      "step": 8859
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6036878769471403e-05,
+      "loss": 0.7607,
+      "step": 8860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603596097831894e-05,
+      "loss": 0.7324,
+      "step": 8861
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6035043107175847e-05,
+      "loss": 0.7793,
+      "step": 8862
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6034125156054295e-05,
+      "loss": 0.748,
+      "step": 8863
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6033207124966447e-05,
+      "loss": 0.7568,
+      "step": 8864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6032289013924465e-05,
+      "loss": 0.748,
+      "step": 8865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603137082294052e-05,
+      "loss": 0.7539,
+      "step": 8866
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030452552026775e-05,
+      "loss": 0.749,
+      "step": 8867
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602953420119541e-05,
+      "loss": 0.7715,
+      "step": 8868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6028615770458586e-05,
+      "loss": 0.7412,
+      "step": 8869
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6027697259828476e-05,
+      "loss": 0.749,
+      "step": 8870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6026778669317256e-05,
+      "loss": 0.7686,
+      "step": 8871
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602585999893709e-05,
+      "loss": 0.751,
+      "step": 8872
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6024941248700168e-05,
+      "loss": 0.7578,
+      "step": 8873
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6024022418618653e-05,
+      "loss": 0.7344,
+      "step": 8874
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6023103508704726e-05,
+      "loss": 0.7324,
+      "step": 8875
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6022184518970565e-05,
+      "loss": 0.7695,
+      "step": 8876
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6021265449428346e-05,
+      "loss": 0.7559,
+      "step": 8877
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602034630009025e-05,
+      "loss": 0.7578,
+      "step": 8878
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6019427070968464e-05,
+      "loss": 0.7275,
+      "step": 8879
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6018507762075158e-05,
+      "loss": 0.7812,
+      "step": 8880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601758837342253e-05,
+      "loss": 0.7676,
+      "step": 8881
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601666890502275e-05,
+      "loss": 0.7402,
+      "step": 8882
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601574935688801e-05,
+      "loss": 0.7314,
+      "step": 8883
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601482972903049e-05,
+      "loss": 0.7334,
+      "step": 8884
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601391002146239e-05,
+      "loss": 0.7471,
+      "step": 8885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601299023419589e-05,
+      "loss": 0.7324,
+      "step": 8886
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6012070367243176e-05,
+      "loss": 0.748,
+      "step": 8887
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6011150420616443e-05,
+      "loss": 0.7334,
+      "step": 8888
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6010230394327884e-05,
+      "loss": 0.748,
+      "step": 8889
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6009310288389687e-05,
+      "loss": 0.7568,
+      "step": 8890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600839010281405e-05,
+      "loss": 0.7588,
+      "step": 8891
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6007469837613166e-05,
+      "loss": 0.7676,
+      "step": 8892
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600654949279923e-05,
+      "loss": 0.751,
+      "step": 8893
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6005629068384437e-05,
+      "loss": 0.7451,
+      "step": 8894
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6004708564380985e-05,
+      "loss": 0.7402,
+      "step": 8895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6003787980801075e-05,
+      "loss": 0.75,
+      "step": 8896
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600286731765691e-05,
+      "loss": 0.7578,
+      "step": 8897
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6001946574960682e-05,
+      "loss": 0.7441,
+      "step": 8898
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6001025752724604e-05,
+      "loss": 0.7578,
+      "step": 8899
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600010485096087e-05,
+      "loss": 0.7539,
+      "step": 8900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5999183869681688e-05,
+      "loss": 0.7617,
+      "step": 8901
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599826280889926e-05,
+      "loss": 0.7441,
+      "step": 8902
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.59973416686258e-05,
+      "loss": 0.7393,
+      "step": 8903
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5996420448873506e-05,
+      "loss": 0.7412,
+      "step": 8904
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5995499149654594e-05,
+      "loss": 0.7559,
+      "step": 8905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599457777098127e-05,
+      "loss": 0.7725,
+      "step": 8906
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599365631286574e-05,
+      "loss": 0.7363,
+      "step": 8907
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599273477532022e-05,
+      "loss": 0.7588,
+      "step": 8908
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5991813158356927e-05,
+      "loss": 0.7754,
+      "step": 8909
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5990891461988065e-05,
+      "loss": 0.7568,
+      "step": 8910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5989969686225856e-05,
+      "loss": 0.7852,
+      "step": 8911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598904783108251e-05,
+      "loss": 0.752,
+      "step": 8912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598812589657025e-05,
+      "loss": 0.749,
+      "step": 8913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5987203882701288e-05,
+      "loss": 0.7549,
+      "step": 8914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598628178948785e-05,
+      "loss": 0.7588,
+      "step": 8915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5985359616942145e-05,
+      "loss": 0.749,
+      "step": 8916
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5984437365076405e-05,
+      "loss": 0.7539,
+      "step": 8917
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5983515033902847e-05,
+      "loss": 0.7588,
+      "step": 8918
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5982592623433695e-05,
+      "loss": 0.7461,
+      "step": 8919
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598167013368117e-05,
+      "loss": 0.7607,
+      "step": 8920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5980747564657502e-05,
+      "loss": 0.7734,
+      "step": 8921
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5979824916374915e-05,
+      "loss": 0.7471,
+      "step": 8922
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978902188845632e-05,
+      "loss": 0.7432,
+      "step": 8923
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5977979382081886e-05,
+      "loss": 0.7705,
+      "step": 8924
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597705649609591e-05,
+      "loss": 0.7529,
+      "step": 8925
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5976133530899927e-05,
+      "loss": 0.7461,
+      "step": 8926
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5975210486506173e-05,
+      "loss": 0.7441,
+      "step": 8927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597428736292688e-05,
+      "loss": 0.751,
+      "step": 8928
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5973364160174276e-05,
+      "loss": 0.7451,
+      "step": 8929
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5972440878260605e-05,
+      "loss": 0.7666,
+      "step": 8930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59715175171981e-05,
+      "loss": 0.7441,
+      "step": 8931
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970594076998987e-05,
+      "loss": 0.7373,
+      "step": 8932
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5969670557675517e-05,
+      "loss": 0.7686,
+      "step": 8933
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5968746959239928e-05,
+      "loss": 0.7695,
+      "step": 8934
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5967823281704452e-05,
+      "loss": 0.8018,
+      "step": 8935
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5966899525081333e-05,
+      "loss": 0.7549,
+      "step": 8936
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5965975689382817e-05,
+      "loss": 0.7783,
+      "step": 8937
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596505177462114e-05,
+      "loss": 0.7568,
+      "step": 8938
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5964127780808553e-05,
+      "loss": 0.7471,
+      "step": 8939
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5963203707957295e-05,
+      "loss": 0.7451,
+      "step": 8940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5962279556079617e-05,
+      "loss": 0.7461,
+      "step": 8941
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5961355325187762e-05,
+      "loss": 0.7461,
+      "step": 8942
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596043101529398e-05,
+      "loss": 0.7334,
+      "step": 8943
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5959506626410526e-05,
+      "loss": 0.7422,
+      "step": 8944
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5958582158549644e-05,
+      "loss": 0.7852,
+      "step": 8945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5957657611723583e-05,
+      "loss": 0.7354,
+      "step": 8946
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5956732985944602e-05,
+      "loss": 0.7402,
+      "step": 8947
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5955808281224946e-05,
+      "loss": 0.7705,
+      "step": 8948
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595488349757688e-05,
+      "loss": 0.7432,
+      "step": 8949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5953958635012654e-05,
+      "loss": 0.752,
+      "step": 8950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5953033693544524e-05,
+      "loss": 0.7432,
+      "step": 8951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595210867318475e-05,
+      "loss": 0.7432,
+      "step": 8952
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5951183573945582e-05,
+      "loss": 0.7549,
+      "step": 8953
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5950258395839293e-05,
+      "loss": 0.7695,
+      "step": 8954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5949333138878136e-05,
+      "loss": 0.749,
+      "step": 8955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594840780307438e-05,
+      "loss": 0.7402,
+      "step": 8956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5947482388440277e-05,
+      "loss": 0.7725,
+      "step": 8957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5946556894988097e-05,
+      "loss": 0.7305,
+      "step": 8958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5945631322730104e-05,
+      "loss": 0.7441,
+      "step": 8959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5944705671678567e-05,
+      "loss": 0.7549,
+      "step": 8960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594377994184575e-05,
+      "loss": 0.75,
+      "step": 8961
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594285413324392e-05,
+      "loss": 0.7627,
+      "step": 8962
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5941928245885344e-05,
+      "loss": 0.7617,
+      "step": 8963
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59410022797823e-05,
+      "loss": 0.75,
+      "step": 8964
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5940076234947056e-05,
+      "loss": 0.7412,
+      "step": 8965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5939150111391878e-05,
+      "loss": 0.7793,
+      "step": 8966
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593822390912905e-05,
+      "loss": 0.749,
+      "step": 8967
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593729762817084e-05,
+      "loss": 0.7607,
+      "step": 8968
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5936371268529525e-05,
+      "loss": 0.7412,
+      "step": 8969
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593544483021738e-05,
+      "loss": 0.7412,
+      "step": 8970
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593451831324668e-05,
+      "loss": 0.7393,
+      "step": 8971
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593359171762971e-05,
+      "loss": 0.7471,
+      "step": 8972
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5932665043378744e-05,
+      "loss": 0.79,
+      "step": 8973
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5931738290506066e-05,
+      "loss": 0.7656,
+      "step": 8974
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593081145902396e-05,
+      "loss": 0.7295,
+      "step": 8975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.59298845489447e-05,
+      "loss": 0.7988,
+      "step": 8976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5928957560280577e-05,
+      "loss": 0.7529,
+      "step": 8977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5928030493043875e-05,
+      "loss": 0.7412,
+      "step": 8978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5927103347246876e-05,
+      "loss": 0.7422,
+      "step": 8979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5926176122901873e-05,
+      "loss": 0.7217,
+      "step": 8980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5925248820021147e-05,
+      "loss": 0.7871,
+      "step": 8981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592432143861699e-05,
+      "loss": 0.7344,
+      "step": 8982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5923393978701693e-05,
+      "loss": 0.7539,
+      "step": 8983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592246644028755e-05,
+      "loss": 0.7285,
+      "step": 8984
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5921538823386844e-05,
+      "loss": 0.7568,
+      "step": 8985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5920611128011877e-05,
+      "loss": 0.7656,
+      "step": 8986
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5919683354174937e-05,
+      "loss": 0.7588,
+      "step": 8987
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5918755501888323e-05,
+      "loss": 0.7793,
+      "step": 8988
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591782757116433e-05,
+      "loss": 0.75,
+      "step": 8989
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5916899562015255e-05,
+      "loss": 0.7451,
+      "step": 8990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5915971474453395e-05,
+      "loss": 0.7373,
+      "step": 8991
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5915043308491053e-05,
+      "loss": 0.7451,
+      "step": 8992
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5914115064140525e-05,
+      "loss": 0.7607,
+      "step": 8993
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5913186741414116e-05,
+      "loss": 0.7568,
+      "step": 8994
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591225834032413e-05,
+      "loss": 0.748,
+      "step": 8995
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591132986088286e-05,
+      "loss": 0.7861,
+      "step": 8996
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5910401303102624e-05,
+      "loss": 0.7803,
+      "step": 8997
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5909472666995722e-05,
+      "loss": 0.752,
+      "step": 8998
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908543952574458e-05,
+      "loss": 0.7529,
+      "step": 8999
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5907615159851144e-05,
+      "loss": 0.748,
+      "step": 9000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5906686288838082e-05,
+      "loss": 0.7393,
+      "step": 9001
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5905757339547592e-05,
+      "loss": 0.7422,
+      "step": 9002
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5904828311991973e-05,
+      "loss": 0.7637,
+      "step": 9003
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5903899206183546e-05,
+      "loss": 0.7295,
+      "step": 9004
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5902970022134622e-05,
+      "loss": 0.749,
+      "step": 9005
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5902040759857513e-05,
+      "loss": 0.7676,
+      "step": 9006
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590111141936453e-05,
+      "loss": 0.7354,
+      "step": 9007
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5900182000668e-05,
+      "loss": 0.7686,
+      "step": 9008
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589925250378023e-05,
+      "loss": 0.7812,
+      "step": 9009
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5898322928713544e-05,
+      "loss": 0.7627,
+      "step": 9010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5897393275480253e-05,
+      "loss": 0.7656,
+      "step": 9011
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589646354409269e-05,
+      "loss": 0.7598,
+      "step": 9012
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589553373456316e-05,
+      "loss": 0.7383,
+      "step": 9013
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5894603846904e-05,
+      "loss": 0.7539,
+      "step": 9014
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5893673881127524e-05,
+      "loss": 0.7627,
+      "step": 9015
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589274383724606e-05,
+      "loss": 0.7412,
+      "step": 9016
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891813715271932e-05,
+      "loss": 0.7539,
+      "step": 9017
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5890883515217464e-05,
+      "loss": 0.7539,
+      "step": 9018
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588995323709499e-05,
+      "loss": 0.749,
+      "step": 9019
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5889022880916837e-05,
+      "loss": 0.7598,
+      "step": 9020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5888092446695326e-05,
+      "loss": 0.7793,
+      "step": 9021
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5887161934442796e-05,
+      "loss": 0.7588,
+      "step": 9022
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5886231344171576e-05,
+      "loss": 0.7588,
+      "step": 9023
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5885300675893998e-05,
+      "loss": 0.7529,
+      "step": 9024
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5884369929622395e-05,
+      "loss": 0.7705,
+      "step": 9025
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5883439105369104e-05,
+      "loss": 0.7432,
+      "step": 9026
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5882508203146457e-05,
+      "loss": 0.7607,
+      "step": 9027
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5881577222966794e-05,
+      "loss": 0.7393,
+      "step": 9028
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5880646164842448e-05,
+      "loss": 0.752,
+      "step": 9029
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5879715028785765e-05,
+      "loss": 0.7656,
+      "step": 9030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5878783814809078e-05,
+      "loss": 0.7412,
+      "step": 9031
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.7441,
+      "step": 9032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5876921153145066e-05,
+      "loss": 0.7529,
+      "step": 9033
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5875989705482427e-05,
+      "loss": 0.7686,
+      "step": 9034
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5875058179949153e-05,
+      "loss": 0.752,
+      "step": 9035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5874126576557593e-05,
+      "loss": 0.7461,
+      "step": 9036
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5873194895320093e-05,
+      "loss": 0.7334,
+      "step": 9037
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5872263136248996e-05,
+      "loss": 0.7432,
+      "step": 9038
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5871331299356653e-05,
+      "loss": 0.748,
+      "step": 9039
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5870399384655412e-05,
+      "loss": 0.749,
+      "step": 9040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5869467392157627e-05,
+      "loss": 0.75,
+      "step": 9041
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5868535321875645e-05,
+      "loss": 0.7588,
+      "step": 9042
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586760317382182e-05,
+      "loss": 0.7646,
+      "step": 9043
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58666709480085e-05,
+      "loss": 0.749,
+      "step": 9044
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5865738644448047e-05,
+      "loss": 0.7627,
+      "step": 9045
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586480626315281e-05,
+      "loss": 0.7607,
+      "step": 9046
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586387380413515e-05,
+      "loss": 0.7432,
+      "step": 9047
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5862941267407423e-05,
+      "loss": 0.7656,
+      "step": 9048
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586200865298198e-05,
+      "loss": 0.7646,
+      "step": 9049
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5861075960871192e-05,
+      "loss": 0.75,
+      "step": 9050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5860143191087413e-05,
+      "loss": 0.7666,
+      "step": 9051
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5859210343643004e-05,
+      "loss": 0.7305,
+      "step": 9052
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585827741855033e-05,
+      "loss": 0.7979,
+      "step": 9053
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5857344415821754e-05,
+      "loss": 0.748,
+      "step": 9054
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5856411335469638e-05,
+      "loss": 0.7393,
+      "step": 9055
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585547817750635e-05,
+      "loss": 0.7227,
+      "step": 9056
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585454494194426e-05,
+      "loss": 0.7549,
+      "step": 9057
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5853611628795725e-05,
+      "loss": 0.7559,
+      "step": 9058
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585267823807312e-05,
+      "loss": 0.7598,
+      "step": 9059
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5851744769788818e-05,
+      "loss": 0.7148,
+      "step": 9060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5850811223955183e-05,
+      "loss": 0.7852,
+      "step": 9061
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5849877600584594e-05,
+      "loss": 0.7334,
+      "step": 9062
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5848943899689415e-05,
+      "loss": 0.749,
+      "step": 9063
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584801012128203e-05,
+      "loss": 0.791,
+      "step": 9064
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5847076265374803e-05,
+      "loss": 0.7891,
+      "step": 9065
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846142331980116e-05,
+      "loss": 0.7559,
+      "step": 9066
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5845208321110343e-05,
+      "loss": 0.749,
+      "step": 9067
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5844274232777867e-05,
+      "loss": 0.7305,
+      "step": 9068
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584334006699506e-05,
+      "loss": 0.7578,
+      "step": 9069
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5842405823774306e-05,
+      "loss": 0.7754,
+      "step": 9070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5841471503127984e-05,
+      "loss": 0.748,
+      "step": 9071
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584053710506848e-05,
+      "loss": 0.7373,
+      "step": 9072
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5839602629608174e-05,
+      "loss": 0.7334,
+      "step": 9073
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5838668076759448e-05,
+      "loss": 0.7451,
+      "step": 9074
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5837733446534688e-05,
+      "loss": 0.7119,
+      "step": 9075
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5836798738946283e-05,
+      "loss": 0.7617,
+      "step": 9076
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5835863954006616e-05,
+      "loss": 0.7559,
+      "step": 9077
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583492909172808e-05,
+      "loss": 0.751,
+      "step": 9078
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583399415212306e-05,
+      "loss": 0.7598,
+      "step": 9079
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583305913520395e-05,
+      "loss": 0.7373,
+      "step": 9080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5832124040983133e-05,
+      "loss": 0.7373,
+      "step": 9081
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5831188869473007e-05,
+      "loss": 0.7373,
+      "step": 9082
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5830253620685972e-05,
+      "loss": 0.749,
+      "step": 9083
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582931829463441e-05,
+      "loss": 0.7529,
+      "step": 9084
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582838289133072e-05,
+      "loss": 0.748,
+      "step": 9085
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5827447410787305e-05,
+      "loss": 0.7646,
+      "step": 9086
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5826511853016552e-05,
+      "loss": 0.75,
+      "step": 9087
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582557621803087e-05,
+      "loss": 0.7871,
+      "step": 9088
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.582464050584265e-05,
+      "loss": 0.7607,
+      "step": 9089
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5823704716464292e-05,
+      "loss": 0.7549,
+      "step": 9090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5822768849908203e-05,
+      "loss": 0.7676,
+      "step": 9091
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5821832906186783e-05,
+      "loss": 0.7783,
+      "step": 9092
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5820896885312435e-05,
+      "loss": 0.7529,
+      "step": 9093
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5819960787297563e-05,
+      "loss": 0.7461,
+      "step": 9094
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5819024612154578e-05,
+      "loss": 0.7275,
+      "step": 9095
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5818088359895877e-05,
+      "loss": 0.749,
+      "step": 9096
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5817152030533874e-05,
+      "loss": 0.7539,
+      "step": 9097
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5816215624080977e-05,
+      "loss": 0.7559,
+      "step": 9098
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5815279140549593e-05,
+      "loss": 0.7568,
+      "step": 9099
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5814342579952136e-05,
+      "loss": 0.7666,
+      "step": 9100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5813405942301013e-05,
+      "loss": 0.7607,
+      "step": 9101
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581246922760864e-05,
+      "loss": 0.7578,
+      "step": 9102
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5811532435887434e-05,
+      "loss": 0.7588,
+      "step": 9103
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5810595567149802e-05,
+      "loss": 0.7383,
+      "step": 9104
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5809658621408165e-05,
+      "loss": 0.7627,
+      "step": 9105
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580872159867494e-05,
+      "loss": 0.7344,
+      "step": 9106
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5807784498962536e-05,
+      "loss": 0.7461,
+      "step": 9107
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5806847322283386e-05,
+      "loss": 0.7471,
+      "step": 9108
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.58059100686499e-05,
+      "loss": 0.7393,
+      "step": 9109
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.58049727380745e-05,
+      "loss": 0.7607,
+      "step": 9110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5804035330569613e-05,
+      "loss": 0.7529,
+      "step": 9111
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5803097846147655e-05,
+      "loss": 0.7744,
+      "step": 9112
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5802160284821052e-05,
+      "loss": 0.7559,
+      "step": 9113
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.580122264660223e-05,
+      "loss": 0.749,
+      "step": 9114
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5800284931503618e-05,
+      "loss": 0.7598,
+      "step": 9115
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5799347139537634e-05,
+      "loss": 0.7285,
+      "step": 9116
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5798409270716716e-05,
+      "loss": 0.7578,
+      "step": 9117
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5797471325053288e-05,
+      "loss": 0.7217,
+      "step": 9118
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796533302559783e-05,
+      "loss": 0.7725,
+      "step": 9119
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5795595203248623e-05,
+      "loss": 0.7412,
+      "step": 9120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579465702713225e-05,
+      "loss": 0.7441,
+      "step": 9121
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5793718774223093e-05,
+      "loss": 0.751,
+      "step": 9122
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5792780444533588e-05,
+      "loss": 0.7412,
+      "step": 9123
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579184203807617e-05,
+      "loss": 0.7725,
+      "step": 9124
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5790903554863267e-05,
+      "loss": 0.7373,
+      "step": 9125
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578996499490733e-05,
+      "loss": 0.749,
+      "step": 9126
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5789026358220786e-05,
+      "loss": 0.7529,
+      "step": 9127
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5788087644816076e-05,
+      "loss": 0.7705,
+      "step": 9128
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5787148854705647e-05,
+      "loss": 0.7354,
+      "step": 9129
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5786209987901932e-05,
+      "loss": 0.7285,
+      "step": 9130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5785271044417378e-05,
+      "loss": 0.7705,
+      "step": 9131
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578433202426443e-05,
+      "loss": 0.7451,
+      "step": 9132
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5783392927455524e-05,
+      "loss": 0.7744,
+      "step": 9133
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578245375400311e-05,
+      "loss": 0.7432,
+      "step": 9134
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578151450391964e-05,
+      "loss": 0.7588,
+      "step": 9135
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5780575177217552e-05,
+      "loss": 0.7363,
+      "step": 9136
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5779635773909297e-05,
+      "loss": 0.749,
+      "step": 9137
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5778696294007326e-05,
+      "loss": 0.7656,
+      "step": 9138
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577775673752409e-05,
+      "loss": 0.7607,
+      "step": 9139
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5776817104472035e-05,
+      "loss": 0.751,
+      "step": 9140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5775877394863622e-05,
+      "loss": 0.7754,
+      "step": 9141
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5774937608711298e-05,
+      "loss": 0.7578,
+      "step": 9142
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5773997746027518e-05,
+      "loss": 0.7666,
+      "step": 9143
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.577305780682474e-05,
+      "loss": 0.7295,
+      "step": 9144
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5772117791115418e-05,
+      "loss": 0.7285,
+      "step": 9145
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.577117769891201e-05,
+      "loss": 0.7412,
+      "step": 9146
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5770237530226977e-05,
+      "loss": 0.7871,
+      "step": 9147
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5769297285072774e-05,
+      "loss": 0.7354,
+      "step": 9148
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576835696346186e-05,
+      "loss": 0.7178,
+      "step": 9149
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5767416565406704e-05,
+      "loss": 0.748,
+      "step": 9150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5766476090919764e-05,
+      "loss": 0.7314,
+      "step": 9151
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5765535540013505e-05,
+      "loss": 0.7607,
+      "step": 9152
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5764594912700388e-05,
+      "loss": 0.7393,
+      "step": 9153
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5763654208992885e-05,
+      "loss": 0.752,
+      "step": 9154
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5762713428903454e-05,
+      "loss": 0.7754,
+      "step": 9155
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576177257244457e-05,
+      "loss": 0.749,
+      "step": 9156
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57608316396287e-05,
+      "loss": 0.7734,
+      "step": 9157
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5759890630468307e-05,
+      "loss": 0.7461,
+      "step": 9158
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575894954497587e-05,
+      "loss": 0.7686,
+      "step": 9159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5758008383163854e-05,
+      "loss": 0.749,
+      "step": 9160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5757067145044738e-05,
+      "loss": 0.7666,
+      "step": 9161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5756125830630993e-05,
+      "loss": 0.751,
+      "step": 9162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575518443993509e-05,
+      "loss": 0.7402,
+      "step": 9163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575424297296951e-05,
+      "loss": 0.7559,
+      "step": 9164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575330142974673e-05,
+      "loss": 0.7744,
+      "step": 9165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575235981027922e-05,
+      "loss": 0.75,
+      "step": 9166
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5751418114579467e-05,
+      "loss": 0.7275,
+      "step": 9167
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5750476342659947e-05,
+      "loss": 0.7686,
+      "step": 9168
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5749534494533137e-05,
+      "loss": 0.7451,
+      "step": 9169
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5748592570211526e-05,
+      "loss": 0.7881,
+      "step": 9170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5747650569707596e-05,
+      "loss": 0.7754,
+      "step": 9171
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746708493033828e-05,
+      "loss": 0.751,
+      "step": 9172
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57457663402027e-05,
+      "loss": 0.7412,
+      "step": 9173
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5744824111226713e-05,
+      "loss": 0.7451,
+      "step": 9174
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574388180611834e-05,
+      "loss": 0.7754,
+      "step": 9175
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5742939424890078e-05,
+      "loss": 0.7705,
+      "step": 9176
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574199696755441e-05,
+      "loss": 0.7461,
+      "step": 9177
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5741054434123828e-05,
+      "loss": 0.7676,
+      "step": 9178
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574011182461082e-05,
+      "loss": 0.7461,
+      "step": 9179
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5739169139027883e-05,
+      "loss": 0.7822,
+      "step": 9180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5738226377387506e-05,
+      "loss": 0.751,
+      "step": 9181
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5737283539702187e-05,
+      "loss": 0.7656,
+      "step": 9182
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5736340625984412e-05,
+      "loss": 0.7695,
+      "step": 9183
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5735397636246688e-05,
+      "loss": 0.7471,
+      "step": 9184
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5734454570501502e-05,
+      "loss": 0.7598,
+      "step": 9185
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5733511428761357e-05,
+      "loss": 0.7334,
+      "step": 9186
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5732568211038755e-05,
+      "loss": 0.7656,
+      "step": 9187
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5731624917346188e-05,
+      "loss": 0.7373,
+      "step": 9188
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5730681547696162e-05,
+      "loss": 0.7656,
+      "step": 9189
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5729738102101175e-05,
+      "loss": 0.7432,
+      "step": 9190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5728794580573734e-05,
+      "loss": 0.7363,
+      "step": 9191
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572785098312634e-05,
+      "loss": 0.752,
+      "step": 9192
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5726907309771498e-05,
+      "loss": 0.7676,
+      "step": 9193
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5725963560521722e-05,
+      "loss": 0.7422,
+      "step": 9194
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5725019735389503e-05,
+      "loss": 0.7617,
+      "step": 9195
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5724075834387364e-05,
+      "loss": 0.751,
+      "step": 9196
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5723131857527808e-05,
+      "loss": 0.7754,
+      "step": 9197
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572218780482334e-05,
+      "loss": 0.7725,
+      "step": 9198
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5721243676286476e-05,
+      "loss": 0.748,
+      "step": 9199
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5720299471929733e-05,
+      "loss": 0.7529,
+      "step": 9200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5719355191765614e-05,
+      "loss": 0.7344,
+      "step": 9201
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571841083580664e-05,
+      "loss": 0.7773,
+      "step": 9202
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571746640406532e-05,
+      "loss": 0.7588,
+      "step": 9203
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5716521896554175e-05,
+      "loss": 0.7568,
+      "step": 9204
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5715577313285716e-05,
+      "loss": 0.7578,
+      "step": 9205
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571463265427247e-05,
+      "loss": 0.7705,
+      "step": 9206
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571368791952695e-05,
+      "loss": 0.7588,
+      "step": 9207
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5712743109061676e-05,
+      "loss": 0.7783,
+      "step": 9208
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571179822288917e-05,
+      "loss": 0.7568,
+      "step": 9209
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571085326102195e-05,
+      "loss": 0.7715,
+      "step": 9210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570990822347255e-05,
+      "loss": 0.749,
+      "step": 9211
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570896311025348e-05,
+      "loss": 0.7539,
+      "step": 9212
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570801792137728e-05,
+      "loss": 0.7598,
+      "step": 9213
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570707265685646e-05,
+      "loss": 0.7568,
+      "step": 9214
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5706127316703558e-05,
+      "loss": 0.7578,
+      "step": 9215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.57051819009311e-05,
+      "loss": 0.7441,
+      "step": 9216
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570423640955161e-05,
+      "loss": 0.748,
+      "step": 9217
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5703290842577624e-05,
+      "loss": 0.7842,
+      "step": 9218
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5702345200021667e-05,
+      "loss": 0.7432,
+      "step": 9219
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5701399481896277e-05,
+      "loss": 0.7441,
+      "step": 9220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5700453688213986e-05,
+      "loss": 0.7461,
+      "step": 9221
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5699507818987324e-05,
+      "loss": 0.7393,
+      "step": 9222
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569856187422883e-05,
+      "loss": 0.7588,
+      "step": 9223
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569761585395104e-05,
+      "loss": 0.7598,
+      "step": 9224
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5696669758166486e-05,
+      "loss": 0.7471,
+      "step": 9225
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5695723586887712e-05,
+      "loss": 0.7373,
+      "step": 9226
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569477734012725e-05,
+      "loss": 0.7559,
+      "step": 9227
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569383101789765e-05,
+      "loss": 0.7393,
+      "step": 9228
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5692884620211442e-05,
+      "loss": 0.7344,
+      "step": 9229
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5691938147081178e-05,
+      "loss": 0.7617,
+      "step": 9230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5690991598519394e-05,
+      "loss": 0.7432,
+      "step": 9231
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5690044974538637e-05,
+      "loss": 0.7393,
+      "step": 9232
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568909827515145e-05,
+      "loss": 0.7705,
+      "step": 9233
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5688151500370384e-05,
+      "loss": 0.7373,
+      "step": 9234
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5687204650207978e-05,
+      "loss": 0.7246,
+      "step": 9235
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5686257724676787e-05,
+      "loss": 0.7354,
+      "step": 9236
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5685310723789356e-05,
+      "loss": 0.7354,
+      "step": 9237
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5684363647558237e-05,
+      "loss": 0.748,
+      "step": 9238
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568341649599598e-05,
+      "loss": 0.7578,
+      "step": 9239
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5682469269115137e-05,
+      "loss": 0.7373,
+      "step": 9240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568152196692826e-05,
+      "loss": 0.7373,
+      "step": 9241
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5680574589447906e-05,
+      "loss": 0.7451,
+      "step": 9242
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567962713668663e-05,
+      "loss": 0.7441,
+      "step": 9243
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5678679608656983e-05,
+      "loss": 0.7461,
+      "step": 9244
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5677732005371525e-05,
+      "loss": 0.7578,
+      "step": 9245
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5676784326842813e-05,
+      "loss": 0.7471,
+      "step": 9246
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5675836573083412e-05,
+      "loss": 0.7402,
+      "step": 9247
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5674888744105873e-05,
+      "loss": 0.7656,
+      "step": 9248
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567394083992276e-05,
+      "loss": 0.748,
+      "step": 9249
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5672992860546638e-05,
+      "loss": 0.7432,
+      "step": 9250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5672044805990066e-05,
+      "loss": 0.7617,
+      "step": 9251
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5671096676265612e-05,
+      "loss": 0.7617,
+      "step": 9252
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5670148471385838e-05,
+      "loss": 0.7705,
+      "step": 9253
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566920019136331e-05,
+      "loss": 0.749,
+      "step": 9254
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5668251836210597e-05,
+      "loss": 0.7607,
+      "step": 9255
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5667303405940262e-05,
+      "loss": 0.75,
+      "step": 9256
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566635490056488e-05,
+      "loss": 0.7432,
+      "step": 9257
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5665406320097017e-05,
+      "loss": 0.7422,
+      "step": 9258
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5664457664549248e-05,
+      "loss": 0.7217,
+      "step": 9259
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566350893393414e-05,
+      "loss": 0.7461,
+      "step": 9260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5662560128264267e-05,
+      "loss": 0.7715,
+      "step": 9261
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5661611247552205e-05,
+      "loss": 0.749,
+      "step": 9262
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5660662291810528e-05,
+      "loss": 0.7617,
+      "step": 9263
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565971326105181e-05,
+      "loss": 0.7607,
+      "step": 9264
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5658764155288632e-05,
+      "loss": 0.7646,
+      "step": 9265
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565781497453357e-05,
+      "loss": 0.7578,
+      "step": 9266
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56568657187992e-05,
+      "loss": 0.7471,
+      "step": 9267
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5655916388098107e-05,
+      "loss": 0.7627,
+      "step": 9268
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5654966982442865e-05,
+      "loss": 0.7695,
+      "step": 9269
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565401750184606e-05,
+      "loss": 0.7314,
+      "step": 9270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5653067946320283e-05,
+      "loss": 0.752,
+      "step": 9271
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56521183158781e-05,
+      "loss": 0.7432,
+      "step": 9272
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5651168610532108e-05,
+      "loss": 0.752,
+      "step": 9273
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5650218830294892e-05,
+      "loss": 0.7656,
+      "step": 9274
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564926897517904e-05,
+      "loss": 0.7656,
+      "step": 9275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564831904519713e-05,
+      "loss": 0.7539,
+      "step": 9276
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564736904036176e-05,
+      "loss": 0.7734,
+      "step": 9277
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564641896068552e-05,
+      "loss": 0.7441,
+      "step": 9278
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5645468806180996e-05,
+      "loss": 0.7441,
+      "step": 9279
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5644518576860782e-05,
+      "loss": 0.7793,
+      "step": 9280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5643568272737473e-05,
+      "loss": 0.7354,
+      "step": 9281
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564261789382366e-05,
+      "loss": 0.7305,
+      "step": 9282
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5641667440131936e-05,
+      "loss": 0.7393,
+      "step": 9283
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5640716911674898e-05,
+      "loss": 0.75,
+      "step": 9284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5639766308465146e-05,
+      "loss": 0.7285,
+      "step": 9285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5638815630515277e-05,
+      "loss": 0.752,
+      "step": 9286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5637864877837885e-05,
+      "loss": 0.7285,
+      "step": 9287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5636914050445577e-05,
+      "loss": 0.7666,
+      "step": 9288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5635963148350946e-05,
+      "loss": 0.7568,
+      "step": 9289
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5635012171566596e-05,
+      "loss": 0.7559,
+      "step": 9290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5634061120105135e-05,
+      "loss": 0.7275,
+      "step": 9291
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563310999397916e-05,
+      "loss": 0.7529,
+      "step": 9292
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5632158793201277e-05,
+      "loss": 0.7461,
+      "step": 9293
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5631207517784098e-05,
+      "loss": 0.7432,
+      "step": 9294
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563025616774022e-05,
+      "loss": 0.7588,
+      "step": 9295
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5629304743082256e-05,
+      "loss": 0.748,
+      "step": 9296
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5628353243822814e-05,
+      "loss": 0.7559,
+      "step": 9297
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56274016699745e-05,
+      "loss": 0.7539,
+      "step": 9298
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5626450021549935e-05,
+      "loss": 0.7275,
+      "step": 9299
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5625498298561715e-05,
+      "loss": 0.749,
+      "step": 9300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562454650102247e-05,
+      "loss": 0.7314,
+      "step": 9301
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56235946289448e-05,
+      "loss": 0.7402,
+      "step": 9302
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5622642682341323e-05,
+      "loss": 0.7754,
+      "step": 9303
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562169066122466e-05,
+      "loss": 0.7451,
+      "step": 9304
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5620738565607418e-05,
+      "loss": 0.7607,
+      "step": 9305
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5619786395502222e-05,
+      "loss": 0.7295,
+      "step": 9306
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5618834150921688e-05,
+      "loss": 0.7373,
+      "step": 9307
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5617881831878433e-05,
+      "loss": 0.7578,
+      "step": 9308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5616929438385082e-05,
+      "loss": 0.7412,
+      "step": 9309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561597697045426e-05,
+      "loss": 0.7832,
+      "step": 9310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5615024428098577e-05,
+      "loss": 0.7285,
+      "step": 9311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5614071811330665e-05,
+      "loss": 0.7568,
+      "step": 9312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5613119120163147e-05,
+      "loss": 0.7422,
+      "step": 9313
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.561216635460865e-05,
+      "loss": 0.7588,
+      "step": 9314
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5611213514679793e-05,
+      "loss": 0.7861,
+      "step": 9315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.561026060038921e-05,
+      "loss": 0.7266,
+      "step": 9316
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5609307611749535e-05,
+      "loss": 0.7393,
+      "step": 9317
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5608354548773387e-05,
+      "loss": 0.7666,
+      "step": 9318
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5607401411473398e-05,
+      "loss": 0.7617,
+      "step": 9319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5606448199862203e-05,
+      "loss": 0.7471,
+      "step": 9320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560549491395243e-05,
+      "loss": 0.7227,
+      "step": 9321
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5604541553756715e-05,
+      "loss": 0.7588,
+      "step": 9322
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5603588119287695e-05,
+      "loss": 0.7441,
+      "step": 9323
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5602634610558003e-05,
+      "loss": 0.7666,
+      "step": 9324
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5601681027580272e-05,
+      "loss": 0.7656,
+      "step": 9325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560072737036714e-05,
+      "loss": 0.7305,
+      "step": 9326
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559977363893125e-05,
+      "loss": 0.7666,
+      "step": 9327
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5598819833285236e-05,
+      "loss": 0.7715,
+      "step": 9328
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5597865953441742e-05,
+      "loss": 0.7549,
+      "step": 9329
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5596911999413405e-05,
+      "loss": 0.7598,
+      "step": 9330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5595957971212873e-05,
+      "loss": 0.7441,
+      "step": 9331
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5595003868852784e-05,
+      "loss": 0.7432,
+      "step": 9332
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5594049692345782e-05,
+      "loss": 0.751,
+      "step": 9333
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5593095441704515e-05,
+      "loss": 0.7363,
+      "step": 9334
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5592141116941628e-05,
+      "loss": 0.7451,
+      "step": 9335
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5591186718069768e-05,
+      "loss": 0.7559,
+      "step": 9336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5590232245101583e-05,
+      "loss": 0.7549,
+      "step": 9337
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5589277698049722e-05,
+      "loss": 0.7686,
+      "step": 9338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558832307692683e-05,
+      "loss": 0.7598,
+      "step": 9339
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5587368381745567e-05,
+      "loss": 0.7861,
+      "step": 9340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558641361251858e-05,
+      "loss": 0.7627,
+      "step": 9341
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5585458769258525e-05,
+      "loss": 0.7686,
+      "step": 9342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5584503851978056e-05,
+      "loss": 0.7568,
+      "step": 9343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5583548860689817e-05,
+      "loss": 0.7676,
+      "step": 9344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5582593795406475e-05,
+      "loss": 0.7676,
+      "step": 9345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558163865614069e-05,
+      "loss": 0.7588,
+      "step": 9346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558068344290511e-05,
+      "loss": 0.7441,
+      "step": 9347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5579728155712394e-05,
+      "loss": 0.7354,
+      "step": 9348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557877279457521e-05,
+      "loss": 0.7393,
+      "step": 9349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557781735950622e-05,
+      "loss": 0.7178,
+      "step": 9350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557686185051807e-05,
+      "loss": 0.7617,
+      "step": 9351
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557590626762344e-05,
+      "loss": 0.7598,
+      "step": 9352
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5574950610834983e-05,
+      "loss": 0.7451,
+      "step": 9353
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5573994880165367e-05,
+      "loss": 0.7412,
+      "step": 9354
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5573039075627257e-05,
+      "loss": 0.7627,
+      "step": 9355
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5572083197233326e-05,
+      "loss": 0.7412,
+      "step": 9356
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557112724499623e-05,
+      "loss": 0.748,
+      "step": 9357
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5570171218928648e-05,
+      "loss": 0.751,
+      "step": 9358
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5569215119043245e-05,
+      "loss": 0.7373,
+      "step": 9359
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556825894535269e-05,
+      "loss": 0.7451,
+      "step": 9360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5567302697869657e-05,
+      "loss": 0.7471,
+      "step": 9361
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556634637660682e-05,
+      "loss": 0.7207,
+      "step": 9362
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556538998157685e-05,
+      "loss": 0.7617,
+      "step": 9363
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556443351279242e-05,
+      "loss": 0.7734,
+      "step": 9364
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556347697026621e-05,
+      "loss": 0.7412,
+      "step": 9365
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5562520354010895e-05,
+      "loss": 0.7617,
+      "step": 9366
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556156366403915e-05,
+      "loss": 0.752,
+      "step": 9367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5560606900363654e-05,
+      "loss": 0.7539,
+      "step": 9368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5559650062997085e-05,
+      "loss": 0.7412,
+      "step": 9369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5558693151952132e-05,
+      "loss": 0.748,
+      "step": 9370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5557736167241465e-05,
+      "loss": 0.7422,
+      "step": 9371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555677910887777e-05,
+      "loss": 0.7432,
+      "step": 9372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5555821976873735e-05,
+      "loss": 0.7422,
+      "step": 9373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5554864771242036e-05,
+      "loss": 0.7363,
+      "step": 9374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5553907491995367e-05,
+      "loss": 0.7471,
+      "step": 9375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5552950139146407e-05,
+      "loss": 0.7617,
+      "step": 9376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5551992712707848e-05,
+      "loss": 0.7422,
+      "step": 9377
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5551035212692375e-05,
+      "loss": 0.7422,
+      "step": 9378
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555007763911268e-05,
+      "loss": 0.7627,
+      "step": 9379
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554911999198145e-05,
+      "loss": 0.7646,
+      "step": 9380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554816227131138e-05,
+      "loss": 0.7529,
+      "step": 9381
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5547204477115157e-05,
+      "loss": 0.7402,
+      "step": 9382
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5546246609405477e-05,
+      "loss": 0.7617,
+      "step": 9383
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5545288668195034e-05,
+      "loss": 0.7334,
+      "step": 9384
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554433065349652e-05,
+      "loss": 0.7227,
+      "step": 9385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5543372565322638e-05,
+      "loss": 0.7412,
+      "step": 9386
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5542414403686078e-05,
+      "loss": 0.7266,
+      "step": 9387
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554145616859954e-05,
+      "loss": 0.7607,
+      "step": 9388
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5540497860075726e-05,
+      "loss": 0.7881,
+      "step": 9389
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539539478127326e-05,
+      "loss": 0.7734,
+      "step": 9390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5538581022767055e-05,
+      "loss": 0.7344,
+      "step": 9391
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5537622494007604e-05,
+      "loss": 0.7529,
+      "step": 9392
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553666389186168e-05,
+      "loss": 0.7461,
+      "step": 9393
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5535705216341987e-05,
+      "loss": 0.748,
+      "step": 9394
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553474646746123e-05,
+      "loss": 0.7646,
+      "step": 9395
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553378764523211e-05,
+      "loss": 0.7432,
+      "step": 9396
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5532828749667338e-05,
+      "loss": 0.752,
+      "step": 9397
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5531869780779624e-05,
+      "loss": 0.7402,
+      "step": 9398
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553091073858167e-05,
+      "loss": 0.7559,
+      "step": 9399
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552995162308619e-05,
+      "loss": 0.7539,
+      "step": 9400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5528992434305893e-05,
+      "loss": 0.7354,
+      "step": 9401
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5528033172253492e-05,
+      "loss": 0.7207,
+      "step": 9402
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5527073836941697e-05,
+      "loss": 0.7656,
+      "step": 9403
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5526114428383222e-05,
+      "loss": 0.7744,
+      "step": 9404
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5525154946590783e-05,
+      "loss": 0.7559,
+      "step": 9405
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5524195391577097e-05,
+      "loss": 0.7441,
+      "step": 9406
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5523235763354874e-05,
+      "loss": 0.7617,
+      "step": 9407
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5522276061936838e-05,
+      "loss": 0.7559,
+      "step": 9408
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5521316287335704e-05,
+      "loss": 0.7354,
+      "step": 9409
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552035643956419e-05,
+      "loss": 0.7578,
+      "step": 9410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551939651863502e-05,
+      "loss": 0.7822,
+      "step": 9411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551843652456091e-05,
+      "loss": 0.749,
+      "step": 9412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551747645735459e-05,
+      "loss": 0.7686,
+      "step": 9413
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5516516317028772e-05,
+      "loss": 0.7686,
+      "step": 9414
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5515556103596196e-05,
+      "loss": 0.7627,
+      "step": 9415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551459581706957e-05,
+      "loss": 0.7461,
+      "step": 9416
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551363545746163e-05,
+      "loss": 0.7354,
+      "step": 9417
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5512675024785102e-05,
+      "loss": 0.7695,
+      "step": 9418
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551171451905271e-05,
+      "loss": 0.7373,
+      "step": 9419
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551075394027719e-05,
+      "loss": 0.7363,
+      "step": 9420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5509793288471266e-05,
+      "loss": 0.7354,
+      "step": 9421
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.550883256364767e-05,
+      "loss": 0.751,
+      "step": 9422
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5507871765819135e-05,
+      "loss": 0.7646,
+      "step": 9423
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5506910894998393e-05,
+      "loss": 0.75,
+      "step": 9424
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5505949951198177e-05,
+      "loss": 0.7568,
+      "step": 9425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5504988934431218e-05,
+      "loss": 0.7842,
+      "step": 9426
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5504027844710264e-05,
+      "loss": 0.7383,
+      "step": 9427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5503066682048044e-05,
+      "loss": 0.7578,
+      "step": 9428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5502105446457293e-05,
+      "loss": 0.7383,
+      "step": 9429
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5501144137950754e-05,
+      "loss": 0.7383,
+      "step": 9430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5500182756541163e-05,
+      "loss": 0.7393,
+      "step": 9431
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5499221302241264e-05,
+      "loss": 0.7402,
+      "step": 9432
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549825977506379e-05,
+      "loss": 0.7559,
+      "step": 9433
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5497298175021502e-05,
+      "loss": 0.7266,
+      "step": 9434
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5496336502127126e-05,
+      "loss": 0.7666,
+      "step": 9435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5495374756393415e-05,
+      "loss": 0.7861,
+      "step": 9436
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5494412937833112e-05,
+      "loss": 0.7441,
+      "step": 9437
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5493451046458964e-05,
+      "loss": 0.7432,
+      "step": 9438
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5492489082283712e-05,
+      "loss": 0.7305,
+      "step": 9439
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5491527045320114e-05,
+      "loss": 0.7568,
+      "step": 9440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5490564935580916e-05,
+      "loss": 0.7549,
+      "step": 9441
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5489602753078864e-05,
+      "loss": 0.7607,
+      "step": 9442
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5488640497826714e-05,
+      "loss": 0.7471,
+      "step": 9443
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5487678169837218e-05,
+      "loss": 0.7451,
+      "step": 9444
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5486715769123124e-05,
+      "loss": 0.7383,
+      "step": 9445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548575329569719e-05,
+      "loss": 0.752,
+      "step": 9446
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5484790749572175e-05,
+      "loss": 0.7588,
+      "step": 9447
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548382813076083e-05,
+      "loss": 0.7412,
+      "step": 9448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5482865439275913e-05,
+      "loss": 0.7432,
+      "step": 9449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548190267513018e-05,
+      "loss": 0.7695,
+      "step": 9450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5480939838336388e-05,
+      "loss": 0.7617,
+      "step": 9451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5479976928907303e-05,
+      "loss": 0.7871,
+      "step": 9452
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5479013946855686e-05,
+      "loss": 0.7725,
+      "step": 9453
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5478050892194294e-05,
+      "loss": 0.7666,
+      "step": 9454
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547708776493589e-05,
+      "loss": 0.7314,
+      "step": 9455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547612456509324e-05,
+      "loss": 0.7617,
+      "step": 9456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547516129267911e-05,
+      "loss": 0.7354,
+      "step": 9457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5474197947706262e-05,
+      "loss": 0.7734,
+      "step": 9458
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5473234530187465e-05,
+      "loss": 0.7627,
+      "step": 9459
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5472271040135486e-05,
+      "loss": 0.7578,
+      "step": 9460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5471307477563093e-05,
+      "loss": 0.7207,
+      "step": 9461
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5470343842483053e-05,
+      "loss": 0.7246,
+      "step": 9462
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5469380134908144e-05,
+      "loss": 0.7549,
+      "step": 9463
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5468416354851127e-05,
+      "loss": 0.752,
+      "step": 9464
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546745250232478e-05,
+      "loss": 0.7656,
+      "step": 9465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5466488577341884e-05,
+      "loss": 0.7725,
+      "step": 9466
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.54655245799152e-05,
+      "loss": 0.7451,
+      "step": 9467
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5464560510057506e-05,
+      "loss": 0.748,
+      "step": 9468
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5463596367781585e-05,
+      "loss": 0.7402,
+      "step": 9469
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5462632153100207e-05,
+      "loss": 0.7598,
+      "step": 9470
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5461667866026154e-05,
+      "loss": 0.7637,
+      "step": 9471
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5460703506572205e-05,
+      "loss": 0.7334,
+      "step": 9472
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5459739074751143e-05,
+      "loss": 0.7383,
+      "step": 9473
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545877457057574e-05,
+      "loss": 0.7441,
+      "step": 9474
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5457809994058782e-05,
+      "loss": 0.7441,
+      "step": 9475
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5456845345213055e-05,
+      "loss": 0.7236,
+      "step": 9476
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5455880624051342e-05,
+      "loss": 0.7773,
+      "step": 9477
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5454915830586425e-05,
+      "loss": 0.7529,
+      "step": 9478
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545395096483109e-05,
+      "loss": 0.7578,
+      "step": 9479
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545298602679813e-05,
+      "loss": 0.7236,
+      "step": 9480
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5452021016500325e-05,
+      "loss": 0.7373,
+      "step": 9481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.545105593395047e-05,
+      "loss": 0.7363,
+      "step": 9482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5450090779161345e-05,
+      "loss": 0.7744,
+      "step": 9483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544912555214575e-05,
+      "loss": 0.7529,
+      "step": 9484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5448160252916472e-05,
+      "loss": 0.7363,
+      "step": 9485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5447194881486304e-05,
+      "loss": 0.748,
+      "step": 9486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5446229437868045e-05,
+      "loss": 0.7393,
+      "step": 9487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5445263922074477e-05,
+      "loss": 0.7539,
+      "step": 9488
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544429833411841e-05,
+      "loss": 0.748,
+      "step": 9489
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5443332674012626e-05,
+      "loss": 0.7725,
+      "step": 9490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5442366941769933e-05,
+      "loss": 0.7783,
+      "step": 9491
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.544140113740313e-05,
+      "loss": 0.7363,
+      "step": 9492
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5440435260925007e-05,
+      "loss": 0.7256,
+      "step": 9493
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543946931234837e-05,
+      "loss": 0.7695,
+      "step": 9494
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5438503291686017e-05,
+      "loss": 0.751,
+      "step": 9495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5437537198950753e-05,
+      "loss": 0.7812,
+      "step": 9496
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5436571034155382e-05,
+      "loss": 0.7461,
+      "step": 9497
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5435604797312702e-05,
+      "loss": 0.7324,
+      "step": 9498
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5434638488435525e-05,
+      "loss": 0.7266,
+      "step": 9499
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543367210753665e-05,
+      "loss": 0.7393,
+      "step": 9500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543270565462889e-05,
+      "loss": 0.7695,
+      "step": 9501
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5431739129725058e-05,
+      "loss": 0.7646,
+      "step": 9502
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5430772532837942e-05,
+      "loss": 0.7578,
+      "step": 9503
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5429805863980375e-05,
+      "loss": 0.748,
+      "step": 9504
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542883912316515e-05,
+      "loss": 0.7549,
+      "step": 9505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5427872310405088e-05,
+      "loss": 0.7686,
+      "step": 9506
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5426905425713003e-05,
+      "loss": 0.7461,
+      "step": 9507
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54259384691017e-05,
+      "loss": 0.7324,
+      "step": 9508
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5424971440584e-05,
+      "loss": 0.748,
+      "step": 9509
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542400434017272e-05,
+      "loss": 0.7559,
+      "step": 9510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542303716788067e-05,
+      "loss": 0.748,
+      "step": 9511
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5422069923720672e-05,
+      "loss": 0.7539,
+      "step": 9512
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542110260770554e-05,
+      "loss": 0.7725,
+      "step": 9513
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5420135219848097e-05,
+      "loss": 0.7402,
+      "step": 9514
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5419167760161163e-05,
+      "loss": 0.7324,
+      "step": 9515
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5418200228657555e-05,
+      "loss": 0.7432,
+      "step": 9516
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54172326253501e-05,
+      "loss": 0.7471,
+      "step": 9517
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541626495025162e-05,
+      "loss": 0.7812,
+      "step": 9518
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5415297203374938e-05,
+      "loss": 0.7559,
+      "step": 9519
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541432938473288e-05,
+      "loss": 0.7695,
+      "step": 9520
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541336149433827e-05,
+      "loss": 0.7295,
+      "step": 9521
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541239353220394e-05,
+      "loss": 0.7334,
+      "step": 9522
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5411425498342705e-05,
+      "loss": 0.7441,
+      "step": 9523
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541045739276741e-05,
+      "loss": 0.7588,
+      "step": 9524
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5409489215490872e-05,
+      "loss": 0.7432,
+      "step": 9525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408520966525926e-05,
+      "loss": 0.7354,
+      "step": 9526
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5407552645885407e-05,
+      "loss": 0.7402,
+      "step": 9527
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5406584253582144e-05,
+      "loss": 0.749,
+      "step": 9528
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540561578962897e-05,
+      "loss": 0.7441,
+      "step": 9529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5404647254038723e-05,
+      "loss": 0.7275,
+      "step": 9530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5403678646824234e-05,
+      "loss": 0.7432,
+      "step": 9531
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5402709967998338e-05,
+      "loss": 0.7539,
+      "step": 9532
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540174121757388e-05,
+      "loss": 0.7656,
+      "step": 9533
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5400772395563692e-05,
+      "loss": 0.7539,
+      "step": 9534
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5399803501980618e-05,
+      "loss": 0.7441,
+      "step": 9535
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539883453683749e-05,
+      "loss": 0.7568,
+      "step": 9536
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5397865500147156e-05,
+      "loss": 0.7715,
+      "step": 9537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5396896391922457e-05,
+      "loss": 0.7354,
+      "step": 9538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5395927212176234e-05,
+      "loss": 0.7627,
+      "step": 9539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5394957960921328e-05,
+      "loss": 0.7354,
+      "step": 9540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5393988638170597e-05,
+      "loss": 0.751,
+      "step": 9541
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5393019243936868e-05,
+      "loss": 0.7598,
+      "step": 9542
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5392049778233e-05,
+      "loss": 0.7539,
+      "step": 9543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.539108024107184e-05,
+      "loss": 0.7451,
+      "step": 9544
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5390110632466235e-05,
+      "loss": 0.752,
+      "step": 9545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538914095242903e-05,
+      "loss": 0.7539,
+      "step": 9546
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5388171200973085e-05,
+      "loss": 0.7529,
+      "step": 9547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5387201378111242e-05,
+      "loss": 0.7627,
+      "step": 9548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5386231483856363e-05,
+      "loss": 0.7285,
+      "step": 9549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538526151822129e-05,
+      "loss": 0.7441,
+      "step": 9550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5384291481218885e-05,
+      "loss": 0.7461,
+      "step": 9551
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5383321372862002e-05,
+      "loss": 0.7539,
+      "step": 9552
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5382351193163495e-05,
+      "loss": 0.7451,
+      "step": 9553
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5381380942136224e-05,
+      "loss": 0.7617,
+      "step": 9554
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538041061979305e-05,
+      "loss": 0.7744,
+      "step": 9555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5379440226146826e-05,
+      "loss": 0.7646,
+      "step": 9556
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537846976121041e-05,
+      "loss": 0.7402,
+      "step": 9557
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537749922499667e-05,
+      "loss": 0.751,
+      "step": 9558
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5376528617518465e-05,
+      "loss": 0.7471,
+      "step": 9559
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5375557938788657e-05,
+      "loss": 0.7295,
+      "step": 9560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537458718882011e-05,
+      "loss": 0.7686,
+      "step": 9561
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537361636762569e-05,
+      "loss": 0.7578,
+      "step": 9562
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5372645475218262e-05,
+      "loss": 0.751,
+      "step": 9563
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537167451161069e-05,
+      "loss": 0.7402,
+      "step": 9564
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537070347681585e-05,
+      "loss": 0.7812,
+      "step": 9565
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53697323708466e-05,
+      "loss": 0.7539,
+      "step": 9566
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5368761193715813e-05,
+      "loss": 0.7646,
+      "step": 9567
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536778994543636e-05,
+      "loss": 0.7256,
+      "step": 9568
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5366818626021115e-05,
+      "loss": 0.7666,
+      "step": 9569
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536584723548295e-05,
+      "loss": 0.7549,
+      "step": 9570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5364875773834732e-05,
+      "loss": 0.7539,
+      "step": 9571
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536390424108934e-05,
+      "loss": 0.7461,
+      "step": 9572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536293263725965e-05,
+      "loss": 0.7822,
+      "step": 9573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5361960962358533e-05,
+      "loss": 0.7529,
+      "step": 9574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536098921639887e-05,
+      "loss": 0.748,
+      "step": 9575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536001739939354e-05,
+      "loss": 0.749,
+      "step": 9576
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5359045511355416e-05,
+      "loss": 0.7412,
+      "step": 9577
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535807355229739e-05,
+      "loss": 0.7363,
+      "step": 9578
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5357101522232324e-05,
+      "loss": 0.7539,
+      "step": 9579
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535612942117312e-05,
+      "loss": 0.7646,
+      "step": 9580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5355157249132645e-05,
+      "loss": 0.7324,
+      "step": 9581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5354185006123786e-05,
+      "loss": 0.7324,
+      "step": 9582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5353212692159438e-05,
+      "loss": 0.7617,
+      "step": 9583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535224030725247e-05,
+      "loss": 0.7725,
+      "step": 9584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.535126785141578e-05,
+      "loss": 0.7686,
+      "step": 9585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5350295324662258e-05,
+      "loss": 0.7422,
+      "step": 9586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534932272700478e-05,
+      "loss": 0.7246,
+      "step": 9587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534835005845624e-05,
+      "loss": 0.7715,
+      "step": 9588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5347377319029534e-05,
+      "loss": 0.752,
+      "step": 9589
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5346404508737546e-05,
+      "loss": 0.7461,
+      "step": 9590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534543162759317e-05,
+      "loss": 0.7354,
+      "step": 9591
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53444586756093e-05,
+      "loss": 0.7402,
+      "step": 9592
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534348565279883e-05,
+      "loss": 0.7539,
+      "step": 9593
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534251255917466e-05,
+      "loss": 0.7539,
+      "step": 9594
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5341539394749675e-05,
+      "loss": 0.7393,
+      "step": 9595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5340566159536776e-05,
+      "loss": 0.7549,
+      "step": 9596
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5339592853548863e-05,
+      "loss": 0.751,
+      "step": 9597
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533861947679883e-05,
+      "loss": 0.748,
+      "step": 9598
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533764602929958e-05,
+      "loss": 0.748,
+      "step": 9599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533667251106402e-05,
+      "loss": 0.7363,
+      "step": 9600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5335698922105038e-05,
+      "loss": 0.7598,
+      "step": 9601
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5334725262435546e-05,
+      "loss": 0.7461,
+      "step": 9602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5333751532068443e-05,
+      "loss": 0.7432,
+      "step": 9603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5332777731016634e-05,
+      "loss": 0.7471,
+      "step": 9604
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5331803859293027e-05,
+      "loss": 0.7393,
+      "step": 9605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5330829916910518e-05,
+      "loss": 0.748,
+      "step": 9606
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532985590388203e-05,
+      "loss": 0.7705,
+      "step": 9607
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5328881820220454e-05,
+      "loss": 0.7617,
+      "step": 9608
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5327907665938715e-05,
+      "loss": 0.7393,
+      "step": 9609
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5326933441049714e-05,
+      "loss": 0.7559,
+      "step": 9610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532595914556636e-05,
+      "loss": 0.75,
+      "step": 9611
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5324984779501568e-05,
+      "loss": 0.7646,
+      "step": 9612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5324010342868253e-05,
+      "loss": 0.7373,
+      "step": 9613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5323035835679322e-05,
+      "loss": 0.7207,
+      "step": 9614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.53220612579477e-05,
+      "loss": 0.7891,
+      "step": 9615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532108660968629e-05,
+      "loss": 0.7549,
+      "step": 9616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5320111890908017e-05,
+      "loss": 0.752,
+      "step": 9617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5319137101625795e-05,
+      "loss": 0.7725,
+      "step": 9618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5318162241852543e-05,
+      "loss": 0.7393,
+      "step": 9619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317187311601176e-05,
+      "loss": 0.7539,
+      "step": 9620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5316212310884623e-05,
+      "loss": 0.7617,
+      "step": 9621
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5315237239715796e-05,
+      "loss": 0.7754,
+      "step": 9622
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5314262098107624e-05,
+      "loss": 0.7783,
+      "step": 9623
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531328688607302e-05,
+      "loss": 0.7324,
+      "step": 9624
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5312311603624927e-05,
+      "loss": 0.7256,
+      "step": 9625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531133625077625e-05,
+      "loss": 0.7705,
+      "step": 9626
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5310360827539923e-05,
+      "loss": 0.7246,
+      "step": 9627
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5309385333928874e-05,
+      "loss": 0.7812,
+      "step": 9628
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5308409769956027e-05,
+      "loss": 0.7539,
+      "step": 9629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530743413563431e-05,
+      "loss": 0.751,
+      "step": 9630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5306458430976657e-05,
+      "loss": 0.749,
+      "step": 9631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5305482655995994e-05,
+      "loss": 0.7402,
+      "step": 9632
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5304506810705258e-05,
+      "loss": 0.7539,
+      "step": 9633
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5303530895117377e-05,
+      "loss": 0.7568,
+      "step": 9634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5302554909245283e-05,
+      "loss": 0.7412,
+      "step": 9635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5301578853101907e-05,
+      "loss": 0.7305,
+      "step": 9636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5300602726700193e-05,
+      "loss": 0.7705,
+      "step": 9637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5299626530053072e-05,
+      "loss": 0.7529,
+      "step": 9638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5298650263173485e-05,
+      "loss": 0.7559,
+      "step": 9639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5297673926074362e-05,
+      "loss": 0.7588,
+      "step": 9640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529669751876865e-05,
+      "loss": 0.7793,
+      "step": 9641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529572104126928e-05,
+      "loss": 0.752,
+      "step": 9642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.52947444935892e-05,
+      "loss": 0.7725,
+      "step": 9643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529376787574135e-05,
+      "loss": 0.7422,
+      "step": 9644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529279118773867e-05,
+      "loss": 0.7832,
+      "step": 9645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5291814429594104e-05,
+      "loss": 0.7344,
+      "step": 9646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5290837601320596e-05,
+      "loss": 0.7656,
+      "step": 9647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5289860702931096e-05,
+      "loss": 0.749,
+      "step": 9648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5288883734438544e-05,
+      "loss": 0.7432,
+      "step": 9649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5287906695855892e-05,
+      "loss": 0.748,
+      "step": 9650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5286929587196084e-05,
+      "loss": 0.7412,
+      "step": 9651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5285952408472072e-05,
+      "loss": 0.75,
+      "step": 9652
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5284975159696802e-05,
+      "loss": 0.7646,
+      "step": 9653
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528399784088323e-05,
+      "loss": 0.7383,
+      "step": 9654
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5283020452044307e-05,
+      "loss": 0.7637,
+      "step": 9655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528204299319298e-05,
+      "loss": 0.7607,
+      "step": 9656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5281065464342207e-05,
+      "loss": 0.7422,
+      "step": 9657
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5280087865504944e-05,
+      "loss": 0.7705,
+      "step": 9658
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279110196694144e-05,
+      "loss": 0.751,
+      "step": 9659
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5278132457922765e-05,
+      "loss": 0.7617,
+      "step": 9660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527715464920376e-05,
+      "loss": 0.7598,
+      "step": 9661
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5276176770550096e-05,
+      "loss": 0.7598,
+      "step": 9662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5275198821974725e-05,
+      "loss": 0.7451,
+      "step": 9663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527422080349061e-05,
+      "loss": 0.7637,
+      "step": 9664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273242715110706e-05,
+      "loss": 0.7432,
+      "step": 9665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527226455684798e-05,
+      "loss": 0.752,
+      "step": 9666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.52712863287154e-05,
+      "loss": 0.7695,
+      "step": 9667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.527030803072592e-05,
+      "loss": 0.7607,
+      "step": 9668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5269329662892513e-05,
+      "loss": 0.7285,
+      "step": 9669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5268351225228138e-05,
+      "loss": 0.7324,
+      "step": 9670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5267372717745767e-05,
+      "loss": 0.7676,
+      "step": 9671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5266394140458365e-05,
+      "loss": 0.7412,
+      "step": 9672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.52654154933789e-05,
+      "loss": 0.7217,
+      "step": 9673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5264436776520342e-05,
+      "loss": 0.7461,
+      "step": 9674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5263457989895658e-05,
+      "loss": 0.7285,
+      "step": 9675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5262479133517825e-05,
+      "loss": 0.7275,
+      "step": 9676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5261500207399815e-05,
+      "loss": 0.7803,
+      "step": 9677
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5260521211554595e-05,
+      "loss": 0.7607,
+      "step": 9678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5259542145995144e-05,
+      "loss": 0.7402,
+      "step": 9679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5258563010734437e-05,
+      "loss": 0.751,
+      "step": 9680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5257583805785448e-05,
+      "loss": 0.7451,
+      "step": 9681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5256604531161152e-05,
+      "loss": 0.7539,
+      "step": 9682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525562518687453e-05,
+      "loss": 0.7666,
+      "step": 9683
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5254645772938558e-05,
+      "loss": 0.7646,
+      "step": 9684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525366628936622e-05,
+      "loss": 0.75,
+      "step": 9685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525268673617049e-05,
+      "loss": 0.7422,
+      "step": 9686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5251707113364358e-05,
+      "loss": 0.7354,
+      "step": 9687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5250727420960796e-05,
+      "loss": 0.7432,
+      "step": 9688
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5249747658972793e-05,
+      "loss": 0.7549,
+      "step": 9689
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5248767827413335e-05,
+      "loss": 0.748,
+      "step": 9690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247787926295403e-05,
+      "loss": 0.7363,
+      "step": 9691
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5246807955631982e-05,
+      "loss": 0.749,
+      "step": 9692
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5245827915436066e-05,
+      "loss": 0.7607,
+      "step": 9693
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5244847805720635e-05,
+      "loss": 0.7363,
+      "step": 9694
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5243867626498683e-05,
+      "loss": 0.7471,
+      "step": 9695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5242887377783197e-05,
+      "loss": 0.7344,
+      "step": 9696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241907059587168e-05,
+      "loss": 0.7119,
+      "step": 9697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5240926671923586e-05,
+      "loss": 0.7305,
+      "step": 9698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5239946214805445e-05,
+      "loss": 0.7559,
+      "step": 9699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5238965688245744e-05,
+      "loss": 0.7197,
+      "step": 9700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5237985092257468e-05,
+      "loss": 0.7451,
+      "step": 9701
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5237004426853615e-05,
+      "loss": 0.7393,
+      "step": 9702
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5236023692047184e-05,
+      "loss": 0.7354,
+      "step": 9703
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5235042887851167e-05,
+      "loss": 0.7656,
+      "step": 9704
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523406201427857e-05,
+      "loss": 0.7441,
+      "step": 9705
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5233081071342386e-05,
+      "loss": 0.7129,
+      "step": 9706
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5232100059055613e-05,
+      "loss": 0.7744,
+      "step": 9707
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5231118977431254e-05,
+      "loss": 0.75,
+      "step": 9708
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5230137826482317e-05,
+      "loss": 0.75,
+      "step": 9709
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5229156606221792e-05,
+      "loss": 0.7471,
+      "step": 9710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5228175316662692e-05,
+      "loss": 0.7627,
+      "step": 9711
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5227193957818021e-05,
+      "loss": 0.7412,
+      "step": 9712
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.522621252970078e-05,
+      "loss": 0.7637,
+      "step": 9713
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225231032323973e-05,
+      "loss": 0.7549,
+      "step": 9714
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5224249465700615e-05,
+      "loss": 0.7344,
+      "step": 9715
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5223267829843708e-05,
+      "loss": 0.7666,
+      "step": 9716
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5222286124766266e-05,
+      "loss": 0.7334,
+      "step": 9717
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5221304350481297e-05,
+      "loss": 0.7373,
+      "step": 9718
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5220322507001809e-05,
+      "loss": 0.7451,
+      "step": 9719
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5219340594340813e-05,
+      "loss": 0.751,
+      "step": 9720
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5218358612511328e-05,
+      "loss": 0.7305,
+      "step": 9721
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5217376561526363e-05,
+      "loss": 0.7256,
+      "step": 9722
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5216394441398934e-05,
+      "loss": 0.7324,
+      "step": 9723
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5215412252142055e-05,
+      "loss": 0.7529,
+      "step": 9724
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5214429993768743e-05,
+      "loss": 0.7539,
+      "step": 9725
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213447666292016e-05,
+      "loss": 0.7646,
+      "step": 9726
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.521246526972489e-05,
+      "loss": 0.7588,
+      "step": 9727
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5211482804080385e-05,
+      "loss": 0.7695,
+      "step": 9728
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5210500269371525e-05,
+      "loss": 0.792,
+      "step": 9729
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5209517665611321e-05,
+      "loss": 0.7578,
+      "step": 9730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5208534992812806e-05,
+      "loss": 0.7607,
+      "step": 9731
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5207552250989e-05,
+      "loss": 0.7646,
+      "step": 9732
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5206569440152919e-05,
+      "loss": 0.7598,
+      "step": 9733
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5205586560317596e-05,
+      "loss": 0.7568,
+      "step": 9734
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5204603611496051e-05,
+      "loss": 0.751,
+      "step": 9735
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5203620593701316e-05,
+      "loss": 0.75,
+      "step": 9736
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5202637506946412e-05,
+      "loss": 0.7441,
+      "step": 9737
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5201654351244375e-05,
+      "loss": 0.7549,
+      "step": 9738
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5200671126608224e-05,
+      "loss": 0.7432,
+      "step": 9739
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5199687833050997e-05,
+      "loss": 0.7461,
+      "step": 9740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5198704470585722e-05,
+      "loss": 0.7168,
+      "step": 9741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5197721039225433e-05,
+      "loss": 0.7646,
+      "step": 9742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5196737538983159e-05,
+      "loss": 0.7676,
+      "step": 9743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5195753969871936e-05,
+      "loss": 0.7695,
+      "step": 9744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.51947703319048e-05,
+      "loss": 0.7432,
+      "step": 9745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193786625094784e-05,
+      "loss": 0.7627,
+      "step": 9746
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5192802849454925e-05,
+      "loss": 0.7666,
+      "step": 9747
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5191819004998261e-05,
+      "loss": 0.7646,
+      "step": 9748
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.519083509173783e-05,
+      "loss": 0.7822,
+      "step": 9749
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5189851109686671e-05,
+      "loss": 0.7666,
+      "step": 9750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5188867058857822e-05,
+      "loss": 0.7637,
+      "step": 9751
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5187882939264328e-05,
+      "loss": 0.7451,
+      "step": 9752
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5186898750919227e-05,
+      "loss": 0.7744,
+      "step": 9753
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185914493835565e-05,
+      "loss": 0.752,
+      "step": 9754
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5184930168026386e-05,
+      "loss": 0.7471,
+      "step": 9755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5183945773504731e-05,
+      "loss": 0.7402,
+      "step": 9756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.518296131028365e-05,
+      "loss": 0.7383,
+      "step": 9757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5181976778376186e-05,
+      "loss": 0.7549,
+      "step": 9758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5180992177795386e-05,
+      "loss": 0.749,
+      "step": 9759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5180007508554302e-05,
+      "loss": 0.7344,
+      "step": 9760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5179022770665978e-05,
+      "loss": 0.751,
+      "step": 9761
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5178037964143469e-05,
+      "loss": 0.7539,
+      "step": 9762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5177053088999824e-05,
+      "loss": 0.7656,
+      "step": 9763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5176068145248093e-05,
+      "loss": 0.7412,
+      "step": 9764
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5175083132901335e-05,
+      "loss": 0.7363,
+      "step": 9765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5174098051972594e-05,
+      "loss": 0.7637,
+      "step": 9766
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5173112902474931e-05,
+      "loss": 0.7461,
+      "step": 9767
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5172127684421402e-05,
+      "loss": 0.7686,
+      "step": 9768
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5171142397825062e-05,
+      "loss": 0.7471,
+      "step": 9769
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5170157042698968e-05,
+      "loss": 0.752,
+      "step": 9770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516917161905618e-05,
+      "loss": 0.7578,
+      "step": 9771
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5168186126909752e-05,
+      "loss": 0.7373,
+      "step": 9772
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5167200566272753e-05,
+      "loss": 0.7383,
+      "step": 9773
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5166214937158236e-05,
+      "loss": 0.7266,
+      "step": 9774
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5165229239579264e-05,
+      "loss": 0.752,
+      "step": 9775
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5164243473548903e-05,
+      "loss": 0.7617,
+      "step": 9776
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5163257639080218e-05,
+      "loss": 0.751,
+      "step": 9777
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5162271736186266e-05,
+      "loss": 0.7607,
+      "step": 9778
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5161285764880122e-05,
+      "loss": 0.7422,
+      "step": 9779
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5160299725174845e-05,
+      "loss": 0.75,
+      "step": 9780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5159313617083507e-05,
+      "loss": 0.7783,
+      "step": 9781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5158327440619172e-05,
+      "loss": 0.749,
+      "step": 9782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5157341195794912e-05,
+      "loss": 0.7402,
+      "step": 9783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5156354882623798e-05,
+      "loss": 0.7207,
+      "step": 9784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51553685011189e-05,
+      "loss": 0.7568,
+      "step": 9785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5154382051293289e-05,
+      "loss": 0.7539,
+      "step": 9786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5153395533160038e-05,
+      "loss": 0.7764,
+      "step": 9787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5152408946732218e-05,
+      "loss": 0.7393,
+      "step": 9788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515142229202291e-05,
+      "loss": 0.7637,
+      "step": 9789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5150435569045186e-05,
+      "loss": 0.7471,
+      "step": 9790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149448777812121e-05,
+      "loss": 0.7773,
+      "step": 9791
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5148461918336791e-05,
+      "loss": 0.7393,
+      "step": 9792
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.514747499063228e-05,
+      "loss": 0.7217,
+      "step": 9793
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5146487994711665e-05,
+      "loss": 0.7373,
+      "step": 9794
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5145500930588024e-05,
+      "loss": 0.7559,
+      "step": 9795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5144513798274439e-05,
+      "loss": 0.7529,
+      "step": 9796
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5143526597783993e-05,
+      "loss": 0.749,
+      "step": 9797
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5142539329129764e-05,
+      "loss": 0.7539,
+      "step": 9798
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141551992324841e-05,
+      "loss": 0.7754,
+      "step": 9799
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5140564587382307e-05,
+      "loss": 0.7773,
+      "step": 9800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5139577114315245e-05,
+      "loss": 0.7246,
+      "step": 9801
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5138589573136745e-05,
+      "loss": 0.7402,
+      "step": 9802
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5137601963859894e-05,
+      "loss": 0.749,
+      "step": 9803
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5136614286497775e-05,
+      "loss": 0.7217,
+      "step": 9804
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5135626541063482e-05,
+      "loss": 0.7354,
+      "step": 9805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5134638727570105e-05,
+      "loss": 0.7324,
+      "step": 9806
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5133650846030731e-05,
+      "loss": 0.7441,
+      "step": 9807
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513266289645846e-05,
+      "loss": 0.7363,
+      "step": 9808
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5131674878866372e-05,
+      "loss": 0.7363,
+      "step": 9809
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513068679326757e-05,
+      "loss": 0.7334,
+      "step": 9810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5129698639675146e-05,
+      "loss": 0.7314,
+      "step": 9811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5128710418102195e-05,
+      "loss": 0.7373,
+      "step": 9812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5127722128561815e-05,
+      "loss": 0.7598,
+      "step": 9813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51267337710671e-05,
+      "loss": 0.7676,
+      "step": 9814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125745345631148e-05,
+      "loss": 0.748,
+      "step": 9815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5124756852267065e-05,
+      "loss": 0.7402,
+      "step": 9816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5123768290987942e-05,
+      "loss": 0.7422,
+      "step": 9817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5122779661806884e-05,
+      "loss": 0.7646,
+      "step": 9818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.512179096473699e-05,
+      "loss": 0.7529,
+      "step": 9819
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5120802199791367e-05,
+      "loss": 0.7363,
+      "step": 9820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5119813366983114e-05,
+      "loss": 0.7588,
+      "step": 9821
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5118824466325336e-05,
+      "loss": 0.751,
+      "step": 9822
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5117835497831143e-05,
+      "loss": 0.7666,
+      "step": 9823
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.511684646151364e-05,
+      "loss": 0.7451,
+      "step": 9824
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5115857357385926e-05,
+      "loss": 0.7441,
+      "step": 9825
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5114868185461117e-05,
+      "loss": 0.7705,
+      "step": 9826
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5113878945752319e-05,
+      "loss": 0.7715,
+      "step": 9827
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5112889638272643e-05,
+      "loss": 0.7686,
+      "step": 9828
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.51119002630352e-05,
+      "loss": 0.7539,
+      "step": 9829
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5110910820053097e-05,
+      "loss": 0.7236,
+      "step": 9830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5109921309339451e-05,
+      "loss": 0.7725,
+      "step": 9831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5108931730907376e-05,
+      "loss": 0.7295,
+      "step": 9832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5107942084769981e-05,
+      "loss": 0.7432,
+      "step": 9833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5106952370940388e-05,
+      "loss": 0.7432,
+      "step": 9834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5105962589431709e-05,
+      "loss": 0.752,
+      "step": 9835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5104972740257062e-05,
+      "loss": 0.7383,
+      "step": 9836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5103982823429562e-05,
+      "loss": 0.7529,
+      "step": 9837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5102992838962332e-05,
+      "loss": 0.7646,
+      "step": 9838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5102002786868491e-05,
+      "loss": 0.7471,
+      "step": 9839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5101012667161157e-05,
+      "loss": 0.7275,
+      "step": 9840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5100022479853451e-05,
+      "loss": 0.7539,
+      "step": 9841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5099032224958498e-05,
+      "loss": 0.7275,
+      "step": 9842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509804190248942e-05,
+      "loss": 0.7588,
+      "step": 9843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509705151245934e-05,
+      "loss": 0.7617,
+      "step": 9844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5096061054881385e-05,
+      "loss": 0.749,
+      "step": 9845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5095070529768683e-05,
+      "loss": 0.7764,
+      "step": 9846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5094079937134357e-05,
+      "loss": 0.752,
+      "step": 9847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509308927699153e-05,
+      "loss": 0.7305,
+      "step": 9848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509209854935334e-05,
+      "loss": 0.7305,
+      "step": 9849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5091107754232911e-05,
+      "loss": 0.7588,
+      "step": 9850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5090116891643378e-05,
+      "loss": 0.749,
+      "step": 9851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.508912596159787e-05,
+      "loss": 0.7578,
+      "step": 9852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5088134964109516e-05,
+      "loss": 0.7383,
+      "step": 9853
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.508714389919145e-05,
+      "loss": 0.7627,
+      "step": 9854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.508615276685681e-05,
+      "loss": 0.7568,
+      "step": 9855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5085161567118725e-05,
+      "loss": 0.7432,
+      "step": 9856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5084170299990335e-05,
+      "loss": 0.7588,
+      "step": 9857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5083178965484777e-05,
+      "loss": 0.7256,
+      "step": 9858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5082187563615187e-05,
+      "loss": 0.7295,
+      "step": 9859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.50811960943947e-05,
+      "loss": 0.7461,
+      "step": 9860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5080204557836462e-05,
+      "loss": 0.7422,
+      "step": 9861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5079212953953609e-05,
+      "loss": 0.7598,
+      "step": 9862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5078221282759283e-05,
+      "loss": 0.7412,
+      "step": 9863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.507722954426662e-05,
+      "loss": 0.748,
+      "step": 9864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5076237738488776e-05,
+      "loss": 0.7295,
+      "step": 9865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5075245865438886e-05,
+      "loss": 0.7432,
+      "step": 9866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5074253925130095e-05,
+      "loss": 0.7725,
+      "step": 9867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5073261917575552e-05,
+      "loss": 0.7783,
+      "step": 9868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5072269842788399e-05,
+      "loss": 0.7451,
+      "step": 9869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5071277700781783e-05,
+      "loss": 0.7666,
+      "step": 9870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5070285491568858e-05,
+      "loss": 0.7646,
+      "step": 9871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5069293215162768e-05,
+      "loss": 0.75,
+      "step": 9872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5068300871576666e-05,
+      "loss": 0.7715,
+      "step": 9873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.50673084608237e-05,
+      "loss": 0.7432,
+      "step": 9874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066315982917026e-05,
+      "loss": 0.7588,
+      "step": 9875
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.506532343786979e-05,
+      "loss": 0.7422,
+      "step": 9876
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5064330825695153e-05,
+      "loss": 0.7441,
+      "step": 9877
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5063338146406266e-05,
+      "loss": 0.7207,
+      "step": 9878
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5062345400016283e-05,
+      "loss": 0.7393,
+      "step": 9879
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.506135258653836e-05,
+      "loss": 0.7529,
+      "step": 9880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5060359705985658e-05,
+      "loss": 0.7451,
+      "step": 9881
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5059366758371333e-05,
+      "loss": 0.7246,
+      "step": 9882
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505837374370854e-05,
+      "loss": 0.7324,
+      "step": 9883
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5057380662010446e-05,
+      "loss": 0.7783,
+      "step": 9884
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5056387513290207e-05,
+      "loss": 0.7646,
+      "step": 9885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5055394297560986e-05,
+      "loss": 0.7578,
+      "step": 9886
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505440101483594e-05,
+      "loss": 0.7393,
+      "step": 9887
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505340766512824e-05,
+      "loss": 0.7588,
+      "step": 9888
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505241424845105e-05,
+      "loss": 0.7363,
+      "step": 9889
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5051420764817533e-05,
+      "loss": 0.7734,
+      "step": 9890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5050427214240851e-05,
+      "loss": 0.7422,
+      "step": 9891
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5049433596734178e-05,
+      "loss": 0.7646,
+      "step": 9892
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5048439912310673e-05,
+      "loss": 0.7461,
+      "step": 9893
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5047446160983514e-05,
+      "loss": 0.75,
+      "step": 9894
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5046452342765863e-05,
+      "loss": 0.7637,
+      "step": 9895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5045458457670896e-05,
+      "loss": 0.7656,
+      "step": 9896
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5044464505711783e-05,
+      "loss": 0.7256,
+      "step": 9897
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5043470486901694e-05,
+      "loss": 0.752,
+      "step": 9898
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5042476401253803e-05,
+      "loss": 0.7441,
+      "step": 9899
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5041482248781285e-05,
+      "loss": 0.7461,
+      "step": 9900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5040488029497312e-05,
+      "loss": 0.7559,
+      "step": 9901
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5039493743415063e-05,
+      "loss": 0.7783,
+      "step": 9902
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5038499390547714e-05,
+      "loss": 0.7676,
+      "step": 9903
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5037504970908443e-05,
+      "loss": 0.7744,
+      "step": 9904
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036510484510425e-05,
+      "loss": 0.7461,
+      "step": 9905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5035515931366844e-05,
+      "loss": 0.7393,
+      "step": 9906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503452131149088e-05,
+      "loss": 0.7344,
+      "step": 9907
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5033526624895706e-05,
+      "loss": 0.7666,
+      "step": 9908
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5032531871594511e-05,
+      "loss": 0.7588,
+      "step": 9909
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503153705160048e-05,
+      "loss": 0.7529,
+      "step": 9910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503054216492679e-05,
+      "loss": 0.7363,
+      "step": 9911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5029547211586632e-05,
+      "loss": 0.7617,
+      "step": 9912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5028552191593187e-05,
+      "loss": 0.7432,
+      "step": 9913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5027557104959643e-05,
+      "loss": 0.7715,
+      "step": 9914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5026561951699185e-05,
+      "loss": 0.7734,
+      "step": 9915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5025566731825004e-05,
+      "loss": 0.75,
+      "step": 9916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5024571445350289e-05,
+      "loss": 0.7559,
+      "step": 9917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.502357609228823e-05,
+      "loss": 0.7637,
+      "step": 9918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5022580672652012e-05,
+      "loss": 0.7451,
+      "step": 9919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5021585186454836e-05,
+      "loss": 0.7461,
+      "step": 9920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5020589633709888e-05,
+      "loss": 0.748,
+      "step": 9921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5019594014430361e-05,
+      "loss": 0.7607,
+      "step": 9922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5018598328629454e-05,
+      "loss": 0.7324,
+      "step": 9923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5017602576320362e-05,
+      "loss": 0.748,
+      "step": 9924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5016606757516274e-05,
+      "loss": 0.752,
+      "step": 9925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5015610872230397e-05,
+      "loss": 0.7422,
+      "step": 9926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5014614920475918e-05,
+      "loss": 0.7324,
+      "step": 9927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5013618902266044e-05,
+      "loss": 0.7432,
+      "step": 9928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5012622817613971e-05,
+      "loss": 0.7256,
+      "step": 9929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5011626666532901e-05,
+      "loss": 0.7793,
+      "step": 9930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5010630449036035e-05,
+      "loss": 0.7334,
+      "step": 9931
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5009634165136573e-05,
+      "loss": 0.7383,
+      "step": 9932
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500863781484772e-05,
+      "loss": 0.7402,
+      "step": 9933
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5007641398182681e-05,
+      "loss": 0.7715,
+      "step": 9934
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500664491515466e-05,
+      "loss": 0.7666,
+      "step": 9935
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5005648365776862e-05,
+      "loss": 0.7402,
+      "step": 9936
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5004651750062493e-05,
+      "loss": 0.7344,
+      "step": 9937
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5003655068024762e-05,
+      "loss": 0.7432,
+      "step": 9938
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500265831967688e-05,
+      "loss": 0.7529,
+      "step": 9939
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5001661505032049e-05,
+      "loss": 0.75,
+      "step": 9940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000664624103484e-05,
+      "loss": 0.748,
+      "step": 9941
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4999667676904395e-05,
+      "loss": 0.7715,
+      "step": 9942
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4998670663448e-05,
+      "loss": 0.7344,
+      "step": 9943
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4997673583747499e-05,
+      "loss": 0.7324,
+      "step": 9944
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4996676437816114e-05,
+      "loss": 0.7744,
+      "step": 9945
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4995679225667059e-05,
+      "loss": 0.7441,
+      "step": 9946
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4994681947313547e-05,
+      "loss": 0.7412,
+      "step": 9947
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4993684602768796e-05,
+      "loss": 0.7412,
+      "step": 9948
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4992687192046028e-05,
+      "loss": 0.7334,
+      "step": 9949
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991689715158448e-05,
+      "loss": 0.7539,
+      "step": 9950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499069217211929e-05,
+      "loss": 0.7803,
+      "step": 9951
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4989694562941763e-05,
+      "loss": 0.7393,
+      "step": 9952
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4988696887639092e-05,
+      "loss": 0.7402,
+      "step": 9953
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4987699146224497e-05,
+      "loss": 0.7754,
+      "step": 9954
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4986701338711202e-05,
+      "loss": 0.79,
+      "step": 9955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4985703465112428e-05,
+      "loss": 0.7314,
+      "step": 9956
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4984705525441406e-05,
+      "loss": 0.7715,
+      "step": 9957
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4983707519711349e-05,
+      "loss": 0.7725,
+      "step": 9958
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4982709447935494e-05,
+      "loss": 0.7451,
+      "step": 9959
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4981711310127062e-05,
+      "loss": 0.748,
+      "step": 9960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4980713106299283e-05,
+      "loss": 0.7695,
+      "step": 9961
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4979714836465383e-05,
+      "loss": 0.7598,
+      "step": 9962
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4978716500638595e-05,
+      "loss": 0.7422,
+      "step": 9963
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4977718098832146e-05,
+      "loss": 0.7383,
+      "step": 9964
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4976719631059272e-05,
+      "loss": 0.7822,
+      "step": 9965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4975721097333193e-05,
+      "loss": 0.748,
+      "step": 9966
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4974722497667158e-05,
+      "loss": 0.7422,
+      "step": 9967
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4973723832074391e-05,
+      "loss": 0.7617,
+      "step": 9968
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497272510056813e-05,
+      "loss": 0.7559,
+      "step": 9969
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4971726303161608e-05,
+      "loss": 0.7666,
+      "step": 9970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4970727439868063e-05,
+      "loss": 0.7568,
+      "step": 9971
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4969728510700735e-05,
+      "loss": 0.7539,
+      "step": 9972
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4968729515672856e-05,
+      "loss": 0.748,
+      "step": 9973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4967730454797669e-05,
+      "loss": 0.7422,
+      "step": 9974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4966731328088414e-05,
+      "loss": 0.7295,
+      "step": 9975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496573213555833e-05,
+      "loss": 0.7539,
+      "step": 9976
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4964732877220658e-05,
+      "loss": 0.7773,
+      "step": 9977
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496373355308865e-05,
+      "loss": 0.7607,
+      "step": 9978
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4962734163175532e-05,
+      "loss": 0.7393,
+      "step": 9979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4961734707494561e-05,
+      "loss": 0.7568,
+      "step": 9980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4960735186058979e-05,
+      "loss": 0.7422,
+      "step": 9981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959735598882036e-05,
+      "loss": 0.7793,
+      "step": 9982
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4958735945976967e-05,
+      "loss": 0.7451,
+      "step": 9983
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4957736227357034e-05,
+      "loss": 0.7578,
+      "step": 9984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4956736443035476e-05,
+      "loss": 0.7471,
+      "step": 9985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4955736593025547e-05,
+      "loss": 0.7598,
+      "step": 9986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4954736677340494e-05,
+      "loss": 0.7471,
+      "step": 9987
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4953736695993573e-05,
+      "loss": 0.7598,
+      "step": 9988
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4952736648998032e-05,
+      "loss": 0.7373,
+      "step": 9989
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4951736536367125e-05,
+      "loss": 0.7607,
+      "step": 9990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4950736358114106e-05,
+      "loss": 0.7715,
+      "step": 9991
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4949736114252231e-05,
+      "loss": 0.7432,
+      "step": 9992
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4948735804794752e-05,
+      "loss": 0.7441,
+      "step": 9993
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494773542975493e-05,
+      "loss": 0.7354,
+      "step": 9994
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494673498914602e-05,
+      "loss": 0.7246,
+      "step": 9995
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494573448298128e-05,
+      "loss": 0.7412,
+      "step": 9996
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4944733911273967e-05,
+      "loss": 0.7529,
+      "step": 9997
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4943733274037348e-05,
+      "loss": 0.7764,
+      "step": 9998
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4942732571284677e-05,
+      "loss": 0.7383,
+      "step": 9999
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4941731803029216e-05,
+      "loss": 0.749,
+      "step": 10000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4940730969284231e-05,
+      "loss": 0.7334,
+      "step": 10001
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4939730070062983e-05,
+      "loss": 0.752,
+      "step": 10002
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4938729105378738e-05,
+      "loss": 0.7578,
+      "step": 10003
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4937728075244762e-05,
+      "loss": 0.7451,
+      "step": 10004
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4936726979674314e-05,
+      "loss": 0.7412,
+      "step": 10005
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493572581868067e-05,
+      "loss": 0.7461,
+      "step": 10006
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493472459227709e-05,
+      "loss": 0.7344,
+      "step": 10007
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4933723300476847e-05,
+      "loss": 0.7334,
+      "step": 10008
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4932721943293211e-05,
+      "loss": 0.7402,
+      "step": 10009
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4931720520739452e-05,
+      "loss": 0.7451,
+      "step": 10010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4930719032828838e-05,
+      "loss": 0.7354,
+      "step": 10011
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4929717479574646e-05,
+      "loss": 0.7393,
+      "step": 10012
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4928715860990144e-05,
+      "loss": 0.7676,
+      "step": 10013
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492771417708861e-05,
+      "loss": 0.7529,
+      "step": 10014
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4926712427883315e-05,
+      "loss": 0.752,
+      "step": 10015
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4925710613387536e-05,
+      "loss": 0.7539,
+      "step": 10016
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4924708733614548e-05,
+      "loss": 0.7373,
+      "step": 10017
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4923706788577632e-05,
+      "loss": 0.7559,
+      "step": 10018
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4922704778290067e-05,
+      "loss": 0.752,
+      "step": 10019
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4921702702765122e-05,
+      "loss": 0.7617,
+      "step": 10020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492070056201609e-05,
+      "loss": 0.7559,
+      "step": 10021
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4919698356056242e-05,
+      "loss": 0.7471,
+      "step": 10022
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4918696084898865e-05,
+      "loss": 0.7402,
+      "step": 10023
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4917693748557238e-05,
+      "loss": 0.7754,
+      "step": 10024
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4916691347044648e-05,
+      "loss": 0.751,
+      "step": 10025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4915688880374372e-05,
+      "loss": 0.749,
+      "step": 10026
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4914686348559706e-05,
+      "loss": 0.7539,
+      "step": 10027
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4913683751613928e-05,
+      "loss": 0.7578,
+      "step": 10028
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4912681089550326e-05,
+      "loss": 0.7588,
+      "step": 10029
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4911678362382187e-05,
+      "loss": 0.7656,
+      "step": 10030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910675570122804e-05,
+      "loss": 0.7578,
+      "step": 10031
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490967271278546e-05,
+      "loss": 0.752,
+      "step": 10032
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4908669790383448e-05,
+      "loss": 0.7686,
+      "step": 10033
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4907666802930061e-05,
+      "loss": 0.7383,
+      "step": 10034
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490666375043859e-05,
+      "loss": 0.7588,
+      "step": 10035
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4905660632922325e-05,
+      "loss": 0.7559,
+      "step": 10036
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4904657450394566e-05,
+      "loss": 0.7393,
+      "step": 10037
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4903654202868597e-05,
+      "loss": 0.7354,
+      "step": 10038
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4902650890357726e-05,
+      "loss": 0.7461,
+      "step": 10039
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490164751287524e-05,
+      "loss": 0.751,
+      "step": 10040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4900644070434438e-05,
+      "loss": 0.748,
+      "step": 10041
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.489964056304862e-05,
+      "loss": 0.7432,
+      "step": 10042
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4898636990731086e-05,
+      "loss": 0.7344,
+      "step": 10043
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4897633353495131e-05,
+      "loss": 0.752,
+      "step": 10044
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4896629651354057e-05,
+      "loss": 0.752,
+      "step": 10045
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4895625884321169e-05,
+      "loss": 0.7471,
+      "step": 10046
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4894622052409767e-05,
+      "loss": 0.7783,
+      "step": 10047
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.489361815563315e-05,
+      "loss": 0.7607,
+      "step": 10048
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4892614194004628e-05,
+      "loss": 0.7432,
+      "step": 10049
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4891610167537507e-05,
+      "loss": 0.7412,
+      "step": 10050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4890606076245087e-05,
+      "loss": 0.7402,
+      "step": 10051
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4889601920140674e-05,
+      "loss": 0.7754,
+      "step": 10052
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4888597699237584e-05,
+      "loss": 0.7607,
+      "step": 10053
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4887593413549115e-05,
+      "loss": 0.7344,
+      "step": 10054
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4886589063088582e-05,
+      "loss": 0.7695,
+      "step": 10055
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4885584647869294e-05,
+      "loss": 0.7617,
+      "step": 10056
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4884580167904563e-05,
+      "loss": 0.7441,
+      "step": 10057
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4883575623207697e-05,
+      "loss": 0.7354,
+      "step": 10058
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4882571013792013e-05,
+      "loss": 0.7637,
+      "step": 10059
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.488156633967082e-05,
+      "loss": 0.7549,
+      "step": 10060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4880561600857437e-05,
+      "loss": 0.7852,
+      "step": 10061
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4879556797365175e-05,
+      "loss": 0.748,
+      "step": 10062
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4878551929207352e-05,
+      "loss": 0.7646,
+      "step": 10063
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4877546996397287e-05,
+      "loss": 0.7451,
+      "step": 10064
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4876541998948292e-05,
+      "loss": 0.749,
+      "step": 10065
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4875536936873692e-05,
+      "loss": 0.752,
+      "step": 10066
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.48745318101868e-05,
+      "loss": 0.7422,
+      "step": 10067
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4873526618900942e-05,
+      "loss": 0.7607,
+      "step": 10068
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4872521363029437e-05,
+      "loss": 0.75,
+      "step": 10069
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4871516042585608e-05,
+      "loss": 0.7334,
+      "step": 10070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4870510657582773e-05,
+      "loss": 0.7529,
+      "step": 10071
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4869505208034262e-05,
+      "loss": 0.7578,
+      "step": 10072
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4868499693953396e-05,
+      "loss": 0.7451,
+      "step": 10073
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.48674941153535e-05,
+      "loss": 0.7705,
+      "step": 10074
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4866488472247902e-05,
+      "loss": 0.7744,
+      "step": 10075
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865482764649933e-05,
+      "loss": 0.748,
+      "step": 10076
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4864476992572909e-05,
+      "loss": 0.7373,
+      "step": 10077
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4863471156030172e-05,
+      "loss": 0.748,
+      "step": 10078
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4862465255035045e-05,
+      "loss": 0.7393,
+      "step": 10079
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4861459289600859e-05,
+      "loss": 0.7402,
+      "step": 10080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4860453259740945e-05,
+      "loss": 0.7617,
+      "step": 10081
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485944716546864e-05,
+      "loss": 0.7539,
+      "step": 10082
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4858441006797272e-05,
+      "loss": 0.7354,
+      "step": 10083
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4857434783740176e-05,
+      "loss": 0.7246,
+      "step": 10084
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4856428496310689e-05,
+      "loss": 0.7617,
+      "step": 10085
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4855422144522144e-05,
+      "loss": 0.7666,
+      "step": 10086
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4854415728387876e-05,
+      "loss": 0.7383,
+      "step": 10087
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4853409247921226e-05,
+      "loss": 0.7393,
+      "step": 10088
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4852402703135533e-05,
+      "loss": 0.7656,
+      "step": 10089
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4851396094044134e-05,
+      "loss": 0.7451,
+      "step": 10090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485038942066037e-05,
+      "loss": 0.7461,
+      "step": 10091
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4849382682997578e-05,
+      "loss": 0.7539,
+      "step": 10092
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4848375881069101e-05,
+      "loss": 0.708,
+      "step": 10093
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.484736901488829e-05,
+      "loss": 0.748,
+      "step": 10094
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4846362084468475e-05,
+      "loss": 0.7324,
+      "step": 10095
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4845355089823008e-05,
+      "loss": 0.7383,
+      "step": 10096
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4844348030965235e-05,
+      "loss": 0.7412,
+      "step": 10097
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4843340907908497e-05,
+      "loss": 0.7744,
+      "step": 10098
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4842333720666142e-05,
+      "loss": 0.7588,
+      "step": 10099
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.484132646925152e-05,
+      "loss": 0.7461,
+      "step": 10100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.484031915367798e-05,
+      "loss": 0.7578,
+      "step": 10101
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4839311773958866e-05,
+      "loss": 0.7617,
+      "step": 10102
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4838304330107536e-05,
+      "loss": 0.7344,
+      "step": 10103
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4837296822137333e-05,
+      "loss": 0.7305,
+      "step": 10104
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4836289250061614e-05,
+      "loss": 0.7402,
+      "step": 10105
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483528161389373e-05,
+      "loss": 0.7539,
+      "step": 10106
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4834273913647034e-05,
+      "loss": 0.7715,
+      "step": 10107
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4833266149334884e-05,
+      "loss": 0.7305,
+      "step": 10108
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483225832097063e-05,
+      "loss": 0.7783,
+      "step": 10109
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4831250428567629e-05,
+      "loss": 0.7471,
+      "step": 10110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4830242472139242e-05,
+      "loss": 0.7432,
+      "step": 10111
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4829234451698822e-05,
+      "loss": 0.752,
+      "step": 10112
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4828226367259731e-05,
+      "loss": 0.7412,
+      "step": 10113
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4827218218835326e-05,
+      "loss": 0.7725,
+      "step": 10114
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826210006438973e-05,
+      "loss": 0.7549,
+      "step": 10115
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4825201730084024e-05,
+      "loss": 0.7539,
+      "step": 10116
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4824193389783848e-05,
+      "loss": 0.7568,
+      "step": 10117
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4823184985551805e-05,
+      "loss": 0.7207,
+      "step": 10118
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482217651740126e-05,
+      "loss": 0.7422,
+      "step": 10119
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4821167985345573e-05,
+      "loss": 0.7656,
+      "step": 10120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4820159389398119e-05,
+      "loss": 0.7539,
+      "step": 10121
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4819150729572256e-05,
+      "loss": 0.7451,
+      "step": 10122
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4818142005881352e-05,
+      "loss": 0.7158,
+      "step": 10123
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481713321833878e-05,
+      "loss": 0.7529,
+      "step": 10124
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4816124366957904e-05,
+      "loss": 0.7725,
+      "step": 10125
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4815115451752094e-05,
+      "loss": 0.7441,
+      "step": 10126
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4814106472734723e-05,
+      "loss": 0.7295,
+      "step": 10127
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481309742991916e-05,
+      "loss": 0.7305,
+      "step": 10128
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481208832331878e-05,
+      "loss": 0.75,
+      "step": 10129
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4811079152946954e-05,
+      "loss": 0.7471,
+      "step": 10130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4810069918817056e-05,
+      "loss": 0.7432,
+      "step": 10131
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4809060620942456e-05,
+      "loss": 0.7588,
+      "step": 10132
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480805125933654e-05,
+      "loss": 0.7363,
+      "step": 10133
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4807041834012678e-05,
+      "loss": 0.7598,
+      "step": 10134
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4806032344984247e-05,
+      "loss": 0.752,
+      "step": 10135
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4805022792264625e-05,
+      "loss": 0.7236,
+      "step": 10136
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4804013175867194e-05,
+      "loss": 0.7354,
+      "step": 10137
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480300349580533e-05,
+      "loss": 0.7725,
+      "step": 10138
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4801993752092416e-05,
+      "loss": 0.7275,
+      "step": 10139
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4800983944741833e-05,
+      "loss": 0.7646,
+      "step": 10140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4799974073766962e-05,
+      "loss": 0.7715,
+      "step": 10141
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4798964139181191e-05,
+      "loss": 0.7412,
+      "step": 10142
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4797954140997894e-05,
+      "loss": 0.7646,
+      "step": 10143
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4796944079230468e-05,
+      "loss": 0.7354,
+      "step": 10144
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479593395389229e-05,
+      "loss": 0.7549,
+      "step": 10145
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4794923764996752e-05,
+      "loss": 0.7441,
+      "step": 10146
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4793913512557236e-05,
+      "loss": 0.7646,
+      "step": 10147
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792903196587137e-05,
+      "loss": 0.7314,
+      "step": 10148
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4791892817099832e-05,
+      "loss": 0.7412,
+      "step": 10149
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4790882374108728e-05,
+      "loss": 0.7432,
+      "step": 10150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4789871867627201e-05,
+      "loss": 0.7568,
+      "step": 10151
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.478886129766865e-05,
+      "loss": 0.7402,
+      "step": 10152
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4787850664246465e-05,
+      "loss": 0.7539,
+      "step": 10153
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4786839967374042e-05,
+      "loss": 0.7393,
+      "step": 10154
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.478582920706477e-05,
+      "loss": 0.748,
+      "step": 10155
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4784818383332049e-05,
+      "loss": 0.7578,
+      "step": 10156
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4783807496189273e-05,
+      "loss": 0.7305,
+      "step": 10157
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4782796545649838e-05,
+      "loss": 0.7725,
+      "step": 10158
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.478178553172714e-05,
+      "loss": 0.7295,
+      "step": 10159
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.478077445443458e-05,
+      "loss": 0.7305,
+      "step": 10160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4779763313785558e-05,
+      "loss": 0.7461,
+      "step": 10161
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4778752109793471e-05,
+      "loss": 0.748,
+      "step": 10162
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4777740842471723e-05,
+      "loss": 0.7539,
+      "step": 10163
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4776729511833712e-05,
+      "loss": 0.7422,
+      "step": 10164
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4775718117892842e-05,
+      "loss": 0.7539,
+      "step": 10165
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4774706660662516e-05,
+      "loss": 0.75,
+      "step": 10166
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.477369514015614e-05,
+      "loss": 0.7324,
+      "step": 10167
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4772683556387117e-05,
+      "loss": 0.7393,
+      "step": 10168
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4771671909368855e-05,
+      "loss": 0.751,
+      "step": 10169
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4770660199114759e-05,
+      "loss": 0.7441,
+      "step": 10170
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4769648425638235e-05,
+      "loss": 0.7256,
+      "step": 10171
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4768636588952694e-05,
+      "loss": 0.7393,
+      "step": 10172
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4767624689071545e-05,
+      "loss": 0.7207,
+      "step": 10173
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766612726008198e-05,
+      "loss": 0.7549,
+      "step": 10174
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4765600699776064e-05,
+      "loss": 0.7402,
+      "step": 10175
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4764588610388554e-05,
+      "loss": 0.7617,
+      "step": 10176
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4763576457859077e-05,
+      "loss": 0.7412,
+      "step": 10177
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4762564242201055e-05,
+      "loss": 0.7451,
+      "step": 10178
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4761551963427894e-05,
+      "loss": 0.7227,
+      "step": 10179
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4760539621553014e-05,
+      "loss": 0.7412,
+      "step": 10180
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4759527216589828e-05,
+      "loss": 0.7529,
+      "step": 10181
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4758514748551753e-05,
+      "loss": 0.7471,
+      "step": 10182
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475750221745221e-05,
+      "loss": 0.7559,
+      "step": 10183
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4756489623304614e-05,
+      "loss": 0.752,
+      "step": 10184
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4755476966122387e-05,
+      "loss": 0.7373,
+      "step": 10185
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4754464245918944e-05,
+      "loss": 0.7676,
+      "step": 10186
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475345146270771e-05,
+      "loss": 0.7812,
+      "step": 10187
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4752438616502107e-05,
+      "loss": 0.7461,
+      "step": 10188
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4751425707315558e-05,
+      "loss": 0.7354,
+      "step": 10189
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475041273516148e-05,
+      "loss": 0.792,
+      "step": 10190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4749399700053309e-05,
+      "loss": 0.792,
+      "step": 10191
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4748386602004458e-05,
+      "loss": 0.7637,
+      "step": 10192
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4747373441028361e-05,
+      "loss": 0.7324,
+      "step": 10193
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474636021713844e-05,
+      "loss": 0.7314,
+      "step": 10194
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4745346930348126e-05,
+      "loss": 0.7656,
+      "step": 10195
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4744333580670844e-05,
+      "loss": 0.7686,
+      "step": 10196
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4743320168120027e-05,
+      "loss": 0.7363,
+      "step": 10197
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4742306692709103e-05,
+      "loss": 0.7471,
+      "step": 10198
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4741293154451506e-05,
+      "loss": 0.7598,
+      "step": 10199
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474027955336066e-05,
+      "loss": 0.752,
+      "step": 10200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4739265889450005e-05,
+      "loss": 0.7578,
+      "step": 10201
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4738252162732974e-05,
+      "loss": 0.7637,
+      "step": 10202
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4737238373222997e-05,
+      "loss": 0.7627,
+      "step": 10203
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4736224520933511e-05,
+      "loss": 0.7383,
+      "step": 10204
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4735210605877954e-05,
+      "loss": 0.7646,
+      "step": 10205
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.473419662806976e-05,
+      "loss": 0.748,
+      "step": 10206
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4733182587522368e-05,
+      "loss": 0.7432,
+      "step": 10207
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4732168484249214e-05,
+      "loss": 0.7568,
+      "step": 10208
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4731154318263743e-05,
+      "loss": 0.7422,
+      "step": 10209
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4730140089579388e-05,
+      "loss": 0.7539,
+      "step": 10210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4729125798209593e-05,
+      "loss": 0.7402,
+      "step": 10211
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4728111444167803e-05,
+      "loss": 0.751,
+      "step": 10212
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.472709702746746e-05,
+      "loss": 0.7471,
+      "step": 10213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4726082548121998e-05,
+      "loss": 0.7432,
+      "step": 10214
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4725068006144873e-05,
+      "loss": 0.7803,
+      "step": 10215
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4724053401549521e-05,
+      "loss": 0.7354,
+      "step": 10216
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4723038734349395e-05,
+      "loss": 0.7539,
+      "step": 10217
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.472202400455794e-05,
+      "loss": 0.7598,
+      "step": 10218
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4721009212188602e-05,
+      "loss": 0.7383,
+      "step": 10219
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4719994357254828e-05,
+      "loss": 0.7637,
+      "step": 10220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4718979439770069e-05,
+      "loss": 0.7295,
+      "step": 10221
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4717964459747778e-05,
+      "loss": 0.75,
+      "step": 10222
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4716949417201401e-05,
+      "loss": 0.7637,
+      "step": 10223
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4715934312144391e-05,
+      "loss": 0.752,
+      "step": 10224
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.47149191445902e-05,
+      "loss": 0.7324,
+      "step": 10225
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4713903914552286e-05,
+      "loss": 0.7529,
+      "step": 10226
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4712888622044096e-05,
+      "loss": 0.751,
+      "step": 10227
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4711873267079094e-05,
+      "loss": 0.7568,
+      "step": 10228
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4710857849670724e-05,
+      "loss": 0.7656,
+      "step": 10229
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4709842369832453e-05,
+      "loss": 0.7236,
+      "step": 10230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4708826827577732e-05,
+      "loss": 0.7705,
+      "step": 10231
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4707811222920025e-05,
+      "loss": 0.7705,
+      "step": 10232
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4706795555872784e-05,
+      "loss": 0.7441,
+      "step": 10233
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4705779826449476e-05,
+      "loss": 0.7656,
+      "step": 10234
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.470476403466356e-05,
+      "loss": 0.7422,
+      "step": 10235
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4703748180528494e-05,
+      "loss": 0.7373,
+      "step": 10236
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4702732264057742e-05,
+      "loss": 0.7256,
+      "step": 10237
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4701716285264773e-05,
+      "loss": 0.7764,
+      "step": 10238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.470070024416304e-05,
+      "loss": 0.7266,
+      "step": 10239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4699684140766019e-05,
+      "loss": 0.7715,
+      "step": 10240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4698667975087172e-05,
+      "loss": 0.7539,
+      "step": 10241
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.469765174713996e-05,
+      "loss": 0.7275,
+      "step": 10242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4696635456937856e-05,
+      "loss": 0.7578,
+      "step": 10243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.469561910449433e-05,
+      "loss": 0.7617,
+      "step": 10244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4694602689822843e-05,
+      "loss": 0.748,
+      "step": 10245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4693586212936874e-05,
+      "loss": 0.752,
+      "step": 10246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4692569673849888e-05,
+      "loss": 0.7666,
+      "step": 10247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4691553072575362e-05,
+      "loss": 0.7432,
+      "step": 10248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.469053640912676e-05,
+      "loss": 0.7334,
+      "step": 10249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4689519683517564e-05,
+      "loss": 0.7295,
+      "step": 10250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468850289576124e-05,
+      "loss": 0.7246,
+      "step": 10251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4687486045871269e-05,
+      "loss": 0.751,
+      "step": 10252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4686469133861124e-05,
+      "loss": 0.7529,
+      "step": 10253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4685452159744282e-05,
+      "loss": 0.7588,
+      "step": 10254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684435123534218e-05,
+      "loss": 0.7412,
+      "step": 10255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4683418025244414e-05,
+      "loss": 0.7627,
+      "step": 10256
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4682400864888346e-05,
+      "loss": 0.749,
+      "step": 10257
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4681383642479498e-05,
+      "loss": 0.7217,
+      "step": 10258
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4680366358031343e-05,
+      "loss": 0.7246,
+      "step": 10259
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467934901155737e-05,
+      "loss": 0.7422,
+      "step": 10260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4678331603071056e-05,
+      "loss": 0.7549,
+      "step": 10261
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4677314132585887e-05,
+      "loss": 0.7236,
+      "step": 10262
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4676296600115348e-05,
+      "loss": 0.7285,
+      "step": 10263
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467527900567292e-05,
+      "loss": 0.7744,
+      "step": 10264
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467426134927209e-05,
+      "loss": 0.7285,
+      "step": 10265
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4673243630926345e-05,
+      "loss": 0.7461,
+      "step": 10266
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4672225850649171e-05,
+      "loss": 0.7529,
+      "step": 10267
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4671208008454054e-05,
+      "loss": 0.7451,
+      "step": 10268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467019010435449e-05,
+      "loss": 0.7402,
+      "step": 10269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4669172138363964e-05,
+      "loss": 0.7227,
+      "step": 10270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4668154110495962e-05,
+      "loss": 0.7598,
+      "step": 10271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4667136020763982e-05,
+      "loss": 0.7549,
+      "step": 10272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4666117869181515e-05,
+      "loss": 0.748,
+      "step": 10273
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4665099655762052e-05,
+      "loss": 0.7617,
+      "step": 10274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4664081380519086e-05,
+      "loss": 0.7383,
+      "step": 10275
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4663063043466115e-05,
+      "loss": 0.7344,
+      "step": 10276
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4662044644616633e-05,
+      "loss": 0.7373,
+      "step": 10277
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4661026183984133e-05,
+      "loss": 0.7471,
+      "step": 10278
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4660007661582117e-05,
+      "loss": 0.7666,
+      "step": 10279
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4658989077424078e-05,
+      "loss": 0.7236,
+      "step": 10280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657970431523519e-05,
+      "loss": 0.7539,
+      "step": 10281
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4656951723893933e-05,
+      "loss": 0.7598,
+      "step": 10282
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465593295454883e-05,
+      "loss": 0.7568,
+      "step": 10283
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4654914123501702e-05,
+      "loss": 0.7334,
+      "step": 10284
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4653895230766058e-05,
+      "loss": 0.7764,
+      "step": 10285
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4652876276355395e-05,
+      "loss": 0.7549,
+      "step": 10286
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465185726028322e-05,
+      "loss": 0.7861,
+      "step": 10287
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4650838182563031e-05,
+      "loss": 0.7236,
+      "step": 10288
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4649819043208343e-05,
+      "loss": 0.7246,
+      "step": 10289
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4648799842232659e-05,
+      "loss": 0.7666,
+      "step": 10290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4647780579649482e-05,
+      "loss": 0.7559,
+      "step": 10291
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4646761255472322e-05,
+      "loss": 0.751,
+      "step": 10292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4645741869714689e-05,
+      "loss": 0.7646,
+      "step": 10293
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4644722422390085e-05,
+      "loss": 0.7373,
+      "step": 10294
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464370291351203e-05,
+      "loss": 0.75,
+      "step": 10295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.464268334309403e-05,
+      "loss": 0.7559,
+      "step": 10296
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4641663711149599e-05,
+      "loss": 0.7285,
+      "step": 10297
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4640644017692242e-05,
+      "loss": 0.7568,
+      "step": 10298
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4639624262735483e-05,
+      "loss": 0.7529,
+      "step": 10299
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4638604446292827e-05,
+      "loss": 0.7246,
+      "step": 10300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4637584568377796e-05,
+      "loss": 0.7305,
+      "step": 10301
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4636564629003902e-05,
+      "loss": 0.7266,
+      "step": 10302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4635544628184663e-05,
+      "loss": 0.791,
+      "step": 10303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634524565933594e-05,
+      "loss": 0.7598,
+      "step": 10304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4633504442264217e-05,
+      "loss": 0.7422,
+      "step": 10305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4632484257190047e-05,
+      "loss": 0.7383,
+      "step": 10306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4631464010724608e-05,
+      "loss": 0.7773,
+      "step": 10307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4630443702881418e-05,
+      "loss": 0.7412,
+      "step": 10308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4629423333674e-05,
+      "loss": 0.75,
+      "step": 10309
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4628402903115872e-05,
+      "loss": 0.7549,
+      "step": 10310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4627382411220561e-05,
+      "loss": 0.7354,
+      "step": 10311
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4626361858001595e-05,
+      "loss": 0.7422,
+      "step": 10312
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4625341243472488e-05,
+      "loss": 0.7529,
+      "step": 10313
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4624320567646775e-05,
+      "loss": 0.749,
+      "step": 10314
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4623299830537975e-05,
+      "loss": 0.749,
+      "step": 10315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4622279032159625e-05,
+      "loss": 0.7637,
+      "step": 10316
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621258172525243e-05,
+      "loss": 0.7539,
+      "step": 10317
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4620237251648363e-05,
+      "loss": 0.7764,
+      "step": 10318
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4619216269542515e-05,
+      "loss": 0.7295,
+      "step": 10319
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4618195226221226e-05,
+      "loss": 0.7217,
+      "step": 10320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4617174121698026e-05,
+      "loss": 0.7725,
+      "step": 10321
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4616152955986456e-05,
+      "loss": 0.7393,
+      "step": 10322
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4615131729100038e-05,
+      "loss": 0.7549,
+      "step": 10323
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4614110441052314e-05,
+      "loss": 0.7715,
+      "step": 10324
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4613089091856814e-05,
+      "loss": 0.751,
+      "step": 10325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4612067681527073e-05,
+      "loss": 0.7334,
+      "step": 10326
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.461104621007663e-05,
+      "loss": 0.7666,
+      "step": 10327
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.461002467751902e-05,
+      "loss": 0.7412,
+      "step": 10328
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4609003083867778e-05,
+      "loss": 0.7656,
+      "step": 10329
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4607981429136449e-05,
+      "loss": 0.7441,
+      "step": 10330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4606959713338565e-05,
+      "loss": 0.7607,
+      "step": 10331
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4605937936487675e-05,
+      "loss": 0.7188,
+      "step": 10332
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.460491609859731e-05,
+      "loss": 0.751,
+      "step": 10333
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4603894199681019e-05,
+      "loss": 0.748,
+      "step": 10334
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4602872239752344e-05,
+      "loss": 0.7725,
+      "step": 10335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601850218824824e-05,
+      "loss": 0.7441,
+      "step": 10336
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4600828136912004e-05,
+      "loss": 0.7383,
+      "step": 10337
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4599805994027434e-05,
+      "loss": 0.7334,
+      "step": 10338
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4598783790184655e-05,
+      "loss": 0.7344,
+      "step": 10339
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4597761525397216e-05,
+      "loss": 0.7598,
+      "step": 10340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4596739199678663e-05,
+      "loss": 0.7441,
+      "step": 10341
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4595716813042544e-05,
+      "loss": 0.749,
+      "step": 10342
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4594694365502409e-05,
+      "loss": 0.751,
+      "step": 10343
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4593671857071805e-05,
+      "loss": 0.7461,
+      "step": 10344
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459264928776429e-05,
+      "loss": 0.7236,
+      "step": 10345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4591626657593407e-05,
+      "loss": 0.7461,
+      "step": 10346
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4590603966572714e-05,
+      "loss": 0.7578,
+      "step": 10347
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458958121471576e-05,
+      "loss": 0.752,
+      "step": 10348
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4588558402036102e-05,
+      "loss": 0.748,
+      "step": 10349
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4587535528547292e-05,
+      "loss": 0.7568,
+      "step": 10350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4586512594262889e-05,
+      "loss": 0.7549,
+      "step": 10351
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4585489599196444e-05,
+      "loss": 0.7373,
+      "step": 10352
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458446654336152e-05,
+      "loss": 0.7363,
+      "step": 10353
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4583443426771671e-05,
+      "loss": 0.75,
+      "step": 10354
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4582420249440458e-05,
+      "loss": 0.7354,
+      "step": 10355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4581397011381437e-05,
+      "loss": 0.7383,
+      "step": 10356
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4580373712608171e-05,
+      "loss": 0.7588,
+      "step": 10357
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4579350353134222e-05,
+      "loss": 0.752,
+      "step": 10358
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457832693297315e-05,
+      "loss": 0.7314,
+      "step": 10359
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4577303452138517e-05,
+      "loss": 0.7666,
+      "step": 10360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457627991064389e-05,
+      "loss": 0.7334,
+      "step": 10361
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457525630850283e-05,
+      "loss": 0.7461,
+      "step": 10362
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4574232645728906e-05,
+      "loss": 0.7461,
+      "step": 10363
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4573208922335677e-05,
+      "loss": 0.7578,
+      "step": 10364
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4572185138336716e-05,
+      "loss": 0.7285,
+      "step": 10365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4571161293745588e-05,
+      "loss": 0.7471,
+      "step": 10366
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4570137388575864e-05,
+      "loss": 0.7852,
+      "step": 10367
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4569113422841109e-05,
+      "loss": 0.7344,
+      "step": 10368
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4568089396554893e-05,
+      "loss": 0.752,
+      "step": 10369
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.456706530973079e-05,
+      "loss": 0.7715,
+      "step": 10370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4566041162382374e-05,
+      "loss": 0.7422,
+      "step": 10371
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4565016954523208e-05,
+      "loss": 0.7529,
+      "step": 10372
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4563992686166875e-05,
+      "loss": 0.7529,
+      "step": 10373
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4562968357326942e-05,
+      "loss": 0.7607,
+      "step": 10374
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4561943968016987e-05,
+      "loss": 0.7471,
+      "step": 10375
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4560919518250585e-05,
+      "loss": 0.7529,
+      "step": 10376
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4559895008041313e-05,
+      "loss": 0.7637,
+      "step": 10377
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4558870437402747e-05,
+      "loss": 0.7432,
+      "step": 10378
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4557845806348467e-05,
+      "loss": 0.7451,
+      "step": 10379
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.455682111489205e-05,
+      "loss": 0.7314,
+      "step": 10380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4555796363047074e-05,
+      "loss": 0.7676,
+      "step": 10381
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4554771550827122e-05,
+      "loss": 0.7461,
+      "step": 10382
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4553746678245773e-05,
+      "loss": 0.7588,
+      "step": 10383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455272174531661e-05,
+      "loss": 0.7441,
+      "step": 10384
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4551696752053219e-05,
+      "loss": 0.7822,
+      "step": 10385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455067169846918e-05,
+      "loss": 0.7295,
+      "step": 10386
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4549646584578079e-05,
+      "loss": 0.7295,
+      "step": 10387
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4548621410393498e-05,
+      "loss": 0.7256,
+      "step": 10388
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547596175929025e-05,
+      "loss": 0.75,
+      "step": 10389
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4546570881198245e-05,
+      "loss": 0.7646,
+      "step": 10390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4545545526214751e-05,
+      "loss": 0.7432,
+      "step": 10391
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4544520110992127e-05,
+      "loss": 0.7227,
+      "step": 10392
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4543494635543963e-05,
+      "loss": 0.7637,
+      "step": 10393
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4542469099883847e-05,
+      "loss": 0.7441,
+      "step": 10394
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4541443504025372e-05,
+      "loss": 0.751,
+      "step": 10395
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4540417847982131e-05,
+      "loss": 0.7256,
+      "step": 10396
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4539392131767713e-05,
+      "loss": 0.7529,
+      "step": 10397
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4538366355395714e-05,
+      "loss": 0.7334,
+      "step": 10398
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4537340518879725e-05,
+      "loss": 0.7373,
+      "step": 10399
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4536314622233346e-05,
+      "loss": 0.7822,
+      "step": 10400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4535288665470165e-05,
+      "loss": 0.7275,
+      "step": 10401
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453426264860378e-05,
+      "loss": 0.7646,
+      "step": 10402
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4533236571647798e-05,
+      "loss": 0.7314,
+      "step": 10403
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4532210434615804e-05,
+      "loss": 0.7539,
+      "step": 10404
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4531184237521401e-05,
+      "loss": 0.7549,
+      "step": 10405
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4530157980378192e-05,
+      "loss": 0.748,
+      "step": 10406
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4529131663199773e-05,
+      "loss": 0.7705,
+      "step": 10407
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4528105285999749e-05,
+      "loss": 0.7383,
+      "step": 10408
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4527078848791718e-05,
+      "loss": 0.7549,
+      "step": 10409
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4526052351589289e-05,
+      "loss": 0.7441,
+      "step": 10410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4525025794406054e-05,
+      "loss": 0.7617,
+      "step": 10411
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4523999177255628e-05,
+      "loss": 0.7295,
+      "step": 10412
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4522972500151613e-05,
+      "loss": 0.7627,
+      "step": 10413
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4521945763107614e-05,
+      "loss": 0.7275,
+      "step": 10414
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4520918966137236e-05,
+      "loss": 0.7607,
+      "step": 10415
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4519892109254092e-05,
+      "loss": 0.7207,
+      "step": 10416
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4518865192471783e-05,
+      "loss": 0.7393,
+      "step": 10417
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4517838215803923e-05,
+      "loss": 0.7383,
+      "step": 10418
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4516811179264121e-05,
+      "loss": 0.7529,
+      "step": 10419
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515784082865987e-05,
+      "loss": 0.7637,
+      "step": 10420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4514756926623131e-05,
+      "loss": 0.7559,
+      "step": 10421
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4513729710549169e-05,
+      "loss": 0.7383,
+      "step": 10422
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4512702434657711e-05,
+      "loss": 0.7432,
+      "step": 10423
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4511675098962373e-05,
+      "loss": 0.7451,
+      "step": 10424
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4510647703476768e-05,
+      "loss": 0.748,
+      "step": 10425
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509620248214512e-05,
+      "loss": 0.7568,
+      "step": 10426
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4508592733189217e-05,
+      "loss": 0.7588,
+      "step": 10427
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4507565158414506e-05,
+      "loss": 0.7412,
+      "step": 10428
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4506537523903997e-05,
+      "loss": 0.7578,
+      "step": 10429
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4505509829671304e-05,
+      "loss": 0.7578,
+      "step": 10430
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4504482075730048e-05,
+      "loss": 0.7939,
+      "step": 10431
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4503454262093853e-05,
+      "loss": 0.7422,
+      "step": 10432
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4502426388776333e-05,
+      "loss": 0.7461,
+      "step": 10433
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4501398455791113e-05,
+      "loss": 0.7754,
+      "step": 10434
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.450037046315182e-05,
+      "loss": 0.7539,
+      "step": 10435
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4499342410872071e-05,
+      "loss": 0.7344,
+      "step": 10436
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4498314298965493e-05,
+      "loss": 0.7275,
+      "step": 10437
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4497286127445706e-05,
+      "loss": 0.7695,
+      "step": 10438
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4496257896326346e-05,
+      "loss": 0.7705,
+      "step": 10439
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4495229605621031e-05,
+      "loss": 0.7451,
+      "step": 10440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.449420125534339e-05,
+      "loss": 0.7559,
+      "step": 10441
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4493172845507053e-05,
+      "loss": 0.7529,
+      "step": 10442
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4492144376125646e-05,
+      "loss": 0.751,
+      "step": 10443
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.44911158472128e-05,
+      "loss": 0.7539,
+      "step": 10444
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4490087258782148e-05,
+      "loss": 0.7451,
+      "step": 10445
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4489058610847315e-05,
+      "loss": 0.7578,
+      "step": 10446
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4488029903421938e-05,
+      "loss": 0.7363,
+      "step": 10447
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4487001136519649e-05,
+      "loss": 0.7363,
+      "step": 10448
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4485972310154083e-05,
+      "loss": 0.749,
+      "step": 10449
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4484943424338868e-05,
+      "loss": 0.751,
+      "step": 10450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4483914479087648e-05,
+      "loss": 0.75,
+      "step": 10451
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4482885474414054e-05,
+      "loss": 0.7783,
+      "step": 10452
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4481856410331722e-05,
+      "loss": 0.7344,
+      "step": 10453
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4480827286854292e-05,
+      "loss": 0.7451,
+      "step": 10454
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4479798103995401e-05,
+      "loss": 0.7383,
+      "step": 10455
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4478768861768686e-05,
+      "loss": 0.7744,
+      "step": 10456
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4477739560187791e-05,
+      "loss": 0.7588,
+      "step": 10457
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4476710199266353e-05,
+      "loss": 0.7471,
+      "step": 10458
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4475680779018017e-05,
+      "loss": 0.7441,
+      "step": 10459
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4474651299456421e-05,
+      "loss": 0.7461,
+      "step": 10460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4473621760595214e-05,
+      "loss": 0.7529,
+      "step": 10461
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4472592162448033e-05,
+      "loss": 0.7334,
+      "step": 10462
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.447156250502853e-05,
+      "loss": 0.7559,
+      "step": 10463
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4470532788350343e-05,
+      "loss": 0.7334,
+      "step": 10464
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446950301242712e-05,
+      "loss": 0.7422,
+      "step": 10465
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4468473177272508e-05,
+      "loss": 0.7363,
+      "step": 10466
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446744328290016e-05,
+      "loss": 0.7354,
+      "step": 10467
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4466413329323721e-05,
+      "loss": 0.7324,
+      "step": 10468
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4465383316556838e-05,
+      "loss": 0.7725,
+      "step": 10469
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4464353244613162e-05,
+      "loss": 0.7432,
+      "step": 10470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4463323113506347e-05,
+      "loss": 0.7412,
+      "step": 10471
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4462292923250041e-05,
+      "loss": 0.749,
+      "step": 10472
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4461262673857896e-05,
+      "loss": 0.7617,
+      "step": 10473
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4460232365343569e-05,
+      "loss": 0.7451,
+      "step": 10474
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4459201997720713e-05,
+      "loss": 0.7705,
+      "step": 10475
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4458171571002981e-05,
+      "loss": 0.7432,
+      "step": 10476
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4457141085204028e-05,
+      "loss": 0.7354,
+      "step": 10477
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4456110540337516e-05,
+      "loss": 0.7324,
+      "step": 10478
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4455079936417094e-05,
+      "loss": 0.749,
+      "step": 10479
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4454049273456425e-05,
+      "loss": 0.7578,
+      "step": 10480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4453018551469167e-05,
+      "loss": 0.7461,
+      "step": 10481
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4451987770468978e-05,
+      "loss": 0.7178,
+      "step": 10482
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4450956930469517e-05,
+      "loss": 0.7422,
+      "step": 10483
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4449926031484452e-05,
+      "loss": 0.7666,
+      "step": 10484
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4448895073527438e-05,
+      "loss": 0.7324,
+      "step": 10485
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4447864056612139e-05,
+      "loss": 0.7393,
+      "step": 10486
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4446832980752219e-05,
+      "loss": 0.7744,
+      "step": 10487
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4445801845961345e-05,
+      "loss": 0.7461,
+      "step": 10488
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4444770652253175e-05,
+      "loss": 0.7354,
+      "step": 10489
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4443739399641383e-05,
+      "loss": 0.7461,
+      "step": 10490
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444270808813963e-05,
+      "loss": 0.75,
+      "step": 10491
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4441676717761586e-05,
+      "loss": 0.752,
+      "step": 10492
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4440645288520913e-05,
+      "loss": 0.7334,
+      "step": 10493
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4439613800431292e-05,
+      "loss": 0.7471,
+      "step": 10494
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4438582253506378e-05,
+      "loss": 0.7422,
+      "step": 10495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4437550647759854e-05,
+      "loss": 0.7578,
+      "step": 10496
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4436518983205385e-05,
+      "loss": 0.7441,
+      "step": 10497
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4435487259856644e-05,
+      "loss": 0.7422,
+      "step": 10498
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4434455477727301e-05,
+      "loss": 0.7598,
+      "step": 10499
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4433423636831036e-05,
+      "loss": 0.751,
+      "step": 10500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4432391737181517e-05,
+      "loss": 0.7568,
+      "step": 10501
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4431359778792422e-05,
+      "loss": 0.7471,
+      "step": 10502
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4430327761677427e-05,
+      "loss": 0.7334,
+      "step": 10503
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4429295685850208e-05,
+      "loss": 0.7773,
+      "step": 10504
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4428263551324442e-05,
+      "loss": 0.7578,
+      "step": 10505
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4427231358113808e-05,
+      "loss": 0.7344,
+      "step": 10506
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4426199106231984e-05,
+      "loss": 0.752,
+      "step": 10507
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4425166795692653e-05,
+      "loss": 0.751,
+      "step": 10508
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4424134426509491e-05,
+      "loss": 0.7266,
+      "step": 10509
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4423101998696182e-05,
+      "loss": 0.7881,
+      "step": 10510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4422069512266408e-05,
+      "loss": 0.7617,
+      "step": 10511
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4421036967233851e-05,
+      "loss": 0.7725,
+      "step": 10512
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4420004363612195e-05,
+      "loss": 0.752,
+      "step": 10513
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4418971701415123e-05,
+      "loss": 0.7588,
+      "step": 10514
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4417938980656323e-05,
+      "loss": 0.7432,
+      "step": 10515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4416906201349476e-05,
+      "loss": 0.7451,
+      "step": 10516
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4415873363508277e-05,
+      "loss": 0.7529,
+      "step": 10517
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4414840467146406e-05,
+      "loss": 0.7373,
+      "step": 10518
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4413807512277556e-05,
+      "loss": 0.7373,
+      "step": 10519
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4412774498915414e-05,
+      "loss": 0.7324,
+      "step": 10520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4411741427073667e-05,
+      "loss": 0.7363,
+      "step": 10521
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4410708296766009e-05,
+      "loss": 0.7461,
+      "step": 10522
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4409675108006135e-05,
+      "loss": 0.7295,
+      "step": 10523
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440864186080773e-05,
+      "loss": 0.7461,
+      "step": 10524
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4407608555184492e-05,
+      "loss": 0.7695,
+      "step": 10525
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4406575191150112e-05,
+      "loss": 0.7539,
+      "step": 10526
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4405541768718285e-05,
+      "loss": 0.7354,
+      "step": 10527
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4404508287902705e-05,
+      "loss": 0.7549,
+      "step": 10528
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4403474748717072e-05,
+      "loss": 0.7334,
+      "step": 10529
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440244115117508e-05,
+      "loss": 0.7471,
+      "step": 10530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4401407495290429e-05,
+      "loss": 0.7578,
+      "step": 10531
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440037378107681e-05,
+      "loss": 0.7344,
+      "step": 10532
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4399340008547934e-05,
+      "loss": 0.7383,
+      "step": 10533
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4398306177717489e-05,
+      "loss": 0.7373,
+      "step": 10534
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4397272288599184e-05,
+      "loss": 0.7451,
+      "step": 10535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4396238341206718e-05,
+      "loss": 0.7529,
+      "step": 10536
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4395204335553791e-05,
+      "loss": 0.748,
+      "step": 10537
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4394170271654108e-05,
+      "loss": 0.7646,
+      "step": 10538
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4393136149521376e-05,
+      "loss": 0.7314,
+      "step": 10539
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4392101969169293e-05,
+      "loss": 0.7334,
+      "step": 10540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4391067730611567e-05,
+      "loss": 0.7598,
+      "step": 10541
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4390033433861907e-05,
+      "loss": 0.7529,
+      "step": 10542
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4388999078934016e-05,
+      "loss": 0.7383,
+      "step": 10543
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.43879646658416e-05,
+      "loss": 0.749,
+      "step": 10544
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4386930194598376e-05,
+      "loss": 0.7266,
+      "step": 10545
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4385895665218044e-05,
+      "loss": 0.7549,
+      "step": 10546
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4384861077714321e-05,
+      "loss": 0.7188,
+      "step": 10547
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4383826432100909e-05,
+      "loss": 0.7588,
+      "step": 10548
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438279172839153e-05,
+      "loss": 0.752,
+      "step": 10549
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4381756966599887e-05,
+      "loss": 0.7646,
+      "step": 10550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4380722146739698e-05,
+      "loss": 0.7646,
+      "step": 10551
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4379687268824676e-05,
+      "loss": 0.75,
+      "step": 10552
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4378652332868534e-05,
+      "loss": 0.7432,
+      "step": 10553
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4377617338884993e-05,
+      "loss": 0.7783,
+      "step": 10554
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4376582286887758e-05,
+      "loss": 0.7725,
+      "step": 10555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.437554717689056e-05,
+      "loss": 0.7432,
+      "step": 10556
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4374512008907104e-05,
+      "loss": 0.7783,
+      "step": 10557
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4373476782951115e-05,
+      "loss": 0.7441,
+      "step": 10558
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4372441499036312e-05,
+      "loss": 0.7656,
+      "step": 10559
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4371406157176414e-05,
+      "loss": 0.7734,
+      "step": 10560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4370370757385137e-05,
+      "loss": 0.7168,
+      "step": 10561
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4369335299676212e-05,
+      "loss": 0.7217,
+      "step": 10562
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4368299784063352e-05,
+      "loss": 0.7432,
+      "step": 10563
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.436726421056029e-05,
+      "loss": 0.7285,
+      "step": 10564
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.436622857918074e-05,
+      "loss": 0.7383,
+      "step": 10565
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4365192889938431e-05,
+      "loss": 0.748,
+      "step": 10566
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4364157142847087e-05,
+      "loss": 0.7461,
+      "step": 10567
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4363121337920434e-05,
+      "loss": 0.7402,
+      "step": 10568
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4362085475172204e-05,
+      "loss": 0.7666,
+      "step": 10569
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4361049554616116e-05,
+      "loss": 0.7461,
+      "step": 10570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4360013576265903e-05,
+      "loss": 0.7637,
+      "step": 10571
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4358977540135297e-05,
+      "loss": 0.7529,
+      "step": 10572
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4357941446238021e-05,
+      "loss": 0.7383,
+      "step": 10573
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4356905294587813e-05,
+      "loss": 0.7549,
+      "step": 10574
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43558690851984e-05,
+      "loss": 0.7344,
+      "step": 10575
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4354832818083515e-05,
+      "loss": 0.7471,
+      "step": 10576
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4353796493256888e-05,
+      "loss": 0.7666,
+      "step": 10577
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.435276011073226e-05,
+      "loss": 0.7461,
+      "step": 10578
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.435172367052336e-05,
+      "loss": 0.7383,
+      "step": 10579
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4350687172643923e-05,
+      "loss": 0.748,
+      "step": 10580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4349650617107686e-05,
+      "loss": 0.7412,
+      "step": 10581
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.434861400392839e-05,
+      "loss": 0.7324,
+      "step": 10582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4347577333119764e-05,
+      "loss": 0.7295,
+      "step": 10583
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4346540604695554e-05,
+      "loss": 0.752,
+      "step": 10584
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345503818669497e-05,
+      "loss": 0.749,
+      "step": 10585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4344466975055331e-05,
+      "loss": 0.7588,
+      "step": 10586
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43434300738668e-05,
+      "loss": 0.7422,
+      "step": 10587
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4342393115117642e-05,
+      "loss": 0.7471,
+      "step": 10588
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4341356098821595e-05,
+      "loss": 0.7461,
+      "step": 10589
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4340319024992413e-05,
+      "loss": 0.7227,
+      "step": 10590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4339281893643833e-05,
+      "loss": 0.7295,
+      "step": 10591
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.43382447047896e-05,
+      "loss": 0.749,
+      "step": 10592
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4337207458443458e-05,
+      "loss": 0.7373,
+      "step": 10593
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4336170154619157e-05,
+      "loss": 0.7588,
+      "step": 10594
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433513279333044e-05,
+      "loss": 0.7451,
+      "step": 10595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4334095374591053e-05,
+      "loss": 0.7324,
+      "step": 10596
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4333057898414748e-05,
+      "loss": 0.7236,
+      "step": 10597
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4332020364815273e-05,
+      "loss": 0.7412,
+      "step": 10598
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433098277380638e-05,
+      "loss": 0.7285,
+      "step": 10599
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4329945125401811e-05,
+      "loss": 0.7451,
+      "step": 10600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432890741961533e-05,
+      "loss": 0.7568,
+      "step": 10601
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4327869656460677e-05,
+      "loss": 0.7422,
+      "step": 10602
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4326831835951613e-05,
+      "loss": 0.7559,
+      "step": 10603
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4325793958101891e-05,
+      "loss": 0.7676,
+      "step": 10604
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4324756022925259e-05,
+      "loss": 0.7412,
+      "step": 10605
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4323718030435477e-05,
+      "loss": 0.748,
+      "step": 10606
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4322679980646301e-05,
+      "loss": 0.7441,
+      "step": 10607
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4321641873571483e-05,
+      "loss": 0.7578,
+      "step": 10608
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4320603709224788e-05,
+      "loss": 0.7354,
+      "step": 10609
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4319565487619966e-05,
+      "loss": 0.748,
+      "step": 10610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4318527208770785e-05,
+      "loss": 0.7236,
+      "step": 10611
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4317488872690995e-05,
+      "loss": 0.7227,
+      "step": 10612
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4316450479394362e-05,
+      "loss": 0.7529,
+      "step": 10613
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4315412028894647e-05,
+      "loss": 0.7549,
+      "step": 10614
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431437352120561e-05,
+      "loss": 0.7363,
+      "step": 10615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4313334956341011e-05,
+      "loss": 0.7539,
+      "step": 10616
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4312296334314623e-05,
+      "loss": 0.7354,
+      "step": 10617
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.43112576551402e-05,
+      "loss": 0.75,
+      "step": 10618
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4310218918831514e-05,
+      "loss": 0.748,
+      "step": 10619
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4309180125402327e-05,
+      "loss": 0.7217,
+      "step": 10620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4308141274866404e-05,
+      "loss": 0.7139,
+      "step": 10621
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307102367237515e-05,
+      "loss": 0.7412,
+      "step": 10622
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4306063402529429e-05,
+      "loss": 0.751,
+      "step": 10623
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4305024380755913e-05,
+      "loss": 0.7383,
+      "step": 10624
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4303985301930736e-05,
+      "loss": 0.7354,
+      "step": 10625
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4302946166067666e-05,
+      "loss": 0.7559,
+      "step": 10626
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4301906973180484e-05,
+      "loss": 0.752,
+      "step": 10627
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4300867723282947e-05,
+      "loss": 0.7383,
+      "step": 10628
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4299828416388839e-05,
+      "loss": 0.7197,
+      "step": 10629
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4298789052511927e-05,
+      "loss": 0.7666,
+      "step": 10630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4297749631665989e-05,
+      "loss": 0.7588,
+      "step": 10631
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.42967101538648e-05,
+      "loss": 0.7559,
+      "step": 10632
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4295670619122128e-05,
+      "loss": 0.7334,
+      "step": 10633
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4294631027451757e-05,
+      "loss": 0.7324,
+      "step": 10634
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4293591378867465e-05,
+      "loss": 0.7617,
+      "step": 10635
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4292551673383025e-05,
+      "loss": 0.7676,
+      "step": 10636
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4291511911012214e-05,
+      "loss": 0.7393,
+      "step": 10637
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290472091768817e-05,
+      "loss": 0.7363,
+      "step": 10638
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428943221566661e-05,
+      "loss": 0.7676,
+      "step": 10639
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428839228271938e-05,
+      "loss": 0.7627,
+      "step": 10640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4287352292940897e-05,
+      "loss": 0.7627,
+      "step": 10641
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4286312246344954e-05,
+      "loss": 0.7441,
+      "step": 10642
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4285272142945328e-05,
+      "loss": 0.7529,
+      "step": 10643
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4284231982755807e-05,
+      "loss": 0.7432,
+      "step": 10644
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4283191765790171e-05,
+      "loss": 0.75,
+      "step": 10645
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4282151492062213e-05,
+      "loss": 0.7393,
+      "step": 10646
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4281111161585709e-05,
+      "loss": 0.751,
+      "step": 10647
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4280070774374453e-05,
+      "loss": 0.7539,
+      "step": 10648
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4279030330442227e-05,
+      "loss": 0.752,
+      "step": 10649
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4277989829802828e-05,
+      "loss": 0.7598,
+      "step": 10650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276949272470036e-05,
+      "loss": 0.7354,
+      "step": 10651
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4275908658457645e-05,
+      "loss": 0.7275,
+      "step": 10652
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4274867987779446e-05,
+      "loss": 0.7344,
+      "step": 10653
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4273827260449229e-05,
+      "loss": 0.7549,
+      "step": 10654
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4272786476480785e-05,
+      "loss": 0.7451,
+      "step": 10655
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4271745635887912e-05,
+      "loss": 0.7197,
+      "step": 10656
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4270704738684395e-05,
+      "loss": 0.7412,
+      "step": 10657
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4269663784884036e-05,
+      "loss": 0.7285,
+      "step": 10658
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.426862277450063e-05,
+      "loss": 0.751,
+      "step": 10659
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4267581707547966e-05,
+      "loss": 0.7422,
+      "step": 10660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4266540584039845e-05,
+      "loss": 0.7676,
+      "step": 10661
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4265499403990063e-05,
+      "loss": 0.7656,
+      "step": 10662
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4264458167412422e-05,
+      "loss": 0.7324,
+      "step": 10663
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4263416874320717e-05,
+      "loss": 0.7578,
+      "step": 10664
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4262375524728746e-05,
+      "loss": 0.751,
+      "step": 10665
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4261334118650318e-05,
+      "loss": 0.7588,
+      "step": 10666
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4260292656099223e-05,
+      "loss": 0.75,
+      "step": 10667
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4259251137089267e-05,
+      "loss": 0.75,
+      "step": 10668
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4258209561634257e-05,
+      "loss": 0.7568,
+      "step": 10669
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425716792974799e-05,
+      "loss": 0.7412,
+      "step": 10670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4256126241444275e-05,
+      "loss": 0.7676,
+      "step": 10671
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4255084496736913e-05,
+      "loss": 0.7402,
+      "step": 10672
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4254042695639709e-05,
+      "loss": 0.7666,
+      "step": 10673
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4253000838166473e-05,
+      "loss": 0.7715,
+      "step": 10674
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4251958924331011e-05,
+      "loss": 0.7412,
+      "step": 10675
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425091695414713e-05,
+      "loss": 0.7559,
+      "step": 10676
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4249874927628638e-05,
+      "loss": 0.7344,
+      "step": 10677
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4248832844789344e-05,
+      "loss": 0.7422,
+      "step": 10678
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424779070564306e-05,
+      "loss": 0.7588,
+      "step": 10679
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4246748510203596e-05,
+      "loss": 0.7549,
+      "step": 10680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4245706258484763e-05,
+      "loss": 0.7334,
+      "step": 10681
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4244663950500376e-05,
+      "loss": 0.7324,
+      "step": 10682
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4243621586264245e-05,
+      "loss": 0.7656,
+      "step": 10683
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424257916579018e-05,
+      "loss": 0.7529,
+      "step": 10684
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4241536689092007e-05,
+      "loss": 0.7441,
+      "step": 10685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4240494156183528e-05,
+      "loss": 0.7324,
+      "step": 10686
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4239451567078569e-05,
+      "loss": 0.748,
+      "step": 10687
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4238408921790941e-05,
+      "loss": 0.7412,
+      "step": 10688
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4237366220334465e-05,
+      "loss": 0.7295,
+      "step": 10689
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4236323462722955e-05,
+      "loss": 0.7461,
+      "step": 10690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4235280648970235e-05,
+      "loss": 0.7549,
+      "step": 10691
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4234237779090122e-05,
+      "loss": 0.75,
+      "step": 10692
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4233194853096437e-05,
+      "loss": 0.7725,
+      "step": 10693
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4232151871002999e-05,
+      "loss": 0.7607,
+      "step": 10694
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231108832823636e-05,
+      "loss": 0.7432,
+      "step": 10695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4230065738572163e-05,
+      "loss": 0.749,
+      "step": 10696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422902258826241e-05,
+      "loss": 0.7461,
+      "step": 10697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4227979381908197e-05,
+      "loss": 0.7402,
+      "step": 10698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4226936119523352e-05,
+      "loss": 0.7646,
+      "step": 10699
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4225892801121695e-05,
+      "loss": 0.7617,
+      "step": 10700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422484942671706e-05,
+      "loss": 0.7432,
+      "step": 10701
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422380599632327e-05,
+      "loss": 0.749,
+      "step": 10702
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4222762509954151e-05,
+      "loss": 0.7549,
+      "step": 10703
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4221718967623534e-05,
+      "loss": 0.7617,
+      "step": 10704
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220675369345253e-05,
+      "loss": 0.75,
+      "step": 10705
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421963171513313e-05,
+      "loss": 0.7451,
+      "step": 10706
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4218588005001e-05,
+      "loss": 0.7686,
+      "step": 10707
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4217544238962695e-05,
+      "loss": 0.7266,
+      "step": 10708
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4216500417032047e-05,
+      "loss": 0.7402,
+      "step": 10709
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4215456539222885e-05,
+      "loss": 0.7646,
+      "step": 10710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421441260554905e-05,
+      "loss": 0.7363,
+      "step": 10711
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421336861602437e-05,
+      "loss": 0.7441,
+      "step": 10712
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4212324570662685e-05,
+      "loss": 0.7539,
+      "step": 10713
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421128046947783e-05,
+      "loss": 0.7676,
+      "step": 10714
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4210236312483641e-05,
+      "loss": 0.7393,
+      "step": 10715
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4209192099693954e-05,
+      "loss": 0.7227,
+      "step": 10716
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4208147831122608e-05,
+      "loss": 0.7422,
+      "step": 10717
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4207103506783444e-05,
+      "loss": 0.7441,
+      "step": 10718
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4206059126690298e-05,
+      "loss": 0.7324,
+      "step": 10719
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4205014690857018e-05,
+      "loss": 0.7539,
+      "step": 10720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4203970199297438e-05,
+      "loss": 0.7471,
+      "step": 10721
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4202925652025403e-05,
+      "loss": 0.7451,
+      "step": 10722
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4201881049054753e-05,
+      "loss": 0.7393,
+      "step": 10723
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4200836390399335e-05,
+      "loss": 0.7354,
+      "step": 10724
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4199791676072989e-05,
+      "loss": 0.749,
+      "step": 10725
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4198746906089566e-05,
+      "loss": 0.748,
+      "step": 10726
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4197702080462907e-05,
+      "loss": 0.7422,
+      "step": 10727
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419665719920686e-05,
+      "loss": 0.749,
+      "step": 10728
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4195612262335268e-05,
+      "loss": 0.7422,
+      "step": 10729
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4194567269861986e-05,
+      "loss": 0.7363,
+      "step": 10730
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4193522221800857e-05,
+      "loss": 0.7451,
+      "step": 10731
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4192477118165735e-05,
+      "loss": 0.7607,
+      "step": 10732
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4191431958970463e-05,
+      "loss": 0.7666,
+      "step": 10733
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4190386744228903e-05,
+      "loss": 0.7412,
+      "step": 10734
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4189341473954894e-05,
+      "loss": 0.7207,
+      "step": 10735
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4188296148162297e-05,
+      "loss": 0.7607,
+      "step": 10736
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4187250766864962e-05,
+      "loss": 0.748,
+      "step": 10737
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4186205330076744e-05,
+      "loss": 0.7559,
+      "step": 10738
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4185159837811491e-05,
+      "loss": 0.7705,
+      "step": 10739
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.418411429008307e-05,
+      "loss": 0.7686,
+      "step": 10740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4183068686905327e-05,
+      "loss": 0.7256,
+      "step": 10741
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182023028292122e-05,
+      "loss": 0.7441,
+      "step": 10742
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4180977314257312e-05,
+      "loss": 0.75,
+      "step": 10743
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.417993154481476e-05,
+      "loss": 0.7354,
+      "step": 10744
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4178885719978317e-05,
+      "loss": 0.7412,
+      "step": 10745
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4177839839761847e-05,
+      "loss": 0.75,
+      "step": 10746
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4176793904179211e-05,
+      "loss": 0.7354,
+      "step": 10747
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4175747913244268e-05,
+      "loss": 0.748,
+      "step": 10748
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4174701866970879e-05,
+      "loss": 0.7432,
+      "step": 10749
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4173655765372913e-05,
+      "loss": 0.749,
+      "step": 10750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4172609608464222e-05,
+      "loss": 0.7266,
+      "step": 10751
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4171563396258683e-05,
+      "loss": 0.7441,
+      "step": 10752
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.417051712877015e-05,
+      "loss": 0.7422,
+      "step": 10753
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4169470806012495e-05,
+      "loss": 0.7539,
+      "step": 10754
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416842442799958e-05,
+      "loss": 0.7363,
+      "step": 10755
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4167377994745274e-05,
+      "loss": 0.7529,
+      "step": 10756
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4166331506263446e-05,
+      "loss": 0.7256,
+      "step": 10757
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4165284962567964e-05,
+      "loss": 0.7314,
+      "step": 10758
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4164238363672696e-05,
+      "loss": 0.7578,
+      "step": 10759
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416319170959151e-05,
+      "loss": 0.7305,
+      "step": 10760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416214500033828e-05,
+      "loss": 0.7256,
+      "step": 10761
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161098235926875e-05,
+      "loss": 0.7305,
+      "step": 10762
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416005141637117e-05,
+      "loss": 0.7402,
+      "step": 10763
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4159004541685036e-05,
+      "loss": 0.7266,
+      "step": 10764
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4157957611882347e-05,
+      "loss": 0.7002,
+      "step": 10765
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4156910626976976e-05,
+      "loss": 0.7451,
+      "step": 10766
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4155863586982799e-05,
+      "loss": 0.7402,
+      "step": 10767
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.415481649191369e-05,
+      "loss": 0.7236,
+      "step": 10768
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4153769341783527e-05,
+      "loss": 0.7451,
+      "step": 10769
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.415272213660619e-05,
+      "loss": 0.7051,
+      "step": 10770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4151674876395552e-05,
+      "loss": 0.7148,
+      "step": 10771
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4150627561165492e-05,
+      "loss": 0.7715,
+      "step": 10772
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4149580190929898e-05,
+      "loss": 0.7461,
+      "step": 10773
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4148532765702635e-05,
+      "loss": 0.7402,
+      "step": 10774
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4147485285497597e-05,
+      "loss": 0.7471,
+      "step": 10775
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414643775032866e-05,
+      "loss": 0.7627,
+      "step": 10776
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4145390160209707e-05,
+      "loss": 0.7617,
+      "step": 10777
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414434251515462e-05,
+      "loss": 0.7402,
+      "step": 10778
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4143294815177284e-05,
+      "loss": 0.7188,
+      "step": 10779
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4142247060291583e-05,
+      "loss": 0.7266,
+      "step": 10780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4141199250511404e-05,
+      "loss": 0.7695,
+      "step": 10781
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4140151385850626e-05,
+      "loss": 0.7656,
+      "step": 10782
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.413910346632315e-05,
+      "loss": 0.748,
+      "step": 10783
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4138055491942847e-05,
+      "loss": 0.7822,
+      "step": 10784
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4137007462723617e-05,
+      "loss": 0.7354,
+      "step": 10785
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4135959378679343e-05,
+      "loss": 0.7568,
+      "step": 10786
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4134911239823915e-05,
+      "loss": 0.7363,
+      "step": 10787
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4133863046171223e-05,
+      "loss": 0.7578,
+      "step": 10788
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4132814797735163e-05,
+      "loss": 0.7002,
+      "step": 10789
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4131766494529621e-05,
+      "loss": 0.7549,
+      "step": 10790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.413071813656849e-05,
+      "loss": 0.7529,
+      "step": 10791
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412966972386567e-05,
+      "loss": 0.7344,
+      "step": 10792
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4128621256435047e-05,
+      "loss": 0.7666,
+      "step": 10793
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4127572734290514e-05,
+      "loss": 0.7383,
+      "step": 10794
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4126524157445976e-05,
+      "loss": 0.7461,
+      "step": 10795
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412547552591532e-05,
+      "loss": 0.7266,
+      "step": 10796
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4124426839712449e-05,
+      "loss": 0.7314,
+      "step": 10797
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4123378098851256e-05,
+      "loss": 0.71,
+      "step": 10798
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4122329303345646e-05,
+      "loss": 0.7344,
+      "step": 10799
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412128045320951e-05,
+      "loss": 0.7598,
+      "step": 10800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412023154845675e-05,
+      "loss": 0.7539,
+      "step": 10801
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4119182589101268e-05,
+      "loss": 0.749,
+      "step": 10802
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4118133575156967e-05,
+      "loss": 0.7266,
+      "step": 10803
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4117084506637747e-05,
+      "loss": 0.7344,
+      "step": 10804
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4116035383557507e-05,
+      "loss": 0.7246,
+      "step": 10805
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4114986205930157e-05,
+      "loss": 0.749,
+      "step": 10806
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4113936973769594e-05,
+      "loss": 0.7529,
+      "step": 10807
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411288768708973e-05,
+      "loss": 0.7471,
+      "step": 10808
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4111838345904462e-05,
+      "loss": 0.7139,
+      "step": 10809
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4110788950227706e-05,
+      "loss": 0.7754,
+      "step": 10810
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4109739500073361e-05,
+      "loss": 0.7178,
+      "step": 10811
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410868999545534e-05,
+      "loss": 0.7285,
+      "step": 10812
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4107640436387549e-05,
+      "loss": 0.7354,
+      "step": 10813
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4106590822883895e-05,
+      "loss": 0.7178,
+      "step": 10814
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4105541154958294e-05,
+      "loss": 0.7373,
+      "step": 10815
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410449143262465e-05,
+      "loss": 0.7373,
+      "step": 10816
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4103441655896874e-05,
+      "loss": 0.7637,
+      "step": 10817
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4102391824788887e-05,
+      "loss": 0.7461,
+      "step": 10818
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4101341939314592e-05,
+      "loss": 0.7402,
+      "step": 10819
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4100291999487907e-05,
+      "loss": 0.7441,
+      "step": 10820
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4099242005322747e-05,
+      "loss": 0.7012,
+      "step": 10821
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4098191956833022e-05,
+      "loss": 0.7627,
+      "step": 10822
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4097141854032654e-05,
+      "loss": 0.7549,
+      "step": 10823
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4096091696935554e-05,
+      "loss": 0.7461,
+      "step": 10824
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4095041485555642e-05,
+      "loss": 0.749,
+      "step": 10825
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4093991219906836e-05,
+      "loss": 0.7686,
+      "step": 10826
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.409294090000305e-05,
+      "loss": 0.7656,
+      "step": 10827
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4091890525858212e-05,
+      "loss": 0.7461,
+      "step": 10828
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4090840097486232e-05,
+      "loss": 0.7354,
+      "step": 10829
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4089789614901036e-05,
+      "loss": 0.7363,
+      "step": 10830
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4088739078116544e-05,
+      "loss": 0.7197,
+      "step": 10831
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.408768848714668e-05,
+      "loss": 0.7646,
+      "step": 10832
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4086637842005364e-05,
+      "loss": 0.749,
+      "step": 10833
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4085587142706524e-05,
+      "loss": 0.7324,
+      "step": 10834
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4084536389264076e-05,
+      "loss": 0.75,
+      "step": 10835
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4083485581691956e-05,
+      "loss": 0.749,
+      "step": 10836
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4082434720004079e-05,
+      "loss": 0.75,
+      "step": 10837
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4081383804214379e-05,
+      "loss": 0.7363,
+      "step": 10838
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4080332834336776e-05,
+      "loss": 0.7451,
+      "step": 10839
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4079281810385208e-05,
+      "loss": 0.7705,
+      "step": 10840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4078230732373594e-05,
+      "loss": 0.7451,
+      "step": 10841
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4077179600315867e-05,
+      "loss": 0.751,
+      "step": 10842
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4076128414225958e-05,
+      "loss": 0.751,
+      "step": 10843
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4075077174117797e-05,
+      "loss": 0.7275,
+      "step": 10844
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4074025880005314e-05,
+      "loss": 0.7441,
+      "step": 10845
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407297453190244e-05,
+      "loss": 0.7402,
+      "step": 10846
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4071923129823111e-05,
+      "loss": 0.7422,
+      "step": 10847
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4070871673781259e-05,
+      "loss": 0.7451,
+      "step": 10848
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4069820163790819e-05,
+      "loss": 0.7559,
+      "step": 10849
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4068768599865724e-05,
+      "loss": 0.7363,
+      "step": 10850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4067716982019914e-05,
+      "loss": 0.7676,
+      "step": 10851
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406666531026732e-05,
+      "loss": 0.752,
+      "step": 10852
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4065613584621885e-05,
+      "loss": 0.7285,
+      "step": 10853
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064561805097541e-05,
+      "loss": 0.7607,
+      "step": 10854
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4063509971708231e-05,
+      "loss": 0.7607,
+      "step": 10855
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4062458084467889e-05,
+      "loss": 0.7471,
+      "step": 10856
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406140614339046e-05,
+      "loss": 0.751,
+      "step": 10857
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4060354148489883e-05,
+      "loss": 0.7246,
+      "step": 10858
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4059302099780102e-05,
+      "loss": 0.7471,
+      "step": 10859
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058249997275053e-05,
+      "loss": 0.7646,
+      "step": 10860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4057197840988686e-05,
+      "loss": 0.7148,
+      "step": 10861
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4056145630934936e-05,
+      "loss": 0.7334,
+      "step": 10862
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4055093367127755e-05,
+      "loss": 0.7168,
+      "step": 10863
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4054041049581084e-05,
+      "loss": 0.7373,
+      "step": 10864
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4052988678308873e-05,
+      "loss": 0.7822,
+      "step": 10865
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4051936253325061e-05,
+      "loss": 0.7305,
+      "step": 10866
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4050883774643602e-05,
+      "loss": 0.7354,
+      "step": 10867
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.404983124227844e-05,
+      "loss": 0.7559,
+      "step": 10868
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4048778656243524e-05,
+      "loss": 0.7568,
+      "step": 10869
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4047726016552805e-05,
+      "loss": 0.7461,
+      "step": 10870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4046673323220232e-05,
+      "loss": 0.7549,
+      "step": 10871
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4045620576259753e-05,
+      "loss": 0.7441,
+      "step": 10872
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4044567775685325e-05,
+      "loss": 0.7051,
+      "step": 10873
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4043514921510894e-05,
+      "loss": 0.7568,
+      "step": 10874
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4042462013750417e-05,
+      "loss": 0.7393,
+      "step": 10875
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4041409052417847e-05,
+      "loss": 0.7236,
+      "step": 10876
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4040356037527136e-05,
+      "loss": 0.7441,
+      "step": 10877
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4039302969092238e-05,
+      "loss": 0.7744,
+      "step": 10878
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4038249847127115e-05,
+      "loss": 0.7275,
+      "step": 10879
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4037196671645718e-05,
+      "loss": 0.748,
+      "step": 10880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4036143442662005e-05,
+      "loss": 0.7266,
+      "step": 10881
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4035090160189934e-05,
+      "loss": 0.7402,
+      "step": 10882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4034036824243465e-05,
+      "loss": 0.7275,
+      "step": 10883
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4032983434836553e-05,
+      "loss": 0.7607,
+      "step": 10884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4031929991983159e-05,
+      "loss": 0.748,
+      "step": 10885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403087649569725e-05,
+      "loss": 0.7363,
+      "step": 10886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.402982294599278e-05,
+      "loss": 0.7275,
+      "step": 10887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4028769342883713e-05,
+      "loss": 0.7666,
+      "step": 10888
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4027715686384012e-05,
+      "loss": 0.7529,
+      "step": 10889
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4026661976507643e-05,
+      "loss": 0.75,
+      "step": 10890
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4025608213268565e-05,
+      "loss": 0.7598,
+      "step": 10891
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4024554396680746e-05,
+      "loss": 0.7441,
+      "step": 10892
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4023500526758152e-05,
+      "loss": 0.7432,
+      "step": 10893
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4022446603514745e-05,
+      "loss": 0.75,
+      "step": 10894
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4021392626964496e-05,
+      "loss": 0.7227,
+      "step": 10895
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4020338597121376e-05,
+      "loss": 0.7529,
+      "step": 10896
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4019284513999348e-05,
+      "loss": 0.7451,
+      "step": 10897
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4018230377612378e-05,
+      "loss": 0.75,
+      "step": 10898
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4017176187974444e-05,
+      "loss": 0.75,
+      "step": 10899
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4016121945099512e-05,
+      "loss": 0.7344,
+      "step": 10900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4015067649001552e-05,
+      "loss": 0.7529,
+      "step": 10901
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.401401329969454e-05,
+      "loss": 0.7207,
+      "step": 10902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4012958897192446e-05,
+      "loss": 0.749,
+      "step": 10903
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4011904441509244e-05,
+      "loss": 0.7441,
+      "step": 10904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4010849932658904e-05,
+      "loss": 0.749,
+      "step": 10905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4009795370655408e-05,
+      "loss": 0.7471,
+      "step": 10906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4008740755512724e-05,
+      "loss": 0.7402,
+      "step": 10907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4007686087244835e-05,
+      "loss": 0.7646,
+      "step": 10908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4006631365865714e-05,
+      "loss": 0.7295,
+      "step": 10909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400557659138934e-05,
+      "loss": 0.7754,
+      "step": 10910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4004521763829689e-05,
+      "loss": 0.75,
+      "step": 10911
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4003466883200742e-05,
+      "loss": 0.75,
+      "step": 10912
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4002411949516477e-05,
+      "loss": 0.748,
+      "step": 10913
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4001356962790875e-05,
+      "loss": 0.7471,
+      "step": 10914
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400030192303792e-05,
+      "loss": 0.7266,
+      "step": 10915
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3999246830271589e-05,
+      "loss": 0.7549,
+      "step": 10916
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3998191684505866e-05,
+      "loss": 0.7295,
+      "step": 10917
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3997136485754737e-05,
+      "loss": 0.748,
+      "step": 10918
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3996081234032182e-05,
+      "loss": 0.7266,
+      "step": 10919
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3995025929352189e-05,
+      "loss": 0.7471,
+      "step": 10920
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3993970571728738e-05,
+      "loss": 0.7627,
+      "step": 10921
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3992915161175822e-05,
+      "loss": 0.7451,
+      "step": 10922
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3991859697707419e-05,
+      "loss": 0.7422,
+      "step": 10923
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3990804181337525e-05,
+      "loss": 0.7529,
+      "step": 10924
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3989748612080125e-05,
+      "loss": 0.751,
+      "step": 10925
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3988692989949207e-05,
+      "loss": 0.7451,
+      "step": 10926
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398763731495876e-05,
+      "loss": 0.7432,
+      "step": 10927
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3986581587122777e-05,
+      "loss": 0.7568,
+      "step": 10928
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3985525806455243e-05,
+      "loss": 0.7354,
+      "step": 10929
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3984469972970153e-05,
+      "loss": 0.7402,
+      "step": 10930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3983414086681502e-05,
+      "loss": 0.7373,
+      "step": 10931
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398235814760328e-05,
+      "loss": 0.75,
+      "step": 10932
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398130215574948e-05,
+      "loss": 0.7617,
+      "step": 10933
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3980246111134099e-05,
+      "loss": 0.7637,
+      "step": 10934
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.397919001377113e-05,
+      "loss": 0.7588,
+      "step": 10935
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3978133863674571e-05,
+      "loss": 0.7598,
+      "step": 10936
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3977077660858416e-05,
+      "loss": 0.7412,
+      "step": 10937
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3976021405336662e-05,
+      "loss": 0.7471,
+      "step": 10938
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.397496509712331e-05,
+      "loss": 0.7432,
+      "step": 10939
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3973908736232351e-05,
+      "loss": 0.7529,
+      "step": 10940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3972852322677798e-05,
+      "loss": 0.7441,
+      "step": 10941
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3971795856473636e-05,
+      "loss": 0.7676,
+      "step": 10942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970739337633873e-05,
+      "loss": 0.7168,
+      "step": 10943
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3969682766172509e-05,
+      "loss": 0.7412,
+      "step": 10944
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3968626142103551e-05,
+      "loss": 0.7236,
+      "step": 10945
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3967569465440996e-05,
+      "loss": 0.7305,
+      "step": 10946
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3966512736198847e-05,
+      "loss": 0.7441,
+      "step": 10947
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3965455954391111e-05,
+      "loss": 0.7598,
+      "step": 10948
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3964399120031792e-05,
+      "loss": 0.7363,
+      "step": 10949
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3963342233134896e-05,
+      "loss": 0.7402,
+      "step": 10950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3962285293714428e-05,
+      "loss": 0.7471,
+      "step": 10951
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3961228301784396e-05,
+      "loss": 0.7139,
+      "step": 10952
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3960171257358806e-05,
+      "loss": 0.7656,
+      "step": 10953
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3959114160451667e-05,
+      "loss": 0.75,
+      "step": 10954
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395805701107699e-05,
+      "loss": 0.7646,
+      "step": 10955
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3956999809248782e-05,
+      "loss": 0.7373,
+      "step": 10956
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3955942554981055e-05,
+      "loss": 0.748,
+      "step": 10957
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3954885248287822e-05,
+      "loss": 0.7363,
+      "step": 10958
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395382788918309e-05,
+      "loss": 0.75,
+      "step": 10959
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3952770477680876e-05,
+      "loss": 0.7607,
+      "step": 10960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3951713013795188e-05,
+      "loss": 0.7432,
+      "step": 10961
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3950655497540042e-05,
+      "loss": 0.7598,
+      "step": 10962
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3949597928929457e-05,
+      "loss": 0.7764,
+      "step": 10963
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3948540307977446e-05,
+      "loss": 0.7559,
+      "step": 10964
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3947482634698019e-05,
+      "loss": 0.7354,
+      "step": 10965
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3946424909105204e-05,
+      "loss": 0.749,
+      "step": 10966
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3945367131213008e-05,
+      "loss": 0.7324,
+      "step": 10967
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3944309301035453e-05,
+      "loss": 0.7646,
+      "step": 10968
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3943251418586554e-05,
+      "loss": 0.7578,
+      "step": 10969
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3942193483880337e-05,
+      "loss": 0.7617,
+      "step": 10970
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3941135496930822e-05,
+      "loss": 0.7168,
+      "step": 10971
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3940077457752023e-05,
+      "loss": 0.7334,
+      "step": 10972
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3939019366357967e-05,
+      "loss": 0.751,
+      "step": 10973
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3937961222762675e-05,
+      "loss": 0.7275,
+      "step": 10974
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3936903026980168e-05,
+      "loss": 0.7412,
+      "step": 10975
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3935844779024472e-05,
+      "loss": 0.7568,
+      "step": 10976
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.393478647890961e-05,
+      "loss": 0.7314,
+      "step": 10977
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3933728126649608e-05,
+      "loss": 0.7383,
+      "step": 10978
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3932669722258489e-05,
+      "loss": 0.7617,
+      "step": 10979
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3931611265750284e-05,
+      "loss": 0.7744,
+      "step": 10980
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3930552757139015e-05,
+      "loss": 0.7197,
+      "step": 10981
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3929494196438714e-05,
+      "loss": 0.7705,
+      "step": 10982
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3928435583663405e-05,
+      "loss": 0.7607,
+      "step": 10983
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3927376918827122e-05,
+      "loss": 0.7578,
+      "step": 10984
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3926318201943891e-05,
+      "loss": 0.7285,
+      "step": 10985
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3925259433027742e-05,
+      "loss": 0.7627,
+      "step": 10986
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3924200612092711e-05,
+      "loss": 0.7451,
+      "step": 10987
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3923141739152826e-05,
+      "loss": 0.7451,
+      "step": 10988
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392208281422212e-05,
+      "loss": 0.7197,
+      "step": 10989
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3921023837314627e-05,
+      "loss": 0.7197,
+      "step": 10990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.391996480844438e-05,
+      "loss": 0.7256,
+      "step": 10991
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3918905727625414e-05,
+      "loss": 0.7539,
+      "step": 10992
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3917846594871764e-05,
+      "loss": 0.7393,
+      "step": 10993
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3916787410197468e-05,
+      "loss": 0.7568,
+      "step": 10994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3915728173616558e-05,
+      "loss": 0.7324,
+      "step": 10995
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3914668885143079e-05,
+      "loss": 0.7705,
+      "step": 10996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3913609544791063e-05,
+      "loss": 0.7617,
+      "step": 10997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3912550152574548e-05,
+      "loss": 0.7334,
+      "step": 10998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3911490708507578e-05,
+      "loss": 0.7393,
+      "step": 10999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.391043121260419e-05,
+      "loss": 0.749,
+      "step": 11000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3909371664878424e-05,
+      "loss": 0.7568,
+      "step": 11001
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3908312065344324e-05,
+      "loss": 0.7676,
+      "step": 11002
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3907252414015932e-05,
+      "loss": 0.7324,
+      "step": 11003
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.390619271090729e-05,
+      "loss": 0.7441,
+      "step": 11004
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3905132956032443e-05,
+      "loss": 0.748,
+      "step": 11005
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3904073149405432e-05,
+      "loss": 0.7363,
+      "step": 11006
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3903013291040303e-05,
+      "loss": 0.7178,
+      "step": 11007
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3901953380951106e-05,
+      "loss": 0.7568,
+      "step": 11008
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3900893419151882e-05,
+      "loss": 0.7559,
+      "step": 11009
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389983340565668e-05,
+      "loss": 0.7393,
+      "step": 11010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3898773340479546e-05,
+      "loss": 0.7568,
+      "step": 11011
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389771322363453e-05,
+      "loss": 0.7256,
+      "step": 11012
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3896653055135682e-05,
+      "loss": 0.7129,
+      "step": 11013
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3895592834997049e-05,
+      "loss": 0.7324,
+      "step": 11014
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3894532563232686e-05,
+      "loss": 0.752,
+      "step": 11015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3893472239856637e-05,
+      "loss": 0.7393,
+      "step": 11016
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389241186488296e-05,
+      "loss": 0.7354,
+      "step": 11017
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3891351438325704e-05,
+      "loss": 0.7549,
+      "step": 11018
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3890290960198925e-05,
+      "loss": 0.7305,
+      "step": 11019
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3889230430516674e-05,
+      "loss": 0.7666,
+      "step": 11020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3888169849293004e-05,
+      "loss": 0.748,
+      "step": 11021
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3887109216541977e-05,
+      "loss": 0.7285,
+      "step": 11022
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3886048532277646e-05,
+      "loss": 0.7686,
+      "step": 11023
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.388498779651406e-05,
+      "loss": 0.7422,
+      "step": 11024
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3883927009265285e-05,
+      "loss": 0.7441,
+      "step": 11025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3882866170545379e-05,
+      "loss": 0.7549,
+      "step": 11026
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3881805280368397e-05,
+      "loss": 0.7627,
+      "step": 11027
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3880744338748397e-05,
+      "loss": 0.7393,
+      "step": 11028
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3879683345699445e-05,
+      "loss": 0.7334,
+      "step": 11029
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3878622301235594e-05,
+      "loss": 0.7549,
+      "step": 11030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3877561205370911e-05,
+      "loss": 0.7393,
+      "step": 11031
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3876500058119457e-05,
+      "loss": 0.7607,
+      "step": 11032
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3875438859495294e-05,
+      "loss": 0.793,
+      "step": 11033
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3874377609512482e-05,
+      "loss": 0.7363,
+      "step": 11034
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3873316308185095e-05,
+      "loss": 0.7529,
+      "step": 11035
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3872254955527185e-05,
+      "loss": 0.7666,
+      "step": 11036
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3871193551552825e-05,
+      "loss": 0.7275,
+      "step": 11037
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387013209627608e-05,
+      "loss": 0.7354,
+      "step": 11038
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3869070589711018e-05,
+      "loss": 0.7578,
+      "step": 11039
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3868009031871701e-05,
+      "loss": 0.75,
+      "step": 11040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3866947422772204e-05,
+      "loss": 0.7402,
+      "step": 11041
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3865885762426592e-05,
+      "loss": 0.7617,
+      "step": 11042
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3864824050848933e-05,
+      "loss": 0.7354,
+      "step": 11043
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3863762288053302e-05,
+      "loss": 0.7627,
+      "step": 11044
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3862700474053768e-05,
+      "loss": 0.7568,
+      "step": 11045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3861638608864398e-05,
+      "loss": 0.7246,
+      "step": 11046
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386057669249927e-05,
+      "loss": 0.7627,
+      "step": 11047
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3859514724972455e-05,
+      "loss": 0.75,
+      "step": 11048
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3858452706298027e-05,
+      "loss": 0.7471,
+      "step": 11049
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.385739063649006e-05,
+      "loss": 0.7803,
+      "step": 11050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3856328515562626e-05,
+      "loss": 0.7227,
+      "step": 11051
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3855266343529805e-05,
+      "loss": 0.7383,
+      "step": 11052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.385420412040567e-05,
+      "loss": 0.7324,
+      "step": 11053
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3853141846204302e-05,
+      "loss": 0.7275,
+      "step": 11054
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3852079520939774e-05,
+      "loss": 0.7344,
+      "step": 11055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3851017144626166e-05,
+      "loss": 0.7783,
+      "step": 11056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3849954717277556e-05,
+      "loss": 0.7354,
+      "step": 11057
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384889223890803e-05,
+      "loss": 0.7168,
+      "step": 11058
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3847829709531659e-05,
+      "loss": 0.7402,
+      "step": 11059
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3846767129162528e-05,
+      "loss": 0.748,
+      "step": 11060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384570449781472e-05,
+      "loss": 0.7568,
+      "step": 11061
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3844641815502317e-05,
+      "loss": 0.7471,
+      "step": 11062
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3843579082239398e-05,
+      "loss": 0.7578,
+      "step": 11063
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3842516298040054e-05,
+      "loss": 0.7246,
+      "step": 11064
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3841453462918364e-05,
+      "loss": 0.7539,
+      "step": 11065
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3840390576888414e-05,
+      "loss": 0.7266,
+      "step": 11066
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.383932763996429e-05,
+      "loss": 0.7334,
+      "step": 11067
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3838264652160081e-05,
+      "loss": 0.7021,
+      "step": 11068
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837201613489868e-05,
+      "loss": 0.7373,
+      "step": 11069
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3836138523967744e-05,
+      "loss": 0.7314,
+      "step": 11070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3835075383607797e-05,
+      "loss": 0.7432,
+      "step": 11071
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3834012192424113e-05,
+      "loss": 0.7549,
+      "step": 11072
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3832948950430782e-05,
+      "loss": 0.7539,
+      "step": 11073
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.38318856576419e-05,
+      "loss": 0.7627,
+      "step": 11074
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3830822314071553e-05,
+      "loss": 0.7617,
+      "step": 11075
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3829758919733834e-05,
+      "loss": 0.7383,
+      "step": 11076
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3828695474642835e-05,
+      "loss": 0.7412,
+      "step": 11077
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3827631978812652e-05,
+      "loss": 0.7676,
+      "step": 11078
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3826568432257372e-05,
+      "loss": 0.7305,
+      "step": 11079
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3825504834991098e-05,
+      "loss": 0.7236,
+      "step": 11080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3824441187027919e-05,
+      "loss": 0.7607,
+      "step": 11081
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3823377488381936e-05,
+      "loss": 0.7461,
+      "step": 11082
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3822313739067236e-05,
+      "loss": 0.7305,
+      "step": 11083
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3821249939097929e-05,
+      "loss": 0.7432,
+      "step": 11084
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3820186088488104e-05,
+      "loss": 0.7734,
+      "step": 11085
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381912218725186e-05,
+      "loss": 0.7627,
+      "step": 11086
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3818058235403303e-05,
+      "loss": 0.7354,
+      "step": 11087
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3816994232956524e-05,
+      "loss": 0.7227,
+      "step": 11088
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3815930179925629e-05,
+      "loss": 0.7275,
+      "step": 11089
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3814866076324715e-05,
+      "loss": 0.7451,
+      "step": 11090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381380192216789e-05,
+      "loss": 0.7451,
+      "step": 11091
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3812737717469253e-05,
+      "loss": 0.7471,
+      "step": 11092
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3811673462242908e-05,
+      "loss": 0.7344,
+      "step": 11093
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3810609156502955e-05,
+      "loss": 0.7402,
+      "step": 11094
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3809544800263504e-05,
+      "loss": 0.75,
+      "step": 11095
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3808480393538655e-05,
+      "loss": 0.7568,
+      "step": 11096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3807415936342523e-05,
+      "loss": 0.7539,
+      "step": 11097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806351428689202e-05,
+      "loss": 0.7393,
+      "step": 11098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.380528687059281e-05,
+      "loss": 0.7471,
+      "step": 11099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3804222262067448e-05,
+      "loss": 0.7627,
+      "step": 11100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.380315760312723e-05,
+      "loss": 0.7432,
+      "step": 11101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3802092893786257e-05,
+      "loss": 0.749,
+      "step": 11102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3801028134058648e-05,
+      "loss": 0.7139,
+      "step": 11103
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.379996332395851e-05,
+      "loss": 0.75,
+      "step": 11104
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3798898463499953e-05,
+      "loss": 0.749,
+      "step": 11105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3797833552697088e-05,
+      "loss": 0.7559,
+      "step": 11106
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3796768591564033e-05,
+      "loss": 0.7393,
+      "step": 11107
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3795703580114897e-05,
+      "loss": 0.7393,
+      "step": 11108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3794638518363792e-05,
+      "loss": 0.7588,
+      "step": 11109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3793573406324838e-05,
+      "loss": 0.7607,
+      "step": 11110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3792508244012146e-05,
+      "loss": 0.7588,
+      "step": 11111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3791443031439831e-05,
+      "loss": 0.7344,
+      "step": 11112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3790377768622016e-05,
+      "loss": 0.7246,
+      "step": 11113
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3789312455572811e-05,
+      "loss": 0.7334,
+      "step": 11114
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.378824709230634e-05,
+      "loss": 0.749,
+      "step": 11115
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3787181678836714e-05,
+      "loss": 0.7549,
+      "step": 11116
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3786116215178063e-05,
+      "loss": 0.7354,
+      "step": 11117
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3785050701344496e-05,
+      "loss": 0.7266,
+      "step": 11118
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3783985137350137e-05,
+      "loss": 0.7383,
+      "step": 11119
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3782919523209113e-05,
+      "loss": 0.7471,
+      "step": 11120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.378185385893554e-05,
+      "loss": 0.7559,
+      "step": 11121
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3780788144543537e-05,
+      "loss": 0.7393,
+      "step": 11122
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3779722380047237e-05,
+      "loss": 0.7373,
+      "step": 11123
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3778656565460758e-05,
+      "loss": 0.7402,
+      "step": 11124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3777590700798224e-05,
+      "loss": 0.7598,
+      "step": 11125
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3776524786073763e-05,
+      "loss": 0.7363,
+      "step": 11126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3775458821301501e-05,
+      "loss": 0.7734,
+      "step": 11127
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3774392806495563e-05,
+      "loss": 0.7539,
+      "step": 11128
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3773326741670072e-05,
+      "loss": 0.7549,
+      "step": 11129
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3772260626839163e-05,
+      "loss": 0.7695,
+      "step": 11130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3771194462016962e-05,
+      "loss": 0.7305,
+      "step": 11131
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3770128247217598e-05,
+      "loss": 0.7256,
+      "step": 11132
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.37690619824552e-05,
+      "loss": 0.7314,
+      "step": 11133
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3767995667743898e-05,
+      "loss": 0.7627,
+      "step": 11134
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3766929303097824e-05,
+      "loss": 0.7314,
+      "step": 11135
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3765862888531112e-05,
+      "loss": 0.7324,
+      "step": 11136
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3764796424057893e-05,
+      "loss": 0.7549,
+      "step": 11137
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3763729909692299e-05,
+      "loss": 0.7402,
+      "step": 11138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3762663345448466e-05,
+      "loss": 0.751,
+      "step": 11139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3761596731340525e-05,
+      "loss": 0.75,
+      "step": 11140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3760530067382613e-05,
+      "loss": 0.7461,
+      "step": 11141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3759463353588869e-05,
+      "loss": 0.7383,
+      "step": 11142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3758396589973421e-05,
+      "loss": 0.7568,
+      "step": 11143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3757329776550416e-05,
+      "loss": 0.7559,
+      "step": 11144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3756262913333987e-05,
+      "loss": 0.7461,
+      "step": 11145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3755196000338274e-05,
+      "loss": 0.7256,
+      "step": 11146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3754129037577411e-05,
+      "loss": 0.7295,
+      "step": 11147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3753062025065545e-05,
+      "loss": 0.7324,
+      "step": 11148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.375199496281681e-05,
+      "loss": 0.751,
+      "step": 11149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3750927850845354e-05,
+      "loss": 0.7383,
+      "step": 11150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.374986068916531e-05,
+      "loss": 0.7559,
+      "step": 11151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3748793477790832e-05,
+      "loss": 0.7295,
+      "step": 11152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.374772621673605e-05,
+      "loss": 0.7393,
+      "step": 11153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3746658906015118e-05,
+      "loss": 0.7451,
+      "step": 11154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3745591545642177e-05,
+      "loss": 0.7314,
+      "step": 11155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.374452413563137e-05,
+      "loss": 0.7393,
+      "step": 11156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.374345667599684e-05,
+      "loss": 0.7285,
+      "step": 11157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3742389166752741e-05,
+      "loss": 0.7295,
+      "step": 11158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3741321607913218e-05,
+      "loss": 0.7451,
+      "step": 11159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3740253999492416e-05,
+      "loss": 0.7295,
+      "step": 11160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3739186341504483e-05,
+      "loss": 0.7373,
+      "step": 11161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3738118633963574e-05,
+      "loss": 0.7871,
+      "step": 11162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3737050876883826e-05,
+      "loss": 0.7422,
+      "step": 11163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3735983070279403e-05,
+      "loss": 0.7559,
+      "step": 11164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373491521416445e-05,
+      "loss": 0.7402,
+      "step": 11165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3733847308553119e-05,
+      "loss": 0.7393,
+      "step": 11166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373277935345956e-05,
+      "loss": 0.749,
+      "step": 11167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3731711348897932e-05,
+      "loss": 0.7637,
+      "step": 11168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373064329488238e-05,
+      "loss": 0.7354,
+      "step": 11169
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3729575191427064e-05,
+      "loss": 0.7715,
+      "step": 11170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372850703854614e-05,
+      "loss": 0.7139,
+      "step": 11171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3727438836253759e-05,
+      "loss": 0.7393,
+      "step": 11172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3726370584564079e-05,
+      "loss": 0.7266,
+      "step": 11173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3725302283491258e-05,
+      "loss": 0.7529,
+      "step": 11174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3724233933049451e-05,
+      "loss": 0.7285,
+      "step": 11175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372316553325282e-05,
+      "loss": 0.7305,
+      "step": 11176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3722097084115522e-05,
+      "loss": 0.7393,
+      "step": 11177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3721028585651717e-05,
+      "loss": 0.7432,
+      "step": 11178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.371996003787556e-05,
+      "loss": 0.71,
+      "step": 11179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3718891440801218e-05,
+      "loss": 0.75,
+      "step": 11180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3717822794442855e-05,
+      "loss": 0.7236,
+      "step": 11181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3716754098814623e-05,
+      "loss": 0.7314,
+      "step": 11182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.371568535393069e-05,
+      "loss": 0.7646,
+      "step": 11183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3714616559805221e-05,
+      "loss": 0.7158,
+      "step": 11184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3713547716452382e-05,
+      "loss": 0.7607,
+      "step": 11185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3712478823886328e-05,
+      "loss": 0.7695,
+      "step": 11186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3711409882121233e-05,
+      "loss": 0.7383,
+      "step": 11187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3710340891171262e-05,
+      "loss": 0.7441,
+      "step": 11188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370927185105058e-05,
+      "loss": 0.7451,
+      "step": 11189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3708202761773353e-05,
+      "loss": 0.7588,
+      "step": 11190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3707133623353755e-05,
+      "loss": 0.7432,
+      "step": 11191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3706064435805943e-05,
+      "loss": 0.7363,
+      "step": 11192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704995199144098e-05,
+      "loss": 0.75,
+      "step": 11193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3703925913382383e-05,
+      "loss": 0.7412,
+      "step": 11194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3702856578534974e-05,
+      "loss": 0.7275,
+      "step": 11195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3701787194616034e-05,
+      "loss": 0.7363,
+      "step": 11196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3700717761639743e-05,
+      "loss": 0.7568,
+      "step": 11197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369964827962027e-05,
+      "loss": 0.7236,
+      "step": 11198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3698578748571787e-05,
+      "loss": 0.75,
+      "step": 11199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3697509168508469e-05,
+      "loss": 0.7539,
+      "step": 11200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3696439539444493e-05,
+      "loss": 0.7305,
+      "step": 11201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3695369861394028e-05,
+      "loss": 0.7598,
+      "step": 11202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3694300134371257e-05,
+      "loss": 0.7568,
+      "step": 11203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3693230358390353e-05,
+      "loss": 0.7129,
+      "step": 11204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369216053346549e-05,
+      "loss": 0.7314,
+      "step": 11205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369109065961085e-05,
+      "loss": 0.7314,
+      "step": 11206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3690020736840612e-05,
+      "loss": 0.7236,
+      "step": 11207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368895076516895e-05,
+      "loss": 0.7295,
+      "step": 11208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3687880744610047e-05,
+      "loss": 0.7627,
+      "step": 11209
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3686810675178083e-05,
+      "loss": 0.751,
+      "step": 11210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3685740556887238e-05,
+      "loss": 0.7354,
+      "step": 11211
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3684670389751698e-05,
+      "loss": 0.7676,
+      "step": 11212
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3683600173785637e-05,
+      "loss": 0.7471,
+      "step": 11213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3682529909003242e-05,
+      "loss": 0.7471,
+      "step": 11214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.36814595954187e-05,
+      "loss": 0.7383,
+      "step": 11215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368038923304619e-05,
+      "loss": 0.7529,
+      "step": 11216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3679318821899901e-05,
+      "loss": 0.7324,
+      "step": 11217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3678248361994016e-05,
+      "loss": 0.7529,
+      "step": 11218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3677177853342718e-05,
+      "loss": 0.748,
+      "step": 11219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3676107295960202e-05,
+      "loss": 0.7266,
+      "step": 11220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3675036689860646e-05,
+      "loss": 0.751,
+      "step": 11221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3673966035058247e-05,
+      "loss": 0.7236,
+      "step": 11222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3672895331567184e-05,
+      "loss": 0.7549,
+      "step": 11223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3671824579401658e-05,
+      "loss": 0.7266,
+      "step": 11224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3670753778575847e-05,
+      "loss": 0.7363,
+      "step": 11225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3669682929103951e-05,
+      "loss": 0.7383,
+      "step": 11226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3668612031000157e-05,
+      "loss": 0.7715,
+      "step": 11227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3667541084278655e-05,
+      "loss": 0.7354,
+      "step": 11228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.366647008895364e-05,
+      "loss": 0.7168,
+      "step": 11229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3665399045039309e-05,
+      "loss": 0.752,
+      "step": 11230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3664327952549847e-05,
+      "loss": 0.7295,
+      "step": 11231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3663256811499457e-05,
+      "loss": 0.7373,
+      "step": 11232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.366218562190233e-05,
+      "loss": 0.7754,
+      "step": 11233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3661114383772661e-05,
+      "loss": 0.7422,
+      "step": 11234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3660043097124649e-05,
+      "loss": 0.7451,
+      "step": 11235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.365897176197249e-05,
+      "loss": 0.7324,
+      "step": 11236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3657900378330381e-05,
+      "loss": 0.7432,
+      "step": 11237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.365682894621252e-05,
+      "loss": 0.7373,
+      "step": 11238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3655757465633108e-05,
+      "loss": 0.7275,
+      "step": 11239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3654685936606347e-05,
+      "loss": 0.7451,
+      "step": 11240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3653614359146428e-05,
+      "loss": 0.7744,
+      "step": 11241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3652542733267559e-05,
+      "loss": 0.7412,
+      "step": 11242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3651471058983943e-05,
+      "loss": 0.7529,
+      "step": 11243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3650399336309776e-05,
+      "loss": 0.7266,
+      "step": 11244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3649327565259266e-05,
+      "loss": 0.7588,
+      "step": 11245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3648255745846616e-05,
+      "loss": 0.7539,
+      "step": 11246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3647183878086026e-05,
+      "loss": 0.7764,
+      "step": 11247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3646111961991706e-05,
+      "loss": 0.7432,
+      "step": 11248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3645039997577859e-05,
+      "loss": 0.7422,
+      "step": 11249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364396798485869e-05,
+      "loss": 0.749,
+      "step": 11250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3642895923848404e-05,
+      "loss": 0.7529,
+      "step": 11251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3641823814561214e-05,
+      "loss": 0.7549,
+      "step": 11252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3640751657011327e-05,
+      "loss": 0.7344,
+      "step": 11253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3639679451212948e-05,
+      "loss": 0.751,
+      "step": 11254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3638607197180287e-05,
+      "loss": 0.7471,
+      "step": 11255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3637534894927555e-05,
+      "loss": 0.7334,
+      "step": 11256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3636462544468965e-05,
+      "loss": 0.7354,
+      "step": 11257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3635390145818722e-05,
+      "loss": 0.748,
+      "step": 11258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363431769899104e-05,
+      "loss": 0.7383,
+      "step": 11259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3633245204000137e-05,
+      "loss": 0.7324,
+      "step": 11260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363217266086022e-05,
+      "loss": 0.7354,
+      "step": 11261
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3631100069585504e-05,
+      "loss": 0.7607,
+      "step": 11262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3630027430190204e-05,
+      "loss": 0.751,
+      "step": 11263
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3628954742688531e-05,
+      "loss": 0.7422,
+      "step": 11264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362788200709471e-05,
+      "loss": 0.7588,
+      "step": 11265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362680922342295e-05,
+      "loss": 0.7549,
+      "step": 11266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362573639168747e-05,
+      "loss": 0.748,
+      "step": 11267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3624663511902484e-05,
+      "loss": 0.7197,
+      "step": 11268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3623590584082216e-05,
+      "loss": 0.7383,
+      "step": 11269
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3622517608240882e-05,
+      "loss": 0.7373,
+      "step": 11270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.36214445843927e-05,
+      "loss": 0.7432,
+      "step": 11271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3620371512551893e-05,
+      "loss": 0.7305,
+      "step": 11272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.361929839273268e-05,
+      "loss": 0.7373,
+      "step": 11273
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3618225224949282e-05,
+      "loss": 0.751,
+      "step": 11274
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3617152009215924e-05,
+      "loss": 0.7236,
+      "step": 11275
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3616078745546823e-05,
+      "loss": 0.7373,
+      "step": 11276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3615005433956205e-05,
+      "loss": 0.7822,
+      "step": 11277
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3613932074458295e-05,
+      "loss": 0.751,
+      "step": 11278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3612858667067321e-05,
+      "loss": 0.7344,
+      "step": 11279
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3611785211797502e-05,
+      "loss": 0.7344,
+      "step": 11280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3610711708663065e-05,
+      "loss": 0.7314,
+      "step": 11281
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3609638157678237e-05,
+      "loss": 0.7334,
+      "step": 11282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3608564558857246e-05,
+      "loss": 0.7568,
+      "step": 11283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3607490912214319e-05,
+      "loss": 0.7314,
+      "step": 11284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3606417217763686e-05,
+      "loss": 0.7305,
+      "step": 11285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360534347551957e-05,
+      "loss": 0.7334,
+      "step": 11286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360426968549621e-05,
+      "loss": 0.7441,
+      "step": 11287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.360319584770783e-05,
+      "loss": 0.7256,
+      "step": 11288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3602121962168662e-05,
+      "loss": 0.751,
+      "step": 11289
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3601048028892933e-05,
+      "loss": 0.7363,
+      "step": 11290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3599974047894886e-05,
+      "loss": 0.7559,
+      "step": 11291
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3598900019188745e-05,
+      "loss": 0.749,
+      "step": 11292
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3597825942788747e-05,
+      "loss": 0.7344,
+      "step": 11293
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3596751818709124e-05,
+      "loss": 0.7334,
+      "step": 11294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3595677646964112e-05,
+      "loss": 0.7266,
+      "step": 11295
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3594603427567942e-05,
+      "loss": 0.7402,
+      "step": 11296
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.359352916053486e-05,
+      "loss": 0.7295,
+      "step": 11297
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3592454845879093e-05,
+      "loss": 0.7285,
+      "step": 11298
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3591380483614883e-05,
+      "loss": 0.7363,
+      "step": 11299
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3590306073756465e-05,
+      "loss": 0.7295,
+      "step": 11300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358923161631808e-05,
+      "loss": 0.7295,
+      "step": 11301
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3588157111313963e-05,
+      "loss": 0.75,
+      "step": 11302
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3587082558758358e-05,
+      "loss": 0.7441,
+      "step": 11303
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3586007958665506e-05,
+      "loss": 0.7559,
+      "step": 11304
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3584933311049646e-05,
+      "loss": 0.748,
+      "step": 11305
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3583858615925016e-05,
+      "loss": 0.7344,
+      "step": 11306
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3582783873305863e-05,
+      "loss": 0.7471,
+      "step": 11307
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3581709083206432e-05,
+      "loss": 0.7354,
+      "step": 11308
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358063424564096e-05,
+      "loss": 0.7354,
+      "step": 11309
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3579559360623695e-05,
+      "loss": 0.7451,
+      "step": 11310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3578484428168882e-05,
+      "loss": 0.7549,
+      "step": 11311
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3577409448290762e-05,
+      "loss": 0.7246,
+      "step": 11312
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3576334421003585e-05,
+      "loss": 0.7246,
+      "step": 11313
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.35752593463216e-05,
+      "loss": 0.79,
+      "step": 11314
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3574184224259048e-05,
+      "loss": 0.7402,
+      "step": 11315
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3573109054830183e-05,
+      "loss": 0.7305,
+      "step": 11316
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3572033838049248e-05,
+      "loss": 0.7422,
+      "step": 11317
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3570958573930497e-05,
+      "loss": 0.7451,
+      "step": 11318
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3569883262488176e-05,
+      "loss": 0.7373,
+      "step": 11319
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3568807903736537e-05,
+      "loss": 0.7383,
+      "step": 11320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3567732497689832e-05,
+      "loss": 0.7246,
+      "step": 11321
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356665704436231e-05,
+      "loss": 0.7471,
+      "step": 11322
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3565581543768224e-05,
+      "loss": 0.7539,
+      "step": 11323
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3564505995921833e-05,
+      "loss": 0.7393,
+      "step": 11324
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.356343040083738e-05,
+      "loss": 0.7646,
+      "step": 11325
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3562354758529126e-05,
+      "loss": 0.7246,
+      "step": 11326
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3561279069011324e-05,
+      "loss": 0.7285,
+      "step": 11327
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3560203332298231e-05,
+      "loss": 0.7236,
+      "step": 11328
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3559127548404098e-05,
+      "loss": 0.7539,
+      "step": 11329
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355805171734319e-05,
+      "loss": 0.7705,
+      "step": 11330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3556975839129758e-05,
+      "loss": 0.7412,
+      "step": 11331
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3555899913778062e-05,
+      "loss": 0.7246,
+      "step": 11332
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355482394130236e-05,
+      "loss": 0.7363,
+      "step": 11333
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3553747921716908e-05,
+      "loss": 0.7305,
+      "step": 11334
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3552671855035971e-05,
+      "loss": 0.7725,
+      "step": 11335
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3551595741273808e-05,
+      "loss": 0.75,
+      "step": 11336
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3550519580444681e-05,
+      "loss": 0.7734,
+      "step": 11337
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.354944337256285e-05,
+      "loss": 0.7588,
+      "step": 11338
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3548367117642573e-05,
+      "loss": 0.7422,
+      "step": 11339
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3547290815698121e-05,
+      "loss": 0.7588,
+      "step": 11340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3546214466743752e-05,
+      "loss": 0.7217,
+      "step": 11341
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3545138070793734e-05,
+      "loss": 0.7461,
+      "step": 11342
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3544061627862326e-05,
+      "loss": 0.7217,
+      "step": 11343
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3542985137963803e-05,
+      "loss": 0.7393,
+      "step": 11344
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3541908601112418e-05,
+      "loss": 0.7178,
+      "step": 11345
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3540832017322449e-05,
+      "loss": 0.7559,
+      "step": 11346
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.353975538660816e-05,
+      "loss": 0.7256,
+      "step": 11347
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3538678708983815e-05,
+      "loss": 0.7598,
+      "step": 11348
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3537601984463686e-05,
+      "loss": 0.7314,
+      "step": 11349
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3536525213062042e-05,
+      "loss": 0.7422,
+      "step": 11350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3535448394793153e-05,
+      "loss": 0.7393,
+      "step": 11351
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3534371529671287e-05,
+      "loss": 0.7285,
+      "step": 11352
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.353329461771072e-05,
+      "loss": 0.7305,
+      "step": 11353
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3532217658925716e-05,
+      "loss": 0.7227,
+      "step": 11354
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3531140653330552e-05,
+      "loss": 0.7246,
+      "step": 11355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.35300636009395e-05,
+      "loss": 0.7227,
+      "step": 11356
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3528986501766836e-05,
+      "loss": 0.7598,
+      "step": 11357
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352790935582683e-05,
+      "loss": 0.7412,
+      "step": 11358
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3526832163133758e-05,
+      "loss": 0.7373,
+      "step": 11359
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.35257549237019e-05,
+      "loss": 0.7422,
+      "step": 11360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3524677637545524e-05,
+      "loss": 0.7402,
+      "step": 11361
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352360030467891e-05,
+      "loss": 0.751,
+      "step": 11362
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3522522925116339e-05,
+      "loss": 0.7422,
+      "step": 11363
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3521445498872081e-05,
+      "loss": 0.7559,
+      "step": 11364
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3520368025960422e-05,
+      "loss": 0.749,
+      "step": 11365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3519290506395638e-05,
+      "loss": 0.7627,
+      "step": 11366
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3518212940192006e-05,
+      "loss": 0.7451,
+      "step": 11367
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3517135327363808e-05,
+      "loss": 0.7578,
+      "step": 11368
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351605766792533e-05,
+      "loss": 0.7285,
+      "step": 11369
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3514979961890846e-05,
+      "loss": 0.7422,
+      "step": 11370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3513902209274643e-05,
+      "loss": 0.7373,
+      "step": 11371
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3512824410091e-05,
+      "loss": 0.7285,
+      "step": 11372
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3511746564354206e-05,
+      "loss": 0.7383,
+      "step": 11373
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3510668672078538e-05,
+      "loss": 0.7393,
+      "step": 11374
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3509590733278284e-05,
+      "loss": 0.7256,
+      "step": 11375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350851274796773e-05,
+      "loss": 0.7451,
+      "step": 11376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3507434716161163e-05,
+      "loss": 0.7256,
+      "step": 11377
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3506356637872867e-05,
+      "loss": 0.7461,
+      "step": 11378
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3505278513117129e-05,
+      "loss": 0.7471,
+      "step": 11379
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3504200341908234e-05,
+      "loss": 0.7129,
+      "step": 11380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3503122124260475e-05,
+      "loss": 0.7734,
+      "step": 11381
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3502043860188141e-05,
+      "loss": 0.7383,
+      "step": 11382
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350096554970552e-05,
+      "loss": 0.7383,
+      "step": 11383
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3499887192826902e-05,
+      "loss": 0.752,
+      "step": 11384
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3498808789566575e-05,
+      "loss": 0.7217,
+      "step": 11385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3497730339938836e-05,
+      "loss": 0.7559,
+      "step": 11386
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3496651843957974e-05,
+      "loss": 0.7354,
+      "step": 11387
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3495573301638283e-05,
+      "loss": 0.7393,
+      "step": 11388
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3494494712994052e-05,
+      "loss": 0.7383,
+      "step": 11389
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.349341607803958e-05,
+      "loss": 0.7334,
+      "step": 11390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3492337396789156e-05,
+      "loss": 0.7402,
+      "step": 11391
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3491258669257085e-05,
+      "loss": 0.751,
+      "step": 11392
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3490179895457649e-05,
+      "loss": 0.7266,
+      "step": 11393
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3489101075405155e-05,
+      "loss": 0.7422,
+      "step": 11394
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3488022209113899e-05,
+      "loss": 0.7158,
+      "step": 11395
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3486943296598169e-05,
+      "loss": 0.7393,
+      "step": 11396
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3485864337872272e-05,
+      "loss": 0.7627,
+      "step": 11397
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3484785332950505e-05,
+      "loss": 0.7422,
+      "step": 11398
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3483706281847168e-05,
+      "loss": 0.7295,
+      "step": 11399
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3482627184576562e-05,
+      "loss": 0.7158,
+      "step": 11400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3481548041152981e-05,
+      "loss": 0.7393,
+      "step": 11401
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3480468851590733e-05,
+      "loss": 0.708,
+      "step": 11402
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3479389615904116e-05,
+      "loss": 0.7441,
+      "step": 11403
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3478310334107433e-05,
+      "loss": 0.7393,
+      "step": 11404
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3477231006214991e-05,
+      "loss": 0.7344,
+      "step": 11405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3476151632241088e-05,
+      "loss": 0.7285,
+      "step": 11406
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3475072212200031e-05,
+      "loss": 0.7441,
+      "step": 11407
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3473992746106125e-05,
+      "loss": 0.7705,
+      "step": 11408
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3472913233973675e-05,
+      "loss": 0.7285,
+      "step": 11409
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3471833675816986e-05,
+      "loss": 0.7295,
+      "step": 11410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3470754071650365e-05,
+      "loss": 0.7383,
+      "step": 11411
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3469674421488124e-05,
+      "loss": 0.7363,
+      "step": 11412
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.346859472534456e-05,
+      "loss": 0.7363,
+      "step": 11413
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3467514983233993e-05,
+      "loss": 0.7295,
+      "step": 11414
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3466435195170729e-05,
+      "loss": 0.7461,
+      "step": 11415
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3465355361169075e-05,
+      "loss": 0.7432,
+      "step": 11416
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3464275481243344e-05,
+      "loss": 0.7686,
+      "step": 11417
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3463195555407841e-05,
+      "loss": 0.7539,
+      "step": 11418
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3462115583676882e-05,
+      "loss": 0.751,
+      "step": 11419
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3461035566064781e-05,
+      "loss": 0.7393,
+      "step": 11420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.345995550258585e-05,
+      "loss": 0.749,
+      "step": 11421
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3458875393254403e-05,
+      "loss": 0.75,
+      "step": 11422
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3457795238084749e-05,
+      "loss": 0.7402,
+      "step": 11423
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3456715037091206e-05,
+      "loss": 0.7412,
+      "step": 11424
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3455634790288093e-05,
+      "loss": 0.751,
+      "step": 11425
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3454554497689716e-05,
+      "loss": 0.7363,
+      "step": 11426
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.34534741593104e-05,
+      "loss": 0.7383,
+      "step": 11427
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3452393775164461e-05,
+      "loss": 0.7119,
+      "step": 11428
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3451313345266213e-05,
+      "loss": 0.7051,
+      "step": 11429
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3450232869629973e-05,
+      "loss": 0.7207,
+      "step": 11430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3449152348270072e-05,
+      "loss": 0.7256,
+      "step": 11431
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3448071781200811e-05,
+      "loss": 0.7383,
+      "step": 11432
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3446991168436526e-05,
+      "loss": 0.7471,
+      "step": 11433
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3445910509991526e-05,
+      "loss": 0.75,
+      "step": 11434
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3444829805880143e-05,
+      "loss": 0.7598,
+      "step": 11435
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3443749056116685e-05,
+      "loss": 0.7578,
+      "step": 11436
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3442668260715491e-05,
+      "loss": 0.7451,
+      "step": 11437
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3441587419690871e-05,
+      "loss": 0.7197,
+      "step": 11438
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3440506533057155e-05,
+      "loss": 0.7617,
+      "step": 11439
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3439425600828664e-05,
+      "loss": 0.751,
+      "step": 11440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343834462301973e-05,
+      "loss": 0.7432,
+      "step": 11441
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3437263599644667e-05,
+      "loss": 0.7363,
+      "step": 11442
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343618253071781e-05,
+      "loss": 0.7568,
+      "step": 11443
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3435101416253482e-05,
+      "loss": 0.7402,
+      "step": 11444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3434020256266012e-05,
+      "loss": 0.75,
+      "step": 11445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3432939050769727e-05,
+      "loss": 0.7578,
+      "step": 11446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3431857799778958e-05,
+      "loss": 0.7344,
+      "step": 11447
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3430776503308027e-05,
+      "loss": 0.7275,
+      "step": 11448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.342969516137127e-05,
+      "loss": 0.7676,
+      "step": 11449
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3428613773983014e-05,
+      "loss": 0.75,
+      "step": 11450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3427532341157596e-05,
+      "loss": 0.7461,
+      "step": 11451
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3426450862909339e-05,
+      "loss": 0.7441,
+      "step": 11452
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3425369339252583e-05,
+      "loss": 0.7568,
+      "step": 11453
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3424287770201656e-05,
+      "loss": 0.748,
+      "step": 11454
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.342320615577089e-05,
+      "loss": 0.7373,
+      "step": 11455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3422124495974622e-05,
+      "loss": 0.7568,
+      "step": 11456
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.342104279082719e-05,
+      "loss": 0.7305,
+      "step": 11457
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3419961040342919e-05,
+      "loss": 0.7168,
+      "step": 11458
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3418879244536154e-05,
+      "loss": 0.7393,
+      "step": 11459
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.341779740342123e-05,
+      "loss": 0.7363,
+      "step": 11460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3416715517012478e-05,
+      "loss": 0.7441,
+      "step": 11461
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3415633585324242e-05,
+      "loss": 0.7227,
+      "step": 11462
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3414551608370853e-05,
+      "loss": 0.7354,
+      "step": 11463
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3413469586166659e-05,
+      "loss": 0.7168,
+      "step": 11464
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3412387518725994e-05,
+      "loss": 0.7441,
+      "step": 11465
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3411305406063197e-05,
+      "loss": 0.75,
+      "step": 11466
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.341022324819261e-05,
+      "loss": 0.752,
+      "step": 11467
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3409141045128578e-05,
+      "loss": 0.7168,
+      "step": 11468
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3408058796885435e-05,
+      "loss": 0.7324,
+      "step": 11469
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3406976503477529e-05,
+      "loss": 0.748,
+      "step": 11470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3405894164919197e-05,
+      "loss": 0.7275,
+      "step": 11471
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3404811781224792e-05,
+      "loss": 0.7178,
+      "step": 11472
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3403729352408652e-05,
+      "loss": 0.749,
+      "step": 11473
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3402646878485122e-05,
+      "loss": 0.7256,
+      "step": 11474
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3401564359468547e-05,
+      "loss": 0.7031,
+      "step": 11475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3400481795373275e-05,
+      "loss": 0.7412,
+      "step": 11476
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3399399186213649e-05,
+      "loss": 0.7275,
+      "step": 11477
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.339831653200402e-05,
+      "loss": 0.7266,
+      "step": 11478
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.339723383275873e-05,
+      "loss": 0.7383,
+      "step": 11479
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3396151088492139e-05,
+      "loss": 0.7236,
+      "step": 11480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3395068299218586e-05,
+      "loss": 0.7197,
+      "step": 11481
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3393985464952419e-05,
+      "loss": 0.7334,
+      "step": 11482
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3392902585707996e-05,
+      "loss": 0.7607,
+      "step": 11483
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3391819661499663e-05,
+      "loss": 0.7617,
+      "step": 11484
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3390736692341769e-05,
+      "loss": 0.7363,
+      "step": 11485
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3389653678248671e-05,
+      "loss": 0.7441,
+      "step": 11486
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3388570619234717e-05,
+      "loss": 0.7148,
+      "step": 11487
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3387487515314265e-05,
+      "loss": 0.748,
+      "step": 11488
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3386404366501667e-05,
+      "loss": 0.7607,
+      "step": 11489
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3385321172811273e-05,
+      "loss": 0.7549,
+      "step": 11490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.338423793425744e-05,
+      "loss": 0.7373,
+      "step": 11491
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3383154650854528e-05,
+      "loss": 0.7334,
+      "step": 11492
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.338207132261689e-05,
+      "loss": 0.7676,
+      "step": 11493
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.338098794955888e-05,
+      "loss": 0.7207,
+      "step": 11494
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3379904531694856e-05,
+      "loss": 0.7158,
+      "step": 11495
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3378821069039183e-05,
+      "loss": 0.7188,
+      "step": 11496
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3377737561606207e-05,
+      "loss": 0.7188,
+      "step": 11497
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3376654009410297e-05,
+      "loss": 0.7344,
+      "step": 11498
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3375570412465808e-05,
+      "loss": 0.7305,
+      "step": 11499
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3374486770787102e-05,
+      "loss": 0.7109,
+      "step": 11500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.337340308438854e-05,
+      "loss": 0.7646,
+      "step": 11501
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3372319353284484e-05,
+      "loss": 0.7412,
+      "step": 11502
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3371235577489292e-05,
+      "loss": 0.7637,
+      "step": 11503
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3370151757017328e-05,
+      "loss": 0.7559,
+      "step": 11504
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.336906789188296e-05,
+      "loss": 0.7598,
+      "step": 11505
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3367983982100548e-05,
+      "loss": 0.7656,
+      "step": 11506
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3366900027684454e-05,
+      "loss": 0.7275,
+      "step": 11507
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3365816028649045e-05,
+      "loss": 0.7451,
+      "step": 11508
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.336473198500869e-05,
+      "loss": 0.7344,
+      "step": 11509
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.336364789677775e-05,
+      "loss": 0.7539,
+      "step": 11510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3362563763970594e-05,
+      "loss": 0.7246,
+      "step": 11511
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3361479586601592e-05,
+      "loss": 0.7363,
+      "step": 11512
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3360395364685105e-05,
+      "loss": 0.7324,
+      "step": 11513
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3359311098235506e-05,
+      "loss": 0.7529,
+      "step": 11514
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3358226787267166e-05,
+      "loss": 0.751,
+      "step": 11515
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3357142431794448e-05,
+      "loss": 0.7314,
+      "step": 11516
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335605803183173e-05,
+      "loss": 0.707,
+      "step": 11517
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3354973587393376e-05,
+      "loss": 0.75,
+      "step": 11518
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3353889098493765e-05,
+      "loss": 0.7432,
+      "step": 11519
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335280456514726e-05,
+      "loss": 0.7549,
+      "step": 11520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3351719987368241e-05,
+      "loss": 0.7393,
+      "step": 11521
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3350635365171079e-05,
+      "loss": 0.7754,
+      "step": 11522
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3349550698570147e-05,
+      "loss": 0.7324,
+      "step": 11523
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3348465987579817e-05,
+      "loss": 0.7344,
+      "step": 11524
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3347381232214472e-05,
+      "loss": 0.7588,
+      "step": 11525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3346296432488477e-05,
+      "loss": 0.7412,
+      "step": 11526
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3345211588416213e-05,
+      "loss": 0.748,
+      "step": 11527
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.334412670001206e-05,
+      "loss": 0.7568,
+      "step": 11528
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3343041767290392e-05,
+      "loss": 0.7344,
+      "step": 11529
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3341956790265587e-05,
+      "loss": 0.7158,
+      "step": 11530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3340871768952024e-05,
+      "loss": 0.7275,
+      "step": 11531
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3339786703364081e-05,
+      "loss": 0.7539,
+      "step": 11532
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333870159351614e-05,
+      "loss": 0.7422,
+      "step": 11533
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3337616439422576e-05,
+      "loss": 0.7334,
+      "step": 11534
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333653124109778e-05,
+      "loss": 0.7598,
+      "step": 11535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333544599855612e-05,
+      "loss": 0.7412,
+      "step": 11536
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3334360711811989e-05,
+      "loss": 0.7344,
+      "step": 11537
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3333275380879764e-05,
+      "loss": 0.7568,
+      "step": 11538
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333219000577383e-05,
+      "loss": 0.752,
+      "step": 11539
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333110458650857e-05,
+      "loss": 0.752,
+      "step": 11540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3330019123098373e-05,
+      "loss": 0.7461,
+      "step": 11541
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3328933615557615e-05,
+      "loss": 0.751,
+      "step": 11542
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3327848063900688e-05,
+      "loss": 0.7637,
+      "step": 11543
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3326762468141978e-05,
+      "loss": 0.7285,
+      "step": 11544
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3325676828295868e-05,
+      "loss": 0.7461,
+      "step": 11545
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3324591144376749e-05,
+      "loss": 0.7246,
+      "step": 11546
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3323505416399002e-05,
+      "loss": 0.7275,
+      "step": 11547
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3322419644377026e-05,
+      "loss": 0.7363,
+      "step": 11548
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3321333828325204e-05,
+      "loss": 0.7363,
+      "step": 11549
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3320247968257926e-05,
+      "loss": 0.7441,
+      "step": 11550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3319162064189584e-05,
+      "loss": 0.7373,
+      "step": 11551
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3318076116134566e-05,
+      "loss": 0.7637,
+      "step": 11552
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3316990124107263e-05,
+      "loss": 0.7031,
+      "step": 11553
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3315904088122072e-05,
+      "loss": 0.7549,
+      "step": 11554
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3314818008193378e-05,
+      "loss": 0.7539,
+      "step": 11555
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331373188433558e-05,
+      "loss": 0.7129,
+      "step": 11556
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3312645716563068e-05,
+      "loss": 0.7383,
+      "step": 11557
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3311559504890244e-05,
+      "loss": 0.751,
+      "step": 11558
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331047324933149e-05,
+      "loss": 0.7178,
+      "step": 11559
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3309386949901213e-05,
+      "loss": 0.7383,
+      "step": 11560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3308300606613802e-05,
+      "loss": 0.749,
+      "step": 11561
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3307214219483657e-05,
+      "loss": 0.7695,
+      "step": 11562
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3306127788525173e-05,
+      "loss": 0.7217,
+      "step": 11563
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3305041313752753e-05,
+      "loss": 0.7129,
+      "step": 11564
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3303954795180786e-05,
+      "loss": 0.7432,
+      "step": 11565
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.330286823282368e-05,
+      "loss": 0.748,
+      "step": 11566
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3301781626695829e-05,
+      "loss": 0.7139,
+      "step": 11567
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3300694976811637e-05,
+      "loss": 0.7314,
+      "step": 11568
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3299608283185499e-05,
+      "loss": 0.7334,
+      "step": 11569
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3298521545831821e-05,
+      "loss": 0.751,
+      "step": 11570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3297434764765005e-05,
+      "loss": 0.7412,
+      "step": 11571
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.329634793999945e-05,
+      "loss": 0.7129,
+      "step": 11572
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.329526107154956e-05,
+      "loss": 0.7432,
+      "step": 11573
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3294174159429745e-05,
+      "loss": 0.7197,
+      "step": 11574
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3293087203654397e-05,
+      "loss": 0.7139,
+      "step": 11575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3292000204237932e-05,
+      "loss": 0.7275,
+      "step": 11576
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3290913161194748e-05,
+      "loss": 0.7217,
+      "step": 11577
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3289826074539254e-05,
+      "loss": 0.7578,
+      "step": 11578
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3288738944285853e-05,
+      "loss": 0.7363,
+      "step": 11579
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3287651770448962e-05,
+      "loss": 0.7275,
+      "step": 11580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3286564553042976e-05,
+      "loss": 0.7217,
+      "step": 11581
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3285477292082309e-05,
+      "loss": 0.7119,
+      "step": 11582
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3284389987581369e-05,
+      "loss": 0.7207,
+      "step": 11583
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3283302639554569e-05,
+      "loss": 0.7227,
+      "step": 11584
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3282215248016315e-05,
+      "loss": 0.7383,
+      "step": 11585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3281127812981014e-05,
+      "loss": 0.7266,
+      "step": 11586
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280040334463085e-05,
+      "loss": 0.7305,
+      "step": 11587
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3278952812476935e-05,
+      "loss": 0.7363,
+      "step": 11588
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3277865247036979e-05,
+      "loss": 0.7598,
+      "step": 11589
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3276777638157624e-05,
+      "loss": 0.7256,
+      "step": 11590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3275689985853291e-05,
+      "loss": 0.7363,
+      "step": 11591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3274602290138385e-05,
+      "loss": 0.7422,
+      "step": 11592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.327351455102733e-05,
+      "loss": 0.7295,
+      "step": 11593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3272426768534536e-05,
+      "loss": 0.748,
+      "step": 11594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.327133894267442e-05,
+      "loss": 0.7295,
+      "step": 11595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3270251073461397e-05,
+      "loss": 0.7158,
+      "step": 11596
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3269163160909885e-05,
+      "loss": 0.7441,
+      "step": 11597
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3268075205034299e-05,
+      "loss": 0.7617,
+      "step": 11598
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3266987205849062e-05,
+      "loss": 0.7637,
+      "step": 11599
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3265899163368588e-05,
+      "loss": 0.7412,
+      "step": 11600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3264811077607299e-05,
+      "loss": 0.7363,
+      "step": 11601
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3263722948579607e-05,
+      "loss": 0.7412,
+      "step": 11602
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3262634776299947e-05,
+      "loss": 0.7354,
+      "step": 11603
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3261546560782726e-05,
+      "loss": 0.7314,
+      "step": 11604
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3260458302042372e-05,
+      "loss": 0.7441,
+      "step": 11605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3259370000093306e-05,
+      "loss": 0.7598,
+      "step": 11606
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3258281654949949e-05,
+      "loss": 0.7334,
+      "step": 11607
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3257193266626724e-05,
+      "loss": 0.7451,
+      "step": 11608
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3256104835138061e-05,
+      "loss": 0.7559,
+      "step": 11609
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3255016360498374e-05,
+      "loss": 0.752,
+      "step": 11610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3253927842722096e-05,
+      "loss": 0.7188,
+      "step": 11611
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3252839281823648e-05,
+      "loss": 0.7549,
+      "step": 11612
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3251750677817461e-05,
+      "loss": 0.7344,
+      "step": 11613
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3250662030717956e-05,
+      "loss": 0.7422,
+      "step": 11614
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3249573340539561e-05,
+      "loss": 0.7393,
+      "step": 11615
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3248484607296706e-05,
+      "loss": 0.752,
+      "step": 11616
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.324739583100382e-05,
+      "loss": 0.751,
+      "step": 11617
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3246307011675326e-05,
+      "loss": 0.7422,
+      "step": 11618
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.324521814932566e-05,
+      "loss": 0.7334,
+      "step": 11619
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3244129243969248e-05,
+      "loss": 0.7373,
+      "step": 11620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3243040295620524e-05,
+      "loss": 0.7266,
+      "step": 11621
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3241951304293917e-05,
+      "loss": 0.75,
+      "step": 11622
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3240862270003858e-05,
+      "loss": 0.7441,
+      "step": 11623
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3239773192764778e-05,
+      "loss": 0.7402,
+      "step": 11624
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3238684072591115e-05,
+      "loss": 0.7461,
+      "step": 11625
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3237594909497297e-05,
+      "loss": 0.7158,
+      "step": 11626
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3236505703497763e-05,
+      "loss": 0.7471,
+      "step": 11627
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3235416454606942e-05,
+      "loss": 0.7461,
+      "step": 11628
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3234327162839274e-05,
+      "loss": 0.7119,
+      "step": 11629
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3233237828209192e-05,
+      "loss": 0.7393,
+      "step": 11630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3232148450731128e-05,
+      "loss": 0.7402,
+      "step": 11631
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3231059030419529e-05,
+      "loss": 0.7334,
+      "step": 11632
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3229969567288824e-05,
+      "loss": 0.7637,
+      "step": 11633
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3228880061353458e-05,
+      "loss": 0.7393,
+      "step": 11634
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3227790512627863e-05,
+      "loss": 0.7324,
+      "step": 11635
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3226700921126478e-05,
+      "loss": 0.7559,
+      "step": 11636
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3225611286863746e-05,
+      "loss": 0.748,
+      "step": 11637
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.322452160985411e-05,
+      "loss": 0.7344,
+      "step": 11638
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3223431890112002e-05,
+      "loss": 0.7422,
+      "step": 11639
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3222342127651872e-05,
+      "loss": 0.7314,
+      "step": 11640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3221252322488156e-05,
+      "loss": 0.7607,
+      "step": 11641
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3220162474635304e-05,
+      "loss": 0.7256,
+      "step": 11642
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3219072584107746e-05,
+      "loss": 0.7451,
+      "step": 11643
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3217982650919936e-05,
+      "loss": 0.7461,
+      "step": 11644
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.321689267508632e-05,
+      "loss": 0.7607,
+      "step": 11645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3215802656621333e-05,
+      "loss": 0.7236,
+      "step": 11646
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3214712595539428e-05,
+      "loss": 0.7393,
+      "step": 11647
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.321362249185505e-05,
+      "loss": 0.7461,
+      "step": 11648
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3212532345582641e-05,
+      "loss": 0.7354,
+      "step": 11649
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3211442156736652e-05,
+      "loss": 0.749,
+      "step": 11650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3210351925331527e-05,
+      "loss": 0.7207,
+      "step": 11651
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3209261651381722e-05,
+      "loss": 0.7734,
+      "step": 11652
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3208171334901675e-05,
+      "loss": 0.752,
+      "step": 11653
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3207080975905842e-05,
+      "loss": 0.7383,
+      "step": 11654
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3205990574408674e-05,
+      "loss": 0.7559,
+      "step": 11655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3204900130424617e-05,
+      "loss": 0.7236,
+      "step": 11656
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320380964396812e-05,
+      "loss": 0.7461,
+      "step": 11657
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3202719115053644e-05,
+      "loss": 0.7373,
+      "step": 11658
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320162854369563e-05,
+      "loss": 0.7451,
+      "step": 11659
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3200537929908541e-05,
+      "loss": 0.7266,
+      "step": 11660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3199447273706823e-05,
+      "loss": 0.7246,
+      "step": 11661
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319835657510493e-05,
+      "loss": 0.7451,
+      "step": 11662
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3197265834117318e-05,
+      "loss": 0.7295,
+      "step": 11663
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3196175050758443e-05,
+      "loss": 0.7451,
+      "step": 11664
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319508422504276e-05,
+      "loss": 0.7324,
+      "step": 11665
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3193993356984724e-05,
+      "loss": 0.7295,
+      "step": 11666
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3192902446598795e-05,
+      "loss": 0.7617,
+      "step": 11667
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3191811493899422e-05,
+      "loss": 0.7676,
+      "step": 11668
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3190720498901068e-05,
+      "loss": 0.7363,
+      "step": 11669
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3189629461618193e-05,
+      "loss": 0.7363,
+      "step": 11670
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3188538382065254e-05,
+      "loss": 0.7617,
+      "step": 11671
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3187447260256711e-05,
+      "loss": 0.7178,
+      "step": 11672
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3186356096207023e-05,
+      "loss": 0.7539,
+      "step": 11673
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3185264889930648e-05,
+      "loss": 0.7656,
+      "step": 11674
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3184173641442053e-05,
+      "loss": 0.751,
+      "step": 11675
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3183082350755691e-05,
+      "loss": 0.7334,
+      "step": 11676
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3181991017886033e-05,
+      "loss": 0.752,
+      "step": 11677
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3180899642847537e-05,
+      "loss": 0.7422,
+      "step": 11678
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3179808225654668e-05,
+      "loss": 0.7324,
+      "step": 11679
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3178716766321888e-05,
+      "loss": 0.7441,
+      "step": 11680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3177625264863668e-05,
+      "loss": 0.751,
+      "step": 11681
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317653372129446e-05,
+      "loss": 0.7295,
+      "step": 11682
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317544213562874e-05,
+      "loss": 0.7129,
+      "step": 11683
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317435050788097e-05,
+      "loss": 0.7402,
+      "step": 11684
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3173258838065621e-05,
+      "loss": 0.7285,
+      "step": 11685
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3172167126197152e-05,
+      "loss": 0.7363,
+      "step": 11686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317107537229004e-05,
+      "loss": 0.7324,
+      "step": 11687
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3169983576358747e-05,
+      "loss": 0.7285,
+      "step": 11688
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3168891738417744e-05,
+      "loss": 0.7314,
+      "step": 11689
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.31677998584815e-05,
+      "loss": 0.7168,
+      "step": 11690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3166707936564487e-05,
+      "loss": 0.7588,
+      "step": 11691
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316561597268117e-05,
+      "loss": 0.7559,
+      "step": 11692
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3164523966846026e-05,
+      "loss": 0.7373,
+      "step": 11693
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3163431919073526e-05,
+      "loss": 0.7334,
+      "step": 11694
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3162339829378139e-05,
+      "loss": 0.7559,
+      "step": 11695
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3161247697774338e-05,
+      "loss": 0.7539,
+      "step": 11696
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.31601555242766e-05,
+      "loss": 0.7529,
+      "step": 11697
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3159063308899393e-05,
+      "loss": 0.7373,
+      "step": 11698
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.31579710516572e-05,
+      "loss": 0.7148,
+      "step": 11699
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3156878752564488e-05,
+      "loss": 0.7354,
+      "step": 11700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3155786411635734e-05,
+      "loss": 0.7324,
+      "step": 11701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3154694028885416e-05,
+      "loss": 0.7588,
+      "step": 11702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3153601604328014e-05,
+      "loss": 0.7275,
+      "step": 11703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3152509137977996e-05,
+      "loss": 0.6934,
+      "step": 11704
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3151416629849849e-05,
+      "loss": 0.7314,
+      "step": 11705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3150324079958046e-05,
+      "loss": 0.7285,
+      "step": 11706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3149231488317067e-05,
+      "loss": 0.75,
+      "step": 11707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314813885494139e-05,
+      "loss": 0.7344,
+      "step": 11708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.31470461798455e-05,
+      "loss": 0.748,
+      "step": 11709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3145953463043872e-05,
+      "loss": 0.7109,
+      "step": 11710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314486070455099e-05,
+      "loss": 0.7383,
+      "step": 11711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3143767904381334e-05,
+      "loss": 0.7295,
+      "step": 11712
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3142675062549385e-05,
+      "loss": 0.749,
+      "step": 11713
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3141582179069632e-05,
+      "loss": 0.7402,
+      "step": 11714
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.314048925395655e-05,
+      "loss": 0.7314,
+      "step": 11715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3139396287224628e-05,
+      "loss": 0.7373,
+      "step": 11716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313830327888835e-05,
+      "loss": 0.7275,
+      "step": 11717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3137210228962199e-05,
+      "loss": 0.7285,
+      "step": 11718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313611713746066e-05,
+      "loss": 0.7744,
+      "step": 11719
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3135024004398225e-05,
+      "loss": 0.7354,
+      "step": 11720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3133930829789369e-05,
+      "loss": 0.7344,
+      "step": 11721
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3132837613648592e-05,
+      "loss": 0.7383,
+      "step": 11722
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3131744355990374e-05,
+      "loss": 0.7275,
+      "step": 11723
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3130651056829206e-05,
+      "loss": 0.7266,
+      "step": 11724
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3129557716179573e-05,
+      "loss": 0.7363,
+      "step": 11725
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3128464334055972e-05,
+      "loss": 0.7227,
+      "step": 11726
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3127370910472883e-05,
+      "loss": 0.7197,
+      "step": 11727
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3126277445444804e-05,
+      "loss": 0.7422,
+      "step": 11728
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3125183938986223e-05,
+      "loss": 0.7275,
+      "step": 11729
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3124090391111632e-05,
+      "loss": 0.7305,
+      "step": 11730
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3122996801835522e-05,
+      "loss": 0.7256,
+      "step": 11731
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3121903171172388e-05,
+      "loss": 0.7266,
+      "step": 11732
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3120809499136722e-05,
+      "loss": 0.7607,
+      "step": 11733
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3119715785743017e-05,
+      "loss": 0.7158,
+      "step": 11734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3118622031005765e-05,
+      "loss": 0.75,
+      "step": 11735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311752823493947e-05,
+      "loss": 0.7422,
+      "step": 11736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3116434397558615e-05,
+      "loss": 0.7461,
+      "step": 11737
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3115340518877703e-05,
+      "loss": 0.7197,
+      "step": 11738
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3114246598911233e-05,
+      "loss": 0.7295,
+      "step": 11739
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3113152637673696e-05,
+      "loss": 0.7227,
+      "step": 11740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311205863517959e-05,
+      "loss": 0.7305,
+      "step": 11741
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3110964591443419e-05,
+      "loss": 0.7227,
+      "step": 11742
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3109870506479674e-05,
+      "loss": 0.7402,
+      "step": 11743
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3108776380302858e-05,
+      "loss": 0.7725,
+      "step": 11744
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3107682212927473e-05,
+      "loss": 0.7383,
+      "step": 11745
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3106588004368016e-05,
+      "loss": 0.7422,
+      "step": 11746
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3105493754638987e-05,
+      "loss": 0.7402,
+      "step": 11747
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.310439946375489e-05,
+      "loss": 0.7314,
+      "step": 11748
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.310330513173023e-05,
+      "loss": 0.7275,
+      "step": 11749
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.31022107585795e-05,
+      "loss": 0.7197,
+      "step": 11750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.310111634431721e-05,
+      "loss": 0.7402,
+      "step": 11751
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3100021888957862e-05,
+      "loss": 0.7217,
+      "step": 11752
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3098927392515959e-05,
+      "loss": 0.7178,
+      "step": 11753
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309783285500601e-05,
+      "loss": 0.7402,
+      "step": 11754
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3096738276442513e-05,
+      "loss": 0.7236,
+      "step": 11755
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3095643656839981e-05,
+      "loss": 0.7246,
+      "step": 11756
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3094548996212914e-05,
+      "loss": 0.7256,
+      "step": 11757
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3093454294575823e-05,
+      "loss": 0.7773,
+      "step": 11758
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3092359551943217e-05,
+      "loss": 0.7627,
+      "step": 11759
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3091264768329595e-05,
+      "loss": 0.7549,
+      "step": 11760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.749,
+      "step": 11761
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3089075078217363e-05,
+      "loss": 0.7422,
+      "step": 11762
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3087980171747768e-05,
+      "loss": 0.7637,
+      "step": 11763
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3086885224355196e-05,
+      "loss": 0.751,
+      "step": 11764
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3085790236054168e-05,
+      "loss": 0.749,
+      "step": 11765
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3084695206859185e-05,
+      "loss": 0.7441,
+      "step": 11766
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3083600136784762e-05,
+      "loss": 0.749,
+      "step": 11767
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3082505025845414e-05,
+      "loss": 0.7412,
+      "step": 11768
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.308140987405565e-05,
+      "loss": 0.7285,
+      "step": 11769
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3080314681429984e-05,
+      "loss": 0.7197,
+      "step": 11770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3079219447982934e-05,
+      "loss": 0.7344,
+      "step": 11771
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307812417372901e-05,
+      "loss": 0.7334,
+      "step": 11772
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3077028858682725e-05,
+      "loss": 0.7256,
+      "step": 11773
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.30759335028586e-05,
+      "loss": 0.7578,
+      "step": 11774
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307483810627115e-05,
+      "loss": 0.7432,
+      "step": 11775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3073742668934888e-05,
+      "loss": 0.749,
+      "step": 11776
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3072647190864336e-05,
+      "loss": 0.7139,
+      "step": 11777
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3071551672074006e-05,
+      "loss": 0.7305,
+      "step": 11778
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307045611257842e-05,
+      "loss": 0.7344,
+      "step": 11779
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3069360512392095e-05,
+      "loss": 0.7529,
+      "step": 11780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3068264871529555e-05,
+      "loss": 0.7412,
+      "step": 11781
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.306716919000531e-05,
+      "loss": 0.7559,
+      "step": 11782
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3066073467833893e-05,
+      "loss": 0.7314,
+      "step": 11783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3064977705029813e-05,
+      "loss": 0.7295,
+      "step": 11784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.30638819016076e-05,
+      "loss": 0.7549,
+      "step": 11785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3062786057581769e-05,
+      "loss": 0.7412,
+      "step": 11786
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.306169017296685e-05,
+      "loss": 0.7314,
+      "step": 11787
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3060594247777361e-05,
+      "loss": 0.7236,
+      "step": 11788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3059498282027825e-05,
+      "loss": 0.7314,
+      "step": 11789
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3058402275732769e-05,
+      "loss": 0.7598,
+      "step": 11790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3057306228906716e-05,
+      "loss": 0.7207,
+      "step": 11791
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3056210141564192e-05,
+      "loss": 0.7305,
+      "step": 11792
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3055114013719723e-05,
+      "loss": 0.7432,
+      "step": 11793
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3054017845387834e-05,
+      "loss": 0.7314,
+      "step": 11794
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3052921636583054e-05,
+      "loss": 0.7451,
+      "step": 11795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3051825387319906e-05,
+      "loss": 0.7041,
+      "step": 11796
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3050729097612922e-05,
+      "loss": 0.7178,
+      "step": 11797
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.304963276747663e-05,
+      "loss": 0.7744,
+      "step": 11798
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3048536396925555e-05,
+      "loss": 0.7422,
+      "step": 11799
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3047439985974233e-05,
+      "loss": 0.7363,
+      "step": 11800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.304634353463719e-05,
+      "loss": 0.7236,
+      "step": 11801
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3045247042928955e-05,
+      "loss": 0.7578,
+      "step": 11802
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3044150510864059e-05,
+      "loss": 0.7539,
+      "step": 11803
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3043053938457044e-05,
+      "loss": 0.7393,
+      "step": 11804
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3041957325722426e-05,
+      "loss": 0.7129,
+      "step": 11805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3040860672674749e-05,
+      "loss": 0.7227,
+      "step": 11806
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3039763979328544e-05,
+      "loss": 0.7549,
+      "step": 11807
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3038667245698341e-05,
+      "loss": 0.7129,
+      "step": 11808
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3037570471798678e-05,
+      "loss": 0.7334,
+      "step": 11809
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3036473657644089e-05,
+      "loss": 0.7549,
+      "step": 11810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3035376803249106e-05,
+      "loss": 0.752,
+      "step": 11811
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3034279908628271e-05,
+      "loss": 0.7373,
+      "step": 11812
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3033182973796117e-05,
+      "loss": 0.7236,
+      "step": 11813
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3032085998767183e-05,
+      "loss": 0.7295,
+      "step": 11814
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3030988983556002e-05,
+      "loss": 0.7148,
+      "step": 11815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3029891928177114e-05,
+      "loss": 0.7266,
+      "step": 11816
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3028794832645063e-05,
+      "loss": 0.7305,
+      "step": 11817
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302769769697438e-05,
+      "loss": 0.7852,
+      "step": 11818
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3026600521179606e-05,
+      "loss": 0.7461,
+      "step": 11819
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302550330527529e-05,
+      "loss": 0.7461,
+      "step": 11820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3024406049275959e-05,
+      "loss": 0.7285,
+      "step": 11821
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3023308753196164e-05,
+      "loss": 0.7324,
+      "step": 11822
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3022211417050445e-05,
+      "loss": 0.7295,
+      "step": 11823
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3021114040853342e-05,
+      "loss": 0.7422,
+      "step": 11824
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3020016624619398e-05,
+      "loss": 0.7188,
+      "step": 11825
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3018919168363157e-05,
+      "loss": 0.7539,
+      "step": 11826
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3017821672099165e-05,
+      "loss": 0.7471,
+      "step": 11827
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3016724135841965e-05,
+      "loss": 0.7363,
+      "step": 11828
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.30156265596061e-05,
+      "loss": 0.7227,
+      "step": 11829
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301452894340612e-05,
+      "loss": 0.7344,
+      "step": 11830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3013431287256562e-05,
+      "loss": 0.7441,
+      "step": 11831
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3012333591171981e-05,
+      "loss": 0.7422,
+      "step": 11832
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3011235855166926e-05,
+      "loss": 0.7344,
+      "step": 11833
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3010138079255936e-05,
+      "loss": 0.7266,
+      "step": 11834
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3009040263453564e-05,
+      "loss": 0.75,
+      "step": 11835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3007942407774357e-05,
+      "loss": 0.7334,
+      "step": 11836
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3006844512232864e-05,
+      "loss": 0.748,
+      "step": 11837
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3005746576843636e-05,
+      "loss": 0.7246,
+      "step": 11838
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3004648601621226e-05,
+      "loss": 0.7305,
+      "step": 11839
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3003550586580177e-05,
+      "loss": 0.751,
+      "step": 11840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.300245253173505e-05,
+      "loss": 0.7559,
+      "step": 11841
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3001354437100386e-05,
+      "loss": 0.7129,
+      "step": 11842
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.300025630269075e-05,
+      "loss": 0.7412,
+      "step": 11843
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2999158128520684e-05,
+      "loss": 0.7461,
+      "step": 11844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2998059914604744e-05,
+      "loss": 0.7441,
+      "step": 11845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2996961660957487e-05,
+      "loss": 0.7354,
+      "step": 11846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2995863367593464e-05,
+      "loss": 0.7471,
+      "step": 11847
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2994765034527232e-05,
+      "loss": 0.7578,
+      "step": 11848
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2993666661773348e-05,
+      "loss": 0.7314,
+      "step": 11849
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2992568249346365e-05,
+      "loss": 0.7158,
+      "step": 11850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2991469797260838e-05,
+      "loss": 0.7178,
+      "step": 11851
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2990371305531329e-05,
+      "loss": 0.7354,
+      "step": 11852
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2989272774172394e-05,
+      "loss": 0.7588,
+      "step": 11853
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2988174203198589e-05,
+      "loss": 0.7148,
+      "step": 11854
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2987075592624473e-05,
+      "loss": 0.7363,
+      "step": 11855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2985976942464609e-05,
+      "loss": 0.7354,
+      "step": 11856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2984878252733552e-05,
+      "loss": 0.7422,
+      "step": 11857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2983779523445863e-05,
+      "loss": 0.7529,
+      "step": 11858
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2982680754616109e-05,
+      "loss": 0.749,
+      "step": 11859
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.298158194625884e-05,
+      "loss": 0.7578,
+      "step": 11860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2980483098388627e-05,
+      "loss": 0.7207,
+      "step": 11861
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.297938421102003e-05,
+      "loss": 0.707,
+      "step": 11862
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2978285284167609e-05,
+      "loss": 0.7471,
+      "step": 11863
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2977186317845927e-05,
+      "loss": 0.7451,
+      "step": 11864
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2976087312069554e-05,
+      "loss": 0.7344,
+      "step": 11865
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2974988266853048e-05,
+      "loss": 0.7402,
+      "step": 11866
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.297388918221098e-05,
+      "loss": 0.7119,
+      "step": 11867
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2972790058157907e-05,
+      "loss": 0.7471,
+      "step": 11868
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2971690894708406e-05,
+      "loss": 0.751,
+      "step": 11869
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2970591691877033e-05,
+      "loss": 0.7236,
+      "step": 11870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296949244967836e-05,
+      "loss": 0.7432,
+      "step": 11871
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2968393168126957e-05,
+      "loss": 0.7529,
+      "step": 11872
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2967293847237388e-05,
+      "loss": 0.7598,
+      "step": 11873
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2966194487024222e-05,
+      "loss": 0.7656,
+      "step": 11874
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2965095087502029e-05,
+      "loss": 0.7285,
+      "step": 11875
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2963995648685374e-05,
+      "loss": 0.7598,
+      "step": 11876
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2962896170588837e-05,
+      "loss": 0.7471,
+      "step": 11877
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2961796653226982e-05,
+      "loss": 0.7188,
+      "step": 11878
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296069709661438e-05,
+      "loss": 0.7275,
+      "step": 11879
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2959597500765606e-05,
+      "loss": 0.708,
+      "step": 11880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2958497865695228e-05,
+      "loss": 0.7451,
+      "step": 11881
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2957398191417823e-05,
+      "loss": 0.7402,
+      "step": 11882
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2956298477947959e-05,
+      "loss": 0.7148,
+      "step": 11883
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2955198725300217e-05,
+      "loss": 0.7285,
+      "step": 11884
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2954098933489165e-05,
+      "loss": 0.7344,
+      "step": 11885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2952999102529383e-05,
+      "loss": 0.7295,
+      "step": 11886
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2951899232435439e-05,
+      "loss": 0.7246,
+      "step": 11887
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2950799323221918e-05,
+      "loss": 0.71,
+      "step": 11888
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2949699374903389e-05,
+      "loss": 0.7363,
+      "step": 11889
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2948599387494431e-05,
+      "loss": 0.7266,
+      "step": 11890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2947499361009621e-05,
+      "loss": 0.7441,
+      "step": 11891
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2946399295463543e-05,
+      "loss": 0.7383,
+      "step": 11892
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2945299190870765e-05,
+      "loss": 0.7412,
+      "step": 11893
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2944199047245876e-05,
+      "loss": 0.7275,
+      "step": 11894
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2943098864603447e-05,
+      "loss": 0.7314,
+      "step": 11895
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2941998642958065e-05,
+      "loss": 0.7529,
+      "step": 11896
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2940898382324302e-05,
+      "loss": 0.748,
+      "step": 11897
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2939798082716752e-05,
+      "loss": 0.7109,
+      "step": 11898
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2938697744149985e-05,
+      "loss": 0.7227,
+      "step": 11899
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2937597366638585e-05,
+      "loss": 0.7256,
+      "step": 11900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2936496950197142e-05,
+      "loss": 0.7373,
+      "step": 11901
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2935396494840231e-05,
+      "loss": 0.7285,
+      "step": 11902
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2934296000582437e-05,
+      "loss": 0.7334,
+      "step": 11903
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2933195467438344e-05,
+      "loss": 0.7363,
+      "step": 11904
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2932094895422542e-05,
+      "loss": 0.7256,
+      "step": 11905
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2930994284549611e-05,
+      "loss": 0.7422,
+      "step": 11906
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2929893634834137e-05,
+      "loss": 0.7305,
+      "step": 11907
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.292879294629071e-05,
+      "loss": 0.7549,
+      "step": 11908
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2927692218933907e-05,
+      "loss": 0.7383,
+      "step": 11909
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2926591452778328e-05,
+      "loss": 0.7627,
+      "step": 11910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2925490647838555e-05,
+      "loss": 0.7305,
+      "step": 11911
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2924389804129174e-05,
+      "loss": 0.7275,
+      "step": 11912
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2923288921664772e-05,
+      "loss": 0.7432,
+      "step": 11913
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.292218800045995e-05,
+      "loss": 0.7422,
+      "step": 11914
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2921087040529283e-05,
+      "loss": 0.7383,
+      "step": 11915
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2919986041887372e-05,
+      "loss": 0.7715,
+      "step": 11916
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2918885004548802e-05,
+      "loss": 0.7373,
+      "step": 11917
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2917783928528168e-05,
+      "loss": 0.751,
+      "step": 11918
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2916682813840057e-05,
+      "loss": 0.7266,
+      "step": 11919
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2915581660499065e-05,
+      "loss": 0.7295,
+      "step": 11920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2914480468519788e-05,
+      "loss": 0.7461,
+      "step": 11921
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2913379237916811e-05,
+      "loss": 0.7402,
+      "step": 11922
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2912277968704736e-05,
+      "loss": 0.7451,
+      "step": 11923
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2911176660898152e-05,
+      "loss": 0.7227,
+      "step": 11924
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2910075314511658e-05,
+      "loss": 0.7158,
+      "step": 11925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2908973929559844e-05,
+      "loss": 0.7178,
+      "step": 11926
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2907872506057316e-05,
+      "loss": 0.7256,
+      "step": 11927
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2906771044018656e-05,
+      "loss": 0.7354,
+      "step": 11928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.290566954345847e-05,
+      "loss": 0.7344,
+      "step": 11929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2904568004391359e-05,
+      "loss": 0.7314,
+      "step": 11930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2903466426831913e-05,
+      "loss": 0.7129,
+      "step": 11931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.290236481079473e-05,
+      "loss": 0.7109,
+      "step": 11932
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2901263156294414e-05,
+      "loss": 0.7578,
+      "step": 11933
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2900161463345566e-05,
+      "loss": 0.7285,
+      "step": 11934
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2899059731962782e-05,
+      "loss": 0.7256,
+      "step": 11935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.289795796216066e-05,
+      "loss": 0.7354,
+      "step": 11936
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2896856153953811e-05,
+      "loss": 0.7178,
+      "step": 11937
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2895754307356825e-05,
+      "loss": 0.749,
+      "step": 11938
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.289465242238431e-05,
+      "loss": 0.751,
+      "step": 11939
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.289355049905087e-05,
+      "loss": 0.7207,
+      "step": 11940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2892448537371105e-05,
+      "loss": 0.7344,
+      "step": 11941
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2891346537359617e-05,
+      "loss": 0.7607,
+      "step": 11942
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2890244499031017e-05,
+      "loss": 0.7266,
+      "step": 11943
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2889142422399901e-05,
+      "loss": 0.7363,
+      "step": 11944
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288804030748088e-05,
+      "loss": 0.7412,
+      "step": 11945
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2886938154288555e-05,
+      "loss": 0.7158,
+      "step": 11946
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2885835962837539e-05,
+      "loss": 0.7266,
+      "step": 11947
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2884733733142431e-05,
+      "loss": 0.7422,
+      "step": 11948
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2883631465217843e-05,
+      "loss": 0.7266,
+      "step": 11949
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2882529159078381e-05,
+      "loss": 0.7256,
+      "step": 11950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2881426814738655e-05,
+      "loss": 0.7275,
+      "step": 11951
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288032443221327e-05,
+      "loss": 0.7148,
+      "step": 11952
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2879222011516838e-05,
+      "loss": 0.7412,
+      "step": 11953
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2878119552663968e-05,
+      "loss": 0.7607,
+      "step": 11954
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2877017055669269e-05,
+      "loss": 0.7363,
+      "step": 11955
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2875914520547355e-05,
+      "loss": 0.7236,
+      "step": 11956
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2874811947312833e-05,
+      "loss": 0.7422,
+      "step": 11957
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2873709335980318e-05,
+      "loss": 0.7402,
+      "step": 11958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2872606686564418e-05,
+      "loss": 0.7373,
+      "step": 11959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2871503999079749e-05,
+      "loss": 0.7236,
+      "step": 11960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2870401273540924e-05,
+      "loss": 0.749,
+      "step": 11961
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2869298509962557e-05,
+      "loss": 0.7598,
+      "step": 11962
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.286819570835926e-05,
+      "loss": 0.7236,
+      "step": 11963
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2867092868745652e-05,
+      "loss": 0.7227,
+      "step": 11964
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2865989991136343e-05,
+      "loss": 0.7236,
+      "step": 11965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2864887075545953e-05,
+      "loss": 0.7344,
+      "step": 11966
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2863784121989094e-05,
+      "loss": 0.709,
+      "step": 11967
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2862681130480388e-05,
+      "loss": 0.7256,
+      "step": 11968
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2861578101034446e-05,
+      "loss": 0.7461,
+      "step": 11969
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2860475033665892e-05,
+      "loss": 0.7412,
+      "step": 11970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2859371928389337e-05,
+      "loss": 0.7119,
+      "step": 11971
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2858268785219408e-05,
+      "loss": 0.7471,
+      "step": 11972
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2857165604170717e-05,
+      "loss": 0.7197,
+      "step": 11973
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2856062385257891e-05,
+      "loss": 0.7363,
+      "step": 11974
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2854959128495541e-05,
+      "loss": 0.7344,
+      "step": 11975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2853855833898298e-05,
+      "loss": 0.7207,
+      "step": 11976
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2852752501480775e-05,
+      "loss": 0.7188,
+      "step": 11977
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2851649131257597e-05,
+      "loss": 0.7207,
+      "step": 11978
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2850545723243388e-05,
+      "loss": 0.7207,
+      "step": 11979
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2849442277452768e-05,
+      "loss": 0.7422,
+      "step": 11980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2848338793900359e-05,
+      "loss": 0.7393,
+      "step": 11981
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2847235272600792e-05,
+      "loss": 0.7383,
+      "step": 11982
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2846131713568682e-05,
+      "loss": 0.7412,
+      "step": 11983
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2845028116818657e-05,
+      "loss": 0.7432,
+      "step": 11984
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2843924482365343e-05,
+      "loss": 0.7324,
+      "step": 11985
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2842820810223369e-05,
+      "loss": 0.7383,
+      "step": 11986
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2841717100407354e-05,
+      "loss": 0.7402,
+      "step": 11987
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2840613352931932e-05,
+      "loss": 0.7539,
+      "step": 11988
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2839509567811725e-05,
+      "loss": 0.7334,
+      "step": 11989
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2838405745061366e-05,
+      "loss": 0.7354,
+      "step": 11990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2837301884695477e-05,
+      "loss": 0.7119,
+      "step": 11991
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2836197986728692e-05,
+      "loss": 0.7305,
+      "step": 11992
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2835094051175636e-05,
+      "loss": 0.7559,
+      "step": 11993
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2833990078050942e-05,
+      "loss": 0.7637,
+      "step": 11994
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2832886067369237e-05,
+      "loss": 0.7451,
+      "step": 11995
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2831782019145156e-05,
+      "loss": 0.7236,
+      "step": 11996
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2830677933393326e-05,
+      "loss": 0.7441,
+      "step": 11997
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.282957381012838e-05,
+      "loss": 0.7314,
+      "step": 11998
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2828469649364954e-05,
+      "loss": 0.7529,
+      "step": 11999
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2827365451117675e-05,
+      "loss": 0.7354,
+      "step": 12000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.282626121540118e-05,
+      "loss": 0.7256,
+      "step": 12001
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2825156942230101e-05,
+      "loss": 0.7246,
+      "step": 12002
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2824052631619072e-05,
+      "loss": 0.7422,
+      "step": 12003
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2822948283582728e-05,
+      "loss": 0.7773,
+      "step": 12004
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2821843898135706e-05,
+      "loss": 0.7305,
+      "step": 12005
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.282073947529264e-05,
+      "loss": 0.7227,
+      "step": 12006
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2819635015068168e-05,
+      "loss": 0.7295,
+      "step": 12007
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2818530517476923e-05,
+      "loss": 0.7412,
+      "step": 12008
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2817425982533545e-05,
+      "loss": 0.7295,
+      "step": 12009
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.281632141025267e-05,
+      "loss": 0.7539,
+      "step": 12010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2815216800648938e-05,
+      "loss": 0.7402,
+      "step": 12011
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.281411215373699e-05,
+      "loss": 0.7314,
+      "step": 12012
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2813007469531455e-05,
+      "loss": 0.7129,
+      "step": 12013
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2811902748046982e-05,
+      "loss": 0.7129,
+      "step": 12014
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2810797989298214e-05,
+      "loss": 0.7129,
+      "step": 12015
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280969319329978e-05,
+      "loss": 0.7461,
+      "step": 12016
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2808588360066332e-05,
+      "loss": 0.7334,
+      "step": 12017
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2807483489612507e-05,
+      "loss": 0.7461,
+      "step": 12018
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2806378581952946e-05,
+      "loss": 0.7393,
+      "step": 12019
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2805273637102291e-05,
+      "loss": 0.7324,
+      "step": 12020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280416865507519e-05,
+      "loss": 0.7285,
+      "step": 12021
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2803063635886282e-05,
+      "loss": 0.7354,
+      "step": 12022
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2801958579550212e-05,
+      "loss": 0.7148,
+      "step": 12023
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280085348608163e-05,
+      "loss": 0.7158,
+      "step": 12024
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2799748355495171e-05,
+      "loss": 0.7295,
+      "step": 12025
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.279864318780549e-05,
+      "loss": 0.7373,
+      "step": 12026
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2797537983027226e-05,
+      "loss": 0.7598,
+      "step": 12027
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2796432741175031e-05,
+      "loss": 0.7344,
+      "step": 12028
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2795327462263547e-05,
+      "loss": 0.7305,
+      "step": 12029
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2794222146307424e-05,
+      "loss": 0.7129,
+      "step": 12030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2793116793321314e-05,
+      "loss": 0.7305,
+      "step": 12031
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.279201140331986e-05,
+      "loss": 0.752,
+      "step": 12032
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.279090597631771e-05,
+      "loss": 0.7148,
+      "step": 12033
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.278980051232952e-05,
+      "loss": 0.7363,
+      "step": 12034
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2788695011369936e-05,
+      "loss": 0.7773,
+      "step": 12035
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2787589473453605e-05,
+      "loss": 0.748,
+      "step": 12036
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2786483898595187e-05,
+      "loss": 0.7627,
+      "step": 12037
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2785378286809325e-05,
+      "loss": 0.7344,
+      "step": 12038
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2784272638110672e-05,
+      "loss": 0.7588,
+      "step": 12039
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2783166952513888e-05,
+      "loss": 0.7549,
+      "step": 12040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2782061230033616e-05,
+      "loss": 0.7354,
+      "step": 12041
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2780955470684515e-05,
+      "loss": 0.7793,
+      "step": 12042
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2779849674481238e-05,
+      "loss": 0.7559,
+      "step": 12043
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277874384143844e-05,
+      "loss": 0.7549,
+      "step": 12044
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2777637971570774e-05,
+      "loss": 0.75,
+      "step": 12045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2776532064892896e-05,
+      "loss": 0.7461,
+      "step": 12046
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2775426121419463e-05,
+      "loss": 0.7451,
+      "step": 12047
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277432014116513e-05,
+      "loss": 0.7266,
+      "step": 12048
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2773214124144552e-05,
+      "loss": 0.7549,
+      "step": 12049
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2772108070372393e-05,
+      "loss": 0.7588,
+      "step": 12050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2771001979863305e-05,
+      "loss": 0.75,
+      "step": 12051
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2769895852631948e-05,
+      "loss": 0.7324,
+      "step": 12052
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.276878968869298e-05,
+      "loss": 0.7432,
+      "step": 12053
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2767683488061065e-05,
+      "loss": 0.7422,
+      "step": 12054
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.276657725075085e-05,
+      "loss": 0.7402,
+      "step": 12055
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.276547097677701e-05,
+      "loss": 0.7354,
+      "step": 12056
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2764364666154199e-05,
+      "loss": 0.7402,
+      "step": 12057
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2763258318897078e-05,
+      "loss": 0.7383,
+      "step": 12058
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2762151935020308e-05,
+      "loss": 0.7441,
+      "step": 12059
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2761045514538556e-05,
+      "loss": 0.7617,
+      "step": 12060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2759939057466479e-05,
+      "loss": 0.7568,
+      "step": 12061
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2758832563818743e-05,
+      "loss": 0.7168,
+      "step": 12062
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2757726033610013e-05,
+      "loss": 0.7471,
+      "step": 12063
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.275661946685495e-05,
+      "loss": 0.7373,
+      "step": 12064
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2755512863568216e-05,
+      "loss": 0.7158,
+      "step": 12065
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2754406223764484e-05,
+      "loss": 0.7246,
+      "step": 12066
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2753299547458413e-05,
+      "loss": 0.75,
+      "step": 12067
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2752192834664677e-05,
+      "loss": 0.7178,
+      "step": 12068
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.275108608539793e-05,
+      "loss": 0.7129,
+      "step": 12069
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.274997929967285e-05,
+      "loss": 0.7217,
+      "step": 12070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27488724775041e-05,
+      "loss": 0.7305,
+      "step": 12071
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2747765618906347e-05,
+      "loss": 0.7471,
+      "step": 12072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2746658723894263e-05,
+      "loss": 0.7354,
+      "step": 12073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2745551792482516e-05,
+      "loss": 0.7266,
+      "step": 12074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2744444824685772e-05,
+      "loss": 0.7334,
+      "step": 12075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2743337820518707e-05,
+      "loss": 0.7354,
+      "step": 12076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2742230779995982e-05,
+      "loss": 0.7061,
+      "step": 12077
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2741123703132279e-05,
+      "loss": 0.7549,
+      "step": 12078
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2740016589942261e-05,
+      "loss": 0.7344,
+      "step": 12079
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2738909440440602e-05,
+      "loss": 0.7715,
+      "step": 12080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2737802254641976e-05,
+      "loss": 0.7275,
+      "step": 12081
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2736695032561057e-05,
+      "loss": 0.7344,
+      "step": 12082
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2735587774212515e-05,
+      "loss": 0.7354,
+      "step": 12083
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2734480479611024e-05,
+      "loss": 0.708,
+      "step": 12084
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2733373148771261e-05,
+      "loss": 0.7188,
+      "step": 12085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27322657817079e-05,
+      "loss": 0.7266,
+      "step": 12086
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2731158378435613e-05,
+      "loss": 0.7324,
+      "step": 12087
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2730050938969077e-05,
+      "loss": 0.7188,
+      "step": 12088
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2728943463322972e-05,
+      "loss": 0.7432,
+      "step": 12089
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2727835951511969e-05,
+      "loss": 0.7256,
+      "step": 12090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272672840355075e-05,
+      "loss": 0.7373,
+      "step": 12091
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2725620819453988e-05,
+      "loss": 0.7275,
+      "step": 12092
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2724513199236368e-05,
+      "loss": 0.7275,
+      "step": 12093
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2723405542912559e-05,
+      "loss": 0.7305,
+      "step": 12094
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2722297850497248e-05,
+      "loss": 0.7305,
+      "step": 12095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2721190122005112e-05,
+      "loss": 0.7402,
+      "step": 12096
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2720082357450829e-05,
+      "loss": 0.7607,
+      "step": 12097
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.271897455684908e-05,
+      "loss": 0.6982,
+      "step": 12098
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2717866720214552e-05,
+      "loss": 0.7197,
+      "step": 12099
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2716758847561916e-05,
+      "loss": 0.7578,
+      "step": 12100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2715650938905863e-05,
+      "loss": 0.7217,
+      "step": 12101
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.271454299426107e-05,
+      "loss": 0.7334,
+      "step": 12102
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2713435013642223e-05,
+      "loss": 0.7422,
+      "step": 12103
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2712326997064002e-05,
+      "loss": 0.7207,
+      "step": 12104
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2711218944541095e-05,
+      "loss": 0.71,
+      "step": 12105
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2710110856088183e-05,
+      "loss": 0.7441,
+      "step": 12106
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2709002731719952e-05,
+      "loss": 0.7041,
+      "step": 12107
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2707894571451087e-05,
+      "loss": 0.7656,
+      "step": 12108
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2706786375296275e-05,
+      "loss": 0.7461,
+      "step": 12109
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27056781432702e-05,
+      "loss": 0.7588,
+      "step": 12110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2704569875387548e-05,
+      "loss": 0.7158,
+      "step": 12111
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2703461571663012e-05,
+      "loss": 0.7451,
+      "step": 12112
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2702353232111273e-05,
+      "loss": 0.7734,
+      "step": 12113
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2701244856747021e-05,
+      "loss": 0.748,
+      "step": 12114
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2700136445584947e-05,
+      "loss": 0.7656,
+      "step": 12115
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2699027998639738e-05,
+      "loss": 0.7529,
+      "step": 12116
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2697919515926082e-05,
+      "loss": 0.7402,
+      "step": 12117
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2696810997458673e-05,
+      "loss": 0.7559,
+      "step": 12118
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2695702443252197e-05,
+      "loss": 0.7256,
+      "step": 12119
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2694593853321347e-05,
+      "loss": 0.7256,
+      "step": 12120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2693485227680815e-05,
+      "loss": 0.7363,
+      "step": 12121
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2692376566345295e-05,
+      "loss": 0.7197,
+      "step": 12122
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2691267869329473e-05,
+      "loss": 0.7461,
+      "step": 12123
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2690159136648048e-05,
+      "loss": 0.7207,
+      "step": 12124
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2689050368315713e-05,
+      "loss": 0.7275,
+      "step": 12125
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2687941564347154e-05,
+      "loss": 0.7119,
+      "step": 12126
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2686832724757075e-05,
+      "loss": 0.7119,
+      "step": 12127
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2685723849560166e-05,
+      "loss": 0.7334,
+      "step": 12128
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2684614938771122e-05,
+      "loss": 0.7461,
+      "step": 12129
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2683505992404642e-05,
+      "loss": 0.7227,
+      "step": 12130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2682397010475418e-05,
+      "loss": 0.7461,
+      "step": 12131
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2681287992998148e-05,
+      "loss": 0.7363,
+      "step": 12132
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2680178939987528e-05,
+      "loss": 0.7188,
+      "step": 12133
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267906985145826e-05,
+      "loss": 0.7412,
+      "step": 12134
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2677960727425036e-05,
+      "loss": 0.7441,
+      "step": 12135
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.267685156790256e-05,
+      "loss": 0.7334,
+      "step": 12136
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2675742372905524e-05,
+      "loss": 0.7324,
+      "step": 12137
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2674633142448639e-05,
+      "loss": 0.7207,
+      "step": 12138
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2673523876546592e-05,
+      "loss": 0.7656,
+      "step": 12139
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2672414575214092e-05,
+      "loss": 0.7236,
+      "step": 12140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2671305238465835e-05,
+      "loss": 0.7188,
+      "step": 12141
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2670195866316525e-05,
+      "loss": 0.71,
+      "step": 12142
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2669086458780861e-05,
+      "loss": 0.7461,
+      "step": 12143
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2667977015873552e-05,
+      "loss": 0.7314,
+      "step": 12144
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2666867537609291e-05,
+      "loss": 0.7549,
+      "step": 12145
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.266575802400279e-05,
+      "loss": 0.7266,
+      "step": 12146
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2664648475068746e-05,
+      "loss": 0.7354,
+      "step": 12147
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.266353889082187e-05,
+      "loss": 0.7539,
+      "step": 12148
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2662429271276857e-05,
+      "loss": 0.7188,
+      "step": 12149
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2661319616448421e-05,
+      "loss": 0.7227,
+      "step": 12150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2660209926351263e-05,
+      "loss": 0.75,
+      "step": 12151
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2659100201000092e-05,
+      "loss": 0.751,
+      "step": 12152
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2657990440409611e-05,
+      "loss": 0.7354,
+      "step": 12153
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2656880644594529e-05,
+      "loss": 0.7275,
+      "step": 12154
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2655770813569552e-05,
+      "loss": 0.7363,
+      "step": 12155
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265466094734939e-05,
+      "loss": 0.7402,
+      "step": 12156
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265355104594875e-05,
+      "loss": 0.7598,
+      "step": 12157
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2652441109382341e-05,
+      "loss": 0.7129,
+      "step": 12158
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265133113766487e-05,
+      "loss": 0.7305,
+      "step": 12159
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265022113081105e-05,
+      "loss": 0.7207,
+      "step": 12160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2649111088835593e-05,
+      "loss": 0.7363,
+      "step": 12161
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2648001011753203e-05,
+      "loss": 0.7119,
+      "step": 12162
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2646890899578599e-05,
+      "loss": 0.7607,
+      "step": 12163
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2645780752326487e-05,
+      "loss": 0.7314,
+      "step": 12164
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2644670570011576e-05,
+      "loss": 0.7363,
+      "step": 12165
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2643560352648587e-05,
+      "loss": 0.7588,
+      "step": 12166
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2642450100252229e-05,
+      "loss": 0.748,
+      "step": 12167
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2641339812837216e-05,
+      "loss": 0.7227,
+      "step": 12168
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2640229490418259e-05,
+      "loss": 0.7441,
+      "step": 12169
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2639119133010076e-05,
+      "loss": 0.7197,
+      "step": 12170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2638008740627381e-05,
+      "loss": 0.7305,
+      "step": 12171
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2636898313284887e-05,
+      "loss": 0.7227,
+      "step": 12172
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2635787850997314e-05,
+      "loss": 0.7666,
+      "step": 12173
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2634677353779377e-05,
+      "loss": 0.7295,
+      "step": 12174
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2633566821645788e-05,
+      "loss": 0.7256,
+      "step": 12175
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2632456254611268e-05,
+      "loss": 0.7412,
+      "step": 12176
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2631345652690538e-05,
+      "loss": 0.7412,
+      "step": 12177
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2630235015898308e-05,
+      "loss": 0.7402,
+      "step": 12178
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2629124344249304e-05,
+      "loss": 0.7383,
+      "step": 12179
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.262801363775824e-05,
+      "loss": 0.7402,
+      "step": 12180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2626902896439838e-05,
+      "loss": 0.7295,
+      "step": 12181
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2625792120308818e-05,
+      "loss": 0.75,
+      "step": 12182
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.26246813093799e-05,
+      "loss": 0.7627,
+      "step": 12183
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2623570463667802e-05,
+      "loss": 0.7549,
+      "step": 12184
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.262245958318725e-05,
+      "loss": 0.7441,
+      "step": 12185
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2621348667952962e-05,
+      "loss": 0.7178,
+      "step": 12186
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2620237717979666e-05,
+      "loss": 0.7373,
+      "step": 12187
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2619126733282076e-05,
+      "loss": 0.749,
+      "step": 12188
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2618015713874922e-05,
+      "loss": 0.7197,
+      "step": 12189
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2616904659772926e-05,
+      "loss": 0.7529,
+      "step": 12190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2615793570990808e-05,
+      "loss": 0.7588,
+      "step": 12191
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2614682447543299e-05,
+      "loss": 0.7314,
+      "step": 12192
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2613571289445122e-05,
+      "loss": 0.7295,
+      "step": 12193
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2612460096710997e-05,
+      "loss": 0.7227,
+      "step": 12194
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2611348869355657e-05,
+      "loss": 0.7412,
+      "step": 12195
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2610237607393824e-05,
+      "loss": 0.7383,
+      "step": 12196
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.260912631084023e-05,
+      "loss": 0.7471,
+      "step": 12197
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2608014979709595e-05,
+      "loss": 0.7295,
+      "step": 12198
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2606903614016653e-05,
+      "loss": 0.7744,
+      "step": 12199
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2605792213776127e-05,
+      "loss": 0.7285,
+      "step": 12200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.260468077900275e-05,
+      "loss": 0.749,
+      "step": 12201
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.260356930971125e-05,
+      "loss": 0.7461,
+      "step": 12202
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2602457805916358e-05,
+      "loss": 0.7344,
+      "step": 12203
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2601346267632797e-05,
+      "loss": 0.7363,
+      "step": 12204
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2600234694875306e-05,
+      "loss": 0.7227,
+      "step": 12205
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2599123087658612e-05,
+      "loss": 0.7402,
+      "step": 12206
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2598011445997447e-05,
+      "loss": 0.7373,
+      "step": 12207
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.259689976990654e-05,
+      "loss": 0.7266,
+      "step": 12208
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2595788059400631e-05,
+      "loss": 0.7588,
+      "step": 12209
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2594676314494442e-05,
+      "loss": 0.7588,
+      "step": 12210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2593564535202715e-05,
+      "loss": 0.75,
+      "step": 12211
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2592452721540183e-05,
+      "loss": 0.7305,
+      "step": 12212
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2591340873521575e-05,
+      "loss": 0.7422,
+      "step": 12213
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.259022899116163e-05,
+      "loss": 0.7393,
+      "step": 12214
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.258911707447508e-05,
+      "loss": 0.7305,
+      "step": 12215
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2588005123476666e-05,
+      "loss": 0.748,
+      "step": 12216
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2586893138181113e-05,
+      "loss": 0.7344,
+      "step": 12217
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2585781118603168e-05,
+      "loss": 0.7217,
+      "step": 12218
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2584669064757566e-05,
+      "loss": 0.7686,
+      "step": 12219
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2583556976659041e-05,
+      "loss": 0.751,
+      "step": 12220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.258244485432233e-05,
+      "loss": 0.7393,
+      "step": 12221
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.258133269776218e-05,
+      "loss": 0.7695,
+      "step": 12222
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2580220506993315e-05,
+      "loss": 0.7432,
+      "step": 12223
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2579108282030487e-05,
+      "loss": 0.7334,
+      "step": 12224
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257799602288843e-05,
+      "loss": 0.7383,
+      "step": 12225
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2576883729581888e-05,
+      "loss": 0.7393,
+      "step": 12226
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2575771402125593e-05,
+      "loss": 0.7627,
+      "step": 12227
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2574659040534294e-05,
+      "loss": 0.751,
+      "step": 12228
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2573546644822732e-05,
+      "loss": 0.7363,
+      "step": 12229
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2572434215005644e-05,
+      "loss": 0.7383,
+      "step": 12230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2571321751097776e-05,
+      "loss": 0.7256,
+      "step": 12231
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2570209253113872e-05,
+      "loss": 0.7295,
+      "step": 12232
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2569096721068667e-05,
+      "loss": 0.7451,
+      "step": 12233
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2567984154976916e-05,
+      "loss": 0.7715,
+      "step": 12234
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2566871554853358e-05,
+      "loss": 0.7637,
+      "step": 12235
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2565758920712734e-05,
+      "loss": 0.7246,
+      "step": 12236
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.256464625256979e-05,
+      "loss": 0.7197,
+      "step": 12237
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.256353355043928e-05,
+      "loss": 0.7451,
+      "step": 12238
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.256242081433594e-05,
+      "loss": 0.7305,
+      "step": 12239
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.256130804427452e-05,
+      "loss": 0.7207,
+      "step": 12240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2560195240269766e-05,
+      "loss": 0.7324,
+      "step": 12241
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255908240233643e-05,
+      "loss": 0.7432,
+      "step": 12242
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2557969530489253e-05,
+      "loss": 0.7383,
+      "step": 12243
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2556856624742985e-05,
+      "loss": 0.7441,
+      "step": 12244
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2555743685112377e-05,
+      "loss": 0.7207,
+      "step": 12245
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255463071161218e-05,
+      "loss": 0.7344,
+      "step": 12246
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2553517704257137e-05,
+      "loss": 0.7227,
+      "step": 12247
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2552404663062002e-05,
+      "loss": 0.748,
+      "step": 12248
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2551291588041519e-05,
+      "loss": 0.7559,
+      "step": 12249
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255017847921045e-05,
+      "loss": 0.7285,
+      "step": 12250
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.254906533658354e-05,
+      "loss": 0.7461,
+      "step": 12251
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2547952160175541e-05,
+      "loss": 0.7529,
+      "step": 12252
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2546838950001205e-05,
+      "loss": 0.7236,
+      "step": 12253
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2545725706075285e-05,
+      "loss": 0.7363,
+      "step": 12254
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2544612428412537e-05,
+      "loss": 0.7461,
+      "step": 12255
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2543499117027707e-05,
+      "loss": 0.7559,
+      "step": 12256
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2542385771935556e-05,
+      "loss": 0.7383,
+      "step": 12257
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2541272393150839e-05,
+      "loss": 0.7588,
+      "step": 12258
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2540158980688306e-05,
+      "loss": 0.7363,
+      "step": 12259
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2539045534562713e-05,
+      "loss": 0.7324,
+      "step": 12260
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.253793205478882e-05,
+      "loss": 0.7363,
+      "step": 12261
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2536818541381377e-05,
+      "loss": 0.7441,
+      "step": 12262
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2535704994355147e-05,
+      "loss": 0.7285,
+      "step": 12263
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2534591413724883e-05,
+      "loss": 0.7402,
+      "step": 12264
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2533477799505347e-05,
+      "loss": 0.7168,
+      "step": 12265
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2532364151711287e-05,
+      "loss": 0.7285,
+      "step": 12266
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2531250470357472e-05,
+      "loss": 0.7871,
+      "step": 12267
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.253013675545866e-05,
+      "loss": 0.751,
+      "step": 12268
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2529023007029604e-05,
+      "loss": 0.7168,
+      "step": 12269
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2527909225085066e-05,
+      "loss": 0.7314,
+      "step": 12270
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2526795409639814e-05,
+      "loss": 0.7295,
+      "step": 12271
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2525681560708593e-05,
+      "loss": 0.7324,
+      "step": 12272
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2524567678306178e-05,
+      "loss": 0.7432,
+      "step": 12273
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2523453762447327e-05,
+      "loss": 0.7109,
+      "step": 12274
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.25223398131468e-05,
+      "loss": 0.7236,
+      "step": 12275
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2521225830419356e-05,
+      "loss": 0.7197,
+      "step": 12276
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252011181427977e-05,
+      "loss": 0.7227,
+      "step": 12277
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.251899776474279e-05,
+      "loss": 0.7354,
+      "step": 12278
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.251788368182319e-05,
+      "loss": 0.7256,
+      "step": 12279
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2516769565535732e-05,
+      "loss": 0.7402,
+      "step": 12280
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2515655415895181e-05,
+      "loss": 0.7402,
+      "step": 12281
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2514541232916297e-05,
+      "loss": 0.7432,
+      "step": 12282
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2513427016613853e-05,
+      "loss": 0.7539,
+      "step": 12283
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2512312767002613e-05,
+      "loss": 0.7354,
+      "step": 12284
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2511198484097338e-05,
+      "loss": 0.7383,
+      "step": 12285
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.25100841679128e-05,
+      "loss": 0.7295,
+      "step": 12286
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2508969818463768e-05,
+      "loss": 0.7266,
+      "step": 12287
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2507855435765003e-05,
+      "loss": 0.7383,
+      "step": 12288
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.250674101983128e-05,
+      "loss": 0.7607,
+      "step": 12289
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2505626570677364e-05,
+      "loss": 0.7314,
+      "step": 12290
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2504512088318026e-05,
+      "loss": 0.7422,
+      "step": 12291
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2503397572768032e-05,
+      "loss": 0.7637,
+      "step": 12292
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2502283024042155e-05,
+      "loss": 0.7236,
+      "step": 12293
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2501168442155165e-05,
+      "loss": 0.7295,
+      "step": 12294
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2500053827121832e-05,
+      "loss": 0.7246,
+      "step": 12295
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2498939178956928e-05,
+      "loss": 0.7373,
+      "step": 12296
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2497824497675226e-05,
+      "loss": 0.749,
+      "step": 12297
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2496709783291497e-05,
+      "loss": 0.7598,
+      "step": 12298
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2495595035820509e-05,
+      "loss": 0.7559,
+      "step": 12299
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2494480255277042e-05,
+      "loss": 0.7275,
+      "step": 12300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2493365441675865e-05,
+      "loss": 0.7373,
+      "step": 12301
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2492250595031754e-05,
+      "loss": 0.7441,
+      "step": 12302
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2491135715359486e-05,
+      "loss": 0.7383,
+      "step": 12303
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.249002080267383e-05,
+      "loss": 0.7588,
+      "step": 12304
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2488905856989562e-05,
+      "loss": 0.7061,
+      "step": 12305
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2487790878321462e-05,
+      "loss": 0.7363,
+      "step": 12306
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2486675866684302e-05,
+      "loss": 0.75,
+      "step": 12307
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2485560822092862e-05,
+      "loss": 0.748,
+      "step": 12308
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2484445744561913e-05,
+      "loss": 0.7324,
+      "step": 12309
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2483330634106244e-05,
+      "loss": 0.749,
+      "step": 12310
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2482215490740617e-05,
+      "loss": 0.7275,
+      "step": 12311
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2481100314479821e-05,
+      "loss": 0.7471,
+      "step": 12312
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2479985105338635e-05,
+      "loss": 0.7432,
+      "step": 12313
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2478869863331835e-05,
+      "loss": 0.7305,
+      "step": 12314
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2477754588474196e-05,
+      "loss": 0.7285,
+      "step": 12315
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2476639280780509e-05,
+      "loss": 0.7432,
+      "step": 12316
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2475523940265544e-05,
+      "loss": 0.7412,
+      "step": 12317
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2474408566944087e-05,
+      "loss": 0.7207,
+      "step": 12318
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.247329316083092e-05,
+      "loss": 0.748,
+      "step": 12319
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2472177721940824e-05,
+      "loss": 0.7617,
+      "step": 12320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2471062250288577e-05,
+      "loss": 0.7285,
+      "step": 12321
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2469946745888968e-05,
+      "loss": 0.7588,
+      "step": 12322
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2468831208756775e-05,
+      "loss": 0.7559,
+      "step": 12323
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2467715638906786e-05,
+      "loss": 0.7334,
+      "step": 12324
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246660003635378e-05,
+      "loss": 0.7529,
+      "step": 12325
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2465484401112546e-05,
+      "loss": 0.7441,
+      "step": 12326
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2464368733197865e-05,
+      "loss": 0.7422,
+      "step": 12327
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2463253032624525e-05,
+      "loss": 0.7422,
+      "step": 12328
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246213729940731e-05,
+      "loss": 0.752,
+      "step": 12329
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.246102153356101e-05,
+      "loss": 0.7344,
+      "step": 12330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2459905735100404e-05,
+      "loss": 0.7354,
+      "step": 12331
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245878990404028e-05,
+      "loss": 0.751,
+      "step": 12332
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2457674040395433e-05,
+      "loss": 0.7441,
+      "step": 12333
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2456558144180646e-05,
+      "loss": 0.7324,
+      "step": 12334
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2455442215410707e-05,
+      "loss": 0.7568,
+      "step": 12335
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2454326254100404e-05,
+      "loss": 0.7549,
+      "step": 12336
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245321026026453e-05,
+      "loss": 0.7178,
+      "step": 12337
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2452094233917868e-05,
+      "loss": 0.7422,
+      "step": 12338
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2450978175075215e-05,
+      "loss": 0.7598,
+      "step": 12339
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2449862083751357e-05,
+      "loss": 0.7314,
+      "step": 12340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2448745959961084e-05,
+      "loss": 0.7451,
+      "step": 12341
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2447629803719193e-05,
+      "loss": 0.7451,
+      "step": 12342
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.244651361504047e-05,
+      "loss": 0.7109,
+      "step": 12343
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2445397393939707e-05,
+      "loss": 0.7363,
+      "step": 12344
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2444281140431702e-05,
+      "loss": 0.749,
+      "step": 12345
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.244316485453124e-05,
+      "loss": 0.7383,
+      "step": 12346
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2442048536253122e-05,
+      "loss": 0.7686,
+      "step": 12347
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2440932185612138e-05,
+      "loss": 0.7607,
+      "step": 12348
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2439815802623084e-05,
+      "loss": 0.7471,
+      "step": 12349
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2438699387300752e-05,
+      "loss": 0.7305,
+      "step": 12350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.243758293965994e-05,
+      "loss": 0.7373,
+      "step": 12351
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2436466459715444e-05,
+      "loss": 0.751,
+      "step": 12352
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2435349947482056e-05,
+      "loss": 0.7412,
+      "step": 12353
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2434233402974573e-05,
+      "loss": 0.748,
+      "step": 12354
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2433116826207799e-05,
+      "loss": 0.7305,
+      "step": 12355
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2432000217196523e-05,
+      "loss": 0.7227,
+      "step": 12356
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2430883575955543e-05,
+      "loss": 0.7402,
+      "step": 12357
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2429766902499663e-05,
+      "loss": 0.7178,
+      "step": 12358
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2428650196843679e-05,
+      "loss": 0.7471,
+      "step": 12359
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2427533459002385e-05,
+      "loss": 0.71,
+      "step": 12360
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2426416688990588e-05,
+      "loss": 0.7129,
+      "step": 12361
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2425299886823084e-05,
+      "loss": 0.7148,
+      "step": 12362
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2424183052514671e-05,
+      "loss": 0.7461,
+      "step": 12363
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2423066186080154e-05,
+      "loss": 0.7314,
+      "step": 12364
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2421949287534334e-05,
+      "loss": 0.7412,
+      "step": 12365
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2420832356892007e-05,
+      "loss": 0.7344,
+      "step": 12366
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2419715394167981e-05,
+      "loss": 0.7559,
+      "step": 12367
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2418598399377057e-05,
+      "loss": 0.7754,
+      "step": 12368
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2417481372534037e-05,
+      "loss": 0.7227,
+      "step": 12369
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2416364313653722e-05,
+      "loss": 0.7275,
+      "step": 12370
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2415247222750922e-05,
+      "loss": 0.75,
+      "step": 12371
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2414130099840433e-05,
+      "loss": 0.7139,
+      "step": 12372
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2413012944937063e-05,
+      "loss": 0.7275,
+      "step": 12373
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2411895758055621e-05,
+      "loss": 0.7373,
+      "step": 12374
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2410778539210908e-05,
+      "loss": 0.7402,
+      "step": 12375
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240966128841773e-05,
+      "loss": 0.7119,
+      "step": 12376
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240854400569089e-05,
+      "loss": 0.7002,
+      "step": 12377
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2407426691045203e-05,
+      "loss": 0.7314,
+      "step": 12378
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240630934449547e-05,
+      "loss": 0.7373,
+      "step": 12379
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.24051919660565e-05,
+      "loss": 0.7334,
+      "step": 12380
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.24040745557431e-05,
+      "loss": 0.7373,
+      "step": 12381
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240295711357008e-05,
+      "loss": 0.7559,
+      "step": 12382
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2401839639552244e-05,
+      "loss": 0.7295,
+      "step": 12383
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240072213370441e-05,
+      "loss": 0.7412,
+      "step": 12384
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.239960459604138e-05,
+      "loss": 0.7666,
+      "step": 12385
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2398487026577965e-05,
+      "loss": 0.7197,
+      "step": 12386
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2397369425328978e-05,
+      "loss": 0.7324,
+      "step": 12387
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2396251792309231e-05,
+      "loss": 0.7305,
+      "step": 12388
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.239513412753353e-05,
+      "loss": 0.7637,
+      "step": 12389
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2394016431016692e-05,
+      "loss": 0.7285,
+      "step": 12390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2392898702773523e-05,
+      "loss": 0.8008,
+      "step": 12391
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2391780942818842e-05,
+      "loss": 0.7637,
+      "step": 12392
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2390663151167456e-05,
+      "loss": 0.7178,
+      "step": 12393
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2389545327834189e-05,
+      "loss": 0.7344,
+      "step": 12394
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2388427472833839e-05,
+      "loss": 0.7109,
+      "step": 12395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2387309586181234e-05,
+      "loss": 0.7568,
+      "step": 12396
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2386191667891182e-05,
+      "loss": 0.7246,
+      "step": 12397
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2385073717978498e-05,
+      "loss": 0.7578,
+      "step": 12398
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2383955736457996e-05,
+      "loss": 0.7432,
+      "step": 12399
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2382837723344496e-05,
+      "loss": 0.7422,
+      "step": 12400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2381719678652815e-05,
+      "loss": 0.7422,
+      "step": 12401
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2380601602397766e-05,
+      "loss": 0.7285,
+      "step": 12402
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2379483494594167e-05,
+      "loss": 0.7471,
+      "step": 12403
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.237836535525684e-05,
+      "loss": 0.7285,
+      "step": 12404
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2377247184400593e-05,
+      "loss": 0.7285,
+      "step": 12405
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2376128982040256e-05,
+      "loss": 0.7451,
+      "step": 12406
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.237501074819064e-05,
+      "loss": 0.7393,
+      "step": 12407
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2373892482866568e-05,
+      "loss": 0.7578,
+      "step": 12408
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2372774186082854e-05,
+      "loss": 0.7344,
+      "step": 12409
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2371655857854328e-05,
+      "loss": 0.7109,
+      "step": 12410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2370537498195798e-05,
+      "loss": 0.7783,
+      "step": 12411
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2369419107122096e-05,
+      "loss": 0.7256,
+      "step": 12412
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.236830068464804e-05,
+      "loss": 0.708,
+      "step": 12413
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2367182230788448e-05,
+      "loss": 0.7305,
+      "step": 12414
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2366063745558145e-05,
+      "loss": 0.7197,
+      "step": 12415
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2364945228971953e-05,
+      "loss": 0.7246,
+      "step": 12416
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2363826681044694e-05,
+      "loss": 0.7061,
+      "step": 12417
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2362708101791196e-05,
+      "loss": 0.7041,
+      "step": 12418
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2361589491226279e-05,
+      "loss": 0.7412,
+      "step": 12419
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2360470849364767e-05,
+      "loss": 0.7334,
+      "step": 12420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2359352176221484e-05,
+      "loss": 0.7412,
+      "step": 12421
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2358233471811256e-05,
+      "loss": 0.7295,
+      "step": 12422
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2357114736148912e-05,
+      "loss": 0.7363,
+      "step": 12423
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.235599596924927e-05,
+      "loss": 0.7344,
+      "step": 12424
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2354877171127164e-05,
+      "loss": 0.7461,
+      "step": 12425
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2353758341797417e-05,
+      "loss": 0.7412,
+      "step": 12426
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2352639481274857e-05,
+      "loss": 0.7441,
+      "step": 12427
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2351520589574311e-05,
+      "loss": 0.7217,
+      "step": 12428
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2350401666710608e-05,
+      "loss": 0.7354,
+      "step": 12429
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2349282712698573e-05,
+      "loss": 0.7422,
+      "step": 12430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234816372755304e-05,
+      "loss": 0.7158,
+      "step": 12431
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2347044711288834e-05,
+      "loss": 0.7422,
+      "step": 12432
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2345925663920789e-05,
+      "loss": 0.7441,
+      "step": 12433
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2344806585463727e-05,
+      "loss": 0.7061,
+      "step": 12434
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2343687475932486e-05,
+      "loss": 0.749,
+      "step": 12435
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2342568335341894e-05,
+      "loss": 0.7412,
+      "step": 12436
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2341449163706782e-05,
+      "loss": 0.7363,
+      "step": 12437
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2340329961041981e-05,
+      "loss": 0.7432,
+      "step": 12438
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2339210727362328e-05,
+      "loss": 0.7422,
+      "step": 12439
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2338091462682649e-05,
+      "loss": 0.7246,
+      "step": 12440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233697216701778e-05,
+      "loss": 0.7559,
+      "step": 12441
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2335852840382555e-05,
+      "loss": 0.7578,
+      "step": 12442
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2334733482791805e-05,
+      "loss": 0.7227,
+      "step": 12443
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2333614094260367e-05,
+      "loss": 0.7324,
+      "step": 12444
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2332494674803073e-05,
+      "loss": 0.7451,
+      "step": 12445
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233137522443476e-05,
+      "loss": 0.7285,
+      "step": 12446
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.233025574317026e-05,
+      "loss": 0.7383,
+      "step": 12447
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2329136231024412e-05,
+      "loss": 0.7275,
+      "step": 12448
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2328016688012057e-05,
+      "loss": 0.7285,
+      "step": 12449
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2326897114148019e-05,
+      "loss": 0.7246,
+      "step": 12450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2325777509447145e-05,
+      "loss": 0.707,
+      "step": 12451
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2324657873924268e-05,
+      "loss": 0.7285,
+      "step": 12452
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.232353820759423e-05,
+      "loss": 0.7402,
+      "step": 12453
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2322418510471863e-05,
+      "loss": 0.7178,
+      "step": 12454
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2321298782572008e-05,
+      "loss": 0.7305,
+      "step": 12455
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.232017902390951e-05,
+      "loss": 0.7441,
+      "step": 12456
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.23190592344992e-05,
+      "loss": 0.7441,
+      "step": 12457
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2317939414355922e-05,
+      "loss": 0.7266,
+      "step": 12458
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2316819563494515e-05,
+      "loss": 0.7686,
+      "step": 12459
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231569968192982e-05,
+      "loss": 0.7158,
+      "step": 12460
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2314579769676676e-05,
+      "loss": 0.75,
+      "step": 12461
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231345982674993e-05,
+      "loss": 0.7393,
+      "step": 12462
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231233985316442e-05,
+      "loss": 0.7656,
+      "step": 12463
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2311219848934987e-05,
+      "loss": 0.7246,
+      "step": 12464
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2310099814076477e-05,
+      "loss": 0.7324,
+      "step": 12465
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.230897974860373e-05,
+      "loss": 0.7324,
+      "step": 12466
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2307859652531591e-05,
+      "loss": 0.7402,
+      "step": 12467
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2306739525874907e-05,
+      "loss": 0.7227,
+      "step": 12468
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2305619368648517e-05,
+      "loss": 0.6982,
+      "step": 12469
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2304499180867267e-05,
+      "loss": 0.7295,
+      "step": 12470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2303378962546003e-05,
+      "loss": 0.707,
+      "step": 12471
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2302258713699573e-05,
+      "loss": 0.7305,
+      "step": 12472
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2301138434342817e-05,
+      "loss": 0.7324,
+      "step": 12473
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2300018124490587e-05,
+      "loss": 0.7383,
+      "step": 12474
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2298897784157729e-05,
+      "loss": 0.7402,
+      "step": 12475
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2297777413359085e-05,
+      "loss": 0.7559,
+      "step": 12476
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2296657012109506e-05,
+      "loss": 0.71,
+      "step": 12477
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2295536580423846e-05,
+      "loss": 0.7217,
+      "step": 12478
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2294416118316941e-05,
+      "loss": 0.7549,
+      "step": 12479
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229329562580365e-05,
+      "loss": 0.748,
+      "step": 12480
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2292175102898813e-05,
+      "loss": 0.7344,
+      "step": 12481
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2291054549617291e-05,
+      "loss": 0.7188,
+      "step": 12482
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2289933965973926e-05,
+      "loss": 0.7617,
+      "step": 12483
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2288813351983571e-05,
+      "loss": 0.7354,
+      "step": 12484
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2287692707661075e-05,
+      "loss": 0.7432,
+      "step": 12485
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2286572033021292e-05,
+      "loss": 0.7373,
+      "step": 12486
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2285451328079066e-05,
+      "loss": 0.7656,
+      "step": 12487
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.228433059284926e-05,
+      "loss": 0.6992,
+      "step": 12488
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.228320982734672e-05,
+      "loss": 0.7441,
+      "step": 12489
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22820890315863e-05,
+      "loss": 0.7324,
+      "step": 12490
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2280968205582852e-05,
+      "loss": 0.7236,
+      "step": 12491
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2279847349351234e-05,
+      "loss": 0.752,
+      "step": 12492
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.227872646290629e-05,
+      "loss": 0.748,
+      "step": 12493
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2277605546262886e-05,
+      "loss": 0.7266,
+      "step": 12494
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.227648459943587e-05,
+      "loss": 0.7207,
+      "step": 12495
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22753636224401e-05,
+      "loss": 0.7246,
+      "step": 12496
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2274242615290428e-05,
+      "loss": 0.7295,
+      "step": 12497
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2273121578001714e-05,
+      "loss": 0.7285,
+      "step": 12498
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2272000510588813e-05,
+      "loss": 0.7695,
+      "step": 12499
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2270879413066581e-05,
+      "loss": 0.7529,
+      "step": 12500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2269758285449875e-05,
+      "loss": 0.7451,
+      "step": 12501
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2268637127753555e-05,
+      "loss": 0.7197,
+      "step": 12502
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2267515939992479e-05,
+      "loss": 0.7393,
+      "step": 12503
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22663947221815e-05,
+      "loss": 0.75,
+      "step": 12504
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2265273474335481e-05,
+      "loss": 0.7578,
+      "step": 12505
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2264152196469281e-05,
+      "loss": 0.7051,
+      "step": 12506
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2263030888597758e-05,
+      "loss": 0.7432,
+      "step": 12507
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2261909550735776e-05,
+      "loss": 0.7197,
+      "step": 12508
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226078818289819e-05,
+      "loss": 0.7393,
+      "step": 12509
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2259666785099862e-05,
+      "loss": 0.7305,
+      "step": 12510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2258545357355658e-05,
+      "loss": 0.7471,
+      "step": 12511
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2257423899680431e-05,
+      "loss": 0.751,
+      "step": 12512
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2256302412089053e-05,
+      "loss": 0.7461,
+      "step": 12513
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.225518089459638e-05,
+      "loss": 0.7305,
+      "step": 12514
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2254059347217273e-05,
+      "loss": 0.7344,
+      "step": 12515
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.22529377699666e-05,
+      "loss": 0.7324,
+      "step": 12516
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2251816162859225e-05,
+      "loss": 0.7568,
+      "step": 12517
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2250694525910004e-05,
+      "loss": 0.7363,
+      "step": 12518
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.224957285913381e-05,
+      "loss": 0.7285,
+      "step": 12519
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2248451162545501e-05,
+      "loss": 0.7539,
+      "step": 12520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2247329436159951e-05,
+      "loss": 0.7344,
+      "step": 12521
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2246207679992015e-05,
+      "loss": 0.7256,
+      "step": 12522
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2245085894056567e-05,
+      "loss": 0.7432,
+      "step": 12523
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224396407836847e-05,
+      "loss": 0.7334,
+      "step": 12524
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224284223294259e-05,
+      "loss": 0.7314,
+      "step": 12525
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2241720357793796e-05,
+      "loss": 0.7275,
+      "step": 12526
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2240598452936956e-05,
+      "loss": 0.748,
+      "step": 12527
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2239476518386935e-05,
+      "loss": 0.7285,
+      "step": 12528
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2238354554158602e-05,
+      "loss": 0.7412,
+      "step": 12529
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2237232560266827e-05,
+      "loss": 0.7168,
+      "step": 12530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2236110536726478e-05,
+      "loss": 0.7197,
+      "step": 12531
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2234988483552423e-05,
+      "loss": 0.7285,
+      "step": 12532
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2233866400759536e-05,
+      "loss": 0.7334,
+      "step": 12533
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2232744288362683e-05,
+      "loss": 0.749,
+      "step": 12534
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2231622146376741e-05,
+      "loss": 0.7422,
+      "step": 12535
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2230499974816571e-05,
+      "loss": 0.7324,
+      "step": 12536
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2229377773697054e-05,
+      "loss": 0.7637,
+      "step": 12537
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2228255543033054e-05,
+      "loss": 0.7285,
+      "step": 12538
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222713328283945e-05,
+      "loss": 0.7588,
+      "step": 12539
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222601099313111e-05,
+      "loss": 0.7568,
+      "step": 12540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2224888673922907e-05,
+      "loss": 0.75,
+      "step": 12541
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222376632522972e-05,
+      "loss": 0.7617,
+      "step": 12542
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2222643947066415e-05,
+      "loss": 0.7188,
+      "step": 12543
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2221521539447871e-05,
+      "loss": 0.7148,
+      "step": 12544
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2220399102388958e-05,
+      "loss": 0.7354,
+      "step": 12545
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2219276635904558e-05,
+      "loss": 0.7393,
+      "step": 12546
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2218154140009542e-05,
+      "loss": 0.7627,
+      "step": 12547
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2217031614718784e-05,
+      "loss": 0.7246,
+      "step": 12548
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2215909060047162e-05,
+      "loss": 0.7275,
+      "step": 12549
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2214786476009557e-05,
+      "loss": 0.7256,
+      "step": 12550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2213663862620837e-05,
+      "loss": 0.7529,
+      "step": 12551
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2212541219895887e-05,
+      "loss": 0.709,
+      "step": 12552
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2211418547849581e-05,
+      "loss": 0.7256,
+      "step": 12553
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2210295846496796e-05,
+      "loss": 0.7188,
+      "step": 12554
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.220917311585241e-05,
+      "loss": 0.7314,
+      "step": 12555
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2208050355931309e-05,
+      "loss": 0.7334,
+      "step": 12556
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2206927566748363e-05,
+      "loss": 0.7412,
+      "step": 12557
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2205804748318457e-05,
+      "loss": 0.7422,
+      "step": 12558
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2204681900656467e-05,
+      "loss": 0.7256,
+      "step": 12559
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2203559023777281e-05,
+      "loss": 0.7344,
+      "step": 12560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2202436117695769e-05,
+      "loss": 0.7383,
+      "step": 12561
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2201313182426821e-05,
+      "loss": 0.7334,
+      "step": 12562
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2200190217985315e-05,
+      "loss": 0.7256,
+      "step": 12563
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2199067224386134e-05,
+      "loss": 0.7324,
+      "step": 12564
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2197944201644156e-05,
+      "loss": 0.7373,
+      "step": 12565
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2196821149774272e-05,
+      "loss": 0.7256,
+      "step": 12566
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2195698068791354e-05,
+      "loss": 0.7344,
+      "step": 12567
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2194574958710298e-05,
+      "loss": 0.7275,
+      "step": 12568
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2193451819545978e-05,
+      "loss": 0.7646,
+      "step": 12569
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2192328651313283e-05,
+      "loss": 0.748,
+      "step": 12570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2191205454027095e-05,
+      "loss": 0.7461,
+      "step": 12571
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2190082227702305e-05,
+      "loss": 0.7324,
+      "step": 12572
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2188958972353787e-05,
+      "loss": 0.7383,
+      "step": 12573
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218783568799644e-05,
+      "loss": 0.7139,
+      "step": 12574
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218671237464514e-05,
+      "loss": 0.7383,
+      "step": 12575
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2185589032314777e-05,
+      "loss": 0.7568,
+      "step": 12576
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2184465661020237e-05,
+      "loss": 0.7314,
+      "step": 12577
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.218334226077641e-05,
+      "loss": 0.75,
+      "step": 12578
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2182218831598183e-05,
+      "loss": 0.7334,
+      "step": 12579
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2181095373500443e-05,
+      "loss": 0.7207,
+      "step": 12580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2179971886498077e-05,
+      "loss": 0.7188,
+      "step": 12581
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2178848370605975e-05,
+      "loss": 0.7314,
+      "step": 12582
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2177724825839029e-05,
+      "loss": 0.7754,
+      "step": 12583
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2176601252212126e-05,
+      "loss": 0.7451,
+      "step": 12584
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2175477649740155e-05,
+      "loss": 0.7441,
+      "step": 12585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2174354018438009e-05,
+      "loss": 0.7285,
+      "step": 12586
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2173230358320576e-05,
+      "loss": 0.7285,
+      "step": 12587
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2172106669402748e-05,
+      "loss": 0.7256,
+      "step": 12588
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.217098295169942e-05,
+      "loss": 0.7432,
+      "step": 12589
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2169859205225476e-05,
+      "loss": 0.7275,
+      "step": 12590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2168735429995817e-05,
+      "loss": 0.7373,
+      "step": 12591
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2167611626025331e-05,
+      "loss": 0.7324,
+      "step": 12592
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2166487793328914e-05,
+      "loss": 0.7471,
+      "step": 12593
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2165363931921452e-05,
+      "loss": 0.7295,
+      "step": 12594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216424004181785e-05,
+      "loss": 0.7412,
+      "step": 12595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2163116123032992e-05,
+      "loss": 0.7217,
+      "step": 12596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2161992175581778e-05,
+      "loss": 0.7607,
+      "step": 12597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2160868199479103e-05,
+      "loss": 0.7197,
+      "step": 12598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.215974419473986e-05,
+      "loss": 0.7471,
+      "step": 12599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2158620161378945e-05,
+      "loss": 0.7471,
+      "step": 12600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2157496099411255e-05,
+      "loss": 0.7354,
+      "step": 12601
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2156372008851687e-05,
+      "loss": 0.749,
+      "step": 12602
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2155247889715138e-05,
+      "loss": 0.7393,
+      "step": 12603
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2154123742016502e-05,
+      "loss": 0.7334,
+      "step": 12604
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2152999565770683e-05,
+      "loss": 0.7207,
+      "step": 12605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2151875360992569e-05,
+      "loss": 0.7334,
+      "step": 12606
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2150751127697068e-05,
+      "loss": 0.7334,
+      "step": 12607
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2149626865899076e-05,
+      "loss": 0.7598,
+      "step": 12608
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2148502575613489e-05,
+      "loss": 0.7432,
+      "step": 12609
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2147378256855208e-05,
+      "loss": 0.7363,
+      "step": 12610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2146253909639137e-05,
+      "loss": 0.7539,
+      "step": 12611
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2145129533980167e-05,
+      "loss": 0.7412,
+      "step": 12612
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2144005129893209e-05,
+      "loss": 0.7256,
+      "step": 12613
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2142880697393156e-05,
+      "loss": 0.7178,
+      "step": 12614
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2141756236494915e-05,
+      "loss": 0.7412,
+      "step": 12615
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2140631747213382e-05,
+      "loss": 0.748,
+      "step": 12616
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2139507229563468e-05,
+      "loss": 0.7285,
+      "step": 12617
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2138382683560065e-05,
+      "loss": 0.75,
+      "step": 12618
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2137258109218085e-05,
+      "loss": 0.7529,
+      "step": 12619
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2136133506552426e-05,
+      "loss": 0.7285,
+      "step": 12620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2135008875577992e-05,
+      "loss": 0.7783,
+      "step": 12621
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2133884216309684e-05,
+      "loss": 0.7188,
+      "step": 12622
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2132759528762414e-05,
+      "loss": 0.7227,
+      "step": 12623
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2131634812951083e-05,
+      "loss": 0.7041,
+      "step": 12624
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2130510068890595e-05,
+      "loss": 0.7334,
+      "step": 12625
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2129385296595857e-05,
+      "loss": 0.7217,
+      "step": 12626
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2128260496081774e-05,
+      "loss": 0.7363,
+      "step": 12627
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2127135667363255e-05,
+      "loss": 0.7256,
+      "step": 12628
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.21260108104552e-05,
+      "loss": 0.7012,
+      "step": 12629
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2124885925372521e-05,
+      "loss": 0.7471,
+      "step": 12630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2123761012130126e-05,
+      "loss": 0.7334,
+      "step": 12631
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2122636070742921e-05,
+      "loss": 0.751,
+      "step": 12632
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2121511101225812e-05,
+      "loss": 0.7422,
+      "step": 12633
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2120386103593713e-05,
+      "loss": 0.7402,
+      "step": 12634
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2119261077861528e-05,
+      "loss": 0.7314,
+      "step": 12635
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2118136024044169e-05,
+      "loss": 0.75,
+      "step": 12636
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2117010942156545e-05,
+      "loss": 0.752,
+      "step": 12637
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2115885832213563e-05,
+      "loss": 0.7256,
+      "step": 12638
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2114760694230137e-05,
+      "loss": 0.748,
+      "step": 12639
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2113635528221175e-05,
+      "loss": 0.7275,
+      "step": 12640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2112510334201593e-05,
+      "loss": 0.7754,
+      "step": 12641
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2111385112186298e-05,
+      "loss": 0.7158,
+      "step": 12642
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2110259862190201e-05,
+      "loss": 0.7422,
+      "step": 12643
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.210913458422822e-05,
+      "loss": 0.7217,
+      "step": 12644
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2108009278315258e-05,
+      "loss": 0.7363,
+      "step": 12645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2106883944466237e-05,
+      "loss": 0.7256,
+      "step": 12646
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2105758582696067e-05,
+      "loss": 0.7305,
+      "step": 12647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2104633193019662e-05,
+      "loss": 0.7373,
+      "step": 12648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2103507775451935e-05,
+      "loss": 0.7227,
+      "step": 12649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2102382330007802e-05,
+      "loss": 0.7354,
+      "step": 12650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2101256856702174e-05,
+      "loss": 0.7422,
+      "step": 12651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2100131355549973e-05,
+      "loss": 0.7471,
+      "step": 12652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2099005826566106e-05,
+      "loss": 0.7432,
+      "step": 12653
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2097880269765498e-05,
+      "loss": 0.7578,
+      "step": 12654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2096754685163057e-05,
+      "loss": 0.7441,
+      "step": 12655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2095629072773707e-05,
+      "loss": 0.7285,
+      "step": 12656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.209450343261236e-05,
+      "loss": 0.7266,
+      "step": 12657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2093377764693934e-05,
+      "loss": 0.7139,
+      "step": 12658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2092252069033346e-05,
+      "loss": 0.7168,
+      "step": 12659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.209112634564552e-05,
+      "loss": 0.7031,
+      "step": 12660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2090000594545366e-05,
+      "loss": 0.7354,
+      "step": 12661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2088874815747808e-05,
+      "loss": 0.7627,
+      "step": 12662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2087749009267764e-05,
+      "loss": 0.7324,
+      "step": 12663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2086623175120155e-05,
+      "loss": 0.7168,
+      "step": 12664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2085497313319898e-05,
+      "loss": 0.7314,
+      "step": 12665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2084371423881916e-05,
+      "loss": 0.7285,
+      "step": 12666
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2083245506821127e-05,
+      "loss": 0.7461,
+      "step": 12667
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2082119562152456e-05,
+      "loss": 0.7646,
+      "step": 12668
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2080993589890822e-05,
+      "loss": 0.7129,
+      "step": 12669
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2079867590051144e-05,
+      "loss": 0.7236,
+      "step": 12670
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2078741562648348e-05,
+      "loss": 0.7158,
+      "step": 12671
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2077615507697356e-05,
+      "loss": 0.7598,
+      "step": 12672
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2076489425213094e-05,
+      "loss": 0.7305,
+      "step": 12673
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2075363315210475e-05,
+      "loss": 0.7402,
+      "step": 12674
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.207423717770443e-05,
+      "loss": 0.7119,
+      "step": 12675
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2073111012709887e-05,
+      "loss": 0.7314,
+      "step": 12676
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2071984820241763e-05,
+      "loss": 0.7324,
+      "step": 12677
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2070858600314982e-05,
+      "loss": 0.7256,
+      "step": 12678
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2069732352944477e-05,
+      "loss": 0.7344,
+      "step": 12679
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2068606078145164e-05,
+      "loss": 0.7334,
+      "step": 12680
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2067479775931977e-05,
+      "loss": 0.7188,
+      "step": 12681
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2066353446319836e-05,
+      "loss": 0.7402,
+      "step": 12682
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2065227089323673e-05,
+      "loss": 0.7295,
+      "step": 12683
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2064100704958409e-05,
+      "loss": 0.752,
+      "step": 12684
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2062974293238976e-05,
+      "loss": 0.7637,
+      "step": 12685
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2061847854180298e-05,
+      "loss": 0.7354,
+      "step": 12686
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2060721387797308e-05,
+      "loss": 0.7363,
+      "step": 12687
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2059594894104925e-05,
+      "loss": 0.7275,
+      "step": 12688
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205846837311809e-05,
+      "loss": 0.749,
+      "step": 12689
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205734182485172e-05,
+      "loss": 0.7363,
+      "step": 12690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2056215249320756e-05,
+      "loss": 0.7441,
+      "step": 12691
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2055088646540116e-05,
+      "loss": 0.7295,
+      "step": 12692
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205396201652474e-05,
+      "loss": 0.7197,
+      "step": 12693
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2052835359289552e-05,
+      "loss": 0.7393,
+      "step": 12694
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2051708674849488e-05,
+      "loss": 0.7285,
+      "step": 12695
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2050581963219475e-05,
+      "loss": 0.7344,
+      "step": 12696
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2049455224414448e-05,
+      "loss": 0.7354,
+      "step": 12697
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2048328458449332e-05,
+      "loss": 0.7441,
+      "step": 12698
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.204720166533907e-05,
+      "loss": 0.7305,
+      "step": 12699
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2046074845098585e-05,
+      "loss": 0.7188,
+      "step": 12700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2044947997742818e-05,
+      "loss": 0.7422,
+      "step": 12701
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2043821123286695e-05,
+      "loss": 0.7334,
+      "step": 12702
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2042694221745154e-05,
+      "loss": 0.7344,
+      "step": 12703
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2041567293133129e-05,
+      "loss": 0.7402,
+      "step": 12704
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2040440337465553e-05,
+      "loss": 0.7324,
+      "step": 12705
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.203931335475736e-05,
+      "loss": 0.7256,
+      "step": 12706
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2038186345023488e-05,
+      "loss": 0.7412,
+      "step": 12707
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2037059308278873e-05,
+      "loss": 0.7354,
+      "step": 12708
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2035932244538446e-05,
+      "loss": 0.7373,
+      "step": 12709
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2034805153817149e-05,
+      "loss": 0.7383,
+      "step": 12710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2033678036129912e-05,
+      "loss": 0.749,
+      "step": 12711
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.203255089149168e-05,
+      "loss": 0.7344,
+      "step": 12712
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2031423719917384e-05,
+      "loss": 0.7432,
+      "step": 12713
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2030296521421962e-05,
+      "loss": 0.7402,
+      "step": 12714
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2029169296020356e-05,
+      "loss": 0.7412,
+      "step": 12715
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2028042043727502e-05,
+      "loss": 0.7207,
+      "step": 12716
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2026914764558337e-05,
+      "loss": 0.7451,
+      "step": 12717
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2025787458527807e-05,
+      "loss": 0.7686,
+      "step": 12718
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.202466012565084e-05,
+      "loss": 0.7061,
+      "step": 12719
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2023532765942385e-05,
+      "loss": 0.7412,
+      "step": 12720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2022405379417376e-05,
+      "loss": 0.7207,
+      "step": 12721
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2021277966090765e-05,
+      "loss": 0.7383,
+      "step": 12722
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2020150525977479e-05,
+      "loss": 0.7393,
+      "step": 12723
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2019023059092465e-05,
+      "loss": 0.7451,
+      "step": 12724
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2017895565450666e-05,
+      "loss": 0.7256,
+      "step": 12725
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2016768045067022e-05,
+      "loss": 0.7305,
+      "step": 12726
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2015640497956473e-05,
+      "loss": 0.7402,
+      "step": 12727
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.201451292413397e-05,
+      "loss": 0.75,
+      "step": 12728
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2013385323614446e-05,
+      "loss": 0.7393,
+      "step": 12729
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.201225769641285e-05,
+      "loss": 0.752,
+      "step": 12730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2011130042544125e-05,
+      "loss": 0.7617,
+      "step": 12731
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2010002362023217e-05,
+      "loss": 0.7383,
+      "step": 12732
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2008874654865061e-05,
+      "loss": 0.7451,
+      "step": 12733
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2007746921084612e-05,
+      "loss": 0.7451,
+      "step": 12734
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2006619160696815e-05,
+      "loss": 0.7246,
+      "step": 12735
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200549137371661e-05,
+      "loss": 0.7324,
+      "step": 12736
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2004363560158943e-05,
+      "loss": 0.7393,
+      "step": 12737
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2003235720038768e-05,
+      "loss": 0.7363,
+      "step": 12738
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.200210785337102e-05,
+      "loss": 0.748,
+      "step": 12739
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2000979960170654e-05,
+      "loss": 0.7344,
+      "step": 12740
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1999852040452615e-05,
+      "loss": 0.7393,
+      "step": 12741
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1998724094231851e-05,
+      "loss": 0.7354,
+      "step": 12742
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1997596121523305e-05,
+      "loss": 0.7441,
+      "step": 12743
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1996468122341936e-05,
+      "loss": 0.7275,
+      "step": 12744
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1995340096702682e-05,
+      "loss": 0.7207,
+      "step": 12745
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1994212044620497e-05,
+      "loss": 0.7236,
+      "step": 12746
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.199308396611033e-05,
+      "loss": 0.7393,
+      "step": 12747
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1991955861187132e-05,
+      "loss": 0.7432,
+      "step": 12748
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.199082772986585e-05,
+      "loss": 0.7305,
+      "step": 12749
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1989699572161435e-05,
+      "loss": 0.7627,
+      "step": 12750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.198857138808884e-05,
+      "loss": 0.7598,
+      "step": 12751
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1987443177663012e-05,
+      "loss": 0.7393,
+      "step": 12752
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1986314940898906e-05,
+      "loss": 0.7334,
+      "step": 12753
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1985186677811475e-05,
+      "loss": 0.7354,
+      "step": 12754
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1984058388415667e-05,
+      "loss": 0.7275,
+      "step": 12755
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1982930072726436e-05,
+      "loss": 0.7295,
+      "step": 12756
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1981801730758738e-05,
+      "loss": 0.7295,
+      "step": 12757
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.198067336252752e-05,
+      "loss": 0.7227,
+      "step": 12758
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1979544968047741e-05,
+      "loss": 0.7295,
+      "step": 12759
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1978416547334348e-05,
+      "loss": 0.7314,
+      "step": 12760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1977288100402307e-05,
+      "loss": 0.7305,
+      "step": 12761
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1976159627266561e-05,
+      "loss": 0.752,
+      "step": 12762
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1975031127942071e-05,
+      "loss": 0.7188,
+      "step": 12763
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1973902602443789e-05,
+      "loss": 0.7559,
+      "step": 12764
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1972774050786675e-05,
+      "loss": 0.7314,
+      "step": 12765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.197164547298568e-05,
+      "loss": 0.7275,
+      "step": 12766
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1970516869055765e-05,
+      "loss": 0.7314,
+      "step": 12767
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1969388239011883e-05,
+      "loss": 0.7354,
+      "step": 12768
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1968259582868994e-05,
+      "loss": 0.7432,
+      "step": 12769
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196713090064205e-05,
+      "loss": 0.7754,
+      "step": 12770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1966002192346013e-05,
+      "loss": 0.7227,
+      "step": 12771
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196487345799584e-05,
+      "loss": 0.7266,
+      "step": 12772
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1963744697606492e-05,
+      "loss": 0.7402,
+      "step": 12773
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1962615911192922e-05,
+      "loss": 0.7188,
+      "step": 12774
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1961487098770096e-05,
+      "loss": 0.7588,
+      "step": 12775
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1960358260352966e-05,
+      "loss": 0.7363,
+      "step": 12776
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1959229395956502e-05,
+      "loss": 0.7422,
+      "step": 12777
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1958100505595652e-05,
+      "loss": 0.749,
+      "step": 12778
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1956971589285385e-05,
+      "loss": 0.7334,
+      "step": 12779
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1955842647040657e-05,
+      "loss": 0.7363,
+      "step": 12780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1954713678876433e-05,
+      "loss": 0.7275,
+      "step": 12781
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1953584684807669e-05,
+      "loss": 0.7432,
+      "step": 12782
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1952455664849335e-05,
+      "loss": 0.7148,
+      "step": 12783
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1951326619016384e-05,
+      "loss": 0.7402,
+      "step": 12784
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1950197547323786e-05,
+      "loss": 0.7266,
+      "step": 12785
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.19490684497865e-05,
+      "loss": 0.7314,
+      "step": 12786
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.194793932641949e-05,
+      "loss": 0.7402,
+      "step": 12787
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.194681017723772e-05,
+      "loss": 0.7559,
+      "step": 12788
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1945681002256151e-05,
+      "loss": 0.7188,
+      "step": 12789
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1944551801489753e-05,
+      "loss": 0.7354,
+      "step": 12790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1943422574953484e-05,
+      "loss": 0.748,
+      "step": 12791
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1942293322662316e-05,
+      "loss": 0.7334,
+      "step": 12792
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1941164044631208e-05,
+      "loss": 0.7393,
+      "step": 12793
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1940034740875128e-05,
+      "loss": 0.7412,
+      "step": 12794
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1938905411409041e-05,
+      "loss": 0.7373,
+      "step": 12795
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1937776056247916e-05,
+      "loss": 0.7383,
+      "step": 12796
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1936646675406717e-05,
+      "loss": 0.7227,
+      "step": 12797
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1935517268900412e-05,
+      "loss": 0.7256,
+      "step": 12798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1934387836743968e-05,
+      "loss": 0.7256,
+      "step": 12799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1933258378952352e-05,
+      "loss": 0.7246,
+      "step": 12800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.193212889554053e-05,
+      "loss": 0.7188,
+      "step": 12801
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1930999386523476e-05,
+      "loss": 0.7295,
+      "step": 12802
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1929869851916156e-05,
+      "loss": 0.7354,
+      "step": 12803
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1928740291733539e-05,
+      "loss": 0.7236,
+      "step": 12804
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1927610705990589e-05,
+      "loss": 0.7549,
+      "step": 12805
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1926481094702285e-05,
+      "loss": 0.7354,
+      "step": 12806
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1925351457883589e-05,
+      "loss": 0.6982,
+      "step": 12807
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1924221795549478e-05,
+      "loss": 0.7412,
+      "step": 12808
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1923092107714918e-05,
+      "loss": 0.7451,
+      "step": 12809
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1921962394394881e-05,
+      "loss": 0.7207,
+      "step": 12810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1920832655604337e-05,
+      "loss": 0.7422,
+      "step": 12811
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1919702891358265e-05,
+      "loss": 0.7461,
+      "step": 12812
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1918573101671627e-05,
+      "loss": 0.7256,
+      "step": 12813
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1917443286559401e-05,
+      "loss": 0.7246,
+      "step": 12814
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1916313446036556e-05,
+      "loss": 0.749,
+      "step": 12815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1915183580118072e-05,
+      "loss": 0.7402,
+      "step": 12816
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1914053688818915e-05,
+      "loss": 0.7178,
+      "step": 12817
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1912923772154063e-05,
+      "loss": 0.7119,
+      "step": 12818
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.191179383013849e-05,
+      "loss": 0.7246,
+      "step": 12819
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1910663862787169e-05,
+      "loss": 0.7168,
+      "step": 12820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1909533870115071e-05,
+      "loss": 0.7402,
+      "step": 12821
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.190840385213718e-05,
+      "loss": 0.7803,
+      "step": 12822
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.190727380886846e-05,
+      "loss": 0.7139,
+      "step": 12823
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1906143740323896e-05,
+      "loss": 0.7451,
+      "step": 12824
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1905013646518464e-05,
+      "loss": 0.707,
+      "step": 12825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1903883527467135e-05,
+      "loss": 0.7412,
+      "step": 12826
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1902753383184887e-05,
+      "loss": 0.7236,
+      "step": 12827
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.19016232136867e-05,
+      "loss": 0.7412,
+      "step": 12828
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1900493018987546e-05,
+      "loss": 0.7197,
+      "step": 12829
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1899362799102412e-05,
+      "loss": 0.7363,
+      "step": 12830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1898232554046267e-05,
+      "loss": 0.708,
+      "step": 12831
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1897102283834094e-05,
+      "loss": 0.7207,
+      "step": 12832
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.189597198848087e-05,
+      "loss": 0.7178,
+      "step": 12833
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.189484166800157e-05,
+      "loss": 0.7578,
+      "step": 12834
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1893711322411185e-05,
+      "loss": 0.7402,
+      "step": 12835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1892580951724684e-05,
+      "loss": 0.7207,
+      "step": 12836
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1891450555957054e-05,
+      "loss": 0.7314,
+      "step": 12837
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.189032013512327e-05,
+      "loss": 0.708,
+      "step": 12838
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1889189689238313e-05,
+      "loss": 0.7324,
+      "step": 12839
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1888059218317167e-05,
+      "loss": 0.7393,
+      "step": 12840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1886928722374813e-05,
+      "loss": 0.7471,
+      "step": 12841
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1885798201426232e-05,
+      "loss": 0.7373,
+      "step": 12842
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1884667655486405e-05,
+      "loss": 0.7158,
+      "step": 12843
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1883537084570316e-05,
+      "loss": 0.7002,
+      "step": 12844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.188240648869295e-05,
+      "loss": 0.75,
+      "step": 12845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1881275867869283e-05,
+      "loss": 0.7334,
+      "step": 12846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1880145222114305e-05,
+      "loss": 0.7217,
+      "step": 12847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1879014551442998e-05,
+      "loss": 0.7373,
+      "step": 12848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1877883855870342e-05,
+      "loss": 0.7402,
+      "step": 12849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1876753135411328e-05,
+      "loss": 0.7539,
+      "step": 12850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1875622390080936e-05,
+      "loss": 0.7373,
+      "step": 12851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1874491619894152e-05,
+      "loss": 0.7246,
+      "step": 12852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1873360824865962e-05,
+      "loss": 0.7285,
+      "step": 12853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1872230005011352e-05,
+      "loss": 0.7559,
+      "step": 12854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1871099160345307e-05,
+      "loss": 0.7256,
+      "step": 12855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1869968290882814e-05,
+      "loss": 0.7178,
+      "step": 12856
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.186883739663886e-05,
+      "loss": 0.7305,
+      "step": 12857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1867706477628431e-05,
+      "loss": 0.7314,
+      "step": 12858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1866575533866516e-05,
+      "loss": 0.7432,
+      "step": 12859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1865444565368099e-05,
+      "loss": 0.7354,
+      "step": 12860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1864313572148172e-05,
+      "loss": 0.7422,
+      "step": 12861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1863182554221719e-05,
+      "loss": 0.7363,
+      "step": 12862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1862051511603737e-05,
+      "loss": 0.7861,
+      "step": 12863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1860920444309205e-05,
+      "loss": 0.7412,
+      "step": 12864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1859789352353119e-05,
+      "loss": 0.7373,
+      "step": 12865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1858658235750463e-05,
+      "loss": 0.7344,
+      "step": 12866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1857527094516232e-05,
+      "loss": 0.7256,
+      "step": 12867
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1856395928665416e-05,
+      "loss": 0.7549,
+      "step": 12868
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1855264738213003e-05,
+      "loss": 0.7168,
+      "step": 12869
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1854133523173985e-05,
+      "loss": 0.7314,
+      "step": 12870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1853002283563355e-05,
+      "loss": 0.7383,
+      "step": 12871
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1851871019396099e-05,
+      "loss": 0.7295,
+      "step": 12872
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1850739730687215e-05,
+      "loss": 0.7373,
+      "step": 12873
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1849608417451695e-05,
+      "loss": 0.7441,
+      "step": 12874
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1848477079704525e-05,
+      "loss": 0.7471,
+      "step": 12875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1847345717460705e-05,
+      "loss": 0.7549,
+      "step": 12876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1846214330735226e-05,
+      "loss": 0.7432,
+      "step": 12877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1845082919543081e-05,
+      "loss": 0.7188,
+      "step": 12878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.184395148389926e-05,
+      "loss": 0.7559,
+      "step": 12879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1842820023818766e-05,
+      "loss": 0.7471,
+      "step": 12880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1841688539316588e-05,
+      "loss": 0.7266,
+      "step": 12881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1840557030407723e-05,
+      "loss": 0.7598,
+      "step": 12882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1839425497107157e-05,
+      "loss": 0.75,
+      "step": 12883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1838293939429904e-05,
+      "loss": 0.7363,
+      "step": 12884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1837162357390941e-05,
+      "loss": 0.7549,
+      "step": 12885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1836030751005275e-05,
+      "loss": 0.7393,
+      "step": 12886
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1834899120287897e-05,
+      "loss": 0.7373,
+      "step": 12887
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1833767465253808e-05,
+      "loss": 0.7373,
+      "step": 12888
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1832635785918e-05,
+      "loss": 0.7305,
+      "step": 12889
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.183150408229548e-05,
+      "loss": 0.7295,
+      "step": 12890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1830372354401234e-05,
+      "loss": 0.7178,
+      "step": 12891
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182924060225027e-05,
+      "loss": 0.7402,
+      "step": 12892
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1828108825857576e-05,
+      "loss": 0.7246,
+      "step": 12893
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1826977025238162e-05,
+      "loss": 0.7158,
+      "step": 12894
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1825845200407017e-05,
+      "loss": 0.7217,
+      "step": 12895
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182471335137915e-05,
+      "loss": 0.7256,
+      "step": 12896
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182358147816955e-05,
+      "loss": 0.7324,
+      "step": 12897
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1822449580793226e-05,
+      "loss": 0.7559,
+      "step": 12898
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1821317659265172e-05,
+      "loss": 0.7422,
+      "step": 12899
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1820185713600396e-05,
+      "loss": 0.7451,
+      "step": 12900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1819053743813888e-05,
+      "loss": 0.7178,
+      "step": 12901
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.181792174992066e-05,
+      "loss": 0.7412,
+      "step": 12902
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1816789731935707e-05,
+      "loss": 0.7227,
+      "step": 12903
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1815657689874036e-05,
+      "loss": 0.7695,
+      "step": 12904
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1814525623750641e-05,
+      "loss": 0.7559,
+      "step": 12905
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1813393533580535e-05,
+      "loss": 0.7412,
+      "step": 12906
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1812261419378713e-05,
+      "loss": 0.7344,
+      "step": 12907
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.181112928116018e-05,
+      "loss": 0.7559,
+      "step": 12908
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1809997118939942e-05,
+      "loss": 0.7559,
+      "step": 12909
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1808864932733001e-05,
+      "loss": 0.7451,
+      "step": 12910
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1807732722554359e-05,
+      "loss": 0.7256,
+      "step": 12911
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1806600488419024e-05,
+      "loss": 0.7432,
+      "step": 12912
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1805468230342e-05,
+      "loss": 0.752,
+      "step": 12913
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1804335948338293e-05,
+      "loss": 0.7373,
+      "step": 12914
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1803203642422905e-05,
+      "loss": 0.7422,
+      "step": 12915
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1802071312610844e-05,
+      "loss": 0.7285,
+      "step": 12916
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1800938958917114e-05,
+      "loss": 0.7178,
+      "step": 12917
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1799806581356723e-05,
+      "loss": 0.7227,
+      "step": 12918
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1798674179944679e-05,
+      "loss": 0.7285,
+      "step": 12919
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1797541754695986e-05,
+      "loss": 0.7422,
+      "step": 12920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1796409305625653e-05,
+      "loss": 0.7188,
+      "step": 12921
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1795276832748688e-05,
+      "loss": 0.7422,
+      "step": 12922
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17941443360801e-05,
+      "loss": 0.7363,
+      "step": 12923
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1793011815634892e-05,
+      "loss": 0.7422,
+      "step": 12924
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1791879271428076e-05,
+      "loss": 0.7207,
+      "step": 12925
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1790746703474666e-05,
+      "loss": 0.7432,
+      "step": 12926
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1789614111789662e-05,
+      "loss": 0.7178,
+      "step": 12927
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1788481496388077e-05,
+      "loss": 0.7324,
+      "step": 12928
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1787348857284925e-05,
+      "loss": 0.7441,
+      "step": 12929
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178621619449521e-05,
+      "loss": 0.7432,
+      "step": 12930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1785083508033945e-05,
+      "loss": 0.7158,
+      "step": 12931
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1783950797916144e-05,
+      "loss": 0.7207,
+      "step": 12932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178281806415681e-05,
+      "loss": 0.7422,
+      "step": 12933
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1781685306770961e-05,
+      "loss": 0.7314,
+      "step": 12934
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1780552525773607e-05,
+      "loss": 0.7207,
+      "step": 12935
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1779419721179762e-05,
+      "loss": 0.7383,
+      "step": 12936
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1778286893004434e-05,
+      "loss": 0.7285,
+      "step": 12937
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1777154041262638e-05,
+      "loss": 0.7529,
+      "step": 12938
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1776021165969389e-05,
+      "loss": 0.7246,
+      "step": 12939
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1774888267139694e-05,
+      "loss": 0.7031,
+      "step": 12940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1773755344788574e-05,
+      "loss": 0.7549,
+      "step": 12941
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1772622398931037e-05,
+      "loss": 0.7412,
+      "step": 12942
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1771489429582101e-05,
+      "loss": 0.7373,
+      "step": 12943
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1770356436756777e-05,
+      "loss": 0.7373,
+      "step": 12944
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1769223420470088e-05,
+      "loss": 0.7354,
+      "step": 12945
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1768090380737037e-05,
+      "loss": 0.7363,
+      "step": 12946
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1766957317572647e-05,
+      "loss": 0.7529,
+      "step": 12947
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1765824230991933e-05,
+      "loss": 0.7344,
+      "step": 12948
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176469112100991e-05,
+      "loss": 0.7422,
+      "step": 12949
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1763557987641593e-05,
+      "loss": 0.7559,
+      "step": 12950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1762424830902004e-05,
+      "loss": 0.7197,
+      "step": 12951
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1761291650806155e-05,
+      "loss": 0.7402,
+      "step": 12952
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1760158447369063e-05,
+      "loss": 0.7539,
+      "step": 12953
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1759025220605749e-05,
+      "loss": 0.7285,
+      "step": 12954
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1757891970531229e-05,
+      "loss": 0.7148,
+      "step": 12955
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1756758697160517e-05,
+      "loss": 0.7041,
+      "step": 12956
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1755625400508641e-05,
+      "loss": 0.749,
+      "step": 12957
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1754492080590613e-05,
+      "loss": 0.7451,
+      "step": 12958
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1753358737421454e-05,
+      "loss": 0.7412,
+      "step": 12959
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1752225371016187e-05,
+      "loss": 0.7432,
+      "step": 12960
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1751091981389822e-05,
+      "loss": 0.7295,
+      "step": 12961
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1749958568557389e-05,
+      "loss": 0.7432,
+      "step": 12962
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17488251325339e-05,
+      "loss": 0.7344,
+      "step": 12963
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1747691673334386e-05,
+      "loss": 0.7363,
+      "step": 12964
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1746558190973858e-05,
+      "loss": 0.7158,
+      "step": 12965
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1745424685467344e-05,
+      "loss": 0.7314,
+      "step": 12966
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1744291156829862e-05,
+      "loss": 0.7451,
+      "step": 12967
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1743157605076436e-05,
+      "loss": 0.7275,
+      "step": 12968
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1742024030222084e-05,
+      "loss": 0.7285,
+      "step": 12969
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1740890432281836e-05,
+      "loss": 0.7139,
+      "step": 12970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1739756811270708e-05,
+      "loss": 0.7529,
+      "step": 12971
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1738623167203727e-05,
+      "loss": 0.7207,
+      "step": 12972
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1737489500095912e-05,
+      "loss": 0.7402,
+      "step": 12973
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1736355809962291e-05,
+      "loss": 0.7188,
+      "step": 12974
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1735222096817891e-05,
+      "loss": 0.7441,
+      "step": 12975
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1734088360677729e-05,
+      "loss": 0.7568,
+      "step": 12976
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1732954601556831e-05,
+      "loss": 0.7568,
+      "step": 12977
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173182081947023e-05,
+      "loss": 0.7334,
+      "step": 12978
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173068701443294e-05,
+      "loss": 0.7314,
+      "step": 12979
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1729553186459992e-05,
+      "loss": 0.71,
+      "step": 12980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1728419335566415e-05,
+      "loss": 0.7578,
+      "step": 12981
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1727285461767229e-05,
+      "loss": 0.748,
+      "step": 12982
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1726151565077464e-05,
+      "loss": 0.7559,
+      "step": 12983
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1725017645512147e-05,
+      "loss": 0.7109,
+      "step": 12984
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1723883703086303e-05,
+      "loss": 0.7344,
+      "step": 12985
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.172274973781496e-05,
+      "loss": 0.7363,
+      "step": 12986
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1721615749713147e-05,
+      "loss": 0.7295,
+      "step": 12987
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1720481738795893e-05,
+      "loss": 0.7344,
+      "step": 12988
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1719347705078221e-05,
+      "loss": 0.7393,
+      "step": 12989
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1718213648575165e-05,
+      "loss": 0.7334,
+      "step": 12990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1717079569301753e-05,
+      "loss": 0.752,
+      "step": 12991
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1715945467273014e-05,
+      "loss": 0.7354,
+      "step": 12992
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1714811342503976e-05,
+      "loss": 0.7158,
+      "step": 12993
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1713677195009668e-05,
+      "loss": 0.7471,
+      "step": 12994
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.171254302480512e-05,
+      "loss": 0.749,
+      "step": 12995
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1711408831905369e-05,
+      "loss": 0.7598,
+      "step": 12996
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1710274616325439e-05,
+      "loss": 0.7188,
+      "step": 12997
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1709140378080361e-05,
+      "loss": 0.7373,
+      "step": 12998
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.170800611718517e-05,
+      "loss": 0.7334,
+      "step": 12999
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1706871833654896e-05,
+      "loss": 0.748,
+      "step": 13000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1705737527504572e-05,
+      "loss": 0.7471,
+      "step": 13001
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1704603198749225e-05,
+      "loss": 0.7402,
+      "step": 13002
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1703468847403893e-05,
+      "loss": 0.71,
+      "step": 13003
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1702334473483609e-05,
+      "loss": 0.7422,
+      "step": 13004
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1701200077003403e-05,
+      "loss": 0.7305,
+      "step": 13005
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1700065657978308e-05,
+      "loss": 0.7549,
+      "step": 13006
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1698931216423365e-05,
+      "loss": 0.7207,
+      "step": 13007
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1697796752353596e-05,
+      "loss": 0.7461,
+      "step": 13008
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1696662265784045e-05,
+      "loss": 0.7412,
+      "step": 13009
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1695527756729744e-05,
+      "loss": 0.7227,
+      "step": 13010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1694393225205727e-05,
+      "loss": 0.7266,
+      "step": 13011
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1693258671227028e-05,
+      "loss": 0.7383,
+      "step": 13012
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1692124094808686e-05,
+      "loss": 0.7314,
+      "step": 13013
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1690989495965733e-05,
+      "loss": 0.7188,
+      "step": 13014
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1689854874713207e-05,
+      "loss": 0.7295,
+      "step": 13015
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1688720231066145e-05,
+      "loss": 0.7354,
+      "step": 13016
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1687585565039582e-05,
+      "loss": 0.7461,
+      "step": 13017
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1686450876648557e-05,
+      "loss": 0.7295,
+      "step": 13018
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1685316165908106e-05,
+      "loss": 0.7314,
+      "step": 13019
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1684181432833269e-05,
+      "loss": 0.7275,
+      "step": 13020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1683046677439078e-05,
+      "loss": 0.7217,
+      "step": 13021
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1681911899740574e-05,
+      "loss": 0.7412,
+      "step": 13022
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1680777099752802e-05,
+      "loss": 0.7441,
+      "step": 13023
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1679642277490791e-05,
+      "loss": 0.7158,
+      "step": 13024
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1678507432969586e-05,
+      "loss": 0.7139,
+      "step": 13025
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1677372566204223e-05,
+      "loss": 0.7334,
+      "step": 13026
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1676237677209743e-05,
+      "loss": 0.7207,
+      "step": 13027
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1675102766001184e-05,
+      "loss": 0.749,
+      "step": 13028
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1673967832593594e-05,
+      "loss": 0.7354,
+      "step": 13029
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1672832877002005e-05,
+      "loss": 0.7344,
+      "step": 13030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1671697899241458e-05,
+      "loss": 0.752,
+      "step": 13031
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1670562899327e-05,
+      "loss": 0.7227,
+      "step": 13032
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166942787727367e-05,
+      "loss": 0.7314,
+      "step": 13033
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1668292833096504e-05,
+      "loss": 0.7393,
+      "step": 13034
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1667157766810553e-05,
+      "loss": 0.7295,
+      "step": 13035
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1666022678430854e-05,
+      "loss": 0.7559,
+      "step": 13036
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.166488756797245e-05,
+      "loss": 0.7295,
+      "step": 13037
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1663752435450389e-05,
+      "loss": 0.7607,
+      "step": 13038
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1662617280879707e-05,
+      "loss": 0.7314,
+      "step": 13039
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1661482104275445e-05,
+      "loss": 0.7256,
+      "step": 13040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1660346905652657e-05,
+      "loss": 0.7295,
+      "step": 13041
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1659211685026383e-05,
+      "loss": 0.7217,
+      "step": 13042
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1658076442411665e-05,
+      "loss": 0.7402,
+      "step": 13043
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1656941177823548e-05,
+      "loss": 0.7246,
+      "step": 13044
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.165580589127708e-05,
+      "loss": 0.7217,
+      "step": 13045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1654670582787304e-05,
+      "loss": 0.7441,
+      "step": 13046
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.165353525236926e-05,
+      "loss": 0.7373,
+      "step": 13047
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1652399900038007e-05,
+      "loss": 0.7168,
+      "step": 13048
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1651264525808582e-05,
+      "loss": 0.7305,
+      "step": 13049
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1650129129696031e-05,
+      "loss": 0.7217,
+      "step": 13050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.16489937117154e-05,
+      "loss": 0.7354,
+      "step": 13051
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1647858271881742e-05,
+      "loss": 0.7246,
+      "step": 13052
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1646722810210097e-05,
+      "loss": 0.7383,
+      "step": 13053
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1645587326715523e-05,
+      "loss": 0.7461,
+      "step": 13054
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1644451821413054e-05,
+      "loss": 0.7334,
+      "step": 13055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.164331629431775e-05,
+      "loss": 0.7344,
+      "step": 13056
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.164218074544465e-05,
+      "loss": 0.7236,
+      "step": 13057
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1641045174808812e-05,
+      "loss": 0.7422,
+      "step": 13058
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1639909582425277e-05,
+      "loss": 0.7168,
+      "step": 13059
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1638773968309099e-05,
+      "loss": 0.7305,
+      "step": 13060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1637638332475323e-05,
+      "loss": 0.7363,
+      "step": 13061
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1636502674939007e-05,
+      "loss": 0.7266,
+      "step": 13062
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163536699571519e-05,
+      "loss": 0.7324,
+      "step": 13063
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1634231294818934e-05,
+      "loss": 0.7383,
+      "step": 13064
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163309557226528e-05,
+      "loss": 0.7109,
+      "step": 13065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1631959828069287e-05,
+      "loss": 0.7168,
+      "step": 13066
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1630824062245998e-05,
+      "loss": 0.7354,
+      "step": 13067
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1629688274810472e-05,
+      "loss": 0.71,
+      "step": 13068
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1628552465777756e-05,
+      "loss": 0.7246,
+      "step": 13069
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1627416635162904e-05,
+      "loss": 0.7637,
+      "step": 13070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1626280782980967e-05,
+      "loss": 0.7266,
+      "step": 13071
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1625144909247004e-05,
+      "loss": 0.7197,
+      "step": 13072
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1624009013976056e-05,
+      "loss": 0.7305,
+      "step": 13073
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1622873097183186e-05,
+      "loss": 0.7344,
+      "step": 13074
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1621737158883449e-05,
+      "loss": 0.7148,
+      "step": 13075
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162060119909189e-05,
+      "loss": 0.7041,
+      "step": 13076
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1619465217823569e-05,
+      "loss": 0.7236,
+      "step": 13077
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.161832921509354e-05,
+      "loss": 0.7402,
+      "step": 13078
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1617193190916853e-05,
+      "loss": 0.7256,
+      "step": 13079
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.161605714530857e-05,
+      "loss": 0.7188,
+      "step": 13080
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1614921078283746e-05,
+      "loss": 0.7627,
+      "step": 13081
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.161378498985743e-05,
+      "loss": 0.7451,
+      "step": 13082
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1612648880044683e-05,
+      "loss": 0.7305,
+      "step": 13083
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1611512748860556e-05,
+      "loss": 0.7236,
+      "step": 13084
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1610376596320112e-05,
+      "loss": 0.7441,
+      "step": 13085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1609240422438406e-05,
+      "loss": 0.7549,
+      "step": 13086
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1608104227230492e-05,
+      "loss": 0.7256,
+      "step": 13087
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.160696801071143e-05,
+      "loss": 0.7324,
+      "step": 13088
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1605831772896275e-05,
+      "loss": 0.7393,
+      "step": 13089
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1604695513800086e-05,
+      "loss": 0.7549,
+      "step": 13090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1603559233437924e-05,
+      "loss": 0.7539,
+      "step": 13091
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1602422931824841e-05,
+      "loss": 0.7227,
+      "step": 13092
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1601286608975901e-05,
+      "loss": 0.748,
+      "step": 13093
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.160015026490616e-05,
+      "loss": 0.7451,
+      "step": 13094
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1599013899630682e-05,
+      "loss": 0.7383,
+      "step": 13095
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.159787751316452e-05,
+      "loss": 0.751,
+      "step": 13096
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.159674110552274e-05,
+      "loss": 0.7188,
+      "step": 13097
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1595604676720396e-05,
+      "loss": 0.7178,
+      "step": 13098
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1594468226772553e-05,
+      "loss": 0.7275,
+      "step": 13099
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1593331755694269e-05,
+      "loss": 0.7236,
+      "step": 13100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1592195263500607e-05,
+      "loss": 0.7139,
+      "step": 13101
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1591058750206626e-05,
+      "loss": 0.7217,
+      "step": 13102
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1589922215827387e-05,
+      "loss": 0.7383,
+      "step": 13103
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1588785660377955e-05,
+      "loss": 0.7344,
+      "step": 13104
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1587649083873388e-05,
+      "loss": 0.7344,
+      "step": 13105
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1586512486328752e-05,
+      "loss": 0.7158,
+      "step": 13106
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1585375867759107e-05,
+      "loss": 0.7432,
+      "step": 13107
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1584239228179519e-05,
+      "loss": 0.7227,
+      "step": 13108
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.158310256760505e-05,
+      "loss": 0.7295,
+      "step": 13109
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1581965886050759e-05,
+      "loss": 0.7256,
+      "step": 13110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1580829183531716e-05,
+      "loss": 0.7402,
+      "step": 13111
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157969246006298e-05,
+      "loss": 0.7207,
+      "step": 13112
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157855571565962e-05,
+      "loss": 0.7539,
+      "step": 13113
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1577418950336697e-05,
+      "loss": 0.7256,
+      "step": 13114
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1576282164109276e-05,
+      "loss": 0.7598,
+      "step": 13115
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.157514535699242e-05,
+      "loss": 0.748,
+      "step": 13116
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1574008529001203e-05,
+      "loss": 0.7236,
+      "step": 13117
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1572871680150682e-05,
+      "loss": 0.7207,
+      "step": 13118
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1571734810455926e-05,
+      "loss": 0.7383,
+      "step": 13119
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1570597919932e-05,
+      "loss": 0.748,
+      "step": 13120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1569461008593974e-05,
+      "loss": 0.752,
+      "step": 13121
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1568324076456909e-05,
+      "loss": 0.7314,
+      "step": 13122
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1567187123535875e-05,
+      "loss": 0.7373,
+      "step": 13123
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156605014984594e-05,
+      "loss": 0.7188,
+      "step": 13124
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1564913155402172e-05,
+      "loss": 0.7168,
+      "step": 13125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.156377614021964e-05,
+      "loss": 0.7588,
+      "step": 13126
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1562639104313407e-05,
+      "loss": 0.7402,
+      "step": 13127
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1561502047698544e-05,
+      "loss": 0.708,
+      "step": 13128
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1560364970390121e-05,
+      "loss": 0.7295,
+      "step": 13129
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1559227872403207e-05,
+      "loss": 0.7207,
+      "step": 13130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1558090753752866e-05,
+      "loss": 0.7275,
+      "step": 13131
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1556953614454176e-05,
+      "loss": 0.7305,
+      "step": 13132
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1555816454522202e-05,
+      "loss": 0.7256,
+      "step": 13133
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1554679273972015e-05,
+      "loss": 0.7266,
+      "step": 13134
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1553542072818679e-05,
+      "loss": 0.7383,
+      "step": 13135
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1552404851077274e-05,
+      "loss": 0.7051,
+      "step": 13136
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1551267608762871e-05,
+      "loss": 0.7432,
+      "step": 13137
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1550130345890534e-05,
+      "loss": 0.7275,
+      "step": 13138
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1548993062475336e-05,
+      "loss": 0.7373,
+      "step": 13139
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1547855758532353e-05,
+      "loss": 0.7412,
+      "step": 13140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1546718434076652e-05,
+      "loss": 0.7461,
+      "step": 13141
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1545581089123307e-05,
+      "loss": 0.7451,
+      "step": 13142
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1544443723687395e-05,
+      "loss": 0.7236,
+      "step": 13143
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1543306337783981e-05,
+      "loss": 0.7432,
+      "step": 13144
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1542168931428144e-05,
+      "loss": 0.748,
+      "step": 13145
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1541031504634957e-05,
+      "loss": 0.7422,
+      "step": 13146
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1539894057419486e-05,
+      "loss": 0.7441,
+      "step": 13147
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1538756589796814e-05,
+      "loss": 0.7471,
+      "step": 13148
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.153761910178201e-05,
+      "loss": 0.751,
+      "step": 13149
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1536481593390153e-05,
+      "loss": 0.7275,
+      "step": 13150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1535344064636312e-05,
+      "loss": 0.7324,
+      "step": 13151
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1534206515535565e-05,
+      "loss": 0.7109,
+      "step": 13152
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1533068946102987e-05,
+      "loss": 0.7158,
+      "step": 13153
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1531931356353654e-05,
+      "loss": 0.75,
+      "step": 13154
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1530793746302638e-05,
+      "loss": 0.7441,
+      "step": 13155
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1529656115965023e-05,
+      "loss": 0.7354,
+      "step": 13156
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1528518465355875e-05,
+      "loss": 0.7178,
+      "step": 13157
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152738079449028e-05,
+      "loss": 0.7461,
+      "step": 13158
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1526243103383308e-05,
+      "loss": 0.7588,
+      "step": 13159
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152510539205004e-05,
+      "loss": 0.748,
+      "step": 13160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152396766050555e-05,
+      "loss": 0.7158,
+      "step": 13161
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1522829908764919e-05,
+      "loss": 0.7441,
+      "step": 13162
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1521692136843223e-05,
+      "loss": 0.7354,
+      "step": 13163
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152055434475554e-05,
+      "loss": 0.7197,
+      "step": 13164
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1519416532516951e-05,
+      "loss": 0.7314,
+      "step": 13165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1518278700142534e-05,
+      "loss": 0.7246,
+      "step": 13166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1517140847647363e-05,
+      "loss": 0.7334,
+      "step": 13167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.151600297504652e-05,
+      "loss": 0.7441,
+      "step": 13168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1514865082355089e-05,
+      "loss": 0.7236,
+      "step": 13169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1513727169588144e-05,
+      "loss": 0.7344,
+      "step": 13170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.151258923676077e-05,
+      "loss": 0.7285,
+      "step": 13171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1511451283888041e-05,
+      "loss": 0.7207,
+      "step": 13172
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1510313310985043e-05,
+      "loss": 0.7363,
+      "step": 13173
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1509175318066852e-05,
+      "loss": 0.7471,
+      "step": 13174
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.150803730514855e-05,
+      "loss": 0.7529,
+      "step": 13175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1506899272245226e-05,
+      "loss": 0.7373,
+      "step": 13176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1505761219371954e-05,
+      "loss": 0.7256,
+      "step": 13177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1504623146543814e-05,
+      "loss": 0.7451,
+      "step": 13178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1503485053775897e-05,
+      "loss": 0.7344,
+      "step": 13179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1502346941083275e-05,
+      "loss": 0.7285,
+      "step": 13180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1501208808481038e-05,
+      "loss": 0.7461,
+      "step": 13181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.150007065598427e-05,
+      "loss": 0.7383,
+      "step": 13182
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1498932483608047e-05,
+      "loss": 0.7529,
+      "step": 13183
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1497794291367457e-05,
+      "loss": 0.7344,
+      "step": 13184
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1496656079277585e-05,
+      "loss": 0.7119,
+      "step": 13185
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.149551784735351e-05,
+      "loss": 0.7363,
+      "step": 13186
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1494379595610321e-05,
+      "loss": 0.7412,
+      "step": 13187
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1493241324063099e-05,
+      "loss": 0.7314,
+      "step": 13188
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1492103032726933e-05,
+      "loss": 0.7373,
+      "step": 13189
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1490964721616904e-05,
+      "loss": 0.7207,
+      "step": 13190
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1489826390748099e-05,
+      "loss": 0.7432,
+      "step": 13191
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1488688040135602e-05,
+      "loss": 0.7305,
+      "step": 13192
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1487549669794504e-05,
+      "loss": 0.7422,
+      "step": 13193
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1486411279739884e-05,
+      "loss": 0.75,
+      "step": 13194
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1485272869986834e-05,
+      "loss": 0.7383,
+      "step": 13195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1484134440550437e-05,
+      "loss": 0.7393,
+      "step": 13196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.148299599144578e-05,
+      "loss": 0.7334,
+      "step": 13197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1481857522687952e-05,
+      "loss": 0.7256,
+      "step": 13198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1480719034292041e-05,
+      "loss": 0.7412,
+      "step": 13199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147958052627313e-05,
+      "loss": 0.7275,
+      "step": 13200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1478441998646313e-05,
+      "loss": 0.7178,
+      "step": 13201
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1477303451426676e-05,
+      "loss": 0.7402,
+      "step": 13202
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1476164884629304e-05,
+      "loss": 0.7363,
+      "step": 13203
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147502629826929e-05,
+      "loss": 0.7578,
+      "step": 13204
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147388769236172e-05,
+      "loss": 0.7539,
+      "step": 13205
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1472749066921686e-05,
+      "loss": 0.7354,
+      "step": 13206
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1471610421964275e-05,
+      "loss": 0.7373,
+      "step": 13207
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1470471757504577e-05,
+      "loss": 0.7285,
+      "step": 13208
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1469333073557685e-05,
+      "loss": 0.75,
+      "step": 13209
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1468194370138685e-05,
+      "loss": 0.7607,
+      "step": 13210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146705564726267e-05,
+      "loss": 0.7402,
+      "step": 13211
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1465916904944732e-05,
+      "loss": 0.7285,
+      "step": 13212
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1464778143199955e-05,
+      "loss": 0.7695,
+      "step": 13213
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146363936204344e-05,
+      "loss": 0.7256,
+      "step": 13214
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1462500561490271e-05,
+      "loss": 0.7217,
+      "step": 13215
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1461361741555543e-05,
+      "loss": 0.7129,
+      "step": 13216
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1460222902254348e-05,
+      "loss": 0.7383,
+      "step": 13217
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1459084043601783e-05,
+      "loss": 0.7617,
+      "step": 13218
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1457945165612928e-05,
+      "loss": 0.7246,
+      "step": 13219
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1456806268302887e-05,
+      "loss": 0.7188,
+      "step": 13220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.145566735168675e-05,
+      "loss": 0.7148,
+      "step": 13221
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1454528415779609e-05,
+      "loss": 0.75,
+      "step": 13222
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1453389460596555e-05,
+      "loss": 0.7715,
+      "step": 13223
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1452250486152692e-05,
+      "loss": 0.7383,
+      "step": 13224
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.14511114924631e-05,
+      "loss": 0.7441,
+      "step": 13225
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1449972479542884e-05,
+      "loss": 0.7441,
+      "step": 13226
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1448833447407134e-05,
+      "loss": 0.7295,
+      "step": 13227
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1447694396070947e-05,
+      "loss": 0.7422,
+      "step": 13228
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1446555325549415e-05,
+      "loss": 0.7314,
+      "step": 13229
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1445416235857638e-05,
+      "loss": 0.7354,
+      "step": 13230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.144427712701071e-05,
+      "loss": 0.7305,
+      "step": 13231
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1443137999023723e-05,
+      "loss": 0.7627,
+      "step": 13232
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1441998851911775e-05,
+      "loss": 0.7373,
+      "step": 13233
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.144085968568997e-05,
+      "loss": 0.7207,
+      "step": 13234
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.143972050037339e-05,
+      "loss": 0.6943,
+      "step": 13235
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1438581295977144e-05,
+      "loss": 0.7285,
+      "step": 13236
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1437442072516325e-05,
+      "loss": 0.7373,
+      "step": 13237
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.143630283000603e-05,
+      "loss": 0.7227,
+      "step": 13238
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1435163568461354e-05,
+      "loss": 0.7676,
+      "step": 13239
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1434024287897403e-05,
+      "loss": 0.7441,
+      "step": 13240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1432884988329268e-05,
+      "loss": 0.71,
+      "step": 13241
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1431745669772047e-05,
+      "loss": 0.7109,
+      "step": 13242
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1430606332240844e-05,
+      "loss": 0.7734,
+      "step": 13243
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1429466975750754e-05,
+      "loss": 0.75,
+      "step": 13244
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1428327600316877e-05,
+      "loss": 0.7441,
+      "step": 13245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1427188205954311e-05,
+      "loss": 0.7441,
+      "step": 13246
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1426048792678159e-05,
+      "loss": 0.7256,
+      "step": 13247
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.142490936050352e-05,
+      "loss": 0.7305,
+      "step": 13248
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1423769909445492e-05,
+      "loss": 0.7178,
+      "step": 13249
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1422630439519176e-05,
+      "loss": 0.7363,
+      "step": 13250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1421490950739673e-05,
+      "loss": 0.7744,
+      "step": 13251
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1420351443122084e-05,
+      "loss": 0.7236,
+      "step": 13252
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1419211916681511e-05,
+      "loss": 0.7344,
+      "step": 13253
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1418072371433055e-05,
+      "loss": 0.7305,
+      "step": 13254
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1416932807391815e-05,
+      "loss": 0.7148,
+      "step": 13255
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1415793224572898e-05,
+      "loss": 0.7178,
+      "step": 13256
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1414653622991404e-05,
+      "loss": 0.7109,
+      "step": 13257
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1413514002662431e-05,
+      "loss": 0.7305,
+      "step": 13258
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1412374363601088e-05,
+      "loss": 0.7432,
+      "step": 13259
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1411234705822473e-05,
+      "loss": 0.7256,
+      "step": 13260
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1410095029341695e-05,
+      "loss": 0.7344,
+      "step": 13261
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.140895533417385e-05,
+      "loss": 0.7383,
+      "step": 13262
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1407815620334051e-05,
+      "loss": 0.7197,
+      "step": 13263
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1406675887837388e-05,
+      "loss": 0.75,
+      "step": 13264
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1405536136698981e-05,
+      "loss": 0.7256,
+      "step": 13265
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1404396366933923e-05,
+      "loss": 0.6953,
+      "step": 13266
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1403256578557323e-05,
+      "loss": 0.7109,
+      "step": 13267
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1402116771584285e-05,
+      "loss": 0.7207,
+      "step": 13268
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1400976946029914e-05,
+      "loss": 0.7412,
+      "step": 13269
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1399837101909317e-05,
+      "loss": 0.7334,
+      "step": 13270
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1398697239237597e-05,
+      "loss": 0.7461,
+      "step": 13271
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1397557358029861e-05,
+      "loss": 0.7129,
+      "step": 13272
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.139641745830122e-05,
+      "loss": 0.7471,
+      "step": 13273
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1395277540066768e-05,
+      "loss": 0.7676,
+      "step": 13274
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1394137603341624e-05,
+      "loss": 0.7207,
+      "step": 13275
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.139299764814089e-05,
+      "loss": 0.7461,
+      "step": 13276
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1391857674479672e-05,
+      "loss": 0.7314,
+      "step": 13277
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1390717682373075e-05,
+      "loss": 0.7383,
+      "step": 13278
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1389577671836219e-05,
+      "loss": 0.7148,
+      "step": 13279
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1388437642884194e-05,
+      "loss": 0.7188,
+      "step": 13280
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1387297595532121e-05,
+      "loss": 0.7334,
+      "step": 13281
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1386157529795104e-05,
+      "loss": 0.7344,
+      "step": 13282
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.138501744568825e-05,
+      "loss": 0.7344,
+      "step": 13283
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.138387734322667e-05,
+      "loss": 0.7285,
+      "step": 13284
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1382737222425475e-05,
+      "loss": 0.7256,
+      "step": 13285
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.138159708329977e-05,
+      "loss": 0.7402,
+      "step": 13286
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1380456925864667e-05,
+      "loss": 0.7422,
+      "step": 13287
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1379316750135273e-05,
+      "loss": 0.7471,
+      "step": 13288
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1378176556126703e-05,
+      "loss": 0.7383,
+      "step": 13289
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1377036343854065e-05,
+      "loss": 0.7188,
+      "step": 13290
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1375896113332463e-05,
+      "loss": 0.7285,
+      "step": 13291
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1374755864577019e-05,
+      "loss": 0.7236,
+      "step": 13292
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1373615597602835e-05,
+      "loss": 0.7246,
+      "step": 13293
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.137247531242503e-05,
+      "loss": 0.7373,
+      "step": 13294
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1371335009058705e-05,
+      "loss": 0.7109,
+      "step": 13295
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1370194687518985e-05,
+      "loss": 0.7422,
+      "step": 13296
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.136905434782097e-05,
+      "loss": 0.7285,
+      "step": 13297
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1367913989979778e-05,
+      "loss": 0.7354,
+      "step": 13298
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.136677361401052e-05,
+      "loss": 0.7305,
+      "step": 13299
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1365633219928312e-05,
+      "loss": 0.7432,
+      "step": 13300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.136449280774826e-05,
+      "loss": 0.7354,
+      "step": 13301
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1363352377485484e-05,
+      "loss": 0.7412,
+      "step": 13302
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1362211929155092e-05,
+      "loss": 0.7432,
+      "step": 13303
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.13610714627722e-05,
+      "loss": 0.7422,
+      "step": 13304
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1359930978351922e-05,
+      "loss": 0.7012,
+      "step": 13305
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1358790475909376e-05,
+      "loss": 0.7197,
+      "step": 13306
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1357649955459667e-05,
+      "loss": 0.7021,
+      "step": 13307
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1356509417017918e-05,
+      "loss": 0.7734,
+      "step": 13308
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.135536886059924e-05,
+      "loss": 0.7246,
+      "step": 13309
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1354228286218747e-05,
+      "loss": 0.7197,
+      "step": 13310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1353087693891556e-05,
+      "loss": 0.7598,
+      "step": 13311
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1351947083632786e-05,
+      "loss": 0.7637,
+      "step": 13312
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1350806455457545e-05,
+      "loss": 0.7246,
+      "step": 13313
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1349665809380959e-05,
+      "loss": 0.7686,
+      "step": 13314
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1348525145418135e-05,
+      "loss": 0.7197,
+      "step": 13315
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1347384463584192e-05,
+      "loss": 0.7598,
+      "step": 13316
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1346243763894249e-05,
+      "loss": 0.7109,
+      "step": 13317
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1345103046363423e-05,
+      "loss": 0.7451,
+      "step": 13318
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1343962311006827e-05,
+      "loss": 0.7236,
+      "step": 13319
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1342821557839584e-05,
+      "loss": 0.7412,
+      "step": 13320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1341680786876807e-05,
+      "loss": 0.7529,
+      "step": 13321
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1340539998133617e-05,
+      "loss": 0.7412,
+      "step": 13322
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1339399191625127e-05,
+      "loss": 0.7441,
+      "step": 13323
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1338258367366464e-05,
+      "loss": 0.7393,
+      "step": 13324
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1337117525372741e-05,
+      "loss": 0.7363,
+      "step": 13325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1335976665659077e-05,
+      "loss": 0.7393,
+      "step": 13326
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1334835788240593e-05,
+      "loss": 0.7168,
+      "step": 13327
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1333694893132405e-05,
+      "loss": 0.7568,
+      "step": 13328
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1332553980349635e-05,
+      "loss": 0.749,
+      "step": 13329
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13314130499074e-05,
+      "loss": 0.7168,
+      "step": 13330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1330272101820827e-05,
+      "loss": 0.7334,
+      "step": 13331
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132913113610503e-05,
+      "loss": 0.7285,
+      "step": 13332
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132799015277513e-05,
+      "loss": 0.7471,
+      "step": 13333
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1326849151846248e-05,
+      "loss": 0.7402,
+      "step": 13334
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132570813333351e-05,
+      "loss": 0.7314,
+      "step": 13335
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1324567097252028e-05,
+      "loss": 0.7334,
+      "step": 13336
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1323426043616932e-05,
+      "loss": 0.7275,
+      "step": 13337
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1322284972443339e-05,
+      "loss": 0.7305,
+      "step": 13338
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132114388374637e-05,
+      "loss": 0.7334,
+      "step": 13339
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1320002777541148e-05,
+      "loss": 0.7256,
+      "step": 13340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13188616538428e-05,
+      "loss": 0.7334,
+      "step": 13341
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1317720512666444e-05,
+      "loss": 0.7168,
+      "step": 13342
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1316579354027202e-05,
+      "loss": 0.7432,
+      "step": 13343
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1315438177940198e-05,
+      "loss": 0.7373,
+      "step": 13344
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1314296984420558e-05,
+      "loss": 0.7275,
+      "step": 13345
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13131557734834e-05,
+      "loss": 0.7266,
+      "step": 13346
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1312014545143857e-05,
+      "loss": 0.752,
+      "step": 13347
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.131087329941704e-05,
+      "loss": 0.7295,
+      "step": 13348
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1309732036318085e-05,
+      "loss": 0.7461,
+      "step": 13349
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1308590755862108e-05,
+      "loss": 0.7559,
+      "step": 13350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130744945806424e-05,
+      "loss": 0.7363,
+      "step": 13351
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1306308142939602e-05,
+      "loss": 0.7334,
+      "step": 13352
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130516681050332e-05,
+      "loss": 0.7275,
+      "step": 13353
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.130402546077052e-05,
+      "loss": 0.7549,
+      "step": 13354
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1302884093756328e-05,
+      "loss": 0.7412,
+      "step": 13355
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1301742709475868e-05,
+      "loss": 0.7158,
+      "step": 13356
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1300601307944269e-05,
+      "loss": 0.7363,
+      "step": 13357
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.129945988917665e-05,
+      "loss": 0.7549,
+      "step": 13358
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1298318453188147e-05,
+      "loss": 0.7129,
+      "step": 13359
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1297176999993882e-05,
+      "loss": 0.7148,
+      "step": 13360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1296035529608979e-05,
+      "loss": 0.7539,
+      "step": 13361
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.129489404204857e-05,
+      "loss": 0.7422,
+      "step": 13362
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1293752537327782e-05,
+      "loss": 0.7363,
+      "step": 13363
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.129261101546174e-05,
+      "loss": 0.7383,
+      "step": 13364
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1291469476465576e-05,
+      "loss": 0.7256,
+      "step": 13365
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1290327920354414e-05,
+      "loss": 0.7393,
+      "step": 13366
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1289186347143382e-05,
+      "loss": 0.7266,
+      "step": 13367
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.128804475684761e-05,
+      "loss": 0.7334,
+      "step": 13368
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1286903149482228e-05,
+      "loss": 0.7188,
+      "step": 13369
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1285761525062364e-05,
+      "loss": 0.75,
+      "step": 13370
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1284619883603146e-05,
+      "loss": 0.7217,
+      "step": 13371
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1283478225119707e-05,
+      "loss": 0.7314,
+      "step": 13372
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1282336549627173e-05,
+      "loss": 0.7383,
+      "step": 13373
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1281194857140674e-05,
+      "loss": 0.7314,
+      "step": 13374
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1280053147675338e-05,
+      "loss": 0.7412,
+      "step": 13375
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1278911421246303e-05,
+      "loss": 0.7451,
+      "step": 13376
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1277769677868695e-05,
+      "loss": 0.7393,
+      "step": 13377
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1276627917557643e-05,
+      "loss": 0.7559,
+      "step": 13378
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1275486140328278e-05,
+      "loss": 0.7451,
+      "step": 13379
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1274344346195739e-05,
+      "loss": 0.7168,
+      "step": 13380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1273202535175145e-05,
+      "loss": 0.7275,
+      "step": 13381
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1272060707281636e-05,
+      "loss": 0.7441,
+      "step": 13382
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1270918862530345e-05,
+      "loss": 0.7422,
+      "step": 13383
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1269777000936397e-05,
+      "loss": 0.7422,
+      "step": 13384
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1268635122514928e-05,
+      "loss": 0.7314,
+      "step": 13385
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1267493227281077e-05,
+      "loss": 0.7461,
+      "step": 13386
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1266351315249964e-05,
+      "loss": 0.7266,
+      "step": 13387
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1265209386436734e-05,
+      "loss": 0.7236,
+      "step": 13388
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.126406744085651e-05,
+      "loss": 0.7266,
+      "step": 13389
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1262925478524435e-05,
+      "loss": 0.7168,
+      "step": 13390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1261783499455633e-05,
+      "loss": 0.7207,
+      "step": 13391
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1260641503665248e-05,
+      "loss": 0.708,
+      "step": 13392
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1259499491168409e-05,
+      "loss": 0.7451,
+      "step": 13393
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.125835746198025e-05,
+      "loss": 0.7256,
+      "step": 13394
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1257215416115903e-05,
+      "loss": 0.7334,
+      "step": 13395
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1256073353590509e-05,
+      "loss": 0.7422,
+      "step": 13396
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.12549312744192e-05,
+      "loss": 0.752,
+      "step": 13397
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1253789178617108e-05,
+      "loss": 0.7051,
+      "step": 13398
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1252647066199374e-05,
+      "loss": 0.748,
+      "step": 13399
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.125150493718113e-05,
+      "loss": 0.7139,
+      "step": 13400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1250362791577513e-05,
+      "loss": 0.748,
+      "step": 13401
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1249220629403661e-05,
+      "loss": 0.7061,
+      "step": 13402
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1248078450674706e-05,
+      "loss": 0.7363,
+      "step": 13403
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.124693625540579e-05,
+      "loss": 0.7305,
+      "step": 13404
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1245794043612046e-05,
+      "loss": 0.7256,
+      "step": 13405
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1244651815308611e-05,
+      "loss": 0.7158,
+      "step": 13406
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1243509570510622e-05,
+      "loss": 0.7266,
+      "step": 13407
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1242367309233217e-05,
+      "loss": 0.7412,
+      "step": 13408
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1241225031491538e-05,
+      "loss": 0.7324,
+      "step": 13409
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1240082737300717e-05,
+      "loss": 0.7227,
+      "step": 13410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1238940426675892e-05,
+      "loss": 0.749,
+      "step": 13411
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1237798099632207e-05,
+      "loss": 0.7393,
+      "step": 13412
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1236655756184793e-05,
+      "loss": 0.7588,
+      "step": 13413
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1235513396348793e-05,
+      "loss": 0.7393,
+      "step": 13414
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1234371020139348e-05,
+      "loss": 0.7617,
+      "step": 13415
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1233228627571596e-05,
+      "loss": 0.75,
+      "step": 13416
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1232086218660671e-05,
+      "loss": 0.7256,
+      "step": 13417
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1230943793421715e-05,
+      "loss": 0.7432,
+      "step": 13418
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1229801351869875e-05,
+      "loss": 0.7256,
+      "step": 13419
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.122865889402028e-05,
+      "loss": 0.7236,
+      "step": 13420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.122751641988808e-05,
+      "loss": 0.7188,
+      "step": 13421
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.122637392948841e-05,
+      "loss": 0.7461,
+      "step": 13422
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1225231422836413e-05,
+      "loss": 0.7617,
+      "step": 13423
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1224088899947225e-05,
+      "loss": 0.7158,
+      "step": 13424
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1222946360835996e-05,
+      "loss": 0.7529,
+      "step": 13425
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1221803805517856e-05,
+      "loss": 0.7188,
+      "step": 13426
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1220661234007957e-05,
+      "loss": 0.7305,
+      "step": 13427
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1219518646321434e-05,
+      "loss": 0.7441,
+      "step": 13428
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1218376042473436e-05,
+      "loss": 0.7266,
+      "step": 13429
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1217233422479094e-05,
+      "loss": 0.7178,
+      "step": 13430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1216090786353562e-05,
+      "loss": 0.7227,
+      "step": 13431
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1214948134111976e-05,
+      "loss": 0.752,
+      "step": 13432
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.121380546576948e-05,
+      "loss": 0.7295,
+      "step": 13433
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1212662781341215e-05,
+      "loss": 0.7529,
+      "step": 13434
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1211520080842333e-05,
+      "loss": 0.7158,
+      "step": 13435
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1210377364287965e-05,
+      "loss": 0.7002,
+      "step": 13436
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1209234631693264e-05,
+      "loss": 0.752,
+      "step": 13437
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1208091883073368e-05,
+      "loss": 0.7793,
+      "step": 13438
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120694911844343e-05,
+      "loss": 0.7227,
+      "step": 13439
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120580633781858e-05,
+      "loss": 0.7422,
+      "step": 13440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1204663541213973e-05,
+      "loss": 0.7207,
+      "step": 13441
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1203520728644755e-05,
+      "loss": 0.7207,
+      "step": 13442
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1202377900126065e-05,
+      "loss": 0.7266,
+      "step": 13443
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1201235055673048e-05,
+      "loss": 0.7285,
+      "step": 13444
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1200092195300857e-05,
+      "loss": 0.7549,
+      "step": 13445
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1198949319024629e-05,
+      "loss": 0.7471,
+      "step": 13446
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1197806426859514e-05,
+      "loss": 0.7188,
+      "step": 13447
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1196663518820656e-05,
+      "loss": 0.7393,
+      "step": 13448
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1195520594923204e-05,
+      "loss": 0.7646,
+      "step": 13449
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1194377655182302e-05,
+      "loss": 0.7266,
+      "step": 13450
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1193234699613098e-05,
+      "loss": 0.7451,
+      "step": 13451
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1192091728230736e-05,
+      "loss": 0.7432,
+      "step": 13452
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.119094874105037e-05,
+      "loss": 0.7227,
+      "step": 13453
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1189805738087139e-05,
+      "loss": 0.7305,
+      "step": 13454
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1188662719356196e-05,
+      "loss": 0.7109,
+      "step": 13455
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1187519684872685e-05,
+      "loss": 0.748,
+      "step": 13456
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1186376634651754e-05,
+      "loss": 0.7246,
+      "step": 13457
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1185233568708557e-05,
+      "loss": 0.7617,
+      "step": 13458
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1184090487058235e-05,
+      "loss": 0.7139,
+      "step": 13459
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1182947389715942e-05,
+      "loss": 0.7295,
+      "step": 13460
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1181804276696822e-05,
+      "loss": 0.7432,
+      "step": 13461
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1180661148016027e-05,
+      "loss": 0.7393,
+      "step": 13462
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1179518003688703e-05,
+      "loss": 0.7324,
+      "step": 13463
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1178374843730007e-05,
+      "loss": 0.7256,
+      "step": 13464
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1177231668155078e-05,
+      "loss": 0.7344,
+      "step": 13465
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1176088476979072e-05,
+      "loss": 0.7402,
+      "step": 13466
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1174945270217139e-05,
+      "loss": 0.7383,
+      "step": 13467
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1173802047884428e-05,
+      "loss": 0.7373,
+      "step": 13468
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1172658809996088e-05,
+      "loss": 0.7324,
+      "step": 13469
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1171515556567271e-05,
+      "loss": 0.7402,
+      "step": 13470
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1170372287613131e-05,
+      "loss": 0.7266,
+      "step": 13471
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1169229003148814e-05,
+      "loss": 0.7334,
+      "step": 13472
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1168085703189471e-05,
+      "loss": 0.7383,
+      "step": 13473
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.116694238775026e-05,
+      "loss": 0.7285,
+      "step": 13474
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1165799056846322e-05,
+      "loss": 0.7207,
+      "step": 13475
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.116465571049282e-05,
+      "loss": 0.7354,
+      "step": 13476
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.11635123487049e-05,
+      "loss": 0.7246,
+      "step": 13477
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1162368971497714e-05,
+      "loss": 0.708,
+      "step": 13478
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1161225578886415e-05,
+      "loss": 0.7188,
+      "step": 13479
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1160082170886157e-05,
+      "loss": 0.7568,
+      "step": 13480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1158938747512092e-05,
+      "loss": 0.709,
+      "step": 13481
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.115779530877937e-05,
+      "loss": 0.7285,
+      "step": 13482
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.115665185470315e-05,
+      "loss": 0.7324,
+      "step": 13483
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1155508385298586e-05,
+      "loss": 0.7461,
+      "step": 13484
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1154364900580821e-05,
+      "loss": 0.7168,
+      "step": 13485
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1153221400565023e-05,
+      "loss": 0.7334,
+      "step": 13486
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1152077885266334e-05,
+      "loss": 0.707,
+      "step": 13487
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1150934354699915e-05,
+      "loss": 0.7393,
+      "step": 13488
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114979080888092e-05,
+      "loss": 0.7607,
+      "step": 13489
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1148647247824504e-05,
+      "loss": 0.7295,
+      "step": 13490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1147503671545818e-05,
+      "loss": 0.708,
+      "step": 13491
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1146360080060022e-05,
+      "loss": 0.7129,
+      "step": 13492
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1145216473382266e-05,
+      "loss": 0.749,
+      "step": 13493
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114407285152771e-05,
+      "loss": 0.7295,
+      "step": 13494
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1142929214511508e-05,
+      "loss": 0.751,
+      "step": 13495
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1141785562348816e-05,
+      "loss": 0.7148,
+      "step": 13496
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1140641895054788e-05,
+      "loss": 0.7539,
+      "step": 13497
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1139498212644584e-05,
+      "loss": 0.7275,
+      "step": 13498
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1138354515133359e-05,
+      "loss": 0.7236,
+      "step": 13499
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1137210802536267e-05,
+      "loss": 0.7422,
+      "step": 13500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1136067074868468e-05,
+      "loss": 0.7168,
+      "step": 13501
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1134923332145117e-05,
+      "loss": 0.7227,
+      "step": 13502
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1133779574381379e-05,
+      "loss": 0.7256,
+      "step": 13503
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1132635801592398e-05,
+      "loss": 0.7139,
+      "step": 13504
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1131492013793342e-05,
+      "loss": 0.7607,
+      "step": 13505
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1130348210999365e-05,
+      "loss": 0.7422,
+      "step": 13506
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1129204393225623e-05,
+      "loss": 0.7207,
+      "step": 13507
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1128060560487276e-05,
+      "loss": 0.7363,
+      "step": 13508
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.112691671279949e-05,
+      "loss": 0.7432,
+      "step": 13509
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1125772850177413e-05,
+      "loss": 0.7227,
+      "step": 13510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1124628972636209e-05,
+      "loss": 0.7266,
+      "step": 13511
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1123485080191032e-05,
+      "loss": 0.7217,
+      "step": 13512
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1122341172857052e-05,
+      "loss": 0.7334,
+      "step": 13513
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1121197250649415e-05,
+      "loss": 0.7227,
+      "step": 13514
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.112005331358329e-05,
+      "loss": 0.6992,
+      "step": 13515
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1118909361673835e-05,
+      "loss": 0.7393,
+      "step": 13516
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1117765394936208e-05,
+      "loss": 0.7412,
+      "step": 13517
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1116621413385569e-05,
+      "loss": 0.7188,
+      "step": 13518
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1115477417037086e-05,
+      "loss": 0.7295,
+      "step": 13519
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1114333405905906e-05,
+      "loss": 0.7207,
+      "step": 13520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.11131893800072e-05,
+      "loss": 0.7256,
+      "step": 13521
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1112045339356126e-05,
+      "loss": 0.748,
+      "step": 13522
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1110901283967848e-05,
+      "loss": 0.7598,
+      "step": 13523
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1109757213857522e-05,
+      "loss": 0.707,
+      "step": 13524
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1108613129040314e-05,
+      "loss": 0.7344,
+      "step": 13525
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1107469029531384e-05,
+      "loss": 0.748,
+      "step": 13526
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1106324915345897e-05,
+      "loss": 0.708,
+      "step": 13527
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1105180786499006e-05,
+      "loss": 0.7227,
+      "step": 13528
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1104036643005886e-05,
+      "loss": 0.7119,
+      "step": 13529
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1102892484881691e-05,
+      "loss": 0.7324,
+      "step": 13530
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1101748312141587e-05,
+      "loss": 0.7188,
+      "step": 13531
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1100604124800736e-05,
+      "loss": 0.7344,
+      "step": 13532
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1099459922874303e-05,
+      "loss": 0.7344,
+      "step": 13533
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1098315706377447e-05,
+      "loss": 0.7412,
+      "step": 13534
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1097171475325335e-05,
+      "loss": 0.7451,
+      "step": 13535
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.109602722973313e-05,
+      "loss": 0.7559,
+      "step": 13536
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1094882969616e-05,
+      "loss": 0.7461,
+      "step": 13537
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1093738694989101e-05,
+      "loss": 0.7432,
+      "step": 13538
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1092594405867603e-05,
+      "loss": 0.7432,
+      "step": 13539
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1091450102266667e-05,
+      "loss": 0.7295,
+      "step": 13540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1090305784201461e-05,
+      "loss": 0.7285,
+      "step": 13541
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108916145168715e-05,
+      "loss": 0.71,
+      "step": 13542
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1088017104738895e-05,
+      "loss": 0.7314,
+      "step": 13543
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1086872743371867e-05,
+      "loss": 0.7246,
+      "step": 13544
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1085728367601225e-05,
+      "loss": 0.7344,
+      "step": 13545
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108458397744214e-05,
+      "loss": 0.749,
+      "step": 13546
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1083439572909775e-05,
+      "loss": 0.7285,
+      "step": 13547
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1082295154019298e-05,
+      "loss": 0.7451,
+      "step": 13548
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1081150720785874e-05,
+      "loss": 0.7285,
+      "step": 13549
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108000627322467e-05,
+      "loss": 0.7188,
+      "step": 13550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1078861811350849e-05,
+      "loss": 0.7148,
+      "step": 13551
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1077717335179588e-05,
+      "loss": 0.7314,
+      "step": 13552
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1076572844726041e-05,
+      "loss": 0.7227,
+      "step": 13553
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1075428340005384e-05,
+      "loss": 0.7051,
+      "step": 13554
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1074283821032778e-05,
+      "loss": 0.7305,
+      "step": 13555
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1073139287823397e-05,
+      "loss": 0.7363,
+      "step": 13556
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1071994740392404e-05,
+      "loss": 0.7246,
+      "step": 13557
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1070850178754975e-05,
+      "loss": 0.7168,
+      "step": 13558
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1069705602926267e-05,
+      "loss": 0.7305,
+      "step": 13559
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1068561012921453e-05,
+      "loss": 0.7412,
+      "step": 13560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1067416408755702e-05,
+      "loss": 0.7363,
+      "step": 13561
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1066271790444182e-05,
+      "loss": 0.7217,
+      "step": 13562
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1065127158002062e-05,
+      "loss": 0.7441,
+      "step": 13563
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1063982511444513e-05,
+      "loss": 0.7188,
+      "step": 13564
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1062837850786702e-05,
+      "loss": 0.7051,
+      "step": 13565
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.10616931760438e-05,
+      "loss": 0.7305,
+      "step": 13566
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1060548487230973e-05,
+      "loss": 0.7295,
+      "step": 13567
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1059403784363398e-05,
+      "loss": 0.7207,
+      "step": 13568
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1058259067456237e-05,
+      "loss": 0.7051,
+      "step": 13569
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1057114336524663e-05,
+      "loss": 0.7246,
+      "step": 13570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.105596959158385e-05,
+      "loss": 0.7158,
+      "step": 13571
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1054824832648964e-05,
+      "loss": 0.7227,
+      "step": 13572
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1053680059735177e-05,
+      "loss": 0.7373,
+      "step": 13573
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1052535272857663e-05,
+      "loss": 0.7461,
+      "step": 13574
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1051390472031587e-05,
+      "loss": 0.7627,
+      "step": 13575
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1050245657272125e-05,
+      "loss": 0.7217,
+      "step": 13576
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1049100828594445e-05,
+      "loss": 0.7207,
+      "step": 13577
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1047955986013725e-05,
+      "loss": 0.7119,
+      "step": 13578
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.104681112954513e-05,
+      "loss": 0.7119,
+      "step": 13579
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1045666259203834e-05,
+      "loss": 0.7129,
+      "step": 13580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1044521375005011e-05,
+      "loss": 0.751,
+      "step": 13581
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1043376476963832e-05,
+      "loss": 0.708,
+      "step": 13582
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1042231565095472e-05,
+      "loss": 0.7256,
+      "step": 13583
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1041086639415099e-05,
+      "loss": 0.7354,
+      "step": 13584
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1039941699937891e-05,
+      "loss": 0.7402,
+      "step": 13585
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1038796746679015e-05,
+      "loss": 0.7148,
+      "step": 13586
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.103765177965365e-05,
+      "loss": 0.7441,
+      "step": 13587
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1036506798876968e-05,
+      "loss": 0.7275,
+      "step": 13588
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1035361804364141e-05,
+      "loss": 0.7344,
+      "step": 13589
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1034216796130344e-05,
+      "loss": 0.7275,
+      "step": 13590
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1033071774190753e-05,
+      "loss": 0.7383,
+      "step": 13591
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1031926738560539e-05,
+      "loss": 0.7373,
+      "step": 13592
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1030781689254878e-05,
+      "loss": 0.7217,
+      "step": 13593
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1029636626288945e-05,
+      "loss": 0.7451,
+      "step": 13594
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1028491549677913e-05,
+      "loss": 0.7764,
+      "step": 13595
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1027346459436957e-05,
+      "loss": 0.7383,
+      "step": 13596
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1026201355581256e-05,
+      "loss": 0.749,
+      "step": 13597
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.102505623812598e-05,
+      "loss": 0.7207,
+      "step": 13598
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1023911107086308e-05,
+      "loss": 0.708,
+      "step": 13599
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1022765962477415e-05,
+      "loss": 0.7256,
+      "step": 13600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1021620804314476e-05,
+      "loss": 0.7939,
+      "step": 13601
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1020475632612666e-05,
+      "loss": 0.7393,
+      "step": 13602
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1019330447387165e-05,
+      "loss": 0.7305,
+      "step": 13603
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1018185248653147e-05,
+      "loss": 0.7383,
+      "step": 13604
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1017040036425787e-05,
+      "loss": 0.7197,
+      "step": 13605
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1015894810720265e-05,
+      "loss": 0.7305,
+      "step": 13606
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1014749571551757e-05,
+      "loss": 0.7363,
+      "step": 13607
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1013604318935437e-05,
+      "loss": 0.7402,
+      "step": 13608
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1012459052886485e-05,
+      "loss": 0.7197,
+      "step": 13609
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.101131377342008e-05,
+      "loss": 0.7461,
+      "step": 13610
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1010168480551397e-05,
+      "loss": 0.7217,
+      "step": 13611
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1009023174295615e-05,
+      "loss": 0.7686,
+      "step": 13612
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1007877854667911e-05,
+      "loss": 0.7451,
+      "step": 13613
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1006732521683464e-05,
+      "loss": 0.7354,
+      "step": 13614
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1005587175357453e-05,
+      "loss": 0.7451,
+      "step": 13615
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1004441815705054e-05,
+      "loss": 0.7168,
+      "step": 13616
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.100329644274145e-05,
+      "loss": 0.7432,
+      "step": 13617
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1002151056481814e-05,
+      "loss": 0.7285,
+      "step": 13618
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1001005656941328e-05,
+      "loss": 0.7227,
+      "step": 13619
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0999860244135173e-05,
+      "loss": 0.7402,
+      "step": 13620
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0998714818078529e-05,
+      "loss": 0.7393,
+      "step": 13621
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0997569378786573e-05,
+      "loss": 0.7363,
+      "step": 13622
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0996423926274482e-05,
+      "loss": 0.752,
+      "step": 13623
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.099527846055744e-05,
+      "loss": 0.7275,
+      "step": 13624
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0994132981650628e-05,
+      "loss": 0.7412,
+      "step": 13625
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0992987489569225e-05,
+      "loss": 0.7217,
+      "step": 13626
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0991841984328408e-05,
+      "loss": 0.7256,
+      "step": 13627
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0990696465943365e-05,
+      "loss": 0.7363,
+      "step": 13628
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0989550934429267e-05,
+      "loss": 0.7549,
+      "step": 13629
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0988405389801306e-05,
+      "loss": 0.7646,
+      "step": 13630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0987259832074656e-05,
+      "loss": 0.7197,
+      "step": 13631
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0986114261264498e-05,
+      "loss": 0.7148,
+      "step": 13632
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0984968677386017e-05,
+      "loss": 0.7139,
+      "step": 13633
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0983823080454392e-05,
+      "loss": 0.7344,
+      "step": 13634
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0982677470484804e-05,
+      "loss": 0.7627,
+      "step": 13635
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0981531847492442e-05,
+      "loss": 0.7129,
+      "step": 13636
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.098038621149248e-05,
+      "loss": 0.7314,
+      "step": 13637
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0979240562500104e-05,
+      "loss": 0.7393,
+      "step": 13638
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0978094900530496e-05,
+      "loss": 0.7363,
+      "step": 13639
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0976949225598839e-05,
+      "loss": 0.749,
+      "step": 13640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0975803537720314e-05,
+      "loss": 0.7432,
+      "step": 13641
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0974657836910105e-05,
+      "loss": 0.7412,
+      "step": 13642
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0973512123183399e-05,
+      "loss": 0.7314,
+      "step": 13643
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0972366396555376e-05,
+      "loss": 0.7227,
+      "step": 13644
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0971220657041217e-05,
+      "loss": 0.7373,
+      "step": 13645
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0970074904656112e-05,
+      "loss": 0.7266,
+      "step": 13646
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0968929139415238e-05,
+      "loss": 0.7295,
+      "step": 13647
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0967783361333786e-05,
+      "loss": 0.7314,
+      "step": 13648
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0966637570426934e-05,
+      "loss": 0.7197,
+      "step": 13649
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0965491766709872e-05,
+      "loss": 0.7451,
+      "step": 13650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0964345950197777e-05,
+      "loss": 0.7178,
+      "step": 13651
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0963200120905846e-05,
+      "loss": 0.7207,
+      "step": 13652
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.096205427884925e-05,
+      "loss": 0.7197,
+      "step": 13653
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0960908424043181e-05,
+      "loss": 0.7588,
+      "step": 13654
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0959762556502827e-05,
+      "loss": 0.7158,
+      "step": 13655
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.095861667624337e-05,
+      "loss": 0.7207,
+      "step": 13656
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.095747078327999e-05,
+      "loss": 0.7578,
+      "step": 13657
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0956324877627887e-05,
+      "loss": 0.7383,
+      "step": 13658
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0955178959302233e-05,
+      "loss": 0.7666,
+      "step": 13659
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0954033028318223e-05,
+      "loss": 0.7285,
+      "step": 13660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0952887084691037e-05,
+      "loss": 0.7227,
+      "step": 13661
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0951741128435866e-05,
+      "loss": 0.7344,
+      "step": 13662
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0950595159567896e-05,
+      "loss": 0.748,
+      "step": 13663
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0949449178102309e-05,
+      "loss": 0.7266,
+      "step": 13664
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09483031840543e-05,
+      "loss": 0.7334,
+      "step": 13665
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0947157177439047e-05,
+      "loss": 0.7363,
+      "step": 13666
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0946011158271745e-05,
+      "loss": 0.7451,
+      "step": 13667
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0944865126567578e-05,
+      "loss": 0.7188,
+      "step": 13668
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0943719082341737e-05,
+      "loss": 0.75,
+      "step": 13669
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0942573025609403e-05,
+      "loss": 0.7383,
+      "step": 13670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0941426956385768e-05,
+      "loss": 0.707,
+      "step": 13671
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.094028087468602e-05,
+      "loss": 0.7383,
+      "step": 13672
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093913478052535e-05,
+      "loss": 0.7529,
+      "step": 13673
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093798867391894e-05,
+      "loss": 0.7402,
+      "step": 13674
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0936842554881985e-05,
+      "loss": 0.7168,
+      "step": 13675
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093569642342967e-05,
+      "loss": 0.7529,
+      "step": 13676
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0934550279577184e-05,
+      "loss": 0.7246,
+      "step": 13677
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0933404123339717e-05,
+      "loss": 0.7295,
+      "step": 13678
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0932257954732459e-05,
+      "loss": 0.752,
+      "step": 13679
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0931111773770597e-05,
+      "loss": 0.752,
+      "step": 13680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0929965580469325e-05,
+      "loss": 0.7412,
+      "step": 13681
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0928819374843829e-05,
+      "loss": 0.7812,
+      "step": 13682
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09276731569093e-05,
+      "loss": 0.7402,
+      "step": 13683
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0926526926680925e-05,
+      "loss": 0.7363,
+      "step": 13684
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0925380684173903e-05,
+      "loss": 0.7178,
+      "step": 13685
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0924234429403413e-05,
+      "loss": 0.7441,
+      "step": 13686
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0923088162384653e-05,
+      "loss": 0.7314,
+      "step": 13687
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0921941883132814e-05,
+      "loss": 0.7295,
+      "step": 13688
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0920795591663085e-05,
+      "loss": 0.7383,
+      "step": 13689
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0919649287990652e-05,
+      "loss": 0.7314,
+      "step": 13690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0918502972130717e-05,
+      "loss": 0.7314,
+      "step": 13691
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.091735664409846e-05,
+      "loss": 0.7178,
+      "step": 13692
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0916210303909084e-05,
+      "loss": 0.7334,
+      "step": 13693
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0915063951577769e-05,
+      "loss": 0.75,
+      "step": 13694
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0913917587119716e-05,
+      "loss": 0.7432,
+      "step": 13695
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0912771210550111e-05,
+      "loss": 0.7139,
+      "step": 13696
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0911624821884147e-05,
+      "loss": 0.7578,
+      "step": 13697
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0910478421137022e-05,
+      "loss": 0.7578,
+      "step": 13698
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0909332008323925e-05,
+      "loss": 0.7227,
+      "step": 13699
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0908185583460047e-05,
+      "loss": 0.752,
+      "step": 13700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.090703914656058e-05,
+      "loss": 0.7295,
+      "step": 13701
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0905892697640721e-05,
+      "loss": 0.7314,
+      "step": 13702
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.090474623671566e-05,
+      "loss": 0.7158,
+      "step": 13703
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0903599763800592e-05,
+      "loss": 0.7363,
+      "step": 13704
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.090245327891071e-05,
+      "loss": 0.7451,
+      "step": 13705
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0901306782061207e-05,
+      "loss": 0.7324,
+      "step": 13706
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.090016027326728e-05,
+      "loss": 0.7344,
+      "step": 13707
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0899013752544117e-05,
+      "loss": 0.7324,
+      "step": 13708
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0897867219906913e-05,
+      "loss": 0.7334,
+      "step": 13709
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0896720675370868e-05,
+      "loss": 0.7598,
+      "step": 13710
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0895574118951173e-05,
+      "loss": 0.7168,
+      "step": 13711
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.089442755066302e-05,
+      "loss": 0.7139,
+      "step": 13712
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0893280970521608e-05,
+      "loss": 0.7402,
+      "step": 13713
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0892134378542131e-05,
+      "loss": 0.7295,
+      "step": 13714
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.089098777473978e-05,
+      "loss": 0.7168,
+      "step": 13715
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0889841159129756e-05,
+      "loss": 0.7256,
+      "step": 13716
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.088869453172725e-05,
+      "loss": 0.748,
+      "step": 13717
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.088754789254746e-05,
+      "loss": 0.707,
+      "step": 13718
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.088640124160558e-05,
+      "loss": 0.7363,
+      "step": 13719
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0885254578916809e-05,
+      "loss": 0.7129,
+      "step": 13720
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0884107904496337e-05,
+      "loss": 0.7344,
+      "step": 13721
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0882961218359367e-05,
+      "loss": 0.7139,
+      "step": 13722
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0881814520521089e-05,
+      "loss": 0.749,
+      "step": 13723
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0880667810996703e-05,
+      "loss": 0.7334,
+      "step": 13724
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0879521089801405e-05,
+      "loss": 0.7295,
+      "step": 13725
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0878374356950392e-05,
+      "loss": 0.7188,
+      "step": 13726
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087722761245886e-05,
+      "loss": 0.7256,
+      "step": 13727
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0876080856342008e-05,
+      "loss": 0.7246,
+      "step": 13728
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087493408861503e-05,
+      "loss": 0.7607,
+      "step": 13729
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0873787309293129e-05,
+      "loss": 0.7285,
+      "step": 13730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0872640518391495e-05,
+      "loss": 0.7178,
+      "step": 13731
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087149371592533e-05,
+      "loss": 0.7275,
+      "step": 13732
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0870346901909833e-05,
+      "loss": 0.7373,
+      "step": 13733
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.08692000763602e-05,
+      "loss": 0.7422,
+      "step": 13734
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0868053239291625e-05,
+      "loss": 0.7344,
+      "step": 13735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0866906390719315e-05,
+      "loss": 0.7344,
+      "step": 13736
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0865759530658462e-05,
+      "loss": 0.7305,
+      "step": 13737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0864612659124267e-05,
+      "loss": 0.7168,
+      "step": 13738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.086346577613193e-05,
+      "loss": 0.7334,
+      "step": 13739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0862318881696648e-05,
+      "loss": 0.7383,
+      "step": 13740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0861171975833617e-05,
+      "loss": 0.7119,
+      "step": 13741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0860025058558042e-05,
+      "loss": 0.7217,
+      "step": 13742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.085887812988512e-05,
+      "loss": 0.7129,
+      "step": 13743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0857731189830049e-05,
+      "loss": 0.752,
+      "step": 13744
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0856584238408033e-05,
+      "loss": 0.7129,
+      "step": 13745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0855437275634266e-05,
+      "loss": 0.7451,
+      "step": 13746
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.085429030152395e-05,
+      "loss": 0.7246,
+      "step": 13747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0853143316092285e-05,
+      "loss": 0.7256,
+      "step": 13748
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0851996319354473e-05,
+      "loss": 0.7598,
+      "step": 13749
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0850849311325715e-05,
+      "loss": 0.7363,
+      "step": 13750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0849702292021211e-05,
+      "loss": 0.7422,
+      "step": 13751
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0848555261456157e-05,
+      "loss": 0.7148,
+      "step": 13752
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0847408219645761e-05,
+      "loss": 0.7393,
+      "step": 13753
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0846261166605216e-05,
+      "loss": 0.7285,
+      "step": 13754
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0845114102349733e-05,
+      "loss": 0.7305,
+      "step": 13755
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0843967026894504e-05,
+      "loss": 0.7529,
+      "step": 13756
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0842819940254736e-05,
+      "loss": 0.7393,
+      "step": 13757
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0841672842445625e-05,
+      "loss": 0.7227,
+      "step": 13758
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0840525733482383e-05,
+      "loss": 0.7549,
+      "step": 13759
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0839378613380203e-05,
+      "loss": 0.7158,
+      "step": 13760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0838231482154288e-05,
+      "loss": 0.7217,
+      "step": 13761
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0837084339819842e-05,
+      "loss": 0.7451,
+      "step": 13762
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0835937186392069e-05,
+      "loss": 0.7246,
+      "step": 13763
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0834790021886169e-05,
+      "loss": 0.7412,
+      "step": 13764
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0833642846317342e-05,
+      "loss": 0.7197,
+      "step": 13765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0832495659700799e-05,
+      "loss": 0.7422,
+      "step": 13766
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0831348462051733e-05,
+      "loss": 0.748,
+      "step": 13767
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0830201253385355e-05,
+      "loss": 0.7207,
+      "step": 13768
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0829054033716867e-05,
+      "loss": 0.708,
+      "step": 13769
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0827906803061466e-05,
+      "loss": 0.7314,
+      "step": 13770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0826759561434361e-05,
+      "loss": 0.7383,
+      "step": 13771
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0825612308850755e-05,
+      "loss": 0.7256,
+      "step": 13772
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0824465045325851e-05,
+      "loss": 0.7324,
+      "step": 13773
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0823317770874853e-05,
+      "loss": 0.708,
+      "step": 13774
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0822170485512969e-05,
+      "loss": 0.748,
+      "step": 13775
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0821023189255395e-05,
+      "loss": 0.7402,
+      "step": 13776
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0819875882117343e-05,
+      "loss": 0.7354,
+      "step": 13777
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0818728564114014e-05,
+      "loss": 0.7119,
+      "step": 13778
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0817581235260612e-05,
+      "loss": 0.7285,
+      "step": 13779
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.081643389557234e-05,
+      "loss": 0.7139,
+      "step": 13780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0815286545064412e-05,
+      "loss": 0.7285,
+      "step": 13781
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0814139183752025e-05,
+      "loss": 0.7119,
+      "step": 13782
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0812991811650384e-05,
+      "loss": 0.7236,
+      "step": 13783
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0811844428774697e-05,
+      "loss": 0.7109,
+      "step": 13784
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.081069703514017e-05,
+      "loss": 0.7402,
+      "step": 13785
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0809549630762006e-05,
+      "loss": 0.7422,
+      "step": 13786
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0808402215655414e-05,
+      "loss": 0.7139,
+      "step": 13787
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0807254789835599e-05,
+      "loss": 0.7002,
+      "step": 13788
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0806107353317765e-05,
+      "loss": 0.7344,
+      "step": 13789
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0804959906117121e-05,
+      "loss": 0.7139,
+      "step": 13790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.080381244824887e-05,
+      "loss": 0.7256,
+      "step": 13791
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0802664979728226e-05,
+      "loss": 0.7412,
+      "step": 13792
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0801517500570385e-05,
+      "loss": 0.7373,
+      "step": 13793
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.080037001079056e-05,
+      "loss": 0.7373,
+      "step": 13794
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0799222510403956e-05,
+      "loss": 0.752,
+      "step": 13795
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0798074999425782e-05,
+      "loss": 0.748,
+      "step": 13796
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0796927477871244e-05,
+      "loss": 0.7129,
+      "step": 13797
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0795779945755554e-05,
+      "loss": 0.7207,
+      "step": 13798
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0794632403093907e-05,
+      "loss": 0.7432,
+      "step": 13799
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0793484849901525e-05,
+      "loss": 0.7295,
+      "step": 13800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0792337286193608e-05,
+      "loss": 0.7266,
+      "step": 13801
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0791189711985364e-05,
+      "loss": 0.7305,
+      "step": 13802
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0790042127292001e-05,
+      "loss": 0.7314,
+      "step": 13803
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0788894532128733e-05,
+      "loss": 0.7363,
+      "step": 13804
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0787746926510761e-05,
+      "loss": 0.7432,
+      "step": 13805
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0786599310453298e-05,
+      "loss": 0.7344,
+      "step": 13806
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0785451683971549e-05,
+      "loss": 0.7344,
+      "step": 13807
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0784304047080728e-05,
+      "loss": 0.7109,
+      "step": 13808
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0783156399796036e-05,
+      "loss": 0.7393,
+      "step": 13809
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.078200874213269e-05,
+      "loss": 0.7188,
+      "step": 13810
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0780861074105895e-05,
+      "loss": 0.7256,
+      "step": 13811
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.077971339573086e-05,
+      "loss": 0.7383,
+      "step": 13812
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0778565707022795e-05,
+      "loss": 0.7422,
+      "step": 13813
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0777418007996913e-05,
+      "loss": 0.7188,
+      "step": 13814
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0776270298668418e-05,
+      "loss": 0.7334,
+      "step": 13815
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0775122579052526e-05,
+      "loss": 0.7549,
+      "step": 13816
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0773974849164439e-05,
+      "loss": 0.7539,
+      "step": 13817
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0772827109019376e-05,
+      "loss": 0.7363,
+      "step": 13818
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.077167935863254e-05,
+      "loss": 0.7217,
+      "step": 13819
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0770531598019145e-05,
+      "loss": 0.7236,
+      "step": 13820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0769383827194401e-05,
+      "loss": 0.748,
+      "step": 13821
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076823604617352e-05,
+      "loss": 0.7412,
+      "step": 13822
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0767088254971711e-05,
+      "loss": 0.7314,
+      "step": 13823
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0765940453604184e-05,
+      "loss": 0.7168,
+      "step": 13824
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076479264208615e-05,
+      "loss": 0.7451,
+      "step": 13825
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0763644820432823e-05,
+      "loss": 0.7686,
+      "step": 13826
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0762496988659414e-05,
+      "loss": 0.752,
+      "step": 13827
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0761349146781132e-05,
+      "loss": 0.7197,
+      "step": 13828
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.076020129481319e-05,
+      "loss": 0.7295,
+      "step": 13829
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0759053432770796e-05,
+      "loss": 0.7305,
+      "step": 13830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0757905560669173e-05,
+      "loss": 0.7275,
+      "step": 13831
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0756757678523519e-05,
+      "loss": 0.7412,
+      "step": 13832
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0755609786349054e-05,
+      "loss": 0.7227,
+      "step": 13833
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0754461884160992e-05,
+      "loss": 0.7432,
+      "step": 13834
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.075331397197454e-05,
+      "loss": 0.7422,
+      "step": 13835
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.075216604980491e-05,
+      "loss": 0.7412,
+      "step": 13836
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0751018117667322e-05,
+      "loss": 0.7188,
+      "step": 13837
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.074987017557698e-05,
+      "loss": 0.7549,
+      "step": 13838
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0748722223549104e-05,
+      "loss": 0.7227,
+      "step": 13839
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0747574261598903e-05,
+      "loss": 0.7256,
+      "step": 13840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0746426289741591e-05,
+      "loss": 0.7246,
+      "step": 13841
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0745278307992381e-05,
+      "loss": 0.751,
+      "step": 13842
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.074413031636649e-05,
+      "loss": 0.7412,
+      "step": 13843
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0742982314879127e-05,
+      "loss": 0.7354,
+      "step": 13844
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0741834303545506e-05,
+      "loss": 0.7402,
+      "step": 13845
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0740686282380844e-05,
+      "loss": 0.7314,
+      "step": 13846
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0739538251400355e-05,
+      "loss": 0.7139,
+      "step": 13847
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0738390210619247e-05,
+      "loss": 0.7188,
+      "step": 13848
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0737242160052741e-05,
+      "loss": 0.7236,
+      "step": 13849
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073609409971605e-05,
+      "loss": 0.7539,
+      "step": 13850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0734946029624385e-05,
+      "loss": 0.7373,
+      "step": 13851
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0733797949792964e-05,
+      "loss": 0.751,
+      "step": 13852
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0732649860237e-05,
+      "loss": 0.7432,
+      "step": 13853
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073150176097171e-05,
+      "loss": 0.7041,
+      "step": 13854
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0730353652012306e-05,
+      "loss": 0.7314,
+      "step": 13855
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0729205533374005e-05,
+      "loss": 0.7568,
+      "step": 13856
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0728057405072025e-05,
+      "loss": 0.7285,
+      "step": 13857
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0726909267121574e-05,
+      "loss": 0.7139,
+      "step": 13858
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0725761119537874e-05,
+      "loss": 0.7266,
+      "step": 13859
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0724612962336139e-05,
+      "loss": 0.7334,
+      "step": 13860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0723464795531584e-05,
+      "loss": 0.7266,
+      "step": 13861
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0722316619139423e-05,
+      "loss": 0.7109,
+      "step": 13862
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.072116843317488e-05,
+      "loss": 0.7432,
+      "step": 13863
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0720020237653161e-05,
+      "loss": 0.7285,
+      "step": 13864
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0718872032589488e-05,
+      "loss": 0.7324,
+      "step": 13865
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0717723817999076e-05,
+      "loss": 0.752,
+      "step": 13866
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0716575593897142e-05,
+      "loss": 0.7285,
+      "step": 13867
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0715427360298903e-05,
+      "loss": 0.7197,
+      "step": 13868
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0714279117219571e-05,
+      "loss": 0.7373,
+      "step": 13869
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.071313086467437e-05,
+      "loss": 0.7383,
+      "step": 13870
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0711982602678515e-05,
+      "loss": 0.7305,
+      "step": 13871
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0710834331247222e-05,
+      "loss": 0.7012,
+      "step": 13872
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0709686050395706e-05,
+      "loss": 0.7197,
+      "step": 13873
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070853776013919e-05,
+      "loss": 0.7158,
+      "step": 13874
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0707389460492886e-05,
+      "loss": 0.7373,
+      "step": 13875
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0706241151472017e-05,
+      "loss": 0.7285,
+      "step": 13876
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0705092833091795e-05,
+      "loss": 0.7334,
+      "step": 13877
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0703944505367442e-05,
+      "loss": 0.7461,
+      "step": 13878
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0702796168314175e-05,
+      "loss": 0.7432,
+      "step": 13879
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0701647821947212e-05,
+      "loss": 0.7383,
+      "step": 13880
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070049946628177e-05,
+      "loss": 0.7217,
+      "step": 13881
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0699351101333071e-05,
+      "loss": 0.7383,
+      "step": 13882
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0698202727116329e-05,
+      "loss": 0.7314,
+      "step": 13883
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0697054343646766e-05,
+      "loss": 0.7168,
+      "step": 13884
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0695905950939598e-05,
+      "loss": 0.7324,
+      "step": 13885
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0694757549010048e-05,
+      "loss": 0.7188,
+      "step": 13886
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0693609137873332e-05,
+      "loss": 0.7373,
+      "step": 13887
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0692460717544668e-05,
+      "loss": 0.7373,
+      "step": 13888
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069131228803928e-05,
+      "loss": 0.752,
+      "step": 13889
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0690163849372381e-05,
+      "loss": 0.7529,
+      "step": 13890
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0689015401559192e-05,
+      "loss": 0.7178,
+      "step": 13891
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0687866944614943e-05,
+      "loss": 0.71,
+      "step": 13892
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0686718478554836e-05,
+      "loss": 0.7314,
+      "step": 13893
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0685570003394106e-05,
+      "loss": 0.7256,
+      "step": 13894
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0684421519147965e-05,
+      "loss": 0.7197,
+      "step": 13895
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0683273025831634e-05,
+      "loss": 0.7617,
+      "step": 13896
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0682124523460335e-05,
+      "loss": 0.7402,
+      "step": 13897
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0680976012049289e-05,
+      "loss": 0.748,
+      "step": 13898
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0679827491613714e-05,
+      "loss": 0.7363,
+      "step": 13899
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0678678962168833e-05,
+      "loss": 0.7412,
+      "step": 13900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0677530423729863e-05,
+      "loss": 0.7314,
+      "step": 13901
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0676381876312032e-05,
+      "loss": 0.7178,
+      "step": 13902
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.067523331993055e-05,
+      "loss": 0.7471,
+      "step": 13903
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.067408475460065e-05,
+      "loss": 0.7559,
+      "step": 13904
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0672936180337544e-05,
+      "loss": 0.7275,
+      "step": 13905
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0671787597156457e-05,
+      "loss": 0.7129,
+      "step": 13906
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0670639005072612e-05,
+      "loss": 0.7197,
+      "step": 13907
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0669490404101228e-05,
+      "loss": 0.707,
+      "step": 13908
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0668341794257526e-05,
+      "loss": 0.7266,
+      "step": 13909
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0667193175556732e-05,
+      "loss": 0.7314,
+      "step": 13910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0666044548014064e-05,
+      "loss": 0.7441,
+      "step": 13911
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0664895911644744e-05,
+      "loss": 0.7305,
+      "step": 13912
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0663747266463995e-05,
+      "loss": 0.7119,
+      "step": 13913
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0662598612487039e-05,
+      "loss": 0.7285,
+      "step": 13914
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.06614499497291e-05,
+      "loss": 0.7295,
+      "step": 13915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0660301278205401e-05,
+      "loss": 0.7324,
+      "step": 13916
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0659152597931162e-05,
+      "loss": 0.7432,
+      "step": 13917
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0658003908921606e-05,
+      "loss": 0.7363,
+      "step": 13918
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0656855211191956e-05,
+      "loss": 0.7344,
+      "step": 13919
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0655706504757433e-05,
+      "loss": 0.7373,
+      "step": 13920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0654557789633269e-05,
+      "loss": 0.6953,
+      "step": 13921
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0653409065834672e-05,
+      "loss": 0.7383,
+      "step": 13922
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0652260333376881e-05,
+      "loss": 0.7363,
+      "step": 13923
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0651111592275107e-05,
+      "loss": 0.7314,
+      "step": 13924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0649962842544583e-05,
+      "loss": 0.71,
+      "step": 13925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0648814084200525e-05,
+      "loss": 0.71,
+      "step": 13926
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0647665317258163e-05,
+      "loss": 0.7344,
+      "step": 13927
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0646516541732716e-05,
+      "loss": 0.7354,
+      "step": 13928
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.064536775763941e-05,
+      "loss": 0.7451,
+      "step": 13929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0644218964993469e-05,
+      "loss": 0.7207,
+      "step": 13930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.064307016381012e-05,
+      "loss": 0.7607,
+      "step": 13931
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0641921354104581e-05,
+      "loss": 0.7432,
+      "step": 13932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.064077253589208e-05,
+      "loss": 0.7334,
+      "step": 13933
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0639623709187842e-05,
+      "loss": 0.7471,
+      "step": 13934
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0638474874007095e-05,
+      "loss": 0.71,
+      "step": 13935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0637326030365054e-05,
+      "loss": 0.7236,
+      "step": 13936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0636177178276953e-05,
+      "loss": 0.71,
+      "step": 13937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0635028317758012e-05,
+      "loss": 0.751,
+      "step": 13938
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0633879448823459e-05,
+      "loss": 0.7207,
+      "step": 13939
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0632730571488516e-05,
+      "loss": 0.7334,
+      "step": 13940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0631581685768414e-05,
+      "loss": 0.7295,
+      "step": 13941
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0630432791678373e-05,
+      "loss": 0.7334,
+      "step": 13942
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062928388923362e-05,
+      "loss": 0.7295,
+      "step": 13943
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0628134978449381e-05,
+      "loss": 0.7363,
+      "step": 13944
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0626986059340882e-05,
+      "loss": 0.7412,
+      "step": 13945
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0625837131923347e-05,
+      "loss": 0.7881,
+      "step": 13946
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0624688196212008e-05,
+      "loss": 0.7314,
+      "step": 13947
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.062353925222208e-05,
+      "loss": 0.7314,
+      "step": 13948
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0622390299968801e-05,
+      "loss": 0.7373,
+      "step": 13949
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0621241339467391e-05,
+      "loss": 0.7529,
+      "step": 13950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0620092370733077e-05,
+      "loss": 0.7373,
+      "step": 13951
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0618943393781087e-05,
+      "loss": 0.7402,
+      "step": 13952
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0617794408626645e-05,
+      "loss": 0.7188,
+      "step": 13953
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0616645415284981e-05,
+      "loss": 0.7285,
+      "step": 13954
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061549641377132e-05,
+      "loss": 0.7305,
+      "step": 13955
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0614347404100889e-05,
+      "loss": 0.7627,
+      "step": 13956
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0613198386288916e-05,
+      "loss": 0.7285,
+      "step": 13957
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061204936035063e-05,
+      "loss": 0.748,
+      "step": 13958
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.061090032630125e-05,
+      "loss": 0.7549,
+      "step": 13959
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0609751284156014e-05,
+      "loss": 0.7441,
+      "step": 13960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0608602233930142e-05,
+      "loss": 0.7373,
+      "step": 13961
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0607453175638864e-05,
+      "loss": 0.7168,
+      "step": 13962
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0606304109297408e-05,
+      "loss": 0.7227,
+      "step": 13963
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0605155034921007e-05,
+      "loss": 0.7227,
+      "step": 13964
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0604005952524876e-05,
+      "loss": 0.7324,
+      "step": 13965
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0602856862124255e-05,
+      "loss": 0.7617,
+      "step": 13966
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.060170776373437e-05,
+      "loss": 0.7285,
+      "step": 13967
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0600558657370446e-05,
+      "loss": 0.7334,
+      "step": 13968
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.059940954304771e-05,
+      "loss": 0.7383,
+      "step": 13969
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0598260420781396e-05,
+      "loss": 0.7197,
+      "step": 13970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0597111290586725e-05,
+      "loss": 0.7168,
+      "step": 13971
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0595962152478935e-05,
+      "loss": 0.7402,
+      "step": 13972
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0594813006473247e-05,
+      "loss": 0.7188,
+      "step": 13973
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0593663852584896e-05,
+      "loss": 0.7002,
+      "step": 13974
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0592514690829105e-05,
+      "loss": 0.7207,
+      "step": 13975
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0591365521221108e-05,
+      "loss": 0.7217,
+      "step": 13976
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0590216343776132e-05,
+      "loss": 0.7412,
+      "step": 13977
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0589067158509405e-05,
+      "loss": 0.7363,
+      "step": 13978
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0587917965436158e-05,
+      "loss": 0.7031,
+      "step": 13979
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0586768764571624e-05,
+      "loss": 0.7119,
+      "step": 13980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0585619555931025e-05,
+      "loss": 0.71,
+      "step": 13981
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0584470339529596e-05,
+      "loss": 0.75,
+      "step": 13982
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0583321115382565e-05,
+      "loss": 0.7324,
+      "step": 13983
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0582171883505164e-05,
+      "loss": 0.7266,
+      "step": 13984
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0581022643912618e-05,
+      "loss": 0.751,
+      "step": 13985
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0579873396620163e-05,
+      "loss": 0.7129,
+      "step": 13986
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0578724141643026e-05,
+      "loss": 0.7471,
+      "step": 13987
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057757487899644e-05,
+      "loss": 0.7393,
+      "step": 13988
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.057642560869563e-05,
+      "loss": 0.7549,
+      "step": 13989
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0575276330755835e-05,
+      "loss": 0.7148,
+      "step": 13990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0574127045192277e-05,
+      "loss": 0.7197,
+      "step": 13991
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0572977752020188e-05,
+      "loss": 0.7373,
+      "step": 13992
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0571828451254806e-05,
+      "loss": 0.7559,
+      "step": 13993
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0570679142911356e-05,
+      "loss": 0.7256,
+      "step": 13994
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0569529827005069e-05,
+      "loss": 0.7295,
+      "step": 13995
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056838050355118e-05,
+      "loss": 0.7393,
+      "step": 13996
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0567231172564914e-05,
+      "loss": 0.7363,
+      "step": 13997
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0566081834061505e-05,
+      "loss": 0.7471,
+      "step": 13998
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0564932488056188e-05,
+      "loss": 0.7422,
+      "step": 13999
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0563783134564191e-05,
+      "loss": 0.7373,
+      "step": 14000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0562633773600747e-05,
+      "loss": 0.7402,
+      "step": 14001
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0561484405181089e-05,
+      "loss": 0.7256,
+      "step": 14002
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0560335029320444e-05,
+      "loss": 0.7012,
+      "step": 14003
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0559185646034044e-05,
+      "loss": 0.7646,
+      "step": 14004
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.055803625533713e-05,
+      "loss": 0.7227,
+      "step": 14005
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0556886857244926e-05,
+      "loss": 0.7324,
+      "step": 14006
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0555737451772665e-05,
+      "loss": 0.7178,
+      "step": 14007
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.055458803893558e-05,
+      "loss": 0.7285,
+      "step": 14008
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0553438618748908e-05,
+      "loss": 0.7021,
+      "step": 14009
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0552289191227873e-05,
+      "loss": 0.7305,
+      "step": 14010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0551139756387714e-05,
+      "loss": 0.7422,
+      "step": 14011
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0549990314243663e-05,
+      "loss": 0.7266,
+      "step": 14012
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0548840864810952e-05,
+      "loss": 0.7246,
+      "step": 14013
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054769140810481e-05,
+      "loss": 0.7363,
+      "step": 14014
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0546541944140479e-05,
+      "loss": 0.7334,
+      "step": 14015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0545392472933181e-05,
+      "loss": 0.7383,
+      "step": 14016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0544242994498156e-05,
+      "loss": 0.7275,
+      "step": 14017
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0543093508850635e-05,
+      "loss": 0.7344,
+      "step": 14018
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0541944016005857e-05,
+      "loss": 0.7266,
+      "step": 14019
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0540794515979047e-05,
+      "loss": 0.7188,
+      "step": 14020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0539645008785444e-05,
+      "loss": 0.7021,
+      "step": 14021
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.053849549444028e-05,
+      "loss": 0.7197,
+      "step": 14022
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0537345972958787e-05,
+      "loss": 0.7148,
+      "step": 14023
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0536196444356201e-05,
+      "loss": 0.7607,
+      "step": 14024
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0535046908647761e-05,
+      "loss": 0.7305,
+      "step": 14025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0533897365848691e-05,
+      "loss": 0.7334,
+      "step": 14026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0532747815974231e-05,
+      "loss": 0.7334,
+      "step": 14027
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0531598259039615e-05,
+      "loss": 0.7383,
+      "step": 14028
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0530448695060076e-05,
+      "loss": 0.7129,
+      "step": 14029
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0529299124050846e-05,
+      "loss": 0.7422,
+      "step": 14030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0528149546027164e-05,
+      "loss": 0.7139,
+      "step": 14031
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0526999961004264e-05,
+      "loss": 0.7383,
+      "step": 14032
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0525850368997381e-05,
+      "loss": 0.7568,
+      "step": 14033
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0524700770021748e-05,
+      "loss": 0.7598,
+      "step": 14034
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0523551164092597e-05,
+      "loss": 0.7324,
+      "step": 14035
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052240155122517e-05,
+      "loss": 0.7295,
+      "step": 14036
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0521251931434695e-05,
+      "loss": 0.7148,
+      "step": 14037
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0520102304736413e-05,
+      "loss": 0.7285,
+      "step": 14038
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0518952671145557e-05,
+      "loss": 0.7256,
+      "step": 14039
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0517803030677361e-05,
+      "loss": 0.7402,
+      "step": 14040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0516653383347061e-05,
+      "loss": 0.7246,
+      "step": 14041
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0515503729169896e-05,
+      "loss": 0.749,
+      "step": 14042
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0514354068161095e-05,
+      "loss": 0.7119,
+      "step": 14043
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.05132044003359e-05,
+      "loss": 0.7412,
+      "step": 14044
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0512054725709541e-05,
+      "loss": 0.7207,
+      "step": 14045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.051090504429726e-05,
+      "loss": 0.7197,
+      "step": 14046
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0509755356114288e-05,
+      "loss": 0.7334,
+      "step": 14047
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0508605661175865e-05,
+      "loss": 0.7275,
+      "step": 14048
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0507455959497224e-05,
+      "loss": 0.7568,
+      "step": 14049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0506306251093603e-05,
+      "loss": 0.7139,
+      "step": 14050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0505156535980238e-05,
+      "loss": 0.7334,
+      "step": 14051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0504006814172365e-05,
+      "loss": 0.749,
+      "step": 14052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0502857085685219e-05,
+      "loss": 0.7188,
+      "step": 14053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0501707350534042e-05,
+      "loss": 0.7139,
+      "step": 14054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0500557608734063e-05,
+      "loss": 0.7139,
+      "step": 14055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0499407860300526e-05,
+      "loss": 0.7363,
+      "step": 14056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0498258105248663e-05,
+      "loss": 0.7217,
+      "step": 14057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0497108343593714e-05,
+      "loss": 0.7266,
+      "step": 14058
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0495958575350913e-05,
+      "loss": 0.7383,
+      "step": 14059
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0494808800535499e-05,
+      "loss": 0.7646,
+      "step": 14060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049365901916271e-05,
+      "loss": 0.7246,
+      "step": 14061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0492509231247783e-05,
+      "loss": 0.7139,
+      "step": 14062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049135943680595e-05,
+      "loss": 0.7119,
+      "step": 14063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049020963585246e-05,
+      "loss": 0.7461,
+      "step": 14064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048905982840254e-05,
+      "loss": 0.7207,
+      "step": 14065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048791001447143e-05,
+      "loss": 0.7393,
+      "step": 14066
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048676019407437e-05,
+      "loss": 0.7451,
+      "step": 14067
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.04856103672266e-05,
+      "loss": 0.7197,
+      "step": 14068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048446053394335e-05,
+      "loss": 0.7295,
+      "step": 14069
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0483310694239866e-05,
+      "loss": 0.7031,
+      "step": 14070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048216084813138e-05,
+      "loss": 0.7393,
+      "step": 14071
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0481010995633137e-05,
+      "loss": 0.7266,
+      "step": 14072
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0479861136760367e-05,
+      "loss": 0.7285,
+      "step": 14073
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0478711271528317e-05,
+      "loss": 0.7188,
+      "step": 14074
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0477561399952218e-05,
+      "loss": 0.7129,
+      "step": 14075
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0476411522047309e-05,
+      "loss": 0.7217,
+      "step": 14076
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0475261637828834e-05,
+      "loss": 0.7139,
+      "step": 14077
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.047411174731203e-05,
+      "loss": 0.7246,
+      "step": 14078
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0472961850512133e-05,
+      "loss": 0.7285,
+      "step": 14079
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0471811947444384e-05,
+      "loss": 0.707,
+      "step": 14080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.047066203812402e-05,
+      "loss": 0.7197,
+      "step": 14081
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.046951212256628e-05,
+      "loss": 0.7412,
+      "step": 14082
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0468362200786406e-05,
+      "loss": 0.7324,
+      "step": 14083
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0467212272799635e-05,
+      "loss": 0.7275,
+      "step": 14084
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0466062338621208e-05,
+      "loss": 0.7559,
+      "step": 14085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.046491239826636e-05,
+      "loss": 0.7256,
+      "step": 14086
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0463762451750337e-05,
+      "loss": 0.7227,
+      "step": 14087
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0462612499088372e-05,
+      "loss": 0.7422,
+      "step": 14088
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0461462540295707e-05,
+      "loss": 0.7363,
+      "step": 14089
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0460312575387585e-05,
+      "loss": 0.7266,
+      "step": 14090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0459162604379242e-05,
+      "loss": 0.7246,
+      "step": 14091
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0458012627285916e-05,
+      "loss": 0.7334,
+      "step": 14092
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0456862644122853e-05,
+      "loss": 0.7227,
+      "step": 14093
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0455712654905285e-05,
+      "loss": 0.7266,
+      "step": 14094
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0454562659648462e-05,
+      "loss": 0.7207,
+      "step": 14095
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0453412658367612e-05,
+      "loss": 0.7266,
+      "step": 14096
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.045226265107799e-05,
+      "loss": 0.7559,
+      "step": 14097
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.045111263779482e-05,
+      "loss": 0.7207,
+      "step": 14098
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0449962618533356e-05,
+      "loss": 0.7344,
+      "step": 14099
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0448812593308833e-05,
+      "loss": 0.7129,
+      "step": 14100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044766256213649e-05,
+      "loss": 0.7373,
+      "step": 14101
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0446512525031567e-05,
+      "loss": 0.7246,
+      "step": 14102
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0445362482009309e-05,
+      "loss": 0.71,
+      "step": 14103
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0444212433084953e-05,
+      "loss": 0.7324,
+      "step": 14104
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0443062378273743e-05,
+      "loss": 0.7256,
+      "step": 14105
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044191231759092e-05,
+      "loss": 0.7061,
+      "step": 14106
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0440762251051722e-05,
+      "loss": 0.7412,
+      "step": 14107
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0439612178671387e-05,
+      "loss": 0.7139,
+      "step": 14108
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0438462100465168e-05,
+      "loss": 0.7109,
+      "step": 14109
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0437312016448293e-05,
+      "loss": 0.7686,
+      "step": 14110
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0436161926636011e-05,
+      "loss": 0.7471,
+      "step": 14111
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0435011831043561e-05,
+      "loss": 0.7441,
+      "step": 14112
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0433861729686185e-05,
+      "loss": 0.7129,
+      "step": 14113
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0432711622579124e-05,
+      "loss": 0.7002,
+      "step": 14114
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0431561509737623e-05,
+      "loss": 0.7285,
+      "step": 14115
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0430411391176918e-05,
+      "loss": 0.7666,
+      "step": 14116
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0429261266912254e-05,
+      "loss": 0.7148,
+      "step": 14117
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0428111136958874e-05,
+      "loss": 0.7529,
+      "step": 14118
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426961001332017e-05,
+      "loss": 0.7295,
+      "step": 14119
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0425810860046928e-05,
+      "loss": 0.7324,
+      "step": 14120
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0424660713118842e-05,
+      "loss": 0.7109,
+      "step": 14121
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.042351056056301e-05,
+      "loss": 0.7236,
+      "step": 14122
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0422360402394673e-05,
+      "loss": 0.71,
+      "step": 14123
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.042121023862907e-05,
+      "loss": 0.7549,
+      "step": 14124
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.042006006928144e-05,
+      "loss": 0.7383,
+      "step": 14125
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0418909894367035e-05,
+      "loss": 0.7422,
+      "step": 14126
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041775971390109e-05,
+      "loss": 0.7598,
+      "step": 14127
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0416609527898848e-05,
+      "loss": 0.7314,
+      "step": 14128
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0415459336375557e-05,
+      "loss": 0.7344,
+      "step": 14129
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0414309139346457e-05,
+      "loss": 0.7354,
+      "step": 14130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0413158936826785e-05,
+      "loss": 0.7119,
+      "step": 14131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0412008728831795e-05,
+      "loss": 0.7285,
+      "step": 14132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041085851537672e-05,
+      "loss": 0.7461,
+      "step": 14133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0409708296476808e-05,
+      "loss": 0.7246,
+      "step": 14134
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0408558072147301e-05,
+      "loss": 0.7236,
+      "step": 14135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0407407842403443e-05,
+      "loss": 0.7363,
+      "step": 14136
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0406257607260474e-05,
+      "loss": 0.7139,
+      "step": 14137
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0405107366733642e-05,
+      "loss": 0.7295,
+      "step": 14138
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.040395712083819e-05,
+      "loss": 0.7256,
+      "step": 14139
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0402806869589356e-05,
+      "loss": 0.7305,
+      "step": 14140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0401656613002386e-05,
+      "loss": 0.7178,
+      "step": 14141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0400506351092532e-05,
+      "loss": 0.708,
+      "step": 14142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0399356083875022e-05,
+      "loss": 0.7432,
+      "step": 14143
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0398205811365113e-05,
+      "loss": 0.7461,
+      "step": 14144
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0397055533578042e-05,
+      "loss": 0.7227,
+      "step": 14145
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0395905250529058e-05,
+      "loss": 0.7441,
+      "step": 14146
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0394754962233396e-05,
+      "loss": 0.7227,
+      "step": 14147
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393604668706312e-05,
+      "loss": 0.7432,
+      "step": 14148
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.039245436996304e-05,
+      "loss": 0.7256,
+      "step": 14149
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0391304066018828e-05,
+      "loss": 0.7344,
+      "step": 14150
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0390153756888922e-05,
+      "loss": 0.751,
+      "step": 14151
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0389003442588564e-05,
+      "loss": 0.71,
+      "step": 14152
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0387853123132996e-05,
+      "loss": 0.7295,
+      "step": 14153
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0386702798537469e-05,
+      "loss": 0.7207,
+      "step": 14154
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0385552468817223e-05,
+      "loss": 0.7295,
+      "step": 14155
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0384402133987504e-05,
+      "loss": 0.7305,
+      "step": 14156
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0383251794063554e-05,
+      "loss": 0.7373,
+      "step": 14157
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0382101449060622e-05,
+      "loss": 0.7627,
+      "step": 14158
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0380951098993947e-05,
+      "loss": 0.7334,
+      "step": 14159
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0379800743878781e-05,
+      "loss": 0.7324,
+      "step": 14160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0378650383730361e-05,
+      "loss": 0.7686,
+      "step": 14161
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0377500018563942e-05,
+      "loss": 0.7363,
+      "step": 14162
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037634964839476e-05,
+      "loss": 0.7393,
+      "step": 14163
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037519927323806e-05,
+      "loss": 0.7188,
+      "step": 14164
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0374048893109099e-05,
+      "loss": 0.7217,
+      "step": 14165
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0372898508023106e-05,
+      "loss": 0.7197,
+      "step": 14166
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0371748117995338e-05,
+      "loss": 0.748,
+      "step": 14167
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0370597723041034e-05,
+      "loss": 0.7383,
+      "step": 14168
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0369447323175443e-05,
+      "loss": 0.7275,
+      "step": 14169
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0368296918413807e-05,
+      "loss": 0.7109,
+      "step": 14170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0367146508771377e-05,
+      "loss": 0.7285,
+      "step": 14171
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0365996094263392e-05,
+      "loss": 0.7217,
+      "step": 14172
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0364845674905106e-05,
+      "loss": 0.7402,
+      "step": 14173
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0363695250711759e-05,
+      "loss": 0.7324,
+      "step": 14174
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0362544821698595e-05,
+      "loss": 0.7168,
+      "step": 14175
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0361394387880861e-05,
+      "loss": 0.7354,
+      "step": 14176
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0360243949273809e-05,
+      "loss": 0.7158,
+      "step": 14177
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0359093505892679e-05,
+      "loss": 0.7422,
+      "step": 14178
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035794305775272e-05,
+      "loss": 0.7461,
+      "step": 14179
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0356792604869172e-05,
+      "loss": 0.7236,
+      "step": 14180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0355642147257292e-05,
+      "loss": 0.7119,
+      "step": 14181
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0354491684932314e-05,
+      "loss": 0.748,
+      "step": 14182
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0353341217909497e-05,
+      "loss": 0.7354,
+      "step": 14183
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0352190746204076e-05,
+      "loss": 0.7344,
+      "step": 14184
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0351040269831304e-05,
+      "loss": 0.7178,
+      "step": 14185
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0349889788806426e-05,
+      "loss": 0.6943,
+      "step": 14186
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034873930314469e-05,
+      "loss": 0.7178,
+      "step": 14187
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0347588812861337e-05,
+      "loss": 0.7432,
+      "step": 14188
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346438317971624e-05,
+      "loss": 0.7246,
+      "step": 14189
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0345287818490783e-05,
+      "loss": 0.7227,
+      "step": 14190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034413731443408e-05,
+      "loss": 0.7061,
+      "step": 14191
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0342986805816742e-05,
+      "loss": 0.7461,
+      "step": 14192
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0341836292654029e-05,
+      "loss": 0.7354,
+      "step": 14193
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0340685774961184e-05,
+      "loss": 0.7207,
+      "step": 14194
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0339535252753455e-05,
+      "loss": 0.6982,
+      "step": 14195
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0338384726046087e-05,
+      "loss": 0.7539,
+      "step": 14196
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0337234194854328e-05,
+      "loss": 0.749,
+      "step": 14197
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0336083659193426e-05,
+      "loss": 0.7256,
+      "step": 14198
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033493311907863e-05,
+      "loss": 0.7646,
+      "step": 14199
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0333782574525184e-05,
+      "loss": 0.7676,
+      "step": 14200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0332632025548337e-05,
+      "loss": 0.7275,
+      "step": 14201
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0331481472163337e-05,
+      "loss": 0.7256,
+      "step": 14202
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0330330914385426e-05,
+      "loss": 0.7393,
+      "step": 14203
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0329180352229862e-05,
+      "loss": 0.7227,
+      "step": 14204
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0328029785711885e-05,
+      "loss": 0.7012,
+      "step": 14205
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0326879214846743e-05,
+      "loss": 0.7598,
+      "step": 14206
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0325728639649688e-05,
+      "loss": 0.7227,
+      "step": 14207
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0324578060135966e-05,
+      "loss": 0.7314,
+      "step": 14208
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.032342747632082e-05,
+      "loss": 0.7158,
+      "step": 14209
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0322276888219506e-05,
+      "loss": 0.7148,
+      "step": 14210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0321126295847266e-05,
+      "loss": 0.7275,
+      "step": 14211
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0319975699219351e-05,
+      "loss": 0.7236,
+      "step": 14212
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0318825098351008e-05,
+      "loss": 0.7178,
+      "step": 14213
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0317674493257489e-05,
+      "loss": 0.749,
+      "step": 14214
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031652388395403e-05,
+      "loss": 0.7295,
+      "step": 14215
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0315373270455898e-05,
+      "loss": 0.7197,
+      "step": 14216
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0314222652778323e-05,
+      "loss": 0.7266,
+      "step": 14217
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0313072030936567e-05,
+      "loss": 0.7451,
+      "step": 14218
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0311921404945869e-05,
+      "loss": 0.7383,
+      "step": 14219
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0310770774821484e-05,
+      "loss": 0.7158,
+      "step": 14220
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0309620140578655e-05,
+      "loss": 0.7285,
+      "step": 14221
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0308469502232636e-05,
+      "loss": 0.7383,
+      "step": 14222
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0307318859798674e-05,
+      "loss": 0.7588,
+      "step": 14223
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0306168213292017e-05,
+      "loss": 0.7285,
+      "step": 14224
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0305017562727913e-05,
+      "loss": 0.7334,
+      "step": 14225
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0303866908121614e-05,
+      "loss": 0.7188,
+      "step": 14226
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030271624948836e-05,
+      "loss": 0.709,
+      "step": 14227
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0301565586843412e-05,
+      "loss": 0.707,
+      "step": 14228
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0300414920202014e-05,
+      "loss": 0.7305,
+      "step": 14229
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0299264249579411e-05,
+      "loss": 0.7295,
+      "step": 14230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0298113574990853e-05,
+      "loss": 0.7256,
+      "step": 14231
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0296962896451596e-05,
+      "loss": 0.7275,
+      "step": 14232
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0295812213976886e-05,
+      "loss": 0.7217,
+      "step": 14233
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0294661527581969e-05,
+      "loss": 0.7139,
+      "step": 14234
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0293510837282097e-05,
+      "loss": 0.7324,
+      "step": 14235
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0292360143092519e-05,
+      "loss": 0.7217,
+      "step": 14236
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.029120944502848e-05,
+      "loss": 0.7227,
+      "step": 14237
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0290058743105239e-05,
+      "loss": 0.7324,
+      "step": 14238
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0288908037338037e-05,
+      "loss": 0.7422,
+      "step": 14239
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0287757327742127e-05,
+      "loss": 0.7119,
+      "step": 14240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028660661433276e-05,
+      "loss": 0.7188,
+      "step": 14241
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028545589712518e-05,
+      "loss": 0.7324,
+      "step": 14242
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0284305176134641e-05,
+      "loss": 0.7393,
+      "step": 14243
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0283154451376394e-05,
+      "loss": 0.7236,
+      "step": 14244
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0282003722865687e-05,
+      "loss": 0.7197,
+      "step": 14245
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0280852990617772e-05,
+      "loss": 0.7373,
+      "step": 14246
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0279702254647894e-05,
+      "loss": 0.7344,
+      "step": 14247
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0278551514971305e-05,
+      "loss": 0.748,
+      "step": 14248
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0277400771603258e-05,
+      "loss": 0.7295,
+      "step": 14249
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0276250024558999e-05,
+      "loss": 0.7227,
+      "step": 14250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0275099273853782e-05,
+      "loss": 0.7207,
+      "step": 14251
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0273948519502852e-05,
+      "loss": 0.7393,
+      "step": 14252
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0272797761521464e-05,
+      "loss": 0.7432,
+      "step": 14253
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0271646999924865e-05,
+      "loss": 0.7266,
+      "step": 14254
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0270496234728312e-05,
+      "loss": 0.7207,
+      "step": 14255
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0269345465947042e-05,
+      "loss": 0.7158,
+      "step": 14256
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0268194693596318e-05,
+      "loss": 0.7441,
+      "step": 14257
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0267043917691383e-05,
+      "loss": 0.7227,
+      "step": 14258
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0265893138247499e-05,
+      "loss": 0.7383,
+      "step": 14259
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0264742355279898e-05,
+      "loss": 0.752,
+      "step": 14260
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0263591568803845e-05,
+      "loss": 0.707,
+      "step": 14261
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0262440778834585e-05,
+      "loss": 0.7285,
+      "step": 14262
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0261289985387371e-05,
+      "loss": 0.7471,
+      "step": 14263
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.026013918847745e-05,
+      "loss": 0.7236,
+      "step": 14264
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025898838812008e-05,
+      "loss": 0.7197,
+      "step": 14265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0257837584330498e-05,
+      "loss": 0.6982,
+      "step": 14266
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0256686777123972e-05,
+      "loss": 0.7256,
+      "step": 14267
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0255535966515741e-05,
+      "loss": 0.7783,
+      "step": 14268
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0254385152521062e-05,
+      "loss": 0.7461,
+      "step": 14269
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0253234335155182e-05,
+      "loss": 0.6943,
+      "step": 14270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0252083514433353e-05,
+      "loss": 0.7139,
+      "step": 14271
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0250932690370829e-05,
+      "loss": 0.7012,
+      "step": 14272
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.024978186298286e-05,
+      "loss": 0.7256,
+      "step": 14273
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.024863103228469e-05,
+      "loss": 0.7666,
+      "step": 14274
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0247480198291584e-05,
+      "loss": 0.7197,
+      "step": 14275
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.024632936101878e-05,
+      "loss": 0.7197,
+      "step": 14276
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0245178520481535e-05,
+      "loss": 0.7256,
+      "step": 14277
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0244027676695104e-05,
+      "loss": 0.752,
+      "step": 14278
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.024287682967473e-05,
+      "loss": 0.7588,
+      "step": 14279
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0241725979435673e-05,
+      "loss": 0.751,
+      "step": 14280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0240575125993178e-05,
+      "loss": 0.7305,
+      "step": 14281
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0239424269362497e-05,
+      "loss": 0.7178,
+      "step": 14282
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0238273409558886e-05,
+      "loss": 0.7168,
+      "step": 14283
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0237122546597594e-05,
+      "loss": 0.7188,
+      "step": 14284
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0235971680493873e-05,
+      "loss": 0.7373,
+      "step": 14285
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0234820811262973e-05,
+      "loss": 0.7305,
+      "step": 14286
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0233669938920145e-05,
+      "loss": 0.7236,
+      "step": 14287
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0232519063480648e-05,
+      "loss": 0.7197,
+      "step": 14288
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0231368184959725e-05,
+      "loss": 0.7256,
+      "step": 14289
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0230217303372631e-05,
+      "loss": 0.7148,
+      "step": 14290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022906641873462e-05,
+      "loss": 0.7422,
+      "step": 14291
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0227915531060944e-05,
+      "loss": 0.7363,
+      "step": 14292
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0226764640366849e-05,
+      "loss": 0.7266,
+      "step": 14293
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0225613746667595e-05,
+      "loss": 0.7256,
+      "step": 14294
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0224462849978428e-05,
+      "loss": 0.7227,
+      "step": 14295
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0223311950314603e-05,
+      "loss": 0.7373,
+      "step": 14296
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0222161047691369e-05,
+      "loss": 0.71,
+      "step": 14297
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0221010142123983e-05,
+      "loss": 0.7549,
+      "step": 14298
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0219859233627694e-05,
+      "loss": 0.7637,
+      "step": 14299
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0218708322217755e-05,
+      "loss": 0.709,
+      "step": 14300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021755740790942e-05,
+      "loss": 0.7246,
+      "step": 14301
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0216406490717939e-05,
+      "loss": 0.7793,
+      "step": 14302
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021525557065856e-05,
+      "loss": 0.7129,
+      "step": 14303
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0214104647746548e-05,
+      "loss": 0.7197,
+      "step": 14304
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021295372199714e-05,
+      "loss": 0.7061,
+      "step": 14305
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0211802793425602e-05,
+      "loss": 0.7119,
+      "step": 14306
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021065186204718e-05,
+      "loss": 0.7002,
+      "step": 14307
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0209500927877127e-05,
+      "loss": 0.7256,
+      "step": 14308
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0208349990930693e-05,
+      "loss": 0.7363,
+      "step": 14309
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0207199051223135e-05,
+      "loss": 0.7246,
+      "step": 14310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0206048108769706e-05,
+      "loss": 0.7285,
+      "step": 14311
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0204897163585655e-05,
+      "loss": 0.7129,
+      "step": 14312
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0203746215686235e-05,
+      "loss": 0.7207,
+      "step": 14313
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0202595265086704e-05,
+      "loss": 0.7217,
+      "step": 14314
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0201444311802307e-05,
+      "loss": 0.7207,
+      "step": 14315
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0200293355848303e-05,
+      "loss": 0.7412,
+      "step": 14316
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0199142397239943e-05,
+      "loss": 0.7363,
+      "step": 14317
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.019799143599248e-05,
+      "loss": 0.7236,
+      "step": 14318
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0196840472121164e-05,
+      "loss": 0.752,
+      "step": 14319
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0195689505641256e-05,
+      "loss": 0.7383,
+      "step": 14320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0194538536567997e-05,
+      "loss": 0.7305,
+      "step": 14321
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.019338756491665e-05,
+      "loss": 0.7227,
+      "step": 14322
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0192236590702465e-05,
+      "loss": 0.7402,
+      "step": 14323
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0191085613940692e-05,
+      "loss": 0.7305,
+      "step": 14324
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.018993463464659e-05,
+      "loss": 0.7256,
+      "step": 14325
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0188783652835406e-05,
+      "loss": 0.7393,
+      "step": 14326
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0187632668522398e-05,
+      "loss": 0.7188,
+      "step": 14327
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0186481681722817e-05,
+      "loss": 0.7363,
+      "step": 14328
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0185330692451917e-05,
+      "loss": 0.749,
+      "step": 14329
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0184179700724952e-05,
+      "loss": 0.7432,
+      "step": 14330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0183028706557174e-05,
+      "loss": 0.7598,
+      "step": 14331
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0181877709963835e-05,
+      "loss": 0.7256,
+      "step": 14332
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0180726710960194e-05,
+      "loss": 0.7236,
+      "step": 14333
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0179575709561495e-05,
+      "loss": 0.7324,
+      "step": 14334
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0178424705783001e-05,
+      "loss": 0.7266,
+      "step": 14335
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0177273699639961e-05,
+      "loss": 0.752,
+      "step": 14336
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.017612269114763e-05,
+      "loss": 0.7324,
+      "step": 14337
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0174971680321256e-05,
+      "loss": 0.7168,
+      "step": 14338
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0173820667176101e-05,
+      "loss": 0.7256,
+      "step": 14339
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0172669651727416e-05,
+      "loss": 0.7158,
+      "step": 14340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0171518633990451e-05,
+      "loss": 0.7246,
+      "step": 14341
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0170367613980463e-05,
+      "loss": 0.7344,
+      "step": 14342
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0169216591712707e-05,
+      "loss": 0.7314,
+      "step": 14343
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0168065567202433e-05,
+      "loss": 0.7168,
+      "step": 14344
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0166914540464894e-05,
+      "loss": 0.7314,
+      "step": 14345
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0165763511515349e-05,
+      "loss": 0.7334,
+      "step": 14346
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.016461248036905e-05,
+      "loss": 0.7119,
+      "step": 14347
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0163461447041245e-05,
+      "loss": 0.7363,
+      "step": 14348
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.01623104115472e-05,
+      "loss": 0.7158,
+      "step": 14349
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0161159373902157e-05,
+      "loss": 0.7168,
+      "step": 14350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0160008334121377e-05,
+      "loss": 0.7539,
+      "step": 14351
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0158857292220105e-05,
+      "loss": 0.7207,
+      "step": 14352
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.015770624821361e-05,
+      "loss": 0.7334,
+      "step": 14353
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0156555202117134e-05,
+      "loss": 0.7422,
+      "step": 14354
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0155404153945936e-05,
+      "loss": 0.7246,
+      "step": 14355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0154253103715268e-05,
+      "loss": 0.7256,
+      "step": 14356
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0153102051440387e-05,
+      "loss": 0.7246,
+      "step": 14357
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0151950997136539e-05,
+      "loss": 0.7031,
+      "step": 14358
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0150799940818992e-05,
+      "loss": 0.7207,
+      "step": 14359
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0149648882502987e-05,
+      "loss": 0.7148,
+      "step": 14360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0148497822203787e-05,
+      "loss": 0.7207,
+      "step": 14361
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.014734675993664e-05,
+      "loss": 0.7197,
+      "step": 14362
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0146195695716806e-05,
+      "loss": 0.7432,
+      "step": 14363
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0145044629559536e-05,
+      "loss": 0.7129,
+      "step": 14364
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.014389356148008e-05,
+      "loss": 0.709,
+      "step": 14365
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0142742491493702e-05,
+      "loss": 0.7266,
+      "step": 14366
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0141591419615651e-05,
+      "loss": 0.7227,
+      "step": 14367
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0140440345861182e-05,
+      "loss": 0.75,
+      "step": 14368
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013928927024555e-05,
+      "loss": 0.7305,
+      "step": 14369
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0138138192784009e-05,
+      "loss": 0.7363,
+      "step": 14370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013698711349181e-05,
+      "loss": 0.71,
+      "step": 14371
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0135836032384211e-05,
+      "loss": 0.7207,
+      "step": 14372
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013468494947647e-05,
+      "loss": 0.7021,
+      "step": 14373
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0133533864783836e-05,
+      "loss": 0.7383,
+      "step": 14374
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0132382778321564e-05,
+      "loss": 0.7344,
+      "step": 14375
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0131231690104912e-05,
+      "loss": 0.7617,
+      "step": 14376
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.013008060014913e-05,
+      "loss": 0.7314,
+      "step": 14377
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0128929508469477e-05,
+      "loss": 0.7305,
+      "step": 14378
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0127778415081206e-05,
+      "loss": 0.7266,
+      "step": 14379
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0126627319999572e-05,
+      "loss": 0.7383,
+      "step": 14380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0125476223239827e-05,
+      "loss": 0.7471,
+      "step": 14381
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0124325124817234e-05,
+      "loss": 0.748,
+      "step": 14382
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0123174024747034e-05,
+      "loss": 0.6953,
+      "step": 14383
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0122022923044494e-05,
+      "loss": 0.7324,
+      "step": 14384
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0120871819724865e-05,
+      "loss": 0.7236,
+      "step": 14385
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0119720714803398e-05,
+      "loss": 0.7041,
+      "step": 14386
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0118569608295353e-05,
+      "loss": 0.7324,
+      "step": 14387
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0117418500215983e-05,
+      "loss": 0.7471,
+      "step": 14388
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.011626739058054e-05,
+      "loss": 0.7451,
+      "step": 14389
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0115116279404285e-05,
+      "loss": 0.7461,
+      "step": 14390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0113965166702467e-05,
+      "loss": 0.7168,
+      "step": 14391
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0112814052490348e-05,
+      "loss": 0.7188,
+      "step": 14392
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0111662936783174e-05,
+      "loss": 0.7051,
+      "step": 14393
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0110511819596204e-05,
+      "loss": 0.7119,
+      "step": 14394
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0109360700944698e-05,
+      "loss": 0.7119,
+      "step": 14395
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0108209580843901e-05,
+      "loss": 0.7275,
+      "step": 14396
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0107058459309076e-05,
+      "loss": 0.7637,
+      "step": 14397
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0105907336355479e-05,
+      "loss": 0.7363,
+      "step": 14398
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0104756211998357e-05,
+      "loss": 0.7295,
+      "step": 14399
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.010360508625297e-05,
+      "loss": 0.7344,
+      "step": 14400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0102453959134575e-05,
+      "loss": 0.707,
+      "step": 14401
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0101302830658425e-05,
+      "loss": 0.6943,
+      "step": 14402
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0100151700839771e-05,
+      "loss": 0.7451,
+      "step": 14403
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0099000569693877e-05,
+      "loss": 0.7246,
+      "step": 14404
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.009784943723599e-05,
+      "loss": 0.7471,
+      "step": 14405
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0096698303481374e-05,
+      "loss": 0.7305,
+      "step": 14406
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0095547168445275e-05,
+      "loss": 0.7373,
+      "step": 14407
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0094396032142952e-05,
+      "loss": 0.7197,
+      "step": 14408
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0093244894589662e-05,
+      "loss": 0.7158,
+      "step": 14409
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0092093755800655e-05,
+      "loss": 0.7197,
+      "step": 14410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0090942615791193e-05,
+      "loss": 0.7158,
+      "step": 14411
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008979147457653e-05,
+      "loss": 0.7354,
+      "step": 14412
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0088640332171917e-05,
+      "loss": 0.7168,
+      "step": 14413
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0087489188592612e-05,
+      "loss": 0.7441,
+      "step": 14414
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0086338043853871e-05,
+      "loss": 0.7471,
+      "step": 14415
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0085186897970947e-05,
+      "loss": 0.7158,
+      "step": 14416
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0084035750959099e-05,
+      "loss": 0.7236,
+      "step": 14417
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0082884602833578e-05,
+      "loss": 0.7324,
+      "step": 14418
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0081733453609646e-05,
+      "loss": 0.7412,
+      "step": 14419
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0080582303302548e-05,
+      "loss": 0.751,
+      "step": 14420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0079431151927552e-05,
+      "loss": 0.748,
+      "step": 14421
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0078279999499903e-05,
+      "loss": 0.7441,
+      "step": 14422
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0077128846034863e-05,
+      "loss": 0.7178,
+      "step": 14423
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0075977691547683e-05,
+      "loss": 0.7295,
+      "step": 14424
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0074826536053622e-05,
+      "loss": 0.7227,
+      "step": 14425
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0073675379567932e-05,
+      "loss": 0.7529,
+      "step": 14426
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0072524222105875e-05,
+      "loss": 0.7725,
+      "step": 14427
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0071373063682696e-05,
+      "loss": 0.7422,
+      "step": 14428
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0070221904313662e-05,
+      "loss": 0.7295,
+      "step": 14429
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.006907074401402e-05,
+      "loss": 0.7383,
+      "step": 14430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.006791958279903e-05,
+      "loss": 0.7295,
+      "step": 14431
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0066768420683945e-05,
+      "loss": 0.7178,
+      "step": 14432
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0065617257684023e-05,
+      "loss": 0.752,
+      "step": 14433
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0064466093814522e-05,
+      "loss": 0.7178,
+      "step": 14434
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0063314929090691e-05,
+      "loss": 0.7441,
+      "step": 14435
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0062163763527786e-05,
+      "loss": 0.7432,
+      "step": 14436
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.006101259714107e-05,
+      "loss": 0.707,
+      "step": 14437
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0059861429945793e-05,
+      "loss": 0.7373,
+      "step": 14438
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0058710261957212e-05,
+      "loss": 0.748,
+      "step": 14439
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0057559093190581e-05,
+      "loss": 0.751,
+      "step": 14440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0056407923661158e-05,
+      "loss": 0.7295,
+      "step": 14441
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0055256753384197e-05,
+      "loss": 0.7275,
+      "step": 14442
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0054105582374958e-05,
+      "loss": 0.7275,
+      "step": 14443
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0052954410648687e-05,
+      "loss": 0.7217,
+      "step": 14444
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0051803238220651e-05,
+      "loss": 0.7686,
+      "step": 14445
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.00506520651061e-05,
+      "loss": 0.7354,
+      "step": 14446
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.004950089132029e-05,
+      "loss": 0.7148,
+      "step": 14447
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0048349716878479e-05,
+      "loss": 0.7451,
+      "step": 14448
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0047198541795914e-05,
+      "loss": 0.7109,
+      "step": 14449
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0046047366087865e-05,
+      "loss": 0.7139,
+      "step": 14450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0044896189769577e-05,
+      "loss": 0.7666,
+      "step": 14451
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0043745012856312e-05,
+      "loss": 0.7168,
+      "step": 14452
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0042593835363322e-05,
+      "loss": 0.7412,
+      "step": 14453
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0041442657305865e-05,
+      "loss": 0.7129,
+      "step": 14454
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0040291478699191e-05,
+      "loss": 0.7275,
+      "step": 14455
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0039140299558563e-05,
+      "loss": 0.7275,
+      "step": 14456
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0037989119899236e-05,
+      "loss": 0.7275,
+      "step": 14457
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0036837939736463e-05,
+      "loss": 0.7344,
+      "step": 14458
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0035686759085499e-05,
+      "loss": 0.7295,
+      "step": 14459
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0034535577961606e-05,
+      "loss": 0.7207,
+      "step": 14460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0033384396380032e-05,
+      "loss": 0.7129,
+      "step": 14461
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.003223321435604e-05,
+      "loss": 0.7207,
+      "step": 14462
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.003108203190488e-05,
+      "loss": 0.7266,
+      "step": 14463
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0029930849041812e-05,
+      "loss": 0.7285,
+      "step": 14464
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0028779665782085e-05,
+      "loss": 0.7197,
+      "step": 14465
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0027628482140968e-05,
+      "loss": 0.7275,
+      "step": 14466
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0026477298133702e-05,
+      "loss": 0.7031,
+      "step": 14467
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0025326113775552e-05,
+      "loss": 0.7256,
+      "step": 14468
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0024174929081773e-05,
+      "loss": 0.7334,
+      "step": 14469
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023023744067619e-05,
+      "loss": 0.7373,
+      "step": 14470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0021872558748345e-05,
+      "loss": 0.7295,
+      "step": 14471
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0020721373139209e-05,
+      "loss": 0.7227,
+      "step": 14472
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0019570187255467e-05,
+      "loss": 0.7266,
+      "step": 14473
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0018419001112374e-05,
+      "loss": 0.7285,
+      "step": 14474
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0017267814725184e-05,
+      "loss": 0.7148,
+      "step": 14475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0016116628109158e-05,
+      "loss": 0.7266,
+      "step": 14476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0014965441279546e-05,
+      "loss": 0.7178,
+      "step": 14477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.001381425425161e-05,
+      "loss": 0.7217,
+      "step": 14478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0012663067040602e-05,
+      "loss": 0.6973,
+      "step": 14479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0011511879661777e-05,
+      "loss": 0.7178,
+      "step": 14480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0010360692130395e-05,
+      "loss": 0.7422,
+      "step": 14481
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.000920950446171e-05,
+      "loss": 0.7402,
+      "step": 14482
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0008058316670972e-05,
+      "loss": 0.7236,
+      "step": 14483
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0006907128773446e-05,
+      "loss": 0.7275,
+      "step": 14484
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0005755940784385e-05,
+      "loss": 0.7158,
+      "step": 14485
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0004604752719046e-05,
+      "loss": 0.7314,
+      "step": 14486
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0003453564592677e-05,
+      "loss": 0.7334,
+      "step": 14487
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0002302376420545e-05,
+      "loss": 0.7295,
+      "step": 14488
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0001151188217901e-05,
+      "loss": 0.7451,
+      "step": 14489
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 0.7373,
+      "step": 14490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.998848811782102e-06,
+      "loss": 0.7549,
+      "step": 14491
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.997697623579455e-06,
+      "loss": 0.7275,
+      "step": 14492
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996546435407323e-06,
+      "loss": 0.7217,
+      "step": 14493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99539524728096e-06,
+      "loss": 0.7314,
+      "step": 14494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.994244059215616e-06,
+      "loss": 0.708,
+      "step": 14495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993092871226555e-06,
+      "loss": 0.7197,
+      "step": 14496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.991941683329031e-06,
+      "loss": 0.7158,
+      "step": 14497
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990790495538295e-06,
+      "loss": 0.7021,
+      "step": 14498
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.989639307869607e-06,
+      "loss": 0.7197,
+      "step": 14499
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.988488120338226e-06,
+      "loss": 0.7178,
+      "step": 14500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.987336932959404e-06,
+      "loss": 0.7246,
+      "step": 14501
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.986185745748393e-06,
+      "loss": 0.7168,
+      "step": 14502
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.985034558720455e-06,
+      "loss": 0.7412,
+      "step": 14503
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983883371890843e-06,
+      "loss": 0.7305,
+      "step": 14504
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.982732185274818e-06,
+      "loss": 0.7158,
+      "step": 14505
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.981580998887626e-06,
+      "loss": 0.7324,
+      "step": 14506
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.980429812744538e-06,
+      "loss": 0.7178,
+      "step": 14507
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.979278626860794e-06,
+      "loss": 0.7441,
+      "step": 14508
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.978127441251659e-06,
+      "loss": 0.7363,
+      "step": 14509
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976976255932385e-06,
+      "loss": 0.7158,
+      "step": 14510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97582507091823e-06,
+      "loss": 0.7139,
+      "step": 14511
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97467388622445e-06,
+      "loss": 0.7207,
+      "step": 14512
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.973522701866303e-06,
+      "loss": 0.7363,
+      "step": 14513
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.972371517859039e-06,
+      "loss": 0.7178,
+      "step": 14514
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.971220334217917e-06,
+      "loss": 0.7441,
+      "step": 14515
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.970069150958193e-06,
+      "loss": 0.7178,
+      "step": 14516
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.968917968095123e-06,
+      "loss": 0.7363,
+      "step": 14517
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967766785643962e-06,
+      "loss": 0.7061,
+      "step": 14518
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.966615603619968e-06,
+      "loss": 0.7383,
+      "step": 14519
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.965464422038397e-06,
+      "loss": 0.7539,
+      "step": 14520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.964313240914504e-06,
+      "loss": 0.7139,
+      "step": 14521
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.963162060263539e-06,
+      "loss": 0.7412,
+      "step": 14522
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962010880100766e-06,
+      "loss": 0.7275,
+      "step": 14523
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960859700441439e-06,
+      "loss": 0.7305,
+      "step": 14524
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.959708521300809e-06,
+      "loss": 0.7246,
+      "step": 14525
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.958557342694142e-06,
+      "loss": 0.7178,
+      "step": 14526
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.957406164636683e-06,
+      "loss": 0.6992,
+      "step": 14527
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.956254987143692e-06,
+      "loss": 0.7256,
+      "step": 14528
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.955103810230425e-06,
+      "loss": 0.708,
+      "step": 14529
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.953952633912138e-06,
+      "loss": 0.7275,
+      "step": 14530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.952801458204086e-06,
+      "loss": 0.7236,
+      "step": 14531
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.951650283121525e-06,
+      "loss": 0.7041,
+      "step": 14532
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.950499108679714e-06,
+      "loss": 0.7139,
+      "step": 14533
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.949347934893903e-06,
+      "loss": 0.7285,
+      "step": 14534
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948196761779352e-06,
+      "loss": 0.752,
+      "step": 14535
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.947045589351314e-06,
+      "loss": 0.7129,
+      "step": 14536
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.945894417625045e-06,
+      "loss": 0.7402,
+      "step": 14537
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.944743246615803e-06,
+      "loss": 0.7432,
+      "step": 14538
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943592076338847e-06,
+      "loss": 0.7363,
+      "step": 14539
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94244090680942e-06,
+      "loss": 0.751,
+      "step": 14540
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.941289738042793e-06,
+      "loss": 0.7197,
+      "step": 14541
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.940138570054209e-06,
+      "loss": 0.748,
+      "step": 14542
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.938987402858931e-06,
+      "loss": 0.7158,
+      "step": 14543
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.937836236472214e-06,
+      "loss": 0.7461,
+      "step": 14544
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93668507090931e-06,
+      "loss": 0.7119,
+      "step": 14545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.935533906185483e-06,
+      "loss": 0.7373,
+      "step": 14546
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.934382742315978e-06,
+      "loss": 0.7207,
+      "step": 14547
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.933231579316058e-06,
+      "loss": 0.7051,
+      "step": 14548
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.932080417200972e-06,
+      "loss": 0.7236,
+      "step": 14549
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.930929255985982e-06,
+      "loss": 0.7051,
+      "step": 14550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.929778095686338e-06,
+      "loss": 0.7295,
+      "step": 14551
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.928626936317307e-06,
+      "loss": 0.707,
+      "step": 14552
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.92747577789413e-06,
+      "loss": 0.7383,
+      "step": 14553
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.92632462043207e-06,
+      "loss": 0.7344,
+      "step": 14554
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.92517346394638e-06,
+      "loss": 0.7178,
+      "step": 14555
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.924022308452318e-06,
+      "loss": 0.7227,
+      "step": 14556
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.922871153965138e-06,
+      "loss": 0.7197,
+      "step": 14557
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.921720000500098e-06,
+      "loss": 0.7119,
+      "step": 14558
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.920568848072454e-06,
+      "loss": 0.7021,
+      "step": 14559
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.919417696697454e-06,
+      "loss": 0.7578,
+      "step": 14560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.918266546390358e-06,
+      "loss": 0.71,
+      "step": 14561
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.917115397166423e-06,
+      "loss": 0.7275,
+      "step": 14562
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.915964249040903e-06,
+      "loss": 0.7344,
+      "step": 14563
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.914813102029055e-06,
+      "loss": 0.708,
+      "step": 14564
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.913661956146134e-06,
+      "loss": 0.7285,
+      "step": 14565
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.912510811407393e-06,
+      "loss": 0.7197,
+      "step": 14566
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.911359667828087e-06,
+      "loss": 0.7178,
+      "step": 14567
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910208525423474e-06,
+      "loss": 0.7383,
+      "step": 14568
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.909057384208809e-06,
+      "loss": 0.7139,
+      "step": 14569
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.907906244199345e-06,
+      "loss": 0.7227,
+      "step": 14570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.90675510541034e-06,
+      "loss": 0.7305,
+      "step": 14571
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.905603967857053e-06,
+      "loss": 0.7402,
+      "step": 14572
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.904452831554728e-06,
+      "loss": 0.7266,
+      "step": 14573
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.90330169651863e-06,
+      "loss": 0.7314,
+      "step": 14574
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.902150562764011e-06,
+      "loss": 0.71,
+      "step": 14575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.900999430306125e-06,
+      "loss": 0.7246,
+      "step": 14576
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.899848299160229e-06,
+      "loss": 0.7227,
+      "step": 14577
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89869716934158e-06,
+      "loss": 0.7139,
+      "step": 14578
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.897546040865429e-06,
+      "loss": 0.749,
+      "step": 14579
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.896394913747032e-06,
+      "loss": 0.7549,
+      "step": 14580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.895243788001646e-06,
+      "loss": 0.7256,
+      "step": 14581
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.894092663644524e-06,
+      "loss": 0.7314,
+      "step": 14582
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.892941540690924e-06,
+      "loss": 0.7402,
+      "step": 14583
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.891790419156097e-06,
+      "loss": 0.7129,
+      "step": 14584
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.890639299055305e-06,
+      "loss": 0.7354,
+      "step": 14585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.889488180403797e-06,
+      "loss": 0.7578,
+      "step": 14586
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88833706321683e-06,
+      "loss": 0.7266,
+      "step": 14587
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.887185947509655e-06,
+      "loss": 0.7178,
+      "step": 14588
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.886034833297533e-06,
+      "loss": 0.707,
+      "step": 14589
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.884883720595717e-06,
+      "loss": 0.7246,
+      "step": 14590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.883732609419464e-06,
+      "loss": 0.7432,
+      "step": 14591
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.882581499784022e-06,
+      "loss": 0.7412,
+      "step": 14592
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.881430391704652e-06,
+      "loss": 0.749,
+      "step": 14593
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.880279285196605e-06,
+      "loss": 0.7441,
+      "step": 14594
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.879128180275138e-06,
+      "loss": 0.7178,
+      "step": 14595
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.877977076955508e-06,
+      "loss": 0.707,
+      "step": 14596
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.876825975252966e-06,
+      "loss": 0.7334,
+      "step": 14597
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.875674875182773e-06,
+      "loss": 0.7383,
+      "step": 14598
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.874523776760175e-06,
+      "loss": 0.7275,
+      "step": 14599
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.87337268000043e-06,
+      "loss": 0.7324,
+      "step": 14600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.872221584918796e-06,
+      "loss": 0.7393,
+      "step": 14601
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.871070491530523e-06,
+      "loss": 0.7441,
+      "step": 14602
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.869919399850872e-06,
+      "loss": 0.7178,
+      "step": 14603
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.868768309895093e-06,
+      "loss": 0.7402,
+      "step": 14604
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.867617221678441e-06,
+      "loss": 0.7246,
+      "step": 14605
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.866466135216169e-06,
+      "loss": 0.7334,
+      "step": 14606
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.865315050523535e-06,
+      "loss": 0.7529,
+      "step": 14607
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86416396761579e-06,
+      "loss": 0.7188,
+      "step": 14608
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.863012886508192e-06,
+      "loss": 0.7256,
+      "step": 14609
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.861861807215998e-06,
+      "loss": 0.7314,
+      "step": 14610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.860710729754455e-06,
+      "loss": 0.7236,
+      "step": 14611
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.859559654138821e-06,
+      "loss": 0.7139,
+      "step": 14612
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.85840858038435e-06,
+      "loss": 0.7568,
+      "step": 14613
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.857257508506301e-06,
+      "loss": 0.7256,
+      "step": 14614
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.85610643851992e-06,
+      "loss": 0.7246,
+      "step": 14615
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.854955370440467e-06,
+      "loss": 0.7217,
+      "step": 14616
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.8538043042832e-06,
+      "loss": 0.7432,
+      "step": 14617
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.852653240063362e-06,
+      "loss": 0.6963,
+      "step": 14618
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.851502177796218e-06,
+      "loss": 0.7178,
+      "step": 14619
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.850351117497016e-06,
+      "loss": 0.71,
+      "step": 14620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.849200059181012e-06,
+      "loss": 0.7197,
+      "step": 14621
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84804900286346e-06,
+      "loss": 0.7119,
+      "step": 14622
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84689794855962e-06,
+      "loss": 0.7275,
+      "step": 14623
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.845746896284736e-06,
+      "loss": 0.7227,
+      "step": 14624
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.844595846054067e-06,
+      "loss": 0.7354,
+      "step": 14625
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.843444797882868e-06,
+      "loss": 0.7197,
+      "step": 14626
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.842293751786391e-06,
+      "loss": 0.7412,
+      "step": 14627
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.841142707779893e-06,
+      "loss": 0.7324,
+      "step": 14628
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.839991665878625e-06,
+      "loss": 0.7305,
+      "step": 14629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838840626097848e-06,
+      "loss": 0.7285,
+      "step": 14630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.837689588452805e-06,
+      "loss": 0.7236,
+      "step": 14631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.836538552958756e-06,
+      "loss": 0.7246,
+      "step": 14632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.835387519630954e-06,
+      "loss": 0.7393,
+      "step": 14633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.834236488484653e-06,
+      "loss": 0.7188,
+      "step": 14634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.833085459535106e-06,
+      "loss": 0.7598,
+      "step": 14635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.831934432797574e-06,
+      "loss": 0.7344,
+      "step": 14636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.830783408287298e-06,
+      "loss": 0.7832,
+      "step": 14637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82963238601954e-06,
+      "loss": 0.7246,
+      "step": 14638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82848136600955e-06,
+      "loss": 0.7422,
+      "step": 14639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.827330348272587e-06,
+      "loss": 0.7109,
+      "step": 14640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.8261793328239e-06,
+      "loss": 0.7568,
+      "step": 14641
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.825028319678745e-06,
+      "loss": 0.7529,
+      "step": 14642
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.823877308852376e-06,
+      "loss": 0.7393,
+      "step": 14643
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.822726300360044e-06,
+      "loss": 0.7568,
+      "step": 14644
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.821575294217002e-06,
+      "loss": 0.7178,
+      "step": 14645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.820424290438508e-06,
+      "loss": 0.7373,
+      "step": 14646
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.81927328903981e-06,
+      "loss": 0.7285,
+      "step": 14647
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.818122290036167e-06,
+      "loss": 0.7266,
+      "step": 14648
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.816971293442832e-06,
+      "loss": 0.7676,
+      "step": 14649
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.815820299275055e-06,
+      "loss": 0.7305,
+      "step": 14650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.814669307548086e-06,
+      "loss": 0.7422,
+      "step": 14651
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.813518318277186e-06,
+      "loss": 0.7285,
+      "step": 14652
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.812367331477605e-06,
+      "loss": 0.7344,
+      "step": 14653
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.811216347164596e-06,
+      "loss": 0.751,
+      "step": 14654
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.810065365353413e-06,
+      "loss": 0.7246,
+      "step": 14655
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.808914386059313e-06,
+      "loss": 0.7344,
+      "step": 14656
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.80776340929754e-06,
+      "loss": 0.7041,
+      "step": 14657
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.806612435083354e-06,
+      "loss": 0.7432,
+      "step": 14658
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.805461463432006e-06,
+      "loss": 0.7334,
+      "step": 14659
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.80431049435875e-06,
+      "loss": 0.7207,
+      "step": 14660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.803159527878838e-06,
+      "loss": 0.7275,
+      "step": 14661
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.802008564007524e-06,
+      "loss": 0.7412,
+      "step": 14662
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.800857602760059e-06,
+      "loss": 0.7148,
+      "step": 14663
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.799706644151699e-06,
+      "loss": 0.7129,
+      "step": 14664
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.798555688197695e-06,
+      "loss": 0.7383,
+      "step": 14665
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.797404734913298e-06,
+      "loss": 0.7051,
+      "step": 14666
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.796253784313765e-06,
+      "loss": 0.7529,
+      "step": 14667
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.795102836414345e-06,
+      "loss": 0.7305,
+      "step": 14668
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.793951891230297e-06,
+      "loss": 0.7402,
+      "step": 14669
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.792800948776867e-06,
+      "loss": 0.7256,
+      "step": 14670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.79165000906931e-06,
+      "loss": 0.7285,
+      "step": 14671
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.790499072122876e-06,
+      "loss": 0.7568,
+      "step": 14672
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.789348137952823e-06,
+      "loss": 0.7246,
+      "step": 14673
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.788197206574398e-06,
+      "loss": 0.7188,
+      "step": 14674
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.787046278002863e-06,
+      "loss": 0.7227,
+      "step": 14675
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.785895352253458e-06,
+      "loss": 0.7197,
+      "step": 14676
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.784744429341441e-06,
+      "loss": 0.7256,
+      "step": 14677
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.783593509282066e-06,
+      "loss": 0.7246,
+      "step": 14678
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.782442592090583e-06,
+      "loss": 0.7363,
+      "step": 14679
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.781291677782245e-06,
+      "loss": 0.749,
+      "step": 14680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.780140766372306e-06,
+      "loss": 0.752,
+      "step": 14681
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.778989857876022e-06,
+      "loss": 0.7275,
+      "step": 14682
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.777838952308635e-06,
+      "loss": 0.7412,
+      "step": 14683
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.7766880496854e-06,
+      "loss": 0.7305,
+      "step": 14684
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.775537150021576e-06,
+      "loss": 0.748,
+      "step": 14685
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.774386253332407e-06,
+      "loss": 0.7334,
+      "step": 14686
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.773235359633151e-06,
+      "loss": 0.709,
+      "step": 14687
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.772084468939061e-06,
+      "loss": 0.7246,
+      "step": 14688
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770933581265384e-06,
+      "loss": 0.7227,
+      "step": 14689
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76978269662737e-06,
+      "loss": 0.7197,
+      "step": 14690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.768631815040279e-06,
+      "loss": 0.75,
+      "step": 14691
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.767480936519355e-06,
+      "loss": 0.7148,
+      "step": 14692
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.766330061079855e-06,
+      "loss": 0.7012,
+      "step": 14693
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76517918873703e-06,
+      "loss": 0.7324,
+      "step": 14694
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.764028319506132e-06,
+      "loss": 0.749,
+      "step": 14695
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76287745340241e-06,
+      "loss": 0.7256,
+      "step": 14696
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.761726590441117e-06,
+      "loss": 0.7266,
+      "step": 14697
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.760575730637506e-06,
+      "loss": 0.7373,
+      "step": 14698
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.759424874006826e-06,
+      "loss": 0.7295,
+      "step": 14699
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.758274020564332e-06,
+      "loss": 0.7236,
+      "step": 14700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.757123170325274e-06,
+      "loss": 0.7441,
+      "step": 14701
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.755972323304901e-06,
+      "loss": 0.709,
+      "step": 14702
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.754821479518468e-06,
+      "loss": 0.7295,
+      "step": 14703
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.753670638981224e-06,
+      "loss": 0.7783,
+      "step": 14704
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75251980170842e-06,
+      "loss": 0.7383,
+      "step": 14705
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75136896771531e-06,
+      "loss": 0.7305,
+      "step": 14706
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.750218137017142e-06,
+      "loss": 0.7197,
+      "step": 14707
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.749067309629173e-06,
+      "loss": 0.7461,
+      "step": 14708
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.747916485566649e-06,
+      "loss": 0.7344,
+      "step": 14709
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.746765664844821e-06,
+      "loss": 0.7393,
+      "step": 14710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.745614847478941e-06,
+      "loss": 0.7285,
+      "step": 14711
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.74446403348426e-06,
+      "loss": 0.7275,
+      "step": 14712
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.743313222876028e-06,
+      "loss": 0.7266,
+      "step": 14713
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.742162415669505e-06,
+      "loss": 0.7461,
+      "step": 14714
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.741011611879927e-06,
+      "loss": 0.7236,
+      "step": 14715
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.739860811522553e-06,
+      "loss": 0.7412,
+      "step": 14716
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.738710014612632e-06,
+      "loss": 0.7617,
+      "step": 14717
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.737559221165418e-06,
+      "loss": 0.7451,
+      "step": 14718
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.736408431196156e-06,
+      "loss": 0.71,
+      "step": 14719
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.735257644720102e-06,
+      "loss": 0.7256,
+      "step": 14720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.734106861752508e-06,
+      "loss": 0.7383,
+      "step": 14721
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.732956082308619e-06,
+      "loss": 0.7578,
+      "step": 14722
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.731805306403683e-06,
+      "loss": 0.7236,
+      "step": 14723
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.73065453405296e-06,
+      "loss": 0.7197,
+      "step": 14724
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.729503765271692e-06,
+      "loss": 0.7256,
+      "step": 14725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.728353000075135e-06,
+      "loss": 0.7188,
+      "step": 14726
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.727202238478541e-06,
+      "loss": 0.7305,
+      "step": 14727
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.726051480497152e-06,
+      "loss": 0.7383,
+      "step": 14728
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.724900726146222e-06,
+      "loss": 0.709,
+      "step": 14729
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.723749975441003e-06,
+      "loss": 0.7178,
+      "step": 14730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.722599228396744e-06,
+      "loss": 0.7188,
+      "step": 14731
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.721448485028697e-06,
+      "loss": 0.7139,
+      "step": 14732
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.720297745352108e-06,
+      "loss": 0.7451,
+      "step": 14733
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.719147009382233e-06,
+      "loss": 0.7285,
+      "step": 14734
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.717996277134315e-06,
+      "loss": 0.7441,
+      "step": 14735
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.716845548623607e-06,
+      "loss": 0.7314,
+      "step": 14736
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.715694823865362e-06,
+      "loss": 0.7354,
+      "step": 14737
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.714544102874822e-06,
+      "loss": 0.752,
+      "step": 14738
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.713393385667245e-06,
+      "loss": 0.71,
+      "step": 14739
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.712242672257878e-06,
+      "loss": 0.7256,
+      "step": 14740
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.711091962661967e-06,
+      "loss": 0.7314,
+      "step": 14741
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.709941256894766e-06,
+      "loss": 0.7451,
+      "step": 14742
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.708790554971522e-06,
+      "loss": 0.7314,
+      "step": 14743
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.707639856907485e-06,
+      "loss": 0.7705,
+      "step": 14744
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.706489162717907e-06,
+      "loss": 0.7061,
+      "step": 14745
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.705338472418033e-06,
+      "loss": 0.7412,
+      "step": 14746
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.704187786023118e-06,
+      "loss": 0.7129,
+      "step": 14747
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.703037103548407e-06,
+      "loss": 0.7451,
+      "step": 14748
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.701886425009148e-06,
+      "loss": 0.7217,
+      "step": 14749
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.700735750420592e-06,
+      "loss": 0.7227,
+      "step": 14750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.699585079797991e-06,
+      "loss": 0.7412,
+      "step": 14751
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.69843441315659e-06,
+      "loss": 0.7188,
+      "step": 14752
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.697283750511642e-06,
+      "loss": 0.7383,
+      "step": 14753
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.696133091878393e-06,
+      "loss": 0.7256,
+      "step": 14754
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.69498243727209e-06,
+      "loss": 0.75,
+      "step": 14755
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.693831786707986e-06,
+      "loss": 0.7109,
+      "step": 14756
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.692681140201328e-06,
+      "loss": 0.6963,
+      "step": 14757
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.691530497767362e-06,
+      "loss": 0.7529,
+      "step": 14758
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.690379859421345e-06,
+      "loss": 0.7354,
+      "step": 14759
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68922922517852e-06,
+      "loss": 0.7178,
+      "step": 14760
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.688078595054136e-06,
+      "loss": 0.7295,
+      "step": 14761
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.686927969063438e-06,
+      "loss": 0.7324,
+      "step": 14762
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68577734722168e-06,
+      "loss": 0.7178,
+      "step": 14763
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.684626729544105e-06,
+      "loss": 0.7432,
+      "step": 14764
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68347611604597e-06,
+      "loss": 0.7031,
+      "step": 14765
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.682325506742518e-06,
+      "loss": 0.7266,
+      "step": 14766
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.681174901648995e-06,
+      "loss": 0.7324,
+      "step": 14767
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.680024300780652e-06,
+      "loss": 0.7529,
+      "step": 14768
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.678873704152737e-06,
+      "loss": 0.7227,
+      "step": 14769
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.677723111780495e-06,
+      "loss": 0.7305,
+      "step": 14770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.67657252367918e-06,
+      "loss": 0.7197,
+      "step": 14771
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.675421939864038e-06,
+      "loss": 0.7422,
+      "step": 14772
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.674271360350317e-06,
+      "loss": 0.7412,
+      "step": 14773
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.673120785153259e-06,
+      "loss": 0.75,
+      "step": 14774
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.671970214288119e-06,
+      "loss": 0.7109,
+      "step": 14775
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.67081964777014e-06,
+      "loss": 0.7207,
+      "step": 14776
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.669669085614575e-06,
+      "loss": 0.7217,
+      "step": 14777
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.668518527836668e-06,
+      "loss": 0.7412,
+      "step": 14778
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.66736797445167e-06,
+      "loss": 0.7129,
+      "step": 14779
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.66621742547482e-06,
+      "loss": 0.7324,
+      "step": 14780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.665066880921374e-06,
+      "loss": 0.7451,
+      "step": 14781
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.663916340806577e-06,
+      "loss": 0.7383,
+      "step": 14782
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662765805145675e-06,
+      "loss": 0.7305,
+      "step": 14783
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.661615273953916e-06,
+      "loss": 0.7422,
+      "step": 14784
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.660464747246547e-06,
+      "loss": 0.7246,
+      "step": 14785
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.659314225038818e-06,
+      "loss": 0.7275,
+      "step": 14786
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.658163707345973e-06,
+      "loss": 0.7539,
+      "step": 14787
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.657013194183262e-06,
+      "loss": 0.7178,
+      "step": 14788
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.655862685565924e-06,
+      "loss": 0.7412,
+      "step": 14789
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.654712181509215e-06,
+      "loss": 0.7168,
+      "step": 14790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.653561682028378e-06,
+      "loss": 0.71,
+      "step": 14791
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.652411187138665e-06,
+      "loss": 0.7383,
+      "step": 14792
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.651260696855314e-06,
+      "loss": 0.749,
+      "step": 14793
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.650110211193578e-06,
+      "loss": 0.7275,
+      "step": 14794
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.648959730168697e-06,
+      "loss": 0.7559,
+      "step": 14795
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.647809253795925e-06,
+      "loss": 0.7549,
+      "step": 14796
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.646658782090505e-06,
+      "loss": 0.7158,
+      "step": 14797
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.645508315067686e-06,
+      "loss": 0.7207,
+      "step": 14798
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.644357852742713e-06,
+      "loss": 0.7441,
+      "step": 14799
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.643207395130831e-06,
+      "loss": 0.7344,
+      "step": 14800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.642056942247284e-06,
+      "loss": 0.7197,
+      "step": 14801
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.640906494107325e-06,
+      "loss": 0.7178,
+      "step": 14802
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.639756050726193e-06,
+      "loss": 0.709,
+      "step": 14803
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.638605612119139e-06,
+      "loss": 0.7334,
+      "step": 14804
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63745517830141e-06,
+      "loss": 0.7188,
+      "step": 14805
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.636304749288248e-06,
+      "loss": 0.75,
+      "step": 14806
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.635154325094897e-06,
+      "loss": 0.7363,
+      "step": 14807
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63400390573661e-06,
+      "loss": 0.7285,
+      "step": 14808
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.632853491228625e-06,
+      "loss": 0.7373,
+      "step": 14809
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.631703081586193e-06,
+      "loss": 0.7314,
+      "step": 14810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63055267682456e-06,
+      "loss": 0.7109,
+      "step": 14811
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62940227695897e-06,
+      "loss": 0.7354,
+      "step": 14812
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.628251882004666e-06,
+      "loss": 0.7295,
+      "step": 14813
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.627101491976898e-06,
+      "loss": 0.7402,
+      "step": 14814
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.625951106890904e-06,
+      "loss": 0.7324,
+      "step": 14815
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62480072676194e-06,
+      "loss": 0.7471,
+      "step": 14816
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.623650351605242e-06,
+      "loss": 0.7314,
+      "step": 14817
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.622499981436065e-06,
+      "loss": 0.7275,
+      "step": 14818
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62134961626964e-06,
+      "loss": 0.7295,
+      "step": 14819
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.620199256121224e-06,
+      "loss": 0.7539,
+      "step": 14820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.619048901006056e-06,
+      "loss": 0.7129,
+      "step": 14821
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.617898550939381e-06,
+      "loss": 0.7549,
+      "step": 14822
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.616748205936447e-06,
+      "loss": 0.7324,
+      "step": 14823
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.615597866012501e-06,
+      "loss": 0.7109,
+      "step": 14824
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.61444753118278e-06,
+      "loss": 0.7275,
+      "step": 14825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.613297201462534e-06,
+      "loss": 0.7197,
+      "step": 14826
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.612146876867006e-06,
+      "loss": 0.7393,
+      "step": 14827
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.61099655741144e-06,
+      "loss": 0.7197,
+      "step": 14828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60984624311108e-06,
+      "loss": 0.7344,
+      "step": 14829
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.608695933981172e-06,
+      "loss": 0.7188,
+      "step": 14830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.607545630036964e-06,
+      "loss": 0.7168,
+      "step": 14831
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.606395331293693e-06,
+      "loss": 0.7012,
+      "step": 14832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.605245037766606e-06,
+      "loss": 0.7451,
+      "step": 14833
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.604094749470946e-06,
+      "loss": 0.7373,
+      "step": 14834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60294446642196e-06,
+      "loss": 0.7314,
+      "step": 14835
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.601794188634888e-06,
+      "loss": 0.7334,
+      "step": 14836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.60064391612498e-06,
+      "loss": 0.7012,
+      "step": 14837
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.599493648907475e-06,
+      "loss": 0.7256,
+      "step": 14838
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.598343386997616e-06,
+      "loss": 0.7119,
+      "step": 14839
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.597193130410645e-06,
+      "loss": 0.7529,
+      "step": 14840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.596042879161814e-06,
+      "loss": 0.7432,
+      "step": 14841
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.594892633266358e-06,
+      "loss": 0.7598,
+      "step": 14842
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.593742392739526e-06,
+      "loss": 0.7031,
+      "step": 14843
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.592592157596562e-06,
+      "loss": 0.7178,
+      "step": 14844
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.591441927852702e-06,
+      "loss": 0.75,
+      "step": 14845
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.590291703523195e-06,
+      "loss": 0.7217,
+      "step": 14846
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.589141484623284e-06,
+      "loss": 0.7158,
+      "step": 14847
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.587991271168207e-06,
+      "loss": 0.7578,
+      "step": 14848
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.586841063173215e-06,
+      "loss": 0.7227,
+      "step": 14849
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.585690860653548e-06,
+      "loss": 0.7324,
+      "step": 14850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.584540663624448e-06,
+      "loss": 0.7168,
+      "step": 14851
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.583390472101153e-06,
+      "loss": 0.7422,
+      "step": 14852
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.582240286098914e-06,
+      "loss": 0.7344,
+      "step": 14853
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.581090105632967e-06,
+      "loss": 0.7051,
+      "step": 14854
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57993993071856e-06,
+      "loss": 0.7412,
+      "step": 14855
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.578789761370935e-06,
+      "loss": 0.7402,
+      "step": 14856
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.577639597605332e-06,
+      "loss": 0.7334,
+      "step": 14857
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576489439436992e-06,
+      "loss": 0.7344,
+      "step": 14858
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57533928688116e-06,
+      "loss": 0.7119,
+      "step": 14859
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.574189139953076e-06,
+      "loss": 0.7598,
+      "step": 14860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573038998667986e-06,
+      "loss": 0.7354,
+      "step": 14861
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.571888863041129e-06,
+      "loss": 0.7373,
+      "step": 14862
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.57073873308775e-06,
+      "loss": 0.7412,
+      "step": 14863
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.569588608823086e-06,
+      "loss": 0.748,
+      "step": 14864
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.56843849026238e-06,
+      "loss": 0.7119,
+      "step": 14865
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.567288377420877e-06,
+      "loss": 0.7217,
+      "step": 14866
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.566138270313816e-06,
+      "loss": 0.7363,
+      "step": 14867
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.564988168956442e-06,
+      "loss": 0.7139,
+      "step": 14868
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.56383807336399e-06,
+      "loss": 0.7402,
+      "step": 14869
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.56268798355171e-06,
+      "loss": 0.7432,
+      "step": 14870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.561537899534839e-06,
+      "loss": 0.7432,
+      "step": 14871
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.560387821328616e-06,
+      "loss": 0.7295,
+      "step": 14872
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.559237748948283e-06,
+      "loss": 0.7285,
+      "step": 14873
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.558087682409084e-06,
+      "loss": 0.7217,
+      "step": 14874
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.556937621726257e-06,
+      "loss": 0.7441,
+      "step": 14875
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.555787566915049e-06,
+      "loss": 0.7432,
+      "step": 14876
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.554637517990694e-06,
+      "loss": 0.7158,
+      "step": 14877
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.553487474968437e-06,
+      "loss": 0.7334,
+      "step": 14878
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.552337437863514e-06,
+      "loss": 0.7373,
+      "step": 14879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55118740669117e-06,
+      "loss": 0.7139,
+      "step": 14880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.550037381466644e-06,
+      "loss": 0.7305,
+      "step": 14881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.54888736220518e-06,
+      "loss": 0.7412,
+      "step": 14882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.547737348922015e-06,
+      "loss": 0.7041,
+      "step": 14883
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.54658734163239e-06,
+      "loss": 0.7236,
+      "step": 14884
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.545437340351543e-06,
+      "loss": 0.7275,
+      "step": 14885
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.544287345094716e-06,
+      "loss": 0.7295,
+      "step": 14886
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.543137355877149e-06,
+      "loss": 0.7432,
+      "step": 14887
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.541987372714086e-06,
+      "loss": 0.7168,
+      "step": 14888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.540837395620763e-06,
+      "loss": 0.7285,
+      "step": 14889
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53968742461242e-06,
+      "loss": 0.7197,
+      "step": 14890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.538537459704294e-06,
+      "loss": 0.7363,
+      "step": 14891
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.537387500911631e-06,
+      "loss": 0.7061,
+      "step": 14892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.536237548249664e-06,
+      "loss": 0.7168,
+      "step": 14893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53508760173364e-06,
+      "loss": 0.7266,
+      "step": 14894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.533937661378795e-06,
+      "loss": 0.7363,
+      "step": 14895
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.53278772720037e-06,
+      "loss": 0.7373,
+      "step": 14896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.531637799213597e-06,
+      "loss": 0.7227,
+      "step": 14897
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.530487877433723e-06,
+      "loss": 0.751,
+      "step": 14898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.529337961875981e-06,
+      "loss": 0.748,
+      "step": 14899
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52818805255562e-06,
+      "loss": 0.7393,
+      "step": 14900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.527038149487869e-06,
+      "loss": 0.7178,
+      "step": 14901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.525888252687975e-06,
+      "loss": 0.7402,
+      "step": 14902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.524738362171167e-06,
+      "loss": 0.7207,
+      "step": 14903
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.523588477952693e-06,
+      "loss": 0.7324,
+      "step": 14904
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.522438600047785e-06,
+      "loss": 0.7373,
+      "step": 14905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.521288728471688e-06,
+      "loss": 0.7021,
+      "step": 14906
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.520138863239635e-06,
+      "loss": 0.7461,
+      "step": 14907
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.518989004366865e-06,
+      "loss": 0.7383,
+      "step": 14908
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.517839151868622e-06,
+      "loss": 0.7529,
+      "step": 14909
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.516689305760139e-06,
+      "loss": 0.7129,
+      "step": 14910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.515539466056654e-06,
+      "loss": 0.7305,
+      "step": 14911
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.514389632773405e-06,
+      "loss": 0.7363,
+      "step": 14912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.513239805925631e-06,
+      "loss": 0.7373,
+      "step": 14913
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.51208998552857e-06,
+      "loss": 0.7334,
+      "step": 14914
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.510940171597464e-06,
+      "loss": 0.7158,
+      "step": 14915
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.509790364147546e-06,
+      "loss": 0.7266,
+      "step": 14916
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.508640563194052e-06,
+      "loss": 0.7295,
+      "step": 14917
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50749076875222e-06,
+      "loss": 0.6973,
+      "step": 14918
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.506340980837294e-06,
+      "loss": 0.7197,
+      "step": 14919
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.505191199464501e-06,
+      "loss": 0.7188,
+      "step": 14920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.504041424649087e-06,
+      "loss": 0.709,
+      "step": 14921
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50289165640629e-06,
+      "loss": 0.7383,
+      "step": 14922
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50174189475134e-06,
+      "loss": 0.7139,
+      "step": 14923
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.500592139699475e-06,
+      "loss": 0.7256,
+      "step": 14924
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.499442391265939e-06,
+      "loss": 0.7354,
+      "step": 14925
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.49829264946596e-06,
+      "loss": 0.7305,
+      "step": 14926
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.497142914314781e-06,
+      "loss": 0.7217,
+      "step": 14927
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.49599318582764e-06,
+      "loss": 0.7178,
+      "step": 14928
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.494843464019767e-06,
+      "loss": 0.7207,
+      "step": 14929
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.493693748906399e-06,
+      "loss": 0.7256,
+      "step": 14930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.492544040502779e-06,
+      "loss": 0.749,
+      "step": 14931
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.491394338824137e-06,
+      "loss": 0.7295,
+      "step": 14932
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.490244643885713e-06,
+      "loss": 0.752,
+      "step": 14933
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.489094955702742e-06,
+      "loss": 0.7305,
+      "step": 14934
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.487945274290464e-06,
+      "loss": 0.7275,
+      "step": 14935
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.486795599664105e-06,
+      "loss": 0.7188,
+      "step": 14936
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.48564593183891e-06,
+      "loss": 0.7305,
+      "step": 14937
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.484496270830107e-06,
+      "loss": 0.7236,
+      "step": 14938
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.48334661665294e-06,
+      "loss": 0.7363,
+      "step": 14939
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.482196969322642e-06,
+      "loss": 0.6982,
+      "step": 14940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.48104732885445e-06,
+      "loss": 0.7432,
+      "step": 14941
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47989769526359e-06,
+      "loss": 0.708,
+      "step": 14942
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.478748068565307e-06,
+      "loss": 0.7334,
+      "step": 14943
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.477598448774834e-06,
+      "loss": 0.7344,
+      "step": 14944
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.476448835907404e-06,
+      "loss": 0.708,
+      "step": 14945
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.475299229978257e-06,
+      "loss": 0.7227,
+      "step": 14946
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47414963100262e-06,
+      "loss": 0.7256,
+      "step": 14947
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.473000038995737e-06,
+      "loss": 0.7363,
+      "step": 14948
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.471850453972839e-06,
+      "loss": 0.71,
+      "step": 14949
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.470700875949157e-06,
+      "loss": 0.7305,
+      "step": 14950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.469551304939927e-06,
+      "loss": 0.7188,
+      "step": 14951
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468401740960389e-06,
+      "loss": 0.7617,
+      "step": 14952
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46725218402577e-06,
+      "loss": 0.7139,
+      "step": 14953
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.466102634151312e-06,
+      "loss": 0.7168,
+      "step": 14954
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.464953091352244e-06,
+      "loss": 0.7324,
+      "step": 14955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.4638035556438e-06,
+      "loss": 0.7471,
+      "step": 14956
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.462654027041214e-06,
+      "loss": 0.7373,
+      "step": 14957
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.461504505559722e-06,
+      "loss": 0.7275,
+      "step": 14958
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.460354991214558e-06,
+      "loss": 0.7061,
+      "step": 14959
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.459205484020955e-06,
+      "loss": 0.7041,
+      "step": 14960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.458055983994148e-06,
+      "loss": 0.7227,
+      "step": 14961
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.456906491149368e-06,
+      "loss": 0.7109,
+      "step": 14962
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.455757005501847e-06,
+      "loss": 0.6904,
+      "step": 14963
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.454607527066824e-06,
+      "loss": 0.7314,
+      "step": 14964
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.453458055859526e-06,
+      "loss": 0.7617,
+      "step": 14965
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.452308591895191e-06,
+      "loss": 0.7324,
+      "step": 14966
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.451159135189053e-06,
+      "loss": 0.7256,
+      "step": 14967
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.450009685756342e-06,
+      "loss": 0.7158,
+      "step": 14968
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.448860243612287e-06,
+      "loss": 0.7295,
+      "step": 14969
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.447710808772129e-06,
+      "loss": 0.7666,
+      "step": 14970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.446561381251095e-06,
+      "loss": 0.7344,
+      "step": 14971
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44541196106442e-06,
+      "loss": 0.71,
+      "step": 14972
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.444262548227338e-06,
+      "loss": 0.749,
+      "step": 14973
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.443113142755079e-06,
+      "loss": 0.7373,
+      "step": 14974
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.441963744662874e-06,
+      "loss": 0.7256,
+      "step": 14975
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.440814353965957e-06,
+      "loss": 0.7246,
+      "step": 14976
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43966497067956e-06,
+      "loss": 0.7373,
+      "step": 14977
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.438515594818915e-06,
+      "loss": 0.7051,
+      "step": 14978
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.437366226399255e-06,
+      "loss": 0.7383,
+      "step": 14979
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.436216865435814e-06,
+      "loss": 0.7393,
+      "step": 14980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.435067511943815e-06,
+      "loss": 0.7373,
+      "step": 14981
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.433918165938498e-06,
+      "loss": 0.7354,
+      "step": 14982
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43276882743509e-06,
+      "loss": 0.7393,
+      "step": 14983
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.431619496448826e-06,
+      "loss": 0.752,
+      "step": 14984
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.430470172994933e-06,
+      "loss": 0.7256,
+      "step": 14985
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.429320857088646e-06,
+      "loss": 0.7158,
+      "step": 14986
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.428171548745197e-06,
+      "loss": 0.751,
+      "step": 14987
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.427022247979813e-06,
+      "loss": 0.751,
+      "step": 14988
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.425872954807726e-06,
+      "loss": 0.7324,
+      "step": 14989
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.424723669244168e-06,
+      "loss": 0.7119,
+      "step": 14990
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.423574391304371e-06,
+      "loss": 0.7275,
+      "step": 14991
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42242512100356e-06,
+      "loss": 0.7207,
+      "step": 14992
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.421275858356977e-06,
+      "loss": 0.7432,
+      "step": 14993
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42012660337984e-06,
+      "loss": 0.7266,
+      "step": 14994
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.418977356087385e-06,
+      "loss": 0.7178,
+      "step": 14995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.41782811649484e-06,
+      "loss": 0.7422,
+      "step": 14996
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.416678884617438e-06,
+      "loss": 0.7285,
+      "step": 14997
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.415529660470404e-06,
+      "loss": 0.71,
+      "step": 14998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.414380444068977e-06,
+      "loss": 0.7285,
+      "step": 14999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.413231235428383e-06,
+      "loss": 0.7305,
+      "step": 15000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.412082034563845e-06,
+      "loss": 0.7207,
+      "step": 15001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.410932841490596e-06,
+      "loss": 0.7051,
+      "step": 15002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.409783656223871e-06,
+      "loss": 0.7324,
+      "step": 15003
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.408634478778893e-06,
+      "loss": 0.7578,
+      "step": 15004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.407485309170895e-06,
+      "loss": 0.7109,
+      "step": 15005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.406336147415109e-06,
+      "loss": 0.7256,
+      "step": 15006
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.405186993526756e-06,
+      "loss": 0.7471,
+      "step": 15007
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.404037847521068e-06,
+      "loss": 0.7109,
+      "step": 15008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.402888709413277e-06,
+      "loss": 0.7275,
+      "step": 15009
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.401739579218607e-06,
+      "loss": 0.7344,
+      "step": 15010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.400590456952292e-06,
+      "loss": 0.7227,
+      "step": 15011
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.399441342629558e-06,
+      "loss": 0.7041,
+      "step": 15012
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.398292236265635e-06,
+      "loss": 0.7598,
+      "step": 15013
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.397143137875747e-06,
+      "loss": 0.7383,
+      "step": 15014
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.395994047475126e-06,
+      "loss": 0.7383,
+      "step": 15015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.394844965078998e-06,
+      "loss": 0.7275,
+      "step": 15016
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393695890702592e-06,
+      "loss": 0.7432,
+      "step": 15017
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39254682436114e-06,
+      "loss": 0.7344,
+      "step": 15018
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.391397766069863e-06,
+      "loss": 0.7588,
+      "step": 15019
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.390248715843991e-06,
+      "loss": 0.7129,
+      "step": 15020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.389099673698754e-06,
+      "loss": 0.7256,
+      "step": 15021
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.387950639649375e-06,
+      "loss": 0.752,
+      "step": 15022
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.386801613711087e-06,
+      "loss": 0.7441,
+      "step": 15023
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.385652595899113e-06,
+      "loss": 0.7139,
+      "step": 15024
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.38450358622868e-06,
+      "loss": 0.7285,
+      "step": 15025
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.383354584715022e-06,
+      "loss": 0.7285,
+      "step": 15026
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.382205591373357e-06,
+      "loss": 0.7344,
+      "step": 15027
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.381056606218914e-06,
+      "loss": 0.7207,
+      "step": 15028
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.379907629266925e-06,
+      "loss": 0.7432,
+      "step": 15029
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.378758660532612e-06,
+      "loss": 0.7217,
+      "step": 15030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.377609700031199e-06,
+      "loss": 0.7402,
+      "step": 15031
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.376460747777921e-06,
+      "loss": 0.7275,
+      "step": 15032
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.375311803787997e-06,
+      "loss": 0.7295,
+      "step": 15033
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.374162868076656e-06,
+      "loss": 0.7578,
+      "step": 15034
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37301394065912e-06,
+      "loss": 0.7275,
+      "step": 15035
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37186502155062e-06,
+      "loss": 0.7129,
+      "step": 15036
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37071611076638e-06,
+      "loss": 0.7383,
+      "step": 15037
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.369567208321628e-06,
+      "loss": 0.7393,
+      "step": 15038
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.368418314231589e-06,
+      "loss": 0.75,
+      "step": 15039
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367269428511486e-06,
+      "loss": 0.7275,
+      "step": 15040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.366120551176543e-06,
+      "loss": 0.7148,
+      "step": 15041
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.36497168224199e-06,
+      "loss": 0.7061,
+      "step": 15042
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.363822821723049e-06,
+      "loss": 0.7383,
+      "step": 15043
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.362673969634948e-06,
+      "loss": 0.7275,
+      "step": 15044
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.361525125992912e-06,
+      "loss": 0.7432,
+      "step": 15045
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360376290812163e-06,
+      "loss": 0.7188,
+      "step": 15046
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.359227464107923e-06,
+      "loss": 0.7236,
+      "step": 15047
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.358078645895424e-06,
+      "loss": 0.7285,
+      "step": 15048
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.356929836189883e-06,
+      "loss": 0.7285,
+      "step": 15049
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.355781035006533e-06,
+      "loss": 0.7168,
+      "step": 15050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.354632242360595e-06,
+      "loss": 0.708,
+      "step": 15051
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.353483458267289e-06,
+      "loss": 0.7305,
+      "step": 15052
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.352334682741842e-06,
+      "loss": 0.7227,
+      "step": 15053
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.351185915799478e-06,
+      "loss": 0.7207,
+      "step": 15054
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.35003715745542e-06,
+      "loss": 0.7148,
+      "step": 15055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.348888407724895e-06,
+      "loss": 0.7236,
+      "step": 15056
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.347739666623124e-06,
+      "loss": 0.7178,
+      "step": 15057
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.346590934165331e-06,
+      "loss": 0.7578,
+      "step": 15058
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.345442210366738e-06,
+      "loss": 0.7344,
+      "step": 15059
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.344293495242568e-06,
+      "loss": 0.7129,
+      "step": 15060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.343144788808047e-06,
+      "loss": 0.749,
+      "step": 15061
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.341996091078399e-06,
+      "loss": 0.7188,
+      "step": 15062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.340847402068843e-06,
+      "loss": 0.7119,
+      "step": 15063
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.339698721794604e-06,
+      "loss": 0.7246,
+      "step": 15064
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.338550050270903e-06,
+      "loss": 0.7041,
+      "step": 15065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.337401387512963e-06,
+      "loss": 0.7148,
+      "step": 15066
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.336252733536007e-06,
+      "loss": 0.7021,
+      "step": 15067
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33510408835526e-06,
+      "loss": 0.7363,
+      "step": 15068
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33395545198594e-06,
+      "loss": 0.7012,
+      "step": 15069
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.33280682444327e-06,
+      "loss": 0.7344,
+      "step": 15070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.331658205742476e-06,
+      "loss": 0.75,
+      "step": 15071
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.330509595898777e-06,
+      "loss": 0.7344,
+      "step": 15072
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32936099492739e-06,
+      "loss": 0.7236,
+      "step": 15073
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.328212402843545e-06,
+      "loss": 0.7773,
+      "step": 15074
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.327063819662458e-06,
+      "loss": 0.7627,
+      "step": 15075
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.325915245399352e-06,
+      "loss": 0.7256,
+      "step": 15076
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.324766680069453e-06,
+      "loss": 0.7363,
+      "step": 15077
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.323618123687973e-06,
+      "loss": 0.7578,
+      "step": 15078
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32246957627014e-06,
+      "loss": 0.7021,
+      "step": 15079
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32132103783117e-06,
+      "loss": 0.7129,
+      "step": 15080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.32017250838629e-06,
+      "loss": 0.7373,
+      "step": 15081
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.319023987950713e-06,
+      "loss": 0.7246,
+      "step": 15082
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.317875476539666e-06,
+      "loss": 0.7227,
+      "step": 15083
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.316726974168369e-06,
+      "loss": 0.708,
+      "step": 15084
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.31557848085204e-06,
+      "loss": 0.7197,
+      "step": 15085
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.314429996605897e-06,
+      "loss": 0.7383,
+      "step": 15086
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.313281521445165e-06,
+      "loss": 0.7031,
+      "step": 15087
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.31213305538506e-06,
+      "loss": 0.7363,
+      "step": 15088
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.310984598440806e-06,
+      "loss": 0.7432,
+      "step": 15089
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.309836150627624e-06,
+      "loss": 0.7256,
+      "step": 15090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.308687711960727e-06,
+      "loss": 0.7344,
+      "step": 15091
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.307539282455334e-06,
+      "loss": 0.7441,
+      "step": 15092
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.306390862126672e-06,
+      "loss": 0.7031,
+      "step": 15093
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.305242450989954e-06,
+      "loss": 0.7295,
+      "step": 15094
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.304094049060402e-06,
+      "loss": 0.7393,
+      "step": 15095
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.302945656353236e-06,
+      "loss": 0.7246,
+      "step": 15096
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.301797272883676e-06,
+      "loss": 0.7275,
+      "step": 15097
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.300648898666932e-06,
+      "loss": 0.7266,
+      "step": 15098
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.299500533718233e-06,
+      "loss": 0.75,
+      "step": 15099
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.298352178052791e-06,
+      "loss": 0.7354,
+      "step": 15100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297203831685828e-06,
+      "loss": 0.7197,
+      "step": 15101
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.296055494632561e-06,
+      "loss": 0.7422,
+      "step": 15102
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29490716690821e-06,
+      "loss": 0.7256,
+      "step": 15103
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.293758848527987e-06,
+      "loss": 0.7441,
+      "step": 15104
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.292610539507116e-06,
+      "loss": 0.7314,
+      "step": 15105
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.291462239860813e-06,
+      "loss": 0.7021,
+      "step": 15106
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.290313949604296e-06,
+      "loss": 0.7129,
+      "step": 15107
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.289165668752781e-06,
+      "loss": 0.7393,
+      "step": 15108
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.288017397321486e-06,
+      "loss": 0.7188,
+      "step": 15109
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.286869135325632e-06,
+      "loss": 0.748,
+      "step": 15110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.28572088278043e-06,
+      "loss": 0.7207,
+      "step": 15111
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2845726397011e-06,
+      "loss": 0.7334,
+      "step": 15112
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.283424406102861e-06,
+      "loss": 0.7168,
+      "step": 15113
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.282276182000926e-06,
+      "loss": 0.752,
+      "step": 15114
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.281127967410512e-06,
+      "loss": 0.7275,
+      "step": 15115
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.279979762346842e-06,
+      "loss": 0.7188,
+      "step": 15116
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.278831566825126e-06,
+      "loss": 0.7217,
+      "step": 15117
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.277683380860578e-06,
+      "loss": 0.7344,
+      "step": 15118
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.276535204468418e-06,
+      "loss": 0.7461,
+      "step": 15119
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.275387037663863e-06,
+      "loss": 0.7363,
+      "step": 15120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.274238880462126e-06,
+      "loss": 0.709,
+      "step": 15121
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.273090732878427e-06,
+      "loss": 0.7314,
+      "step": 15122
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.27194259492798e-06,
+      "loss": 0.7197,
+      "step": 15123
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.270794466625998e-06,
+      "loss": 0.7188,
+      "step": 15124
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.269646347987697e-06,
+      "loss": 0.7305,
+      "step": 15125
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.268498239028294e-06,
+      "loss": 0.7402,
+      "step": 15126
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.267350139763001e-06,
+      "loss": 0.7148,
+      "step": 15127
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.266202050207038e-06,
+      "loss": 0.7373,
+      "step": 15128
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.265053970375619e-06,
+      "loss": 0.7285,
+      "step": 15129
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.263905900283955e-06,
+      "loss": 0.7256,
+      "step": 15130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.262757839947262e-06,
+      "loss": 0.7168,
+      "step": 15131
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.261609789380755e-06,
+      "loss": 0.7637,
+      "step": 15132
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.260461748599649e-06,
+      "loss": 0.7324,
+      "step": 15133
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.259313717619157e-06,
+      "loss": 0.7197,
+      "step": 15134
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.258165696454492e-06,
+      "loss": 0.7227,
+      "step": 15135
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.257017685120878e-06,
+      "loss": 0.707,
+      "step": 15136
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255869683633515e-06,
+      "loss": 0.7109,
+      "step": 15137
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.25472169200762e-06,
+      "loss": 0.7451,
+      "step": 15138
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.25357371025841e-06,
+      "loss": 0.7334,
+      "step": 15139
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2524257384011e-06,
+      "loss": 0.7334,
+      "step": 15140
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.251277776450899e-06,
+      "loss": 0.7373,
+      "step": 15141
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250129824423025e-06,
+      "loss": 0.7139,
+      "step": 15142
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248981882332683e-06,
+      "loss": 0.7432,
+      "step": 15143
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.247833950195092e-06,
+      "loss": 0.7354,
+      "step": 15144
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.246686028025465e-06,
+      "loss": 0.7266,
+      "step": 15145
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.245538115839012e-06,
+      "loss": 0.7256,
+      "step": 15146
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244390213650947e-06,
+      "loss": 0.7158,
+      "step": 15147
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.243242321476483e-06,
+      "loss": 0.7188,
+      "step": 15148
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.242094439330832e-06,
+      "loss": 0.7227,
+      "step": 15149
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.240946567229205e-06,
+      "loss": 0.7266,
+      "step": 15150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.239798705186813e-06,
+      "loss": 0.7441,
+      "step": 15151
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.238650853218871e-06,
+      "loss": 0.7334,
+      "step": 15152
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.23750301134059e-06,
+      "loss": 0.7119,
+      "step": 15153
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.236355179567179e-06,
+      "loss": 0.7441,
+      "step": 15154
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.235207357913853e-06,
+      "loss": 0.7012,
+      "step": 15155
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234059546395821e-06,
+      "loss": 0.749,
+      "step": 15156
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.232911745028292e-06,
+      "loss": 0.7275,
+      "step": 15157
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231763953826483e-06,
+      "loss": 0.748,
+      "step": 15158
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.2306161728056e-06,
+      "loss": 0.7275,
+      "step": 15159
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.229468401980855e-06,
+      "loss": 0.7236,
+      "step": 15160
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.228320641367462e-06,
+      "loss": 0.7061,
+      "step": 15161
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227172890980629e-06,
+      "loss": 0.7344,
+      "step": 15162
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.226025150835564e-06,
+      "loss": 0.7275,
+      "step": 15163
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.224877420947478e-06,
+      "loss": 0.7383,
+      "step": 15164
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.223729701331583e-06,
+      "loss": 0.7305,
+      "step": 15165
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.222581992003089e-06,
+      "loss": 0.7354,
+      "step": 15166
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.221434292977205e-06,
+      "loss": 0.71,
+      "step": 15167
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.220286604269144e-06,
+      "loss": 0.7363,
+      "step": 15168
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.219138925894112e-06,
+      "loss": 0.7441,
+      "step": 15169
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.217991257867313e-06,
+      "loss": 0.7334,
+      "step": 15170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.216843600203967e-06,
+      "loss": 0.7119,
+      "step": 15171
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.215695952919276e-06,
+      "loss": 0.7012,
+      "step": 15172
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.214548316028453e-06,
+      "loss": 0.7461,
+      "step": 15173
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.213400689546704e-06,
+      "loss": 0.7383,
+      "step": 15174
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.212253073489244e-06,
+      "loss": 0.7432,
+      "step": 15175
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.21110546787127e-06,
+      "loss": 0.7285,
+      "step": 15176
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.209957872708e-06,
+      "loss": 0.7227,
+      "step": 15177
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208810288014637e-06,
+      "loss": 0.7363,
+      "step": 15178
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.207662713806395e-06,
+      "loss": 0.7129,
+      "step": 15179
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.206515150098477e-06,
+      "loss": 0.7031,
+      "step": 15180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.205367596906096e-06,
+      "loss": 0.7334,
+      "step": 15181
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.204220054244453e-06,
+      "loss": 0.7109,
+      "step": 15182
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.20307252212876e-06,
+      "loss": 0.7197,
+      "step": 15183
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.20192500057422e-06,
+      "loss": 0.7344,
+      "step": 15184
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.200777489596045e-06,
+      "loss": 0.7217,
+      "step": 15185
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.199629989209444e-06,
+      "loss": 0.7363,
+      "step": 15186
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.198482499429617e-06,
+      "loss": 0.7383,
+      "step": 15187
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19733502027178e-06,
+      "loss": 0.7139,
+      "step": 15188
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.196187551751132e-06,
+      "loss": 0.7119,
+      "step": 15189
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19504009388288e-06,
+      "loss": 0.7031,
+      "step": 15190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.193892646682237e-06,
+      "loss": 0.7393,
+      "step": 15191
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.192745210164404e-06,
+      "loss": 0.7178,
+      "step": 15192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.191597784344586e-06,
+      "loss": 0.7295,
+      "step": 15193
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.190450369237997e-06,
+      "loss": 0.7324,
+      "step": 15194
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.189302964859834e-06,
+      "loss": 0.7109,
+      "step": 15195
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.188155571225306e-06,
+      "loss": 0.7324,
+      "step": 15196
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18700818834962e-06,
+      "loss": 0.7334,
+      "step": 15197
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18586081624798e-06,
+      "loss": 0.7031,
+      "step": 15198
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18471345493559e-06,
+      "loss": 0.7354,
+      "step": 15199
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18356610442766e-06,
+      "loss": 0.6992,
+      "step": 15200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.182418764739393e-06,
+      "loss": 0.7197,
+      "step": 15201
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18127143588599e-06,
+      "loss": 0.7334,
+      "step": 15202
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18012411788266e-06,
+      "loss": 0.7393,
+      "step": 15203
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.178976810744608e-06,
+      "loss": 0.7109,
+      "step": 15204
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.177829514487033e-06,
+      "loss": 0.7451,
+      "step": 15205
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.176682229125147e-06,
+      "loss": 0.7354,
+      "step": 15206
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175534954674152e-06,
+      "loss": 0.7246,
+      "step": 15207
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.17438769114925e-06,
+      "loss": 0.7188,
+      "step": 15208
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.173240438565642e-06,
+      "loss": 0.7266,
+      "step": 15209
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.172093196938539e-06,
+      "loss": 0.7158,
+      "step": 15210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.170945966283137e-06,
+      "loss": 0.7168,
+      "step": 15211
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.169798746614646e-06,
+      "loss": 0.7451,
+      "step": 15212
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.168651537948265e-06,
+      "loss": 0.749,
+      "step": 15213
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.167504340299208e-06,
+      "loss": 0.7275,
+      "step": 15214
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.16635715368266e-06,
+      "loss": 0.709,
+      "step": 15215
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.165209978113836e-06,
+      "loss": 0.7305,
+      "step": 15216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164062813607933e-06,
+      "loss": 0.7285,
+      "step": 15217
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.16291566018016e-06,
+      "loss": 0.7383,
+      "step": 15218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.161768517845713e-06,
+      "loss": 0.7236,
+      "step": 15219
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.160621386619804e-06,
+      "loss": 0.7451,
+      "step": 15220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.15947426651762e-06,
+      "loss": 0.7012,
+      "step": 15221
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.158327157554376e-06,
+      "loss": 0.7451,
+      "step": 15222
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.157180059745267e-06,
+      "loss": 0.7295,
+      "step": 15223
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.156032973105498e-06,
+      "loss": 0.6943,
+      "step": 15224
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154885897650272e-06,
+      "loss": 0.7246,
+      "step": 15225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.153738833394784e-06,
+      "loss": 0.707,
+      "step": 15226
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.152591780354244e-06,
+      "loss": 0.7373,
+      "step": 15227
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.151444738543846e-06,
+      "loss": 0.7236,
+      "step": 15228
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.150297707978792e-06,
+      "loss": 0.7354,
+      "step": 15229
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.149150688674287e-06,
+      "loss": 0.7207,
+      "step": 15230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.148003680645528e-06,
+      "loss": 0.7617,
+      "step": 15231
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.146856683907714e-06,
+      "loss": 0.7393,
+      "step": 15232
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.145709698476054e-06,
+      "loss": 0.7021,
+      "step": 15233
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.144562724365739e-06,
+      "loss": 0.7373,
+      "step": 15234
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.143415761591972e-06,
+      "loss": 0.7324,
+      "step": 15235
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.142268810169953e-06,
+      "loss": 0.7217,
+      "step": 15236
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.141121870114883e-06,
+      "loss": 0.7393,
+      "step": 15237
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.139974941441958e-06,
+      "loss": 0.7441,
+      "step": 15238
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.138828024166385e-06,
+      "loss": 0.7432,
+      "step": 15239
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.137681118303357e-06,
+      "loss": 0.7393,
+      "step": 15240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.136534223868071e-06,
+      "loss": 0.7236,
+      "step": 15241
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.135387340875735e-06,
+      "loss": 0.7246,
+      "step": 15242
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13424046934154e-06,
+      "loss": 0.7266,
+      "step": 15243
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.133093609280687e-06,
+      "loss": 0.7363,
+      "step": 15244
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.131946760708376e-06,
+      "loss": 0.7129,
+      "step": 15245
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.130799923639806e-06,
+      "loss": 0.7373,
+      "step": 15246
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.129653098090173e-06,
+      "loss": 0.7236,
+      "step": 15247
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.128506284074671e-06,
+      "loss": 0.7695,
+      "step": 15248
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.127359481608508e-06,
+      "loss": 0.7266,
+      "step": 15249
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.126212690706874e-06,
+      "loss": 0.7432,
+      "step": 15250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.12506591138497e-06,
+      "loss": 0.7148,
+      "step": 15251
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.123919143657992e-06,
+      "loss": 0.7549,
+      "step": 15252
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.122772387541143e-06,
+      "loss": 0.7451,
+      "step": 15253
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.121625643049612e-06,
+      "loss": 0.7549,
+      "step": 15254
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.120478910198598e-06,
+      "loss": 0.751,
+      "step": 15255
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.119332189003298e-06,
+      "loss": 0.7295,
+      "step": 15256
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.118185479478913e-06,
+      "loss": 0.7607,
+      "step": 15257
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.117038781640635e-06,
+      "loss": 0.7031,
+      "step": 15258
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.115892095503668e-06,
+      "loss": 0.748,
+      "step": 15259
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.114745421083196e-06,
+      "loss": 0.7178,
+      "step": 15260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.113598758394424e-06,
+      "loss": 0.7295,
+      "step": 15261
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.112452107452541e-06,
+      "loss": 0.7422,
+      "step": 15262
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.111305468272752e-06,
+      "loss": 0.7129,
+      "step": 15263
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.110158840870245e-06,
+      "loss": 0.7441,
+      "step": 15264
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.109012225260225e-06,
+      "loss": 0.7402,
+      "step": 15265
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.107865621457872e-06,
+      "loss": 0.7363,
+      "step": 15266
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.106719029478396e-06,
+      "loss": 0.7256,
+      "step": 15267
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.105572449336982e-06,
+      "loss": 0.7285,
+      "step": 15268
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.10442588104883e-06,
+      "loss": 0.7109,
+      "step": 15269
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.103279324629135e-06,
+      "loss": 0.7246,
+      "step": 15270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.102132780093087e-06,
+      "loss": 0.7393,
+      "step": 15271
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.100986247455888e-06,
+      "loss": 0.7578,
+      "step": 15272
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.099839726732726e-06,
+      "loss": 0.7236,
+      "step": 15273
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.098693217938795e-06,
+      "loss": 0.7314,
+      "step": 15274
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.097546721089293e-06,
+      "loss": 0.7246,
+      "step": 15275
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.096400236199411e-06,
+      "loss": 0.7314,
+      "step": 15276
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.09525376328434e-06,
+      "loss": 0.7109,
+      "step": 15277
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.094107302359284e-06,
+      "loss": 0.7256,
+      "step": 15278
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.092960853439423e-06,
+      "loss": 0.7266,
+      "step": 15279
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.091814416539956e-06,
+      "loss": 0.752,
+      "step": 15280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.090667991676078e-06,
+      "loss": 0.7295,
+      "step": 15281
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08952157886298e-06,
+      "loss": 0.7461,
+      "step": 15282
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.088375178115851e-06,
+      "loss": 0.7129,
+      "step": 15283
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08722878944989e-06,
+      "loss": 0.7314,
+      "step": 15284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08608241288029e-06,
+      "loss": 0.7266,
+      "step": 15285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.084936048422234e-06,
+      "loss": 0.7256,
+      "step": 15286
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.083789696090921e-06,
+      "loss": 0.7275,
+      "step": 15287
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.082643355901541e-06,
+      "loss": 0.7119,
+      "step": 15288
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.081497027869286e-06,
+      "loss": 0.7363,
+      "step": 15289
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.080350712009348e-06,
+      "loss": 0.7148,
+      "step": 15290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.079204408336922e-06,
+      "loss": 0.7139,
+      "step": 15291
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07805811686719e-06,
+      "loss": 0.6992,
+      "step": 15292
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.076911837615348e-06,
+      "loss": 0.7148,
+      "step": 15293
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.075765570596589e-06,
+      "loss": 0.7646,
+      "step": 15294
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0746193158261e-06,
+      "loss": 0.71,
+      "step": 15295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.073473073319075e-06,
+      "loss": 0.7285,
+      "step": 15296
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.0723268430907e-06,
+      "loss": 0.7305,
+      "step": 15297
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.071180625156176e-06,
+      "loss": 0.7617,
+      "step": 15298
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.070034419530678e-06,
+      "loss": 0.708,
+      "step": 15299
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.068888226229406e-06,
+      "loss": 0.751,
+      "step": 15300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.067742045267543e-06,
+      "loss": 0.7314,
+      "step": 15301
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.066595876660285e-06,
+      "loss": 0.7266,
+      "step": 15302
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.065449720422816e-06,
+      "loss": 0.7607,
+      "step": 15303
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.064303576570337e-06,
+      "loss": 0.708,
+      "step": 15304
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.06315744511802e-06,
+      "loss": 0.7354,
+      "step": 15305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.062011326081063e-06,
+      "loss": 0.708,
+      "step": 15306
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.060865219474654e-06,
+      "loss": 0.7744,
+      "step": 15307
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.059719125313982e-06,
+      "loss": 0.7373,
+      "step": 15308
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.058573043614232e-06,
+      "loss": 0.7109,
+      "step": 15309
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.057426974390599e-06,
+      "loss": 0.7217,
+      "step": 15310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.056280917658268e-06,
+      "loss": 0.7275,
+      "step": 15311
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.055134873432425e-06,
+      "loss": 0.7334,
+      "step": 15312
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.053988841728257e-06,
+      "loss": 0.7451,
+      "step": 15313
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.052842822560955e-06,
+      "loss": 0.7168,
+      "step": 15314
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.051696815945706e-06,
+      "loss": 0.709,
+      "step": 15315
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.050550821897693e-06,
+      "loss": 0.7178,
+      "step": 15316
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.04940484043211e-06,
+      "loss": 0.7275,
+      "step": 15317
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.048258871564139e-06,
+      "loss": 0.7305,
+      "step": 15318
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.047112915308966e-06,
+      "loss": 0.7021,
+      "step": 15319
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.04596697168178e-06,
+      "loss": 0.7148,
+      "step": 15320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.044821040697769e-06,
+      "loss": 0.7139,
+      "step": 15321
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.043675122372115e-06,
+      "loss": 0.7324,
+      "step": 15322
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.042529216720008e-06,
+      "loss": 0.7207,
+      "step": 15323
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.041383323756636e-06,
+      "loss": 0.7422,
+      "step": 15324
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.040237443497176e-06,
+      "loss": 0.7383,
+      "step": 15325
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.03909157595682e-06,
+      "loss": 0.7559,
+      "step": 15326
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.037945721150755e-06,
+      "loss": 0.7285,
+      "step": 15327
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.036799879094158e-06,
+      "loss": 0.7305,
+      "step": 15328
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.035654049802223e-06,
+      "loss": 0.7061,
+      "step": 15329
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.034508233290133e-06,
+      "loss": 0.7324,
+      "step": 15330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.033362429573071e-06,
+      "loss": 0.7236,
+      "step": 15331
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.032216638666219e-06,
+      "loss": 0.7227,
+      "step": 15332
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.031070860584765e-06,
+      "loss": 0.7256,
+      "step": 15333
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.02992509534389e-06,
+      "loss": 0.7324,
+      "step": 15334
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.028779342958785e-06,
+      "loss": 0.7295,
+      "step": 15335
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.027633603444625e-06,
+      "loss": 0.7412,
+      "step": 15336
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.026487876816605e-06,
+      "loss": 0.7139,
+      "step": 15337
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.025342163089897e-06,
+      "loss": 0.7158,
+      "step": 15338
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.02419646227969e-06,
+      "loss": 0.7217,
+      "step": 15339
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.023050774401165e-06,
+      "loss": 0.7227,
+      "step": 15340
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.021905099469508e-06,
+      "loss": 0.7373,
+      "step": 15341
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.020759437499898e-06,
+      "loss": 0.707,
+      "step": 15342
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.019613788507526e-06,
+      "loss": 0.7266,
+      "step": 15343
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.018468152507563e-06,
+      "loss": 0.7383,
+      "step": 15344
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.017322529515198e-06,
+      "loss": 0.7412,
+      "step": 15345
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.016176919545611e-06,
+      "loss": 0.7227,
+      "step": 15346
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.015031322613988e-06,
+      "loss": 0.7314,
+      "step": 15347
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.013885738735503e-06,
+      "loss": 0.7314,
+      "step": 15348
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.012740167925346e-06,
+      "loss": 0.7451,
+      "step": 15349
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.011594610198699e-06,
+      "loss": 0.7354,
+      "step": 15350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.010449065570735e-06,
+      "loss": 0.708,
+      "step": 15351
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.009303534056638e-06,
+      "loss": 0.7236,
+      "step": 15352
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.008158015671593e-06,
+      "loss": 0.751,
+      "step": 15353
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00701251043078e-06,
+      "loss": 0.7207,
+      "step": 15354
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.005867018349372e-06,
+      "loss": 0.7207,
+      "step": 15355
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.004721539442563e-06,
+      "loss": 0.7529,
+      "step": 15356
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.003576073725521e-06,
+      "loss": 0.7393,
+      "step": 15357
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.002430621213432e-06,
+      "loss": 0.7236,
+      "step": 15358
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.001285181921474e-06,
+      "loss": 0.7139,
+      "step": 15359
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.000139755864828e-06,
+      "loss": 0.7314,
+      "step": 15360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.998994343058672e-06,
+      "loss": 0.7109,
+      "step": 15361
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.997848943518187e-06,
+      "loss": 0.7061,
+      "step": 15362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.996703557258556e-06,
+      "loss": 0.7422,
+      "step": 15363
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99555818429495e-06,
+      "loss": 0.7236,
+      "step": 15364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99441282464255e-06,
+      "loss": 0.7266,
+      "step": 15365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99326747831654e-06,
+      "loss": 0.7285,
+      "step": 15366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99212214533209e-06,
+      "loss": 0.7461,
+      "step": 15367
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.990976825704387e-06,
+      "loss": 0.749,
+      "step": 15368
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.989831519448608e-06,
+      "loss": 0.7246,
+      "step": 15369
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.988686226579922e-06,
+      "loss": 0.7119,
+      "step": 15370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.987540947113516e-06,
+      "loss": 0.7461,
+      "step": 15371
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.986395681064567e-06,
+      "loss": 0.7227,
+      "step": 15372
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.985250428448245e-06,
+      "loss": 0.7129,
+      "step": 15373
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.984105189279737e-06,
+      "loss": 0.7266,
+      "step": 15374
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.982959963574213e-06,
+      "loss": 0.7666,
+      "step": 15375
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.981814751346858e-06,
+      "loss": 0.7275,
+      "step": 15376
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.980669552612838e-06,
+      "loss": 0.7236,
+      "step": 15377
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.979524367387337e-06,
+      "loss": 0.7324,
+      "step": 15378
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.978379195685527e-06,
+      "loss": 0.7139,
+      "step": 15379
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.977234037522588e-06,
+      "loss": 0.7305,
+      "step": 15380
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.976088892913692e-06,
+      "loss": 0.7305,
+      "step": 15381
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.974943761874025e-06,
+      "loss": 0.7129,
+      "step": 15382
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.973798644418748e-06,
+      "loss": 0.749,
+      "step": 15383
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.972653540563047e-06,
+      "loss": 0.75,
+      "step": 15384
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.97150845032209e-06,
+      "loss": 0.7539,
+      "step": 15385
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.970363373711058e-06,
+      "loss": 0.7295,
+      "step": 15386
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.969218310745124e-06,
+      "loss": 0.7344,
+      "step": 15387
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.968073261439463e-06,
+      "loss": 0.7373,
+      "step": 15388
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.966928225809252e-06,
+      "loss": 0.7373,
+      "step": 15389
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.96578320386966e-06,
+      "loss": 0.7305,
+      "step": 15390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.964638195635862e-06,
+      "loss": 0.7197,
+      "step": 15391
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.963493201123036e-06,
+      "loss": 0.75,
+      "step": 15392
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.962348220346352e-06,
+      "loss": 0.7129,
+      "step": 15393
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.961203253320986e-06,
+      "loss": 0.7275,
+      "step": 15394
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.960058300062115e-06,
+      "loss": 0.7393,
+      "step": 15395
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.958913360584906e-06,
+      "loss": 0.7129,
+      "step": 15396
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.957768434904532e-06,
+      "loss": 0.7158,
+      "step": 15397
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.956623523036172e-06,
+      "loss": 0.7275,
+      "step": 15398
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.955478624994992e-06,
+      "loss": 0.7012,
+      "step": 15399
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.954333740796168e-06,
+      "loss": 0.7256,
+      "step": 15400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.953188870454872e-06,
+      "loss": 0.7246,
+      "step": 15401
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.952044013986281e-06,
+      "loss": 0.7393,
+      "step": 15402
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.950899171405557e-06,
+      "loss": 0.7305,
+      "step": 15403
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.949754342727878e-06,
+      "loss": 0.7559,
+      "step": 15404
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.948609527968417e-06,
+      "loss": 0.7344,
+      "step": 15405
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94746472714234e-06,
+      "loss": 0.7559,
+      "step": 15406
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.946319940264823e-06,
+      "loss": 0.7217,
+      "step": 15407
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.94517516735104e-06,
+      "loss": 0.7285,
+      "step": 15408
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.944030408416153e-06,
+      "loss": 0.7275,
+      "step": 15409
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.942885663475339e-06,
+      "loss": 0.6982,
+      "step": 15410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.941740932543766e-06,
+      "loss": 0.7021,
+      "step": 15411
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.940596215636604e-06,
+      "loss": 0.7383,
+      "step": 15412
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.939451512769027e-06,
+      "loss": 0.7383,
+      "step": 15413
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9383068239562e-06,
+      "loss": 0.7676,
+      "step": 15414
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9371621492133e-06,
+      "loss": 0.6914,
+      "step": 15415
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.936017488555489e-06,
+      "loss": 0.7656,
+      "step": 15416
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.934872841997941e-06,
+      "loss": 0.7236,
+      "step": 15417
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.93372820955582e-06,
+      "loss": 0.7607,
+      "step": 15418
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.932583591244301e-06,
+      "loss": 0.7275,
+      "step": 15419
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.931438987078549e-06,
+      "loss": 0.7334,
+      "step": 15420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.93029439707374e-06,
+      "loss": 0.7207,
+      "step": 15421
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92914982124503e-06,
+      "loss": 0.7178,
+      "step": 15422
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.928005259607598e-06,
+      "loss": 0.7324,
+      "step": 15423
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.926860712176604e-06,
+      "loss": 0.7129,
+      "step": 15424
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.925716178967224e-06,
+      "loss": 0.7178,
+      "step": 15425
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.924571659994618e-06,
+      "loss": 0.7061,
+      "step": 15426
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92342715527396e-06,
+      "loss": 0.7139,
+      "step": 15427
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.922282664820417e-06,
+      "loss": 0.7383,
+      "step": 15428
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.921138188649153e-06,
+      "loss": 0.7393,
+      "step": 15429
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.919993726775333e-06,
+      "loss": 0.7344,
+      "step": 15430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.918849279214129e-06,
+      "loss": 0.7178,
+      "step": 15431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.917704845980702e-06,
+      "loss": 0.7324,
+      "step": 15432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.916560427090227e-06,
+      "loss": 0.7246,
+      "step": 15433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.915416022557864e-06,
+      "loss": 0.7207,
+      "step": 15434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.914271632398779e-06,
+      "loss": 0.752,
+      "step": 15435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.913127256628137e-06,
+      "loss": 0.7295,
+      "step": 15436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.911982895261107e-06,
+      "loss": 0.7256,
+      "step": 15437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.910838548312855e-06,
+      "loss": 0.7236,
+      "step": 15438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90969421579854e-06,
+      "loss": 0.7168,
+      "step": 15439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.908549897733334e-06,
+      "loss": 0.7178,
+      "step": 15440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.907405594132403e-06,
+      "loss": 0.708,
+      "step": 15441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.906261305010902e-06,
+      "loss": 0.7227,
+      "step": 15442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.905117030384006e-06,
+      "loss": 0.7354,
+      "step": 15443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.903972770266873e-06,
+      "loss": 0.7432,
+      "step": 15444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.902828524674666e-06,
+      "loss": 0.6865,
+      "step": 15445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.901684293622554e-06,
+      "loss": 0.7158,
+      "step": 15446
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.900540077125702e-06,
+      "loss": 0.7178,
+      "step": 15447
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.899395875199266e-06,
+      "loss": 0.7422,
+      "step": 15448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.898251687858416e-06,
+      "loss": 0.7178,
+      "step": 15449
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.897107515118312e-06,
+      "loss": 0.7529,
+      "step": 15450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.895963356994116e-06,
+      "loss": 0.7197,
+      "step": 15451
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.894819213500994e-06,
+      "loss": 0.7305,
+      "step": 15452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.893675084654105e-06,
+      "loss": 0.7344,
+      "step": 15453
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.892530970468618e-06,
+      "loss": 0.7734,
+      "step": 15454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.891386870959688e-06,
+      "loss": 0.7402,
+      "step": 15455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.890242786142481e-06,
+      "loss": 0.7256,
+      "step": 15456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.889098716032154e-06,
+      "loss": 0.7402,
+      "step": 15457
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.887954660643875e-06,
+      "loss": 0.7441,
+      "step": 15458
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.8868106199928e-06,
+      "loss": 0.7305,
+      "step": 15459
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.885666594094099e-06,
+      "loss": 0.7275,
+      "step": 15460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.88452258296292e-06,
+      "loss": 0.7207,
+      "step": 15461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.883378586614433e-06,
+      "loss": 0.7803,
+      "step": 15462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.882234605063794e-06,
+      "loss": 0.7637,
+      "step": 15463
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.881090638326167e-06,
+      "loss": 0.7314,
+      "step": 15464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87994668641671e-06,
+      "loss": 0.7188,
+      "step": 15465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.878802749350586e-06,
+      "loss": 0.7422,
+      "step": 15466
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.877658827142955e-06,
+      "loss": 0.7539,
+      "step": 15467
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87651491980897e-06,
+      "loss": 0.7383,
+      "step": 15468
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.875371027363795e-06,
+      "loss": 0.7256,
+      "step": 15469
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87422714982259e-06,
+      "loss": 0.71,
+      "step": 15470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.873083287200512e-06,
+      "loss": 0.7285,
+      "step": 15471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.871939439512722e-06,
+      "loss": 0.7119,
+      "step": 15472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87079560677438e-06,
+      "loss": 0.7295,
+      "step": 15473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.869651789000642e-06,
+      "loss": 0.7373,
+      "step": 15474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.868507986206663e-06,
+      "loss": 0.748,
+      "step": 15475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.867364198407606e-06,
+      "loss": 0.7383,
+      "step": 15476
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.866220425618624e-06,
+      "loss": 0.7314,
+      "step": 15477
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.865076667854883e-06,
+      "loss": 0.7266,
+      "step": 15478
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.863932925131534e-06,
+      "loss": 0.7061,
+      "step": 15479
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.862789197463738e-06,
+      "loss": 0.748,
+      "step": 15480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.861645484866646e-06,
+      "loss": 0.7314,
+      "step": 15481
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86050178735542e-06,
+      "loss": 0.75,
+      "step": 15482
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.859358104945216e-06,
+      "loss": 0.7314,
+      "step": 15483
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.858214437651186e-06,
+      "loss": 0.7158,
+      "step": 15484
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.857070785488495e-06,
+      "loss": 0.7295,
+      "step": 15485
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.855927148472294e-06,
+      "loss": 0.7246,
+      "step": 15486
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.854783526617736e-06,
+      "loss": 0.7197,
+      "step": 15487
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.853639919939983e-06,
+      "loss": 0.7412,
+      "step": 15488
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.852496328454186e-06,
+      "loss": 0.7061,
+      "step": 15489
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.851352752175498e-06,
+      "loss": 0.71,
+      "step": 15490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.850209191119081e-06,
+      "loss": 0.7109,
+      "step": 15491
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.849065645300088e-06,
+      "loss": 0.7178,
+      "step": 15492
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.847922114733668e-06,
+      "loss": 0.71,
+      "step": 15493
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.846778599434982e-06,
+      "loss": 0.7295,
+      "step": 15494
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.84563509941918e-06,
+      "loss": 0.7432,
+      "step": 15495
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.844491614701419e-06,
+      "loss": 0.7148,
+      "step": 15496
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.84334814529685e-06,
+      "loss": 0.7432,
+      "step": 15497
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.84220469122063e-06,
+      "loss": 0.7363,
+      "step": 15498
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.841061252487912e-06,
+      "loss": 0.7295,
+      "step": 15499
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.839917829113848e-06,
+      "loss": 0.7207,
+      "step": 15500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.83877442111359e-06,
+      "loss": 0.7236,
+      "step": 15501
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.837631028502289e-06,
+      "loss": 0.7432,
+      "step": 15502
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.836487651295104e-06,
+      "loss": 0.708,
+      "step": 15503
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.835344289507181e-06,
+      "loss": 0.7314,
+      "step": 15504
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.834200943153681e-06,
+      "loss": 0.748,
+      "step": 15505
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.833057612249745e-06,
+      "loss": 0.7314,
+      "step": 15506
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.831914296810532e-06,
+      "loss": 0.7275,
+      "step": 15507
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.83077099685119e-06,
+      "loss": 0.7461,
+      "step": 15508
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.829627712386872e-06,
+      "loss": 0.7285,
+      "step": 15509
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.828484443432727e-06,
+      "loss": 0.7363,
+      "step": 15510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.827341190003913e-06,
+      "loss": 0.7246,
+      "step": 15511
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.826197952115576e-06,
+      "loss": 0.7393,
+      "step": 15512
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.825054729782866e-06,
+      "loss": 0.7334,
+      "step": 15513
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.823911523020931e-06,
+      "loss": 0.7227,
+      "step": 15514
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.822768331844926e-06,
+      "loss": 0.7246,
+      "step": 15515
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.821625156269997e-06,
+      "loss": 0.7246,
+      "step": 15516
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.820481996311297e-06,
+      "loss": 0.7383,
+      "step": 15517
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.819338851983978e-06,
+      "loss": 0.7441,
+      "step": 15518
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.818195723303183e-06,
+      "loss": 0.7168,
+      "step": 15519
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.817052610284063e-06,
+      "loss": 0.7402,
+      "step": 15520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.815909512941768e-06,
+      "loss": 0.7168,
+      "step": 15521
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.814766431291444e-06,
+      "loss": 0.7314,
+      "step": 15522
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.813623365348245e-06,
+      "loss": 0.752,
+      "step": 15523
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.812480315127318e-06,
+      "loss": 0.7227,
+      "step": 15524
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81133728064381e-06,
+      "loss": 0.7471,
+      "step": 15525
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.810194261912865e-06,
+      "loss": 0.7061,
+      "step": 15526
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.809051258949635e-06,
+      "loss": 0.6924,
+      "step": 15527
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.807908271769265e-06,
+      "loss": 0.7383,
+      "step": 15528
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.806765300386903e-06,
+      "loss": 0.7471,
+      "step": 15529
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.8056223448177e-06,
+      "loss": 0.7119,
+      "step": 15530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.8044794050768e-06,
+      "loss": 0.7236,
+      "step": 15531
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.803336481179347e-06,
+      "loss": 0.7354,
+      "step": 15532
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.80219357314049e-06,
+      "loss": 0.752,
+      "step": 15533
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.801050680975375e-06,
+      "loss": 0.7354,
+      "step": 15534
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.799907804699146e-06,
+      "loss": 0.7246,
+      "step": 15535
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.798764944326952e-06,
+      "loss": 0.6992,
+      "step": 15536
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.797622099873936e-06,
+      "loss": 0.7334,
+      "step": 15537
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796479271355248e-06,
+      "loss": 0.7334,
+      "step": 15538
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.795336458786028e-06,
+      "loss": 0.71,
+      "step": 15539
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.794193662181422e-06,
+      "loss": 0.7236,
+      "step": 15540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.793050881556576e-06,
+      "loss": 0.7031,
+      "step": 15541
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.791908116926634e-06,
+      "loss": 0.7295,
+      "step": 15542
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.790765368306737e-06,
+      "loss": 0.7295,
+      "step": 15543
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78962263571204e-06,
+      "loss": 0.71,
+      "step": 15544
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.788479919157674e-06,
+      "loss": 0.7021,
+      "step": 15545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.787337218658786e-06,
+      "loss": 0.7061,
+      "step": 15546
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.786194534230523e-06,
+      "loss": 0.749,
+      "step": 15547
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.785051865888029e-06,
+      "loss": 0.7432,
+      "step": 15548
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78390921364644e-06,
+      "loss": 0.7363,
+      "step": 15549
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.782766577520906e-06,
+      "loss": 0.7461,
+      "step": 15550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.781623957526571e-06,
+      "loss": 0.7021,
+      "step": 15551
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78048135367857e-06,
+      "loss": 0.7139,
+      "step": 15552
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.779338765992047e-06,
+      "loss": 0.7393,
+      "step": 15553
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.778196194482145e-06,
+      "loss": 0.71,
+      "step": 15554
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.777053639164007e-06,
+      "loss": 0.7549,
+      "step": 15555
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.775911100052775e-06,
+      "loss": 0.7451,
+      "step": 15556
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.774768577163593e-06,
+      "loss": 0.7422,
+      "step": 15557
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.773626070511594e-06,
+      "loss": 0.7324,
+      "step": 15558
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.772483580111923e-06,
+      "loss": 0.7383,
+      "step": 15559
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.771341105979721e-06,
+      "loss": 0.7441,
+      "step": 15560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.770198648130127e-06,
+      "loss": 0.7471,
+      "step": 15561
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.769056206578285e-06,
+      "loss": 0.7324,
+      "step": 15562
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.767913781339332e-06,
+      "loss": 0.7266,
+      "step": 15563
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.766771372428411e-06,
+      "loss": 0.749,
+      "step": 15564
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.765628979860654e-06,
+      "loss": 0.7471,
+      "step": 15565
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.764486603651209e-06,
+      "loss": 0.7305,
+      "step": 15566
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.76334424381521e-06,
+      "loss": 0.6895,
+      "step": 15567
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.762201900367796e-06,
+      "loss": 0.7246,
+      "step": 15568
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.76105957332411e-06,
+      "loss": 0.7412,
+      "step": 15569
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.759917262699288e-06,
+      "loss": 0.709,
+      "step": 15570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.758774968508465e-06,
+      "loss": 0.7227,
+      "step": 15571
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.757632690766785e-06,
+      "loss": 0.7441,
+      "step": 15572
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756490429489382e-06,
+      "loss": 0.7344,
+      "step": 15573
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.755348184691392e-06,
+      "loss": 0.7373,
+      "step": 15574
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.754205956387957e-06,
+      "loss": 0.7432,
+      "step": 15575
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.75306374459421e-06,
+      "loss": 0.7168,
+      "step": 15576
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.751921549325296e-06,
+      "loss": 0.7275,
+      "step": 15577
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.750779370596342e-06,
+      "loss": 0.7207,
+      "step": 15578
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74963720842249e-06,
+      "loss": 0.7275,
+      "step": 15579
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.748495062818873e-06,
+      "loss": 0.7607,
+      "step": 15580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74735293380063e-06,
+      "loss": 0.71,
+      "step": 15581
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.746210821382892e-06,
+      "loss": 0.7539,
+      "step": 15582
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.745068725580806e-06,
+      "loss": 0.7305,
+      "step": 15583
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.743926646409494e-06,
+      "loss": 0.7324,
+      "step": 15584
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.7427845838841e-06,
+      "loss": 0.7207,
+      "step": 15585
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.741642538019754e-06,
+      "loss": 0.71,
+      "step": 15586
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.740500508831594e-06,
+      "loss": 0.7168,
+      "step": 15587
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.739358496334752e-06,
+      "loss": 0.7305,
+      "step": 15588
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.738216500544365e-06,
+      "loss": 0.7646,
+      "step": 15589
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73707452147557e-06,
+      "loss": 0.7256,
+      "step": 15590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.735932559143494e-06,
+      "loss": 0.7148,
+      "step": 15591
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73479061356327e-06,
+      "loss": 0.71,
+      "step": 15592
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.733648684750037e-06,
+      "loss": 0.7051,
+      "step": 15593
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.732506772718926e-06,
+      "loss": 0.7021,
+      "step": 15594
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.731364877485072e-06,
+      "loss": 0.7256,
+      "step": 15595
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.730222999063606e-06,
+      "loss": 0.7402,
+      "step": 15596
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.729081137469662e-06,
+      "loss": 0.7148,
+      "step": 15597
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.727939292718366e-06,
+      "loss": 0.7334,
+      "step": 15598
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.726797464824857e-06,
+      "loss": 0.7256,
+      "step": 15599
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.725655653804265e-06,
+      "loss": 0.7041,
+      "step": 15600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.724513859671722e-06,
+      "loss": 0.7441,
+      "step": 15601
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.723372082442359e-06,
+      "loss": 0.7637,
+      "step": 15602
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.72223032213131e-06,
+      "loss": 0.7197,
+      "step": 15603
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.721088578753698e-06,
+      "loss": 0.71,
+      "step": 15604
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.719946852324664e-06,
+      "loss": 0.7021,
+      "step": 15605
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.718805142859329e-06,
+      "loss": 0.7256,
+      "step": 15606
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71766345037283e-06,
+      "loss": 0.7324,
+      "step": 15607
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.716521774880297e-06,
+      "loss": 0.7373,
+      "step": 15608
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.715380116396859e-06,
+      "loss": 0.7178,
+      "step": 15609
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71423847493764e-06,
+      "loss": 0.7158,
+      "step": 15610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.713096850517775e-06,
+      "loss": 0.7383,
+      "step": 15611
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.711955243152394e-06,
+      "loss": 0.7373,
+      "step": 15612
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71081365285662e-06,
+      "loss": 0.708,
+      "step": 15613
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.70967207964559e-06,
+      "loss": 0.7197,
+      "step": 15614
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.708530523534426e-06,
+      "loss": 0.7324,
+      "step": 15615
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.707388984538261e-06,
+      "loss": 0.7158,
+      "step": 15616
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.706247462672222e-06,
+      "loss": 0.708,
+      "step": 15617
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.705105957951433e-06,
+      "loss": 0.7393,
+      "step": 15618
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.703964470391023e-06,
+      "loss": 0.7598,
+      "step": 15619
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.702823000006123e-06,
+      "loss": 0.7295,
+      "step": 15620
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.701681546811855e-06,
+      "loss": 0.7393,
+      "step": 15621
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.700540110823352e-06,
+      "loss": 0.7148,
+      "step": 15622
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.699398692055736e-06,
+      "loss": 0.6865,
+      "step": 15623
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.698257290524136e-06,
+      "loss": 0.75,
+      "step": 15624
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.697115906243674e-06,
+      "loss": 0.7168,
+      "step": 15625
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.695974539229482e-06,
+      "loss": 0.7285,
+      "step": 15626
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.69483318949668e-06,
+      "loss": 0.7227,
+      "step": 15627
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.6936918570604e-06,
+      "loss": 0.7393,
+      "step": 15628
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.692550541935764e-06,
+      "loss": 0.7363,
+      "step": 15629
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.691409244137895e-06,
+      "loss": 0.6963,
+      "step": 15630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.690267963681919e-06,
+      "loss": 0.7363,
+      "step": 15631
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.689126700582962e-06,
+      "loss": 0.7451,
+      "step": 15632
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.687985454856147e-06,
+      "loss": 0.7188,
+      "step": 15633
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.6868442265166e-06,
+      "loss": 0.7354,
+      "step": 15634
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.685703015579447e-06,
+      "loss": 0.7314,
+      "step": 15635
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.684561822059805e-06,
+      "loss": 0.7314,
+      "step": 15636
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.683420645972803e-06,
+      "loss": 0.7305,
+      "step": 15637
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.68227948733356e-06,
+      "loss": 0.7041,
+      "step": 15638
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.681138346157201e-06,
+      "loss": 0.7148,
+      "step": 15639
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.679997222458852e-06,
+      "loss": 0.749,
+      "step": 15640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.678856116253633e-06,
+      "loss": 0.7373,
+      "step": 15641
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.677715027556666e-06,
+      "loss": 0.7295,
+      "step": 15642
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.676573956383072e-06,
+      "loss": 0.7373,
+      "step": 15643
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.675432902747973e-06,
+      "loss": 0.7217,
+      "step": 15644
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.674291866666491e-06,
+      "loss": 0.7148,
+      "step": 15645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.673150848153752e-06,
+      "loss": 0.7412,
+      "step": 15646
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.672009847224872e-06,
+      "loss": 0.7432,
+      "step": 15647
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.670868863894974e-06,
+      "loss": 0.75,
+      "step": 15648
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.669727898179175e-06,
+      "loss": 0.7354,
+      "step": 15649
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.668586950092602e-06,
+      "loss": 0.7314,
+      "step": 15650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.66744601965037e-06,
+      "loss": 0.7305,
+      "step": 15651
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.666305106867598e-06,
+      "loss": 0.7637,
+      "step": 15652
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.665164211759412e-06,
+      "loss": 0.7197,
+      "step": 15653
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.664023334340925e-06,
+      "loss": 0.7344,
+      "step": 15654
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.662882474627264e-06,
+      "loss": 0.7314,
+      "step": 15655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.661741632633539e-06,
+      "loss": 0.7285,
+      "step": 15656
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.660600808374875e-06,
+      "loss": 0.7256,
+      "step": 15657
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.659460001866386e-06,
+      "loss": 0.7178,
+      "step": 15658
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.658319213123197e-06,
+      "loss": 0.7207,
+      "step": 15659
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.657178442160418e-06,
+      "loss": 0.7568,
+      "step": 15660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.656037688993177e-06,
+      "loss": 0.7178,
+      "step": 15661
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.654896953636582e-06,
+      "loss": 0.7148,
+      "step": 15662
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.653756236105755e-06,
+      "loss": 0.7139,
+      "step": 15663
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65261553641581e-06,
+      "loss": 0.749,
+      "step": 15664
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.651474854581868e-06,
+      "loss": 0.7256,
+      "step": 15665
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.650334190619043e-06,
+      "loss": 0.7305,
+      "step": 15666
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.649193544542453e-06,
+      "loss": 0.7275,
+      "step": 15667
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.648052916367219e-06,
+      "loss": 0.7109,
+      "step": 15668
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.646912306108445e-06,
+      "loss": 0.7363,
+      "step": 15669
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.645771713781255e-06,
+      "loss": 0.7266,
+      "step": 15670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.644631139400764e-06,
+      "loss": 0.7178,
+      "step": 15671
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.643490582982083e-06,
+      "loss": 0.7217,
+      "step": 15672
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.642350044540333e-06,
+      "loss": 0.7393,
+      "step": 15673
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.64120952409063e-06,
+      "loss": 0.7324,
+      "step": 15674
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.64006902164808e-06,
+      "loss": 0.7148,
+      "step": 15675
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.638928537227802e-06,
+      "loss": 0.7158,
+      "step": 15676
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.637788070844912e-06,
+      "loss": 0.7275,
+      "step": 15677
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63664762251452e-06,
+      "loss": 0.7568,
+      "step": 15678
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.635507192251742e-06,
+      "loss": 0.7412,
+      "step": 15679
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.634366780071693e-06,
+      "loss": 0.7236,
+      "step": 15680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.633226385989484e-06,
+      "loss": 0.7188,
+      "step": 15681
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.632086010020225e-06,
+      "loss": 0.7451,
+      "step": 15682
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.630945652179034e-06,
+      "loss": 0.7188,
+      "step": 15683
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.629805312481019e-06,
+      "loss": 0.7295,
+      "step": 15684
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.628664990941296e-06,
+      "loss": 0.7295,
+      "step": 15685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.627524687574974e-06,
+      "loss": 0.7168,
+      "step": 15686
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.62638440239717e-06,
+      "loss": 0.7197,
+      "step": 15687
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.625244135422986e-06,
+      "loss": 0.7412,
+      "step": 15688
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624103886667539e-06,
+      "loss": 0.707,
+      "step": 15689
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.62296365614594e-06,
+      "loss": 0.707,
+      "step": 15690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.6218234438733e-06,
+      "loss": 0.7207,
+      "step": 15691
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.620683249864728e-06,
+      "loss": 0.7041,
+      "step": 15692
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.619543074135334e-06,
+      "loss": 0.7188,
+      "step": 15693
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.618402916700234e-06,
+      "loss": 0.7285,
+      "step": 15694
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.617262777574529e-06,
+      "loss": 0.7168,
+      "step": 15695
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.616122656773332e-06,
+      "loss": 0.7363,
+      "step": 15696
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.614982554311751e-06,
+      "loss": 0.7441,
+      "step": 15697
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.6138424702049e-06,
+      "loss": 0.7305,
+      "step": 15698
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.61270240446788e-06,
+      "loss": 0.7334,
+      "step": 15699
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.61156235711581e-06,
+      "loss": 0.7217,
+      "step": 15700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.610422328163788e-06,
+      "loss": 0.7197,
+      "step": 15701
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.609282317626926e-06,
+      "loss": 0.7402,
+      "step": 15702
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.608142325520331e-06,
+      "loss": 0.7031,
+      "step": 15703
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.607002351859114e-06,
+      "loss": 0.7402,
+      "step": 15704
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.605862396658378e-06,
+      "loss": 0.7256,
+      "step": 15705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.604722459933235e-06,
+      "loss": 0.7129,
+      "step": 15706
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.603582541698787e-06,
+      "loss": 0.7109,
+      "step": 15707
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.602442641970142e-06,
+      "loss": 0.7344,
+      "step": 15708
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.601302760762405e-06,
+      "loss": 0.6973,
+      "step": 15709
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.600162898090688e-06,
+      "loss": 0.7207,
+      "step": 15710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.599023053970086e-06,
+      "loss": 0.7422,
+      "step": 15711
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.597883228415717e-06,
+      "loss": 0.7061,
+      "step": 15712
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59674342144268e-06,
+      "loss": 0.7227,
+      "step": 15713
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.595603633066082e-06,
+      "loss": 0.7207,
+      "step": 15714
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.594463863301024e-06,
+      "loss": 0.752,
+      "step": 15715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.593324112162614e-06,
+      "loss": 0.7646,
+      "step": 15716
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.592184379665952e-06,
+      "loss": 0.7031,
+      "step": 15717
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.59104466582615e-06,
+      "loss": 0.7178,
+      "step": 15718
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.589904970658309e-06,
+      "loss": 0.7236,
+      "step": 15719
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58876529417753e-06,
+      "loss": 0.7354,
+      "step": 15720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.587625636398915e-06,
+      "loss": 0.7393,
+      "step": 15721
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.586485997337572e-06,
+      "loss": 0.7275,
+      "step": 15722
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.585346377008598e-06,
+      "loss": 0.7354,
+      "step": 15723
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.584206775427104e-06,
+      "loss": 0.7256,
+      "step": 15724
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.583067192608186e-06,
+      "loss": 0.7295,
+      "step": 15725
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58192762856695e-06,
+      "loss": 0.7178,
+      "step": 15726
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.580788083318492e-06,
+      "loss": 0.7607,
+      "step": 15727
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.57964855687792e-06,
+      "loss": 0.7188,
+      "step": 15728
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.578509049260328e-06,
+      "loss": 0.7344,
+      "step": 15729
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.577369560480828e-06,
+      "loss": 0.7129,
+      "step": 15730
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.576230090554512e-06,
+      "loss": 0.7305,
+      "step": 15731
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.575090639496487e-06,
+      "loss": 0.7324,
+      "step": 15732
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.573951207321843e-06,
+      "loss": 0.7129,
+      "step": 15733
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.572811794045692e-06,
+      "loss": 0.7217,
+      "step": 15734
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.571672399683126e-06,
+      "loss": 0.7324,
+      "step": 15735
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.570533024249248e-06,
+      "loss": 0.7344,
+      "step": 15736
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.569393667759158e-06,
+      "loss": 0.7119,
+      "step": 15737
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.568254330227953e-06,
+      "loss": 0.7129,
+      "step": 15738
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.567115011670737e-06,
+      "loss": 0.7148,
+      "step": 15739
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.565975712102602e-06,
+      "loss": 0.7256,
+      "step": 15740
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.564836431538648e-06,
+      "loss": 0.7002,
+      "step": 15741
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.563697169993974e-06,
+      "loss": 0.7451,
+      "step": 15742
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.562557927483678e-06,
+      "loss": 0.7188,
+      "step": 15743
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.561418704022857e-06,
+      "loss": 0.7275,
+      "step": 15744
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.560279499626612e-06,
+      "loss": 0.7197,
+      "step": 15745
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.559140314310035e-06,
+      "loss": 0.7266,
+      "step": 15746
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.558001148088227e-06,
+      "loss": 0.7168,
+      "step": 15747
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.556862000976279e-06,
+      "loss": 0.71,
+      "step": 15748
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.555722872989293e-06,
+      "loss": 0.7451,
+      "step": 15749
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.554583764142362e-06,
+      "loss": 0.7354,
+      "step": 15750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.553444674450584e-06,
+      "loss": 0.7246,
+      "step": 15751
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.552305603929057e-06,
+      "loss": 0.7275,
+      "step": 15752
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55116655259287e-06,
+      "loss": 0.7178,
+      "step": 15753
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55002752045712e-06,
+      "loss": 0.7041,
+      "step": 15754
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.548888507536903e-06,
+      "loss": 0.6963,
+      "step": 15755
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.547749513847311e-06,
+      "loss": 0.7275,
+      "step": 15756
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.546610539403447e-06,
+      "loss": 0.7344,
+      "step": 15757
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.545471584220398e-06,
+      "loss": 0.7266,
+      "step": 15758
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.544332648313256e-06,
+      "loss": 0.7021,
+      "step": 15759
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.543193731697117e-06,
+      "loss": 0.7148,
+      "step": 15760
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.542054834387074e-06,
+      "loss": 0.7129,
+      "step": 15761
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.540915956398222e-06,
+      "loss": 0.7451,
+      "step": 15762
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.539777097745652e-06,
+      "loss": 0.7461,
+      "step": 15763
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.538638258444459e-06,
+      "loss": 0.7227,
+      "step": 15764
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.537499438509734e-06,
+      "loss": 0.7041,
+      "step": 15765
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.536360637956565e-06,
+      "loss": 0.7344,
+      "step": 15766
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.535221856800049e-06,
+      "loss": 0.7188,
+      "step": 15767
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.534083095055273e-06,
+      "loss": 0.7012,
+      "step": 15768
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.532944352737333e-06,
+      "loss": 0.7061,
+      "step": 15769
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.531805629861318e-06,
+      "loss": 0.7598,
+      "step": 15770
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.530666926442321e-06,
+      "loss": 0.7295,
+      "step": 15771
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.529528242495424e-06,
+      "loss": 0.7236,
+      "step": 15772
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.528389578035728e-06,
+      "loss": 0.7539,
+      "step": 15773
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.527250933078315e-06,
+      "loss": 0.7422,
+      "step": 15774
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.526112307638281e-06,
+      "loss": 0.7041,
+      "step": 15775
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.524973701730713e-06,
+      "loss": 0.7139,
+      "step": 15776
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.523835115370697e-06,
+      "loss": 0.7227,
+      "step": 15777
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.522696548573329e-06,
+      "loss": 0.7031,
+      "step": 15778
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.52155800135369e-06,
+      "loss": 0.7021,
+      "step": 15779
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.520419473726873e-06,
+      "loss": 0.7178,
+      "step": 15780
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.519280965707962e-06,
+      "loss": 0.7324,
+      "step": 15781
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.51814247731205e-06,
+      "loss": 0.7119,
+      "step": 15782
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.51700400855422e-06,
+      "loss": 0.7295,
+      "step": 15783
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.515865559449568e-06,
+      "loss": 0.7217,
+      "step": 15784
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.514727130013171e-06,
+      "loss": 0.75,
+      "step": 15785
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.51358872026012e-06,
+      "loss": 0.7256,
+      "step": 15786
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.5124503302055e-06,
+      "loss": 0.7158,
+      "step": 15787
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.5113119598644e-06,
+      "loss": 0.7285,
+      "step": 15788
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.510173609251901e-06,
+      "loss": 0.7334,
+      "step": 15789
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.509035278383097e-06,
+      "loss": 0.7363,
+      "step": 15790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.507896967273072e-06,
+      "loss": 0.7471,
+      "step": 15791
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.506758675936905e-06,
+      "loss": 0.7422,
+      "step": 15792
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.505620404389682e-06,
+      "loss": 0.7109,
+      "step": 15793
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.504482152646495e-06,
+      "loss": 0.7842,
+      "step": 15794
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.503343920722418e-06,
+      "loss": 0.7002,
+      "step": 15795
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.502205708632544e-06,
+      "loss": 0.7432,
+      "step": 15796
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.501067516391958e-06,
+      "loss": 0.7344,
+      "step": 15797
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.499929344015736e-06,
+      "loss": 0.7295,
+      "step": 15798
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.498791191518963e-06,
+      "loss": 0.7275,
+      "step": 15799
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.497653058916727e-06,
+      "loss": 0.7344,
+      "step": 15800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.496514946224104e-06,
+      "loss": 0.7129,
+      "step": 15801
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.495376853456185e-06,
+      "loss": 0.7188,
+      "step": 15802
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.494238780628049e-06,
+      "loss": 0.7139,
+      "step": 15803
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.493100727754779e-06,
+      "loss": 0.7148,
+      "step": 15804
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.49196269485145e-06,
+      "loss": 0.7256,
+      "step": 15805
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.490824681933152e-06,
+      "loss": 0.7295,
+      "step": 15806
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.48968668901496e-06,
+      "loss": 0.7402,
+      "step": 15807
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.488548716111962e-06,
+      "loss": 0.7178,
+      "step": 15808
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.487410763239234e-06,
+      "loss": 0.7451,
+      "step": 15809
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.48627283041186e-06,
+      "loss": 0.7031,
+      "step": 15810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.485134917644915e-06,
+      "loss": 0.7295,
+      "step": 15811
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.483997024953483e-06,
+      "loss": 0.7314,
+      "step": 15812
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.482859152352639e-06,
+      "loss": 0.7227,
+      "step": 15813
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.481721299857469e-06,
+      "loss": 0.7393,
+      "step": 15814
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.480583467483052e-06,
+      "loss": 0.7363,
+      "step": 15815
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.47944565524446e-06,
+      "loss": 0.7168,
+      "step": 15816
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.478307863156779e-06,
+      "loss": 0.7334,
+      "step": 15817
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.477170091235085e-06,
+      "loss": 0.7227,
+      "step": 15818
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.476032339494452e-06,
+      "loss": 0.7188,
+      "step": 15819
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.474894607949962e-06,
+      "loss": 0.7139,
+      "step": 15820
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.473756896616695e-06,
+      "loss": 0.7363,
+      "step": 15821
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.472619205509722e-06,
+      "loss": 0.7461,
+      "step": 15822
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.471481534644126e-06,
+      "loss": 0.7461,
+      "step": 15823
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.470343884034982e-06,
+      "loss": 0.708,
+      "step": 15824
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.469206253697364e-06,
+      "loss": 0.7109,
+      "step": 15825
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.468068643646349e-06,
+      "loss": 0.7119,
+      "step": 15826
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.466931053897016e-06,
+      "loss": 0.7607,
+      "step": 15827
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.465793484464437e-06,
+      "loss": 0.7178,
+      "step": 15828
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.46465593536369e-06,
+      "loss": 0.7344,
+      "step": 15829
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.463518406609852e-06,
+      "loss": 0.7285,
+      "step": 15830
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.462380898217994e-06,
+      "loss": 0.7305,
+      "step": 15831
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.46124341020319e-06,
+      "loss": 0.7393,
+      "step": 15832
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.460105942580516e-06,
+      "loss": 0.7373,
+      "step": 15833
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.458968495365048e-06,
+      "loss": 0.7236,
+      "step": 15834
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.457831068571858e-06,
+      "loss": 0.7344,
+      "step": 15835
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.456693662216022e-06,
+      "loss": 0.7227,
+      "step": 15836
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.45555627631261e-06,
+      "loss": 0.7148,
+      "step": 15837
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.454418910876694e-06,
+      "loss": 0.7139,
+      "step": 15838
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.453281565923351e-06,
+      "loss": 0.6934,
+      "step": 15839
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.45214424146765e-06,
+      "loss": 0.7275,
+      "step": 15840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.451006937524666e-06,
+      "loss": 0.7139,
+      "step": 15841
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.449869654109471e-06,
+      "loss": 0.749,
+      "step": 15842
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.448732391237136e-06,
+      "loss": 0.7432,
+      "step": 15843
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.447595148922727e-06,
+      "loss": 0.7432,
+      "step": 15844
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.446457927181323e-06,
+      "loss": 0.7217,
+      "step": 15845
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.445320726027989e-06,
+      "loss": 0.71,
+      "step": 15846
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.444183545477801e-06,
+      "loss": 0.7529,
+      "step": 15847
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.443046385545828e-06,
+      "loss": 0.7363,
+      "step": 15848
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.441909246247137e-06,
+      "loss": 0.7324,
+      "step": 15849
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.440772127596798e-06,
+      "loss": 0.71,
+      "step": 15850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.439635029609884e-06,
+      "loss": 0.7227,
+      "step": 15851
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.438497952301457e-06,
+      "loss": 0.7354,
+      "step": 15852
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.437360895686596e-06,
+      "loss": 0.7178,
+      "step": 15853
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.436223859780363e-06,
+      "loss": 0.7178,
+      "step": 15854
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.435086844597827e-06,
+      "loss": 0.7148,
+      "step": 15855
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.433949850154062e-06,
+      "loss": 0.7051,
+      "step": 15856
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.432812876464128e-06,
+      "loss": 0.7051,
+      "step": 15857
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.431675923543093e-06,
+      "loss": 0.7354,
+      "step": 15858
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.430538991406029e-06,
+      "loss": 0.7441,
+      "step": 15859
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.429402080068002e-06,
+      "loss": 0.7012,
+      "step": 15860
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.428265189544076e-06,
+      "loss": 0.752,
+      "step": 15861
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.427128319849322e-06,
+      "loss": 0.7285,
+      "step": 15862
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.425991470998802e-06,
+      "loss": 0.7188,
+      "step": 15863
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.424854643007581e-06,
+      "loss": 0.709,
+      "step": 15864
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.423717835890727e-06,
+      "loss": 0.7227,
+      "step": 15865
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.422581049663306e-06,
+      "loss": 0.7344,
+      "step": 15866
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42144428434038e-06,
+      "loss": 0.7139,
+      "step": 15867
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.420307539937021e-06,
+      "loss": 0.7285,
+      "step": 15868
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.419170816468287e-06,
+      "loss": 0.7266,
+      "step": 15869
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.418034113949245e-06,
+      "loss": 0.7168,
+      "step": 15870
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.416897432394952e-06,
+      "loss": 0.7207,
+      "step": 15871
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.415760771820483e-06,
+      "loss": 0.7314,
+      "step": 15872
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.414624132240893e-06,
+      "loss": 0.7217,
+      "step": 15873
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41348751367125e-06,
+      "loss": 0.7461,
+      "step": 15874
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.412350916126615e-06,
+      "loss": 0.7422,
+      "step": 15875
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41121433962205e-06,
+      "loss": 0.7939,
+      "step": 15876
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.410077784172616e-06,
+      "loss": 0.7266,
+      "step": 15877
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.40894124979338e-06,
+      "loss": 0.7002,
+      "step": 15878
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.407804736499397e-06,
+      "loss": 0.7129,
+      "step": 15879
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.406668244305733e-06,
+      "loss": 0.7266,
+      "step": 15880
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.405531773227447e-06,
+      "loss": 0.7158,
+      "step": 15881
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.404395323279608e-06,
+      "loss": 0.7217,
+      "step": 15882
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.403258894477265e-06,
+      "loss": 0.7158,
+      "step": 15883
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.402122486835482e-06,
+      "loss": 0.7295,
+      "step": 15884
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.40098610036932e-06,
+      "loss": 0.7139,
+      "step": 15885
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39984973509384e-06,
+      "loss": 0.7354,
+      "step": 15886
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.398713391024102e-06,
+      "loss": 0.7168,
+      "step": 15887
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.397577068175164e-06,
+      "loss": 0.7422,
+      "step": 15888
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.396440766562082e-06,
+      "loss": 0.7285,
+      "step": 15889
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.395304486199917e-06,
+      "loss": 0.7109,
+      "step": 15890
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.394168227103727e-06,
+      "loss": 0.7168,
+      "step": 15891
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.393031989288574e-06,
+      "loss": 0.7188,
+      "step": 15892
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.391895772769511e-06,
+      "loss": 0.7109,
+      "step": 15893
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.390759577561596e-06,
+      "loss": 0.7012,
+      "step": 15894
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.38962340367989e-06,
+      "loss": 0.7051,
+      "step": 15895
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.388487251139447e-06,
+      "loss": 0.7236,
+      "step": 15896
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38735111995532e-06,
+      "loss": 0.7402,
+      "step": 15897
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.386215010142573e-06,
+      "loss": 0.7197,
+      "step": 15898
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38507892171626e-06,
+      "loss": 0.7256,
+      "step": 15899
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.383942854691429e-06,
+      "loss": 0.7354,
+      "step": 15900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.382806809083149e-06,
+      "loss": 0.7197,
+      "step": 15901
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.381670784906465e-06,
+      "loss": 0.7354,
+      "step": 15902
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.380534782176433e-06,
+      "loss": 0.7207,
+      "step": 15903
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.379398800908113e-06,
+      "loss": 0.707,
+      "step": 15904
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.378262841116556e-06,
+      "loss": 0.7295,
+      "step": 15905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.377126902816812e-06,
+      "loss": 0.7432,
+      "step": 15906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.375990986023944e-06,
+      "loss": 0.7314,
+      "step": 15907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.374855090753003e-06,
+      "loss": 0.7227,
+      "step": 15908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.373719217019036e-06,
+      "loss": 0.7598,
+      "step": 15909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.3725833648371e-06,
+      "loss": 0.7588,
+      "step": 15910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.371447534222248e-06,
+      "loss": 0.7129,
+      "step": 15911
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.37031172518953e-06,
+      "loss": 0.6934,
+      "step": 15912
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.369175937754004e-06,
+      "loss": 0.7314,
+      "step": 15913
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.368040171930718e-06,
+      "loss": 0.7295,
+      "step": 15914
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.366904427734724e-06,
+      "loss": 0.7148,
+      "step": 15915
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36576870518107e-06,
+      "loss": 0.7158,
+      "step": 15916
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36463300428481e-06,
+      "loss": 0.7314,
+      "step": 15917
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.363497325060997e-06,
+      "loss": 0.7285,
+      "step": 15918
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.362361667524678e-06,
+      "loss": 0.7031,
+      "step": 15919
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.361226031690903e-06,
+      "loss": 0.7129,
+      "step": 15920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.360090417574728e-06,
+      "loss": 0.7295,
+      "step": 15921
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.358954825191193e-06,
+      "loss": 0.7178,
+      "step": 15922
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.357819254555353e-06,
+      "loss": 0.7168,
+      "step": 15923
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.356683705682253e-06,
+      "loss": 0.7266,
+      "step": 15924
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.355548178586947e-06,
+      "loss": 0.7676,
+      "step": 15925
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.354412673284479e-06,
+      "loss": 0.7529,
+      "step": 15926
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.353277189789906e-06,
+      "loss": 0.7119,
+      "step": 15927
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35214172811826e-06,
+      "loss": 0.7607,
+      "step": 15928
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.351006288284604e-06,
+      "loss": 0.7158,
+      "step": 15929
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.349870870303974e-06,
+      "loss": 0.7451,
+      "step": 15930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348735474191423e-06,
+      "loss": 0.7178,
+      "step": 15931
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.347600099961996e-06,
+      "loss": 0.6963,
+      "step": 15932
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.346464747630743e-06,
+      "loss": 0.7305,
+      "step": 15933
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.345329417212703e-06,
+      "loss": 0.71,
+      "step": 15934
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.344194108722926e-06,
+      "loss": 0.7217,
+      "step": 15935
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.343058822176454e-06,
+      "loss": 0.7148,
+      "step": 15936
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.341923557588337e-06,
+      "loss": 0.7139,
+      "step": 15937
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.34078831497362e-06,
+      "loss": 0.7314,
+      "step": 15938
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.339653094347341e-06,
+      "loss": 0.7051,
+      "step": 15939
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.338517895724556e-06,
+      "loss": 0.7383,
+      "step": 15940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.3373827191203e-06,
+      "loss": 0.7148,
+      "step": 15941
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.336247564549614e-06,
+      "loss": 0.7178,
+      "step": 15942
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.33511243202755e-06,
+      "loss": 0.7354,
+      "step": 15943
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.333977321569147e-06,
+      "loss": 0.7314,
+      "step": 15944
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.332842233189447e-06,
+      "loss": 0.7422,
+      "step": 15945
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.331707166903498e-06,
+      "loss": 0.7412,
+      "step": 15946
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.330572122726334e-06,
+      "loss": 0.7314,
+      "step": 15947
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.329437100673001e-06,
+      "loss": 0.7285,
+      "step": 15948
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.328302100758543e-06,
+      "loss": 0.7129,
+      "step": 15949
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.327167122997998e-06,
+      "loss": 0.71,
+      "step": 15950
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.326032167406408e-06,
+      "loss": 0.7158,
+      "step": 15951
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.324897233998815e-06,
+      "loss": 0.7266,
+      "step": 15952
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.32376232279026e-06,
+      "loss": 0.7451,
+      "step": 15953
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.322627433795782e-06,
+      "loss": 0.7314,
+      "step": 15954
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.321492567030417e-06,
+      "loss": 0.7178,
+      "step": 15955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.320357722509212e-06,
+      "loss": 0.7422,
+      "step": 15956
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.319222900247201e-06,
+      "loss": 0.7432,
+      "step": 15957
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.318088100259426e-06,
+      "loss": 0.6963,
+      "step": 15958
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.316953322560927e-06,
+      "loss": 0.7305,
+      "step": 15959
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.315818567166736e-06,
+      "loss": 0.7227,
+      "step": 15960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.314683834091897e-06,
+      "loss": 0.7266,
+      "step": 15961
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.313549123351445e-06,
+      "loss": 0.6973,
+      "step": 15962
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.31241443496042e-06,
+      "loss": 0.7178,
+      "step": 15963
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.311279768933857e-06,
+      "loss": 0.7383,
+      "step": 15964
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.310145125286793e-06,
+      "loss": 0.7344,
+      "step": 15965
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.309010504034272e-06,
+      "loss": 0.748,
+      "step": 15966
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.307875905191317e-06,
+      "loss": 0.7168,
+      "step": 15967
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.306741328772976e-06,
+      "loss": 0.7178,
+      "step": 15968
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.305606774794276e-06,
+      "loss": 0.7383,
+      "step": 15969
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.304472243270258e-06,
+      "loss": 0.7295,
+      "step": 15970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.303337734215957e-06,
+      "loss": 0.7314,
+      "step": 15971
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30220324764641e-06,
+      "loss": 0.7227,
+      "step": 15972
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30106878357664e-06,
+      "loss": 0.7129,
+      "step": 15973
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.299934342021694e-06,
+      "loss": 0.7236,
+      "step": 15974
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.2987999229966e-06,
+      "loss": 0.7334,
+      "step": 15975
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.297665526516394e-06,
+      "loss": 0.7168,
+      "step": 15976
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.296531152596108e-06,
+      "loss": 0.7061,
+      "step": 15977
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.295396801250775e-06,
+      "loss": 0.7041,
+      "step": 15978
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.294262472495433e-06,
+      "loss": 0.7227,
+      "step": 15979
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.293128166345108e-06,
+      "loss": 0.7275,
+      "step": 15980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.291993882814831e-06,
+      "loss": 0.7607,
+      "step": 15981
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.290859621919642e-06,
+      "loss": 0.7295,
+      "step": 15982
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.289725383674564e-06,
+      "loss": 0.7334,
+      "step": 15983
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.288591168094633e-06,
+      "loss": 0.7256,
+      "step": 15984
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.287456975194881e-06,
+      "loss": 0.7295,
+      "step": 15985
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.286322804990337e-06,
+      "loss": 0.7148,
+      "step": 15986
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.285188657496029e-06,
+      "loss": 0.7051,
+      "step": 15987
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.284054532726991e-06,
+      "loss": 0.7178,
+      "step": 15988
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.28292043069825e-06,
+      "loss": 0.7422,
+      "step": 15989
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.281786351424835e-06,
+      "loss": 0.7178,
+      "step": 15990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.28065229492178e-06,
+      "loss": 0.75,
+      "step": 15991
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.279518261204112e-06,
+      "loss": 0.7363,
+      "step": 15992
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.278384250286857e-06,
+      "loss": 0.7168,
+      "step": 15993
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.277250262185043e-06,
+      "loss": 0.7178,
+      "step": 15994
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.2761162969137e-06,
+      "loss": 0.752,
+      "step": 15995
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.274982354487854e-06,
+      "loss": 0.7285,
+      "step": 15996
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.273848434922538e-06,
+      "loss": 0.7031,
+      "step": 15997
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.272714538232776e-06,
+      "loss": 0.7012,
+      "step": 15998
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27158066443359e-06,
+      "loss": 0.7129,
+      "step": 15999
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27044681354001e-06,
+      "loss": 0.749,
+      "step": 16000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.269312985567065e-06,
+      "loss": 0.7451,
+      "step": 16001
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.268179180529773e-06,
+      "loss": 0.71,
+      "step": 16002
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.267045398443169e-06,
+      "loss": 0.7334,
+      "step": 16003
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.265911639322271e-06,
+      "loss": 0.7324,
+      "step": 16004
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.264777903182116e-06,
+      "loss": 0.7285,
+      "step": 16005
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26364419003771e-06,
+      "loss": 0.749,
+      "step": 16006
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26251049990409e-06,
+      "loss": 0.7227,
+      "step": 16007
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.261376832796276e-06,
+      "loss": 0.7471,
+      "step": 16008
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.260243188729295e-06,
+      "loss": 0.751,
+      "step": 16009
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.259109567718166e-06,
+      "loss": 0.7402,
+      "step": 16010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.257975969777921e-06,
+      "loss": 0.7148,
+      "step": 16011
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.256842394923569e-06,
+      "loss": 0.7539,
+      "step": 16012
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.255708843170143e-06,
+      "loss": 0.7168,
+      "step": 16013
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.25457531453266e-06,
+      "loss": 0.7266,
+      "step": 16014
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.253441809026144e-06,
+      "loss": 0.7197,
+      "step": 16015
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.252308326665618e-06,
+      "loss": 0.7041,
+      "step": 16016
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2511748674661e-06,
+      "loss": 0.7236,
+      "step": 16017
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.250041431442616e-06,
+      "loss": 0.7295,
+      "step": 16018
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.248908018610183e-06,
+      "loss": 0.7197,
+      "step": 16019
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.247774628983819e-06,
+      "loss": 0.7266,
+      "step": 16020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.246641262578547e-06,
+      "loss": 0.7158,
+      "step": 16021
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.245507919409389e-06,
+      "loss": 0.707,
+      "step": 16022
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.244374599491359e-06,
+      "loss": 0.7354,
+      "step": 16023
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.243241302839484e-06,
+      "loss": 0.7109,
+      "step": 16024
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.242108029468776e-06,
+      "loss": 0.7344,
+      "step": 16025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.240974779394255e-06,
+      "loss": 0.7412,
+      "step": 16026
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23984155263094e-06,
+      "loss": 0.7383,
+      "step": 16027
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.238708349193849e-06,
+      "loss": 0.708,
+      "step": 16028
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.237575169097997e-06,
+      "loss": 0.7334,
+      "step": 16029
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.236442012358407e-06,
+      "loss": 0.7461,
+      "step": 16030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.235308878990093e-06,
+      "loss": 0.7402,
+      "step": 16031
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23417576900807e-06,
+      "loss": 0.7363,
+      "step": 16032
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.233042682427356e-06,
+      "loss": 0.7578,
+      "step": 16033
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.231909619262967e-06,
+      "loss": 0.7158,
+      "step": 16034
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.230776579529916e-06,
+      "loss": 0.7295,
+      "step": 16035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.229643563243223e-06,
+      "loss": 0.7285,
+      "step": 16036
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228510570417904e-06,
+      "loss": 0.709,
+      "step": 16037
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.227377601068968e-06,
+      "loss": 0.71,
+      "step": 16038
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.22624465521143e-06,
+      "loss": 0.7412,
+      "step": 16039
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.22511173286031e-06,
+      "loss": 0.7334,
+      "step": 16040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.223978834030615e-06,
+      "loss": 0.7305,
+      "step": 16041
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.222845958737364e-06,
+      "loss": 0.7275,
+      "step": 16042
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.221713106995566e-06,
+      "loss": 0.7275,
+      "step": 16043
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.220580278820243e-06,
+      "loss": 0.71,
+      "step": 16044
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.219447474226395e-06,
+      "loss": 0.7158,
+      "step": 16045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.21831469322904e-06,
+      "loss": 0.707,
+      "step": 16046
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.217181935843191e-06,
+      "loss": 0.7383,
+      "step": 16047
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.21604920208386e-06,
+      "loss": 0.7188,
+      "step": 16048
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.214916491966053e-06,
+      "loss": 0.7422,
+      "step": 16049
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.213783805504795e-06,
+      "loss": 0.7217,
+      "step": 16050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.21265114271508e-06,
+      "loss": 0.7236,
+      "step": 16051
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.211518503611925e-06,
+      "loss": 0.709,
+      "step": 16052
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.21038588821034e-06,
+      "loss": 0.7324,
+      "step": 16053
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.209253296525336e-06,
+      "loss": 0.7285,
+      "step": 16054
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.208120728571925e-06,
+      "loss": 0.7373,
+      "step": 16055
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.206988184365108e-06,
+      "loss": 0.7236,
+      "step": 16056
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.205855663919905e-06,
+      "loss": 0.7354,
+      "step": 16057
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.204723167251317e-06,
+      "loss": 0.7246,
+      "step": 16058
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.203590694374349e-06,
+      "loss": 0.75,
+      "step": 16059
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.202458245304016e-06,
+      "loss": 0.7393,
+      "step": 16060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.201325820055325e-06,
+      "loss": 0.7305,
+      "step": 16061
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.200193418643279e-06,
+      "loss": 0.7051,
+      "step": 16062
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.199061041082891e-06,
+      "loss": 0.7236,
+      "step": 16063
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.197928687389161e-06,
+      "loss": 0.7236,
+      "step": 16064
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.196796357577098e-06,
+      "loss": 0.7529,
+      "step": 16065
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.19566405166171e-06,
+      "loss": 0.7324,
+      "step": 16066
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.194531769658001e-06,
+      "loss": 0.7354,
+      "step": 16067
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.193399511580975e-06,
+      "loss": 0.71,
+      "step": 16068
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.192267277445641e-06,
+      "loss": 0.7256,
+      "step": 16069
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.191135067267004e-06,
+      "loss": 0.7109,
+      "step": 16070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.19000288106006e-06,
+      "loss": 0.7041,
+      "step": 16071
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.188870718839822e-06,
+      "loss": 0.7461,
+      "step": 16072
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18773858062129e-06,
+      "loss": 0.7295,
+      "step": 16073
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.186606466419467e-06,
+      "loss": 0.7197,
+      "step": 16074
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.185474376249359e-06,
+      "loss": 0.7227,
+      "step": 16075
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18434231012597e-06,
+      "loss": 0.751,
+      "step": 16076
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.183210268064297e-06,
+      "loss": 0.7344,
+      "step": 16077
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.182078250079343e-06,
+      "loss": 0.6924,
+      "step": 16078
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.180946256186113e-06,
+      "loss": 0.7207,
+      "step": 16079
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.179814286399608e-06,
+      "loss": 0.7109,
+      "step": 16080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.17868234073483e-06,
+      "loss": 0.709,
+      "step": 16081
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.177550419206774e-06,
+      "loss": 0.7314,
+      "step": 16082
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.176418521830456e-06,
+      "loss": 0.7188,
+      "step": 16083
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.175286648620854e-06,
+      "loss": 0.7451,
+      "step": 16084
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.174154799592986e-06,
+      "loss": 0.7334,
+      "step": 16085
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.173022974761841e-06,
+      "loss": 0.75,
+      "step": 16086
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.171891174142424e-06,
+      "loss": 0.7158,
+      "step": 16087
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.170759397749732e-06,
+      "loss": 0.7197,
+      "step": 16088
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16962764559877e-06,
+      "loss": 0.7178,
+      "step": 16089
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.168495917704524e-06,
+      "loss": 0.7236,
+      "step": 16090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.167364214082001e-06,
+      "loss": 0.7119,
+      "step": 16091
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.166232534746195e-06,
+      "loss": 0.6973,
+      "step": 16092
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.165100879712106e-06,
+      "loss": 0.7256,
+      "step": 16093
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.163969248994727e-06,
+      "loss": 0.7178,
+      "step": 16094
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16283764260906e-06,
+      "loss": 0.752,
+      "step": 16095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.161706060570103e-06,
+      "loss": 0.7207,
+      "step": 16096
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.160574502892844e-06,
+      "loss": 0.709,
+      "step": 16097
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.159442969592282e-06,
+      "loss": 0.7021,
+      "step": 16098
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.158311460683415e-06,
+      "loss": 0.7256,
+      "step": 16099
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.157179976181236e-06,
+      "loss": 0.707,
+      "step": 16100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.156048516100738e-06,
+      "loss": 0.7393,
+      "step": 16101
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.154917080456924e-06,
+      "loss": 0.7314,
+      "step": 16102
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.153785669264778e-06,
+      "loss": 0.7402,
+      "step": 16103
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.152654282539296e-06,
+      "loss": 0.7207,
+      "step": 16104
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.151522920295476e-06,
+      "loss": 0.7275,
+      "step": 16105
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.15039158254831e-06,
+      "loss": 0.7188,
+      "step": 16106
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.149260269312785e-06,
+      "loss": 0.7197,
+      "step": 16107
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.148128980603903e-06,
+      "loss": 0.7471,
+      "step": 16108
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.14699771643665e-06,
+      "loss": 0.7236,
+      "step": 16109
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.145866476826017e-06,
+      "loss": 0.748,
+      "step": 16110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.144735261786999e-06,
+      "loss": 0.7246,
+      "step": 16111
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.143604071334586e-06,
+      "loss": 0.7236,
+      "step": 16112
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.142472905483767e-06,
+      "loss": 0.7314,
+      "step": 16113
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.141341764249537e-06,
+      "loss": 0.7129,
+      "step": 16114
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.140210647646886e-06,
+      "loss": 0.7363,
+      "step": 16115
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.139079555690797e-06,
+      "loss": 0.7402,
+      "step": 16116
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.137948488396267e-06,
+      "loss": 0.7227,
+      "step": 16117
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.136817445778283e-06,
+      "loss": 0.7207,
+      "step": 16118
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.13568642785183e-06,
+      "loss": 0.7383,
+      "step": 16119
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.134555434631903e-06,
+      "loss": 0.7129,
+      "step": 16120
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.133424466133486e-06,
+      "loss": 0.7373,
+      "step": 16121
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.132293522371576e-06,
+      "loss": 0.7246,
+      "step": 16122
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.131162603361144e-06,
+      "loss": 0.7207,
+      "step": 16123
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.13003170911719e-06,
+      "loss": 0.7432,
+      "step": 16124
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.128900839654695e-06,
+      "loss": 0.7256,
+      "step": 16125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12776999498865e-06,
+      "loss": 0.7529,
+      "step": 16126
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.126639175134038e-06,
+      "loss": 0.7432,
+      "step": 16127
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.125508380105853e-06,
+      "loss": 0.7158,
+      "step": 16128
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.124377609919067e-06,
+      "loss": 0.7539,
+      "step": 16129
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.123246864588677e-06,
+      "loss": 0.708,
+      "step": 16130
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.12211614412966e-06,
+      "loss": 0.7354,
+      "step": 16131
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.120985448557007e-06,
+      "loss": 0.7441,
+      "step": 16132
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.119854777885696e-06,
+      "loss": 0.7324,
+      "step": 16133
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.118724132130717e-06,
+      "loss": 0.7139,
+      "step": 16134
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.117593511307055e-06,
+      "loss": 0.7061,
+      "step": 16135
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.116462915429687e-06,
+      "loss": 0.7373,
+      "step": 16136
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.115332344513596e-06,
+      "loss": 0.7158,
+      "step": 16137
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.114201798573771e-06,
+      "loss": 0.709,
+      "step": 16138
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.113071277625188e-06,
+      "loss": 0.7354,
+      "step": 16139
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.111940781682835e-06,
+      "loss": 0.7324,
+      "step": 16140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.110810310761692e-06,
+      "loss": 0.7539,
+      "step": 16141
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.109679864876735e-06,
+      "loss": 0.71,
+      "step": 16142
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10854944404295e-06,
+      "loss": 0.708,
+      "step": 16143
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.107419048275318e-06,
+      "loss": 0.749,
+      "step": 16144
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.106288677588818e-06,
+      "loss": 0.7549,
+      "step": 16145
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.105158331998428e-06,
+      "loss": 0.7148,
+      "step": 16146
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.104028011519135e-06,
+      "loss": 0.7295,
+      "step": 16147
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.102897716165911e-06,
+      "loss": 0.7422,
+      "step": 16148
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.101767445953737e-06,
+      "loss": 0.7461,
+      "step": 16149
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.100637200897592e-06,
+      "loss": 0.7305,
+      "step": 16150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.099506981012456e-06,
+      "loss": 0.7432,
+      "step": 16151
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.098376786313302e-06,
+      "loss": 0.7383,
+      "step": 16152
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.097246616815115e-06,
+      "loss": 0.7344,
+      "step": 16153
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.09611647253287e-06,
+      "loss": 0.6904,
+      "step": 16154
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.09498635348154e-06,
+      "loss": 0.7539,
+      "step": 16155
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.093856259676105e-06,
+      "loss": 0.7207,
+      "step": 16156
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.092726191131542e-06,
+      "loss": 0.7207,
+      "step": 16157
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.091596147862823e-06,
+      "loss": 0.7139,
+      "step": 16158
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.09046612988493e-06,
+      "loss": 0.7051,
+      "step": 16159
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.089336137212838e-06,
+      "loss": 0.7549,
+      "step": 16160
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.088206169861512e-06,
+      "loss": 0.7256,
+      "step": 16161
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.087076227845939e-06,
+      "loss": 0.7354,
+      "step": 16162
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.085946311181086e-06,
+      "loss": 0.7432,
+      "step": 16163
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.08481641988193e-06,
+      "loss": 0.7129,
+      "step": 16164
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.083686553963444e-06,
+      "loss": 0.7334,
+      "step": 16165
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.082556713440599e-06,
+      "loss": 0.7354,
+      "step": 16166
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.081426898328378e-06,
+      "loss": 0.7305,
+      "step": 16167
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.08029710864174e-06,
+      "loss": 0.7402,
+      "step": 16168
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.079167344395665e-06,
+      "loss": 0.7275,
+      "step": 16169
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.07803760560512e-06,
+      "loss": 0.7266,
+      "step": 16170
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076907892285086e-06,
+      "loss": 0.7363,
+      "step": 16171
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.075778204450522e-06,
+      "loss": 0.7188,
+      "step": 16172
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.074648542116416e-06,
+      "loss": 0.7256,
+      "step": 16173
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.073518905297719e-06,
+      "loss": 0.7021,
+      "step": 16174
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.072389294009414e-06,
+      "loss": 0.7393,
+      "step": 16175
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.071259708266466e-06,
+      "loss": 0.7227,
+      "step": 16176
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.070130148083847e-06,
+      "loss": 0.7129,
+      "step": 16177
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.069000613476524e-06,
+      "loss": 0.7109,
+      "step": 16178
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.067871104459471e-06,
+      "loss": 0.7197,
+      "step": 16179
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.066741621047653e-06,
+      "loss": 0.7148,
+      "step": 16180
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.065612163256037e-06,
+      "loss": 0.7148,
+      "step": 16181
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.064482731099591e-06,
+      "loss": 0.7188,
+      "step": 16182
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.063353324593286e-06,
+      "loss": 0.7373,
+      "step": 16183
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.062223943752087e-06,
+      "loss": 0.7363,
+      "step": 16184
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06109458859096e-06,
+      "loss": 0.71,
+      "step": 16185
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.059965259124875e-06,
+      "loss": 0.7324,
+      "step": 16186
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.058835955368797e-06,
+      "loss": 0.7383,
+      "step": 16187
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.057706677337686e-06,
+      "loss": 0.7344,
+      "step": 16188
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.056577425046517e-06,
+      "loss": 0.75,
+      "step": 16189
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05544819851025e-06,
+      "loss": 0.7402,
+      "step": 16190
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05431899774385e-06,
+      "loss": 0.7139,
+      "step": 16191
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053189822762283e-06,
+      "loss": 0.7188,
+      "step": 16192
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.052060673580514e-06,
+      "loss": 0.6924,
+      "step": 16193
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.050931550213503e-06,
+      "loss": 0.707,
+      "step": 16194
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.049802452676217e-06,
+      "loss": 0.6943,
+      "step": 16195
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.048673380983618e-06,
+      "loss": 0.7275,
+      "step": 16196
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.047544335150667e-06,
+      "loss": 0.7451,
+      "step": 16197
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.046415315192333e-06,
+      "loss": 0.7285,
+      "step": 16198
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.045286321123572e-06,
+      "loss": 0.751,
+      "step": 16199
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.044157352959345e-06,
+      "loss": 0.7471,
+      "step": 16200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.04302841071462e-06,
+      "loss": 0.7393,
+      "step": 16201
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.04189949440435e-06,
+      "loss": 0.7148,
+      "step": 16202
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.0407706040435e-06,
+      "loss": 0.7197,
+      "step": 16203
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.039641739647032e-06,
+      "loss": 0.7197,
+      "step": 16204
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.038512901229905e-06,
+      "loss": 0.7451,
+      "step": 16205
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.037384088807081e-06,
+      "loss": 0.7471,
+      "step": 16206
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036255302393511e-06,
+      "loss": 0.7188,
+      "step": 16207
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.035126542004162e-06,
+      "loss": 0.7148,
+      "step": 16208
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.033997807653989e-06,
+      "loss": 0.7129,
+      "step": 16209
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.032869099357953e-06,
+      "loss": 0.7334,
+      "step": 16210
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.031740417131008e-06,
+      "loss": 0.7119,
+      "step": 16211
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.030611760988123e-06,
+      "loss": 0.7012,
+      "step": 16212
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02948313094424e-06,
+      "loss": 0.7354,
+      "step": 16213
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.028354527014323e-06,
+      "loss": 0.7227,
+      "step": 16214
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.027225949213327e-06,
+      "loss": 0.7031,
+      "step": 16215
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.026097397556213e-06,
+      "loss": 0.7656,
+      "step": 16216
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02496887205793e-06,
+      "loss": 0.75,
+      "step": 16217
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.02384037273344e-06,
+      "loss": 0.7461,
+      "step": 16218
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.022711899597696e-06,
+      "loss": 0.7129,
+      "step": 16219
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.021583452665654e-06,
+      "loss": 0.748,
+      "step": 16220
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.020455031952264e-06,
+      "loss": 0.7441,
+      "step": 16221
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.019326637472484e-06,
+      "loss": 0.7422,
+      "step": 16222
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.018198269241264e-06,
+      "loss": 0.7412,
+      "step": 16223
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.017069927273564e-06,
+      "loss": 0.7295,
+      "step": 16224
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.015941611584337e-06,
+      "loss": 0.7246,
+      "step": 16225
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.01481332218853e-06,
+      "loss": 0.707,
+      "step": 16226
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.013685059101095e-06,
+      "loss": 0.7373,
+      "step": 16227
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.01255682233699e-06,
+      "loss": 0.7354,
+      "step": 16228
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.011428611911164e-06,
+      "loss": 0.7656,
+      "step": 16229
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.010300427838566e-06,
+      "loss": 0.7422,
+      "step": 16230
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.009172270134153e-06,
+      "loss": 0.7471,
+      "step": 16231
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.008044138812873e-06,
+      "loss": 0.7129,
+      "step": 16232
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.006916033889671e-06,
+      "loss": 0.7539,
+      "step": 16233
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.005787955379505e-06,
+      "loss": 0.7129,
+      "step": 16234
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.004659903297322e-06,
+      "loss": 0.7344,
+      "step": 16235
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.003531877658068e-06,
+      "loss": 0.71,
+      "step": 16236
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.002403878476695e-06,
+      "loss": 0.7139,
+      "step": 16237
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.001275905768156e-06,
+      "loss": 0.709,
+      "step": 16238
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.000147959547388e-06,
+      "loss": 0.7529,
+      "step": 16239
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.99902003982935e-06,
+      "loss": 0.7109,
+      "step": 16240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.997892146628982e-06,
+      "loss": 0.7373,
+      "step": 16241
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.996764279961236e-06,
+      "loss": 0.7061,
+      "step": 16242
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.995636439841057e-06,
+      "loss": 0.7227,
+      "step": 16243
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.994508626283391e-06,
+      "loss": 0.7432,
+      "step": 16244
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.993380839303188e-06,
+      "loss": 0.7188,
+      "step": 16245
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.99225307891539e-06,
+      "loss": 0.7051,
+      "step": 16246
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.991125345134942e-06,
+      "loss": 0.748,
+      "step": 16247
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.989997637976788e-06,
+      "loss": 0.7266,
+      "step": 16248
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.988869957455877e-06,
+      "loss": 0.7266,
+      "step": 16249
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98774230358715e-06,
+      "loss": 0.7256,
+      "step": 16250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.986614676385559e-06,
+      "loss": 0.7158,
+      "step": 16251
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.985487075866033e-06,
+      "loss": 0.7109,
+      "step": 16252
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.984359502043528e-06,
+      "loss": 0.7246,
+      "step": 16253
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.983231954932981e-06,
+      "loss": 0.7129,
+      "step": 16254
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.982104434549337e-06,
+      "loss": 0.7432,
+      "step": 16255
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.980976940907535e-06,
+      "loss": 0.7139,
+      "step": 16256
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.979849474022523e-06,
+      "loss": 0.7158,
+      "step": 16257
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.97872203390924e-06,
+      "loss": 0.7236,
+      "step": 16258
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.977594620582626e-06,
+      "loss": 0.7109,
+      "step": 16259
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.976467234057618e-06,
+      "loss": 0.7256,
+      "step": 16260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.975339874349163e-06,
+      "loss": 0.7148,
+      "step": 16261
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.974212541472197e-06,
+      "loss": 0.7422,
+      "step": 16262
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.973085235441665e-06,
+      "loss": 0.7324,
+      "step": 16263
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.971957956272503e-06,
+      "loss": 0.709,
+      "step": 16264
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.970830703979649e-06,
+      "loss": 0.7012,
+      "step": 16265
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.969703478578041e-06,
+      "loss": 0.7217,
+      "step": 16266
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.968576280082621e-06,
+      "loss": 0.6992,
+      "step": 16267
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.967449108508326e-06,
+      "loss": 0.7041,
+      "step": 16268
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.966321963870088e-06,
+      "loss": 0.7354,
+      "step": 16269
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.965194846182855e-06,
+      "loss": 0.7529,
+      "step": 16270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.964067755461557e-06,
+      "loss": 0.7295,
+      "step": 16271
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96294069172113e-06,
+      "loss": 0.7275,
+      "step": 16272
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.961813654976513e-06,
+      "loss": 0.7109,
+      "step": 16273
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.960686645242641e-06,
+      "loss": 0.7334,
+      "step": 16274
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.959559662534449e-06,
+      "loss": 0.7402,
+      "step": 16275
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.958432706866873e-06,
+      "loss": 0.749,
+      "step": 16276
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95730577825485e-06,
+      "loss": 0.7158,
+      "step": 16277
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.956178876713307e-06,
+      "loss": 0.751,
+      "step": 16278
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.955052002257187e-06,
+      "loss": 0.7305,
+      "step": 16279
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.953925154901418e-06,
+      "loss": 0.7305,
+      "step": 16280
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.952798334660931e-06,
+      "loss": 0.7451,
+      "step": 16281
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.951671541550668e-06,
+      "loss": 0.7402,
+      "step": 16282
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.950544775585554e-06,
+      "loss": 0.7188,
+      "step": 16283
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.949418036780527e-06,
+      "loss": 0.71,
+      "step": 16284
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.948291325150516e-06,
+      "loss": 0.7178,
+      "step": 16285
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.94716464071045e-06,
+      "loss": 0.7217,
+      "step": 16286
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.946037983475262e-06,
+      "loss": 0.7168,
+      "step": 16287
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.944911353459885e-06,
+      "loss": 0.7197,
+      "step": 16288
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.943784750679246e-06,
+      "loss": 0.7197,
+      "step": 16289
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.942658175148285e-06,
+      "loss": 0.7188,
+      "step": 16290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.941531626881915e-06,
+      "loss": 0.7471,
+      "step": 16291
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.940405105895078e-06,
+      "loss": 0.7227,
+      "step": 16292
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.939278612202697e-06,
+      "loss": 0.7188,
+      "step": 16293
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.938152145819703e-06,
+      "loss": 0.7344,
+      "step": 16294
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.937025706761025e-06,
+      "loss": 0.7246,
+      "step": 16295
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.935899295041593e-06,
+      "loss": 0.7441,
+      "step": 16296
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.934772910676332e-06,
+      "loss": 0.7168,
+      "step": 16297
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.933646553680168e-06,
+      "loss": 0.7324,
+      "step": 16298
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.932520224068025e-06,
+      "loss": 0.7373,
+      "step": 16299
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.931393921854838e-06,
+      "loss": 0.7148,
+      "step": 16300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.930267647055525e-06,
+      "loss": 0.7197,
+      "step": 16301
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92914139968502e-06,
+      "loss": 0.7285,
+      "step": 16302
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.928015179758242e-06,
+      "loss": 0.7393,
+      "step": 16303
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.926888987290118e-06,
+      "loss": 0.7568,
+      "step": 16304
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.925762822295571e-06,
+      "loss": 0.708,
+      "step": 16305
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.924636684789528e-06,
+      "loss": 0.7275,
+      "step": 16306
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92351057478691e-06,
+      "loss": 0.7461,
+      "step": 16307
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.922384492302646e-06,
+      "loss": 0.71,
+      "step": 16308
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.921258437351654e-06,
+      "loss": 0.7383,
+      "step": 16309
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.920132409948859e-06,
+      "loss": 0.6826,
+      "step": 16310
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.919006410109182e-06,
+      "loss": 0.7393,
+      "step": 16311
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.917880437847547e-06,
+      "loss": 0.71,
+      "step": 16312
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.916754493178875e-06,
+      "loss": 0.7275,
+      "step": 16313
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.915628576118086e-06,
+      "loss": 0.7129,
+      "step": 16314
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.914502686680104e-06,
+      "loss": 0.7256,
+      "step": 16315
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91337682487985e-06,
+      "loss": 0.752,
+      "step": 16316
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.912250990732238e-06,
+      "loss": 0.7158,
+      "step": 16317
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.911125184252195e-06,
+      "loss": 0.7207,
+      "step": 16318
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.909999405454637e-06,
+      "loss": 0.7188,
+      "step": 16319
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.908873654354483e-06,
+      "loss": 0.7354,
+      "step": 16320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.907747930966656e-06,
+      "loss": 0.7354,
+      "step": 16321
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.906622235306068e-06,
+      "loss": 0.7324,
+      "step": 16322
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.905496567387644e-06,
+      "loss": 0.7461,
+      "step": 16323
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.904370927226298e-06,
+      "loss": 0.7363,
+      "step": 16324
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.903245314836945e-06,
+      "loss": 0.7139,
+      "step": 16325
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.902119730234504e-06,
+      "loss": 0.7363,
+      "step": 16326
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.900994173433895e-06,
+      "loss": 0.7236,
+      "step": 16327
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.899868644450028e-06,
+      "loss": 0.709,
+      "step": 16328
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.898743143297828e-06,
+      "loss": 0.7178,
+      "step": 16329
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.897617669992203e-06,
+      "loss": 0.749,
+      "step": 16330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.89649222454807e-06,
+      "loss": 0.7266,
+      "step": 16331
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.89536680698034e-06,
+      "loss": 0.7412,
+      "step": 16332
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.894241417303934e-06,
+      "loss": 0.7344,
+      "step": 16333
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.893116055533763e-06,
+      "loss": 0.7148,
+      "step": 16334
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.891990721684742e-06,
+      "loss": 0.7275,
+      "step": 16335
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.890865415771786e-06,
+      "loss": 0.7178,
+      "step": 16336
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.889740137809802e-06,
+      "loss": 0.7197,
+      "step": 16337
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.888614887813705e-06,
+      "loss": 0.7334,
+      "step": 16338
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.88748966579841e-06,
+      "loss": 0.7188,
+      "step": 16339
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.886364471778825e-06,
+      "loss": 0.7549,
+      "step": 16340
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.885239305769865e-06,
+      "loss": 0.7363,
+      "step": 16341
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.884114167786442e-06,
+      "loss": 0.7393,
+      "step": 16342
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.882989057843462e-06,
+      "loss": 0.709,
+      "step": 16343
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.881863975955834e-06,
+      "loss": 0.7051,
+      "step": 16344
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.880738922138476e-06,
+      "loss": 0.7207,
+      "step": 16345
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.879613896406288e-06,
+      "loss": 0.7227,
+      "step": 16346
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.87848889877419e-06,
+      "loss": 0.7344,
+      "step": 16347
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.877363929257082e-06,
+      "loss": 0.7168,
+      "step": 16348
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.87623898786988e-06,
+      "loss": 0.7764,
+      "step": 16349
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.87511407462748e-06,
+      "loss": 0.7178,
+      "step": 16350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.873989189544804e-06,
+      "loss": 0.7295,
+      "step": 16351
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.872864332636749e-06,
+      "loss": 0.7168,
+      "step": 16352
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.871739503918228e-06,
+      "loss": 0.7295,
+      "step": 16353
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.870614703404144e-06,
+      "loss": 0.7109,
+      "step": 16354
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.869489931109408e-06,
+      "loss": 0.7158,
+      "step": 16355
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.86836518704892e-06,
+      "loss": 0.7168,
+      "step": 16356
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.867240471237587e-06,
+      "loss": 0.748,
+      "step": 16357
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.866115783690318e-06,
+      "loss": 0.7393,
+      "step": 16358
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.864991124422012e-06,
+      "loss": 0.7217,
+      "step": 16359
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.863866493447578e-06,
+      "loss": 0.7295,
+      "step": 16360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.86274189078192e-06,
+      "loss": 0.7246,
+      "step": 16361
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.861617316439936e-06,
+      "loss": 0.7197,
+      "step": 16362
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.860492770436536e-06,
+      "loss": 0.7578,
+      "step": 16363
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.85936825278662e-06,
+      "loss": 0.7236,
+      "step": 16364
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.858243763505087e-06,
+      "loss": 0.7393,
+      "step": 16365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.857119302606845e-06,
+      "loss": 0.7217,
+      "step": 16366
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.855994870106793e-06,
+      "loss": 0.7383,
+      "step": 16367
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.854870466019834e-06,
+      "loss": 0.7539,
+      "step": 16368
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.85374609036087e-06,
+      "loss": 0.7246,
+      "step": 16369
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.852621743144796e-06,
+      "loss": 0.7188,
+      "step": 16370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.851497424386513e-06,
+      "loss": 0.7373,
+      "step": 16371
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.850373134100928e-06,
+      "loss": 0.7148,
+      "step": 16372
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.849248872302932e-06,
+      "loss": 0.7119,
+      "step": 16373
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.848124639007437e-06,
+      "loss": 0.7012,
+      "step": 16374
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.847000434229324e-06,
+      "loss": 0.7441,
+      "step": 16375
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.845876257983503e-06,
+      "loss": 0.7422,
+      "step": 16376
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.844752110284866e-06,
+      "loss": 0.7178,
+      "step": 16377
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.843627991148316e-06,
+      "loss": 0.7354,
+      "step": 16378
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.842503900588746e-06,
+      "loss": 0.75,
+      "step": 16379
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.841379838621056e-06,
+      "loss": 0.7207,
+      "step": 16380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.840255805260146e-06,
+      "loss": 0.7686,
+      "step": 16381
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.839131800520903e-06,
+      "loss": 0.7285,
+      "step": 16382
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.838007824418226e-06,
+      "loss": 0.7383,
+      "step": 16383
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.836883876967012e-06,
+      "loss": 0.709,
+      "step": 16384
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.835759958182154e-06,
+      "loss": 0.7236,
+      "step": 16385
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.834636068078548e-06,
+      "loss": 0.7402,
+      "step": 16386
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.833512206671092e-06,
+      "loss": 0.7441,
+      "step": 16387
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.832388373974674e-06,
+      "loss": 0.709,
+      "step": 16388
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.831264570004185e-06,
+      "loss": 0.7334,
+      "step": 16389
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.830140794774525e-06,
+      "loss": 0.7266,
+      "step": 16390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.829017048300584e-06,
+      "loss": 0.7217,
+      "step": 16391
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.827893330597253e-06,
+      "loss": 0.7373,
+      "step": 16392
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.826769641679428e-06,
+      "loss": 0.7344,
+      "step": 16393
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.825645981561996e-06,
+      "loss": 0.7236,
+      "step": 16394
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82452235025985e-06,
+      "loss": 0.7412,
+      "step": 16395
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.823398747787879e-06,
+      "loss": 0.6953,
+      "step": 16396
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.822275174160975e-06,
+      "loss": 0.7334,
+      "step": 16397
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.821151629394027e-06,
+      "loss": 0.7148,
+      "step": 16398
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.820028113501925e-06,
+      "loss": 0.7178,
+      "step": 16399
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.818904626499563e-06,
+      "loss": 0.7188,
+      "step": 16400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.817781168401822e-06,
+      "loss": 0.7178,
+      "step": 16401
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.816657739223593e-06,
+      "loss": 0.7246,
+      "step": 16402
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.815534338979766e-06,
+      "loss": 0.7236,
+      "step": 16403
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.814410967685227e-06,
+      "loss": 0.7559,
+      "step": 16404
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.813287625354863e-06,
+      "loss": 0.71,
+      "step": 16405
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.812164312003562e-06,
+      "loss": 0.7471,
+      "step": 16406
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.811041027646214e-06,
+      "loss": 0.7188,
+      "step": 16407
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.8099177722977e-06,
+      "loss": 0.7295,
+      "step": 16408
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.808794545972907e-06,
+      "loss": 0.7139,
+      "step": 16409
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.807671348686719e-06,
+      "loss": 0.7334,
+      "step": 16410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.806548180454024e-06,
+      "loss": 0.7119,
+      "step": 16411
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.805425041289703e-06,
+      "loss": 0.7402,
+      "step": 16412
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.804301931208647e-06,
+      "loss": 0.709,
+      "step": 16413
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.803178850225733e-06,
+      "loss": 0.7109,
+      "step": 16414
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.802055798355847e-06,
+      "loss": 0.7334,
+      "step": 16415
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.80093277561387e-06,
+      "loss": 0.707,
+      "step": 16416
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.799809782014687e-06,
+      "loss": 0.7051,
+      "step": 16417
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.798686817573179e-06,
+      "loss": 0.7207,
+      "step": 16418
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.797563882304231e-06,
+      "loss": 0.708,
+      "step": 16419
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.796440976222724e-06,
+      "loss": 0.7354,
+      "step": 16420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.795318099343536e-06,
+      "loss": 0.7266,
+      "step": 16421
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.794195251681546e-06,
+      "loss": 0.71,
+      "step": 16422
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.79307243325164e-06,
+      "loss": 0.749,
+      "step": 16423
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.791949644068694e-06,
+      "loss": 0.708,
+      "step": 16424
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.79082688414759e-06,
+      "loss": 0.708,
+      "step": 16425
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.789704153503209e-06,
+      "loss": 0.7246,
+      "step": 16426
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.788581452150426e-06,
+      "loss": 0.6924,
+      "step": 16427
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.787458780104115e-06,
+      "loss": 0.7314,
+      "step": 16428
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.786336137379165e-06,
+      "loss": 0.7109,
+      "step": 16429
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.785213523990445e-06,
+      "loss": 0.7334,
+      "step": 16430
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.784090939952838e-06,
+      "loss": 0.7256,
+      "step": 16431
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.782968385281217e-06,
+      "loss": 0.7188,
+      "step": 16432
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.781845859990463e-06,
+      "loss": 0.7393,
+      "step": 16433
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.780723364095445e-06,
+      "loss": 0.7188,
+      "step": 16434
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.779600897611044e-06,
+      "loss": 0.7461,
+      "step": 16435
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.778478460552132e-06,
+      "loss": 0.7207,
+      "step": 16436
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.777356052933588e-06,
+      "loss": 0.7002,
+      "step": 16437
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.776233674770283e-06,
+      "loss": 0.7354,
+      "step": 16438
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.775111326077096e-06,
+      "loss": 0.7197,
+      "step": 16439
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.773989006868894e-06,
+      "loss": 0.751,
+      "step": 16440
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.772866717160555e-06,
+      "loss": 0.7295,
+      "step": 16441
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.771744456966949e-06,
+      "loss": 0.7285,
+      "step": 16442
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.770622226302948e-06,
+      "loss": 0.7129,
+      "step": 16443
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76950002518343e-06,
+      "loss": 0.7344,
+      "step": 16444
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76837785362326e-06,
+      "loss": 0.7188,
+      "step": 16445
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.767255711637319e-06,
+      "loss": 0.7197,
+      "step": 16446
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.766133599240466e-06,
+      "loss": 0.7207,
+      "step": 16447
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76501151644758e-06,
+      "loss": 0.7217,
+      "step": 16448
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.763889463273524e-06,
+      "loss": 0.7207,
+      "step": 16449
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.762767439733176e-06,
+      "loss": 0.7402,
+      "step": 16450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.7616454458414e-06,
+      "loss": 0.7256,
+      "step": 16451
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.760523481613068e-06,
+      "loss": 0.7217,
+      "step": 16452
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.759401547063048e-06,
+      "loss": 0.707,
+      "step": 16453
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.758279642206207e-06,
+      "loss": 0.708,
+      "step": 16454
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.757157767057411e-06,
+      "loss": 0.7217,
+      "step": 16455
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.756035921631531e-06,
+      "loss": 0.7656,
+      "step": 16456
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.754914105943433e-06,
+      "loss": 0.7168,
+      "step": 16457
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.753792320007985e-06,
+      "loss": 0.7334,
+      "step": 16458
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.752670563840054e-06,
+      "loss": 0.7393,
+      "step": 16459
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.7515488374545e-06,
+      "loss": 0.7314,
+      "step": 16460
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.750427140866194e-06,
+      "loss": 0.7334,
+      "step": 16461
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74930547409e-06,
+      "loss": 0.7295,
+      "step": 16462
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74818383714078e-06,
+      "loss": 0.7354,
+      "step": 16463
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.747062230033401e-06,
+      "loss": 0.7031,
+      "step": 16464
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.745940652782732e-06,
+      "loss": 0.7236,
+      "step": 16465
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.744819105403626e-06,
+      "loss": 0.7158,
+      "step": 16466
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.743697587910949e-06,
+      "loss": 0.6934,
+      "step": 16467
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74257610031957e-06,
+      "loss": 0.7451,
+      "step": 16468
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.741454642644344e-06,
+      "loss": 0.7188,
+      "step": 16469
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74033321490014e-06,
+      "loss": 0.7383,
+      "step": 16470
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.739211817101812e-06,
+      "loss": 0.7139,
+      "step": 16471
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.73809044926423e-06,
+      "loss": 0.7383,
+      "step": 16472
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.736969111402243e-06,
+      "loss": 0.7197,
+      "step": 16473
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.735847803530722e-06,
+      "loss": 0.7236,
+      "step": 16474
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.73472652566452e-06,
+      "loss": 0.7256,
+      "step": 16475
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.733605277818503e-06,
+      "loss": 0.7334,
+      "step": 16476
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.732484060007526e-06,
+      "loss": 0.7334,
+      "step": 16477
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.73136287224645e-06,
+      "loss": 0.709,
+      "step": 16478
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.730241714550126e-06,
+      "loss": 0.7432,
+      "step": 16479
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.729120586933422e-06,
+      "loss": 0.7305,
+      "step": 16480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72799948941119e-06,
+      "loss": 0.7451,
+      "step": 16481
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.726878421998288e-06,
+      "loss": 0.7412,
+      "step": 16482
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.725757384709573e-06,
+      "loss": 0.7314,
+      "step": 16483
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.724636377559902e-06,
+      "loss": 0.7402,
+      "step": 16484
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.723515400564133e-06,
+      "loss": 0.7236,
+      "step": 16485
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72239445373712e-06,
+      "loss": 0.7178,
+      "step": 16486
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.721273537093713e-06,
+      "loss": 0.7432,
+      "step": 16487
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.720152650648772e-06,
+      "loss": 0.7266,
+      "step": 16488
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.719031794417151e-06,
+      "loss": 0.7109,
+      "step": 16489
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.7179109684137e-06,
+      "loss": 0.7373,
+      "step": 16490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.716790172653283e-06,
+      "loss": 0.7139,
+      "step": 16491
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.715669407150743e-06,
+      "loss": 0.749,
+      "step": 16492
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.714548671920935e-06,
+      "loss": 0.7334,
+      "step": 16493
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.713427966978713e-06,
+      "loss": 0.7188,
+      "step": 16494
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.712307292338929e-06,
+      "loss": 0.7285,
+      "step": 16495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71118664801643e-06,
+      "loss": 0.7119,
+      "step": 16496
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.710066034026075e-06,
+      "loss": 0.7432,
+      "step": 16497
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.708945450382712e-06,
+      "loss": 0.7305,
+      "step": 16498
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.707824897101188e-06,
+      "loss": 0.7305,
+      "step": 16499
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.706704374196353e-06,
+      "loss": 0.7471,
+      "step": 16500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.705583881683062e-06,
+      "loss": 0.7109,
+      "step": 16501
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.704463419576159e-06,
+      "loss": 0.7109,
+      "step": 16502
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.703342987890494e-06,
+      "loss": 0.7178,
+      "step": 16503
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.70222258664092e-06,
+      "loss": 0.7227,
+      "step": 16504
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.701102215842278e-06,
+      "loss": 0.7188,
+      "step": 16505
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.699981875509416e-06,
+      "loss": 0.7246,
+      "step": 16506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.698861565657186e-06,
+      "loss": 0.7334,
+      "step": 16507
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.697741286300429e-06,
+      "loss": 0.708,
+      "step": 16508
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.696621037453999e-06,
+      "loss": 0.7051,
+      "step": 16509
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.695500819132737e-06,
+      "loss": 0.7168,
+      "step": 16510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.694380631351488e-06,
+      "loss": 0.7334,
+      "step": 16511
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.693260474125097e-06,
+      "loss": 0.6992,
+      "step": 16512
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.69214034746841e-06,
+      "loss": 0.7139,
+      "step": 16513
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.691020251396271e-06,
+      "loss": 0.7041,
+      "step": 16514
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.689900185923526e-06,
+      "loss": 0.7344,
+      "step": 16515
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.688780151065015e-06,
+      "loss": 0.71,
+      "step": 16516
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.687660146835585e-06,
+      "loss": 0.7178,
+      "step": 16517
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.686540173250074e-06,
+      "loss": 0.7002,
+      "step": 16518
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.685420230323326e-06,
+      "loss": 0.7344,
+      "step": 16519
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.684300318070183e-06,
+      "loss": 0.7422,
+      "step": 16520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.683180436505488e-06,
+      "loss": 0.6992,
+      "step": 16521
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.682060585644081e-06,
+      "loss": 0.7256,
+      "step": 16522
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.680940765500802e-06,
+      "loss": 0.7207,
+      "step": 16523
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.679820976090494e-06,
+      "loss": 0.6992,
+      "step": 16524
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.678701217427993e-06,
+      "loss": 0.7139,
+      "step": 16525
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67758148952814e-06,
+      "loss": 0.707,
+      "step": 16526
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.676461792405773e-06,
+      "loss": 0.7451,
+      "step": 16527
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.675342126075734e-06,
+      "loss": 0.7441,
+      "step": 16528
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.674222490552855e-06,
+      "loss": 0.7256,
+      "step": 16529
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.673102885851984e-06,
+      "loss": 0.6992,
+      "step": 16530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.671983311987948e-06,
+      "loss": 0.7412,
+      "step": 16531
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.67086376897559e-06,
+      "loss": 0.7256,
+      "step": 16532
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.669744256829742e-06,
+      "loss": 0.6895,
+      "step": 16533
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.668624775565242e-06,
+      "loss": 0.7207,
+      "step": 16534
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.667505325196927e-06,
+      "loss": 0.7314,
+      "step": 16535
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.666385905739634e-06,
+      "loss": 0.7021,
+      "step": 16536
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.665266517208199e-06,
+      "loss": 0.6904,
+      "step": 16537
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.66414715961745e-06,
+      "loss": 0.7061,
+      "step": 16538
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.663027832982222e-06,
+      "loss": 0.7344,
+      "step": 16539
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.661908537317355e-06,
+      "loss": 0.7598,
+      "step": 16540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.660789272637674e-06,
+      "loss": 0.7178,
+      "step": 16541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.659670038958018e-06,
+      "loss": 0.7354,
+      "step": 16542
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.658550836293223e-06,
+      "loss": 0.7461,
+      "step": 16543
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.657431664658111e-06,
+      "loss": 0.7256,
+      "step": 16544
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.656312524067517e-06,
+      "loss": 0.7197,
+      "step": 16545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.655193414536278e-06,
+      "loss": 0.7285,
+      "step": 16546
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.654074336079216e-06,
+      "loss": 0.7529,
+      "step": 16547
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.652955288711168e-06,
+      "loss": 0.7158,
+      "step": 16548
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.651836272446964e-06,
+      "loss": 0.7178,
+      "step": 16549
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.650717287301432e-06,
+      "loss": 0.7471,
+      "step": 16550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.649598333289396e-06,
+      "loss": 0.7363,
+      "step": 16551
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.648479410425692e-06,
+      "loss": 0.708,
+      "step": 16552
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.647360518725145e-06,
+      "loss": 0.75,
+      "step": 16553
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.646241658202585e-06,
+      "loss": 0.7637,
+      "step": 16554
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.645122828872838e-06,
+      "loss": 0.7334,
+      "step": 16555
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.644004030750734e-06,
+      "loss": 0.752,
+      "step": 16556
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.642885263851092e-06,
+      "loss": 0.7441,
+      "step": 16557
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.641766528188747e-06,
+      "loss": 0.7344,
+      "step": 16558
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.640647823778518e-06,
+      "loss": 0.7207,
+      "step": 16559
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.639529150635236e-06,
+      "loss": 0.7295,
+      "step": 16560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.638410508773723e-06,
+      "loss": 0.752,
+      "step": 16561
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.637291898208804e-06,
+      "loss": 0.7109,
+      "step": 16562
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.636173318955307e-06,
+      "loss": 0.7051,
+      "step": 16563
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.635054771028049e-06,
+      "loss": 0.7178,
+      "step": 16564
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.633936254441857e-06,
+      "loss": 0.7354,
+      "step": 16565
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.632817769211555e-06,
+      "loss": 0.7422,
+      "step": 16566
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.631699315351963e-06,
+      "loss": 0.7285,
+      "step": 16567
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.630580892877904e-06,
+      "loss": 0.7217,
+      "step": 16568
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.629462501804203e-06,
+      "loss": 0.709,
+      "step": 16569
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.628344142145678e-06,
+      "loss": 0.7236,
+      "step": 16570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6272258139171494e-06,
+      "loss": 0.71,
+      "step": 16571
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.626107517133436e-06,
+      "loss": 0.7266,
+      "step": 16572
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.624989251809363e-06,
+      "loss": 0.7275,
+      "step": 16573
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.623871017959745e-06,
+      "loss": 0.7314,
+      "step": 16574
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.622752815599406e-06,
+      "loss": 0.7412,
+      "step": 16575
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6216346447431654e-06,
+      "loss": 0.7168,
+      "step": 16576
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.620516505405835e-06,
+      "loss": 0.7139,
+      "step": 16577
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.619398397602236e-06,
+      "loss": 0.7363,
+      "step": 16578
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.618280321347188e-06,
+      "loss": 0.7021,
+      "step": 16579
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617162276655503e-06,
+      "loss": 0.7129,
+      "step": 16580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.616044263542005e-06,
+      "loss": 0.7285,
+      "step": 16581
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.614926282021509e-06,
+      "loss": 0.7227,
+      "step": 16582
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.613808332108825e-06,
+      "loss": 0.7324,
+      "step": 16583
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.61269041381877e-06,
+      "loss": 0.7227,
+      "step": 16584
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.611572527166163e-06,
+      "loss": 0.7314,
+      "step": 16585
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6104546721658145e-06,
+      "loss": 0.7344,
+      "step": 16586
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.609336848832543e-06,
+      "loss": 0.7305,
+      "step": 16587
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.608219057181162e-06,
+      "loss": 0.7324,
+      "step": 16588
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.607101297226481e-06,
+      "loss": 0.748,
+      "step": 16589
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.605983568983313e-06,
+      "loss": 0.7441,
+      "step": 16590
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6048658724664734e-06,
+      "loss": 0.7168,
+      "step": 16591
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.603748207690771e-06,
+      "loss": 0.7119,
+      "step": 16592
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.602630574671023e-06,
+      "loss": 0.708,
+      "step": 16593
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.601512973422036e-06,
+      "loss": 0.7217,
+      "step": 16594
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.600395403958626e-06,
+      "loss": 0.7305,
+      "step": 16595
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5992778662955935e-06,
+      "loss": 0.7207,
+      "step": 16596
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.598160360447758e-06,
+      "loss": 0.708,
+      "step": 16597
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.597042886429923e-06,
+      "loss": 0.7334,
+      "step": 16598
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.595925444256902e-06,
+      "loss": 0.7334,
+      "step": 16599
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5948080339435036e-06,
+      "loss": 0.7275,
+      "step": 16600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5936906555045355e-06,
+      "loss": 0.7354,
+      "step": 16601
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5925733089548e-06,
+      "loss": 0.7256,
+      "step": 16602
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5914559943091114e-06,
+      "loss": 0.7188,
+      "step": 16603
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.590338711582274e-06,
+      "loss": 0.7246,
+      "step": 16604
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5892214607890955e-06,
+      "loss": 0.7461,
+      "step": 16605
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5881042419443814e-06,
+      "loss": 0.7305,
+      "step": 16606
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.586987055062935e-06,
+      "loss": 0.7031,
+      "step": 16607
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.58586990015957e-06,
+      "loss": 0.7412,
+      "step": 16608
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.584752777249083e-06,
+      "loss": 0.7227,
+      "step": 16609
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.583635686346281e-06,
+      "loss": 0.7266,
+      "step": 16610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.582518627465965e-06,
+      "loss": 0.7295,
+      "step": 16611
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.581401600622945e-06,
+      "loss": 0.7227,
+      "step": 16612
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.580284605832019e-06,
+      "loss": 0.7148,
+      "step": 16613
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.579167643107995e-06,
+      "loss": 0.7354,
+      "step": 16614
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.578050712465671e-06,
+      "loss": 0.7139,
+      "step": 16615
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57693381391985e-06,
+      "loss": 0.7295,
+      "step": 16616
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.575816947485331e-06,
+      "loss": 0.7051,
+      "step": 16617
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.57470011317692e-06,
+      "loss": 0.751,
+      "step": 16618
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.573583311009414e-06,
+      "loss": 0.7383,
+      "step": 16619
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.572466540997616e-06,
+      "loss": 0.7197,
+      "step": 16620
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.571349803156327e-06,
+      "loss": 0.7334,
+      "step": 16621
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5702330975003416e-06,
+      "loss": 0.7324,
+      "step": 16622
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.569116424044458e-06,
+      "loss": 0.7266,
+      "step": 16623
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5679997828034815e-06,
+      "loss": 0.7314,
+      "step": 16624
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.566883173792203e-06,
+      "loss": 0.709,
+      "step": 16625
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5657665970254266e-06,
+      "loss": 0.7305,
+      "step": 16626
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.564650052517949e-06,
+      "loss": 0.7363,
+      "step": 16627
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.563533540284562e-06,
+      "loss": 0.708,
+      "step": 16628
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562417060340061e-06,
+      "loss": 0.7188,
+      "step": 16629
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.56130061269925e-06,
+      "loss": 0.7236,
+      "step": 16630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.560184197376917e-06,
+      "loss": 0.7412,
+      "step": 16631
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.559067814387864e-06,
+      "loss": 0.7178,
+      "step": 16632
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55795146374688e-06,
+      "loss": 0.7598,
+      "step": 16633
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.556835145468764e-06,
+      "loss": 0.7344,
+      "step": 16634
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.555718859568304e-06,
+      "loss": 0.7402,
+      "step": 16635
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.554602606060297e-06,
+      "loss": 0.7344,
+      "step": 16636
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5534863849595345e-06,
+      "loss": 0.7334,
+      "step": 16637
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5523701962808115e-06,
+      "loss": 0.7148,
+      "step": 16638
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.551254040038918e-06,
+      "loss": 0.7178,
+      "step": 16639
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.55013791624865e-06,
+      "loss": 0.7109,
+      "step": 16640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.549021824924789e-06,
+      "loss": 0.7422,
+      "step": 16641
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.547905766082135e-06,
+      "loss": 0.7227,
+      "step": 16642
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.546789739735472e-06,
+      "loss": 0.7061,
+      "step": 16643
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.545673745899598e-06,
+      "loss": 0.7148,
+      "step": 16644
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.544557784589296e-06,
+      "loss": 0.7246,
+      "step": 16645
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.543441855819355e-06,
+      "loss": 0.7227,
+      "step": 16646
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.542325959604569e-06,
+      "loss": 0.7148,
+      "step": 16647
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.541210095959722e-06,
+      "loss": 0.7119,
+      "step": 16648
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5400942648996e-06,
+      "loss": 0.7393,
+      "step": 16649
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.538978466438996e-06,
+      "loss": 0.748,
+      "step": 16650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.537862700592692e-06,
+      "loss": 0.7109,
+      "step": 16651
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.536746967375475e-06,
+      "loss": 0.7275,
+      "step": 16652
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.535631266802137e-06,
+      "loss": 0.707,
+      "step": 16653
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.534515598887457e-06,
+      "loss": 0.7324,
+      "step": 16654
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.533399963646223e-06,
+      "loss": 0.7041,
+      "step": 16655
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.532284361093216e-06,
+      "loss": 0.7314,
+      "step": 16656
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.531168791243226e-06,
+      "loss": 0.7402,
+      "step": 16657
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5300532541110335e-06,
+      "loss": 0.7266,
+      "step": 16658
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5289377497114235e-06,
+      "loss": 0.6943,
+      "step": 16659
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.527822278059181e-06,
+      "loss": 0.7617,
+      "step": 16660
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.526706839169085e-06,
+      "loss": 0.7246,
+      "step": 16661
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5255914330559145e-06,
+      "loss": 0.7012,
+      "step": 16662
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.524476059734459e-06,
+      "loss": 0.7275,
+      "step": 16663
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5233607192194935e-06,
+      "loss": 0.7354,
+      "step": 16664
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5222454115258036e-06,
+      "loss": 0.6963,
+      "step": 16665
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.521130136668171e-06,
+      "loss": 0.709,
+      "step": 16666
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.520014894661369e-06,
+      "loss": 0.7227,
+      "step": 16667
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5188996855201804e-06,
+      "loss": 0.7236,
+      "step": 16668
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.517784509259386e-06,
+      "loss": 0.7598,
+      "step": 16669
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5166693658937605e-06,
+      "loss": 0.707,
+      "step": 16670
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5155542554380866e-06,
+      "loss": 0.7373,
+      "step": 16671
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5144391779071415e-06,
+      "loss": 0.7207,
+      "step": 16672
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.513324133315703e-06,
+      "loss": 0.7354,
+      "step": 16673
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.512209121678542e-06,
+      "loss": 0.7549,
+      "step": 16674
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.511094143010441e-06,
+      "loss": 0.7422,
+      "step": 16675
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.509979197326174e-06,
+      "loss": 0.7227,
+      "step": 16676
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.508864284640518e-06,
+      "loss": 0.7383,
+      "step": 16677
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.507749404968248e-06,
+      "loss": 0.7314,
+      "step": 16678
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.50663455832414e-06,
+      "loss": 0.7441,
+      "step": 16679
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.505519744722962e-06,
+      "loss": 0.71,
+      "step": 16680
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.504404964179495e-06,
+      "loss": 0.7314,
+      "step": 16681
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.503290216708507e-06,
+      "loss": 0.7256,
+      "step": 16682
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.502175502324776e-06,
+      "loss": 0.7168,
+      "step": 16683
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5010608210430735e-06,
+      "loss": 0.7236,
+      "step": 16684
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4999461728781675e-06,
+      "loss": 0.7451,
+      "step": 16685
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.498831557844839e-06,
+      "loss": 0.7441,
+      "step": 16686
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.497716975957848e-06,
+      "loss": 0.7402,
+      "step": 16687
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.496602427231969e-06,
+      "loss": 0.7188,
+      "step": 16688
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.495487911681978e-06,
+      "loss": 0.707,
+      "step": 16689
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.494373429322639e-06,
+      "loss": 0.709,
+      "step": 16690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.49325898016872e-06,
+      "loss": 0.7207,
+      "step": 16691
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.492144564235e-06,
+      "loss": 0.7129,
+      "step": 16692
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.491030181536236e-06,
+      "loss": 0.7305,
+      "step": 16693
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.489915832087202e-06,
+      "loss": 0.7314,
+      "step": 16694
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.488801515902664e-06,
+      "loss": 0.7266,
+      "step": 16695
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.487687232997391e-06,
+      "loss": 0.7344,
+      "step": 16696
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4865729833861475e-06,
+      "loss": 0.7207,
+      "step": 16697
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.485458767083703e-06,
+      "loss": 0.7266,
+      "step": 16698
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.484344584104824e-06,
+      "loss": 0.7256,
+      "step": 16699
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.483230434464272e-06,
+      "loss": 0.7227,
+      "step": 16700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.482116318176812e-06,
+      "loss": 0.752,
+      "step": 16701
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.481002235257213e-06,
+      "loss": 0.6992,
+      "step": 16702
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.479888185720233e-06,
+      "loss": 0.7119,
+      "step": 16703
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.478774169580643e-06,
+      "loss": 0.7334,
+      "step": 16704
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.477660186853205e-06,
+      "loss": 0.7275,
+      "step": 16705
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.476546237552677e-06,
+      "loss": 0.7256,
+      "step": 16706
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.475432321693823e-06,
+      "loss": 0.6982,
+      "step": 16707
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.474318439291408e-06,
+      "loss": 0.7129,
+      "step": 16708
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.47320459036019e-06,
+      "loss": 0.7686,
+      "step": 16709
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.472090774914934e-06,
+      "loss": 0.7275,
+      "step": 16710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.470976992970395e-06,
+      "loss": 0.7246,
+      "step": 16711
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.469863244541347e-06,
+      "loss": 0.7275,
+      "step": 16712
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.468749529642529e-06,
+      "loss": 0.7168,
+      "step": 16713
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4676358482887145e-06,
+      "loss": 0.7217,
+      "step": 16714
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.466522200494657e-06,
+      "loss": 0.7178,
+      "step": 16715
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.465408586275118e-06,
+      "loss": 0.7256,
+      "step": 16716
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.464295005644855e-06,
+      "loss": 0.7363,
+      "step": 16717
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4631814586186265e-06,
+      "loss": 0.7168,
+      "step": 16718
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.462067945211184e-06,
+      "loss": 0.7568,
+      "step": 16719
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4609544654372915e-06,
+      "loss": 0.7275,
+      "step": 16720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.459841019311698e-06,
+      "loss": 0.7148,
+      "step": 16721
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.458727606849166e-06,
+      "loss": 0.7158,
+      "step": 16722
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4576142280644456e-06,
+      "loss": 0.7119,
+      "step": 16723
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.456500882972293e-06,
+      "loss": 0.7412,
+      "step": 16724
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.455387571587469e-06,
+      "loss": 0.71,
+      "step": 16725
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.454274293924719e-06,
+      "loss": 0.7227,
+      "step": 16726
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.453161049998798e-06,
+      "loss": 0.7227,
+      "step": 16727
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4520478398244625e-06,
+      "loss": 0.7471,
+      "step": 16728
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4509346634164625e-06,
+      "loss": 0.7334,
+      "step": 16729
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.449821520789551e-06,
+      "loss": 0.7197,
+      "step": 16730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.448708411958483e-06,
+      "loss": 0.7354,
+      "step": 16731
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.447595336938005e-06,
+      "loss": 0.7188,
+      "step": 16732
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.446482295742867e-06,
+      "loss": 0.7217,
+      "step": 16733
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4453692883878245e-06,
+      "loss": 0.7363,
+      "step": 16734
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.444256314887623e-06,
+      "loss": 0.7109,
+      "step": 16735
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.443143375257014e-06,
+      "loss": 0.7119,
+      "step": 16736
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4420304695107485e-06,
+      "loss": 0.7217,
+      "step": 16737
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.440917597663574e-06,
+      "loss": 0.7275,
+      "step": 16738
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.439804759730236e-06,
+      "loss": 0.7441,
+      "step": 16739
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.438691955725482e-06,
+      "loss": 0.7383,
+      "step": 16740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.437579185664064e-06,
+      "loss": 0.7275,
+      "step": 16741
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.436466449560722e-06,
+      "loss": 0.7256,
+      "step": 16742
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.43535374743021e-06,
+      "loss": 0.709,
+      "step": 16743
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4342410792872724e-06,
+      "loss": 0.7471,
+      "step": 16744
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.433128445146649e-06,
+      "loss": 0.7217,
+      "step": 16745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4320158450230875e-06,
+      "loss": 0.7021,
+      "step": 16746
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.430903278931335e-06,
+      "loss": 0.7422,
+      "step": 16747
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.429790746886132e-06,
+      "loss": 0.7295,
+      "step": 16748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.428678248902226e-06,
+      "loss": 0.7217,
+      "step": 16749
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.427565784994355e-06,
+      "loss": 0.7061,
+      "step": 16750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.426453355177273e-06,
+      "loss": 0.7285,
+      "step": 16751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.425340959465708e-06,
+      "loss": 0.7256,
+      "step": 16752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.42422859787441e-06,
+      "loss": 0.708,
+      "step": 16753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.423116270418116e-06,
+      "loss": 0.7295,
+      "step": 16754
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.422003977111572e-06,
+      "loss": 0.7275,
+      "step": 16755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.420891717969513e-06,
+      "loss": 0.7148,
+      "step": 16756
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.419779493006689e-06,
+      "loss": 0.7217,
+      "step": 16757
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4186673022378265e-06,
+      "loss": 0.7393,
+      "step": 16758
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.417555145677673e-06,
+      "loss": 0.7432,
+      "step": 16759
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.416443023340963e-06,
+      "loss": 0.7217,
+      "step": 16760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.415330935242438e-06,
+      "loss": 0.7129,
+      "step": 16761
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.414218881396834e-06,
+      "loss": 0.71,
+      "step": 16762
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.413106861818887e-06,
+      "loss": 0.7188,
+      "step": 16763
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.411994876523341e-06,
+      "loss": 0.7598,
+      "step": 16764
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4108829255249235e-06,
+      "loss": 0.7295,
+      "step": 16765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4097710088383725e-06,
+      "loss": 0.7354,
+      "step": 16766
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.408659126478428e-06,
+      "loss": 0.7285,
+      "step": 16767
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.40754727845982e-06,
+      "loss": 0.7178,
+      "step": 16768
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.406435464797285e-06,
+      "loss": 0.7236,
+      "step": 16769
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4053236855055596e-06,
+      "loss": 0.7158,
+      "step": 16770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.404211940599375e-06,
+      "loss": 0.7236,
+      "step": 16771
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4031002300934604e-06,
+      "loss": 0.7344,
+      "step": 16772
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.401988554002557e-06,
+      "loss": 0.7529,
+      "step": 16773
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.400876912341391e-06,
+      "loss": 0.7109,
+      "step": 16774
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.399765305124696e-06,
+      "loss": 0.7588,
+      "step": 16775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.398653732367205e-06,
+      "loss": 0.7236,
+      "step": 16776
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.397542194083648e-06,
+      "loss": 0.7119,
+      "step": 16777
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.396430690288752e-06,
+      "loss": 0.7148,
+      "step": 16778
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3953192209972525e-06,
+      "loss": 0.7227,
+      "step": 16779
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3942077862238746e-06,
+      "loss": 0.7139,
+      "step": 16780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.39309638598335e-06,
+      "loss": 0.7051,
+      "step": 16781
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.391985020290406e-06,
+      "loss": 0.7285,
+      "step": 16782
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390873689159775e-06,
+      "loss": 0.7227,
+      "step": 16783
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.389762392606179e-06,
+      "loss": 0.7285,
+      "step": 16784
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.388651130644346e-06,
+      "loss": 0.7354,
+      "step": 16785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.387539903289006e-06,
+      "loss": 0.7002,
+      "step": 16786
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.386428710554881e-06,
+      "loss": 0.7363,
+      "step": 16787
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.385317552456704e-06,
+      "loss": 0.7402,
+      "step": 16788
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.384206429009191e-06,
+      "loss": 0.7197,
+      "step": 16789
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.38309534022708e-06,
+      "loss": 0.7012,
+      "step": 16790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.381984286125081e-06,
+      "loss": 0.7109,
+      "step": 16791
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3808732667179275e-06,
+      "loss": 0.7188,
+      "step": 16792
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3797622820203375e-06,
+      "loss": 0.7314,
+      "step": 16793
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3786513320470385e-06,
+      "loss": 0.7168,
+      "step": 16794
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3775404168127495e-06,
+      "loss": 0.709,
+      "step": 16795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.376429536332203e-06,
+      "loss": 0.6963,
+      "step": 16796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3753186906201045e-06,
+      "loss": 0.7236,
+      "step": 16797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3742078796911865e-06,
+      "loss": 0.7354,
+      "step": 16798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.373097103560164e-06,
+      "loss": 0.7373,
+      "step": 16799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.371986362241763e-06,
+      "loss": 0.7344,
+      "step": 16800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3708756557507e-06,
+      "loss": 0.75,
+      "step": 16801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3697649841016975e-06,
+      "loss": 0.7432,
+      "step": 16802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.368654347309468e-06,
+      "loss": 0.7061,
+      "step": 16803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.367543745388735e-06,
+      "loss": 0.7305,
+      "step": 16804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.366433178354214e-06,
+      "loss": 0.7227,
+      "step": 16805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.365322646220627e-06,
+      "loss": 0.6934,
+      "step": 16806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.364212149002688e-06,
+      "loss": 0.7324,
+      "step": 16807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.363101686715113e-06,
+      "loss": 0.709,
+      "step": 16808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.361991259372623e-06,
+      "loss": 0.7354,
+      "step": 16809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.360880866989928e-06,
+      "loss": 0.7168,
+      "step": 16810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.359770509581743e-06,
+      "loss": 0.752,
+      "step": 16811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.358660187162787e-06,
+      "loss": 0.7314,
+      "step": 16812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3575498997477735e-06,
+      "loss": 0.71,
+      "step": 16813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3564396473514136e-06,
+      "loss": 0.6973,
+      "step": 16814
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.355329429988427e-06,
+      "loss": 0.7256,
+      "step": 16815
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.35421924767352e-06,
+      "loss": 0.7158,
+      "step": 16816
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.353109100421405e-06,
+      "loss": 0.7314,
+      "step": 16817
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.351998988246798e-06,
+      "loss": 0.7207,
+      "step": 16818
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.35088891116441e-06,
+      "loss": 0.7412,
+      "step": 16819
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.349778869188948e-06,
+      "loss": 0.708,
+      "step": 16820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.348668862335131e-06,
+      "loss": 0.7451,
+      "step": 16821
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3475588906176635e-06,
+      "loss": 0.7188,
+      "step": 16822
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.346448954051255e-06,
+      "loss": 0.7148,
+      "step": 16823
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.345339052650613e-06,
+      "loss": 0.7588,
+      "step": 16824
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.344229186430451e-06,
+      "loss": 0.7441,
+      "step": 16825
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.343119355405472e-06,
+      "loss": 0.7334,
+      "step": 16826
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.34200955959039e-06,
+      "loss": 0.748,
+      "step": 16827
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.340899798999913e-06,
+      "loss": 0.7129,
+      "step": 16828
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.33979007364874e-06,
+      "loss": 0.7393,
+      "step": 16829
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.338680383551582e-06,
+      "loss": 0.7285,
+      "step": 16830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3375707287231455e-06,
+      "loss": 0.7256,
+      "step": 16831
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.336461109178135e-06,
+      "loss": 0.7324,
+      "step": 16832
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.335351524931256e-06,
+      "loss": 0.7266,
+      "step": 16833
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.334241975997211e-06,
+      "loss": 0.7188,
+      "step": 16834
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.333132462390713e-06,
+      "loss": 0.709,
+      "step": 16835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.332022984126453e-06,
+      "loss": 0.7451,
+      "step": 16836
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.330913541219142e-06,
+      "loss": 0.752,
+      "step": 16837
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.329804133683478e-06,
+      "loss": 0.6982,
+      "step": 16838
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.328694761534168e-06,
+      "loss": 0.6982,
+      "step": 16839
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.32758542478591e-06,
+      "loss": 0.7129,
+      "step": 16840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.326476123453414e-06,
+      "loss": 0.7197,
+      "step": 16841
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.325366857551367e-06,
+      "loss": 0.708,
+      "step": 16842
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.324257627094477e-06,
+      "loss": 0.7412,
+      "step": 16843
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.323148432097444e-06,
+      "loss": 0.7188,
+      "step": 16844
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3220392725749665e-06,
+      "loss": 0.7236,
+      "step": 16845
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.320930148541745e-06,
+      "loss": 0.7451,
+      "step": 16846
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.319821060012473e-06,
+      "loss": 0.7227,
+      "step": 16847
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.318712007001857e-06,
+      "loss": 0.7383,
+      "step": 16848
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3176029895245884e-06,
+      "loss": 0.7246,
+      "step": 16849
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.316494007595362e-06,
+      "loss": 0.7295,
+      "step": 16850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.31538506122888e-06,
+      "loss": 0.7256,
+      "step": 16851
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.314276150439836e-06,
+      "loss": 0.7188,
+      "step": 16852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.313167275242926e-06,
+      "loss": 0.7236,
+      "step": 16853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.312058435652849e-06,
+      "loss": 0.7168,
+      "step": 16854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.310949631684292e-06,
+      "loss": 0.7109,
+      "step": 16855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.309840863351954e-06,
+      "loss": 0.7393,
+      "step": 16856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3087321306705285e-06,
+      "loss": 0.7246,
+      "step": 16857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.307623433654709e-06,
+      "loss": 0.7119,
+      "step": 16858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.306514772319184e-06,
+      "loss": 0.7373,
+      "step": 16859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.305406146678653e-06,
+      "loss": 0.7607,
+      "step": 16860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.304297556747808e-06,
+      "loss": 0.71,
+      "step": 16861
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.303189002541331e-06,
+      "loss": 0.7422,
+      "step": 16862
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.302080484073921e-06,
+      "loss": 0.7109,
+      "step": 16863
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.300972001360266e-06,
+      "loss": 0.7373,
+      "step": 16864
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2998635544150545e-06,
+      "loss": 0.7119,
+      "step": 16865
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.29875514325298e-06,
+      "loss": 0.7646,
+      "step": 16866
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.297646767888731e-06,
+      "loss": 0.708,
+      "step": 16867
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.296538428336993e-06,
+      "loss": 0.7197,
+      "step": 16868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.295430124612452e-06,
+      "loss": 0.7246,
+      "step": 16869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.294321856729804e-06,
+      "loss": 0.7285,
+      "step": 16870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.293213624703728e-06,
+      "loss": 0.7285,
+      "step": 16871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2921054285489144e-06,
+      "loss": 0.7129,
+      "step": 16872
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2909972682800486e-06,
+      "loss": 0.7275,
+      "step": 16873
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.289889143911821e-06,
+      "loss": 0.7168,
+      "step": 16874
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.288781055458909e-06,
+      "loss": 0.7207,
+      "step": 16875
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.287673002936001e-06,
+      "loss": 0.7129,
+      "step": 16876
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2865649863577804e-06,
+      "loss": 0.7275,
+      "step": 16877
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.285457005738932e-06,
+      "loss": 0.7002,
+      "step": 16878
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.284349061094138e-06,
+      "loss": 0.7363,
+      "step": 16879
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.283241152438089e-06,
+      "loss": 0.7275,
+      "step": 16880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.282133279785453e-06,
+      "loss": 0.752,
+      "step": 16881
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.281025443150922e-06,
+      "loss": 0.7148,
+      "step": 16882
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.279917642549174e-06,
+      "loss": 0.7344,
+      "step": 16883
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2788098779948925e-06,
+      "loss": 0.7148,
+      "step": 16884
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2777021495027546e-06,
+      "loss": 0.7021,
+      "step": 16885
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.276594457087441e-06,
+      "loss": 0.7119,
+      "step": 16886
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.275486800763637e-06,
+      "loss": 0.7363,
+      "step": 16887
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.274379180546014e-06,
+      "loss": 0.7227,
+      "step": 16888
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.273271596449252e-06,
+      "loss": 0.7041,
+      "step": 16889
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.272164048488033e-06,
+      "loss": 0.7148,
+      "step": 16890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.271056536677031e-06,
+      "loss": 0.7041,
+      "step": 16891
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.269949061030923e-06,
+      "loss": 0.7305,
+      "step": 16892
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.268841621564391e-06,
+      "loss": 0.7324,
+      "step": 16893
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.267734218292105e-06,
+      "loss": 0.7266,
+      "step": 16894
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2666268512287405e-06,
+      "loss": 0.7129,
+      "step": 16895
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.265519520388977e-06,
+      "loss": 0.7344,
+      "step": 16896
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2644122257874874e-06,
+      "loss": 0.7119,
+      "step": 16897
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.263304967438945e-06,
+      "loss": 0.7148,
+      "step": 16898
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.262197745358024e-06,
+      "loss": 0.6982,
+      "step": 16899
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.261090559559402e-06,
+      "loss": 0.6973,
+      "step": 16900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.259983410057743e-06,
+      "loss": 0.7129,
+      "step": 16901
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.258876296867725e-06,
+      "loss": 0.7227,
+      "step": 16902
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.257769220004021e-06,
+      "loss": 0.7363,
+      "step": 16903
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.256662179481297e-06,
+      "loss": 0.7393,
+      "step": 16904
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.255555175314229e-06,
+      "loss": 0.6963,
+      "step": 16905
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2544482075174884e-06,
+      "loss": 0.7148,
+      "step": 16906
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.253341276105741e-06,
+      "loss": 0.7158,
+      "step": 16907
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.252234381093655e-06,
+      "loss": 0.7012,
+      "step": 16908
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2511275224959035e-06,
+      "loss": 0.7012,
+      "step": 16909
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.250020700327151e-06,
+      "loss": 0.7383,
+      "step": 16910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2489139146020715e-06,
+      "loss": 0.707,
+      "step": 16911
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.247807165335326e-06,
+      "loss": 0.7129,
+      "step": 16912
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.246700452541591e-06,
+      "loss": 0.7129,
+      "step": 16913
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.24559377623552e-06,
+      "loss": 0.7061,
+      "step": 16914
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.244487136431786e-06,
+      "loss": 0.7178,
+      "step": 16915
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.243380533145055e-06,
+      "loss": 0.6914,
+      "step": 16916
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.242273966389991e-06,
+      "loss": 0.7227,
+      "step": 16917
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.241167436181258e-06,
+      "loss": 0.7061,
+      "step": 16918
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.240060942533527e-06,
+      "loss": 0.7461,
+      "step": 16919
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.238954485461448e-06,
+      "loss": 0.6875,
+      "step": 16920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.237848064979695e-06,
+      "loss": 0.7266,
+      "step": 16921
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.236741681102923e-06,
+      "loss": 0.7041,
+      "step": 16922
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.235635333845803e-06,
+      "loss": 0.708,
+      "step": 16923
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.234529023222991e-06,
+      "loss": 0.7227,
+      "step": 16924
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.233422749249149e-06,
+      "loss": 0.7168,
+      "step": 16925
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.232316511938942e-06,
+      "loss": 0.7051,
+      "step": 16926
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.231210311307024e-06,
+      "loss": 0.707,
+      "step": 16927
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.230104147368053e-06,
+      "loss": 0.708,
+      "step": 16928
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.228998020136697e-06,
+      "loss": 0.7598,
+      "step": 16929
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.22789192962761e-06,
+      "loss": 0.7188,
+      "step": 16930
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.226785875855446e-06,
+      "loss": 0.6943,
+      "step": 16931
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.225679858834874e-06,
+      "loss": 0.7158,
+      "step": 16932
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.224573878580542e-06,
+      "loss": 0.7305,
+      "step": 16933
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.223467935107107e-06,
+      "loss": 0.7324,
+      "step": 16934
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.222362028429231e-06,
+      "loss": 0.7246,
+      "step": 16935
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.221256158561564e-06,
+      "loss": 0.7412,
+      "step": 16936
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2201503255187645e-06,
+      "loss": 0.7207,
+      "step": 16937
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.219044529315487e-06,
+      "loss": 0.7422,
+      "step": 16938
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.217938769966389e-06,
+      "loss": 0.7041,
+      "step": 16939
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.216833047486116e-06,
+      "loss": 0.7275,
+      "step": 16940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2157273618893285e-06,
+      "loss": 0.7441,
+      "step": 16941
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.21462171319068e-06,
+      "loss": 0.71,
+      "step": 16942
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.213516101404816e-06,
+      "loss": 0.7334,
+      "step": 16943
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.212410526546395e-06,
+      "loss": 0.707,
+      "step": 16944
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.211304988630069e-06,
+      "loss": 0.7139,
+      "step": 16945
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.210199487670482e-06,
+      "loss": 0.7295,
+      "step": 16946
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.209094023682291e-06,
+      "loss": 0.7227,
+      "step": 16947
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2079885966801446e-06,
+      "loss": 0.7266,
+      "step": 16948
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2068832066786875e-06,
+      "loss": 0.7363,
+      "step": 16949
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.205777853692576e-06,
+      "loss": 0.7178,
+      "step": 16950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.204672537736452e-06,
+      "loss": 0.707,
+      "step": 16951
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.203567258824975e-06,
+      "loss": 0.7129,
+      "step": 16952
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.202462016972778e-06,
+      "loss": 0.7129,
+      "step": 16953
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.201356812194516e-06,
+      "loss": 0.6982,
+      "step": 16954
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.20025164450483e-06,
+      "loss": 0.7109,
+      "step": 16955
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.199146513918373e-06,
+      "loss": 0.6982,
+      "step": 16956
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.198041420449786e-06,
+      "loss": 0.7012,
+      "step": 16957
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1969363641137225e-06,
+      "loss": 0.7236,
+      "step": 16958
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1958313449248126e-06,
+      "loss": 0.7178,
+      "step": 16959
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.194726362897712e-06,
+      "loss": 0.6953,
+      "step": 16960
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1936214180470565e-06,
+      "loss": 0.71,
+      "step": 16961
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192516510387497e-06,
+      "loss": 0.7021,
+      "step": 16962
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.191411639933669e-06,
+      "loss": 0.7256,
+      "step": 16963
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.190306806700219e-06,
+      "loss": 0.7402,
+      "step": 16964
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1892020107017905e-06,
+      "loss": 0.7217,
+      "step": 16965
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.188097251953019e-06,
+      "loss": 0.6973,
+      "step": 16966
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.186992530468546e-06,
+      "loss": 0.7188,
+      "step": 16967
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.185887846263016e-06,
+      "loss": 0.709,
+      "step": 16968
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.184783199351062e-06,
+      "loss": 0.7246,
+      "step": 16969
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.183678589747332e-06,
+      "loss": 0.7412,
+      "step": 16970
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.182574017466459e-06,
+      "loss": 0.7266,
+      "step": 16971
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1814694825230825e-06,
+      "loss": 0.6934,
+      "step": 16972
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.180364984931836e-06,
+      "loss": 0.709,
+      "step": 16973
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.179260524707362e-06,
+      "loss": 0.7178,
+      "step": 16974
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.178156101864296e-06,
+      "loss": 0.7256,
+      "step": 16975
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.177051716417271e-06,
+      "loss": 0.7275,
+      "step": 16976
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.17594736838093e-06,
+      "loss": 0.7178,
+      "step": 16977
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.174843057769905e-06,
+      "loss": 0.7275,
+      "step": 16978
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.173738784598822e-06,
+      "loss": 0.7314,
+      "step": 16979
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.172634548882328e-06,
+      "loss": 0.7256,
+      "step": 16980
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1715303506350495e-06,
+      "loss": 0.7109,
+      "step": 16981
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.17042618987162e-06,
+      "loss": 0.7158,
+      "step": 16982
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.169322066606677e-06,
+      "loss": 0.6982,
+      "step": 16983
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.16821798085485e-06,
+      "loss": 0.7295,
+      "step": 16984
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.167113932630765e-06,
+      "loss": 0.7275,
+      "step": 16985
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.166009921949062e-06,
+      "loss": 0.7227,
+      "step": 16986
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.164905948824368e-06,
+      "loss": 0.7119,
+      "step": 16987
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.16380201327131e-06,
+      "loss": 0.7168,
+      "step": 16988
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.162698115304525e-06,
+      "loss": 0.7207,
+      "step": 16989
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.161594254938635e-06,
+      "loss": 0.6807,
+      "step": 16990
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.160490432188276e-06,
+      "loss": 0.7275,
+      "step": 16991
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.159386647068071e-06,
+      "loss": 0.7119,
+      "step": 16992
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.158282899592648e-06,
+      "loss": 0.707,
+      "step": 16993
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.157179189776633e-06,
+      "loss": 0.7227,
+      "step": 16994
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.156075517634657e-06,
+      "loss": 0.7285,
+      "step": 16995
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.154971883181343e-06,
+      "loss": 0.7188,
+      "step": 16996
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.153868286431325e-06,
+      "loss": 0.7031,
+      "step": 16997
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1527647273992126e-06,
+      "loss": 0.7324,
+      "step": 16998
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.151661206099644e-06,
+      "loss": 0.7168,
+      "step": 16999
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1505577225472345e-06,
+      "loss": 0.7012,
+      "step": 17000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.149454276756614e-06,
+      "loss": 0.6924,
+      "step": 17001
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.148350868742402e-06,
+      "loss": 0.7344,
+      "step": 17002
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.147247498519225e-06,
+      "loss": 0.7217,
+      "step": 17003
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.146144166101706e-06,
+      "loss": 0.7441,
+      "step": 17004
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1450408715044605e-06,
+      "loss": 0.709,
+      "step": 17005
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.143937614742112e-06,
+      "loss": 0.7041,
+      "step": 17006
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.142834395829283e-06,
+      "loss": 0.7158,
+      "step": 17007
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.141731214780593e-06,
+      "loss": 0.7236,
+      "step": 17008
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.140628071610662e-06,
+      "loss": 0.7285,
+      "step": 17009
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.139524966334112e-06,
+      "loss": 0.7158,
+      "step": 17010
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.138421898965557e-06,
+      "loss": 0.7188,
+      "step": 17011
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.137318869519616e-06,
+      "loss": 0.7168,
+      "step": 17012
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.136215878010908e-06,
+      "loss": 0.6846,
+      "step": 17013
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.135112924454051e-06,
+      "loss": 0.71,
+      "step": 17014
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.134010008863657e-06,
+      "loss": 0.748,
+      "step": 17015
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.132907131254349e-06,
+      "loss": 0.709,
+      "step": 17016
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.131804291640743e-06,
+      "loss": 0.708,
+      "step": 17017
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.130701490037445e-06,
+      "loss": 0.7314,
+      "step": 17018
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1295987264590784e-06,
+      "loss": 0.71,
+      "step": 17019
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.128496000920253e-06,
+      "loss": 0.7041,
+      "step": 17020
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1273933134355845e-06,
+      "loss": 0.7373,
+      "step": 17021
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.126290664019686e-06,
+      "loss": 0.7236,
+      "step": 17022
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.125188052687172e-06,
+      "loss": 0.7266,
+      "step": 17023
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.124085479452649e-06,
+      "loss": 0.71,
+      "step": 17024
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.122982944330733e-06,
+      "loss": 0.7178,
+      "step": 17025
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.121880447336036e-06,
+      "loss": 0.7246,
+      "step": 17026
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.120777988483163e-06,
+      "loss": 0.7197,
+      "step": 17027
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1196755677867324e-06,
+      "loss": 0.7227,
+      "step": 17028
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.118573185261351e-06,
+      "loss": 0.7178,
+      "step": 17029
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.117470840921621e-06,
+      "loss": 0.6914,
+      "step": 17030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1163685347821584e-06,
+      "loss": 0.6973,
+      "step": 17031
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.115266266857572e-06,
+      "loss": 0.7412,
+      "step": 17032
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.114164037162463e-06,
+      "loss": 0.7119,
+      "step": 17033
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.113061845711447e-06,
+      "loss": 0.7158,
+      "step": 17034
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.111959692519122e-06,
+      "loss": 0.6953,
+      "step": 17035
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.110857577600106e-06,
+      "loss": 0.7393,
+      "step": 17036
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.109755500968989e-06,
+      "loss": 0.7373,
+      "step": 17037
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.108653462640386e-06,
+      "loss": 0.7031,
+      "step": 17038
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.107551462628898e-06,
+      "loss": 0.7197,
+      "step": 17039
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.106449500949133e-06,
+      "loss": 0.7236,
+      "step": 17040
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.105347577615688e-06,
+      "loss": 0.7236,
+      "step": 17041
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.10424569264318e-06,
+      "loss": 0.7324,
+      "step": 17042
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1031438460461945e-06,
+      "loss": 0.7031,
+      "step": 17043
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.102042037839342e-06,
+      "loss": 0.7207,
+      "step": 17044
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.100940268037222e-06,
+      "loss": 0.7197,
+      "step": 17045
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.099838536654437e-06,
+      "loss": 0.7334,
+      "step": 17046
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.098736843705586e-06,
+      "loss": 0.6973,
+      "step": 17047
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.097635189205272e-06,
+      "loss": 0.7295,
+      "step": 17048
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.096533573168094e-06,
+      "loss": 0.7139,
+      "step": 17049
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.095431995608648e-06,
+      "loss": 0.7383,
+      "step": 17050
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.094330456541531e-06,
+      "loss": 0.709,
+      "step": 17051
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.093228955981347e-06,
+      "loss": 0.7178,
+      "step": 17052
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.092127493942689e-06,
+      "loss": 0.7178,
+      "step": 17053
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.091026070440156e-06,
+      "loss": 0.7295,
+      "step": 17054
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.089924685488345e-06,
+      "loss": 0.7314,
+      "step": 17055
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.088823339101851e-06,
+      "loss": 0.7217,
+      "step": 17056
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.087722031295265e-06,
+      "loss": 0.7207,
+      "step": 17057
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.086620762083189e-06,
+      "loss": 0.709,
+      "step": 17058
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.085519531480216e-06,
+      "loss": 0.748,
+      "step": 17059
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0844183395009345e-06,
+      "loss": 0.7256,
+      "step": 17060
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.083317186159943e-06,
+      "loss": 0.7119,
+      "step": 17061
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.082216071471838e-06,
+      "loss": 0.7354,
+      "step": 17062
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.081114995451202e-06,
+      "loss": 0.7373,
+      "step": 17063
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0800139581126325e-06,
+      "loss": 0.7051,
+      "step": 17064
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.078912959470721e-06,
+      "loss": 0.7158,
+      "step": 17065
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.077811999540054e-06,
+      "loss": 0.7295,
+      "step": 17066
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.076711078335228e-06,
+      "loss": 0.7393,
+      "step": 17067
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.075610195870832e-06,
+      "loss": 0.7227,
+      "step": 17068
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.074509352161449e-06,
+      "loss": 0.7266,
+      "step": 17069
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.073408547221675e-06,
+      "loss": 0.7119,
+      "step": 17070
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.072307781066094e-06,
+      "loss": 0.7119,
+      "step": 17071
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.071207053709295e-06,
+      "loss": 0.7373,
+      "step": 17072
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0701063651658655e-06,
+      "loss": 0.7266,
+      "step": 17073
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0690057154503895e-06,
+      "loss": 0.7021,
+      "step": 17074
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.067905104577462e-06,
+      "loss": 0.7217,
+      "step": 17075
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.066804532561657e-06,
+      "loss": 0.707,
+      "step": 17076
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.065703999417568e-06,
+      "loss": 0.7178,
+      "step": 17077
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.064603505159773e-06,
+      "loss": 0.7266,
+      "step": 17078
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0635030498028625e-06,
+      "loss": 0.7266,
+      "step": 17079
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.062402633361414e-06,
+      "loss": 0.7158,
+      "step": 17080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.06130225585002e-06,
+      "loss": 0.6973,
+      "step": 17081
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.060201917283251e-06,
+      "loss": 0.6953,
+      "step": 17082
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.059101617675699e-06,
+      "loss": 0.7158,
+      "step": 17083
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.058001357041938e-06,
+      "loss": 0.6963,
+      "step": 17084
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.056901135396554e-06,
+      "loss": 0.6924,
+      "step": 17085
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.055800952754126e-06,
+      "loss": 0.7422,
+      "step": 17086
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0547008091292355e-06,
+      "loss": 0.7314,
+      "step": 17087
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.053600704536462e-06,
+      "loss": 0.7051,
+      "step": 17088
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.052500638990381e-06,
+      "loss": 0.6934,
+      "step": 17089
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.051400612505572e-06,
+      "loss": 0.7334,
+      "step": 17090
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.050300625096615e-06,
+      "loss": 0.7451,
+      "step": 17091
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.049200676778085e-06,
+      "loss": 0.7012,
+      "step": 17092
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048100767564562e-06,
+      "loss": 0.7168,
+      "step": 17093
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.047000897470623e-06,
+      "loss": 0.7178,
+      "step": 17094
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.045901066510839e-06,
+      "loss": 0.7061,
+      "step": 17095
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0448012746997865e-06,
+      "loss": 0.708,
+      "step": 17096
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.043701522052042e-06,
+      "loss": 0.7168,
+      "step": 17097
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.042601808582181e-06,
+      "loss": 0.7012,
+      "step": 17098
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.041502134304774e-06,
+      "loss": 0.7402,
+      "step": 17099
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.040402499234398e-06,
+      "loss": 0.7256,
+      "step": 17100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.039302903385624e-06,
+      "loss": 0.7295,
+      "step": 17101
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.038203346773022e-06,
+      "loss": 0.7188,
+      "step": 17102
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.037103829411166e-06,
+      "loss": 0.7354,
+      "step": 17103
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.036004351314627e-06,
+      "loss": 0.748,
+      "step": 17104
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.034904912497975e-06,
+      "loss": 0.7314,
+      "step": 17105
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.033805512975782e-06,
+      "loss": 0.7041,
+      "step": 17106
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.032706152762618e-06,
+      "loss": 0.7061,
+      "step": 17107
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.031606831873046e-06,
+      "loss": 0.7148,
+      "step": 17108
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.030507550321642e-06,
+      "loss": 0.7314,
+      "step": 17109
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.029408308122969e-06,
+      "loss": 0.7471,
+      "step": 17110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.028309105291597e-06,
+      "loss": 0.7461,
+      "step": 17111
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.027209941842093e-06,
+      "loss": 0.7266,
+      "step": 17112
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.026110817789021e-06,
+      "loss": 0.7305,
+      "step": 17113
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.025011733146953e-06,
+      "loss": 0.7002,
+      "step": 17114
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.02391268793045e-06,
+      "loss": 0.71,
+      "step": 17115
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.022813682154076e-06,
+      "loss": 0.7285,
+      "step": 17116
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.021714715832395e-06,
+      "loss": 0.7314,
+      "step": 17117
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.020615788979974e-06,
+      "loss": 0.7266,
+      "step": 17118
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.019516901611374e-06,
+      "loss": 0.7539,
+      "step": 17119
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0184180537411654e-06,
+      "loss": 0.7578,
+      "step": 17120
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017319245383896e-06,
+      "loss": 0.7236,
+      "step": 17121
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.016220476554138e-06,
+      "loss": 0.7148,
+      "step": 17122
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.015121747266449e-06,
+      "loss": 0.7236,
+      "step": 17123
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.014023057535393e-06,
+      "loss": 0.7305,
+      "step": 17124
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0129244073755265e-06,
+      "loss": 0.7188,
+      "step": 17125
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.011825796801412e-06,
+      "loss": 0.7354,
+      "step": 17126
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.010727225827609e-06,
+      "loss": 0.7334,
+      "step": 17127
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.009628694468675e-06,
+      "loss": 0.7051,
+      "step": 17128
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.008530202739163e-06,
+      "loss": 0.7129,
+      "step": 17129
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0074317506536394e-06,
+      "loss": 0.6875,
+      "step": 17130
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.006333338226654e-06,
+      "loss": 0.7119,
+      "step": 17131
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.005234965472769e-06,
+      "loss": 0.7158,
+      "step": 17132
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.00413663240654e-06,
+      "loss": 0.7051,
+      "step": 17133
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0030383390425185e-06,
+      "loss": 0.7305,
+      "step": 17134
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.001940085395259e-06,
+      "loss": 0.7197,
+      "step": 17135
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.000841871479321e-06,
+      "loss": 0.6963,
+      "step": 17136
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.999743697309252e-06,
+      "loss": 0.7227,
+      "step": 17137
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9986455628996135e-06,
+      "loss": 0.7334,
+      "step": 17138
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.997547468264952e-06,
+      "loss": 0.7129,
+      "step": 17139
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.996449413419826e-06,
+      "loss": 0.7334,
+      "step": 17140
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.995351398378778e-06,
+      "loss": 0.707,
+      "step": 17141
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.994253423156365e-06,
+      "loss": 0.7129,
+      "step": 17142
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9931554877671385e-06,
+      "loss": 0.6973,
+      "step": 17143
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.992057592225645e-06,
+      "loss": 0.707,
+      "step": 17144
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.990959736546439e-06,
+      "loss": 0.7197,
+      "step": 17145
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98986192074407e-06,
+      "loss": 0.7129,
+      "step": 17146
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9887641448330775e-06,
+      "loss": 0.7109,
+      "step": 17147
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98766640882802e-06,
+      "loss": 0.6953,
+      "step": 17148
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98656871274344e-06,
+      "loss": 0.7061,
+      "step": 17149
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.985471056593885e-06,
+      "loss": 0.6924,
+      "step": 17150
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9843734403939014e-06,
+      "loss": 0.7178,
+      "step": 17151
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.983275864158035e-06,
+      "loss": 0.6895,
+      "step": 17152
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.982178327900837e-06,
+      "loss": 0.7012,
+      "step": 17153
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.981080831636845e-06,
+      "loss": 0.7178,
+      "step": 17154
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.979983375380605e-06,
+      "loss": 0.7031,
+      "step": 17155
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.978885959146661e-06,
+      "loss": 0.7217,
+      "step": 17156
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9777885829495585e-06,
+      "loss": 0.7109,
+      "step": 17157
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.976691246803836e-06,
+      "loss": 0.7188,
+      "step": 17158
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.975593950724045e-06,
+      "loss": 0.7227,
+      "step": 17159
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.974496694724716e-06,
+      "loss": 0.7314,
+      "step": 17160
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.973399478820396e-06,
+      "loss": 0.7383,
+      "step": 17161
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.972302303025624e-06,
+      "loss": 0.707,
+      "step": 17162
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.971205167354942e-06,
+      "loss": 0.6924,
+      "step": 17163
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9701080718228854e-06,
+      "loss": 0.6982,
+      "step": 17164
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.969011016444001e-06,
+      "loss": 0.7031,
+      "step": 17165
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.967914001232823e-06,
+      "loss": 0.7207,
+      "step": 17166
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.966817026203887e-06,
+      "loss": 0.7129,
+      "step": 17167
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.965720091371731e-06,
+      "loss": 0.7188,
+      "step": 17168
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.964623196750896e-06,
+      "loss": 0.7266,
+      "step": 17169
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.963526342355913e-06,
+      "loss": 0.7109,
+      "step": 17170
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9624295282013245e-06,
+      "loss": 0.7549,
+      "step": 17171
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.961332754301664e-06,
+      "loss": 0.7402,
+      "step": 17172
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.960236020671462e-06,
+      "loss": 0.7217,
+      "step": 17173
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.959139327325253e-06,
+      "loss": 0.7061,
+      "step": 17174
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.958042674277577e-06,
+      "loss": 0.7148,
+      "step": 17175
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9569460615429595e-06,
+      "loss": 0.7158,
+      "step": 17176
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.955849489135939e-06,
+      "loss": 0.7324,
+      "step": 17177
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.954752957071047e-06,
+      "loss": 0.7119,
+      "step": 17178
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.953656465362817e-06,
+      "loss": 0.7109,
+      "step": 17179
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.952560014025771e-06,
+      "loss": 0.6982,
+      "step": 17180
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.951463603074448e-06,
+      "loss": 0.71,
+      "step": 17181
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9503672325233725e-06,
+      "loss": 0.7393,
+      "step": 17182
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.94927090238708e-06,
+      "loss": 0.7275,
+      "step": 17183
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.948174612680097e-06,
+      "loss": 0.7041,
+      "step": 17184
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.947078363416952e-06,
+      "loss": 0.7432,
+      "step": 17185
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.94598215461217e-06,
+      "loss": 0.7021,
+      "step": 17186
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.944885986280281e-06,
+      "loss": 0.6992,
+      "step": 17187
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.943789858435811e-06,
+      "loss": 0.7285,
+      "step": 17188
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9426937710932854e-06,
+      "loss": 0.7305,
+      "step": 17189
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9415977242672325e-06,
+      "loss": 0.709,
+      "step": 17190
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.940501717972175e-06,
+      "loss": 0.7061,
+      "step": 17191
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.939405752222643e-06,
+      "loss": 0.7275,
+      "step": 17192
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.938309827033154e-06,
+      "loss": 0.7168,
+      "step": 17193
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.937213942418233e-06,
+      "loss": 0.7324,
+      "step": 17194
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.936118098392402e-06,
+      "loss": 0.7441,
+      "step": 17195
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9350222949701884e-06,
+      "loss": 0.7402,
+      "step": 17196
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.933926532166109e-06,
+      "loss": 0.7178,
+      "step": 17197
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9328308099946905e-06,
+      "loss": 0.7412,
+      "step": 17198
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.931735128470449e-06,
+      "loss": 0.7275,
+      "step": 17199
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.930639487607908e-06,
+      "loss": 0.7207,
+      "step": 17200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.929543887421582e-06,
+      "loss": 0.7354,
+      "step": 17201
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.928448327925996e-06,
+      "loss": 0.7148,
+      "step": 17202
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.927352809135666e-06,
+      "loss": 0.7197,
+      "step": 17203
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.926257331065113e-06,
+      "loss": 0.7256,
+      "step": 17204
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925161893728853e-06,
+      "loss": 0.7373,
+      "step": 17205
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.924066497141403e-06,
+      "loss": 0.7363,
+      "step": 17206
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.922971141317276e-06,
+      "loss": 0.7627,
+      "step": 17207
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.921875826270994e-06,
+      "loss": 0.708,
+      "step": 17208
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.920780552017068e-06,
+      "loss": 0.7158,
+      "step": 17209
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.919685318570016e-06,
+      "loss": 0.7412,
+      "step": 17210
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9185901259443534e-06,
+      "loss": 0.7021,
+      "step": 17211
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.91749497415459e-06,
+      "loss": 0.7197,
+      "step": 17212
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.91639986321524e-06,
+      "loss": 0.7139,
+      "step": 17213
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.915304793140818e-06,
+      "loss": 0.7266,
+      "step": 17214
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.914209763945834e-06,
+      "loss": 0.6904,
+      "step": 17215
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.913114775644804e-06,
+      "loss": 0.6963,
+      "step": 17216
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.912019828252234e-06,
+      "loss": 0.6934,
+      "step": 17217
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.910924921782643e-06,
+      "loss": 0.7061,
+      "step": 17218
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.7402,
+      "step": 17219
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.908735231670407e-06,
+      "loss": 0.7041,
+      "step": 17220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9076404480567875e-06,
+      "loss": 0.708,
+      "step": 17221
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.906545705424178e-06,
+      "loss": 0.7461,
+      "step": 17222
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.905451003787087e-06,
+      "loss": 0.7451,
+      "step": 17223
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.904356343160025e-06,
+      "loss": 0.708,
+      "step": 17224
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.90326172355749e-06,
+      "loss": 0.7168,
+      "step": 17225
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.902167144993994e-06,
+      "loss": 0.7432,
+      "step": 17226
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.901072607484044e-06,
+      "loss": 0.7324,
+      "step": 17227
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.899978111042141e-06,
+      "loss": 0.7119,
+      "step": 17228
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.898883655682793e-06,
+      "loss": 0.707,
+      "step": 17229
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.897789241420501e-06,
+      "loss": 0.7285,
+      "step": 17230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.896694868269776e-06,
+      "loss": 0.6846,
+      "step": 17231
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.895600536245112e-06,
+      "loss": 0.7119,
+      "step": 17232
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.894506245361016e-06,
+      "loss": 0.7393,
+      "step": 17233
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.893411995631987e-06,
+      "loss": 0.7246,
+      "step": 17234
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.892317787072529e-06,
+      "loss": 0.71,
+      "step": 17235
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8912236196971416e-06,
+      "loss": 0.7207,
+      "step": 17236
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.890129493520329e-06,
+      "loss": 0.7275,
+      "step": 17237
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8890354085565855e-06,
+      "loss": 0.7061,
+      "step": 17238
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.887941364820413e-06,
+      "loss": 0.7119,
+      "step": 17239
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.886847362326308e-06,
+      "loss": 0.7061,
+      "step": 17240
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.885753401088771e-06,
+      "loss": 0.7207,
+      "step": 17241
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.884659481122296e-06,
+      "loss": 0.7168,
+      "step": 17242
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883565602441388e-06,
+      "loss": 0.7314,
+      "step": 17243
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8824717650605355e-06,
+      "loss": 0.7031,
+      "step": 17244
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.8813779689942375e-06,
+      "loss": 0.7275,
+      "step": 17245
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.880284214256987e-06,
+      "loss": 0.71,
+      "step": 17246
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.879190500863282e-06,
+      "loss": 0.7041,
+      "step": 17247
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.878096828827614e-06,
+      "loss": 0.7334,
+      "step": 17248
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.877003198164479e-06,
+      "loss": 0.7324,
+      "step": 17249
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.875909608888372e-06,
+      "loss": 0.7158,
+      "step": 17250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8748160610137824e-06,
+      "loss": 0.7246,
+      "step": 17251
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8737225545552e-06,
+      "loss": 0.7168,
+      "step": 17252
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8726290895271205e-06,
+      "loss": 0.7168,
+      "step": 17253
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.871535665944031e-06,
+      "loss": 0.7529,
+      "step": 17254
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.870442283820427e-06,
+      "loss": 0.7021,
+      "step": 17255
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8693489431708e-06,
+      "loss": 0.7178,
+      "step": 17256
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.868255644009631e-06,
+      "loss": 0.6953,
+      "step": 17257
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.867162386351412e-06,
+      "loss": 0.7285,
+      "step": 17258
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.866069170210632e-06,
+      "loss": 0.7266,
+      "step": 17259
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.864975995601779e-06,
+      "loss": 0.709,
+      "step": 17260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.863882862539342e-06,
+      "loss": 0.7012,
+      "step": 17261
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.862789771037803e-06,
+      "loss": 0.7305,
+      "step": 17262
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.861696721111654e-06,
+      "loss": 0.71,
+      "step": 17263
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.860603712775376e-06,
+      "loss": 0.6787,
+      "step": 17264
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.859510746043453e-06,
+      "loss": 0.7158,
+      "step": 17265
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.858417820930371e-06,
+      "loss": 0.7197,
+      "step": 17266
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.857324937450615e-06,
+      "loss": 0.7275,
+      "step": 17267
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.856232095618669e-06,
+      "loss": 0.7227,
+      "step": 17268
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.855139295449016e-06,
+      "loss": 0.7266,
+      "step": 17269
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.854046536956132e-06,
+      "loss": 0.7188,
+      "step": 17270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.852953820154504e-06,
+      "loss": 0.6797,
+      "step": 17271
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8518611450586125e-06,
+      "loss": 0.7246,
+      "step": 17272
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.850768511682935e-06,
+      "loss": 0.708,
+      "step": 17273
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.849675920041956e-06,
+      "loss": 0.7197,
+      "step": 17274
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.848583370150152e-06,
+      "loss": 0.708,
+      "step": 17275
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.847490862022006e-06,
+      "loss": 0.7236,
+      "step": 17276
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8463983956719915e-06,
+      "loss": 0.7139,
+      "step": 17277
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.845305971114587e-06,
+      "loss": 0.6914,
+      "step": 17278
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.844213588364268e-06,
+      "loss": 0.7012,
+      "step": 17279
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.843121247435516e-06,
+      "loss": 0.707,
+      "step": 17280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.842028948342803e-06,
+      "loss": 0.7217,
+      "step": 17281
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.840936691100609e-06,
+      "loss": 0.7217,
+      "step": 17282
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.839844475723405e-06,
+      "loss": 0.7256,
+      "step": 17283
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.838752302225666e-06,
+      "loss": 0.6943,
+      "step": 17284
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.837660170621864e-06,
+      "loss": 0.709,
+      "step": 17285
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.836568080926478e-06,
+      "loss": 0.75,
+      "step": 17286
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8354760331539735e-06,
+      "loss": 0.7266,
+      "step": 17287
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.83438402731883e-06,
+      "loss": 0.7305,
+      "step": 17288
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.833292063435518e-06,
+      "loss": 0.7275,
+      "step": 17289
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8322001415185036e-06,
+      "loss": 0.7441,
+      "step": 17290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.831108261582259e-06,
+      "loss": 0.6982,
+      "step": 17291
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.830016423641257e-06,
+      "loss": 0.7061,
+      "step": 17292
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.828924627709962e-06,
+      "loss": 0.7168,
+      "step": 17293
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8278328738028485e-06,
+      "loss": 0.7002,
+      "step": 17294
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.826741161934385e-06,
+      "loss": 0.7295,
+      "step": 17295
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.825649492119033e-06,
+      "loss": 0.7266,
+      "step": 17296
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8245578643712625e-06,
+      "loss": 0.6846,
+      "step": 17297
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8234662787055436e-06,
+      "loss": 0.7334,
+      "step": 17298
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.822374735136336e-06,
+      "loss": 0.7051,
+      "step": 17299
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.821283233678113e-06,
+      "loss": 0.7373,
+      "step": 17300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.820191774345334e-06,
+      "loss": 0.749,
+      "step": 17301
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.819100357152468e-06,
+      "loss": 0.6816,
+      "step": 17302
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8180089821139695e-06,
+      "loss": 0.7207,
+      "step": 17303
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.816917649244311e-06,
+      "loss": 0.7285,
+      "step": 17304
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.815826358557951e-06,
+      "loss": 0.7266,
+      "step": 17305
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.814735110069354e-06,
+      "loss": 0.7266,
+      "step": 17306
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.81364390379298e-06,
+      "loss": 0.7383,
+      "step": 17307
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.812552739743295e-06,
+      "loss": 0.7031,
+      "step": 17308
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.811461617934747e-06,
+      "loss": 0.7021,
+      "step": 17309
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.810370538381808e-06,
+      "loss": 0.7314,
+      "step": 17310
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.809279501098933e-06,
+      "loss": 0.7393,
+      "step": 17311
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.808188506100579e-06,
+      "loss": 0.7275,
+      "step": 17312
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.807097553401209e-06,
+      "loss": 0.709,
+      "step": 17313
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.806006643015275e-06,
+      "loss": 0.7354,
+      "step": 17314
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.804915774957242e-06,
+      "loss": 0.7285,
+      "step": 17315
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.803824949241558e-06,
+      "loss": 0.7217,
+      "step": 17316
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.802734165882685e-06,
+      "loss": 0.7041,
+      "step": 17317
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.801643424895073e-06,
+      "loss": 0.7246,
+      "step": 17318
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.800552726293181e-06,
+      "loss": 0.7266,
+      "step": 17319
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.799462070091461e-06,
+      "loss": 0.7461,
+      "step": 17320
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.798371456304371e-06,
+      "loss": 0.708,
+      "step": 17321
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.79728088494636e-06,
+      "loss": 0.7227,
+      "step": 17322
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.796190356031882e-06,
+      "loss": 0.7168,
+      "step": 17323
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795099869575388e-06,
+      "loss": 0.7021,
+      "step": 17324
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.79400942559133e-06,
+      "loss": 0.7012,
+      "step": 17325
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.792919024094158e-06,
+      "loss": 0.7256,
+      "step": 17326
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.791828665098327e-06,
+      "loss": 0.7041,
+      "step": 17327
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.790738348618284e-06,
+      "loss": 0.6924,
+      "step": 17328
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.789648074668477e-06,
+      "loss": 0.7119,
+      "step": 17329
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.788557843263353e-06,
+      "loss": 0.7246,
+      "step": 17330
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.787467654417363e-06,
+      "loss": 0.7197,
+      "step": 17331
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.786377508144954e-06,
+      "loss": 0.709,
+      "step": 17332
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.785287404460575e-06,
+      "loss": 0.7158,
+      "step": 17333
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7841973433786716e-06,
+      "loss": 0.7197,
+      "step": 17334
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.783107324913687e-06,
+      "loss": 0.7324,
+      "step": 17335
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7820173490800655e-06,
+      "loss": 0.7461,
+      "step": 17336
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7809274158922575e-06,
+      "loss": 0.7021,
+      "step": 17337
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7798375253647006e-06,
+      "loss": 0.708,
+      "step": 17338
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.778747677511845e-06,
+      "loss": 0.7061,
+      "step": 17339
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.777657872348131e-06,
+      "loss": 0.6826,
+      "step": 17340
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.776568109888002e-06,
+      "loss": 0.7061,
+      "step": 17341
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.775478390145895e-06,
+      "loss": 0.7256,
+      "step": 17342
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.774388713136256e-06,
+      "loss": 0.7461,
+      "step": 17343
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.773299078873523e-06,
+      "loss": 0.6895,
+      "step": 17344
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.772209487372141e-06,
+      "loss": 0.7266,
+      "step": 17345
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7711199386465466e-06,
+      "loss": 0.7432,
+      "step": 17346
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.77003043271118e-06,
+      "loss": 0.6973,
+      "step": 17347
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7689409695804755e-06,
+      "loss": 0.7168,
+      "step": 17348
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.767851549268875e-06,
+      "loss": 0.7246,
+      "step": 17349
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.766762171790813e-06,
+      "loss": 0.7236,
+      "step": 17350
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.76567283716073e-06,
+      "loss": 0.7168,
+      "step": 17351
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.764583545393062e-06,
+      "loss": 0.7246,
+      "step": 17352
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.76349429650224e-06,
+      "loss": 0.7305,
+      "step": 17353
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.762405090502706e-06,
+      "loss": 0.7207,
+      "step": 17354
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.761315927408889e-06,
+      "loss": 0.7178,
+      "step": 17355
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.760226807235225e-06,
+      "loss": 0.7334,
+      "step": 17356
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.759137729996146e-06,
+      "loss": 0.7021,
+      "step": 17357
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.758048695706086e-06,
+      "loss": 0.7178,
+      "step": 17358
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.756959704379477e-06,
+      "loss": 0.6895,
+      "step": 17359
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7558707560307534e-06,
+      "loss": 0.7236,
+      "step": 17360
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.754781850674343e-06,
+      "loss": 0.7012,
+      "step": 17361
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7536929883246785e-06,
+      "loss": 0.7061,
+      "step": 17362
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.752604168996184e-06,
+      "loss": 0.709,
+      "step": 17363
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.751515392703296e-06,
+      "loss": 0.71,
+      "step": 17364
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.75042665946044e-06,
+      "loss": 0.7324,
+      "step": 17365
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.749337969282046e-06,
+      "loss": 0.7139,
+      "step": 17366
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.748249322182543e-06,
+      "loss": 0.7266,
+      "step": 17367
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.747160718176355e-06,
+      "loss": 0.7158,
+      "step": 17368
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.746072157277907e-06,
+      "loss": 0.7021,
+      "step": 17369
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.744983639501628e-06,
+      "loss": 0.6865,
+      "step": 17370
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.743895164861942e-06,
+      "loss": 0.7246,
+      "step": 17371
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.742806733373277e-06,
+      "loss": 0.7197,
+      "step": 17372
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.741718345050056e-06,
+      "loss": 0.707,
+      "step": 17373
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7406299999067e-06,
+      "loss": 0.7207,
+      "step": 17374
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.739541697957632e-06,
+      "loss": 0.7188,
+      "step": 17375
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7384534392172785e-06,
+      "loss": 0.7539,
+      "step": 17376
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7373652237000564e-06,
+      "loss": 0.7178,
+      "step": 17377
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.736277051420393e-06,
+      "loss": 0.7207,
+      "step": 17378
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.735188922392706e-06,
+      "loss": 0.7393,
+      "step": 17379
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.734100836631416e-06,
+      "loss": 0.7344,
+      "step": 17380
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7330127941509415e-06,
+      "loss": 0.7275,
+      "step": 17381
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.731924794965703e-06,
+      "loss": 0.7031,
+      "step": 17382
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.730836839090116e-06,
+      "loss": 0.7236,
+      "step": 17383
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.729748926538604e-06,
+      "loss": 0.7188,
+      "step": 17384
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.728661057325582e-06,
+      "loss": 0.7217,
+      "step": 17385
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.727573231465467e-06,
+      "loss": 0.7266,
+      "step": 17386
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.726485448972671e-06,
+      "loss": 0.7344,
+      "step": 17387
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.725397709861616e-06,
+      "loss": 0.748,
+      "step": 17388
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.724310014146712e-06,
+      "loss": 0.708,
+      "step": 17389
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.723222361842377e-06,
+      "loss": 0.7344,
+      "step": 17390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.722134752963025e-06,
+      "loss": 0.7051,
+      "step": 17391
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.721047187523066e-06,
+      "loss": 0.749,
+      "step": 17392
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.719959665536917e-06,
+      "loss": 0.708,
+      "step": 17393
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.718872187018988e-06,
+      "loss": 0.7275,
+      "step": 17394
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7177847519836896e-06,
+      "loss": 0.7246,
+      "step": 17395
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7166973604454345e-06,
+      "loss": 0.7295,
+      "step": 17396
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.715610012418632e-06,
+      "loss": 0.7031,
+      "step": 17397
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.714522707917693e-06,
+      "loss": 0.708,
+      "step": 17398
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.713435446957029e-06,
+      "loss": 0.7139,
+      "step": 17399
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7123482295510445e-06,
+      "loss": 0.709,
+      "step": 17400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.711261055714149e-06,
+      "loss": 0.7275,
+      "step": 17401
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7101739254607486e-06,
+      "loss": 0.7217,
+      "step": 17402
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.709086838805255e-06,
+      "loss": 0.7285,
+      "step": 17403
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.70799979576207e-06,
+      "loss": 0.7295,
+      "step": 17404
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.706912796345603e-06,
+      "loss": 0.7207,
+      "step": 17405
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7058258405702615e-06,
+      "loss": 0.7334,
+      "step": 17406
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.704738928450443e-06,
+      "loss": 0.7305,
+      "step": 17407
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7036520600005515e-06,
+      "loss": 0.7236,
+      "step": 17408
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.702565235234999e-06,
+      "loss": 0.7207,
+      "step": 17409
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.70147845416818e-06,
+      "loss": 0.7012,
+      "step": 17410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.700391716814503e-06,
+      "loss": 0.7236,
+      "step": 17411
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.699305023188368e-06,
+      "loss": 0.7471,
+      "step": 17412
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.698218373304175e-06,
+      "loss": 0.7295,
+      "step": 17413
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.697131767176322e-06,
+      "loss": 0.6953,
+      "step": 17414
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.6960452048192155e-06,
+      "loss": 0.751,
+      "step": 17415
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.69495868624725e-06,
+      "loss": 0.6963,
+      "step": 17416
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.693872211474828e-06,
+      "loss": 0.6934,
+      "step": 17417
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.692785780516345e-06,
+      "loss": 0.7188,
+      "step": 17418
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.691699393386202e-06,
+      "loss": 0.7295,
+      "step": 17419
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.690613050098791e-06,
+      "loss": 0.7158,
+      "step": 17420
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.689526750668512e-06,
+      "loss": 0.7002,
+      "step": 17421
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6884404951097605e-06,
+      "loss": 0.7207,
+      "step": 17422
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6873542834369335e-06,
+      "loss": 0.7109,
+      "step": 17423
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.686268115664422e-06,
+      "loss": 0.6963,
+      "step": 17424
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.685181991806627e-06,
+      "loss": 0.7139,
+      "step": 17425
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.684095911877934e-06,
+      "loss": 0.7119,
+      "step": 17426
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.683009875892741e-06,
+      "loss": 0.7246,
+      "step": 17427
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.681923883865438e-06,
+      "loss": 0.6924,
+      "step": 17428
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.680837935810419e-06,
+      "loss": 0.71,
+      "step": 17429
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679752031742076e-06,
+      "loss": 0.7031,
+      "step": 17430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.678666171674797e-06,
+      "loss": 0.7441,
+      "step": 17431
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.677580355622977e-06,
+      "loss": 0.6904,
+      "step": 17432
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6764945836009986e-06,
+      "loss": 0.6895,
+      "step": 17433
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6754088556232554e-06,
+      "loss": 0.7256,
+      "step": 17434
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.674323171704136e-06,
+      "loss": 0.7383,
+      "step": 17435
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6732375318580255e-06,
+      "loss": 0.7344,
+      "step": 17436
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6721519360993125e-06,
+      "loss": 0.708,
+      "step": 17437
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.671066384442389e-06,
+      "loss": 0.708,
+      "step": 17438
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.669980876901632e-06,
+      "loss": 0.7305,
+      "step": 17439
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6688954134914316e-06,
+      "loss": 0.6885,
+      "step": 17440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.667809994226171e-06,
+      "loss": 0.7227,
+      "step": 17441
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6667246191202385e-06,
+      "loss": 0.6895,
+      "step": 17442
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.665639288188013e-06,
+      "loss": 0.7002,
+      "step": 17443
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.664554001443882e-06,
+      "loss": 0.6992,
+      "step": 17444
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.663468758902226e-06,
+      "loss": 0.7354,
+      "step": 17445
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.662383560577427e-06,
+      "loss": 0.7314,
+      "step": 17446
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.661298406483864e-06,
+      "loss": 0.7236,
+      "step": 17447
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.660213296635922e-06,
+      "loss": 0.7227,
+      "step": 17448
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6591282310479776e-06,
+      "loss": 0.7227,
+      "step": 17449
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.658043209734415e-06,
+      "loss": 0.7188,
+      "step": 17450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.656958232709611e-06,
+      "loss": 0.749,
+      "step": 17451
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6558732999879436e-06,
+      "loss": 0.7188,
+      "step": 17452
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.654788411583788e-06,
+      "loss": 0.7119,
+      "step": 17453
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.653703567511527e-06,
+      "loss": 0.7139,
+      "step": 17454
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.652618767785533e-06,
+      "loss": 0.7324,
+      "step": 17455
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.651534012420184e-06,
+      "loss": 0.7217,
+      "step": 17456
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.650449301429855e-06,
+      "loss": 0.7158,
+      "step": 17457
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.649364634828926e-06,
+      "loss": 0.7373,
+      "step": 17458
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6482800126317615e-06,
+      "loss": 0.7158,
+      "step": 17459
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.647195434852743e-06,
+      "loss": 0.7295,
+      "step": 17460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.646110901506237e-06,
+      "loss": 0.7363,
+      "step": 17461
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.645026412606625e-06,
+      "loss": 0.7432,
+      "step": 17462
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.643941968168274e-06,
+      "loss": 0.7295,
+      "step": 17463
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.642857568205556e-06,
+      "loss": 0.7275,
+      "step": 17464
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.64177321273284e-06,
+      "loss": 0.7168,
+      "step": 17465
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.640688901764498e-06,
+      "loss": 0.7061,
+      "step": 17466
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.639604635314898e-06,
+      "loss": 0.7109,
+      "step": 17467
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.638520413398412e-06,
+      "loss": 0.7041,
+      "step": 17468
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.637436236029407e-06,
+      "loss": 0.708,
+      "step": 17469
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.636352103222255e-06,
+      "loss": 0.7041,
+      "step": 17470
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.635268014991313e-06,
+      "loss": 0.7188,
+      "step": 17471
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6341839713509574e-06,
+      "loss": 0.7061,
+      "step": 17472
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.633099972315547e-06,
+      "loss": 0.7363,
+      "step": 17473
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.632016017899455e-06,
+      "loss": 0.7178,
+      "step": 17474
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.630932108117042e-06,
+      "loss": 0.7178,
+      "step": 17475
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.629848242982671e-06,
+      "loss": 0.7305,
+      "step": 17476
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.628764422510712e-06,
+      "loss": 0.7041,
+      "step": 17477
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6276806467155205e-06,
+      "loss": 0.7217,
+      "step": 17478
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6265969156114605e-06,
+      "loss": 0.7119,
+      "step": 17479
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.625513229212899e-06,
+      "loss": 0.7041,
+      "step": 17480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6244295875341935e-06,
+      "loss": 0.7129,
+      "step": 17481
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.623345990589704e-06,
+      "loss": 0.6963,
+      "step": 17482
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.622262438393796e-06,
+      "loss": 0.7021,
+      "step": 17483
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.621178930960823e-06,
+      "loss": 0.7119,
+      "step": 17484
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.620095468305147e-06,
+      "loss": 0.6992,
+      "step": 17485
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.619012050441123e-06,
+      "loss": 0.7363,
+      "step": 17486
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.617928677383114e-06,
+      "loss": 0.7158,
+      "step": 17487
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.616845349145473e-06,
+      "loss": 0.7217,
+      "step": 17488
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.615762065742559e-06,
+      "loss": 0.7275,
+      "step": 17489
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6146788271887316e-06,
+      "loss": 0.7266,
+      "step": 17490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.613595633498339e-06,
+      "loss": 0.6943,
+      "step": 17491
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.612512484685738e-06,
+      "loss": 0.7324,
+      "step": 17492
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.611429380765284e-06,
+      "loss": 0.7217,
+      "step": 17493
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.610346321751331e-06,
+      "loss": 0.7168,
+      "step": 17494
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.609263307658232e-06,
+      "loss": 0.709,
+      "step": 17495
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6081803385003425e-06,
+      "loss": 0.7129,
+      "step": 17496
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6070974142920095e-06,
+      "loss": 0.7246,
+      "step": 17497
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.606014535047583e-06,
+      "loss": 0.7119,
+      "step": 17498
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.604931700781419e-06,
+      "loss": 0.7305,
+      "step": 17499
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.603848911507862e-06,
+      "loss": 0.7139,
+      "step": 17500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.602766167241268e-06,
+      "loss": 0.7207,
+      "step": 17501
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.60168346799598e-06,
+      "loss": 0.7236,
+      "step": 17502
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.600600813786355e-06,
+      "loss": 0.7178,
+      "step": 17503
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5995182046267295e-06,
+      "loss": 0.7314,
+      "step": 17504
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.598435640531457e-06,
+      "loss": 0.7158,
+      "step": 17505
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.597353121514881e-06,
+      "loss": 0.7246,
+      "step": 17506
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.596270647591352e-06,
+      "loss": 0.6914,
+      "step": 17507
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595188218775211e-06,
+      "loss": 0.7139,
+      "step": 17508
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.594105835080806e-06,
+      "loss": 0.6992,
+      "step": 17509
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.593023496522476e-06,
+      "loss": 0.7207,
+      "step": 17510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.591941203114569e-06,
+      "loss": 0.7158,
+      "step": 17511
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5908589548714265e-06,
+      "loss": 0.7119,
+      "step": 17512
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.589776751807391e-06,
+      "loss": 0.7119,
+      "step": 17513
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.588694593936805e-06,
+      "loss": 0.7227,
+      "step": 17514
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.587612481274007e-06,
+      "loss": 0.7188,
+      "step": 17515
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5865304138333445e-06,
+      "loss": 0.7393,
+      "step": 17516
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.585448391629149e-06,
+      "loss": 0.7422,
+      "step": 17517
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5843664146757616e-06,
+      "loss": 0.6963,
+      "step": 17518
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.583284482987524e-06,
+      "loss": 0.7109,
+      "step": 17519
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.582202596578774e-06,
+      "loss": 0.7227,
+      "step": 17520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.581120755463846e-06,
+      "loss": 0.7285,
+      "step": 17521
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5800389596570825e-06,
+      "loss": 0.7227,
+      "step": 17522
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.578957209172815e-06,
+      "loss": 0.7188,
+      "step": 17523
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.577875504025379e-06,
+      "loss": 0.7119,
+      "step": 17524
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.576793844229111e-06,
+      "loss": 0.709,
+      "step": 17525
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.575712229798348e-06,
+      "loss": 0.7256,
+      "step": 17526
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.574630660747419e-06,
+      "loss": 0.7148,
+      "step": 17527
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.57354913709066e-06,
+      "loss": 0.7109,
+      "step": 17528
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.572467658842408e-06,
+      "loss": 0.7168,
+      "step": 17529
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.571386226016989e-06,
+      "loss": 0.7217,
+      "step": 17530
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.570304838628733e-06,
+      "loss": 0.708,
+      "step": 17531
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.569223496691976e-06,
+      "loss": 0.7314,
+      "step": 17532
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5681422002210456e-06,
+      "loss": 0.7061,
+      "step": 17533
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.567060949230275e-06,
+      "loss": 0.7207,
+      "step": 17534
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.565979743733992e-06,
+      "loss": 0.7139,
+      "step": 17535
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5648985837465216e-06,
+      "loss": 0.7275,
+      "step": 17536
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.563817469282194e-06,
+      "loss": 0.6953,
+      "step": 17537
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562736400355336e-06,
+      "loss": 0.7021,
+      "step": 17538
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.561655376980274e-06,
+      "loss": 0.7354,
+      "step": 17539
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.560574399171336e-06,
+      "loss": 0.7432,
+      "step": 17540
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.559493466942845e-06,
+      "loss": 0.7246,
+      "step": 17541
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.558412580309133e-06,
+      "loss": 0.7002,
+      "step": 17542
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.557331739284514e-06,
+      "loss": 0.7188,
+      "step": 17543
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.556250943883316e-06,
+      "loss": 0.7188,
+      "step": 17544
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.555170194119862e-06,
+      "loss": 0.7617,
+      "step": 17545
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5540894900084765e-06,
+      "loss": 0.7334,
+      "step": 17546
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.553008831563479e-06,
+      "loss": 0.7354,
+      "step": 17547
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.551928218799193e-06,
+      "loss": 0.7061,
+      "step": 17548
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.550847651729934e-06,
+      "loss": 0.7119,
+      "step": 17549
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.549767130370028e-06,
+      "loss": 0.6943,
+      "step": 17550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.54868665473379e-06,
+      "loss": 0.7275,
+      "step": 17551
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.547606224835543e-06,
+      "loss": 0.7051,
+      "step": 17552
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.546525840689603e-06,
+      "loss": 0.7188,
+      "step": 17553
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.545445502310284e-06,
+      "loss": 0.6982,
+      "step": 17554
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.544365209711913e-06,
+      "loss": 0.7178,
+      "step": 17555
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.543284962908798e-06,
+      "loss": 0.7275,
+      "step": 17556
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.542204761915254e-06,
+      "loss": 0.7217,
+      "step": 17557
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.541124606745601e-06,
+      "loss": 0.7217,
+      "step": 17558
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.540044497414152e-06,
+      "loss": 0.7197,
+      "step": 17559
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.538964433935218e-06,
+      "loss": 0.7207,
+      "step": 17560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.53788441632312e-06,
+      "loss": 0.7041,
+      "step": 17561
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.536804444592164e-06,
+      "loss": 0.7061,
+      "step": 17562
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.535724518756661e-06,
+      "loss": 0.7021,
+      "step": 17563
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.534644638830928e-06,
+      "loss": 0.7344,
+      "step": 17564
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5335648048292734e-06,
+      "loss": 0.7188,
+      "step": 17565
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.532485016766006e-06,
+      "loss": 0.7441,
+      "step": 17566
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.531405274655439e-06,
+      "loss": 0.7119,
+      "step": 17567
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5303255785118826e-06,
+      "loss": 0.7178,
+      "step": 17568
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.529245928349639e-06,
+      "loss": 0.7285,
+      "step": 17569
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5281663241830166e-06,
+      "loss": 0.6904,
+      "step": 17570
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.527086766026329e-06,
+      "loss": 0.7051,
+      "step": 17571
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.526007253893877e-06,
+      "loss": 0.708,
+      "step": 17572
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.52492778779997e-06,
+      "loss": 0.6895,
+      "step": 17573
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.523848367758916e-06,
+      "loss": 0.7158,
+      "step": 17574
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.522768993785013e-06,
+      "loss": 0.6885,
+      "step": 17575
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5216896658925684e-06,
+      "loss": 0.7236,
+      "step": 17576
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.520610384095887e-06,
+      "loss": 0.7314,
+      "step": 17577
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.519531148409269e-06,
+      "loss": 0.7227,
+      "step": 17578
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.518451958847021e-06,
+      "loss": 0.707,
+      "step": 17579
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.517372815423439e-06,
+      "loss": 0.709,
+      "step": 17580
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.516293718152835e-06,
+      "loss": 0.7197,
+      "step": 17581
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.515214667049496e-06,
+      "loss": 0.7324,
+      "step": 17582
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.51413566212773e-06,
+      "loss": 0.7109,
+      "step": 17583
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.513056703401832e-06,
+      "loss": 0.7129,
+      "step": 17584
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.511977790886106e-06,
+      "loss": 0.7188,
+      "step": 17585
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.510898924594845e-06,
+      "loss": 0.7363,
+      "step": 17586
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.509820104542355e-06,
+      "loss": 0.7041,
+      "step": 17587
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5087413307429206e-06,
+      "loss": 0.707,
+      "step": 17588
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.507662603210845e-06,
+      "loss": 0.7383,
+      "step": 17589
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.506583921960422e-06,
+      "loss": 0.7617,
+      "step": 17590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.50550528700595e-06,
+      "loss": 0.7314,
+      "step": 17591
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.504426698361721e-06,
+      "loss": 0.7197,
+      "step": 17592
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.503348156042026e-06,
+      "loss": 0.7412,
+      "step": 17593
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.502269660061167e-06,
+      "loss": 0.7109,
+      "step": 17594
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.501191210433428e-06,
+      "loss": 0.6934,
+      "step": 17595
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5001128071731016e-06,
+      "loss": 0.7012,
+      "step": 17596
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.499034450294484e-06,
+      "loss": 0.7227,
+      "step": 17597
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.497956139811862e-06,
+      "loss": 0.7197,
+      "step": 17598
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.496877875739526e-06,
+      "loss": 0.7383,
+      "step": 17599
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.49579965809177e-06,
+      "loss": 0.6973,
+      "step": 17600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.494721486882878e-06,
+      "loss": 0.7275,
+      "step": 17601
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4936433621271375e-06,
+      "loss": 0.7178,
+      "step": 17602
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.492565283838841e-06,
+      "loss": 0.7393,
+      "step": 17603
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4914872520322715e-06,
+      "loss": 0.7129,
+      "step": 17604
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.490409266721717e-06,
+      "loss": 0.6992,
+      "step": 17605
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.489331327921464e-06,
+      "loss": 0.71,
+      "step": 17606
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.488253435645799e-06,
+      "loss": 0.7148,
+      "step": 17607
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.487175589909001e-06,
+      "loss": 0.6875,
+      "step": 17608
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4860977907253595e-06,
+      "loss": 0.7246,
+      "step": 17609
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.485020038109157e-06,
+      "loss": 0.7236,
+      "step": 17610
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.483942332074671e-06,
+      "loss": 0.7051,
+      "step": 17611
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.482864672636191e-06,
+      "loss": 0.7207,
+      "step": 17612
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4817870598079975e-06,
+      "loss": 0.6963,
+      "step": 17613
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.480709493604368e-06,
+      "loss": 0.6914,
+      "step": 17614
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.479631974039581e-06,
+      "loss": 0.7178,
+      "step": 17615
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.478554501127921e-06,
+      "loss": 0.7139,
+      "step": 17616
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4774770748836645e-06,
+      "loss": 0.6943,
+      "step": 17617
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.476399695321091e-06,
+      "loss": 0.7129,
+      "step": 17618
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.475322362454477e-06,
+      "loss": 0.7109,
+      "step": 17619
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.474245076298107e-06,
+      "loss": 0.709,
+      "step": 17620
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4731678368662436e-06,
+      "loss": 0.7031,
+      "step": 17621
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.472090644173172e-06,
+      "loss": 0.7344,
+      "step": 17622
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.471013498233166e-06,
+      "loss": 0.7139,
+      "step": 17623
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.469936399060501e-06,
+      "loss": 0.7158,
+      "step": 17624
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46885934666945e-06,
+      "loss": 0.7129,
+      "step": 17625
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46778234107429e-06,
+      "loss": 0.7207,
+      "step": 17626
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.466705382289287e-06,
+      "loss": 0.7178,
+      "step": 17627
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.465628470328718e-06,
+      "loss": 0.7344,
+      "step": 17628
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46455160520685e-06,
+      "loss": 0.7061,
+      "step": 17629
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46347478693796e-06,
+      "loss": 0.709,
+      "step": 17630
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.462398015536316e-06,
+      "loss": 0.7197,
+      "step": 17631
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.461321291016187e-06,
+      "loss": 0.7168,
+      "step": 17632
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4602446133918454e-06,
+      "loss": 0.7158,
+      "step": 17633
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459167982677554e-06,
+      "loss": 0.7266,
+      "step": 17634
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4580913988875825e-06,
+      "loss": 0.7285,
+      "step": 17635
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4570148620362014e-06,
+      "loss": 0.71,
+      "step": 17636
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.455938372137675e-06,
+      "loss": 0.7236,
+      "step": 17637
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.454861929206269e-06,
+      "loss": 0.7266,
+      "step": 17638
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.453785533256252e-06,
+      "loss": 0.7227,
+      "step": 17639
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.452709184301884e-06,
+      "loss": 0.7207,
+      "step": 17640
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.451632882357428e-06,
+      "loss": 0.7295,
+      "step": 17641
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4505566274371546e-06,
+      "loss": 0.7354,
+      "step": 17642
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4494804195553225e-06,
+      "loss": 0.7256,
+      "step": 17643
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.448404258726192e-06,
+      "loss": 0.7236,
+      "step": 17644
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.447328144964029e-06,
+      "loss": 0.7305,
+      "step": 17645
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.446252078283095e-06,
+      "loss": 0.7324,
+      "step": 17646
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.445176058697644e-06,
+      "loss": 0.7285,
+      "step": 17647
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.444100086221941e-06,
+      "loss": 0.6943,
+      "step": 17648
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4430241608702446e-06,
+      "loss": 0.7227,
+      "step": 17649
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.441948282656811e-06,
+      "loss": 0.6992,
+      "step": 17650
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.440872451595901e-06,
+      "loss": 0.71,
+      "step": 17651
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.439796667701774e-06,
+      "loss": 0.7178,
+      "step": 17652
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.43872093098868e-06,
+      "loss": 0.7373,
+      "step": 17653
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.437645241470877e-06,
+      "loss": 0.7061,
+      "step": 17654
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.436569599162623e-06,
+      "loss": 0.7285,
+      "step": 17655
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.435494004078171e-06,
+      "loss": 0.7041,
+      "step": 17656
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.434418456231777e-06,
+      "loss": 0.7041,
+      "step": 17657
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.43334295563769e-06,
+      "loss": 0.7168,
+      "step": 17658
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.432267502310174e-06,
+      "loss": 0.7178,
+      "step": 17659
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.431192096263466e-06,
+      "loss": 0.7295,
+      "step": 17660
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.430116737511828e-06,
+      "loss": 0.6973,
+      "step": 17661
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.429041426069506e-06,
+      "loss": 0.7305,
+      "step": 17662
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4279661619507535e-06,
+      "loss": 0.7207,
+      "step": 17663
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.426890945169818e-06,
+      "loss": 0.6982,
+      "step": 17664
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.425815775740955e-06,
+      "loss": 0.7344,
+      "step": 17665
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.424740653678404e-06,
+      "loss": 0.7061,
+      "step": 17666
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.423665578996418e-06,
+      "loss": 0.6982,
+      "step": 17667
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4225905517092404e-06,
+      "loss": 0.7334,
+      "step": 17668
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.421515571831123e-06,
+      "loss": 0.7217,
+      "step": 17669
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.420440639376306e-06,
+      "loss": 0.6924,
+      "step": 17670
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.419365754359041e-06,
+      "loss": 0.7305,
+      "step": 17671
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.418290916793572e-06,
+      "loss": 0.7432,
+      "step": 17672
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.417216126694139e-06,
+      "loss": 0.7256,
+      "step": 17673
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.416141384074985e-06,
+      "loss": 0.7129,
+      "step": 17674
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4150666889503575e-06,
+      "loss": 0.7217,
+      "step": 17675
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413992041334497e-06,
+      "loss": 0.6943,
+      "step": 17676
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.412917441241642e-06,
+      "loss": 0.7109,
+      "step": 17677
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.41184288868604e-06,
+      "loss": 0.709,
+      "step": 17678
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.410768383681926e-06,
+      "loss": 0.71,
+      "step": 17679
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.409693926243539e-06,
+      "loss": 0.7012,
+      "step": 17680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.408619516385121e-06,
+      "loss": 0.7012,
+      "step": 17681
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.407545154120911e-06,
+      "loss": 0.709,
+      "step": 17682
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.406470839465142e-06,
+      "loss": 0.7188,
+      "step": 17683
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.40539657243206e-06,
+      "loss": 0.7021,
+      "step": 17684
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.404322353035895e-06,
+      "loss": 0.6992,
+      "step": 17685
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.40324818129088e-06,
+      "loss": 0.7051,
+      "step": 17686
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4021740572112565e-06,
+      "loss": 0.7109,
+      "step": 17687
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4010999808112584e-06,
+      "loss": 0.7217,
+      "step": 17688
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.400025952105116e-06,
+      "loss": 0.751,
+      "step": 17689
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.398951971107067e-06,
+      "loss": 0.7451,
+      "step": 17690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.397878037831344e-06,
+      "loss": 0.7295,
+      "step": 17691
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.396804152292173e-06,
+      "loss": 0.707,
+      "step": 17692
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.395730314503793e-06,
+      "loss": 0.71,
+      "step": 17693
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.394656524480431e-06,
+      "loss": 0.7012,
+      "step": 17694
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.393582782236316e-06,
+      "loss": 0.7305,
+      "step": 17695
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.392509087785683e-06,
+      "loss": 0.7188,
+      "step": 17696
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3914354411427586e-06,
+      "loss": 0.7363,
+      "step": 17697
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.390361842321768e-06,
+      "loss": 0.7061,
+      "step": 17698
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3892882913369395e-06,
+      "loss": 0.7246,
+      "step": 17699
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.388214788202503e-06,
+      "loss": 0.7168,
+      "step": 17700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3871413329326805e-06,
+      "loss": 0.7119,
+      "step": 17701
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.386067925541703e-06,
+      "loss": 0.7178,
+      "step": 17702
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.384994566043794e-06,
+      "loss": 0.71,
+      "step": 17703
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.383921254453182e-06,
+      "loss": 0.707,
+      "step": 17704
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.382847990784081e-06,
+      "loss": 0.6875,
+      "step": 17705
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.381774775050722e-06,
+      "loss": 0.7031,
+      "step": 17706
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.380701607267322e-06,
+      "loss": 0.7188,
+      "step": 17707
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.37962848744811e-06,
+      "loss": 0.7207,
+      "step": 17708
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3785554156073006e-06,
+      "loss": 0.707,
+      "step": 17709
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.377482391759124e-06,
+      "loss": 0.7168,
+      "step": 17710
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.376409415917788e-06,
+      "loss": 0.7246,
+      "step": 17711
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.375336488097519e-06,
+      "loss": 0.7051,
+      "step": 17712
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.374263608312534e-06,
+      "loss": 0.7432,
+      "step": 17713
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.373190776577054e-06,
+      "loss": 0.7129,
+      "step": 17714
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.372117992905293e-06,
+      "loss": 0.7148,
+      "step": 17715
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.371045257311468e-06,
+      "loss": 0.6943,
+      "step": 17716
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3699725698098015e-06,
+      "loss": 0.6973,
+      "step": 17717
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.368899930414502e-06,
+      "loss": 0.7051,
+      "step": 17718
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.367827339139784e-06,
+      "loss": 0.7354,
+      "step": 17719
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.366754795999867e-06,
+      "loss": 0.6904,
+      "step": 17720
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3656823010089616e-06,
+      "loss": 0.7402,
+      "step": 17721
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.364609854181281e-06,
+      "loss": 0.709,
+      "step": 17722
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3635374555310415e-06,
+      "loss": 0.708,
+      "step": 17723
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.362465105072448e-06,
+      "loss": 0.7256,
+      "step": 17724
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.361392802819715e-06,
+      "loss": 0.7383,
+      "step": 17725
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360320548787056e-06,
+      "loss": 0.7178,
+      "step": 17726
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.359248342988677e-06,
+      "loss": 0.7275,
+      "step": 17727
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.358176185438786e-06,
+      "loss": 0.7334,
+      "step": 17728
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.357104076151596e-06,
+      "loss": 0.707,
+      "step": 17729
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.356032015141316e-06,
+      "loss": 0.708,
+      "step": 17730
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.354960002422145e-06,
+      "loss": 0.7061,
+      "step": 17731
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.353888038008298e-06,
+      "loss": 0.7119,
+      "step": 17732
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.352816121913976e-06,
+      "loss": 0.7041,
+      "step": 17733
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.351744254153387e-06,
+      "loss": 0.7188,
+      "step": 17734
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.350672434740735e-06,
+      "loss": 0.7129,
+      "step": 17735
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.349600663690227e-06,
+      "loss": 0.7197,
+      "step": 17736
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.348528941016063e-06,
+      "loss": 0.7207,
+      "step": 17737
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.347457266732443e-06,
+      "loss": 0.71,
+      "step": 17738
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.346385640853575e-06,
+      "loss": 0.7256,
+      "step": 17739
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.345314063393657e-06,
+      "loss": 0.6914,
+      "step": 17740
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.344242534366892e-06,
+      "loss": 0.6758,
+      "step": 17741
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.343171053787478e-06,
+      "loss": 0.7012,
+      "step": 17742
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.342099621669624e-06,
+      "loss": 0.708,
+      "step": 17743
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.341028238027513e-06,
+      "loss": 0.7031,
+      "step": 17744
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3399569028753535e-06,
+      "loss": 0.7314,
+      "step": 17745
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338885616227339e-06,
+      "loss": 0.6826,
+      "step": 17746
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.337814378097672e-06,
+      "loss": 0.7207,
+      "step": 17747
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3367431885005425e-06,
+      "loss": 0.7373,
+      "step": 17748
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.335672047450157e-06,
+      "loss": 0.6992,
+      "step": 17749
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.334600954960696e-06,
+      "loss": 0.7178,
+      "step": 17750
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3335299110463625e-06,
+      "loss": 0.6885,
+      "step": 17751
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.332458915721347e-06,
+      "loss": 0.7256,
+      "step": 17752
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.331387968999847e-06,
+      "loss": 0.709,
+      "step": 17753
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.33031707089605e-06,
+      "loss": 0.7314,
+      "step": 17754
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.329246221424154e-06,
+      "loss": 0.7061,
+      "step": 17755
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.328175420598347e-06,
+      "loss": 0.707,
+      "step": 17756
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.327104668432817e-06,
+      "loss": 0.7246,
+      "step": 17757
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.326033964941757e-06,
+      "loss": 0.7119,
+      "step": 17758
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.324963310139356e-06,
+      "loss": 0.7354,
+      "step": 17759
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3238927040398024e-06,
+      "loss": 0.7266,
+      "step": 17760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.322822146657283e-06,
+      "loss": 0.7051,
+      "step": 17761
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.32175163800599e-06,
+      "loss": 0.7324,
+      "step": 17762
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.320681178100103e-06,
+      "loss": 0.7227,
+      "step": 17763
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.319610766953812e-06,
+      "loss": 0.7402,
+      "step": 17764
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.318540404581304e-06,
+      "loss": 0.7383,
+      "step": 17765
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31747009099676e-06,
+      "loss": 0.7207,
+      "step": 17766
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.316399826214366e-06,
+      "loss": 0.7285,
+      "step": 17767
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.315329610248306e-06,
+      "loss": 0.707,
+      "step": 17768
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3142594431127665e-06,
+      "loss": 0.7021,
+      "step": 17769
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.31318932482192e-06,
+      "loss": 0.7305,
+      "step": 17770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.312119255389957e-06,
+      "loss": 0.7139,
+      "step": 17771
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.311049234831054e-06,
+      "loss": 0.7559,
+      "step": 17772
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.309979263159392e-06,
+      "loss": 0.7207,
+      "step": 17773
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.308909340389151e-06,
+      "loss": 0.7285,
+      "step": 17774
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.307839466534513e-06,
+      "loss": 0.7207,
+      "step": 17775
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.306769641609651e-06,
+      "loss": 0.7285,
+      "step": 17776
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.305699865628745e-06,
+      "loss": 0.7227,
+      "step": 17777
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.304630138605974e-06,
+      "loss": 0.7197,
+      "step": 17778
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.303560460555509e-06,
+      "loss": 0.7139,
+      "step": 17779
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.302490831491532e-06,
+      "loss": 0.7119,
+      "step": 17780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.301421251428214e-06,
+      "loss": 0.7305,
+      "step": 17781
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.300351720379736e-06,
+      "loss": 0.7109,
+      "step": 17782
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29928223836026e-06,
+      "loss": 0.7188,
+      "step": 17783
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29821280538397e-06,
+      "loss": 0.7217,
+      "step": 17784
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29714342146503e-06,
+      "loss": 0.7031,
+      "step": 17785
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.296074086617618e-06,
+      "loss": 0.7119,
+      "step": 17786
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2950048008559026e-06,
+      "loss": 0.7148,
+      "step": 17787
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.293935564194061e-06,
+      "loss": 0.707,
+      "step": 17788
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.292866376646252e-06,
+      "loss": 0.7002,
+      "step": 17789
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.291797238226649e-06,
+      "loss": 0.7393,
+      "step": 17790
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.290728148949422e-06,
+      "loss": 0.6846,
+      "step": 17791
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.289659108828739e-06,
+      "loss": 0.708,
+      "step": 17792
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.288590117878765e-06,
+      "loss": 0.7002,
+      "step": 17793
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.287521176113672e-06,
+      "loss": 0.6982,
+      "step": 17794
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.286452283547623e-06,
+      "loss": 0.707,
+      "step": 17795
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.285383440194781e-06,
+      "loss": 0.6934,
+      "step": 17796
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.284314646069312e-06,
+      "loss": 0.7178,
+      "step": 17797
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.28324590118538e-06,
+      "loss": 0.7109,
+      "step": 17798
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.282177205557149e-06,
+      "loss": 0.6973,
+      "step": 17799
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.281108559198781e-06,
+      "loss": 0.6963,
+      "step": 17800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.280039962124443e-06,
+      "loss": 0.7256,
+      "step": 17801
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.278971414348288e-06,
+      "loss": 0.7031,
+      "step": 17802
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.27790291588448e-06,
+      "loss": 0.7129,
+      "step": 17803
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.276834466747181e-06,
+      "loss": 0.6992,
+      "step": 17804
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.27576606695055e-06,
+      "loss": 0.7188,
+      "step": 17805
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.274697716508744e-06,
+      "loss": 0.7363,
+      "step": 17806
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.273629415435923e-06,
+      "loss": 0.7236,
+      "step": 17807
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.272561163746247e-06,
+      "loss": 0.7178,
+      "step": 17808
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.271492961453864e-06,
+      "loss": 0.6992,
+      "step": 17809
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.270424808572939e-06,
+      "loss": 0.7334,
+      "step": 17810
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.269356705117624e-06,
+      "loss": 0.7031,
+      "step": 17811
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.268288651102073e-06,
+      "loss": 0.6924,
+      "step": 17812
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.267220646540441e-06,
+      "loss": 0.7334,
+      "step": 17813
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.266152691446886e-06,
+      "loss": 0.6963,
+      "step": 17814
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.265084785835552e-06,
+      "loss": 0.707,
+      "step": 17815
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2640169297205984e-06,
+      "loss": 0.751,
+      "step": 17816
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.262949123116175e-06,
+      "loss": 0.7041,
+      "step": 17817
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2618813660364306e-06,
+      "loss": 0.7012,
+      "step": 17818
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.260813658495518e-06,
+      "loss": 0.7109,
+      "step": 17819
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.259746000507585e-06,
+      "loss": 0.7373,
+      "step": 17820
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.258678392086786e-06,
+      "loss": 0.7402,
+      "step": 17821
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.257610833247261e-06,
+      "loss": 0.751,
+      "step": 17822
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.256543324003162e-06,
+      "loss": 0.707,
+      "step": 17823
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.255475864368634e-06,
+      "loss": 0.7402,
+      "step": 17824
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2544084543578275e-06,
+      "loss": 0.7314,
+      "step": 17825
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.253341093984882e-06,
+      "loss": 0.7158,
+      "step": 17826
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.252273783263954e-06,
+      "loss": 0.6982,
+      "step": 17827
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.251206522209172e-06,
+      "loss": 0.7119,
+      "step": 17828
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.250139310834691e-06,
+      "loss": 0.6953,
+      "step": 17829
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.249072149154649e-06,
+      "loss": 0.6953,
+      "step": 17830
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.248005037183191e-06,
+      "loss": 0.707,
+      "step": 17831
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.246937974934457e-06,
+      "loss": 0.7461,
+      "step": 17832
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24587096242259e-06,
+      "loss": 0.6992,
+      "step": 17833
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2448039996617315e-06,
+      "loss": 0.7178,
+      "step": 17834
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.243737086666017e-06,
+      "loss": 0.6914,
+      "step": 17835
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.242670223449585e-06,
+      "loss": 0.7041,
+      "step": 17836
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24160341002658e-06,
+      "loss": 0.7207,
+      "step": 17837
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.240536646411134e-06,
+      "loss": 0.6982,
+      "step": 17838
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.239469932617388e-06,
+      "loss": 0.7158,
+      "step": 17839
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.238403268659479e-06,
+      "loss": 0.7412,
+      "step": 17840
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.23733665455154e-06,
+      "loss": 0.7188,
+      "step": 17841
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.236270090307703e-06,
+      "loss": 0.6982,
+      "step": 17842
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.235203575942111e-06,
+      "loss": 0.7275,
+      "step": 17843
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.23413711146889e-06,
+      "loss": 0.7178,
+      "step": 17844
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2330706969021774e-06,
+      "loss": 0.7295,
+      "step": 17845
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2320043322561045e-06,
+      "loss": 0.7178,
+      "step": 17846
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.230938017544805e-06,
+      "loss": 0.7012,
+      "step": 17847
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.229871752782406e-06,
+      "loss": 0.7285,
+      "step": 17848
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2288055379830415e-06,
+      "loss": 0.7207,
+      "step": 17849
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.227739373160839e-06,
+      "loss": 0.7129,
+      "step": 17850
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.22667325832993e-06,
+      "loss": 0.748,
+      "step": 17851
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2256071935044415e-06,
+      "loss": 0.7461,
+      "step": 17852
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.224541178698504e-06,
+      "loss": 0.7256,
+      "step": 17853
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.223475213926238e-06,
+      "loss": 0.7422,
+      "step": 17854
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.222409299201777e-06,
+      "loss": 0.7178,
+      "step": 17855
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.221343434539246e-06,
+      "loss": 0.7461,
+      "step": 17856
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.220277619952764e-06,
+      "loss": 0.6992,
+      "step": 17857
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.219211855456464e-06,
+      "loss": 0.7031,
+      "step": 17858
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.218146141064463e-06,
+      "loss": 0.7295,
+      "step": 17859
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.217080476790891e-06,
+      "loss": 0.708,
+      "step": 17860
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.216014862649864e-06,
+      "loss": 0.7266,
+      "step": 17861
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.214949298655508e-06,
+      "loss": 0.7197,
+      "step": 17862
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.213883784821942e-06,
+      "loss": 0.7207,
+      "step": 17863
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.212818321163287e-06,
+      "loss": 0.7324,
+      "step": 17864
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.211752907693662e-06,
+      "loss": 0.7256,
+      "step": 17865
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.210687544427193e-06,
+      "loss": 0.6992,
+      "step": 17866
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.209622231377987e-06,
+      "loss": 0.709,
+      "step": 17867
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.208556968560171e-06,
+      "loss": 0.6992,
+      "step": 17868
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.207491755987857e-06,
+      "loss": 0.6924,
+      "step": 17869
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2064265936751655e-06,
+      "loss": 0.7217,
+      "step": 17870
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.205361481636208e-06,
+      "loss": 0.7168,
+      "step": 17871
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.204296419885105e-06,
+      "loss": 0.7324,
+      "step": 17872
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.203231408435971e-06,
+      "loss": 0.7217,
+      "step": 17873
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2021664473029155e-06,
+      "loss": 0.7285,
+      "step": 17874
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.2011015365000504e-06,
+      "loss": 0.7178,
+      "step": 17875
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.200036676041494e-06,
+      "loss": 0.7012,
+      "step": 17876
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.198971865941352e-06,
+      "loss": 0.708,
+      "step": 17877
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.197907106213743e-06,
+      "loss": 0.7256,
+      "step": 17878
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1968423968727766e-06,
+      "loss": 0.7188,
+      "step": 17879
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.195777737932557e-06,
+      "loss": 0.7207,
+      "step": 17880
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.194713129407194e-06,
+      "loss": 0.7402,
+      "step": 17881
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1936485713108e-06,
+      "loss": 0.7139,
+      "step": 17882
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.192584063657481e-06,
+      "loss": 0.7314,
+      "step": 17883
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.191519606461345e-06,
+      "loss": 0.6875,
+      "step": 17884
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.190455199736498e-06,
+      "loss": 0.7119,
+      "step": 17885
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1893908434970494e-06,
+      "loss": 0.7549,
+      "step": 17886
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.188326537757096e-06,
+      "loss": 0.708,
+      "step": 17887
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.18726228253075e-06,
+      "loss": 0.7041,
+      "step": 17888
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.186198077832113e-06,
+      "loss": 0.7295,
+      "step": 17889
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.185133923675285e-06,
+      "loss": 0.7031,
+      "step": 17890
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.184069820074375e-06,
+      "loss": 0.7295,
+      "step": 17891
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.183005767043481e-06,
+      "loss": 0.7217,
+      "step": 17892
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.181941764596702e-06,
+      "loss": 0.7236,
+      "step": 17893
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.180877812748142e-06,
+      "loss": 0.709,
+      "step": 17894
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.179813911511901e-06,
+      "loss": 0.7402,
+      "step": 17895
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.178750060902074e-06,
+      "loss": 0.7578,
+      "step": 17896
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.177686260932764e-06,
+      "loss": 0.7217,
+      "step": 17897
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.176622511618067e-06,
+      "loss": 0.7041,
+      "step": 17898
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1755588129720845e-06,
+      "loss": 0.7158,
+      "step": 17899
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.174495165008906e-06,
+      "loss": 0.7188,
+      "step": 17900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.17343156774263e-06,
+      "loss": 0.707,
+      "step": 17901
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.172368021187351e-06,
+      "loss": 0.7178,
+      "step": 17902
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.171304525357167e-06,
+      "loss": 0.6963,
+      "step": 17903
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.170241080266166e-06,
+      "loss": 0.7119,
+      "step": 17904
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.169177685928449e-06,
+      "loss": 0.7344,
+      "step": 17905
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.168114342358103e-06,
+      "loss": 0.7158,
+      "step": 17906
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.167051049569219e-06,
+      "loss": 0.7002,
+      "step": 17907
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1659878075758895e-06,
+      "loss": 0.708,
+      "step": 17908
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.164924616392207e-06,
+      "loss": 0.7109,
+      "step": 17909
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.163861476032257e-06,
+      "loss": 0.7422,
+      "step": 17910
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.162798386510137e-06,
+      "loss": 0.7539,
+      "step": 17911
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.161735347839923e-06,
+      "loss": 0.71,
+      "step": 17912
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.160672360035713e-06,
+      "loss": 0.7295,
+      "step": 17913
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1596094231115875e-06,
+      "loss": 0.7441,
+      "step": 17914
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.158546537081639e-06,
+      "loss": 0.7158,
+      "step": 17915
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.157483701959947e-06,
+      "loss": 0.7021,
+      "step": 17916
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.156420917760601e-06,
+      "loss": 0.7041,
+      "step": 17917
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.155358184497687e-06,
+      "loss": 0.7227,
+      "step": 17918
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.154295502185284e-06,
+      "loss": 0.7168,
+      "step": 17919
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.153232870837475e-06,
+      "loss": 0.7295,
+      "step": 17920
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.152170290468345e-06,
+      "loss": 0.7393,
+      "step": 17921
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1511077610919725e-06,
+      "loss": 0.6973,
+      "step": 17922
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.150045282722443e-06,
+      "loss": 0.7451,
+      "step": 17923
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.148982855373837e-06,
+      "loss": 0.7295,
+      "step": 17924
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.147920479060231e-06,
+      "loss": 0.708,
+      "step": 17925
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.146858153795701e-06,
+      "loss": 0.7441,
+      "step": 17926
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.145795879594332e-06,
+      "loss": 0.7119,
+      "step": 17927
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.144733656470199e-06,
+      "loss": 0.7559,
+      "step": 17928
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.143671484437374e-06,
+      "loss": 0.7129,
+      "step": 17929
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.142609363509944e-06,
+      "loss": 0.7119,
+      "step": 17930
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.141547293701977e-06,
+      "loss": 0.7002,
+      "step": 17931
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.140485275027548e-06,
+      "loss": 0.7305,
+      "step": 17932
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.139423307500733e-06,
+      "loss": 0.7256,
+      "step": 17933
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1383613911356055e-06,
+      "loss": 0.7178,
+      "step": 17934
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.137299525946235e-06,
+      "loss": 0.751,
+      "step": 17935
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1362377119467e-06,
+      "loss": 0.6982,
+      "step": 17936
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.135175949151071e-06,
+      "loss": 0.7324,
+      "step": 17937
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.134114237573412e-06,
+      "loss": 0.707,
+      "step": 17938
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1330525772278e-06,
+      "loss": 0.7051,
+      "step": 17939
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.131990968128303e-06,
+      "loss": 0.7305,
+      "step": 17940
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.130929410288986e-06,
+      "loss": 0.7061,
+      "step": 17941
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.129867903723923e-06,
+      "loss": 0.7188,
+      "step": 17942
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.128806448447175e-06,
+      "loss": 0.7188,
+      "step": 17943
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.127745044472818e-06,
+      "loss": 0.6953,
+      "step": 17944
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1266836918149104e-06,
+      "loss": 0.707,
+      "step": 17945
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.12562239048752e-06,
+      "loss": 0.7227,
+      "step": 17946
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.124561140504708e-06,
+      "loss": 0.7314,
+      "step": 17947
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.123499941880545e-06,
+      "loss": 0.7285,
+      "step": 17948
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.122438794629089e-06,
+      "loss": 0.7188,
+      "step": 17949
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1213776987644105e-06,
+      "loss": 0.7344,
+      "step": 17950
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.12031665430056e-06,
+      "loss": 0.7188,
+      "step": 17951
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.119255661251606e-06,
+      "loss": 0.708,
+      "step": 17952
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.118194719631606e-06,
+      "loss": 0.7256,
+      "step": 17953
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.117133829454624e-06,
+      "loss": 0.7207,
+      "step": 17954
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.116072990734714e-06,
+      "loss": 0.7412,
+      "step": 17955
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1150122034859405e-06,
+      "loss": 0.6953,
+      "step": 17956
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113951467722361e-06,
+      "loss": 0.7412,
+      "step": 17957
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.112890783458028e-06,
+      "loss": 0.7295,
+      "step": 17958
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.111830150706996e-06,
+      "loss": 0.7256,
+      "step": 17959
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.110769569483331e-06,
+      "loss": 0.7061,
+      "step": 17960
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.109709039801078e-06,
+      "loss": 0.7256,
+      "step": 17961
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.108648561674297e-06,
+      "loss": 0.7217,
+      "step": 17962
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.107588135117044e-06,
+      "loss": 0.7148,
+      "step": 17963
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.106527760143368e-06,
+      "loss": 0.7158,
+      "step": 17964
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.105467436767318e-06,
+      "loss": 0.7256,
+      "step": 17965
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.104407165002954e-06,
+      "loss": 0.7109,
+      "step": 17966
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.103346944864319e-06,
+      "loss": 0.7012,
+      "step": 17967
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.102286776365471e-06,
+      "loss": 0.7012,
+      "step": 17968
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.101226659520456e-06,
+      "loss": 0.7246,
+      "step": 17969
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1001665943433255e-06,
+      "loss": 0.6982,
+      "step": 17970
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.099106580848122e-06,
+      "loss": 0.71,
+      "step": 17971
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.098046619048897e-06,
+      "loss": 0.7441,
+      "step": 17972
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0969867089596975e-06,
+      "loss": 0.6895,
+      "step": 17973
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.095926850594569e-06,
+      "loss": 0.7012,
+      "step": 17974
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.09486704396756e-06,
+      "loss": 0.7314,
+      "step": 17975
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.093807289092713e-06,
+      "loss": 0.707,
+      "step": 17976
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0927475859840704e-06,
+      "loss": 0.6895,
+      "step": 17977
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.091687934655679e-06,
+      "loss": 0.7109,
+      "step": 17978
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.09062833512158e-06,
+      "loss": 0.709,
+      "step": 17979
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.089568787395813e-06,
+      "loss": 0.709,
+      "step": 17980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.088509291492426e-06,
+      "loss": 0.7393,
+      "step": 17981
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.087449847425453e-06,
+      "loss": 0.6914,
+      "step": 17982
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0863904552089415e-06,
+      "loss": 0.7188,
+      "step": 17983
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.085331114856925e-06,
+      "loss": 0.7178,
+      "step": 17984
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.084271826383443e-06,
+      "loss": 0.7324,
+      "step": 17985
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.083212589802534e-06,
+      "loss": 0.7148,
+      "step": 17986
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.082153405128237e-06,
+      "loss": 0.7012,
+      "step": 17987
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.081094272374587e-06,
+      "loss": 0.7188,
+      "step": 17988
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0800351915556245e-06,
+      "loss": 0.71,
+      "step": 17989
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.078976162685377e-06,
+      "loss": 0.7168,
+      "step": 17990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.077917185777884e-06,
+      "loss": 0.6904,
+      "step": 17991
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.076858260847176e-06,
+      "loss": 0.7002,
+      "step": 17992
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.075799387907291e-06,
+      "loss": 0.7002,
+      "step": 17993
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.074740566972257e-06,
+      "loss": 0.6953,
+      "step": 17994
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07368179805611e-06,
+      "loss": 0.6748,
+      "step": 17995
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.072623081172881e-06,
+      "loss": 0.7393,
+      "step": 17996
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.071564416336598e-06,
+      "loss": 0.7363,
+      "step": 17997
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.070505803561289e-06,
+      "loss": 0.7256,
+      "step": 17998
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.069447242860988e-06,
+      "loss": 0.7471,
+      "step": 17999
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.068388734249717e-06,
+      "loss": 0.7295,
+      "step": 18000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.067330277741512e-06,
+      "loss": 0.709,
+      "step": 18001
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.066271873350395e-06,
+      "loss": 0.7266,
+      "step": 18002
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.065213521090394e-06,
+      "loss": 0.708,
+      "step": 18003
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0641552209755305e-06,
+      "loss": 0.707,
+      "step": 18004
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.063096973019835e-06,
+      "loss": 0.7383,
+      "step": 18005
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.062038777237327e-06,
+      "loss": 0.7246,
+      "step": 18006
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.060980633642035e-06,
+      "loss": 0.7305,
+      "step": 18007
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.059922542247978e-06,
+      "loss": 0.7246,
+      "step": 18008
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.058864503069184e-06,
+      "loss": 0.7295,
+      "step": 18009
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.057806516119664e-06,
+      "loss": 0.6846,
+      "step": 18010
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.056748581413448e-06,
+      "loss": 0.7568,
+      "step": 18011
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.055690698964551e-06,
+      "loss": 0.71,
+      "step": 18012
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.054632868786996e-06,
+      "loss": 0.7158,
+      "step": 18013
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.053575090894801e-06,
+      "loss": 0.6846,
+      "step": 18014
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0525173653019845e-06,
+      "loss": 0.7119,
+      "step": 18015
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0514596920225585e-06,
+      "loss": 0.7031,
+      "step": 18016
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.050402071070546e-06,
+      "loss": 0.7402,
+      "step": 18017
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.04934450245996e-06,
+      "loss": 0.7295,
+      "step": 18018
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0482869862048145e-06,
+      "loss": 0.7246,
+      "step": 18019
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0472295223191294e-06,
+      "loss": 0.7266,
+      "step": 18020
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.046172110816911e-06,
+      "loss": 0.71,
+      "step": 18021
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0451147517121825e-06,
+      "loss": 0.7031,
+      "step": 18022
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.044057445018947e-06,
+      "loss": 0.6885,
+      "step": 18023
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.043000190751221e-06,
+      "loss": 0.7324,
+      "step": 18024
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.041942988923013e-06,
+      "loss": 0.7305,
+      "step": 18025
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.040885839548336e-06,
+      "loss": 0.7197,
+      "step": 18026
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.039828742641196e-06,
+      "loss": 0.7129,
+      "step": 18027
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.038771698215609e-06,
+      "loss": 0.7217,
+      "step": 18028
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.037714706285576e-06,
+      "loss": 0.7021,
+      "step": 18029
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.036657766865108e-06,
+      "loss": 0.7217,
+      "step": 18030
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.035600879968211e-06,
+      "loss": 0.7109,
+      "step": 18031
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.034544045608892e-06,
+      "loss": 0.7061,
+      "step": 18032
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.033487263801154e-06,
+      "loss": 0.7148,
+      "step": 18033
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.032430534559006e-06,
+      "loss": 0.709,
+      "step": 18034
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.031373857896452e-06,
+      "loss": 0.7002,
+      "step": 18035
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0303172338274926e-06,
+      "loss": 0.7129,
+      "step": 18036
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.02926066236613e-06,
+      "loss": 0.7236,
+      "step": 18037
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.028204143526368e-06,
+      "loss": 0.7178,
+      "step": 18038
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.027147677322206e-06,
+      "loss": 0.7373,
+      "step": 18039
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.026091263767648e-06,
+      "loss": 0.707,
+      "step": 18040
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0250349028766966e-06,
+      "loss": 0.71,
+      "step": 18041
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.023978594663342e-06,
+      "loss": 0.7168,
+      "step": 18042
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.022922339141588e-06,
+      "loss": 0.707,
+      "step": 18043
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0218661363254325e-06,
+      "loss": 0.6982,
+      "step": 18044
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0208099862288705e-06,
+      "loss": 0.7021,
+      "step": 18045
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.019753888865902e-06,
+      "loss": 0.7119,
+      "step": 18046
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018697844250521e-06,
+      "loss": 0.7119,
+      "step": 18047
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.017641852396725e-06,
+      "loss": 0.707,
+      "step": 18048
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0165859133185e-06,
+      "loss": 0.7246,
+      "step": 18049
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.01553002702985e-06,
+      "loss": 0.7168,
+      "step": 18050
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0144741935447596e-06,
+      "loss": 0.7344,
+      "step": 18051
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.013418412877227e-06,
+      "loss": 0.6797,
+      "step": 18052
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.012362685041242e-06,
+      "loss": 0.7334,
+      "step": 18053
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.011307010050798e-06,
+      "loss": 0.7197,
+      "step": 18054
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.010251387919878e-06,
+      "loss": 0.7178,
+      "step": 18055
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.009195818662476e-06,
+      "loss": 0.7334,
+      "step": 18056
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.008140302292582e-06,
+      "loss": 0.7344,
+      "step": 18057
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0070848388241824e-06,
+      "loss": 0.7148,
+      "step": 18058
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.006029428271264e-06,
+      "loss": 0.6982,
+      "step": 18059
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.004974070647814e-06,
+      "loss": 0.7197,
+      "step": 18060
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.003918765967821e-06,
+      "loss": 0.7012,
+      "step": 18061
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.002863514245267e-06,
+      "loss": 0.7197,
+      "step": 18062
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0018083154941375e-06,
+      "loss": 0.752,
+      "step": 18063
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.000753169728413e-06,
+      "loss": 0.708,
+      "step": 18064
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9996980769620835e-06,
+      "loss": 0.7256,
+      "step": 18065
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.998643037209125e-06,
+      "loss": 0.7451,
+      "step": 18066
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.997588050483526e-06,
+      "loss": 0.7168,
+      "step": 18067
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9965331167992615e-06,
+      "loss": 0.7129,
+      "step": 18068
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.995478236170315e-06,
+      "loss": 0.6934,
+      "step": 18069
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.994423408610663e-06,
+      "loss": 0.7217,
+      "step": 18070
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.993368634134289e-06,
+      "loss": 0.709,
+      "step": 18071
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.992313912755165e-06,
+      "loss": 0.7051,
+      "step": 18072
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.991259244487275e-06,
+      "loss": 0.6963,
+      "step": 18073
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990204629344596e-06,
+      "loss": 0.7158,
+      "step": 18074
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9891500673411e-06,
+      "loss": 0.7324,
+      "step": 18075
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98809555849076e-06,
+      "loss": 0.7129,
+      "step": 18076
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9870411028075585e-06,
+      "loss": 0.7373,
+      "step": 18077
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.985986700305462e-06,
+      "loss": 0.7305,
+      "step": 18078
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.984932350998449e-06,
+      "loss": 0.7119,
+      "step": 18079
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9838780549004915e-06,
+      "loss": 0.7334,
+      "step": 18080
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98282381202556e-06,
+      "loss": 0.6963,
+      "step": 18081
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.981769622387623e-06,
+      "loss": 0.7295,
+      "step": 18082
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.980715486000656e-06,
+      "loss": 0.7207,
+      "step": 18083
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.979661402878626e-06,
+      "loss": 0.7314,
+      "step": 18084
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.978607373035504e-06,
+      "loss": 0.7129,
+      "step": 18085
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.977553396485256e-06,
+      "loss": 0.7197,
+      "step": 18086
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9764994732418546e-06,
+      "loss": 0.6982,
+      "step": 18087
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.975445603319258e-06,
+      "loss": 0.7061,
+      "step": 18088
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.97439178673144e-06,
+      "loss": 0.7031,
+      "step": 18089
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9733380234923595e-06,
+      "loss": 0.6973,
+      "step": 18090
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.972284313615989e-06,
+      "loss": 0.6973,
+      "step": 18091
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.971230657116289e-06,
+      "loss": 0.7109,
+      "step": 18092
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.970177054007224e-06,
+      "loss": 0.7207,
+      "step": 18093
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.969123504302754e-06,
+      "loss": 0.7129,
+      "step": 18094
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.968070008016842e-06,
+      "loss": 0.7334,
+      "step": 18095
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.967016565163448e-06,
+      "loss": 0.707,
+      "step": 18096
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9659631757565386e-06,
+      "loss": 0.6914,
+      "step": 18097
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.964909839810068e-06,
+      "loss": 0.6943,
+      "step": 18098
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9638565573379946e-06,
+      "loss": 0.7256,
+      "step": 18099
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.962803328354285e-06,
+      "loss": 0.709,
+      "step": 18100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.961750152872887e-06,
+      "loss": 0.7236,
+      "step": 18101
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.960697030907763e-06,
+      "loss": 0.71,
+      "step": 18102
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.959643962472866e-06,
+      "loss": 0.71,
+      "step": 18103
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.958590947582156e-06,
+      "loss": 0.7207,
+      "step": 18104
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.957537986249584e-06,
+      "loss": 0.71,
+      "step": 18105
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.956485078489109e-06,
+      "loss": 0.7246,
+      "step": 18106
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.955432224314679e-06,
+      "loss": 0.7002,
+      "step": 18107
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.95437942374025e-06,
+      "loss": 0.71,
+      "step": 18108
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.953326676779772e-06,
+      "loss": 0.7129,
+      "step": 18109
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.952273983447198e-06,
+      "loss": 0.6992,
+      "step": 18110
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.951221343756477e-06,
+      "loss": 0.7178,
+      "step": 18111
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.950168757721561e-06,
+      "loss": 0.7002,
+      "step": 18112
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.949116225356402e-06,
+      "loss": 0.7295,
+      "step": 18113
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9480637466749415e-06,
+      "loss": 0.7031,
+      "step": 18114
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.94701132169113e-06,
+      "loss": 0.7051,
+      "step": 18115
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.945958950418918e-06,
+      "loss": 0.7188,
+      "step": 18116
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.944906632872245e-06,
+      "loss": 0.7168,
+      "step": 18117
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.943854369065066e-06,
+      "loss": 0.7168,
+      "step": 18118
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.94280215901132e-06,
+      "loss": 0.6846,
+      "step": 18119
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.941750002724951e-06,
+      "loss": 0.7227,
+      "step": 18120
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.940697900219902e-06,
+      "loss": 0.7256,
+      "step": 18121
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.939645851510118e-06,
+      "loss": 0.7188,
+      "step": 18122
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93859385660954e-06,
+      "loss": 0.71,
+      "step": 18123
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.937541915532112e-06,
+      "loss": 0.7148,
+      "step": 18124
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9364900282917746e-06,
+      "loss": 0.6973,
+      "step": 18125
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.935438194902463e-06,
+      "loss": 0.7139,
+      "step": 18126
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.93438641537812e-06,
+      "loss": 0.709,
+      "step": 18127
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.933334689732681e-06,
+      "loss": 0.7207,
+      "step": 18128
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.932283017980087e-06,
+      "loss": 0.7041,
+      "step": 18129
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.931231400134276e-06,
+      "loss": 0.7041,
+      "step": 18130
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.930179836209185e-06,
+      "loss": 0.7373,
+      "step": 18131
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9291283262187464e-06,
+      "loss": 0.7148,
+      "step": 18132
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.928076870176893e-06,
+      "loss": 0.7217,
+      "step": 18133
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9270254680975645e-06,
+      "loss": 0.7002,
+      "step": 18134
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.925974119994691e-06,
+      "loss": 0.71,
+      "step": 18135
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.924922825882207e-06,
+      "loss": 0.7451,
+      "step": 18136
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.923871585774045e-06,
+      "loss": 0.6963,
+      "step": 18137
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.922820399684137e-06,
+      "loss": 0.7285,
+      "step": 18138
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.921769267626409e-06,
+      "loss": 0.7236,
+      "step": 18139
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.920718189614796e-06,
+      "loss": 0.7031,
+      "step": 18140
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.919667165663226e-06,
+      "loss": 0.7178,
+      "step": 18141
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.918616195785624e-06,
+      "loss": 0.7197,
+      "step": 18142
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9175652799959225e-06,
+      "loss": 0.6865,
+      "step": 18143
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.916514418308046e-06,
+      "loss": 0.7158,
+      "step": 18144
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.915463610735925e-06,
+      "loss": 0.7285,
+      "step": 18145
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.914412857293481e-06,
+      "loss": 0.7158,
+      "step": 18146
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9133621579946384e-06,
+      "loss": 0.7324,
+      "step": 18147
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.912311512853322e-06,
+      "loss": 0.7227,
+      "step": 18148
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9112609218834575e-06,
+      "loss": 0.7168,
+      "step": 18149
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.910210385098964e-06,
+      "loss": 0.7471,
+      "step": 18150
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.909159902513773e-06,
+      "loss": 0.7275,
+      "step": 18151
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.908109474141795e-06,
+      "loss": 0.6943,
+      "step": 18152
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.907059099996952e-06,
+      "loss": 0.7148,
+      "step": 18153
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.906008780093168e-06,
+      "loss": 0.7109,
+      "step": 18154
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.904958514444361e-06,
+      "loss": 0.707,
+      "step": 18155
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.903908303064447e-06,
+      "loss": 0.7363,
+      "step": 18156
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.902858145967347e-06,
+      "loss": 0.7324,
+      "step": 18157
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.901808043166981e-06,
+      "loss": 0.7041,
+      "step": 18158
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.900757994677257e-06,
+      "loss": 0.7178,
+      "step": 18159
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.899708000512095e-06,
+      "loss": 0.7334,
+      "step": 18160
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8986580606854104e-06,
+      "loss": 0.7158,
+      "step": 18161
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.897608175211115e-06,
+      "loss": 0.7158,
+      "step": 18162
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.896558344103125e-06,
+      "loss": 0.7021,
+      "step": 18163
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.895508567375355e-06,
+      "loss": 0.709,
+      "step": 18164
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.894458845041713e-06,
+      "loss": 0.7148,
+      "step": 18165
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.893409177116107e-06,
+      "loss": 0.7119,
+      "step": 18166
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.892359563612455e-06,
+      "loss": 0.7412,
+      "step": 18167
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.891310004544662e-06,
+      "loss": 0.7031,
+      "step": 18168
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8902604999266415e-06,
+      "loss": 0.7246,
+      "step": 18169
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8892110497722976e-06,
+      "loss": 0.7158,
+      "step": 18170
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.888161654095543e-06,
+      "loss": 0.709,
+      "step": 18171
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.887112312910275e-06,
+      "loss": 0.7178,
+      "step": 18172
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8860630262304105e-06,
+      "loss": 0.6943,
+      "step": 18173
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8850137940698474e-06,
+      "loss": 0.6875,
+      "step": 18174
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.883964616442495e-06,
+      "loss": 0.6924,
+      "step": 18175
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.882915493362258e-06,
+      "loss": 0.7158,
+      "step": 18176
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.881866424843038e-06,
+      "loss": 0.6924,
+      "step": 18177
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.880817410898734e-06,
+      "loss": 0.708,
+      "step": 18178
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.879768451543253e-06,
+      "loss": 0.7148,
+      "step": 18179
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.878719546790492e-06,
+      "loss": 0.7158,
+      "step": 18180
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.877670696654358e-06,
+      "loss": 0.7217,
+      "step": 18181
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.876621901148744e-06,
+      "loss": 0.7168,
+      "step": 18182
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.875573160287551e-06,
+      "loss": 0.7324,
+      "step": 18183
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.874524474084682e-06,
+      "loss": 0.7324,
+      "step": 18184
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.873475842554028e-06,
+      "loss": 0.7246,
+      "step": 18185
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.872427265709488e-06,
+      "loss": 0.709,
+      "step": 18186
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.871378743564957e-06,
+      "loss": 0.7266,
+      "step": 18187
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.870330276134334e-06,
+      "loss": 0.7051,
+      "step": 18188
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.869281863431509e-06,
+      "loss": 0.7051,
+      "step": 18189
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.868233505470382e-06,
+      "loss": 0.7285,
+      "step": 18190
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.86718520226484e-06,
+      "loss": 0.7207,
+      "step": 18191
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.866136953828778e-06,
+      "loss": 0.7109,
+      "step": 18192
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.865088760176088e-06,
+      "loss": 0.6992,
+      "step": 18193
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.86404062132066e-06,
+      "loss": 0.6973,
+      "step": 18194
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.862992537276384e-06,
+      "loss": 0.7051,
+      "step": 18195
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.861944508057153e-06,
+      "loss": 0.71,
+      "step": 18196
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.860896533676856e-06,
+      "loss": 0.7207,
+      "step": 18197
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.859848614149375e-06,
+      "loss": 0.748,
+      "step": 18198
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8588007494886e-06,
+      "loss": 0.7266,
+      "step": 18199
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.85775293970842e-06,
+      "loss": 0.7217,
+      "step": 18200
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8567051848227175e-06,
+      "loss": 0.7295,
+      "step": 18201
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.855657484845383e-06,
+      "loss": 0.7432,
+      "step": 18202
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.854609839790298e-06,
+      "loss": 0.7139,
+      "step": 18203
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8535622496713455e-06,
+      "loss": 0.7207,
+      "step": 18204
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.852514714502407e-06,
+      "loss": 0.7373,
+      "step": 18205
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.851467234297368e-06,
+      "loss": 0.71,
+      "step": 18206
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.850419809070107e-06,
+      "loss": 0.7139,
+      "step": 18207
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.849372438834508e-06,
+      "loss": 0.7148,
+      "step": 18208
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8483251236044505e-06,
+      "loss": 0.7148,
+      "step": 18209
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.847277863393815e-06,
+      "loss": 0.7363,
+      "step": 18210
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8462306582164754e-06,
+      "loss": 0.7363,
+      "step": 18211
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.845183508086314e-06,
+      "loss": 0.709,
+      "step": 18212
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8441364130172044e-06,
+      "loss": 0.7227,
+      "step": 18213
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.843089373023028e-06,
+      "loss": 0.7236,
+      "step": 18214
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.842042388117657e-06,
+      "loss": 0.7363,
+      "step": 18215
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.840995458314968e-06,
+      "loss": 0.7109,
+      "step": 18216
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.839948583628832e-06,
+      "loss": 0.7061,
+      "step": 18217
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.838901764073126e-06,
+      "loss": 0.6973,
+      "step": 18218
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83785499966172e-06,
+      "loss": 0.6904,
+      "step": 18219
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.836808290408491e-06,
+      "loss": 0.7129,
+      "step": 18220
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.835761636327307e-06,
+      "loss": 0.7021,
+      "step": 18221
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83471503743204e-06,
+      "loss": 0.7051,
+      "step": 18222
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.833668493736556e-06,
+      "loss": 0.7129,
+      "step": 18223
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.832622005254728e-06,
+      "loss": 0.7197,
+      "step": 18224
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.831575572000423e-06,
+      "loss": 0.7217,
+      "step": 18225
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83052919398751e-06,
+      "loss": 0.7217,
+      "step": 18226
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.82948287122985e-06,
+      "loss": 0.752,
+      "step": 18227
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.828436603741319e-06,
+      "loss": 0.7031,
+      "step": 18228
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.827390391535781e-06,
+      "loss": 0.7207,
+      "step": 18229
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8263442346270904e-06,
+      "loss": 0.7393,
+      "step": 18230
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.825298133029122e-06,
+      "loss": 0.7021,
+      "step": 18231
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.824252086755735e-06,
+      "loss": 0.6992,
+      "step": 18232
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.823206095820793e-06,
+      "loss": 0.7188,
+      "step": 18233
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.822160160238155e-06,
+      "loss": 0.7285,
+      "step": 18234
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8211142800216825e-06,
+      "loss": 0.7051,
+      "step": 18235
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.820068455185245e-06,
+      "loss": 0.6953,
+      "step": 18236
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.81902268574269e-06,
+      "loss": 0.7148,
+      "step": 18237
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.817976971707883e-06,
+      "loss": 0.7031,
+      "step": 18238
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.816931313094676e-06,
+      "loss": 0.7246,
+      "step": 18239
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.815885709916933e-06,
+      "loss": 0.7158,
+      "step": 18240
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.81484016218851e-06,
+      "loss": 0.7012,
+      "step": 18241
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.813794669923265e-06,
+      "loss": 0.7109,
+      "step": 18242
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.812749233135041e-06,
+      "loss": 0.7471,
+      "step": 18243
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8117038518377066e-06,
+      "loss": 0.7041,
+      "step": 18244
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8106585260451085e-06,
+      "loss": 0.7236,
+      "step": 18245
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.809613255771103e-06,
+      "loss": 0.7002,
+      "step": 18246
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8085680410295345e-06,
+      "loss": 0.709,
+      "step": 18247
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.807522881834267e-06,
+      "loss": 0.6992,
+      "step": 18248
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.806477778199147e-06,
+      "loss": 0.7295,
+      "step": 18249
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.805432730138018e-06,
+      "loss": 0.6992,
+      "step": 18250
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8043877376647326e-06,
+      "loss": 0.7354,
+      "step": 18251
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.803342800793143e-06,
+      "loss": 0.6934,
+      "step": 18252
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802297919537097e-06,
+      "loss": 0.7373,
+      "step": 18253
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8012530939104374e-06,
+      "loss": 0.7275,
+      "step": 18254
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.800208323927012e-06,
+      "loss": 0.7246,
+      "step": 18255
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.799163609600669e-06,
+      "loss": 0.7461,
+      "step": 18256
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.798118950945251e-06,
+      "loss": 0.7334,
+      "step": 18257
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.797074347974601e-06,
+      "loss": 0.7256,
+      "step": 18258
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.796029800702566e-06,
+      "loss": 0.7119,
+      "step": 18259
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.794985309142982e-06,
+      "loss": 0.7324,
+      "step": 18260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7939408733097e-06,
+      "loss": 0.7266,
+      "step": 18261
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.792896493216561e-06,
+      "loss": 0.7305,
+      "step": 18262
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7918521688773945e-06,
+      "loss": 0.7246,
+      "step": 18263
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.790807900306049e-06,
+      "loss": 0.709,
+      "step": 18264
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.789763687516363e-06,
+      "loss": 0.7285,
+      "step": 18265
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7887195305221734e-06,
+      "loss": 0.6924,
+      "step": 18266
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7876754293373175e-06,
+      "loss": 0.71,
+      "step": 18267
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.786631383975632e-06,
+      "loss": 0.7217,
+      "step": 18268
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7855873944509535e-06,
+      "loss": 0.7246,
+      "step": 18269
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.784543460777118e-06,
+      "loss": 0.7139,
+      "step": 18270
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.783499582967959e-06,
+      "loss": 0.7197,
+      "step": 18271
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.782455761037306e-06,
+      "loss": 0.7217,
+      "step": 18272
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.781411994999e-06,
+      "loss": 0.6943,
+      "step": 18273
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.780368284866871e-06,
+      "loss": 0.707,
+      "step": 18274
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.779324630654749e-06,
+      "loss": 0.7217,
+      "step": 18275
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.778281032376466e-06,
+      "loss": 0.7227,
+      "step": 18276
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7772374900458505e-06,
+      "loss": 0.7158,
+      "step": 18277
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.776194003676734e-06,
+      "loss": 0.7002,
+      "step": 18278
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.775150573282944e-06,
+      "loss": 0.7402,
+      "step": 18279
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.774107198878304e-06,
+      "loss": 0.7148,
+      "step": 18280
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7730638804766535e-06,
+      "loss": 0.6963,
+      "step": 18281
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.772020618091807e-06,
+      "loss": 0.6992,
+      "step": 18282
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.770977411737595e-06,
+      "loss": 0.7197,
+      "step": 18283
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7699342614278364e-06,
+      "loss": 0.7158,
+      "step": 18284
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.768891167176366e-06,
+      "loss": 0.71,
+      "step": 18285
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.767848128997001e-06,
+      "loss": 0.6924,
+      "step": 18286
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.766805146903565e-06,
+      "loss": 0.7178,
+      "step": 18287
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.765762220909881e-06,
+      "loss": 0.7363,
+      "step": 18288
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.764719351029769e-06,
+      "loss": 0.7139,
+      "step": 18289
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.763676537277048e-06,
+      "loss": 0.7334,
+      "step": 18290
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7626337796655395e-06,
+      "loss": 0.7178,
+      "step": 18291
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.761591078209059e-06,
+      "loss": 0.7441,
+      "step": 18292
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.760548432921431e-06,
+      "loss": 0.7393,
+      "step": 18293
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.759505843816475e-06,
+      "loss": 0.6982,
+      "step": 18294
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.758463310908e-06,
+      "loss": 0.71,
+      "step": 18295
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7574208342098195e-06,
+      "loss": 0.7139,
+      "step": 18296
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.756378413735758e-06,
+      "loss": 0.7227,
+      "step": 18297
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.755336049499627e-06,
+      "loss": 0.7393,
+      "step": 18298
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.754293741515238e-06,
+      "loss": 0.7002,
+      "step": 18299
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.753251489796405e-06,
+      "loss": 0.7051,
+      "step": 18300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.752209294356943e-06,
+      "loss": 0.7227,
+      "step": 18301
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751167155210659e-06,
+      "loss": 0.7217,
+      "step": 18302
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.750125072371366e-06,
+      "loss": 0.7324,
+      "step": 18303
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.749083045852875e-06,
+      "loss": 0.7178,
+      "step": 18304
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.748041075668991e-06,
+      "loss": 0.7295,
+      "step": 18305
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.746999161833529e-06,
+      "loss": 0.7529,
+      "step": 18306
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.745957304360297e-06,
+      "loss": 0.7363,
+      "step": 18307
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.744915503263091e-06,
+      "loss": 0.7354,
+      "step": 18308
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.743873758555728e-06,
+      "loss": 0.7363,
+      "step": 18309
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.742832070252011e-06,
+      "loss": 0.7012,
+      "step": 18310
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.741790438365746e-06,
+      "loss": 0.71,
+      "step": 18311
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7407488629107344e-06,
+      "loss": 0.6963,
+      "step": 18312
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.739707343900776e-06,
+      "loss": 0.6934,
+      "step": 18313
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7386658813496865e-06,
+      "loss": 0.7344,
+      "step": 18314
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.737624475271255e-06,
+      "loss": 0.7139,
+      "step": 18315
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.736583125679287e-06,
+      "loss": 0.7344,
+      "step": 18316
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.735541832587578e-06,
+      "loss": 0.7188,
+      "step": 18317
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7345005960099365e-06,
+      "loss": 0.7139,
+      "step": 18318
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.733459415960157e-06,
+      "loss": 0.7119,
+      "step": 18319
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.732418292452036e-06,
+      "loss": 0.707,
+      "step": 18320
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.731377225499375e-06,
+      "loss": 0.7051,
+      "step": 18321
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.730336215115966e-06,
+      "loss": 0.7002,
+      "step": 18322
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.729295261315606e-06,
+      "loss": 0.709,
+      "step": 18323
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.728254364112093e-06,
+      "loss": 0.7051,
+      "step": 18324
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.727213523519214e-06,
+      "loss": 0.71,
+      "step": 18325
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.726172739550772e-06,
+      "loss": 0.7217,
+      "step": 18326
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.725132012220558e-06,
+      "loss": 0.7393,
+      "step": 18327
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.724091341542359e-06,
+      "loss": 0.6953,
+      "step": 18328
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723050727529965e-06,
+      "loss": 0.7188,
+      "step": 18329
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.722010170197174e-06,
+      "loss": 0.7158,
+      "step": 18330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.720969669557773e-06,
+      "loss": 0.7246,
+      "step": 18331
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7199292256255504e-06,
+      "loss": 0.707,
+      "step": 18332
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.718888838414294e-06,
+      "loss": 0.7178,
+      "step": 18333
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.717848507937791e-06,
+      "loss": 0.7158,
+      "step": 18334
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7168082342098305e-06,
+      "loss": 0.7002,
+      "step": 18335
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.715768017244196e-06,
+      "loss": 0.7119,
+      "step": 18336
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.714727857054672e-06,
+      "loss": 0.7275,
+      "step": 18337
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.713687753655047e-06,
+      "loss": 0.6963,
+      "step": 18338
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.712647707059105e-06,
+      "loss": 0.7705,
+      "step": 18339
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.711607717280628e-06,
+      "loss": 0.7422,
+      "step": 18340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.710567784333391e-06,
+      "loss": 0.6992,
+      "step": 18341
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.709527908231185e-06,
+      "loss": 0.7061,
+      "step": 18342
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.708488088987789e-06,
+      "loss": 0.6855,
+      "step": 18343
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.70744832661698e-06,
+      "loss": 0.6973,
+      "step": 18344
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7064086211325394e-06,
+      "loss": 0.7139,
+      "step": 18345
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.705368972548246e-06,
+      "loss": 0.7002,
+      "step": 18346
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.704329380877875e-06,
+      "loss": 0.7275,
+      "step": 18347
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.703289846135207e-06,
+      "loss": 0.7246,
+      "step": 18348
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.702250368334015e-06,
+      "loss": 0.6953,
+      "step": 18349
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.701210947488073e-06,
+      "loss": 0.7393,
+      "step": 18350
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.700171583611162e-06,
+      "loss": 0.7012,
+      "step": 18351
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.699132276717057e-06,
+      "loss": 0.6963,
+      "step": 18352
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.69809302681952e-06,
+      "loss": 0.7275,
+      "step": 18353
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.697053833932333e-06,
+      "loss": 0.7441,
+      "step": 18354
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.696014698069267e-06,
+      "loss": 0.6992,
+      "step": 18355
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.69497561924409e-06,
+      "loss": 0.7373,
+      "step": 18356
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.693936597470573e-06,
+      "loss": 0.7002,
+      "step": 18357
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.692897632762483e-06,
+      "loss": 0.7021,
+      "step": 18358
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.691858725133599e-06,
+      "loss": 0.7158,
+      "step": 18359
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.690819874597677e-06,
+      "loss": 0.6924,
+      "step": 18360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6897810811684905e-06,
+      "loss": 0.7139,
+      "step": 18361
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.688742344859799e-06,
+      "loss": 0.708,
+      "step": 18362
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.687703665685378e-06,
+      "loss": 0.708,
+      "step": 18363
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.686665043658988e-06,
+      "loss": 0.6992,
+      "step": 18364
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.685626478794397e-06,
+      "loss": 0.7129,
+      "step": 18365
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.684587971105357e-06,
+      "loss": 0.6953,
+      "step": 18366
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.683549520605641e-06,
+      "loss": 0.7197,
+      "step": 18367
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.682511127309009e-06,
+      "loss": 0.7227,
+      "step": 18368
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.681472791229221e-06,
+      "loss": 0.7139,
+      "step": 18369
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.680434512380033e-06,
+      "loss": 0.7256,
+      "step": 18370
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679396290775214e-06,
+      "loss": 0.7207,
+      "step": 18371
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.67835812642852e-06,
+      "loss": 0.7031,
+      "step": 18372
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.677320019353706e-06,
+      "loss": 0.7324,
+      "step": 18373
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.676281969564526e-06,
+      "loss": 0.7168,
+      "step": 18374
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.675243977074744e-06,
+      "loss": 0.7012,
+      "step": 18375
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.674206041898114e-06,
+      "loss": 0.7314,
+      "step": 18376
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.673168164048388e-06,
+      "loss": 0.7441,
+      "step": 18377
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.672130343539324e-06,
+      "loss": 0.7373,
+      "step": 18378
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.671092580384674e-06,
+      "loss": 0.7002,
+      "step": 18379
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.67005487459819e-06,
+      "loss": 0.7002,
+      "step": 18380
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.669017226193625e-06,
+      "loss": 0.7324,
+      "step": 18381
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.667979635184726e-06,
+      "loss": 0.7227,
+      "step": 18382
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.666942101585253e-06,
+      "loss": 0.7158,
+      "step": 18383
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.665904625408949e-06,
+      "loss": 0.6953,
+      "step": 18384
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.664867206669567e-06,
+      "loss": 0.7227,
+      "step": 18385
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.663829845380846e-06,
+      "loss": 0.6973,
+      "step": 18386
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.662792541556543e-06,
+      "loss": 0.7549,
+      "step": 18387
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.661755295210403e-06,
+      "loss": 0.7158,
+      "step": 18388
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.66071810635617e-06,
+      "loss": 0.7197,
+      "step": 18389
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.65968097500759e-06,
+      "loss": 0.7139,
+      "step": 18390
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6586439011784066e-06,
+      "loss": 0.7188,
+      "step": 18391
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.657606884882364e-06,
+      "loss": 0.7197,
+      "step": 18392
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.656569926133207e-06,
+      "loss": 0.6943,
+      "step": 18393
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.655533024944673e-06,
+      "loss": 0.7295,
+      "step": 18394
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.654496181330504e-06,
+      "loss": 0.7178,
+      "step": 18395
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.653459395304446e-06,
+      "loss": 0.6963,
+      "step": 18396
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.652422666880236e-06,
+      "loss": 0.752,
+      "step": 18397
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6513859960716135e-06,
+      "loss": 0.7461,
+      "step": 18398
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.650349382892315e-06,
+      "loss": 0.7314,
+      "step": 18399
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.64931282735608e-06,
+      "loss": 0.707,
+      "step": 18400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6482763294766455e-06,
+      "loss": 0.7314,
+      "step": 18401
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.647239889267744e-06,
+      "loss": 0.6992,
+      "step": 18402
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6462035067431105e-06,
+      "loss": 0.7236,
+      "step": 18403
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.645167181916491e-06,
+      "loss": 0.7441,
+      "step": 18404
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.644130914801604e-06,
+      "loss": 0.6973,
+      "step": 18405
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.643094705412191e-06,
+      "loss": 0.7334,
+      "step": 18406
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.642058553761978e-06,
+      "loss": 0.707,
+      "step": 18407
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.641022459864705e-06,
+      "loss": 0.7305,
+      "step": 18408
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.639986423734097e-06,
+      "loss": 0.709,
+      "step": 18409
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.638950445383886e-06,
+      "loss": 0.7344,
+      "step": 18410
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6379145248278e-06,
+      "loss": 0.7178,
+      "step": 18411
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.636878662079568e-06,
+      "loss": 0.709,
+      "step": 18412
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.635842857152917e-06,
+      "loss": 0.7051,
+      "step": 18413
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.634807110061575e-06,
+      "loss": 0.7275,
+      "step": 18414
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.633771420819261e-06,
+      "loss": 0.7451,
+      "step": 18415
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.632735789439711e-06,
+      "loss": 0.7197,
+      "step": 18416
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.631700215936651e-06,
+      "loss": 0.7178,
+      "step": 18417
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.630664700323793e-06,
+      "loss": 0.7227,
+      "step": 18418
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.629629242614862e-06,
+      "loss": 0.748,
+      "step": 18419
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.62859384282359e-06,
+      "loss": 0.6904,
+      "step": 18420
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.627558500963691e-06,
+      "loss": 0.7139,
+      "step": 18421
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6265232170488874e-06,
+      "loss": 0.7275,
+      "step": 18422
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.625487991092899e-06,
+      "loss": 0.7178,
+      "step": 18423
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624452823109445e-06,
+      "loss": 0.7188,
+      "step": 18424
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.623417713112243e-06,
+      "loss": 0.7285,
+      "step": 18425
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.622382661115014e-06,
+      "loss": 0.7344,
+      "step": 18426
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.621347667131465e-06,
+      "loss": 0.6992,
+      "step": 18427
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.620312731175325e-06,
+      "loss": 0.7051,
+      "step": 18428
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.619277853260303e-06,
+      "loss": 0.7158,
+      "step": 18429
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.618243033400119e-06,
+      "loss": 0.7021,
+      "step": 18430
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.617208271608474e-06,
+      "loss": 0.7207,
+      "step": 18431
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6161735678990925e-06,
+      "loss": 0.7012,
+      "step": 18432
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.615138922285683e-06,
+      "loss": 0.6963,
+      "step": 18433
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.614104334781958e-06,
+      "loss": 0.7295,
+      "step": 18434
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.613069805401628e-06,
+      "loss": 0.71,
+      "step": 18435
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.612035334158396e-06,
+      "loss": 0.7422,
+      "step": 18436
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6110009210659885e-06,
+      "loss": 0.7051,
+      "step": 18437
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.609966566138098e-06,
+      "loss": 0.7002,
+      "step": 18438
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6089322693884364e-06,
+      "loss": 0.6943,
+      "step": 18439
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.607898030830709e-06,
+      "loss": 0.7012,
+      "step": 18440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606863850478626e-06,
+      "loss": 0.7383,
+      "step": 18441
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.605829728345892e-06,
+      "loss": 0.7178,
+      "step": 18442
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.60479566444621e-06,
+      "loss": 0.75,
+      "step": 18443
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.603761658793285e-06,
+      "loss": 0.7314,
+      "step": 18444
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.602727711400817e-06,
+      "loss": 0.7197,
+      "step": 18445
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6016938222825125e-06,
+      "loss": 0.7021,
+      "step": 18446
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.600659991452071e-06,
+      "loss": 0.7109,
+      "step": 18447
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5996262189231886e-06,
+      "loss": 0.7529,
+      "step": 18448
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.598592504709572e-06,
+      "loss": 0.7266,
+      "step": 18449
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.597558848824925e-06,
+      "loss": 0.7266,
+      "step": 18450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.596525251282933e-06,
+      "loss": 0.6914,
+      "step": 18451
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.595491712097295e-06,
+      "loss": 0.71,
+      "step": 18452
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.594458231281717e-06,
+      "loss": 0.6924,
+      "step": 18453
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.593424808849891e-06,
+      "loss": 0.7119,
+      "step": 18454
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5923914448155105e-06,
+      "loss": 0.7412,
+      "step": 18455
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5913581391922725e-06,
+      "loss": 0.707,
+      "step": 18456
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.590324891993868e-06,
+      "loss": 0.709,
+      "step": 18457
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.589291703233992e-06,
+      "loss": 0.7109,
+      "step": 18458
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.588258572926337e-06,
+      "loss": 0.7158,
+      "step": 18459
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.587225501084588e-06,
+      "loss": 0.7051,
+      "step": 18460
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.586192487722446e-06,
+      "loss": 0.7266,
+      "step": 18461
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5851595328535946e-06,
+      "loss": 0.7217,
+      "step": 18462
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.584126636491729e-06,
+      "loss": 0.7314,
+      "step": 18463
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.583093798650525e-06,
+      "loss": 0.7217,
+      "step": 18464
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.582061019343681e-06,
+      "loss": 0.7256,
+      "step": 18465
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.581028298584881e-06,
+      "loss": 0.7061,
+      "step": 18466
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.57999563638781e-06,
+      "loss": 0.7061,
+      "step": 18467
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.578963032766154e-06,
+      "loss": 0.7119,
+      "step": 18468
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.577930487733596e-06,
+      "loss": 0.7314,
+      "step": 18469
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.576898001303822e-06,
+      "loss": 0.7266,
+      "step": 18470
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.575865573490513e-06,
+      "loss": 0.7305,
+      "step": 18471
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.574833204307352e-06,
+      "loss": 0.709,
+      "step": 18472
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5738008937680154e-06,
+      "loss": 0.7295,
+      "step": 18473
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5727686418861924e-06,
+      "loss": 0.7021,
+      "step": 18474
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.57173644867556e-06,
+      "loss": 0.7148,
+      "step": 18475
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.570704314149795e-06,
+      "loss": 0.707,
+      "step": 18476
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.569672238322577e-06,
+      "loss": 0.7227,
+      "step": 18477
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.568640221207581e-06,
+      "loss": 0.7148,
+      "step": 18478
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.567608262818487e-06,
+      "loss": 0.7051,
+      "step": 18479
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.56657636316897e-06,
+      "loss": 0.7197,
+      "step": 18480
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.565544522272699e-06,
+      "loss": 0.7402,
+      "step": 18481
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.564512740143362e-06,
+      "loss": 0.7314,
+      "step": 18482
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.563481016794621e-06,
+      "loss": 0.7041,
+      "step": 18483
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.562449352240151e-06,
+      "loss": 0.7178,
+      "step": 18484
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5614177464936224e-06,
+      "loss": 0.7285,
+      "step": 18485
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.560386199568713e-06,
+      "loss": 0.7148,
+      "step": 18486
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.559354711479088e-06,
+      "loss": 0.7197,
+      "step": 18487
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.558323282238418e-06,
+      "loss": 0.7021,
+      "step": 18488
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.557291911860374e-06,
+      "loss": 0.7207,
+      "step": 18489
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.556260600358621e-06,
+      "loss": 0.7295,
+      "step": 18490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.555229347746827e-06,
+      "loss": 0.7314,
+      "step": 18491
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.55419815403866e-06,
+      "loss": 0.7119,
+      "step": 18492
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.55316701924778e-06,
+      "loss": 0.7012,
+      "step": 18493
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.552135943387862e-06,
+      "loss": 0.7041,
+      "step": 18494
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.551104926472568e-06,
+      "loss": 0.7266,
+      "step": 18495
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.550073968515555e-06,
+      "loss": 0.7148,
+      "step": 18496
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.549043069530484e-06,
+      "loss": 0.7207,
+      "step": 18497
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5480122295310256e-06,
+      "loss": 0.7324,
+      "step": 18498
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.546981448530837e-06,
+      "loss": 0.7314,
+      "step": 18499
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5459507265435784e-06,
+      "loss": 0.7002,
+      "step": 18500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.54492006358291e-06,
+      "loss": 0.6953,
+      "step": 18501
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.543889459662489e-06,
+      "loss": 0.7168,
+      "step": 18502
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.542858914795974e-06,
+      "loss": 0.7188,
+      "step": 18503
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.541828428997022e-06,
+      "loss": 0.7178,
+      "step": 18504
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.540798002279288e-06,
+      "loss": 0.7139,
+      "step": 18505
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.539767634656431e-06,
+      "loss": 0.7217,
+      "step": 18506
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.538737326142104e-06,
+      "loss": 0.7148,
+      "step": 18507
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.537707076749965e-06,
+      "loss": 0.7178,
+      "step": 18508
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.536676886493656e-06,
+      "loss": 0.7129,
+      "step": 18509
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.53564675538684e-06,
+      "loss": 0.7324,
+      "step": 18510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.534616683443166e-06,
+      "loss": 0.6865,
+      "step": 18511
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.533586670676283e-06,
+      "loss": 0.7158,
+      "step": 18512
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.532556717099842e-06,
+      "loss": 0.7119,
+      "step": 18513
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.531526822727489e-06,
+      "loss": 0.7158,
+      "step": 18514
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.530496987572884e-06,
+      "loss": 0.7197,
+      "step": 18515
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.529467211649664e-06,
+      "loss": 0.709,
+      "step": 18516
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.528437494971477e-06,
+      "loss": 0.6816,
+      "step": 18517
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.527407837551968e-06,
+      "loss": 0.7021,
+      "step": 18518
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.526378239404788e-06,
+      "loss": 0.709,
+      "step": 18519
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5253487005435795e-06,
+      "loss": 0.7217,
+      "step": 18520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.524319220981985e-06,
+      "loss": 0.7031,
+      "step": 18521
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.523289800733649e-06,
+      "loss": 0.7197,
+      "step": 18522
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.522260439812213e-06,
+      "loss": 0.7354,
+      "step": 18523
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.521231138231318e-06,
+      "loss": 0.7227,
+      "step": 18524
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.520201896004605e-06,
+      "loss": 0.7324,
+      "step": 18525
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.519172713145708e-06,
+      "loss": 0.7217,
+      "step": 18526
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.518143589668278e-06,
+      "loss": 0.7188,
+      "step": 18527
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.517114525585951e-06,
+      "loss": 0.7168,
+      "step": 18528
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.516085520912356e-06,
+      "loss": 0.7109,
+      "step": 18529
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.515056575661131e-06,
+      "loss": 0.7256,
+      "step": 18530
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.51402768984592e-06,
+      "loss": 0.7119,
+      "step": 18531
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.512998863480353e-06,
+      "loss": 0.7002,
+      "step": 18532
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.511970096578063e-06,
+      "loss": 0.7178,
+      "step": 18533
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.510941389152687e-06,
+      "loss": 0.7197,
+      "step": 18534
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.509912741217858e-06,
+      "loss": 0.7021,
+      "step": 18535
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.508884152787204e-06,
+      "loss": 0.7061,
+      "step": 18536
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.507855623874359e-06,
+      "loss": 0.7061,
+      "step": 18537
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.506827154492948e-06,
+      "loss": 0.707,
+      "step": 18538
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.505798744656611e-06,
+      "loss": 0.7275,
+      "step": 18539
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.504770394378971e-06,
+      "loss": 0.7109,
+      "step": 18540
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.503742103673659e-06,
+      "loss": 0.7324,
+      "step": 18541
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.502713872554293e-06,
+      "loss": 0.7461,
+      "step": 18542
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.501685701034509e-06,
+      "loss": 0.7227,
+      "step": 18543
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.5006575891279315e-06,
+      "loss": 0.7061,
+      "step": 18544
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4996295368481835e-06,
+      "loss": 0.7061,
+      "step": 18545
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.498601544208888e-06,
+      "loss": 0.7148,
+      "step": 18546
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.49757361122367e-06,
+      "loss": 0.7344,
+      "step": 18547
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.496545737906152e-06,
+      "loss": 0.7119,
+      "step": 18548
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.495517924269955e-06,
+      "loss": 0.7432,
+      "step": 18549
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4944901703286964e-06,
+      "loss": 0.6836,
+      "step": 18550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.493462476096004e-06,
+      "loss": 0.6973,
+      "step": 18551
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.492434841585494e-06,
+      "loss": 0.7148,
+      "step": 18552
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4914072668107845e-06,
+      "loss": 0.7236,
+      "step": 18553
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4903797517854925e-06,
+      "loss": 0.7207,
+      "step": 18554
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.489352296523236e-06,
+      "loss": 0.709,
+      "step": 18555
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4883249010376295e-06,
+      "loss": 0.7178,
+      "step": 18556
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4872975653422914e-06,
+      "loss": 0.7178,
+      "step": 18557
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4862702894508346e-06,
+      "loss": 0.7031,
+      "step": 18558
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.485243073376868e-06,
+      "loss": 0.709,
+      "step": 18559
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484215917134017e-06,
+      "loss": 0.6992,
+      "step": 18560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.483188820735884e-06,
+      "loss": 0.707,
+      "step": 18561
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4821617841960814e-06,
+      "loss": 0.7344,
+      "step": 18562
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.481134807528219e-06,
+      "loss": 0.6973,
+      "step": 18563
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4801078907459115e-06,
+      "loss": 0.7119,
+      "step": 18564
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.479081033862766e-06,
+      "loss": 0.7129,
+      "step": 18565
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4780542368923896e-06,
+      "loss": 0.7334,
+      "step": 18566
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.47702749984839e-06,
+      "loss": 0.7021,
+      "step": 18567
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.476000822744374e-06,
+      "loss": 0.6924,
+      "step": 18568
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.474974205593948e-06,
+      "loss": 0.6924,
+      "step": 18569
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.473947648410718e-06,
+      "loss": 0.7158,
+      "step": 18570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.472921151208281e-06,
+      "loss": 0.7363,
+      "step": 18571
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.471894714000252e-06,
+      "loss": 0.6865,
+      "step": 18572
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.47086833680023e-06,
+      "loss": 0.7188,
+      "step": 18573
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4698420196218134e-06,
+      "loss": 0.6992,
+      "step": 18574
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.468815762478601e-06,
+      "loss": 0.7256,
+      "step": 18575
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4677895653841995e-06,
+      "loss": 0.7197,
+      "step": 18576
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.466763428352206e-06,
+      "loss": 0.7148,
+      "step": 18577
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4657373513962205e-06,
+      "loss": 0.7324,
+      "step": 18578
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.46471133452984e-06,
+      "loss": 0.7129,
+      "step": 18579
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.46368537776666e-06,
+      "loss": 0.7031,
+      "step": 18580
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.462659481120278e-06,
+      "loss": 0.7129,
+      "step": 18581
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.46163364460429e-06,
+      "loss": 0.7314,
+      "step": 18582
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.460607868232287e-06,
+      "loss": 0.7051,
+      "step": 18583
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4595821520178705e-06,
+      "loss": 0.7236,
+      "step": 18584
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.458556495974628e-06,
+      "loss": 0.6836,
+      "step": 18585
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.457530900116159e-06,
+      "loss": 0.6943,
+      "step": 18586
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.456505364456041e-06,
+      "loss": 0.7314,
+      "step": 18587
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4554798890078765e-06,
+      "loss": 0.7148,
+      "step": 18588
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.454454473785252e-06,
+      "loss": 0.7354,
+      "step": 18589
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.453429118801757e-06,
+      "loss": 0.7168,
+      "step": 18590
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.45240382407098e-06,
+      "loss": 0.7139,
+      "step": 18591
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.451378589606507e-06,
+      "loss": 0.71,
+      "step": 18592
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.450353415421927e-06,
+      "loss": 0.7129,
+      "step": 18593
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.449328301530825e-06,
+      "loss": 0.6992,
+      "step": 18594
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.448303247946781e-06,
+      "loss": 0.7305,
+      "step": 18595
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.447278254683389e-06,
+      "loss": 0.7432,
+      "step": 18596
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.446253321754228e-06,
+      "loss": 0.6992,
+      "step": 18597
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.445228449172881e-06,
+      "loss": 0.7227,
+      "step": 18598
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.444203636952929e-06,
+      "loss": 0.6982,
+      "step": 18599
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.443178885107954e-06,
+      "loss": 0.7432,
+      "step": 18600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.442154193651536e-06,
+      "loss": 0.707,
+      "step": 18601
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.441129562597256e-06,
+      "loss": 0.7256,
+      "step": 18602
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.44010499195869e-06,
+      "loss": 0.7031,
+      "step": 18603
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.439080481749415e-06,
+      "loss": 0.6953,
+      "step": 18604
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.438056031983016e-06,
+      "loss": 0.7119,
+      "step": 18605
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4370316426730625e-06,
+      "loss": 0.7236,
+      "step": 18606
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4360073138331305e-06,
+      "loss": 0.6982,
+      "step": 18607
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.434983045476791e-06,
+      "loss": 0.6963,
+      "step": 18608
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.433958837617629e-06,
+      "loss": 0.6914,
+      "step": 18609
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.43293469026921e-06,
+      "loss": 0.7158,
+      "step": 18610
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431910603445108e-06,
+      "loss": 0.6914,
+      "step": 18611
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.430886577158895e-06,
+      "loss": 0.7148,
+      "step": 18612
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.42986261142414e-06,
+      "loss": 0.7119,
+      "step": 18613
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.428838706254416e-06,
+      "loss": 0.7217,
+      "step": 18614
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.427814861663287e-06,
+      "loss": 0.6797,
+      "step": 18615
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.426791077664323e-06,
+      "loss": 0.7188,
+      "step": 18616
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.425767354271097e-06,
+      "loss": 0.7354,
+      "step": 18617
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.424743691497174e-06,
+      "loss": 0.6943,
+      "step": 18618
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4237200893561144e-06,
+      "loss": 0.6895,
+      "step": 18619
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.422696547861483e-06,
+      "loss": 0.7188,
+      "step": 18620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4216730670268526e-06,
+      "loss": 0.7197,
+      "step": 18621
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.420649646865781e-06,
+      "loss": 0.709,
+      "step": 18622
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4196262873918314e-06,
+      "loss": 0.7197,
+      "step": 18623
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4186029886185665e-06,
+      "loss": 0.708,
+      "step": 18624
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4175797505595475e-06,
+      "loss": 0.6992,
+      "step": 18625
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.416556573228333e-06,
+      "loss": 0.7256,
+      "step": 18626
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4155334566384844e-06,
+      "loss": 0.7393,
+      "step": 18627
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4145104008035565e-06,
+      "loss": 0.709,
+      "step": 18628
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.413487405737113e-06,
+      "loss": 0.7148,
+      "step": 18629
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.412464471452709e-06,
+      "loss": 0.6914,
+      "step": 18630
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4114415979639044e-06,
+      "loss": 0.7109,
+      "step": 18631
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.410418785284242e-06,
+      "loss": 0.7207,
+      "step": 18632
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.40939603342729e-06,
+      "loss": 0.707,
+      "step": 18633
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.408373342406597e-06,
+      "loss": 0.7402,
+      "step": 18634
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.407350712235714e-06,
+      "loss": 0.709,
+      "step": 18635
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4063281429281965e-06,
+      "loss": 0.7354,
+      "step": 18636
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.405305634497592e-06,
+      "loss": 0.6934,
+      "step": 18637
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4042831869574605e-06,
+      "loss": 0.7207,
+      "step": 18638
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.403260800321343e-06,
+      "loss": 0.7051,
+      "step": 18639
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.402238474602785e-06,
+      "loss": 0.708,
+      "step": 18640
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.401216209815345e-06,
+      "loss": 0.7256,
+      "step": 18641
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.400194005972568e-06,
+      "loss": 0.6924,
+      "step": 18642
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3991718630879966e-06,
+      "loss": 0.7617,
+      "step": 18643
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.398149781175179e-06,
+      "loss": 0.709,
+      "step": 18644
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.39712776024766e-06,
+      "loss": 0.709,
+      "step": 18645
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.396105800318982e-06,
+      "loss": 0.7188,
+      "step": 18646
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.395083901402692e-06,
+      "loss": 0.7373,
+      "step": 18647
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3940620635123294e-06,
+      "loss": 0.7246,
+      "step": 18648
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3930402866614325e-06,
+      "loss": 0.7197,
+      "step": 18649
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.392018570863553e-06,
+      "loss": 0.7158,
+      "step": 18650
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.390996916132226e-06,
+      "loss": 0.7129,
+      "step": 18651
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.389975322480987e-06,
+      "loss": 0.7188,
+      "step": 18652
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.388953789923372e-06,
+      "loss": 0.7256,
+      "step": 18653
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.387932318472929e-06,
+      "loss": 0.709,
+      "step": 18654
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3869109081431894e-06,
+      "loss": 0.7119,
+      "step": 18655
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.385889558947689e-06,
+      "loss": 0.7197,
+      "step": 18656
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.384868270899963e-06,
+      "loss": 0.6836,
+      "step": 18657
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.383847044013548e-06,
+      "loss": 0.6943,
+      "step": 18658
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.382825878301976e-06,
+      "loss": 0.7158,
+      "step": 18659
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.381804773778779e-06,
+      "loss": 0.7334,
+      "step": 18660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.380783730457486e-06,
+      "loss": 0.7197,
+      "step": 18661
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.379762748351638e-06,
+      "loss": 0.7188,
+      "step": 18662
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.378741827474759e-06,
+      "loss": 0.6992,
+      "step": 18663
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.377720967840381e-06,
+      "loss": 0.7246,
+      "step": 18664
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.376700169462023e-06,
+      "loss": 0.6914,
+      "step": 18665
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.375679432353227e-06,
+      "loss": 0.7188,
+      "step": 18666
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.374658756527513e-06,
+      "loss": 0.7168,
+      "step": 18667
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3736381419984095e-06,
+      "loss": 0.7451,
+      "step": 18668
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3726175887794404e-06,
+      "loss": 0.7305,
+      "step": 18669
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.371597096884131e-06,
+      "loss": 0.7168,
+      "step": 18670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.370576666326006e-06,
+      "loss": 0.7188,
+      "step": 18671
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.369556297118587e-06,
+      "loss": 0.7246,
+      "step": 18672
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.368535989275392e-06,
+      "loss": 0.7354,
+      "step": 18673
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.367515742809952e-06,
+      "loss": 0.6963,
+      "step": 18674
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3664955577357846e-06,
+      "loss": 0.7227,
+      "step": 18675
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3654754340664075e-06,
+      "loss": 0.6963,
+      "step": 18676
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36445537181534e-06,
+      "loss": 0.7217,
+      "step": 18677
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.363435370996099e-06,
+      "loss": 0.7227,
+      "step": 18678
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.362415431622207e-06,
+      "loss": 0.6963,
+      "step": 18679
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3613955537071746e-06,
+      "loss": 0.7188,
+      "step": 18680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3603757372645215e-06,
+      "loss": 0.71,
+      "step": 18681
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.359355982307756e-06,
+      "loss": 0.7109,
+      "step": 18682
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.358336288850408e-06,
+      "loss": 0.7031,
+      "step": 18683
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.357316656905974e-06,
+      "loss": 0.707,
+      "step": 18684
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.356297086487974e-06,
+      "loss": 0.7109,
+      "step": 18685
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.355277577609914e-06,
+      "loss": 0.6982,
+      "step": 18686
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.354258130285315e-06,
+      "loss": 0.6855,
+      "step": 18687
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.35323874452768e-06,
+      "loss": 0.6865,
+      "step": 18688
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.35221942035052e-06,
+      "loss": 0.7158,
+      "step": 18689
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.351200157767344e-06,
+      "loss": 0.7432,
+      "step": 18690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.350180956791658e-06,
+      "loss": 0.707,
+      "step": 18691
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.349161817436969e-06,
+      "loss": 0.7119,
+      "step": 18692
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.348142739716786e-06,
+      "loss": 0.6982,
+      "step": 18693
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.347123723644607e-06,
+      "loss": 0.6797,
+      "step": 18694
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.346104769233944e-06,
+      "loss": 0.7217,
+      "step": 18695
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.345085876498303e-06,
+      "loss": 0.7363,
+      "step": 18696
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3440670454511755e-06,
+      "loss": 0.7207,
+      "step": 18697
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.343048276106067e-06,
+      "loss": 0.7246,
+      "step": 18698
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.342029568476485e-06,
+      "loss": 0.7422,
+      "step": 18699
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.341010922575925e-06,
+      "loss": 0.7197,
+      "step": 18700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.339992338417888e-06,
+      "loss": 0.71,
+      "step": 18701
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3389738160158665e-06,
+      "loss": 0.7266,
+      "step": 18702
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.337955355383371e-06,
+      "loss": 0.7051,
+      "step": 18703
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3369369565338894e-06,
+      "loss": 0.6875,
+      "step": 18704
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.335918619480917e-06,
+      "loss": 0.71,
+      "step": 18705
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.334900344237948e-06,
+      "loss": 0.7148,
+      "step": 18706
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.333882130818486e-06,
+      "loss": 0.7012,
+      "step": 18707
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.332863979236018e-06,
+      "loss": 0.7305,
+      "step": 18708
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.331845889504043e-06,
+      "loss": 0.6973,
+      "step": 18709
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.330827861636041e-06,
+      "loss": 0.7324,
+      "step": 18710
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.329809895645512e-06,
+      "loss": 0.7402,
+      "step": 18711
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3287919915459474e-06,
+      "loss": 0.6943,
+      "step": 18712
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.327774149350834e-06,
+      "loss": 0.7168,
+      "step": 18713
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3267563690736605e-06,
+      "loss": 0.6963,
+      "step": 18714
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.325738650727912e-06,
+      "loss": 0.7012,
+      "step": 18715
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.324720994327086e-06,
+      "loss": 0.7275,
+      "step": 18716
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.323703399884658e-06,
+      "loss": 0.708,
+      "step": 18717
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.322685867414112e-06,
+      "loss": 0.6992,
+      "step": 18718
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.321668396928945e-06,
+      "loss": 0.6963,
+      "step": 18719
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.320650988442632e-06,
+      "loss": 0.7178,
+      "step": 18720
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.319633641968658e-06,
+      "loss": 0.7305,
+      "step": 18721
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.318616357520505e-06,
+      "loss": 0.6992,
+      "step": 18722
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3175991351116555e-06,
+      "loss": 0.7256,
+      "step": 18723
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.316581974755588e-06,
+      "loss": 0.7363,
+      "step": 18724
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.315564876465784e-06,
+      "loss": 0.7012,
+      "step": 18725
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.314547840255721e-06,
+      "loss": 0.708,
+      "step": 18726
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.313530866138876e-06,
+      "loss": 0.7461,
+      "step": 18727
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.31251395412873e-06,
+      "loss": 0.71,
+      "step": 18728
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.311497104238765e-06,
+      "loss": 0.6895,
+      "step": 18729
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.310480316482442e-06,
+      "loss": 0.7061,
+      "step": 18730
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.309463590873241e-06,
+      "loss": 0.7188,
+      "step": 18731
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.308446927424641e-06,
+      "loss": 0.71,
+      "step": 18732
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.307430326150112e-06,
+      "loss": 0.7148,
+      "step": 18733
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3064137870631275e-06,
+      "loss": 0.7119,
+      "step": 18734
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3053973101771585e-06,
+      "loss": 0.7207,
+      "step": 18735
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3043808955056755e-06,
+      "loss": 0.7305,
+      "step": 18736
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.303364543062148e-06,
+      "loss": 0.6992,
+      "step": 18737
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.302348252860044e-06,
+      "loss": 0.7061,
+      "step": 18738
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.301332024912831e-06,
+      "loss": 0.6934,
+      "step": 18739
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.300315859233982e-06,
+      "loss": 0.7236,
+      "step": 18740
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2992997558369595e-06,
+      "loss": 0.7266,
+      "step": 18741
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.298283714735234e-06,
+      "loss": 0.7119,
+      "step": 18742
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2972677359422586e-06,
+      "loss": 0.7031,
+      "step": 18743
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.296251819471509e-06,
+      "loss": 0.7393,
+      "step": 18744
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.295235965336444e-06,
+      "loss": 0.6875,
+      "step": 18745
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.294220173550526e-06,
+      "loss": 0.7109,
+      "step": 18746
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.293204444127214e-06,
+      "loss": 0.7129,
+      "step": 18747
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.29218877707998e-06,
+      "loss": 0.7197,
+      "step": 18748
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.291173172422273e-06,
+      "loss": 0.6855,
+      "step": 18749
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.290157630167553e-06,
+      "loss": 0.7373,
+      "step": 18750
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.289142150329277e-06,
+      "loss": 0.6953,
+      "step": 18751
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2881267329209105e-06,
+      "loss": 0.7158,
+      "step": 18752
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2871113779559045e-06,
+      "loss": 0.7227,
+      "step": 18753
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.286096085447717e-06,
+      "loss": 0.7139,
+      "step": 18754
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.285080855409801e-06,
+      "loss": 0.7227,
+      "step": 18755
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.284065687855613e-06,
+      "loss": 0.707,
+      "step": 18756
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.283050582798604e-06,
+      "loss": 0.6875,
+      "step": 18757
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.282035540252226e-06,
+      "loss": 0.7168,
+      "step": 18758
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.281020560229934e-06,
+      "loss": 0.6963,
+      "step": 18759
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2800056427451715e-06,
+      "loss": 0.7139,
+      "step": 18760
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.278990787811404e-06,
+      "loss": 0.7129,
+      "step": 18761
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.277975995442064e-06,
+      "loss": 0.6992,
+      "step": 18762
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.276961265650604e-06,
+      "loss": 0.7256,
+      "step": 18763
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.275946598450479e-06,
+      "loss": 0.7256,
+      "step": 18764
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.27493199385513e-06,
+      "loss": 0.7031,
+      "step": 18765
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.273917451878003e-06,
+      "loss": 0.7217,
+      "step": 18766
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.272902972532545e-06,
+      "loss": 0.6924,
+      "step": 18767
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.271888555832198e-06,
+      "loss": 0.7188,
+      "step": 18768
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.270874201790408e-06,
+      "loss": 0.749,
+      "step": 18769
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.269859910420615e-06,
+      "loss": 0.7178,
+      "step": 18770
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2688456817362625e-06,
+      "loss": 0.7227,
+      "step": 18771
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.267831515750785e-06,
+      "loss": 0.7285,
+      "step": 18772
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.266817412477634e-06,
+      "loss": 0.7285,
+      "step": 18773
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.265803371930245e-06,
+      "loss": 0.7041,
+      "step": 18774
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.264789394122052e-06,
+      "loss": 0.7129,
+      "step": 18775
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.26377547906649e-06,
+      "loss": 0.7129,
+      "step": 18776
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2627616267770055e-06,
+      "loss": 0.7158,
+      "step": 18777
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.261747837267029e-06,
+      "loss": 0.7334,
+      "step": 18778
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.260734110549997e-06,
+      "loss": 0.7061,
+      "step": 18779
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.259720446639338e-06,
+      "loss": 0.6982,
+      "step": 18780
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.258706845548499e-06,
+      "loss": 0.7236,
+      "step": 18781
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2576933072909e-06,
+      "loss": 0.7402,
+      "step": 18782
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.256679831879976e-06,
+      "loss": 0.7148,
+      "step": 18783
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.255666419329155e-06,
+      "loss": 0.7021,
+      "step": 18784
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.254653069651875e-06,
+      "loss": 0.7051,
+      "step": 18785
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.253639782861562e-06,
+      "loss": 0.7285,
+      "step": 18786
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.252626558971645e-06,
+      "loss": 0.6982,
+      "step": 18787
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.251613397995543e-06,
+      "loss": 0.708,
+      "step": 18788
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.250600299946695e-06,
+      "loss": 0.6846,
+      "step": 18789
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.24958726483852e-06,
+      "loss": 0.7197,
+      "step": 18790
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.248574292684446e-06,
+      "loss": 0.7002,
+      "step": 18791
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.247561383497896e-06,
+      "loss": 0.7217,
+      "step": 18792
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.246548537292293e-06,
+      "loss": 0.708,
+      "step": 18793
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.245535754081059e-06,
+      "loss": 0.7031,
+      "step": 18794
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.244523033877619e-06,
+      "loss": 0.7393,
+      "step": 18795
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.243510376695386e-06,
+      "loss": 0.7402,
+      "step": 18796
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.242497782547791e-06,
+      "loss": 0.7012,
+      "step": 18797
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.241485251448247e-06,
+      "loss": 0.7061,
+      "step": 18798
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.240472783410174e-06,
+      "loss": 0.7227,
+      "step": 18799
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.23946037844699e-06,
+      "loss": 0.7383,
+      "step": 18800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.238448036572109e-06,
+      "loss": 0.7236,
+      "step": 18801
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2374357577989495e-06,
+      "loss": 0.7041,
+      "step": 18802
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.236423542140925e-06,
+      "loss": 0.71,
+      "step": 18803
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2354113896114515e-06,
+      "loss": 0.7266,
+      "step": 18804
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.234399300223937e-06,
+      "loss": 0.6943,
+      "step": 18805
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233387273991806e-06,
+      "loss": 0.6982,
+      "step": 18806
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.23237531092846e-06,
+      "loss": 0.6875,
+      "step": 18807
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.231363411047307e-06,
+      "loss": 0.709,
+      "step": 18808
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.230351574361766e-06,
+      "loss": 0.752,
+      "step": 18809
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.229339800885245e-06,
+      "loss": 0.7178,
+      "step": 18810
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2283280906311476e-06,
+      "loss": 0.7305,
+      "step": 18811
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2273164436128855e-06,
+      "loss": 0.7363,
+      "step": 18812
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.226304859843863e-06,
+      "loss": 0.709,
+      "step": 18813
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.225293339337487e-06,
+      "loss": 0.7051,
+      "step": 18814
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.224281882107163e-06,
+      "loss": 0.7266,
+      "step": 18815
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.223270488166293e-06,
+      "loss": 0.708,
+      "step": 18816
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.222259157528279e-06,
+      "loss": 0.7227,
+      "step": 18817
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.221247890206529e-06,
+      "loss": 0.6943,
+      "step": 18818
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2202366862144475e-06,
+      "loss": 0.7012,
+      "step": 18819
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.219225545565425e-06,
+      "loss": 0.7227,
+      "step": 18820
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.218214468272862e-06,
+      "loss": 0.6943,
+      "step": 18821
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.217203454350166e-06,
+      "loss": 0.7139,
+      "step": 18822
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2161925038107305e-06,
+      "loss": 0.7227,
+      "step": 18823
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.215181616667954e-06,
+      "loss": 0.7061,
+      "step": 18824
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.214170792935229e-06,
+      "loss": 0.7275,
+      "step": 18825
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.213160032625963e-06,
+      "loss": 0.7363,
+      "step": 18826
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.212149335753538e-06,
+      "loss": 0.7246,
+      "step": 18827
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.211138702331354e-06,
+      "loss": 0.7422,
+      "step": 18828
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.210128132372801e-06,
+      "loss": 0.7129,
+      "step": 18829
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.209117625891275e-06,
+      "loss": 0.6924,
+      "step": 18830
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2081071829001675e-06,
+      "loss": 0.7334,
+      "step": 18831
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.207096803412872e-06,
+      "loss": 0.7109,
+      "step": 18832
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.206086487442765e-06,
+      "loss": 0.6953,
+      "step": 18833
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2050762350032516e-06,
+      "loss": 0.7041,
+      "step": 18834
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.204066046107712e-06,
+      "loss": 0.7129,
+      "step": 18835
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.203055920769535e-06,
+      "loss": 0.7051,
+      "step": 18836
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.202045859002107e-06,
+      "loss": 0.71,
+      "step": 18837
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2010358608188106e-06,
+      "loss": 0.7227,
+      "step": 18838
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.200025926233041e-06,
+      "loss": 0.7139,
+      "step": 18839
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.199016055258172e-06,
+      "loss": 0.7109,
+      "step": 18840
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.198006247907585e-06,
+      "loss": 0.6992,
+      "step": 18841
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1969965041946715e-06,
+      "loss": 0.7012,
+      "step": 18842
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.195986824132808e-06,
+      "loss": 0.7266,
+      "step": 18843
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1949772077353765e-06,
+      "loss": 0.748,
+      "step": 18844
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1939676550157555e-06,
+      "loss": 0.7002,
+      "step": 18845
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.192958165987325e-06,
+      "loss": 0.6973,
+      "step": 18846
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.191948740663462e-06,
+      "loss": 0.7266,
+      "step": 18847
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.190939379057544e-06,
+      "loss": 0.7061,
+      "step": 18848
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.189930081182949e-06,
+      "loss": 0.7129,
+      "step": 18849
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1889208470530475e-06,
+      "loss": 0.7227,
+      "step": 18850
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.18791167668122e-06,
+      "loss": 0.7354,
+      "step": 18851
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1869025700808426e-06,
+      "loss": 0.7002,
+      "step": 18852
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.185893527265278e-06,
+      "loss": 0.6982,
+      "step": 18853
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.184884548247906e-06,
+      "loss": 0.7256,
+      "step": 18854
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.183875633042098e-06,
+      "loss": 0.7236,
+      "step": 18855
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.182866781661222e-06,
+      "loss": 0.6904,
+      "step": 18856
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.18185799411865e-06,
+      "loss": 0.7207,
+      "step": 18857
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.180849270427748e-06,
+      "loss": 0.7207,
+      "step": 18858
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.179840610601885e-06,
+      "loss": 0.7402,
+      "step": 18859
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.178832014654429e-06,
+      "loss": 0.7188,
+      "step": 18860
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1778234825987465e-06,
+      "loss": 0.7227,
+      "step": 18861
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.176815014448196e-06,
+      "loss": 0.7148,
+      "step": 18862
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.175806610216154e-06,
+      "loss": 0.6953,
+      "step": 18863
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.174798269915977e-06,
+      "loss": 0.7012,
+      "step": 18864
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.173789993561034e-06,
+      "loss": 0.6953,
+      "step": 18865
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.172781781164676e-06,
+      "loss": 0.7051,
+      "step": 18866
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1717736327402715e-06,
+      "loss": 0.7197,
+      "step": 18867
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.17076554830118e-06,
+      "loss": 0.7178,
+      "step": 18868
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1697575278607615e-06,
+      "loss": 0.7275,
+      "step": 18869
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.16874957143237e-06,
+      "loss": 0.7197,
+      "step": 18870
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.167741679029374e-06,
+      "loss": 0.7305,
+      "step": 18871
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.166733850665122e-06,
+      "loss": 0.708,
+      "step": 18872
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.165726086352969e-06,
+      "loss": 0.708,
+      "step": 18873
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.164718386106271e-06,
+      "loss": 0.7207,
+      "step": 18874
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1637107499383864e-06,
+      "loss": 0.7236,
+      "step": 18875
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.162703177862669e-06,
+      "loss": 0.7578,
+      "step": 18876
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.161695669892467e-06,
+      "loss": 0.7119,
+      "step": 18877
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.160688226041135e-06,
+      "loss": 0.7178,
+      "step": 18878
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.159680846322024e-06,
+      "loss": 0.7031,
+      "step": 18879
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.158673530748483e-06,
+      "loss": 0.7178,
+      "step": 18880
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.157666279333861e-06,
+      "loss": 0.7256,
+      "step": 18881
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.156659092091508e-06,
+      "loss": 0.7236,
+      "step": 18882
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1556519690347675e-06,
+      "loss": 0.7324,
+      "step": 18883
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.154644910176995e-06,
+      "loss": 0.6982,
+      "step": 18884
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.153637915531531e-06,
+      "loss": 0.7109,
+      "step": 18885
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.152630985111713e-06,
+      "loss": 0.6982,
+      "step": 18886
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.151624118930898e-06,
+      "loss": 0.6963,
+      "step": 18887
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1506173170024245e-06,
+      "loss": 0.7148,
+      "step": 18888
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.149610579339634e-06,
+      "loss": 0.7246,
+      "step": 18889
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.148603905955869e-06,
+      "loss": 0.7012,
+      "step": 18890
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.147597296864468e-06,
+      "loss": 0.6826,
+      "step": 18891
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.146590752078775e-06,
+      "loss": 0.7256,
+      "step": 18892
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1455842716121276e-06,
+      "loss": 0.6963,
+      "step": 18893
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.144577855477862e-06,
+      "loss": 0.708,
+      "step": 18894
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.143571503689312e-06,
+      "loss": 0.75,
+      "step": 18895
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.142565216259824e-06,
+      "loss": 0.7217,
+      "step": 18896
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.141558993202733e-06,
+      "loss": 0.708,
+      "step": 18897
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.140552834531365e-06,
+      "loss": 0.709,
+      "step": 18898
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.139546740259053e-06,
+      "loss": 0.7305,
+      "step": 18899
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.138540710399143e-06,
+      "loss": 0.748,
+      "step": 18900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.137534744964958e-06,
+      "loss": 0.6934,
+      "step": 18901
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.136528843969831e-06,
+      "loss": 0.7402,
+      "step": 18902
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1355230074270904e-06,
+      "loss": 0.6992,
+      "step": 18903
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134517235350074e-06,
+      "loss": 0.7012,
+      "step": 18904
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.133511527752102e-06,
+      "loss": 0.7061,
+      "step": 18905
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.132505884646505e-06,
+      "loss": 0.7354,
+      "step": 18906
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.131500306046606e-06,
+      "loss": 0.75,
+      "step": 18907
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.130494791965741e-06,
+      "loss": 0.7295,
+      "step": 18908
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.129489342417228e-06,
+      "loss": 0.7061,
+      "step": 18909
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.128483957414399e-06,
+      "loss": 0.7275,
+      "step": 18910
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1274786369705655e-06,
+      "loss": 0.707,
+      "step": 18911
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1264733810990595e-06,
+      "loss": 0.7275,
+      "step": 18912
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.125468189813202e-06,
+      "loss": 0.7373,
+      "step": 18913
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.124463063126312e-06,
+      "loss": 0.7197,
+      "step": 18914
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.123458001051707e-06,
+      "loss": 0.7217,
+      "step": 18915
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122453003602714e-06,
+      "loss": 0.7031,
+      "step": 18916
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.121448070792652e-06,
+      "loss": 0.7422,
+      "step": 18917
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.120443202634829e-06,
+      "loss": 0.7168,
+      "step": 18918
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.119438399142565e-06,
+      "loss": 0.7236,
+      "step": 18919
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.11843366032918e-06,
+      "loss": 0.6787,
+      "step": 18920
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.117428986207989e-06,
+      "loss": 0.6963,
+      "step": 18921
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.116424376792305e-06,
+      "loss": 0.7031,
+      "step": 18922
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.11541983209544e-06,
+      "loss": 0.7002,
+      "step": 18923
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1144153521307085e-06,
+      "loss": 0.7188,
+      "step": 18924
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.11341093691142e-06,
+      "loss": 0.7129,
+      "step": 18925
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.112406586450889e-06,
+      "loss": 0.7227,
+      "step": 18926
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.111402300762421e-06,
+      "loss": 0.7148,
+      "step": 18927
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.110398079859325e-06,
+      "loss": 0.7012,
+      "step": 18928
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.109393923754915e-06,
+      "loss": 0.7422,
+      "step": 18929
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.108389832462499e-06,
+      "loss": 0.7139,
+      "step": 18930
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.107385805995371e-06,
+      "loss": 0.7139,
+      "step": 18931
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.10638184436685e-06,
+      "loss": 0.7227,
+      "step": 18932
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1053779475902354e-06,
+      "loss": 0.6963,
+      "step": 18933
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.104374115678834e-06,
+      "loss": 0.7295,
+      "step": 18934
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.103370348645944e-06,
+      "loss": 0.71,
+      "step": 18935
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.102366646504872e-06,
+      "loss": 0.71,
+      "step": 18936
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.1013630092689185e-06,
+      "loss": 0.7207,
+      "step": 18937
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.100359436951383e-06,
+      "loss": 0.7158,
+      "step": 18938
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099355929565566e-06,
+      "loss": 0.7227,
+      "step": 18939
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0983524871247615e-06,
+      "loss": 0.7363,
+      "step": 18940
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.097349109642276e-06,
+      "loss": 0.709,
+      "step": 18941
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0963457971314026e-06,
+      "loss": 0.6963,
+      "step": 18942
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.095342549605441e-06,
+      "loss": 0.7266,
+      "step": 18943
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.094339367077676e-06,
+      "loss": 0.7422,
+      "step": 18944
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.093336249561412e-06,
+      "loss": 0.7119,
+      "step": 18945
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.092333197069941e-06,
+      "loss": 0.7373,
+      "step": 18946
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.091330209616554e-06,
+      "loss": 0.71,
+      "step": 18947
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0903272872145405e-06,
+      "loss": 0.7148,
+      "step": 18948
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.089324429877203e-06,
+      "loss": 0.71,
+      "step": 18949
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.088321637617817e-06,
+      "loss": 0.7334,
+      "step": 18950
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.08731891044968e-06,
+      "loss": 0.7354,
+      "step": 18951
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.086316248386074e-06,
+      "loss": 0.7051,
+      "step": 18952
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085313651440295e-06,
+      "loss": 0.7168,
+      "step": 18953
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.084311119625628e-06,
+      "loss": 0.7129,
+      "step": 18954
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0833086529553555e-06,
+      "loss": 0.7236,
+      "step": 18955
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.082306251442765e-06,
+      "loss": 0.7266,
+      "step": 18956
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.081303915101139e-06,
+      "loss": 0.7129,
+      "step": 18957
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.080301643943761e-06,
+      "loss": 0.6826,
+      "step": 18958
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.079299437983914e-06,
+      "loss": 0.6934,
+      "step": 18959
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0782972972348754e-06,
+      "loss": 0.709,
+      "step": 18960
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.077295221709935e-06,
+      "loss": 0.7314,
+      "step": 18961
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.076293211422371e-06,
+      "loss": 0.7295,
+      "step": 18962
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.075291266385456e-06,
+      "loss": 0.709,
+      "step": 18963
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.074289386612465e-06,
+      "loss": 0.7217,
+      "step": 18964
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.073287572116688e-06,
+      "loss": 0.6953,
+      "step": 18965
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.072285822911393e-06,
+      "loss": 0.7246,
+      "step": 18966
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.071284139009858e-06,
+      "loss": 0.7412,
+      "step": 18967
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.070282520425357e-06,
+      "loss": 0.7178,
+      "step": 18968
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.069280967171164e-06,
+      "loss": 0.7334,
+      "step": 18969
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.068279479260552e-06,
+      "loss": 0.7227,
+      "step": 18970
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.067278056706791e-06,
+      "loss": 0.7041,
+      "step": 18971
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.066276699523156e-06,
+      "loss": 0.7148,
+      "step": 18972
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.06527540772291e-06,
+      "loss": 0.709,
+      "step": 18973
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.064274181319333e-06,
+      "loss": 0.7031,
+      "step": 18974
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.063273020325692e-06,
+      "loss": 0.7285,
+      "step": 18975
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.062271924755242e-06,
+      "loss": 0.7227,
+      "step": 18976
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.061270894621264e-06,
+      "loss": 0.7275,
+      "step": 18977
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0602699299370185e-06,
+      "loss": 0.7227,
+      "step": 18978
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.059269030715771e-06,
+      "loss": 0.707,
+      "step": 18979
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.058268196970786e-06,
+      "loss": 0.7324,
+      "step": 18980
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0572674287153226e-06,
+      "loss": 0.7129,
+      "step": 18981
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.056266725962655e-06,
+      "loss": 0.6943,
+      "step": 18982
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.055266088726034e-06,
+      "loss": 0.7295,
+      "step": 18983
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.054265517018724e-06,
+      "loss": 0.7139,
+      "step": 18984
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.053265010853981e-06,
+      "loss": 0.7168,
+      "step": 18985
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.052264570245071e-06,
+      "loss": 0.7109,
+      "step": 18986
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.051264195205249e-06,
+      "loss": 0.7383,
+      "step": 18987
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.050263885747775e-06,
+      "loss": 0.7012,
+      "step": 18988
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0492636418858955e-06,
+      "loss": 0.7227,
+      "step": 18989
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.048263463632879e-06,
+      "loss": 0.7197,
+      "step": 18990
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0472633510019716e-06,
+      "loss": 0.7393,
+      "step": 18991
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.046263304006432e-06,
+      "loss": 0.7109,
+      "step": 18992
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.045263322659505e-06,
+      "loss": 0.7324,
+      "step": 18993
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.044263406974454e-06,
+      "loss": 0.7207,
+      "step": 18994
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.043263556964528e-06,
+      "loss": 0.7158,
+      "step": 18995
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.042263772642972e-06,
+      "loss": 0.7061,
+      "step": 18996
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.041264054023033e-06,
+      "loss": 0.7158,
+      "step": 18997
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0402644011179694e-06,
+      "loss": 0.7207,
+      "step": 18998
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.039264813941023e-06,
+      "loss": 0.7285,
+      "step": 18999
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.038265292505441e-06,
+      "loss": 0.6963,
+      "step": 19000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.037265836824471e-06,
+      "loss": 0.6914,
+      "step": 19001
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0362664469113575e-06,
+      "loss": 0.7061,
+      "step": 19002
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.035267122779344e-06,
+      "loss": 0.7041,
+      "step": 19003
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.034267864441673e-06,
+      "loss": 0.7158,
+      "step": 19004
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0332686719115905e-06,
+      "loss": 0.7119,
+      "step": 19005
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.032269545202332e-06,
+      "loss": 0.7041,
+      "step": 19006
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.031270484327148e-06,
+      "loss": 0.6895,
+      "step": 19007
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.030271489299272e-06,
+      "loss": 0.7236,
+      "step": 19008
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.029272560131937e-06,
+      "loss": 0.71,
+      "step": 19009
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0282736968383924e-06,
+      "loss": 0.7236,
+      "step": 19010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.027274899431873e-06,
+      "loss": 0.7129,
+      "step": 19011
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0262761679256115e-06,
+      "loss": 0.7158,
+      "step": 19012
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.025277502332844e-06,
+      "loss": 0.7109,
+      "step": 19013
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.024278902666808e-06,
+      "loss": 0.7168,
+      "step": 19014
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.023280368940735e-06,
+      "loss": 0.7197,
+      "step": 19015
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.022281901167858e-06,
+      "loss": 0.708,
+      "step": 19016
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.021283499361409e-06,
+      "loss": 0.7168,
+      "step": 19017
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.020285163534617e-06,
+      "loss": 0.7305,
+      "step": 19018
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.019286893700719e-06,
+      "loss": 0.6924,
+      "step": 19019
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0182886898729435e-06,
+      "loss": 0.7061,
+      "step": 19020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.017290552064507e-06,
+      "loss": 0.7012,
+      "step": 19021
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0162924802886514e-06,
+      "loss": 0.7002,
+      "step": 19022
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.015294474558599e-06,
+      "loss": 0.7109,
+      "step": 19023
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.014296534887573e-06,
+      "loss": 0.708,
+      "step": 19024
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.013298661288801e-06,
+      "loss": 0.7061,
+      "step": 19025
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.012300853775503e-06,
+      "loss": 0.6963,
+      "step": 19026
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.011303112360912e-06,
+      "loss": 0.6963,
+      "step": 19027
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.010305437058241e-06,
+      "loss": 0.7246,
+      "step": 19028
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.009307827880716e-06,
+      "loss": 0.7158,
+      "step": 19029
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.008310284841551e-06,
+      "loss": 0.71,
+      "step": 19030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.007312807953976e-06,
+      "loss": 0.7148,
+      "step": 19031
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.006315397231204e-06,
+      "loss": 0.7012,
+      "step": 19032
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.005318052686458e-06,
+      "loss": 0.7031,
+      "step": 19033
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004320774332945e-06,
+      "loss": 0.7031,
+      "step": 19034
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.00332356218389e-06,
+      "loss": 0.7373,
+      "step": 19035
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.002326416252505e-06,
+      "loss": 0.749,
+      "step": 19036
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0013293365520075e-06,
+      "loss": 0.6973,
+      "step": 19037
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000332323095604e-06,
+      "loss": 0.6992,
+      "step": 19038
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.999335375896516e-06,
+      "loss": 0.7119,
+      "step": 19039
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9983384949679555e-06,
+      "loss": 0.7334,
+      "step": 19040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.997341680323127e-06,
+      "loss": 0.7061,
+      "step": 19041
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9963449319752385e-06,
+      "loss": 0.7334,
+      "step": 19042
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995348249937508e-06,
+      "loss": 0.7246,
+      "step": 19043
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.99435163422314e-06,
+      "loss": 0.7402,
+      "step": 19044
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.993355084845343e-06,
+      "loss": 0.7188,
+      "step": 19045
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.992358601817321e-06,
+      "loss": 0.7051,
+      "step": 19046
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.991362185152282e-06,
+      "loss": 0.7041,
+      "step": 19047
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.99036583486343e-06,
+      "loss": 0.7246,
+      "step": 19048
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989369550963969e-06,
+      "loss": 0.7256,
+      "step": 19049
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.988373333467102e-06,
+      "loss": 0.6895,
+      "step": 19050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.98737718238603e-06,
+      "loss": 0.7178,
+      "step": 19051
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.986381097733956e-06,
+      "loss": 0.7324,
+      "step": 19052
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.985385079524087e-06,
+      "loss": 0.7051,
+      "step": 19053
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.984389127769607e-06,
+      "loss": 0.7148,
+      "step": 19054
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983393242483727e-06,
+      "loss": 0.7119,
+      "step": 19055
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.982397423679641e-06,
+      "loss": 0.7061,
+      "step": 19056
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.981401671370547e-06,
+      "loss": 0.7383,
+      "step": 19057
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9804059855696405e-06,
+      "loss": 0.7148,
+      "step": 19058
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.979410366290116e-06,
+      "loss": 0.7275,
+      "step": 19059
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.978414813545168e-06,
+      "loss": 0.7188,
+      "step": 19060
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.977419327347991e-06,
+      "loss": 0.7197,
+      "step": 19061
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.976423907711776e-06,
+      "loss": 0.7256,
+      "step": 19062
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.975428554649711e-06,
+      "loss": 0.7021,
+      "step": 19063
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.974433268174996e-06,
+      "loss": 0.7207,
+      "step": 19064
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.973438048300816e-06,
+      "loss": 0.7119,
+      "step": 19065
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.972442895040361e-06,
+      "loss": 0.7227,
+      "step": 19066
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.971447808406815e-06,
+      "loss": 0.7188,
+      "step": 19067
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.97045278841337e-06,
+      "loss": 0.7168,
+      "step": 19068
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.969457835073212e-06,
+      "loss": 0.7129,
+      "step": 19069
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.968462948399523e-06,
+      "loss": 0.7002,
+      "step": 19070
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.967468128405487e-06,
+      "loss": 0.6924,
+      "step": 19071
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.966473375104297e-06,
+      "loss": 0.7285,
+      "step": 19072
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.965478688509127e-06,
+      "loss": 0.6992,
+      "step": 19073
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9644840686331595e-06,
+      "loss": 0.7109,
+      "step": 19074
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.963489515489574e-06,
+      "loss": 0.7227,
+      "step": 19075
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.962495029091558e-06,
+      "loss": 0.6953,
+      "step": 19076
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.961500609452287e-06,
+      "loss": 0.7246,
+      "step": 19077
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.960506256584938e-06,
+      "loss": 0.7197,
+      "step": 19078
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.959511970502689e-06,
+      "loss": 0.7363,
+      "step": 19079
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.958517751218719e-06,
+      "loss": 0.7119,
+      "step": 19080
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.957523598746201e-06,
+      "loss": 0.7324,
+      "step": 19081
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9565295130983106e-06,
+      "loss": 0.7402,
+      "step": 19082
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.955535494288219e-06,
+      "loss": 0.7363,
+      "step": 19083
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.954541542329104e-06,
+      "loss": 0.7051,
+      "step": 19084
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9535476572341416e-06,
+      "loss": 0.7021,
+      "step": 19085
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9525538390164915e-06,
+      "loss": 0.6943,
+      "step": 19086
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.951560087689329e-06,
+      "loss": 0.7178,
+      "step": 19087
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.950566403265826e-06,
+      "loss": 0.7061,
+      "step": 19088
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9495727857591514e-06,
+      "loss": 0.7383,
+      "step": 19089
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.948579235182471e-06,
+      "loss": 0.7344,
+      "step": 19090
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.947585751548952e-06,
+      "loss": 0.709,
+      "step": 19091
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.946592334871762e-06,
+      "loss": 0.7227,
+      "step": 19092
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9455989851640615e-06,
+      "loss": 0.7012,
+      "step": 19093
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.94460570243902e-06,
+      "loss": 0.7158,
+      "step": 19094
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.943612486709798e-06,
+      "loss": 0.71,
+      "step": 19095
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.942619337989555e-06,
+      "loss": 0.7031,
+      "step": 19096
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.94162625629146e-06,
+      "loss": 0.7246,
+      "step": 19097
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9406332416286725e-06,
+      "loss": 0.7383,
+      "step": 19098
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.939640294014344e-06,
+      "loss": 0.7148,
+      "step": 19099
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.938647413461641e-06,
+      "loss": 0.6973,
+      "step": 19100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.93765459998372e-06,
+      "loss": 0.708,
+      "step": 19101
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.936661853593737e-06,
+      "loss": 0.7422,
+      "step": 19102
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.935669174304849e-06,
+      "loss": 0.7324,
+      "step": 19103
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9346765621302074e-06,
+      "loss": 0.7119,
+      "step": 19104
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.933684017082978e-06,
+      "loss": 0.71,
+      "step": 19105
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.932691539176303e-06,
+      "loss": 0.7285,
+      "step": 19106
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9316991284233385e-06,
+      "loss": 0.7275,
+      "step": 19107
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.930706784837231e-06,
+      "loss": 0.7285,
+      "step": 19108
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.929714508431142e-06,
+      "loss": 0.7158,
+      "step": 19109
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.928722299218217e-06,
+      "loss": 0.7168,
+      "step": 19110
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.927730157211608e-06,
+      "loss": 0.6982,
+      "step": 19111
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.926738082424451e-06,
+      "loss": 0.7393,
+      "step": 19112
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.925746074869907e-06,
+      "loss": 0.6934,
+      "step": 19113
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9247541345611175e-06,
+      "loss": 0.6924,
+      "step": 19114
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.923762261511227e-06,
+      "loss": 0.6914,
+      "step": 19115
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.922770455733378e-06,
+      "loss": 0.6934,
+      "step": 19116
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.92177871724072e-06,
+      "loss": 0.6953,
+      "step": 19117
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.920787046046397e-06,
+      "loss": 0.7246,
+      "step": 19118
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.919795442163544e-06,
+      "loss": 0.6973,
+      "step": 19119
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.918803905605302e-06,
+      "loss": 0.7197,
+      "step": 19120
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.917812436384817e-06,
+      "loss": 0.7217,
+      "step": 19121
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9168210345152255e-06,
+      "loss": 0.708,
+      "step": 19122
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9158297000096665e-06,
+      "loss": 0.7178,
+      "step": 19123
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.914838432881278e-06,
+      "loss": 0.708,
+      "step": 19124
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.913847233143195e-06,
+      "loss": 0.71,
+      "step": 19125
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.912856100808554e-06,
+      "loss": 0.7119,
+      "step": 19126
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.91186503589049e-06,
+      "loss": 0.7021,
+      "step": 19127
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.910874038402131e-06,
+      "loss": 0.7305,
+      "step": 19128
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9098831083566215e-06,
+      "loss": 0.7373,
+      "step": 19129
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.908892245767088e-06,
+      "loss": 0.707,
+      "step": 19130
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.907901450646664e-06,
+      "loss": 0.7373,
+      "step": 19131
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.90691072300847e-06,
+      "loss": 0.7051,
+      "step": 19132
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.905920062865648e-06,
+      "loss": 0.7051,
+      "step": 19133
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.90492947023132e-06,
+      "loss": 0.7246,
+      "step": 19134
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.903938945118616e-06,
+      "loss": 0.7373,
+      "step": 19135
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9029484875406615e-06,
+      "loss": 0.7109,
+      "step": 19136
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.901958097510583e-06,
+      "loss": 0.7168,
+      "step": 19137
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.900967775041506e-06,
+      "loss": 0.7002,
+      "step": 19138
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.899977520146553e-06,
+      "loss": 0.7227,
+      "step": 19139
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.898987332838848e-06,
+      "loss": 0.7119,
+      "step": 19140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89799721313151e-06,
+      "loss": 0.7256,
+      "step": 19141
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.897007161037668e-06,
+      "loss": 0.7256,
+      "step": 19142
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8960171765704376e-06,
+      "loss": 0.7441,
+      "step": 19143
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.895027259742941e-06,
+      "loss": 0.7324,
+      "step": 19144
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.894037410568293e-06,
+      "loss": 0.7061,
+      "step": 19145
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8930476290596145e-06,
+      "loss": 0.709,
+      "step": 19146
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8920579152300195e-06,
+      "loss": 0.7256,
+      "step": 19147
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.891068269092628e-06,
+      "loss": 0.6992,
+      "step": 19148
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.890078690660548e-06,
+      "loss": 0.7148,
+      "step": 19149
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.889089179946908e-06,
+      "loss": 0.7002,
+      "step": 19150
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.888099736964807e-06,
+      "loss": 0.707,
+      "step": 19151
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8871103617273616e-06,
+      "loss": 0.7158,
+      "step": 19152
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.886121054247683e-06,
+      "loss": 0.6982,
+      "step": 19153
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8851318145388845e-06,
+      "loss": 0.7168,
+      "step": 19154
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.884142642614076e-06,
+      "loss": 0.7334,
+      "step": 19155
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8831535384863645e-06,
+      "loss": 0.6904,
+      "step": 19156
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.882164502168859e-06,
+      "loss": 0.7021,
+      "step": 19157
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.881175533674664e-06,
+      "loss": 0.7266,
+      "step": 19158
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.880186633016889e-06,
+      "loss": 0.7236,
+      "step": 19159
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.879197800208637e-06,
+      "loss": 0.7246,
+      "step": 19160
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.87820903526301e-06,
+      "loss": 0.7129,
+      "step": 19161
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.877220338193118e-06,
+      "loss": 0.6973,
+      "step": 19162
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.876231709012064e-06,
+      "loss": 0.7051,
+      "step": 19163
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.875243147732941e-06,
+      "loss": 0.7158,
+      "step": 19164
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.874254654368852e-06,
+      "loss": 0.7266,
+      "step": 19165
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.873266228932902e-06,
+      "loss": 0.7012,
+      "step": 19166
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.872277871438188e-06,
+      "loss": 0.7148,
+      "step": 19167
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.871289581897808e-06,
+      "loss": 0.7158,
+      "step": 19168
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.870301360324857e-06,
+      "loss": 0.6973,
+      "step": 19169
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.869313206732434e-06,
+      "loss": 0.7148,
+      "step": 19170
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.868325121133633e-06,
+      "loss": 0.7109,
+      "step": 19171
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8673371035415475e-06,
+      "loss": 0.7168,
+      "step": 19172
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8663491539692685e-06,
+      "loss": 0.7041,
+      "step": 19173
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.865361272429897e-06,
+      "loss": 0.7354,
+      "step": 19174
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.864373458936519e-06,
+      "loss": 0.7217,
+      "step": 19175
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.863385713502231e-06,
+      "loss": 0.7109,
+      "step": 19176
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8623980361401106e-06,
+      "loss": 0.707,
+      "step": 19177
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.861410426863258e-06,
+      "loss": 0.709,
+      "step": 19178
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.860422885684757e-06,
+      "loss": 0.7285,
+      "step": 19179
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.859435412617697e-06,
+      "loss": 0.7285,
+      "step": 19180
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8584480076751625e-06,
+      "loss": 0.7217,
+      "step": 19181
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.857460670870235e-06,
+      "loss": 0.75,
+      "step": 19182
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.856473402216012e-06,
+      "loss": 0.7119,
+      "step": 19183
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.855486201725565e-06,
+      "loss": 0.6875,
+      "step": 19184
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.85449906941198e-06,
+      "loss": 0.7188,
+      "step": 19185
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.853512005288335e-06,
+      "loss": 0.71,
+      "step": 19186
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.852525009367719e-06,
+      "loss": 0.6855,
+      "step": 19187
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.851538081663209e-06,
+      "loss": 0.7305,
+      "step": 19188
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.850551222187881e-06,
+      "loss": 0.7334,
+      "step": 19189
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.849564430954817e-06,
+      "loss": 0.707,
+      "step": 19190
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8485777079770925e-06,
+      "loss": 0.7266,
+      "step": 19191
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.847591053267784e-06,
+      "loss": 0.7129,
+      "step": 19192
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.846604466839967e-06,
+      "loss": 0.708,
+      "step": 19193
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.845617948706712e-06,
+      "loss": 0.7041,
+      "step": 19194
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8446314988811005e-06,
+      "loss": 0.7129,
+      "step": 19195
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.843645117376205e-06,
+      "loss": 0.7656,
+      "step": 19196
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8426588042050915e-06,
+      "loss": 0.7051,
+      "step": 19197
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.841672559380829e-06,
+      "loss": 0.7314,
+      "step": 19198
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8406863829164965e-06,
+      "loss": 0.6943,
+      "step": 19199
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.839700274825158e-06,
+      "loss": 0.7266,
+      "step": 19200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8387142351198814e-06,
+      "loss": 0.7383,
+      "step": 19201
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.837728263813736e-06,
+      "loss": 0.7012,
+      "step": 19202
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8367423609197864e-06,
+      "loss": 0.7109,
+      "step": 19203
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8357565264511e-06,
+      "loss": 0.7119,
+      "step": 19204
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8347707604207396e-06,
+      "loss": 0.7441,
+      "step": 19205
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.833785062841766e-06,
+      "loss": 0.7207,
+      "step": 19206
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.832799433727249e-06,
+      "loss": 0.7168,
+      "step": 19207
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.831813873090248e-06,
+      "loss": 0.7275,
+      "step": 19208
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.830828380943826e-06,
+      "loss": 0.7275,
+      "step": 19209
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.829842957301033e-06,
+      "loss": 0.7109,
+      "step": 19210
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.82885760217494e-06,
+      "loss": 0.7178,
+      "step": 19211
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8278723155786e-06,
+      "loss": 0.6992,
+      "step": 19212
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.826887097525071e-06,
+      "loss": 0.7295,
+      "step": 19213
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.82590194802741e-06,
+      "loss": 0.6904,
+      "step": 19214
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.824916867098671e-06,
+      "loss": 0.71,
+      "step": 19215
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.82393185475191e-06,
+      "loss": 0.6953,
+      "step": 19216
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.82294691100018e-06,
+      "loss": 0.7119,
+      "step": 19217
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.821962035856535e-06,
+      "loss": 0.7295,
+      "step": 19218
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.820977229334022e-06,
+      "loss": 0.7129,
+      "step": 19219
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8199924914457e-06,
+      "loss": 0.7139,
+      "step": 19220
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.819007822204616e-06,
+      "loss": 0.7266,
+      "step": 19221
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.818023221623817e-06,
+      "loss": 0.708,
+      "step": 19222
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8170386897163525e-06,
+      "loss": 0.709,
+      "step": 19223
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8160542264952704e-06,
+      "loss": 0.7129,
+      "step": 19224
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.815069831973616e-06,
+      "loss": 0.7061,
+      "step": 19225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.814085506164436e-06,
+      "loss": 0.6914,
+      "step": 19226
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.813101249080772e-06,
+      "loss": 0.7275,
+      "step": 19227
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.812117060735676e-06,
+      "loss": 0.71,
+      "step": 19228
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.811132941142182e-06,
+      "loss": 0.6924,
+      "step": 19229
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.810148890313335e-06,
+      "loss": 0.6943,
+      "step": 19230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.809164908262172e-06,
+      "loss": 0.709,
+      "step": 19231
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.808180995001741e-06,
+      "loss": 0.7197,
+      "step": 19232
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.807197150545076e-06,
+      "loss": 0.6982,
+      "step": 19233
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8062133749052185e-06,
+      "loss": 0.7109,
+      "step": 19234
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.805229668095203e-06,
+      "loss": 0.751,
+      "step": 19235
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8042460301280654e-06,
+      "loss": 0.7051,
+      "step": 19236
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.803262461016844e-06,
+      "loss": 0.6934,
+      "step": 19237
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8022789607745715e-06,
+      "loss": 0.7148,
+      "step": 19238
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.801295529414277e-06,
+      "loss": 0.7256,
+      "step": 19239
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.800312166949003e-06,
+      "loss": 0.7266,
+      "step": 19240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.79932887339178e-06,
+      "loss": 0.7227,
+      "step": 19241
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.798345648755632e-06,
+      "loss": 0.7148,
+      "step": 19242
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.797362493053589e-06,
+      "loss": 0.7256,
+      "step": 19243
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.796379406298687e-06,
+      "loss": 0.7354,
+      "step": 19244
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7953963885039515e-06,
+      "loss": 0.6924,
+      "step": 19245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.794413439682408e-06,
+      "loss": 0.7197,
+      "step": 19246
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.793430559847084e-06,
+      "loss": 0.6934,
+      "step": 19247
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.792447749011007e-06,
+      "loss": 0.6924,
+      "step": 19248
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.791465007187197e-06,
+      "loss": 0.7295,
+      "step": 19249
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.790482334388682e-06,
+      "loss": 0.7256,
+      "step": 19250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.789499730628477e-06,
+      "loss": 0.6914,
+      "step": 19251
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.788517195919615e-06,
+      "loss": 0.7178,
+      "step": 19252
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.787534730275112e-06,
+      "loss": 0.7197,
+      "step": 19253
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7865523337079915e-06,
+      "loss": 0.7061,
+      "step": 19254
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.78557000623126e-06,
+      "loss": 0.7324,
+      "step": 19255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.784587747857948e-06,
+      "loss": 0.7197,
+      "step": 19256
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.783605558601068e-06,
+      "loss": 0.7109,
+      "step": 19257
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.782623438473639e-06,
+      "loss": 0.7324,
+      "step": 19258
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.781641387488675e-06,
+      "loss": 0.7422,
+      "step": 19259
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.78065940565919e-06,
+      "loss": 0.7168,
+      "step": 19260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7796774929981955e-06,
+      "loss": 0.7002,
+      "step": 19261
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.778695649518709e-06,
+      "loss": 0.7305,
+      "step": 19262
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.777713875233738e-06,
+      "loss": 0.6982,
+      "step": 19263
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.776732170156292e-06,
+      "loss": 0.7051,
+      "step": 19264
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.775750534299387e-06,
+      "loss": 0.7207,
+      "step": 19265
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774768967676028e-06,
+      "loss": 0.6914,
+      "step": 19266
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.773787470299225e-06,
+      "loss": 0.7021,
+      "step": 19267
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.772806042181983e-06,
+      "loss": 0.7295,
+      "step": 19268
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.77182468333731e-06,
+      "loss": 0.71,
+      "step": 19269
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.770843393778211e-06,
+      "loss": 0.7354,
+      "step": 19270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7698621735176885e-06,
+      "loss": 0.6943,
+      "step": 19271
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.768881022568745e-06,
+      "loss": 0.7051,
+      "step": 19272
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.767899940944392e-06,
+      "loss": 0.7148,
+      "step": 19273
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.76691892865762e-06,
+      "loss": 0.7041,
+      "step": 19274
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.765937985721435e-06,
+      "loss": 0.7324,
+      "step": 19275
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7649571121488326e-06,
+      "loss": 0.6963,
+      "step": 19276
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.763976307952818e-06,
+      "loss": 0.7412,
+      "step": 19277
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7629955731463875e-06,
+      "loss": 0.7285,
+      "step": 19278
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.762014907742536e-06,
+      "loss": 0.7354,
+      "step": 19279
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.76103431175426e-06,
+      "loss": 0.7305,
+      "step": 19280
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.760053785194556e-06,
+      "loss": 0.7051,
+      "step": 19281
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7590733280764165e-06,
+      "loss": 0.7363,
+      "step": 19282
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7580929404128375e-06,
+      "loss": 0.7354,
+      "step": 19283
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.757112622216804e-06,
+      "loss": 0.7129,
+      "step": 19284
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.756132373501319e-06,
+      "loss": 0.7168,
+      "step": 19285
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.75515219427937e-06,
+      "loss": 0.7148,
+      "step": 19286
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.754172084563939e-06,
+      "loss": 0.708,
+      "step": 19287
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.753192044368018e-06,
+      "loss": 0.7266,
+      "step": 19288
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.752212073704599e-06,
+      "loss": 0.7109,
+      "step": 19289
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.751232172586668e-06,
+      "loss": 0.7002,
+      "step": 19290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.750252341027209e-06,
+      "loss": 0.7207,
+      "step": 19291
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.749272579039207e-06,
+      "loss": 0.7207,
+      "step": 19292
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.748292886635647e-06,
+      "loss": 0.7061,
+      "step": 19293
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.747313263829512e-06,
+      "loss": 0.6807,
+      "step": 19294
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746333710633783e-06,
+      "loss": 0.708,
+      "step": 19295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.745354227061442e-06,
+      "loss": 0.6855,
+      "step": 19296
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.744374813125471e-06,
+      "loss": 0.7041,
+      "step": 19297
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.743395468838849e-06,
+      "loss": 0.7412,
+      "step": 19298
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.742416194214558e-06,
+      "loss": 0.6914,
+      "step": 19299
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7414369892655654e-06,
+      "loss": 0.7266,
+      "step": 19300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.740457854004857e-06,
+      "loss": 0.708,
+      "step": 19301
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.739478788445406e-06,
+      "loss": 0.71,
+      "step": 19302
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.738499792600189e-06,
+      "loss": 0.7051,
+      "step": 19303
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.737520866482176e-06,
+      "loss": 0.7158,
+      "step": 19304
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.73654201010434e-06,
+      "loss": 0.7188,
+      "step": 19305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735563223479663e-06,
+      "loss": 0.6846,
+      "step": 19306
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7345845066211046e-06,
+      "loss": 0.6992,
+      "step": 19307
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.73360585954164e-06,
+      "loss": 0.7197,
+      "step": 19308
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.732627282254234e-06,
+      "loss": 0.707,
+      "step": 19309
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.731648774771862e-06,
+      "loss": 0.7061,
+      "step": 19310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.730670337107489e-06,
+      "loss": 0.7129,
+      "step": 19311
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.729691969274081e-06,
+      "loss": 0.7217,
+      "step": 19312
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7287136712846025e-06,
+      "loss": 0.7021,
+      "step": 19313
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.727735443152021e-06,
+      "loss": 0.7324,
+      "step": 19314
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.726757284889297e-06,
+      "loss": 0.7158,
+      "step": 19315
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.725779196509397e-06,
+      "loss": 0.7139,
+      "step": 19316
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.724801178025275e-06,
+      "loss": 0.7344,
+      "step": 19317
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.723823229449904e-06,
+      "loss": 0.7275,
+      "step": 19318
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.722845350796241e-06,
+      "loss": 0.6924,
+      "step": 19319
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.72186754207724e-06,
+      "loss": 0.708,
+      "step": 19320
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7208898033058555e-06,
+      "loss": 0.7168,
+      "step": 19321
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.719912134495057e-06,
+      "loss": 0.7012,
+      "step": 19322
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.718934535657793e-06,
+      "loss": 0.7451,
+      "step": 19323
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.717957006807022e-06,
+      "loss": 0.7158,
+      "step": 19324
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.716979547955698e-06,
+      "loss": 0.7314,
+      "step": 19325
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.716002159116773e-06,
+      "loss": 0.708,
+      "step": 19326
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.715024840303201e-06,
+      "loss": 0.7412,
+      "step": 19327
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.714047591527932e-06,
+      "loss": 0.7256,
+      "step": 19328
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.713070412803917e-06,
+      "loss": 0.7041,
+      "step": 19329
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7120933041441095e-06,
+      "loss": 0.7471,
+      "step": 19330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.711116265561457e-06,
+      "loss": 0.7061,
+      "step": 19331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.71013929706891e-06,
+      "loss": 0.7197,
+      "step": 19332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.709162398679406e-06,
+      "loss": 0.7256,
+      "step": 19333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7081855704059e-06,
+      "loss": 0.6992,
+      "step": 19334
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.707208812261335e-06,
+      "loss": 0.7227,
+      "step": 19335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.706232124258655e-06,
+      "loss": 0.7109,
+      "step": 19336
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.705255506410804e-06,
+      "loss": 0.7227,
+      "step": 19337
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.704278958730724e-06,
+      "loss": 0.7285,
+      "step": 19338
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.703302481231356e-06,
+      "loss": 0.707,
+      "step": 19339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7023260739256426e-06,
+      "loss": 0.6855,
+      "step": 19340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.701349736826517e-06,
+      "loss": 0.708,
+      "step": 19341
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.700373469946927e-06,
+      "loss": 0.7168,
+      "step": 19342
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.699397273299807e-06,
+      "loss": 0.7021,
+      "step": 19343
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.698421146898093e-06,
+      "loss": 0.7314,
+      "step": 19344
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.697445090754722e-06,
+      "loss": 0.7256,
+      "step": 19345
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.696469104882627e-06,
+      "loss": 0.7188,
+      "step": 19346
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.695493189294745e-06,
+      "loss": 0.709,
+      "step": 19347
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.694517344004007e-06,
+      "loss": 0.7236,
+      "step": 19348
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6935415690233455e-06,
+      "loss": 0.7061,
+      "step": 19349
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.692565864365689e-06,
+      "loss": 0.7393,
+      "step": 19350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6915902300439776e-06,
+      "loss": 0.71,
+      "step": 19351
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.690614666071131e-06,
+      "loss": 0.6982,
+      "step": 19352
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6896391724600805e-06,
+      "loss": 0.7236,
+      "step": 19353
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.688663749223752e-06,
+      "loss": 0.7061,
+      "step": 19354
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.687688396375076e-06,
+      "loss": 0.709,
+      "step": 19355
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.686713113926978e-06,
+      "loss": 0.71,
+      "step": 19356
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.68573790189238e-06,
+      "loss": 0.6924,
+      "step": 19357
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6847627602842075e-06,
+      "loss": 0.6836,
+      "step": 19358
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6837876891153825e-06,
+      "loss": 0.6836,
+      "step": 19359
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.682812688398828e-06,
+      "loss": 0.7256,
+      "step": 19360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.681837758147464e-06,
+      "loss": 0.7031,
+      "step": 19361
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6808628983742075e-06,
+      "loss": 0.7041,
+      "step": 19362
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.679888109091986e-06,
+      "loss": 0.7119,
+      "step": 19363
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6789133903137155e-06,
+      "loss": 0.71,
+      "step": 19364
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.677938742052307e-06,
+      "loss": 0.7139,
+      "step": 19365
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.676964164320678e-06,
+      "loss": 0.7002,
+      "step": 19366
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.675989657131749e-06,
+      "loss": 0.7363,
+      "step": 19367
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.675015220498433e-06,
+      "loss": 0.7148,
+      "step": 19368
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6740408544336426e-06,
+      "loss": 0.7041,
+      "step": 19369
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.67306655895029e-06,
+      "loss": 0.7266,
+      "step": 19370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.672092334061288e-06,
+      "loss": 0.6904,
+      "step": 19371
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6711181797795465e-06,
+      "loss": 0.7314,
+      "step": 19372
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.670144096117977e-06,
+      "loss": 0.6904,
+      "step": 19373
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6691700830894815e-06,
+      "loss": 0.7314,
+      "step": 19374
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.668196140706977e-06,
+      "loss": 0.7158,
+      "step": 19375
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.667222268983368e-06,
+      "loss": 0.7129,
+      "step": 19376
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6662484679315635e-06,
+      "loss": 0.7344,
+      "step": 19377
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.665274737564457e-06,
+      "loss": 0.7188,
+      "step": 19378
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.664301077894964e-06,
+      "loss": 0.7363,
+      "step": 19379
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.663327488935984e-06,
+      "loss": 0.6914,
+      "step": 19380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.662353970700419e-06,
+      "loss": 0.7285,
+      "step": 19381
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.661380523201173e-06,
+      "loss": 0.7227,
+      "step": 19382
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.660407146451138e-06,
+      "loss": 0.6992,
+      "step": 19383
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6594338404632285e-06,
+      "loss": 0.7031,
+      "step": 19384
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.658460605250331e-06,
+      "loss": 0.7383,
+      "step": 19385
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.657487440825342e-06,
+      "loss": 0.7109,
+      "step": 19386
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6565143472011685e-06,
+      "loss": 0.7422,
+      "step": 19387
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.655541324390699e-06,
+      "loss": 0.7285,
+      "step": 19388
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.65456837240683e-06,
+      "loss": 0.708,
+      "step": 19389
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.653595491262456e-06,
+      "loss": 0.7012,
+      "step": 19390
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.652622680970469e-06,
+      "loss": 0.7168,
+      "step": 19391
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.651649941543762e-06,
+      "loss": 0.7305,
+      "step": 19392
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.650677272995225e-06,
+      "loss": 0.7246,
+      "step": 19393
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6497046753377474e-06,
+      "loss": 0.7139,
+      "step": 19394
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6487321485842175e-06,
+      "loss": 0.7129,
+      "step": 19395
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6477596927475274e-06,
+      "loss": 0.7139,
+      "step": 19396
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6467873078405675e-06,
+      "loss": 0.7207,
+      "step": 19397
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.645814993876217e-06,
+      "loss": 0.7197,
+      "step": 19398
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.644842750867358e-06,
+      "loss": 0.6953,
+      "step": 19399
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643870578826885e-06,
+      "loss": 0.7178,
+      "step": 19400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.642898477767676e-06,
+      "loss": 0.7314,
+      "step": 19401
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.641926447702616e-06,
+      "loss": 0.7422,
+      "step": 19402
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.640954488644585e-06,
+      "loss": 0.71,
+      "step": 19403
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.639982600606464e-06,
+      "loss": 0.7236,
+      "step": 19404
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.639010783601133e-06,
+      "loss": 0.7246,
+      "step": 19405
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.638039037641472e-06,
+      "loss": 0.7051,
+      "step": 19406
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6370673627403526e-06,
+      "loss": 0.7158,
+      "step": 19407
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.636095758910662e-06,
+      "loss": 0.6895,
+      "step": 19408
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.63512422616527e-06,
+      "loss": 0.7402,
+      "step": 19409
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.634152764517057e-06,
+      "loss": 0.6992,
+      "step": 19410
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6331813739788855e-06,
+      "loss": 0.7148,
+      "step": 19411
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.632210054563639e-06,
+      "loss": 0.7285,
+      "step": 19412
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.631238806284188e-06,
+      "loss": 0.7422,
+      "step": 19413
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.630267629153403e-06,
+      "loss": 0.7178,
+      "step": 19414
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.629296523184154e-06,
+      "loss": 0.6963,
+      "step": 19415
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.62832548838931e-06,
+      "loss": 0.7109,
+      "step": 19416
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.627354524781741e-06,
+      "loss": 0.6953,
+      "step": 19417
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6263836323743125e-06,
+      "loss": 0.6855,
+      "step": 19418
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.625412811179889e-06,
+      "loss": 0.7236,
+      "step": 19419
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.624442061211344e-06,
+      "loss": 0.7207,
+      "step": 19420
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.623471382481536e-06,
+      "loss": 0.7021,
+      "step": 19421
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6225007750033315e-06,
+      "loss": 0.6943,
+      "step": 19422
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.621530238789591e-06,
+      "loss": 0.6924,
+      "step": 19423
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.620559773853178e-06,
+      "loss": 0.6953,
+      "step": 19424
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.619589380206954e-06,
+      "loss": 0.7061,
+      "step": 19425
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.618619057863777e-06,
+      "loss": 0.6797,
+      "step": 19426
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617648806836507e-06,
+      "loss": 0.7178,
+      "step": 19427
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6166786271379985e-06,
+      "loss": 0.7119,
+      "step": 19428
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.615708518781119e-06,
+      "loss": 0.707,
+      "step": 19429
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.614738481778714e-06,
+      "loss": 0.7061,
+      "step": 19430
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.613768516143644e-06,
+      "loss": 0.7471,
+      "step": 19431
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.612798621888758e-06,
+      "loss": 0.7041,
+      "step": 19432
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6118287990269174e-06,
+      "loss": 0.7178,
+      "step": 19433
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.61085904757097e-06,
+      "loss": 0.7168,
+      "step": 19434
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.609889367533768e-06,
+      "loss": 0.7178,
+      "step": 19435
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.608919758928162e-06,
+      "loss": 0.7168,
+      "step": 19436
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.607950221767001e-06,
+      "loss": 0.7256,
+      "step": 19437
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6069807560631355e-06,
+      "loss": 0.7246,
+      "step": 19438
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.60601136182941e-06,
+      "loss": 0.7148,
+      "step": 19439
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6050420390786695e-06,
+      "loss": 0.7197,
+      "step": 19440
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.604072787823768e-06,
+      "loss": 0.7217,
+      "step": 19441
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.60310360807755e-06,
+      "loss": 0.7373,
+      "step": 19442
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.60213449985285e-06,
+      "loss": 0.7051,
+      "step": 19443
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.6011654631625124e-06,
+      "loss": 0.7217,
+      "step": 19444
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.6001964980193854e-06,
+      "loss": 0.708,
+      "step": 19445
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.599227604436309e-06,
+      "loss": 0.7051,
+      "step": 19446
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.598258782426123e-06,
+      "loss": 0.7334,
+      "step": 19447
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.597290032001663e-06,
+      "loss": 0.7197,
+      "step": 19448
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.596321353175771e-06,
+      "loss": 0.7188,
+      "step": 19449
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5953527459612825e-06,
+      "loss": 0.7168,
+      "step": 19450
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.594384210371033e-06,
+      "loss": 0.7363,
+      "step": 19451
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5934157464178575e-06,
+      "loss": 0.6924,
+      "step": 19452
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.592447354114594e-06,
+      "loss": 0.7275,
+      "step": 19453
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591479033474075e-06,
+      "loss": 0.7129,
+      "step": 19454
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.590510784509134e-06,
+      "loss": 0.6953,
+      "step": 19455
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.589542607232594e-06,
+      "loss": 0.7227,
+      "step": 19456
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.588574501657296e-06,
+      "loss": 0.7148,
+      "step": 19457
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.587606467796065e-06,
+      "loss": 0.6953,
+      "step": 19458
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5866385056617314e-06,
+      "loss": 0.7158,
+      "step": 19459
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.585670615267123e-06,
+      "loss": 0.7246,
+      "step": 19460
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.584702796625064e-06,
+      "loss": 0.6943,
+      "step": 19461
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.583735049748383e-06,
+      "loss": 0.7314,
+      "step": 19462
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.582767374649902e-06,
+      "loss": 0.7188,
+      "step": 19463
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5817997713424444e-06,
+      "loss": 0.707,
+      "step": 19464
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.58083223983884e-06,
+      "loss": 0.7061,
+      "step": 19465
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.579864780151905e-06,
+      "loss": 0.6895,
+      "step": 19466
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.578897392294463e-06,
+      "loss": 0.7129,
+      "step": 19467
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.577930076279332e-06,
+      "loss": 0.6973,
+      "step": 19468
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.576962832119334e-06,
+      "loss": 0.707,
+      "step": 19469
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.575995659827284e-06,
+      "loss": 0.7334,
+      "step": 19470
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.575028559416002e-06,
+      "loss": 0.707,
+      "step": 19471
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.574061530898303e-06,
+      "loss": 0.7168,
+      "step": 19472
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.573094574286998e-06,
+      "loss": 0.6982,
+      "step": 19473
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.572127689594915e-06,
+      "loss": 0.7217,
+      "step": 19474
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.571160876834854e-06,
+      "loss": 0.71,
+      "step": 19475
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.570194136019632e-06,
+      "loss": 0.7402,
+      "step": 19476
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.569227467162058e-06,
+      "loss": 0.7158,
+      "step": 19477
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.568260870274947e-06,
+      "loss": 0.7061,
+      "step": 19478
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5672943453711084e-06,
+      "loss": 0.7227,
+      "step": 19479
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.566327892463349e-06,
+      "loss": 0.7246,
+      "step": 19480
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.565361511564478e-06,
+      "loss": 0.7139,
+      "step": 19481
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5643952026873e-06,
+      "loss": 0.7188,
+      "step": 19482
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.563428965844623e-06,
+      "loss": 0.709,
+      "step": 19483
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.562462801049251e-06,
+      "loss": 0.7227,
+      "step": 19484
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.561496708313984e-06,
+      "loss": 0.7471,
+      "step": 19485
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560530687651633e-06,
+      "loss": 0.7178,
+      "step": 19486
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.559564739075e-06,
+      "loss": 0.7344,
+      "step": 19487
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.558598862596877e-06,
+      "loss": 0.7188,
+      "step": 19488
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5576330582300665e-06,
+      "loss": 0.707,
+      "step": 19489
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.556667325987374e-06,
+      "loss": 0.6934,
+      "step": 19490
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.555701665881594e-06,
+      "loss": 0.7041,
+      "step": 19491
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.554736077925524e-06,
+      "loss": 0.7021,
+      "step": 19492
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.55377056213196e-06,
+      "loss": 0.7227,
+      "step": 19493
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5528051185136975e-06,
+      "loss": 0.71,
+      "step": 19494
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5518397470835315e-06,
+      "loss": 0.7412,
+      "step": 19495
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.550874447854254e-06,
+      "loss": 0.6943,
+      "step": 19496
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.549909220838655e-06,
+      "loss": 0.7324,
+      "step": 19497
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.548944066049533e-06,
+      "loss": 0.6904,
+      "step": 19498
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.547978983499676e-06,
+      "loss": 0.7256,
+      "step": 19499
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.547013973201875e-06,
+      "loss": 0.708,
+      "step": 19500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.546049035168909e-06,
+      "loss": 0.6904,
+      "step": 19501
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.545084169413576e-06,
+      "loss": 0.7031,
+      "step": 19502
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.54411937594866e-06,
+      "loss": 0.707,
+      "step": 19503
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.543154654786946e-06,
+      "loss": 0.71,
+      "step": 19504
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.542190005941221e-06,
+      "loss": 0.7061,
+      "step": 19505
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.541225429424261e-06,
+      "loss": 0.7109,
+      "step": 19506
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5402609252488625e-06,
+      "loss": 0.6973,
+      "step": 19507
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.539296493427797e-06,
+      "loss": 0.7051,
+      "step": 19508
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5383321339738454e-06,
+      "loss": 0.7422,
+      "step": 19509
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5373678468997925e-06,
+      "loss": 0.7012,
+      "step": 19510
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.536403632218417e-06,
+      "loss": 0.7188,
+      "step": 19511
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.535439489942495e-06,
+      "loss": 0.7285,
+      "step": 19512
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.534475420084804e-06,
+      "loss": 0.7324,
+      "step": 19513
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.533511422658121e-06,
+      "loss": 0.7168,
+      "step": 19514
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.532547497675219e-06,
+      "loss": 0.7334,
+      "step": 19515
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.531583645148876e-06,
+      "loss": 0.7178,
+      "step": 19516
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.530619865091862e-06,
+      "loss": 0.7178,
+      "step": 19517
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.529656157516948e-06,
+      "loss": 0.7266,
+      "step": 19518
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.528692522436909e-06,
+      "loss": 0.7314,
+      "step": 19519
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.52772895986452e-06,
+      "loss": 0.7158,
+      "step": 19520
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5267654698125405e-06,
+      "loss": 0.7129,
+      "step": 19521
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.525802052293739e-06,
+      "loss": 0.7119,
+      "step": 19522
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.524838707320892e-06,
+      "loss": 0.7158,
+      "step": 19523
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.523875434906761e-06,
+      "loss": 0.708,
+      "step": 19524
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.522912235064112e-06,
+      "loss": 0.6855,
+      "step": 19525
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.52194910780571e-06,
+      "loss": 0.7451,
+      "step": 19526
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.520986053144318e-06,
+      "loss": 0.7021,
+      "step": 19527
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5200230710926996e-06,
+      "loss": 0.7246,
+      "step": 19528
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.519060161663615e-06,
+      "loss": 0.7285,
+      "step": 19529
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.518097324869823e-06,
+      "loss": 0.7383,
+      "step": 19530
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.51713456072409e-06,
+      "loss": 0.7344,
+      "step": 19531
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.516171869239172e-06,
+      "loss": 0.7217,
+      "step": 19532
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.515209250427829e-06,
+      "loss": 0.7529,
+      "step": 19533
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.514246704302808e-06,
+      "loss": 0.71,
+      "step": 19534
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.513284230876877e-06,
+      "loss": 0.7314,
+      "step": 19535
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.512321830162786e-06,
+      "loss": 0.708,
+      "step": 19536
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.511359502173288e-06,
+      "loss": 0.7285,
+      "step": 19537
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.510397246921139e-06,
+      "loss": 0.71,
+      "step": 19538
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.509435064419089e-06,
+      "loss": 0.7354,
+      "step": 19539
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.508472954679889e-06,
+      "loss": 0.7061,
+      "step": 19540
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.507510917716291e-06,
+      "loss": 0.708,
+      "step": 19541
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.506548953541039e-06,
+      "loss": 0.7021,
+      "step": 19542
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.50558706216689e-06,
+      "loss": 0.6904,
+      "step": 19543
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.504625243606585e-06,
+      "loss": 0.7178,
+      "step": 19544
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.503663497872873e-06,
+      "loss": 0.7139,
+      "step": 19545
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.502701824978499e-06,
+      "loss": 0.7051,
+      "step": 19546
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5017402249362076e-06,
+      "loss": 0.7236,
+      "step": 19547
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.500778697758739e-06,
+      "loss": 0.7012,
+      "step": 19548
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.499817243458842e-06,
+      "loss": 0.7158,
+      "step": 19549
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.498855862049252e-06,
+      "loss": 0.7236,
+      "step": 19550
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.497894553542709e-06,
+      "loss": 0.7178,
+      "step": 19551
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.496933317951961e-06,
+      "loss": 0.7158,
+      "step": 19552
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.495972155289739e-06,
+      "loss": 0.7051,
+      "step": 19553
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.495011065568781e-06,
+      "loss": 0.752,
+      "step": 19554
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.494050048801827e-06,
+      "loss": 0.7334,
+      "step": 19555
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.493089105001611e-06,
+      "loss": 0.7158,
+      "step": 19556
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.492128234180869e-06,
+      "loss": 0.7109,
+      "step": 19557
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.491167436352334e-06,
+      "loss": 0.7139,
+      "step": 19558
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.490206711528738e-06,
+      "loss": 0.6914,
+      "step": 19559
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4892460597228135e-06,
+      "loss": 0.7344,
+      "step": 19560
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.488285480947292e-06,
+      "loss": 0.7334,
+      "step": 19561
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4873249752149014e-06,
+      "loss": 0.7363,
+      "step": 19562
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.486364542538369e-06,
+      "loss": 0.7246,
+      "step": 19563
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4854041829304305e-06,
+      "loss": 0.7256,
+      "step": 19564
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.484443896403812e-06,
+      "loss": 0.7031,
+      "step": 19565
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.48348368297123e-06,
+      "loss": 0.7207,
+      "step": 19566
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.482523542645414e-06,
+      "loss": 0.6865,
+      "step": 19567
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.481563475439091e-06,
+      "loss": 0.6904,
+      "step": 19568
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.480603481364985e-06,
+      "loss": 0.7031,
+      "step": 19569
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.479643560435814e-06,
+      "loss": 0.6992,
+      "step": 19570
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.478683712664298e-06,
+      "loss": 0.7041,
+      "step": 19571
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.477723938063166e-06,
+      "loss": 0.7148,
+      "step": 19572
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.47676423664513e-06,
+      "loss": 0.7031,
+      "step": 19573
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.475804608422909e-06,
+      "loss": 0.7109,
+      "step": 19574
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474845053409217e-06,
+      "loss": 0.7119,
+      "step": 19575
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.473885571616778e-06,
+      "loss": 0.7363,
+      "step": 19576
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.472926163058306e-06,
+      "loss": 0.7354,
+      "step": 19577
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4719668277465145e-06,
+      "loss": 0.7031,
+      "step": 19578
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.471007565694109e-06,
+      "loss": 0.7285,
+      "step": 19579
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.470048376913813e-06,
+      "loss": 0.71,
+      "step": 19580
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.469089261418333e-06,
+      "loss": 0.707,
+      "step": 19581
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.468130219220379e-06,
+      "loss": 0.7119,
+      "step": 19582
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.467171250332664e-06,
+      "loss": 0.7109,
+      "step": 19583
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.466212354767888e-06,
+      "loss": 0.7041,
+      "step": 19584
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.465253532538775e-06,
+      "loss": 0.6973,
+      "step": 19585
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.464294783658017e-06,
+      "loss": 0.7363,
+      "step": 19586
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.46333610813832e-06,
+      "loss": 0.6875,
+      "step": 19587
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.462377505992397e-06,
+      "loss": 0.7109,
+      "step": 19588
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.461418977232947e-06,
+      "loss": 0.7139,
+      "step": 19589
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.460460521872674e-06,
+      "loss": 0.708,
+      "step": 19590
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.459502139924279e-06,
+      "loss": 0.6982,
+      "step": 19591
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.458543831400463e-06,
+      "loss": 0.707,
+      "step": 19592
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.457585596313926e-06,
+      "loss": 0.707,
+      "step": 19593
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.456627434677366e-06,
+      "loss": 0.7148,
+      "step": 19594
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.455669346503483e-06,
+      "loss": 0.6914,
+      "step": 19595
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.454711331804968e-06,
+      "loss": 0.7393,
+      "step": 19596
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.453753390594525e-06,
+      "loss": 0.7373,
+      "step": 19597
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.45279552288485e-06,
+      "loss": 0.7109,
+      "step": 19598
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4518377286886276e-06,
+      "loss": 0.7119,
+      "step": 19599
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.450880008018552e-06,
+      "loss": 0.6904,
+      "step": 19600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.449922360887323e-06,
+      "loss": 0.7119,
+      "step": 19601
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.448964787307627e-06,
+      "loss": 0.6953,
+      "step": 19602
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4480072872921555e-06,
+      "loss": 0.7031,
+      "step": 19603
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.447049860853596e-06,
+      "loss": 0.7021,
+      "step": 19604
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.446092508004637e-06,
+      "loss": 0.7129,
+      "step": 19605
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.445135228757966e-06,
+      "loss": 0.7383,
+      "step": 19606
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.444178023126271e-06,
+      "loss": 0.6963,
+      "step": 19607
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.443220891122231e-06,
+      "loss": 0.7002,
+      "step": 19608
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.442263832758537e-06,
+      "loss": 0.7393,
+      "step": 19609
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.441306848047871e-06,
+      "loss": 0.7168,
+      "step": 19610
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.440349937002918e-06,
+      "loss": 0.7236,
+      "step": 19611
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.439393099636348e-06,
+      "loss": 0.7021,
+      "step": 19612
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.438436335960853e-06,
+      "loss": 0.6963,
+      "step": 19613
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4374796459891075e-06,
+      "loss": 0.7061,
+      "step": 19614
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.436523029733792e-06,
+      "loss": 0.7002,
+      "step": 19615
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.435566487207577e-06,
+      "loss": 0.7061,
+      "step": 19616
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4346100184231535e-06,
+      "loss": 0.7188,
+      "step": 19617
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.433653623393183e-06,
+      "loss": 0.6934,
+      "step": 19618
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.432697302130345e-06,
+      "loss": 0.6963,
+      "step": 19619
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.43174105464731e-06,
+      "loss": 0.7197,
+      "step": 19620
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.430784880956757e-06,
+      "loss": 0.71,
+      "step": 19621
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.429828781071354e-06,
+      "loss": 0.7139,
+      "step": 19622
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.428872755003771e-06,
+      "loss": 0.7109,
+      "step": 19623
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.427916802766678e-06,
+      "loss": 0.7256,
+      "step": 19624
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.426960924372744e-06,
+      "loss": 0.7295,
+      "step": 19625
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.426005119834637e-06,
+      "loss": 0.7275,
+      "step": 19626
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.425049389165022e-06,
+      "loss": 0.6875,
+      "step": 19627
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4240937323765665e-06,
+      "loss": 0.7051,
+      "step": 19628
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.423138149481931e-06,
+      "loss": 0.7148,
+      "step": 19629
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.422182640493788e-06,
+      "loss": 0.7217,
+      "step": 19630
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4212272054247926e-06,
+      "loss": 0.7334,
+      "step": 19631
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.420271844287605e-06,
+      "loss": 0.7246,
+      "step": 19632
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.419316557094892e-06,
+      "loss": 0.6982,
+      "step": 19633
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.418361343859313e-06,
+      "loss": 0.7012,
+      "step": 19634
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.417406204593525e-06,
+      "loss": 0.6963,
+      "step": 19635
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.416451139310185e-06,
+      "loss": 0.7158,
+      "step": 19636
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.41549614802195e-06,
+      "loss": 0.7344,
+      "step": 19637
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.414541230741478e-06,
+      "loss": 0.708,
+      "step": 19638
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4135863874814224e-06,
+      "loss": 0.6992,
+      "step": 19639
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.412631618254436e-06,
+      "loss": 0.6973,
+      "step": 19640
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.411676923073169e-06,
+      "loss": 0.6924,
+      "step": 19641
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.410722301950282e-06,
+      "loss": 0.7119,
+      "step": 19642
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.409767754898424e-06,
+      "loss": 0.7051,
+      "step": 19643
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.408813281930239e-06,
+      "loss": 0.708,
+      "step": 19644
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407858883058375e-06,
+      "loss": 0.6963,
+      "step": 19645
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.406904558295487e-06,
+      "loss": 0.7227,
+      "step": 19646
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.405950307654221e-06,
+      "loss": 0.7207,
+      "step": 19647
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.40499613114722e-06,
+      "loss": 0.7012,
+      "step": 19648
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4040420287871275e-06,
+      "loss": 0.6904,
+      "step": 19649
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.403088000586597e-06,
+      "loss": 0.7188,
+      "step": 19650
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.402134046558261e-06,
+      "loss": 0.7266,
+      "step": 19651
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.401180166714767e-06,
+      "loss": 0.7178,
+      "step": 19652
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.400226361068751e-06,
+      "loss": 0.7295,
+      "step": 19653
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.39927262963286e-06,
+      "loss": 0.7158,
+      "step": 19654
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.398318972419731e-06,
+      "loss": 0.6992,
+      "step": 19655
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.397365389442003e-06,
+      "loss": 0.7354,
+      "step": 19656
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3964118807123054e-06,
+      "loss": 0.7266,
+      "step": 19657
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.395458446243285e-06,
+      "loss": 0.7002,
+      "step": 19658
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.394505086047573e-06,
+      "loss": 0.6924,
+      "step": 19659
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393551800137801e-06,
+      "loss": 0.6953,
+      "step": 19660
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.392598588526602e-06,
+      "loss": 0.7178,
+      "step": 19661
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.391645451226618e-06,
+      "loss": 0.7119,
+      "step": 19662
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.39069238825047e-06,
+      "loss": 0.7344,
+      "step": 19663
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.38973939961079e-06,
+      "loss": 0.7305,
+      "step": 19664
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3887864853202075e-06,
+      "loss": 0.7061,
+      "step": 19665
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.387833645391354e-06,
+      "loss": 0.7041,
+      "step": 19666
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.386880879836856e-06,
+      "loss": 0.6807,
+      "step": 19667
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.385928188669338e-06,
+      "loss": 0.7168,
+      "step": 19668
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.384975571901427e-06,
+      "loss": 0.7236,
+      "step": 19669
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.384023029545746e-06,
+      "loss": 0.7324,
+      "step": 19670
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3830705616149195e-06,
+      "loss": 0.7021,
+      "step": 19671
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.382118168121569e-06,
+      "loss": 0.6963,
+      "step": 19672
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.381165849078317e-06,
+      "loss": 0.7129,
+      "step": 19673
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.380213604497779e-06,
+      "loss": 0.7168,
+      "step": 19674
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3792614343925864e-06,
+      "loss": 0.6875,
+      "step": 19675
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3783093387753476e-06,
+      "loss": 0.7441,
+      "step": 19676
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.377357317658678e-06,
+      "loss": 0.7021,
+      "step": 19677
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.376405371055202e-06,
+      "loss": 0.7119,
+      "step": 19678
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.375453498977533e-06,
+      "loss": 0.6875,
+      "step": 19679
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.374501701438284e-06,
+      "loss": 0.7285,
+      "step": 19680
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373549978450069e-06,
+      "loss": 0.7002,
+      "step": 19681
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.372598330025501e-06,
+      "loss": 0.7119,
+      "step": 19682
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.37164675617719e-06,
+      "loss": 0.7363,
+      "step": 19683
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3706952569177484e-06,
+      "loss": 0.7207,
+      "step": 19684
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3697438322597855e-06,
+      "loss": 0.7197,
+      "step": 19685
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3687924822159045e-06,
+      "loss": 0.7305,
+      "step": 19686
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.367841206798723e-06,
+      "loss": 0.708,
+      "step": 19687
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3668900060208455e-06,
+      "loss": 0.7295,
+      "step": 19688
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.365938879894872e-06,
+      "loss": 0.7236,
+      "step": 19689
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.364987828433406e-06,
+      "loss": 0.707,
+      "step": 19690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.364036851649057e-06,
+      "loss": 0.7051,
+      "step": 19691
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3630859495544275e-06,
+      "loss": 0.7031,
+      "step": 19692
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.362135122162118e-06,
+      "loss": 0.708,
+      "step": 19693
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.361184369484723e-06,
+      "loss": 0.7041,
+      "step": 19694
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.360233691534856e-06,
+      "loss": 0.707,
+      "step": 19695
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.359283088325104e-06,
+      "loss": 0.6992,
+      "step": 19696
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.358332559868069e-06,
+      "loss": 0.708,
+      "step": 19697
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.357382106176343e-06,
+      "loss": 0.7139,
+      "step": 19698
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.356431727262529e-06,
+      "loss": 0.7207,
+      "step": 19699
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3554814231392184e-06,
+      "loss": 0.7412,
+      "step": 19700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.354531193819009e-06,
+      "loss": 0.708,
+      "step": 19701
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.353581039314483e-06,
+      "loss": 0.7012,
+      "step": 19702
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.352630959638241e-06,
+      "loss": 0.7012,
+      "step": 19703
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.351680954802872e-06,
+      "loss": 0.6943,
+      "step": 19704
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.350731024820966e-06,
+      "loss": 0.709,
+      "step": 19705
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349781169705111e-06,
+      "loss": 0.709,
+      "step": 19706
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.34883138946789e-06,
+      "loss": 0.6904,
+      "step": 19707
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.347881684121903e-06,
+      "loss": 0.7188,
+      "step": 19708
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.346932053679724e-06,
+      "loss": 0.71,
+      "step": 19709
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3459824981539375e-06,
+      "loss": 0.7158,
+      "step": 19710
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.345033017557136e-06,
+      "loss": 0.7168,
+      "step": 19711
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.344083611901896e-06,
+      "loss": 0.7285,
+      "step": 19712
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.343134281200801e-06,
+      "loss": 0.707,
+      "step": 19713
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.342185025466433e-06,
+      "loss": 0.7002,
+      "step": 19714
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.34123584471137e-06,
+      "loss": 0.7002,
+      "step": 19715
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.340286738948192e-06,
+      "loss": 0.708,
+      "step": 19716
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.339337708189476e-06,
+      "loss": 0.6992,
+      "step": 19717
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.338388752447799e-06,
+      "loss": 0.708,
+      "step": 19718
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.337439871735733e-06,
+      "loss": 0.6943,
+      "step": 19719
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.336491066065862e-06,
+      "loss": 0.7148,
+      "step": 19720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.335542335450759e-06,
+      "loss": 0.7148,
+      "step": 19721
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.334593679902984e-06,
+      "loss": 0.7168,
+      "step": 19722
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.333645099435122e-06,
+      "loss": 0.7148,
+      "step": 19723
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.33269659405974e-06,
+      "loss": 0.6973,
+      "step": 19724
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3317481637894075e-06,
+      "loss": 0.7002,
+      "step": 19725
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.330799808636693e-06,
+      "loss": 0.708,
+      "step": 19726
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.329851528614166e-06,
+      "loss": 0.709,
+      "step": 19727
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.328903323734392e-06,
+      "loss": 0.71,
+      "step": 19728
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.327955194009937e-06,
+      "loss": 0.6943,
+      "step": 19729
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.327007139453367e-06,
+      "loss": 0.7119,
+      "step": 19730
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.326059160077241e-06,
+      "loss": 0.7158,
+      "step": 19731
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.325111255894129e-06,
+      "loss": 0.6963,
+      "step": 19732
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3241634269165914e-06,
+      "loss": 0.708,
+      "step": 19733
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.323215673157191e-06,
+      "loss": 0.7021,
+      "step": 19734
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.322267994628477e-06,
+      "loss": 0.7246,
+      "step": 19735
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.32132039134302e-06,
+      "loss": 0.7139,
+      "step": 19736
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.320372863313373e-06,
+      "loss": 0.6992,
+      "step": 19737
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.319425410552096e-06,
+      "loss": 0.708,
+      "step": 19738
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.318478033071738e-06,
+      "loss": 0.7354,
+      "step": 19739
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.317530730884866e-06,
+      "loss": 0.7168,
+      "step": 19740
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.316583504004024e-06,
+      "loss": 0.7295,
+      "step": 19741
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.315636352441766e-06,
+      "loss": 0.7012,
+      "step": 19742
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.314689276210644e-06,
+      "loss": 0.7207,
+      "step": 19743
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.313742275323214e-06,
+      "loss": 0.6885,
+      "step": 19744
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.312795349792022e-06,
+      "loss": 0.7168,
+      "step": 19745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.311848499629618e-06,
+      "loss": 0.7051,
+      "step": 19746
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.31090172484855e-06,
+      "loss": 0.708,
+      "step": 19747
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309955025461365e-06,
+      "loss": 0.7031,
+      "step": 19748
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309008401480609e-06,
+      "loss": 0.7285,
+      "step": 19749
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.308061852918825e-06,
+      "loss": 0.7041,
+      "step": 19750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.30711537978856e-06,
+      "loss": 0.7119,
+      "step": 19751
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.306168982102351e-06,
+      "loss": 0.7236,
+      "step": 19752
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305222659872753e-06,
+      "loss": 0.708,
+      "step": 19753
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.304276413112295e-06,
+      "loss": 0.7305,
+      "step": 19754
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.303330241833517e-06,
+      "loss": 0.7002,
+      "step": 19755
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.302384146048963e-06,
+      "loss": 0.7461,
+      "step": 19756
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.301438125771172e-06,
+      "loss": 0.709,
+      "step": 19757
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.300492181012678e-06,
+      "loss": 0.7236,
+      "step": 19758
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.299546311786017e-06,
+      "loss": 0.7061,
+      "step": 19759
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.298600518103725e-06,
+      "loss": 0.6904,
+      "step": 19760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.297654799978336e-06,
+      "loss": 0.7217,
+      "step": 19761
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.296709157422381e-06,
+      "loss": 0.7012,
+      "step": 19762
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.295763590448395e-06,
+      "loss": 0.6924,
+      "step": 19763
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.294818099068902e-06,
+      "loss": 0.7148,
+      "step": 19764
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.293872683296443e-06,
+      "loss": 0.7314,
+      "step": 19765
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.292927343143545e-06,
+      "loss": 0.7188,
+      "step": 19766
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.291982078622724e-06,
+      "loss": 0.6924,
+      "step": 19767
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.291036889746519e-06,
+      "loss": 0.7246,
+      "step": 19768
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.290091776527453e-06,
+      "loss": 0.709,
+      "step": 19769
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2891467389780495e-06,
+      "loss": 0.71,
+      "step": 19770
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.288201777110834e-06,
+      "loss": 0.7178,
+      "step": 19771
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.287256890938325e-06,
+      "loss": 0.7207,
+      "step": 19772
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.286312080473055e-06,
+      "loss": 0.6914,
+      "step": 19773
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2853673457275345e-06,
+      "loss": 0.6934,
+      "step": 19774
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.284422686714287e-06,
+      "loss": 0.7246,
+      "step": 19775
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.283478103445829e-06,
+      "loss": 0.7246,
+      "step": 19776
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.282533595934682e-06,
+      "loss": 0.7295,
+      "step": 19777
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.281589164193364e-06,
+      "loss": 0.7246,
+      "step": 19778
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.280644808234392e-06,
+      "loss": 0.7236,
+      "step": 19779
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.27970052807027e-06,
+      "loss": 0.7139,
+      "step": 19780
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.278756323713523e-06,
+      "loss": 0.709,
+      "step": 19781
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.277812195176662e-06,
+      "loss": 0.7031,
+      "step": 19782
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.276868142472197e-06,
+      "loss": 0.6895,
+      "step": 19783
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.275924165612635e-06,
+      "loss": 0.7041,
+      "step": 19784
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.274980264610494e-06,
+      "loss": 0.7148,
+      "step": 19785
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.274036439478284e-06,
+      "loss": 0.7236,
+      "step": 19786
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.273092690228504e-06,
+      "loss": 0.7266,
+      "step": 19787
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.272149016873661e-06,
+      "loss": 0.7139,
+      "step": 19788
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.271205419426269e-06,
+      "loss": 0.7041,
+      "step": 19789
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.270261897898828e-06,
+      "loss": 0.6963,
+      "step": 19790
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.269318452303843e-06,
+      "loss": 0.7197,
+      "step": 19791
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.268375082653816e-06,
+      "loss": 0.7305,
+      "step": 19792
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.267431788961251e-06,
+      "loss": 0.7217,
+      "step": 19793
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.266488571238645e-06,
+      "loss": 0.7051,
+      "step": 19794
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.265545429498501e-06,
+      "loss": 0.7324,
+      "step": 19795
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.264602363753317e-06,
+      "loss": 0.7197,
+      "step": 19796
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.263659374015588e-06,
+      "loss": 0.7207,
+      "step": 19797
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.262716460297816e-06,
+      "loss": 0.6953,
+      "step": 19798
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.261773622612499e-06,
+      "loss": 0.7012,
+      "step": 19799
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.260830860972118e-06,
+      "loss": 0.6914,
+      "step": 19800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.259888175389181e-06,
+      "loss": 0.7471,
+      "step": 19801
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.258945565876176e-06,
+      "loss": 0.709,
+      "step": 19802
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.258003032445593e-06,
+      "loss": 0.7021,
+      "step": 19803
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2570605751099255e-06,
+      "loss": 0.6904,
+      "step": 19804
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.256118193881662e-06,
+      "loss": 0.7197,
+      "step": 19805
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.255175888773292e-06,
+      "loss": 0.7236,
+      "step": 19806
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.254233659797301e-06,
+      "loss": 0.7158,
+      "step": 19807
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253291506966178e-06,
+      "loss": 0.7275,
+      "step": 19808
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.252349430292405e-06,
+      "loss": 0.6992,
+      "step": 19809
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2514074297884736e-06,
+      "loss": 0.7471,
+      "step": 19810
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.250465505466863e-06,
+      "loss": 0.7178,
+      "step": 19811
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.24952365734006e-06,
+      "loss": 0.7285,
+      "step": 19812
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.248581885420536e-06,
+      "loss": 0.7119,
+      "step": 19813
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.247640189720783e-06,
+      "loss": 0.6836,
+      "step": 19814
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.246698570253276e-06,
+      "loss": 0.6855,
+      "step": 19815
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2457570270304925e-06,
+      "loss": 0.707,
+      "step": 19816
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.24481556006491e-06,
+      "loss": 0.6934,
+      "step": 19817
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.243874169369012e-06,
+      "loss": 0.708,
+      "step": 19818
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.242932854955266e-06,
+      "loss": 0.6914,
+      "step": 19819
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2419916168361506e-06,
+      "loss": 0.7236,
+      "step": 19820
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.241050455024134e-06,
+      "loss": 0.7109,
+      "step": 19821
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.240109369531696e-06,
+      "loss": 0.6963,
+      "step": 19822
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.239168360371305e-06,
+      "loss": 0.7129,
+      "step": 19823
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.238227427555433e-06,
+      "loss": 0.7051,
+      "step": 19824
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.237286571096548e-06,
+      "loss": 0.7119,
+      "step": 19825
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2363457910071195e-06,
+      "loss": 0.709,
+      "step": 19826
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.235405087299613e-06,
+      "loss": 0.7227,
+      "step": 19827
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.234464459986498e-06,
+      "loss": 0.7568,
+      "step": 19828
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.233523909080235e-06,
+      "loss": 0.71,
+      "step": 19829
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.232583434593295e-06,
+      "loss": 0.7188,
+      "step": 19830
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.231643036538142e-06,
+      "loss": 0.7012,
+      "step": 19831
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.230702714927232e-06,
+      "loss": 0.7061,
+      "step": 19832
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.229762469773026e-06,
+      "loss": 0.7168,
+      "step": 19833
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.228822301087991e-06,
+      "loss": 0.7061,
+      "step": 19834
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.227882208884584e-06,
+      "loss": 0.7158,
+      "step": 19835
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2269421931752615e-06,
+      "loss": 0.7295,
+      "step": 19836
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.226002253972483e-06,
+      "loss": 0.71,
+      "step": 19837
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2250623912887045e-06,
+      "loss": 0.7021,
+      "step": 19838
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.224122605136381e-06,
+      "loss": 0.7021,
+      "step": 19839
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.223182895527967e-06,
+      "loss": 0.6943,
+      "step": 19840
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.222243262475914e-06,
+      "loss": 0.6973,
+      "step": 19841
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.221303705992675e-06,
+      "loss": 0.7432,
+      "step": 19842
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.220364226090705e-06,
+      "loss": 0.7227,
+      "step": 19843
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.219424822782455e-06,
+      "loss": 0.7041,
+      "step": 19844
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.218485496080363e-06,
+      "loss": 0.7109,
+      "step": 19845
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.217546245996891e-06,
+      "loss": 0.7012,
+      "step": 19846
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2166070725444785e-06,
+      "loss": 0.7178,
+      "step": 19847
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.215667975735574e-06,
+      "loss": 0.7148,
+      "step": 19848
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.214728955582625e-06,
+      "loss": 0.7383,
+      "step": 19849
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.213790012098067e-06,
+      "loss": 0.707,
+      "step": 19850
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.212851145294357e-06,
+      "loss": 0.7344,
+      "step": 19851
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.211912355183926e-06,
+      "loss": 0.7314,
+      "step": 19852
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.210973641779219e-06,
+      "loss": 0.7031,
+      "step": 19853
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.210035005092673e-06,
+      "loss": 0.7188,
+      "step": 19854
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.209096445136733e-06,
+      "loss": 0.7168,
+      "step": 19855
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.208157961923834e-06,
+      "loss": 0.7139,
+      "step": 19856
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.207219555466418e-06,
+      "loss": 0.7188,
+      "step": 19857
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.206281225776908e-06,
+      "loss": 0.7207,
+      "step": 19858
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.205342972867752e-06,
+      "loss": 0.71,
+      "step": 19859
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.204404796751379e-06,
+      "loss": 0.7266,
+      "step": 19860
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.203466697440223e-06,
+      "loss": 0.6826,
+      "step": 19861
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2025286749467106e-06,
+      "loss": 0.7178,
+      "step": 19862
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.201590729283283e-06,
+      "loss": 0.7236,
+      "step": 19863
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.200652860462367e-06,
+      "loss": 0.7275,
+      "step": 19864
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.199715068496387e-06,
+      "loss": 0.7266,
+      "step": 19865
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.198777353397769e-06,
+      "loss": 0.6836,
+      "step": 19866
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.197839715178949e-06,
+      "loss": 0.708,
+      "step": 19867
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.196902153852348e-06,
+      "loss": 0.707,
+      "step": 19868
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.19596466943039e-06,
+      "loss": 0.7227,
+      "step": 19869
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.195027261925501e-06,
+      "loss": 0.7197,
+      "step": 19870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.194089931350103e-06,
+      "loss": 0.6963,
+      "step": 19871
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.193152677716618e-06,
+      "loss": 0.709,
+      "step": 19872
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.192215501037466e-06,
+      "loss": 0.7207,
+      "step": 19873
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.191278401325063e-06,
+      "loss": 0.7402,
+      "step": 19874
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.190341378591836e-06,
+      "loss": 0.6924,
+      "step": 19875
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.189404432850198e-06,
+      "loss": 0.7188,
+      "step": 19876
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.188467564112572e-06,
+      "loss": 0.71,
+      "step": 19877
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.187530772391359e-06,
+      "loss": 0.7139,
+      "step": 19878
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.186594057698988e-06,
+      "loss": 0.708,
+      "step": 19879
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.185657420047867e-06,
+      "loss": 0.6973,
+      "step": 19880
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.184720859450409e-06,
+      "loss": 0.7061,
+      "step": 19881
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.183784375919027e-06,
+      "loss": 0.7041,
+      "step": 19882
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.182847969466128e-06,
+      "loss": 0.71,
+      "step": 19883
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.181911640104126e-06,
+      "loss": 0.7217,
+      "step": 19884
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.180975387845427e-06,
+      "loss": 0.7471,
+      "step": 19885
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1800392127024395e-06,
+      "loss": 0.7041,
+      "step": 19886
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.179103114687565e-06,
+      "loss": 0.7168,
+      "step": 19887
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.178167093813218e-06,
+      "loss": 0.7256,
+      "step": 19888
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.177231150091802e-06,
+      "loss": 0.6943,
+      "step": 19889
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.17629528353571e-06,
+      "loss": 0.7021,
+      "step": 19890
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.175359494157355e-06,
+      "loss": 0.7285,
+      "step": 19891
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.174423781969134e-06,
+      "loss": 0.7383,
+      "step": 19892
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.17348814698345e-06,
+      "loss": 0.7051,
+      "step": 19893
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1725525892127e-06,
+      "loss": 0.7275,
+      "step": 19894
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.171617108669278e-06,
+      "loss": 0.7158,
+      "step": 19895
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.170681705365594e-06,
+      "loss": 0.7363,
+      "step": 19896
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.169746379314034e-06,
+      "loss": 0.7178,
+      "step": 19897
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.168811130526995e-06,
+      "loss": 0.7227,
+      "step": 19898
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.167875959016868e-06,
+      "loss": 0.7266,
+      "step": 19899
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.166940864796054e-06,
+      "loss": 0.71,
+      "step": 19900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.166005847876942e-06,
+      "loss": 0.6885,
+      "step": 19901
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.165070908271926e-06,
+      "loss": 0.7129,
+      "step": 19902
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.164136045993386e-06,
+      "loss": 0.708,
+      "step": 19903
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.16320126105372e-06,
+      "loss": 0.6992,
+      "step": 19904
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.162266553465315e-06,
+      "loss": 0.7148,
+      "step": 19905
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.161331923240557e-06,
+      "loss": 0.7207,
+      "step": 19906
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.160397370391828e-06,
+      "loss": 0.7051,
+      "step": 19907
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.159462894931521e-06,
+      "loss": 0.7207,
+      "step": 19908
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.158528496872019e-06,
+      "loss": 0.7129,
+      "step": 19909
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1575941762256986e-06,
+      "loss": 0.7188,
+      "step": 19910
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.156659933004941e-06,
+      "loss": 0.6973,
+      "step": 19911
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.155725767222136e-06,
+      "loss": 0.7227,
+      "step": 19912
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.154791678889659e-06,
+      "loss": 0.7168,
+      "step": 19913
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.153857668019889e-06,
+      "loss": 0.7002,
+      "step": 19914
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.152923734625201e-06,
+      "loss": 0.708,
+      "step": 19915
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.151989878717976e-06,
+      "loss": 0.7324,
+      "step": 19916
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.151056100310588e-06,
+      "loss": 0.6973,
+      "step": 19917
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.15012239941541e-06,
+      "loss": 0.7148,
+      "step": 19918
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.14918877604482e-06,
+      "loss": 0.707,
+      "step": 19919
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.148255230211183e-06,
+      "loss": 0.7188,
+      "step": 19920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.14732176192688e-06,
+      "loss": 0.7285,
+      "step": 19921
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.146388371204281e-06,
+      "loss": 0.7236,
+      "step": 19922
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.145455058055745e-06,
+      "loss": 0.7041,
+      "step": 19923
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.144521822493651e-06,
+      "loss": 0.75,
+      "step": 19924
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.143588664530364e-06,
+      "loss": 0.748,
+      "step": 19925
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.142655584178249e-06,
+      "loss": 0.7021,
+      "step": 19926
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.141722581449673e-06,
+      "loss": 0.709,
+      "step": 19927
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.140789656356999e-06,
+      "loss": 0.6865,
+      "step": 19928
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.139856808912592e-06,
+      "loss": 0.7227,
+      "step": 19929
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.138924039128813e-06,
+      "loss": 0.709,
+      "step": 19930
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.137991347018024e-06,
+      "loss": 0.707,
+      "step": 19931
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.137058732592581e-06,
+      "loss": 0.7109,
+      "step": 19932
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.136126195864853e-06,
+      "loss": 0.7217,
+      "step": 19933
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.135193736847191e-06,
+      "loss": 0.71,
+      "step": 19934
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.134261355551956e-06,
+      "loss": 0.7158,
+      "step": 19935
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.133329051991502e-06,
+      "loss": 0.71,
+      "step": 19936
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.132396826178186e-06,
+      "loss": 0.7217,
+      "step": 19937
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131464678124358e-06,
+      "loss": 0.7236,
+      "step": 19938
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.130532607842376e-06,
+      "loss": 0.7324,
+      "step": 19939
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.129600615344587e-06,
+      "loss": 0.7363,
+      "step": 19940
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1286687006433515e-06,
+      "loss": 0.7148,
+      "step": 19941
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.127736863751008e-06,
+      "loss": 0.707,
+      "step": 19942
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.126805104679913e-06,
+      "loss": 0.6807,
+      "step": 19943
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.125873423442408e-06,
+      "loss": 0.7451,
+      "step": 19944
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.124941820050849e-06,
+      "loss": 0.709,
+      "step": 19945
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.124010294517575e-06,
+      "loss": 0.7148,
+      "step": 19946
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.123078846854934e-06,
+      "loss": 0.7217,
+      "step": 19947
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.7354,
+      "step": 19948
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.121216185190924e-06,
+      "loss": 0.6924,
+      "step": 19949
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1202849712142375e-06,
+      "loss": 0.7051,
+      "step": 19950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1193538351575535e-06,
+      "loss": 0.709,
+      "step": 19951
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.118422777033207e-06,
+      "loss": 0.7139,
+      "step": 19952
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1174917968535435e-06,
+      "loss": 0.7051,
+      "step": 19953
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.116560894630901e-06,
+      "loss": 0.7373,
+      "step": 19954
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.11563007037761e-06,
+      "loss": 0.7188,
+      "step": 19955
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.114699324106003e-06,
+      "loss": 0.7266,
+      "step": 19956
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.113768655828425e-06,
+      "loss": 0.707,
+      "step": 19957
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1128380655572055e-06,
+      "loss": 0.7236,
+      "step": 19958
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1119075533046745e-06,
+      "loss": 0.6846,
+      "step": 19959
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1109771190831674e-06,
+      "loss": 0.7324,
+      "step": 19960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.110046762905011e-06,
+      "loss": 0.708,
+      "step": 19961
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.109116484782537e-06,
+      "loss": 0.7344,
+      "step": 19962
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.108186284728072e-06,
+      "loss": 0.7139,
+      "step": 19963
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.107256162753945e-06,
+      "loss": 0.708,
+      "step": 19964
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.106326118872478e-06,
+      "loss": 0.7314,
+      "step": 19965
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.105396153096003e-06,
+      "loss": 0.7119,
+      "step": 19966
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.104466265436845e-06,
+      "loss": 0.71,
+      "step": 19967
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.103536455907316e-06,
+      "loss": 0.7021,
+      "step": 19968
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1026067245197485e-06,
+      "loss": 0.7236,
+      "step": 19969
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.101677071286461e-06,
+      "loss": 0.7158,
+      "step": 19970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.100747496219773e-06,
+      "loss": 0.7275,
+      "step": 19971
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.099817999332003e-06,
+      "loss": 0.7305,
+      "step": 19972
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.098888580635467e-06,
+      "loss": 0.6934,
+      "step": 19973
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.097959240142492e-06,
+      "loss": 0.7041,
+      "step": 19974
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.097029977865382e-06,
+      "loss": 0.7207,
+      "step": 19975
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.096100793816458e-06,
+      "loss": 0.7334,
+      "step": 19976
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.095171688008027e-06,
+      "loss": 0.7119,
+      "step": 19977
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.094242660452411e-06,
+      "loss": 0.6992,
+      "step": 19978
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.093313711161919e-06,
+      "loss": 0.7158,
+      "step": 19979
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.092384840148859e-06,
+      "loss": 0.7148,
+      "step": 19980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.091456047425545e-06,
+      "loss": 0.7305,
+      "step": 19981
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.090527333004282e-06,
+      "loss": 0.6924,
+      "step": 19982
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.089598696897378e-06,
+      "loss": 0.7031,
+      "step": 19983
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.088670139117141e-06,
+      "loss": 0.6914,
+      "step": 19984
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0877416596758724e-06,
+      "loss": 0.7041,
+      "step": 19985
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.086813258585883e-06,
+      "loss": 0.7002,
+      "step": 19986
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.085884935859479e-06,
+      "loss": 0.6953,
+      "step": 19987
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.084956691508952e-06,
+      "loss": 0.7178,
+      "step": 19988
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.084028525546606e-06,
+      "loss": 0.71,
+      "step": 19989
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0831004379847475e-06,
+      "loss": 0.7197,
+      "step": 19990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0821724288356725e-06,
+      "loss": 0.7246,
+      "step": 19991
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08124449811168e-06,
+      "loss": 0.7324,
+      "step": 19992
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.080316645825065e-06,
+      "loss": 0.6973,
+      "step": 19993
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.079388871988127e-06,
+      "loss": 0.7295,
+      "step": 19994
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0784611766131585e-06,
+      "loss": 0.6953,
+      "step": 19995
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.077533559712455e-06,
+      "loss": 0.7168,
+      "step": 19996
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.076606021298306e-06,
+      "loss": 0.7266,
+      "step": 19997
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.07567856138301e-06,
+      "loss": 0.71,
+      "step": 19998
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0747511799788545e-06,
+      "loss": 0.7168,
+      "step": 19999
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.073823877098133e-06,
+      "loss": 0.7139,
+      "step": 20000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0728966527531255e-06,
+      "loss": 0.7109,
+      "step": 20001
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0719695069561285e-06,
+      "loss": 0.7021,
+      "step": 20002
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.071042439719425e-06,
+      "loss": 0.7031,
+      "step": 20003
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.070115451055302e-06,
+      "loss": 0.6953,
+      "step": 20004
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.069188540976045e-06,
+      "loss": 0.6973,
+      "step": 20005
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.068261709493936e-06,
+      "loss": 0.7061,
+      "step": 20006
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.067334956621258e-06,
+      "loss": 0.7168,
+      "step": 20007
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.066408282370295e-06,
+      "loss": 0.6973,
+      "step": 20008
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.065481686753324e-06,
+      "loss": 0.6934,
+      "step": 20009
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.064555169782623e-06,
+      "loss": 0.7402,
+      "step": 20010
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.063628731470477e-06,
+      "loss": 0.7432,
+      "step": 20011
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.062702371829162e-06,
+      "loss": 0.7139,
+      "step": 20012
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.06177609087095e-06,
+      "loss": 0.7109,
+      "step": 20013
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.060849888608121e-06,
+      "loss": 0.7217,
+      "step": 20014
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.059923765052947e-06,
+      "loss": 0.7109,
+      "step": 20015
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0589977202177025e-06,
+      "loss": 0.7197,
+      "step": 20016
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.058071754114658e-06,
+      "loss": 0.7041,
+      "step": 20017
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.057145866756082e-06,
+      "loss": 0.7314,
+      "step": 20018
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.056220058154256e-06,
+      "loss": 0.6953,
+      "step": 20019
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.055294328321438e-06,
+      "loss": 0.6963,
+      "step": 20020
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0543686772699e-06,
+      "loss": 0.6992,
+      "step": 20021
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.053443105011904e-06,
+      "loss": 0.707,
+      "step": 20022
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0525176115597255e-06,
+      "loss": 0.707,
+      "step": 20023
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.051592196925624e-06,
+      "loss": 0.6973,
+      "step": 20024
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0506668611218635e-06,
+      "loss": 0.7207,
+      "step": 20025
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.049741604160708e-06,
+      "loss": 0.7051,
+      "step": 20026
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.048816426054419e-06,
+      "loss": 0.7207,
+      "step": 20027
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.047891326815256e-06,
+      "loss": 0.707,
+      "step": 20028
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.046966306455481e-06,
+      "loss": 0.7188,
+      "step": 20029
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.046041364987348e-06,
+      "loss": 0.748,
+      "step": 20030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.04511650242312e-06,
+      "loss": 0.7168,
+      "step": 20031
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.044191718775057e-06,
+      "loss": 0.7275,
+      "step": 20032
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.043267014055404e-06,
+      "loss": 0.7051,
+      "step": 20033
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.042342388276418e-06,
+      "loss": 0.7197,
+      "step": 20034
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.041417841450359e-06,
+      "loss": 0.7031,
+      "step": 20035
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.040493373589476e-06,
+      "loss": 0.7041,
+      "step": 20036
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.039568984706019e-06,
+      "loss": 0.709,
+      "step": 20037
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.03864467481224e-06,
+      "loss": 0.7148,
+      "step": 20038
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.037720443920387e-06,
+      "loss": 0.6885,
+      "step": 20039
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.036796292042709e-06,
+      "loss": 0.7295,
+      "step": 20040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.035872219191452e-06,
+      "loss": 0.7373,
+      "step": 20041
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.034948225378861e-06,
+      "loss": 0.7197,
+      "step": 20042
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0340243106171865e-06,
+      "loss": 0.6924,
+      "step": 20043
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0331004749186684e-06,
+      "loss": 0.6855,
+      "step": 20044
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.032176718295555e-06,
+      "loss": 0.6924,
+      "step": 20045
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0312530407600754e-06,
+      "loss": 0.7217,
+      "step": 20046
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.030329442324483e-06,
+      "loss": 0.7275,
+      "step": 20047
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.029405923001014e-06,
+      "loss": 0.709,
+      "step": 20048
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.028482482801907e-06,
+      "loss": 0.7139,
+      "step": 20049
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.027559121739398e-06,
+      "loss": 0.7227,
+      "step": 20050
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.026635839825722e-06,
+      "loss": 0.7012,
+      "step": 20051
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0257126370731255e-06,
+      "loss": 0.7207,
+      "step": 20052
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.024789513493831e-06,
+      "loss": 0.709,
+      "step": 20053
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.023866469100077e-06,
+      "loss": 0.7275,
+      "step": 20054
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0229435039040925e-06,
+      "loss": 0.7314,
+      "step": 20055
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.022020617918115e-06,
+      "loss": 0.7207,
+      "step": 20056
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.021097811154369e-06,
+      "loss": 0.7236,
+      "step": 20057
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0201750836250896e-06,
+      "loss": 0.7275,
+      "step": 20058
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.019252435342502e-06,
+      "loss": 0.7129,
+      "step": 20059
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.018329866318833e-06,
+      "loss": 0.7119,
+      "step": 20060
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.017407376566309e-06,
+      "loss": 0.7246,
+      "step": 20061
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.016484966097156e-06,
+      "loss": 0.7012,
+      "step": 20062
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0155626349235945e-06,
+      "loss": 0.7295,
+      "step": 20063
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0146403830578525e-06,
+      "loss": 0.7324,
+      "step": 20064
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.013718210512155e-06,
+      "loss": 0.7178,
+      "step": 20065
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.012796117298715e-06,
+      "loss": 0.6992,
+      "step": 20066
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.011874103429752e-06,
+      "loss": 0.7031,
+      "step": 20067
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.010952168917491e-06,
+      "loss": 0.6895,
+      "step": 20068
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.010030313774147e-06,
+      "loss": 0.7168,
+      "step": 20069
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.009108538011938e-06,
+      "loss": 0.7275,
+      "step": 20070
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.0081868416430785e-06,
+      "loss": 0.7451,
+      "step": 20071
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.007265224679782e-06,
+      "loss": 0.6963,
+      "step": 20072
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.006343687134263e-06,
+      "loss": 0.7129,
+      "step": 20073
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005422229018736e-06,
+      "loss": 0.7129,
+      "step": 20074
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.004500850345406e-06,
+      "loss": 0.7314,
+      "step": 20075
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.003579551126493e-06,
+      "loss": 0.7031,
+      "step": 20076
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.002658331374201e-06,
+      "loss": 0.7139,
+      "step": 20077
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.001737191100742e-06,
+      "loss": 0.7334,
+      "step": 20078
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.000816130318314e-06,
+      "loss": 0.7246,
+      "step": 20079
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.999895149039132e-06,
+      "loss": 0.6934,
+      "step": 20080
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.998974247275399e-06,
+      "loss": 0.6934,
+      "step": 20081
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.998053425039319e-06,
+      "loss": 0.6758,
+      "step": 20082
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.997132682343093e-06,
+      "loss": 0.7373,
+      "step": 20083
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.996212019198927e-06,
+      "loss": 0.7236,
+      "step": 20084
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9952914356190185e-06,
+      "loss": 0.7471,
+      "step": 20085
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9943709316155684e-06,
+      "loss": 0.7178,
+      "step": 20086
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.993450507200773e-06,
+      "loss": 0.6973,
+      "step": 20087
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.992530162386836e-06,
+      "loss": 0.7051,
+      "step": 20088
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.991609897185951e-06,
+      "loss": 0.7188,
+      "step": 20089
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.990689711610313e-06,
+      "loss": 0.7139,
+      "step": 20090
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.989769605672118e-06,
+      "loss": 0.7148,
+      "step": 20091
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.988849579383558e-06,
+      "loss": 0.7314,
+      "step": 20092
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.987929632756826e-06,
+      "loss": 0.7148,
+      "step": 20093
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.987009765804115e-06,
+      "loss": 0.7031,
+      "step": 20094
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.986089978537613e-06,
+      "loss": 0.7158,
+      "step": 20095
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9851702709695075e-06,
+      "loss": 0.7051,
+      "step": 20096
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.984250643111996e-06,
+      "loss": 0.7158,
+      "step": 20097
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.983331094977257e-06,
+      "loss": 0.6846,
+      "step": 20098
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.982411626577478e-06,
+      "loss": 0.7031,
+      "step": 20099
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.981492237924842e-06,
+      "loss": 0.7363,
+      "step": 20100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98057292903154e-06,
+      "loss": 0.7197,
+      "step": 20101
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9796536999097516e-06,
+      "loss": 0.71,
+      "step": 20102
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.978734550571658e-06,
+      "loss": 0.7002,
+      "step": 20103
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.97781548102944e-06,
+      "loss": 0.708,
+      "step": 20104
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.976896491295278e-06,
+      "loss": 0.7109,
+      "step": 20105
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.975977581381352e-06,
+      "loss": 0.7256,
+      "step": 20106
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.975058751299837e-06,
+      "loss": 0.708,
+      "step": 20107
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.974140001062907e-06,
+      "loss": 0.6904,
+      "step": 20108
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.973221330682747e-06,
+      "loss": 0.71,
+      "step": 20109
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.972302740171529e-06,
+      "loss": 0.708,
+      "step": 20110
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.971384229541419e-06,
+      "loss": 0.7119,
+      "step": 20111
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.970465798804592e-06,
+      "loss": 0.6934,
+      "step": 20112
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9695474479732245e-06,
+      "loss": 0.7109,
+      "step": 20113
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.968629177059483e-06,
+      "loss": 0.6934,
+      "step": 20114
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.967710986075538e-06,
+      "loss": 0.7305,
+      "step": 20115
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.966792875033557e-06,
+      "loss": 0.6973,
+      "step": 20116
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.965874843945707e-06,
+      "loss": 0.7285,
+      "step": 20117
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9649568928241555e-06,
+      "loss": 0.707,
+      "step": 20118
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.964039021681065e-06,
+      "loss": 0.6973,
+      "step": 20119
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.963121230528598e-06,
+      "loss": 0.7041,
+      "step": 20120
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.962203519378923e-06,
+      "loss": 0.6982,
+      "step": 20121
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.961285888244201e-06,
+      "loss": 0.7148,
+      "step": 20122
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.960368337136592e-06,
+      "loss": 0.7158,
+      "step": 20123
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.959450866068248e-06,
+      "loss": 0.7471,
+      "step": 20124
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9585334750513385e-06,
+      "loss": 0.707,
+      "step": 20125
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.957616164098015e-06,
+      "loss": 0.6699,
+      "step": 20126
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.956698933220436e-06,
+      "loss": 0.7412,
+      "step": 20127
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.955781782430757e-06,
+      "loss": 0.7275,
+      "step": 20128
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.954864711741133e-06,
+      "loss": 0.7109,
+      "step": 20129
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9539477211637155e-06,
+      "loss": 0.7158,
+      "step": 20130
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.953030810710659e-06,
+      "loss": 0.7012,
+      "step": 20131
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.952113980394111e-06,
+      "loss": 0.7197,
+      "step": 20132
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.951197230226223e-06,
+      "loss": 0.7236,
+      "step": 20133
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.950280560219146e-06,
+      "loss": 0.71,
+      "step": 20134
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.949363970385028e-06,
+      "loss": 0.7275,
+      "step": 20135
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.948447460736016e-06,
+      "loss": 0.7334,
+      "step": 20136
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9475310312842544e-06,
+      "loss": 0.7275,
+      "step": 20137
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9466146820418885e-06,
+      "loss": 0.7295,
+      "step": 20138
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9456984130210626e-06,
+      "loss": 0.7041,
+      "step": 20139
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.944782224233918e-06,
+      "loss": 0.6924,
+      "step": 20140
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.943866115692595e-06,
+      "loss": 0.7539,
+      "step": 20141
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.942950087409244e-06,
+      "loss": 0.7334,
+      "step": 20142
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.942034139395993e-06,
+      "loss": 0.7334,
+      "step": 20143
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.941118271664983e-06,
+      "loss": 0.7275,
+      "step": 20144
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.940202484228352e-06,
+      "loss": 0.6943,
+      "step": 20145
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.939286777098239e-06,
+      "loss": 0.707,
+      "step": 20146
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.938371150286777e-06,
+      "loss": 0.7061,
+      "step": 20147
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.937455603806102e-06,
+      "loss": 0.7256,
+      "step": 20148
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.936540137668345e-06,
+      "loss": 0.7002,
+      "step": 20149
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9356247518856405e-06,
+      "loss": 0.7139,
+      "step": 20150
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.934709446470116e-06,
+      "loss": 0.6846,
+      "step": 20151
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.933794221433905e-06,
+      "loss": 0.7246,
+      "step": 20152
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.93287907678913e-06,
+      "loss": 0.7178,
+      "step": 20153
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.931964012547927e-06,
+      "loss": 0.6943,
+      "step": 20154
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.931049028722425e-06,
+      "loss": 0.7217,
+      "step": 20155
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.930134125324738e-06,
+      "loss": 0.6904,
+      "step": 20156
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9292193023669935e-06,
+      "loss": 0.7324,
+      "step": 20157
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.928304559861323e-06,
+      "loss": 0.7383,
+      "step": 20158
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.927389897819842e-06,
+      "loss": 0.7119,
+      "step": 20159
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.926475316254675e-06,
+      "loss": 0.7354,
+      "step": 20160
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.925560815177942e-06,
+      "loss": 0.7178,
+      "step": 20161
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.924646394601761e-06,
+      "loss": 0.6973,
+      "step": 20162
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.923732054538252e-06,
+      "loss": 0.6875,
+      "step": 20163
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.922817794999529e-06,
+      "loss": 0.7236,
+      "step": 20164
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.921903615997709e-06,
+      "loss": 0.6963,
+      "step": 20165
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.92098951754491e-06,
+      "loss": 0.7236,
+      "step": 20166
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.920075499653244e-06,
+      "loss": 0.7344,
+      "step": 20167
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.919161562334828e-06,
+      "loss": 0.7021,
+      "step": 20168
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.918247705601763e-06,
+      "loss": 0.7354,
+      "step": 20169
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.917333929466169e-06,
+      "loss": 0.708,
+      "step": 20170
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.916420233940154e-06,
+      "loss": 0.7129,
+      "step": 20171
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.915506619035826e-06,
+      "loss": 0.7373,
+      "step": 20172
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.914593084765292e-06,
+      "loss": 0.6963,
+      "step": 20173
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9136796311406556e-06,
+      "loss": 0.7129,
+      "step": 20174
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9127662581740325e-06,
+      "loss": 0.7266,
+      "step": 20175
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.911852965877515e-06,
+      "loss": 0.6953,
+      "step": 20176
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.910939754263212e-06,
+      "loss": 0.7012,
+      "step": 20177
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.910026623343222e-06,
+      "loss": 0.7021,
+      "step": 20178
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.909113573129652e-06,
+      "loss": 0.7061,
+      "step": 20179
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9082006036345995e-06,
+      "loss": 0.6904,
+      "step": 20180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.907287714870164e-06,
+      "loss": 0.7051,
+      "step": 20181
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.906374906848441e-06,
+      "loss": 0.6934,
+      "step": 20182
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.905462179581531e-06,
+      "loss": 0.7119,
+      "step": 20183
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.904549533081526e-06,
+      "loss": 0.7422,
+      "step": 20184
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.903636967360522e-06,
+      "loss": 0.7266,
+      "step": 20185
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.902724482430611e-06,
+      "loss": 0.7061,
+      "step": 20186
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.9018120783038905e-06,
+      "loss": 0.7383,
+      "step": 20187
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.900899754992452e-06,
+      "loss": 0.7275,
+      "step": 20188
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.899987512508379e-06,
+      "loss": 0.6924,
+      "step": 20189
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.899075350863764e-06,
+      "loss": 0.7256,
+      "step": 20190
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.898163270070697e-06,
+      "loss": 0.7324,
+      "step": 20191
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.897251270141266e-06,
+      "loss": 0.6963,
+      "step": 20192
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.896339351087555e-06,
+      "loss": 0.6973,
+      "step": 20193
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.895427512921649e-06,
+      "loss": 0.7109,
+      "step": 20194
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8945157556556325e-06,
+      "loss": 0.7197,
+      "step": 20195
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.893604079301588e-06,
+      "loss": 0.7578,
+      "step": 20196
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.892692483871599e-06,
+      "loss": 0.7383,
+      "step": 20197
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.891780969377741e-06,
+      "loss": 0.71,
+      "step": 20198
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.890869535832101e-06,
+      "loss": 0.7266,
+      "step": 20199
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.889958183246756e-06,
+      "loss": 0.6914,
+      "step": 20200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.889046911633784e-06,
+      "loss": 0.7002,
+      "step": 20201
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.888135721005253e-06,
+      "loss": 0.7148,
+      "step": 20202
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.887224611373247e-06,
+      "loss": 0.7188,
+      "step": 20203
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88631358274984e-06,
+      "loss": 0.6953,
+      "step": 20204
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.885402635147103e-06,
+      "loss": 0.7188,
+      "step": 20205
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.884491768577108e-06,
+      "loss": 0.6943,
+      "step": 20206
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.883580983051927e-06,
+      "loss": 0.7217,
+      "step": 20207
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88267027858363e-06,
+      "loss": 0.6973,
+      "step": 20208
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.881759655184286e-06,
+      "loss": 0.6963,
+      "step": 20209
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.880849112865959e-06,
+      "loss": 0.7217,
+      "step": 20210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8799386516407224e-06,
+      "loss": 0.7246,
+      "step": 20211
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.879028271520639e-06,
+      "loss": 0.7109,
+      "step": 20212
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.878117972517776e-06,
+      "loss": 0.7139,
+      "step": 20213
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.877207754644191e-06,
+      "loss": 0.7314,
+      "step": 20214
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.876297617911952e-06,
+      "loss": 0.708,
+      "step": 20215
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.875387562333118e-06,
+      "loss": 0.7246,
+      "step": 20216
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.874477587919751e-06,
+      "loss": 0.7139,
+      "step": 20217
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8735676946839075e-06,
+      "loss": 0.7275,
+      "step": 20218
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8726578826376435e-06,
+      "loss": 0.7002,
+      "step": 20219
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.871748151793027e-06,
+      "loss": 0.7031,
+      "step": 20220
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8708385021621056e-06,
+      "loss": 0.7188,
+      "step": 20221
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.869928933756935e-06,
+      "loss": 0.7129,
+      "step": 20222
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.869019446589565e-06,
+      "loss": 0.7305,
+      "step": 20223
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.868110040672058e-06,
+      "loss": 0.709,
+      "step": 20224
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.86720071601646e-06,
+      "loss": 0.7236,
+      "step": 20225
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.866291472634824e-06,
+      "loss": 0.7275,
+      "step": 20226
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.865382310539197e-06,
+      "loss": 0.7344,
+      "step": 20227
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.864473229741631e-06,
+      "loss": 0.7178,
+      "step": 20228
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.86356423025417e-06,
+      "loss": 0.7373,
+      "step": 20229
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.862655312088861e-06,
+      "loss": 0.708,
+      "step": 20230
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.861746475257747e-06,
+      "loss": 0.707,
+      "step": 20231
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.860837719772878e-06,
+      "loss": 0.7314,
+      "step": 20232
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8599290456463e-06,
+      "loss": 0.7266,
+      "step": 20233
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.859020452890044e-06,
+      "loss": 0.7217,
+      "step": 20234
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8581119415161535e-06,
+      "loss": 0.7139,
+      "step": 20235
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.857203511536675e-06,
+      "loss": 0.7441,
+      "step": 20236
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.856295162963644e-06,
+      "loss": 0.7324,
+      "step": 20237
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.855386895809097e-06,
+      "loss": 0.7285,
+      "step": 20238
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.854478710085071e-06,
+      "loss": 0.7363,
+      "step": 20239
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.853570605803603e-06,
+      "loss": 0.7334,
+      "step": 20240
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.852662582976726e-06,
+      "loss": 0.749,
+      "step": 20241
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.851754641616475e-06,
+      "loss": 0.7051,
+      "step": 20242
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850846781734877e-06,
+      "loss": 0.7109,
+      "step": 20243
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.849939003343972e-06,
+      "loss": 0.7139,
+      "step": 20244
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.849031306455786e-06,
+      "loss": 0.7002,
+      "step": 20245
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.84812369108235e-06,
+      "loss": 0.709,
+      "step": 20246
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.847216157235685e-06,
+      "loss": 0.7236,
+      "step": 20247
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.846308704927825e-06,
+      "loss": 0.708,
+      "step": 20248
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8454013341707945e-06,
+      "loss": 0.6904,
+      "step": 20249
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.844494044976617e-06,
+      "loss": 0.7148,
+      "step": 20250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.843586837357317e-06,
+      "loss": 0.7266,
+      "step": 20251
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.842679711324914e-06,
+      "loss": 0.7266,
+      "step": 20252
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.841772666891439e-06,
+      "loss": 0.6992,
+      "step": 20253
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.840865704068904e-06,
+      "loss": 0.7393,
+      "step": 20254
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.839958822869324e-06,
+      "loss": 0.6963,
+      "step": 20255
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.839052023304729e-06,
+      "loss": 0.709,
+      "step": 20256
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8381453053871305e-06,
+      "loss": 0.707,
+      "step": 20257
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8372386691285456e-06,
+      "loss": 0.7051,
+      "step": 20258
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.836332114540988e-06,
+      "loss": 0.7236,
+      "step": 20259
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.835425641636472e-06,
+      "loss": 0.7168,
+      "step": 20260
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.834519250427013e-06,
+      "loss": 0.6855,
+      "step": 20261
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.833612940924618e-06,
+      "loss": 0.748,
+      "step": 20262
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.832706713141302e-06,
+      "loss": 0.7314,
+      "step": 20263
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.831800567089069e-06,
+      "loss": 0.6924,
+      "step": 20264
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.830894502779936e-06,
+      "loss": 0.7002,
+      "step": 20265
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.829988520225909e-06,
+      "loss": 0.6963,
+      "step": 20266
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.829082619438988e-06,
+      "loss": 0.7188,
+      "step": 20267
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8281768004311785e-06,
+      "loss": 0.709,
+      "step": 20268
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.827271063214492e-06,
+      "loss": 0.7188,
+      "step": 20269
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.826365407800926e-06,
+      "loss": 0.7168,
+      "step": 20270
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.825459834202485e-06,
+      "loss": 0.6963,
+      "step": 20271
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.824554342431171e-06,
+      "loss": 0.7148,
+      "step": 20272
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.82364893249898e-06,
+      "loss": 0.7139,
+      "step": 20273
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.822743604417913e-06,
+      "loss": 0.6885,
+      "step": 20274
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.821838358199968e-06,
+      "loss": 0.7197,
+      "step": 20275
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8209331938571395e-06,
+      "loss": 0.708,
+      "step": 20276
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.820028111401426e-06,
+      "loss": 0.7148,
+      "step": 20277
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.819123110844822e-06,
+      "loss": 0.6768,
+      "step": 20278
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.818218192199322e-06,
+      "loss": 0.7266,
+      "step": 20279
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.81731335547691e-06,
+      "loss": 0.7207,
+      "step": 20280
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.816408600689588e-06,
+      "loss": 0.7129,
+      "step": 20281
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.815503927849339e-06,
+      "loss": 0.7051,
+      "step": 20282
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8145993369681544e-06,
+      "loss": 0.7158,
+      "step": 20283
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.813694828058022e-06,
+      "loss": 0.6865,
+      "step": 20284
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.81279040113093e-06,
+      "loss": 0.6982,
+      "step": 20285
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8118860561988623e-06,
+      "loss": 0.7129,
+      "step": 20286
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8109817932738038e-06,
+      "loss": 0.6924,
+      "step": 20287
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8100776123677354e-06,
+      "loss": 0.7168,
+      "step": 20288
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8091735134926465e-06,
+      "loss": 0.7012,
+      "step": 20289
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8082694966605137e-06,
+      "loss": 0.7109,
+      "step": 20290
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8073655618833195e-06,
+      "loss": 0.7266,
+      "step": 20291
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8064617091730414e-06,
+      "loss": 0.7188,
+      "step": 20292
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8055579385416586e-06,
+      "loss": 0.7188,
+      "step": 20293
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8046542500011473e-06,
+      "loss": 0.6934,
+      "step": 20294
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.803750643563484e-06,
+      "loss": 0.7178,
+      "step": 20295
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8028471192406435e-06,
+      "loss": 0.6895,
+      "step": 20296
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.801943677044597e-06,
+      "loss": 0.7061,
+      "step": 20297
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8010403169873256e-06,
+      "loss": 0.6963,
+      "step": 20298
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.800137039080792e-06,
+      "loss": 0.751,
+      "step": 20299
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.799233843336968e-06,
+      "loss": 0.7051,
+      "step": 20300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7983307297678274e-06,
+      "loss": 0.7012,
+      "step": 20301
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7974276983853366e-06,
+      "loss": 0.7334,
+      "step": 20302
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7965247492014624e-06,
+      "loss": 0.7041,
+      "step": 20303
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79562188222817e-06,
+      "loss": 0.6875,
+      "step": 20304
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.794719097477427e-06,
+      "loss": 0.7227,
+      "step": 20305
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.793816394961195e-06,
+      "loss": 0.708,
+      "step": 20306
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7929137746914393e-06,
+      "loss": 0.7285,
+      "step": 20307
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7920112366801187e-06,
+      "loss": 0.7432,
+      "step": 20308
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.791108780939193e-06,
+      "loss": 0.6748,
+      "step": 20309
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7902064074806287e-06,
+      "loss": 0.7285,
+      "step": 20310
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7893041163163825e-06,
+      "loss": 0.7168,
+      "step": 20311
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7884019074584064e-06,
+      "loss": 0.6963,
+      "step": 20312
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7874997809186566e-06,
+      "loss": 0.707,
+      "step": 20313
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7865977367090946e-06,
+      "loss": 0.6924,
+      "step": 20314
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7856957748416722e-06,
+      "loss": 0.7188,
+      "step": 20315
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7847938953283415e-06,
+      "loss": 0.6992,
+      "step": 20316
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.783892098181051e-06,
+      "loss": 0.6982,
+      "step": 20317
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7829903834117643e-06,
+      "loss": 0.7021,
+      "step": 20318
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7820887510324178e-06,
+      "loss": 0.7139,
+      "step": 20319
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7811872010549656e-06,
+      "loss": 0.7441,
+      "step": 20320
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7802857334913514e-06,
+      "loss": 0.7197,
+      "step": 20321
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7793843483535276e-06,
+      "loss": 0.7139,
+      "step": 20322
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.778483045653438e-06,
+      "loss": 0.709,
+      "step": 20323
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7775818254030294e-06,
+      "loss": 0.7188,
+      "step": 20324
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.776680687614235e-06,
+      "loss": 0.7021,
+      "step": 20325
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.775779632299007e-06,
+      "loss": 0.7119,
+      "step": 20326
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.774878659469283e-06,
+      "loss": 0.6953,
+      "step": 20327
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7739777691370037e-06,
+      "loss": 0.7285,
+      "step": 20328
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7730769613141083e-06,
+      "loss": 0.7178,
+      "step": 20329
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7721762360125325e-06,
+      "loss": 0.7256,
+      "step": 20330
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7712755932442158e-06,
+      "loss": 0.7139,
+      "step": 20331
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.770375033021092e-06,
+      "loss": 0.7285,
+      "step": 20332
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7694745553550925e-06,
+      "loss": 0.7139,
+      "step": 20333
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.768574160258157e-06,
+      "loss": 0.7246,
+      "step": 20334
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.767673847742217e-06,
+      "loss": 0.7207,
+      "step": 20335
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7667736178192003e-06,
+      "loss": 0.707,
+      "step": 20336
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7658734705010402e-06,
+      "loss": 0.7139,
+      "step": 20337
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7649734057996634e-06,
+      "loss": 0.7324,
+      "step": 20338
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.764073423726998e-06,
+      "loss": 0.709,
+      "step": 20339
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7631735242949728e-06,
+      "loss": 0.7119,
+      "step": 20340
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7622737075155123e-06,
+      "loss": 0.7275,
+      "step": 20341
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7613739734005373e-06,
+      "loss": 0.7217,
+      "step": 20342
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7604743219619823e-06,
+      "loss": 0.6895,
+      "step": 20343
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.75957475321176e-06,
+      "loss": 0.6914,
+      "step": 20344
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7586752671617933e-06,
+      "loss": 0.7227,
+      "step": 20345
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7577758638240013e-06,
+      "loss": 0.7344,
+      "step": 20346
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7568765432103095e-06,
+      "loss": 0.708,
+      "step": 20347
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.755977305332631e-06,
+      "loss": 0.749,
+      "step": 20348
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7550781502028856e-06,
+      "loss": 0.708,
+      "step": 20349
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.754179077832987e-06,
+      "loss": 0.707,
+      "step": 20350
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7532800882348495e-06,
+      "loss": 0.7119,
+      "step": 20351
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7523811814203893e-06,
+      "loss": 0.7422,
+      "step": 20352
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.751482357401518e-06,
+      "loss": 0.7168,
+      "step": 20353
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.750583616190142e-06,
+      "loss": 0.7422,
+      "step": 20354
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.749684957798181e-06,
+      "loss": 0.7422,
+      "step": 20355
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7487863822375436e-06,
+      "loss": 0.7158,
+      "step": 20356
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7478878895201296e-06,
+      "loss": 0.7148,
+      "step": 20357
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.746989479657849e-06,
+      "loss": 0.7031,
+      "step": 20358
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.746091152662612e-06,
+      "loss": 0.7207,
+      "step": 20359
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7451929085463213e-06,
+      "loss": 0.7061,
+      "step": 20360
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7442947473208815e-06,
+      "loss": 0.6992,
+      "step": 20361
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7433966689981904e-06,
+      "loss": 0.6953,
+      "step": 20362
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7424986735901604e-06,
+      "loss": 0.7119,
+      "step": 20363
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7416007611086824e-06,
+      "loss": 0.707,
+      "step": 20364
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7407029315656585e-06,
+      "loss": 0.707,
+      "step": 20365
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7398051849729846e-06,
+      "loss": 0.6826,
+      "step": 20366
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7389075213425642e-06,
+      "loss": 0.7178,
+      "step": 20367
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.738009940686289e-06,
+      "loss": 0.7129,
+      "step": 20368
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.737112443016059e-06,
+      "loss": 0.7031,
+      "step": 20369
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.736215028343757e-06,
+      "loss": 0.7178,
+      "step": 20370
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7353176966812867e-06,
+      "loss": 0.71,
+      "step": 20371
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7344204480405367e-06,
+      "loss": 0.709,
+      "step": 20372
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7335232824333966e-06,
+      "loss": 0.708,
+      "step": 20373
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.732626199871756e-06,
+      "loss": 0.7168,
+      "step": 20374
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.731729200367501e-06,
+      "loss": 0.7158,
+      "step": 20375
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7308322839325285e-06,
+      "loss": 0.7246,
+      "step": 20376
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7299354505787146e-06,
+      "loss": 0.7129,
+      "step": 20377
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.729038700317944e-06,
+      "loss": 0.7217,
+      "step": 20378
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.728142033162109e-06,
+      "loss": 0.7197,
+      "step": 20379
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.727245449123087e-06,
+      "loss": 0.708,
+      "step": 20380
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7263489482127614e-06,
+      "loss": 0.7061,
+      "step": 20381
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7254525304430123e-06,
+      "loss": 0.7061,
+      "step": 20382
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7245561958257194e-06,
+      "loss": 0.6992,
+      "step": 20383
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.723659944372762e-06,
+      "loss": 0.7188,
+      "step": 20384
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7227637760960155e-06,
+      "loss": 0.708,
+      "step": 20385
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.721867691007359e-06,
+      "loss": 0.7451,
+      "step": 20386
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.720971689118663e-06,
+      "loss": 0.7266,
+      "step": 20387
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.720075770441808e-06,
+      "loss": 0.7236,
+      "step": 20388
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.719179934988667e-06,
+      "loss": 0.7012,
+      "step": 20389
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.718284182771106e-06,
+      "loss": 0.7119,
+      "step": 20390
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7173885138009958e-06,
+      "loss": 0.6982,
+      "step": 20391
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7164929280902116e-06,
+      "loss": 0.6982,
+      "step": 20392
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.71559742565062e-06,
+      "loss": 0.7031,
+      "step": 20393
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7147020064940875e-06,
+      "loss": 0.7197,
+      "step": 20394
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7138066706324805e-06,
+      "loss": 0.7236,
+      "step": 20395
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.712911418077666e-06,
+      "loss": 0.7314,
+      "step": 20396
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7120162488415055e-06,
+      "loss": 0.7402,
+      "step": 20397
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7111211629358645e-06,
+      "loss": 0.707,
+      "step": 20398
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.710226160372601e-06,
+      "loss": 0.7412,
+      "step": 20399
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.709331241163582e-06,
+      "loss": 0.7275,
+      "step": 20400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7084364053206636e-06,
+      "loss": 0.7031,
+      "step": 20401
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7075416528557095e-06,
+      "loss": 0.71,
+      "step": 20402
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.706646983780565e-06,
+      "loss": 0.7168,
+      "step": 20403
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7057523981070985e-06,
+      "loss": 0.709,
+      "step": 20404
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7048578958471606e-06,
+      "loss": 0.6963,
+      "step": 20405
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7039634770126064e-06,
+      "loss": 0.7041,
+      "step": 20406
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7030691416152864e-06,
+      "loss": 0.7471,
+      "step": 20407
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7021748896670606e-06,
+      "loss": 0.7148,
+      "step": 20408
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.701280721179772e-06,
+      "loss": 0.7188,
+      "step": 20409
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7003866361652718e-06,
+      "loss": 0.6992,
+      "step": 20410
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6994926346354065e-06,
+      "loss": 0.7256,
+      "step": 20411
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.698598716602031e-06,
+      "loss": 0.709,
+      "step": 20412
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.697704882076988e-06,
+      "loss": 0.7188,
+      "step": 20413
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.696811131072122e-06,
+      "loss": 0.7373,
+      "step": 20414
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.695917463599278e-06,
+      "loss": 0.7275,
+      "step": 20415
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6950238796702996e-06,
+      "loss": 0.7061,
+      "step": 20416
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6941303792970283e-06,
+      "loss": 0.7139,
+      "step": 20417
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6932369624913055e-06,
+      "loss": 0.7295,
+      "step": 20418
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6923436292649705e-06,
+      "loss": 0.7012,
+      "step": 20419
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6914503796298586e-06,
+      "loss": 0.7139,
+      "step": 20420
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6905572135978185e-06,
+      "loss": 0.7217,
+      "step": 20421
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.689664131180676e-06,
+      "loss": 0.6982,
+      "step": 20422
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6887711323902662e-06,
+      "loss": 0.709,
+      "step": 20423
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.687878217238431e-06,
+      "loss": 0.6982,
+      "step": 20424
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.686985385737e-06,
+      "loss": 0.7148,
+      "step": 20425
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6860926378978046e-06,
+      "loss": 0.707,
+      "step": 20426
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.685199973732676e-06,
+      "loss": 0.7314,
+      "step": 20427
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6843073932534446e-06,
+      "loss": 0.6963,
+      "step": 20428
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6834148964719397e-06,
+      "loss": 0.751,
+      "step": 20429
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6825224833999883e-06,
+      "loss": 0.709,
+      "step": 20430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.681630154049416e-06,
+      "loss": 0.7061,
+      "step": 20431
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.680737908432046e-06,
+      "loss": 0.7393,
+      "step": 20432
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6798457465597105e-06,
+      "loss": 0.7119,
+      "step": 20433
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6789536684442306e-06,
+      "loss": 0.7598,
+      "step": 20434
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6780616740974216e-06,
+      "loss": 0.7188,
+      "step": 20435
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6771697635311064e-06,
+      "loss": 0.7051,
+      "step": 20436
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.676277936757111e-06,
+      "loss": 0.6943,
+      "step": 20437
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6753861937872504e-06,
+      "loss": 0.7305,
+      "step": 20438
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6744945346333426e-06,
+      "loss": 0.7412,
+      "step": 20439
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6736029593072e-06,
+      "loss": 0.7031,
+      "step": 20440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.672711467820649e-06,
+      "loss": 0.7344,
+      "step": 20441
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6718200601854935e-06,
+      "loss": 0.7295,
+      "step": 20442
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.67092873641355e-06,
+      "loss": 0.7188,
+      "step": 20443
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6700374965166275e-06,
+      "loss": 0.7178,
+      "step": 20444
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.669146340506544e-06,
+      "loss": 0.7188,
+      "step": 20445
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6682552683951066e-06,
+      "loss": 0.7188,
+      "step": 20446
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6673642801941254e-06,
+      "loss": 0.7188,
+      "step": 20447
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6664733759154004e-06,
+      "loss": 0.708,
+      "step": 20448
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6655825555707457e-06,
+      "loss": 0.7178,
+      "step": 20449
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6646918191719662e-06,
+      "loss": 0.7285,
+      "step": 20450
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.663801166730864e-06,
+      "loss": 0.7246,
+      "step": 20451
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6629105982592427e-06,
+      "loss": 0.7178,
+      "step": 20452
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6620201137689025e-06,
+      "loss": 0.71,
+      "step": 20453
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.661129713271653e-06,
+      "loss": 0.6992,
+      "step": 20454
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.660239396779285e-06,
+      "loss": 0.7627,
+      "step": 20455
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.659349164303596e-06,
+      "loss": 0.7178,
+      "step": 20456
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6584590158563917e-06,
+      "loss": 0.7256,
+      "step": 20457
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.657568951449465e-06,
+      "loss": 0.7334,
+      "step": 20458
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6566789710946103e-06,
+      "loss": 0.708,
+      "step": 20459
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6557890748036227e-06,
+      "loss": 0.7148,
+      "step": 20460
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.654899262588296e-06,
+      "loss": 0.6816,
+      "step": 20461
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.654009534460422e-06,
+      "loss": 0.7266,
+      "step": 20462
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.653119890431791e-06,
+      "loss": 0.7061,
+      "step": 20463
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6522303305141935e-06,
+      "loss": 0.7109,
+      "step": 20464
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.651340854719415e-06,
+      "loss": 0.7188,
+      "step": 20465
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.650451463059249e-06,
+      "loss": 0.7461,
+      "step": 20466
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.649562155545483e-06,
+      "loss": 0.7197,
+      "step": 20467
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6486729321898916e-06,
+      "loss": 0.7031,
+      "step": 20468
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.647783793004269e-06,
+      "loss": 0.708,
+      "step": 20469
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.646894738000396e-06,
+      "loss": 0.6836,
+      "step": 20470
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.646005767190055e-06,
+      "loss": 0.7158,
+      "step": 20471
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6451168805850244e-06,
+      "loss": 0.707,
+      "step": 20472
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6442280781970874e-06,
+      "loss": 0.7197,
+      "step": 20473
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6433393600380195e-06,
+      "loss": 0.7334,
+      "step": 20474
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6424507261196007e-06,
+      "loss": 0.7178,
+      "step": 20475
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.641562176453607e-06,
+      "loss": 0.7158,
+      "step": 20476
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.64067371105181e-06,
+      "loss": 0.7275,
+      "step": 20477
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.63978532992599e-06,
+      "loss": 0.7188,
+      "step": 20478
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.638897033087917e-06,
+      "loss": 0.7275,
+      "step": 20479
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.638008820549368e-06,
+      "loss": 0.6934,
+      "step": 20480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6371206923221025e-06,
+      "loss": 0.7168,
+      "step": 20481
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6362326484179e-06,
+      "loss": 0.7002,
+      "step": 20482
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6353446888485267e-06,
+      "loss": 0.7021,
+      "step": 20483
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6344568136257494e-06,
+      "loss": 0.7334,
+      "step": 20484
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6335690227613317e-06,
+      "loss": 0.7334,
+      "step": 20485
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6326813162670495e-06,
+      "loss": 0.7012,
+      "step": 20486
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.631793694154655e-06,
+      "loss": 0.6816,
+      "step": 20487
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.630906156435916e-06,
+      "loss": 0.7207,
+      "step": 20488
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.630018703122592e-06,
+      "loss": 0.6934,
+      "step": 20489
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.629131334226449e-06,
+      "loss": 0.7041,
+      "step": 20490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.628244049759244e-06,
+      "loss": 0.6914,
+      "step": 20491
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.627356849732735e-06,
+      "loss": 0.6904,
+      "step": 20492
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6264697341586806e-06,
+      "loss": 0.6826,
+      "step": 20493
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6255827030488366e-06,
+      "loss": 0.7285,
+      "step": 20494
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6246957564149577e-06,
+      "loss": 0.7354,
+      "step": 20495
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6238088942687977e-06,
+      "loss": 0.7002,
+      "step": 20496
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6229221166221117e-06,
+      "loss": 0.7148,
+      "step": 20497
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6220354234866463e-06,
+      "loss": 0.7012,
+      "step": 20498
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6211488148741636e-06,
+      "loss": 0.7041,
+      "step": 20499
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6202622907964012e-06,
+      "loss": 0.6973,
+      "step": 20500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.61937585126511e-06,
+      "loss": 0.7188,
+      "step": 20501
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6184894962920423e-06,
+      "loss": 0.7305,
+      "step": 20502
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6176032258889425e-06,
+      "loss": 0.7178,
+      "step": 20503
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6167170400675543e-06,
+      "loss": 0.7168,
+      "step": 20504
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.615830938839623e-06,
+      "loss": 0.7236,
+      "step": 20505
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.614944922216891e-06,
+      "loss": 0.7207,
+      "step": 20506
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6140589902111e-06,
+      "loss": 0.7051,
+      "step": 20507
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.613173142833991e-06,
+      "loss": 0.6836,
+      "step": 20508
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6122873800973026e-06,
+      "loss": 0.7275,
+      "step": 20509
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.611401702012771e-06,
+      "loss": 0.6992,
+      "step": 20510
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6105161085921404e-06,
+      "loss": 0.7158,
+      "step": 20511
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.609630599847146e-06,
+      "loss": 0.6934,
+      "step": 20512
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.608745175789512e-06,
+      "loss": 0.7158,
+      "step": 20513
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6078598364309846e-06,
+      "loss": 0.7168,
+      "step": 20514
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6069745817832924e-06,
+      "loss": 0.7246,
+      "step": 20515
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6060894118581667e-06,
+      "loss": 0.7129,
+      "step": 20516
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6052043266673388e-06,
+      "loss": 0.6846,
+      "step": 20517
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.604319326222534e-06,
+      "loss": 0.7051,
+      "step": 20518
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6034344105354915e-06,
+      "loss": 0.7217,
+      "step": 20519
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.602549579617927e-06,
+      "loss": 0.7129,
+      "step": 20520
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.601664833481571e-06,
+      "loss": 0.7188,
+      "step": 20521
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.600780172138145e-06,
+      "loss": 0.708,
+      "step": 20522
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5998955955993807e-06,
+      "loss": 0.6914,
+      "step": 20523
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.599011103876996e-06,
+      "loss": 0.7285,
+      "step": 20524
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5981266969827155e-06,
+      "loss": 0.7207,
+      "step": 20525
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5972423749282505e-06,
+      "loss": 0.7002,
+      "step": 20526
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5963581377253296e-06,
+      "loss": 0.7109,
+      "step": 20527
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.595473985385669e-06,
+      "loss": 0.7129,
+      "step": 20528
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5945899179209853e-06,
+      "loss": 0.7148,
+      "step": 20529
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.593705935342989e-06,
+      "loss": 0.7051,
+      "step": 20530
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.592822037663406e-06,
+      "loss": 0.7148,
+      "step": 20531
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.591938224893946e-06,
+      "loss": 0.7236,
+      "step": 20532
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.591054497046317e-06,
+      "loss": 0.7061,
+      "step": 20533
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.590170854132229e-06,
+      "loss": 0.7324,
+      "step": 20534
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5892872961634005e-06,
+      "loss": 0.7012,
+      "step": 20535
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.588403823151536e-06,
+      "loss": 0.7236,
+      "step": 20536
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5875204351083458e-06,
+      "loss": 0.7266,
+      "step": 20537
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.586637132045534e-06,
+      "loss": 0.7236,
+      "step": 20538
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5857539139748088e-06,
+      "loss": 0.7305,
+      "step": 20539
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5848707809078733e-06,
+      "loss": 0.71,
+      "step": 20540
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.583987732856432e-06,
+      "loss": 0.708,
+      "step": 20541
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5831047698321873e-06,
+      "loss": 0.7256,
+      "step": 20542
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5822218918468365e-06,
+      "loss": 0.7168,
+      "step": 20543
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.58133909891209e-06,
+      "loss": 0.7148,
+      "step": 20544
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5804563910396383e-06,
+      "loss": 0.7139,
+      "step": 20545
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5795737682411767e-06,
+      "loss": 0.7207,
+      "step": 20546
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5786912305284116e-06,
+      "loss": 0.7178,
+      "step": 20547
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.577808777913033e-06,
+      "loss": 0.7314,
+      "step": 20548
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.576926410406737e-06,
+      "loss": 0.708,
+      "step": 20549
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5760441280212153e-06,
+      "loss": 0.7061,
+      "step": 20550
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.575161930768163e-06,
+      "loss": 0.6943,
+      "step": 20551
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5742798186592687e-06,
+      "loss": 0.7129,
+      "step": 20552
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5733977917062236e-06,
+      "loss": 0.7275,
+      "step": 20553
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.572515849920716e-06,
+      "loss": 0.7197,
+      "step": 20554
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5716339933144315e-06,
+      "loss": 0.7061,
+      "step": 20555
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.570752221899062e-06,
+      "loss": 0.7178,
+      "step": 20556
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.569870535686294e-06,
+      "loss": 0.7031,
+      "step": 20557
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5689889346878047e-06,
+      "loss": 0.6836,
+      "step": 20558
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5681074189152786e-06,
+      "loss": 0.7002,
+      "step": 20559
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5672259883804027e-06,
+      "loss": 0.7148,
+      "step": 20560
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5663446430948557e-06,
+      "loss": 0.6973,
+      "step": 20561
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.565463383070318e-06,
+      "loss": 0.7031,
+      "step": 20562
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.564582208318463e-06,
+      "loss": 0.6885,
+      "step": 20563
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.563701118850981e-06,
+      "loss": 0.7227,
+      "step": 20564
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.562820114679537e-06,
+      "loss": 0.7207,
+      "step": 20565
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.56193919581581e-06,
+      "loss": 0.7148,
+      "step": 20566
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.56105836227147e-06,
+      "loss": 0.7051,
+      "step": 20567
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5601776140581976e-06,
+      "loss": 0.6748,
+      "step": 20568
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.559296951187663e-06,
+      "loss": 0.7002,
+      "step": 20569
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.558416373671537e-06,
+      "loss": 0.7119,
+      "step": 20570
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.557535881521482e-06,
+      "loss": 0.707,
+      "step": 20571
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5566554747491755e-06,
+      "loss": 0.7031,
+      "step": 20572
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.555775153366282e-06,
+      "loss": 0.6982,
+      "step": 20573
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5548949173844684e-06,
+      "loss": 0.6855,
+      "step": 20574
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5540147668153948e-06,
+      "loss": 0.6846,
+      "step": 20575
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.553134701670733e-06,
+      "loss": 0.7129,
+      "step": 20576
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5522547219621462e-06,
+      "loss": 0.7158,
+      "step": 20577
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5513748277012893e-06,
+      "loss": 0.6748,
+      "step": 20578
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5504950188998222e-06,
+      "loss": 0.6992,
+      "step": 20579
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.549615295569412e-06,
+      "loss": 0.7188,
+      "step": 20580
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5487356577217136e-06,
+      "loss": 0.6924,
+      "step": 20581
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5478561053683846e-06,
+      "loss": 0.7148,
+      "step": 20582
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.54697663852108e-06,
+      "loss": 0.7236,
+      "step": 20583
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5460972571914565e-06,
+      "loss": 0.71,
+      "step": 20584
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5452179613911664e-06,
+      "loss": 0.7178,
+      "step": 20585
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.544338751131863e-06,
+      "loss": 0.7197,
+      "step": 20586
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5434596264251973e-06,
+      "loss": 0.6797,
+      "step": 20587
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.542580587282817e-06,
+      "loss": 0.6904,
+      "step": 20588
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.541701633716379e-06,
+      "loss": 0.6992,
+      "step": 20589
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5408227657375306e-06,
+      "loss": 0.7266,
+      "step": 20590
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5399439833579076e-06,
+      "loss": 0.7148,
+      "step": 20591
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5390652865891684e-06,
+      "loss": 0.7295,
+      "step": 20592
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.538186675442953e-06,
+      "loss": 0.71,
+      "step": 20593
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5373081499309047e-06,
+      "loss": 0.6963,
+      "step": 20594
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5364297100646673e-06,
+      "loss": 0.7148,
+      "step": 20595
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.535551355855882e-06,
+      "loss": 0.7158,
+      "step": 20596
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5346730873161873e-06,
+      "loss": 0.7236,
+      "step": 20597
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.533794904457224e-06,
+      "loss": 0.6914,
+      "step": 20598
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5329168072906307e-06,
+      "loss": 0.71,
+      "step": 20599
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5320387958280387e-06,
+      "loss": 0.7266,
+      "step": 20600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.531160870081092e-06,
+      "loss": 0.7139,
+      "step": 20601
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.530283030061421e-06,
+      "loss": 0.7207,
+      "step": 20602
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.529405275780663e-06,
+      "loss": 0.709,
+      "step": 20603
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5285276072504406e-06,
+      "loss": 0.7012,
+      "step": 20604
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5276500244823943e-06,
+      "loss": 0.7168,
+      "step": 20605
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5267725274881504e-06,
+      "loss": 0.709,
+      "step": 20606
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.525895116279339e-06,
+      "loss": 0.7207,
+      "step": 20607
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.525017790867584e-06,
+      "loss": 0.7285,
+      "step": 20608
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.524140551264522e-06,
+      "loss": 0.6982,
+      "step": 20609
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5232633974817677e-06,
+      "loss": 0.7139,
+      "step": 20610
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.52238632953095e-06,
+      "loss": 0.7217,
+      "step": 20611
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.521509347423688e-06,
+      "loss": 0.6846,
+      "step": 20612
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5206324511716117e-06,
+      "loss": 0.7021,
+      "step": 20613
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5197556407863364e-06,
+      "loss": 0.7246,
+      "step": 20614
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5188789162794845e-06,
+      "loss": 0.709,
+      "step": 20615
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.518002277662672e-06,
+      "loss": 0.7031,
+      "step": 20616
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.517125724947519e-06,
+      "loss": 0.7148,
+      "step": 20617
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5162492581456398e-06,
+      "loss": 0.7051,
+      "step": 20618
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5153728772686513e-06,
+      "loss": 0.7002,
+      "step": 20619
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.514496582328163e-06,
+      "loss": 0.7109,
+      "step": 20620
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.513620373335794e-06,
+      "loss": 0.7041,
+      "step": 20621
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5127442503031595e-06,
+      "loss": 0.708,
+      "step": 20622
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5118682132418592e-06,
+      "loss": 0.6855,
+      "step": 20623
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.510992262163505e-06,
+      "loss": 0.7041,
+      "step": 20624
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5101163970797115e-06,
+      "loss": 0.7021,
+      "step": 20625
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5092406180020832e-06,
+      "loss": 0.7002,
+      "step": 20626
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5083649249422246e-06,
+      "loss": 0.7227,
+      "step": 20627
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5074893179117433e-06,
+      "loss": 0.6904,
+      "step": 20628
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5066137969222402e-06,
+      "loss": 0.7217,
+      "step": 20629
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5057383619853203e-06,
+      "loss": 0.7148,
+      "step": 20630
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.504863013112584e-06,
+      "loss": 0.7227,
+      "step": 20631
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5039877503156316e-06,
+      "loss": 0.7227,
+      "step": 20632
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.50311257360606e-06,
+      "loss": 0.7197,
+      "step": 20633
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.502237482995473e-06,
+      "loss": 0.7148,
+      "step": 20634
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5013624784954682e-06,
+      "loss": 0.7305,
+      "step": 20635
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5004875601176312e-06,
+      "loss": 0.7148,
+      "step": 20636
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.499612727873567e-06,
+      "loss": 0.7256,
+      "step": 20637
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4987379817748667e-06,
+      "loss": 0.6797,
+      "step": 20638
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4978633218331203e-06,
+      "loss": 0.7197,
+      "step": 20639
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4969887480599216e-06,
+      "loss": 0.7012,
+      "step": 20640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4961142604668553e-06,
+      "loss": 0.707,
+      "step": 20641
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4952398590655225e-06,
+      "loss": 0.6992,
+      "step": 20642
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4943655438674994e-06,
+      "loss": 0.7178,
+      "step": 20643
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4934913148843762e-06,
+      "loss": 0.7041,
+      "step": 20644
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.492617172127737e-06,
+      "loss": 0.7129,
+      "step": 20645
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4917431156091706e-06,
+      "loss": 0.6914,
+      "step": 20646
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4908691453402575e-06,
+      "loss": 0.7275,
+      "step": 20647
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.489995261332585e-06,
+      "loss": 0.7012,
+      "step": 20648
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.489121463597721e-06,
+      "loss": 0.7139,
+      "step": 20649
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.488247752147258e-06,
+      "loss": 0.6924,
+      "step": 20650
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4873741269927707e-06,
+      "loss": 0.6963,
+      "step": 20651
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4865005881458358e-06,
+      "loss": 0.6895,
+      "step": 20652
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.485627135618027e-06,
+      "loss": 0.7285,
+      "step": 20653
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4847537694209275e-06,
+      "loss": 0.6758,
+      "step": 20654
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4838804895661095e-06,
+      "loss": 0.7002,
+      "step": 20655
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4830072960651395e-06,
+      "loss": 0.7236,
+      "step": 20656
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4821341889295903e-06,
+      "loss": 0.7168,
+      "step": 20657
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4812611681710394e-06,
+      "loss": 0.6963,
+      "step": 20658
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.480388233801052e-06,
+      "loss": 0.7002,
+      "step": 20659
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.479515385831197e-06,
+      "loss": 0.7285,
+      "step": 20660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.478642624273042e-06,
+      "loss": 0.7246,
+      "step": 20661
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4777699491381533e-06,
+      "loss": 0.7119,
+      "step": 20662
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.476897360438095e-06,
+      "loss": 0.7363,
+      "step": 20663
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4760248581844313e-06,
+      "loss": 0.6875,
+      "step": 20664
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.475152442388726e-06,
+      "loss": 0.7148,
+      "step": 20665
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.474280113062535e-06,
+      "loss": 0.6904,
+      "step": 20666
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.473407870217428e-06,
+      "loss": 0.7295,
+      "step": 20667
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4725357138649617e-06,
+      "loss": 0.7168,
+      "step": 20668
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.471663644016686e-06,
+      "loss": 0.7168,
+      "step": 20669
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4707916606841685e-06,
+      "loss": 0.7285,
+      "step": 20670
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4699197638789596e-06,
+      "loss": 0.7119,
+      "step": 20671
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4690479536126152e-06,
+      "loss": 0.7012,
+      "step": 20672
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4681762298966883e-06,
+      "loss": 0.7256,
+      "step": 20673
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.467304592742732e-06,
+      "loss": 0.6836,
+      "step": 20674
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.466433042162296e-06,
+      "loss": 0.7129,
+      "step": 20675
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.465561578166933e-06,
+      "loss": 0.71,
+      "step": 20676
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4646902007681895e-06,
+      "loss": 0.7041,
+      "step": 20677
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4638189099776108e-06,
+      "loss": 0.7334,
+      "step": 20678
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4629477058067507e-06,
+      "loss": 0.6992,
+      "step": 20679
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4620765882671504e-06,
+      "loss": 0.708,
+      "step": 20680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.461205557370355e-06,
+      "loss": 0.7041,
+      "step": 20681
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.460334613127908e-06,
+      "loss": 0.707,
+      "step": 20682
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.45946375555135e-06,
+      "loss": 0.7139,
+      "step": 20683
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4585929846522236e-06,
+      "loss": 0.7148,
+      "step": 20684
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4577223004420677e-06,
+      "loss": 0.7266,
+      "step": 20685
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.456851702932418e-06,
+      "loss": 0.6953,
+      "step": 20686
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.455981192134821e-06,
+      "loss": 0.7021,
+      "step": 20687
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.455110768060804e-06,
+      "loss": 0.7109,
+      "step": 20688
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.454240430721905e-06,
+      "loss": 0.7295,
+      "step": 20689
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.453370180129655e-06,
+      "loss": 0.707,
+      "step": 20690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4525000162955934e-06,
+      "loss": 0.7119,
+      "step": 20691
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.451629939231248e-06,
+      "loss": 0.7236,
+      "step": 20692
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.45075994894815e-06,
+      "loss": 0.7422,
+      "step": 20693
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.449890045457829e-06,
+      "loss": 0.7021,
+      "step": 20694
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4490202287718135e-06,
+      "loss": 0.7168,
+      "step": 20695
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.448150498901629e-06,
+      "loss": 0.7188,
+      "step": 20696
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4472808558588034e-06,
+      "loss": 0.6943,
+      "step": 20697
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.446411299654856e-06,
+      "loss": 0.7393,
+      "step": 20698
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.44554183030132e-06,
+      "loss": 0.6914,
+      "step": 20699
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4446724478097147e-06,
+      "loss": 0.7285,
+      "step": 20700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4438031521915573e-06,
+      "loss": 0.7246,
+      "step": 20701
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4429339434583673e-06,
+      "loss": 0.7158,
+      "step": 20702
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.44206482162167e-06,
+      "loss": 0.7334,
+      "step": 20703
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.44119578669298e-06,
+      "loss": 0.7119,
+      "step": 20704
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4403268386838153e-06,
+      "loss": 0.7227,
+      "step": 20705
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4394579776056893e-06,
+      "loss": 0.6924,
+      "step": 20706
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4385892034701176e-06,
+      "loss": 0.7168,
+      "step": 20707
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4377205162886142e-06,
+      "loss": 0.7139,
+      "step": 20708
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.43685191607269e-06,
+      "loss": 0.7334,
+      "step": 20709
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4359834028338556e-06,
+      "loss": 0.7217,
+      "step": 20710
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.43511497658362e-06,
+      "loss": 0.7432,
+      "step": 20711
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.434246637333496e-06,
+      "loss": 0.708,
+      "step": 20712
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433378385094993e-06,
+      "loss": 0.7041,
+      "step": 20713
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4325102198796057e-06,
+      "loss": 0.7061,
+      "step": 20714
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.43164214169885e-06,
+      "loss": 0.7412,
+      "step": 20715
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.430774150564228e-06,
+      "loss": 0.7021,
+      "step": 20716
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.42990624648724e-06,
+      "loss": 0.7031,
+      "step": 20717
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.429038429479389e-06,
+      "loss": 0.7129,
+      "step": 20718
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4281706995521733e-06,
+      "loss": 0.7129,
+      "step": 20719
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4273030567171005e-06,
+      "loss": 0.6992,
+      "step": 20720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.426435500985661e-06,
+      "loss": 0.7119,
+      "step": 20721
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.425568032369353e-06,
+      "loss": 0.7363,
+      "step": 20722
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.424700650879671e-06,
+      "loss": 0.7158,
+      "step": 20723
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4238333565281158e-06,
+      "loss": 0.7344,
+      "step": 20724
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4229661493261767e-06,
+      "loss": 0.7236,
+      "step": 20725
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4220990292853485e-06,
+      "loss": 0.7139,
+      "step": 20726
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4212319964171205e-06,
+      "loss": 0.7295,
+      "step": 20727
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4203650507329846e-06,
+      "loss": 0.7178,
+      "step": 20728
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.419498192244428e-06,
+      "loss": 0.748,
+      "step": 20729
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.41863142096294e-06,
+      "loss": 0.7061,
+      "step": 20730
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4177647369000044e-06,
+      "loss": 0.7178,
+      "step": 20731
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.416898140067112e-06,
+      "loss": 0.7168,
+      "step": 20732
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.416031630475748e-06,
+      "loss": 0.7178,
+      "step": 20733
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4151652081373887e-06,
+      "loss": 0.708,
+      "step": 20734
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4142988730635174e-06,
+      "loss": 0.7207,
+      "step": 20735
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4134326252656192e-06,
+      "loss": 0.6865,
+      "step": 20736
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4125664647551735e-06,
+      "loss": 0.7168,
+      "step": 20737
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4117003915436576e-06,
+      "loss": 0.7334,
+      "step": 20738
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.410834405642548e-06,
+      "loss": 0.7139,
+      "step": 20739
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4099685070633235e-06,
+      "loss": 0.7295,
+      "step": 20740
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4091026958174565e-06,
+      "loss": 0.6953,
+      "step": 20741
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4082369719164234e-06,
+      "loss": 0.7324,
+      "step": 20742
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.407371335371692e-06,
+      "loss": 0.6846,
+      "step": 20743
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4065057861947426e-06,
+      "loss": 0.7197,
+      "step": 20744
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4056403243970406e-06,
+      "loss": 0.7012,
+      "step": 20745
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4047749499900583e-06,
+      "loss": 0.7207,
+      "step": 20746
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4039096629852563e-06,
+      "loss": 0.7139,
+      "step": 20747
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4030444633941108e-06,
+      "loss": 0.7344,
+      "step": 20748
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4021793512280822e-06,
+      "loss": 0.7178,
+      "step": 20749
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.401314326498638e-06,
+      "loss": 0.7217,
+      "step": 20750
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4004493892172407e-06,
+      "loss": 0.7314,
+      "step": 20751
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3995845393953543e-06,
+      "loss": 0.7012,
+      "step": 20752
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.398719777044437e-06,
+      "loss": 0.7188,
+      "step": 20753
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.397855102175951e-06,
+      "loss": 0.7285,
+      "step": 20754
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.396990514801355e-06,
+      "loss": 0.708,
+      "step": 20755
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.396126014932103e-06,
+      "loss": 0.7227,
+      "step": 20756
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.395261602579659e-06,
+      "loss": 0.7256,
+      "step": 20757
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3943972777554744e-06,
+      "loss": 0.7139,
+      "step": 20758
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3935330404710043e-06,
+      "loss": 0.7275,
+      "step": 20759
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3926688907377015e-06,
+      "loss": 0.7129,
+      "step": 20760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.391804828567018e-06,
+      "loss": 0.6963,
+      "step": 20761
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.390940853970405e-06,
+      "loss": 0.7139,
+      "step": 20762
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3900769669593114e-06,
+      "loss": 0.7061,
+      "step": 20763
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3892131675451835e-06,
+      "loss": 0.7129,
+      "step": 20764
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.388349455739478e-06,
+      "loss": 0.7061,
+      "step": 20765
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3874858315536306e-06,
+      "loss": 0.7021,
+      "step": 20766
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3866222949990903e-06,
+      "loss": 0.7148,
+      "step": 20767
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.385758846087297e-06,
+      "loss": 0.7412,
+      "step": 20768
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.384895484829701e-06,
+      "loss": 0.7188,
+      "step": 20769
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.384032211237741e-06,
+      "loss": 0.6895,
+      "step": 20770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3831690253228587e-06,
+      "loss": 0.7168,
+      "step": 20771
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3823059270964853e-06,
+      "loss": 0.708,
+      "step": 20772
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3814429165700678e-06,
+      "loss": 0.6904,
+      "step": 20773
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3805799937550397e-06,
+      "loss": 0.707,
+      "step": 20774
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3797171586628386e-06,
+      "loss": 0.71,
+      "step": 20775
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3788544113048928e-06,
+      "loss": 0.7188,
+      "step": 20776
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.377991751692644e-06,
+      "loss": 0.7344,
+      "step": 20777
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.377129179837525e-06,
+      "loss": 0.7119,
+      "step": 20778
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3762666957509593e-06,
+      "loss": 0.7266,
+      "step": 20779
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.375404299444378e-06,
+      "loss": 0.7109,
+      "step": 20780
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3745419909292143e-06,
+      "loss": 0.7041,
+      "step": 20781
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3736797702168943e-06,
+      "loss": 0.7158,
+      "step": 20782
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3728176373188436e-06,
+      "loss": 0.6943,
+      "step": 20783
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3719555922464886e-06,
+      "loss": 0.6895,
+      "step": 20784
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.371093635011252e-06,
+      "loss": 0.707,
+      "step": 20785
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3702317656245575e-06,
+      "loss": 0.7236,
+      "step": 20786
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.369369984097828e-06,
+      "loss": 0.7031,
+      "step": 20787
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3685082904424784e-06,
+      "loss": 0.7041,
+      "step": 20788
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3676466846699362e-06,
+      "loss": 0.6934,
+      "step": 20789
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.366785166791616e-06,
+      "loss": 0.7158,
+      "step": 20790
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.365923736818939e-06,
+      "loss": 0.7148,
+      "step": 20791
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.36506239476331e-06,
+      "loss": 0.7256,
+      "step": 20792
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.364201140636154e-06,
+      "loss": 0.7178,
+      "step": 20793
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3633399744488826e-06,
+      "loss": 0.7041,
+      "step": 20794
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3624788962129073e-06,
+      "loss": 0.6934,
+      "step": 20795
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3616179059396405e-06,
+      "loss": 0.7139,
+      "step": 20796
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.360757003640489e-06,
+      "loss": 0.7031,
+      "step": 20797
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3598961893268656e-06,
+      "loss": 0.7197,
+      "step": 20798
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.359035463010176e-06,
+      "loss": 0.7178,
+      "step": 20799
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3581748247018287e-06,
+      "loss": 0.7148,
+      "step": 20800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3573142744132237e-06,
+      "loss": 0.6982,
+      "step": 20801
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.356453812155772e-06,
+      "loss": 0.7002,
+      "step": 20802
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.355593437940876e-06,
+      "loss": 0.708,
+      "step": 20803
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.354733151779935e-06,
+      "loss": 0.7041,
+      "step": 20804
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3538729536843508e-06,
+      "loss": 0.7129,
+      "step": 20805
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.353012843665523e-06,
+      "loss": 0.7119,
+      "step": 20806
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.35215282173485e-06,
+      "loss": 0.7119,
+      "step": 20807
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3512928879037297e-06,
+      "loss": 0.7061,
+      "step": 20808
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.350433042183554e-06,
+      "loss": 0.6992,
+      "step": 20809
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3495732845857277e-06,
+      "loss": 0.7285,
+      "step": 20810
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.348713615121636e-06,
+      "loss": 0.6953,
+      "step": 20811
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.347854033802673e-06,
+      "loss": 0.7178,
+      "step": 20812
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3469945406402294e-06,
+      "loss": 0.7012,
+      "step": 20813
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3461351356456997e-06,
+      "loss": 0.7275,
+      "step": 20814
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.34527581883047e-06,
+      "loss": 0.7217,
+      "step": 20815
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.34441659020593e-06,
+      "loss": 0.7168,
+      "step": 20816
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3435574497834654e-06,
+      "loss": 0.6973,
+      "step": 20817
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.342698397574461e-06,
+      "loss": 0.7207,
+      "step": 20818
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3418394335903025e-06,
+      "loss": 0.7041,
+      "step": 20819
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3409805578423725e-06,
+      "loss": 0.71,
+      "step": 20820
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3401217703420497e-06,
+      "loss": 0.7139,
+      "step": 20821
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3392630711007236e-06,
+      "loss": 0.7168,
+      "step": 20822
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3384044601297703e-06,
+      "loss": 0.7246,
+      "step": 20823
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.337545937440565e-06,
+      "loss": 0.7002,
+      "step": 20824
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3366875030444843e-06,
+      "loss": 0.7217,
+      "step": 20825
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.33582915695291e-06,
+      "loss": 0.6992,
+      "step": 20826
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3349708991772144e-06,
+      "loss": 0.7256,
+      "step": 20827
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3341127297287735e-06,
+      "loss": 0.7197,
+      "step": 20828
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3332546486189563e-06,
+      "loss": 0.7061,
+      "step": 20829
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.332396655859137e-06,
+      "loss": 0.7168,
+      "step": 20830
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.331538751460686e-06,
+      "loss": 0.749,
+      "step": 20831
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.330680935434971e-06,
+      "loss": 0.7227,
+      "step": 20832
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.329823207793359e-06,
+      "loss": 0.7334,
+      "step": 20833
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3289655685472212e-06,
+      "loss": 0.7168,
+      "step": 20834
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3281080177079206e-06,
+      "loss": 0.709,
+      "step": 20835
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3272505552868263e-06,
+      "loss": 0.707,
+      "step": 20836
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.32639318129529e-06,
+      "loss": 0.7275,
+      "step": 20837
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.325535895744686e-06,
+      "loss": 0.7031,
+      "step": 20838
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3246786986463697e-06,
+      "loss": 0.6973,
+      "step": 20839
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.323821590011703e-06,
+      "loss": 0.7197,
+      "step": 20840
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.322964569852043e-06,
+      "loss": 0.7305,
+      "step": 20841
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3221076381787454e-06,
+      "loss": 0.7021,
+      "step": 20842
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.321250795003175e-06,
+      "loss": 0.7158,
+      "step": 20843
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3203940403366774e-06,
+      "loss": 0.7227,
+      "step": 20844
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.319537374190611e-06,
+      "loss": 0.7266,
+      "step": 20845
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3186807965763245e-06,
+      "loss": 0.7129,
+      "step": 20846
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3178243075051764e-06,
+      "loss": 0.707,
+      "step": 20847
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.316967906988513e-06,
+      "loss": 0.6924,
+      "step": 20848
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.316111595037684e-06,
+      "loss": 0.6934,
+      "step": 20849
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.315255371664039e-06,
+      "loss": 0.7139,
+      "step": 20850
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3143992368789235e-06,
+      "loss": 0.7158,
+      "step": 20851
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3135431906936843e-06,
+      "loss": 0.7061,
+      "step": 20852
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.312687233119665e-06,
+      "loss": 0.6904,
+      "step": 20853
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3118313641682056e-06,
+      "loss": 0.7158,
+      "step": 20854
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.310975583850655e-06,
+      "loss": 0.7061,
+      "step": 20855
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.310119892178355e-06,
+      "loss": 0.7031,
+      "step": 20856
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3092642891626394e-06,
+      "loss": 0.7129,
+      "step": 20857
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3084087748148463e-06,
+      "loss": 0.6943,
+      "step": 20858
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3075533491463186e-06,
+      "loss": 0.6953,
+      "step": 20859
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3066980121683923e-06,
+      "loss": 0.7021,
+      "step": 20860
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3058427638923996e-06,
+      "loss": 0.6875,
+      "step": 20861
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.304987604329676e-06,
+      "loss": 0.71,
+      "step": 20862
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.304132533491554e-06,
+      "loss": 0.7002,
+      "step": 20863
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.303277551389367e-06,
+      "loss": 0.7402,
+      "step": 20864
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.302422658034443e-06,
+      "loss": 0.7002,
+      "step": 20865
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.301567853438109e-06,
+      "loss": 0.7178,
+      "step": 20866
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3007131376117e-06,
+      "loss": 0.71,
+      "step": 20867
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2998585105665393e-06,
+      "loss": 0.7207,
+      "step": 20868
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2990039723139567e-06,
+      "loss": 0.7188,
+      "step": 20869
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.298149522865266e-06,
+      "loss": 0.7227,
+      "step": 20870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2972951622318005e-06,
+      "loss": 0.7041,
+      "step": 20871
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2964408904248813e-06,
+      "loss": 0.7197,
+      "step": 20872
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2955867074558265e-06,
+      "loss": 0.7031,
+      "step": 20873
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2947326133359583e-06,
+      "loss": 0.7109,
+      "step": 20874
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.293878608076594e-06,
+      "loss": 0.6973,
+      "step": 20875
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.293024691689052e-06,
+      "loss": 0.7246,
+      "step": 20876
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2921708641846483e-06,
+      "loss": 0.7031,
+      "step": 20877
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2913171255746978e-06,
+      "loss": 0.6914,
+      "step": 20878
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.290463475870512e-06,
+      "loss": 0.7256,
+      "step": 20879
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2896099150834115e-06,
+      "loss": 0.7109,
+      "step": 20880
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.288756443224702e-06,
+      "loss": 0.6934,
+      "step": 20881
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.287903060305695e-06,
+      "loss": 0.7324,
+      "step": 20882
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2870497663376997e-06,
+      "loss": 0.7266,
+      "step": 20883
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2861965613320256e-06,
+      "loss": 0.7012,
+      "step": 20884
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.285343445299979e-06,
+      "loss": 0.7139,
+      "step": 20885
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2844904182528646e-06,
+      "loss": 0.7178,
+      "step": 20886
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2836374802019845e-06,
+      "loss": 0.7041,
+      "step": 20887
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.282784631158652e-06,
+      "loss": 0.7139,
+      "step": 20888
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.28193187113416e-06,
+      "loss": 0.748,
+      "step": 20889
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.281079200139812e-06,
+      "loss": 0.7217,
+      "step": 20890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2802266181869035e-06,
+      "loss": 0.7051,
+      "step": 20891
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2793741252867426e-06,
+      "loss": 0.7275,
+      "step": 20892
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2785217214506227e-06,
+      "loss": 0.7129,
+      "step": 20893
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2776694066898386e-06,
+      "loss": 0.7275,
+      "step": 20894
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2768171810156867e-06,
+      "loss": 0.7178,
+      "step": 20895
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.275965044439461e-06,
+      "loss": 0.7031,
+      "step": 20896
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.275112996972454e-06,
+      "loss": 0.7061,
+      "step": 20897
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2742610386259566e-06,
+      "loss": 0.6953,
+      "step": 20898
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.273409169411257e-06,
+      "loss": 0.6787,
+      "step": 20899
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.272557389339651e-06,
+      "loss": 0.7256,
+      "step": 20900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2717056984224273e-06,
+      "loss": 0.7129,
+      "step": 20901
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2708540966708647e-06,
+      "loss": 0.7002,
+      "step": 20902
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.270002584096249e-06,
+      "loss": 0.7129,
+      "step": 20903
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2691511607098723e-06,
+      "loss": 0.7188,
+      "step": 20904
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.268299826523015e-06,
+      "loss": 0.6904,
+      "step": 20905
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.267448581546957e-06,
+      "loss": 0.7461,
+      "step": 20906
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.266597425792982e-06,
+      "loss": 0.7129,
+      "step": 20907
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.265746359272368e-06,
+      "loss": 0.7012,
+      "step": 20908
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.264895381996395e-06,
+      "loss": 0.6943,
+      "step": 20909
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2640444939763392e-06,
+      "loss": 0.7061,
+      "step": 20910
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2631936952234744e-06,
+      "loss": 0.6963,
+      "step": 20911
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2623429857490817e-06,
+      "loss": 0.7148,
+      "step": 20912
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2614923655644324e-06,
+      "loss": 0.7227,
+      "step": 20913
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.260641834680801e-06,
+      "loss": 0.7031,
+      "step": 20914
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2597913931094516e-06,
+      "loss": 0.6943,
+      "step": 20915
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2589410408616617e-06,
+      "loss": 0.7129,
+      "step": 20916
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.258090777948699e-06,
+      "loss": 0.7373,
+      "step": 20917
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2572406043818317e-06,
+      "loss": 0.7178,
+      "step": 20918
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2563905201723244e-06,
+      "loss": 0.709,
+      "step": 20919
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2555405253314422e-06,
+      "loss": 0.7012,
+      "step": 20920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2546906198704575e-06,
+      "loss": 0.7197,
+      "step": 20921
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.253840803800624e-06,
+      "loss": 0.7188,
+      "step": 20922
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2529910771332075e-06,
+      "loss": 0.7334,
+      "step": 20923
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.252141439879466e-06,
+      "loss": 0.6904,
+      "step": 20924
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2512918920506643e-06,
+      "loss": 0.7021,
+      "step": 20925
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2504424336580585e-06,
+      "loss": 0.6943,
+      "step": 20926
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2495930647129057e-06,
+      "loss": 0.7051,
+      "step": 20927
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2487437852264625e-06,
+      "loss": 0.6943,
+      "step": 20928
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.247894595209984e-06,
+      "loss": 0.7012,
+      "step": 20929
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2470454946747232e-06,
+      "loss": 0.707,
+      "step": 20930
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2461964836319327e-06,
+      "loss": 0.7422,
+      "step": 20931
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.245347562092861e-06,
+      "loss": 0.6846,
+      "step": 20932
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2444987300687637e-06,
+      "loss": 0.7041,
+      "step": 20933
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2436499875708916e-06,
+      "loss": 0.7402,
+      "step": 20934
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.242801334610486e-06,
+      "loss": 0.707,
+      "step": 20935
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2419527711987918e-06,
+      "loss": 0.6963,
+      "step": 20936
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.241104297347062e-06,
+      "loss": 0.7002,
+      "step": 20937
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.240255913066538e-06,
+      "loss": 0.7207,
+      "step": 20938
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.239407618368462e-06,
+      "loss": 0.7051,
+      "step": 20939
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2385594132640763e-06,
+      "loss": 0.708,
+      "step": 20940
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.237711297764622e-06,
+      "loss": 0.7031,
+      "step": 20941
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2368632718813375e-06,
+      "loss": 0.7334,
+      "step": 20942
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.236015335625462e-06,
+      "loss": 0.7266,
+      "step": 20943
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2351674890082294e-06,
+      "loss": 0.6924,
+      "step": 20944
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2343197320408826e-06,
+      "loss": 0.7012,
+      "step": 20945
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.233472064734652e-06,
+      "loss": 0.7285,
+      "step": 20946
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.232624487100775e-06,
+      "loss": 0.7119,
+      "step": 20947
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.231776999150473e-06,
+      "loss": 0.6963,
+      "step": 20948
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.23092960089499e-06,
+      "loss": 0.7051,
+      "step": 20949
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.23008229234555e-06,
+      "loss": 0.7021,
+      "step": 20950
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2292350735133837e-06,
+      "loss": 0.7139,
+      "step": 20951
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.228387944409717e-06,
+      "loss": 0.7051,
+      "step": 20952
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2275409050457763e-06,
+      "loss": 0.7383,
+      "step": 20953
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.226693955432789e-06,
+      "loss": 0.7129,
+      "step": 20954
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2258470955819764e-06,
+      "loss": 0.6943,
+      "step": 20955
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2250003255045613e-06,
+      "loss": 0.7061,
+      "step": 20956
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2241536452117695e-06,
+      "loss": 0.6973,
+      "step": 20957
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2233070547148184e-06,
+      "loss": 0.6895,
+      "step": 20958
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2224605540249277e-06,
+      "loss": 0.7217,
+      "step": 20959
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2216141431533167e-06,
+      "loss": 0.7002,
+      "step": 20960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2207678221111994e-06,
+      "loss": 0.6865,
+      "step": 20961
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2199215909097947e-06,
+      "loss": 0.7021,
+      "step": 20962
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.219075449560316e-06,
+      "loss": 0.7012,
+      "step": 20963
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2182293980739765e-06,
+      "loss": 0.7031,
+      "step": 20964
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.217383436461984e-06,
+      "loss": 0.7207,
+      "step": 20965
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2165375647355612e-06,
+      "loss": 0.7012,
+      "step": 20966
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2156917829059077e-06,
+      "loss": 0.7344,
+      "step": 20967
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.214846090984234e-06,
+      "loss": 0.7373,
+      "step": 20968
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2140004889817444e-06,
+      "loss": 0.6963,
+      "step": 20969
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.213154976909654e-06,
+      "loss": 0.6953,
+      "step": 20970
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2123095547791626e-06,
+      "loss": 0.7354,
+      "step": 20971
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2114642226014735e-06,
+      "loss": 0.7188,
+      "step": 20972
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2106189803877908e-06,
+      "loss": 0.7158,
+      "step": 20973
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.209773828149315e-06,
+      "loss": 0.71,
+      "step": 20974
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2089287658972467e-06,
+      "loss": 0.6865,
+      "step": 20975
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2080837936427856e-06,
+      "loss": 0.6963,
+      "step": 20976
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.207238911397125e-06,
+      "loss": 0.7305,
+      "step": 20977
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2063941191714686e-06,
+      "loss": 0.7061,
+      "step": 20978
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.205549416977012e-06,
+      "loss": 0.7168,
+      "step": 20979
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.204704804824943e-06,
+      "loss": 0.7207,
+      "step": 20980
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2038602827264553e-06,
+      "loss": 0.6826,
+      "step": 20981
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2030158506927457e-06,
+      "loss": 0.7275,
+      "step": 20982
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2021715087350035e-06,
+      "loss": 0.7139,
+      "step": 20983
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.201327256864416e-06,
+      "loss": 0.7246,
+      "step": 20984
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.200483095092174e-06,
+      "loss": 0.708,
+      "step": 20985
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1996390234294627e-06,
+      "loss": 0.707,
+      "step": 20986
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1987950418874693e-06,
+      "loss": 0.7168,
+      "step": 20987
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1979511504773786e-06,
+      "loss": 0.7129,
+      "step": 20988
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.197107349210369e-06,
+      "loss": 0.7256,
+      "step": 20989
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.196263638097631e-06,
+      "loss": 0.7305,
+      "step": 20990
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1954200171503424e-06,
+      "loss": 0.7256,
+      "step": 20991
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.194576486379686e-06,
+      "loss": 0.6982,
+      "step": 20992
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.193733045796831e-06,
+      "loss": 0.7021,
+      "step": 20993
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.192889695412965e-06,
+      "loss": 0.6787,
+      "step": 20994
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.19204643523926e-06,
+      "loss": 0.7324,
+      "step": 20995
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.191203265286893e-06,
+      "loss": 0.7119,
+      "step": 20996
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1903601855670363e-06,
+      "loss": 0.6885,
+      "step": 20997
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.189517196090863e-06,
+      "loss": 0.6875,
+      "step": 20998
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.188674296869545e-06,
+      "loss": 0.7363,
+      "step": 20999
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1878314879142537e-06,
+      "loss": 0.7109,
+      "step": 21000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1869887692361536e-06,
+      "loss": 0.7334,
+      "step": 21001
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.186146140846419e-06,
+      "loss": 0.6934,
+      "step": 21002
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1853036027562145e-06,
+      "loss": 0.7129,
+      "step": 21003
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1844611549767057e-06,
+      "loss": 0.7158,
+      "step": 21004
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1836187975190556e-06,
+      "loss": 0.7012,
+      "step": 21005
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.182776530394429e-06,
+      "loss": 0.7178,
+      "step": 21006
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1819343536139866e-06,
+      "loss": 0.7246,
+      "step": 21007
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1810922671888912e-06,
+      "loss": 0.7148,
+      "step": 21008
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1802502711303008e-06,
+      "loss": 0.7109,
+      "step": 21009
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.17940836544937e-06,
+      "loss": 0.709,
+      "step": 21010
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.178566550157268e-06,
+      "loss": 0.7207,
+      "step": 21011
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1777248252651393e-06,
+      "loss": 0.6865,
+      "step": 21012
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1768831907841423e-06,
+      "loss": 0.7061,
+      "step": 21013
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1760416467254272e-06,
+      "loss": 0.7021,
+      "step": 21014
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.175200193100153e-06,
+      "loss": 0.7207,
+      "step": 21015
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1743588299194696e-06,
+      "loss": 0.707,
+      "step": 21016
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.173517557194523e-06,
+      "loss": 0.7217,
+      "step": 21017
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1726763749364663e-06,
+      "loss": 0.7285,
+      "step": 21018
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.171835283156445e-06,
+      "loss": 0.6885,
+      "step": 21019
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1709942818656047e-06,
+      "loss": 0.6914,
+      "step": 21020
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.170153371075093e-06,
+      "loss": 0.7031,
+      "step": 21021
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1693125507960486e-06,
+      "loss": 0.7061,
+      "step": 21022
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1684718210396216e-06,
+      "loss": 0.6973,
+      "step": 21023
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1676311818169534e-06,
+      "loss": 0.6875,
+      "step": 21024
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1667906331391795e-06,
+      "loss": 0.7188,
+      "step": 21025
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.165950175017437e-06,
+      "loss": 0.7158,
+      "step": 21026
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.165109807462872e-06,
+      "loss": 0.7148,
+      "step": 21027
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1642695304866168e-06,
+      "loss": 0.7148,
+      "step": 21028
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1634293440998065e-06,
+      "loss": 0.7119,
+      "step": 21029
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.162589248313578e-06,
+      "loss": 0.7373,
+      "step": 21030
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1617492431390626e-06,
+      "loss": 0.7031,
+      "step": 21031
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1609093285873937e-06,
+      "loss": 0.7256,
+      "step": 21032
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1600695046697015e-06,
+      "loss": 0.7324,
+      "step": 21033
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.159229771397112e-06,
+      "loss": 0.7139,
+      "step": 21034
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1583901287807593e-06,
+      "loss": 0.7168,
+      "step": 21035
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1575505768317693e-06,
+      "loss": 0.7012,
+      "step": 21036
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1567111155612707e-06,
+      "loss": 0.6973,
+      "step": 21037
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.155871744980379e-06,
+      "loss": 0.7246,
+      "step": 21038
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.155032465100225e-06,
+      "loss": 0.7031,
+      "step": 21039
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1541932759319316e-06,
+      "loss": 0.7188,
+      "step": 21040
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1533541774866173e-06,
+      "loss": 0.6914,
+      "step": 21041
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.152515169775403e-06,
+      "loss": 0.6992,
+      "step": 21042
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.151676252809405e-06,
+      "loss": 0.7305,
+      "step": 21043
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.150837426599749e-06,
+      "loss": 0.7256,
+      "step": 21044
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.149998691157543e-06,
+      "loss": 0.7236,
+      "step": 21045
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.149160046493902e-06,
+      "loss": 0.7285,
+      "step": 21046
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.148321492619947e-06,
+      "loss": 0.6992,
+      "step": 21047
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1474830295467853e-06,
+      "loss": 0.7168,
+      "step": 21048
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.146644657285531e-06,
+      "loss": 0.7041,
+      "step": 21049
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1458063758472934e-06,
+      "loss": 0.7129,
+      "step": 21050
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1449681852431824e-06,
+      "loss": 0.7207,
+      "step": 21051
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.144130085484305e-06,
+      "loss": 0.6963,
+      "step": 21052
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1432920765817686e-06,
+      "loss": 0.6904,
+      "step": 21053
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1424541585466784e-06,
+      "loss": 0.7256,
+      "step": 21054
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.141616331390136e-06,
+      "loss": 0.6904,
+      "step": 21055
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.140778595123252e-06,
+      "loss": 0.7021,
+      "step": 21056
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1399409497571255e-06,
+      "loss": 0.7266,
+      "step": 21057
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1391033953028527e-06,
+      "loss": 0.6973,
+      "step": 21058
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.138265931771534e-06,
+      "loss": 0.6963,
+      "step": 21059
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1374285591742725e-06,
+      "loss": 0.6982,
+      "step": 21060
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1365912775221617e-06,
+      "loss": 0.7148,
+      "step": 21061
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1357540868263003e-06,
+      "loss": 0.7061,
+      "step": 21062
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134916987097781e-06,
+      "loss": 0.7158,
+      "step": 21063
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.134079978347697e-06,
+      "loss": 0.7236,
+      "step": 21064
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1332430605871423e-06,
+      "loss": 0.7012,
+      "step": 21065
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1324062338272064e-06,
+      "loss": 0.7109,
+      "step": 21066
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131569498078977e-06,
+      "loss": 0.6973,
+      "step": 21067
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1307328533535486e-06,
+      "loss": 0.709,
+      "step": 21068
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1298962996620063e-06,
+      "loss": 0.7471,
+      "step": 21069
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1290598370154377e-06,
+      "loss": 0.7129,
+      "step": 21070
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.128223465424921e-06,
+      "loss": 0.709,
+      "step": 21071
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.127387184901548e-06,
+      "loss": 0.709,
+      "step": 21072
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1265509954563988e-06,
+      "loss": 0.7061,
+      "step": 21073
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.125714897100555e-06,
+      "loss": 0.6797,
+      "step": 21074
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1248788898450953e-06,
+      "loss": 0.7012,
+      "step": 21075
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1240429737011e-06,
+      "loss": 0.6982,
+      "step": 21076
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.123207148679648e-06,
+      "loss": 0.6885,
+      "step": 21077
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.122371414791814e-06,
+      "loss": 0.71,
+      "step": 21078
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1215357720486704e-06,
+      "loss": 0.707,
+      "step": 21079
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1207002204613e-06,
+      "loss": 0.7109,
+      "step": 21080
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1198647600407707e-06,
+      "loss": 0.7158,
+      "step": 21081
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.119029390798154e-06,
+      "loss": 0.7266,
+      "step": 21082
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1181941127445215e-06,
+      "loss": 0.7012,
+      "step": 21083
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.117358925890942e-06,
+      "loss": 0.7061,
+      "step": 21084
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1165238302484834e-06,
+      "loss": 0.7061,
+      "step": 21085
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.115688825828215e-06,
+      "loss": 0.7451,
+      "step": 21086
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.114853912641199e-06,
+      "loss": 0.7324,
+      "step": 21087
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1140190906985e-06,
+      "loss": 0.7061,
+      "step": 21088
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.113184360011189e-06,
+      "loss": 0.7393,
+      "step": 21089
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1123497205903186e-06,
+      "loss": 0.6924,
+      "step": 21090
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.111515172446954e-06,
+      "loss": 0.7119,
+      "step": 21091
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1106807155921503e-06,
+      "loss": 0.7148,
+      "step": 21092
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1098463500369735e-06,
+      "loss": 0.7178,
+      "step": 21093
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.109012075792478e-06,
+      "loss": 0.7168,
+      "step": 21094
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.108177892869718e-06,
+      "loss": 0.7217,
+      "step": 21095
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1073438012797507e-06,
+      "loss": 0.7295,
+      "step": 21096
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.106509801033629e-06,
+      "loss": 0.7207,
+      "step": 21097
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1056758921424044e-06,
+      "loss": 0.7559,
+      "step": 21098
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1048420746171294e-06,
+      "loss": 0.7139,
+      "step": 21099
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1040083484688497e-06,
+      "loss": 0.7207,
+      "step": 21100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.103174713708621e-06,
+      "loss": 0.709,
+      "step": 21101
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1023411703474914e-06,
+      "loss": 0.7451,
+      "step": 21102
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1015077183965005e-06,
+      "loss": 0.7158,
+      "step": 21103
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1006743578666933e-06,
+      "loss": 0.707,
+      "step": 21104
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.09984108876912e-06,
+      "loss": 0.6943,
+      "step": 21105
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.099007911114822e-06,
+      "loss": 0.709,
+      "step": 21106
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.098174824914838e-06,
+      "loss": 0.709,
+      "step": 21107
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.097341830180207e-06,
+      "loss": 0.7236,
+      "step": 21108
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.096508926921977e-06,
+      "loss": 0.7188,
+      "step": 21109
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0956761151511773e-06,
+      "loss": 0.6953,
+      "step": 21110
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.094843394878846e-06,
+      "loss": 0.7012,
+      "step": 21111
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.094010766116018e-06,
+      "loss": 0.6787,
+      "step": 21112
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.093178228873731e-06,
+      "loss": 0.7139,
+      "step": 21113
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0923457831630176e-06,
+      "loss": 0.7227,
+      "step": 21114
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0915134289949113e-06,
+      "loss": 0.6836,
+      "step": 21115
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0906811663804325e-06,
+      "loss": 0.7188,
+      "step": 21116
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.089848995330622e-06,
+      "loss": 0.709,
+      "step": 21117
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.089016915856504e-06,
+      "loss": 0.708,
+      "step": 21118
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0881849279691056e-06,
+      "loss": 0.707,
+      "step": 21119
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.087353031679452e-06,
+      "loss": 0.7021,
+      "step": 21120
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0865212269985656e-06,
+      "loss": 0.7139,
+      "step": 21121
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0856895139374778e-06,
+      "loss": 0.7266,
+      "step": 21122
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0848578925072038e-06,
+      "loss": 0.7129,
+      "step": 21123
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.084026362718762e-06,
+      "loss": 0.6904,
+      "step": 21124
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0831949245831794e-06,
+      "loss": 0.7217,
+      "step": 21125
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0823635781114714e-06,
+      "loss": 0.7217,
+      "step": 21126
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.081532323314654e-06,
+      "loss": 0.7295,
+      "step": 21127
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.080701160203746e-06,
+      "loss": 0.7246,
+      "step": 21128
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0798700887897602e-06,
+      "loss": 0.7129,
+      "step": 21129
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0790391090837114e-06,
+      "loss": 0.7021,
+      "step": 21130
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0782082210966114e-06,
+      "loss": 0.7236,
+      "step": 21131
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0773774248394704e-06,
+      "loss": 0.7266,
+      "step": 21132
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.076546720323298e-06,
+      "loss": 0.7295,
+      "step": 21133
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.075716107559106e-06,
+      "loss": 0.7266,
+      "step": 21134
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.074885586557904e-06,
+      "loss": 0.7295,
+      "step": 21135
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.074055157330692e-06,
+      "loss": 0.707,
+      "step": 21136
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.073224819888474e-06,
+      "loss": 0.7217,
+      "step": 21137
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.072394574242261e-06,
+      "loss": 0.7197,
+      "step": 21138
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0715644204030535e-06,
+      "loss": 0.7041,
+      "step": 21139
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0707343583818505e-06,
+      "loss": 0.6904,
+      "step": 21140
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.069904388189655e-06,
+      "loss": 0.7148,
+      "step": 21141
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.069074509837464e-06,
+      "loss": 0.7256,
+      "step": 21142
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0682447233362766e-06,
+      "loss": 0.7607,
+      "step": 21143
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.067415028697088e-06,
+      "loss": 0.6895,
+      "step": 21144
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0665854259308926e-06,
+      "loss": 0.7041,
+      "step": 21145
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0657559150486894e-06,
+      "loss": 0.707,
+      "step": 21146
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0649264960614675e-06,
+      "loss": 0.7002,
+      "step": 21147
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0640971689802233e-06,
+      "loss": 0.7217,
+      "step": 21148
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0632679338159375e-06,
+      "loss": 0.6943,
+      "step": 21149
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0624387905796094e-06,
+      "loss": 0.708,
+      "step": 21150
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.061609739282222e-06,
+      "loss": 0.7139,
+      "step": 21151
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.060780779934764e-06,
+      "loss": 0.6904,
+      "step": 21152
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.05995191254822e-06,
+      "loss": 0.6846,
+      "step": 21153
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0591231371335763e-06,
+      "loss": 0.7002,
+      "step": 21154
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0582944537018135e-06,
+      "loss": 0.6777,
+      "step": 21155
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0574658622639143e-06,
+      "loss": 0.7227,
+      "step": 21156
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0566373628308575e-06,
+      "loss": 0.6855,
+      "step": 21157
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0558089554136282e-06,
+      "loss": 0.7344,
+      "step": 21158
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054980640023203e-06,
+      "loss": 0.7178,
+      "step": 21159
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0541524166705563e-06,
+      "loss": 0.7422,
+      "step": 21160
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0533242853666657e-06,
+      "loss": 0.7158,
+      "step": 21161
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.052496246122505e-06,
+      "loss": 0.6973,
+      "step": 21162
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0516682989490487e-06,
+      "loss": 0.7227,
+      "step": 21163
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0508404438572693e-06,
+      "loss": 0.7168,
+      "step": 21164
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0500126808581366e-06,
+      "loss": 0.6963,
+      "step": 21165
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0491850099626175e-06,
+      "loss": 0.707,
+      "step": 21166
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.04835743118169e-06,
+      "loss": 0.7217,
+      "step": 21167
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.047529944526313e-06,
+      "loss": 0.6963,
+      "step": 21168
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.046702550007452e-06,
+      "loss": 0.7256,
+      "step": 21169
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0458752476360777e-06,
+      "loss": 0.7236,
+      "step": 21170
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0450480374231507e-06,
+      "loss": 0.6875,
+      "step": 21171
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0442209193796344e-06,
+      "loss": 0.7275,
+      "step": 21172
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.04339389351649e-06,
+      "loss": 0.7402,
+      "step": 21173
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.042566959844676e-06,
+      "loss": 0.7051,
+      "step": 21174
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0417401183751527e-06,
+      "loss": 0.7246,
+      "step": 21175
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0409133691188773e-06,
+      "loss": 0.6973,
+      "step": 21176
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0400867120868073e-06,
+      "loss": 0.6982,
+      "step": 21177
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0392601472898918e-06,
+      "loss": 0.7393,
+      "step": 21178
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.038433674739093e-06,
+      "loss": 0.7227,
+      "step": 21179
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.037607294445364e-06,
+      "loss": 0.7256,
+      "step": 21180
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.036781006419649e-06,
+      "loss": 0.751,
+      "step": 21181
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0359548106728986e-06,
+      "loss": 0.7002,
+      "step": 21182
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0351287072160675e-06,
+      "loss": 0.7227,
+      "step": 21183
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.034302696060102e-06,
+      "loss": 0.7129,
+      "step": 21184
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.033476777215947e-06,
+      "loss": 0.7051,
+      "step": 21185
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0326509506945458e-06,
+      "loss": 0.6865,
+      "step": 21186
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.031825216506851e-06,
+      "loss": 0.7031,
+      "step": 21187
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0309995746637967e-06,
+      "loss": 0.7129,
+      "step": 21188
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0301740251763268e-06,
+      "loss": 0.7285,
+      "step": 21189
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0293485680553802e-06,
+      "loss": 0.7168,
+      "step": 21190
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0285232033119016e-06,
+      "loss": 0.6973,
+      "step": 21191
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0276979309568268e-06,
+      "loss": 0.7246,
+      "step": 21192
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0268727510010932e-06,
+      "loss": 0.7188,
+      "step": 21193
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.02604766345563e-06,
+      "loss": 0.7275,
+      "step": 21194
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.025222668331379e-06,
+      "loss": 0.707,
+      "step": 21195
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.024397765639271e-06,
+      "loss": 0.7207,
+      "step": 21196
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.023572955390237e-06,
+      "loss": 0.7109,
+      "step": 21197
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.022748237595209e-06,
+      "loss": 0.7402,
+      "step": 21198
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.021923612265113e-06,
+      "loss": 0.6885,
+      "step": 21199
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.021099079410885e-06,
+      "loss": 0.7168,
+      "step": 21200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0202746390434445e-06,
+      "loss": 0.7246,
+      "step": 21201
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0194502911737166e-06,
+      "loss": 0.7158,
+      "step": 21202
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0186260358126327e-06,
+      "loss": 0.7295,
+      "step": 21203
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0178018729711124e-06,
+      "loss": 0.7324,
+      "step": 21204
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.016977802660077e-06,
+      "loss": 0.6992,
+      "step": 21205
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0161538248904487e-06,
+      "loss": 0.7227,
+      "step": 21206
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.015329939673146e-06,
+      "loss": 0.6807,
+      "step": 21207
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0145061470190884e-06,
+      "loss": 0.7188,
+      "step": 21208
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0136824469391924e-06,
+      "loss": 0.7217,
+      "step": 21209
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0128588394443737e-06,
+      "loss": 0.7061,
+      "step": 21210
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0120353245455446e-06,
+      "loss": 0.6953,
+      "step": 21211
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0112119022536278e-06,
+      "loss": 0.7207,
+      "step": 21212
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.010388572579527e-06,
+      "loss": 0.709,
+      "step": 21213
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0095653355341514e-06,
+      "loss": 0.7246,
+      "step": 21214
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0087421911284178e-06,
+      "loss": 0.7061,
+      "step": 21215
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.007919139373232e-06,
+      "loss": 0.6855,
+      "step": 21216
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.007096180279502e-06,
+      "loss": 0.7031,
+      "step": 21217
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0062733138581333e-06,
+      "loss": 0.7178,
+      "step": 21218
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0054505401200297e-06,
+      "loss": 0.7344,
+      "step": 21219
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.004627859076097e-06,
+      "loss": 0.7266,
+      "step": 21220
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0038052707372357e-06,
+      "loss": 0.7002,
+      "step": 21221
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.002982775114348e-06,
+      "loss": 0.7314,
+      "step": 21222
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0021603722183313e-06,
+      "loss": 0.6953,
+      "step": 21223
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0013380620600887e-06,
+      "loss": 0.7109,
+      "step": 21224
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0005158446505202e-06,
+      "loss": 0.7119,
+      "step": 21225
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9996937200005137e-06,
+      "loss": 0.7236,
+      "step": 21226
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9988716881209656e-06,
+      "loss": 0.7119,
+      "step": 21227
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9980497490227757e-06,
+      "loss": 0.6973,
+      "step": 21228
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.997227902716834e-06,
+      "loss": 0.709,
+      "step": 21229
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.996406149214031e-06,
+      "loss": 0.7148,
+      "step": 21230
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9955844885252526e-06,
+      "loss": 0.7217,
+      "step": 21231
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9947629206614005e-06,
+      "loss": 0.7383,
+      "step": 21232
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.993941445633349e-06,
+      "loss": 0.7139,
+      "step": 21233
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.993120063451992e-06,
+      "loss": 0.7119,
+      "step": 21234
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9922987741282083e-06,
+      "loss": 0.707,
+      "step": 21235
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.991477577672889e-06,
+      "loss": 0.7158,
+      "step": 21236
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9906564740969144e-06,
+      "loss": 0.7158,
+      "step": 21237
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9898354634111694e-06,
+      "loss": 0.708,
+      "step": 21238
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.989014545626523e-06,
+      "loss": 0.7363,
+      "step": 21239
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.988193720753866e-06,
+      "loss": 0.7285,
+      "step": 21240
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9873729888040713e-06,
+      "loss": 0.7217,
+      "step": 21241
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9865523497880165e-06,
+      "loss": 0.7051,
+      "step": 21242
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.985731803716577e-06,
+      "loss": 0.7031,
+      "step": 21243
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9849113506006233e-06,
+      "loss": 0.6943,
+      "step": 21244
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9840909904510383e-06,
+      "loss": 0.7109,
+      "step": 21245
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.983270723278684e-06,
+      "loss": 0.7197,
+      "step": 21246
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.982450549094431e-06,
+      "loss": 0.708,
+      "step": 21247
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9816304679091544e-06,
+      "loss": 0.71,
+      "step": 21248
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.98081047973372e-06,
+      "loss": 0.7246,
+      "step": 21249
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.979990584578993e-06,
+      "loss": 0.7295,
+      "step": 21250
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9791707824558404e-06,
+      "loss": 0.6934,
+      "step": 21251
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9783510733751263e-06,
+      "loss": 0.6875,
+      "step": 21252
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9775314573477133e-06,
+      "loss": 0.707,
+      "step": 21253
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.976711934384462e-06,
+      "loss": 0.7158,
+      "step": 21254
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.975892504496236e-06,
+      "loss": 0.709,
+      "step": 21255
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9750731676938893e-06,
+      "loss": 0.7178,
+      "step": 21256
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9742539239882874e-06,
+      "loss": 0.707,
+      "step": 21257
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9734347733902857e-06,
+      "loss": 0.6855,
+      "step": 21258
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9726157159107316e-06,
+      "loss": 0.6855,
+      "step": 21259
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9717967515604896e-06,
+      "loss": 0.7334,
+      "step": 21260
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9709778803504085e-06,
+      "loss": 0.7217,
+      "step": 21261
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.97015910229134e-06,
+      "loss": 0.7314,
+      "step": 21262
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9693404173941364e-06,
+      "loss": 0.7188,
+      "step": 21263
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.968521825669647e-06,
+      "loss": 0.6973,
+      "step": 21264
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.967703327128718e-06,
+      "loss": 0.7158,
+      "step": 21265
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9668849217821983e-06,
+      "loss": 0.7148,
+      "step": 21266
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.966066609640934e-06,
+      "loss": 0.7451,
+      "step": 21267
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.965248390715765e-06,
+      "loss": 0.6963,
+      "step": 21268
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9644302650175416e-06,
+      "loss": 0.707,
+      "step": 21269
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9636122325571027e-06,
+      "loss": 0.7227,
+      "step": 21270
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9627942933452923e-06,
+      "loss": 0.7148,
+      "step": 21271
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9619764473929403e-06,
+      "loss": 0.6924,
+      "step": 21272
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.961158694710895e-06,
+      "loss": 0.7051,
+      "step": 21273
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9603410353099905e-06,
+      "loss": 0.708,
+      "step": 21274
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9595234692010633e-06,
+      "loss": 0.6943,
+      "step": 21275
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9587059963949425e-06,
+      "loss": 0.7061,
+      "step": 21276
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9578886169024732e-06,
+      "loss": 0.7178,
+      "step": 21277
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.957071330734478e-06,
+      "loss": 0.7207,
+      "step": 21278
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9562541379017893e-06,
+      "loss": 0.7002,
+      "step": 21279
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9554370384152355e-06,
+      "loss": 0.7148,
+      "step": 21280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.954620032285651e-06,
+      "loss": 0.6934,
+      "step": 21281
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9538031195238594e-06,
+      "loss": 0.6943,
+      "step": 21282
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9529863001406865e-06,
+      "loss": 0.7031,
+      "step": 21283
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.952169574146958e-06,
+      "loss": 0.6865,
+      "step": 21284
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9513529415534967e-06,
+      "loss": 0.6992,
+      "step": 21285
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.950536402371126e-06,
+      "loss": 0.708,
+      "step": 21286
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.949719956610665e-06,
+      "loss": 0.7207,
+      "step": 21287
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9489036042829346e-06,
+      "loss": 0.7305,
+      "step": 21288
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.94808734539875e-06,
+      "loss": 0.709,
+      "step": 21289
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.947271179968939e-06,
+      "loss": 0.7256,
+      "step": 21290
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.946455108004308e-06,
+      "loss": 0.7158,
+      "step": 21291
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9456391295156705e-06,
+      "loss": 0.6846,
+      "step": 21292
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.944823244513847e-06,
+      "loss": 0.7139,
+      "step": 21293
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.944007453009647e-06,
+      "loss": 0.7285,
+      "step": 21294
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.943191755013883e-06,
+      "loss": 0.7002,
+      "step": 21295
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.942376150537363e-06,
+      "loss": 0.708,
+      "step": 21296
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9415606395908957e-06,
+      "loss": 0.7109,
+      "step": 21297
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9407452221852893e-06,
+      "loss": 0.7441,
+      "step": 21298
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9399298983313494e-06,
+      "loss": 0.7061,
+      "step": 21299
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.939114668039882e-06,
+      "loss": 0.7148,
+      "step": 21300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9382995313216866e-06,
+      "loss": 0.7168,
+      "step": 21301
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9374844881875732e-06,
+      "loss": 0.7383,
+      "step": 21302
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.936669538648341e-06,
+      "loss": 0.7275,
+      "step": 21303
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9358546827147847e-06,
+      "loss": 0.7314,
+      "step": 21304
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9350399203977042e-06,
+      "loss": 0.7275,
+      "step": 21305
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.934225251707902e-06,
+      "loss": 0.6953,
+      "step": 21306
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.933410676656172e-06,
+      "loss": 0.7217,
+      "step": 21307
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9325961952533077e-06,
+      "loss": 0.7041,
+      "step": 21308
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9317818075101012e-06,
+      "loss": 0.7031,
+      "step": 21309
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.930967513437355e-06,
+      "loss": 0.7158,
+      "step": 21310
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9301533130458493e-06,
+      "loss": 0.7178,
+      "step": 21311
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9293392063463786e-06,
+      "loss": 0.7207,
+      "step": 21312
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.928525193349727e-06,
+      "loss": 0.7031,
+      "step": 21313
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9277112740666903e-06,
+      "loss": 0.7178,
+      "step": 21314
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9268974485080503e-06,
+      "loss": 0.7441,
+      "step": 21315
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9260837166845966e-06,
+      "loss": 0.7148,
+      "step": 21316
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.925270078607102e-06,
+      "loss": 0.7178,
+      "step": 21317
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9244565342863596e-06,
+      "loss": 0.7207,
+      "step": 21318
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9236430837331476e-06,
+      "loss": 0.7168,
+      "step": 21319
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.922829726958246e-06,
+      "loss": 0.7002,
+      "step": 21320
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.92201646397243e-06,
+      "loss": 0.707,
+      "step": 21321
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.921203294786483e-06,
+      "loss": 0.7041,
+      "step": 21322
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9203902194111843e-06,
+      "loss": 0.6982,
+      "step": 21323
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9195772378572997e-06,
+      "loss": 0.7178,
+      "step": 21324
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9187643501356044e-06,
+      "loss": 0.7256,
+      "step": 21325
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.917951556256877e-06,
+      "loss": 0.6982,
+      "step": 21326
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9171388562318858e-06,
+      "loss": 0.7119,
+      "step": 21327
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.916326250071402e-06,
+      "loss": 0.709,
+      "step": 21328
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.915513737786193e-06,
+      "loss": 0.7021,
+      "step": 21329
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9147013193870268e-06,
+      "loss": 0.7031,
+      "step": 21330
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9138889948846703e-06,
+      "loss": 0.7002,
+      "step": 21331
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9130767642898895e-06,
+      "loss": 0.7217,
+      "step": 21332
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.912264627613447e-06,
+      "loss": 0.7168,
+      "step": 21333
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9114525848661026e-06,
+      "loss": 0.7061,
+      "step": 21334
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9106406360586246e-06,
+      "loss": 0.7168,
+      "step": 21335
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.909828781201773e-06,
+      "loss": 0.707,
+      "step": 21336
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.909017020306297e-06,
+      "loss": 0.6973,
+      "step": 21337
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9082053533829646e-06,
+      "loss": 0.6943,
+      "step": 21338
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.907393780442528e-06,
+      "loss": 0.7021,
+      "step": 21339
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.906582301495743e-06,
+      "loss": 0.7021,
+      "step": 21340
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9057709165533634e-06,
+      "loss": 0.7041,
+      "step": 21341
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9049596256261425e-06,
+      "loss": 0.7061,
+      "step": 21342
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9041484287248313e-06,
+      "loss": 0.6973,
+      "step": 21343
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9033373258601806e-06,
+      "loss": 0.7129,
+      "step": 21344
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9025263170429384e-06,
+      "loss": 0.7217,
+      "step": 21345
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.901715402283851e-06,
+      "loss": 0.6973,
+      "step": 21346
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9009045815936698e-06,
+      "loss": 0.6934,
+      "step": 21347
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.900093854983137e-06,
+      "loss": 0.748,
+      "step": 21348
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8992832224629995e-06,
+      "loss": 0.7061,
+      "step": 21349
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8984726840439926e-06,
+      "loss": 0.6953,
+      "step": 21350
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8976622397368646e-06,
+      "loss": 0.7061,
+      "step": 21351
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8968518895523544e-06,
+      "loss": 0.7139,
+      "step": 21352
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.896041633501201e-06,
+      "loss": 0.7354,
+      "step": 21353
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8952314715941387e-06,
+      "loss": 0.7285,
+      "step": 21354
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8944214038419126e-06,
+      "loss": 0.707,
+      "step": 21355
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.89361143025525e-06,
+      "loss": 0.7148,
+      "step": 21356
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8928015508448882e-06,
+      "loss": 0.7451,
+      "step": 21357
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8919917656215546e-06,
+      "loss": 0.7266,
+      "step": 21358
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8911820745959897e-06,
+      "loss": 0.71,
+      "step": 21359
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8903724777789187e-06,
+      "loss": 0.7119,
+      "step": 21360
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.889562975181072e-06,
+      "loss": 0.7256,
+      "step": 21361
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8887535668131763e-06,
+      "loss": 0.7158,
+      "step": 21362
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.887944252685959e-06,
+      "loss": 0.7002,
+      "step": 21363
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8871350328101454e-06,
+      "loss": 0.7051,
+      "step": 21364
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.886325907196459e-06,
+      "loss": 0.7266,
+      "step": 21365
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8855168758556228e-06,
+      "loss": 0.6875,
+      "step": 21366
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8847079387983555e-06,
+      "loss": 0.708,
+      "step": 21367
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8838990960353862e-06,
+      "loss": 0.7236,
+      "step": 21368
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.883090347577425e-06,
+      "loss": 0.6943,
+      "step": 21369
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8822816934351906e-06,
+      "loss": 0.71,
+      "step": 21370
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.881473133619405e-06,
+      "loss": 0.7178,
+      "step": 21371
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8806646681407802e-06,
+      "loss": 0.7021,
+      "step": 21372
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8798562970100296e-06,
+      "loss": 0.7246,
+      "step": 21373
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8790480202378668e-06,
+      "loss": 0.7207,
+      "step": 21374
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.878239837835004e-06,
+      "loss": 0.6934,
+      "step": 21375
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8774317498121505e-06,
+      "loss": 0.7148,
+      "step": 21376
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.876623756180016e-06,
+      "loss": 0.7305,
+      "step": 21377
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.875815856949308e-06,
+      "loss": 0.708,
+      "step": 21378
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8750080521307298e-06,
+      "loss": 0.7188,
+      "step": 21379
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.874200341734994e-06,
+      "loss": 0.7061,
+      "step": 21380
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8733927257728024e-06,
+      "loss": 0.7266,
+      "step": 21381
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.872585204254851e-06,
+      "loss": 0.6992,
+      "step": 21382
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8717777771918487e-06,
+      "loss": 0.7012,
+      "step": 21383
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.870970444594494e-06,
+      "loss": 0.7021,
+      "step": 21384
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8701632064734853e-06,
+      "loss": 0.708,
+      "step": 21385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8693560628395203e-06,
+      "loss": 0.6914,
+      "step": 21386
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8685490137032935e-06,
+      "loss": 0.7295,
+      "step": 21387
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8677420590755078e-06,
+      "loss": 0.71,
+      "step": 21388
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.866935198966849e-06,
+      "loss": 0.6895,
+      "step": 21389
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8661284333880136e-06,
+      "loss": 0.6982,
+      "step": 21390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8653217623496885e-06,
+      "loss": 0.7012,
+      "step": 21391
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8645151858625707e-06,
+      "loss": 0.6924,
+      "step": 21392
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8637087039373467e-06,
+      "loss": 0.6992,
+      "step": 21393
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.862902316584708e-06,
+      "loss": 0.7705,
+      "step": 21394
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8620960238153295e-06,
+      "loss": 0.7129,
+      "step": 21395
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.861289825639908e-06,
+      "loss": 0.7314,
+      "step": 21396
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.860483722069123e-06,
+      "loss": 0.7021,
+      "step": 21397
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8596777131136577e-06,
+      "loss": 0.6963,
+      "step": 21398
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.858871798784192e-06,
+      "loss": 0.7139,
+      "step": 21399
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.858065979091409e-06,
+      "loss": 0.6875,
+      "step": 21400
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8572602540459903e-06,
+      "loss": 0.7295,
+      "step": 21401
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.856454623658608e-06,
+      "loss": 0.7051,
+      "step": 21402
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8556490879399356e-06,
+      "loss": 0.6953,
+      "step": 21403
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.854843646900658e-06,
+      "loss": 0.7002,
+      "step": 21404
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.854038300551444e-06,
+      "loss": 0.7139,
+      "step": 21405
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.853233048902966e-06,
+      "loss": 0.71,
+      "step": 21406
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8524278919658967e-06,
+      "loss": 0.7051,
+      "step": 21407
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.851622829750906e-06,
+      "loss": 0.7227,
+      "step": 21408
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.850817862268662e-06,
+      "loss": 0.7344,
+      "step": 21409
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8500129895298333e-06,
+      "loss": 0.7109,
+      "step": 21410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.849208211545087e-06,
+      "loss": 0.7207,
+      "step": 21411
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8484035283250843e-06,
+      "loss": 0.6943,
+      "step": 21412
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8475989398804947e-06,
+      "loss": 0.6963,
+      "step": 21413
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.846794446221982e-06,
+      "loss": 0.7148,
+      "step": 21414
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8459900473601977e-06,
+      "loss": 0.707,
+      "step": 21415
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8451857433058106e-06,
+      "loss": 0.6719,
+      "step": 21416
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.844381534069478e-06,
+      "loss": 0.7178,
+      "step": 21417
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.843577419661858e-06,
+      "loss": 0.7002,
+      "step": 21418
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8427734000936045e-06,
+      "loss": 0.7246,
+      "step": 21419
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8419694753753745e-06,
+      "loss": 0.7266,
+      "step": 21420
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8411656455178215e-06,
+      "loss": 0.708,
+      "step": 21421
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8403619105315984e-06,
+      "loss": 0.71,
+      "step": 21422
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.839558270427356e-06,
+      "loss": 0.708,
+      "step": 21423
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.838754725215742e-06,
+      "loss": 0.6963,
+      "step": 21424
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.837951274907411e-06,
+      "loss": 0.7314,
+      "step": 21425
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8371479195130104e-06,
+      "loss": 0.7383,
+      "step": 21426
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8363446590431777e-06,
+      "loss": 0.7051,
+      "step": 21427
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8355414935085656e-06,
+      "loss": 0.7021,
+      "step": 21428
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8347384229198185e-06,
+      "loss": 0.6953,
+      "step": 21429
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8339354472875758e-06,
+      "loss": 0.7344,
+      "step": 21430
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8331325666224796e-06,
+      "loss": 0.6865,
+      "step": 21431
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8323297809351667e-06,
+      "loss": 0.7529,
+      "step": 21432
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8315270902362855e-06,
+      "loss": 0.6797,
+      "step": 21433
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8307244945364653e-06,
+      "loss": 0.7119,
+      "step": 21434
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.829921993846343e-06,
+      "loss": 0.7373,
+      "step": 21435
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.829119588176553e-06,
+      "loss": 0.6992,
+      "step": 21436
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8283172775377333e-06,
+      "loss": 0.6973,
+      "step": 21437
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8275150619405157e-06,
+      "loss": 0.7119,
+      "step": 21438
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.826712941395532e-06,
+      "loss": 0.7246,
+      "step": 21439
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8259109159134033e-06,
+      "loss": 0.6846,
+      "step": 21440
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8251089855047687e-06,
+      "loss": 0.7217,
+      "step": 21441
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.824307150180252e-06,
+      "loss": 0.7139,
+      "step": 21442
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.82350540995048e-06,
+      "loss": 0.7021,
+      "step": 21443
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8227037648260737e-06,
+      "loss": 0.7168,
+      "step": 21444
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.821902214817663e-06,
+      "loss": 0.708,
+      "step": 21445
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.821100759935871e-06,
+      "loss": 0.6992,
+      "step": 21446
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.820299400191312e-06,
+      "loss": 0.7236,
+      "step": 21447
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.819498135594606e-06,
+      "loss": 0.7412,
+      "step": 21448
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.818696966156378e-06,
+      "loss": 0.7031,
+      "step": 21449
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.817895891887242e-06,
+      "loss": 0.7471,
+      "step": 21450
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8170949127978155e-06,
+      "loss": 0.7119,
+      "step": 21451
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.816294028898711e-06,
+      "loss": 0.7207,
+      "step": 21452
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.815493240200544e-06,
+      "loss": 0.7051,
+      "step": 21453
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8146925467139253e-06,
+      "loss": 0.7002,
+      "step": 21454
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8138919484494677e-06,
+      "loss": 0.7002,
+      "step": 21455
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8130914454177795e-06,
+      "loss": 0.7051,
+      "step": 21456
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8122910376294676e-06,
+      "loss": 0.6865,
+      "step": 21457
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.811490725095144e-06,
+      "loss": 0.6924,
+      "step": 21458
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.810690507825414e-06,
+      "loss": 0.7148,
+      "step": 21459
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.809890385830877e-06,
+      "loss": 0.7178,
+      "step": 21460
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8090903591221407e-06,
+      "loss": 0.71,
+      "step": 21461
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8082904277098077e-06,
+      "loss": 0.667,
+      "step": 21462
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.807490591604478e-06,
+      "loss": 0.7109,
+      "step": 21463
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.806690850816752e-06,
+      "loss": 0.7178,
+      "step": 21464
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8058912053572263e-06,
+      "loss": 0.71,
+      "step": 21465
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8050916552365004e-06,
+      "loss": 0.7051,
+      "step": 21466
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8042922004651673e-06,
+      "loss": 0.7041,
+      "step": 21467
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8034928410538244e-06,
+      "loss": 0.6895,
+      "step": 21468
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8026935770130614e-06,
+      "loss": 0.6895,
+      "step": 21469
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8018944083534747e-06,
+      "loss": 0.7227,
+      "step": 21470
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8010953350856552e-06,
+      "loss": 0.7246,
+      "step": 21471
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.800296357220189e-06,
+      "loss": 0.6953,
+      "step": 21472
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7994974747676672e-06,
+      "loss": 0.7197,
+      "step": 21473
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7986986877386767e-06,
+      "loss": 0.6914,
+      "step": 21474
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7978999961438015e-06,
+      "loss": 0.7441,
+      "step": 21475
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7971013999936268e-06,
+      "loss": 0.7422,
+      "step": 21476
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.796302899298734e-06,
+      "loss": 0.7012,
+      "step": 21477
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.795504494069713e-06,
+      "loss": 0.7158,
+      "step": 21478
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.794706184317135e-06,
+      "loss": 0.7012,
+      "step": 21479
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7939079700515847e-06,
+      "loss": 0.7246,
+      "step": 21480
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.793109851283635e-06,
+      "loss": 0.6865,
+      "step": 21481
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7923118280238703e-06,
+      "loss": 0.7021,
+      "step": 21482
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7915139002828627e-06,
+      "loss": 0.6904,
+      "step": 21483
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7907160680711864e-06,
+      "loss": 0.7002,
+      "step": 21484
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.789918331399415e-06,
+      "loss": 0.6855,
+      "step": 21485
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.78912069027812e-06,
+      "loss": 0.7158,
+      "step": 21486
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.788323144717873e-06,
+      "loss": 0.7012,
+      "step": 21487
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.787525694729242e-06,
+      "loss": 0.7031,
+      "step": 21488
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.786728340322792e-06,
+      "loss": 0.7061,
+      "step": 21489
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7859310815090967e-06,
+      "loss": 0.7314,
+      "step": 21490
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7851339182987212e-06,
+      "loss": 0.6992,
+      "step": 21491
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.784336850702225e-06,
+      "loss": 0.6924,
+      "step": 21492
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7835398787301692e-06,
+      "loss": 0.7021,
+      "step": 21493
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7827430023931224e-06,
+      "loss": 0.7031,
+      "step": 21494
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7819462217016426e-06,
+      "loss": 0.7266,
+      "step": 21495
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.781149536666289e-06,
+      "loss": 0.6953,
+      "step": 21496
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7803529472976175e-06,
+      "loss": 0.7012,
+      "step": 21497
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.779556453606188e-06,
+      "loss": 0.7324,
+      "step": 21498
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.778760055602553e-06,
+      "loss": 0.7217,
+      "step": 21499
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.777963753297268e-06,
+      "loss": 0.7305,
+      "step": 21500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7771675467008864e-06,
+      "loss": 0.7139,
+      "step": 21501
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7763714358239556e-06,
+      "loss": 0.7119,
+      "step": 21502
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7755754206770334e-06,
+      "loss": 0.7217,
+      "step": 21503
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.774779501270667e-06,
+      "loss": 0.7217,
+      "step": 21504
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.773983677615396e-06,
+      "loss": 0.7012,
+      "step": 21505
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7731879497217773e-06,
+      "loss": 0.707,
+      "step": 21506
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.772392317600351e-06,
+      "loss": 0.7012,
+      "step": 21507
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.771596781261663e-06,
+      "loss": 0.7158,
+      "step": 21508
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7708013407162536e-06,
+      "loss": 0.6914,
+      "step": 21509
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770005995974664e-06,
+      "loss": 0.7041,
+      "step": 21510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7692107470474404e-06,
+      "loss": 0.7168,
+      "step": 21511
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.768415593945116e-06,
+      "loss": 0.6836,
+      "step": 21512
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7676205366782294e-06,
+      "loss": 0.7207,
+      "step": 21513
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7668255752573147e-06,
+      "loss": 0.7041,
+      "step": 21514
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7660307096929117e-06,
+      "loss": 0.6924,
+      "step": 21515
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7652359399955532e-06,
+      "loss": 0.6836,
+      "step": 21516
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7644412661757735e-06,
+      "loss": 0.7061,
+      "step": 21517
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.763646688244096e-06,
+      "loss": 0.7021,
+      "step": 21518
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7628522062110576e-06,
+      "loss": 0.71,
+      "step": 21519
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.762057820087186e-06,
+      "loss": 0.7344,
+      "step": 21520
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.761263529883008e-06,
+      "loss": 0.7129,
+      "step": 21521
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.760469335609047e-06,
+      "loss": 0.7139,
+      "step": 21522
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7596752372758328e-06,
+      "loss": 0.6973,
+      "step": 21523
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.758881234893892e-06,
+      "loss": 0.6846,
+      "step": 21524
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7580873284737376e-06,
+      "loss": 0.7383,
+      "step": 21525
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.757293518025892e-06,
+      "loss": 0.7217,
+      "step": 21526
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7564998035608815e-06,
+      "loss": 0.7246,
+      "step": 21527
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.755706185089222e-06,
+      "loss": 0.7129,
+      "step": 21528
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7549126626214296e-06,
+      "loss": 0.7334,
+      "step": 21529
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.754119236168021e-06,
+      "loss": 0.7031,
+      "step": 21530
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7533259057395112e-06,
+      "loss": 0.7158,
+      "step": 21531
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7525326713464127e-06,
+      "loss": 0.7354,
+      "step": 21532
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.751739532999239e-06,
+      "loss": 0.7051,
+      "step": 21533
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.750946490708497e-06,
+      "loss": 0.7178,
+      "step": 21534
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.750153544484704e-06,
+      "loss": 0.7285,
+      "step": 21535
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7493606943383624e-06,
+      "loss": 0.7178,
+      "step": 21536
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7485679402799846e-06,
+      "loss": 0.7188,
+      "step": 21537
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7477752823200676e-06,
+      "loss": 0.708,
+      "step": 21538
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7469827204691247e-06,
+      "loss": 0.6777,
+      "step": 21539
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7461902547376552e-06,
+      "loss": 0.7148,
+      "step": 21540
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.745397885136161e-06,
+      "loss": 0.6943,
+      "step": 21541
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.744605611675144e-06,
+      "loss": 0.6904,
+      "step": 21542
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7438134343651048e-06,
+      "loss": 0.7324,
+      "step": 21543
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7430213532165386e-06,
+      "loss": 0.7158,
+      "step": 21544
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7422293682399437e-06,
+      "loss": 0.7334,
+      "step": 21545
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7414374794458166e-06,
+      "loss": 0.7148,
+      "step": 21546
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7406456868446473e-06,
+      "loss": 0.6963,
+      "step": 21547
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.739853990446937e-06,
+      "loss": 0.6934,
+      "step": 21548
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7390623902631718e-06,
+      "loss": 0.7148,
+      "step": 21549
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.738270886303843e-06,
+      "loss": 0.7109,
+      "step": 21550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7374794785794425e-06,
+      "loss": 0.7031,
+      "step": 21551
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7366881671004542e-06,
+      "loss": 0.708,
+      "step": 21552
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.735896951877368e-06,
+      "loss": 0.707,
+      "step": 21553
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.735105832920668e-06,
+      "loss": 0.7539,
+      "step": 21554
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7343148102408356e-06,
+      "loss": 0.708,
+      "step": 21555
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7335238838483637e-06,
+      "loss": 0.6904,
+      "step": 21556
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.732733053753722e-06,
+      "loss": 0.6875,
+      "step": 21557
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.731942319967398e-06,
+      "loss": 0.7129,
+      "step": 21558
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7311516824998643e-06,
+      "loss": 0.7188,
+      "step": 21559
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.730361141361606e-06,
+      "loss": 0.6973,
+      "step": 21560
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.729570696563096e-06,
+      "loss": 0.6904,
+      "step": 21561
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.728780348114811e-06,
+      "loss": 0.7207,
+      "step": 21562
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.727990096027224e-06,
+      "loss": 0.6904,
+      "step": 21563
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7271999403108087e-06,
+      "loss": 0.7422,
+      "step": 21564
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.726409880976034e-06,
+      "loss": 0.6895,
+      "step": 21565
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.725619918033373e-06,
+      "loss": 0.7197,
+      "step": 21566
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.724830051493289e-06,
+      "loss": 0.7158,
+      "step": 21567
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.724040281366258e-06,
+      "loss": 0.6875,
+      "step": 21568
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.723250607662744e-06,
+      "loss": 0.7246,
+      "step": 21569
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7224610303932084e-06,
+      "loss": 0.7139,
+      "step": 21570
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7216715495681124e-06,
+      "loss": 0.6973,
+      "step": 21571
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7208821651979267e-06,
+      "loss": 0.6963,
+      "step": 21572
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.720092877293108e-06,
+      "loss": 0.6963,
+      "step": 21573
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7193036858641177e-06,
+      "loss": 0.6973,
+      "step": 21574
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7185145909214127e-06,
+      "loss": 0.6914,
+      "step": 21575
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.717725592475451e-06,
+      "loss": 0.7061,
+      "step": 21576
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.716936690536689e-06,
+      "loss": 0.6846,
+      "step": 21577
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.716147885115582e-06,
+      "loss": 0.7041,
+      "step": 21578
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7153591762225816e-06,
+      "loss": 0.7402,
+      "step": 21579
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7145705638681396e-06,
+      "loss": 0.71,
+      "step": 21580
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7137820480627108e-06,
+      "loss": 0.6982,
+      "step": 21581
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.712993628816746e-06,
+      "loss": 0.7305,
+      "step": 21582
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7122053061406837e-06,
+      "loss": 0.708,
+      "step": 21583
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7114170800449814e-06,
+      "loss": 0.7324,
+      "step": 21584
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.710628950540082e-06,
+      "loss": 0.7158,
+      "step": 21585
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7098409176364283e-06,
+      "loss": 0.6846,
+      "step": 21586
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.709052981344464e-06,
+      "loss": 0.6973,
+      "step": 21587
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7082651416746296e-06,
+      "loss": 0.7256,
+      "step": 21588
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7074773986373735e-06,
+      "loss": 0.6865,
+      "step": 21589
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7066897522431268e-06,
+      "loss": 0.7188,
+      "step": 21590
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7059022025023294e-06,
+      "loss": 0.6982,
+      "step": 21591
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.705114749425417e-06,
+      "loss": 0.7314,
+      "step": 21592
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7043273930228308e-06,
+      "loss": 0.7344,
+      "step": 21593
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7035401333050016e-06,
+      "loss": 0.7139,
+      "step": 21594
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.702752970282362e-06,
+      "loss": 0.7217,
+      "step": 21595
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7019659039653433e-06,
+      "loss": 0.7188,
+      "step": 21596
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.701178934364379e-06,
+      "loss": 0.6992,
+      "step": 21597
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.700392061489895e-06,
+      "loss": 0.7031,
+      "step": 21598
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6996052853523204e-06,
+      "loss": 0.6836,
+      "step": 21599
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.698818605962079e-06,
+      "loss": 0.71,
+      "step": 21600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.698032023329602e-06,
+      "loss": 0.7119,
+      "step": 21601
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.697245537465314e-06,
+      "loss": 0.6924,
+      "step": 21602
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.696459148379631e-06,
+      "loss": 0.6914,
+      "step": 21603
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6956728560829746e-06,
+      "loss": 0.6846,
+      "step": 21604
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6948866605857704e-06,
+      "loss": 0.7129,
+      "step": 21605
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.694100561898436e-06,
+      "loss": 0.7139,
+      "step": 21606
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.693314560031388e-06,
+      "loss": 0.7139,
+      "step": 21607
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6925286549950435e-06,
+      "loss": 0.707,
+      "step": 21608
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.691742846799816e-06,
+      "loss": 0.7168,
+      "step": 21609
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6909571354561215e-06,
+      "loss": 0.6982,
+      "step": 21610
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6901715209743696e-06,
+      "loss": 0.7324,
+      "step": 21611
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6893860033649723e-06,
+      "loss": 0.7109,
+      "step": 21612
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.688600582638342e-06,
+      "loss": 0.707,
+      "step": 21613
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6878152588048854e-06,
+      "loss": 0.7275,
+      "step": 21614
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.687030031875015e-06,
+      "loss": 0.7285,
+      "step": 21615
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.686244901859125e-06,
+      "loss": 0.6914,
+      "step": 21616
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6854598687676313e-06,
+      "loss": 0.6982,
+      "step": 21617
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6846749326109335e-06,
+      "loss": 0.6992,
+      "step": 21618
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6838900933994326e-06,
+      "loss": 0.7393,
+      "step": 21619
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6831053511435325e-06,
+      "loss": 0.6836,
+      "step": 21620
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6823207058536307e-06,
+      "loss": 0.7158,
+      "step": 21621
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6815361575401256e-06,
+      "loss": 0.7129,
+      "step": 21622
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.680751706213416e-06,
+      "loss": 0.7061,
+      "step": 21623
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6799673518838954e-06,
+      "loss": 0.6992,
+      "step": 21624
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.679183094561957e-06,
+      "loss": 0.7051,
+      "step": 21625
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6783989342579997e-06,
+      "loss": 0.7275,
+      "step": 21626
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6776148709824125e-06,
+      "loss": 0.7305,
+      "step": 21627
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6768309047455853e-06,
+      "loss": 0.7148,
+      "step": 21628
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6760470355579083e-06,
+      "loss": 0.7051,
+      "step": 21629
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.675263263429769e-06,
+      "loss": 0.7148,
+      "step": 21630
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6744795883715557e-06,
+      "loss": 0.6953,
+      "step": 21631
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6736960103936514e-06,
+      "loss": 0.6885,
+      "step": 21632
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6729125295064396e-06,
+      "loss": 0.6865,
+      "step": 21633
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6721291457203113e-06,
+      "loss": 0.6973,
+      "step": 21634
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.67134585904564e-06,
+      "loss": 0.7305,
+      "step": 21635
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.670562669492808e-06,
+      "loss": 0.6738,
+      "step": 21636
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6697795770721903e-06,
+      "loss": 0.7197,
+      "step": 21637
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6689965817941734e-06,
+      "loss": 0.707,
+      "step": 21638
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6682136836691297e-06,
+      "loss": 0.6865,
+      "step": 21639
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.667430882707435e-06,
+      "loss": 0.7207,
+      "step": 21640
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.666648178919461e-06,
+      "loss": 0.7012,
+      "step": 21641
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6658655723155823e-06,
+      "loss": 0.7002,
+      "step": 21642
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.665083062906171e-06,
+      "loss": 0.7041,
+      "step": 21643
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6643006507015958e-06,
+      "loss": 0.707,
+      "step": 21644
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.663518335712222e-06,
+      "loss": 0.7197,
+      "step": 21645
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6627361179484245e-06,
+      "loss": 0.6875,
+      "step": 21646
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6619539974205688e-06,
+      "loss": 0.6992,
+      "step": 21647
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.661171974139014e-06,
+      "loss": 0.6953,
+      "step": 21648
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.660390048114123e-06,
+      "loss": 0.6982,
+      "step": 21649
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6596082193562665e-06,
+      "loss": 0.7246,
+      "step": 21650
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6588264878757998e-06,
+      "loss": 0.6904,
+      "step": 21651
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.658044853683084e-06,
+      "loss": 0.7432,
+      "step": 21652
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6572633167884766e-06,
+      "loss": 0.6982,
+      "step": 21653
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.656481877202336e-06,
+      "loss": 0.7227,
+      "step": 21654
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6557005349350184e-06,
+      "loss": 0.7266,
+      "step": 21655
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6549192899968766e-06,
+      "loss": 0.7324,
+      "step": 21656
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6541381423982637e-06,
+      "loss": 0.7266,
+      "step": 21657
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6533570921495344e-06,
+      "loss": 0.7148,
+      "step": 21658
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6525761392610384e-06,
+      "loss": 0.7139,
+      "step": 21659
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6517952837431284e-06,
+      "loss": 0.7236,
+      "step": 21660
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6510145256061425e-06,
+      "loss": 0.6943,
+      "step": 21661
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6502338648604377e-06,
+      "loss": 0.7061,
+      "step": 21662
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6494533015163548e-06,
+      "loss": 0.7207,
+      "step": 21663
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6486728355842405e-06,
+      "loss": 0.7158,
+      "step": 21664
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6478924670744364e-06,
+      "loss": 0.7109,
+      "step": 21665
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6471121959972834e-06,
+      "loss": 0.7168,
+      "step": 21666
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6463320223631237e-06,
+      "loss": 0.7002,
+      "step": 21667
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6455519461822956e-06,
+      "loss": 0.7168,
+      "step": 21668
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6447719674651373e-06,
+      "loss": 0.7188,
+      "step": 21669
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6439920862219814e-06,
+      "loss": 0.7324,
+      "step": 21670
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6432123024631694e-06,
+      "loss": 0.7031,
+      "step": 21671
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6424326161990333e-06,
+      "loss": 0.7109,
+      "step": 21672
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6416530274399053e-06,
+      "loss": 0.6973,
+      "step": 21673
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6408735361961158e-06,
+      "loss": 0.7324,
+      "step": 21674
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.640094142477997e-06,
+      "loss": 0.6934,
+      "step": 21675
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.639314846295875e-06,
+      "loss": 0.7363,
+      "step": 21676
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.638535647660079e-06,
+      "loss": 0.7178,
+      "step": 21677
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6377565465809317e-06,
+      "loss": 0.7295,
+      "step": 21678
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6369775430687684e-06,
+      "loss": 0.6816,
+      "step": 21679
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.636198637133901e-06,
+      "loss": 0.7021,
+      "step": 21680
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6354198287866574e-06,
+      "loss": 0.7031,
+      "step": 21681
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6346411180373533e-06,
+      "loss": 0.6875,
+      "step": 21682
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.633862504896316e-06,
+      "loss": 0.7197,
+      "step": 21683
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6330839893738614e-06,
+      "loss": 0.6807,
+      "step": 21684
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.632305571480306e-06,
+      "loss": 0.7051,
+      "step": 21685
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6315272512259647e-06,
+      "loss": 0.7188,
+      "step": 21686
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6307490286211547e-06,
+      "loss": 0.7363,
+      "step": 21687
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6299709036761854e-06,
+      "loss": 0.7402,
+      "step": 21688
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.629192876401372e-06,
+      "loss": 0.708,
+      "step": 21689
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6284149468070207e-06,
+      "loss": 0.7168,
+      "step": 21690
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6276371149034473e-06,
+      "loss": 0.6787,
+      "step": 21691
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6268593807009602e-06,
+      "loss": 0.6963,
+      "step": 21692
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6260817442098596e-06,
+      "loss": 0.7168,
+      "step": 21693
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6253042054404508e-06,
+      "loss": 0.7139,
+      "step": 21694
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6245267644030447e-06,
+      "loss": 0.7168,
+      "step": 21695
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6237494211079416e-06,
+      "loss": 0.7207,
+      "step": 21696
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6229721755654424e-06,
+      "loss": 0.7256,
+      "step": 21697
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6221950277858466e-06,
+      "loss": 0.7168,
+      "step": 21698
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6214179777794544e-06,
+      "loss": 0.6904,
+      "step": 21699
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6206410255565638e-06,
+      "loss": 0.708,
+      "step": 21700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6198641711274707e-06,
+      "loss": 0.7109,
+      "step": 21701
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.619087414502468e-06,
+      "loss": 0.708,
+      "step": 21702
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.618310755691854e-06,
+      "loss": 0.709,
+      "step": 21703
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6175341947059186e-06,
+      "loss": 0.7178,
+      "step": 21704
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6167577315549575e-06,
+      "loss": 0.7207,
+      "step": 21705
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6159813662492504e-06,
+      "loss": 0.7217,
+      "step": 21706
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.615205098799095e-06,
+      "loss": 0.7275,
+      "step": 21707
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6144289292147763e-06,
+      "loss": 0.6973,
+      "step": 21708
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.613652857506579e-06,
+      "loss": 0.7041,
+      "step": 21709
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6128768836847894e-06,
+      "loss": 0.71,
+      "step": 21710
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6121010077596876e-06,
+      "loss": 0.6943,
+      "step": 21711
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.611325229741565e-06,
+      "loss": 0.6973,
+      "step": 21712
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.610549549640692e-06,
+      "loss": 0.7207,
+      "step": 21713
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6097739674673526e-06,
+      "loss": 0.7139,
+      "step": 21714
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6089984832318215e-06,
+      "loss": 0.6982,
+      "step": 21715
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6082230969443835e-06,
+      "loss": 0.6914,
+      "step": 21716
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6074478086153076e-06,
+      "loss": 0.7021,
+      "step": 21717
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6066726182548717e-06,
+      "loss": 0.71,
+      "step": 21718
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6058975258733466e-06,
+      "loss": 0.7012,
+      "step": 21719
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6051225314810045e-06,
+      "loss": 0.7236,
+      "step": 21720
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.604347635088118e-06,
+      "loss": 0.7188,
+      "step": 21721
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.603572836704954e-06,
+      "loss": 0.7266,
+      "step": 21722
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6027981363417777e-06,
+      "loss": 0.7139,
+      "step": 21723
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.602023534008862e-06,
+      "loss": 0.7012,
+      "step": 21724
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6012490297164718e-06,
+      "loss": 0.7031,
+      "step": 21725
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6004746234748656e-06,
+      "loss": 0.6992,
+      "step": 21726
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.599700315294307e-06,
+      "loss": 0.6982,
+      "step": 21727
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5989261051850622e-06,
+      "loss": 0.7061,
+      "step": 21728
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5981519931573874e-06,
+      "loss": 0.7051,
+      "step": 21729
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5973779792215438e-06,
+      "loss": 0.7021,
+      "step": 21730
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5966040633877864e-06,
+      "loss": 0.6768,
+      "step": 21731
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.595830245666374e-06,
+      "loss": 0.7002,
+      "step": 21732
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5950565260675598e-06,
+      "loss": 0.7324,
+      "step": 21733
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5942829046015972e-06,
+      "loss": 0.7354,
+      "step": 21734
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.593509381278736e-06,
+      "loss": 0.7109,
+      "step": 21735
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5927359561092335e-06,
+      "loss": 0.6797,
+      "step": 21736
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5919626291033353e-06,
+      "loss": 0.7158,
+      "step": 21737
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5911894002712944e-06,
+      "loss": 0.7012,
+      "step": 21738
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5904162696233483e-06,
+      "loss": 0.7041,
+      "step": 21739
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.589643237169751e-06,
+      "loss": 0.6875,
+      "step": 21740
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5888703029207442e-06,
+      "loss": 0.7139,
+      "step": 21741
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.588097466886572e-06,
+      "loss": 0.7021,
+      "step": 21742
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.587324729077476e-06,
+      "loss": 0.7295,
+      "step": 21743
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5865520895036954e-06,
+      "loss": 0.71,
+      "step": 21744
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.585779548175472e-06,
+      "loss": 0.6943,
+      "step": 21745
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5850071051030425e-06,
+      "loss": 0.7031,
+      "step": 21746
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.584234760296639e-06,
+      "loss": 0.7031,
+      "step": 21747
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.583462513766505e-06,
+      "loss": 0.6689,
+      "step": 21748
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5826903655228717e-06,
+      "loss": 0.7148,
+      "step": 21749
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5819183155759695e-06,
+      "loss": 0.7012,
+      "step": 21750
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5811463639360335e-06,
+      "loss": 0.7021,
+      "step": 21751
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5803745106132904e-06,
+      "loss": 0.7012,
+      "step": 21752
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5796027556179713e-06,
+      "loss": 0.7002,
+      "step": 21753
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5788310989603026e-06,
+      "loss": 0.7158,
+      "step": 21754
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5780595406505104e-06,
+      "loss": 0.7051,
+      "step": 21755
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5772880806988177e-06,
+      "loss": 0.6758,
+      "step": 21756
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5765167191154574e-06,
+      "loss": 0.7217,
+      "step": 21757
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.575745455910641e-06,
+      "loss": 0.7031,
+      "step": 21758
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.574974291094594e-06,
+      "loss": 0.7227,
+      "step": 21759
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.574203224677533e-06,
+      "loss": 0.7002,
+      "step": 21760
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.573432256669681e-06,
+      "loss": 0.7383,
+      "step": 21761
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5726613870812543e-06,
+      "loss": 0.6738,
+      "step": 21762
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5718906159224687e-06,
+      "loss": 0.7178,
+      "step": 21763
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.571119943203536e-06,
+      "loss": 0.6846,
+      "step": 21764
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5703493689346726e-06,
+      "loss": 0.7227,
+      "step": 21765
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.569578893126089e-06,
+      "loss": 0.7246,
+      "step": 21766
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.568808515787995e-06,
+      "loss": 0.7109,
+      "step": 21767
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.568038236930598e-06,
+      "loss": 0.7305,
+      "step": 21768
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.567268056564112e-06,
+      "loss": 0.7363,
+      "step": 21769
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.566497974698743e-06,
+      "loss": 0.7148,
+      "step": 21770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.565727991344691e-06,
+      "loss": 0.7246,
+      "step": 21771
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5649581065121596e-06,
+      "loss": 0.6934,
+      "step": 21772
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5641883202113592e-06,
+      "loss": 0.7295,
+      "step": 21773
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5634186324524856e-06,
+      "loss": 0.7383,
+      "step": 21774
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5626490432457407e-06,
+      "loss": 0.7012,
+      "step": 21775
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.561879552601323e-06,
+      "loss": 0.6787,
+      "step": 21776
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5611101605294307e-06,
+      "loss": 0.7012,
+      "step": 21777
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5603408670402586e-06,
+      "loss": 0.7139,
+      "step": 21778
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.559571672144002e-06,
+      "loss": 0.708,
+      "step": 21779
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5588025758508527e-06,
+      "loss": 0.7217,
+      "step": 21780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5580335781710087e-06,
+      "loss": 0.7139,
+      "step": 21781
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5572646791146573e-06,
+      "loss": 0.709,
+      "step": 21782
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5564958786919913e-06,
+      "loss": 0.7139,
+      "step": 21783
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.55572717691319e-06,
+      "loss": 0.6816,
+      "step": 21784
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5549585737884497e-06,
+      "loss": 0.7324,
+      "step": 21785
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.554190069327954e-06,
+      "loss": 0.7031,
+      "step": 21786
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5534216635418863e-06,
+      "loss": 0.6973,
+      "step": 21787
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.552653356440431e-06,
+      "loss": 0.6943,
+      "step": 21788
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5518851480337657e-06,
+      "loss": 0.6924,
+      "step": 21789
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5511170383320803e-06,
+      "loss": 0.7266,
+      "step": 21790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5503490273455446e-06,
+      "loss": 0.6953,
+      "step": 21791
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5495811150843407e-06,
+      "loss": 0.7051,
+      "step": 21792
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5488133015586414e-06,
+      "loss": 0.7266,
+      "step": 21793
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.548045586778629e-06,
+      "loss": 0.7246,
+      "step": 21794
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5472779707544735e-06,
+      "loss": 0.6924,
+      "step": 21795
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.546510453496348e-06,
+      "loss": 0.7188,
+      "step": 21796
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.545743035014423e-06,
+      "loss": 0.7012,
+      "step": 21797
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5449757153188705e-06,
+      "loss": 0.6855,
+      "step": 21798
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5442084944198575e-06,
+      "loss": 0.7529,
+      "step": 21799
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.543441372327552e-06,
+      "loss": 0.7256,
+      "step": 21800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5426743490521178e-06,
+      "loss": 0.6973,
+      "step": 21801
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5419074246037244e-06,
+      "loss": 0.7168,
+      "step": 21802
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5411405989925373e-06,
+      "loss": 0.6729,
+      "step": 21803
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.540373872228712e-06,
+      "loss": 0.7002,
+      "step": 21804
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5396072443224085e-06,
+      "loss": 0.6914,
+      "step": 21805
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.538840715283792e-06,
+      "loss": 0.708,
+      "step": 21806
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.53807428512302e-06,
+      "loss": 0.7354,
+      "step": 21807
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.537307953850249e-06,
+      "loss": 0.7021,
+      "step": 21808
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5365417214756325e-06,
+      "loss": 0.7168,
+      "step": 21809
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.535775588009327e-06,
+      "loss": 0.6992,
+      "step": 21810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.535009553461485e-06,
+      "loss": 0.6895,
+      "step": 21811
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.534243617842258e-06,
+      "loss": 0.6885,
+      "step": 21812
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.533477781161794e-06,
+      "loss": 0.7109,
+      "step": 21813
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5327120434302467e-06,
+      "loss": 0.7109,
+      "step": 21814
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5319464046577635e-06,
+      "loss": 0.7012,
+      "step": 21815
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.531180864854491e-06,
+      "loss": 0.7236,
+      "step": 21816
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5304154240305677e-06,
+      "loss": 0.7402,
+      "step": 21817
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.529650082196146e-06,
+      "loss": 0.7227,
+      "step": 21818
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.528884839361364e-06,
+      "loss": 0.6875,
+      "step": 21819
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.528119695536364e-06,
+      "loss": 0.7061,
+      "step": 21820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.527354650731286e-06,
+      "loss": 0.7197,
+      "step": 21821
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5265897049562693e-06,
+      "loss": 0.7031,
+      "step": 21822
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5258248582214493e-06,
+      "loss": 0.71,
+      "step": 21823
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5250601105369644e-06,
+      "loss": 0.7207,
+      "step": 21824
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5242954619129445e-06,
+      "loss": 0.7129,
+      "step": 21825
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5235309123595307e-06,
+      "loss": 0.707,
+      "step": 21826
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.52276646188685e-06,
+      "loss": 0.7109,
+      "step": 21827
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.522002110505034e-06,
+      "loss": 0.6904,
+      "step": 21828
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5212378582242126e-06,
+      "loss": 0.708,
+      "step": 21829
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5204737050545134e-06,
+      "loss": 0.7139,
+      "step": 21830
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5197096510060637e-06,
+      "loss": 0.707,
+      "step": 21831
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.518945696088989e-06,
+      "loss": 0.7119,
+      "step": 21832
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.518181840313413e-06,
+      "loss": 0.7246,
+      "step": 21833
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.517418083689456e-06,
+      "loss": 0.7207,
+      "step": 21834
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5166544262272475e-06,
+      "loss": 0.7246,
+      "step": 21835
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5158908679368997e-06,
+      "loss": 0.6943,
+      "step": 21836
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.515127408828536e-06,
+      "loss": 0.6973,
+      "step": 21837
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5143640489122677e-06,
+      "loss": 0.7119,
+      "step": 21838
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5136007881982204e-06,
+      "loss": 0.6855,
+      "step": 21839
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.512837626696504e-06,
+      "loss": 0.7402,
+      "step": 21840
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.512074564417233e-06,
+      "loss": 0.7051,
+      "step": 21841
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5113116013705185e-06,
+      "loss": 0.6973,
+      "step": 21842
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.510548737566474e-06,
+      "loss": 0.7021,
+      "step": 21843
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5097859730152063e-06,
+      "loss": 0.7207,
+      "step": 21844
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5090233077268267e-06,
+      "loss": 0.7441,
+      "step": 21845
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.508260741711437e-06,
+      "loss": 0.6963,
+      "step": 21846
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.507498274979151e-06,
+      "loss": 0.7402,
+      "step": 21847
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.506735907540071e-06,
+      "loss": 0.6992,
+      "step": 21848
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5059736394042956e-06,
+      "loss": 0.7051,
+      "step": 21849
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5052114705819265e-06,
+      "loss": 0.7012,
+      "step": 21850
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.50444940108307e-06,
+      "loss": 0.7002,
+      "step": 21851
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.503687430917822e-06,
+      "loss": 0.7275,
+      "step": 21852
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.502925560096281e-06,
+      "loss": 0.6924,
+      "step": 21853
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.50216378862854e-06,
+      "loss": 0.7168,
+      "step": 21854
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.501402116524704e-06,
+      "loss": 0.708,
+      "step": 21855
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5006405437948566e-06,
+      "loss": 0.7061,
+      "step": 21856
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.499879070449095e-06,
+      "loss": 0.6953,
+      "step": 21857
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4991176964975062e-06,
+      "loss": 0.7178,
+      "step": 21858
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.498356421950188e-06,
+      "loss": 0.6787,
+      "step": 21859
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4975952468172237e-06,
+      "loss": 0.707,
+      "step": 21860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.496834171108705e-06,
+      "loss": 0.7168,
+      "step": 21861
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4960731948347095e-06,
+      "loss": 0.7168,
+      "step": 21862
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4953123180053296e-06,
+      "loss": 0.7158,
+      "step": 21863
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4945515406306463e-06,
+      "loss": 0.7021,
+      "step": 21864
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.493790862720742e-06,
+      "loss": 0.6904,
+      "step": 21865
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.493030284285697e-06,
+      "loss": 0.6992,
+      "step": 21866
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.492269805335591e-06,
+      "loss": 0.7031,
+      "step": 21867
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4915094258805005e-06,
+      "loss": 0.708,
+      "step": 21868
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4907491459305055e-06,
+      "loss": 0.7295,
+      "step": 21869
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4899889654956765e-06,
+      "loss": 0.6738,
+      "step": 21870
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4892288845860934e-06,
+      "loss": 0.6895,
+      "step": 21871
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.488468903211827e-06,
+      "loss": 0.7207,
+      "step": 21872
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.487709021382949e-06,
+      "loss": 0.7158,
+      "step": 21873
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4869492391095284e-06,
+      "loss": 0.6924,
+      "step": 21874
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.486189556401635e-06,
+      "loss": 0.7324,
+      "step": 21875
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4854299732693367e-06,
+      "loss": 0.7021,
+      "step": 21876
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.484670489722698e-06,
+      "loss": 0.7334,
+      "step": 21877
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.483911105771786e-06,
+      "loss": 0.7383,
+      "step": 21878
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4831518214266592e-06,
+      "loss": 0.7305,
+      "step": 21879
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4823926366973916e-06,
+      "loss": 0.7158,
+      "step": 21880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4816335515940326e-06,
+      "loss": 0.7256,
+      "step": 21881
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4808745661266464e-06,
+      "loss": 0.7002,
+      "step": 21882
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.480115680305287e-06,
+      "loss": 0.7207,
+      "step": 21883
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4793568941400193e-06,
+      "loss": 0.7119,
+      "step": 21884
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.478598207640894e-06,
+      "loss": 0.7285,
+      "step": 21885
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4778396208179665e-06,
+      "loss": 0.7246,
+      "step": 21886
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4770811336812907e-06,
+      "loss": 0.7002,
+      "step": 21887
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4763227462409165e-06,
+      "loss": 0.7061,
+      "step": 21888
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.475564458506896e-06,
+      "loss": 0.7051,
+      "step": 21889
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.474806270489276e-06,
+      "loss": 0.7246,
+      "step": 21890
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4740481821981045e-06,
+      "loss": 0.6943,
+      "step": 21891
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4732901936434316e-06,
+      "loss": 0.6992,
+      "step": 21892
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4725323048353023e-06,
+      "loss": 0.6885,
+      "step": 21893
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.471774515783756e-06,
+      "loss": 0.7227,
+      "step": 21894
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4710168264988334e-06,
+      "loss": 0.7168,
+      "step": 21895
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4702592369905832e-06,
+      "loss": 0.6943,
+      "step": 21896
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4695017472690417e-06,
+      "loss": 0.6855,
+      "step": 21897
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.468744357344247e-06,
+      "loss": 0.7051,
+      "step": 21898
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4679870672262375e-06,
+      "loss": 0.6992,
+      "step": 21899
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.467229876925047e-06,
+      "loss": 0.6943,
+      "step": 21900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.466472786450712e-06,
+      "loss": 0.7314,
+      "step": 21901
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4657157958132636e-06,
+      "loss": 0.708,
+      "step": 21902
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.464958905022734e-06,
+      "loss": 0.6963,
+      "step": 21903
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.464202114089157e-06,
+      "loss": 0.707,
+      "step": 21904
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.463445423022559e-06,
+      "loss": 0.7041,
+      "step": 21905
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.462688831832972e-06,
+      "loss": 0.7314,
+      "step": 21906
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4619323405304142e-06,
+      "loss": 0.7158,
+      "step": 21907
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4611759491249176e-06,
+      "loss": 0.6943,
+      "step": 21908
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4604196576265048e-06,
+      "loss": 0.7324,
+      "step": 21909
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.459663466045199e-06,
+      "loss": 0.7158,
+      "step": 21910
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4589073743910198e-06,
+      "loss": 0.709,
+      "step": 21911
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4581513826739846e-06,
+      "loss": 0.7148,
+      "step": 21912
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.457395490904122e-06,
+      "loss": 0.7188,
+      "step": 21913
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4566396990914398e-06,
+      "loss": 0.7246,
+      "step": 21914
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.455884007245953e-06,
+      "loss": 0.7002,
+      "step": 21915
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4551284153776842e-06,
+      "loss": 0.7188,
+      "step": 21916
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.454372923496643e-06,
+      "loss": 0.7109,
+      "step": 21917
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.453617531612841e-06,
+      "loss": 0.7021,
+      "step": 21918
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.452862239736289e-06,
+      "loss": 0.7217,
+      "step": 21919
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.452107047876997e-06,
+      "loss": 0.7285,
+      "step": 21920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.451351956044973e-06,
+      "loss": 0.6924,
+      "step": 21921
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4505969642502227e-06,
+      "loss": 0.708,
+      "step": 21922
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4498420725027517e-06,
+      "loss": 0.7344,
+      "step": 21923
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4490872808125634e-06,
+      "loss": 0.7383,
+      "step": 21924
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4483325891896624e-06,
+      "loss": 0.7158,
+      "step": 21925
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.447577997644054e-06,
+      "loss": 0.6992,
+      "step": 21926
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4468235061857306e-06,
+      "loss": 0.7012,
+      "step": 21927
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4460691148246905e-06,
+      "loss": 0.6846,
+      "step": 21928
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.445314823570939e-06,
+      "loss": 0.6953,
+      "step": 21929
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.444560632434467e-06,
+      "loss": 0.7275,
+      "step": 21930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.44380654142527e-06,
+      "loss": 0.7041,
+      "step": 21931
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4430525505533433e-06,
+      "loss": 0.7256,
+      "step": 21932
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.442298659828676e-06,
+      "loss": 0.7041,
+      "step": 21933
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.441544869261261e-06,
+      "loss": 0.7227,
+      "step": 21934
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4407911788610873e-06,
+      "loss": 0.748,
+      "step": 21935
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.44003758863814e-06,
+      "loss": 0.71,
+      "step": 21936
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4392840986024126e-06,
+      "loss": 0.7021,
+      "step": 21937
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4385307087638865e-06,
+      "loss": 0.709,
+      "step": 21938
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.437777419132549e-06,
+      "loss": 0.7227,
+      "step": 21939
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4370242297183744e-06,
+      "loss": 0.7002,
+      "step": 21940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4362711405313533e-06,
+      "loss": 0.7119,
+      "step": 21941
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.435518151581463e-06,
+      "loss": 0.7002,
+      "step": 21942
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4347652628786823e-06,
+      "loss": 0.7207,
+      "step": 21943
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.434012474432987e-06,
+      "loss": 0.71,
+      "step": 21944
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.433259786254356e-06,
+      "loss": 0.7432,
+      "step": 21945
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.432507198352764e-06,
+      "loss": 0.7158,
+      "step": 21946
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4317547107381824e-06,
+      "loss": 0.71,
+      "step": 21947
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4310023234205817e-06,
+      "loss": 0.6895,
+      "step": 21948
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.430250036409938e-06,
+      "loss": 0.7012,
+      "step": 21949
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4294978497162192e-06,
+      "loss": 0.7266,
+      "step": 21950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.428745763349393e-06,
+      "loss": 0.6904,
+      "step": 21951
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4279937773194264e-06,
+      "loss": 0.7197,
+      "step": 21952
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.427241891636284e-06,
+      "loss": 0.7139,
+      "step": 21953
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.426490106309931e-06,
+      "loss": 0.7275,
+      "step": 21954
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.425738421350331e-06,
+      "loss": 0.7188,
+      "step": 21955
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4249868367674446e-06,
+      "loss": 0.7275,
+      "step": 21956
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.424235352571228e-06,
+      "loss": 0.7031,
+      "step": 21957
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4234839687716517e-06,
+      "loss": 0.7207,
+      "step": 21958
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4227326853786627e-06,
+      "loss": 0.7324,
+      "step": 21959
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4219815024022164e-06,
+      "loss": 0.71,
+      "step": 21960
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.421230419852276e-06,
+      "loss": 0.709,
+      "step": 21961
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.420479437738791e-06,
+      "loss": 0.7109,
+      "step": 21962
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.419728556071712e-06,
+      "loss": 0.7041,
+      "step": 21963
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4189777748609934e-06,
+      "loss": 0.7246,
+      "step": 21964
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.418227094116582e-06,
+      "loss": 0.7256,
+      "step": 21965
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4174765138484267e-06,
+      "loss": 0.7158,
+      "step": 21966
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4167260340664754e-06,
+      "loss": 0.6992,
+      "step": 21967
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4159756547806724e-06,
+      "loss": 0.6982,
+      "step": 21968
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4152253760009604e-06,
+      "loss": 0.7246,
+      "step": 21969
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.414475197737287e-06,
+      "loss": 0.7031,
+      "step": 21970
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4137251199995946e-06,
+      "loss": 0.7129,
+      "step": 21971
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4129751427978176e-06,
+      "loss": 0.6914,
+      "step": 21972
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4122252661418953e-06,
+      "loss": 0.7188,
+      "step": 21973
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.41147549004177e-06,
+      "loss": 0.7129,
+      "step": 21974
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.410725814507375e-06,
+      "loss": 0.7305,
+      "step": 21975
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.409976239548647e-06,
+      "loss": 0.7109,
+      "step": 21976
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.409226765175515e-06,
+      "loss": 0.71,
+      "step": 21977
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.408477391397921e-06,
+      "loss": 0.7266,
+      "step": 21978
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.407728118225787e-06,
+      "loss": 0.7129,
+      "step": 21979
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4069789456690452e-06,
+      "loss": 0.7275,
+      "step": 21980
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4062298737376223e-06,
+      "loss": 0.7217,
+      "step": 21981
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4054809024414482e-06,
+      "loss": 0.7061,
+      "step": 21982
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.404732031790449e-06,
+      "loss": 0.6865,
+      "step": 21983
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.403983261794549e-06,
+      "loss": 0.7119,
+      "step": 21984
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.403234592463665e-06,
+      "loss": 0.7266,
+      "step": 21985
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.402486023807725e-06,
+      "loss": 0.7148,
+      "step": 21986
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4017375558366485e-06,
+      "loss": 0.7031,
+      "step": 21987
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.400989188560352e-06,
+      "loss": 0.6973,
+      "step": 21988
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4002409219887555e-06,
+      "loss": 0.7256,
+      "step": 21989
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3994927561317717e-06,
+      "loss": 0.6982,
+      "step": 21990
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.398744690999324e-06,
+      "loss": 0.7168,
+      "step": 21991
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.397996726601317e-06,
+      "loss": 0.7197,
+      "step": 21992
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.397248862947663e-06,
+      "loss": 0.707,
+      "step": 21993
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3965011000482796e-06,
+      "loss": 0.7041,
+      "step": 21994
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.395753437913072e-06,
+      "loss": 0.7197,
+      "step": 21995
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.39500587655195e-06,
+      "loss": 0.6992,
+      "step": 21996
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3942584159748207e-06,
+      "loss": 0.7021,
+      "step": 21997
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3935110561915886e-06,
+      "loss": 0.7129,
+      "step": 21998
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3927637972121577e-06,
+      "loss": 0.7188,
+      "step": 21999
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.392016639046432e-06,
+      "loss": 0.7246,
+      "step": 22000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.391269581704313e-06,
+      "loss": 0.7051,
+      "step": 22001
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.390522625195697e-06,
+      "loss": 0.7236,
+      "step": 22002
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.389775769530489e-06,
+      "loss": 0.7012,
+      "step": 22003
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.389029014718588e-06,
+      "loss": 0.6973,
+      "step": 22004
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3882823607698835e-06,
+      "loss": 0.7178,
+      "step": 22005
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38753580769427e-06,
+      "loss": 0.7129,
+      "step": 22006
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3867893555016474e-06,
+      "loss": 0.6904,
+      "step": 22007
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.386043004201906e-06,
+      "loss": 0.6846,
+      "step": 22008
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3852967538049344e-06,
+      "loss": 0.7168,
+      "step": 22009
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3845506043206235e-06,
+      "loss": 0.6865,
+      "step": 22010
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3838045557588607e-06,
+      "loss": 0.7305,
+      "step": 22011
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.383058608129535e-06,
+      "loss": 0.7109,
+      "step": 22012
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38231276144253e-06,
+      "loss": 0.7041,
+      "step": 22013
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3815670157077275e-06,
+      "loss": 0.6875,
+      "step": 22014
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.380821370935016e-06,
+      "loss": 0.7031,
+      "step": 22015
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.380075827134275e-06,
+      "loss": 0.708,
+      "step": 22016
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.379330384315387e-06,
+      "loss": 0.6992,
+      "step": 22017
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.378585042488222e-06,
+      "loss": 0.7002,
+      "step": 22018
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.377839801662666e-06,
+      "loss": 0.7197,
+      "step": 22019
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3770946618485924e-06,
+      "loss": 0.7148,
+      "step": 22020
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3763496230558768e-06,
+      "loss": 0.7588,
+      "step": 22021
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3756046852943902e-06,
+      "loss": 0.6904,
+      "step": 22022
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3748598485740117e-06,
+      "loss": 0.7119,
+      "step": 22023
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3741151129046045e-06,
+      "loss": 0.7129,
+      "step": 22024
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.373370478296041e-06,
+      "loss": 0.7119,
+      "step": 22025
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3726259447581856e-06,
+      "loss": 0.7129,
+      "step": 22026
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3718815123009122e-06,
+      "loss": 0.71,
+      "step": 22027
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3711371809340823e-06,
+      "loss": 0.7363,
+      "step": 22028
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3703929506675617e-06,
+      "loss": 0.6875,
+      "step": 22029
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3696488215112103e-06,
+      "loss": 0.7256,
+      "step": 22030
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.368904793474892e-06,
+      "loss": 0.6904,
+      "step": 22031
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3681608665684664e-06,
+      "loss": 0.7109,
+      "step": 22032
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.367417040801793e-06,
+      "loss": 0.7129,
+      "step": 22033
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3666733161847268e-06,
+      "loss": 0.7275,
+      "step": 22034
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3659296927271225e-06,
+      "loss": 0.7021,
+      "step": 22035
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3651861704388434e-06,
+      "loss": 0.6924,
+      "step": 22036
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3644427493297352e-06,
+      "loss": 0.7178,
+      "step": 22037
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3636994294096496e-06,
+      "loss": 0.6992,
+      "step": 22038
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3629562106884407e-06,
+      "loss": 0.7197,
+      "step": 22039
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.362213093175959e-06,
+      "loss": 0.7031,
+      "step": 22040
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3614700768820485e-06,
+      "loss": 0.7021,
+      "step": 22041
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3607271618165595e-06,
+      "loss": 0.7051,
+      "step": 22042
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.359984347989335e-06,
+      "loss": 0.7041,
+      "step": 22043
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.35924163541022e-06,
+      "loss": 0.7168,
+      "step": 22044
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3584990240890573e-06,
+      "loss": 0.7021,
+      "step": 22045
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3577565140356873e-06,
+      "loss": 0.7002,
+      "step": 22046
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3570141052599483e-06,
+      "loss": 0.7012,
+      "step": 22047
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3562717977716833e-06,
+      "loss": 0.7021,
+      "step": 22048
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.355529591580731e-06,
+      "loss": 0.6982,
+      "step": 22049
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.354787486696921e-06,
+      "loss": 0.7148,
+      "step": 22050
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3540454831300875e-06,
+      "loss": 0.7031,
+      "step": 22051
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3533035808900705e-06,
+      "loss": 0.7031,
+      "step": 22052
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3525617799866986e-06,
+      "loss": 0.7178,
+      "step": 22053
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3518200804298032e-06,
+      "loss": 0.7109,
+      "step": 22054
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3510784822292087e-06,
+      "loss": 0.7432,
+      "step": 22055
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3503369853947533e-06,
+      "loss": 0.708,
+      "step": 22056
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3495955899362542e-06,
+      "loss": 0.7158,
+      "step": 22057
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3488542958635395e-06,
+      "loss": 0.7266,
+      "step": 22058
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3481131031864313e-06,
+      "loss": 0.6992,
+      "step": 22059
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3473720119147557e-06,
+      "loss": 0.7402,
+      "step": 22060
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3466310220583345e-06,
+      "loss": 0.7148,
+      "step": 22061
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.345890133626987e-06,
+      "loss": 0.7012,
+      "step": 22062
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.345149346630524e-06,
+      "loss": 0.6895,
+      "step": 22063
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.344408661078772e-06,
+      "loss": 0.6953,
+      "step": 22064
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3436680769815434e-06,
+      "loss": 0.7295,
+      "step": 22065
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.342927594348654e-06,
+      "loss": 0.7139,
+      "step": 22066
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3421872131899127e-06,
+      "loss": 0.7061,
+      "step": 22067
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.341446933515136e-06,
+      "loss": 0.7314,
+      "step": 22068
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3407067553341366e-06,
+      "loss": 0.7344,
+      "step": 22069
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3399666786567166e-06,
+      "loss": 0.7031,
+      "step": 22070
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3392267034926843e-06,
+      "loss": 0.7256,
+      "step": 22071
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3384868298518514e-06,
+      "loss": 0.7246,
+      "step": 22072
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.337747057744021e-06,
+      "loss": 0.7158,
+      "step": 22073
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3370073871789946e-06,
+      "loss": 0.7295,
+      "step": 22074
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.336267818166578e-06,
+      "loss": 0.709,
+      "step": 22075
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3355283507165683e-06,
+      "loss": 0.6885,
+      "step": 22076
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3347889848387684e-06,
+      "loss": 0.707,
+      "step": 22077
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3340497205429748e-06,
+      "loss": 0.7432,
+      "step": 22078
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3333105578389847e-06,
+      "loss": 0.7002,
+      "step": 22079
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.332571496736592e-06,
+      "loss": 0.75,
+      "step": 22080
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.331832537245595e-06,
+      "loss": 0.6982,
+      "step": 22081
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3310936793757888e-06,
+      "loss": 0.6953,
+      "step": 22082
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.330354923136955e-06,
+      "loss": 0.6914,
+      "step": 22083
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3296162685388924e-06,
+      "loss": 0.7158,
+      "step": 22084
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3288777155913865e-06,
+      "loss": 0.7295,
+      "step": 22085
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3281392643042266e-06,
+      "loss": 0.7119,
+      "step": 22086
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.327400914687198e-06,
+      "loss": 0.6953,
+      "step": 22087
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3266626667500858e-06,
+      "loss": 0.6836,
+      "step": 22088
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3259245205026726e-06,
+      "loss": 0.7031,
+      "step": 22089
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3251864759547403e-06,
+      "loss": 0.7129,
+      "step": 22090
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3244485331160726e-06,
+      "loss": 0.707,
+      "step": 22091
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3237106919964426e-06,
+      "loss": 0.6924,
+      "step": 22092
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3229729526056368e-06,
+      "loss": 0.7227,
+      "step": 22093
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.322235314953429e-06,
+      "loss": 0.7119,
+      "step": 22094
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3214977790495927e-06,
+      "loss": 0.708,
+      "step": 22095
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3207603449038994e-06,
+      "loss": 0.7031,
+      "step": 22096
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.320023012526128e-06,
+      "loss": 0.7207,
+      "step": 22097
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.319285781926047e-06,
+      "loss": 0.71,
+      "step": 22098
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.318548653113427e-06,
+      "loss": 0.7021,
+      "step": 22099
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3178116260980344e-06,
+      "loss": 0.7002,
+      "step": 22100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.317074700889642e-06,
+      "loss": 0.7002,
+      "step": 22101
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3163378774980116e-06,
+      "loss": 0.7227,
+      "step": 22102
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3156011559329072e-06,
+      "loss": 0.71,
+      "step": 22103
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.314864536204091e-06,
+      "loss": 0.6992,
+      "step": 22104
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3141280183213298e-06,
+      "loss": 0.7031,
+      "step": 22105
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3133916022943816e-06,
+      "loss": 0.7178,
+      "step": 22106
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.312655288133009e-06,
+      "loss": 0.7363,
+      "step": 22107
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.311919075846961e-06,
+      "loss": 0.7383,
+      "step": 22108
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3111829654460017e-06,
+      "loss": 0.7314,
+      "step": 22109
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3104469569398856e-06,
+      "loss": 0.709,
+      "step": 22110
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.309711050338365e-06,
+      "loss": 0.7314,
+      "step": 22111
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3089752456511914e-06,
+      "loss": 0.6924,
+      "step": 22112
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3082395428881156e-06,
+      "loss": 0.7021,
+      "step": 22113
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.307503942058895e-06,
+      "loss": 0.7314,
+      "step": 22114
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3067684431732683e-06,
+      "loss": 0.7051,
+      "step": 22115
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.306033046240984e-06,
+      "loss": 0.6963,
+      "step": 22116
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3052977512717923e-06,
+      "loss": 0.7061,
+      "step": 22117
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.304562558275436e-06,
+      "loss": 0.7148,
+      "step": 22118
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.303827467261658e-06,
+      "loss": 0.6885,
+      "step": 22119
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3030924782401987e-06,
+      "loss": 0.7129,
+      "step": 22120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3023575912207997e-06,
+      "loss": 0.6816,
+      "step": 22121
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3016228062131982e-06,
+      "loss": 0.7471,
+      "step": 22122
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3008881232271354e-06,
+      "loss": 0.7051,
+      "step": 22123
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3001535422723442e-06,
+      "loss": 0.7207,
+      "step": 22124
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2994190633585577e-06,
+      "loss": 0.7031,
+      "step": 22125
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.298684686495515e-06,
+      "loss": 0.7178,
+      "step": 22126
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2979504116929497e-06,
+      "loss": 0.6973,
+      "step": 22127
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.297216238960581e-06,
+      "loss": 0.6953,
+      "step": 22128
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2964821683081506e-06,
+      "loss": 0.71,
+      "step": 22129
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.295748199745382e-06,
+      "loss": 0.7061,
+      "step": 22130
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2950143332820017e-06,
+      "loss": 0.6875,
+      "step": 22131
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2942805689277357e-06,
+      "loss": 0.6895,
+      "step": 22132
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2935469066923087e-06,
+      "loss": 0.6924,
+      "step": 22133
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.292813346585442e-06,
+      "loss": 0.708,
+      "step": 22134
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2920798886168585e-06,
+      "loss": 0.7188,
+      "step": 22135
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.291346532796276e-06,
+      "loss": 0.7061,
+      "step": 22136
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2906132791334133e-06,
+      "loss": 0.6924,
+      "step": 22137
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.289880127637991e-06,
+      "loss": 0.7236,
+      "step": 22138
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2891470783197233e-06,
+      "loss": 0.7002,
+      "step": 22139
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2884141311883277e-06,
+      "loss": 0.7061,
+      "step": 22140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.287681286253509e-06,
+      "loss": 0.7012,
+      "step": 22141
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2869485435249873e-06,
+      "loss": 0.7207,
+      "step": 22142
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.286215903012471e-06,
+      "loss": 0.7051,
+      "step": 22143
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.285483364725667e-06,
+      "loss": 0.7139,
+      "step": 22144
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.284750928674283e-06,
+      "loss": 0.7275,
+      "step": 22145
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.284018594868034e-06,
+      "loss": 0.6865,
+      "step": 22146
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2832863633166146e-06,
+      "loss": 0.7148,
+      "step": 22147
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2825542340297323e-06,
+      "loss": 0.7344,
+      "step": 22148
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.281822207017087e-06,
+      "loss": 0.6924,
+      "step": 22149
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.281090282288386e-06,
+      "loss": 0.6904,
+      "step": 22150
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2803584598533255e-06,
+      "loss": 0.7393,
+      "step": 22151
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2796267397216033e-06,
+      "loss": 0.7139,
+      "step": 22152
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2788951219029175e-06,
+      "loss": 0.7227,
+      "step": 22153
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2781636064069624e-06,
+      "loss": 0.6924,
+      "step": 22154
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.277432193243434e-06,
+      "loss": 0.6934,
+      "step": 22155
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2767008824220238e-06,
+      "loss": 0.7188,
+      "step": 22156
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.275969673952424e-06,
+      "loss": 0.7158,
+      "step": 22157
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.275238567844322e-06,
+      "loss": 0.7109,
+      "step": 22158
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2745075641074155e-06,
+      "loss": 0.6865,
+      "step": 22159
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2737766627513826e-06,
+      "loss": 0.6992,
+      "step": 22160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.27304586378591e-06,
+      "loss": 0.7051,
+      "step": 22161
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.272315167220688e-06,
+      "loss": 0.7393,
+      "step": 22162
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2715845730653984e-06,
+      "loss": 0.7402,
+      "step": 22163
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.270854081329721e-06,
+      "loss": 0.6963,
+      "step": 22164
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2701236920233384e-06,
+      "loss": 0.7139,
+      "step": 22165
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2693934051559285e-06,
+      "loss": 0.71,
+      "step": 22166
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.268663220737172e-06,
+      "loss": 0.7178,
+      "step": 22167
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2679331387767424e-06,
+      "loss": 0.7148,
+      "step": 22168
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2672031592843157e-06,
+      "loss": 0.6973,
+      "step": 22169
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2664732822695645e-06,
+      "loss": 0.6943,
+      "step": 22170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2657435077421653e-06,
+      "loss": 0.6973,
+      "step": 22171
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2650138357117912e-06,
+      "loss": 0.709,
+      "step": 22172
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2642842661881027e-06,
+      "loss": 0.6973,
+      "step": 22173
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.263554799180775e-06,
+      "loss": 0.7061,
+      "step": 22174
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.262825434699476e-06,
+      "loss": 0.7119,
+      "step": 22175
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2620961727538682e-06,
+      "loss": 0.7139,
+      "step": 22176
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.261367013353618e-06,
+      "loss": 0.7246,
+      "step": 22177
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2606379565083846e-06,
+      "loss": 0.7119,
+      "step": 22178
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.259909002227839e-06,
+      "loss": 0.7354,
+      "step": 22179
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2591801505216326e-06,
+      "loss": 0.6895,
+      "step": 22180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2584514013994275e-06,
+      "loss": 0.7148,
+      "step": 22181
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.257722754870877e-06,
+      "loss": 0.7178,
+      "step": 22182
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2569942109456455e-06,
+      "loss": 0.7129,
+      "step": 22183
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2562657696333847e-06,
+      "loss": 0.7148,
+      "step": 22184
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2555374309437483e-06,
+      "loss": 0.6982,
+      "step": 22185
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.254809194886383e-06,
+      "loss": 0.7188,
+      "step": 22186
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2540810614709473e-06,
+      "loss": 0.7354,
+      "step": 22187
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2533530307070874e-06,
+      "loss": 0.6787,
+      "step": 22188
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.252625102604451e-06,
+      "loss": 0.6973,
+      "step": 22189
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2518972771726822e-06,
+      "loss": 0.6973,
+      "step": 22190
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.251169554421434e-06,
+      "loss": 0.7061,
+      "step": 22191
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2504419343603477e-06,
+      "loss": 0.709,
+      "step": 22192
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2497144169990614e-06,
+      "loss": 0.7178,
+      "step": 22193
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.248987002347217e-06,
+      "loss": 0.708,
+      "step": 22194
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2482596904144607e-06,
+      "loss": 0.6934,
+      "step": 22195
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.247532481210426e-06,
+      "loss": 0.7236,
+      "step": 22196
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2468053747447516e-06,
+      "loss": 0.7188,
+      "step": 22197
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.246078371027073e-06,
+      "loss": 0.7197,
+      "step": 22198
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.245351470067024e-06,
+      "loss": 0.7188,
+      "step": 22199
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.244624671874239e-06,
+      "loss": 0.6963,
+      "step": 22200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2438979764583504e-06,
+      "loss": 0.7266,
+      "step": 22201
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.243171383828986e-06,
+      "loss": 0.7129,
+      "step": 22202
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2424448939957745e-06,
+      "loss": 0.7344,
+      "step": 22203
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2417185069683476e-06,
+      "loss": 0.7002,
+      "step": 22204
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.240992222756333e-06,
+      "loss": 0.7217,
+      "step": 22205
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2402660413693457e-06,
+      "loss": 0.7148,
+      "step": 22206
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.239539962817019e-06,
+      "loss": 0.7031,
+      "step": 22207
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.238813987108972e-06,
+      "loss": 0.7168,
+      "step": 22208
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.238088114254825e-06,
+      "loss": 0.707,
+      "step": 22209
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.237362344264199e-06,
+      "loss": 0.7031,
+      "step": 22210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2366366771467106e-06,
+      "loss": 0.7305,
+      "step": 22211
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2359111129119772e-06,
+      "loss": 0.7012,
+      "step": 22212
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2351856515696135e-06,
+      "loss": 0.7012,
+      "step": 22213
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2344602931292346e-06,
+      "loss": 0.708,
+      "step": 22214
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.233735037600451e-06,
+      "loss": 0.7139,
+      "step": 22215
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.233009884992877e-06,
+      "loss": 0.7148,
+      "step": 22216
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2322848353161232e-06,
+      "loss": 0.7275,
+      "step": 22217
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.231559888579797e-06,
+      "loss": 0.7139,
+      "step": 22218
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2308350447935013e-06,
+      "loss": 0.6982,
+      "step": 22219
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.230110303966848e-06,
+      "loss": 0.7139,
+      "step": 22220
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2293856661094406e-06,
+      "loss": 0.7031,
+      "step": 22221
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2286611312308792e-06,
+      "loss": 0.71,
+      "step": 22222
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2279366993407657e-06,
+      "loss": 0.7139,
+      "step": 22223
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.227212370448707e-06,
+      "loss": 0.6982,
+      "step": 22224
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.226488144564295e-06,
+      "loss": 0.7256,
+      "step": 22225
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.22576402169713e-06,
+      "loss": 0.7217,
+      "step": 22226
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.225040001856805e-06,
+      "loss": 0.6963,
+      "step": 22227
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.224316085052921e-06,
+      "loss": 0.6875,
+      "step": 22228
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.223592271295069e-06,
+      "loss": 0.7041,
+      "step": 22229
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2228685605928403e-06,
+      "loss": 0.708,
+      "step": 22230
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.222144952955827e-06,
+      "loss": 0.6895,
+      "step": 22231
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.221421448393617e-06,
+      "loss": 0.6865,
+      "step": 22232
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2206980469158002e-06,
+      "loss": 0.7344,
+      "step": 22233
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2199747485319623e-06,
+      "loss": 0.71,
+      "step": 22234
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2192515532516857e-06,
+      "loss": 0.707,
+      "step": 22235
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2185284610845604e-06,
+      "loss": 0.7139,
+      "step": 22236
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.21780547204017e-06,
+      "loss": 0.7207,
+      "step": 22237
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.217082586128089e-06,
+      "loss": 0.7041,
+      "step": 22238
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.216359803357897e-06,
+      "loss": 0.7002,
+      "step": 22239
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2156371237391795e-06,
+      "loss": 0.71,
+      "step": 22240
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2149145472815102e-06,
+      "loss": 0.7012,
+      "step": 22241
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2141920739944657e-06,
+      "loss": 0.6738,
+      "step": 22242
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2134697038876185e-06,
+      "loss": 0.7129,
+      "step": 22243
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2127474369705428e-06,
+      "loss": 0.6953,
+      "step": 22244
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2120252732528112e-06,
+      "loss": 0.707,
+      "step": 22245
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2113032127439936e-06,
+      "loss": 0.6982,
+      "step": 22246
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2105812554536577e-06,
+      "loss": 0.7129,
+      "step": 22247
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.20985940139137e-06,
+      "loss": 0.6943,
+      "step": 22248
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2091376505667016e-06,
+      "loss": 0.7275,
+      "step": 22249
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2084160029892177e-06,
+      "loss": 0.7246,
+      "step": 22250
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.207694458668472e-06,
+      "loss": 0.7236,
+      "step": 22251
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.206973017614038e-06,
+      "loss": 0.7051,
+      "step": 22252
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2062516798354706e-06,
+      "loss": 0.6963,
+      "step": 22253
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.205530445342331e-06,
+      "loss": 0.6973,
+      "step": 22254
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2048093141441775e-06,
+      "loss": 0.7324,
+      "step": 22255
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.204088286250562e-06,
+      "loss": 0.7109,
+      "step": 22256
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2033673616710504e-06,
+      "loss": 0.7207,
+      "step": 22257
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2026465404151865e-06,
+      "loss": 0.7168,
+      "step": 22258
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2019258224925275e-06,
+      "loss": 0.7061,
+      "step": 22259
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.20120520791262e-06,
+      "loss": 0.7285,
+      "step": 22260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.20048469668502e-06,
+      "loss": 0.707,
+      "step": 22261
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.199764288819275e-06,
+      "loss": 0.7217,
+      "step": 22262
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1990439843249323e-06,
+      "loss": 0.7109,
+      "step": 22263
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.19832378321153e-06,
+      "loss": 0.6963,
+      "step": 22264
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1976036854886208e-06,
+      "loss": 0.7041,
+      "step": 22265
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.196883691165745e-06,
+      "loss": 0.6885,
+      "step": 22266
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1961638002524453e-06,
+      "loss": 0.7109,
+      "step": 22267
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1954440127582577e-06,
+      "loss": 0.6904,
+      "step": 22268
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1947243286927278e-06,
+      "loss": 0.707,
+      "step": 22269
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1940047480653913e-06,
+      "loss": 0.7305,
+      "step": 22270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.193285270885781e-06,
+      "loss": 0.7285,
+      "step": 22271
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.19256589716343e-06,
+      "loss": 0.7246,
+      "step": 22272
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1918466269078777e-06,
+      "loss": 0.7246,
+      "step": 22273
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1911274601286535e-06,
+      "loss": 0.7197,
+      "step": 22274
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1904083968352886e-06,
+      "loss": 0.7012,
+      "step": 22275
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.189689437037311e-06,
+      "loss": 0.7217,
+      "step": 22276
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1889705807442494e-06,
+      "loss": 0.7002,
+      "step": 22277
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.188251827965631e-06,
+      "loss": 0.7246,
+      "step": 22278
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.187533178710979e-06,
+      "loss": 0.7314,
+      "step": 22279
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1868146329898153e-06,
+      "loss": 0.7119,
+      "step": 22280
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1860961908116694e-06,
+      "loss": 0.71,
+      "step": 22281
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.185377852186058e-06,
+      "loss": 0.7197,
+      "step": 22282
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1846596171225034e-06,
+      "loss": 0.7295,
+      "step": 22283
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1839414856305165e-06,
+      "loss": 0.7256,
+      "step": 22284
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1832234577196222e-06,
+      "loss": 0.7197,
+      "step": 22285
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1825055333993328e-06,
+      "loss": 0.7197,
+      "step": 22286
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.181787712679162e-06,
+      "loss": 0.6924,
+      "step": 22287
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1810699955686245e-06,
+      "loss": 0.71,
+      "step": 22288
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.180352382077231e-06,
+      "loss": 0.7188,
+      "step": 22289
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.17963487221449e-06,
+      "loss": 0.7285,
+      "step": 22290
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1789174659899127e-06,
+      "loss": 0.7021,
+      "step": 22291
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1782001634130044e-06,
+      "loss": 0.6992,
+      "step": 22292
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.177482964493269e-06,
+      "loss": 0.7334,
+      "step": 22293
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.176765869240217e-06,
+      "loss": 0.7061,
+      "step": 22294
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.176048877663348e-06,
+      "loss": 0.6963,
+      "step": 22295
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1753319897721646e-06,
+      "loss": 0.7324,
+      "step": 22296
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.174615205576167e-06,
+      "loss": 0.7041,
+      "step": 22297
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.173898525084854e-06,
+      "loss": 0.7148,
+      "step": 22298
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1731819483077232e-06,
+      "loss": 0.6934,
+      "step": 22299
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.172465475254272e-06,
+      "loss": 0.7305,
+      "step": 22300
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1717491059339902e-06,
+      "loss": 0.71,
+      "step": 22301
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.171032840356383e-06,
+      "loss": 0.6992,
+      "step": 22302
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1703166785309316e-06,
+      "loss": 0.7188,
+      "step": 22303
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1696006204671307e-06,
+      "loss": 0.6934,
+      "step": 22304
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1688846661744665e-06,
+      "loss": 0.7275,
+      "step": 22305
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1681688156624326e-06,
+      "loss": 0.7021,
+      "step": 22306
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1674530689405136e-06,
+      "loss": 0.6797,
+      "step": 22307
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1667374260181985e-06,
+      "loss": 0.708,
+      "step": 22308
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.16602188690496e-06,
+      "loss": 0.7129,
+      "step": 22309
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1653064516102907e-06,
+      "loss": 0.7256,
+      "step": 22310
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.16459112014367e-06,
+      "loss": 0.7285,
+      "step": 22311
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1638758925145763e-06,
+      "loss": 0.7383,
+      "step": 22312
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.163160768732485e-06,
+      "loss": 0.7129,
+      "step": 22313
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.162445748806881e-06,
+      "loss": 0.7041,
+      "step": 22314
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1617308327472364e-06,
+      "loss": 0.7139,
+      "step": 22315
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1610160205630226e-06,
+      "loss": 0.7139,
+      "step": 22316
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.160301312263712e-06,
+      "loss": 0.709,
+      "step": 22317
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.159586707858782e-06,
+      "loss": 0.6924,
+      "step": 22318
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1588722073576993e-06,
+      "loss": 0.6963,
+      "step": 22319
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.158157810769933e-06,
+      "loss": 0.71,
+      "step": 22320
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1574435181049503e-06,
+      "loss": 0.7197,
+      "step": 22321
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1567293293722168e-06,
+      "loss": 0.6846,
+      "step": 22322
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.156015244581199e-06,
+      "loss": 0.6895,
+      "step": 22323
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1553012637413583e-06,
+      "loss": 0.7148,
+      "step": 22324
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1545873868621583e-06,
+      "loss": 0.7041,
+      "step": 22325
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.153873613953055e-06,
+      "loss": 0.7266,
+      "step": 22326
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.153159945023514e-06,
+      "loss": 0.7441,
+      "step": 22327
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.152446380082993e-06,
+      "loss": 0.6914,
+      "step": 22328
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1517329191409397e-06,
+      "loss": 0.6885,
+      "step": 22329
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1510195622068186e-06,
+      "loss": 0.7041,
+      "step": 22330
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.15030630929008e-06,
+      "loss": 0.7295,
+      "step": 22331
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.149593160400175e-06,
+      "loss": 0.6963,
+      "step": 22332
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1488801155465567e-06,
+      "loss": 0.7061,
+      "step": 22333
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.148167174738672e-06,
+      "loss": 0.7012,
+      "step": 22334
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1474543379859714e-06,
+      "loss": 0.6973,
+      "step": 22335
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.146741605297901e-06,
+      "loss": 0.6982,
+      "step": 22336
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.146028976683905e-06,
+      "loss": 0.7246,
+      "step": 22337
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1453164521534265e-06,
+      "loss": 0.7129,
+      "step": 22338
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1446040317159123e-06,
+      "loss": 0.6943,
+      "step": 22339
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1438917153808002e-06,
+      "loss": 0.7275,
+      "step": 22340
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.143179503157532e-06,
+      "loss": 0.7119,
+      "step": 22341
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1424673950555453e-06,
+      "loss": 0.7344,
+      "step": 22342
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.141755391084278e-06,
+      "loss": 0.7188,
+      "step": 22343
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.141043491253164e-06,
+      "loss": 0.7061,
+      "step": 22344
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.140331695571638e-06,
+      "loss": 0.7021,
+      "step": 22345
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.139620004049132e-06,
+      "loss": 0.6846,
+      "step": 22346
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1389084166950846e-06,
+      "loss": 0.7139,
+      "step": 22347
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1381969335189177e-06,
+      "loss": 0.7412,
+      "step": 22348
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.137485554530062e-06,
+      "loss": 0.6992,
+      "step": 22349
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1367742797379444e-06,
+      "loss": 0.7129,
+      "step": 22350
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1360631091519946e-06,
+      "loss": 0.7188,
+      "step": 22351
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1353520427816355e-06,
+      "loss": 0.7158,
+      "step": 22352
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1346410806362905e-06,
+      "loss": 0.6846,
+      "step": 22353
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.133930222725381e-06,
+      "loss": 0.7344,
+      "step": 22354
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1332194690583275e-06,
+      "loss": 0.6982,
+      "step": 22355
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1325088196445487e-06,
+      "loss": 0.71,
+      "step": 22356
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1317982744934627e-06,
+      "loss": 0.707,
+      "step": 22357
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1310878336144847e-06,
+      "loss": 0.7207,
+      "step": 22358
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1303774970170334e-06,
+      "loss": 0.7314,
+      "step": 22359
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.129667264710522e-06,
+      "loss": 0.7109,
+      "step": 22360
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1289571367043594e-06,
+      "loss": 0.708,
+      "step": 22361
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.128247113007954e-06,
+      "loss": 0.7061,
+      "step": 22362
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.127537193630722e-06,
+      "loss": 0.7266,
+      "step": 22363
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.12682737858207e-06,
+      "loss": 0.7207,
+      "step": 22364
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.126117667871402e-06,
+      "loss": 0.707,
+      "step": 22365
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.125408061508126e-06,
+      "loss": 0.7031,
+      "step": 22366
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1246985595016444e-06,
+      "loss": 0.7227,
+      "step": 22367
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.123989161861361e-06,
+      "loss": 0.6963,
+      "step": 22368
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1232798685966736e-06,
+      "loss": 0.7402,
+      "step": 22369
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1225706797169866e-06,
+      "loss": 0.7031,
+      "step": 22370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1218615952316932e-06,
+      "loss": 0.708,
+      "step": 22371
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.121152615150196e-06,
+      "loss": 0.6885,
+      "step": 22372
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.12044373948189e-06,
+      "loss": 0.6953,
+      "step": 22373
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1197349682361634e-06,
+      "loss": 0.7178,
+      "step": 22374
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1190263014224153e-06,
+      "loss": 0.6973,
+      "step": 22375
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1183177390500365e-06,
+      "loss": 0.7061,
+      "step": 22376
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1176092811284144e-06,
+      "loss": 0.7197,
+      "step": 22377
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1169009276669394e-06,
+      "loss": 0.6982,
+      "step": 22378
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1161926786749964e-06,
+      "loss": 0.708,
+      "step": 22379
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.115484534161978e-06,
+      "loss": 0.6953,
+      "step": 22380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1147764941372627e-06,
+      "loss": 0.6875,
+      "step": 22381
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.114068558610235e-06,
+      "loss": 0.7061,
+      "step": 22382
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1133607275902734e-06,
+      "loss": 0.7002,
+      "step": 22383
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.112653001086765e-06,
+      "loss": 0.7002,
+      "step": 22384
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1119453791090873e-06,
+      "loss": 0.7363,
+      "step": 22385
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.111237861666615e-06,
+      "loss": 0.6982,
+      "step": 22386
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1105304487687263e-06,
+      "loss": 0.6885,
+      "step": 22387
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.109823140424795e-06,
+      "loss": 0.7002,
+      "step": 22388
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.109115936644195e-06,
+      "loss": 0.7168,
+      "step": 22389
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108408837436299e-06,
+      "loss": 0.7021,
+      "step": 22390
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1077018428104744e-06,
+      "loss": 0.6865,
+      "step": 22391
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1069949527760967e-06,
+      "loss": 0.7002,
+      "step": 22392
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.106288167342533e-06,
+      "loss": 0.7217,
+      "step": 22393
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1055814865191437e-06,
+      "loss": 0.6875,
+      "step": 22394
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1048749103152964e-06,
+      "loss": 0.7344,
+      "step": 22395
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1041684387403582e-06,
+      "loss": 0.6963,
+      "step": 22396
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1034620718036903e-06,
+      "loss": 0.7061,
+      "step": 22397
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.102755809514652e-06,
+      "loss": 0.6943,
+      "step": 22398
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.102049651882604e-06,
+      "loss": 0.707,
+      "step": 22399
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.101343598916904e-06,
+      "loss": 0.709,
+      "step": 22400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.10063765062691e-06,
+      "loss": 0.7041,
+      "step": 22401
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0999318070219766e-06,
+      "loss": 0.7119,
+      "step": 22402
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.099226068111455e-06,
+      "loss": 0.7285,
+      "step": 22403
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.098520433904704e-06,
+      "loss": 0.7119,
+      "step": 22404
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.097814904411071e-06,
+      "loss": 0.7051,
+      "step": 22405
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0971094796399105e-06,
+      "loss": 0.7061,
+      "step": 22406
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0964041596005615e-06,
+      "loss": 0.6934,
+      "step": 22407
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.095698944302379e-06,
+      "loss": 0.7148,
+      "step": 22408
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.094993833754707e-06,
+      "loss": 0.7207,
+      "step": 22409
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.094288827966889e-06,
+      "loss": 0.7021,
+      "step": 22410
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0935839269482696e-06,
+      "loss": 0.6953,
+      "step": 22411
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.092879130708189e-06,
+      "loss": 0.708,
+      "step": 22412
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0921744392559863e-06,
+      "loss": 0.709,
+      "step": 22413
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0914698526010035e-06,
+      "loss": 0.7002,
+      "step": 22414
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0907653707525756e-06,
+      "loss": 0.6943,
+      "step": 22415
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0900609937200356e-06,
+      "loss": 0.7148,
+      "step": 22416
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0893567215127254e-06,
+      "loss": 0.6904,
+      "step": 22417
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.088652554139975e-06,
+      "loss": 0.7119,
+      "step": 22418
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.087948491611115e-06,
+      "loss": 0.7236,
+      "step": 22419
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0872445339354773e-06,
+      "loss": 0.7012,
+      "step": 22420
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.086540681122391e-06,
+      "loss": 0.6904,
+      "step": 22421
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0858369331811825e-06,
+      "loss": 0.7285,
+      "step": 22422
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0851332901211795e-06,
+      "loss": 0.7148,
+      "step": 22423
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0844297519517033e-06,
+      "loss": 0.7275,
+      "step": 22424
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0837263186820856e-06,
+      "loss": 0.6816,
+      "step": 22425
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0830229903216404e-06,
+      "loss": 0.7197,
+      "step": 22426
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0823197668796923e-06,
+      "loss": 0.707,
+      "step": 22427
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0816166483655565e-06,
+      "loss": 0.707,
+      "step": 22428
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.080913634788557e-06,
+      "loss": 0.7021,
+      "step": 22429
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0802107261580087e-06,
+      "loss": 0.7002,
+      "step": 22430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.079507922483225e-06,
+      "loss": 0.7051,
+      "step": 22431
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.07880522377352e-06,
+      "loss": 0.7344,
+      "step": 22432
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0781026300382068e-06,
+      "loss": 0.7109,
+      "step": 22433
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0774001412865963e-06,
+      "loss": 0.7217,
+      "step": 22434
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0766977575279977e-06,
+      "loss": 0.7256,
+      "step": 22435
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.075995478771716e-06,
+      "loss": 0.709,
+      "step": 22436
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.075293305027066e-06,
+      "loss": 0.7061,
+      "step": 22437
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0745912363033504e-06,
+      "loss": 0.7148,
+      "step": 22438
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.073889272609869e-06,
+      "loss": 0.709,
+      "step": 22439
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.073187413955924e-06,
+      "loss": 0.7334,
+      "step": 22440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0724856603508225e-06,
+      "loss": 0.7021,
+      "step": 22441
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.071784011803862e-06,
+      "loss": 0.6943,
+      "step": 22442
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0710824683243402e-06,
+      "loss": 0.7275,
+      "step": 22443
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0703810299215543e-06,
+      "loss": 0.6797,
+      "step": 22444
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0696796966048005e-06,
+      "loss": 0.7305,
+      "step": 22445
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0689784683833724e-06,
+      "loss": 0.7275,
+      "step": 22446
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.068277345266564e-06,
+      "loss": 0.6904,
+      "step": 22447
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.067576327263664e-06,
+      "loss": 0.7031,
+      "step": 22448
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0668754143839664e-06,
+      "loss": 0.7246,
+      "step": 22449
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.066174606636758e-06,
+      "loss": 0.7021,
+      "step": 22450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.06547390403133e-06,
+      "loss": 0.6904,
+      "step": 22451
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.064773306576959e-06,
+      "loss": 0.7109,
+      "step": 22452
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.064072814282938e-06,
+      "loss": 0.7148,
+      "step": 22453
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0633724271585477e-06,
+      "loss": 0.7061,
+      "step": 22454
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0626721452130693e-06,
+      "loss": 0.7031,
+      "step": 22455
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.061971968455783e-06,
+      "loss": 0.7139,
+      "step": 22456
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0612718968959667e-06,
+      "loss": 0.7197,
+      "step": 22457
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.060571930542904e-06,
+      "loss": 0.6895,
+      "step": 22458
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.059872069405865e-06,
+      "loss": 0.6904,
+      "step": 22459
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0591723134941255e-06,
+      "loss": 0.7334,
+      "step": 22460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.058472662816957e-06,
+      "loss": 0.6943,
+      "step": 22461
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0577731173836367e-06,
+      "loss": 0.7051,
+      "step": 22462
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0570736772034326e-06,
+      "loss": 0.6826,
+      "step": 22463
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0563743422856142e-06,
+      "loss": 0.6982,
+      "step": 22464
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0556751126394493e-06,
+      "loss": 0.7061,
+      "step": 22465
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0549759882742036e-06,
+      "loss": 0.6855,
+      "step": 22466
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0542769691991427e-06,
+      "loss": 0.709,
+      "step": 22467
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0535780554235294e-06,
+      "loss": 0.6846,
+      "step": 22468
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0528792469566248e-06,
+      "loss": 0.6992,
+      "step": 22469
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.052180543807695e-06,
+      "loss": 0.6973,
+      "step": 22470
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0514819459859982e-06,
+      "loss": 0.6914,
+      "step": 22471
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0507834535007874e-06,
+      "loss": 0.7305,
+      "step": 22472
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0500850663613204e-06,
+      "loss": 0.709,
+      "step": 22473
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0493867845768567e-06,
+      "loss": 0.7305,
+      "step": 22474
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.048688608156647e-06,
+      "loss": 0.6953,
+      "step": 22475
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0479905371099453e-06,
+      "loss": 0.7119,
+      "step": 22476
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0472925714460024e-06,
+      "loss": 0.6982,
+      "step": 22477
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0465947111740657e-06,
+      "loss": 0.6865,
+      "step": 22478
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.045896956303387e-06,
+      "loss": 0.7236,
+      "step": 22479
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.045199306843211e-06,
+      "loss": 0.7148,
+      "step": 22480
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.044501762802781e-06,
+      "loss": 0.7148,
+      "step": 22481
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0438043241913463e-06,
+      "loss": 0.6934,
+      "step": 22482
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.043106991018147e-06,
+      "loss": 0.7559,
+      "step": 22483
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0424097632924276e-06,
+      "loss": 0.6973,
+      "step": 22484
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.041712641023419e-06,
+      "loss": 0.709,
+      "step": 22485
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0410156242203683e-06,
+      "loss": 0.7061,
+      "step": 22486
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0403187128925096e-06,
+      "loss": 0.7363,
+      "step": 22487
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0396219070490795e-06,
+      "loss": 0.6943,
+      "step": 22488
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0389252066993115e-06,
+      "loss": 0.708,
+      "step": 22489
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0382286118524387e-06,
+      "loss": 0.7207,
+      "step": 22490
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0375321225176924e-06,
+      "loss": 0.7188,
+      "step": 22491
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0368357387043024e-06,
+      "loss": 0.6992,
+      "step": 22492
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0361394604214966e-06,
+      "loss": 0.7031,
+      "step": 22493
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0354432876785056e-06,
+      "loss": 0.71,
+      "step": 22494
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.034747220484553e-06,
+      "loss": 0.6885,
+      "step": 22495
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.034051258848865e-06,
+      "loss": 0.6953,
+      "step": 22496
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0333554027806625e-06,
+      "loss": 0.7422,
+      "step": 22497
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.032659652289167e-06,
+      "loss": 0.7021,
+      "step": 22498
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0319640073836023e-06,
+      "loss": 0.7148,
+      "step": 22499
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.031268468073183e-06,
+      "loss": 0.7246,
+      "step": 22500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0305730343671294e-06,
+      "loss": 0.7051,
+      "step": 22501
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0298777062746543e-06,
+      "loss": 0.6807,
+      "step": 22502
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.02918248380498e-06,
+      "loss": 0.7168,
+      "step": 22503
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.028487366967313e-06,
+      "loss": 0.7061,
+      "step": 22504
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0277923557708667e-06,
+      "loss": 0.7051,
+      "step": 22505
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0270974502248486e-06,
+      "loss": 0.707,
+      "step": 22506
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0264026503384746e-06,
+      "loss": 0.6904,
+      "step": 22507
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.025707956120949e-06,
+      "loss": 0.7139,
+      "step": 22508
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0250133675814775e-06,
+      "loss": 0.7051,
+      "step": 22509
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.024318884729266e-06,
+      "loss": 0.7158,
+      "step": 22510
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.023624507573517e-06,
+      "loss": 0.6846,
+      "step": 22511
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.022930236123435e-06,
+      "loss": 0.7217,
+      "step": 22512
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.022236070388218e-06,
+      "loss": 0.7197,
+      "step": 22513
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0215420103770632e-06,
+      "loss": 0.6982,
+      "step": 22514
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.020848056099175e-06,
+      "loss": 0.6865,
+      "step": 22515
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0201542075637493e-06,
+      "loss": 0.7129,
+      "step": 22516
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0194604647799755e-06,
+      "loss": 0.709,
+      "step": 22517
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.018766827757048e-06,
+      "loss": 0.7305,
+      "step": 22518
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0180732965041637e-06,
+      "loss": 0.6914,
+      "step": 22519
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0173798710305113e-06,
+      "loss": 0.7363,
+      "step": 22520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0166865513452817e-06,
+      "loss": 0.6914,
+      "step": 22521
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.01599333745766e-06,
+      "loss": 0.7041,
+      "step": 22522
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0153002293768342e-06,
+      "loss": 0.7109,
+      "step": 22523
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.014607227111991e-06,
+      "loss": 0.6816,
+      "step": 22524
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.013914330672313e-06,
+      "loss": 0.6973,
+      "step": 22525
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.01322154006698e-06,
+      "loss": 0.7031,
+      "step": 22526
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.012528855305179e-06,
+      "loss": 0.7188,
+      "step": 22527
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0118362763960864e-06,
+      "loss": 0.71,
+      "step": 22528
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.011143803348884e-06,
+      "loss": 0.7061,
+      "step": 22529
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.010451436172739e-06,
+      "loss": 0.7256,
+      "step": 22530
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0097591748768364e-06,
+      "loss": 0.7002,
+      "step": 22531
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0090670194703475e-06,
+      "loss": 0.71,
+      "step": 22532
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0083749699624457e-06,
+      "loss": 0.7197,
+      "step": 22533
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0076830263623005e-06,
+      "loss": 0.6924,
+      "step": 22534
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.006991188679082e-06,
+      "loss": 0.6875,
+      "step": 22535
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0062994569219597e-06,
+      "loss": 0.7051,
+      "step": 22536
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0056078311001003e-06,
+      "loss": 0.7139,
+      "step": 22537
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0049163112226698e-06,
+      "loss": 0.6934,
+      "step": 22538
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.004224897298829e-06,
+      "loss": 0.6875,
+      "step": 22539
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.003533589337746e-06,
+      "loss": 0.7197,
+      "step": 22540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.00284238734858e-06,
+      "loss": 0.6943,
+      "step": 22541
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.002151291340493e-06,
+      "loss": 0.7129,
+      "step": 22542
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0014603013226396e-06,
+      "loss": 0.7148,
+      "step": 22543
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0007694173041804e-06,
+      "loss": 0.6914,
+      "step": 22544
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0000786392942697e-06,
+      "loss": 0.6865,
+      "step": 22545
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.999387967302061e-06,
+      "loss": 0.6924,
+      "step": 22546
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9986974013367077e-06,
+      "loss": 0.7041,
+      "step": 22547
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9980069414073668e-06,
+      "loss": 0.7246,
+      "step": 22548
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9973165875231814e-06,
+      "loss": 0.6885,
+      "step": 22549
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.996626339693303e-06,
+      "loss": 0.7041,
+      "step": 22550
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9959361979268753e-06,
+      "loss": 0.6904,
+      "step": 22551
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.995246162233052e-06,
+      "loss": 0.6709,
+      "step": 22552
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9945562326209734e-06,
+      "loss": 0.707,
+      "step": 22553
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993866409099782e-06,
+      "loss": 0.7197,
+      "step": 22554
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993176691678621e-06,
+      "loss": 0.6729,
+      "step": 22555
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9924870803666294e-06,
+      "loss": 0.7031,
+      "step": 22556
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.991797575172948e-06,
+      "loss": 0.7061,
+      "step": 22557
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.991108176106713e-06,
+      "loss": 0.7051,
+      "step": 22558
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9904188831770587e-06,
+      "loss": 0.7285,
+      "step": 22559
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9897296963931244e-06,
+      "loss": 0.7139,
+      "step": 22560
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.989040615764044e-06,
+      "loss": 0.7246,
+      "step": 22561
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9883516412989435e-06,
+      "loss": 0.6934,
+      "step": 22562
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9876627730069533e-06,
+      "loss": 0.7188,
+      "step": 22563
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9869740108972092e-06,
+      "loss": 0.7061,
+      "step": 22564
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.986285354978834e-06,
+      "loss": 0.6885,
+      "step": 22565
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9855968052609563e-06,
+      "loss": 0.6992,
+      "step": 22566
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.984908361752701e-06,
+      "loss": 0.6875,
+      "step": 22567
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.984220024463189e-06,
+      "loss": 0.7139,
+      "step": 22568
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.983531793401545e-06,
+      "loss": 0.7227,
+      "step": 22569
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.982843668576889e-06,
+      "loss": 0.709,
+      "step": 22570
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.982155649998336e-06,
+      "loss": 0.6689,
+      "step": 22571
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9814677376750126e-06,
+      "loss": 0.7051,
+      "step": 22572
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9807799316160283e-06,
+      "loss": 0.707,
+      "step": 22573
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9800922318305037e-06,
+      "loss": 0.6836,
+      "step": 22574
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9794046383275444e-06,
+      "loss": 0.6934,
+      "step": 22575
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9787171511162695e-06,
+      "loss": 0.6973,
+      "step": 22576
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.978029770205788e-06,
+      "loss": 0.6855,
+      "step": 22577
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.977342495605209e-06,
+      "loss": 0.6807,
+      "step": 22578
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9766553273236398e-06,
+      "loss": 0.7295,
+      "step": 22579
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.975968265370185e-06,
+      "loss": 0.6758,
+      "step": 22580
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9752813097539582e-06,
+      "loss": 0.7031,
+      "step": 22581
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.974594460484054e-06,
+      "loss": 0.7002,
+      "step": 22582
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9739077175695786e-06,
+      "loss": 0.6865,
+      "step": 22583
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9732210810196297e-06,
+      "loss": 0.6963,
+      "step": 22584
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9725345508433123e-06,
+      "loss": 0.7246,
+      "step": 22585
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9718481270497214e-06,
+      "loss": 0.71,
+      "step": 22586
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9711618096479545e-06,
+      "loss": 0.7178,
+      "step": 22587
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9704755986471067e-06,
+      "loss": 0.707,
+      "step": 22588
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.969789494056272e-06,
+      "loss": 0.7158,
+      "step": 22589
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.969103495884542e-06,
+      "loss": 0.709,
+      "step": 22590
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.96841760414101e-06,
+      "loss": 0.7041,
+      "step": 22591
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9677318188347593e-06,
+      "loss": 0.6914,
+      "step": 22592
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9670461399748876e-06,
+      "loss": 0.707,
+      "step": 22593
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.966360567570479e-06,
+      "loss": 0.71,
+      "step": 22594
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.965675101630614e-06,
+      "loss": 0.6914,
+      "step": 22595
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.964989742164377e-06,
+      "loss": 0.6914,
+      "step": 22596
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9643044891808573e-06,
+      "loss": 0.6953,
+      "step": 22597
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9636193426891314e-06,
+      "loss": 0.7217,
+      "step": 22598
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9629343026982796e-06,
+      "loss": 0.6777,
+      "step": 22599
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9622493692173807e-06,
+      "loss": 0.71,
+      "step": 22600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.961564542255512e-06,
+      "loss": 0.707,
+      "step": 22601
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9608798218217474e-06,
+      "loss": 0.6934,
+      "step": 22602
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.960195207925163e-06,
+      "loss": 0.6855,
+      "step": 22603
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9595107005748282e-06,
+      "loss": 0.6914,
+      "step": 22604
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.958826299779819e-06,
+      "loss": 0.7061,
+      "step": 22605
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.958142005549204e-06,
+      "loss": 0.6982,
+      "step": 22606
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9574578178920533e-06,
+      "loss": 0.6982,
+      "step": 22607
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9567737368174254e-06,
+      "loss": 0.7266,
+      "step": 22608
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9560897623343965e-06,
+      "loss": 0.7256,
+      "step": 22609
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.955405894452025e-06,
+      "loss": 0.7051,
+      "step": 22610
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9547221331793764e-06,
+      "loss": 0.7217,
+      "step": 22611
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9540384785255105e-06,
+      "loss": 0.6973,
+      "step": 22612
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.953354930499487e-06,
+      "loss": 0.7305,
+      "step": 22613
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.952671489110367e-06,
+      "loss": 0.6807,
+      "step": 22614
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9519881543672046e-06,
+      "loss": 0.7031,
+      "step": 22615
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.951304926279055e-06,
+      "loss": 0.6934,
+      "step": 22616
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9506218048549776e-06,
+      "loss": 0.709,
+      "step": 22617
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9499387901040225e-06,
+      "loss": 0.6738,
+      "step": 22618
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9492558820352403e-06,
+      "loss": 0.7012,
+      "step": 22619
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.948573080657683e-06,
+      "loss": 0.7061,
+      "step": 22620
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9478903859803976e-06,
+      "loss": 0.6953,
+      "step": 22621
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9472077980124314e-06,
+      "loss": 0.7012,
+      "step": 22622
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9465253167628318e-06,
+      "loss": 0.7393,
+      "step": 22623
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9458429422406423e-06,
+      "loss": 0.6934,
+      "step": 22624
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9451606744549036e-06,
+      "loss": 0.708,
+      "step": 22625
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.944478513414665e-06,
+      "loss": 0.7021,
+      "step": 22626
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9437964591289594e-06,
+      "loss": 0.6943,
+      "step": 22627
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9431145116068274e-06,
+      "loss": 0.7031,
+      "step": 22628
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9424326708573036e-06,
+      "loss": 0.7061,
+      "step": 22629
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9417509368894304e-06,
+      "loss": 0.6982,
+      "step": 22630
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9410693097122402e-06,
+      "loss": 0.7012,
+      "step": 22631
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9403877893347632e-06,
+      "loss": 0.7275,
+      "step": 22632
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.939706375766035e-06,
+      "loss": 0.7061,
+      "step": 22633
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9390250690150825e-06,
+      "loss": 0.7031,
+      "step": 22634
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.938343869090936e-06,
+      "loss": 0.7021,
+      "step": 22635
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9376627760026247e-06,
+      "loss": 0.6797,
+      "step": 22636
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.936981789759169e-06,
+      "loss": 0.6982,
+      "step": 22637
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.936300910369601e-06,
+      "loss": 0.6904,
+      "step": 22638
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.935620137842943e-06,
+      "loss": 0.6982,
+      "step": 22639
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.934939472188211e-06,
+      "loss": 0.7041,
+      "step": 22640
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9342589134144276e-06,
+      "loss": 0.6904,
+      "step": 22641
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9335784615306154e-06,
+      "loss": 0.7158,
+      "step": 22642
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9328981165457883e-06,
+      "loss": 0.707,
+      "step": 22643
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.932217878468965e-06,
+      "loss": 0.6855,
+      "step": 22644
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.931537747309159e-06,
+      "loss": 0.7129,
+      "step": 22645
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9308577230753835e-06,
+      "loss": 0.7295,
+      "step": 22646
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9301778057766496e-06,
+      "loss": 0.6953,
+      "step": 22647
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9294979954219693e-06,
+      "loss": 0.6953,
+      "step": 22648
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.928818292020349e-06,
+      "loss": 0.7354,
+      "step": 22649
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9281386955808013e-06,
+      "loss": 0.7168,
+      "step": 22650
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.92745920611233e-06,
+      "loss": 0.71,
+      "step": 22651
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.926779823623941e-06,
+      "loss": 0.7227,
+      "step": 22652
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.926100548124632e-06,
+      "loss": 0.7041,
+      "step": 22653
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.925421379623411e-06,
+      "loss": 0.6904,
+      "step": 22654
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.924742318129278e-06,
+      "loss": 0.6943,
+      "step": 22655
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9240633636512295e-06,
+      "loss": 0.7041,
+      "step": 22656
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9233845161982665e-06,
+      "loss": 0.7363,
+      "step": 22657
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9227057757793797e-06,
+      "loss": 0.7148,
+      "step": 22658
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9220271424035743e-06,
+      "loss": 0.6963,
+      "step": 22659
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.921348616079833e-06,
+      "loss": 0.7109,
+      "step": 22660
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9206701968171513e-06,
+      "loss": 0.6924,
+      "step": 22661
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9199918846245225e-06,
+      "loss": 0.7061,
+      "step": 22662
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9193136795109347e-06,
+      "loss": 0.6904,
+      "step": 22663
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.918635581485374e-06,
+      "loss": 0.6738,
+      "step": 22664
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.917957590556829e-06,
+      "loss": 0.6904,
+      "step": 22665
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.917279706734283e-06,
+      "loss": 0.7188,
+      "step": 22666
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.91660193002672e-06,
+      "loss": 0.7119,
+      "step": 22667
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.915924260443123e-06,
+      "loss": 0.7256,
+      "step": 22668
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9152466979924712e-06,
+      "loss": 0.6934,
+      "step": 22669
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9145692426837427e-06,
+      "loss": 0.7178,
+      "step": 22670
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.913891894525919e-06,
+      "loss": 0.6924,
+      "step": 22671
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9132146535279784e-06,
+      "loss": 0.6924,
+      "step": 22672
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9125375196988883e-06,
+      "loss": 0.6914,
+      "step": 22673
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.911860493047625e-06,
+      "loss": 0.7158,
+      "step": 22674
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9111835735831653e-06,
+      "loss": 0.6953,
+      "step": 22675
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.910506761314476e-06,
+      "loss": 0.7305,
+      "step": 22676
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.7168,
+      "step": 22677
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9091534584002856e-06,
+      "loss": 0.7217,
+      "step": 22678
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.908476967772721e-06,
+      "loss": 0.7236,
+      "step": 22679
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9078005843767956e-06,
+      "loss": 0.7305,
+      "step": 22680
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.907124308221473e-06,
+      "loss": 0.7041,
+      "step": 22681
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9064481393157153e-06,
+      "loss": 0.7158,
+      "step": 22682
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9057720776684863e-06,
+      "loss": 0.6904,
+      "step": 22683
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9050961232887434e-06,
+      "loss": 0.6973,
+      "step": 22684
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9044202761854469e-06,
+      "loss": 0.6943,
+      "step": 22685
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9037445363675467e-06,
+      "loss": 0.6953,
+      "step": 22686
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9030689038440042e-06,
+      "loss": 0.709,
+      "step": 22687
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9023933786237715e-06,
+      "loss": 0.7266,
+      "step": 22688
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9017179607157999e-06,
+      "loss": 0.6992,
+      "step": 22689
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9010426501290425e-06,
+      "loss": 0.7227,
+      "step": 22690
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9003674468724464e-06,
+      "loss": 0.7246,
+      "step": 22691
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8996923509549604e-06,
+      "loss": 0.7012,
+      "step": 22692
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8990173623855313e-06,
+      "loss": 0.6826,
+      "step": 22693
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8983424811731021e-06,
+      "loss": 0.7256,
+      "step": 22694
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8976677073266203e-06,
+      "loss": 0.7227,
+      "step": 22695
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8969930408550274e-06,
+      "loss": 0.6846,
+      "step": 22696
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8963184817672632e-06,
+      "loss": 0.6836,
+      "step": 22697
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8956440300722667e-06,
+      "loss": 0.7148,
+      "step": 22698
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.894969685778978e-06,
+      "loss": 0.6973,
+      "step": 22699
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8942954488963316e-06,
+      "loss": 0.7109,
+      "step": 22700
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8936213194332632e-06,
+      "loss": 0.7324,
+      "step": 22701
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8929472973987074e-06,
+      "loss": 0.6924,
+      "step": 22702
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8922733828015937e-06,
+      "loss": 0.7139,
+      "step": 22703
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.891599575650861e-06,
+      "loss": 0.7207,
+      "step": 22704
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8909258759554306e-06,
+      "loss": 0.709,
+      "step": 22705
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8902522837242298e-06,
+      "loss": 0.6924,
+      "step": 22706
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.889578798966193e-06,
+      "loss": 0.6992,
+      "step": 22707
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8889054216902403e-06,
+      "loss": 0.7002,
+      "step": 22708
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.888232151905296e-06,
+      "loss": 0.708,
+      "step": 22709
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8875589896202838e-06,
+      "loss": 0.7129,
+      "step": 22710
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8868859348441237e-06,
+      "loss": 0.7021,
+      "step": 22711
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.886212987585735e-06,
+      "loss": 0.6885,
+      "step": 22712
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8855401478540369e-06,
+      "loss": 0.7236,
+      "step": 22713
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.884867415657945e-06,
+      "loss": 0.7266,
+      "step": 22714
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8841947910063718e-06,
+      "loss": 0.7002,
+      "step": 22715
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8835222739082381e-06,
+      "loss": 0.7158,
+      "step": 22716
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8828498643724546e-06,
+      "loss": 0.6963,
+      "step": 22717
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8821775624079276e-06,
+      "loss": 0.7061,
+      "step": 22718
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8815053680235673e-06,
+      "loss": 0.6699,
+      "step": 22719
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.880833281228286e-06,
+      "loss": 0.7021,
+      "step": 22720
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8801613020309883e-06,
+      "loss": 0.6992,
+      "step": 22721
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8794894304405797e-06,
+      "loss": 0.7148,
+      "step": 22722
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8788176664659618e-06,
+      "loss": 0.709,
+      "step": 22723
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8781460101160443e-06,
+      "loss": 0.6836,
+      "step": 22724
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8774744613997197e-06,
+      "loss": 0.6914,
+      "step": 22725
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8768030203258914e-06,
+      "loss": 0.7314,
+      "step": 22726
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8761316869034541e-06,
+      "loss": 0.7139,
+      "step": 22727
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8754604611413108e-06,
+      "loss": 0.7109,
+      "step": 22728
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8747893430483533e-06,
+      "loss": 0.6885,
+      "step": 22729
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8741183326334788e-06,
+      "loss": 0.6992,
+      "step": 22730
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8734474299055715e-06,
+      "loss": 0.7031,
+      "step": 22731
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8727766348735299e-06,
+      "loss": 0.7461,
+      "step": 22732
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8721059475462421e-06,
+      "loss": 0.6953,
+      "step": 22733
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8714353679325948e-06,
+      "loss": 0.7021,
+      "step": 22734
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.870764896041476e-06,
+      "loss": 0.707,
+      "step": 22735
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8700945318817675e-06,
+      "loss": 0.6865,
+      "step": 22736
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8694242754623615e-06,
+      "loss": 0.6982,
+      "step": 22737
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8687541267921328e-06,
+      "loss": 0.7285,
+      "step": 22738
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8680840858799621e-06,
+      "loss": 0.7002,
+      "step": 22739
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8674141527347346e-06,
+      "loss": 0.7012,
+      "step": 22740
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8667443273653251e-06,
+      "loss": 0.7227,
+      "step": 22741
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8660746097806115e-06,
+      "loss": 0.6943,
+      "step": 22742
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8654049999894685e-06,
+      "loss": 0.7129,
+      "step": 22743
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8647354980007692e-06,
+      "loss": 0.6875,
+      "step": 22744
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8640661038233876e-06,
+      "loss": 0.709,
+      "step": 22745
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8633968174661942e-06,
+      "loss": 0.7168,
+      "step": 22746
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8627276389380577e-06,
+      "loss": 0.6826,
+      "step": 22747
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8620585682478454e-06,
+      "loss": 0.7188,
+      "step": 22748
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.86138960540443e-06,
+      "loss": 0.7168,
+      "step": 22749
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8607207504166702e-06,
+      "loss": 0.6797,
+      "step": 22750
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8600520032934332e-06,
+      "loss": 0.7051,
+      "step": 22751
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8593833640435766e-06,
+      "loss": 0.7139,
+      "step": 22752
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8587148326759686e-06,
+      "loss": 0.71,
+      "step": 22753
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8580464091994654e-06,
+      "loss": 0.6807,
+      "step": 22754
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8573780936229246e-06,
+      "loss": 0.707,
+      "step": 22755
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8567098859552047e-06,
+      "loss": 0.6846,
+      "step": 22756
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8560417862051595e-06,
+      "loss": 0.7139,
+      "step": 22757
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.855373794381643e-06,
+      "loss": 0.6943,
+      "step": 22758
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8547059104935083e-06,
+      "loss": 0.6982,
+      "step": 22759
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8540381345496028e-06,
+      "loss": 0.6992,
+      "step": 22760
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8533704665587837e-06,
+      "loss": 0.6943,
+      "step": 22761
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8527029065298962e-06,
+      "loss": 0.7119,
+      "step": 22762
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8520354544717832e-06,
+      "loss": 0.7207,
+      "step": 22763
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8513681103932902e-06,
+      "loss": 0.7041,
+      "step": 22764
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.850700874303266e-06,
+      "loss": 0.6885,
+      "step": 22765
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.85003374621055e-06,
+      "loss": 0.6826,
+      "step": 22766
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8493667261239845e-06,
+      "loss": 0.6943,
+      "step": 22767
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8486998140524048e-06,
+      "loss": 0.6836,
+      "step": 22768
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8480330100046585e-06,
+      "loss": 0.708,
+      "step": 22769
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8473663139895737e-06,
+      "loss": 0.7061,
+      "step": 22770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.846699726015988e-06,
+      "loss": 0.7217,
+      "step": 22771
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8460332460927322e-06,
+      "loss": 0.7119,
+      "step": 22772
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8453668742286456e-06,
+      "loss": 0.6904,
+      "step": 22773
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.844700610432555e-06,
+      "loss": 0.6855,
+      "step": 22774
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8440344547132937e-06,
+      "loss": 0.6904,
+      "step": 22775
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8433684070796809e-06,
+      "loss": 0.7236,
+      "step": 22776
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8427024675405524e-06,
+      "loss": 0.6973,
+      "step": 22777
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8420366361047304e-06,
+      "loss": 0.7129,
+      "step": 22778
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8413709127810374e-06,
+      "loss": 0.6992,
+      "step": 22779
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.840705297578298e-06,
+      "loss": 0.7002,
+      "step": 22780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8400397905053291e-06,
+      "loss": 0.6992,
+      "step": 22781
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8393743915709573e-06,
+      "loss": 0.6914,
+      "step": 22782
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8387091007839953e-06,
+      "loss": 0.707,
+      "step": 22783
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8380439181532584e-06,
+      "loss": 0.7148,
+      "step": 22784
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.837378843687566e-06,
+      "loss": 0.7031,
+      "step": 22785
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8367138773957316e-06,
+      "loss": 0.6953,
+      "step": 22786
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8360490192865653e-06,
+      "loss": 0.7119,
+      "step": 22787
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8353842693688795e-06,
+      "loss": 0.7021,
+      "step": 22788
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8347196276514835e-06,
+      "loss": 0.7236,
+      "step": 22789
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8340550941431855e-06,
+      "loss": 0.7061,
+      "step": 22790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8333906688527914e-06,
+      "loss": 0.7266,
+      "step": 22791
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8327263517891059e-06,
+      "loss": 0.7021,
+      "step": 22792
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.832062142960932e-06,
+      "loss": 0.7344,
+      "step": 22793
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.831398042377076e-06,
+      "loss": 0.7178,
+      "step": 22794
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8307340500463377e-06,
+      "loss": 0.6826,
+      "step": 22795
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8300701659775145e-06,
+      "loss": 0.6924,
+      "step": 22796
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8294063901794013e-06,
+      "loss": 0.6982,
+      "step": 22797
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8287427226608023e-06,
+      "loss": 0.6855,
+      "step": 22798
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8280791634305083e-06,
+      "loss": 0.7041,
+      "step": 22799
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8274157124973135e-06,
+      "loss": 0.7021,
+      "step": 22800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8267523698700095e-06,
+      "loss": 0.6953,
+      "step": 22801
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8260891355573894e-06,
+      "loss": 0.6963,
+      "step": 22802
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8254260095682409e-06,
+      "loss": 0.71,
+      "step": 22803
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.824762991911352e-06,
+      "loss": 0.7002,
+      "step": 22804
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8241000825955068e-06,
+      "loss": 0.6943,
+      "step": 22805
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8234372816294954e-06,
+      "loss": 0.7061,
+      "step": 22806
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8227745890220993e-06,
+      "loss": 0.7021,
+      "step": 22807
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8221120047821028e-06,
+      "loss": 0.7012,
+      "step": 22808
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8214495289182798e-06,
+      "loss": 0.6855,
+      "step": 22809
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8207871614394169e-06,
+      "loss": 0.6943,
+      "step": 22810
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8201249023542888e-06,
+      "loss": 0.6641,
+      "step": 22811
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8194627516716734e-06,
+      "loss": 0.7031,
+      "step": 22812
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8188007094003423e-06,
+      "loss": 0.7031,
+      "step": 22813
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8181387755490766e-06,
+      "loss": 0.7305,
+      "step": 22814
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.817476950126641e-06,
+      "loss": 0.6992,
+      "step": 22815
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8168152331418098e-06,
+      "loss": 0.6924,
+      "step": 22816
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8161536246033473e-06,
+      "loss": 0.7148,
+      "step": 22817
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8154921245200296e-06,
+      "loss": 0.7197,
+      "step": 22818
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8148307329006176e-06,
+      "loss": 0.7041,
+      "step": 22819
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8141694497538786e-06,
+      "loss": 0.7119,
+      "step": 22820
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8135082750885746e-06,
+      "loss": 0.7227,
+      "step": 22821
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8128472089134685e-06,
+      "loss": 0.7119,
+      "step": 22822
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8121862512373213e-06,
+      "loss": 0.6826,
+      "step": 22823
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.811525402068891e-06,
+      "loss": 0.7119,
+      "step": 22824
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8108646614169366e-06,
+      "loss": 0.709,
+      "step": 22825
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8102040292902123e-06,
+      "loss": 0.6895,
+      "step": 22826
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8095435056974786e-06,
+      "loss": 0.7158,
+      "step": 22827
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.808883090647483e-06,
+      "loss": 0.7207,
+      "step": 22828
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8082227841489763e-06,
+      "loss": 0.7158,
+      "step": 22829
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.807562586210716e-06,
+      "loss": 0.7148,
+      "step": 22830
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8069024968414472e-06,
+      "loss": 0.6768,
+      "step": 22831
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8062425160499186e-06,
+      "loss": 0.6992,
+      "step": 22832
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8055826438448766e-06,
+      "loss": 0.6982,
+      "step": 22833
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8049228802350648e-06,
+      "loss": 0.7178,
+      "step": 22834
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.804263225229227e-06,
+      "loss": 0.7139,
+      "step": 22835
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8036036788361056e-06,
+      "loss": 0.6982,
+      "step": 22836
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8029442410644416e-06,
+      "loss": 0.6963,
+      "step": 22837
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8022849119229702e-06,
+      "loss": 0.7031,
+      "step": 22838
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8016256914204356e-06,
+      "loss": 0.6953,
+      "step": 22839
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8009665795655728e-06,
+      "loss": 0.6875,
+      "step": 22840
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8003075763671129e-06,
+      "loss": 0.6934,
+      "step": 22841
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7996486818337877e-06,
+      "loss": 0.71,
+      "step": 22842
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7989898959743358e-06,
+      "loss": 0.6963,
+      "step": 22843
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7983312187974834e-06,
+      "loss": 0.7383,
+      "step": 22844
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7976726503119602e-06,
+      "loss": 0.7061,
+      "step": 22845
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7970141905264915e-06,
+      "loss": 0.6973,
+      "step": 22846
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7963558394498104e-06,
+      "loss": 0.7109,
+      "step": 22847
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7956975970906343e-06,
+      "loss": 0.7139,
+      "step": 22848
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7950394634576907e-06,
+      "loss": 0.7109,
+      "step": 22849
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7943814385596958e-06,
+      "loss": 0.6895,
+      "step": 22850
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7937235224053773e-06,
+      "loss": 0.71,
+      "step": 22851
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7930657150034503e-06,
+      "loss": 0.6836,
+      "step": 22852
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7924080163626356e-06,
+      "loss": 0.6992,
+      "step": 22853
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.791750426491642e-06,
+      "loss": 0.7031,
+      "step": 22854
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.791092945399191e-06,
+      "loss": 0.7148,
+      "step": 22855
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7904355730939927e-06,
+      "loss": 0.7031,
+      "step": 22856
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7897783095847598e-06,
+      "loss": 0.7168,
+      "step": 22857
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7891211548802024e-06,
+      "loss": 0.7266,
+      "step": 22858
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7884641089890265e-06,
+      "loss": 0.6836,
+      "step": 22859
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7878071719199474e-06,
+      "loss": 0.709,
+      "step": 22860
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.787150343681664e-06,
+      "loss": 0.707,
+      "step": 22861
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.786493624282879e-06,
+      "loss": 0.7236,
+      "step": 22862
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7858370137323021e-06,
+      "loss": 0.7109,
+      "step": 22863
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7851805120386322e-06,
+      "loss": 0.707,
+      "step": 22864
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.78452411921057e-06,
+      "loss": 0.6924,
+      "step": 22865
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7838678352568129e-06,
+      "loss": 0.7207,
+      "step": 22866
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7832116601860583e-06,
+      "loss": 0.6865,
+      "step": 22867
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7825555940070028e-06,
+      "loss": 0.7119,
+      "step": 22868
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7818996367283403e-06,
+      "loss": 0.7344,
+      "step": 22869
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7812437883587642e-06,
+      "loss": 0.7197,
+      "step": 22870
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7805880489069637e-06,
+      "loss": 0.6885,
+      "step": 22871
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7799324183816336e-06,
+      "loss": 0.6963,
+      "step": 22872
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.779276896791463e-06,
+      "loss": 0.6992,
+      "step": 22873
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.77862148414513e-06,
+      "loss": 0.709,
+      "step": 22874
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7779661804513293e-06,
+      "loss": 0.7139,
+      "step": 22875
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.777310985718743e-06,
+      "loss": 0.6924,
+      "step": 22876
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7766558999560523e-06,
+      "loss": 0.6904,
+      "step": 22877
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.77600092317194e-06,
+      "loss": 0.6885,
+      "step": 22878
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7753460553750868e-06,
+      "loss": 0.6992,
+      "step": 22879
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7746912965741692e-06,
+      "loss": 0.6953,
+      "step": 22880
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7740366467778647e-06,
+      "loss": 0.6953,
+      "step": 22881
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7733821059948508e-06,
+      "loss": 0.6982,
+      "step": 22882
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7727276742337961e-06,
+      "loss": 0.7324,
+      "step": 22883
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7720733515033817e-06,
+      "loss": 0.6924,
+      "step": 22884
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7714191378122747e-06,
+      "loss": 0.6943,
+      "step": 22885
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7707650331691473e-06,
+      "loss": 0.6943,
+      "step": 22886
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7701110375826603e-06,
+      "loss": 0.7021,
+      "step": 22887
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.76945715106149e-06,
+      "loss": 0.6885,
+      "step": 22888
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7688033736142962e-06,
+      "loss": 0.6992,
+      "step": 22889
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7681497052497464e-06,
+      "loss": 0.7021,
+      "step": 22890
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7674961459764982e-06,
+      "loss": 0.7139,
+      "step": 22891
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7668426958032214e-06,
+      "loss": 0.7002,
+      "step": 22892
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.766189354738569e-06,
+      "loss": 0.7373,
+      "step": 22893
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7655361227912004e-06,
+      "loss": 0.7119,
+      "step": 22894
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7648829999697702e-06,
+      "loss": 0.7129,
+      "step": 22895
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7642299862829381e-06,
+      "loss": 0.6855,
+      "step": 22896
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7635770817393582e-06,
+      "loss": 0.6992,
+      "step": 22897
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7629242863476802e-06,
+      "loss": 0.708,
+      "step": 22898
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7622716001165574e-06,
+      "loss": 0.7334,
+      "step": 22899
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.761619023054637e-06,
+      "loss": 0.707,
+      "step": 22900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.76096655517057e-06,
+      "loss": 0.708,
+      "step": 22901
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.760314196473001e-06,
+      "loss": 0.7314,
+      "step": 22902
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7596619469705755e-06,
+      "loss": 0.6914,
+      "step": 22903
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7590098066719352e-06,
+      "loss": 0.6992,
+      "step": 22904
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7583577755857307e-06,
+      "loss": 0.7109,
+      "step": 22905
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.757705853720595e-06,
+      "loss": 0.7227,
+      "step": 22906
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7570540410851665e-06,
+      "loss": 0.6934,
+      "step": 22907
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.756402337688089e-06,
+      "loss": 0.707,
+      "step": 22908
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7557507435379971e-06,
+      "loss": 0.707,
+      "step": 22909
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7550992586435257e-06,
+      "loss": 0.7275,
+      "step": 22910
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7544478830133082e-06,
+      "loss": 0.7119,
+      "step": 22911
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7537966166559773e-06,
+      "loss": 0.709,
+      "step": 22912
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7531454595801634e-06,
+      "loss": 0.6836,
+      "step": 22913
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7524944117944965e-06,
+      "loss": 0.7109,
+      "step": 22914
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.751843473307603e-06,
+      "loss": 0.7109,
+      "step": 22915
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7511926441281079e-06,
+      "loss": 0.7031,
+      "step": 22916
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7505419242646415e-06,
+      "loss": 0.6875,
+      "step": 22917
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7498913137258267e-06,
+      "loss": 0.6934,
+      "step": 22918
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7492408125202787e-06,
+      "loss": 0.7051,
+      "step": 22919
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7485904206566261e-06,
+      "loss": 0.708,
+      "step": 22920
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7479401381434835e-06,
+      "loss": 0.6963,
+      "step": 22921
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7472899649894715e-06,
+      "loss": 0.6924,
+      "step": 22922
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7466399012032042e-06,
+      "loss": 0.7139,
+      "step": 22923
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7459899467932951e-06,
+      "loss": 0.6973,
+      "step": 22924
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7453401017683647e-06,
+      "loss": 0.7041,
+      "step": 22925
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7446903661370173e-06,
+      "loss": 0.6865,
+      "step": 22926
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7440407399078662e-06,
+      "loss": 0.7178,
+      "step": 22927
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7433912230895178e-06,
+      "loss": 0.7324,
+      "step": 22928
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7427418156905851e-06,
+      "loss": 0.6992,
+      "step": 22929
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7420925177196723e-06,
+      "loss": 0.7236,
+      "step": 22930
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.741443329185385e-06,
+      "loss": 0.6914,
+      "step": 22931
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7407942500963193e-06,
+      "loss": 0.71,
+      "step": 22932
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7401452804610864e-06,
+      "loss": 0.7002,
+      "step": 22933
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7394964202882813e-06,
+      "loss": 0.7021,
+      "step": 22934
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7388476695865053e-06,
+      "loss": 0.6934,
+      "step": 22935
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7381990283643523e-06,
+      "loss": 0.7217,
+      "step": 22936
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7375504966304247e-06,
+      "loss": 0.7002,
+      "step": 22937
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.736902074393314e-06,
+      "loss": 0.6953,
+      "step": 22938
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7362537616616116e-06,
+      "loss": 0.7061,
+      "step": 22939
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7356055584439069e-06,
+      "loss": 0.7139,
+      "step": 22940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7349574647487966e-06,
+      "loss": 0.71,
+      "step": 22941
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.734309480584866e-06,
+      "loss": 0.7207,
+      "step": 22942
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7336616059607026e-06,
+      "loss": 0.6963,
+      "step": 22943
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.733013840884893e-06,
+      "loss": 0.6904,
+      "step": 22944
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.73236618536602e-06,
+      "loss": 0.7236,
+      "step": 22945
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7317186394126683e-06,
+      "loss": 0.6865,
+      "step": 22946
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7310712030334176e-06,
+      "loss": 0.7178,
+      "step": 22947
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7304238762368497e-06,
+      "loss": 0.7217,
+      "step": 22948
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7297766590315402e-06,
+      "loss": 0.7119,
+      "step": 22949
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.729129551426071e-06,
+      "loss": 0.7139,
+      "step": 22950
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7284825534290183e-06,
+      "loss": 0.7373,
+      "step": 22951
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.727835665048948e-06,
+      "loss": 0.7178,
+      "step": 22952
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7271888862944408e-06,
+      "loss": 0.7334,
+      "step": 22953
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7265422171740653e-06,
+      "loss": 0.6973,
+      "step": 22954
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7258956576963914e-06,
+      "loss": 0.7295,
+      "step": 22955
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7252492078699879e-06,
+      "loss": 0.6924,
+      "step": 22956
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.724602867703421e-06,
+      "loss": 0.7158,
+      "step": 22957
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7239566372052573e-06,
+      "loss": 0.709,
+      "step": 22958
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7233105163840614e-06,
+      "loss": 0.6973,
+      "step": 22959
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7226645052483936e-06,
+      "loss": 0.6973,
+      "step": 22960
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7220186038068144e-06,
+      "loss": 0.6973,
+      "step": 22961
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7213728120678874e-06,
+      "loss": 0.6992,
+      "step": 22962
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7207271300401695e-06,
+      "loss": 0.708,
+      "step": 22963
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7200815577322182e-06,
+      "loss": 0.7041,
+      "step": 22964
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7194360951525834e-06,
+      "loss": 0.7129,
+      "step": 22965
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.718790742309826e-06,
+      "loss": 0.6963,
+      "step": 22966
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7181454992124957e-06,
+      "loss": 0.7197,
+      "step": 22967
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7175003658691424e-06,
+      "loss": 0.6973,
+      "step": 22968
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.716855342288315e-06,
+      "loss": 0.6719,
+      "step": 22969
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7162104284785674e-06,
+      "loss": 0.7256,
+      "step": 22970
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7155656244484408e-06,
+      "loss": 0.6963,
+      "step": 22971
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7149209302064806e-06,
+      "loss": 0.7109,
+      "step": 22972
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7142763457612288e-06,
+      "loss": 0.7168,
+      "step": 22973
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.713631871121233e-06,
+      "loss": 0.6904,
+      "step": 22974
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7129875062950319e-06,
+      "loss": 0.708,
+      "step": 22975
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7123432512911652e-06,
+      "loss": 0.7168,
+      "step": 22976
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7116991061181664e-06,
+      "loss": 0.6846,
+      "step": 22977
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7110550707845775e-06,
+      "loss": 0.7236,
+      "step": 22978
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7104111452989313e-06,
+      "loss": 0.6963,
+      "step": 22979
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7097673296697603e-06,
+      "loss": 0.709,
+      "step": 22980
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7091236239055963e-06,
+      "loss": 0.7061,
+      "step": 22981
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7084800280149739e-06,
+      "loss": 0.7031,
+      "step": 22982
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7078365420064213e-06,
+      "loss": 0.7168,
+      "step": 22983
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7071931658884633e-06,
+      "loss": 0.7021,
+      "step": 22984
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.706549899669624e-06,
+      "loss": 0.7139,
+      "step": 22985
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.705906743358433e-06,
+      "loss": 0.6797,
+      "step": 22986
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.705263696963414e-06,
+      "loss": 0.6914,
+      "step": 22987
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7046207604930865e-06,
+      "loss": 0.7021,
+      "step": 22988
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7039779339559703e-06,
+      "loss": 0.6895,
+      "step": 22989
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7033352173605866e-06,
+      "loss": 0.7119,
+      "step": 22990
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7026926107154518e-06,
+      "loss": 0.7109,
+      "step": 22991
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7020501140290812e-06,
+      "loss": 0.7139,
+      "step": 22992
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7014077273099904e-06,
+      "loss": 0.7178,
+      "step": 22993
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7007654505666904e-06,
+      "loss": 0.6973,
+      "step": 22994
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7001232838076975e-06,
+      "loss": 0.7021,
+      "step": 22995
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6994812270415207e-06,
+      "loss": 0.6992,
+      "step": 22996
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6988392802766618e-06,
+      "loss": 0.7139,
+      "step": 22997
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6981974435216374e-06,
+      "loss": 0.7178,
+      "step": 22998
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6975557167849488e-06,
+      "loss": 0.7061,
+      "step": 22999
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.696914100075101e-06,
+      "loss": 0.6992,
+      "step": 23000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6962725934005964e-06,
+      "loss": 0.6855,
+      "step": 23001
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.695631196769938e-06,
+      "loss": 0.6982,
+      "step": 23002
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6949899101916233e-06,
+      "loss": 0.7109,
+      "step": 23003
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6943487336741538e-06,
+      "loss": 0.6904,
+      "step": 23004
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6937076672260256e-06,
+      "loss": 0.7051,
+      "step": 23005
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6930667108557297e-06,
+      "loss": 0.6953,
+      "step": 23006
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6924258645717683e-06,
+      "loss": 0.6943,
+      "step": 23007
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6917851283826292e-06,
+      "loss": 0.7012,
+      "step": 23008
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6911445022968087e-06,
+      "loss": 0.7246,
+      "step": 23009
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6905039863227867e-06,
+      "loss": 0.7646,
+      "step": 23010
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6898635804690599e-06,
+      "loss": 0.7012,
+      "step": 23011
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6892232847441126e-06,
+      "loss": 0.71,
+      "step": 23012
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6885830991564311e-06,
+      "loss": 0.7197,
+      "step": 23013
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6879430237144957e-06,
+      "loss": 0.707,
+      "step": 23014
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6873030584267968e-06,
+      "loss": 0.7158,
+      "step": 23015
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6866632033018081e-06,
+      "loss": 0.7168,
+      "step": 23016
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6860234583480117e-06,
+      "loss": 0.7217,
+      "step": 23017
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6853838235738829e-06,
+      "loss": 0.71,
+      "step": 23018
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6847442989879037e-06,
+      "loss": 0.6738,
+      "step": 23019
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6841048845985463e-06,
+      "loss": 0.6855,
+      "step": 23020
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6834655804142852e-06,
+      "loss": 0.7197,
+      "step": 23021
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6828263864435912e-06,
+      "loss": 0.7168,
+      "step": 23022
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6821873026949377e-06,
+      "loss": 0.6768,
+      "step": 23023
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.681548329176792e-06,
+      "loss": 0.7256,
+      "step": 23024
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6809094658976222e-06,
+      "loss": 0.6953,
+      "step": 23025
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6802707128658924e-06,
+      "loss": 0.7051,
+      "step": 23026
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6796320700900725e-06,
+      "loss": 0.7246,
+      "step": 23027
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.678993537578627e-06,
+      "loss": 0.709,
+      "step": 23028
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6783551153400124e-06,
+      "loss": 0.6934,
+      "step": 23029
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6777168033826885e-06,
+      "loss": 0.7324,
+      "step": 23030
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6770786017151196e-06,
+      "loss": 0.7168,
+      "step": 23031
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6764405103457626e-06,
+      "loss": 0.6953,
+      "step": 23032
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6758025292830715e-06,
+      "loss": 0.709,
+      "step": 23033
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6751646585355018e-06,
+      "loss": 0.7012,
+      "step": 23034
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.674526898111507e-06,
+      "loss": 0.7285,
+      "step": 23035
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.673889248019539e-06,
+      "loss": 0.7344,
+      "step": 23036
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6732517082680477e-06,
+      "loss": 0.7207,
+      "step": 23037
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.672614278865483e-06,
+      "loss": 0.6914,
+      "step": 23038
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6719769598202883e-06,
+      "loss": 0.6943,
+      "step": 23039
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6713397511409157e-06,
+      "loss": 0.7178,
+      "step": 23040
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6707026528358094e-06,
+      "loss": 0.707,
+      "step": 23041
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.670065664913405e-06,
+      "loss": 0.6797,
+      "step": 23042
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6694287873821514e-06,
+      "loss": 0.6797,
+      "step": 23043
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.668792020250487e-06,
+      "loss": 0.6982,
+      "step": 23044
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6681553635268487e-06,
+      "loss": 0.6787,
+      "step": 23045
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6675188172196755e-06,
+      "loss": 0.707,
+      "step": 23046
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6668823813374001e-06,
+      "loss": 0.7012,
+      "step": 23047
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6662460558884642e-06,
+      "loss": 0.7178,
+      "step": 23048
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6656098408812926e-06,
+      "loss": 0.7207,
+      "step": 23049
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.664973736324319e-06,
+      "loss": 0.7119,
+      "step": 23050
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.664337742225972e-06,
+      "loss": 0.6885,
+      "step": 23051
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.663701858594684e-06,
+      "loss": 0.7021,
+      "step": 23052
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6630660854388803e-06,
+      "loss": 0.7158,
+      "step": 23053
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6624304227669875e-06,
+      "loss": 0.7051,
+      "step": 23054
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6617948705874243e-06,
+      "loss": 0.6992,
+      "step": 23055
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6611594289086187e-06,
+      "loss": 0.7012,
+      "step": 23056
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6605240977389891e-06,
+      "loss": 0.6885,
+      "step": 23057
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.659888877086957e-06,
+      "loss": 0.7227,
+      "step": 23058
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6592537669609376e-06,
+      "loss": 0.6982,
+      "step": 23059
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.658618767369351e-06,
+      "loss": 0.6904,
+      "step": 23060
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6579838783206136e-06,
+      "loss": 0.6846,
+      "step": 23061
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6573490998231345e-06,
+      "loss": 0.7061,
+      "step": 23062
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6567144318853257e-06,
+      "loss": 0.6992,
+      "step": 23063
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6560798745156037e-06,
+      "loss": 0.7021,
+      "step": 23064
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.655445427722373e-06,
+      "loss": 0.7275,
+      "step": 23065
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6548110915140447e-06,
+      "loss": 0.7002,
+      "step": 23066
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6541768658990232e-06,
+      "loss": 0.7158,
+      "step": 23067
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6535427508857127e-06,
+      "loss": 0.708,
+      "step": 23068
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.65290874648252e-06,
+      "loss": 0.6973,
+      "step": 23069
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6522748526978439e-06,
+      "loss": 0.7197,
+      "step": 23070
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6516410695400864e-06,
+      "loss": 0.7129,
+      "step": 23071
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6510073970176433e-06,
+      "loss": 0.7158,
+      "step": 23072
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6503738351389187e-06,
+      "loss": 0.708,
+      "step": 23073
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6497403839123071e-06,
+      "loss": 0.6973,
+      "step": 23074
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6491070433461974e-06,
+      "loss": 0.7119,
+      "step": 23075
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6484738134489886e-06,
+      "loss": 0.6963,
+      "step": 23076
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6478406942290715e-06,
+      "loss": 0.6953,
+      "step": 23077
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.647207685694835e-06,
+      "loss": 0.6934,
+      "step": 23078
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.64657478785467e-06,
+      "loss": 0.7031,
+      "step": 23079
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6459420007169614e-06,
+      "loss": 0.7139,
+      "step": 23080
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6453093242900974e-06,
+      "loss": 0.7256,
+      "step": 23081
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6446767585824608e-06,
+      "loss": 0.708,
+      "step": 23082
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6440443036024345e-06,
+      "loss": 0.6963,
+      "step": 23083
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6434119593583986e-06,
+      "loss": 0.6885,
+      "step": 23084
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6427797258587375e-06,
+      "loss": 0.7012,
+      "step": 23085
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6421476031118277e-06,
+      "loss": 0.6943,
+      "step": 23086
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6415155911260461e-06,
+      "loss": 0.7178,
+      "step": 23087
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6408836899097691e-06,
+      "loss": 0.7178,
+      "step": 23088
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.640251899471369e-06,
+      "loss": 0.7139,
+      "step": 23089
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.63962021981922e-06,
+      "loss": 0.708,
+      "step": 23090
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6389886509616926e-06,
+      "loss": 0.6914,
+      "step": 23091
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.638357192907154e-06,
+      "loss": 0.707,
+      "step": 23092
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6377258456639811e-06,
+      "loss": 0.7129,
+      "step": 23093
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6370946092405326e-06,
+      "loss": 0.7217,
+      "step": 23094
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6364634836451754e-06,
+      "loss": 0.6924,
+      "step": 23095
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6358324688862725e-06,
+      "loss": 0.6914,
+      "step": 23096
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.635201564972191e-06,
+      "loss": 0.7139,
+      "step": 23097
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.634570771911288e-06,
+      "loss": 0.7041,
+      "step": 23098
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.633940089711925e-06,
+      "loss": 0.7178,
+      "step": 23099
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6333095183824577e-06,
+      "loss": 0.7158,
+      "step": 23100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6326790579312445e-06,
+      "loss": 0.7148,
+      "step": 23101
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6320487083666404e-06,
+      "loss": 0.7197,
+      "step": 23102
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.631418469696997e-06,
+      "loss": 0.7129,
+      "step": 23103
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.630788341930667e-06,
+      "loss": 0.7168,
+      "step": 23104
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6301583250760034e-06,
+      "loss": 0.7129,
+      "step": 23105
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6295284191413574e-06,
+      "loss": 0.7041,
+      "step": 23106
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.62889862413507e-06,
+      "loss": 0.709,
+      "step": 23107
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6282689400654883e-06,
+      "loss": 0.7051,
+      "step": 23108
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.627639366940963e-06,
+      "loss": 0.7061,
+      "step": 23109
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6270099047698329e-06,
+      "loss": 0.7236,
+      "step": 23110
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6263805535604415e-06,
+      "loss": 0.6904,
+      "step": 23111
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6257513133211289e-06,
+      "loss": 0.7061,
+      "step": 23112
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6251221840602326e-06,
+      "loss": 0.7168,
+      "step": 23113
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.624493165786093e-06,
+      "loss": 0.7031,
+      "step": 23114
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.623864258507043e-06,
+      "loss": 0.6621,
+      "step": 23115
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6232354622314184e-06,
+      "loss": 0.71,
+      "step": 23116
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6226067769675492e-06,
+      "loss": 0.7266,
+      "step": 23117
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.621978202723773e-06,
+      "loss": 0.6904,
+      "step": 23118
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6213497395084188e-06,
+      "loss": 0.7031,
+      "step": 23119
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.620721387329809e-06,
+      "loss": 0.6836,
+      "step": 23120
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6200931461962777e-06,
+      "loss": 0.7324,
+      "step": 23121
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6194650161161463e-06,
+      "loss": 0.709,
+      "step": 23122
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6188369970977414e-06,
+      "loss": 0.7178,
+      "step": 23123
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6182090891493851e-06,
+      "loss": 0.6816,
+      "step": 23124
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6175812922793954e-06,
+      "loss": 0.71,
+      "step": 23125
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6169536064961e-06,
+      "loss": 0.6855,
+      "step": 23126
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.61632603180781e-06,
+      "loss": 0.6768,
+      "step": 23127
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6156985682228444e-06,
+      "loss": 0.6904,
+      "step": 23128
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6150712157495163e-06,
+      "loss": 0.7568,
+      "step": 23129
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6144439743961438e-06,
+      "loss": 0.6924,
+      "step": 23130
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6138168441710378e-06,
+      "loss": 0.6875,
+      "step": 23131
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6131898250825096e-06,
+      "loss": 0.6973,
+      "step": 23132
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.612562917138867e-06,
+      "loss": 0.7246,
+      "step": 23133
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6119361203484197e-06,
+      "loss": 0.6924,
+      "step": 23134
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6113094347194724e-06,
+      "loss": 0.6973,
+      "step": 23135
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.610682860260332e-06,
+      "loss": 0.6904,
+      "step": 23136
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6100563969792993e-06,
+      "loss": 0.6953,
+      "step": 23137
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.60943004488468e-06,
+      "loss": 0.6973,
+      "step": 23138
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6088038039847765e-06,
+      "loss": 0.6992,
+      "step": 23139
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6081776742878817e-06,
+      "loss": 0.6826,
+      "step": 23140
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.607551655802294e-06,
+      "loss": 0.7129,
+      "step": 23141
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6069257485363144e-06,
+      "loss": 0.6914,
+      "step": 23142
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6062999524982347e-06,
+      "loss": 0.6982,
+      "step": 23143
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6056742676963478e-06,
+      "loss": 0.7012,
+      "step": 23144
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6050486941389476e-06,
+      "loss": 0.7119,
+      "step": 23145
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6044232318343223e-06,
+      "loss": 0.7256,
+      "step": 23146
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6037978807907617e-06,
+      "loss": 0.6846,
+      "step": 23147
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6031726410165526e-06,
+      "loss": 0.6973,
+      "step": 23148
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6025475125199796e-06,
+      "loss": 0.707,
+      "step": 23149
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6019224953093303e-06,
+      "loss": 0.7168,
+      "step": 23150
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6012975893928861e-06,
+      "loss": 0.6982,
+      "step": 23151
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6006727947789313e-06,
+      "loss": 0.7178,
+      "step": 23152
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6000481114757382e-06,
+      "loss": 0.7236,
+      "step": 23153
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5994235394915936e-06,
+      "loss": 0.709,
+      "step": 23154
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5987990788347697e-06,
+      "loss": 0.7168,
+      "step": 23155
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5981747295135441e-06,
+      "loss": 0.6738,
+      "step": 23156
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5975504915361895e-06,
+      "loss": 0.6943,
+      "step": 23157
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5969263649109802e-06,
+      "loss": 0.7158,
+      "step": 23158
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.596302349646186e-06,
+      "loss": 0.7227,
+      "step": 23159
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5956784457500774e-06,
+      "loss": 0.7168,
+      "step": 23160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5950546532309219e-06,
+      "loss": 0.7295,
+      "step": 23161
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.594430972096983e-06,
+      "loss": 0.7305,
+      "step": 23162
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5938074023565332e-06,
+      "loss": 0.7041,
+      "step": 23163
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.593183944017832e-06,
+      "loss": 0.6885,
+      "step": 23164
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5925605970891423e-06,
+      "loss": 0.6973,
+      "step": 23165
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5919373615787247e-06,
+      "loss": 0.7021,
+      "step": 23166
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5913142374948398e-06,
+      "loss": 0.6807,
+      "step": 23167
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.590691224845743e-06,
+      "loss": 0.6729,
+      "step": 23168
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5900683236396918e-06,
+      "loss": 0.6992,
+      "step": 23169
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.589445533884939e-06,
+      "loss": 0.6865,
+      "step": 23170
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5888228555897456e-06,
+      "loss": 0.6982,
+      "step": 23171
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.588200288762355e-06,
+      "loss": 0.6973,
+      "step": 23172
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5875778334110215e-06,
+      "loss": 0.6973,
+      "step": 23173
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5869554895439908e-06,
+      "loss": 0.7197,
+      "step": 23174
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5863332571695156e-06,
+      "loss": 0.6943,
+      "step": 23175
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5857111362958399e-06,
+      "loss": 0.7002,
+      "step": 23176
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5850891269312075e-06,
+      "loss": 0.7031,
+      "step": 23177
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5844672290838614e-06,
+      "loss": 0.7051,
+      "step": 23178
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5838454427620443e-06,
+      "loss": 0.6729,
+      "step": 23179
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5832237679739948e-06,
+      "loss": 0.7295,
+      "step": 23180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5826022047279533e-06,
+      "loss": 0.7236,
+      "step": 23181
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.581980753032152e-06,
+      "loss": 0.6836,
+      "step": 23182
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.581359412894834e-06,
+      "loss": 0.6982,
+      "step": 23183
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5807381843242332e-06,
+      "loss": 0.6904,
+      "step": 23184
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.580117067328576e-06,
+      "loss": 0.6895,
+      "step": 23185
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5794960619160948e-06,
+      "loss": 0.6855,
+      "step": 23186
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.578875168095023e-06,
+      "loss": 0.6729,
+      "step": 23187
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5782543858735887e-06,
+      "loss": 0.7373,
+      "step": 23188
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5776337152600162e-06,
+      "loss": 0.7041,
+      "step": 23189
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5770131562625334e-06,
+      "loss": 0.6914,
+      "step": 23190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5763927088893616e-06,
+      "loss": 0.6982,
+      "step": 23191
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.575772373148725e-06,
+      "loss": 0.7061,
+      "step": 23192
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.575152149048844e-06,
+      "loss": 0.7227,
+      "step": 23193
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5745320365979355e-06,
+      "loss": 0.7119,
+      "step": 23194
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5739120358042214e-06,
+      "loss": 0.707,
+      "step": 23195
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5732921466759176e-06,
+      "loss": 0.7021,
+      "step": 23196
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5726723692212397e-06,
+      "loss": 0.707,
+      "step": 23197
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5720527034483957e-06,
+      "loss": 0.6934,
+      "step": 23198
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5714331493656031e-06,
+      "loss": 0.7158,
+      "step": 23199
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5708137069810715e-06,
+      "loss": 0.7373,
+      "step": 23200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5701943763030093e-06,
+      "loss": 0.7314,
+      "step": 23201
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5695751573396246e-06,
+      "loss": 0.707,
+      "step": 23202
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5689560500991219e-06,
+      "loss": 0.7012,
+      "step": 23203
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.568337054589707e-06,
+      "loss": 0.6982,
+      "step": 23204
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5677181708195832e-06,
+      "loss": 0.6963,
+      "step": 23205
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.567099398796953e-06,
+      "loss": 0.6992,
+      "step": 23206
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5664807385300118e-06,
+      "loss": 0.6953,
+      "step": 23207
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5658621900269644e-06,
+      "loss": 0.6895,
+      "step": 23208
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5652437532960063e-06,
+      "loss": 0.7031,
+      "step": 23209
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.564625428345332e-06,
+      "loss": 0.7305,
+      "step": 23210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5640072151831376e-06,
+      "loss": 0.6875,
+      "step": 23211
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5633891138176137e-06,
+      "loss": 0.7109,
+      "step": 23212
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.562771124256952e-06,
+      "loss": 0.6982,
+      "step": 23213
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5621532465093436e-06,
+      "loss": 0.7275,
+      "step": 23214
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.561535480582973e-06,
+      "loss": 0.7041,
+      "step": 23215
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5609178264860347e-06,
+      "loss": 0.6934,
+      "step": 23216
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5603002842267078e-06,
+      "loss": 0.707,
+      "step": 23217
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5596828538131781e-06,
+      "loss": 0.7031,
+      "step": 23218
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5590655352536244e-06,
+      "loss": 0.707,
+      "step": 23219
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5584483285562336e-06,
+      "loss": 0.7314,
+      "step": 23220
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5578312337291823e-06,
+      "loss": 0.7275,
+      "step": 23221
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5572142507806498e-06,
+      "loss": 0.71,
+      "step": 23222
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5565973797188095e-06,
+      "loss": 0.6953,
+      "step": 23223
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5559806205518403e-06,
+      "loss": 0.7227,
+      "step": 23224
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5553639732879123e-06,
+      "loss": 0.6875,
+      "step": 23225
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.554747437935199e-06,
+      "loss": 0.7178,
+      "step": 23226
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5541310145018684e-06,
+      "loss": 0.7188,
+      "step": 23227
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.553514702996095e-06,
+      "loss": 0.7207,
+      "step": 23228
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5528985034260457e-06,
+      "loss": 0.6943,
+      "step": 23229
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5522824157998818e-06,
+      "loss": 0.7051,
+      "step": 23230
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5516664401257676e-06,
+      "loss": 0.6953,
+      "step": 23231
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.551050576411871e-06,
+      "loss": 0.709,
+      "step": 23232
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5504348246663525e-06,
+      "loss": 0.6875,
+      "step": 23233
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.549819184897371e-06,
+      "loss": 0.7031,
+      "step": 23234
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5492036571130865e-06,
+      "loss": 0.7139,
+      "step": 23235
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5485882413216535e-06,
+      "loss": 0.6914,
+      "step": 23236
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5479729375312303e-06,
+      "loss": 0.7158,
+      "step": 23237
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5473577457499712e-06,
+      "loss": 0.6855,
+      "step": 23238
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5467426659860262e-06,
+      "loss": 0.6992,
+      "step": 23239
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.546127698247547e-06,
+      "loss": 0.7256,
+      "step": 23240
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5455128425426869e-06,
+      "loss": 0.7178,
+      "step": 23241
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5448980988795947e-06,
+      "loss": 0.6953,
+      "step": 23242
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5442834672664086e-06,
+      "loss": 0.7012,
+      "step": 23243
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5436689477112833e-06,
+      "loss": 0.6904,
+      "step": 23244
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5430545402223584e-06,
+      "loss": 0.6895,
+      "step": 23245
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.542440244807777e-06,
+      "loss": 0.71,
+      "step": 23246
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5418260614756808e-06,
+      "loss": 0.7207,
+      "step": 23247
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5412119902342049e-06,
+      "loss": 0.71,
+      "step": 23248
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.540598031091496e-06,
+      "loss": 0.7139,
+      "step": 23249
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5399841840556817e-06,
+      "loss": 0.6797,
+      "step": 23250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5393704491349005e-06,
+      "loss": 0.7051,
+      "step": 23251
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5387568263372844e-06,
+      "loss": 0.6914,
+      "step": 23252
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5381433156709679e-06,
+      "loss": 0.7002,
+      "step": 23253
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5375299171440793e-06,
+      "loss": 0.7021,
+      "step": 23254
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5369166307647499e-06,
+      "loss": 0.7031,
+      "step": 23255
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.536303456541105e-06,
+      "loss": 0.6914,
+      "step": 23256
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5356903944812706e-06,
+      "loss": 0.7031,
+      "step": 23257
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5350774445933725e-06,
+      "loss": 0.7041,
+      "step": 23258
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.534464606885533e-06,
+      "loss": 0.6797,
+      "step": 23259
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5338518813658709e-06,
+      "loss": 0.6875,
+      "step": 23260
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5332392680425123e-06,
+      "loss": 0.6982,
+      "step": 23261
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5326267669235739e-06,
+      "loss": 0.6875,
+      "step": 23262
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5320143780171693e-06,
+      "loss": 0.7285,
+      "step": 23263
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5314021013314139e-06,
+      "loss": 0.7168,
+      "step": 23264
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5307899368744272e-06,
+      "loss": 0.6992,
+      "step": 23265
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5301778846543169e-06,
+      "loss": 0.7002,
+      "step": 23266
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5295659446791977e-06,
+      "loss": 0.7051,
+      "step": 23267
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5289541169571763e-06,
+      "loss": 0.6904,
+      "step": 23268
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5283424014963622e-06,
+      "loss": 0.6836,
+      "step": 23269
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5277307983048617e-06,
+      "loss": 0.707,
+      "step": 23270
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5271193073907808e-06,
+      "loss": 0.6875,
+      "step": 23271
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5265079287622198e-06,
+      "loss": 0.6953,
+      "step": 23272
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5258966624272854e-06,
+      "loss": 0.7012,
+      "step": 23273
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5252855083940776e-06,
+      "loss": 0.6992,
+      "step": 23274
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5246744666706958e-06,
+      "loss": 0.708,
+      "step": 23275
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5240635372652324e-06,
+      "loss": 0.6797,
+      "step": 23276
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5234527201857907e-06,
+      "loss": 0.6592,
+      "step": 23277
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.522842015440461e-06,
+      "loss": 0.7148,
+      "step": 23278
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.522231423037339e-06,
+      "loss": 0.6875,
+      "step": 23279
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5216209429845152e-06,
+      "loss": 0.708,
+      "step": 23280
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5210105752900805e-06,
+      "loss": 0.7109,
+      "step": 23281
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5204003199621232e-06,
+      "loss": 0.7129,
+      "step": 23282
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5197901770087309e-06,
+      "loss": 0.665,
+      "step": 23283
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5191801464379897e-06,
+      "loss": 0.6992,
+      "step": 23284
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5185702282579806e-06,
+      "loss": 0.707,
+      "step": 23285
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.517960422476793e-06,
+      "loss": 0.7236,
+      "step": 23286
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5173507291025037e-06,
+      "loss": 0.7207,
+      "step": 23287
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5167411481431937e-06,
+      "loss": 0.7021,
+      "step": 23288
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5161316796069415e-06,
+      "loss": 0.6904,
+      "step": 23289
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5155223235018235e-06,
+      "loss": 0.6992,
+      "step": 23290
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5149130798359146e-06,
+      "loss": 0.7178,
+      "step": 23291
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5143039486172905e-06,
+      "loss": 0.6992,
+      "step": 23292
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5136949298540194e-06,
+      "loss": 0.6768,
+      "step": 23293
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.513086023554181e-06,
+      "loss": 0.7305,
+      "step": 23294
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.512477229725836e-06,
+      "loss": 0.7051,
+      "step": 23295
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5118685483770556e-06,
+      "loss": 0.6875,
+      "step": 23296
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5112599795159045e-06,
+      "loss": 0.6973,
+      "step": 23297
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5106515231504504e-06,
+      "loss": 0.7012,
+      "step": 23298
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.510043179288756e-06,
+      "loss": 0.7158,
+      "step": 23299
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5094349479388836e-06,
+      "loss": 0.6865,
+      "step": 23300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5088268291088925e-06,
+      "loss": 0.6973,
+      "step": 23301
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5082188228068418e-06,
+      "loss": 0.7119,
+      "step": 23302
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5076109290407903e-06,
+      "loss": 0.6904,
+      "step": 23303
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5070031478187929e-06,
+      "loss": 0.7139,
+      "step": 23304
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5063954791489021e-06,
+      "loss": 0.708,
+      "step": 23305
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5057879230391748e-06,
+      "loss": 0.6963,
+      "step": 23306
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5051804794976643e-06,
+      "loss": 0.708,
+      "step": 23307
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5045731485324144e-06,
+      "loss": 0.6953,
+      "step": 23308
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5039659301514743e-06,
+      "loss": 0.7041,
+      "step": 23309
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.503358824362895e-06,
+      "loss": 0.6904,
+      "step": 23310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5027518311747214e-06,
+      "loss": 0.7529,
+      "step": 23311
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.502144950594996e-06,
+      "loss": 0.6973,
+      "step": 23312
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.501538182631762e-06,
+      "loss": 0.7041,
+      "step": 23313
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5009315272930603e-06,
+      "loss": 0.7021,
+      "step": 23314
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5003249845869306e-06,
+      "loss": 0.6943,
+      "step": 23315
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.499718554521412e-06,
+      "loss": 0.6836,
+      "step": 23316
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4991122371045374e-06,
+      "loss": 0.7041,
+      "step": 23317
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4985060323443468e-06,
+      "loss": 0.7158,
+      "step": 23318
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4978999402488715e-06,
+      "loss": 0.6973,
+      "step": 23319
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4972939608261461e-06,
+      "loss": 0.7383,
+      "step": 23320
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4966880940841955e-06,
+      "loss": 0.7109,
+      "step": 23321
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4960823400310542e-06,
+      "loss": 0.7422,
+      "step": 23322
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4954766986747482e-06,
+      "loss": 0.6953,
+      "step": 23323
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4948711700233033e-06,
+      "loss": 0.7012,
+      "step": 23324
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4942657540847438e-06,
+      "loss": 0.6875,
+      "step": 23325
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4936604508670916e-06,
+      "loss": 0.7266,
+      "step": 23326
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4930552603783754e-06,
+      "loss": 0.6807,
+      "step": 23327
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4924501826266068e-06,
+      "loss": 0.6992,
+      "step": 23328
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4918452176198094e-06,
+      "loss": 0.7197,
+      "step": 23329
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4912403653659957e-06,
+      "loss": 0.7021,
+      "step": 23330
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.490635625873187e-06,
+      "loss": 0.7344,
+      "step": 23331
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4900309991493955e-06,
+      "loss": 0.6973,
+      "step": 23332
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4894264852026341e-06,
+      "loss": 0.7188,
+      "step": 23333
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4888220840409129e-06,
+      "loss": 0.7188,
+      "step": 23334
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.488217795672242e-06,
+      "loss": 0.6836,
+      "step": 23335
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4876136201046298e-06,
+      "loss": 0.7178,
+      "step": 23336
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.487009557346084e-06,
+      "loss": 0.7168,
+      "step": 23337
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4864056074046062e-06,
+      "loss": 0.7188,
+      "step": 23338
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4858017702882054e-06,
+      "loss": 0.6924,
+      "step": 23339
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4851980460048832e-06,
+      "loss": 0.7148,
+      "step": 23340
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4845944345626362e-06,
+      "loss": 0.7119,
+      "step": 23341
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.483990935969465e-06,
+      "loss": 0.6924,
+      "step": 23342
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4833875502333694e-06,
+      "loss": 0.6855,
+      "step": 23343
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4827842773623447e-06,
+      "loss": 0.7031,
+      "step": 23344
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4821811173643864e-06,
+      "loss": 0.6865,
+      "step": 23345
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.481578070247487e-06,
+      "loss": 0.6826,
+      "step": 23346
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.480975136019638e-06,
+      "loss": 0.6855,
+      "step": 23347
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4803723146888294e-06,
+      "loss": 0.7031,
+      "step": 23348
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4797696062630506e-06,
+      "loss": 0.6807,
+      "step": 23349
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4791670107502875e-06,
+      "loss": 0.6787,
+      "step": 23350
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4785645281585292e-06,
+      "loss": 0.6992,
+      "step": 23351
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4779621584957582e-06,
+      "loss": 0.6973,
+      "step": 23352
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4773599017699603e-06,
+      "loss": 0.7012,
+      "step": 23353
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4767577579891079e-06,
+      "loss": 0.6885,
+      "step": 23354
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4761557271611893e-06,
+      "loss": 0.7363,
+      "step": 23355
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4755538092941802e-06,
+      "loss": 0.6943,
+      "step": 23356
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4749520043960574e-06,
+      "loss": 0.6787,
+      "step": 23357
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.474350312474796e-06,
+      "loss": 0.7227,
+      "step": 23358
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4737487335383704e-06,
+      "loss": 0.7002,
+      "step": 23359
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4731472675947522e-06,
+      "loss": 0.7109,
+      "step": 23360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4725459146519138e-06,
+      "loss": 0.7061,
+      "step": 23361
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.47194467471782e-06,
+      "loss": 0.7266,
+      "step": 23362
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4713435478004445e-06,
+      "loss": 0.7148,
+      "step": 23363
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4707425339077508e-06,
+      "loss": 0.6973,
+      "step": 23364
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4701416330477037e-06,
+      "loss": 0.6787,
+      "step": 23365
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4695408452282679e-06,
+      "loss": 0.7158,
+      "step": 23366
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4689401704574036e-06,
+      "loss": 0.6953,
+      "step": 23367
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4683396087430724e-06,
+      "loss": 0.6982,
+      "step": 23368
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4677391600932322e-06,
+      "loss": 0.7129,
+      "step": 23369
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4671388245158402e-06,
+      "loss": 0.6934,
+      "step": 23370
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.466538602018851e-06,
+      "loss": 0.6895,
+      "step": 23371
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.465938492610226e-06,
+      "loss": 0.71,
+      "step": 23372
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.465338496297909e-06,
+      "loss": 0.7158,
+      "step": 23373
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4647386130898566e-06,
+      "loss": 0.7207,
+      "step": 23374
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.464138842994014e-06,
+      "loss": 0.6904,
+      "step": 23375
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4635391860183346e-06,
+      "loss": 0.6982,
+      "step": 23376
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4629396421707642e-06,
+      "loss": 0.6865,
+      "step": 23377
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4623402114592467e-06,
+      "loss": 0.7002,
+      "step": 23378
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4617408938917276e-06,
+      "loss": 0.7188,
+      "step": 23379
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4611416894761477e-06,
+      "loss": 0.6875,
+      "step": 23380
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.460542598220448e-06,
+      "loss": 0.7041,
+      "step": 23381
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.459943620132569e-06,
+      "loss": 0.709,
+      "step": 23382
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4593447552204454e-06,
+      "loss": 0.7178,
+      "step": 23383
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4587460034920186e-06,
+      "loss": 0.6875,
+      "step": 23384
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4581473649552225e-06,
+      "loss": 0.7012,
+      "step": 23385
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.457548839617987e-06,
+      "loss": 0.6982,
+      "step": 23386
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.456950427488244e-06,
+      "loss": 0.6807,
+      "step": 23387
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4563521285739268e-06,
+      "loss": 0.6836,
+      "step": 23388
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4557539428829648e-06,
+      "loss": 0.6699,
+      "step": 23389
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4551558704232828e-06,
+      "loss": 0.7158,
+      "step": 23390
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4545579112028073e-06,
+      "loss": 0.7051,
+      "step": 23391
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4539600652294628e-06,
+      "loss": 0.7061,
+      "step": 23392
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4533623325111724e-06,
+      "loss": 0.6982,
+      "step": 23393
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4527647130558575e-06,
+      "loss": 0.7012,
+      "step": 23394
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4521672068714365e-06,
+      "loss": 0.6855,
+      "step": 23395
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4515698139658296e-06,
+      "loss": 0.7129,
+      "step": 23396
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.450972534346955e-06,
+      "loss": 0.7275,
+      "step": 23397
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4503753680227273e-06,
+      "loss": 0.7041,
+      "step": 23398
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4497783150010546e-06,
+      "loss": 0.7012,
+      "step": 23399
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4491813752898565e-06,
+      "loss": 0.7168,
+      "step": 23400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4485845488970408e-06,
+      "loss": 0.7197,
+      "step": 23401
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4479878358305177e-06,
+      "loss": 0.6816,
+      "step": 23402
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4473912360981945e-06,
+      "loss": 0.7061,
+      "step": 23403
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4467947497079783e-06,
+      "loss": 0.6904,
+      "step": 23404
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4461983766677722e-06,
+      "loss": 0.7373,
+      "step": 23405
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4456021169854818e-06,
+      "loss": 0.6836,
+      "step": 23406
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4450059706690044e-06,
+      "loss": 0.6973,
+      "step": 23407
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4444099377262465e-06,
+      "loss": 0.7266,
+      "step": 23408
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.443814018165104e-06,
+      "loss": 0.7148,
+      "step": 23409
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4432182119934745e-06,
+      "loss": 0.707,
+      "step": 23410
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4426225192192545e-06,
+      "loss": 0.7139,
+      "step": 23411
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.442026939850336e-06,
+      "loss": 0.7148,
+      "step": 23412
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4414314738946144e-06,
+      "loss": 0.7129,
+      "step": 23413
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4408361213599808e-06,
+      "loss": 0.6797,
+      "step": 23414
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4402408822543223e-06,
+      "loss": 0.7158,
+      "step": 23415
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4396457565855282e-06,
+      "loss": 0.7051,
+      "step": 23416
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.439050744361491e-06,
+      "loss": 0.6777,
+      "step": 23417
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4384558455900877e-06,
+      "loss": 0.6943,
+      "step": 23418
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4378610602792064e-06,
+      "loss": 0.6797,
+      "step": 23419
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4372663884367256e-06,
+      "loss": 0.6973,
+      "step": 23420
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4366718300705329e-06,
+      "loss": 0.7188,
+      "step": 23421
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4360773851885024e-06,
+      "loss": 0.7002,
+      "step": 23422
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4354830537985144e-06,
+      "loss": 0.6865,
+      "step": 23423
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4348888359084433e-06,
+      "loss": 0.6982,
+      "step": 23424
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4342947315261646e-06,
+      "loss": 0.6973,
+      "step": 23425
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4337007406595516e-06,
+      "loss": 0.6924,
+      "step": 23426
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4331068633164768e-06,
+      "loss": 0.7139,
+      "step": 23427
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4325130995048075e-06,
+      "loss": 0.7236,
+      "step": 23428
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4319194492324162e-06,
+      "loss": 0.71,
+      "step": 23429
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4313259125071721e-06,
+      "loss": 0.7012,
+      "step": 23430
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4307324893369346e-06,
+      "loss": 0.6973,
+      "step": 23431
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4301391797295683e-06,
+      "loss": 0.7188,
+      "step": 23432
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4295459836929414e-06,
+      "loss": 0.7119,
+      "step": 23433
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4289529012349124e-06,
+      "loss": 0.6895,
+      "step": 23434
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4283599323633402e-06,
+      "loss": 0.7188,
+      "step": 23435
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4277670770860842e-06,
+      "loss": 0.7051,
+      "step": 23436
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4271743354110001e-06,
+      "loss": 0.7002,
+      "step": 23437
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4265817073459442e-06,
+      "loss": 0.6924,
+      "step": 23438
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.42598919289877e-06,
+      "loss": 0.6797,
+      "step": 23439
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.425396792077327e-06,
+      "loss": 0.7051,
+      "step": 23440
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4248045048894699e-06,
+      "loss": 0.7207,
+      "step": 23441
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4242123313430467e-06,
+      "loss": 0.7373,
+      "step": 23442
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.423620271445908e-06,
+      "loss": 0.709,
+      "step": 23443
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4230283252058908e-06,
+      "loss": 0.7256,
+      "step": 23444
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4224364926308476e-06,
+      "loss": 0.7178,
+      "step": 23445
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4218447737286211e-06,
+      "loss": 0.709,
+      "step": 23446
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4212531685070496e-06,
+      "loss": 0.71,
+      "step": 23447
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4206616769739766e-06,
+      "loss": 0.7188,
+      "step": 23448
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4200702991372362e-06,
+      "loss": 0.7295,
+      "step": 23449
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.419479035004674e-06,
+      "loss": 0.7324,
+      "step": 23450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.418887884584117e-06,
+      "loss": 0.6904,
+      "step": 23451
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4182968478834036e-06,
+      "loss": 0.6846,
+      "step": 23452
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4177059249103619e-06,
+      "loss": 0.6816,
+      "step": 23453
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.417115115672829e-06,
+      "loss": 0.7129,
+      "step": 23454
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4165244201786333e-06,
+      "loss": 0.707,
+      "step": 23455
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4159338384356002e-06,
+      "loss": 0.7119,
+      "step": 23456
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4153433704515584e-06,
+      "loss": 0.7148,
+      "step": 23457
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4147530162343315e-06,
+      "loss": 0.7285,
+      "step": 23458
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4141627757917453e-06,
+      "loss": 0.7012,
+      "step": 23459
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4135726491316193e-06,
+      "loss": 0.708,
+      "step": 23460
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4129826362617728e-06,
+      "loss": 0.6992,
+      "step": 23461
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4123927371900304e-06,
+      "loss": 0.6943,
+      "step": 23462
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4118029519242082e-06,
+      "loss": 0.7002,
+      "step": 23463
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.411213280472118e-06,
+      "loss": 0.7168,
+      "step": 23464
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4106237228415753e-06,
+      "loss": 0.7139,
+      "step": 23465
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4100342790403965e-06,
+      "loss": 0.7314,
+      "step": 23466
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4094449490763917e-06,
+      "loss": 0.707,
+      "step": 23467
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4088557329573705e-06,
+      "loss": 0.7158,
+      "step": 23468
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4082666306911408e-06,
+      "loss": 0.7031,
+      "step": 23469
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.40767764228551e-06,
+      "loss": 0.6943,
+      "step": 23470
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4070887677482846e-06,
+      "loss": 0.6973,
+      "step": 23471
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.406500007087267e-06,
+      "loss": 0.7129,
+      "step": 23472
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.405911360310258e-06,
+      "loss": 0.6953,
+      "step": 23473
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.405322827425064e-06,
+      "loss": 0.6855,
+      "step": 23474
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4047344084394809e-06,
+      "loss": 0.6943,
+      "step": 23475
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.404146103361309e-06,
+      "loss": 0.6865,
+      "step": 23476
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.403557912198339e-06,
+      "loss": 0.7021,
+      "step": 23477
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.402969834958372e-06,
+      "loss": 0.6895,
+      "step": 23478
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4023818716491977e-06,
+      "loss": 0.7129,
+      "step": 23479
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.401794022278611e-06,
+      "loss": 0.7012,
+      "step": 23480
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4012062868543996e-06,
+      "loss": 0.6992,
+      "step": 23481
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4006186653843535e-06,
+      "loss": 0.7246,
+      "step": 23482
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4000311578762604e-06,
+      "loss": 0.7051,
+      "step": 23483
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3994437643379056e-06,
+      "loss": 0.6816,
+      "step": 23484
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3988564847770703e-06,
+      "loss": 0.7373,
+      "step": 23485
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3982693192015439e-06,
+      "loss": 0.6953,
+      "step": 23486
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3976822676191037e-06,
+      "loss": 0.6836,
+      "step": 23487
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.39709533003753e-06,
+      "loss": 0.6797,
+      "step": 23488
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3965085064646023e-06,
+      "loss": 0.6846,
+      "step": 23489
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.395921796908095e-06,
+      "loss": 0.7021,
+      "step": 23490
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3953352013757859e-06,
+      "loss": 0.7031,
+      "step": 23491
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3947487198754473e-06,
+      "loss": 0.707,
+      "step": 23492
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3941623524148517e-06,
+      "loss": 0.6934,
+      "step": 23493
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3935760990017677e-06,
+      "loss": 0.6904,
+      "step": 23494
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3929899596439711e-06,
+      "loss": 0.7051,
+      "step": 23495
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3924039343492235e-06,
+      "loss": 0.6934,
+      "step": 23496
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.391818023125292e-06,
+      "loss": 0.7012,
+      "step": 23497
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3912322259799405e-06,
+      "loss": 0.6836,
+      "step": 23498
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3906465429209359e-06,
+      "loss": 0.6895,
+      "step": 23499
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.390060973956039e-06,
+      "loss": 0.7197,
+      "step": 23500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3894755190930076e-06,
+      "loss": 0.7139,
+      "step": 23501
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3888901783396015e-06,
+      "loss": 0.7168,
+      "step": 23502
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3883049517035773e-06,
+      "loss": 0.6836,
+      "step": 23503
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3877198391926927e-06,
+      "loss": 0.707,
+      "step": 23504
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.387134840814699e-06,
+      "loss": 0.7354,
+      "step": 23505
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3865499565773488e-06,
+      "loss": 0.7109,
+      "step": 23506
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.385965186488396e-06,
+      "loss": 0.707,
+      "step": 23507
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3853805305555911e-06,
+      "loss": 0.6963,
+      "step": 23508
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.384795988786678e-06,
+      "loss": 0.6982,
+      "step": 23509
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3842115611894014e-06,
+      "loss": 0.6787,
+      "step": 23510
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.383627247771514e-06,
+      "loss": 0.6846,
+      "step": 23511
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3830430485407541e-06,
+      "loss": 0.6953,
+      "step": 23512
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3824589635048647e-06,
+      "loss": 0.6992,
+      "step": 23513
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3818749926715847e-06,
+      "loss": 0.7051,
+      "step": 23514
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3812911360486591e-06,
+      "loss": 0.7139,
+      "step": 23515
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3807073936438187e-06,
+      "loss": 0.7119,
+      "step": 23516
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3801237654648026e-06,
+      "loss": 0.7168,
+      "step": 23517
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3795402515193412e-06,
+      "loss": 0.6846,
+      "step": 23518
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.378956851815173e-06,
+      "loss": 0.6816,
+      "step": 23519
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3783735663600274e-06,
+      "loss": 0.6943,
+      "step": 23520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3777903951616367e-06,
+      "loss": 0.6748,
+      "step": 23521
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3772073382277217e-06,
+      "loss": 0.6777,
+      "step": 23522
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3766243955660164e-06,
+      "loss": 0.7158,
+      "step": 23523
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3760415671842442e-06,
+      "loss": 0.7402,
+      "step": 23524
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.375458853090128e-06,
+      "loss": 0.708,
+      "step": 23525
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3748762532913917e-06,
+      "loss": 0.7061,
+      "step": 23526
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3742937677957524e-06,
+      "loss": 0.7012,
+      "step": 23527
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3737113966109372e-06,
+      "loss": 0.7256,
+      "step": 23528
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3731291397446567e-06,
+      "loss": 0.7236,
+      "step": 23529
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3725469972046268e-06,
+      "loss": 0.6953,
+      "step": 23530
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.371964968998567e-06,
+      "loss": 0.6865,
+      "step": 23531
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3713830551341878e-06,
+      "loss": 0.7002,
+      "step": 23532
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.370801255619202e-06,
+      "loss": 0.6973,
+      "step": 23533
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3702195704613196e-06,
+      "loss": 0.6729,
+      "step": 23534
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3696379996682485e-06,
+      "loss": 0.71,
+      "step": 23535
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3690565432476965e-06,
+      "loss": 0.7441,
+      "step": 23536
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3684752012073687e-06,
+      "loss": 0.6953,
+      "step": 23537
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3678939735549702e-06,
+      "loss": 0.709,
+      "step": 23538
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.367312860298201e-06,
+      "loss": 0.7002,
+      "step": 23539
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3667318614447667e-06,
+      "loss": 0.6973,
+      "step": 23540
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3661509770023663e-06,
+      "loss": 0.6914,
+      "step": 23541
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3655702069786937e-06,
+      "loss": 0.7373,
+      "step": 23542
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.364989551381445e-06,
+      "loss": 0.7227,
+      "step": 23543
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3644090102183216e-06,
+      "loss": 0.6855,
+      "step": 23544
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3638285834970123e-06,
+      "loss": 0.707,
+      "step": 23545
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3632482712252115e-06,
+      "loss": 0.7021,
+      "step": 23546
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.362668073410608e-06,
+      "loss": 0.7031,
+      "step": 23547
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3620879900608908e-06,
+      "loss": 0.7031,
+      "step": 23548
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3615080211837483e-06,
+      "loss": 0.7139,
+      "step": 23549
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3609281667868657e-06,
+      "loss": 0.6953,
+      "step": 23550
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3603484268779266e-06,
+      "loss": 0.6885,
+      "step": 23551
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3597688014646159e-06,
+      "loss": 0.6855,
+      "step": 23552
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3591892905546155e-06,
+      "loss": 0.7109,
+      "step": 23553
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3586098941556047e-06,
+      "loss": 0.7139,
+      "step": 23554
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3580306122752574e-06,
+      "loss": 0.7129,
+      "step": 23555
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.357451444921256e-06,
+      "loss": 0.7324,
+      "step": 23556
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.356872392101275e-06,
+      "loss": 0.6865,
+      "step": 23557
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3562934538229866e-06,
+      "loss": 0.7129,
+      "step": 23558
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3557146300940627e-06,
+      "loss": 0.707,
+      "step": 23559
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.355135920922176e-06,
+      "loss": 0.708,
+      "step": 23560
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3545573263149947e-06,
+      "loss": 0.6885,
+      "step": 23561
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3539788462801862e-06,
+      "loss": 0.6904,
+      "step": 23562
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3534004808254153e-06,
+      "loss": 0.6982,
+      "step": 23563
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3528222299583505e-06,
+      "loss": 0.7158,
+      "step": 23564
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3522440936866533e-06,
+      "loss": 0.7002,
+      "step": 23565
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3516660720179853e-06,
+      "loss": 0.6924,
+      "step": 23566
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.351088164960006e-06,
+      "loss": 0.6865,
+      "step": 23567
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3505103725203738e-06,
+      "loss": 0.7148,
+      "step": 23568
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3499326947067471e-06,
+      "loss": 0.7256,
+      "step": 23569
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3493551315267805e-06,
+      "loss": 0.708,
+      "step": 23570
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3487776829881284e-06,
+      "loss": 0.7285,
+      "step": 23571
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.348200349098442e-06,
+      "loss": 0.7109,
+      "step": 23572
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3476231298653775e-06,
+      "loss": 0.7119,
+      "step": 23573
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3470460252965777e-06,
+      "loss": 0.6875,
+      "step": 23574
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3464690353996924e-06,
+      "loss": 0.7031,
+      "step": 23575
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3458921601823705e-06,
+      "loss": 0.6943,
+      "step": 23576
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.345315399652255e-06,
+      "loss": 0.7041,
+      "step": 23577
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.344738753816991e-06,
+      "loss": 0.7012,
+      "step": 23578
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3441622226842188e-06,
+      "loss": 0.7236,
+      "step": 23579
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3435858062615791e-06,
+      "loss": 0.7334,
+      "step": 23580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3430095045567116e-06,
+      "loss": 0.7246,
+      "step": 23581
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.342433317577253e-06,
+      "loss": 0.7305,
+      "step": 23582
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3418572453308388e-06,
+      "loss": 0.7012,
+      "step": 23583
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3412812878251013e-06,
+      "loss": 0.7139,
+      "step": 23584
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3407054450676782e-06,
+      "loss": 0.6953,
+      "step": 23585
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3401297170661999e-06,
+      "loss": 0.6982,
+      "step": 23586
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3395541038282922e-06,
+      "loss": 0.6973,
+      "step": 23587
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3389786053615838e-06,
+      "loss": 0.6729,
+      "step": 23588
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3384032216737053e-06,
+      "loss": 0.7207,
+      "step": 23589
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3378279527722792e-06,
+      "loss": 0.7002,
+      "step": 23590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3372527986649297e-06,
+      "loss": 0.6982,
+      "step": 23591
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3366777593592772e-06,
+      "loss": 0.7227,
+      "step": 23592
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.336102834862949e-06,
+      "loss": 0.71,
+      "step": 23593
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3355280251835568e-06,
+      "loss": 0.7139,
+      "step": 23594
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.33495333032872e-06,
+      "loss": 0.7061,
+      "step": 23595
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3343787503060534e-06,
+      "loss": 0.6797,
+      "step": 23596
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3338042851231758e-06,
+      "loss": 0.6865,
+      "step": 23597
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3332299347876976e-06,
+      "loss": 0.71,
+      "step": 23598
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3326556993072325e-06,
+      "loss": 0.6846,
+      "step": 23599
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3320815786893849e-06,
+      "loss": 0.7139,
+      "step": 23600
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.331507572941768e-06,
+      "loss": 0.7334,
+      "step": 23601
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3309336820719877e-06,
+      "loss": 0.6934,
+      "step": 23602
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3303599060876504e-06,
+      "loss": 0.6826,
+      "step": 23603
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3297862449963573e-06,
+      "loss": 0.7275,
+      "step": 23604
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3292126988057109e-06,
+      "loss": 0.7148,
+      "step": 23605
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3286392675233172e-06,
+      "loss": 0.6982,
+      "step": 23606
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3280659511567706e-06,
+      "loss": 0.7217,
+      "step": 23607
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.327492749713667e-06,
+      "loss": 0.6953,
+      "step": 23608
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3269196632016078e-06,
+      "loss": 0.6846,
+      "step": 23609
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3263466916281865e-06,
+      "loss": 0.6982,
+      "step": 23610
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3257738350009941e-06,
+      "loss": 0.71,
+      "step": 23611
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.325201093327625e-06,
+      "loss": 0.7012,
+      "step": 23612
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3246284666156673e-06,
+      "loss": 0.6963,
+      "step": 23613
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3240559548727106e-06,
+      "loss": 0.707,
+      "step": 23614
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.323483558106341e-06,
+      "loss": 0.7002,
+      "step": 23615
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3229112763241448e-06,
+      "loss": 0.7041,
+      "step": 23616
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3223391095337046e-06,
+      "loss": 0.6875,
+      "step": 23617
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3217670577426067e-06,
+      "loss": 0.7119,
+      "step": 23618
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3211951209584318e-06,
+      "loss": 0.6943,
+      "step": 23619
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3206232991887535e-06,
+      "loss": 0.7188,
+      "step": 23620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.320051592441155e-06,
+      "loss": 0.7148,
+      "step": 23621
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3194800007232123e-06,
+      "loss": 0.7148,
+      "step": 23622
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3189085240424992e-06,
+      "loss": 0.6777,
+      "step": 23623
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3183371624065899e-06,
+      "loss": 0.6934,
+      "step": 23624
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.317765915823056e-06,
+      "loss": 0.6836,
+      "step": 23625
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.317194784299467e-06,
+      "loss": 0.6982,
+      "step": 23626
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3166237678433936e-06,
+      "loss": 0.71,
+      "step": 23627
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3160528664624007e-06,
+      "loss": 0.7129,
+      "step": 23628
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3154820801640533e-06,
+      "loss": 0.7256,
+      "step": 23629
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.31491140895592e-06,
+      "loss": 0.7148,
+      "step": 23630
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3143408528455636e-06,
+      "loss": 0.7246,
+      "step": 23631
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3137704118405404e-06,
+      "loss": 0.7188,
+      "step": 23632
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3132000859484107e-06,
+      "loss": 0.6934,
+      "step": 23633
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3126298751767374e-06,
+      "loss": 0.6709,
+      "step": 23634
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3120597795330736e-06,
+      "loss": 0.7051,
+      "step": 23635
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3114897990249741e-06,
+      "loss": 0.7188,
+      "step": 23636
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.310919933659993e-06,
+      "loss": 0.7148,
+      "step": 23637
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3103501834456854e-06,
+      "loss": 0.6855,
+      "step": 23638
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3097805483895986e-06,
+      "loss": 0.7061,
+      "step": 23639
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.309211028499282e-06,
+      "loss": 0.7275,
+      "step": 23640
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3086416237822808e-06,
+      "loss": 0.707,
+      "step": 23641
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3080723342461455e-06,
+      "loss": 0.6914,
+      "step": 23642
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.307503159898419e-06,
+      "loss": 0.6826,
+      "step": 23643
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.306934100746645e-06,
+      "loss": 0.707,
+      "step": 23644
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3063651567983594e-06,
+      "loss": 0.7227,
+      "step": 23645
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3057963280611086e-06,
+      "loss": 0.6895,
+      "step": 23646
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.305227614542428e-06,
+      "loss": 0.7256,
+      "step": 23647
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3046590162498563e-06,
+      "loss": 0.707,
+      "step": 23648
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3040905331909259e-06,
+      "loss": 0.6992,
+      "step": 23649
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.30352216537317e-06,
+      "loss": 0.7129,
+      "step": 23650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.302953912804128e-06,
+      "loss": 0.6914,
+      "step": 23651
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3023857754913217e-06,
+      "loss": 0.6895,
+      "step": 23652
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3018177534422826e-06,
+      "loss": 0.71,
+      "step": 23653
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3012498466645417e-06,
+      "loss": 0.7119,
+      "step": 23654
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.300682055165624e-06,
+      "loss": 0.6992,
+      "step": 23655
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3001143789530513e-06,
+      "loss": 0.709,
+      "step": 23656
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2995468180343495e-06,
+      "loss": 0.6904,
+      "step": 23657
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2989793724170385e-06,
+      "loss": 0.7031,
+      "step": 23658
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.29841204210864e-06,
+      "loss": 0.6953,
+      "step": 23659
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2978448271166699e-06,
+      "loss": 0.6982,
+      "step": 23660
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.297277727448647e-06,
+      "loss": 0.707,
+      "step": 23661
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.296710743112084e-06,
+      "loss": 0.6787,
+      "step": 23662
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2961438741144995e-06,
+      "loss": 0.7119,
+      "step": 23663
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2955771204634049e-06,
+      "loss": 0.7119,
+      "step": 23664
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.295010482166308e-06,
+      "loss": 0.7109,
+      "step": 23665
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2944439592307167e-06,
+      "loss": 0.6992,
+      "step": 23666
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2938775516641443e-06,
+      "loss": 0.6943,
+      "step": 23667
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2933112594740937e-06,
+      "loss": 0.6816,
+      "step": 23668
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.292745082668071e-06,
+      "loss": 0.7012,
+      "step": 23669
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2921790212535779e-06,
+      "loss": 0.7285,
+      "step": 23670
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2916130752381161e-06,
+      "loss": 0.7158,
+      "step": 23671
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2910472446291878e-06,
+      "loss": 0.6865,
+      "step": 23672
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2904815294342899e-06,
+      "loss": 0.7412,
+      "step": 23673
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2899159296609165e-06,
+      "loss": 0.7021,
+      "step": 23674
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2893504453165684e-06,
+      "loss": 0.7158,
+      "step": 23675
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2887850764087384e-06,
+      "loss": 0.6758,
+      "step": 23676
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2882198229449194e-06,
+      "loss": 0.7041,
+      "step": 23677
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2876546849325978e-06,
+      "loss": 0.6924,
+      "step": 23678
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2870896623792672e-06,
+      "loss": 0.7012,
+      "step": 23679
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2865247552924142e-06,
+      "loss": 0.7148,
+      "step": 23680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2859599636795262e-06,
+      "loss": 0.6953,
+      "step": 23681
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2853952875480847e-06,
+      "loss": 0.7139,
+      "step": 23682
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2848307269055803e-06,
+      "loss": 0.6865,
+      "step": 23683
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2842662817594875e-06,
+      "loss": 0.7139,
+      "step": 23684
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2837019521172888e-06,
+      "loss": 0.6963,
+      "step": 23685
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2831377379864607e-06,
+      "loss": 0.7012,
+      "step": 23686
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2825736393744847e-06,
+      "loss": 0.7354,
+      "step": 23687
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2820096562888351e-06,
+      "loss": 0.707,
+      "step": 23688
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2814457887369847e-06,
+      "loss": 0.7021,
+      "step": 23689
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2808820367264073e-06,
+      "loss": 0.7031,
+      "step": 23690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.280318400264573e-06,
+      "loss": 0.6934,
+      "step": 23691
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.279754879358952e-06,
+      "loss": 0.6992,
+      "step": 23692
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.279191474017012e-06,
+      "loss": 0.7021,
+      "step": 23693
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2786281842462177e-06,
+      "loss": 0.6787,
+      "step": 23694
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2780650100540348e-06,
+      "loss": 0.7051,
+      "step": 23695
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2775019514479324e-06,
+      "loss": 0.6953,
+      "step": 23696
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.276939008435365e-06,
+      "loss": 0.7129,
+      "step": 23697
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2763761810237928e-06,
+      "loss": 0.7041,
+      "step": 23698
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2758134692206792e-06,
+      "loss": 0.7041,
+      "step": 23699
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2752508730334788e-06,
+      "loss": 0.709,
+      "step": 23700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2746883924696496e-06,
+      "loss": 0.7012,
+      "step": 23701
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2741260275366429e-06,
+      "loss": 0.7236,
+      "step": 23702
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2735637782419131e-06,
+      "loss": 0.6982,
+      "step": 23703
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2730016445929105e-06,
+      "loss": 0.708,
+      "step": 23704
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.272439626597085e-06,
+      "loss": 0.7109,
+      "step": 23705
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2718777242618852e-06,
+      "loss": 0.6973,
+      "step": 23706
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2713159375947549e-06,
+      "loss": 0.7148,
+      "step": 23707
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2707542666031424e-06,
+      "loss": 0.7178,
+      "step": 23708
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2701927112944933e-06,
+      "loss": 0.7158,
+      "step": 23709
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2696312716762437e-06,
+      "loss": 0.6934,
+      "step": 23710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2690699477558344e-06,
+      "loss": 0.6943,
+      "step": 23711
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.268508739540708e-06,
+      "loss": 0.7168,
+      "step": 23712
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2679476470383002e-06,
+      "loss": 0.6816,
+      "step": 23713
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2673866702560477e-06,
+      "loss": 0.7314,
+      "step": 23714
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2668258092013808e-06,
+      "loss": 0.7041,
+      "step": 23715
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2662650638817398e-06,
+      "loss": 0.7129,
+      "step": 23716
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2657044343045499e-06,
+      "loss": 0.7109,
+      "step": 23717
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.265143920477242e-06,
+      "loss": 0.7207,
+      "step": 23718
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2645835224072423e-06,
+      "loss": 0.7402,
+      "step": 23719
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2640232401019803e-06,
+      "loss": 0.6934,
+      "step": 23720
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2634630735688813e-06,
+      "loss": 0.6924,
+      "step": 23721
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2629030228153704e-06,
+      "loss": 0.6738,
+      "step": 23722
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2623430878488628e-06,
+      "loss": 0.7061,
+      "step": 23723
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.261783268676785e-06,
+      "loss": 0.7031,
+      "step": 23724
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.261223565306553e-06,
+      "loss": 0.6865,
+      "step": 23725
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2606639777455865e-06,
+      "loss": 0.6992,
+      "step": 23726
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2601045060012973e-06,
+      "loss": 0.7246,
+      "step": 23727
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2595451500811052e-06,
+      "loss": 0.7148,
+      "step": 23728
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2589859099924217e-06,
+      "loss": 0.7061,
+      "step": 23729
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2584267857426547e-06,
+      "loss": 0.7178,
+      "step": 23730
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2578677773392144e-06,
+      "loss": 0.6943,
+      "step": 23731
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.257308884789512e-06,
+      "loss": 0.6904,
+      "step": 23732
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2567501081009526e-06,
+      "loss": 0.7285,
+      "step": 23733
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.256191447280941e-06,
+      "loss": 0.7139,
+      "step": 23734
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2556329023368807e-06,
+      "loss": 0.7227,
+      "step": 23735
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2550744732761744e-06,
+      "loss": 0.7051,
+      "step": 23736
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2545161601062228e-06,
+      "loss": 0.709,
+      "step": 23737
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2539579628344233e-06,
+      "loss": 0.7021,
+      "step": 23738
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2533998814681746e-06,
+      "loss": 0.7246,
+      "step": 23739
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2528419160148708e-06,
+      "loss": 0.6797,
+      "step": 23740
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.252284066481909e-06,
+      "loss": 0.6992,
+      "step": 23741
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2517263328766827e-06,
+      "loss": 0.6875,
+      "step": 23742
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2511687152065777e-06,
+      "loss": 0.7129,
+      "step": 23743
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2506112134789894e-06,
+      "loss": 0.7461,
+      "step": 23744
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.250053827701303e-06,
+      "loss": 0.6729,
+      "step": 23745
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.249496557880907e-06,
+      "loss": 0.6826,
+      "step": 23746
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2489394040251857e-06,
+      "loss": 0.7256,
+      "step": 23747
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2483823661415218e-06,
+      "loss": 0.7061,
+      "step": 23748
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2478254442372983e-06,
+      "loss": 0.707,
+      "step": 23749
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2472686383198952e-06,
+      "loss": 0.71,
+      "step": 23750
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2467119483966928e-06,
+      "loss": 0.7402,
+      "step": 23751
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2461553744750643e-06,
+      "loss": 0.6982,
+      "step": 23752
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2455989165623916e-06,
+      "loss": 0.71,
+      "step": 23753
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2450425746660466e-06,
+      "loss": 0.7061,
+      "step": 23754
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2444863487934034e-06,
+      "loss": 0.7148,
+      "step": 23755
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2439302389518281e-06,
+      "loss": 0.7031,
+      "step": 23756
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2433742451486952e-06,
+      "loss": 0.7266,
+      "step": 23757
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2428183673913729e-06,
+      "loss": 0.6719,
+      "step": 23758
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2422626056872268e-06,
+      "loss": 0.6953,
+      "step": 23759
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2417069600436194e-06,
+      "loss": 0.6943,
+      "step": 23760
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2411514304679218e-06,
+      "loss": 0.6787,
+      "step": 23761
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.240596016967489e-06,
+      "loss": 0.7051,
+      "step": 23762
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2400407195496833e-06,
+      "loss": 0.7119,
+      "step": 23763
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.239485538221863e-06,
+      "loss": 0.6953,
+      "step": 23764
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2389304729913876e-06,
+      "loss": 0.7227,
+      "step": 23765
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2383755238656125e-06,
+      "loss": 0.6846,
+      "step": 23766
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2378206908518919e-06,
+      "loss": 0.6934,
+      "step": 23767
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2372659739575787e-06,
+      "loss": 0.708,
+      "step": 23768
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2367113731900238e-06,
+      "loss": 0.7021,
+      "step": 23769
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2361568885565777e-06,
+      "loss": 0.7314,
+      "step": 23770
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2356025200645872e-06,
+      "loss": 0.709,
+      "step": 23771
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2350482677214005e-06,
+      "loss": 0.7051,
+      "step": 23772
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2344941315343594e-06,
+      "loss": 0.71,
+      "step": 23773
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.233940111510814e-06,
+      "loss": 0.7051,
+      "step": 23774
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2333862076581016e-06,
+      "loss": 0.7139,
+      "step": 23775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2328324199835605e-06,
+      "loss": 0.6904,
+      "step": 23776
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2322787484945353e-06,
+      "loss": 0.6865,
+      "step": 23777
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.231725193198361e-06,
+      "loss": 0.7158,
+      "step": 23778
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2311717541023739e-06,
+      "loss": 0.6758,
+      "step": 23779
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.230618431213908e-06,
+      "loss": 0.7275,
+      "step": 23780
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2300652245402956e-06,
+      "loss": 0.71,
+      "step": 23781
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2295121340888682e-06,
+      "loss": 0.6904,
+      "step": 23782
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.228959159866956e-06,
+      "loss": 0.6934,
+      "step": 23783
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.228406301881887e-06,
+      "loss": 0.7256,
+      "step": 23784
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2278535601409857e-06,
+      "loss": 0.7061,
+      "step": 23785
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2273009346515819e-06,
+      "loss": 0.6846,
+      "step": 23786
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2267484254209983e-06,
+      "loss": 0.6758,
+      "step": 23787
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2261960324565514e-06,
+      "loss": 0.7168,
+      "step": 23788
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2256437557655676e-06,
+      "loss": 0.6807,
+      "step": 23789
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2250915953553645e-06,
+      "loss": 0.7158,
+      "step": 23790
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2245395512332581e-06,
+      "loss": 0.7139,
+      "step": 23791
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2239876234065652e-06,
+      "loss": 0.7344,
+      "step": 23792
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2234358118825984e-06,
+      "loss": 0.7178,
+      "step": 23793
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2228841166686755e-06,
+      "loss": 0.7012,
+      "step": 23794
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2223325377721028e-06,
+      "loss": 0.7139,
+      "step": 23795
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2217810752001924e-06,
+      "loss": 0.6924,
+      "step": 23796
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2212297289602481e-06,
+      "loss": 0.7051,
+      "step": 23797
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2206784990595833e-06,
+      "loss": 0.7236,
+      "step": 23798
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2201273855054996e-06,
+      "loss": 0.6895,
+      "step": 23799
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2195763883053025e-06,
+      "loss": 0.7178,
+      "step": 23800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2190255074662882e-06,
+      "loss": 0.6729,
+      "step": 23801
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2184747429957634e-06,
+      "loss": 0.6875,
+      "step": 23802
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2179240949010251e-06,
+      "loss": 0.7227,
+      "step": 23803
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2173735631893702e-06,
+      "loss": 0.6992,
+      "step": 23804
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2168231478680925e-06,
+      "loss": 0.708,
+      "step": 23805
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2162728489444907e-06,
+      "loss": 0.6787,
+      "step": 23806
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2157226664258582e-06,
+      "loss": 0.7412,
+      "step": 23807
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2151726003194809e-06,
+      "loss": 0.7021,
+      "step": 23808
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2146226506326497e-06,
+      "loss": 0.707,
+      "step": 23809
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2140728173726546e-06,
+      "loss": 0.7031,
+      "step": 23810
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.213523100546783e-06,
+      "loss": 0.7119,
+      "step": 23811
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2129735001623176e-06,
+      "loss": 0.7178,
+      "step": 23812
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2124240162265444e-06,
+      "loss": 0.7344,
+      "step": 23813
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2118746487467425e-06,
+      "loss": 0.7139,
+      "step": 23814
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2113253977301941e-06,
+      "loss": 0.6865,
+      "step": 23815
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2107762631841768e-06,
+      "loss": 0.6904,
+      "step": 23816
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.210227245115969e-06,
+      "loss": 0.6777,
+      "step": 23817
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2096783435328453e-06,
+      "loss": 0.7188,
+      "step": 23818
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2091295584420814e-06,
+      "loss": 0.7021,
+      "step": 23819
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2085808898509522e-06,
+      "loss": 0.707,
+      "step": 23820
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2080323377667213e-06,
+      "loss": 0.6865,
+      "step": 23821
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2074839021966655e-06,
+      "loss": 0.7041,
+      "step": 23822
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2069355831480512e-06,
+      "loss": 0.7109,
+      "step": 23823
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2063873806281435e-06,
+      "loss": 0.7139,
+      "step": 23824
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2058392946442077e-06,
+      "loss": 0.71,
+      "step": 23825
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.205291325203507e-06,
+      "loss": 0.6943,
+      "step": 23826
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2047434723133044e-06,
+      "loss": 0.7031,
+      "step": 23827
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2041957359808598e-06,
+      "loss": 0.6777,
+      "step": 23828
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2036481162134316e-06,
+      "loss": 0.7119,
+      "step": 23829
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2031006130182743e-06,
+      "loss": 0.7246,
+      "step": 23830
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2025532264026497e-06,
+      "loss": 0.7041,
+      "step": 23831
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2020059563738084e-06,
+      "loss": 0.7119,
+      "step": 23832
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.201458802939004e-06,
+      "loss": 0.7227,
+      "step": 23833
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.200911766105487e-06,
+      "loss": 0.6953,
+      "step": 23834
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2003648458805062e-06,
+      "loss": 0.7324,
+      "step": 23835
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.19981804227131e-06,
+      "loss": 0.709,
+      "step": 23836
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1992713552851465e-06,
+      "loss": 0.6953,
+      "step": 23837
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.198724784929256e-06,
+      "loss": 0.7031,
+      "step": 23838
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1981783312108886e-06,
+      "loss": 0.7178,
+      "step": 23839
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1976319941372816e-06,
+      "loss": 0.6904,
+      "step": 23840
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.197085773715676e-06,
+      "loss": 0.6816,
+      "step": 23841
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1965396699533084e-06,
+      "loss": 0.71,
+      "step": 23842
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1959936828574203e-06,
+      "loss": 0.7012,
+      "step": 23843
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1954478124352454e-06,
+      "loss": 0.7168,
+      "step": 23844
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1949020586940186e-06,
+      "loss": 0.6865,
+      "step": 23845
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1943564216409686e-06,
+      "loss": 0.709,
+      "step": 23846
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.193810901283331e-06,
+      "loss": 0.7051,
+      "step": 23847
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.193265497628333e-06,
+      "loss": 0.6982,
+      "step": 23848
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1927202106832025e-06,
+      "loss": 0.6963,
+      "step": 23849
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1921750404551647e-06,
+      "loss": 0.6865,
+      "step": 23850
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.191629986951448e-06,
+      "loss": 0.7285,
+      "step": 23851
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.191085050179276e-06,
+      "loss": 0.7021,
+      "step": 23852
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1905402301458647e-06,
+      "loss": 0.7227,
+      "step": 23853
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1899955268584362e-06,
+      "loss": 0.6904,
+      "step": 23854
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1894509403242128e-06,
+      "loss": 0.7285,
+      "step": 23855
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1889064705504083e-06,
+      "loss": 0.6885,
+      "step": 23856
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1883621175442384e-06,
+      "loss": 0.709,
+      "step": 23857
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1878178813129192e-06,
+      "loss": 0.6865,
+      "step": 23858
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1872737618636609e-06,
+      "loss": 0.6973,
+      "step": 23859
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1867297592036742e-06,
+      "loss": 0.6963,
+      "step": 23860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1861858733401698e-06,
+      "loss": 0.666,
+      "step": 23861
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1856421042803556e-06,
+      "loss": 0.6924,
+      "step": 23862
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1850984520314347e-06,
+      "loss": 0.7178,
+      "step": 23863
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1845549166006154e-06,
+      "loss": 0.7031,
+      "step": 23864
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1840114979951023e-06,
+      "loss": 0.7051,
+      "step": 23865
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1834681962220895e-06,
+      "loss": 0.7041,
+      "step": 23866
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1829250112887846e-06,
+      "loss": 0.7168,
+      "step": 23867
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.182381943202383e-06,
+      "loss": 0.7119,
+      "step": 23868
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1818389919700824e-06,
+      "loss": 0.6885,
+      "step": 23869
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1812961575990767e-06,
+      "loss": 0.7051,
+      "step": 23870
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1807534400965614e-06,
+      "loss": 0.7139,
+      "step": 23871
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1802108394697276e-06,
+      "loss": 0.7051,
+      "step": 23872
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.179668355725766e-06,
+      "loss": 0.6885,
+      "step": 23873
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1791259888718676e-06,
+      "loss": 0.7236,
+      "step": 23874
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1785837389152154e-06,
+      "loss": 0.6992,
+      "step": 23875
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1780416058630017e-06,
+      "loss": 0.7148,
+      "step": 23876
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1774995897224073e-06,
+      "loss": 0.7129,
+      "step": 23877
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1769576905006187e-06,
+      "loss": 0.7305,
+      "step": 23878
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1764159082048098e-06,
+      "loss": 0.7031,
+      "step": 23879
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1758742428421688e-06,
+      "loss": 0.6982,
+      "step": 23880
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1753326944198696e-06,
+      "loss": 0.7012,
+      "step": 23881
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1747912629450897e-06,
+      "loss": 0.7148,
+      "step": 23882
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1742499484250024e-06,
+      "loss": 0.6885,
+      "step": 23883
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1737087508667877e-06,
+      "loss": 0.7012,
+      "step": 23884
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.173167670277613e-06,
+      "loss": 0.7129,
+      "step": 23885
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.172626706664648e-06,
+      "loss": 0.6934,
+      "step": 23886
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1720858600350616e-06,
+      "loss": 0.7236,
+      "step": 23887
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1715451303960245e-06,
+      "loss": 0.7021,
+      "step": 23888
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1710045177547013e-06,
+      "loss": 0.6934,
+      "step": 23889
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1704640221182562e-06,
+      "loss": 0.7334,
+      "step": 23890
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.169923643493852e-06,
+      "loss": 0.7139,
+      "step": 23891
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1693833818886502e-06,
+      "loss": 0.6904,
+      "step": 23892
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.168843237309809e-06,
+      "loss": 0.6943,
+      "step": 23893
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1683032097644898e-06,
+      "loss": 0.7119,
+      "step": 23894
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.167763299259843e-06,
+      "loss": 0.7041,
+      "step": 23895
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1672235058030324e-06,
+      "loss": 0.7354,
+      "step": 23896
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1666838294012072e-06,
+      "loss": 0.7168,
+      "step": 23897
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1661442700615177e-06,
+      "loss": 0.7178,
+      "step": 23898
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1656048277911147e-06,
+      "loss": 0.6943,
+      "step": 23899
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1650655025971492e-06,
+      "loss": 0.6787,
+      "step": 23900
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1645262944867686e-06,
+      "loss": 0.7354,
+      "step": 23901
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1639872034671173e-06,
+      "loss": 0.6738,
+      "step": 23902
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1634482295453398e-06,
+      "loss": 0.7139,
+      "step": 23903
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1629093727285777e-06,
+      "loss": 0.7109,
+      "step": 23904
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1623706330239748e-06,
+      "loss": 0.7236,
+      "step": 23905
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1618320104386683e-06,
+      "loss": 0.7246,
+      "step": 23906
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1612935049797979e-06,
+      "loss": 0.7119,
+      "step": 23907
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1607551166544962e-06,
+      "loss": 0.7109,
+      "step": 23908
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1602168454699037e-06,
+      "loss": 0.71,
+      "step": 23909
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1596786914331525e-06,
+      "loss": 0.6992,
+      "step": 23910
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1591406545513695e-06,
+      "loss": 0.71,
+      "step": 23911
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1586027348316908e-06,
+      "loss": 0.6895,
+      "step": 23912
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1580649322812431e-06,
+      "loss": 0.6807,
+      "step": 23913
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1575272469071518e-06,
+      "loss": 0.7236,
+      "step": 23914
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1569896787165447e-06,
+      "loss": 0.7051,
+      "step": 23915
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1564522277165436e-06,
+      "loss": 0.7031,
+      "step": 23916
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.155914893914276e-06,
+      "loss": 0.6914,
+      "step": 23917
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1553776773168579e-06,
+      "loss": 0.7031,
+      "step": 23918
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1548405779314087e-06,
+      "loss": 0.6914,
+      "step": 23919
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1543035957650462e-06,
+      "loss": 0.6992,
+      "step": 23920
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1537667308248901e-06,
+      "loss": 0.6904,
+      "step": 23921
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1532299831180538e-06,
+      "loss": 0.6904,
+      "step": 23922
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1526933526516515e-06,
+      "loss": 0.6982,
+      "step": 23923
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1521568394327887e-06,
+      "loss": 0.6797,
+      "step": 23924
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1516204434685819e-06,
+      "loss": 0.7012,
+      "step": 23925
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1510841647661375e-06,
+      "loss": 0.6973,
+      "step": 23926
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1505480033325623e-06,
+      "loss": 0.6904,
+      "step": 23927
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1500119591749592e-06,
+      "loss": 0.6797,
+      "step": 23928
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1494760323004373e-06,
+      "loss": 0.6807,
+      "step": 23929
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1489402227160996e-06,
+      "loss": 0.7178,
+      "step": 23930
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1484045304290404e-06,
+      "loss": 0.6885,
+      "step": 23931
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1478689554463597e-06,
+      "loss": 0.6943,
+      "step": 23932
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1473334977751593e-06,
+      "loss": 0.71,
+      "step": 23933
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1467981574225339e-06,
+      "loss": 0.6797,
+      "step": 23934
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1462629343955779e-06,
+      "loss": 0.7656,
+      "step": 23935
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.145727828701383e-06,
+      "loss": 0.6729,
+      "step": 23936
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1451928403470414e-06,
+      "loss": 0.7002,
+      "step": 23937
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1446579693396442e-06,
+      "loss": 0.708,
+      "step": 23938
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.144123215686278e-06,
+      "loss": 0.708,
+      "step": 23939
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1435885793940271e-06,
+      "loss": 0.6963,
+      "step": 23940
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1430540604699825e-06,
+      "loss": 0.6885,
+      "step": 23941
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.142519658921225e-06,
+      "loss": 0.7109,
+      "step": 23942
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1419853747548382e-06,
+      "loss": 0.7197,
+      "step": 23943
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1414512079778971e-06,
+      "loss": 0.6777,
+      "step": 23944
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1409171585974877e-06,
+      "loss": 0.7041,
+      "step": 23945
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1403832266206838e-06,
+      "loss": 0.7031,
+      "step": 23946
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1398494120545612e-06,
+      "loss": 0.71,
+      "step": 23947
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.139315714906195e-06,
+      "loss": 0.7031,
+      "step": 23948
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1387821351826579e-06,
+      "loss": 0.7412,
+      "step": 23949
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1382486728910203e-06,
+      "loss": 0.7383,
+      "step": 23950
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1377153280383535e-06,
+      "loss": 0.6924,
+      "step": 23951
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.137182100631723e-06,
+      "loss": 0.6982,
+      "step": 23952
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1366489906781963e-06,
+      "loss": 0.7002,
+      "step": 23953
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1361159981848402e-06,
+      "loss": 0.7051,
+      "step": 23954
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1355831231587167e-06,
+      "loss": 0.6865,
+      "step": 23955
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1350503656068868e-06,
+      "loss": 0.6963,
+      "step": 23956
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1345177255364125e-06,
+      "loss": 0.7061,
+      "step": 23957
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.133985202954352e-06,
+      "loss": 0.6729,
+      "step": 23958
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1334527978677624e-06,
+      "loss": 0.7109,
+      "step": 23959
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1329205102836983e-06,
+      "loss": 0.6699,
+      "step": 23960
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.132388340209213e-06,
+      "loss": 0.7051,
+      "step": 23961
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.131856287651365e-06,
+      "loss": 0.6768,
+      "step": 23962
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1313243526171979e-06,
+      "loss": 0.7178,
+      "step": 23963
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.130792535113765e-06,
+      "loss": 0.6895,
+      "step": 23964
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1302608351481104e-06,
+      "loss": 0.6982,
+      "step": 23965
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.129729252727284e-06,
+      "loss": 0.7051,
+      "step": 23966
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1291977878583316e-06,
+      "loss": 0.7314,
+      "step": 23967
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1286664405482928e-06,
+      "loss": 0.7285,
+      "step": 23968
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1281352108042121e-06,
+      "loss": 0.7002,
+      "step": 23969
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1276040986331271e-06,
+      "loss": 0.6953,
+      "step": 23970
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1270731040420779e-06,
+      "loss": 0.6748,
+      "step": 23971
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.126542227038101e-06,
+      "loss": 0.6943,
+      "step": 23972
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1260114676282285e-06,
+      "loss": 0.7324,
+      "step": 23973
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1254808258195004e-06,
+      "loss": 0.7139,
+      "step": 23974
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1249503016189478e-06,
+      "loss": 0.7119,
+      "step": 23975
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1244198950335982e-06,
+      "loss": 0.668,
+      "step": 23976
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1238896060704786e-06,
+      "loss": 0.6934,
+      "step": 23977
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1233594347366228e-06,
+      "loss": 0.707,
+      "step": 23978
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1228293810390533e-06,
+      "loss": 0.708,
+      "step": 23979
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1222994449847969e-06,
+      "loss": 0.7178,
+      "step": 23980
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.121769626580873e-06,
+      "loss": 0.7031,
+      "step": 23981
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1212399258343055e-06,
+      "loss": 0.7051,
+      "step": 23982
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1207103427521138e-06,
+      "loss": 0.7158,
+      "step": 23983
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1201808773413158e-06,
+      "loss": 0.708,
+      "step": 23984
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1196515296089272e-06,
+      "loss": 0.7197,
+      "step": 23985
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1191222995619633e-06,
+      "loss": 0.7129,
+      "step": 23986
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1185931872074396e-06,
+      "loss": 0.6846,
+      "step": 23987
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1180641925523695e-06,
+      "loss": 0.7119,
+      "step": 23988
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.117535315603756e-06,
+      "loss": 0.6914,
+      "step": 23989
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1170065563686149e-06,
+      "loss": 0.71,
+      "step": 23990
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1164779148539517e-06,
+      "loss": 0.7012,
+      "step": 23991
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1159493910667707e-06,
+      "loss": 0.6982,
+      "step": 23992
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1154209850140785e-06,
+      "loss": 0.6953,
+      "step": 23993
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.114892696702874e-06,
+      "loss": 0.7471,
+      "step": 23994
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1143645261401636e-06,
+      "loss": 0.708,
+      "step": 23995
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1138364733329431e-06,
+      "loss": 0.6953,
+      "step": 23996
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1133085382882102e-06,
+      "loss": 0.7129,
+      "step": 23997
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1127807210129615e-06,
+      "loss": 0.6943,
+      "step": 23998
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1122530215141935e-06,
+      "loss": 0.7021,
+      "step": 23999
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1117254397988985e-06,
+      "loss": 0.708,
+      "step": 24000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1111979758740688e-06,
+      "loss": 0.6836,
+      "step": 24001
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.110670629746694e-06,
+      "loss": 0.6904,
+      "step": 24002
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1101434014237623e-06,
+      "loss": 0.6895,
+      "step": 24003
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.109616290912261e-06,
+      "loss": 0.71,
+      "step": 24004
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1090892982191747e-06,
+      "loss": 0.71,
+      "step": 24005
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.108562423351487e-06,
+      "loss": 0.7041,
+      "step": 24006
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.108035666316184e-06,
+      "loss": 0.7021,
+      "step": 24007
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1075090271202448e-06,
+      "loss": 0.6895,
+      "step": 24008
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1069825057706462e-06,
+      "loss": 0.6865,
+      "step": 24009
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1064561022743647e-06,
+      "loss": 0.7002,
+      "step": 24010
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1059298166383814e-06,
+      "loss": 0.6895,
+      "step": 24011
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1054036488696684e-06,
+      "loss": 0.7139,
+      "step": 24012
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1048775989751992e-06,
+      "loss": 0.7021,
+      "step": 24013
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1043516669619435e-06,
+      "loss": 0.7041,
+      "step": 24014
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1038258528368729e-06,
+      "loss": 0.7158,
+      "step": 24015
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1033001566069557e-06,
+      "loss": 0.6904,
+      "step": 24016
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1027745782791566e-06,
+      "loss": 0.7285,
+      "step": 24017
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.102249117860441e-06,
+      "loss": 0.7324,
+      "step": 24018
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1017237753577748e-06,
+      "loss": 0.7061,
+      "step": 24019
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1011985507781198e-06,
+      "loss": 0.6982,
+      "step": 24020
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1006734441284372e-06,
+      "loss": 0.7148,
+      "step": 24021
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1001484554156795e-06,
+      "loss": 0.6738,
+      "step": 24022
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0996235846468107e-06,
+      "loss": 0.7051,
+      "step": 24023
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0990988318287854e-06,
+      "loss": 0.7051,
+      "step": 24024
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0985741969685559e-06,
+      "loss": 0.6787,
+      "step": 24025
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0980496800730766e-06,
+      "loss": 0.708,
+      "step": 24026
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0975252811492976e-06,
+      "loss": 0.6855,
+      "step": 24027
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0970010002041687e-06,
+      "loss": 0.6982,
+      "step": 24028
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0964768372446377e-06,
+      "loss": 0.7051,
+      "step": 24029
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0959527922776513e-06,
+      "loss": 0.6895,
+      "step": 24030
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0954288653101518e-06,
+      "loss": 0.7344,
+      "step": 24031
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.094905056349087e-06,
+      "loss": 0.7305,
+      "step": 24032
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0943813654013967e-06,
+      "loss": 0.7021,
+      "step": 24033
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0938577924740202e-06,
+      "loss": 0.7227,
+      "step": 24034
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.093334337573897e-06,
+      "loss": 0.6943,
+      "step": 24035
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.092811000707964e-06,
+      "loss": 0.7002,
+      "step": 24036
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0922877818831557e-06,
+      "loss": 0.7236,
+      "step": 24037
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0917646811064064e-06,
+      "loss": 0.7188,
+      "step": 24038
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0912416983846475e-06,
+      "loss": 0.709,
+      "step": 24039
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0907188337248142e-06,
+      "loss": 0.7061,
+      "step": 24040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0901960871338302e-06,
+      "loss": 0.6914,
+      "step": 24041
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0896734586186265e-06,
+      "loss": 0.6924,
+      "step": 24042
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.089150948186124e-06,
+      "loss": 0.7061,
+      "step": 24043
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.088628555843254e-06,
+      "loss": 0.6943,
+      "step": 24044
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0881062815969378e-06,
+      "loss": 0.6748,
+      "step": 24045
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0875841254540942e-06,
+      "loss": 0.7002,
+      "step": 24046
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0870620874216443e-06,
+      "loss": 0.6953,
+      "step": 24047
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0865401675065069e-06,
+      "loss": 0.7178,
+      "step": 24048
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0860183657155977e-06,
+      "loss": 0.7158,
+      "step": 24049
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.085496682055831e-06,
+      "loss": 0.6865,
+      "step": 24050
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0849751165341215e-06,
+      "loss": 0.7021,
+      "step": 24051
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0844536691573814e-06,
+      "loss": 0.7119,
+      "step": 24052
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0839323399325242e-06,
+      "loss": 0.6992,
+      "step": 24053
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0834111288664529e-06,
+      "loss": 0.7246,
+      "step": 24054
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0828900359660754e-06,
+      "loss": 0.7061,
+      "step": 24055
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.082369061238301e-06,
+      "loss": 0.7041,
+      "step": 24056
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0818482046900325e-06,
+      "loss": 0.6787,
+      "step": 24057
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0813274663281715e-06,
+      "loss": 0.6982,
+      "step": 24058
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.08080684615962e-06,
+      "loss": 0.7021,
+      "step": 24059
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.080286344191277e-06,
+      "loss": 0.7139,
+      "step": 24060
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0797659604300403e-06,
+      "loss": 0.7041,
+      "step": 24061
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0792456948828056e-06,
+      "loss": 0.6875,
+      "step": 24062
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0787255475564674e-06,
+      "loss": 0.7285,
+      "step": 24063
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0782055184579221e-06,
+      "loss": 0.7207,
+      "step": 24064
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0776856075940577e-06,
+      "loss": 0.7402,
+      "step": 24065
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0771658149717678e-06,
+      "loss": 0.6934,
+      "step": 24066
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0766461405979357e-06,
+      "loss": 0.6797,
+      "step": 24067
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0761265844794522e-06,
+      "loss": 0.71,
+      "step": 24068
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0756071466232022e-06,
+      "loss": 0.7061,
+      "step": 24069
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0750878270360687e-06,
+      "loss": 0.6904,
+      "step": 24070
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0745686257249332e-06,
+      "loss": 0.7012,
+      "step": 24071
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.074049542696678e-06,
+      "loss": 0.7168,
+      "step": 24072
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0735305779581817e-06,
+      "loss": 0.7129,
+      "step": 24073
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0730117315163212e-06,
+      "loss": 0.6943,
+      "step": 24074
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0724930033779734e-06,
+      "loss": 0.7314,
+      "step": 24075
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0719743935500092e-06,
+      "loss": 0.7051,
+      "step": 24076
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0714559020393067e-06,
+      "loss": 0.6943,
+      "step": 24077
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0709375288527347e-06,
+      "loss": 0.6934,
+      "step": 24078
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.070419273997162e-06,
+      "loss": 0.6973,
+      "step": 24079
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0699011374794576e-06,
+      "loss": 0.7002,
+      "step": 24080
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0693831193064884e-06,
+      "loss": 0.6924,
+      "step": 24081
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.068865219485119e-06,
+      "loss": 0.6826,
+      "step": 24082
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0683474380222125e-06,
+      "loss": 0.7002,
+      "step": 24083
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0678297749246292e-06,
+      "loss": 0.7031,
+      "step": 24084
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0673122301992345e-06,
+      "loss": 0.6934,
+      "step": 24085
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0667948038528809e-06,
+      "loss": 0.7158,
+      "step": 24086
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0662774958924293e-06,
+      "loss": 0.7031,
+      "step": 24087
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0657603063247313e-06,
+      "loss": 0.6865,
+      "step": 24088
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0652432351566467e-06,
+      "loss": 0.7031,
+      "step": 24089
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0647262823950244e-06,
+      "loss": 0.7305,
+      "step": 24090
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0642094480467159e-06,
+      "loss": 0.6875,
+      "step": 24091
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0636927321185698e-06,
+      "loss": 0.6973,
+      "step": 24092
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0631761346174352e-06,
+      "loss": 0.7217,
+      "step": 24093
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0626596555501567e-06,
+      "loss": 0.707,
+      "step": 24094
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0621432949235798e-06,
+      "loss": 0.707,
+      "step": 24095
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.061627052744545e-06,
+      "loss": 0.7236,
+      "step": 24096
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0611109290198983e-06,
+      "loss": 0.7021,
+      "step": 24097
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0605949237564794e-06,
+      "loss": 0.7139,
+      "step": 24098
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0600790369611213e-06,
+      "loss": 0.7285,
+      "step": 24099
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.059563268640662e-06,
+      "loss": 0.6895,
+      "step": 24100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0590476188019417e-06,
+      "loss": 0.6953,
+      "step": 24101
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.058532087451789e-06,
+      "loss": 0.6943,
+      "step": 24102
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0580166745970388e-06,
+      "loss": 0.6836,
+      "step": 24103
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0575013802445189e-06,
+      "loss": 0.6904,
+      "step": 24104
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0569862044010603e-06,
+      "loss": 0.6904,
+      "step": 24105
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.05647114707349e-06,
+      "loss": 0.7129,
+      "step": 24106
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0559562082686326e-06,
+      "loss": 0.7021,
+      "step": 24107
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0554413879933113e-06,
+      "loss": 0.6885,
+      "step": 24108
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0549266862543516e-06,
+      "loss": 0.7168,
+      "step": 24109
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.054412103058574e-06,
+      "loss": 0.6816,
+      "step": 24110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0538976384127986e-06,
+      "loss": 0.7295,
+      "step": 24111
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0533832923238386e-06,
+      "loss": 0.7139,
+      "step": 24112
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0528690647985152e-06,
+      "loss": 0.6924,
+      "step": 24113
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0523549558436408e-06,
+      "loss": 0.7061,
+      "step": 24114
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.051840965466029e-06,
+      "loss": 0.709,
+      "step": 24115
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0513270936724918e-06,
+      "loss": 0.6846,
+      "step": 24116
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0508133404698374e-06,
+      "loss": 0.6865,
+      "step": 24117
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0502997058648789e-06,
+      "loss": 0.6855,
+      "step": 24118
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.049786189864418e-06,
+      "loss": 0.6904,
+      "step": 24119
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0492727924752622e-06,
+      "loss": 0.7197,
+      "step": 24120
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0487595137042139e-06,
+      "loss": 0.7109,
+      "step": 24121
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0482463535580766e-06,
+      "loss": 0.7109,
+      "step": 24122
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0477333120436528e-06,
+      "loss": 0.7041,
+      "step": 24123
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.047220389167738e-06,
+      "loss": 0.7051,
+      "step": 24124
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.046707584937131e-06,
+      "loss": 0.6924,
+      "step": 24125
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0461948993586279e-06,
+      "loss": 0.7002,
+      "step": 24126
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.045682332439022e-06,
+      "loss": 0.7041,
+      "step": 24127
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0451698841851076e-06,
+      "loss": 0.7139,
+      "step": 24128
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.044657554603673e-06,
+      "loss": 0.6963,
+      "step": 24129
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0441453437015113e-06,
+      "loss": 0.6973,
+      "step": 24130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.043633251485411e-06,
+      "loss": 0.7295,
+      "step": 24131
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0431212779621547e-06,
+      "loss": 0.7021,
+      "step": 24132
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0426094231385265e-06,
+      "loss": 0.7373,
+      "step": 24133
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.042097687021314e-06,
+      "loss": 0.7178,
+      "step": 24134
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0415860696172974e-06,
+      "loss": 0.7148,
+      "step": 24135
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.041074570933257e-06,
+      "loss": 0.7012,
+      "step": 24136
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0405631909759706e-06,
+      "loss": 0.6973,
+      "step": 24137
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0400519297522148e-06,
+      "loss": 0.7109,
+      "step": 24138
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.039540787268767e-06,
+      "loss": 0.6992,
+      "step": 24139
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0390297635323987e-06,
+      "loss": 0.6904,
+      "step": 24140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0385188585498808e-06,
+      "loss": 0.7188,
+      "step": 24141
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0380080723279895e-06,
+      "loss": 0.6914,
+      "step": 24142
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0374974048734897e-06,
+      "loss": 0.6875,
+      "step": 24143
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0369868561931529e-06,
+      "loss": 0.7051,
+      "step": 24144
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0364764262937377e-06,
+      "loss": 0.6943,
+      "step": 24145
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0359661151820144e-06,
+      "loss": 0.7061,
+      "step": 24146
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0354559228647453e-06,
+      "loss": 0.6963,
+      "step": 24147
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0349458493486896e-06,
+      "loss": 0.6855,
+      "step": 24148
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0344358946406087e-06,
+      "loss": 0.7314,
+      "step": 24149
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.03392605874726e-06,
+      "loss": 0.7119,
+      "step": 24150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0334163416754006e-06,
+      "loss": 0.7041,
+      "step": 24151
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.032906743431784e-06,
+      "loss": 0.7178,
+      "step": 24152
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.032397264023164e-06,
+      "loss": 0.7197,
+      "step": 24153
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0318879034562933e-06,
+      "loss": 0.6943,
+      "step": 24154
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0313786617379228e-06,
+      "loss": 0.7227,
+      "step": 24155
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.030869538874799e-06,
+      "loss": 0.6875,
+      "step": 24156
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0303605348736712e-06,
+      "loss": 0.7139,
+      "step": 24157
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0298516497412836e-06,
+      "loss": 0.7119,
+      "step": 24158
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0293428834843799e-06,
+      "loss": 0.7109,
+      "step": 24159
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0288342361097025e-06,
+      "loss": 0.7188,
+      "step": 24160
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0283257076239928e-06,
+      "loss": 0.6982,
+      "step": 24161
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0278172980339874e-06,
+      "loss": 0.7031,
+      "step": 24162
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0273090073464308e-06,
+      "loss": 0.7129,
+      "step": 24163
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0268008355680525e-06,
+      "loss": 0.7002,
+      "step": 24164
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0262927827055879e-06,
+      "loss": 0.6836,
+      "step": 24165
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0257848487657695e-06,
+      "loss": 0.6709,
+      "step": 24166
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0252770337553319e-06,
+      "loss": 0.7266,
+      "step": 24167
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.024769337681003e-06,
+      "loss": 0.7197,
+      "step": 24168
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0242617605495108e-06,
+      "loss": 0.7021,
+      "step": 24169
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0237543023675823e-06,
+      "loss": 0.707,
+      "step": 24170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0232469631419428e-06,
+      "loss": 0.6895,
+      "step": 24171
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.022739742879314e-06,
+      "loss": 0.7021,
+      "step": 24172
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0222326415864192e-06,
+      "loss": 0.7051,
+      "step": 24173
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0217256592699764e-06,
+      "loss": 0.7031,
+      "step": 24174
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.021218795936708e-06,
+      "loss": 0.6895,
+      "step": 24175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0207120515933322e-06,
+      "loss": 0.7354,
+      "step": 24176
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0202054262465588e-06,
+      "loss": 0.6943,
+      "step": 24177
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0196989199031037e-06,
+      "loss": 0.6895,
+      "step": 24178
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0191925325696817e-06,
+      "loss": 0.707,
+      "step": 24179
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0186862642530026e-06,
+      "loss": 0.7148,
+      "step": 24180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0181801149597749e-06,
+      "loss": 0.7109,
+      "step": 24181
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.017674084696707e-06,
+      "loss": 0.7002,
+      "step": 24182
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0171681734705042e-06,
+      "loss": 0.7021,
+      "step": 24183
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0166623812878717e-06,
+      "loss": 0.7021,
+      "step": 24184
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0161567081555123e-06,
+      "loss": 0.6963,
+      "step": 24185
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.015651154080125e-06,
+      "loss": 0.709,
+      "step": 24186
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0151457190684133e-06,
+      "loss": 0.707,
+      "step": 24187
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0146404031270752e-06,
+      "loss": 0.6973,
+      "step": 24188
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0141352062628063e-06,
+      "loss": 0.6973,
+      "step": 24189
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0136301284822992e-06,
+      "loss": 0.6943,
+      "step": 24190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0131251697922507e-06,
+      "loss": 0.6641,
+      "step": 24191
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.012620330199351e-06,
+      "loss": 0.7021,
+      "step": 24192
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0121156097102913e-06,
+      "loss": 0.7139,
+      "step": 24193
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0116110083317598e-06,
+      "loss": 0.6934,
+      "step": 24194
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0111065260704423e-06,
+      "loss": 0.6953,
+      "step": 24195
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0106021629330298e-06,
+      "loss": 0.7139,
+      "step": 24196
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0100979189262006e-06,
+      "loss": 0.7354,
+      "step": 24197
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.009593794056638e-06,
+      "loss": 0.7295,
+      "step": 24198
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0090897883310224e-06,
+      "loss": 0.6807,
+      "step": 24199
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0085859017560362e-06,
+      "loss": 0.7266,
+      "step": 24200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.008082134338355e-06,
+      "loss": 0.7002,
+      "step": 24201
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.007578486084655e-06,
+      "loss": 0.709,
+      "step": 24202
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0070749570016114e-06,
+      "loss": 0.7275,
+      "step": 24203
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.006571547095897e-06,
+      "loss": 0.6768,
+      "step": 24204
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.006068256374182e-06,
+      "loss": 0.7158,
+      "step": 24205
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0055650848431365e-06,
+      "loss": 0.709,
+      "step": 24206
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0050620325094285e-06,
+      "loss": 0.709,
+      "step": 24207
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.004559099379726e-06,
+      "loss": 0.7178,
+      "step": 24208
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0040562854606961e-06,
+      "loss": 0.6895,
+      "step": 24209
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0035535907589976e-06,
+      "loss": 0.7051,
+      "step": 24210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0030510152812923e-06,
+      "loss": 0.7109,
+      "step": 24211
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0025485590342432e-06,
+      "loss": 0.7158,
+      "step": 24212
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0020462220245098e-06,
+      "loss": 0.7158,
+      "step": 24213
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0015440042587466e-06,
+      "loss": 0.666,
+      "step": 24214
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0010419057436116e-06,
+      "loss": 0.7109,
+      "step": 24215
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0005399264857562e-06,
+      "loss": 0.7148,
+      "step": 24216
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0000380664918352e-06,
+      "loss": 0.6934,
+      "step": 24217
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.995363257684976e-07,
+      "loss": 0.7158,
+      "step": 24218
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.990347043223913e-07,
+      "loss": 0.7119,
+      "step": 24219
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.98533202160168e-07,
+      "loss": 0.7129,
+      "step": 24220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.98031819288472e-07,
+      "loss": 0.7344,
+      "step": 24221
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.975305557139492e-07,
+      "loss": 0.71,
+      "step": 24222
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.970294114432377e-07,
+      "loss": 0.7051,
+      "step": 24223
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.965283864829834e-07,
+      "loss": 0.7197,
+      "step": 24224
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.96027480839824e-07,
+      "loss": 0.7314,
+      "step": 24225
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.955266945204e-07,
+      "loss": 0.7012,
+      "step": 24226
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.950260275313451e-07,
+      "loss": 0.6895,
+      "step": 24227
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.945254798792958e-07,
+      "loss": 0.6924,
+      "step": 24228
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.94025051570886e-07,
+      "loss": 0.6797,
+      "step": 24229
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.935247426127458e-07,
+      "loss": 0.6953,
+      "step": 24230
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.930245530115045e-07,
+      "loss": 0.7119,
+      "step": 24231
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.925244827737956e-07,
+      "loss": 0.7002,
+      "step": 24232
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.920245319062439e-07,
+      "loss": 0.6885,
+      "step": 24233
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.915247004154726e-07,
+      "loss": 0.6992,
+      "step": 24234
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.910249883081092e-07,
+      "loss": 0.6855,
+      "step": 24235
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.905253955907735e-07,
+      "loss": 0.6943,
+      "step": 24236
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.900259222700869e-07,
+      "loss": 0.6963,
+      "step": 24237
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.89526568352669e-07,
+      "loss": 0.7285,
+      "step": 24238
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.890273338451372e-07,
+      "loss": 0.7021,
+      "step": 24239
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.885282187541046e-07,
+      "loss": 0.7324,
+      "step": 24240
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.880292230861942e-07,
+      "loss": 0.7041,
+      "step": 24241
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.875303468480103e-07,
+      "loss": 0.6885,
+      "step": 24242
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.87031590046167e-07,
+      "loss": 0.7109,
+      "step": 24243
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.865329526872725e-07,
+      "loss": 0.6992,
+      "step": 24244
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.860344347779383e-07,
+      "loss": 0.7246,
+      "step": 24245
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.85536036324769e-07,
+      "loss": 0.7012,
+      "step": 24246
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.850377573343694e-07,
+      "loss": 0.6807,
+      "step": 24247
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.84539597813343e-07,
+      "loss": 0.71,
+      "step": 24248
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.840415577682927e-07,
+      "loss": 0.7217,
+      "step": 24249
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.835436372058171e-07,
+      "loss": 0.7041,
+      "step": 24250
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.830458361325158e-07,
+      "loss": 0.6934,
+      "step": 24251
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.82548154554983e-07,
+      "loss": 0.6738,
+      "step": 24252
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.820505924798196e-07,
+      "loss": 0.71,
+      "step": 24253
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.815531499136178e-07,
+      "loss": 0.708,
+      "step": 24254
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.810558268629666e-07,
+      "loss": 0.7051,
+      "step": 24255
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.805586233344566e-07,
+      "loss": 0.7324,
+      "step": 24256
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.800615393346802e-07,
+      "loss": 0.6875,
+      "step": 24257
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.795645748702242e-07,
+      "loss": 0.6885,
+      "step": 24258
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.790677299476747e-07,
+      "loss": 0.6914,
+      "step": 24259
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.78571004573613e-07,
+      "loss": 0.6982,
+      "step": 24260
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.78074398754628e-07,
+      "loss": 0.7148,
+      "step": 24261
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.77577912497295e-07,
+      "loss": 0.7188,
+      "step": 24262
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.77081545808196e-07,
+      "loss": 0.7109,
+      "step": 24263
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.765852986939073e-07,
+      "loss": 0.7129,
+      "step": 24264
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.760891711610087e-07,
+      "loss": 0.7188,
+      "step": 24265
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.755931632160731e-07,
+      "loss": 0.7129,
+      "step": 24266
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.750972748656751e-07,
+      "loss": 0.7012,
+      "step": 24267
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.746015061163827e-07,
+      "loss": 0.7051,
+      "step": 24268
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.741058569747685e-07,
+      "loss": 0.6963,
+      "step": 24269
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.736103274474017e-07,
+      "loss": 0.7197,
+      "step": 24270
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.731149175408494e-07,
+      "loss": 0.7207,
+      "step": 24271
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.726196272616751e-07,
+      "loss": 0.7227,
+      "step": 24272
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.721244566164411e-07,
+      "loss": 0.7051,
+      "step": 24273
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.71629405611717e-07,
+      "loss": 0.6787,
+      "step": 24274
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.71134474254055e-07,
+      "loss": 0.6709,
+      "step": 24275
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.70639662550017e-07,
+      "loss": 0.6826,
+      "step": 24276
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.701449705061616e-07,
+      "loss": 0.7051,
+      "step": 24277
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.696503981290451e-07,
+      "loss": 0.6992,
+      "step": 24278
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.691559454252198e-07,
+      "loss": 0.7051,
+      "step": 24279
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.686616124012383e-07,
+      "loss": 0.709,
+      "step": 24280
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.681673990636532e-07,
+      "loss": 0.6943,
+      "step": 24281
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.676733054190135e-07,
+      "loss": 0.6934,
+      "step": 24282
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.671793314738664e-07,
+      "loss": 0.707,
+      "step": 24283
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.666854772347579e-07,
+      "loss": 0.709,
+      "step": 24284
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.661917427082323e-07,
+      "loss": 0.7178,
+      "step": 24285
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.656981279008382e-07,
+      "loss": 0.6992,
+      "step": 24286
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.6520463281911e-07,
+      "loss": 0.7051,
+      "step": 24287
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.647112574695916e-07,
+      "loss": 0.7217,
+      "step": 24288
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.642180018588177e-07,
+      "loss": 0.6943,
+      "step": 24289
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.637248659933296e-07,
+      "loss": 0.7012,
+      "step": 24290
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.6323184987966e-07,
+      "loss": 0.7021,
+      "step": 24291
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.627389535243437e-07,
+      "loss": 0.6797,
+      "step": 24292
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.62246176933913e-07,
+      "loss": 0.7168,
+      "step": 24293
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.617535201148964e-07,
+      "loss": 0.7246,
+      "step": 24294
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.61260983073824e-07,
+      "loss": 0.6934,
+      "step": 24295
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.607685658172229e-07,
+      "loss": 0.7031,
+      "step": 24296
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.602762683516176e-07,
+      "loss": 0.7295,
+      "step": 24297
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.597840906835342e-07,
+      "loss": 0.6807,
+      "step": 24298
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.592920328194965e-07,
+      "loss": 0.6914,
+      "step": 24299
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.588000947660225e-07,
+      "loss": 0.6885,
+      "step": 24300
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.583082765296291e-07,
+      "loss": 0.7051,
+      "step": 24301
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.578165781168403e-07,
+      "loss": 0.7061,
+      "step": 24302
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.573249995341683e-07,
+      "loss": 0.6973,
+      "step": 24303
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.56833540788129e-07,
+      "loss": 0.7139,
+      "step": 24304
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.563422018852353e-07,
+      "loss": 0.6855,
+      "step": 24305
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.558509828319984e-07,
+      "loss": 0.7129,
+      "step": 24306
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.553598836349277e-07,
+      "loss": 0.6992,
+      "step": 24307
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.548689043005311e-07,
+      "loss": 0.7168,
+      "step": 24308
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.543780448353146e-07,
+      "loss": 0.708,
+      "step": 24309
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.538873052457864e-07,
+      "loss": 0.6895,
+      "step": 24310
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.53396685538448e-07,
+      "loss": 0.6973,
+      "step": 24311
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.52906185719803e-07,
+      "loss": 0.7178,
+      "step": 24312
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.524158057963462e-07,
+      "loss": 0.7148,
+      "step": 24313
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.519255457745824e-07,
+      "loss": 0.7129,
+      "step": 24314
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.514354056610054e-07,
+      "loss": 0.7021,
+      "step": 24315
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.50945385462112e-07,
+      "loss": 0.7031,
+      "step": 24316
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.504554851843961e-07,
+      "loss": 0.6924,
+      "step": 24317
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.49965704834348e-07,
+      "loss": 0.7031,
+      "step": 24318
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.494760444184636e-07,
+      "loss": 0.7061,
+      "step": 24319
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.489865039432267e-07,
+      "loss": 0.7119,
+      "step": 24320
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.484970834151242e-07,
+      "loss": 0.7031,
+      "step": 24321
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.480077828406475e-07,
+      "loss": 0.707,
+      "step": 24322
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.475186022262772e-07,
+      "loss": 0.7041,
+      "step": 24323
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.470295415784969e-07,
+      "loss": 0.7061,
+      "step": 24324
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.46540600903788e-07,
+      "loss": 0.7158,
+      "step": 24325
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.460517802086311e-07,
+      "loss": 0.6914,
+      "step": 24326
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.455630794995019e-07,
+      "loss": 0.709,
+      "step": 24327
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.450744987828786e-07,
+      "loss": 0.6924,
+      "step": 24328
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.44586038065235e-07,
+      "loss": 0.71,
+      "step": 24329
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.440976973530435e-07,
+      "loss": 0.6826,
+      "step": 24330
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.436094766527781e-07,
+      "loss": 0.71,
+      "step": 24331
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.431213759709101e-07,
+      "loss": 0.6914,
+      "step": 24332
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.426333953139044e-07,
+      "loss": 0.6699,
+      "step": 24333
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.421455346882269e-07,
+      "loss": 0.7178,
+      "step": 24334
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.416577941003468e-07,
+      "loss": 0.708,
+      "step": 24335
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.411701735567269e-07,
+      "loss": 0.6973,
+      "step": 24336
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.406826730638274e-07,
+      "loss": 0.6973,
+      "step": 24337
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.40195292628111e-07,
+      "loss": 0.6787,
+      "step": 24338
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.397080322560359e-07,
+      "loss": 0.7275,
+      "step": 24339
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.392208919540591e-07,
+      "loss": 0.7129,
+      "step": 24340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.387338717286354e-07,
+      "loss": 0.7334,
+      "step": 24341
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.382469715862186e-07,
+      "loss": 0.7051,
+      "step": 24342
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.377601915332657e-07,
+      "loss": 0.7178,
+      "step": 24343
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.372735315762238e-07,
+      "loss": 0.7178,
+      "step": 24344
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.367869917215444e-07,
+      "loss": 0.6963,
+      "step": 24345
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.363005719756713e-07,
+      "loss": 0.7109,
+      "step": 24346
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.358142723450548e-07,
+      "loss": 0.7275,
+      "step": 24347
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.353280928361385e-07,
+      "loss": 0.7158,
+      "step": 24348
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.348420334553654e-07,
+      "loss": 0.6738,
+      "step": 24349
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.343560942091756e-07,
+      "loss": 0.7041,
+      "step": 24350
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.338702751040107e-07,
+      "loss": 0.7129,
+      "step": 24351
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.333845761463079e-07,
+      "loss": 0.7012,
+      "step": 24352
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.32898997342504e-07,
+      "loss": 0.6992,
+      "step": 24353
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.324135386990329e-07,
+      "loss": 0.708,
+      "step": 24354
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.319282002223306e-07,
+      "loss": 0.6836,
+      "step": 24355
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.314429819188275e-07,
+      "loss": 0.6953,
+      "step": 24356
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.309578837949551e-07,
+      "loss": 0.7119,
+      "step": 24357
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.304729058571405e-07,
+      "loss": 0.7012,
+      "step": 24358
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.299880481118106e-07,
+      "loss": 0.6826,
+      "step": 24359
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.295033105653928e-07,
+      "loss": 0.6787,
+      "step": 24360
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.290186932243095e-07,
+      "loss": 0.6953,
+      "step": 24361
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.285341960949835e-07,
+      "loss": 0.6934,
+      "step": 24362
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.280498191838338e-07,
+      "loss": 0.6885,
+      "step": 24363
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.275655624972846e-07,
+      "loss": 0.7168,
+      "step": 24364
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.270814260417482e-07,
+      "loss": 0.6826,
+      "step": 24365
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.265974098236397e-07,
+      "loss": 0.6816,
+      "step": 24366
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.261135138493793e-07,
+      "loss": 0.6904,
+      "step": 24367
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.256297381253753e-07,
+      "loss": 0.7021,
+      "step": 24368
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.251460826580416e-07,
+      "loss": 0.6963,
+      "step": 24369
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.246625474537852e-07,
+      "loss": 0.7119,
+      "step": 24370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.241791325190153e-07,
+      "loss": 0.7139,
+      "step": 24371
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.236958378601379e-07,
+      "loss": 0.708,
+      "step": 24372
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.232126634835581e-07,
+      "loss": 0.6758,
+      "step": 24373
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.227296093956795e-07,
+      "loss": 0.6953,
+      "step": 24374
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.222466756029003e-07,
+      "loss": 0.7129,
+      "step": 24375
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.217638621116264e-07,
+      "loss": 0.7412,
+      "step": 24376
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.212811689282541e-07,
+      "loss": 0.7324,
+      "step": 24377
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.207985960591781e-07,
+      "loss": 0.6934,
+      "step": 24378
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.203161435107933e-07,
+      "loss": 0.6943,
+      "step": 24379
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.198338112894955e-07,
+      "loss": 0.7168,
+      "step": 24380
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.193515994016778e-07,
+      "loss": 0.6875,
+      "step": 24381
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.188695078537279e-07,
+      "loss": 0.6963,
+      "step": 24382
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.183875366520334e-07,
+      "loss": 0.7178,
+      "step": 24383
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.179056858029889e-07,
+      "loss": 0.7236,
+      "step": 24384
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.174239553129727e-07,
+      "loss": 0.7021,
+      "step": 24385
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.169423451883707e-07,
+      "loss": 0.6924,
+      "step": 24386
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.164608554355636e-07,
+      "loss": 0.7148,
+      "step": 24387
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.159794860609373e-07,
+      "loss": 0.6719,
+      "step": 24388
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.154982370708676e-07,
+      "loss": 0.7061,
+      "step": 24389
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.150171084717352e-07,
+      "loss": 0.7021,
+      "step": 24390
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.145361002699093e-07,
+      "loss": 0.7148,
+      "step": 24391
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.140552124717716e-07,
+      "loss": 0.6904,
+      "step": 24392
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.135744450836914e-07,
+      "loss": 0.708,
+      "step": 24393
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.130937981120413e-07,
+      "loss": 0.7041,
+      "step": 24394
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.126132715631907e-07,
+      "loss": 0.7197,
+      "step": 24395
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.121328654435057e-07,
+      "loss": 0.7139,
+      "step": 24396
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.116525797593589e-07,
+      "loss": 0.6836,
+      "step": 24397
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.111724145171086e-07,
+      "loss": 0.708,
+      "step": 24398
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.106923697231185e-07,
+      "loss": 0.6836,
+      "step": 24399
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.102124453837547e-07,
+      "loss": 0.6865,
+      "step": 24400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.097326415053742e-07,
+      "loss": 0.7021,
+      "step": 24401
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.092529580943365e-07,
+      "loss": 0.6904,
+      "step": 24402
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.087733951569988e-07,
+      "loss": 0.7012,
+      "step": 24403
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.082939526997158e-07,
+      "loss": 0.7178,
+      "step": 24404
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.078146307288416e-07,
+      "loss": 0.7285,
+      "step": 24405
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.073354292507275e-07,
+      "loss": 0.6777,
+      "step": 24406
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.068563482717252e-07,
+      "loss": 0.6865,
+      "step": 24407
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.063773877981808e-07,
+      "loss": 0.709,
+      "step": 24408
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.058985478364457e-07,
+      "loss": 0.6973,
+      "step": 24409
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.054198283928661e-07,
+      "loss": 0.6826,
+      "step": 24410
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.049412294737824e-07,
+      "loss": 0.6943,
+      "step": 24411
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.044627510855364e-07,
+      "loss": 0.7129,
+      "step": 24412
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.039843932344738e-07,
+      "loss": 0.7207,
+      "step": 24413
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.035061559269309e-07,
+      "loss": 0.7012,
+      "step": 24414
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.030280391692459e-07,
+      "loss": 0.707,
+      "step": 24415
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.025500429677559e-07,
+      "loss": 0.6885,
+      "step": 24416
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.020721673287946e-07,
+      "loss": 0.7012,
+      "step": 24417
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.01594412258695e-07,
+      "loss": 0.6914,
+      "step": 24418
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.011167777637886e-07,
+      "loss": 0.6865,
+      "step": 24419
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.006392638504025e-07,
+      "loss": 0.6846,
+      "step": 24420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.001618705248705e-07,
+      "loss": 0.6885,
+      "step": 24421
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.996845977935154e-07,
+      "loss": 0.6943,
+      "step": 24422
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.992074456626653e-07,
+      "loss": 0.7061,
+      "step": 24423
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.987304141386377e-07,
+      "loss": 0.6973,
+      "step": 24424
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.982535032277595e-07,
+      "loss": 0.6924,
+      "step": 24425
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.977767129363502e-07,
+      "loss": 0.6777,
+      "step": 24426
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.973000432707269e-07,
+      "loss": 0.71,
+      "step": 24427
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.968234942372056e-07,
+      "loss": 0.7139,
+      "step": 24428
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.96347065842107e-07,
+      "loss": 0.6904,
+      "step": 24429
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.958707580917392e-07,
+      "loss": 0.6904,
+      "step": 24430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.953945709924161e-07,
+      "loss": 0.7041,
+      "step": 24431
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.949185045504471e-07,
+      "loss": 0.7256,
+      "step": 24432
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.944425587721439e-07,
+      "loss": 0.7129,
+      "step": 24433
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.939667336638135e-07,
+      "loss": 0.6758,
+      "step": 24434
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.934910292317611e-07,
+      "loss": 0.7051,
+      "step": 24435
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.930154454822904e-07,
+      "loss": 0.7061,
+      "step": 24436
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.92539982421704e-07,
+      "loss": 0.6924,
+      "step": 24437
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.920646400563038e-07,
+      "loss": 0.6895,
+      "step": 24438
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.915894183923879e-07,
+      "loss": 0.6865,
+      "step": 24439
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.911143174362557e-07,
+      "loss": 0.7188,
+      "step": 24440
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.906393371942002e-07,
+      "loss": 0.6963,
+      "step": 24441
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.901644776725216e-07,
+      "loss": 0.7012,
+      "step": 24442
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.896897388775083e-07,
+      "loss": 0.6924,
+      "step": 24443
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.892151208154509e-07,
+      "loss": 0.7305,
+      "step": 24444
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.887406234926421e-07,
+      "loss": 0.71,
+      "step": 24445
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.882662469153713e-07,
+      "loss": 0.7334,
+      "step": 24446
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.877919910899213e-07,
+      "loss": 0.7227,
+      "step": 24447
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.873178560225803e-07,
+      "loss": 0.6865,
+      "step": 24448
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.868438417196301e-07,
+      "loss": 0.7178,
+      "step": 24449
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.863699481873522e-07,
+      "loss": 0.7012,
+      "step": 24450
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.858961754320272e-07,
+      "loss": 0.6836,
+      "step": 24451
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.854225234599345e-07,
+      "loss": 0.707,
+      "step": 24452
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.84948992277348e-07,
+      "loss": 0.6807,
+      "step": 24453
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.84475581890547e-07,
+      "loss": 0.6973,
+      "step": 24454
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.840022923058067e-07,
+      "loss": 0.7217,
+      "step": 24455
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.835291235293941e-07,
+      "loss": 0.7021,
+      "step": 24456
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.830560755675799e-07,
+      "loss": 0.7119,
+      "step": 24457
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.825831484266379e-07,
+      "loss": 0.6797,
+      "step": 24458
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.821103421128319e-07,
+      "loss": 0.709,
+      "step": 24459
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.816376566324292e-07,
+      "loss": 0.6953,
+      "step": 24460
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.811650919916914e-07,
+      "loss": 0.6836,
+      "step": 24461
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.806926481968869e-07,
+      "loss": 0.6777,
+      "step": 24462
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.802203252542718e-07,
+      "loss": 0.6934,
+      "step": 24463
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.797481231701055e-07,
+      "loss": 0.6816,
+      "step": 24464
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.792760419506463e-07,
+      "loss": 0.7051,
+      "step": 24465
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.788040816021514e-07,
+      "loss": 0.6797,
+      "step": 24466
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.78332242130876e-07,
+      "loss": 0.6875,
+      "step": 24467
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.778605235430737e-07,
+      "loss": 0.7412,
+      "step": 24468
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.773889258449908e-07,
+      "loss": 0.709,
+      "step": 24469
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.769174490428812e-07,
+      "loss": 0.7168,
+      "step": 24470
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.764460931429941e-07,
+      "loss": 0.7109,
+      "step": 24471
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.759748581515725e-07,
+      "loss": 0.7119,
+      "step": 24472
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.755037440748626e-07,
+      "loss": 0.7119,
+      "step": 24473
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.750327509191104e-07,
+      "loss": 0.7139,
+      "step": 24474
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.745618786905563e-07,
+      "loss": 0.6914,
+      "step": 24475
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.740911273954388e-07,
+      "loss": 0.7188,
+      "step": 24476
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.736204970399953e-07,
+      "loss": 0.6895,
+      "step": 24477
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.731499876304672e-07,
+      "loss": 0.6816,
+      "step": 24478
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.726795991730874e-07,
+      "loss": 0.7178,
+      "step": 24479
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.722093316740898e-07,
+      "loss": 0.6885,
+      "step": 24480
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.717391851397061e-07,
+      "loss": 0.7139,
+      "step": 24481
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.71269159576168e-07,
+      "loss": 0.6963,
+      "step": 24482
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.707992549897027e-07,
+      "loss": 0.707,
+      "step": 24483
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.703294713865396e-07,
+      "loss": 0.6943,
+      "step": 24484
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.698598087729027e-07,
+      "loss": 0.7188,
+      "step": 24485
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.693902671550159e-07,
+      "loss": 0.707,
+      "step": 24486
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.68920846539103e-07,
+      "loss": 0.6953,
+      "step": 24487
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.684515469313881e-07,
+      "loss": 0.6973,
+      "step": 24488
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.679823683380828e-07,
+      "loss": 0.6875,
+      "step": 24489
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.6751331076541e-07,
+      "loss": 0.6904,
+      "step": 24490
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.670443742195855e-07,
+      "loss": 0.7021,
+      "step": 24491
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.665755587068236e-07,
+      "loss": 0.6992,
+      "step": 24492
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.66106864233336e-07,
+      "loss": 0.6797,
+      "step": 24493
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.656382908053351e-07,
+      "loss": 0.6846,
+      "step": 24494
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.651698384290308e-07,
+      "loss": 0.6875,
+      "step": 24495
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.647015071106301e-07,
+      "loss": 0.71,
+      "step": 24496
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.642332968563394e-07,
+      "loss": 0.6934,
+      "step": 24497
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.637652076723635e-07,
+      "loss": 0.7031,
+      "step": 24498
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.632972395649076e-07,
+      "loss": 0.6846,
+      "step": 24499
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.628293925401721e-07,
+      "loss": 0.71,
+      "step": 24500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.623616666043588e-07,
+      "loss": 0.708,
+      "step": 24501
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.618940617636617e-07,
+      "loss": 0.6982,
+      "step": 24502
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.614265780242814e-07,
+      "loss": 0.7139,
+      "step": 24503
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.609592153924129e-07,
+      "loss": 0.7031,
+      "step": 24504
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.60491973874249e-07,
+      "loss": 0.6797,
+      "step": 24505
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.600248534759792e-07,
+      "loss": 0.7061,
+      "step": 24506
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.595578542038018e-07,
+      "loss": 0.7373,
+      "step": 24507
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.590909760638977e-07,
+      "loss": 0.7139,
+      "step": 24508
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.586242190624571e-07,
+      "loss": 0.6895,
+      "step": 24509
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.581575832056633e-07,
+      "loss": 0.7129,
+      "step": 24510
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.576910684997042e-07,
+      "loss": 0.6943,
+      "step": 24511
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.572246749507607e-07,
+      "loss": 0.6865,
+      "step": 24512
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.567584025650155e-07,
+      "loss": 0.7158,
+      "step": 24513
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.562922513486415e-07,
+      "loss": 0.7031,
+      "step": 24514
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.558262213078227e-07,
+      "loss": 0.6982,
+      "step": 24515
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.553603124487331e-07,
+      "loss": 0.6885,
+      "step": 24516
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.548945247775464e-07,
+      "loss": 0.6846,
+      "step": 24517
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.544288583004368e-07,
+      "loss": 0.7188,
+      "step": 24518
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.539633130235713e-07,
+      "loss": 0.7021,
+      "step": 24519
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.534978889531265e-07,
+      "loss": 0.6992,
+      "step": 24520
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.530325860952648e-07,
+      "loss": 0.6973,
+      "step": 24521
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.525674044561538e-07,
+      "loss": 0.7139,
+      "step": 24522
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.521023440419585e-07,
+      "loss": 0.6992,
+      "step": 24523
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.516374048588438e-07,
+      "loss": 0.7461,
+      "step": 24524
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.511725869129683e-07,
+      "loss": 0.7148,
+      "step": 24525
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.507078902104948e-07,
+      "loss": 0.6973,
+      "step": 24526
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.502433147575795e-07,
+      "loss": 0.7188,
+      "step": 24527
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.497788605603796e-07,
+      "loss": 0.7217,
+      "step": 24528
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.493145276250503e-07,
+      "loss": 0.7012,
+      "step": 24529
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.488503159577455e-07,
+      "loss": 0.6973,
+      "step": 24530
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.483862255646147e-07,
+      "loss": 0.7139,
+      "step": 24531
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.479222564518119e-07,
+      "loss": 0.6895,
+      "step": 24532
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.474584086254867e-07,
+      "loss": 0.6924,
+      "step": 24533
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.469946820917784e-07,
+      "loss": 0.7119,
+      "step": 24534
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.465310768568413e-07,
+      "loss": 0.6914,
+      "step": 24535
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.460675929268148e-07,
+      "loss": 0.7051,
+      "step": 24536
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.456042303078415e-07,
+      "loss": 0.6953,
+      "step": 24537
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.451409890060624e-07,
+      "loss": 0.7041,
+      "step": 24538
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.446778690276181e-07,
+      "loss": 0.7168,
+      "step": 24539
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.442148703786435e-07,
+      "loss": 0.7217,
+      "step": 24540
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.437519930652749e-07,
+      "loss": 0.7441,
+      "step": 24541
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.432892370936474e-07,
+      "loss": 0.6934,
+      "step": 24542
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.428266024698927e-07,
+      "loss": 0.6934,
+      "step": 24543
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.423640892001428e-07,
+      "loss": 0.7031,
+      "step": 24544
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.41901697290527e-07,
+      "loss": 0.6992,
+      "step": 24545
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.414394267471748e-07,
+      "loss": 0.6973,
+      "step": 24546
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.409772775762071e-07,
+      "loss": 0.6836,
+      "step": 24547
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.40515249783752e-07,
+      "loss": 0.6953,
+      "step": 24548
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.400533433759339e-07,
+      "loss": 0.6914,
+      "step": 24549
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.39591558358871e-07,
+      "loss": 0.6865,
+      "step": 24550
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.391298947386828e-07,
+      "loss": 0.7031,
+      "step": 24551
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.386683525214934e-07,
+      "loss": 0.708,
+      "step": 24552
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.382069317134123e-07,
+      "loss": 0.7148,
+      "step": 24553
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.377456323205568e-07,
+      "loss": 0.7119,
+      "step": 24554
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.372844543490399e-07,
+      "loss": 0.6875,
+      "step": 24555
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.368233978049744e-07,
+      "loss": 0.6895,
+      "step": 24556
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.3636246269447e-07,
+      "loss": 0.7227,
+      "step": 24557
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.35901649023636e-07,
+      "loss": 0.6846,
+      "step": 24558
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.354409567985766e-07,
+      "loss": 0.7061,
+      "step": 24559
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.349803860254002e-07,
+      "loss": 0.6895,
+      "step": 24560
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.345199367102086e-07,
+      "loss": 0.7129,
+      "step": 24561
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.340596088591035e-07,
+      "loss": 0.7197,
+      "step": 24562
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.335994024781857e-07,
+      "loss": 0.6992,
+      "step": 24563
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.331393175735536e-07,
+      "loss": 0.6914,
+      "step": 24564
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.326793541513079e-07,
+      "loss": 0.6895,
+      "step": 24565
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.322195122175403e-07,
+      "loss": 0.6846,
+      "step": 24566
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.317597917783437e-07,
+      "loss": 0.6943,
+      "step": 24567
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.313001928398146e-07,
+      "loss": 0.7148,
+      "step": 24568
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.308407154080411e-07,
+      "loss": 0.6973,
+      "step": 24569
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.30381359489113e-07,
+      "loss": 0.7148,
+      "step": 24570
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.299221250891188e-07,
+      "loss": 0.6885,
+      "step": 24571
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.294630122141434e-07,
+      "loss": 0.6807,
+      "step": 24572
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.29004020870271e-07,
+      "loss": 0.7285,
+      "step": 24573
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.285451510635844e-07,
+      "loss": 0.7012,
+      "step": 24574
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.280864028001656e-07,
+      "loss": 0.6973,
+      "step": 24575
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.276277760860907e-07,
+      "loss": 0.6914,
+      "step": 24576
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.271692709274425e-07,
+      "loss": 0.7002,
+      "step": 24577
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.267108873302964e-07,
+      "loss": 0.6982,
+      "step": 24578
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.262526253007219e-07,
+      "loss": 0.7344,
+      "step": 24579
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.257944848447985e-07,
+      "loss": 0.709,
+      "step": 24580
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.253364659685948e-07,
+      "loss": 0.7217,
+      "step": 24581
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.248785686781802e-07,
+      "loss": 0.7178,
+      "step": 24582
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.244207929796233e-07,
+      "loss": 0.6992,
+      "step": 24583
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.239631388789893e-07,
+      "loss": 0.709,
+      "step": 24584
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.235056063823488e-07,
+      "loss": 0.6777,
+      "step": 24585
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.230481954957592e-07,
+      "loss": 0.7217,
+      "step": 24586
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.225909062252846e-07,
+      "loss": 0.7061,
+      "step": 24587
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.221337385769824e-07,
+      "loss": 0.71,
+      "step": 24588
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.216766925569153e-07,
+      "loss": 0.6992,
+      "step": 24589
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.212197681711376e-07,
+      "loss": 0.709,
+      "step": 24590
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.207629654257088e-07,
+      "loss": 0.709,
+      "step": 24591
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.203062843266751e-07,
+      "loss": 0.7217,
+      "step": 24592
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.19849724880094e-07,
+      "loss": 0.708,
+      "step": 24593
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.193932870920151e-07,
+      "loss": 0.7002,
+      "step": 24594
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.189369709684869e-07,
+      "loss": 0.7002,
+      "step": 24595
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.184807765155556e-07,
+      "loss": 0.708,
+      "step": 24596
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.180247037392685e-07,
+      "loss": 0.7021,
+      "step": 24597
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.17568752645671e-07,
+      "loss": 0.7061,
+      "step": 24598
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.171129232408026e-07,
+      "loss": 0.7031,
+      "step": 24599
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.166572155307028e-07,
+      "loss": 0.6963,
+      "step": 24600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.162016295214148e-07,
+      "loss": 0.7178,
+      "step": 24601
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.157461652189747e-07,
+      "loss": 0.7031,
+      "step": 24602
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.152908226294188e-07,
+      "loss": 0.6768,
+      "step": 24603
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.148356017587799e-07,
+      "loss": 0.7021,
+      "step": 24604
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.143805026130913e-07,
+      "loss": 0.7324,
+      "step": 24605
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.139255251983857e-07,
+      "loss": 0.6982,
+      "step": 24606
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.134706695206907e-07,
+      "loss": 0.7256,
+      "step": 24607
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.130159355860356e-07,
+      "loss": 0.71,
+      "step": 24608
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.125613234004437e-07,
+      "loss": 0.7031,
+      "step": 24609
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.121068329699444e-07,
+      "loss": 0.6963,
+      "step": 24610
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.116524643005596e-07,
+      "loss": 0.7236,
+      "step": 24611
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.111982173983058e-07,
+      "loss": 0.6816,
+      "step": 24612
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.107440922692089e-07,
+      "loss": 0.7031,
+      "step": 24613
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.102900889192844e-07,
+      "loss": 0.7148,
+      "step": 24614
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.098362073545496e-07,
+      "loss": 0.7373,
+      "step": 24615
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.093824475810197e-07,
+      "loss": 0.6973,
+      "step": 24616
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.089288096047066e-07,
+      "loss": 0.6904,
+      "step": 24617
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.084752934316242e-07,
+      "loss": 0.7109,
+      "step": 24618
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.080218990677802e-07,
+      "loss": 0.7021,
+      "step": 24619
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.075686265191852e-07,
+      "loss": 0.7197,
+      "step": 24620
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.071154757918431e-07,
+      "loss": 0.71,
+      "step": 24621
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.066624468917639e-07,
+      "loss": 0.6982,
+      "step": 24622
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.062095398249492e-07,
+      "loss": 0.7041,
+      "step": 24623
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.05756754597401e-07,
+      "loss": 0.6953,
+      "step": 24624
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.053040912151177e-07,
+      "loss": 0.7178,
+      "step": 24625
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.048515496841002e-07,
+      "loss": 0.6895,
+      "step": 24626
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.04399130010347e-07,
+      "loss": 0.7031,
+      "step": 24627
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.039468321998512e-07,
+      "loss": 0.7246,
+      "step": 24628
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.034946562586066e-07,
+      "loss": 0.6865,
+      "step": 24629
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.030426021926096e-07,
+      "loss": 0.7246,
+      "step": 24630
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.025906700078468e-07,
+      "loss": 0.7295,
+      "step": 24631
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.021388597103085e-07,
+      "loss": 0.6865,
+      "step": 24632
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.016871713059815e-07,
+      "loss": 0.6826,
+      "step": 24633
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.01235604800854e-07,
+      "loss": 0.7031,
+      "step": 24634
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.007841602009092e-07,
+      "loss": 0.7256,
+      "step": 24635
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.0033283751213e-07,
+      "loss": 0.6924,
+      "step": 24636
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.99881636740496e-07,
+      "loss": 0.6973,
+      "step": 24637
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.994305578919892e-07,
+      "loss": 0.6729,
+      "step": 24638
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.989796009725847e-07,
+      "loss": 0.6885,
+      "step": 24639
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.985287659882612e-07,
+      "loss": 0.7344,
+      "step": 24640
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.980780529449894e-07,
+      "loss": 0.7012,
+      "step": 24641
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.976274618487478e-07,
+      "loss": 0.6904,
+      "step": 24642
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.971769927055062e-07,
+      "loss": 0.7217,
+      "step": 24643
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.967266455212319e-07,
+      "loss": 0.6963,
+      "step": 24644
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.962764203018924e-07,
+      "loss": 0.6943,
+      "step": 24645
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.958263170534586e-07,
+      "loss": 0.6865,
+      "step": 24646
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.953763357818923e-07,
+      "loss": 0.7119,
+      "step": 24647
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.949264764931574e-07,
+      "loss": 0.6836,
+      "step": 24648
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.944767391932173e-07,
+      "loss": 0.7178,
+      "step": 24649
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.940271238880293e-07,
+      "loss": 0.7051,
+      "step": 24650
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.935776305835529e-07,
+      "loss": 0.707,
+      "step": 24651
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.931282592857459e-07,
+      "loss": 0.6836,
+      "step": 24652
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.926790100005621e-07,
+      "loss": 0.6816,
+      "step": 24653
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.922298827339548e-07,
+      "loss": 0.6943,
+      "step": 24654
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.917808774918778e-07,
+      "loss": 0.7031,
+      "step": 24655
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.913319942802833e-07,
+      "loss": 0.7012,
+      "step": 24656
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.908832331051131e-07,
+      "loss": 0.6943,
+      "step": 24657
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.904345939723213e-07,
+      "loss": 0.6973,
+      "step": 24658
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.899860768878498e-07,
+      "loss": 0.7109,
+      "step": 24659
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.895376818576439e-07,
+      "loss": 0.6826,
+      "step": 24660
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.890894088876455e-07,
+      "loss": 0.7314,
+      "step": 24661
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.886412579837932e-07,
+      "loss": 0.7002,
+      "step": 24662
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.881932291520322e-07,
+      "loss": 0.7002,
+      "step": 24663
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.877453223982934e-07,
+      "loss": 0.7236,
+      "step": 24664
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.872975377285153e-07,
+      "loss": 0.7246,
+      "step": 24665
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.868498751486297e-07,
+      "loss": 0.6816,
+      "step": 24666
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.864023346645733e-07,
+      "loss": 0.6709,
+      "step": 24667
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.859549162822744e-07,
+      "loss": 0.7051,
+      "step": 24668
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.85507620007665e-07,
+      "loss": 0.6963,
+      "step": 24669
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.850604458466682e-07,
+      "loss": 0.7021,
+      "step": 24670
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.846133938052136e-07,
+      "loss": 0.6895,
+      "step": 24671
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.841664638892243e-07,
+      "loss": 0.7295,
+      "step": 24672
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.837196561046245e-07,
+      "loss": 0.7295,
+      "step": 24673
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.832729704573328e-07,
+      "loss": 0.6914,
+      "step": 24674
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.828264069532721e-07,
+      "loss": 0.6924,
+      "step": 24675
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.823799655983611e-07,
+      "loss": 0.708,
+      "step": 24676
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819336463985117e-07,
+      "loss": 0.6885,
+      "step": 24677
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.814874493596403e-07,
+      "loss": 0.7354,
+      "step": 24678
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.810413744876621e-07,
+      "loss": 0.6836,
+      "step": 24679
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.80595421788487e-07,
+      "loss": 0.6963,
+      "step": 24680
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.801495912680246e-07,
+      "loss": 0.709,
+      "step": 24681
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.797038829321846e-07,
+      "loss": 0.6914,
+      "step": 24682
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.792582967868734e-07,
+      "loss": 0.7051,
+      "step": 24683
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.788128328379951e-07,
+      "loss": 0.7441,
+      "step": 24684
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.783674910914529e-07,
+      "loss": 0.7129,
+      "step": 24685
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.779222715531487e-07,
+      "loss": 0.6904,
+      "step": 24686
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.774771742289844e-07,
+      "loss": 0.6855,
+      "step": 24687
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.770321991248575e-07,
+      "loss": 0.6875,
+      "step": 24688
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.765873462466678e-07,
+      "loss": 0.6865,
+      "step": 24689
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.761426156003027e-07,
+      "loss": 0.7002,
+      "step": 24690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.756980071916642e-07,
+      "loss": 0.6934,
+      "step": 24691
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.752535210266399e-07,
+      "loss": 0.6787,
+      "step": 24692
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.748091571111227e-07,
+      "loss": 0.6963,
+      "step": 24693
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.743649154510002e-07,
+      "loss": 0.7168,
+      "step": 24694
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.739207960521599e-07,
+      "loss": 0.6846,
+      "step": 24695
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.734767989204872e-07,
+      "loss": 0.7275,
+      "step": 24696
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.730329240618662e-07,
+      "loss": 0.7051,
+      "step": 24697
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.725891714821787e-07,
+      "loss": 0.6719,
+      "step": 24698
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.721455411873058e-07,
+      "loss": 0.6855,
+      "step": 24699
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.717020331831271e-07,
+      "loss": 0.6738,
+      "step": 24700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.712586474755213e-07,
+      "loss": 0.7031,
+      "step": 24701
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.708153840703625e-07,
+      "loss": 0.6836,
+      "step": 24702
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.70372242973525e-07,
+      "loss": 0.71,
+      "step": 24703
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.699292241908828e-07,
+      "loss": 0.6992,
+      "step": 24704
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.69486327728306e-07,
+      "loss": 0.6807,
+      "step": 24705
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.69043553591663e-07,
+      "loss": 0.71,
+      "step": 24706
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.686009017868224e-07,
+      "loss": 0.6963,
+      "step": 24707
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.681583723196529e-07,
+      "loss": 0.6729,
+      "step": 24708
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.677159651960154e-07,
+      "loss": 0.7021,
+      "step": 24709
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.67273680421774e-07,
+      "loss": 0.6973,
+      "step": 24710
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.668315180027885e-07,
+      "loss": 0.7021,
+      "step": 24711
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.663894779449221e-07,
+      "loss": 0.6992,
+      "step": 24712
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.659475602540312e-07,
+      "loss": 0.7109,
+      "step": 24713
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.655057649359721e-07,
+      "loss": 0.7002,
+      "step": 24714
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.650640919966002e-07,
+      "loss": 0.6895,
+      "step": 24715
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.646225414417674e-07,
+      "loss": 0.7012,
+      "step": 24716
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.641811132773258e-07,
+      "loss": 0.7012,
+      "step": 24717
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.637398075091263e-07,
+      "loss": 0.6963,
+      "step": 24718
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.632986241430151e-07,
+      "loss": 0.7129,
+      "step": 24719
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.628575631848412e-07,
+      "loss": 0.6855,
+      "step": 24720
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.624166246404508e-07,
+      "loss": 0.6914,
+      "step": 24721
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.619758085156836e-07,
+      "loss": 0.7109,
+      "step": 24722
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.615351148163818e-07,
+      "loss": 0.6924,
+      "step": 24723
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.610945435483885e-07,
+      "loss": 0.6973,
+      "step": 24724
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.606540947175412e-07,
+      "loss": 0.6865,
+      "step": 24725
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.602137683296761e-07,
+      "loss": 0.7012,
+      "step": 24726
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.59773564390629e-07,
+      "loss": 0.6875,
+      "step": 24727
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.593334829062338e-07,
+      "loss": 0.7148,
+      "step": 24728
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.588935238823225e-07,
+      "loss": 0.6787,
+      "step": 24729
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.584536873247261e-07,
+      "loss": 0.6982,
+      "step": 24730
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.580139732392722e-07,
+      "loss": 0.7158,
+      "step": 24731
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.575743816317871e-07,
+      "loss": 0.6963,
+      "step": 24732
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.571349125080995e-07,
+      "loss": 0.7275,
+      "step": 24733
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.566955658740338e-07,
+      "loss": 0.6924,
+      "step": 24734
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.562563417354074e-07,
+      "loss": 0.6816,
+      "step": 24735
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.558172400980456e-07,
+      "loss": 0.6943,
+      "step": 24736
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.553782609677662e-07,
+      "loss": 0.7031,
+      "step": 24737
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.549394043503866e-07,
+      "loss": 0.6904,
+      "step": 24738
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.545006702517232e-07,
+      "loss": 0.6826,
+      "step": 24739
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.540620586775893e-07,
+      "loss": 0.6855,
+      "step": 24740
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.536235696337979e-07,
+      "loss": 0.7207,
+      "step": 24741
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.5318520312616e-07,
+      "loss": 0.6826,
+      "step": 24742
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.527469591604852e-07,
+      "loss": 0.7031,
+      "step": 24743
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.523088377425791e-07,
+      "loss": 0.6973,
+      "step": 24744
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.518708388782525e-07,
+      "loss": 0.7393,
+      "step": 24745
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.514329625733064e-07,
+      "loss": 0.7031,
+      "step": 24746
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.509952088335449e-07,
+      "loss": 0.7246,
+      "step": 24747
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.505575776647678e-07,
+      "loss": 0.7031,
+      "step": 24748
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.501200690727773e-07,
+      "loss": 0.6943,
+      "step": 24749
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.496826830633697e-07,
+      "loss": 0.6982,
+      "step": 24750
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.492454196423416e-07,
+      "loss": 0.6953,
+      "step": 24751
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.48808278815486e-07,
+      "loss": 0.6973,
+      "step": 24752
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.483712605886007e-07,
+      "loss": 0.708,
+      "step": 24753
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.479343649674731e-07,
+      "loss": 0.7031,
+      "step": 24754
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.474975919578942e-07,
+      "loss": 0.6982,
+      "step": 24755
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.470609415656516e-07,
+      "loss": 0.6943,
+      "step": 24756
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.466244137965339e-07,
+      "loss": 0.6992,
+      "step": 24757
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.461880086563245e-07,
+      "loss": 0.7119,
+      "step": 24758
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.457517261508074e-07,
+      "loss": 0.7197,
+      "step": 24759
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.453155662857647e-07,
+      "loss": 0.6973,
+      "step": 24760
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.448795290669764e-07,
+      "loss": 0.6934,
+      "step": 24761
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.444436145002198e-07,
+      "loss": 0.7275,
+      "step": 24762
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.440078225912739e-07,
+      "loss": 0.6738,
+      "step": 24763
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.435721533459095e-07,
+      "loss": 0.7207,
+      "step": 24764
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.431366067699064e-07,
+      "loss": 0.7041,
+      "step": 24765
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.427011828690345e-07,
+      "loss": 0.6973,
+      "step": 24766
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.422658816490625e-07,
+      "loss": 0.6797,
+      "step": 24767
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.418307031157579e-07,
+      "loss": 0.6904,
+      "step": 24768
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.413956472748917e-07,
+      "loss": 0.7139,
+      "step": 24769
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.409607141322284e-07,
+      "loss": 0.7148,
+      "step": 24770
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.405259036935309e-07,
+      "loss": 0.6963,
+      "step": 24771
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.400912159645612e-07,
+      "loss": 0.751,
+      "step": 24772
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.396566509510805e-07,
+      "loss": 0.7051,
+      "step": 24773
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.392222086588475e-07,
+      "loss": 0.6631,
+      "step": 24774
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.387878890936207e-07,
+      "loss": 0.7168,
+      "step": 24775
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.383536922611545e-07,
+      "loss": 0.7148,
+      "step": 24776
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.379196181672022e-07,
+      "loss": 0.6777,
+      "step": 24777
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.37485666817519e-07,
+      "loss": 0.6816,
+      "step": 24778
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.37051838217856e-07,
+      "loss": 0.6865,
+      "step": 24779
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.366181323739574e-07,
+      "loss": 0.6768,
+      "step": 24780
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.361845492915765e-07,
+      "loss": 0.6895,
+      "step": 24781
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.357510889764563e-07,
+      "loss": 0.7002,
+      "step": 24782
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.353177514343424e-07,
+      "loss": 0.708,
+      "step": 24783
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.348845366709779e-07,
+      "loss": 0.6943,
+      "step": 24784
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.344514446921003e-07,
+      "loss": 0.6885,
+      "step": 24785
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.340184755034552e-07,
+      "loss": 0.71,
+      "step": 24786
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.335856291107757e-07,
+      "loss": 0.7051,
+      "step": 24787
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.331529055197994e-07,
+      "loss": 0.7002,
+      "step": 24788
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.327203047362596e-07,
+      "loss": 0.7021,
+      "step": 24789
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.322878267658917e-07,
+      "loss": 0.6943,
+      "step": 24790
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.318554716144266e-07,
+      "loss": 0.665,
+      "step": 24791
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.31423239287592e-07,
+      "loss": 0.6875,
+      "step": 24792
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.309911297911188e-07,
+      "loss": 0.7188,
+      "step": 24793
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.305591431307312e-07,
+      "loss": 0.6992,
+      "step": 24794
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.301272793121561e-07,
+      "loss": 0.7168,
+      "step": 24795
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.296955383411141e-07,
+      "loss": 0.6729,
+      "step": 24796
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.292639202233276e-07,
+      "loss": 0.7158,
+      "step": 24797
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.288324249645174e-07,
+      "loss": 0.6914,
+      "step": 24798
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.284010525704044e-07,
+      "loss": 0.708,
+      "step": 24799
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.279698030466997e-07,
+      "loss": 0.6836,
+      "step": 24800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.27538676399121e-07,
+      "loss": 0.6846,
+      "step": 24801
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.271076726333825e-07,
+      "loss": 0.7119,
+      "step": 24802
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.266767917551954e-07,
+      "loss": 0.7041,
+      "step": 24803
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.262460337702693e-07,
+      "loss": 0.7217,
+      "step": 24804
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.258153986843142e-07,
+      "loss": 0.708,
+      "step": 24805
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.253848865030355e-07,
+      "loss": 0.6748,
+      "step": 24806
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.249544972321398e-07,
+      "loss": 0.6924,
+      "step": 24807
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.245242308773292e-07,
+      "loss": 0.7158,
+      "step": 24808
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.240940874443048e-07,
+      "loss": 0.6748,
+      "step": 24809
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.236640669387717e-07,
+      "loss": 0.7188,
+      "step": 24810
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.232341693664246e-07,
+      "loss": 0.6973,
+      "step": 24811
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.228043947329643e-07,
+      "loss": 0.6846,
+      "step": 24812
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.223747430440797e-07,
+      "loss": 0.7051,
+      "step": 24813
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.219452143054706e-07,
+      "loss": 0.7139,
+      "step": 24814
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.21515808522828e-07,
+      "loss": 0.709,
+      "step": 24815
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.210865257018429e-07,
+      "loss": 0.6865,
+      "step": 24816
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.206573658482019e-07,
+      "loss": 0.7041,
+      "step": 24817
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.202283289675949e-07,
+      "loss": 0.7012,
+      "step": 24818
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.197994150657061e-07,
+      "loss": 0.6992,
+      "step": 24819
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.193706241482201e-07,
+      "loss": 0.7012,
+      "step": 24820
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.1894195622082e-07,
+      "loss": 0.6748,
+      "step": 24821
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.185134112891834e-07,
+      "loss": 0.7051,
+      "step": 24822
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.180849893589947e-07,
+      "loss": 0.6963,
+      "step": 24823
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.176566904359295e-07,
+      "loss": 0.7168,
+      "step": 24824
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.17228514525663e-07,
+      "loss": 0.6719,
+      "step": 24825
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.168004616338697e-07,
+      "loss": 0.7334,
+      "step": 24826
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.163725317662218e-07,
+      "loss": 0.6963,
+      "step": 24827
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.159447249283913e-07,
+      "loss": 0.7197,
+      "step": 24828
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.155170411260481e-07,
+      "loss": 0.7051,
+      "step": 24829
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.150894803648567e-07,
+      "loss": 0.7178,
+      "step": 24830
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.146620426504902e-07,
+      "loss": 0.7217,
+      "step": 24831
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.142347279886064e-07,
+      "loss": 0.71,
+      "step": 24832
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.138075363848707e-07,
+      "loss": 0.6699,
+      "step": 24833
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.13380467844943e-07,
+      "loss": 0.6943,
+      "step": 24834
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.129535223744855e-07,
+      "loss": 0.7383,
+      "step": 24835
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.125266999791547e-07,
+      "loss": 0.7041,
+      "step": 24836
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.121000006646083e-07,
+      "loss": 0.6826,
+      "step": 24837
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.116734244364987e-07,
+      "loss": 0.6875,
+      "step": 24838
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.112469713004822e-07,
+      "loss": 0.71,
+      "step": 24839
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.108206412622065e-07,
+      "loss": 0.7148,
+      "step": 24840
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.10394434327325e-07,
+      "loss": 0.7129,
+      "step": 24841
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.099683505014821e-07,
+      "loss": 0.7119,
+      "step": 24842
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.095423897903286e-07,
+      "loss": 0.7021,
+      "step": 24843
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.091165521995091e-07,
+      "loss": 0.7227,
+      "step": 24844
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.086908377346635e-07,
+      "loss": 0.6924,
+      "step": 24845
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.082652464014339e-07,
+      "loss": 0.6846,
+      "step": 24846
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.078397782054647e-07,
+      "loss": 0.7109,
+      "step": 24847
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.074144331523903e-07,
+      "loss": 0.7139,
+      "step": 24848
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.069892112478494e-07,
+      "loss": 0.707,
+      "step": 24849
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.065641124974765e-07,
+      "loss": 0.6934,
+      "step": 24850
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.061391369069048e-07,
+      "loss": 0.6953,
+      "step": 24851
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.057142844817666e-07,
+      "loss": 0.7158,
+      "step": 24852
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.052895552276928e-07,
+      "loss": 0.71,
+      "step": 24853
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.048649491503102e-07,
+      "loss": 0.7227,
+      "step": 24854
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.044404662552484e-07,
+      "loss": 0.7021,
+      "step": 24855
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.04016106548131e-07,
+      "loss": 0.7119,
+      "step": 24856
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.035918700345845e-07,
+      "loss": 0.7021,
+      "step": 24857
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.031677567202255e-07,
+      "loss": 0.7275,
+      "step": 24858
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.027437666106795e-07,
+      "loss": 0.7021,
+      "step": 24859
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.023198997115633e-07,
+      "loss": 0.6904,
+      "step": 24860
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.018961560284954e-07,
+      "loss": 0.7227,
+      "step": 24861
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.014725355670893e-07,
+      "loss": 0.6973,
+      "step": 24862
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.010490383329594e-07,
+      "loss": 0.6748,
+      "step": 24863
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.006256643317222e-07,
+      "loss": 0.7041,
+      "step": 24864
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.002024135689833e-07,
+      "loss": 0.7207,
+      "step": 24865
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.997792860503527e-07,
+      "loss": 0.7109,
+      "step": 24866
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.99356281781437e-07,
+      "loss": 0.707,
+      "step": 24867
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.989334007678461e-07,
+      "loss": 0.7129,
+      "step": 24868
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.985106430151811e-07,
+      "loss": 0.6934,
+      "step": 24869
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.980880085290454e-07,
+      "loss": 0.709,
+      "step": 24870
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.976654973150398e-07,
+      "loss": 0.7178,
+      "step": 24871
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.972431093787624e-07,
+      "loss": 0.7109,
+      "step": 24872
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.96820844725813e-07,
+      "loss": 0.7217,
+      "step": 24873
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.963987033617847e-07,
+      "loss": 0.7139,
+      "step": 24874
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.959766852922734e-07,
+      "loss": 0.709,
+      "step": 24875
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.955547905228732e-07,
+      "loss": 0.7207,
+      "step": 24876
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.951330190591754e-07,
+      "loss": 0.6719,
+      "step": 24877
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.947113709067665e-07,
+      "loss": 0.7441,
+      "step": 24878
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.942898460712344e-07,
+      "loss": 0.7246,
+      "step": 24879
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.938684445581678e-07,
+      "loss": 0.6855,
+      "step": 24880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.934471663731513e-07,
+      "loss": 0.6768,
+      "step": 24881
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.930260115217657e-07,
+      "loss": 0.6709,
+      "step": 24882
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.926049800095935e-07,
+      "loss": 0.7031,
+      "step": 24883
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.921840718422145e-07,
+      "loss": 0.7012,
+      "step": 24884
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.917632870252056e-07,
+      "loss": 0.6973,
+      "step": 24885
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.913426255641453e-07,
+      "loss": 0.6797,
+      "step": 24886
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.909220874646039e-07,
+      "loss": 0.709,
+      "step": 24887
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.905016727321601e-07,
+      "loss": 0.7031,
+      "step": 24888
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.900813813723828e-07,
+      "loss": 0.6904,
+      "step": 24889
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.896612133908442e-07,
+      "loss": 0.6924,
+      "step": 24890
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.892411687931056e-07,
+      "loss": 0.6865,
+      "step": 24891
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.888212475847411e-07,
+      "loss": 0.7373,
+      "step": 24892
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.88401449771312e-07,
+      "loss": 0.7051,
+      "step": 24893
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.879817753583828e-07,
+      "loss": 0.7002,
+      "step": 24894
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.875622243515145e-07,
+      "loss": 0.7051,
+      "step": 24895
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.871427967562683e-07,
+      "loss": 0.7021,
+      "step": 24896
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.867234925782007e-07,
+      "loss": 0.7061,
+      "step": 24897
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.863043118228696e-07,
+      "loss": 0.6777,
+      "step": 24898
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.858852544958284e-07,
+      "loss": 0.7129,
+      "step": 24899
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.854663206026347e-07,
+      "loss": 0.6992,
+      "step": 24900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.850475101488363e-07,
+      "loss": 0.6914,
+      "step": 24901
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.846288231399856e-07,
+      "loss": 0.6934,
+      "step": 24902
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.842102595816314e-07,
+      "loss": 0.6953,
+      "step": 24903
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.83791819479318e-07,
+      "loss": 0.7021,
+      "step": 24904
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.833735028385935e-07,
+      "loss": 0.7158,
+      "step": 24905
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.82955309665001e-07,
+      "loss": 0.7051,
+      "step": 24906
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.825372399640817e-07,
+      "loss": 0.6904,
+      "step": 24907
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.821192937413746e-07,
+      "loss": 0.7119,
+      "step": 24908
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.817014710024239e-07,
+      "loss": 0.71,
+      "step": 24909
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.81283771752761e-07,
+      "loss": 0.707,
+      "step": 24910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.808661959979235e-07,
+      "loss": 0.7188,
+      "step": 24911
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.804487437434438e-07,
+      "loss": 0.7012,
+      "step": 24912
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.800314149948572e-07,
+      "loss": 0.6846,
+      "step": 24913
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.79614209757693e-07,
+      "loss": 0.6934,
+      "step": 24914
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.791971280374798e-07,
+      "loss": 0.6885,
+      "step": 24915
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.787801698397445e-07,
+      "loss": 0.7275,
+      "step": 24916
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.783633351700136e-07,
+      "loss": 0.6982,
+      "step": 24917
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.779466240338095e-07,
+      "loss": 0.6924,
+      "step": 24918
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.775300364366577e-07,
+      "loss": 0.7139,
+      "step": 24919
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.771135723840738e-07,
+      "loss": 0.6826,
+      "step": 24920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.766972318815835e-07,
+      "loss": 0.7256,
+      "step": 24921
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.76281014934701e-07,
+      "loss": 0.7012,
+      "step": 24922
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.75864921548941e-07,
+      "loss": 0.6836,
+      "step": 24923
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.754489517298179e-07,
+      "loss": 0.7158,
+      "step": 24924
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.750331054828463e-07,
+      "loss": 0.6865,
+      "step": 24925
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.746173828135361e-07,
+      "loss": 0.7354,
+      "step": 24926
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.742017837273973e-07,
+      "loss": 0.6943,
+      "step": 24927
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.737863082299357e-07,
+      "loss": 0.708,
+      "step": 24928
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.733709563266589e-07,
+      "loss": 0.6963,
+      "step": 24929
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.729557280230715e-07,
+      "loss": 0.6953,
+      "step": 24930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.725406233246745e-07,
+      "loss": 0.6924,
+      "step": 24931
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.721256422369693e-07,
+      "loss": 0.7002,
+      "step": 24932
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.717107847654581e-07,
+      "loss": 0.6787,
+      "step": 24933
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.712960509156352e-07,
+      "loss": 0.7373,
+      "step": 24934
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.70881440693002e-07,
+      "loss": 0.6963,
+      "step": 24935
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.704669541030451e-07,
+      "loss": 0.7061,
+      "step": 24936
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.700525911512634e-07,
+      "loss": 0.7061,
+      "step": 24937
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.696383518431471e-07,
+      "loss": 0.7021,
+      "step": 24938
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.692242361841838e-07,
+      "loss": 0.7129,
+      "step": 24939
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.68810244179865e-07,
+      "loss": 0.6943,
+      "step": 24940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.683963758356727e-07,
+      "loss": 0.6973,
+      "step": 24941
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.67982631157097e-07,
+      "loss": 0.6699,
+      "step": 24942
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.675690101496158e-07,
+      "loss": 0.6885,
+      "step": 24943
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.671555128187135e-07,
+      "loss": 0.7139,
+      "step": 24944
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.667421391698681e-07,
+      "loss": 0.7178,
+      "step": 24945
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.663288892085596e-07,
+      "loss": 0.6973,
+      "step": 24946
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.659157629402635e-07,
+      "loss": 0.7012,
+      "step": 24947
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.655027603704556e-07,
+      "loss": 0.6865,
+      "step": 24948
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.65089881504608e-07,
+      "loss": 0.6934,
+      "step": 24949
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.646771263481932e-07,
+      "loss": 0.7158,
+      "step": 24950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.642644949066802e-07,
+      "loss": 0.7188,
+      "step": 24951
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.638519871855387e-07,
+      "loss": 0.71,
+      "step": 24952
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.634396031902323e-07,
+      "loss": 0.7021,
+      "step": 24953
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.630273429262313e-07,
+      "loss": 0.7041,
+      "step": 24954
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.626152063989954e-07,
+      "loss": 0.71,
+      "step": 24955
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.622031936139872e-07,
+      "loss": 0.7158,
+      "step": 24956
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.617913045766644e-07,
+      "loss": 0.6816,
+      "step": 24957
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.613795392924893e-07,
+      "loss": 0.6982,
+      "step": 24958
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.609678977669176e-07,
+      "loss": 0.6729,
+      "step": 24959
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.60556380005405e-07,
+      "loss": 0.6895,
+      "step": 24960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.601449860134035e-07,
+      "loss": 0.7236,
+      "step": 24961
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.597337157963668e-07,
+      "loss": 0.7031,
+      "step": 24962
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.593225693597427e-07,
+      "loss": 0.6982,
+      "step": 24963
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.589115467089824e-07,
+      "loss": 0.7109,
+      "step": 24964
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.585006478495304e-07,
+      "loss": 0.7021,
+      "step": 24965
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.580898727868345e-07,
+      "loss": 0.6934,
+      "step": 24966
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.576792215263395e-07,
+      "loss": 0.6855,
+      "step": 24967
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.57268694073484e-07,
+      "loss": 0.7061,
+      "step": 24968
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.568582904337084e-07,
+      "loss": 0.7158,
+      "step": 24969
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.564480106124549e-07,
+      "loss": 0.6885,
+      "step": 24970
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.56037854615158e-07,
+      "loss": 0.7041,
+      "step": 24971
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.556278224472545e-07,
+      "loss": 0.6953,
+      "step": 24972
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.552179141141779e-07,
+      "loss": 0.6934,
+      "step": 24973
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.548081296213604e-07,
+      "loss": 0.6758,
+      "step": 24974
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.543984689742322e-07,
+      "loss": 0.6943,
+      "step": 24975
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.539889321782233e-07,
+      "loss": 0.7041,
+      "step": 24976
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.535795192387584e-07,
+      "loss": 0.7002,
+      "step": 24977
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.531702301612675e-07,
+      "loss": 0.7119,
+      "step": 24978
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.527610649511718e-07,
+      "loss": 0.6914,
+      "step": 24979
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.523520236138947e-07,
+      "loss": 0.709,
+      "step": 24980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.519431061548542e-07,
+      "loss": 0.7139,
+      "step": 24981
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.515343125794738e-07,
+      "loss": 0.6982,
+      "step": 24982
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.511256428931678e-07,
+      "loss": 0.6689,
+      "step": 24983
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.507170971013543e-07,
+      "loss": 0.6924,
+      "step": 24984
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.503086752094456e-07,
+      "loss": 0.7197,
+      "step": 24985
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.499003772228529e-07,
+      "loss": 0.7168,
+      "step": 24986
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.49492203146993e-07,
+      "loss": 0.7178,
+      "step": 24987
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.490841529872694e-07,
+      "loss": 0.6885,
+      "step": 24988
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.486762267490909e-07,
+      "loss": 0.7129,
+      "step": 24989
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.482684244378634e-07,
+      "loss": 0.7051,
+      "step": 24990
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.478607460589936e-07,
+      "loss": 0.708,
+      "step": 24991
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.474531916178828e-07,
+      "loss": 0.708,
+      "step": 24992
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.47045761119931e-07,
+      "loss": 0.7012,
+      "step": 24993
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.466384545705395e-07,
+      "loss": 0.709,
+      "step": 24994
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.46231271975104e-07,
+      "loss": 0.7041,
+      "step": 24995
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.458242133390225e-07,
+      "loss": 0.6826,
+      "step": 24996
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.454172786676882e-07,
+      "loss": 0.7256,
+      "step": 24997
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.450104679664926e-07,
+      "loss": 0.6943,
+      "step": 24998
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.446037812408313e-07,
+      "loss": 0.7295,
+      "step": 24999
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.44197218496091e-07,
+      "loss": 0.7002,
+      "step": 25000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.437907797376597e-07,
+      "loss": 0.7109,
+      "step": 25001
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.43384464970922e-07,
+      "loss": 0.6855,
+      "step": 25002
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.429782742012658e-07,
+      "loss": 0.6914,
+      "step": 25003
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.425722074340724e-07,
+      "loss": 0.7061,
+      "step": 25004
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.421662646747228e-07,
+      "loss": 0.7051,
+      "step": 25005
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.417604459285986e-07,
+      "loss": 0.7109,
+      "step": 25006
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.413547512010753e-07,
+      "loss": 0.6855,
+      "step": 25007
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.409491804975309e-07,
+      "loss": 0.7158,
+      "step": 25008
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.4054373382334e-07,
+      "loss": 0.7168,
+      "step": 25009
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.401384111838737e-07,
+      "loss": 0.6895,
+      "step": 25010
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.397332125845068e-07,
+      "loss": 0.708,
+      "step": 25011
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393281380306071e-07,
+      "loss": 0.6885,
+      "step": 25012
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.389231875275459e-07,
+      "loss": 0.7197,
+      "step": 25013
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.385183610806833e-07,
+      "loss": 0.709,
+      "step": 25014
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.381136586953896e-07,
+      "loss": 0.7178,
+      "step": 25015
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.377090803770259e-07,
+      "loss": 0.7227,
+      "step": 25016
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.373046261309557e-07,
+      "loss": 0.7119,
+      "step": 25017
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.36900295962537e-07,
+      "loss": 0.6982,
+      "step": 25018
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.364960898771278e-07,
+      "loss": 0.707,
+      "step": 25019
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.36092007880087e-07,
+      "loss": 0.7168,
+      "step": 25020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.356880499767682e-07,
+      "loss": 0.6934,
+      "step": 25021
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.352842161725225e-07,
+      "loss": 0.7178,
+      "step": 25022
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.34880506472706e-07,
+      "loss": 0.6875,
+      "step": 25023
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.344769208826674e-07,
+      "loss": 0.7139,
+      "step": 25024
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.340734594077546e-07,
+      "loss": 0.7031,
+      "step": 25025
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.336701220533137e-07,
+      "loss": 0.7236,
+      "step": 25026
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.332669088246912e-07,
+      "loss": 0.7178,
+      "step": 25027
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.328638197272308e-07,
+      "loss": 0.6826,
+      "step": 25028
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.324608547662726e-07,
+      "loss": 0.6826,
+      "step": 25029
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.320580139471577e-07,
+      "loss": 0.7061,
+      "step": 25030
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.316552972752244e-07,
+      "loss": 0.6992,
+      "step": 25031
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.312527047558115e-07,
+      "loss": 0.6914,
+      "step": 25032
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.308502363942526e-07,
+      "loss": 0.7012,
+      "step": 25033
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.304478921958813e-07,
+      "loss": 0.6895,
+      "step": 25034
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.300456721660286e-07,
+      "loss": 0.7021,
+      "step": 25035
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.296435763100273e-07,
+      "loss": 0.7129,
+      "step": 25036
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.292416046332051e-07,
+      "loss": 0.6992,
+      "step": 25037
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.288397571408888e-07,
+      "loss": 0.6855,
+      "step": 25038
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.284380338384044e-07,
+      "loss": 0.7197,
+      "step": 25039
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.280364347310752e-07,
+      "loss": 0.6738,
+      "step": 25040
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.276349598242237e-07,
+      "loss": 0.7275,
+      "step": 25041
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.27233609123169e-07,
+      "loss": 0.7207,
+      "step": 25042
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.268323826332301e-07,
+      "loss": 0.6768,
+      "step": 25043
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.264312803597272e-07,
+      "loss": 0.7188,
+      "step": 25044
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.260303023079739e-07,
+      "loss": 0.6992,
+      "step": 25045
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.256294484832836e-07,
+      "loss": 0.6797,
+      "step": 25046
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.252287188909667e-07,
+      "loss": 0.7393,
+      "step": 25047
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.248281135363376e-07,
+      "loss": 0.7246,
+      "step": 25048
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.244276324247045e-07,
+      "loss": 0.6963,
+      "step": 25049
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.24027275561373e-07,
+      "loss": 0.7178,
+      "step": 25050
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2362704295165e-07,
+      "loss": 0.6846,
+      "step": 25051
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.23226934600839e-07,
+      "loss": 0.7031,
+      "step": 25052
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.228269505142426e-07,
+      "loss": 0.7002,
+      "step": 25053
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.224270906971608e-07,
+      "loss": 0.7168,
+      "step": 25054
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.220273551548927e-07,
+      "loss": 0.6973,
+      "step": 25055
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.216277438927376e-07,
+      "loss": 0.6914,
+      "step": 25056
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2122825691599e-07,
+      "loss": 0.6953,
+      "step": 25057
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.208288942299457e-07,
+      "loss": 0.7207,
+      "step": 25058
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.204296558398926e-07,
+      "loss": 0.707,
+      "step": 25059
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.200305417511265e-07,
+      "loss": 0.7129,
+      "step": 25060
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.196315519689333e-07,
+      "loss": 0.6875,
+      "step": 25061
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.192326864986031e-07,
+      "loss": 0.7012,
+      "step": 25062
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.188339453454195e-07,
+      "loss": 0.6934,
+      "step": 25063
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.184353285146672e-07,
+      "loss": 0.7012,
+      "step": 25064
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.180368360116329e-07,
+      "loss": 0.6895,
+      "step": 25065
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.176384678415914e-07,
+      "loss": 0.7051,
+      "step": 25066
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.172402240098241e-07,
+      "loss": 0.6865,
+      "step": 25067
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.1684210452161e-07,
+      "loss": 0.6904,
+      "step": 25068
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.164441093822248e-07,
+      "loss": 0.6729,
+      "step": 25069
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.160462385969424e-07,
+      "loss": 0.6914,
+      "step": 25070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.15648492171036e-07,
+      "loss": 0.707,
+      "step": 25071
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.152508701097759e-07,
+      "loss": 0.6885,
+      "step": 25072
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.148533724184314e-07,
+      "loss": 0.667,
+      "step": 25073
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.144559991022713e-07,
+      "loss": 0.6895,
+      "step": 25074
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.140587501665607e-07,
+      "loss": 0.7031,
+      "step": 25075
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.136616256165639e-07,
+      "loss": 0.6973,
+      "step": 25076
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.132646254575448e-07,
+      "loss": 0.7021,
+      "step": 25077
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.128677496947666e-07,
+      "loss": 0.6914,
+      "step": 25078
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.124709983334842e-07,
+      "loss": 0.7061,
+      "step": 25079
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.120743713789567e-07,
+      "loss": 0.6924,
+      "step": 25080
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.116778688364433e-07,
+      "loss": 0.7148,
+      "step": 25081
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.112814907111953e-07,
+      "loss": 0.6982,
+      "step": 25082
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.108852370084684e-07,
+      "loss": 0.7148,
+      "step": 25083
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.104891077335106e-07,
+      "loss": 0.708,
+      "step": 25084
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.100931028915747e-07,
+      "loss": 0.7324,
+      "step": 25085
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.096972224879072e-07,
+      "loss": 0.7012,
+      "step": 25086
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.093014665277553e-07,
+      "loss": 0.708,
+      "step": 25087
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.089058350163601e-07,
+      "loss": 0.7041,
+      "step": 25088
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.085103279589699e-07,
+      "loss": 0.6895,
+      "step": 25089
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.081149453608237e-07,
+      "loss": 0.7158,
+      "step": 25090
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.077196872271629e-07,
+      "loss": 0.6729,
+      "step": 25091
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0732455356322e-07,
+      "loss": 0.71,
+      "step": 25092
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.069295443742384e-07,
+      "loss": 0.6924,
+      "step": 25093
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.065346596654486e-07,
+      "loss": 0.7158,
+      "step": 25094
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.061398994420864e-07,
+      "loss": 0.7305,
+      "step": 25095
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.057452637093808e-07,
+      "loss": 0.7129,
+      "step": 25096
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.053507524725632e-07,
+      "loss": 0.6836,
+      "step": 25097
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.049563657368617e-07,
+      "loss": 0.6865,
+      "step": 25098
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.045621035075022e-07,
+      "loss": 0.7109,
+      "step": 25099
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.041679657897093e-07,
+      "loss": 0.6875,
+      "step": 25100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.037739525887076e-07,
+      "loss": 0.6973,
+      "step": 25101
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.033800639097187e-07,
+      "loss": 0.6943,
+      "step": 25102
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.029862997579617e-07,
+      "loss": 0.6914,
+      "step": 25103
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.025926601386544e-07,
+      "loss": 0.7217,
+      "step": 25104
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.021991450570141e-07,
+      "loss": 0.7139,
+      "step": 25105
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.018057545182565e-07,
+      "loss": 0.6885,
+      "step": 25106
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.01412488527594e-07,
+      "loss": 0.709,
+      "step": 25107
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.010193470902382e-07,
+      "loss": 0.7012,
+      "step": 25108
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.00626330211399e-07,
+      "loss": 0.7012,
+      "step": 25109
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.002334378962871e-07,
+      "loss": 0.709,
+      "step": 25110
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.998406701501069e-07,
+      "loss": 0.7031,
+      "step": 25111
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.994480269780622e-07,
+      "loss": 0.7031,
+      "step": 25112
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.990555083853599e-07,
+      "loss": 0.7109,
+      "step": 25113
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.986631143771992e-07,
+      "loss": 0.6816,
+      "step": 25114
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.982708449587826e-07,
+      "loss": 0.7031,
+      "step": 25115
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.978787001353059e-07,
+      "loss": 0.6904,
+      "step": 25116
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.974866799119682e-07,
+      "loss": 0.707,
+      "step": 25117
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.970947842939634e-07,
+      "loss": 0.6699,
+      "step": 25118
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.96703013286486e-07,
+      "loss": 0.6875,
+      "step": 25119
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.963113668947263e-07,
+      "loss": 0.6855,
+      "step": 25120
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.959198451238746e-07,
+      "loss": 0.7188,
+      "step": 25121
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.955284479791212e-07,
+      "loss": 0.7012,
+      "step": 25122
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.951371754656543e-07,
+      "loss": 0.7168,
+      "step": 25123
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.947460275886552e-07,
+      "loss": 0.7188,
+      "step": 25124
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.943550043533075e-07,
+      "loss": 0.71,
+      "step": 25125
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.939641057647971e-07,
+      "loss": 0.6904,
+      "step": 25126
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.935733318283021e-07,
+      "loss": 0.7236,
+      "step": 25127
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.931826825490017e-07,
+      "loss": 0.7002,
+      "step": 25128
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.927921579320706e-07,
+      "loss": 0.7012,
+      "step": 25129
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.92401757982689e-07,
+      "loss": 0.6865,
+      "step": 25130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.920114827060264e-07,
+      "loss": 0.7236,
+      "step": 25131
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.916213321072572e-07,
+      "loss": 0.7109,
+      "step": 25132
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.912313061915487e-07,
+      "loss": 0.6836,
+      "step": 25133
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.90841404964072e-07,
+      "loss": 0.709,
+      "step": 25134
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.904516284299955e-07,
+      "loss": 0.7012,
+      "step": 25135
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.900619765944848e-07,
+      "loss": 0.6973,
+      "step": 25136
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.896724494626982e-07,
+      "loss": 0.6943,
+      "step": 25137
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.892830470398037e-07,
+      "loss": 0.7002,
+      "step": 25138
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.888937693309593e-07,
+      "loss": 0.6885,
+      "step": 25139
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.885046163413244e-07,
+      "loss": 0.6953,
+      "step": 25140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.881155880760559e-07,
+      "loss": 0.6992,
+      "step": 25141
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.877266845403085e-07,
+      "loss": 0.6641,
+      "step": 25142
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.873379057392403e-07,
+      "loss": 0.7148,
+      "step": 25143
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.869492516779984e-07,
+      "loss": 0.7012,
+      "step": 25144
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.865607223617343e-07,
+      "loss": 0.7002,
+      "step": 25145
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.861723177955991e-07,
+      "loss": 0.7129,
+      "step": 25146
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.857840379847391e-07,
+      "loss": 0.6836,
+      "step": 25147
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.853958829342988e-07,
+      "loss": 0.6953,
+      "step": 25148
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.850078526494241e-07,
+      "loss": 0.709,
+      "step": 25149
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.846199471352565e-07,
+      "loss": 0.7188,
+      "step": 25150
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.842321663969363e-07,
+      "loss": 0.708,
+      "step": 25151
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.838445104396018e-07,
+      "loss": 0.7109,
+      "step": 25152
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.834569792683919e-07,
+      "loss": 0.6768,
+      "step": 25153
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.830695728884406e-07,
+      "loss": 0.7021,
+      "step": 25154
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.826822913048836e-07,
+      "loss": 0.6836,
+      "step": 25155
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.822951345228544e-07,
+      "loss": 0.6777,
+      "step": 25156
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.819081025474815e-07,
+      "loss": 0.6973,
+      "step": 25157
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.815211953838917e-07,
+      "loss": 0.6973,
+      "step": 25158
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.811344130372176e-07,
+      "loss": 0.7031,
+      "step": 25159
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.807477555125818e-07,
+      "loss": 0.7041,
+      "step": 25160
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.803612228151089e-07,
+      "loss": 0.6914,
+      "step": 25161
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.79974814949923e-07,
+      "loss": 0.7002,
+      "step": 25162
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.795885319221428e-07,
+      "loss": 0.6953,
+      "step": 25163
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.792023737368879e-07,
+      "loss": 0.6885,
+      "step": 25164
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.788163403992764e-07,
+      "loss": 0.7041,
+      "step": 25165
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.784304319144229e-07,
+      "loss": 0.7109,
+      "step": 25166
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.780446482874436e-07,
+      "loss": 0.7061,
+      "step": 25167
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.776589895234508e-07,
+      "loss": 0.6826,
+      "step": 25168
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.772734556275539e-07,
+      "loss": 0.7012,
+      "step": 25169
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.76888046604861e-07,
+      "loss": 0.707,
+      "step": 25170
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.765027624604835e-07,
+      "loss": 0.7012,
+      "step": 25171
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.761176031995253e-07,
+      "loss": 0.6924,
+      "step": 25172
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.757325688270898e-07,
+      "loss": 0.7119,
+      "step": 25173
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.753476593482798e-07,
+      "loss": 0.7217,
+      "step": 25174
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.74962874768199e-07,
+      "loss": 0.707,
+      "step": 25175
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.745782150919432e-07,
+      "loss": 0.6973,
+      "step": 25176
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.741936803246129e-07,
+      "loss": 0.6836,
+      "step": 25177
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.738092704712994e-07,
+      "loss": 0.7031,
+      "step": 25178
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.734249855371033e-07,
+      "loss": 0.6787,
+      "step": 25179
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.730408255271136e-07,
+      "loss": 0.6777,
+      "step": 25180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.726567904464231e-07,
+      "loss": 0.707,
+      "step": 25181
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.722728803001165e-07,
+      "loss": 0.6924,
+      "step": 25182
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.718890950932876e-07,
+      "loss": 0.7178,
+      "step": 25183
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.715054348310201e-07,
+      "loss": 0.6807,
+      "step": 25184
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.711218995183976e-07,
+      "loss": 0.6914,
+      "step": 25185
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.707384891605039e-07,
+      "loss": 0.7266,
+      "step": 25186
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.703552037624171e-07,
+      "loss": 0.6738,
+      "step": 25187
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.699720433292232e-07,
+      "loss": 0.6836,
+      "step": 25188
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.695890078659938e-07,
+      "loss": 0.6934,
+      "step": 25189
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.692060973778058e-07,
+      "loss": 0.668,
+      "step": 25190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.688233118697373e-07,
+      "loss": 0.71,
+      "step": 25191
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.684406513468577e-07,
+      "loss": 0.7061,
+      "step": 25192
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.680581158142395e-07,
+      "loss": 0.7217,
+      "step": 25193
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.676757052769522e-07,
+      "loss": 0.6973,
+      "step": 25194
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.672934197400637e-07,
+      "loss": 0.7012,
+      "step": 25195
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.66911259208639e-07,
+      "loss": 0.6914,
+      "step": 25196
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.66529223687744e-07,
+      "loss": 0.708,
+      "step": 25197
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.661473131824413e-07,
+      "loss": 0.7051,
+      "step": 25198
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.657655276977892e-07,
+      "loss": 0.707,
+      "step": 25199
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.653838672388523e-07,
+      "loss": 0.7139,
+      "step": 25200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.650023318106879e-07,
+      "loss": 0.6914,
+      "step": 25201
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.646209214183484e-07,
+      "loss": 0.6865,
+      "step": 25202
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.64239636066889e-07,
+      "loss": 0.709,
+      "step": 25203
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.638584757613663e-07,
+      "loss": 0.7021,
+      "step": 25204
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.634774405068288e-07,
+      "loss": 0.7178,
+      "step": 25205
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.630965303083258e-07,
+      "loss": 0.6973,
+      "step": 25206
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.627157451709075e-07,
+      "loss": 0.7061,
+      "step": 25207
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.623350850996179e-07,
+      "loss": 0.6865,
+      "step": 25208
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.619545500995027e-07,
+      "loss": 0.7168,
+      "step": 25209
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.615741401756047e-07,
+      "loss": 0.6992,
+      "step": 25210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.611938553329644e-07,
+      "loss": 0.6846,
+      "step": 25211
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.60813695576623e-07,
+      "loss": 0.707,
+      "step": 25212
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.604336609116179e-07,
+      "loss": 0.6973,
+      "step": 25213
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.60053751342986e-07,
+      "loss": 0.6943,
+      "step": 25214
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5967396687576e-07,
+      "loss": 0.7275,
+      "step": 25215
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.592943075149748e-07,
+      "loss": 0.6777,
+      "step": 25216
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.589147732656608e-07,
+      "loss": 0.7119,
+      "step": 25217
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.585353641328495e-07,
+      "loss": 0.7178,
+      "step": 25218
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.581560801215668e-07,
+      "loss": 0.6904,
+      "step": 25219
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.577769212368401e-07,
+      "loss": 0.707,
+      "step": 25220
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.573978874836938e-07,
+      "loss": 0.7051,
+      "step": 25221
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.57018978867151e-07,
+      "loss": 0.6758,
+      "step": 25222
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.566401953922318e-07,
+      "loss": 0.6924,
+      "step": 25223
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.56261537063959e-07,
+      "loss": 0.7012,
+      "step": 25224
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.558830038873497e-07,
+      "loss": 0.6934,
+      "step": 25225
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.555045958674199e-07,
+      "loss": 0.708,
+      "step": 25226
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.551263130091833e-07,
+      "loss": 0.709,
+      "step": 25227
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.547481553176548e-07,
+      "loss": 0.6631,
+      "step": 25228
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.543701227978449e-07,
+      "loss": 0.7168,
+      "step": 25229
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.539922154547639e-07,
+      "loss": 0.7119,
+      "step": 25230
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.53614433293419e-07,
+      "loss": 0.7021,
+      "step": 25231
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.532367763188163e-07,
+      "loss": 0.7178,
+      "step": 25232
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.528592445359649e-07,
+      "loss": 0.7217,
+      "step": 25233
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.524818379498631e-07,
+      "loss": 0.6836,
+      "step": 25234
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.521045565655125e-07,
+      "loss": 0.667,
+      "step": 25235
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.517274003879158e-07,
+      "loss": 0.7148,
+      "step": 25236
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.513503694220712e-07,
+      "loss": 0.7012,
+      "step": 25237
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.509734636729724e-07,
+      "loss": 0.7129,
+      "step": 25238
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.505966831456178e-07,
+      "loss": 0.7324,
+      "step": 25239
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.502200278449976e-07,
+      "loss": 0.6973,
+      "step": 25240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.498434977761047e-07,
+      "loss": 0.7129,
+      "step": 25241
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.494670929439283e-07,
+      "loss": 0.7021,
+      "step": 25242
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.490908133534578e-07,
+      "loss": 0.7207,
+      "step": 25243
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.48714659009677e-07,
+      "loss": 0.7373,
+      "step": 25244
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.483386299175753e-07,
+      "loss": 0.7178,
+      "step": 25245
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.479627260821352e-07,
+      "loss": 0.6855,
+      "step": 25246
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.475869475083339e-07,
+      "loss": 0.7031,
+      "step": 25247
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.472112942011543e-07,
+      "loss": 0.6934,
+      "step": 25248
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.468357661655754e-07,
+      "loss": 0.6904,
+      "step": 25249
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.464603634065735e-07,
+      "loss": 0.7285,
+      "step": 25250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.460850859291222e-07,
+      "loss": 0.7256,
+      "step": 25251
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.457099337381944e-07,
+      "loss": 0.6934,
+      "step": 25252
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.453349068387659e-07,
+      "loss": 0.6963,
+      "step": 25253
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.449600052358028e-07,
+      "loss": 0.6914,
+      "step": 25254
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.445852289342745e-07,
+      "loss": 0.6865,
+      "step": 25255
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.44210577939146e-07,
+      "loss": 0.6904,
+      "step": 25256
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.438360522553854e-07,
+      "loss": 0.7275,
+      "step": 25257
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.434616518879543e-07,
+      "loss": 0.7061,
+      "step": 25258
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.430873768418166e-07,
+      "loss": 0.6973,
+      "step": 25259
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.427132271219271e-07,
+      "loss": 0.6924,
+      "step": 25260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.423392027332486e-07,
+      "loss": 0.7275,
+      "step": 25261
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.419653036807371e-07,
+      "loss": 0.7207,
+      "step": 25262
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.415915299693464e-07,
+      "loss": 0.7031,
+      "step": 25263
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.412178816040314e-07,
+      "loss": 0.6943,
+      "step": 25264
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.408443585897405e-07,
+      "loss": 0.7051,
+      "step": 25265
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.404709609314307e-07,
+      "loss": 0.6934,
+      "step": 25266
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.400976886340437e-07,
+      "loss": 0.709,
+      "step": 25267
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.397245417025276e-07,
+      "loss": 0.7051,
+      "step": 25268
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.393515201418287e-07,
+      "loss": 0.6943,
+      "step": 25269
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.389786239568918e-07,
+      "loss": 0.7197,
+      "step": 25270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.38605853152655e-07,
+      "loss": 0.7002,
+      "step": 25271
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.382332077340613e-07,
+      "loss": 0.7021,
+      "step": 25272
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.378606877060488e-07,
+      "loss": 0.708,
+      "step": 25273
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.374882930735526e-07,
+      "loss": 0.7168,
+      "step": 25274
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.371160238415097e-07,
+      "loss": 0.7061,
+      "step": 25275
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.367438800148517e-07,
+      "loss": 0.7109,
+      "step": 25276
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.363718615985114e-07,
+      "loss": 0.6826,
+      "step": 25277
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.359999685974194e-07,
+      "loss": 0.6963,
+      "step": 25278
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.356282010165048e-07,
+      "loss": 0.6865,
+      "step": 25279
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.352565588606907e-07,
+      "loss": 0.7168,
+      "step": 25280
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.348850421349061e-07,
+      "loss": 0.6865,
+      "step": 25281
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.345136508440729e-07,
+      "loss": 0.6982,
+      "step": 25282
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.341423849931127e-07,
+      "loss": 0.6943,
+      "step": 25283
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.33771244586947e-07,
+      "loss": 0.7061,
+      "step": 25284
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.33400229630493e-07,
+      "loss": 0.7217,
+      "step": 25285
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.33029340128668e-07,
+      "loss": 0.7021,
+      "step": 25286
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.326585760863867e-07,
+      "loss": 0.6904,
+      "step": 25287
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.322879375085621e-07,
+      "loss": 0.7197,
+      "step": 25288
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.319174244001069e-07,
+      "loss": 0.7188,
+      "step": 25289
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.315470367659314e-07,
+      "loss": 0.6924,
+      "step": 25290
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.311767746109453e-07,
+      "loss": 0.6855,
+      "step": 25291
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.308066379400545e-07,
+      "loss": 0.7207,
+      "step": 25292
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.304366267581607e-07,
+      "loss": 0.7031,
+      "step": 25293
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.300667410701721e-07,
+      "loss": 0.71,
+      "step": 25294
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.296969808809893e-07,
+      "loss": 0.6973,
+      "step": 25295
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.293273461955118e-07,
+      "loss": 0.7061,
+      "step": 25296
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.289578370186366e-07,
+      "loss": 0.7021,
+      "step": 25297
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.285884533552655e-07,
+      "loss": 0.6914,
+      "step": 25298
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.282191952102889e-07,
+      "loss": 0.665,
+      "step": 25299
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.27850062588603e-07,
+      "loss": 0.6699,
+      "step": 25300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.27481055495096e-07,
+      "loss": 0.7139,
+      "step": 25301
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.27112173934663e-07,
+      "loss": 0.6934,
+      "step": 25302
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.267434179121899e-07,
+      "loss": 0.7139,
+      "step": 25303
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.263747874325642e-07,
+      "loss": 0.6846,
+      "step": 25304
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.260062825006706e-07,
+      "loss": 0.6963,
+      "step": 25305
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.256379031213932e-07,
+      "loss": 0.71,
+      "step": 25306
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.252696492996134e-07,
+      "loss": 0.7207,
+      "step": 25307
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.24901521040212e-07,
+      "loss": 0.7109,
+      "step": 25308
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.245335183480682e-07,
+      "loss": 0.6973,
+      "step": 25309
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.241656412280549e-07,
+      "loss": 0.6924,
+      "step": 25310
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.237978896850537e-07,
+      "loss": 0.7012,
+      "step": 25311
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.234302637239341e-07,
+      "loss": 0.7012,
+      "step": 25312
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.230627633495666e-07,
+      "loss": 0.751,
+      "step": 25313
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.226953885668251e-07,
+      "loss": 0.6934,
+      "step": 25314
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.223281393805768e-07,
+      "loss": 0.708,
+      "step": 25315
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.21961015795689e-07,
+      "loss": 0.707,
+      "step": 25316
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.215940178170265e-07,
+      "loss": 0.6982,
+      "step": 25317
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.212271454494522e-07,
+      "loss": 0.7129,
+      "step": 25318
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.208603986978289e-07,
+      "loss": 0.71,
+      "step": 25319
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.20493777567016e-07,
+      "loss": 0.7021,
+      "step": 25320
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.20127282061873e-07,
+      "loss": 0.6865,
+      "step": 25321
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.197609121872549e-07,
+      "loss": 0.7139,
+      "step": 25322
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.193946679480211e-07,
+      "loss": 0.6865,
+      "step": 25323
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.190285493490232e-07,
+      "loss": 0.71,
+      "step": 25324
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.186625563951098e-07,
+      "loss": 0.7158,
+      "step": 25325
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.182966890911356e-07,
+      "loss": 0.7002,
+      "step": 25326
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.179309474419469e-07,
+      "loss": 0.6865,
+      "step": 25327
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.175653314523921e-07,
+      "loss": 0.7041,
+      "step": 25328
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.171998411273149e-07,
+      "loss": 0.6904,
+      "step": 25329
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.168344764715583e-07,
+      "loss": 0.7041,
+      "step": 25330
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.164692374899693e-07,
+      "loss": 0.7021,
+      "step": 25331
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.161041241873821e-07,
+      "loss": 0.6875,
+      "step": 25332
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.157391365686382e-07,
+      "loss": 0.6953,
+      "step": 25333
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.153742746385726e-07,
+      "loss": 0.707,
+      "step": 25334
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.150095384020238e-07,
+      "loss": 0.6914,
+      "step": 25335
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.146449278638221e-07,
+      "loss": 0.7051,
+      "step": 25336
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.14280443028804e-07,
+      "loss": 0.71,
+      "step": 25337
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.139160839017932e-07,
+      "loss": 0.6885,
+      "step": 25338
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.135518504876236e-07,
+      "loss": 0.7129,
+      "step": 25339
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.131877427911192e-07,
+      "loss": 0.6943,
+      "step": 25340
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.128237608171083e-07,
+      "loss": 0.6963,
+      "step": 25341
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.124599045704104e-07,
+      "loss": 0.6963,
+      "step": 25342
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.120961740558505e-07,
+      "loss": 0.7031,
+      "step": 25343
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.117325692782504e-07,
+      "loss": 0.7119,
+      "step": 25344
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.11369090242424e-07,
+      "loss": 0.6895,
+      "step": 25345
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.110057369531906e-07,
+      "loss": 0.6943,
+      "step": 25346
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.106425094153655e-07,
+      "loss": 0.6924,
+      "step": 25347
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.102794076337636e-07,
+      "loss": 0.7041,
+      "step": 25348
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.099164316131955e-07,
+      "loss": 0.6758,
+      "step": 25349
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.095535813584706e-07,
+      "loss": 0.7246,
+      "step": 25350
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.091908568743997e-07,
+      "loss": 0.6943,
+      "step": 25351
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.088282581657878e-07,
+      "loss": 0.6895,
+      "step": 25352
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.084657852374419e-07,
+      "loss": 0.6885,
+      "step": 25353
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.08103438094163e-07,
+      "loss": 0.6836,
+      "step": 25354
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.077412167407547e-07,
+      "loss": 0.7061,
+      "step": 25355
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.073791211820178e-07,
+      "loss": 0.7031,
+      "step": 25356
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.070171514227529e-07,
+      "loss": 0.7363,
+      "step": 25357
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.066553074677505e-07,
+      "loss": 0.6885,
+      "step": 25358
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.062935893218123e-07,
+      "loss": 0.7158,
+      "step": 25359
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.059319969897292e-07,
+      "loss": 0.7119,
+      "step": 25360
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.055705304762926e-07,
+      "loss": 0.6953,
+      "step": 25361
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.052091897862943e-07,
+      "loss": 0.6982,
+      "step": 25362
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.048479749245216e-07,
+      "loss": 0.7285,
+      "step": 25363
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.04486885895763e-07,
+      "loss": 0.7168,
+      "step": 25364
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.041259227048023e-07,
+      "loss": 0.6904,
+      "step": 25365
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.037650853564236e-07,
+      "loss": 0.7168,
+      "step": 25366
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.034043738554074e-07,
+      "loss": 0.6953,
+      "step": 25367
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.030437882065364e-07,
+      "loss": 0.7002,
+      "step": 25368
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.026833284145893e-07,
+      "loss": 0.707,
+      "step": 25369
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.023229944843421e-07,
+      "loss": 0.6797,
+      "step": 25370
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.019627864205679e-07,
+      "loss": 0.709,
+      "step": 25371
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.016027042280436e-07,
+      "loss": 0.7148,
+      "step": 25372
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0124274791154e-07,
+      "loss": 0.6855,
+      "step": 25373
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.008829174758267e-07,
+      "loss": 0.7412,
+      "step": 25374
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.005232129256721e-07,
+      "loss": 0.7041,
+      "step": 25375
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.001636342658467e-07,
+      "loss": 0.6943,
+      "step": 25376
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.998041815011112e-07,
+      "loss": 0.6875,
+      "step": 25377
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.994448546362307e-07,
+      "loss": 0.6816,
+      "step": 25378
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.990856536759658e-07,
+      "loss": 0.7012,
+      "step": 25379
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.987265786250806e-07,
+      "loss": 0.6924,
+      "step": 25380
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.983676294883299e-07,
+      "loss": 0.6826,
+      "step": 25381
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.980088062704735e-07,
+      "loss": 0.6826,
+      "step": 25382
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.976501089762642e-07,
+      "loss": 0.6865,
+      "step": 25383
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.972915376104571e-07,
+      "loss": 0.7178,
+      "step": 25384
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.969330921778037e-07,
+      "loss": 0.6846,
+      "step": 25385
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.96574772683055e-07,
+      "loss": 0.6865,
+      "step": 25386
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.962165791309559e-07,
+      "loss": 0.7168,
+      "step": 25387
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.958585115262582e-07,
+      "loss": 0.6924,
+      "step": 25388
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.955005698737059e-07,
+      "loss": 0.7002,
+      "step": 25389
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.951427541780418e-07,
+      "loss": 0.7207,
+      "step": 25390
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.947850644440044e-07,
+      "loss": 0.6982,
+      "step": 25391
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.944275006763399e-07,
+      "loss": 0.6836,
+      "step": 25392
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.940700628797845e-07,
+      "loss": 0.7129,
+      "step": 25393
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.937127510590744e-07,
+      "loss": 0.7002,
+      "step": 25394
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.933555652189448e-07,
+      "loss": 0.7168,
+      "step": 25395
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.929985053641295e-07,
+      "loss": 0.6963,
+      "step": 25396
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.926415714993616e-07,
+      "loss": 0.6748,
+      "step": 25397
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.922847636293692e-07,
+      "loss": 0.6924,
+      "step": 25398
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.919280817588834e-07,
+      "loss": 0.7119,
+      "step": 25399
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.915715258926269e-07,
+      "loss": 0.7266,
+      "step": 25400
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.91215096035329e-07,
+      "loss": 0.709,
+      "step": 25401
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.908587921917141e-07,
+      "loss": 0.7178,
+      "step": 25402
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.905026143664982e-07,
+      "loss": 0.6934,
+      "step": 25403
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.901465625644064e-07,
+      "loss": 0.6953,
+      "step": 25404
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.897906367901573e-07,
+      "loss": 0.6797,
+      "step": 25405
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.7188,
+      "step": 25406
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.890791633440462e-07,
+      "loss": 0.7148,
+      "step": 25407
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.887236156816144e-07,
+      "loss": 0.6992,
+      "step": 25408
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.883681940658814e-07,
+      "loss": 0.6973,
+      "step": 25409
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.880128985015564e-07,
+      "loss": 0.7051,
+      "step": 25410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.876577289933493e-07,
+      "loss": 0.6875,
+      "step": 25411
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.87302685545965e-07,
+      "loss": 0.7178,
+      "step": 25412
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.86947768164111e-07,
+      "loss": 0.6748,
+      "step": 25413
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.865929768524913e-07,
+      "loss": 0.6992,
+      "step": 25414
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.862383116158065e-07,
+      "loss": 0.7139,
+      "step": 25415
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.858837724587528e-07,
+      "loss": 0.7041,
+      "step": 25416
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.855293593860355e-07,
+      "loss": 0.6836,
+      "step": 25417
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.851750724023475e-07,
+      "loss": 0.6982,
+      "step": 25418
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.848209115123848e-07,
+      "loss": 0.6973,
+      "step": 25419
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.844668767208383e-07,
+      "loss": 0.7041,
+      "step": 25420
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.841129680324064e-07,
+      "loss": 0.6895,
+      "step": 25421
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.837591854517732e-07,
+      "loss": 0.6748,
+      "step": 25422
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.834055289836292e-07,
+      "loss": 0.7324,
+      "step": 25423
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.830519986326588e-07,
+      "loss": 0.6885,
+      "step": 25424
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.826985944035512e-07,
+      "loss": 0.7119,
+      "step": 25425
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.823453163009884e-07,
+      "loss": 0.6943,
+      "step": 25426
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.81992164329651e-07,
+      "loss": 0.6865,
+      "step": 25427
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.816391384942199e-07,
+      "loss": 0.7021,
+      "step": 25428
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.812862387993733e-07,
+      "loss": 0.6943,
+      "step": 25429
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.809334652497888e-07,
+      "loss": 0.6709,
+      "step": 25430
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.805808178501404e-07,
+      "loss": 0.7031,
+      "step": 25431
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.802282966051009e-07,
+      "loss": 0.707,
+      "step": 25432
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.798759015193421e-07,
+      "loss": 0.7236,
+      "step": 25433
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.795236325975383e-07,
+      "loss": 0.7207,
+      "step": 25434
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.791714898443522e-07,
+      "loss": 0.6895,
+      "step": 25435
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.788194732644525e-07,
+      "loss": 0.7197,
+      "step": 25436
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.784675828625052e-07,
+      "loss": 0.71,
+      "step": 25437
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.781158186431733e-07,
+      "loss": 0.7002,
+      "step": 25438
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.777641806111177e-07,
+      "loss": 0.7051,
+      "step": 25439
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.77412668770999e-07,
+      "loss": 0.6953,
+      "step": 25440
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.770612831274768e-07,
+      "loss": 0.6895,
+      "step": 25441
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7671002368520515e-07,
+      "loss": 0.7168,
+      "step": 25442
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7635889044884144e-07,
+      "loss": 0.7236,
+      "step": 25443
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.760078834230364e-07,
+      "loss": 0.6846,
+      "step": 25444
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.75657002612443e-07,
+      "loss": 0.7158,
+      "step": 25445
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.75306248021713e-07,
+      "loss": 0.6943,
+      "step": 25446
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.74955619655495e-07,
+      "loss": 0.6934,
+      "step": 25447
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7460511751842965e-07,
+      "loss": 0.6982,
+      "step": 25448
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7425474161516884e-07,
+      "loss": 0.7305,
+      "step": 25449
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7390449195035325e-07,
+      "loss": 0.7002,
+      "step": 25450
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.735543685286237e-07,
+      "loss": 0.7305,
+      "step": 25451
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7320437135462083e-07,
+      "loss": 0.6797,
+      "step": 25452
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.728545004329821e-07,
+      "loss": 0.707,
+      "step": 25453
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.725047557683471e-07,
+      "loss": 0.6934,
+      "step": 25454
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.721551373653477e-07,
+      "loss": 0.7012,
+      "step": 25455
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7180564522861903e-07,
+      "loss": 0.7344,
+      "step": 25456
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7145627936278857e-07,
+      "loss": 0.6826,
+      "step": 25457
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.711070397724926e-07,
+      "loss": 0.7217,
+      "step": 25458
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7075792646235517e-07,
+      "loss": 0.7021,
+      "step": 25459
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7040893943700595e-07,
+      "loss": 0.6973,
+      "step": 25460
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7006007870106453e-07,
+      "loss": 0.6895,
+      "step": 25461
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.697113442591583e-07,
+      "loss": 0.6953,
+      "step": 25462
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.693627361159081e-07,
+      "loss": 0.7168,
+      "step": 25463
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.690142542759346e-07,
+      "loss": 0.6777,
+      "step": 25464
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6866589874385303e-07,
+      "loss": 0.7002,
+      "step": 25465
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.68317669524283e-07,
+      "loss": 0.7158,
+      "step": 25466
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.679695666218398e-07,
+      "loss": 0.6865,
+      "step": 25467
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.676215900411329e-07,
+      "loss": 0.7012,
+      "step": 25468
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.672737397867755e-07,
+      "loss": 0.6729,
+      "step": 25469
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6692601586337926e-07,
+      "loss": 0.667,
+      "step": 25470
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6657841827555064e-07,
+      "loss": 0.7041,
+      "step": 25471
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6623094702789693e-07,
+      "loss": 0.6982,
+      "step": 25472
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6588360212502125e-07,
+      "loss": 0.6953,
+      "step": 25473
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6553638357152985e-07,
+      "loss": 0.6992,
+      "step": 25474
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6518929137202017e-07,
+      "loss": 0.7197,
+      "step": 25475
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.648423255310952e-07,
+      "loss": 0.7217,
+      "step": 25476
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.644954860533524e-07,
+      "loss": 0.7266,
+      "step": 25477
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.641487729433858e-07,
+      "loss": 0.7031,
+      "step": 25478
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6380218620579396e-07,
+      "loss": 0.7314,
+      "step": 25479
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6345572584516884e-07,
+      "loss": 0.6963,
+      "step": 25480
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6310939186609895e-07,
+      "loss": 0.7148,
+      "step": 25481
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6276318427317724e-07,
+      "loss": 0.7041,
+      "step": 25482
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.624171030709901e-07,
+      "loss": 0.709,
+      "step": 25483
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6207114826412604e-07,
+      "loss": 0.6982,
+      "step": 25484
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6172531985716696e-07,
+      "loss": 0.6904,
+      "step": 25485
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6137961785469807e-07,
+      "loss": 0.7021,
+      "step": 25486
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.610340422612991e-07,
+      "loss": 0.7178,
+      "step": 25487
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.606885930815519e-07,
+      "loss": 0.7119,
+      "step": 25488
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6034327032003166e-07,
+      "loss": 0.6836,
+      "step": 25489
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5999807398131593e-07,
+      "loss": 0.6924,
+      "step": 25490
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.596530040699798e-07,
+      "loss": 0.6846,
+      "step": 25491
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5930806059059643e-07,
+      "loss": 0.7236,
+      "step": 25492
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.589632435477376e-07,
+      "loss": 0.6963,
+      "step": 25493
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.586185529459719e-07,
+      "loss": 0.7197,
+      "step": 25494
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5827398878986685e-07,
+      "loss": 0.6836,
+      "step": 25495
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5792955108398987e-07,
+      "loss": 0.7119,
+      "step": 25496
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.575852398329039e-07,
+      "loss": 0.6982,
+      "step": 25497
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.572410550411732e-07,
+      "loss": 0.6924,
+      "step": 25498
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5689699671336075e-07,
+      "loss": 0.6963,
+      "step": 25499
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5655306485402287e-07,
+      "loss": 0.708,
+      "step": 25500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.562092594677192e-07,
+      "loss": 0.6816,
+      "step": 25501
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.558655805590051e-07,
+      "loss": 0.7021,
+      "step": 25502
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.555220281324357e-07,
+      "loss": 0.6934,
+      "step": 25503
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5517860219256414e-07,
+      "loss": 0.6982,
+      "step": 25504
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5483530274394117e-07,
+      "loss": 0.6934,
+      "step": 25505
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5449212979111647e-07,
+      "loss": 0.6826,
+      "step": 25506
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.541490833386375e-07,
+      "loss": 0.708,
+      "step": 25507
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.538061633910518e-07,
+      "loss": 0.7139,
+      "step": 25508
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.534633699529023e-07,
+      "loss": 0.6895,
+      "step": 25509
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.53120703028731e-07,
+      "loss": 0.709,
+      "step": 25510
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5277816262308094e-07,
+      "loss": 0.7002,
+      "step": 25511
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.52435748740494e-07,
+      "loss": 0.7012,
+      "step": 25512
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5209346138550214e-07,
+      "loss": 0.7295,
+      "step": 25513
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.517513005626439e-07,
+      "loss": 0.6904,
+      "step": 25514
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5140926627645464e-07,
+      "loss": 0.7109,
+      "step": 25515
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.510673585314673e-07,
+      "loss": 0.7139,
+      "step": 25516
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.507255773322128e-07,
+      "loss": 0.7168,
+      "step": 25517
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.503839226832185e-07,
+      "loss": 0.6689,
+      "step": 25518
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5004239458901424e-07,
+      "loss": 0.6826,
+      "step": 25519
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4970099305412626e-07,
+      "loss": 0.709,
+      "step": 25520
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4935971808307776e-07,
+      "loss": 0.7129,
+      "step": 25521
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.490185696803917e-07,
+      "loss": 0.6904,
+      "step": 25522
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.486775478505878e-07,
+      "loss": 0.6885,
+      "step": 25523
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4833665259818804e-07,
+      "loss": 0.7344,
+      "step": 25524
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4799588392770985e-07,
+      "loss": 0.7197,
+      "step": 25525
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.476552418436664e-07,
+      "loss": 0.7109,
+      "step": 25526
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4731472635057396e-07,
+      "loss": 0.7217,
+      "step": 25527
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4697433745294673e-07,
+      "loss": 0.71,
+      "step": 25528
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4663407515529225e-07,
+      "loss": 0.6943,
+      "step": 25529
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.462939394621224e-07,
+      "loss": 0.7305,
+      "step": 25530
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4595393037794256e-07,
+      "loss": 0.6885,
+      "step": 25531
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.456140479072624e-07,
+      "loss": 0.6924,
+      "step": 25532
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.452742920545816e-07,
+      "loss": 0.7168,
+      "step": 25533
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4493466282440555e-07,
+      "loss": 0.6709,
+      "step": 25534
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4459516022123285e-07,
+      "loss": 0.7334,
+      "step": 25535
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.442557842495665e-07,
+      "loss": 0.7178,
+      "step": 25536
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4391653491390074e-07,
+      "loss": 0.7158,
+      "step": 25537
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4357741221873306e-07,
+      "loss": 0.708,
+      "step": 25538
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4323841616855655e-07,
+      "loss": 0.7285,
+      "step": 25539
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.428995467678643e-07,
+      "loss": 0.6865,
+      "step": 25540
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.425608040211471e-07,
+      "loss": 0.707,
+      "step": 25541
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.422221879328936e-07,
+      "loss": 0.6836,
+      "step": 25542
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.418836985075914e-07,
+      "loss": 0.6924,
+      "step": 25543
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.415453357497268e-07,
+      "loss": 0.7061,
+      "step": 25544
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.412070996637852e-07,
+      "loss": 0.71,
+      "step": 25545
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4086899025424625e-07,
+      "loss": 0.7031,
+      "step": 25546
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4053100752559087e-07,
+      "loss": 0.7109,
+      "step": 25547
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.401931514822999e-07,
+      "loss": 0.71,
+      "step": 25548
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.398554221288498e-07,
+      "loss": 0.7129,
+      "step": 25549
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3951781946971697e-07,
+      "loss": 0.7061,
+      "step": 25550
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3918034350937445e-07,
+      "loss": 0.7041,
+      "step": 25551
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3884299425229537e-07,
+      "loss": 0.6836,
+      "step": 25552
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.385057717029506e-07,
+      "loss": 0.7285,
+      "step": 25553
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3816867586580767e-07,
+      "loss": 0.7188,
+      "step": 25554
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3783170674533414e-07,
+      "loss": 0.7031,
+      "step": 25555
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3749486434599755e-07,
+      "loss": 0.7012,
+      "step": 25556
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.371581486722598e-07,
+      "loss": 0.7109,
+      "step": 25557
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.368215597285863e-07,
+      "loss": 0.707,
+      "step": 25558
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3648509751943233e-07,
+      "loss": 0.6992,
+      "step": 25559
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3614876204926106e-07,
+      "loss": 0.6787,
+      "step": 25560
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.358125533225288e-07,
+      "loss": 0.708,
+      "step": 25561
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3547647134369097e-07,
+      "loss": 0.7168,
+      "step": 25562
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3514051611720065e-07,
+      "loss": 0.7197,
+      "step": 25563
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3480468764750985e-07,
+      "loss": 0.6982,
+      "step": 25564
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.344689859390716e-07,
+      "loss": 0.6914,
+      "step": 25565
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.341334109963313e-07,
+      "loss": 0.6807,
+      "step": 25566
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3379796282373875e-07,
+      "loss": 0.7148,
+      "step": 25567
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3346264142573704e-07,
+      "loss": 0.7061,
+      "step": 25568
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3312744680677364e-07,
+      "loss": 0.7012,
+      "step": 25569
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3279237897128733e-07,
+      "loss": 0.7012,
+      "step": 25570
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.324574379237201e-07,
+      "loss": 0.6797,
+      "step": 25571
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3212262366850944e-07,
+      "loss": 0.7109,
+      "step": 25572
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.317879362100941e-07,
+      "loss": 0.6836,
+      "step": 25573
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3145337555290825e-07,
+      "loss": 0.708,
+      "step": 25574
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3111894170138503e-07,
+      "loss": 0.7197,
+      "step": 25575
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.307846346599576e-07,
+      "loss": 0.7207,
+      "step": 25576
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.304504544330579e-07,
+      "loss": 0.7002,
+      "step": 25577
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.301164010251113e-07,
+      "loss": 0.6875,
+      "step": 25578
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2978247444054766e-07,
+      "loss": 0.6865,
+      "step": 25579
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.294486746837878e-07,
+      "loss": 0.709,
+      "step": 25580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2911500175926047e-07,
+      "loss": 0.7119,
+      "step": 25581
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.287814556713865e-07,
+      "loss": 0.6992,
+      "step": 25582
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2844803642458466e-07,
+      "loss": 0.6982,
+      "step": 25583
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.281147440232736e-07,
+      "loss": 0.7061,
+      "step": 25584
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.277815784718709e-07,
+      "loss": 0.7432,
+      "step": 25585
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2744853977479186e-07,
+      "loss": 0.7061,
+      "step": 25586
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2711562793645077e-07,
+      "loss": 0.6982,
+      "step": 25587
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.267828429612564e-07,
+      "loss": 0.7061,
+      "step": 25588
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.264501848536218e-07,
+      "loss": 0.7119,
+      "step": 25589
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2611765361795677e-07,
+      "loss": 0.7061,
+      "step": 25590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.257852492586645e-07,
+      "loss": 0.7129,
+      "step": 25591
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2545297178015034e-07,
+      "loss": 0.7168,
+      "step": 25592
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2512082118682074e-07,
+      "loss": 0.7021,
+      "step": 25593
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2478879748307555e-07,
+      "loss": 0.6963,
+      "step": 25594
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.244569006733157e-07,
+      "loss": 0.6836,
+      "step": 25595
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2412513076193874e-07,
+      "loss": 0.71,
+      "step": 25596
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2379348775334227e-07,
+      "loss": 0.6992,
+      "step": 25597
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.234619716519206e-07,
+      "loss": 0.7002,
+      "step": 25598
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2313058246206685e-07,
+      "loss": 0.6924,
+      "step": 25599
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2279932018817194e-07,
+      "loss": 0.7002,
+      "step": 25600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.22468184834629e-07,
+      "loss": 0.6826,
+      "step": 25601
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.221371764058235e-07,
+      "loss": 0.7207,
+      "step": 25602
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.218062949061452e-07,
+      "loss": 0.6992,
+      "step": 25603
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2147554033997283e-07,
+      "loss": 0.6924,
+      "step": 25604
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.211449127116962e-07,
+      "loss": 0.7021,
+      "step": 25605
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2081441202569404e-07,
+      "loss": 0.6875,
+      "step": 25606
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.204840382863473e-07,
+      "loss": 0.7041,
+      "step": 25607
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.201537914980336e-07,
+      "loss": 0.6787,
+      "step": 25608
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1982367166512827e-07,
+      "loss": 0.6953,
+      "step": 25609
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1949367879200784e-07,
+      "loss": 0.6865,
+      "step": 25610
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.191638128830455e-07,
+      "loss": 0.7314,
+      "step": 25611
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.188340739426111e-07,
+      "loss": 0.6943,
+      "step": 25612
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1850446197507556e-07,
+      "loss": 0.7119,
+      "step": 25613
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1817497698480756e-07,
+      "loss": 0.6992,
+      "step": 25614
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.178456189761726e-07,
+      "loss": 0.6924,
+      "step": 25615
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.17516387953536e-07,
+      "loss": 0.7012,
+      "step": 25616
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1718728392126097e-07,
+      "loss": 0.6885,
+      "step": 25617
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1685830688370845e-07,
+      "loss": 0.7051,
+      "step": 25618
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.165294568452394e-07,
+      "loss": 0.7236,
+      "step": 25619
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.162007338102092e-07,
+      "loss": 0.6758,
+      "step": 25620
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1587213778297555e-07,
+      "loss": 0.7041,
+      "step": 25621
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.15543668767896e-07,
+      "loss": 0.6787,
+      "step": 25622
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.152153267693193e-07,
+      "loss": 0.6807,
+      "step": 25623
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.148871117915976e-07,
+      "loss": 0.6855,
+      "step": 25624
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1455902383908066e-07,
+      "loss": 0.7207,
+      "step": 25625
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1423106291611834e-07,
+      "loss": 0.6797,
+      "step": 25626
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1390322902705506e-07,
+      "loss": 0.707,
+      "step": 25627
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.135755221762372e-07,
+      "loss": 0.6992,
+      "step": 25628
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.132479423680047e-07,
+      "loss": 0.6865,
+      "step": 25629
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1292048960670185e-07,
+      "loss": 0.7393,
+      "step": 25630
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1259316389666516e-07,
+      "loss": 0.6895,
+      "step": 25631
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.122659652422356e-07,
+      "loss": 0.748,
+      "step": 25632
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.119388936477453e-07,
+      "loss": 0.6885,
+      "step": 25633
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.11611949117533e-07,
+      "loss": 0.707,
+      "step": 25634
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1128513165593077e-07,
+      "loss": 0.7109,
+      "step": 25635
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1095844126726735e-07,
+      "loss": 0.7139,
+      "step": 25636
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1063187795587154e-07,
+      "loss": 0.6865,
+      "step": 25637
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1030544172607543e-07,
+      "loss": 0.7119,
+      "step": 25638
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.099791325822011e-07,
+      "loss": 0.7314,
+      "step": 25639
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.096529505285751e-07,
+      "loss": 0.7178,
+      "step": 25640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0932689556951953e-07,
+      "loss": 0.708,
+      "step": 25641
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.090009677093554e-07,
+      "loss": 0.7285,
+      "step": 25642
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.086751669524014e-07,
+      "loss": 0.7285,
+      "step": 25643
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0834949330297526e-07,
+      "loss": 0.7061,
+      "step": 25644
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0802394676539347e-07,
+      "loss": 0.7158,
+      "step": 25645
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.076985273439693e-07,
+      "loss": 0.6895,
+      "step": 25646
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0737323504301705e-07,
+      "loss": 0.7041,
+      "step": 25647
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0704806986684777e-07,
+      "loss": 0.6758,
+      "step": 25648
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.067230318197668e-07,
+      "loss": 0.6934,
+      "step": 25649
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.063981209060852e-07,
+      "loss": 0.7061,
+      "step": 25650
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0607333713010957e-07,
+      "loss": 0.6924,
+      "step": 25651
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.057486804961408e-07,
+      "loss": 0.7002,
+      "step": 25652
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.054241510084844e-07,
+      "loss": 0.6982,
+      "step": 25653
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0509974867143696e-07,
+      "loss": 0.6924,
+      "step": 25654
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.047754734893039e-07,
+      "loss": 0.6914,
+      "step": 25655
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0445132546637845e-07,
+      "loss": 0.7031,
+      "step": 25656
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0412730460695604e-07,
+      "loss": 0.6914,
+      "step": 25657
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0380341091533105e-07,
+      "loss": 0.667,
+      "step": 25658
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.034796443957967e-07,
+      "loss": 0.7168,
+      "step": 25659
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.03156005052644e-07,
+      "loss": 0.6875,
+      "step": 25660
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.028324928901606e-07,
+      "loss": 0.6777,
+      "step": 25661
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.025091079126353e-07,
+      "loss": 0.7197,
+      "step": 25662
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.021858501243536e-07,
+      "loss": 0.6982,
+      "step": 25663
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0186271952959765e-07,
+      "loss": 0.7129,
+      "step": 25664
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.015397161326506e-07,
+      "loss": 0.7139,
+      "step": 25665
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.012168399377925e-07,
+      "loss": 0.6904,
+      "step": 25666
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.008940909493042e-07,
+      "loss": 0.6729,
+      "step": 25667
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0057146917146237e-07,
+      "loss": 0.6934,
+      "step": 25668
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0024897460854027e-07,
+      "loss": 0.707,
+      "step": 25669
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9992660726481223e-07,
+      "loss": 0.7168,
+      "step": 25670
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9960436714455265e-07,
+      "loss": 0.6826,
+      "step": 25671
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9928225425203135e-07,
+      "loss": 0.7285,
+      "step": 25672
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9896026859151507e-07,
+      "loss": 0.7041,
+      "step": 25673
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9863841016727247e-07,
+      "loss": 0.7158,
+      "step": 25674
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9831667898356905e-07,
+      "loss": 0.6924,
+      "step": 25675
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9799507504466814e-07,
+      "loss": 0.6836,
+      "step": 25676
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9767359835483075e-07,
+      "loss": 0.6943,
+      "step": 25677
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.973522489183179e-07,
+      "loss": 0.6943,
+      "step": 25678
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.970310267393884e-07,
+      "loss": 0.6846,
+      "step": 25679
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.967099318223e-07,
+      "loss": 0.6836,
+      "step": 25680
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9638896417130926e-07,
+      "loss": 0.707,
+      "step": 25681
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9606812379066385e-07,
+      "loss": 0.7041,
+      "step": 25682
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9574741068462153e-07,
+      "loss": 0.6748,
+      "step": 25683
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9542682485743e-07,
+      "loss": 0.71,
+      "step": 25684
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.951063663133381e-07,
+      "loss": 0.7295,
+      "step": 25685
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.947860350565935e-07,
+      "loss": 0.6895,
+      "step": 25686
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9446583109143956e-07,
+      "loss": 0.7148,
+      "step": 25687
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9414575442212164e-07,
+      "loss": 0.6855,
+      "step": 25688
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.938258050528809e-07,
+      "loss": 0.7324,
+      "step": 25689
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9350598298795617e-07,
+      "loss": 0.6934,
+      "step": 25690
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9318628823158624e-07,
+      "loss": 0.7021,
+      "step": 25691
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.928667207880099e-07,
+      "loss": 0.6924,
+      "step": 25692
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.925472806614605e-07,
+      "loss": 0.7021,
+      "step": 25693
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.922279678561713e-07,
+      "loss": 0.6924,
+      "step": 25694
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9190878237637453e-07,
+      "loss": 0.7188,
+      "step": 25695
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.915897242262989e-07,
+      "loss": 0.7051,
+      "step": 25696
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.912707934101745e-07,
+      "loss": 0.7119,
+      "step": 25697
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9095198993222673e-07,
+      "loss": 0.7246,
+      "step": 25698
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9063331379668e-07,
+      "loss": 0.6973,
+      "step": 25699
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.90314765007761e-07,
+      "loss": 0.6875,
+      "step": 25700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.899963435696863e-07,
+      "loss": 0.6934,
+      "step": 25701
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.896780494866781e-07,
+      "loss": 0.6943,
+      "step": 25702
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.893598827629541e-07,
+      "loss": 0.6895,
+      "step": 25703
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.89041843402731e-07,
+      "loss": 0.6836,
+      "step": 25704
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.887239314102242e-07,
+      "loss": 0.6807,
+      "step": 25705
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8840614678964717e-07,
+      "loss": 0.7002,
+      "step": 25706
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8808848954520864e-07,
+      "loss": 0.7129,
+      "step": 25707
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8777095968112076e-07,
+      "loss": 0.6768,
+      "step": 25708
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.874535572015914e-07,
+      "loss": 0.6855,
+      "step": 25709
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8713628211082597e-07,
+      "loss": 0.7178,
+      "step": 25710
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.868191344130279e-07,
+      "loss": 0.7217,
+      "step": 25711
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8650211411240256e-07,
+      "loss": 0.6963,
+      "step": 25712
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8618522121315226e-07,
+      "loss": 0.6895,
+      "step": 25713
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.858684557194725e-07,
+      "loss": 0.7275,
+      "step": 25714
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8555181763556327e-07,
+      "loss": 0.708,
+      "step": 25715
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8523530696562115e-07,
+      "loss": 0.7285,
+      "step": 25716
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8491892371384066e-07,
+      "loss": 0.7139,
+      "step": 25717
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.846026678844139e-07,
+      "loss": 0.7021,
+      "step": 25718
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8428653948153204e-07,
+      "loss": 0.7188,
+      "step": 25719
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8397053850938503e-07,
+      "loss": 0.7285,
+      "step": 25720
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8365466497216063e-07,
+      "loss": 0.6904,
+      "step": 25721
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.833389188740455e-07,
+      "loss": 0.6855,
+      "step": 25722
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8302330021922076e-07,
+      "loss": 0.6885,
+      "step": 25723
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.82707809011873e-07,
+      "loss": 0.7002,
+      "step": 25724
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8239244525618226e-07,
+      "loss": 0.7314,
+      "step": 25725
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.820772089563285e-07,
+      "loss": 0.6816,
+      "step": 25726
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.817621001164851e-07,
+      "loss": 0.7158,
+      "step": 25727
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8144711874083306e-07,
+      "loss": 0.7168,
+      "step": 25728
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.811322648335447e-07,
+      "loss": 0.6875,
+      "step": 25729
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.808175383987922e-07,
+      "loss": 0.6992,
+      "step": 25730
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.805029394407467e-07,
+      "loss": 0.7002,
+      "step": 25731
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.801884679635759e-07,
+      "loss": 0.6904,
+      "step": 25732
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7987412397145097e-07,
+      "loss": 0.71,
+      "step": 25733
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7955990746853524e-07,
+      "loss": 0.71,
+      "step": 25734
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.792458184589931e-07,
+      "loss": 0.6855,
+      "step": 25735
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7893185694698466e-07,
+      "loss": 0.6826,
+      "step": 25736
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7861802293667535e-07,
+      "loss": 0.7285,
+      "step": 25737
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.783043164322209e-07,
+      "loss": 0.6836,
+      "step": 25738
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7799073743778e-07,
+      "loss": 0.709,
+      "step": 25739
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.776772859575084e-07,
+      "loss": 0.6758,
+      "step": 25740
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.773639619955582e-07,
+      "loss": 0.7012,
+      "step": 25741
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.77050765556084e-07,
+      "loss": 0.7012,
+      "step": 25742
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.767376966432357e-07,
+      "loss": 0.6885,
+      "step": 25743
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7642475526115997e-07,
+      "loss": 0.6934,
+      "step": 25744
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.76111941414008e-07,
+      "loss": 0.7227,
+      "step": 25745
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7579925510592307e-07,
+      "loss": 0.7012,
+      "step": 25746
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.754866963410486e-07,
+      "loss": 0.6787,
+      "step": 25747
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7517426512352684e-07,
+      "loss": 0.71,
+      "step": 25748
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.748619614574989e-07,
+      "loss": 0.7031,
+      "step": 25749
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.745497853471047e-07,
+      "loss": 0.7012,
+      "step": 25750
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.742377367964789e-07,
+      "loss": 0.6904,
+      "step": 25751
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7392581580975807e-07,
+      "loss": 0.708,
+      "step": 25752
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.736140223910756e-07,
+      "loss": 0.6748,
+      "step": 25753
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.733023565445637e-07,
+      "loss": 0.7266,
+      "step": 25754
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7299081827435245e-07,
+      "loss": 0.6963,
+      "step": 25755
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7267940758456967e-07,
+      "loss": 0.7041,
+      "step": 25756
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.723681244793442e-07,
+      "loss": 0.6973,
+      "step": 25757
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.720569689628006e-07,
+      "loss": 0.6748,
+      "step": 25758
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.717459410390634e-07,
+      "loss": 0.71,
+      "step": 25759
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.714350407122513e-07,
+      "loss": 0.6797,
+      "step": 25760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7112426798648683e-07,
+      "loss": 0.7305,
+      "step": 25761
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.708136228658876e-07,
+      "loss": 0.7227,
+      "step": 25762
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7050310535457046e-07,
+      "loss": 0.7051,
+      "step": 25763
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7019271545665204e-07,
+      "loss": 0.7246,
+      "step": 25764
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.698824531762435e-07,
+      "loss": 0.6855,
+      "step": 25765
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6957231851745824e-07,
+      "loss": 0.7012,
+      "step": 25766
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6926231148440515e-07,
+      "loss": 0.6904,
+      "step": 25767
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6895243208119103e-07,
+      "loss": 0.7344,
+      "step": 25768
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6864268031192584e-07,
+      "loss": 0.7061,
+      "step": 25769
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6833305618071415e-07,
+      "loss": 0.7305,
+      "step": 25770
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6802355969165706e-07,
+      "loss": 0.6885,
+      "step": 25771
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6771419084885794e-07,
+      "loss": 0.6973,
+      "step": 25772
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6740494965641583e-07,
+      "loss": 0.6846,
+      "step": 25773
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.670958361184285e-07,
+      "loss": 0.6748,
+      "step": 25774
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6678685023899265e-07,
+      "loss": 0.6699,
+      "step": 25775
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.664779920222039e-07,
+      "loss": 0.7324,
+      "step": 25776
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.661692614721535e-07,
+      "loss": 0.7275,
+      "step": 25777
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6586065859293696e-07,
+      "loss": 0.751,
+      "step": 25778
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6555218338863885e-07,
+      "loss": 0.6973,
+      "step": 25779
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6524383586335034e-07,
+      "loss": 0.71,
+      "step": 25780
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.649356160211559e-07,
+      "loss": 0.6992,
+      "step": 25781
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.646275238661423e-07,
+      "loss": 0.7021,
+      "step": 25782
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6431955940239074e-07,
+      "loss": 0.6953,
+      "step": 25783
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6401172263398345e-07,
+      "loss": 0.7285,
+      "step": 25784
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.637040135649994e-07,
+      "loss": 0.7158,
+      "step": 25785
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6339643219951756e-07,
+      "loss": 0.6875,
+      "step": 25786
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.630889785416125e-07,
+      "loss": 0.7168,
+      "step": 25787
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.627816525953598e-07,
+      "loss": 0.6934,
+      "step": 25788
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6247445436482953e-07,
+      "loss": 0.6826,
+      "step": 25789
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.621673838540973e-07,
+      "loss": 0.7148,
+      "step": 25790
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.618604410672322e-07,
+      "loss": 0.7129,
+      "step": 25791
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6155362600829746e-07,
+      "loss": 0.6777,
+      "step": 25792
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.612469386813622e-07,
+      "loss": 0.6895,
+      "step": 25793
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6094037909048975e-07,
+      "loss": 0.6953,
+      "step": 25794
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6063394723974467e-07,
+      "loss": 0.6855,
+      "step": 25795
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.603276431331859e-07,
+      "loss": 0.7051,
+      "step": 25796
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.600214667748725e-07,
+      "loss": 0.7188,
+      "step": 25797
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.597154181688645e-07,
+      "loss": 0.7061,
+      "step": 25798
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.5940949731921526e-07,
+      "loss": 0.709,
+      "step": 25799
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.591037042299794e-07,
+      "loss": 0.7178,
+      "step": 25800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5879803890520816e-07,
+      "loss": 0.7119,
+      "step": 25801
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5849250134895597e-07,
+      "loss": 0.667,
+      "step": 25802
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5818709156526965e-07,
+      "loss": 0.6807,
+      "step": 25803
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.578818095581982e-07,
+      "loss": 0.6943,
+      "step": 25804
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5757665533178386e-07,
+      "loss": 0.7295,
+      "step": 25805
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.572716288900735e-07,
+      "loss": 0.6846,
+      "step": 25806
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5696673023710824e-07,
+      "loss": 0.7139,
+      "step": 25807
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.566619593769294e-07,
+      "loss": 0.7285,
+      "step": 25808
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5635731631357583e-07,
+      "loss": 0.7051,
+      "step": 25809
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.560528010510833e-07,
+      "loss": 0.7168,
+      "step": 25810
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.557484135934908e-07,
+      "loss": 0.6973,
+      "step": 25811
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5544415394482945e-07,
+      "loss": 0.6787,
+      "step": 25812
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5514002210912946e-07,
+      "loss": 0.707,
+      "step": 25813
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5483601809042646e-07,
+      "loss": 0.7256,
+      "step": 25814
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.54532141892745e-07,
+      "loss": 0.6895,
+      "step": 25815
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.542283935201152e-07,
+      "loss": 0.7148,
+      "step": 25816
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5392477297655935e-07,
+      "loss": 0.6973,
+      "step": 25817
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.536212802661043e-07,
+      "loss": 0.7109,
+      "step": 25818
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5331791539276907e-07,
+      "loss": 0.708,
+      "step": 25819
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5301467836057593e-07,
+      "loss": 0.6865,
+      "step": 25820
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5271156917354277e-07,
+      "loss": 0.7314,
+      "step": 25821
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.524085878356853e-07,
+      "loss": 0.6875,
+      "step": 25822
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5210573435102144e-07,
+      "loss": 0.7041,
+      "step": 25823
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.518030087235647e-07,
+      "loss": 0.6904,
+      "step": 25824
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.515004109573239e-07,
+      "loss": 0.7012,
+      "step": 25825
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5119794105630934e-07,
+      "loss": 0.7041,
+      "step": 25826
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5089559902453327e-07,
+      "loss": 0.6973,
+      "step": 25827
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5059338486599813e-07,
+      "loss": 0.749,
+      "step": 25828
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5029129858471287e-07,
+      "loss": 0.707,
+      "step": 25829
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4998934018467765e-07,
+      "loss": 0.6846,
+      "step": 25830
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4968750966989593e-07,
+      "loss": 0.6973,
+      "step": 25831
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4938580704436676e-07,
+      "loss": 0.7041,
+      "step": 25832
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4908423231208796e-07,
+      "loss": 0.6924,
+      "step": 25833
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4878278547705644e-07,
+      "loss": 0.6973,
+      "step": 25834
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4848146654326896e-07,
+      "loss": 0.7109,
+      "step": 25835
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.481802755147179e-07,
+      "loss": 0.6924,
+      "step": 25836
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4787921239539335e-07,
+      "loss": 0.7178,
+      "step": 25837
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.475782771892844e-07,
+      "loss": 0.6963,
+      "step": 25838
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4727746990038226e-07,
+      "loss": 0.6982,
+      "step": 25839
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4697679053267044e-07,
+      "loss": 0.7363,
+      "step": 25840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.466762390901357e-07,
+      "loss": 0.6885,
+      "step": 25841
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.463758155767605e-07,
+      "loss": 0.6934,
+      "step": 25842
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.46075519996526e-07,
+      "loss": 0.7275,
+      "step": 25843
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.457753523534102e-07,
+      "loss": 0.7109,
+      "step": 25844
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.454753126513943e-07,
+      "loss": 0.6836,
+      "step": 25845
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4517540089445077e-07,
+      "loss": 0.6904,
+      "step": 25846
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.448756170865575e-07,
+      "loss": 0.6924,
+      "step": 25847
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.445759612316857e-07,
+      "loss": 0.7188,
+      "step": 25848
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.442764333338078e-07,
+      "loss": 0.7109,
+      "step": 25849
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.439770333968895e-07,
+      "loss": 0.7041,
+      "step": 25850
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4367776142490317e-07,
+      "loss": 0.7051,
+      "step": 25851
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4337861742181236e-07,
+      "loss": 0.7051,
+      "step": 25852
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.430796013915827e-07,
+      "loss": 0.7266,
+      "step": 25853
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.427807133381755e-07,
+      "loss": 0.6914,
+      "step": 25854
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.424819532655521e-07,
+      "loss": 0.6914,
+      "step": 25855
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.421833211776737e-07,
+      "loss": 0.7051,
+      "step": 25856
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4188481707849496e-07,
+      "loss": 0.7031,
+      "step": 25857
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.415864409719738e-07,
+      "loss": 0.6973,
+      "step": 25858
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.412881928620615e-07,
+      "loss": 0.709,
+      "step": 25859
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.409900727527149e-07,
+      "loss": 0.6816,
+      "step": 25860
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4069208064788194e-07,
+      "loss": 0.7021,
+      "step": 25861
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.403942165515117e-07,
+      "loss": 0.71,
+      "step": 25862
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4009648046755326e-07,
+      "loss": 0.7012,
+      "step": 25863
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.397988723999512e-07,
+      "loss": 0.6963,
+      "step": 25864
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.395013923526491e-07,
+      "loss": 0.6875,
+      "step": 25865
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3920404032959043e-07,
+      "loss": 0.7021,
+      "step": 25866
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3890681633471423e-07,
+      "loss": 0.7227,
+      "step": 25867
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3860972037196073e-07,
+      "loss": 0.6943,
+      "step": 25868
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3831275244526785e-07,
+      "loss": 0.7217,
+      "step": 25869
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3801591255856915e-07,
+      "loss": 0.6904,
+      "step": 25870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.377192007157981e-07,
+      "loss": 0.6895,
+      "step": 25871
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.374226169208894e-07,
+      "loss": 0.7178,
+      "step": 25872
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.371261611777721e-07,
+      "loss": 0.6953,
+      "step": 25873
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.368298334903741e-07,
+      "loss": 0.7021,
+      "step": 25874
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.365336338626246e-07,
+      "loss": 0.7031,
+      "step": 25875
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.362375622984471e-07,
+      "loss": 0.6914,
+      "step": 25876
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.359416188017661e-07,
+      "loss": 0.7041,
+      "step": 25877
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.35645803376502e-07,
+      "loss": 0.7188,
+      "step": 25878
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3535011602657597e-07,
+      "loss": 0.7178,
+      "step": 25879
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.350545567559083e-07,
+      "loss": 0.7109,
+      "step": 25880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.347591255684146e-07,
+      "loss": 0.6953,
+      "step": 25881
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3446382246800967e-07,
+      "loss": 0.6846,
+      "step": 25882
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3416864745860587e-07,
+      "loss": 0.6777,
+      "step": 25883
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3387360054411787e-07,
+      "loss": 0.7109,
+      "step": 25884
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3357868172845255e-07,
+      "loss": 0.6865,
+      "step": 25885
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3328389101552115e-07,
+      "loss": 0.7539,
+      "step": 25886
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3298922840922843e-07,
+      "loss": 0.7188,
+      "step": 25887
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3269469391348006e-07,
+      "loss": 0.6914,
+      "step": 25888
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.324002875321786e-07,
+      "loss": 0.6895,
+      "step": 25889
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.321060092692274e-07,
+      "loss": 0.71,
+      "step": 25890
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3181185912852356e-07,
+      "loss": 0.6953,
+      "step": 25891
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.315178371139671e-07,
+      "loss": 0.7041,
+      "step": 25892
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3122394322945506e-07,
+      "loss": 0.6904,
+      "step": 25893
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.309301774788809e-07,
+      "loss": 0.7109,
+      "step": 25894
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.306365398661382e-07,
+      "loss": 0.6846,
+      "step": 25895
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.303430303951194e-07,
+      "loss": 0.6924,
+      "step": 25896
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3004964906971247e-07,
+      "loss": 0.7178,
+      "step": 25897
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.297563958938055e-07,
+      "loss": 0.7178,
+      "step": 25898
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.294632708712864e-07,
+      "loss": 0.7197,
+      "step": 25899
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.291702740060365e-07,
+      "loss": 0.6787,
+      "step": 25900
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.288774053019439e-07,
+      "loss": 0.7168,
+      "step": 25901
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2858466476288544e-07,
+      "loss": 0.6963,
+      "step": 25902
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.282920523927413e-07,
+      "loss": 0.6963,
+      "step": 25903
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.279995681953896e-07,
+      "loss": 0.7188,
+      "step": 25904
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2770721217470823e-07,
+      "loss": 0.6836,
+      "step": 25905
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.274149843345698e-07,
+      "loss": 0.7012,
+      "step": 25906
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.271228846788477e-07,
+      "loss": 0.6904,
+      "step": 25907
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.268309132114122e-07,
+      "loss": 0.7295,
+      "step": 25908
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.265390699361326e-07,
+      "loss": 0.6885,
+      "step": 25909
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.262473548568779e-07,
+      "loss": 0.71,
+      "step": 25910
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.259557679775127e-07,
+      "loss": 0.7109,
+      "step": 25911
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.256643093018996e-07,
+      "loss": 0.7109,
+      "step": 25912
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.253729788339055e-07,
+      "loss": 0.7139,
+      "step": 25913
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2508177657738947e-07,
+      "loss": 0.6904,
+      "step": 25914
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.247907025362085e-07,
+      "loss": 0.6973,
+      "step": 25915
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.244997567142205e-07,
+      "loss": 0.6953,
+      "step": 25916
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.242089391152825e-07,
+      "loss": 0.71,
+      "step": 25917
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2391824974324913e-07,
+      "loss": 0.6904,
+      "step": 25918
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2362768860197067e-07,
+      "loss": 0.7139,
+      "step": 25919
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.233372556952985e-07,
+      "loss": 0.6914,
+      "step": 25920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2304695102708396e-07,
+      "loss": 0.7246,
+      "step": 25921
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2275677460117063e-07,
+      "loss": 0.6748,
+      "step": 25922
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2246672642140655e-07,
+      "loss": 0.7227,
+      "step": 25923
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2217680649163195e-07,
+      "loss": 0.7109,
+      "step": 25924
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2188701481569384e-07,
+      "loss": 0.6904,
+      "step": 25925
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2159735139742907e-07,
+      "loss": 0.6992,
+      "step": 25926
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.213078162406791e-07,
+      "loss": 0.7021,
+      "step": 25927
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.210184093492774e-07,
+      "loss": 0.7217,
+      "step": 25928
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.207291307270621e-07,
+      "loss": 0.7188,
+      "step": 25929
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.204399803778668e-07,
+      "loss": 0.7051,
+      "step": 25930
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2015095830552177e-07,
+      "loss": 0.6855,
+      "step": 25931
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.198620645138595e-07,
+      "loss": 0.7012,
+      "step": 25932
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1957329900670463e-07,
+      "loss": 0.71,
+      "step": 25933
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1928466178788866e-07,
+      "loss": 0.6982,
+      "step": 25934
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.18996152861234e-07,
+      "loss": 0.7275,
+      "step": 25935
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1870777223056315e-07,
+      "loss": 0.7109,
+      "step": 25936
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.184195198997009e-07,
+      "loss": 0.709,
+      "step": 25937
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.181313958724652e-07,
+      "loss": 0.6875,
+      "step": 25938
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1784340015267535e-07,
+      "loss": 0.7207,
+      "step": 25939
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1755553274414707e-07,
+      "loss": 0.7119,
+      "step": 25940
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.172677936506952e-07,
+      "loss": 0.7236,
+      "step": 25941
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.169801828761343e-07,
+      "loss": 0.6982,
+      "step": 25942
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.166927004242737e-07,
+      "loss": 0.7002,
+      "step": 25943
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1640534629892583e-07,
+      "loss": 0.7197,
+      "step": 25944
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1611812050389546e-07,
+      "loss": 0.6924,
+      "step": 25945
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1583102304299285e-07,
+      "loss": 0.7002,
+      "step": 25946
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1554405392002163e-07,
+      "loss": 0.7041,
+      "step": 25947
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.152572131387832e-07,
+      "loss": 0.7051,
+      "step": 25948
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.149705007030779e-07,
+      "loss": 0.6885,
+      "step": 25949
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1468391661670815e-07,
+      "loss": 0.71,
+      "step": 25950
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.143974608834721e-07,
+      "loss": 0.6875,
+      "step": 25951
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.141111335071645e-07,
+      "loss": 0.6924,
+      "step": 25952
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.138249344915789e-07,
+      "loss": 0.7314,
+      "step": 25953
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1353886384051015e-07,
+      "loss": 0.6934,
+      "step": 25954
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1325292155774847e-07,
+      "loss": 0.707,
+      "step": 25955
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1296710764708417e-07,
+      "loss": 0.7119,
+      "step": 25956
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.12681422112302e-07,
+      "loss": 0.71,
+      "step": 25957
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.123958649571912e-07,
+      "loss": 0.6836,
+      "step": 25958
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1211043618553537e-07,
+      "loss": 0.7012,
+      "step": 25959
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.118251358011171e-07,
+      "loss": 0.6807,
+      "step": 25960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1153996380771433e-07,
+      "loss": 0.6943,
+      "step": 25961
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.112549202091109e-07,
+      "loss": 0.7207,
+      "step": 25962
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1097000500908247e-07,
+      "loss": 0.7256,
+      "step": 25963
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1068521821140397e-07,
+      "loss": 0.6914,
+      "step": 25964
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1040055981985006e-07,
+      "loss": 0.6914,
+      "step": 25965
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1011602983819333e-07,
+      "loss": 0.7178,
+      "step": 25966
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.098316282702041e-07,
+      "loss": 0.7061,
+      "step": 25967
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.095473551196515e-07,
+      "loss": 0.6807,
+      "step": 25968
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.092632103903015e-07,
+      "loss": 0.6875,
+      "step": 25969
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.089791940859221e-07,
+      "loss": 0.7061,
+      "step": 25970
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.086953062102771e-07,
+      "loss": 0.7158,
+      "step": 25971
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.084115467671267e-07,
+      "loss": 0.7012,
+      "step": 25972
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.081279157602335e-07,
+      "loss": 0.7109,
+      "step": 25973
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0784441319335444e-07,
+      "loss": 0.7158,
+      "step": 25974
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.075610390702466e-07,
+      "loss": 0.6963,
+      "step": 25975
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.072777933946669e-07,
+      "loss": 0.7002,
+      "step": 25976
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.06994676170369e-07,
+      "loss": 0.668,
+      "step": 25977
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.067116874011011e-07,
+      "loss": 0.6953,
+      "step": 25978
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0642882709062014e-07,
+      "loss": 0.7051,
+      "step": 25979
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.061460952426687e-07,
+      "loss": 0.7041,
+      "step": 25980
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0586349186099594e-07,
+      "loss": 0.6953,
+      "step": 25981
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0558101694934785e-07,
+      "loss": 0.7148,
+      "step": 25982
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0529867051146576e-07,
+      "loss": 0.7178,
+      "step": 25983
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0501645255109346e-07,
+      "loss": 0.6943,
+      "step": 25984
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.047343630719701e-07,
+      "loss": 0.6797,
+      "step": 25985
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.04452402077835e-07,
+      "loss": 0.6963,
+      "step": 25986
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0417056957242286e-07,
+      "loss": 0.7139,
+      "step": 25987
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0388886555947073e-07,
+      "loss": 0.6895,
+      "step": 25988
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0360729004271007e-07,
+      "loss": 0.6953,
+      "step": 25989
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.033258430258712e-07,
+      "loss": 0.7207,
+      "step": 25990
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0304452451268786e-07,
+      "loss": 0.6875,
+      "step": 25991
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.027633345068881e-07,
+      "loss": 0.7119,
+      "step": 25992
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0248227301219455e-07,
+      "loss": 0.7285,
+      "step": 25993
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0220134003233314e-07,
+      "loss": 0.7021,
+      "step": 25994
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.019205355710275e-07,
+      "loss": 0.7178,
+      "step": 25995
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0163985963200025e-07,
+      "loss": 0.6875,
+      "step": 25996
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0135931221896954e-07,
+      "loss": 0.7119,
+      "step": 25997
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.010788933356523e-07,
+      "loss": 0.6846,
+      "step": 25998
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.00798602985769e-07,
+      "loss": 0.6865,
+      "step": 25999
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.005184411730289e-07,
+      "loss": 0.7139,
+      "step": 26000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.002384079011478e-07,
+      "loss": 0.6689,
+      "step": 26001
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.99958503173835e-07,
+      "loss": 0.6904,
+      "step": 26002
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9967872699480095e-07,
+      "loss": 0.7109,
+      "step": 26003
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.993990793677537e-07,
+      "loss": 0.7012,
+      "step": 26004
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.991195602964003e-07,
+      "loss": 0.6973,
+      "step": 26005
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.988401697844412e-07,
+      "loss": 0.6865,
+      "step": 26006
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9856090783558334e-07,
+      "loss": 0.6963,
+      "step": 26007
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.982817744535249e-07,
+      "loss": 0.7129,
+      "step": 26008
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.980027696419652e-07,
+      "loss": 0.6875,
+      "step": 26009
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.977238934046034e-07,
+      "loss": 0.707,
+      "step": 26010
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9744514574513215e-07,
+      "loss": 0.7031,
+      "step": 26011
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.971665266672497e-07,
+      "loss": 0.7021,
+      "step": 26012
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9688803617464515e-07,
+      "loss": 0.6992,
+      "step": 26013
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.966096742710101e-07,
+      "loss": 0.6973,
+      "step": 26014
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9633144096003375e-07,
+      "loss": 0.7168,
+      "step": 26015
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.960533362454032e-07,
+      "loss": 0.7031,
+      "step": 26016
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.957753601308044e-07,
+      "loss": 0.6943,
+      "step": 26017
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9549751261992e-07,
+      "loss": 0.7109,
+      "step": 26018
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.952197937164336e-07,
+      "loss": 0.6982,
+      "step": 26019
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9494220342402455e-07,
+      "loss": 0.7002,
+      "step": 26020
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9466474174637215e-07,
+      "loss": 0.7188,
+      "step": 26021
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.943874086871534e-07,
+      "loss": 0.7129,
+      "step": 26022
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.941102042500421e-07,
+      "loss": 0.7168,
+      "step": 26023
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9383312843871415e-07,
+      "loss": 0.707,
+      "step": 26024
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9355618125684215e-07,
+      "loss": 0.7002,
+      "step": 26025
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9327936270809207e-07,
+      "loss": 0.6865,
+      "step": 26026
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9300267279613547e-07,
+      "loss": 0.6982,
+      "step": 26027
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9272611152463937e-07,
+      "loss": 0.7148,
+      "step": 26028
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.924496788972686e-07,
+      "loss": 0.708,
+      "step": 26029
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.921733749176847e-07,
+      "loss": 0.7275,
+      "step": 26030
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.918971995895514e-07,
+      "loss": 0.7051,
+      "step": 26031
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9162115291652804e-07,
+      "loss": 0.6934,
+      "step": 26032
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.913452349022727e-07,
+      "loss": 0.71,
+      "step": 26033
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.910694455504426e-07,
+      "loss": 0.7344,
+      "step": 26034
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9079378486469024e-07,
+      "loss": 0.7129,
+      "step": 26035
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9051825284867275e-07,
+      "loss": 0.7109,
+      "step": 26036
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9024284950603834e-07,
+      "loss": 0.6934,
+      "step": 26037
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8996757484043957e-07,
+      "loss": 0.7158,
+      "step": 26038
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.896924288555203e-07,
+      "loss": 0.6953,
+      "step": 26039
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8941741155492976e-07,
+      "loss": 0.6865,
+      "step": 26040
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.891425229423128e-07,
+      "loss": 0.7373,
+      "step": 26041
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8886776302131103e-07,
+      "loss": 0.7158,
+      "step": 26042
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.885931317955659e-07,
+      "loss": 0.7217,
+      "step": 26043
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.88318629268719e-07,
+      "loss": 0.71,
+      "step": 26044
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8804425544440627e-07,
+      "loss": 0.7158,
+      "step": 26045
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.877700103262626e-07,
+      "loss": 0.6885,
+      "step": 26046
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8749589391792397e-07,
+      "loss": 0.7129,
+      "step": 26047
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.87221906223023e-07,
+      "loss": 0.7227,
+      "step": 26048
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.869480472451902e-07,
+      "loss": 0.6963,
+      "step": 26049
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8667431698805705e-07,
+      "loss": 0.6914,
+      "step": 26050
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.864007154552473e-07,
+      "loss": 0.6924,
+      "step": 26051
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8612724265038915e-07,
+      "loss": 0.6846,
+      "step": 26052
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8585389857710646e-07,
+      "loss": 0.6885,
+      "step": 26053
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8558068323902066e-07,
+      "loss": 0.6758,
+      "step": 26054
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.853075966397545e-07,
+      "loss": 0.7041,
+      "step": 26055
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8503463878292504e-07,
+      "loss": 0.7314,
+      "step": 26056
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8476180967215163e-07,
+      "loss": 0.7119,
+      "step": 26057
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.844891093110491e-07,
+      "loss": 0.7031,
+      "step": 26058
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.842165377032291e-07,
+      "loss": 0.6729,
+      "step": 26059
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8394409485230643e-07,
+      "loss": 0.7139,
+      "step": 26060
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8367178076189163e-07,
+      "loss": 0.6943,
+      "step": 26061
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.833995954355917e-07,
+      "loss": 0.7041,
+      "step": 26062
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8312753887701604e-07,
+      "loss": 0.7129,
+      "step": 26063
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8285561108976844e-07,
+      "loss": 0.7266,
+      "step": 26064
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.825838120774527e-07,
+      "loss": 0.7012,
+      "step": 26065
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.823121418436714e-07,
+      "loss": 0.7373,
+      "step": 26066
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8204060039202396e-07,
+      "loss": 0.6777,
+      "step": 26067
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.817691877261086e-07,
+      "loss": 0.6816,
+      "step": 26068
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.814979038495236e-07,
+      "loss": 0.6914,
+      "step": 26069
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8122674876586484e-07,
+      "loss": 0.6973,
+      "step": 26070
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8095572247872406e-07,
+      "loss": 0.6826,
+      "step": 26071
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.806848249916916e-07,
+      "loss": 0.7002,
+      "step": 26072
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.804140563083613e-07,
+      "loss": 0.7031,
+      "step": 26073
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.80143416432318e-07,
+      "loss": 0.707,
+      "step": 26074
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.798729053671512e-07,
+      "loss": 0.7109,
+      "step": 26075
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.796025231164434e-07,
+      "loss": 0.6895,
+      "step": 26076
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.793322696837786e-07,
+      "loss": 0.7158,
+      "step": 26077
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7906214507273934e-07,
+      "loss": 0.7139,
+      "step": 26078
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.787921492869039e-07,
+      "loss": 0.6895,
+      "step": 26079
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7852228232985057e-07,
+      "loss": 0.7158,
+      "step": 26080
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7825254420515647e-07,
+      "loss": 0.6875,
+      "step": 26081
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7798293491639537e-07,
+      "loss": 0.6934,
+      "step": 26082
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7771345446714227e-07,
+      "loss": 0.7129,
+      "step": 26083
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7744410286096535e-07,
+      "loss": 0.7119,
+      "step": 26084
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.771748801014362e-07,
+      "loss": 0.7061,
+      "step": 26085
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7690578619212314e-07,
+      "loss": 0.7109,
+      "step": 26086
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7663682113658997e-07,
+      "loss": 0.6846,
+      "step": 26087
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.763679849384027e-07,
+      "loss": 0.6787,
+      "step": 26088
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.760992776011251e-07,
+      "loss": 0.6963,
+      "step": 26089
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.758306991283155e-07,
+      "loss": 0.6738,
+      "step": 26090
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.755622495235355e-07,
+      "loss": 0.7021,
+      "step": 26091
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7529392879034e-07,
+      "loss": 0.6807,
+      "step": 26092
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.750257369322873e-07,
+      "loss": 0.7236,
+      "step": 26093
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7475767395293227e-07,
+      "loss": 0.6816,
+      "step": 26094
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.744897398558244e-07,
+      "loss": 0.6895,
+      "step": 26095
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7422193464451743e-07,
+      "loss": 0.6729,
+      "step": 26096
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.739542583225585e-07,
+      "loss": 0.6943,
+      "step": 26097
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7368671089349486e-07,
+      "loss": 0.6807,
+      "step": 26098
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.734192923608736e-07,
+      "loss": 0.6787,
+      "step": 26099
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.731520027282375e-07,
+      "loss": 0.6865,
+      "step": 26100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.728848419991281e-07,
+      "loss": 0.7158,
+      "step": 26101
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7261781017708824e-07,
+      "loss": 0.7109,
+      "step": 26102
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.723509072656549e-07,
+      "loss": 0.6904,
+      "step": 26103
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7208413326836544e-07,
+      "loss": 0.6934,
+      "step": 26104
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7181748818875473e-07,
+      "loss": 0.7373,
+      "step": 26105
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.715509720303588e-07,
+      "loss": 0.7119,
+      "step": 26106
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7128458479670716e-07,
+      "loss": 0.7129,
+      "step": 26107
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.710183264913302e-07,
+      "loss": 0.7021,
+      "step": 26108
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7075219711775737e-07,
+      "loss": 0.7129,
+      "step": 26109
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.704861966795158e-07,
+      "loss": 0.6826,
+      "step": 26110
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.702203251801305e-07,
+      "loss": 0.709,
+      "step": 26111
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.699545826231231e-07,
+      "loss": 0.6914,
+      "step": 26112
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.696889690120164e-07,
+      "loss": 0.7139,
+      "step": 26113
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6942348435033185e-07,
+      "loss": 0.7295,
+      "step": 26114
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6915812864158785e-07,
+      "loss": 0.6836,
+      "step": 26115
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6889290188929826e-07,
+      "loss": 0.7061,
+      "step": 26116
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.686278040969781e-07,
+      "loss": 0.7119,
+      "step": 26117
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6836283526814333e-07,
+      "loss": 0.7051,
+      "step": 26118
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6809799540630345e-07,
+      "loss": 0.6934,
+      "step": 26119
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.678332845149689e-07,
+      "loss": 0.6943,
+      "step": 26120
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.67568702597647e-07,
+      "loss": 0.7227,
+      "step": 26121
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6730424965784594e-07,
+      "loss": 0.7109,
+      "step": 26122
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6703992569906855e-07,
+      "loss": 0.668,
+      "step": 26123
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6677573072481756e-07,
+      "loss": 0.6943,
+      "step": 26124
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.665116647385935e-07,
+      "loss": 0.707,
+      "step": 26125
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6624772774389907e-07,
+      "loss": 0.7012,
+      "step": 26126
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6598391974422933e-07,
+      "loss": 0.7168,
+      "step": 26127
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6572024074308257e-07,
+      "loss": 0.7041,
+      "step": 26128
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.654566907439493e-07,
+      "loss": 0.7051,
+      "step": 26129
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.651932697503268e-07,
+      "loss": 0.708,
+      "step": 26130
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6492997776570217e-07,
+      "loss": 0.708,
+      "step": 26131
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6466681479356715e-07,
+      "loss": 0.7383,
+      "step": 26132
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.644037808374078e-07,
+      "loss": 0.707,
+      "step": 26133
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6414087590071136e-07,
+      "loss": 0.707,
+      "step": 26134
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6387809998696167e-07,
+      "loss": 0.7197,
+      "step": 26135
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6361545309963934e-07,
+      "loss": 0.6904,
+      "step": 26136
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6335293524222595e-07,
+      "loss": 0.71,
+      "step": 26137
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6309054641820206e-07,
+      "loss": 0.7217,
+      "step": 26138
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.628282866310428e-07,
+      "loss": 0.7139,
+      "step": 26139
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.625661558842252e-07,
+      "loss": 0.6846,
+      "step": 26140
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.623041541812232e-07,
+      "loss": 0.6963,
+      "step": 26141
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.620422815255075e-07,
+      "loss": 0.7148,
+      "step": 26142
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6178053792054847e-07,
+      "loss": 0.707,
+      "step": 26143
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.615189233698168e-07,
+      "loss": 0.7354,
+      "step": 26144
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.612574378767785e-07,
+      "loss": 0.7129,
+      "step": 26145
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6099608144489754e-07,
+      "loss": 0.708,
+      "step": 26146
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.607348540776389e-07,
+      "loss": 0.7275,
+      "step": 26147
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6047375577846533e-07,
+      "loss": 0.7051,
+      "step": 26148
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.602127865508353e-07,
+      "loss": 0.6953,
+      "step": 26149
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.59951946398207e-07,
+      "loss": 0.7227,
+      "step": 26150
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5969123532403997e-07,
+      "loss": 0.7207,
+      "step": 26151
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5943065333178585e-07,
+      "loss": 0.6934,
+      "step": 26152
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.591702004248997e-07,
+      "loss": 0.6963,
+      "step": 26153
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.589098766068332e-07,
+      "loss": 0.6807,
+      "step": 26154
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5864968188103577e-07,
+      "loss": 0.6865,
+      "step": 26155
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.583896162509558e-07,
+      "loss": 0.7207,
+      "step": 26156
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5812967972003946e-07,
+      "loss": 0.6924,
+      "step": 26157
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5786987229173057e-07,
+      "loss": 0.7109,
+      "step": 26158
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5761019396947527e-07,
+      "loss": 0.71,
+      "step": 26159
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.57350644756712e-07,
+      "loss": 0.7021,
+      "step": 26160
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.570912246568835e-07,
+      "loss": 0.7051,
+      "step": 26161
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.568319336734226e-07,
+      "loss": 0.707,
+      "step": 26162
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.565727718097699e-07,
+      "loss": 0.6807,
+      "step": 26163
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.563137390693582e-07,
+      "loss": 0.7246,
+      "step": 26164
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.560548354556214e-07,
+      "loss": 0.7061,
+      "step": 26165
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5579606097198894e-07,
+      "loss": 0.7119,
+      "step": 26166
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.555374156218915e-07,
+      "loss": 0.6943,
+      "step": 26167
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5527889940875626e-07,
+      "loss": 0.7002,
+      "step": 26168
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.550205123360094e-07,
+      "loss": 0.709,
+      "step": 26169
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5476225440707383e-07,
+      "loss": 0.7119,
+      "step": 26170
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.545041256253744e-07,
+      "loss": 0.6729,
+      "step": 26171
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.542461259943296e-07,
+      "loss": 0.707,
+      "step": 26172
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.539882555173612e-07,
+      "loss": 0.7227,
+      "step": 26173
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.537305141978841e-07,
+      "loss": 0.7031,
+      "step": 26174
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.534729020393145e-07,
+      "loss": 0.7129,
+      "step": 26175
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5321541904506755e-07,
+      "loss": 0.709,
+      "step": 26176
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.529580652185537e-07,
+      "loss": 0.6924,
+      "step": 26177
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5270084056318586e-07,
+      "loss": 0.71,
+      "step": 26178
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.524437450823691e-07,
+      "loss": 0.7031,
+      "step": 26179
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5218677877951626e-07,
+      "loss": 0.7197,
+      "step": 26180
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5192994165802787e-07,
+      "loss": 0.6895,
+      "step": 26181
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.516732337213079e-07,
+      "loss": 0.7119,
+      "step": 26182
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.514166549727615e-07,
+      "loss": 0.7012,
+      "step": 26183
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.511602054157858e-07,
+      "loss": 0.6934,
+      "step": 26184
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5090388505378147e-07,
+      "loss": 0.6963,
+      "step": 26185
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.506476938901437e-07,
+      "loss": 0.6943,
+      "step": 26186
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5039163192826954e-07,
+      "loss": 0.7061,
+      "step": 26187
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.501356991715509e-07,
+      "loss": 0.7109,
+      "step": 26188
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4987989562337946e-07,
+      "loss": 0.708,
+      "step": 26189
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.496242212871458e-07,
+      "loss": 0.7227,
+      "step": 26190
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.493686761662373e-07,
+      "loss": 0.6963,
+      "step": 26191
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4911326026404225e-07,
+      "loss": 0.7041,
+      "step": 26192
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.488579735839447e-07,
+      "loss": 0.6875,
+      "step": 26193
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4860281612932635e-07,
+      "loss": 0.6904,
+      "step": 26194
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4834778790357117e-07,
+      "loss": 0.707,
+      "step": 26195
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.480928889100576e-07,
+      "loss": 0.7148,
+      "step": 26196
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4783811915216283e-07,
+      "loss": 0.709,
+      "step": 26197
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4758347863326424e-07,
+      "loss": 0.707,
+      "step": 26198
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4732896735673586e-07,
+      "loss": 0.6992,
+      "step": 26199
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4707458532595264e-07,
+      "loss": 0.6904,
+      "step": 26200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4682033254428307e-07,
+      "loss": 0.7012,
+      "step": 26201
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4656620901509664e-07,
+      "loss": 0.7646,
+      "step": 26202
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.463122147417618e-07,
+      "loss": 0.6904,
+      "step": 26203
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.460583497276447e-07,
+      "loss": 0.7148,
+      "step": 26204
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4580461397611055e-07,
+      "loss": 0.7129,
+      "step": 26205
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.455510074905221e-07,
+      "loss": 0.7119,
+      "step": 26206
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4529753027423665e-07,
+      "loss": 0.7246,
+      "step": 26207
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4504418233061713e-07,
+      "loss": 0.6855,
+      "step": 26208
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.447909636630197e-07,
+      "loss": 0.6787,
+      "step": 26209
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.445378742748006e-07,
+      "loss": 0.7041,
+      "step": 26210
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.442849141693127e-07,
+      "loss": 0.6836,
+      "step": 26211
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.440320833499088e-07,
+      "loss": 0.6836,
+      "step": 26212
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.437793818199419e-07,
+      "loss": 0.6797,
+      "step": 26213
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.43526809582757e-07,
+      "loss": 0.6826,
+      "step": 26214
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4327436664170146e-07,
+      "loss": 0.6875,
+      "step": 26215
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.430220530001237e-07,
+      "loss": 0.71,
+      "step": 26216
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4276986866136665e-07,
+      "loss": 0.6846,
+      "step": 26217
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.42517813628772e-07,
+      "loss": 0.6807,
+      "step": 26218
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.422658879056783e-07,
+      "loss": 0.709,
+      "step": 26219
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.420140914954272e-07,
+      "loss": 0.6846,
+      "step": 26220
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4176242440135277e-07,
+      "loss": 0.6846,
+      "step": 26221
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4151088662679235e-07,
+      "loss": 0.7002,
+      "step": 26222
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.412594781750777e-07,
+      "loss": 0.7061,
+      "step": 26223
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.410081990495416e-07,
+      "loss": 0.7109,
+      "step": 26224
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.407570492535138e-07,
+      "loss": 0.6855,
+      "step": 26225
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.405060287903238e-07,
+      "loss": 0.709,
+      "step": 26226
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4025513766329556e-07,
+      "loss": 0.7139,
+      "step": 26227
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4000437587575644e-07,
+      "loss": 0.7236,
+      "step": 26228
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.397537434310293e-07,
+      "loss": 0.7188,
+      "step": 26229
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.395032403324338e-07,
+      "loss": 0.71,
+      "step": 26230
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.392528665832916e-07,
+      "loss": 0.6924,
+      "step": 26231
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3900262218692015e-07,
+      "loss": 0.7158,
+      "step": 26232
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.387525071466357e-07,
+      "loss": 0.6865,
+      "step": 26233
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3850252146575214e-07,
+      "loss": 0.71,
+      "step": 26234
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3825266514758362e-07,
+      "loss": 0.6904,
+      "step": 26235
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3800293819543962e-07,
+      "loss": 0.7031,
+      "step": 26236
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.37753340612632e-07,
+      "loss": 0.709,
+      "step": 26237
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3750387240246698e-07,
+      "loss": 0.7012,
+      "step": 26238
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3725453356825078e-07,
+      "loss": 0.6943,
+      "step": 26239
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3700532411328857e-07,
+      "loss": 0.707,
+      "step": 26240
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3675624404088106e-07,
+      "loss": 0.7188,
+      "step": 26241
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3650729335433108e-07,
+      "loss": 0.7256,
+      "step": 26242
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.362584720569372e-07,
+      "loss": 0.6982,
+      "step": 26243
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.360097801519945e-07,
+      "loss": 0.7188,
+      "step": 26244
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3576121764280368e-07,
+      "loss": 0.6953,
+      "step": 26245
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3551278453265546e-07,
+      "loss": 0.6914,
+      "step": 26246
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3526448082484278e-07,
+      "loss": 0.709,
+      "step": 26247
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3501630652265405e-07,
+      "loss": 0.7041,
+      "step": 26248
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3476826162938227e-07,
+      "loss": 0.7148,
+      "step": 26249
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3452034614831253e-07,
+      "loss": 0.6768,
+      "step": 26250
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3427256008273113e-07,
+      "loss": 0.6777,
+      "step": 26251
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.340249034359199e-07,
+      "loss": 0.7119,
+      "step": 26252
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.337773762111628e-07,
+      "loss": 0.6973,
+      "step": 26253
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.335299784117395e-07,
+      "loss": 0.7109,
+      "step": 26254
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.332827100409274e-07,
+      "loss": 0.6797,
+      "step": 26255
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3303557110200492e-07,
+      "loss": 0.6963,
+      "step": 26256
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3278856159824725e-07,
+      "loss": 0.7129,
+      "step": 26257
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3254168153292843e-07,
+      "loss": 0.6816,
+      "step": 26258
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.322949309093181e-07,
+      "loss": 0.7148,
+      "step": 26259
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3204830973068582e-07,
+      "loss": 0.6904,
+      "step": 26260
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3180181800030342e-07,
+      "loss": 0.7119,
+      "step": 26261
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.31555455721435e-07,
+      "loss": 0.7119,
+      "step": 26262
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.313092228973457e-07,
+      "loss": 0.7227,
+      "step": 26263
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3106311953129955e-07,
+      "loss": 0.7119,
+      "step": 26264
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3081714562655733e-07,
+      "loss": 0.7168,
+      "step": 26265
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.305713011863786e-07,
+      "loss": 0.6973,
+      "step": 26266
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.303255862140219e-07,
+      "loss": 0.7119,
+      "step": 26267
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.300800007127435e-07,
+      "loss": 0.6953,
+      "step": 26268
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2983454468579526e-07,
+      "loss": 0.6738,
+      "step": 26269
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2958921813643454e-07,
+      "loss": 0.6787,
+      "step": 26270
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.293440210679121e-07,
+      "loss": 0.7275,
+      "step": 26271
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.290989534834731e-07,
+      "loss": 0.6816,
+      "step": 26272
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.288540153863683e-07,
+      "loss": 0.7061,
+      "step": 26273
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.28609206779844e-07,
+      "loss": 0.7217,
+      "step": 26274
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2836452766714313e-07,
+      "loss": 0.6982,
+      "step": 26275
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2811997805150975e-07,
+      "loss": 0.7119,
+      "step": 26276
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2787555793618354e-07,
+      "loss": 0.6709,
+      "step": 26277
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2763126732440411e-07,
+      "loss": 0.7061,
+      "step": 26278
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2738710621940886e-07,
+      "loss": 0.6973,
+      "step": 26279
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2714307462443297e-07,
+      "loss": 0.6943,
+      "step": 26280
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2689917254270943e-07,
+      "loss": 0.6953,
+      "step": 26281
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2665539997747343e-07,
+      "loss": 0.6846,
+      "step": 26282
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.264117569319546e-07,
+      "loss": 0.6934,
+      "step": 26283
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2616824340938148e-07,
+      "loss": 0.7207,
+      "step": 26284
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2592485941297925e-07,
+      "loss": 0.7207,
+      "step": 26285
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2568160494597536e-07,
+      "loss": 0.6855,
+      "step": 26286
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2543848001159385e-07,
+      "loss": 0.7139,
+      "step": 26287
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.251954846130555e-07,
+      "loss": 0.6924,
+      "step": 26288
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.249526187535811e-07,
+      "loss": 0.7012,
+      "step": 26289
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2470988243639024e-07,
+      "loss": 0.6865,
+      "step": 26290
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2446727566469705e-07,
+      "loss": 0.6914,
+      "step": 26291
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2422479844171897e-07,
+      "loss": 0.6943,
+      "step": 26292
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2398245077066784e-07,
+      "loss": 0.7148,
+      "step": 26293
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2374023265475775e-07,
+      "loss": 0.6836,
+      "step": 26294
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2349814409719616e-07,
+      "loss": 0.7158,
+      "step": 26295
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2325618510119162e-07,
+      "loss": 0.6885,
+      "step": 26296
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2301435566995267e-07,
+      "loss": 0.7021,
+      "step": 26297
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2277265580668117e-07,
+      "loss": 0.6963,
+      "step": 26298
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2253108551458346e-07,
+      "loss": 0.7021,
+      "step": 26299
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.222896447968581e-07,
+      "loss": 0.709,
+      "step": 26300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2204833365670474e-07,
+      "loss": 0.7061,
+      "step": 26301
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2180715209732307e-07,
+      "loss": 0.6807,
+      "step": 26302
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.215661001219105e-07,
+      "loss": 0.7012,
+      "step": 26303
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2132517773365892e-07,
+      "loss": 0.7178,
+      "step": 26304
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2108438493576024e-07,
+      "loss": 0.7031,
+      "step": 26305
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2084372173140745e-07,
+      "loss": 0.7188,
+      "step": 26306
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2060318812379132e-07,
+      "loss": 0.708,
+      "step": 26307
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.20362784116096e-07,
+      "loss": 0.7041,
+      "step": 26308
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2012250971151007e-07,
+      "loss": 0.7031,
+      "step": 26309
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1988236491321757e-07,
+      "loss": 0.6924,
+      "step": 26310
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1964234972439936e-07,
+      "loss": 0.6953,
+      "step": 26311
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.194024641482373e-07,
+      "loss": 0.7441,
+      "step": 26312
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1916270818790996e-07,
+      "loss": 0.7061,
+      "step": 26313
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1892308184659483e-07,
+      "loss": 0.6973,
+      "step": 26314
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1868358512746713e-07,
+      "loss": 0.7168,
+      "step": 26315
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.184442180337032e-07,
+      "loss": 0.7188,
+      "step": 26316
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1820498056847163e-07,
+      "loss": 0.7158,
+      "step": 26317
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.179658727349443e-07,
+      "loss": 0.7051,
+      "step": 26318
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1772689453629093e-07,
+      "loss": 0.6973,
+      "step": 26319
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1748804597567785e-07,
+      "loss": 0.6924,
+      "step": 26320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1724932705627034e-07,
+      "loss": 0.6865,
+      "step": 26321
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.170107377812314e-07,
+      "loss": 0.7041,
+      "step": 26322
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1677227815372405e-07,
+      "loss": 0.7178,
+      "step": 26323
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.16533948176908e-07,
+      "loss": 0.6846,
+      "step": 26324
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.162957478539407e-07,
+      "loss": 0.7061,
+      "step": 26325
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1605767718797965e-07,
+      "loss": 0.6904,
+      "step": 26326
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1581973618218012e-07,
+      "loss": 0.6914,
+      "step": 26327
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.155819248396951e-07,
+      "loss": 0.707,
+      "step": 26328
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1534424316367764e-07,
+      "loss": 0.7168,
+      "step": 26329
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1510669115727413e-07,
+      "loss": 0.6885,
+      "step": 26330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1486926882363534e-07,
+      "loss": 0.7197,
+      "step": 26331
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1463197616590657e-07,
+      "loss": 0.6816,
+      "step": 26332
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1439481318723422e-07,
+      "loss": 0.6973,
+      "step": 26333
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1415777989075792e-07,
+      "loss": 0.7158,
+      "step": 26334
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1392087627962188e-07,
+      "loss": 0.709,
+      "step": 26335
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.136841023569658e-07,
+      "loss": 0.6738,
+      "step": 26336
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1344745812592604e-07,
+      "loss": 0.7168,
+      "step": 26337
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.132109435896379e-07,
+      "loss": 0.752,
+      "step": 26338
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1297455875123773e-07,
+      "loss": 0.6875,
+      "step": 26339
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.127383036138575e-07,
+      "loss": 0.7119,
+      "step": 26340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.125021781806269e-07,
+      "loss": 0.6914,
+      "step": 26341
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.122661824546779e-07,
+      "loss": 0.7012,
+      "step": 26342
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1203031643913574e-07,
+      "loss": 0.707,
+      "step": 26343
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1179458013712572e-07,
+      "loss": 0.7061,
+      "step": 26344
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1155897355177423e-07,
+      "loss": 0.707,
+      "step": 26345
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.11323496686201e-07,
+      "loss": 0.7354,
+      "step": 26346
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1108814954352908e-07,
+      "loss": 0.7217,
+      "step": 26347
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1085293212687707e-07,
+      "loss": 0.6826,
+      "step": 26348
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.106178444393614e-07,
+      "loss": 0.7041,
+      "step": 26349
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.103828864840962e-07,
+      "loss": 0.7119,
+      "step": 26350
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.101480582641968e-07,
+      "loss": 0.6953,
+      "step": 26351
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0991335978277516e-07,
+      "loss": 0.6836,
+      "step": 26352
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.09678791042941e-07,
+      "loss": 0.6943,
+      "step": 26353
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.09444352047804e-07,
+      "loss": 0.7246,
+      "step": 26354
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0921004280047064e-07,
+      "loss": 0.7002,
+      "step": 26355
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0897586330404508e-07,
+      "loss": 0.7119,
+      "step": 26356
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0874181356163148e-07,
+      "loss": 0.7275,
+      "step": 26357
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0850789357633073e-07,
+      "loss": 0.7207,
+      "step": 26358
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0827410335124366e-07,
+      "loss": 0.6875,
+      "step": 26359
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.080404428894689e-07,
+      "loss": 0.7051,
+      "step": 26360
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0780691219410287e-07,
+      "loss": 0.709,
+      "step": 26361
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0757351126823977e-07,
+      "loss": 0.7314,
+      "step": 26362
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.073402401149738e-07,
+      "loss": 0.7148,
+      "step": 26363
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0710709873739466e-07,
+      "loss": 0.7012,
+      "step": 26364
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0687408713859325e-07,
+      "loss": 0.6865,
+      "step": 26365
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.066412053216571e-07,
+      "loss": 0.6895,
+      "step": 26366
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.064084532896715e-07,
+      "loss": 0.7139,
+      "step": 26367
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0617583104572403e-07,
+      "loss": 0.7119,
+      "step": 26368
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0594333859289438e-07,
+      "loss": 0.71,
+      "step": 26369
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0571097593426458e-07,
+      "loss": 0.7021,
+      "step": 26370
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0547874307291326e-07,
+      "loss": 0.6953,
+      "step": 26371
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.052466400119202e-07,
+      "loss": 0.707,
+      "step": 26372
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0501466675435954e-07,
+      "loss": 0.7246,
+      "step": 26373
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0478282330330557e-07,
+      "loss": 0.7285,
+      "step": 26374
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0455110966183244e-07,
+      "loss": 0.7178,
+      "step": 26375
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0431952583300774e-07,
+      "loss": 0.6768,
+      "step": 26376
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0408807181990342e-07,
+      "loss": 0.71,
+      "step": 26377
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0385674762558593e-07,
+      "loss": 0.7041,
+      "step": 26378
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.036255532531184e-07,
+      "loss": 0.6953,
+      "step": 26379
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0339448870556944e-07,
+      "loss": 0.7041,
+      "step": 26380
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0316355398599884e-07,
+      "loss": 0.707,
+      "step": 26381
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.029327490974653e-07,
+      "loss": 0.7451,
+      "step": 26382
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0270207404302855e-07,
+      "loss": 0.6982,
+      "step": 26383
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0247152882574727e-07,
+      "loss": 0.7207,
+      "step": 26384
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0224111344867569e-07,
+      "loss": 0.7129,
+      "step": 26385
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0201082791486692e-07,
+      "loss": 0.6992,
+      "step": 26386
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0178067222737297e-07,
+      "loss": 0.7002,
+      "step": 26387
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.015506463892436e-07,
+      "loss": 0.7275,
+      "step": 26388
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0132075040352861e-07,
+      "loss": 0.6963,
+      "step": 26389
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0109098427327335e-07,
+      "loss": 0.7119,
+      "step": 26390
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0086134800152313e-07,
+      "loss": 0.7363,
+      "step": 26391
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0063184159132e-07,
+      "loss": 0.6953,
+      "step": 26392
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0040246504570814e-07,
+      "loss": 0.6982,
+      "step": 26393
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.001732183677263e-07,
+      "loss": 0.6895,
+      "step": 26394
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9994410156040979e-07,
+      "loss": 0.6807,
+      "step": 26395
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9971511462679837e-07,
+      "loss": 0.7129,
+      "step": 26396
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9948625756992524e-07,
+      "loss": 0.6934,
+      "step": 26397
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9925753039282348e-07,
+      "loss": 0.709,
+      "step": 26398
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9902893309852512e-07,
+      "loss": 0.7061,
+      "step": 26399
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9880046569005662e-07,
+      "loss": 0.6943,
+      "step": 26400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9857212817045114e-07,
+      "loss": 0.7012,
+      "step": 26401
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9834392054272955e-07,
+      "loss": 0.6963,
+      "step": 26402
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9811584280991835e-07,
+      "loss": 0.7041,
+      "step": 26403
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9788789497503846e-07,
+      "loss": 0.6895,
+      "step": 26404
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9766007704111302e-07,
+      "loss": 0.6748,
+      "step": 26405
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.974323890111607e-07,
+      "loss": 0.6885,
+      "step": 26406
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9720483088819797e-07,
+      "loss": 0.7031,
+      "step": 26407
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9697740267524025e-07,
+      "loss": 0.7129,
+      "step": 26408
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9675010437530285e-07,
+      "loss": 0.7012,
+      "step": 26409
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9652293599139782e-07,
+      "loss": 0.7275,
+      "step": 26410
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.962958975265339e-07,
+      "loss": 0.7285,
+      "step": 26411
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9606898898372084e-07,
+      "loss": 0.7002,
+      "step": 26412
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.958422103659674e-07,
+      "loss": 0.6914,
+      "step": 26413
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9561556167627782e-07,
+      "loss": 0.6904,
+      "step": 26414
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9538904291765416e-07,
+      "loss": 0.6982,
+      "step": 26415
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9516265409309954e-07,
+      "loss": 0.6738,
+      "step": 26416
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9493639520561491e-07,
+      "loss": 0.7207,
+      "step": 26417
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.947102662581979e-07,
+      "loss": 0.7373,
+      "step": 26418
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9448426725384495e-07,
+      "loss": 0.707,
+      "step": 26419
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9425839819555148e-07,
+      "loss": 0.6963,
+      "step": 26420
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9403265908631174e-07,
+      "loss": 0.6885,
+      "step": 26421
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9380704992911558e-07,
+      "loss": 0.6865,
+      "step": 26422
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9358157072695392e-07,
+      "loss": 0.7236,
+      "step": 26423
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.933562214828133e-07,
+      "loss": 0.7266,
+      "step": 26424
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.931310021996824e-07,
+      "loss": 0.6934,
+      "step": 26425
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.929059128805455e-07,
+      "loss": 0.6953,
+      "step": 26426
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9268095352838468e-07,
+      "loss": 0.6719,
+      "step": 26427
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9245612414618088e-07,
+      "loss": 0.6816,
+      "step": 26428
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9223142473691504e-07,
+      "loss": 0.7109,
+      "step": 26429
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9200685530356367e-07,
+      "loss": 0.6836,
+      "step": 26430
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.917824158491033e-07,
+      "loss": 0.707,
+      "step": 26431
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9155810637650817e-07,
+      "loss": 0.6992,
+      "step": 26432
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9133392688875152e-07,
+      "loss": 0.6992,
+      "step": 26433
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9110987738880316e-07,
+      "loss": 0.71,
+      "step": 26434
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9088595787963404e-07,
+      "loss": 0.6855,
+      "step": 26435
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.906621683642096e-07,
+      "loss": 0.7051,
+      "step": 26436
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.904385088454952e-07,
+      "loss": 0.6982,
+      "step": 26437
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.902149793264574e-07,
+      "loss": 0.7041,
+      "step": 26438
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.899915798100571e-07,
+      "loss": 0.6885,
+      "step": 26439
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8976831029925536e-07,
+      "loss": 0.6953,
+      "step": 26440
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8954517079700972e-07,
+      "loss": 0.6895,
+      "step": 26441
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.89322161306279e-07,
+      "loss": 0.6963,
+      "step": 26442
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8909928183001635e-07,
+      "loss": 0.6904,
+      "step": 26443
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.888765323711783e-07,
+      "loss": 0.6787,
+      "step": 26444
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8865391293271363e-07,
+      "loss": 0.6904,
+      "step": 26445
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.884314235175766e-07,
+      "loss": 0.707,
+      "step": 26446
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8820906412871265e-07,
+      "loss": 0.7051,
+      "step": 26447
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8798683476906942e-07,
+      "loss": 0.7031,
+      "step": 26448
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8776473544159012e-07,
+      "loss": 0.6865,
+      "step": 26449
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.875427661492213e-07,
+      "loss": 0.7158,
+      "step": 26450
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.873209268949039e-07,
+      "loss": 0.7197,
+      "step": 26451
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8709921768157556e-07,
+      "loss": 0.6895,
+      "step": 26452
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8687763851217732e-07,
+      "loss": 0.6973,
+      "step": 26453
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8665618938964347e-07,
+      "loss": 0.6895,
+      "step": 26454
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8643487031690943e-07,
+      "loss": 0.6904,
+      "step": 26455
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8621368129690843e-07,
+      "loss": 0.6885,
+      "step": 26456
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.859926223325703e-07,
+      "loss": 0.71,
+      "step": 26457
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.857716934268272e-07,
+      "loss": 0.7324,
+      "step": 26458
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.855508945826068e-07,
+      "loss": 0.6943,
+      "step": 26459
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8533022580283223e-07,
+      "loss": 0.7061,
+      "step": 26460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8510968709042897e-07,
+      "loss": 0.7119,
+      "step": 26461
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.848892784483225e-07,
+      "loss": 0.6953,
+      "step": 26462
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8466899987943044e-07,
+      "loss": 0.6953,
+      "step": 26463
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8444885138667269e-07,
+      "loss": 0.6904,
+      "step": 26464
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.84228832972968e-07,
+      "loss": 0.6992,
+      "step": 26465
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8400894464123075e-07,
+      "loss": 0.7168,
+      "step": 26466
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8378918639437636e-07,
+      "loss": 0.7119,
+      "step": 26467
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8356955823531587e-07,
+      "loss": 0.6914,
+      "step": 26468
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8335006016696022e-07,
+      "loss": 0.6953,
+      "step": 26469
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8313069219221824e-07,
+      "loss": 0.6709,
+      "step": 26470
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.829114543139976e-07,
+      "loss": 0.708,
+      "step": 26471
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8269234653520374e-07,
+      "loss": 0.6846,
+      "step": 26472
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.824733688587388e-07,
+      "loss": 0.6904,
+      "step": 26473
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.822545212875071e-07,
+      "loss": 0.6895,
+      "step": 26474
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8203580382440743e-07,
+      "loss": 0.6963,
+      "step": 26475
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8181721647233753e-07,
+      "loss": 0.7266,
+      "step": 26476
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8159875923419612e-07,
+      "loss": 0.707,
+      "step": 26477
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.813804321128765e-07,
+      "loss": 0.7197,
+      "step": 26478
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8116223511127516e-07,
+      "loss": 0.7061,
+      "step": 26479
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.809441682322799e-07,
+      "loss": 0.6895,
+      "step": 26480
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8072623147878166e-07,
+      "loss": 0.6953,
+      "step": 26481
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8050842485366926e-07,
+      "loss": 0.6924,
+      "step": 26482
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.802907483598293e-07,
+      "loss": 0.71,
+      "step": 26483
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.800732020001461e-07,
+      "loss": 0.7109,
+      "step": 26484
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7985578577750407e-07,
+      "loss": 0.6982,
+      "step": 26485
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7963849969478197e-07,
+      "loss": 0.708,
+      "step": 26486
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7942134375486086e-07,
+      "loss": 0.7314,
+      "step": 26487
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7920431796061732e-07,
+      "loss": 0.6836,
+      "step": 26488
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7898742231493017e-07,
+      "loss": 0.7158,
+      "step": 26489
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.787706568206704e-07,
+      "loss": 0.6982,
+      "step": 26490
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7855402148071353e-07,
+      "loss": 0.707,
+      "step": 26491
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.783375162979284e-07,
+      "loss": 0.7119,
+      "step": 26492
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.78121141275186e-07,
+      "loss": 0.6885,
+      "step": 26493
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.779048964153507e-07,
+      "loss": 0.7061,
+      "step": 26494
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7768878172129244e-07,
+      "loss": 0.6973,
+      "step": 26495
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.774727971958723e-07,
+      "loss": 0.6982,
+      "step": 26496
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.772569428419535e-07,
+      "loss": 0.7119,
+      "step": 26497
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.770412186623971e-07,
+      "loss": 0.6738,
+      "step": 26498
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7682562466006192e-07,
+      "loss": 0.7256,
+      "step": 26499
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7661016083780348e-07,
+      "loss": 0.7129,
+      "step": 26500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7639482719847945e-07,
+      "loss": 0.6914,
+      "step": 26501
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.761796237449409e-07,
+      "loss": 0.6826,
+      "step": 26502
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7596455048004225e-07,
+      "loss": 0.7012,
+      "step": 26503
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7574960740663339e-07,
+      "loss": 0.707,
+      "step": 26504
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7553479452756096e-07,
+      "loss": 0.6855,
+      "step": 26505
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7532011184567266e-07,
+      "loss": 0.6963,
+      "step": 26506
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.75105559363814e-07,
+      "loss": 0.7197,
+      "step": 26507
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7489113708482942e-07,
+      "loss": 0.6924,
+      "step": 26508
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7467684501155768e-07,
+      "loss": 0.7432,
+      "step": 26509
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.74462683146841e-07,
+      "loss": 0.6934,
+      "step": 26510
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.742486514935171e-07,
+      "loss": 0.6836,
+      "step": 26511
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.740347500544215e-07,
+      "loss": 0.7344,
+      "step": 26512
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7382097883238968e-07,
+      "loss": 0.6797,
+      "step": 26513
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7360733783025275e-07,
+      "loss": 0.7363,
+      "step": 26514
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7339382705084506e-07,
+      "loss": 0.6885,
+      "step": 26515
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7318044649699438e-07,
+      "loss": 0.7051,
+      "step": 26516
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7296719617152958e-07,
+      "loss": 0.6816,
+      "step": 26517
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7275407607727502e-07,
+      "loss": 0.6963,
+      "step": 26518
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7254108621705622e-07,
+      "loss": 0.7256,
+      "step": 26519
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.723282265936954e-07,
+      "loss": 0.7139,
+      "step": 26520
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7211549721001253e-07,
+      "loss": 0.7012,
+      "step": 26521
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7190289806882865e-07,
+      "loss": 0.6875,
+      "step": 26522
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7169042917296041e-07,
+      "loss": 0.7002,
+      "step": 26523
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7147809052522335e-07,
+      "loss": 0.6865,
+      "step": 26524
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7126588212843186e-07,
+      "loss": 0.6846,
+      "step": 26525
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7105380398539707e-07,
+      "loss": 0.7207,
+      "step": 26526
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7084185609893e-07,
+      "loss": 0.7031,
+      "step": 26527
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7063003847184067e-07,
+      "loss": 0.7119,
+      "step": 26528
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7041835110693573e-07,
+      "loss": 0.6934,
+      "step": 26529
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7020679400701955e-07,
+      "loss": 0.6953,
+      "step": 26530
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.699953671748955e-07,
+      "loss": 0.7012,
+      "step": 26531
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.697840706133669e-07,
+      "loss": 0.6816,
+      "step": 26532
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6957290432523365e-07,
+      "loss": 0.6934,
+      "step": 26533
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.693618683132936e-07,
+      "loss": 0.709,
+      "step": 26534
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6915096258034224e-07,
+      "loss": 0.7412,
+      "step": 26535
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6894018712917736e-07,
+      "loss": 0.7178,
+      "step": 26536
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6872954196259228e-07,
+      "loss": 0.7129,
+      "step": 26537
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6851902708337586e-07,
+      "loss": 0.7148,
+      "step": 26538
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.683086424943181e-07,
+      "loss": 0.7217,
+      "step": 26539
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.68098388198209e-07,
+      "loss": 0.6865,
+      "step": 26540
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.678882641978341e-07,
+      "loss": 0.709,
+      "step": 26541
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6767827049597784e-07,
+      "loss": 0.6787,
+      "step": 26542
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6746840709542466e-07,
+      "loss": 0.7051,
+      "step": 26543
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6725867399895345e-07,
+      "loss": 0.7324,
+      "step": 26544
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6704907120934421e-07,
+      "loss": 0.7119,
+      "step": 26545
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6683959872937584e-07,
+      "loss": 0.709,
+      "step": 26546
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6663025656182163e-07,
+      "loss": 0.6992,
+      "step": 26547
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.664210447094605e-07,
+      "loss": 0.7334,
+      "step": 26548
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6621196317506139e-07,
+      "loss": 0.7129,
+      "step": 26549
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6600301196139645e-07,
+      "loss": 0.6865,
+      "step": 26550
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6579419107123241e-07,
+      "loss": 0.7139,
+      "step": 26551
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6558550050734033e-07,
+      "loss": 0.7041,
+      "step": 26552
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6537694027248362e-07,
+      "loss": 0.6973,
+      "step": 26553
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6516851036942782e-07,
+      "loss": 0.7002,
+      "step": 26554
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6496021080093295e-07,
+      "loss": 0.6895,
+      "step": 26555
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.647520415697601e-07,
+      "loss": 0.7227,
+      "step": 26556
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6454400267866932e-07,
+      "loss": 0.6836,
+      "step": 26557
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6433609413041617e-07,
+      "loss": 0.6797,
+      "step": 26558
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6412831592775625e-07,
+      "loss": 0.707,
+      "step": 26559
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6392066807344288e-07,
+      "loss": 0.6846,
+      "step": 26560
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.637131505702294e-07,
+      "loss": 0.6807,
+      "step": 26561
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6350576342086478e-07,
+      "loss": 0.7158,
+      "step": 26562
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6329850662809676e-07,
+      "loss": 0.6904,
+      "step": 26563
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6309138019467318e-07,
+      "loss": 0.7061,
+      "step": 26564
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6288438412333851e-07,
+      "loss": 0.6914,
+      "step": 26565
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.62677518416835e-07,
+      "loss": 0.6826,
+      "step": 26566
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6247078307790597e-07,
+      "loss": 0.6885,
+      "step": 26567
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6226417810928817e-07,
+      "loss": 0.7041,
+      "step": 26568
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6205770351372386e-07,
+      "loss": 0.709,
+      "step": 26569
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6185135929394636e-07,
+      "loss": 0.6973,
+      "step": 26570
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6164514545269015e-07,
+      "loss": 0.7041,
+      "step": 26571
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6143906199268866e-07,
+      "loss": 0.708,
+      "step": 26572
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.612331089166741e-07,
+      "loss": 0.6973,
+      "step": 26573
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.610272862273743e-07,
+      "loss": 0.6982,
+      "step": 26574
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6082159392751705e-07,
+      "loss": 0.7051,
+      "step": 26575
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6061603201982912e-07,
+      "loss": 0.7041,
+      "step": 26576
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6041060050703384e-07,
+      "loss": 0.6709,
+      "step": 26577
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6020529939185348e-07,
+      "loss": 0.7188,
+      "step": 26578
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6000012867701032e-07,
+      "loss": 0.7061,
+      "step": 26579
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5979508836522106e-07,
+      "loss": 0.7363,
+      "step": 26580
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5959017845920467e-07,
+      "loss": 0.6924,
+      "step": 26581
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.593853989616767e-07,
+      "loss": 0.6865,
+      "step": 26582
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.591807498753506e-07,
+      "loss": 0.7109,
+      "step": 26583
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5897623120293747e-07,
+      "loss": 0.7158,
+      "step": 26584
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5877184294714853e-07,
+      "loss": 0.6885,
+      "step": 26585
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.585675851106938e-07,
+      "loss": 0.7002,
+      "step": 26586
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.583634576962778e-07,
+      "loss": 0.709,
+      "step": 26587
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5815946070660614e-07,
+      "loss": 0.7051,
+      "step": 26588
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5795559414438333e-07,
+      "loss": 0.6855,
+      "step": 26589
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5775185801231052e-07,
+      "loss": 0.6924,
+      "step": 26590
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.575482523130878e-07,
+      "loss": 0.7119,
+      "step": 26591
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5734477704941187e-07,
+      "loss": 0.7061,
+      "step": 26592
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5714143222398282e-07,
+      "loss": 0.7012,
+      "step": 26593
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.569382178394918e-07,
+      "loss": 0.6943,
+      "step": 26594
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5673513389863558e-07,
+      "loss": 0.6943,
+      "step": 26595
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.565321804041009e-07,
+      "loss": 0.7197,
+      "step": 26596
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5632935735858112e-07,
+      "loss": 0.6953,
+      "step": 26597
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.561266647647619e-07,
+      "loss": 0.7168,
+      "step": 26598
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5592410262533109e-07,
+      "loss": 0.7148,
+      "step": 26599
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5572167094297207e-07,
+      "loss": 0.6875,
+      "step": 26600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5551936972036718e-07,
+      "loss": 0.71,
+      "step": 26601
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.553171989601987e-07,
+      "loss": 0.7178,
+      "step": 26602
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.551151586651456e-07,
+      "loss": 0.6816,
+      "step": 26603
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.549132488378835e-07,
+      "loss": 0.6758,
+      "step": 26604
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5471146948109028e-07,
+      "loss": 0.7178,
+      "step": 26605
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5450982059743934e-07,
+      "loss": 0.7246,
+      "step": 26606
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5430830218960302e-07,
+      "loss": 0.6904,
+      "step": 26607
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5410691426025137e-07,
+      "loss": 0.7041,
+      "step": 26608
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5390565681205448e-07,
+      "loss": 0.7031,
+      "step": 26609
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5370452984767915e-07,
+      "loss": 0.7041,
+      "step": 26610
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5350353336978986e-07,
+      "loss": 0.7119,
+      "step": 26611
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5330266738105004e-07,
+      "loss": 0.7246,
+      "step": 26612
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5310193188412204e-07,
+      "loss": 0.6992,
+      "step": 26613
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5290132688166814e-07,
+      "loss": 0.6875,
+      "step": 26614
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5270085237634513e-07,
+      "loss": 0.7002,
+      "step": 26615
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5250050837080864e-07,
+      "loss": 0.7021,
+      "step": 26616
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5230029486771547e-07,
+      "loss": 0.6895,
+      "step": 26617
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.521002118697179e-07,
+      "loss": 0.6953,
+      "step": 26618
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.519002593794683e-07,
+      "loss": 0.7227,
+      "step": 26619
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.517004373996156e-07,
+      "loss": 0.6865,
+      "step": 26620
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5150074593280883e-07,
+      "loss": 0.7148,
+      "step": 26621
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5130118498169365e-07,
+      "loss": 0.707,
+      "step": 26622
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5110175454891463e-07,
+      "loss": 0.6836,
+      "step": 26623
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5090245463711517e-07,
+      "loss": 0.7002,
+      "step": 26624
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5070328524893652e-07,
+      "loss": 0.7227,
+      "step": 26625
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5050424638701767e-07,
+      "loss": 0.7012,
+      "step": 26626
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.503053380539976e-07,
+      "loss": 0.7031,
+      "step": 26627
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5010656025251092e-07,
+      "loss": 0.7129,
+      "step": 26628
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4990791298519214e-07,
+      "loss": 0.6924,
+      "step": 26629
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4970939625467363e-07,
+      "loss": 0.748,
+      "step": 26630
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.495110100635877e-07,
+      "loss": 0.7031,
+      "step": 26631
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4931275441456115e-07,
+      "loss": 0.666,
+      "step": 26632
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4911462931022304e-07,
+      "loss": 0.7158,
+      "step": 26633
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4891663475319896e-07,
+      "loss": 0.7051,
+      "step": 26634
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4871877074611128e-07,
+      "loss": 0.6934,
+      "step": 26635
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.485210372915835e-07,
+      "loss": 0.7021,
+      "step": 26636
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4832343439223463e-07,
+      "loss": 0.7197,
+      "step": 26637
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4812596205068585e-07,
+      "loss": 0.6846,
+      "step": 26638
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.479286202695529e-07,
+      "loss": 0.6953,
+      "step": 26639
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4773140905145033e-07,
+      "loss": 0.7021,
+      "step": 26640
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4753432839899162e-07,
+      "loss": 0.7051,
+      "step": 26641
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4733737831479023e-07,
+      "loss": 0.6973,
+      "step": 26642
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4714055880145407e-07,
+      "loss": 0.6924,
+      "step": 26643
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4694386986159326e-07,
+      "loss": 0.7109,
+      "step": 26644
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4674731149781352e-07,
+      "loss": 0.6992,
+      "step": 26645
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.465508837127183e-07,
+      "loss": 0.6777,
+      "step": 26646
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4635458650891444e-07,
+      "loss": 0.7129,
+      "step": 26647
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4615841988899982e-07,
+      "loss": 0.6904,
+      "step": 26648
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.459623838555757e-07,
+      "loss": 0.7031,
+      "step": 26649
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4576647841124002e-07,
+      "loss": 0.6914,
+      "step": 26650
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4557070355858848e-07,
+      "loss": 0.7041,
+      "step": 26651
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4537505930021568e-07,
+      "loss": 0.7188,
+      "step": 26652
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4517954563871395e-07,
+      "loss": 0.71,
+      "step": 26653
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.449841625766757e-07,
+      "loss": 0.6992,
+      "step": 26654
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4478891011668883e-07,
+      "loss": 0.7041,
+      "step": 26655
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.445937882613413e-07,
+      "loss": 0.7012,
+      "step": 26656
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.443987970132199e-07,
+      "loss": 0.6992,
+      "step": 26657
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4420393637490593e-07,
+      "loss": 0.7314,
+      "step": 26658
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4400920634898507e-07,
+      "loss": 0.6982,
+      "step": 26659
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.438146069380364e-07,
+      "loss": 0.7139,
+      "step": 26660
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4362013814463893e-07,
+      "loss": 0.707,
+      "step": 26661
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.434257999713695e-07,
+      "loss": 0.708,
+      "step": 26662
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4323159242080387e-07,
+      "loss": 0.7129,
+      "step": 26663
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4303751549551658e-07,
+      "loss": 0.707,
+      "step": 26664
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4284356919807784e-07,
+      "loss": 0.6738,
+      "step": 26665
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4264975353105893e-07,
+      "loss": 0.7051,
+      "step": 26666
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4245606849702887e-07,
+      "loss": 0.6953,
+      "step": 26667
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4226251409855453e-07,
+      "loss": 0.7178,
+      "step": 26668
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4206909033819938e-07,
+      "loss": 0.7275,
+      "step": 26669
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4187579721852695e-07,
+      "loss": 0.7002,
+      "step": 26670
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4168263474209965e-07,
+      "loss": 0.7236,
+      "step": 26671
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4148960291147761e-07,
+      "loss": 0.7129,
+      "step": 26672
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4129670172921995e-07,
+      "loss": 0.7051,
+      "step": 26673
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4110393119787902e-07,
+      "loss": 0.6895,
+      "step": 26674
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.409112913200139e-07,
+      "loss": 0.6885,
+      "step": 26675
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.407187820981748e-07,
+      "loss": 0.6953,
+      "step": 26676
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.405264035349141e-07,
+      "loss": 0.6924,
+      "step": 26677
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4033415563278197e-07,
+      "loss": 0.6934,
+      "step": 26678
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4014203839432305e-07,
+      "loss": 0.6943,
+      "step": 26679
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3995005182208865e-07,
+      "loss": 0.6924,
+      "step": 26680
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3975819591861784e-07,
+      "loss": 0.7061,
+      "step": 26681
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3956647068645411e-07,
+      "loss": 0.7129,
+      "step": 26682
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3937487612814104e-07,
+      "loss": 0.7236,
+      "step": 26683
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3918341224621545e-07,
+      "loss": 0.7021,
+      "step": 26684
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3899207904321644e-07,
+      "loss": 0.7051,
+      "step": 26685
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.388008765216775e-07,
+      "loss": 0.708,
+      "step": 26686
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3860980468413332e-07,
+      "loss": 0.7061,
+      "step": 26687
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3841886353311629e-07,
+      "loss": 0.7021,
+      "step": 26688
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.382280530711566e-07,
+      "loss": 0.6875,
+      "step": 26689
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3803737330078337e-07,
+      "loss": 0.7041,
+      "step": 26690
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3784682422452234e-07,
+      "loss": 0.7031,
+      "step": 26691
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3765640584490037e-07,
+      "loss": 0.7031,
+      "step": 26692
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.37466118164441e-07,
+      "loss": 0.7002,
+      "step": 26693
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3727596118566444e-07,
+      "loss": 0.6992,
+      "step": 26694
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.370859349110909e-07,
+      "loss": 0.6865,
+      "step": 26695
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3689603934323947e-07,
+      "loss": 0.6934,
+      "step": 26696
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3670627448462704e-07,
+      "loss": 0.7119,
+      "step": 26697
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3651664033776823e-07,
+      "loss": 0.6924,
+      "step": 26698
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3632713690517442e-07,
+      "loss": 0.7178,
+      "step": 26699
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.361377641893591e-07,
+      "loss": 0.6885,
+      "step": 26700
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.359485221928314e-07,
+      "loss": 0.7305,
+      "step": 26701
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.357594109180993e-07,
+      "loss": 0.709,
+      "step": 26702
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3557043036766748e-07,
+      "loss": 0.6738,
+      "step": 26703
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3538158054404172e-07,
+      "loss": 0.6924,
+      "step": 26704
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3519286144972665e-07,
+      "loss": 0.7109,
+      "step": 26705
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.350042730872192e-07,
+      "loss": 0.6836,
+      "step": 26706
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3481581545901957e-07,
+      "loss": 0.7207,
+      "step": 26707
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3462748856762796e-07,
+      "loss": 0.6973,
+      "step": 26708
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3443929241553798e-07,
+      "loss": 0.6836,
+      "step": 26709
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3425122700524428e-07,
+      "loss": 0.6904,
+      "step": 26710
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.340632923392382e-07,
+      "loss": 0.6963,
+      "step": 26711
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.338754884200122e-07,
+      "loss": 0.707,
+      "step": 26712
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3368781525005314e-07,
+      "loss": 0.7109,
+      "step": 26713
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3350027283184907e-07,
+      "loss": 0.6904,
+      "step": 26714
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3331286116788576e-07,
+      "loss": 0.6836,
+      "step": 26715
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.331255802606457e-07,
+      "loss": 0.6943,
+      "step": 26716
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3293843011261242e-07,
+      "loss": 0.709,
+      "step": 26717
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3275141072626508e-07,
+      "loss": 0.7178,
+      "step": 26718
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3256452210408167e-07,
+      "loss": 0.7266,
+      "step": 26719
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.323777642485402e-07,
+      "loss": 0.6768,
+      "step": 26720
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.321911371621143e-07,
+      "loss": 0.6934,
+      "step": 26721
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.320046408472775e-07,
+      "loss": 0.7334,
+      "step": 26722
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3181827530650227e-07,
+      "loss": 0.6855,
+      "step": 26723
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3163204054225774e-07,
+      "loss": 0.7021,
+      "step": 26724
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.31445936557012e-07,
+      "loss": 0.6797,
+      "step": 26725
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3125996335323187e-07,
+      "loss": 0.6963,
+      "step": 26726
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.310741209333799e-07,
+      "loss": 0.7207,
+      "step": 26727
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3088840929992074e-07,
+      "loss": 0.7207,
+      "step": 26728
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.307028284553158e-07,
+      "loss": 0.7217,
+      "step": 26729
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3051737840202418e-07,
+      "loss": 0.7021,
+      "step": 26730
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.303320591425028e-07,
+      "loss": 0.6777,
+      "step": 26731
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3014687067920863e-07,
+      "loss": 0.7002,
+      "step": 26732
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.299618130145941e-07,
+      "loss": 0.7275,
+      "step": 26733
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2977688615111395e-07,
+      "loss": 0.7217,
+      "step": 26734
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2959209009121732e-07,
+      "loss": 0.7041,
+      "step": 26735
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2940742483735225e-07,
+      "loss": 0.6924,
+      "step": 26736
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.292228903919679e-07,
+      "loss": 0.7236,
+      "step": 26737
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2903848675751118e-07,
+      "loss": 0.6973,
+      "step": 26738
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.288542139364224e-07,
+      "loss": 0.6914,
+      "step": 26739
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.28670071931144e-07,
+      "loss": 0.6982,
+      "step": 26740
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.284860607441185e-07,
+      "loss": 0.7051,
+      "step": 26741
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2830218037778397e-07,
+      "loss": 0.707,
+      "step": 26742
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.281184308345762e-07,
+      "loss": 0.7012,
+      "step": 26743
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2793481211692993e-07,
+      "loss": 0.7354,
+      "step": 26744
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2775132422727988e-07,
+      "loss": 0.6992,
+      "step": 26745
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.275679671680574e-07,
+      "loss": 0.7021,
+      "step": 26746
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2738474094169283e-07,
+      "loss": 0.6992,
+      "step": 26747
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2720164555061198e-07,
+      "loss": 0.6885,
+      "step": 26748
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.27018680997244e-07,
+      "loss": 0.7031,
+      "step": 26749
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2683584728401254e-07,
+      "loss": 0.6836,
+      "step": 26750
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2665314441334232e-07,
+      "loss": 0.7002,
+      "step": 26751
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2647057238765027e-07,
+      "loss": 0.7012,
+      "step": 26752
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2628813120936002e-07,
+      "loss": 0.7256,
+      "step": 26753
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.261058208808874e-07,
+      "loss": 0.7227,
+      "step": 26754
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.259236414046483e-07,
+      "loss": 0.6914,
+      "step": 26755
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2574159278305852e-07,
+      "loss": 0.6836,
+      "step": 26756
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2555967501852951e-07,
+      "loss": 0.6973,
+      "step": 26757
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2537788811347263e-07,
+      "loss": 0.6797,
+      "step": 26758
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2519623207029706e-07,
+      "loss": 0.7188,
+      "step": 26759
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.250147068914076e-07,
+      "loss": 0.6758,
+      "step": 26760
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.248333125792145e-07,
+      "loss": 0.7227,
+      "step": 26761
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2465204913611805e-07,
+      "loss": 0.6963,
+      "step": 26762
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2447091656452193e-07,
+      "loss": 0.708,
+      "step": 26763
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.242899148668264e-07,
+      "loss": 0.6787,
+      "step": 26764
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2410904404542957e-07,
+      "loss": 0.7363,
+      "step": 26765
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.239283041027295e-07,
+      "loss": 0.7041,
+      "step": 26766
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2374769504111983e-07,
+      "loss": 0.6875,
+      "step": 26767
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2356721686299645e-07,
+      "loss": 0.7021,
+      "step": 26768
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2338686957074742e-07,
+      "loss": 0.7158,
+      "step": 26769
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.232066531667675e-07,
+      "loss": 0.6953,
+      "step": 26770
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2302656765344145e-07,
+      "loss": 0.707,
+      "step": 26771
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2284661303315627e-07,
+      "loss": 0.7139,
+      "step": 26772
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2266678930829778e-07,
+      "loss": 0.6895,
+      "step": 26773
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2248709648124858e-07,
+      "loss": 0.7129,
+      "step": 26774
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2230753455439004e-07,
+      "loss": 0.7012,
+      "step": 26775
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.221281035301014e-07,
+      "loss": 0.7002,
+      "step": 26776
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2194880341076188e-07,
+      "loss": 0.6904,
+      "step": 26777
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.217696341987462e-07,
+      "loss": 0.7373,
+      "step": 26778
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.215905958964292e-07,
+      "loss": 0.71,
+      "step": 26779
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2141168850618446e-07,
+      "loss": 0.6914,
+      "step": 26780
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2123291203038123e-07,
+      "loss": 0.6885,
+      "step": 26781
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2105426647138985e-07,
+      "loss": 0.7012,
+      "step": 26782
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.208757518315784e-07,
+      "loss": 0.7012,
+      "step": 26783
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2069736811331056e-07,
+      "loss": 0.7168,
+      "step": 26784
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2051911531895112e-07,
+      "loss": 0.7002,
+      "step": 26785
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.203409934508637e-07,
+      "loss": 0.6836,
+      "step": 26786
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2016300251140755e-07,
+      "loss": 0.6904,
+      "step": 26787
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1998514250294192e-07,
+      "loss": 0.7197,
+      "step": 26788
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1980741342782265e-07,
+      "loss": 0.7012,
+      "step": 26789
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.196298152884079e-07,
+      "loss": 0.7178,
+      "step": 26790
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.194523480870491e-07,
+      "loss": 0.71,
+      "step": 26791
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.192750118260988e-07,
+      "loss": 0.6924,
+      "step": 26792
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1909780650790515e-07,
+      "loss": 0.6826,
+      "step": 26793
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1892073213481958e-07,
+      "loss": 0.709,
+      "step": 26794
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.187437887091869e-07,
+      "loss": 0.6943,
+      "step": 26795
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1856697623335412e-07,
+      "loss": 0.7061,
+      "step": 26796
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1839029470966045e-07,
+      "loss": 0.6953,
+      "step": 26797
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1821374414045184e-07,
+      "loss": 0.7246,
+      "step": 26798
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1803732452806416e-07,
+      "loss": 0.71,
+      "step": 26799
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1786103587483778e-07,
+      "loss": 0.6846,
+      "step": 26800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1768487818310858e-07,
+      "loss": 0.6914,
+      "step": 26801
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1750885145521029e-07,
+      "loss": 0.7012,
+      "step": 26802
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1733295569347659e-07,
+      "loss": 0.6865,
+      "step": 26803
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1715719090023781e-07,
+      "loss": 0.6777,
+      "step": 26804
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1698155707782211e-07,
+      "loss": 0.7041,
+      "step": 26805
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1680605422855984e-07,
+      "loss": 0.6875,
+      "step": 26806
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1663068235477581e-07,
+      "loss": 0.7217,
+      "step": 26807
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1645544145879262e-07,
+      "loss": 0.6953,
+      "step": 26808
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1628033154293395e-07,
+      "loss": 0.7178,
+      "step": 26809
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1610535260952127e-07,
+      "loss": 0.7354,
+      "step": 26810
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1593050466087163e-07,
+      "loss": 0.6943,
+      "step": 26811
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1575578769930207e-07,
+      "loss": 0.6943,
+      "step": 26812
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.155812017271296e-07,
+      "loss": 0.708,
+      "step": 26813
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1540674674666685e-07,
+      "loss": 0.6982,
+      "step": 26814
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.152324227602264e-07,
+      "loss": 0.7402,
+      "step": 26815
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.150582297701186e-07,
+      "loss": 0.6992,
+      "step": 26816
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1488416777865053e-07,
+      "loss": 0.6914,
+      "step": 26817
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.147102367881292e-07,
+      "loss": 0.7002,
+      "step": 26818
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1453643680086057e-07,
+      "loss": 0.6963,
+      "step": 26819
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1436276781914835e-07,
+      "loss": 0.7012,
+      "step": 26820
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.141892298452929e-07,
+      "loss": 0.6982,
+      "step": 26821
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1401582288159352e-07,
+      "loss": 0.7021,
+      "step": 26822
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1384254693035058e-07,
+      "loss": 0.7061,
+      "step": 26823
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1366940199385779e-07,
+      "loss": 0.7041,
+      "step": 26824
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1349638807441111e-07,
+      "loss": 0.7051,
+      "step": 26825
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1332350517430202e-07,
+      "loss": 0.7021,
+      "step": 26826
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1315075329582425e-07,
+      "loss": 0.7207,
+      "step": 26827
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1297813244126488e-07,
+      "loss": 0.6875,
+      "step": 26828
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1280564261291316e-07,
+      "loss": 0.71,
+      "step": 26829
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1263328381305283e-07,
+      "loss": 0.7012,
+      "step": 26830
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1246105604396985e-07,
+      "loss": 0.7012,
+      "step": 26831
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1228895930794681e-07,
+      "loss": 0.7051,
+      "step": 26832
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1211699360726302e-07,
+      "loss": 0.6924,
+      "step": 26833
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1194515894419666e-07,
+      "loss": 0.7119,
+      "step": 26834
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1177345532102923e-07,
+      "loss": 0.7188,
+      "step": 26835
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1160188274003114e-07,
+      "loss": 0.6953,
+      "step": 26836
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1143044120347945e-07,
+      "loss": 0.6934,
+      "step": 26837
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1125913071364458e-07,
+      "loss": 0.7051,
+      "step": 26838
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1108795127279693e-07,
+      "loss": 0.707,
+      "step": 26839
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1091690288320578e-07,
+      "loss": 0.6855,
+      "step": 26840
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1074598554713822e-07,
+      "loss": 0.6797,
+      "step": 26841
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.10575199266858e-07,
+      "loss": 0.71,
+      "step": 26842
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1040454404462886e-07,
+      "loss": 0.7021,
+      "step": 26843
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1023401988271231e-07,
+      "loss": 0.6963,
+      "step": 26844
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.100636267833699e-07,
+      "loss": 0.6924,
+      "step": 26845
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0989336474885648e-07,
+      "loss": 0.7188,
+      "step": 26846
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0972323378143024e-07,
+      "loss": 0.6934,
+      "step": 26847
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0955323388334716e-07,
+      "loss": 0.6875,
+      "step": 26848
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0938336505685876e-07,
+      "loss": 0.6865,
+      "step": 26849
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0921362730421437e-07,
+      "loss": 0.707,
+      "step": 26850
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0904402062766661e-07,
+      "loss": 0.7217,
+      "step": 26851
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0887454502946149e-07,
+      "loss": 0.6973,
+      "step": 26852
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0870520051184608e-07,
+      "loss": 0.6934,
+      "step": 26853
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0853598707706304e-07,
+      "loss": 0.7256,
+      "step": 26854
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0836690472735611e-07,
+      "loss": 0.6865,
+      "step": 26855
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0819795346496464e-07,
+      "loss": 0.7227,
+      "step": 26856
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0802913329212905e-07,
+      "loss": 0.6953,
+      "step": 26857
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0786044421108643e-07,
+      "loss": 0.6963,
+      "step": 26858
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0769188622407057e-07,
+      "loss": 0.7207,
+      "step": 26859
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0752345933331743e-07,
+      "loss": 0.7109,
+      "step": 26860
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0735516354105968e-07,
+      "loss": 0.6875,
+      "step": 26861
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0718699884952444e-07,
+      "loss": 0.6914,
+      "step": 26862
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0701896526094214e-07,
+      "loss": 0.7148,
+      "step": 26863
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0685106277753987e-07,
+      "loss": 0.708,
+      "step": 26864
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0668329140154255e-07,
+      "loss": 0.7031,
+      "step": 26865
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0651565113517282e-07,
+      "loss": 0.7031,
+      "step": 26866
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0634814198065335e-07,
+      "loss": 0.7041,
+      "step": 26867
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0618076394020349e-07,
+      "loss": 0.7197,
+      "step": 26868
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0601351701604146e-07,
+      "loss": 0.7002,
+      "step": 26869
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0584640121038437e-07,
+      "loss": 0.6689,
+      "step": 26870
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0567941652544489e-07,
+      "loss": 0.6914,
+      "step": 26871
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0551256296343793e-07,
+      "loss": 0.6885,
+      "step": 26872
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0534584052657392e-07,
+      "loss": 0.6855,
+      "step": 26873
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0517924921706225e-07,
+      "loss": 0.7148,
+      "step": 26874
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.050127890371111e-07,
+      "loss": 0.7109,
+      "step": 26875
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.048464599889254e-07,
+      "loss": 0.709,
+      "step": 26876
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0468026207471116e-07,
+      "loss": 0.7002,
+      "step": 26877
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0451419529666995e-07,
+      "loss": 0.7031,
+      "step": 26878
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0434825965700112e-07,
+      "loss": 0.751,
+      "step": 26879
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0418245515790626e-07,
+      "loss": 0.6904,
+      "step": 26880
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0401678180158248e-07,
+      "loss": 0.6924,
+      "step": 26881
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0385123959022247e-07,
+      "loss": 0.6904,
+      "step": 26882
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0368582852602227e-07,
+      "loss": 0.6982,
+      "step": 26883
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0352054861117455e-07,
+      "loss": 0.6992,
+      "step": 26884
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0335539984786757e-07,
+      "loss": 0.6934,
+      "step": 26885
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.031903822382918e-07,
+      "loss": 0.6807,
+      "step": 26886
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0302549578463439e-07,
+      "loss": 0.6963,
+      "step": 26887
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0286074048907801e-07,
+      "loss": 0.7031,
+      "step": 26888
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.026961163538076e-07,
+      "loss": 0.6787,
+      "step": 26889
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0253162338100587e-07,
+      "loss": 0.7158,
+      "step": 26890
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0236726157285104e-07,
+      "loss": 0.7295,
+      "step": 26891
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.022030309315214e-07,
+      "loss": 0.7002,
+      "step": 26892
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0203893145919408e-07,
+      "loss": 0.6924,
+      "step": 26893
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0187496315804402e-07,
+      "loss": 0.7139,
+      "step": 26894
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.017111260302428e-07,
+      "loss": 0.7051,
+      "step": 26895
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0154742007796314e-07,
+      "loss": 0.7119,
+      "step": 26896
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0138384530337442e-07,
+      "loss": 0.6904,
+      "step": 26897
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0122040170864378e-07,
+      "loss": 0.7266,
+      "step": 26898
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0105708929593617e-07,
+      "loss": 0.7129,
+      "step": 26899
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0089390806741873e-07,
+      "loss": 0.6934,
+      "step": 26900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0073085802525195e-07,
+      "loss": 0.7188,
+      "step": 26901
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0056793917159635e-07,
+      "loss": 0.71,
+      "step": 26902
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.004051515086124e-07,
+      "loss": 0.6963,
+      "step": 26903
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0024249503845507e-07,
+      "loss": 0.6875,
+      "step": 26904
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.000799697632826e-07,
+      "loss": 0.6914,
+      "step": 26905
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.991757568524885e-08,
+      "loss": 0.6953,
+      "step": 26906
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.97553128065043e-08,
+      "loss": 0.7178,
+      "step": 26907
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.959318112919835e-08,
+      "loss": 0.6797,
+      "step": 26908
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.943118065548263e-08,
+      "loss": 0.6768,
+      "step": 26909
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.926931138750207e-08,
+      "loss": 0.7139,
+      "step": 26910
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.910757332740273e-08,
+      "loss": 0.7012,
+      "step": 26911
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.894596647732624e-08,
+      "loss": 0.7002,
+      "step": 26912
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.878449083941755e-08,
+      "loss": 0.7295,
+      "step": 26913
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.862314641581272e-08,
+      "loss": 0.707,
+      "step": 26914
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.846193320865338e-08,
+      "loss": 0.6953,
+      "step": 26915
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.830085122007338e-08,
+      "loss": 0.7031,
+      "step": 26916
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.813990045220879e-08,
+      "loss": 0.6953,
+      "step": 26917
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.797908090719233e-08,
+      "loss": 0.7061,
+      "step": 26918
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.781839258715564e-08,
+      "loss": 0.707,
+      "step": 26919
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.765783549422703e-08,
+      "loss": 0.6982,
+      "step": 26920
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.749740963053588e-08,
+      "loss": 0.7129,
+      "step": 26921
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.733711499820608e-08,
+      "loss": 0.6914,
+      "step": 26922
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.717695159936479e-08,
+      "loss": 0.6973,
+      "step": 26923
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.701691943613145e-08,
+      "loss": 0.6738,
+      "step": 26924
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.685701851062879e-08,
+      "loss": 0.6953,
+      "step": 26925
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.669724882497622e-08,
+      "loss": 0.6914,
+      "step": 26926
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.653761038128983e-08,
+      "loss": 0.6982,
+      "step": 26927
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.637810318168462e-08,
+      "loss": 0.7061,
+      "step": 26928
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.621872722827664e-08,
+      "loss": 0.6953,
+      "step": 26929
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.605948252317643e-08,
+      "loss": 0.6953,
+      "step": 26930
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.590036906849564e-08,
+      "loss": 0.6924,
+      "step": 26931
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.574138686634038e-08,
+      "loss": 0.7207,
+      "step": 26932
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.558253591882005e-08,
+      "loss": 0.7197,
+      "step": 26933
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.542381622803853e-08,
+      "loss": 0.6963,
+      "step": 26934
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.52652277961008e-08,
+      "loss": 0.7148,
+      "step": 26935
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.510677062510631e-08,
+      "loss": 0.7002,
+      "step": 26936
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.494844471715558e-08,
+      "loss": 0.6924,
+      "step": 26937
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.479025007434695e-08,
+      "loss": 0.7383,
+      "step": 26938
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.463218669877871e-08,
+      "loss": 0.7139,
+      "step": 26939
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.447425459254256e-08,
+      "loss": 0.6992,
+      "step": 26940
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.431645375773346e-08,
+      "loss": 0.6924,
+      "step": 26941
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.415878419644309e-08,
+      "loss": 0.709,
+      "step": 26942
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.400124591075865e-08,
+      "loss": 0.7227,
+      "step": 26943
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.38438389027696e-08,
+      "loss": 0.7334,
+      "step": 26944
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.368656317456093e-08,
+      "loss": 0.708,
+      "step": 26945
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.352941872821875e-08,
+      "loss": 0.7158,
+      "step": 26946
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.33724055658236e-08,
+      "loss": 0.6982,
+      "step": 26947
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.32155236894572e-08,
+      "loss": 0.6934,
+      "step": 26948
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.305877310119782e-08,
+      "loss": 0.7031,
+      "step": 26949
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.290215380312384e-08,
+      "loss": 0.7168,
+      "step": 26950
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.274566579731025e-08,
+      "loss": 0.6904,
+      "step": 26951
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.258930908583208e-08,
+      "loss": 0.7012,
+      "step": 26952
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.243308367075876e-08,
+      "loss": 0.6992,
+      "step": 26953
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.227698955416309e-08,
+      "loss": 0.7012,
+      "step": 26954
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.21210267381123e-08,
+      "loss": 0.6953,
+      "step": 26955
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.196519522467473e-08,
+      "loss": 0.7129,
+      "step": 26956
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.180949501591319e-08,
+      "loss": 0.7061,
+      "step": 26957
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.165392611389379e-08,
+      "loss": 0.6895,
+      "step": 26958
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.1498488520676e-08,
+      "loss": 0.7012,
+      "step": 26959
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.13431822383215e-08,
+      "loss": 0.7002,
+      "step": 26960
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.118800726888643e-08,
+      "loss": 0.7021,
+      "step": 26961
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.103296361442915e-08,
+      "loss": 0.6787,
+      "step": 26962
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.087805127700356e-08,
+      "loss": 0.7256,
+      "step": 26963
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.072327025866135e-08,
+      "loss": 0.7148,
+      "step": 26964
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.056862056145643e-08,
+      "loss": 0.7031,
+      "step": 26965
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.041410218743607e-08,
+      "loss": 0.6865,
+      "step": 26966
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.025971513864973e-08,
+      "loss": 0.6963,
+      "step": 26967
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.010545941714133e-08,
+      "loss": 0.7051,
+      "step": 26968
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.995133502495701e-08,
+      "loss": 0.6914,
+      "step": 26969
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.979734196413737e-08,
+      "loss": 0.6904,
+      "step": 26970
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.96434802367252e-08,
+      "loss": 0.7139,
+      "step": 26971
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.948974984475778e-08,
+      "loss": 0.6924,
+      "step": 26972
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.933615079027347e-08,
+      "loss": 0.7002,
+      "step": 26973
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.918268307530842e-08,
+      "loss": 0.6914,
+      "step": 26974
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.902934670189434e-08,
+      "loss": 0.6943,
+      "step": 26975
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.887614167206515e-08,
+      "loss": 0.6865,
+      "step": 26976
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.872306798785146e-08,
+      "loss": 0.7197,
+      "step": 26977
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.857012565127942e-08,
+      "loss": 0.6963,
+      "step": 26978
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.841731466437853e-08,
+      "loss": 0.7021,
+      "step": 26979
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.82646350291727e-08,
+      "loss": 0.7139,
+      "step": 26980
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.811208674768479e-08,
+      "loss": 0.7178,
+      "step": 26981
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.79596698219376e-08,
+      "loss": 0.6924,
+      "step": 26982
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.780738425394952e-08,
+      "loss": 0.7188,
+      "step": 26983
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.765523004574117e-08,
+      "loss": 0.7607,
+      "step": 26984
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.750320719932536e-08,
+      "loss": 0.7031,
+      "step": 26985
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.73513157167205e-08,
+      "loss": 0.6797,
+      "step": 26986
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.71995555999372e-08,
+      "loss": 0.6982,
+      "step": 26987
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.704792685098717e-08,
+      "loss": 0.6973,
+      "step": 26988
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.689642947187993e-08,
+      "loss": 0.709,
+      "step": 26989
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.674506346462164e-08,
+      "loss": 0.7012,
+      "step": 26990
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.65938288312218e-08,
+      "loss": 0.7158,
+      "step": 26991
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.644272557368216e-08,
+      "loss": 0.6875,
+      "step": 26992
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.629175369400556e-08,
+      "loss": 0.7129,
+      "step": 26993
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.61409131941915e-08,
+      "loss": 0.6953,
+      "step": 26994
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.59902040762417e-08,
+      "loss": 0.6826,
+      "step": 26995
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.583962634215126e-08,
+      "loss": 0.6807,
+      "step": 26996
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.568917999391745e-08,
+      "loss": 0.7305,
+      "step": 26997
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.553886503353203e-08,
+      "loss": 0.7197,
+      "step": 26998
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.538868146298784e-08,
+      "loss": 0.7139,
+      "step": 26999
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.523862928427551e-08,
+      "loss": 0.7012,
+      "step": 27000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.508870849938344e-08,
+      "loss": 0.708,
+      "step": 27001
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.493891911029784e-08,
+      "loss": 0.6953,
+      "step": 27002
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.478926111900487e-08,
+      "loss": 0.6953,
+      "step": 27003
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.463973452748742e-08,
+      "loss": 0.708,
+      "step": 27004
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.449033933772721e-08,
+      "loss": 0.7012,
+      "step": 27005
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.434107555170156e-08,
+      "loss": 0.6963,
+      "step": 27006
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.419194317139334e-08,
+      "loss": 0.6914,
+      "step": 27007
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.40429421987754e-08,
+      "loss": 0.7139,
+      "step": 27008
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.389407263582283e-08,
+      "loss": 0.7061,
+      "step": 27009
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.374533448450962e-08,
+      "loss": 0.6895,
+      "step": 27010
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.359672774680638e-08,
+      "loss": 0.6904,
+      "step": 27011
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.344825242468158e-08,
+      "loss": 0.7178,
+      "step": 27012
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.329990852010361e-08,
+      "loss": 0.6934,
+      "step": 27013
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.31516960350398e-08,
+      "loss": 0.6963,
+      "step": 27014
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.300361497145082e-08,
+      "loss": 0.6855,
+      "step": 27015
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.285566533130284e-08,
+      "loss": 0.6768,
+      "step": 27016
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.270784711655433e-08,
+      "loss": 0.6953,
+      "step": 27017
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.25601603291648e-08,
+      "loss": 0.7461,
+      "step": 27018
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.241260497109161e-08,
+      "loss": 0.7002,
+      "step": 27019
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.226518104428981e-08,
+      "loss": 0.6895,
+      "step": 27020
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.211788855071345e-08,
+      "loss": 0.708,
+      "step": 27021
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.197072749231428e-08,
+      "loss": 0.6895,
+      "step": 27022
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.182369787104293e-08,
+      "loss": 0.7148,
+      "step": 27023
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.167679968884679e-08,
+      "loss": 0.6992,
+      "step": 27024
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.153003294767425e-08,
+      "loss": 0.7324,
+      "step": 27025
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.138339764946935e-08,
+      "loss": 0.6943,
+      "step": 27026
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.123689379617494e-08,
+      "loss": 0.7109,
+      "step": 27027
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.109052138973395e-08,
+      "loss": 0.7031,
+      "step": 27028
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.09442804320848e-08,
+      "loss": 0.7168,
+      "step": 27029
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.079817092516707e-08,
+      "loss": 0.7051,
+      "step": 27030
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.065219287091363e-08,
+      "loss": 0.6729,
+      "step": 27031
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.050634627126297e-08,
+      "loss": 0.6943,
+      "step": 27032
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.036063112814685e-08,
+      "loss": 0.709,
+      "step": 27033
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.021504744349484e-08,
+      "loss": 0.7314,
+      "step": 27034
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.006959521923652e-08,
+      "loss": 0.6982,
+      "step": 27035
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.992427445730144e-08,
+      "loss": 0.7129,
+      "step": 27036
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.977908515961475e-08,
+      "loss": 0.708,
+      "step": 27037
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.963402732809933e-08,
+      "loss": 0.7002,
+      "step": 27038
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.948910096467699e-08,
+      "loss": 0.6895,
+      "step": 27039
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.934430607127063e-08,
+      "loss": 0.709,
+      "step": 27040
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.919964264979763e-08,
+      "loss": 0.6982,
+      "step": 27041
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.905511070217642e-08,
+      "loss": 0.7061,
+      "step": 27042
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.891071023032105e-08,
+      "loss": 0.7188,
+      "step": 27043
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.876644123614552e-08,
+      "loss": 0.7305,
+      "step": 27044
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.862230372156165e-08,
+      "loss": 0.7236,
+      "step": 27045
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.847829768848015e-08,
+      "loss": 0.7021,
+      "step": 27046
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.833442313880835e-08,
+      "loss": 0.7021,
+      "step": 27047
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.819068007445363e-08,
+      "loss": 0.6797,
+      "step": 27048
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.804706849732225e-08,
+      "loss": 0.7168,
+      "step": 27049
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.790358840931489e-08,
+      "loss": 0.6982,
+      "step": 27050
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.776023981233449e-08,
+      "loss": 0.7246,
+      "step": 27051
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.761702270828064e-08,
+      "loss": 0.7031,
+      "step": 27052
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.747393709905072e-08,
+      "loss": 0.7188,
+      "step": 27053
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.733098298654207e-08,
+      "loss": 0.6885,
+      "step": 27054
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.718816037264765e-08,
+      "loss": 0.6855,
+      "step": 27055
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.704546925926148e-08,
+      "loss": 0.7061,
+      "step": 27056
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.690290964827318e-08,
+      "loss": 0.7061,
+      "step": 27057
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.676048154157455e-08,
+      "loss": 0.71,
+      "step": 27058
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.661818494104967e-08,
+      "loss": 0.7207,
+      "step": 27059
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.6476019848587e-08,
+      "loss": 0.749,
+      "step": 27060
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.633398626606948e-08,
+      "loss": 0.6953,
+      "step": 27061
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.619208419538005e-08,
+      "loss": 0.7031,
+      "step": 27062
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.605031363839721e-08,
+      "loss": 0.6924,
+      "step": 27063
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.590867459700169e-08,
+      "loss": 0.6855,
+      "step": 27064
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.576716707307086e-08,
+      "loss": 0.7246,
+      "step": 27065
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.562579106847878e-08,
+      "loss": 0.6836,
+      "step": 27066
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.548454658510063e-08,
+      "loss": 0.6924,
+      "step": 27067
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.534343362480489e-08,
+      "loss": 0.7031,
+      "step": 27068
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.520245218946565e-08,
+      "loss": 0.7227,
+      "step": 27069
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.506160228094917e-08,
+      "loss": 0.6914,
+      "step": 27070
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.492088390112284e-08,
+      "loss": 0.707,
+      "step": 27071
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.478029705184964e-08,
+      "loss": 0.7158,
+      "step": 27072
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.463984173499473e-08,
+      "loss": 0.6973,
+      "step": 27073
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.449951795241883e-08,
+      "loss": 0.71,
+      "step": 27074
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.435932570598159e-08,
+      "loss": 0.6953,
+      "step": 27075
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.42192649975404e-08,
+      "loss": 0.7012,
+      "step": 27076
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.407933582895154e-08,
+      "loss": 0.7217,
+      "step": 27077
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.393953820207022e-08,
+      "loss": 0.7002,
+      "step": 27078
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.379987211874829e-08,
+      "loss": 0.707,
+      "step": 27079
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.366033758083646e-08,
+      "loss": 0.6895,
+      "step": 27080
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.352093459018439e-08,
+      "loss": 0.7139,
+      "step": 27081
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.33816631486406e-08,
+      "loss": 0.6846,
+      "step": 27082
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.324252325804804e-08,
+      "loss": 0.7539,
+      "step": 27083
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.310351492025303e-08,
+      "loss": 0.7031,
+      "step": 27084
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.296463813709631e-08,
+      "loss": 0.6914,
+      "step": 27085
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.282589291041974e-08,
+      "loss": 0.7061,
+      "step": 27086
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.268727924206076e-08,
+      "loss": 0.7012,
+      "step": 27087
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.254879713385676e-08,
+      "loss": 0.6992,
+      "step": 27088
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.241044658764295e-08,
+      "loss": 0.6777,
+      "step": 27089
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.227222760525232e-08,
+      "loss": 0.7129,
+      "step": 27090
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.213414018851783e-08,
+      "loss": 0.7051,
+      "step": 27091
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.199618433926913e-08,
+      "loss": 0.7158,
+      "step": 27092
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.185836005933256e-08,
+      "loss": 0.6934,
+      "step": 27093
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.172066735053662e-08,
+      "loss": 0.7061,
+      "step": 27094
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.158310621470655e-08,
+      "loss": 0.6992,
+      "step": 27095
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.144567665366309e-08,
+      "loss": 0.7129,
+      "step": 27096
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.130837866922924e-08,
+      "loss": 0.6943,
+      "step": 27097
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.117121226322355e-08,
+      "loss": 0.6885,
+      "step": 27098
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.103417743746454e-08,
+      "loss": 0.7061,
+      "step": 27099
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.089727419376857e-08,
+      "loss": 0.7188,
+      "step": 27100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.076050253394862e-08,
+      "loss": 0.7178,
+      "step": 27101
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.062386245981878e-08,
+      "loss": 0.6992,
+      "step": 27102
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.048735397318874e-08,
+      "loss": 0.7012,
+      "step": 27103
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.035097707586702e-08,
+      "loss": 0.7002,
+      "step": 27104
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.02147317696622e-08,
+      "loss": 0.6973,
+      "step": 27105
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.00786180563795e-08,
+      "loss": 0.6953,
+      "step": 27106
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.994263593782191e-08,
+      "loss": 0.6865,
+      "step": 27107
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.980678541579356e-08,
+      "loss": 0.7178,
+      "step": 27108
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.967106649209299e-08,
+      "loss": 0.6973,
+      "step": 27109
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.953547916851878e-08,
+      "loss": 0.6865,
+      "step": 27110
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.940002344686947e-08,
+      "loss": 0.7178,
+      "step": 27111
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.926469932893808e-08,
+      "loss": 0.6875,
+      "step": 27112
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.912950681651764e-08,
+      "loss": 0.7051,
+      "step": 27113
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.899444591140336e-08,
+      "loss": 0.6816,
+      "step": 27114
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.88595166153816e-08,
+      "loss": 0.7041,
+      "step": 27115
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.872471893024202e-08,
+      "loss": 0.6807,
+      "step": 27116
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.859005285776987e-08,
+      "loss": 0.6992,
+      "step": 27117
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.845551839975151e-08,
+      "loss": 0.7334,
+      "step": 27118
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.832111555796772e-08,
+      "loss": 0.7207,
+      "step": 27119
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.818684433420152e-08,
+      "loss": 0.6963,
+      "step": 27120
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.805270473023152e-08,
+      "loss": 0.6982,
+      "step": 27121
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.791869674783624e-08,
+      "loss": 0.6924,
+      "step": 27122
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.778482038878986e-08,
+      "loss": 0.7129,
+      "step": 27123
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.765107565486761e-08,
+      "loss": 0.7158,
+      "step": 27124
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.751746254784142e-08,
+      "loss": 0.6885,
+      "step": 27125
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.738398106948318e-08,
+      "loss": 0.6982,
+      "step": 27126
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.72506312215615e-08,
+      "loss": 0.7139,
+      "step": 27127
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.711741300584162e-08,
+      "loss": 0.7168,
+      "step": 27128
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.698432642409103e-08,
+      "loss": 0.6963,
+      "step": 27129
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.685137147807386e-08,
+      "loss": 0.6807,
+      "step": 27130
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.671854816954981e-08,
+      "loss": 0.7227,
+      "step": 27131
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.658585650028082e-08,
+      "loss": 0.6953,
+      "step": 27132
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.645329647202548e-08,
+      "loss": 0.709,
+      "step": 27133
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.632086808654015e-08,
+      "loss": 0.7207,
+      "step": 27134
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.618857134557899e-08,
+      "loss": 0.7217,
+      "step": 27135
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.605640625089616e-08,
+      "loss": 0.6846,
+      "step": 27136
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.592437280424246e-08,
+      "loss": 0.7139,
+      "step": 27137
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.579247100736874e-08,
+      "loss": 0.71,
+      "step": 27138
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.566070086202136e-08,
+      "loss": 0.6787,
+      "step": 27139
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.552906236994894e-08,
+      "loss": 0.6973,
+      "step": 27140
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.53975555328934e-08,
+      "loss": 0.7051,
+      "step": 27141
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.526618035260002e-08,
+      "loss": 0.7061,
+      "step": 27142
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.513493683080741e-08,
+      "loss": 0.709,
+      "step": 27143
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.500382496925639e-08,
+      "loss": 0.6738,
+      "step": 27144
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.487284476968447e-08,
+      "loss": 0.7061,
+      "step": 27145
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.474199623382694e-08,
+      "loss": 0.6973,
+      "step": 27146
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.461127936341794e-08,
+      "loss": 0.7148,
+      "step": 27147
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.448069416018942e-08,
+      "loss": 0.6914,
+      "step": 27148
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.435024062587336e-08,
+      "loss": 0.6758,
+      "step": 27149
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.421991876219613e-08,
+      "loss": 0.7109,
+      "step": 27150
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.408972857088747e-08,
+      "loss": 0.6963,
+      "step": 27151
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.395967005367044e-08,
+      "loss": 0.7021,
+      "step": 27152
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.382974321227031e-08,
+      "loss": 0.6816,
+      "step": 27153
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.369994804840906e-08,
+      "loss": 0.6934,
+      "step": 27154
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.357028456380532e-08,
+      "loss": 0.7061,
+      "step": 27155
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.344075276017769e-08,
+      "loss": 0.6855,
+      "step": 27156
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.33113526392437e-08,
+      "loss": 0.6963,
+      "step": 27157
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.318208420271643e-08,
+      "loss": 0.707,
+      "step": 27158
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.305294745231228e-08,
+      "loss": 0.6895,
+      "step": 27159
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.292394238973876e-08,
+      "loss": 0.6875,
+      "step": 27160
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.279506901670784e-08,
+      "loss": 0.6973,
+      "step": 27161
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.266632733492705e-08,
+      "loss": 0.7217,
+      "step": 27162
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.253771734610282e-08,
+      "loss": 0.7041,
+      "step": 27163
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.240923905193819e-08,
+      "loss": 0.6895,
+      "step": 27164
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.228089245413738e-08,
+      "loss": 0.7002,
+      "step": 27165
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.215267755440013e-08,
+      "loss": 0.707,
+      "step": 27166
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.20245943544262e-08,
+      "loss": 0.6943,
+      "step": 27167
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.189664285591201e-08,
+      "loss": 0.6846,
+      "step": 27168
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.17688230605551e-08,
+      "loss": 0.7334,
+      "step": 27169
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.164113497004742e-08,
+      "loss": 0.6846,
+      "step": 27170
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.151357858608209e-08,
+      "loss": 0.7236,
+      "step": 27171
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.138615391035107e-08,
+      "loss": 0.7021,
+      "step": 27172
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.12588609445397e-08,
+      "loss": 0.6777,
+      "step": 27173
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.11316996903366e-08,
+      "loss": 0.7227,
+      "step": 27174
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.100467014942712e-08,
+      "loss": 0.6855,
+      "step": 27175
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.087777232349545e-08,
+      "loss": 0.7217,
+      "step": 27176
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.075100621422136e-08,
+      "loss": 0.6787,
+      "step": 27177
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.062437182328684e-08,
+      "loss": 0.7041,
+      "step": 27178
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.049786915236833e-08,
+      "loss": 0.7002,
+      "step": 27179
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.037149820314337e-08,
+      "loss": 0.6865,
+      "step": 27180
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.024525897728617e-08,
+      "loss": 0.6865,
+      "step": 27181
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.011915147646985e-08,
+      "loss": 0.6973,
+      "step": 27182
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.999317570236529e-08,
+      "loss": 0.6709,
+      "step": 27183
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.986733165664338e-08,
+      "loss": 0.7314,
+      "step": 27184
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.974161934097056e-08,
+      "loss": 0.7031,
+      "step": 27185
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.961603875701217e-08,
+      "loss": 0.6885,
+      "step": 27186
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.9490589906433526e-08,
+      "loss": 0.7012,
+      "step": 27187
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.936527279089777e-08,
+      "loss": 0.6719,
+      "step": 27188
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.9240087412063554e-08,
+      "loss": 0.7109,
+      "step": 27189
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.911503377159178e-08,
+      "loss": 0.6641,
+      "step": 27190
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.89901118711389e-08,
+      "loss": 0.71,
+      "step": 27191
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.886532171236137e-08,
+      "loss": 0.708,
+      "step": 27192
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.874066329691119e-08,
+      "loss": 0.6914,
+      "step": 27193
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.86161366264415e-08,
+      "loss": 0.6963,
+      "step": 27194
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8491741702602076e-08,
+      "loss": 0.6973,
+      "step": 27195
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.836747852704161e-08,
+      "loss": 0.7041,
+      "step": 27196
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.824334710140767e-08,
+      "loss": 0.6738,
+      "step": 27197
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.811934742734449e-08,
+      "loss": 0.6992,
+      "step": 27198
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.799547950649631e-08,
+      "loss": 0.6875,
+      "step": 27199
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.787174334050294e-08,
+      "loss": 0.6953,
+      "step": 27200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7748138931005285e-08,
+      "loss": 0.707,
+      "step": 27201
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.762466627964092e-08,
+      "loss": 0.6758,
+      "step": 27202
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7501325388046315e-08,
+      "loss": 0.7158,
+      "step": 27203
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.737811625785683e-08,
+      "loss": 0.6846,
+      "step": 27204
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7255038890704495e-08,
+      "loss": 0.7002,
+      "step": 27205
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7132093288221335e-08,
+      "loss": 0.6826,
+      "step": 27206
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.700927945203383e-08,
+      "loss": 0.7051,
+      "step": 27207
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.688659738377289e-08,
+      "loss": 0.6934,
+      "step": 27208
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.676404708506278e-08,
+      "loss": 0.7266,
+      "step": 27209
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.664162855752775e-08,
+      "loss": 0.7129,
+      "step": 27210
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.651934180278984e-08,
+      "loss": 0.6885,
+      "step": 27211
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6397186822469976e-08,
+      "loss": 0.6943,
+      "step": 27212
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6275163618186855e-08,
+      "loss": 0.6953,
+      "step": 27213
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.615327219155808e-08,
+      "loss": 0.6855,
+      "step": 27214
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.603151254419792e-08,
+      "loss": 0.6924,
+      "step": 27215
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.590988467772174e-08,
+      "loss": 0.7021,
+      "step": 27216
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.578838859373936e-08,
+      "loss": 0.71,
+      "step": 27217
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5667024293862834e-08,
+      "loss": 0.7197,
+      "step": 27218
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5545791779698655e-08,
+      "loss": 0.6777,
+      "step": 27219
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.542469105285553e-08,
+      "loss": 0.6943,
+      "step": 27220
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.530372211493662e-08,
+      "loss": 0.707,
+      "step": 27221
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.518288496754509e-08,
+      "loss": 0.7109,
+      "step": 27222
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5062179612282995e-08,
+      "loss": 0.7021,
+      "step": 27223
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.494160605074905e-08,
+      "loss": 0.7012,
+      "step": 27224
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4821164284543094e-08,
+      "loss": 0.7246,
+      "step": 27225
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.47008543152594e-08,
+      "loss": 0.7119,
+      "step": 27226
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4580676144492254e-08,
+      "loss": 0.6963,
+      "step": 27227
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4460629773835925e-08,
+      "loss": 0.7314,
+      "step": 27228
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4340715204880265e-08,
+      "loss": 0.6846,
+      "step": 27229
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4220932439214004e-08,
+      "loss": 0.7119,
+      "step": 27230
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.410128147842586e-08,
+      "loss": 0.708,
+      "step": 27231
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.398176232410013e-08,
+      "loss": 0.6826,
+      "step": 27232
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.38623749778211e-08,
+      "loss": 0.708,
+      "step": 27233
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3743119441171944e-08,
+      "loss": 0.7246,
+      "step": 27234
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3623995715731405e-08,
+      "loss": 0.6855,
+      "step": 27235
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.350500380307933e-08,
+      "loss": 0.7256,
+      "step": 27236
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.338614370479333e-08,
+      "loss": 0.7012,
+      "step": 27237
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3267415422446624e-08,
+      "loss": 0.7266,
+      "step": 27238
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.314881895761348e-08,
+      "loss": 0.7139,
+      "step": 27239
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.303035431186598e-08,
+      "loss": 0.6992,
+      "step": 27240
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.291202148677288e-08,
+      "loss": 0.709,
+      "step": 27241
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2793820483904025e-08,
+      "loss": 0.6865,
+      "step": 27242
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.267575130482594e-08,
+      "loss": 0.6787,
+      "step": 27243
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.255781395110182e-08,
+      "loss": 0.6826,
+      "step": 27244
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2440008424294865e-08,
+      "loss": 0.7139,
+      "step": 27245
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.232233472596715e-08,
+      "loss": 0.6895,
+      "step": 27246
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2204792857677435e-08,
+      "loss": 0.6914,
+      "step": 27247
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.208738282098225e-08,
+      "loss": 0.7021,
+      "step": 27248
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.197010461744034e-08,
+      "loss": 0.71,
+      "step": 27249
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1852958248604925e-08,
+      "loss": 0.6875,
+      "step": 27250
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.173594371602808e-08,
+      "loss": 0.7197,
+      "step": 27251
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1619061021259683e-08,
+      "loss": 0.71,
+      "step": 27252
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1502310165849613e-08,
+      "loss": 0.7148,
+      "step": 27253
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.138569115134551e-08,
+      "loss": 0.6816,
+      "step": 27254
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.126920397929169e-08,
+      "loss": 0.7188,
+      "step": 27255
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1152848651232486e-08,
+      "loss": 0.7139,
+      "step": 27256
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.103662516871111e-08,
+      "loss": 0.7197,
+      "step": 27257
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.0920533533265206e-08,
+      "loss": 0.7188,
+      "step": 27258
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.080457374643577e-08,
+      "loss": 0.6885,
+      "step": 27259
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.068874580975713e-08,
+      "loss": 0.6855,
+      "step": 27260
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.0573049724765844e-08,
+      "loss": 0.6895,
+      "step": 27261
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.045748549299623e-08,
+      "loss": 0.7148,
+      "step": 27262
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.0342053115978175e-08,
+      "loss": 0.7031,
+      "step": 27263
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.0226752595240457e-08,
+      "loss": 0.7334,
+      "step": 27264
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.011158393231297e-08,
+      "loss": 0.708,
+      "step": 27265
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9996547128721154e-08,
+      "loss": 0.7148,
+      "step": 27266
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9881642185989345e-08,
+      "loss": 0.6924,
+      "step": 27267
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.976686910564188e-08,
+      "loss": 0.7051,
+      "step": 27268
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.965222788919755e-08,
+      "loss": 0.7266,
+      "step": 27269
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.953771853817735e-08,
+      "loss": 0.7051,
+      "step": 27270
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.942334105409674e-08,
+      "loss": 0.7041,
+      "step": 27271
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.930909543847451e-08,
+      "loss": 0.6797,
+      "step": 27272
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.919498169282166e-08,
+      "loss": 0.7148,
+      "step": 27273
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9080999818651445e-08,
+      "loss": 0.7012,
+      "step": 27274
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8967149817475966e-08,
+      "loss": 0.6865,
+      "step": 27275
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.885343169080181e-08,
+      "loss": 0.6816,
+      "step": 27276
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8739845440137767e-08,
+      "loss": 0.6855,
+      "step": 27277
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8626391066988187e-08,
+      "loss": 0.7236,
+      "step": 27278
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.851306857285632e-08,
+      "loss": 0.7148,
+      "step": 27279
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.839987795924428e-08,
+      "loss": 0.7363,
+      "step": 27280
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8286819227651996e-08,
+      "loss": 0.6816,
+      "step": 27281
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.817389237957826e-08,
+      "loss": 0.707,
+      "step": 27282
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.8061097416519654e-08,
+      "loss": 0.708,
+      "step": 27283
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7948434339970537e-08,
+      "loss": 0.6914,
+      "step": 27284
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.783590315142417e-08,
+      "loss": 0.708,
+      "step": 27285
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7723503852371566e-08,
+      "loss": 0.7002,
+      "step": 27286
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.761123644430155e-08,
+      "loss": 0.7041,
+      "step": 27287
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.749910092870402e-08,
+      "loss": 0.6709,
+      "step": 27288
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.738709730706337e-08,
+      "loss": 0.7002,
+      "step": 27289
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.727522558086506e-08,
+      "loss": 0.6953,
+      "step": 27290
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.716348575159013e-08,
+      "loss": 0.71,
+      "step": 27291
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7051877820720736e-08,
+      "loss": 0.7109,
+      "step": 27292
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.694040178973458e-08,
+      "loss": 0.7188,
+      "step": 27293
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6829057660110474e-08,
+      "loss": 0.6914,
+      "step": 27294
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.671784543332392e-08,
+      "loss": 0.7178,
+      "step": 27295
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.660676511084816e-08,
+      "loss": 0.6973,
+      "step": 27296
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.649581669415426e-08,
+      "loss": 0.7012,
+      "step": 27297
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.638500018471437e-08,
+      "loss": 0.7051,
+      "step": 27298
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.62743155839962e-08,
+      "loss": 0.6934,
+      "step": 27299
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.616376289346636e-08,
+      "loss": 0.7158,
+      "step": 27300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6053342114590335e-08,
+      "loss": 0.6924,
+      "step": 27301
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5943053248831417e-08,
+      "loss": 0.7217,
+      "step": 27302
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.583289629765175e-08,
+      "loss": 0.7051,
+      "step": 27303
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.572287126250907e-08,
+      "loss": 0.6953,
+      "step": 27304
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.561297814486443e-08,
+      "loss": 0.6934,
+      "step": 27305
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.55032169461711e-08,
+      "loss": 0.6787,
+      "step": 27306
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5393587667886816e-08,
+      "loss": 0.7148,
+      "step": 27307
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5284090311462637e-08,
+      "loss": 0.7139,
+      "step": 27308
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.517472487834962e-08,
+      "loss": 0.7236,
+      "step": 27309
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.5065491369996605e-08,
+      "loss": 0.6982,
+      "step": 27310
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.495638978785355e-08,
+      "loss": 0.6973,
+      "step": 27311
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.484742013336374e-08,
+      "loss": 0.708,
+      "step": 27312
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4738582407972685e-08,
+      "loss": 0.7148,
+      "step": 27313
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.462987661312146e-08,
+      "loss": 0.7188,
+      "step": 27314
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4521302750252236e-08,
+      "loss": 0.7246,
+      "step": 27315
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.441286082080276e-08,
+      "loss": 0.7119,
+      "step": 27316
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.430455082621077e-08,
+      "loss": 0.7217,
+      "step": 27317
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.419637276791178e-08,
+      "loss": 0.7432,
+      "step": 27318
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.408832664733909e-08,
+      "loss": 0.709,
+      "step": 27319
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3980412465923773e-08,
+      "loss": 0.709,
+      "step": 27320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3872630225096914e-08,
+      "loss": 0.7129,
+      "step": 27321
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3764979926287363e-08,
+      "loss": 0.7139,
+      "step": 27322
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.365746157092066e-08,
+      "loss": 0.7002,
+      "step": 27323
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.35500751604212e-08,
+      "loss": 0.7188,
+      "step": 27324
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.344282069621342e-08,
+      "loss": 0.7002,
+      "step": 27325
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.333569817971839e-08,
+      "loss": 0.6973,
+      "step": 27326
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.322870761235609e-08,
+      "loss": 0.6953,
+      "step": 27327
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.312184899554317e-08,
+      "loss": 0.6914,
+      "step": 27328
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.301512233069627e-08,
+      "loss": 0.6943,
+      "step": 27329
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.290852761922981e-08,
+      "loss": 0.7129,
+      "step": 27330
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.280206486255711e-08,
+      "loss": 0.6963,
+      "step": 27331
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2695734062089264e-08,
+      "loss": 0.7178,
+      "step": 27332
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.258953521923404e-08,
+      "loss": 0.7041,
+      "step": 27333
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.248346833539918e-08,
+      "loss": 0.6992,
+      "step": 27334
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2377533411990245e-08,
+      "loss": 0.7109,
+      "step": 27335
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2271730450411665e-08,
+      "loss": 0.7178,
+      "step": 27336
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2166059452065644e-08,
+      "loss": 0.71,
+      "step": 27337
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.206052041835107e-08,
+      "loss": 0.6953,
+      "step": 27338
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.195511335066904e-08,
+      "loss": 0.6953,
+      "step": 27339
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.184983825041511e-08,
+      "loss": 0.7188,
+      "step": 27340
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.174469511898482e-08,
+      "loss": 0.6973,
+      "step": 27341
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1639683957770404e-08,
+      "loss": 0.6895,
+      "step": 27342
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1534804768165185e-08,
+      "loss": 0.7158,
+      "step": 27343
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1430057551556935e-08,
+      "loss": 0.6963,
+      "step": 27344
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.132544230933677e-08,
+      "loss": 0.7061,
+      "step": 27345
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.122095904288803e-08,
+      "loss": 0.6807,
+      "step": 27346
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.111660775359738e-08,
+      "loss": 0.7246,
+      "step": 27347
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.101238844284705e-08,
+      "loss": 0.7148,
+      "step": 27348
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.090830111201815e-08,
+      "loss": 0.6934,
+      "step": 27349
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0804345762489596e-08,
+      "loss": 0.7246,
+      "step": 27350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0700522395640263e-08,
+      "loss": 0.6748,
+      "step": 27351
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.059683101284462e-08,
+      "loss": 0.6982,
+      "step": 27352
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.049327161547822e-08,
+      "loss": 0.6855,
+      "step": 27353
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.038984420491332e-08,
+      "loss": 0.7021,
+      "step": 27354
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.028654878251992e-08,
+      "loss": 0.6943,
+      "step": 27355
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0183385349666924e-08,
+      "loss": 0.6963,
+      "step": 27356
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.008035390772103e-08,
+      "loss": 0.7236,
+      "step": 27357
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.99774544580489e-08,
+      "loss": 0.7002,
+      "step": 27358
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.98746870020128e-08,
+      "loss": 0.7334,
+      "step": 27359
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9772051540976074e-08,
+      "loss": 0.71,
+      "step": 27360
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.966954807629875e-08,
+      "loss": 0.7188,
+      "step": 27361
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.956717660933751e-08,
+      "loss": 0.7051,
+      "step": 27362
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.946493714145017e-08,
+      "loss": 0.7041,
+      "step": 27363
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.93628296739923e-08,
+      "loss": 0.6934,
+      "step": 27364
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.926085420831727e-08,
+      "loss": 0.6895,
+      "step": 27365
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9159010745775105e-08,
+      "loss": 0.7109,
+      "step": 27366
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9057299287714735e-08,
+      "loss": 0.6904,
+      "step": 27367
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8955719835487294e-08,
+      "loss": 0.71,
+      "step": 27368
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.885427239043616e-08,
+      "loss": 0.7041,
+      "step": 27369
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8752956953905796e-08,
+      "loss": 0.709,
+      "step": 27370
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.86517735272407e-08,
+      "loss": 0.6992,
+      "step": 27371
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.85507221117809e-08,
+      "loss": 0.71,
+      "step": 27372
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8449802708865337e-08,
+      "loss": 0.6895,
+      "step": 27373
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8349015319830704e-08,
+      "loss": 0.6982,
+      "step": 27374
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.824835994601372e-08,
+      "loss": 0.7324,
+      "step": 27375
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.814783658874888e-08,
+      "loss": 0.71,
+      "step": 27376
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.804744524936732e-08,
+      "loss": 0.6963,
+      "step": 27377
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7947185929199105e-08,
+      "loss": 0.6836,
+      "step": 27378
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.784705862957427e-08,
+      "loss": 0.6963,
+      "step": 27379
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7747063351818434e-08,
+      "loss": 0.709,
+      "step": 27380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7647200097258305e-08,
+      "loss": 0.7148,
+      "step": 27381
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.754746886721505e-08,
+      "loss": 0.6953,
+      "step": 27382
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.744786966301206e-08,
+      "loss": 0.707,
+      "step": 27383
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.734840248596827e-08,
+      "loss": 0.7012,
+      "step": 27384
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7249067337402634e-08,
+      "loss": 0.6816,
+      "step": 27385
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.714986421863187e-08,
+      "loss": 0.7012,
+      "step": 27386
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7050793130969375e-08,
+      "loss": 0.7324,
+      "step": 27387
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6951854075729654e-08,
+      "loss": 0.7021,
+      "step": 27388
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6853047054222765e-08,
+      "loss": 0.6709,
+      "step": 27389
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.675437206775767e-08,
+      "loss": 0.7158,
+      "step": 27390
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.665582911764332e-08,
+      "loss": 0.7051,
+      "step": 27391
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.655741820518532e-08,
+      "loss": 0.7021,
+      "step": 27392
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.64591393316871e-08,
+      "loss": 0.7373,
+      "step": 27393
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6360992498452044e-08,
+      "loss": 0.6807,
+      "step": 27394
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.626297770678022e-08,
+      "loss": 0.7109,
+      "step": 27395
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.61650949579706e-08,
+      "loss": 0.6641,
+      "step": 27396
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.606734425332104e-08,
+      "loss": 0.6963,
+      "step": 27397
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5969725594126034e-08,
+      "loss": 0.6826,
+      "step": 27398
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.587223898167902e-08,
+      "loss": 0.7324,
+      "step": 27399
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5774884417273395e-08,
+      "loss": 0.7061,
+      "step": 27400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.567766190219812e-08,
+      "loss": 0.6963,
+      "step": 27401
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5580571437741073e-08,
+      "loss": 0.6836,
+      "step": 27402
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.548361302519121e-08,
+      "loss": 0.6992,
+      "step": 27403
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5386786665830844e-08,
+      "loss": 0.7002,
+      "step": 27404
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5290092360945604e-08,
+      "loss": 0.709,
+      "step": 27405
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.519353011181559e-08,
+      "loss": 0.6768,
+      "step": 27406
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.509709991971977e-08,
+      "loss": 0.6787,
+      "step": 27407
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.500080178593712e-08,
+      "loss": 0.7246,
+      "step": 27408
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4904635711743296e-08,
+      "loss": 0.6953,
+      "step": 27409
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4808601698413935e-08,
+      "loss": 0.7031,
+      "step": 27410
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.471269974722025e-08,
+      "loss": 0.71,
+      "step": 27411
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4616929859433436e-08,
+      "loss": 0.7275,
+      "step": 27412
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4521292036323594e-08,
+      "loss": 0.6914,
+      "step": 27413
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4425786279157493e-08,
+      "loss": 0.7139,
+      "step": 27414
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.433041258919967e-08,
+      "loss": 0.7031,
+      "step": 27415
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.423517096771578e-08,
+      "loss": 0.7012,
+      "step": 27416
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.414006141596815e-08,
+      "loss": 0.7197,
+      "step": 27417
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4045083935215775e-08,
+      "loss": 0.7324,
+      "step": 27418
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.395023852671875e-08,
+      "loss": 0.7012,
+      "step": 27419
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.385552519173163e-08,
+      "loss": 0.6992,
+      "step": 27420
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.37609439315123e-08,
+      "loss": 0.6895,
+      "step": 27421
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.366649474731309e-08,
+      "loss": 0.7109,
+      "step": 27422
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.357217764038523e-08,
+      "loss": 0.6914,
+      "step": 27423
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.347799261197993e-08,
+      "loss": 0.7158,
+      "step": 27424
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3383939663343966e-08,
+      "loss": 0.7051,
+      "step": 27425
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.329001879572413e-08,
+      "loss": 0.7197,
+      "step": 27426
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.319623001036609e-08,
+      "loss": 0.7305,
+      "step": 27427
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.310257330851108e-08,
+      "loss": 0.6953,
+      "step": 27428
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3009048691401426e-08,
+      "loss": 0.7021,
+      "step": 27429
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.291565616027614e-08,
+      "loss": 0.6904,
+      "step": 27430
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.282239571637314e-08,
+      "loss": 0.7012,
+      "step": 27431
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2729267360928075e-08,
+      "loss": 0.6992,
+      "step": 27432
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.263627109517553e-08,
+      "loss": 0.6953,
+      "step": 27433
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.254340692034785e-08,
+      "loss": 0.6846,
+      "step": 27434
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2450674837675165e-08,
+      "loss": 0.7139,
+      "step": 27435
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.235807484838649e-08,
+      "loss": 0.7012,
+      "step": 27436
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2265606953709725e-08,
+      "loss": 0.709,
+      "step": 27437
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.217327115487057e-08,
+      "loss": 0.707,
+      "step": 27438
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.208106745309025e-08,
+      "loss": 0.7168,
+      "step": 27439
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.198899584959336e-08,
+      "loss": 0.7217,
+      "step": 27440
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.189705634559892e-08,
+      "loss": 0.6836,
+      "step": 27441
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.180524894232484e-08,
+      "loss": 0.7012,
+      "step": 27442
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1713573640989034e-08,
+      "loss": 0.7178,
+      "step": 27443
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1622030442806094e-08,
+      "loss": 0.7119,
+      "step": 27444
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.153061934898838e-08,
+      "loss": 0.6904,
+      "step": 27445
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.143934036074825e-08,
+      "loss": 0.7002,
+      "step": 27446
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.134819347929474e-08,
+      "loss": 0.6846,
+      "step": 27447
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1257178705835775e-08,
+      "loss": 0.6816,
+      "step": 27448
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.116629604157817e-08,
+      "loss": 0.6973,
+      "step": 27449
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.10755454877254e-08,
+      "loss": 0.6924,
+      "step": 27450
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0984927045480953e-08,
+      "loss": 0.6904,
+      "step": 27451
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.089444071604497e-08,
+      "loss": 0.7129,
+      "step": 27452
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.080408650061761e-08,
+      "loss": 0.6982,
+      "step": 27453
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.071386440039459e-08,
+      "loss": 0.6777,
+      "step": 27454
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0623774416573825e-08,
+      "loss": 0.7061,
+      "step": 27455
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0533816550346596e-08,
+      "loss": 0.7129,
+      "step": 27456
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0443990802907495e-08,
+      "loss": 0.6973,
+      "step": 27457
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.035429717544669e-08,
+      "loss": 0.7178,
+      "step": 27458
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0264735669151004e-08,
+      "loss": 0.7031,
+      "step": 27459
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.017530628520837e-08,
+      "loss": 0.709,
+      "step": 27460
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.008600902480341e-08,
+      "loss": 0.7344,
+      "step": 27461
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.999684388912072e-08,
+      "loss": 0.7188,
+      "step": 27462
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.990781087934158e-08,
+      "loss": 0.7031,
+      "step": 27463
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.981890999664616e-08,
+      "loss": 0.6777,
+      "step": 27464
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9730141242211296e-08,
+      "loss": 0.7021,
+      "step": 27465
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9641504617213823e-08,
+      "loss": 0.707,
+      "step": 27466
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.955300012282947e-08,
+      "loss": 0.6846,
+      "step": 27467
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9464627760229513e-08,
+      "loss": 0.6924,
+      "step": 27468
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.937638753058636e-08,
+      "loss": 0.6738,
+      "step": 27469
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9288279435069067e-08,
+      "loss": 0.7158,
+      "step": 27470
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9200303474845592e-08,
+      "loss": 0.7197,
+      "step": 27471
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.911245965108056e-08,
+      "loss": 0.7207,
+      "step": 27472
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9024747964938592e-08,
+      "loss": 0.7119,
+      "step": 27473
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8937168417583207e-08,
+      "loss": 0.6982,
+      "step": 27474
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.884972101017458e-08,
+      "loss": 0.6846,
+      "step": 27475
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.876240574387068e-08,
+      "loss": 0.7188,
+      "step": 27476
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.867522261983058e-08,
+      "loss": 0.6816,
+      "step": 27477
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8588171639206686e-08,
+      "loss": 0.7227,
+      "step": 27478
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8501252803154744e-08,
+      "loss": 0.7002,
+      "step": 27479
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8414466112827166e-08,
+      "loss": 0.6914,
+      "step": 27480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.832781156937192e-08,
+      "loss": 0.6895,
+      "step": 27481
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.824128917393809e-08,
+      "loss": 0.7012,
+      "step": 27482
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8154898927673647e-08,
+      "loss": 0.7227,
+      "step": 27483
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8068640831723226e-08,
+      "loss": 0.7061,
+      "step": 27484
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7982514887228142e-08,
+      "loss": 0.6787,
+      "step": 27485
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.789652109533081e-08,
+      "loss": 0.7119,
+      "step": 27486
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.781065945717143e-08,
+      "loss": 0.6914,
+      "step": 27487
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.772492997388687e-08,
+      "loss": 0.7119,
+      "step": 27488
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7639332646615112e-08,
+      "loss": 0.6846,
+      "step": 27489
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7553867476487473e-08,
+      "loss": 0.7051,
+      "step": 27490
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7468534464639708e-08,
+      "loss": 0.6904,
+      "step": 27491
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7383333612200912e-08,
+      "loss": 0.7012,
+      "step": 27492
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7298264920300188e-08,
+      "loss": 0.7012,
+      "step": 27493
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.721332839006441e-08,
+      "loss": 0.7051,
+      "step": 27494
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.712852402262156e-08,
+      "loss": 0.7344,
+      "step": 27495
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.704385181909297e-08,
+      "loss": 0.7031,
+      "step": 27496
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.695931178060218e-08,
+      "loss": 0.6982,
+      "step": 27497
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.68749039082683e-08,
+      "loss": 0.6973,
+      "step": 27498
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6790628203211544e-08,
+      "loss": 0.7061,
+      "step": 27499
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6706484666547683e-08,
+      "loss": 0.6855,
+      "step": 27500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6622473299392494e-08,
+      "loss": 0.708,
+      "step": 27501
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6538594102858417e-08,
+      "loss": 0.6973,
+      "step": 27502
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.64548470780579e-08,
+      "loss": 0.6875,
+      "step": 27503
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6371232226100053e-08,
+      "loss": 0.7197,
+      "step": 27504
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6287749548092878e-08,
+      "loss": 0.708,
+      "step": 27505
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.620439904514327e-08,
+      "loss": 0.7217,
+      "step": 27506
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6121180718355898e-08,
+      "loss": 0.7305,
+      "step": 27507
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6038094568834328e-08,
+      "loss": 0.71,
+      "step": 27508
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5955140597677676e-08,
+      "loss": 0.6865,
+      "step": 27509
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5872318805987283e-08,
+      "loss": 0.6865,
+      "step": 27510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5789629194858946e-08,
+      "loss": 0.709,
+      "step": 27511
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.570707176538956e-08,
+      "loss": 0.7021,
+      "step": 27512
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.562464651867269e-08,
+      "loss": 0.7031,
+      "step": 27513
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5542353455801917e-08,
+      "loss": 0.6973,
+      "step": 27514
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5460192577865252e-08,
+      "loss": 0.7109,
+      "step": 27515
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5378163885955153e-08,
+      "loss": 0.7119,
+      "step": 27516
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5296267381155204e-08,
+      "loss": 0.7002,
+      "step": 27517
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.521450306455231e-08,
+      "loss": 0.6982,
+      "step": 27518
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5132870937230047e-08,
+      "loss": 0.6885,
+      "step": 27519
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5051371000270885e-08,
+      "loss": 0.7344,
+      "step": 27520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.497000325475285e-08,
+      "loss": 0.6992,
+      "step": 27521
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.488876770175619e-08,
+      "loss": 0.7168,
+      "step": 27522
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4807664342356706e-08,
+      "loss": 0.6875,
+      "step": 27523
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.47266931776291e-08,
+      "loss": 0.6963,
+      "step": 27524
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4645854208646957e-08,
+      "loss": 0.7031,
+      "step": 27525
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.456514743648164e-08,
+      "loss": 0.6943,
+      "step": 27526
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4484572862201183e-08,
+      "loss": 0.7041,
+      "step": 27527
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4404130486874733e-08,
+      "loss": 0.6934,
+      "step": 27528
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4323820311569213e-08,
+      "loss": 0.7021,
+      "step": 27529
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4243642337347107e-08,
+      "loss": 0.6943,
+      "step": 27530
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4163596565270896e-08,
+      "loss": 0.6904,
+      "step": 27531
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.408368299640307e-08,
+      "loss": 0.6797,
+      "step": 27532
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4003901631801663e-08,
+      "loss": 0.71,
+      "step": 27533
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3924252472523614e-08,
+      "loss": 0.6904,
+      "step": 27534
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.384473551962474e-08,
+      "loss": 0.7148,
+      "step": 27535
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3765350774159755e-08,
+      "loss": 0.6855,
+      "step": 27536
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3686098237180044e-08,
+      "loss": 0.6953,
+      "step": 27537
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3606977909734763e-08,
+      "loss": 0.6934,
+      "step": 27538
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3527989792874184e-08,
+      "loss": 0.6934,
+      "step": 27539
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3449133887644137e-08,
+      "loss": 0.7002,
+      "step": 27540
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3370410195090453e-08,
+      "loss": 0.7178,
+      "step": 27541
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3291818716255632e-08,
+      "loss": 0.6963,
+      "step": 27542
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3213359452179952e-08,
+      "loss": 0.7021,
+      "step": 27543
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.313503240390591e-08,
+      "loss": 0.6816,
+      "step": 27544
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.3056837572470457e-08,
+      "loss": 0.7002,
+      "step": 27545
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2978774958909434e-08,
+      "loss": 0.6973,
+      "step": 27546
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.290084456425645e-08,
+      "loss": 0.6846,
+      "step": 27547
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2823046389546244e-08,
+      "loss": 0.7148,
+      "step": 27548
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2745380435809094e-08,
+      "loss": 0.6787,
+      "step": 27549
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2667846704074183e-08,
+      "loss": 0.6875,
+      "step": 27550
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.259044519536846e-08,
+      "loss": 0.7051,
+      "step": 27551
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2513175910718887e-08,
+      "loss": 0.7109,
+      "step": 27552
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2436038851147978e-08,
+      "loss": 0.6865,
+      "step": 27553
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.235903401767825e-08,
+      "loss": 0.7168,
+      "step": 27554
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2282161411332215e-08,
+      "loss": 0.6943,
+      "step": 27555
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.220542103312573e-08,
+      "loss": 0.7041,
+      "step": 27556
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2128812884077978e-08,
+      "loss": 0.7314,
+      "step": 27557
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2052336965202592e-08,
+      "loss": 0.7041,
+      "step": 27558
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1975993277514317e-08,
+      "loss": 0.6963,
+      "step": 27559
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.189978182202346e-08,
+      "loss": 0.7051,
+      "step": 27560
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1823702599742537e-08,
+      "loss": 0.7266,
+      "step": 27561
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1747755611677413e-08,
+      "loss": 0.7021,
+      "step": 27562
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1671940858835062e-08,
+      "loss": 0.6846,
+      "step": 27563
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1596258342220233e-08,
+      "loss": 0.6934,
+      "step": 27564
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.152070806283657e-08,
+      "loss": 0.7021,
+      "step": 27565
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1445290021685494e-08,
+      "loss": 0.7285,
+      "step": 27566
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1370004219765094e-08,
+      "loss": 0.709,
+      "step": 27567
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1294850658073464e-08,
+      "loss": 0.7119,
+      "step": 27568
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.121982933760758e-08,
+      "loss": 0.6875,
+      "step": 27569
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1144940259359982e-08,
+      "loss": 0.6973,
+      "step": 27570
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.107018342432543e-08,
+      "loss": 0.6904,
+      "step": 27571
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0995558833492023e-08,
+      "loss": 0.6934,
+      "step": 27572
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0921066487851194e-08,
+      "loss": 0.6826,
+      "step": 27573
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0846706388387705e-08,
+      "loss": 0.6924,
+      "step": 27574
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0772478536089656e-08,
+      "loss": 0.7168,
+      "step": 27575
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0698382931938488e-08,
+      "loss": 0.6895,
+      "step": 27576
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0624419576916742e-08,
+      "loss": 0.7041,
+      "step": 27577
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.055058847200586e-08,
+      "loss": 0.6934,
+      "step": 27578
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.047688961818284e-08,
+      "loss": 0.7373,
+      "step": 27579
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.040332301642356e-08,
+      "loss": 0.7119,
+      "step": 27580
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.032988866770613e-08,
+      "loss": 0.7051,
+      "step": 27581
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0256586573000892e-08,
+      "loss": 0.6992,
+      "step": 27582
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0183416733280393e-08,
+      "loss": 0.6982,
+      "step": 27583
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.011037914951386e-08,
+      "loss": 0.7246,
+      "step": 27584
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0037473822669406e-08,
+      "loss": 0.6699,
+      "step": 27585
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.996470075371293e-08,
+      "loss": 0.7109,
+      "step": 27586
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.989205994360921e-08,
+      "loss": 0.7158,
+      "step": 27587
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9819551393320812e-08,
+      "loss": 0.7002,
+      "step": 27588
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.974717510380919e-08,
+      "loss": 0.6992,
+      "step": 27589
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9674931076032467e-08,
+      "loss": 0.6982,
+      "step": 27590
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.960281931094876e-08,
+      "loss": 0.7041,
+      "step": 27591
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.953083980951398e-08,
+      "loss": 0.6797,
+      "step": 27592
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.945899257268069e-08,
+      "loss": 0.6943,
+      "step": 27593
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9387277601402577e-08,
+      "loss": 0.709,
+      "step": 27594
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9315694896628878e-08,
+      "loss": 0.7031,
+      "step": 27595
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9244244459308835e-08,
+      "loss": 0.6807,
+      "step": 27596
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9172926290389472e-08,
+      "loss": 0.7002,
+      "step": 27597
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9101740390815583e-08,
+      "loss": 0.7256,
+      "step": 27598
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9030686761529748e-08,
+      "loss": 0.6816,
+      "step": 27599
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.895976540347566e-08,
+      "loss": 0.707,
+      "step": 27600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8888976317591457e-08,
+      "loss": 0.7158,
+      "step": 27601
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8818319504815274e-08,
+      "loss": 0.7158,
+      "step": 27602
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8747794966085252e-08,
+      "loss": 0.709,
+      "step": 27603
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.867740270233398e-08,
+      "loss": 0.6943,
+      "step": 27604
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8607142714495152e-08,
+      "loss": 0.6826,
+      "step": 27605
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.853701500350025e-08,
+      "loss": 0.7031,
+      "step": 27606
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.846701957027741e-08,
+      "loss": 0.6875,
+      "step": 27607
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.83971564157559e-08,
+      "loss": 0.7041,
+      "step": 27608
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.832742554085942e-08,
+      "loss": 0.6719,
+      "step": 27609
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8257826946515013e-08,
+      "loss": 0.6992,
+      "step": 27610
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.818836063364193e-08,
+      "loss": 0.6787,
+      "step": 27611
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.811902660316278e-08,
+      "loss": 0.6982,
+      "step": 27612
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8049824855995714e-08,
+      "loss": 0.6885,
+      "step": 27613
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7980755393058878e-08,
+      "loss": 0.7295,
+      "step": 27614
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7911818215265996e-08,
+      "loss": 0.7148,
+      "step": 27615
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.784301332353078e-08,
+      "loss": 0.7051,
+      "step": 27616
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7774340718765825e-08,
+      "loss": 0.708,
+      "step": 27617
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7705800401880412e-08,
+      "loss": 0.7158,
+      "step": 27618
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.763739237378381e-08,
+      "loss": 0.7178,
+      "step": 27619
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7569116635383075e-08,
+      "loss": 0.7197,
+      "step": 27620
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7500973187580817e-08,
+      "loss": 0.7129,
+      "step": 27621
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7432962031282974e-08,
+      "loss": 0.6924,
+      "step": 27622
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.736508316738772e-08,
+      "loss": 0.6934,
+      "step": 27623
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7297336596796555e-08,
+      "loss": 0.7061,
+      "step": 27624
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.722972232040765e-08,
+      "loss": 0.709,
+      "step": 27625
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7162240339115843e-08,
+      "loss": 0.707,
+      "step": 27626
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7094890653815977e-08,
+      "loss": 0.6875,
+      "step": 27627
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7027673265399557e-08,
+      "loss": 0.6943,
+      "step": 27628
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6960588174759206e-08,
+      "loss": 0.7139,
+      "step": 27629
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.68936353827831e-08,
+      "loss": 0.7178,
+      "step": 27630
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.682681489035831e-08,
+      "loss": 0.7236,
+      "step": 27631
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.676012669836968e-08,
+      "loss": 0.707,
+      "step": 27632
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6693570807703175e-08,
+      "loss": 0.7012,
+      "step": 27633
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6627147219239193e-08,
+      "loss": 0.6846,
+      "step": 27634
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6560855933858144e-08,
+      "loss": 0.7178,
+      "step": 27635
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6494696952438216e-08,
+      "loss": 0.6689,
+      "step": 27636
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6428670275857596e-08,
+      "loss": 0.7012,
+      "step": 27637
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6362775904988914e-08,
+      "loss": 0.6924,
+      "step": 27638
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6297013840708143e-08,
+      "loss": 0.7227,
+      "step": 27639
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6231384083885694e-08,
+      "loss": 0.6963,
+      "step": 27640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.616588663538976e-08,
+      "loss": 0.7041,
+      "step": 27641
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.610052149609076e-08,
+      "loss": 0.7041,
+      "step": 27642
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6035288666853555e-08,
+      "loss": 0.7207,
+      "step": 27643
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.597018814854301e-08,
+      "loss": 0.7061,
+      "step": 27644
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5905219942021766e-08,
+      "loss": 0.6885,
+      "step": 27645
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5840384048150248e-08,
+      "loss": 0.6943,
+      "step": 27646
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.577568046778888e-08,
+      "loss": 0.7031,
+      "step": 27647
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.571110920179475e-08,
+      "loss": 0.7139,
+      "step": 27648
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5646670251022732e-08,
+      "loss": 0.7158,
+      "step": 27649
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5582363616327702e-08,
+      "loss": 0.6875,
+      "step": 27650
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.551818929856119e-08,
+      "loss": 0.6963,
+      "step": 27651
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.545414729857475e-08,
+      "loss": 0.6846,
+      "step": 27652
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5390237617216587e-08,
+      "loss": 0.6855,
+      "step": 27653
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5326460255332688e-08,
+      "loss": 0.708,
+      "step": 27654
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5262815213769043e-08,
+      "loss": 0.6914,
+      "step": 27655
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5199302493369428e-08,
+      "loss": 0.708,
+      "step": 27656
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5135922094975385e-08,
+      "loss": 0.71,
+      "step": 27657
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5072674019426248e-08,
+      "loss": 0.7246,
+      "step": 27658
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.500955826756134e-08,
+      "loss": 0.71,
+      "step": 27659
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4946574840215554e-08,
+      "loss": 0.6943,
+      "step": 27660
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4883723738224887e-08,
+      "loss": 0.6982,
+      "step": 27661
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4821004962420893e-08,
+      "loss": 0.6895,
+      "step": 27662
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4758418513636241e-08,
+      "loss": 0.7266,
+      "step": 27663
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4695964392699158e-08,
+      "loss": 0.7158,
+      "step": 27664
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.463364260043898e-08,
+      "loss": 0.7109,
+      "step": 27665
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4571453137679492e-08,
+      "loss": 0.6885,
+      "step": 27666
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.450939600524559e-08,
+      "loss": 0.7344,
+      "step": 27667
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.444747120396106e-08,
+      "loss": 0.6992,
+      "step": 27668
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4385678734644138e-08,
+      "loss": 0.6992,
+      "step": 27669
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4324018598116385e-08,
+      "loss": 0.6934,
+      "step": 27670
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4262490795192707e-08,
+      "loss": 0.6963,
+      "step": 27671
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4201095326689118e-08,
+      "loss": 0.7217,
+      "step": 27672
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4139832193420522e-08,
+      "loss": 0.6934,
+      "step": 27673
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.407870139619738e-08,
+      "loss": 0.7236,
+      "step": 27674
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4017702935829047e-08,
+      "loss": 0.7031,
+      "step": 27675
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3956836813125985e-08,
+      "loss": 0.7236,
+      "step": 27676
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3896103028894215e-08,
+      "loss": 0.7002,
+      "step": 27677
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3835501583937538e-08,
+      "loss": 0.7139,
+      "step": 27678
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3775032479060869e-08,
+      "loss": 0.6953,
+      "step": 27679
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3714695715063564e-08,
+      "loss": 0.6885,
+      "step": 27680
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3654491292746096e-08,
+      "loss": 0.6875,
+      "step": 27681
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3594419212906718e-08,
+      "loss": 0.6934,
+      "step": 27682
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3534479476341456e-08,
+      "loss": 0.6992,
+      "step": 27683
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.347467208384412e-08,
+      "loss": 0.707,
+      "step": 27684
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3414997036207411e-08,
+      "loss": 0.6934,
+      "step": 27685
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3355454334221807e-08,
+      "loss": 0.7227,
+      "step": 27686
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3296043978677786e-08,
+      "loss": 0.6738,
+      "step": 27687
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3236765970361387e-08,
+      "loss": 0.708,
+      "step": 27688
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3177620310058647e-08,
+      "loss": 0.6924,
+      "step": 27689
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3118606998552274e-08,
+      "loss": 0.7051,
+      "step": 27690
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3059726036627196e-08,
+      "loss": 0.6953,
+      "step": 27691
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3000977425060568e-08,
+      "loss": 0.6943,
+      "step": 27692
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2942361164632877e-08,
+      "loss": 0.6846,
+      "step": 27693
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.288387725612017e-08,
+      "loss": 0.7002,
+      "step": 27694
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2825525700297381e-08,
+      "loss": 0.6875,
+      "step": 27695
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2767306497938336e-08,
+      "loss": 0.6953,
+      "step": 27696
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2709219649814642e-08,
+      "loss": 0.6787,
+      "step": 27697
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2651265156694569e-08,
+      "loss": 0.7246,
+      "step": 27698
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2593443019348616e-08,
+      "loss": 0.6875,
+      "step": 27699
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2535753238540616e-08,
+      "loss": 0.71,
+      "step": 27700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.247819581503773e-08,
+      "loss": 0.6943,
+      "step": 27701
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2420770749600463e-08,
+      "loss": 0.7061,
+      "step": 27702
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2363478042990429e-08,
+      "loss": 0.6973,
+      "step": 27703
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2306317695967018e-08,
+      "loss": 0.6963,
+      "step": 27704
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2249289709288514e-08,
+      "loss": 0.6934,
+      "step": 27705
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2192394083708758e-08,
+      "loss": 0.6826,
+      "step": 27706
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2135630819983812e-08,
+      "loss": 0.6826,
+      "step": 27707
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2078999918865297e-08,
+      "loss": 0.7168,
+      "step": 27708
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2022501381102613e-08,
+      "loss": 0.7168,
+      "step": 27709
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1966135207446273e-08,
+      "loss": 0.7109,
+      "step": 27710
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1909901398642342e-08,
+      "loss": 0.6738,
+      "step": 27711
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1853799955434676e-08,
+      "loss": 0.751,
+      "step": 27712
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1797830878569338e-08,
+      "loss": 0.6865,
+      "step": 27713
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1741994168786852e-08,
+      "loss": 0.6914,
+      "step": 27714
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1686289826826625e-08,
+      "loss": 0.7158,
+      "step": 27715
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1630717853428063e-08,
+      "loss": 0.6914,
+      "step": 27716
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1575278249326139e-08,
+      "loss": 0.7158,
+      "step": 27717
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1519971015256925e-08,
+      "loss": 0.6855,
+      "step": 27718
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1464796151953173e-08,
+      "loss": 0.709,
+      "step": 27719
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1409753660145406e-08,
+      "loss": 0.7041,
+      "step": 27720
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1354843540563043e-08,
+      "loss": 0.6934,
+      "step": 27721
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.130006579393439e-08,
+      "loss": 0.709,
+      "step": 27722
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1245420420985532e-08,
+      "loss": 0.7021,
+      "step": 27723
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1190907422440334e-08,
+      "loss": 0.7148,
+      "step": 27724
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1136526799020442e-08,
+      "loss": 0.6924,
+      "step": 27725
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1082278551447501e-08,
+      "loss": 0.6924,
+      "step": 27726
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1028162680439824e-08,
+      "loss": 0.6914,
+      "step": 27727
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0974179186715728e-08,
+      "loss": 0.7051,
+      "step": 27728
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0920328070989083e-08,
+      "loss": 0.6904,
+      "step": 27729
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0866609333974875e-08,
+      "loss": 0.7021,
+      "step": 27730
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0813022976383647e-08,
+      "loss": 0.7061,
+      "step": 27731
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.075956899892594e-08,
+      "loss": 0.7051,
+      "step": 27732
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0706247402311187e-08,
+      "loss": 0.7021,
+      "step": 27733
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.065305818724549e-08,
+      "loss": 0.7246,
+      "step": 27734
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0600001354432732e-08,
+      "loss": 0.7021,
+      "step": 27735
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0547076904577902e-08,
+      "loss": 0.6904,
+      "step": 27736
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0494284838380442e-08,
+      "loss": 0.7334,
+      "step": 27737
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0441625156542012e-08,
+      "loss": 0.709,
+      "step": 27738
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0389097859758723e-08,
+      "loss": 0.6895,
+      "step": 27739
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0336702948726685e-08,
+      "loss": 0.6924,
+      "step": 27740
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0284440424140895e-08,
+      "loss": 0.7119,
+      "step": 27741
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0232310286694136e-08,
+      "loss": 0.6924,
+      "step": 27742
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0180312537078074e-08,
+      "loss": 0.7012,
+      "step": 27743
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0128447175979939e-08,
+      "loss": 0.6865,
+      "step": 27744
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0076714204086957e-08,
+      "loss": 0.6748,
+      "step": 27745
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0025113622087468e-08,
+      "loss": 0.7021,
+      "step": 27746
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.973645430662038e-09,
+      "loss": 0.6943,
+      "step": 27747
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.922309630494565e-09,
+      "loss": 0.6953,
+      "step": 27748
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.871106222265615e-09,
+      "loss": 0.6953,
+      "step": 27749
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.820035206652422e-09,
+      "loss": 0.7012,
+      "step": 27750
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.769096584332227e-09,
+      "loss": 0.7031,
+      "step": 27751
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.718290355981153e-09,
+      "loss": 0.6855,
+      "step": 27752
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.667616522270885e-09,
+      "loss": 0.71,
+      "step": 27753
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.617075083874217e-09,
+      "loss": 0.709,
+      "step": 27754
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.566666041460615e-09,
+      "loss": 0.7129,
+      "step": 27755
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.516389395698433e-09,
+      "loss": 0.7012,
+      "step": 27756
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.466245147252695e-09,
+      "loss": 0.7158,
+      "step": 27757
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.416233296789535e-09,
+      "loss": 0.708,
+      "step": 27758
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.366353844970643e-09,
+      "loss": 0.6963,
+      "step": 27759
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.316606792456606e-09,
+      "loss": 0.7129,
+      "step": 27760
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.266992139908004e-09,
+      "loss": 0.7031,
+      "step": 27761
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.217509887982091e-09,
+      "loss": 0.7012,
+      "step": 27762
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.168160037335005e-09,
+      "loss": 0.6826,
+      "step": 27763
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.118942588618452e-09,
+      "loss": 0.7021,
+      "step": 27764
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.06985754248746e-09,
+      "loss": 0.7168,
+      "step": 27765
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.0209048995904e-09,
+      "loss": 0.709,
+      "step": 27766
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.972084660577862e-09,
+      "loss": 0.6943,
+      "step": 27767
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.923396826094887e-09,
+      "loss": 0.6768,
+      "step": 27768
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.874841396788735e-09,
+      "loss": 0.7246,
+      "step": 27769
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.826418373301116e-09,
+      "loss": 0.7158,
+      "step": 27770
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.778127756274846e-09,
+      "loss": 0.6963,
+      "step": 27771
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.729969546350526e-09,
+      "loss": 0.7148,
+      "step": 27772
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.681943744164311e-09,
+      "loss": 0.7021,
+      "step": 27773
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.634050350353473e-09,
+      "loss": 0.7012,
+      "step": 27774
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.586289365553057e-09,
+      "loss": 0.7021,
+      "step": 27775
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.538660790397003e-09,
+      "loss": 0.7158,
+      "step": 27776
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.491164625514802e-09,
+      "loss": 0.708,
+      "step": 27777
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.443800871535956e-09,
+      "loss": 0.7158,
+      "step": 27778
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.39656952908885e-09,
+      "loss": 0.709,
+      "step": 27779
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.349470598799647e-09,
+      "loss": 0.6797,
+      "step": 27780
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.302504081292295e-09,
+      "loss": 0.6992,
+      "step": 27781
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.255669977188518e-09,
+      "loss": 0.6934,
+      "step": 27782
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.208968287110042e-09,
+      "loss": 0.6973,
+      "step": 27783
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.16239901167526e-09,
+      "loss": 0.709,
+      "step": 27784
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.115962151501455e-09,
+      "loss": 0.7061,
+      "step": 27785
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.069657707202583e-09,
+      "loss": 0.6865,
+      "step": 27786
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.023485679394816e-09,
+      "loss": 0.7158,
+      "step": 27787
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.977446068688776e-09,
+      "loss": 0.71,
+      "step": 27788
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.931538875693977e-09,
+      "loss": 0.7002,
+      "step": 27789
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.885764101019933e-09,
+      "loss": 0.708,
+      "step": 27790
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.840121745272821e-09,
+      "loss": 0.6973,
+      "step": 27791
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.794611809057717e-09,
+      "loss": 0.6855,
+      "step": 27792
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.74923429297636e-09,
+      "loss": 0.7178,
+      "step": 27793
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.7039891976316e-09,
+      "loss": 0.7246,
+      "step": 27794
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.65887652362407e-09,
+      "loss": 0.6904,
+      "step": 27795
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.613896271548848e-09,
+      "loss": 0.6816,
+      "step": 27796
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.569048442003234e-09,
+      "loss": 0.7012,
+      "step": 27797
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.52433303558231e-09,
+      "loss": 0.6855,
+      "step": 27798
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.479750052877821e-09,
+      "loss": 0.7061,
+      "step": 27799
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.435299494481518e-09,
+      "loss": 0.7051,
+      "step": 27800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.390981360980709e-09,
+      "loss": 0.71,
+      "step": 27801
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.346795652964922e-09,
+      "loss": 0.707,
+      "step": 27802
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.302742371018135e-09,
+      "loss": 0.7158,
+      "step": 27803
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.258821515724324e-09,
+      "loss": 0.7197,
+      "step": 27804
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.215033087666357e-09,
+      "loss": 0.6914,
+      "step": 27805
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.17137708742377e-09,
+      "loss": 0.7178,
+      "step": 27806
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.1278535155760994e-09,
+      "loss": 0.7041,
+      "step": 27807
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.084462372698442e-09,
+      "loss": 0.6973,
+      "step": 27808
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.041203659367002e-09,
+      "loss": 0.7119,
+      "step": 27809
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.998077376154655e-09,
+      "loss": 0.7051,
+      "step": 27810
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.955083523634276e-09,
+      "loss": 0.7031,
+      "step": 27811
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.912222102373189e-09,
+      "loss": 0.6846,
+      "step": 27812
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.869493112940939e-09,
+      "loss": 0.6973,
+      "step": 27813
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.826896555903739e-09,
+      "loss": 0.7119,
+      "step": 27814
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.784432431826693e-09,
+      "loss": 0.6797,
+      "step": 27815
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.7421007412704635e-09,
+      "loss": 0.6992,
+      "step": 27816
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.699901484797933e-09,
+      "loss": 0.7002,
+      "step": 27817
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.6578346629675436e-09,
+      "loss": 0.7021,
+      "step": 27818
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.615900276337739e-09,
+      "loss": 0.7002,
+      "step": 27819
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.574098325462519e-09,
+      "loss": 0.7217,
+      "step": 27820
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.532428810898106e-09,
+      "loss": 0.6836,
+      "step": 27821
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.49089173319406e-09,
+      "loss": 0.709,
+      "step": 27822
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.449487092903273e-09,
+      "loss": 0.7061,
+      "step": 27823
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.408214890571973e-09,
+      "loss": 0.7148,
+      "step": 27824
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.367075126749722e-09,
+      "loss": 0.6992,
+      "step": 27825
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.3260678019805284e-09,
+      "loss": 0.7178,
+      "step": 27826
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.285192916807293e-09,
+      "loss": 0.6992,
+      "step": 27827
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.244450471772912e-09,
+      "loss": 0.7314,
+      "step": 27828
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.203840467416955e-09,
+      "loss": 0.7119,
+      "step": 27829
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.16336290427566e-09,
+      "loss": 0.6963,
+      "step": 27830
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.1230177828885964e-09,
+      "loss": 0.6846,
+      "step": 27831
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.082805103788669e-09,
+      "loss": 0.708,
+      "step": 27832
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.042724867508787e-09,
+      "loss": 0.6787,
+      "step": 27833
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.002777074580746e-09,
+      "loss": 0.7021,
+      "step": 27834
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.962961725533012e-09,
+      "loss": 0.7178,
+      "step": 27835
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.923278820894052e-09,
+      "loss": 0.6885,
+      "step": 27836
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.883728361190111e-09,
+      "loss": 0.6895,
+      "step": 27837
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.844310346944104e-09,
+      "loss": 0.6846,
+      "step": 27838
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.805024778678947e-09,
+      "loss": 0.749,
+      "step": 27839
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.765871656915334e-09,
+      "loss": 0.6992,
+      "step": 27840
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.726850982172849e-09,
+      "loss": 0.7061,
+      "step": 27841
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.687962754967746e-09,
+      "loss": 0.6953,
+      "step": 27842
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.649206975815169e-09,
+      "loss": 0.7178,
+      "step": 27843
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.610583645229151e-09,
+      "loss": 0.6885,
+      "step": 27844
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.572092763721504e-09,
+      "loss": 0.7139,
+      "step": 27845
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.533734331801821e-09,
+      "loss": 0.7041,
+      "step": 27846
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.495508349979695e-09,
+      "loss": 0.6787,
+      "step": 27847
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.457414818761386e-09,
+      "loss": 0.7256,
+      "step": 27848
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.419453738649827e-09,
+      "loss": 0.6934,
+      "step": 27849
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.381625110150168e-09,
+      "loss": 0.6826,
+      "step": 27850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.343928933763121e-09,
+      "loss": 0.6924,
+      "step": 27851
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.306365209988285e-09,
+      "loss": 0.709,
+      "step": 27852
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.26893393932415e-09,
+      "loss": 0.7002,
+      "step": 27853
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.231635122265877e-09,
+      "loss": 0.7188,
+      "step": 27854
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.194468759306404e-09,
+      "loss": 0.707,
+      "step": 27855
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.15743485094089e-09,
+      "loss": 0.6982,
+      "step": 27856
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.120533397658944e-09,
+      "loss": 0.6729,
+      "step": 27857
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.083764399950175e-09,
+      "loss": 0.6992,
+      "step": 27858
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.04712785830086e-09,
+      "loss": 0.7109,
+      "step": 27859
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.010623773197276e-09,
+      "loss": 0.7295,
+      "step": 27860
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.974252145122371e-09,
+      "loss": 0.7051,
+      "step": 27861
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.938012974559092e-09,
+      "loss": 0.6914,
+      "step": 27862
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.901906261987055e-09,
+      "loss": 0.6768,
+      "step": 27863
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.865932007885876e-09,
+      "loss": 0.6855,
+      "step": 27864
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.830090212730731e-09,
+      "loss": 0.6855,
+      "step": 27865
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.794380876996796e-09,
+      "loss": 0.6934,
+      "step": 27866
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.758804001158135e-09,
+      "loss": 0.6982,
+      "step": 27867
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.723359585686593e-09,
+      "loss": 0.7129,
+      "step": 27868
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.688047631050685e-09,
+      "loss": 0.7412,
+      "step": 27869
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.6528681377189244e-09,
+      "loss": 0.7061,
+      "step": 27870
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.617821106157605e-09,
+      "loss": 0.6953,
+      "step": 27871
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.5829065368308e-09,
+      "loss": 0.6924,
+      "step": 27872
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.548124430202583e-09,
+      "loss": 0.6934,
+      "step": 27873
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.513474786731476e-09,
+      "loss": 0.709,
+      "step": 27874
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.478957606878221e-09,
+      "loss": 0.6973,
+      "step": 27875
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4445728911002296e-09,
+      "loss": 0.7031,
+      "step": 27876
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.410320639853804e-09,
+      "loss": 0.707,
+      "step": 27877
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.376200853590806e-09,
+      "loss": 0.7207,
+      "step": 27878
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.342213532765316e-09,
+      "loss": 0.708,
+      "step": 27879
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.3083586778269735e-09,
+      "loss": 0.7217,
+      "step": 27880
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.27463628922431e-09,
+      "loss": 0.7051,
+      "step": 27881
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.241046367404744e-09,
+      "loss": 0.7031,
+      "step": 27882
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.207588912812366e-09,
+      "loss": 0.7119,
+      "step": 27883
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.174263925892375e-09,
+      "loss": 0.6924,
+      "step": 27884
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.141071407084418e-09,
+      "loss": 0.7158,
+      "step": 27885
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.108011356829256e-09,
+      "loss": 0.7178,
+      "step": 27886
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.075083775565425e-09,
+      "loss": 0.7119,
+      "step": 27887
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.042288663729244e-09,
+      "loss": 0.6973,
+      "step": 27888
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.009626021754809e-09,
+      "loss": 0.7393,
+      "step": 27889
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.977095850075108e-09,
+      "loss": 0.7061,
+      "step": 27890
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.944698149122017e-09,
+      "loss": 0.6982,
+      "step": 27891
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.912432919322973e-09,
+      "loss": 0.7041,
+      "step": 27892
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.880300161107631e-09,
+      "loss": 0.7012,
+      "step": 27893
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8482998749000964e-09,
+      "loss": 0.6846,
+      "step": 27894
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8164320611266956e-09,
+      "loss": 0.6963,
+      "step": 27895
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.784696720207093e-09,
+      "loss": 0.7109,
+      "step": 27896
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.753093852563172e-09,
+      "loss": 0.7129,
+      "step": 27897
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.721623458614598e-09,
+      "loss": 0.6807,
+      "step": 27898
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6902855387765946e-09,
+      "loss": 0.7178,
+      "step": 27899
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6590800934666047e-09,
+      "loss": 0.6992,
+      "step": 27900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6280071230954118e-09,
+      "loss": 0.6953,
+      "step": 27901
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5970666280771285e-09,
+      "loss": 0.6904,
+      "step": 27902
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5662586088214267e-09,
+      "loss": 0.7041,
+      "step": 27903
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.535583065735759e-09,
+      "loss": 0.6973,
+      "step": 27904
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5050399992275776e-09,
+      "loss": 0.6895,
+      "step": 27905
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4746294096998923e-09,
+      "loss": 0.7041,
+      "step": 27906
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4443512975579354e-09,
+      "loss": 0.6943,
+      "step": 27907
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4142056632024968e-09,
+      "loss": 0.707,
+      "step": 27908
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3841925070310365e-09,
+      "loss": 0.6719,
+      "step": 27909
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3543118294443454e-09,
+      "loss": 0.7119,
+      "step": 27910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3245636308365523e-09,
+      "loss": 0.7021,
+      "step": 27911
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.294947911601787e-09,
+      "loss": 0.7012,
+      "step": 27912
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2654646721330675e-09,
+      "loss": 0.707,
+      "step": 27913
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2361139128211927e-09,
+      "loss": 0.6855,
+      "step": 27914
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.206895634054741e-09,
+      "loss": 0.6943,
+      "step": 27915
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1778098362211796e-09,
+      "loss": 0.7139,
+      "step": 27916
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.148856519705756e-09,
+      "loss": 0.71,
+      "step": 27917
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1200356848937185e-09,
+      "loss": 0.707,
+      "step": 27918
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.091347332163652e-09,
+      "loss": 0.7119,
+      "step": 27919
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.0627914618985845e-09,
+      "loss": 0.7002,
+      "step": 27920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.034368074477101e-09,
+      "loss": 0.6914,
+      "step": 27921
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.0060771702733472e-09,
+      "loss": 0.6973,
+      "step": 27922
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9779187496647987e-09,
+      "loss": 0.7148,
+      "step": 27923
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9498928130222706e-09,
+      "loss": 0.7275,
+      "step": 27924
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.921999360719907e-09,
+      "loss": 0.6963,
+      "step": 27925
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8942383931251928e-09,
+      "loss": 0.6787,
+      "step": 27926
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8666099106078314e-09,
+      "loss": 0.7002,
+      "step": 27927
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.839113913531977e-09,
+      "loss": 0.6855,
+      "step": 27928
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8117504022628916e-09,
+      "loss": 0.7266,
+      "step": 27929
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.784519377163619e-09,
+      "loss": 0.6875,
+      "step": 27930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7574208385938717e-09,
+      "loss": 0.7119,
+      "step": 27931
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7304547869144714e-09,
+      "loss": 0.6943,
+      "step": 27932
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7036212224818e-09,
+      "loss": 0.7275,
+      "step": 27933
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6769201456511297e-09,
+      "loss": 0.7051,
+      "step": 27934
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6503515567766202e-09,
+      "loss": 0.7246,
+      "step": 27935
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.623915456211323e-09,
+      "loss": 0.6885,
+      "step": 27936
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.597611844303849e-09,
+      "loss": 0.7158,
+      "step": 27937
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5714407214050273e-09,
+      "loss": 0.7178,
+      "step": 27938
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5454020878590278e-09,
+      "loss": 0.7012,
+      "step": 27939
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5194959440122403e-09,
+      "loss": 0.6904,
+      "step": 27940
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4937222902088333e-09,
+      "loss": 0.6992,
+      "step": 27941
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4680811267885353e-09,
+      "loss": 0.708,
+      "step": 27942
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4425724540932947e-09,
+      "loss": 0.7051,
+      "step": 27943
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4171962724584e-09,
+      "loss": 0.7031,
+      "step": 27944
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.391952582223578e-09,
+      "loss": 0.6914,
+      "step": 27945
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3668413837196757e-09,
+      "loss": 0.6904,
+      "step": 27946
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.341862677283091e-09,
+      "loss": 0.6885,
+      "step": 27947
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3170164632413394e-09,
+      "loss": 0.707,
+      "step": 27948
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2923027419263777e-09,
+      "loss": 0.7031,
+      "step": 27949
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2677215136646115e-09,
+      "loss": 0.6982,
+      "step": 27950
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2432727787824458e-09,
+      "loss": 0.7021,
+      "step": 27951
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2189565376029566e-09,
+      "loss": 0.7295,
+      "step": 27952
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.194772790449218e-09,
+      "loss": 0.7061,
+      "step": 27953
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.170721537640974e-09,
+      "loss": 0.7139,
+      "step": 27954
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.146802779496859e-09,
+      "loss": 0.7148,
+      "step": 27955
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1230165163355078e-09,
+      "loss": 0.6807,
+      "step": 27956
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0993627484700018e-09,
+      "loss": 0.7002,
+      "step": 27957
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.075841476214535e-09,
+      "loss": 0.7031,
+      "step": 27958
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0524526998821904e-09,
+      "loss": 0.6846,
+      "step": 27959
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.029196419780499e-09,
+      "loss": 0.6934,
+      "step": 27960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0060726362203244e-09,
+      "loss": 0.7139,
+      "step": 27961
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9830813495058665e-09,
+      "loss": 0.6924,
+      "step": 27962
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9602225599424372e-09,
+      "loss": 0.6963,
+      "step": 27963
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9374962678331277e-09,
+      "loss": 0.7188,
+      "step": 27964
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9149024734799184e-09,
+      "loss": 0.7256,
+      "step": 27965
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.892441177181459e-09,
+      "loss": 0.6855,
+      "step": 27966
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8701123792352894e-09,
+      "loss": 0.7012,
+      "step": 27967
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.847916079936729e-09,
+      "loss": 0.6963,
+      "step": 27968
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8258522795822075e-09,
+      "loss": 0.708,
+      "step": 27969
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.803920978461493e-09,
+      "loss": 0.7129,
+      "step": 27970
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7821221768665741e-09,
+      "loss": 0.7041,
+      "step": 27971
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.760455875086109e-09,
+      "loss": 0.7021,
+      "step": 27972
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7389220734076451e-09,
+      "loss": 0.6846,
+      "step": 27973
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7175207721153996e-09,
+      "loss": 0.7129,
+      "step": 27974
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6962519714935899e-09,
+      "loss": 0.707,
+      "step": 27975
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6751156718253225e-09,
+      "loss": 0.7236,
+      "step": 27976
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6541118733892636e-09,
+      "loss": 0.71,
+      "step": 27977
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6332405764629688e-09,
+      "loss": 0.6953,
+      "step": 27978
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6125017813251043e-09,
+      "loss": 0.6709,
+      "step": 27979
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.591895488249895e-09,
+      "loss": 0.6846,
+      "step": 27980
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5714216975093454e-09,
+      "loss": 0.71,
+      "step": 27981
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5510804093754606e-09,
+      "loss": 0.7031,
+      "step": 27982
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5308716241191345e-09,
+      "loss": 0.6787,
+      "step": 27983
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5107953420057109e-09,
+      "loss": 0.7109,
+      "step": 27984
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.490851563302753e-09,
+      "loss": 0.6973,
+      "step": 27985
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4710402882744946e-09,
+      "loss": 0.7373,
+      "step": 27986
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4513615171829477e-09,
+      "loss": 0.708,
+      "step": 27987
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.431815250289015e-09,
+      "loss": 0.7383,
+      "step": 27988
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4124014878513781e-09,
+      "loss": 0.6934,
+      "step": 27989
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3931202301287195e-09,
+      "loss": 0.6924,
+      "step": 27990
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3739714773763902e-09,
+      "loss": 0.707,
+      "step": 27991
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3549552298464109e-09,
+      "loss": 0.7109,
+      "step": 27992
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3360714877919122e-09,
+      "loss": 0.6953,
+      "step": 27993
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3173202514626948e-09,
+      "loss": 0.6943,
+      "step": 27994
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2987015211085585e-09,
+      "loss": 0.7061,
+      "step": 27995
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2802152969748627e-09,
+      "loss": 0.7041,
+      "step": 27996
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2618615793080769e-09,
+      "loss": 0.7012,
+      "step": 27997
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2436403683491194e-09,
+      "loss": 0.6963,
+      "step": 27998
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2255516643411292e-09,
+      "loss": 0.6963,
+      "step": 27999
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.207595467522804e-09,
+      "loss": 0.6807,
+      "step": 28000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1897717781339525e-09,
+      "loss": 0.6885,
+      "step": 28001
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.172080596408831e-09,
+      "loss": 0.7314,
+      "step": 28002
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1545219225828075e-09,
+      "loss": 0.7178,
+      "step": 28003
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1370957568890284e-09,
+      "loss": 0.7139,
+      "step": 28004
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.11980209955731e-09,
+      "loss": 0.6826,
+      "step": 28005
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1026409508163583e-09,
+      "loss": 0.7227,
+      "step": 28006
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0856123108959893e-09,
+      "loss": 0.6982,
+      "step": 28007
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.068716180019358e-09,
+      "loss": 0.6934,
+      "step": 28008
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0519525584129497e-09,
+      "loss": 0.7021,
+      "step": 28009
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0353214462954786e-09,
+      "loss": 0.7031,
+      "step": 28010
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0188228438912096e-09,
+      "loss": 0.7002,
+      "step": 28011
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.002456751415526e-09,
+      "loss": 0.6982,
+      "step": 28012
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.862231690871415e-10,
+      "loss": 0.6924,
+      "step": 28013
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.701220971214398e-10,
+      "loss": 0.7031,
+      "step": 28014
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.541535357304733e-10,
+      "loss": 0.6904,
+      "step": 28015
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.383174851262943e-10,
+      "loss": 0.6924,
+      "step": 28016
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.226139455187355e-10,
+      "loss": 0.7021,
+      "step": 28017
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.070429171165185e-10,
+      "loss": 0.7168,
+      "step": 28018
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.916044001261448e-10,
+      "loss": 0.7012,
+      "step": 28019
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.762983947507852e-10,
+      "loss": 0.6953,
+      "step": 28020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.611249011947209e-10,
+      "loss": 0.71,
+      "step": 28021
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.460839196577919e-10,
+      "loss": 0.7158,
+      "step": 28022
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.311754503409486e-10,
+      "loss": 0.7021,
+      "step": 28023
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.163994934395903e-10,
+      "loss": 0.7012,
+      "step": 28024
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.017560491513365e-10,
+      "loss": 0.7148,
+      "step": 28025
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.872451176704765e-10,
+      "loss": 0.7158,
+      "step": 28026
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.728666991879685e-10,
+      "loss": 0.7119,
+      "step": 28027
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.586207938947709e-10,
+      "loss": 0.6953,
+      "step": 28028
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.445074019796217e-10,
+      "loss": 0.6885,
+      "step": 28029
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.305265236301485e-10,
+      "loss": 0.7275,
+      "step": 28030
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.166781590306482e-10,
+      "loss": 0.7061,
+      "step": 28031
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.029623083665283e-10,
+      "loss": 0.708,
+      "step": 28032
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.893789718165344e-10,
+      "loss": 0.6924,
+      "step": 28033
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.759281495638537e-10,
+      "loss": 0.6914,
+      "step": 28034
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.626098417839011e-10,
+      "loss": 0.6934,
+      "step": 28035
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.494240486554226e-10,
+      "loss": 0.7051,
+      "step": 28036
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.363707703527233e-10,
+      "loss": 0.6836,
+      "step": 28037
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.234500070478877e-10,
+      "loss": 0.7246,
+      "step": 28038
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.106617589118901e-10,
+      "loss": 0.6904,
+      "step": 28039
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.98006026115705e-10,
+      "loss": 0.708,
+      "step": 28040
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.854828088269759e-10,
+      "loss": 0.7227,
+      "step": 28041
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.73092107210016e-10,
+      "loss": 0.7207,
+      "step": 28042
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.608339214302483e-10,
+      "loss": 0.6982,
+      "step": 28043
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.487082516497655e-10,
+      "loss": 0.7148,
+      "step": 28044
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.367150980295499e-10,
+      "loss": 0.7021,
+      "step": 28045
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.248544607272532e-10,
+      "loss": 0.7002,
+      "step": 28046
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.131263399027475e-10,
+      "loss": 0.7031,
+      "step": 28047
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.015307357092436e-10,
+      "loss": 0.6904,
+      "step": 28048
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.900676483010624e-10,
+      "loss": 0.6904,
+      "step": 28049
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.787370778291944e-10,
+      "loss": 0.7197,
+      "step": 28050
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.675390244457401e-10,
+      "loss": 0.7021,
+      "step": 28051
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.564734882972488e-10,
+      "loss": 0.707,
+      "step": 28052
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.455404695324905e-10,
+      "loss": 0.6924,
+      "step": 28053
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.3473996829357376e-10,
+      "loss": 0.7002,
+      "step": 28054
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.240719847259378e-10,
+      "loss": 0.708,
+      "step": 28055
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.135365189694707e-10,
+      "loss": 0.709,
+      "step": 28056
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.031335711651707e-10,
+      "loss": 0.6943,
+      "step": 28057
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.9286314144959536e-10,
+      "loss": 0.6924,
+      "step": 28058
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.8272522995930207e-10,
+      "loss": 0.6963,
+      "step": 28059
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.72719836829738e-10,
+      "loss": 0.6846,
+      "step": 28060
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.628469621919095e-10,
+      "loss": 0.6953,
+      "step": 28061
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.531066061768229e-10,
+      "loss": 0.7275,
+      "step": 28062
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.434987689143743e-10,
+      "loss": 0.708,
+      "step": 28063
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3402345053112907e-10,
+      "loss": 0.6992,
+      "step": 28064
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.246806511536527e-10,
+      "loss": 0.7305,
+      "step": 28065
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.1547037090406963e-10,
+      "loss": 0.7002,
+      "step": 28066
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.063926099067249e-10,
+      "loss": 0.6865,
+      "step": 28067
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.9744736828041244e-10,
+      "loss": 0.7363,
+      "step": 28068
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.88634646143926e-10,
+      "loss": 0.709,
+      "step": 28069
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.7995444361383906e-10,
+      "loss": 0.7119,
+      "step": 28070
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.714067608056148e-10,
+      "loss": 0.7217,
+      "step": 28071
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6299159783249597e-10,
+      "loss": 0.6826,
+      "step": 28072
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5470895480550486e-10,
+      "loss": 0.7344,
+      "step": 28073
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.4655883183455356e-10,
+      "loss": 0.7021,
+      "step": 28074
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.385412290284439e-10,
+      "loss": 0.6943,
+      "step": 28075
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.306561464926471e-10,
+      "loss": 0.7109,
+      "step": 28076
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.2290358433152414e-10,
+      "loss": 0.6963,
+      "step": 28077
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.1528354264832573e-10,
+      "loss": 0.707,
+      "step": 28078
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0779602154408218e-10,
+      "loss": 0.707,
+      "step": 28079
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0044102111760334e-10,
+      "loss": 0.6973,
+      "step": 28080
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.9321854146769902e-10,
+      "loss": 0.6719,
+      "step": 28081
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.86128582687628e-10,
+      "loss": 0.6895,
+      "step": 28082
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7917114487286947e-10,
+      "loss": 0.7158,
+      "step": 28083
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.723462281155719e-10,
+      "loss": 0.6982,
+      "step": 28084
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6565383250566335e-10,
+      "loss": 0.708,
+      "step": 28085
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5909395813307194e-10,
+      "loss": 0.7256,
+      "step": 28086
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5266660508328478e-10,
+      "loss": 0.7285,
+      "step": 28087
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.463717734417891e-10,
+      "loss": 0.7178,
+      "step": 28088
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4020946329185158e-10,
+      "loss": 0.7129,
+      "step": 28089
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3417967471673897e-10,
+      "loss": 0.7031,
+      "step": 28090
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.282824077941669e-10,
+      "loss": 0.7021,
+      "step": 28091
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2251766260407138e-10,
+      "loss": 0.6924,
+      "step": 28092
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.168854392219476e-10,
+      "loss": 0.6943,
+      "step": 28093
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1138573772218053e-10,
+      "loss": 0.708,
+      "step": 28094
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0601855817804486e-10,
+      "loss": 0.6963,
+      "step": 28095
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0078390066059484e-10,
+      "loss": 0.7139,
+      "step": 28096
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.568176523977458e-11,
+      "loss": 0.6777,
+      "step": 28097
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.071215198219741e-11,
+      "loss": 0.7158,
+      "step": 28098
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.587506095447673e-11,
+      "loss": 0.7305,
+      "step": 28099
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.117049222100548e-11,
+      "loss": 0.7002,
+      "step": 28100
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.659844584284593e-11,
+      "loss": 0.7148,
+      "step": 28101
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.215892188106033e-11,
+      "loss": 0.6914,
+      "step": 28102
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.785192039560073e-11,
+      "loss": 0.7051,
+      "step": 28103
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.367744144197829e-11,
+      "loss": 0.7207,
+      "step": 28104
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.963548507681438e-11,
+      "loss": 0.7061,
+      "step": 28105
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.5726051352289476e-11,
+      "loss": 0.6914,
+      "step": 28106
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.194914032169429e-11,
+      "loss": 0.7598,
+      "step": 28107
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.830475203387863e-11,
+      "loss": 0.6777,
+      "step": 28108
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.4792886538802535e-11,
+      "loss": 0.7158,
+      "step": 28109
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.141354388087493e-11,
+      "loss": 0.7051,
+      "step": 28110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.816672410561495e-11,
+      "loss": 0.6904,
+      "step": 28111
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.5052427256321296e-11,
+      "loss": 0.6777,
+      "step": 28112
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.207065337518245e-11,
+      "loss": 0.7021,
+      "step": 28113
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.9221402499945983e-11,
+      "loss": 0.6992,
+      "step": 28114
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.650467466946971e-11,
+      "loss": 0.6953,
+      "step": 28115
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.3920469919280766e-11,
+      "loss": 0.7314,
+      "step": 28116
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.146878828379606e-11,
+      "loss": 0.707,
+      "step": 28117
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.914962979632229e-11,
+      "loss": 0.7061,
+      "step": 28118
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.696299448572525e-11,
+      "loss": 0.7256,
+      "step": 28119
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4908882383091182e-11,
+      "loss": 0.7139,
+      "step": 28120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2987293513955224e-11,
+      "loss": 0.71,
+      "step": 28121
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.1198227904962722e-11,
+      "loss": 0.7041,
+      "step": 28122
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.541685578318138e-12,
+      "loss": 0.6826,
+      "step": 28123
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.017666557336157e-12,
+      "loss": 0.6943,
+      "step": 28124
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.626170863111014e-12,
+      "loss": 0.6963,
+      "step": 28125
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.367198511185834e-12,
+      "loss": 0.6934,
+      "step": 28126
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.240749520434406e-12,
+      "loss": 0.6826,
+      "step": 28127
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.246823905289631e-12,
+      "loss": 0.7236,
+      "step": 28128
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.3854216790741847e-12,
+      "loss": 0.7236,
+      "step": 28129
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.656542852890297e-12,
+      "loss": 0.7051,
+      "step": 28130
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.060187436729976e-12,
+      "loss": 0.7246,
+      "step": 28131
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.963554372545588e-13,
+      "loss": 0.7012,
+      "step": 28132
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.6504686223560727e-13,
+      "loss": 0.709,
+      "step": 28133
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.626171611401333e-14,
+      "loss": 0.7207,
+      "step": 28134
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.7227,
+      "step": 28135
+    },
+    {
+      "epoch": 5.0,
+      "step": 28135,
+      "total_flos": 1.0130367629525778e+19,
+      "train_loss": 0.7608149047005509,
+      "train_runtime": 27187.6938,
+      "train_samples_per_second": 66.227,
+      "train_steps_per_second": 1.035
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 28135,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 1.0130367629525778e+19,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dcce45792f17c6f4a217b759836daa4a81605ce7
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md
@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.5.0
diff --git a/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..cb6c83aa528ef6fc8db74323ae14e83f72a24f97
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "vicuna-v1-3-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "q_proj",
+    "down_proj",
+    "v_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..07ded93858a29daab1afab78462cac5633001857
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6dfdb3407a8583bee602ab6f9006ae849e94c47cdde8560dbaa8efd18a05b716
+size 319970957
diff --git a/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8a11c13945ca6691233666c289a41f105fe5499a
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json
@@ -0,0 +1,36 @@
+{
+  "_name_or_path": "vicuna-v1-3-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "mm_graph_tower": "hvqvae2",
+  "mm_hidden_size": 308,
+  "mm_projector_type": "hlinear",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "model_type": "llava_graph",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.33.2",
+  "tune_mm_mlp_adapter": false,
+  "use_cache": true,
+  "use_lap_pe": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}
diff --git a/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..6eecd3c92cc7eaabe4048ef1bee1aa96d51fdb22
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b6107c13a4144d5e3d3b5d6829d32e150a30000801cbe900f70439ba4124db9
+size 11335231
diff --git a/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..71e5161404d60474203182d5e69edc60bb75e0c6
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/reagent_pred-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json
@@ -0,0 +1,58348 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 9720,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.84931506849315e-08,
+      "loss": 1.4941,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.36986301369863e-07,
+      "loss": 1.6074,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0547945205479452e-07,
+      "loss": 1.6055,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.73972602739726e-07,
+      "loss": 1.8652,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4246575342465755e-07,
+      "loss": 1.8867,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 1.6035,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.794520547945206e-07,
+      "loss": 1.5391,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.47945205479452e-07,
+      "loss": 1.8281,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.164383561643836e-07,
+      "loss": 1.8516,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.849315068493151e-07,
+      "loss": 1.6543,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.534246575342466e-07,
+      "loss": 1.627,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 1.7715,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.904109589041097e-07,
+      "loss": 1.6699,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.589041095890411e-07,
+      "loss": 1.7461,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0273972602739727e-06,
+      "loss": 1.6094,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.095890410958904e-06,
+      "loss": 1.7539,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1643835616438357e-06,
+      "loss": 1.6406,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2328767123287673e-06,
+      "loss": 1.6855,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3013698630136986e-06,
+      "loss": 1.6719,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3698630136986302e-06,
+      "loss": 1.7676,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4383561643835616e-06,
+      "loss": 1.666,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5068493150684932e-06,
+      "loss": 1.6602,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5753424657534248e-06,
+      "loss": 1.7383,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 1.7988,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7123287671232877e-06,
+      "loss": 1.7129,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7808219178082193e-06,
+      "loss": 1.9531,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8493150684931507e-06,
+      "loss": 1.666,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 1.832,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9863013698630136e-06,
+      "loss": 2.0137,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0547945205479454e-06,
+      "loss": 1.918,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.123287671232877e-06,
+      "loss": 1.6602,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.191780821917808e-06,
+      "loss": 1.9531,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.26027397260274e-06,
+      "loss": 1.8438,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3287671232876713e-06,
+      "loss": 1.748,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3972602739726027e-06,
+      "loss": 1.9375,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4657534246575345e-06,
+      "loss": 1.7207,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.534246575342466e-06,
+      "loss": 1.7617,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 1.8184,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.671232876712329e-06,
+      "loss": 1.7285,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7397260273972604e-06,
+      "loss": 1.7578,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8082191780821922e-06,
+      "loss": 1.7617,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 1.6699,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.945205479452055e-06,
+      "loss": 1.3789,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0136986301369864e-06,
+      "loss": 1.7695,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.082191780821918e-06,
+      "loss": 1.6719,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1506849315068495e-06,
+      "loss": 1.4023,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2191780821917813e-06,
+      "loss": 1.5898,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 1.8164,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.356164383561644e-06,
+      "loss": 1.7852,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 1.5898,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4931506849315072e-06,
+      "loss": 1.5117,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5616438356164386e-06,
+      "loss": 1.752,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6301369863013704e-06,
+      "loss": 1.6777,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6986301369863014e-06,
+      "loss": 1.5586,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.767123287671233e-06,
+      "loss": 1.8066,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 1.5586,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.904109589041096e-06,
+      "loss": 1.457,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.972602739726027e-06,
+      "loss": 1.8145,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0410958904109595e-06,
+      "loss": 1.6191,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.109589041095891e-06,
+      "loss": 1.6777,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.178082191780822e-06,
+      "loss": 1.6602,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.246575342465754e-06,
+      "loss": 1.4863,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.315068493150685e-06,
+      "loss": 1.7422,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.383561643835616e-06,
+      "loss": 1.6836,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4520547945205486e-06,
+      "loss": 1.6133,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.52054794520548e-06,
+      "loss": 1.6758,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.589041095890411e-06,
+      "loss": 1.6777,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.657534246575343e-06,
+      "loss": 1.4863,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.726027397260274e-06,
+      "loss": 1.4727,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7945205479452054e-06,
+      "loss": 1.6191,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.863013698630138e-06,
+      "loss": 1.5293,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.931506849315069e-06,
+      "loss": 1.3984,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 1.6699,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.068493150684932e-06,
+      "loss": 1.627,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.136986301369864e-06,
+      "loss": 1.5625,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 1.7012,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.273972602739727e-06,
+      "loss": 1.4297,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.342465753424658e-06,
+      "loss": 1.6934,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.41095890410959e-06,
+      "loss": 1.6875,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.479452054794521e-06,
+      "loss": 1.6836,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.547945205479452e-06,
+      "loss": 1.5703,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6164383561643845e-06,
+      "loss": 1.5332,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.684931506849316e-06,
+      "loss": 1.7012,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 1.623,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.821917808219179e-06,
+      "loss": 1.6035,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.89041095890411e-06,
+      "loss": 1.3711,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.958904109589042e-06,
+      "loss": 1.6641,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.027397260273973e-06,
+      "loss": 1.6367,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.095890410958905e-06,
+      "loss": 1.373,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.164383561643836e-06,
+      "loss": 1.5078,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2328767123287685e-06,
+      "loss": 1.4043,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.301369863013699e-06,
+      "loss": 1.5293,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.36986301369863e-06,
+      "loss": 1.4648,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.438356164383563e-06,
+      "loss": 1.4863,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.506849315068494e-06,
+      "loss": 1.584,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 1.4863,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.643835616438357e-06,
+      "loss": 1.3789,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.712328767123288e-06,
+      "loss": 1.3848,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.78082191780822e-06,
+      "loss": 1.2812,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.849315068493151e-06,
+      "loss": 1.3457,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.917808219178082e-06,
+      "loss": 1.3613,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9863013698630145e-06,
+      "loss": 1.2363,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.054794520547946e-06,
+      "loss": 1.291,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.123287671232877e-06,
+      "loss": 1.4121,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.191780821917809e-06,
+      "loss": 1.3633,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.260273972602741e-06,
+      "loss": 1.3809,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.328767123287672e-06,
+      "loss": 1.2461,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.397260273972603e-06,
+      "loss": 1.3672,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.465753424657535e-06,
+      "loss": 1.1816,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.534246575342466e-06,
+      "loss": 1.1279,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6027397260273985e-06,
+      "loss": 1.3242,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 1.3594,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.739726027397261e-06,
+      "loss": 1.2148,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 1.2695,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.876712328767124e-06,
+      "loss": 1.249,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 1.2598,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.013698630136987e-06,
+      "loss": 1.1191,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.082191780821919e-06,
+      "loss": 1.1719,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.150684931506851e-06,
+      "loss": 1.1865,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.219178082191782e-06,
+      "loss": 1.0244,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.287671232876712e-06,
+      "loss": 1.3008,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.356164383561644e-06,
+      "loss": 1.0752,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.424657534246577e-06,
+      "loss": 1.1348,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.493150684931507e-06,
+      "loss": 1.1553,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.56164383561644e-06,
+      "loss": 1.0557,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.63013698630137e-06,
+      "loss": 1.1641,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.698630136986302e-06,
+      "loss": 1.2031,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.767123287671233e-06,
+      "loss": 1.1289,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.835616438356165e-06,
+      "loss": 1.0752,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.904109589041097e-06,
+      "loss": 1.2383,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.972602739726028e-06,
+      "loss": 1.1328,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.04109589041096e-06,
+      "loss": 1.1523,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.10958904109589e-06,
+      "loss": 1.1113,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.178082191780823e-06,
+      "loss": 1.1191,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.246575342465755e-06,
+      "loss": 1.0605,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.315068493150685e-06,
+      "loss": 0.9746,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.383561643835618e-06,
+      "loss": 1.0068,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 1.1895,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.52054794520548e-06,
+      "loss": 1.0117,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.589041095890411e-06,
+      "loss": 1.0537,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.657534246575343e-06,
+      "loss": 1.0977,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.726027397260275e-06,
+      "loss": 1.0449,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.794520547945206e-06,
+      "loss": 1.0586,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.863013698630138e-06,
+      "loss": 1.0391,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.931506849315069e-06,
+      "loss": 1.0195,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.9971,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0068493150684933e-05,
+      "loss": 0.9678,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 0.9863,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0205479452054796e-05,
+      "loss": 1.0127,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0273972602739728e-05,
+      "loss": 1.0137,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0342465753424657e-05,
+      "loss": 0.9717,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 1.0176,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0479452054794521e-05,
+      "loss": 0.9658,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0547945205479453e-05,
+      "loss": 1.041,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0616438356164384e-05,
+      "loss": 0.9268,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0684931506849316e-05,
+      "loss": 0.8604,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0753424657534248e-05,
+      "loss": 0.9697,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.082191780821918e-05,
+      "loss": 0.9971,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.089041095890411e-05,
+      "loss": 0.9824,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0958904109589042e-05,
+      "loss": 0.9775,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1027397260273974e-05,
+      "loss": 0.916,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1095890410958904e-05,
+      "loss": 0.9209,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1164383561643837e-05,
+      "loss": 0.9795,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1232876712328769e-05,
+      "loss": 0.9424,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1301369863013701e-05,
+      "loss": 0.9336,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1369863013698632e-05,
+      "loss": 0.9219,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1438356164383562e-05,
+      "loss": 0.9482,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.9131,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1575342465753425e-05,
+      "loss": 0.8848,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1643835616438357e-05,
+      "loss": 0.8916,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.171232876712329e-05,
+      "loss": 0.7852,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.178082191780822e-05,
+      "loss": 0.917,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1849315068493152e-05,
+      "loss": 0.9258,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1917808219178084e-05,
+      "loss": 0.8477,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1986301369863013e-05,
+      "loss": 0.8477,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2054794520547945e-05,
+      "loss": 0.9102,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2123287671232878e-05,
+      "loss": 0.8057,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.219178082191781e-05,
+      "loss": 0.8896,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.226027397260274e-05,
+      "loss": 0.8379,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2328767123287673e-05,
+      "loss": 0.8564,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2397260273972605e-05,
+      "loss": 0.8691,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2465753424657537e-05,
+      "loss": 0.8984,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2534246575342466e-05,
+      "loss": 0.8975,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2602739726027398e-05,
+      "loss": 0.8398,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2671232876712329e-05,
+      "loss": 0.8398,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.273972602739726e-05,
+      "loss": 0.8359,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2808219178082193e-05,
+      "loss": 0.8076,
+      "step": 187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2876712328767125e-05,
+      "loss": 0.7197,
+      "step": 188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2945205479452056e-05,
+      "loss": 0.8936,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3013698630136988e-05,
+      "loss": 0.8818,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3082191780821919e-05,
+      "loss": 0.8066,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.8584,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3219178082191781e-05,
+      "loss": 0.7676,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3287671232876714e-05,
+      "loss": 0.877,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3356164383561646e-05,
+      "loss": 0.7695,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3424657534246576e-05,
+      "loss": 0.7842,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3493150684931508e-05,
+      "loss": 0.6631,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.356164383561644e-05,
+      "loss": 0.7969,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.363013698630137e-05,
+      "loss": 0.8496,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3698630136986302e-05,
+      "loss": 0.7725,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3767123287671234e-05,
+      "loss": 0.7949,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3835616438356164e-05,
+      "loss": 0.748,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3904109589041097e-05,
+      "loss": 0.7637,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3972602739726029e-05,
+      "loss": 0.7793,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4041095890410961e-05,
+      "loss": 0.7529,
+      "step": 205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4109589041095892e-05,
+      "loss": 0.8115,
+      "step": 206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4178082191780822e-05,
+      "loss": 0.7744,
+      "step": 207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4246575342465754e-05,
+      "loss": 0.7568,
+      "step": 208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4315068493150685e-05,
+      "loss": 0.7744,
+      "step": 209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4383561643835617e-05,
+      "loss": 0.7764,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.445205479452055e-05,
+      "loss": 0.7432,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4520547945205482e-05,
+      "loss": 0.7822,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4589041095890412e-05,
+      "loss": 0.7871,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4657534246575344e-05,
+      "loss": 0.7812,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4726027397260275e-05,
+      "loss": 0.7266,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4794520547945205e-05,
+      "loss": 0.6777,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4863013698630138e-05,
+      "loss": 0.7002,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.493150684931507e-05,
+      "loss": 0.7959,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7432,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5068493150684933e-05,
+      "loss": 0.7266,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5136986301369865e-05,
+      "loss": 0.7402,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5205479452054797e-05,
+      "loss": 0.7646,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5273972602739728e-05,
+      "loss": 0.7422,
+      "step": 223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 0.7217,
+      "step": 224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.541095890410959e-05,
+      "loss": 0.6621,
+      "step": 225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5479452054794523e-05,
+      "loss": 0.6787,
+      "step": 226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5547945205479453e-05,
+      "loss": 0.7568,
+      "step": 227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.7412,
+      "step": 228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5684931506849318e-05,
+      "loss": 0.7461,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5753424657534248e-05,
+      "loss": 0.6855,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5821917808219182e-05,
+      "loss": 0.708,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.6553,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5958904109589043e-05,
+      "loss": 0.6797,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6027397260273974e-05,
+      "loss": 0.6797,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6095890410958904e-05,
+      "loss": 0.7363,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6164383561643838e-05,
+      "loss": 0.7334,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.623287671232877e-05,
+      "loss": 0.6904,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6301369863013702e-05,
+      "loss": 0.6611,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6369863013698633e-05,
+      "loss": 0.7246,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6438356164383563e-05,
+      "loss": 0.6167,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6506849315068494e-05,
+      "loss": 0.6338,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6575342465753425e-05,
+      "loss": 0.6582,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.664383561643836e-05,
+      "loss": 0.6416,
+      "step": 243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.671232876712329e-05,
+      "loss": 0.5791,
+      "step": 244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.678082191780822e-05,
+      "loss": 0.6201,
+      "step": 245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6849315068493153e-05,
+      "loss": 0.6592,
+      "step": 246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6917808219178084e-05,
+      "loss": 0.6455,
+      "step": 247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6986301369863014e-05,
+      "loss": 0.666,
+      "step": 248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7054794520547945e-05,
+      "loss": 0.6172,
+      "step": 249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.712328767123288e-05,
+      "loss": 0.6162,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.719178082191781e-05,
+      "loss": 0.5967,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.726027397260274e-05,
+      "loss": 0.5381,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7328767123287674e-05,
+      "loss": 0.5986,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7397260273972604e-05,
+      "loss": 0.582,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7465753424657538e-05,
+      "loss": 0.6309,
+      "step": 255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7534246575342465e-05,
+      "loss": 0.5625,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.76027397260274e-05,
+      "loss": 0.6045,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.767123287671233e-05,
+      "loss": 0.6309,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.773972602739726e-05,
+      "loss": 0.6025,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7808219178082194e-05,
+      "loss": 0.5977,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7876712328767125e-05,
+      "loss": 0.5771,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7945205479452055e-05,
+      "loss": 0.5234,
+      "step": 262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.801369863013699e-05,
+      "loss": 0.5566,
+      "step": 263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.808219178082192e-05,
+      "loss": 0.519,
+      "step": 264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.815068493150685e-05,
+      "loss": 0.5723,
+      "step": 265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.821917808219178e-05,
+      "loss": 0.5703,
+      "step": 266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8287671232876715e-05,
+      "loss": 0.6006,
+      "step": 267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8356164383561645e-05,
+      "loss": 0.4883,
+      "step": 268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8424657534246576e-05,
+      "loss": 0.5869,
+      "step": 269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.849315068493151e-05,
+      "loss": 0.5439,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.856164383561644e-05,
+      "loss": 0.5752,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.863013698630137e-05,
+      "loss": 0.5132,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.86986301369863e-05,
+      "loss": 0.4878,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8767123287671235e-05,
+      "loss": 0.4932,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8835616438356166e-05,
+      "loss": 0.5742,
+      "step": 275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 0.5083,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.897260273972603e-05,
+      "loss": 0.5103,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904109589041096e-05,
+      "loss": 0.5122,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910958904109589e-05,
+      "loss": 0.5508,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9178082191780822e-05,
+      "loss": 0.4565,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9246575342465756e-05,
+      "loss": 0.5664,
+      "step": 281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315068493150686e-05,
+      "loss": 0.4688,
+      "step": 282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9383561643835617e-05,
+      "loss": 0.5259,
+      "step": 283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.945205479452055e-05,
+      "loss": 0.5381,
+      "step": 284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.952054794520548e-05,
+      "loss": 0.5381,
+      "step": 285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9589041095890412e-05,
+      "loss": 0.5288,
+      "step": 286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9657534246575346e-05,
+      "loss": 0.5098,
+      "step": 287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9726027397260276e-05,
+      "loss": 0.5063,
+      "step": 288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9794520547945207e-05,
+      "loss": 0.5205,
+      "step": 289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9863013698630137e-05,
+      "loss": 0.4995,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.993150684931507e-05,
+      "loss": 0.4878,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.4839,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999999444824108e-05,
+      "loss": 0.4434,
+      "step": 293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997779296502e-05,
+      "loss": 0.4619,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995003417356e-05,
+      "loss": 0.4722,
+      "step": 295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999111718698e-05,
+      "loss": 0.4966,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999986120605816e-05,
+      "loss": 0.4595,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999998001367441e-05,
+      "loss": 0.4062,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997279639344e-05,
+      "loss": 0.4785,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999996446876371e-05,
+      "loss": 0.4346,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999955030786143e-05,
+      "loss": 0.4346,
+      "step": 301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999994448246179e-05,
+      "loss": 0.4463,
+      "step": 302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999932823791816e-05,
+      "loss": 0.4199,
+      "step": 303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999920054777522e-05,
+      "loss": 0.4272,
+      "step": 304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999990617542032e-05,
+      "loss": 0.4849,
+      "step": 305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999989118572176e-05,
+      "loss": 0.4346,
+      "step": 306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999875085683498e-05,
+      "loss": 0.4287,
+      "step": 307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999857875307324e-05,
+      "loss": 0.4355,
+      "step": 308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999839554595152e-05,
+      "loss": 0.4517,
+      "step": 309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999820123549014e-05,
+      "loss": 0.436,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999799582171066e-05,
+      "loss": 0.4209,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999977793046359e-05,
+      "loss": 0.4521,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999755168428986e-05,
+      "loss": 0.479,
+      "step": 313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999731296069788e-05,
+      "loss": 0.4863,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999706313388645e-05,
+      "loss": 0.4062,
+      "step": 315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999968022038833e-05,
+      "loss": 0.4185,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999965301707174e-05,
+      "loss": 0.4438,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99996247034419e-05,
+      "loss": 0.4023,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999595279501944e-05,
+      "loss": 0.3687,
+      "step": 319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999564745255148e-05,
+      "loss": 0.4385,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99995331007049e-05,
+      "loss": 0.4131,
+      "step": 321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999950034585471e-05,
+      "loss": 0.4282,
+      "step": 322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946648070822e-05,
+      "loss": 0.437,
+      "step": 323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999431505269185e-05,
+      "loss": 0.4463,
+      "step": 324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999395419541494e-05,
+      "loss": 0.3965,
+      "step": 325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999935822352915e-05,
+      "loss": 0.3955,
+      "step": 326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999319917236287e-05,
+      "loss": 0.3672,
+      "step": 327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999280500667154e-05,
+      "loss": 0.4028,
+      "step": 328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999923997382613e-05,
+      "loss": 0.4136,
+      "step": 329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999198336717712e-05,
+      "loss": 0.4033,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999155589346528e-05,
+      "loss": 0.4966,
+      "step": 331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999911173171732e-05,
+      "loss": 0.4028,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999906676383496e-05,
+      "loss": 0.4517,
+      "step": 333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999902068570444e-05,
+      "loss": 0.3838,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998973497330878e-05,
+      "loss": 0.4546,
+      "step": 335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998925198719514e-05,
+      "loss": 0.376,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998875789875707e-05,
+      "loss": 0.3472,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999882527080495e-05,
+      "loss": 0.4165,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998773641512842e-05,
+      "loss": 0.3857,
+      "step": 339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998720902005125e-05,
+      "loss": 0.4126,
+      "step": 340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998667052287647e-05,
+      "loss": 0.4233,
+      "step": 341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998612092366396e-05,
+      "loss": 0.3818,
+      "step": 342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998556022247468e-05,
+      "loss": 0.4209,
+      "step": 343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999849884193709e-05,
+      "loss": 0.4263,
+      "step": 344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998440551441618e-05,
+      "loss": 0.4282,
+      "step": 345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998381150767514e-05,
+      "loss": 0.3643,
+      "step": 346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998320639921377e-05,
+      "loss": 0.3833,
+      "step": 347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999825901890993e-05,
+      "loss": 0.3979,
+      "step": 348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998196287740006e-05,
+      "loss": 0.3696,
+      "step": 349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998132446418583e-05,
+      "loss": 0.3809,
+      "step": 350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998067494952736e-05,
+      "loss": 0.3687,
+      "step": 351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999800143334969e-05,
+      "loss": 0.395,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997934261616768e-05,
+      "loss": 0.4014,
+      "step": 353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997865979761436e-05,
+      "loss": 0.354,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997796587791276e-05,
+      "loss": 0.4121,
+      "step": 355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997726085713993e-05,
+      "loss": 0.4043,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999765447353741e-05,
+      "loss": 0.4009,
+      "step": 357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999758175126948e-05,
+      "loss": 0.3726,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999750791891828e-05,
+      "loss": 0.416,
+      "step": 359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997432976492006e-05,
+      "loss": 0.3486,
+      "step": 360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999735692399898e-05,
+      "loss": 0.3442,
+      "step": 361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997279761447652e-05,
+      "loss": 0.3354,
+      "step": 362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997201488846585e-05,
+      "loss": 0.3594,
+      "step": 363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997122106204466e-05,
+      "loss": 0.3579,
+      "step": 364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997041613530115e-05,
+      "loss": 0.375,
+      "step": 365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996960010832466e-05,
+      "loss": 0.3398,
+      "step": 366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996877298120583e-05,
+      "loss": 0.3672,
+      "step": 367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996793475403647e-05,
+      "loss": 0.436,
+      "step": 368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996708542690966e-05,
+      "loss": 0.3652,
+      "step": 369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996622499991973e-05,
+      "loss": 0.3794,
+      "step": 370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996535347316217e-05,
+      "loss": 0.3364,
+      "step": 371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996447084673383e-05,
+      "loss": 0.3599,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996357712073263e-05,
+      "loss": 0.377,
+      "step": 373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996267229525783e-05,
+      "loss": 0.3696,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999617563704099e-05,
+      "loss": 0.4487,
+      "step": 375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996082934629057e-05,
+      "loss": 0.3779,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995989122300275e-05,
+      "loss": 0.3643,
+      "step": 377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995894200065055e-05,
+      "loss": 0.3503,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995798167933945e-05,
+      "loss": 0.3398,
+      "step": 379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995701025917607e-05,
+      "loss": 0.3672,
+      "step": 380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995602774026826e-05,
+      "loss": 0.3667,
+      "step": 381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995503412272504e-05,
+      "loss": 0.3535,
+      "step": 382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995402940665684e-05,
+      "loss": 0.3228,
+      "step": 383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999530135921752e-05,
+      "loss": 0.3682,
+      "step": 384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995198667939285e-05,
+      "loss": 0.3188,
+      "step": 385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995094866842386e-05,
+      "loss": 0.3516,
+      "step": 386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994989955938352e-05,
+      "loss": 0.3345,
+      "step": 387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994883935238822e-05,
+      "loss": 0.3872,
+      "step": 388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994776804755576e-05,
+      "loss": 0.3633,
+      "step": 389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999466856450051e-05,
+      "loss": 0.3633,
+      "step": 390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994559214485637e-05,
+      "loss": 0.3286,
+      "step": 391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994448754723098e-05,
+      "loss": 0.303,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994337185225164e-05,
+      "loss": 0.4048,
+      "step": 393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999422450600422e-05,
+      "loss": 0.3716,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999411071707278e-05,
+      "loss": 0.354,
+      "step": 395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993995818443473e-05,
+      "loss": 0.3613,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999387981012906e-05,
+      "loss": 0.3818,
+      "step": 397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993762692142423e-05,
+      "loss": 0.3247,
+      "step": 398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993644464496566e-05,
+      "loss": 0.3662,
+      "step": 399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993525127204615e-05,
+      "loss": 0.3789,
+      "step": 400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999340468027982e-05,
+      "loss": 0.4297,
+      "step": 401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999328312373556e-05,
+      "loss": 0.2925,
+      "step": 402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993160457585325e-05,
+      "loss": 0.3105,
+      "step": 403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993036681842737e-05,
+      "loss": 0.3643,
+      "step": 404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992911796521543e-05,
+      "loss": 0.3218,
+      "step": 405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999278580163561e-05,
+      "loss": 0.2988,
+      "step": 406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992658697198922e-05,
+      "loss": 0.3135,
+      "step": 407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992530483225596e-05,
+      "loss": 0.3516,
+      "step": 408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992401159729866e-05,
+      "loss": 0.3091,
+      "step": 409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992270726726096e-05,
+      "loss": 0.3521,
+      "step": 410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992139184228766e-05,
+      "loss": 0.3279,
+      "step": 411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999200653225248e-05,
+      "loss": 0.3477,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999187277081197e-05,
+      "loss": 0.4175,
+      "step": 413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991737899922086e-05,
+      "loss": 0.3301,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991601919597802e-05,
+      "loss": 0.2893,
+      "step": 415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991464829854223e-05,
+      "loss": 0.3098,
+      "step": 416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991326630706564e-05,
+      "loss": 0.394,
+      "step": 417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991187322170175e-05,
+      "loss": 0.3623,
+      "step": 418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999104690426052e-05,
+      "loss": 0.3438,
+      "step": 419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999090537699319e-05,
+      "loss": 0.3447,
+      "step": 420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990762740383908e-05,
+      "loss": 0.3442,
+      "step": 421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990618994448502e-05,
+      "loss": 0.3403,
+      "step": 422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990474139202936e-05,
+      "loss": 0.2891,
+      "step": 423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990328174663297e-05,
+      "loss": 0.3076,
+      "step": 424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990181100845786e-05,
+      "loss": 0.3062,
+      "step": 425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999003291776674e-05,
+      "loss": 0.3291,
+      "step": 426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989883625442603e-05,
+      "loss": 0.3872,
+      "step": 427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989733223889964e-05,
+      "loss": 0.3379,
+      "step": 428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989581713125516e-05,
+      "loss": 0.2891,
+      "step": 429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998942909316608e-05,
+      "loss": 0.3176,
+      "step": 430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989275364028608e-05,
+      "loss": 0.3398,
+      "step": 431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989120525730165e-05,
+      "loss": 0.3901,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998896457828795e-05,
+      "loss": 0.3394,
+      "step": 433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988807521719264e-05,
+      "loss": 0.3696,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988649356041562e-05,
+      "loss": 0.3291,
+      "step": 435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988490081272397e-05,
+      "loss": 0.3145,
+      "step": 436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998832969742946e-05,
+      "loss": 0.3447,
+      "step": 437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998816820453055e-05,
+      "loss": 0.3418,
+      "step": 438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998800560259361e-05,
+      "loss": 0.3276,
+      "step": 439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987841891636687e-05,
+      "loss": 0.3506,
+      "step": 440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998767707167796e-05,
+      "loss": 0.3452,
+      "step": 441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998751114273573e-05,
+      "loss": 0.3682,
+      "step": 442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998734410482842e-05,
+      "loss": 0.3179,
+      "step": 443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987175957974577e-05,
+      "loss": 0.354,
+      "step": 444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987006702192875e-05,
+      "loss": 0.3071,
+      "step": 445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986836337502106e-05,
+      "loss": 0.2764,
+      "step": 446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986664863921183e-05,
+      "loss": 0.311,
+      "step": 447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986492281469147e-05,
+      "loss": 0.3311,
+      "step": 448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986318590165162e-05,
+      "loss": 0.3052,
+      "step": 449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986143790028513e-05,
+      "loss": 0.3184,
+      "step": 450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998596788107861e-05,
+      "loss": 0.3157,
+      "step": 451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985790863334988e-05,
+      "loss": 0.2825,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985612736817295e-05,
+      "loss": 0.313,
+      "step": 453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985433501545312e-05,
+      "loss": 0.3467,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998525315753894e-05,
+      "loss": 0.3262,
+      "step": 455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998507170481821e-05,
+      "loss": 0.2844,
+      "step": 456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984889143403263e-05,
+      "loss": 0.3486,
+      "step": 457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998470547331437e-05,
+      "loss": 0.3337,
+      "step": 458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984520694571926e-05,
+      "loss": 0.3525,
+      "step": 459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998433480719645e-05,
+      "loss": 0.2629,
+      "step": 460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984147811208576e-05,
+      "loss": 0.3384,
+      "step": 461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983959706629076e-05,
+      "loss": 0.2861,
+      "step": 462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983770493478828e-05,
+      "loss": 0.314,
+      "step": 463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983580171778846e-05,
+      "loss": 0.312,
+      "step": 464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998338874155026e-05,
+      "loss": 0.2419,
+      "step": 465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998319620281433e-05,
+      "loss": 0.3428,
+      "step": 466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983002555592428e-05,
+      "loss": 0.2485,
+      "step": 467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982807799906057e-05,
+      "loss": 0.2761,
+      "step": 468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982611935776847e-05,
+      "loss": 0.29,
+      "step": 469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982414963226544e-05,
+      "loss": 0.3267,
+      "step": 470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982216882277013e-05,
+      "loss": 0.2598,
+      "step": 471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982017692950254e-05,
+      "loss": 0.3296,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981817395268384e-05,
+      "loss": 0.2864,
+      "step": 473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998161598925364e-05,
+      "loss": 0.3491,
+      "step": 474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981413474928386e-05,
+      "loss": 0.2734,
+      "step": 475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981209852315108e-05,
+      "loss": 0.3398,
+      "step": 476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9981005121436418e-05,
+      "loss": 0.3257,
+      "step": 477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980799282315045e-05,
+      "loss": 0.3726,
+      "step": 478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980592334973847e-05,
+      "loss": 0.3018,
+      "step": 479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99803842794358e-05,
+      "loss": 0.3081,
+      "step": 480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980175115724007e-05,
+      "loss": 0.3364,
+      "step": 481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979964843861693e-05,
+      "loss": 0.3281,
+      "step": 482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979753463872203e-05,
+      "loss": 0.2954,
+      "step": 483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997954097577901e-05,
+      "loss": 0.272,
+      "step": 484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979327379605707e-05,
+      "loss": 0.3481,
+      "step": 485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979112675376014e-05,
+      "loss": 0.333,
+      "step": 486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997889686311376e-05,
+      "loss": 0.3247,
+      "step": 487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978679942842922e-05,
+      "loss": 0.3159,
+      "step": 488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978461914587577e-05,
+      "loss": 0.2373,
+      "step": 489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978242778371934e-05,
+      "loss": 0.3262,
+      "step": 490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997802253422033e-05,
+      "loss": 0.3071,
+      "step": 491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977801182157213e-05,
+      "loss": 0.2817,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997757872220717e-05,
+      "loss": 0.3286,
+      "step": 493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977355154394885e-05,
+      "loss": 0.2949,
+      "step": 494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99771304787452e-05,
+      "loss": 0.3071,
+      "step": 495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976904695283057e-05,
+      "loss": 0.2876,
+      "step": 496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997667780403352e-05,
+      "loss": 0.2935,
+      "step": 497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976449805021788e-05,
+      "loss": 0.2988,
+      "step": 498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976220698273177e-05,
+      "loss": 0.3022,
+      "step": 499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997599048381312e-05,
+      "loss": 0.2886,
+      "step": 500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975759161667182e-05,
+      "loss": 0.3164,
+      "step": 501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997552673186105e-05,
+      "loss": 0.2979,
+      "step": 502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975293194420532e-05,
+      "loss": 0.3115,
+      "step": 503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975058549371553e-05,
+      "loss": 0.3096,
+      "step": 504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974822796740174e-05,
+      "loss": 0.3074,
+      "step": 505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974585936552565e-05,
+      "loss": 0.3379,
+      "step": 506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974347968835036e-05,
+      "loss": 0.2979,
+      "step": 507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974108893614e-05,
+      "loss": 0.29,
+      "step": 508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973868710916004e-05,
+      "loss": 0.2769,
+      "step": 509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997362742076772e-05,
+      "loss": 0.3164,
+      "step": 510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973385023195943e-05,
+      "loss": 0.3354,
+      "step": 511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997314151822758e-05,
+      "loss": 0.3008,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972896905889674e-05,
+      "loss": 0.2793,
+      "step": 513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997265118620938e-05,
+      "loss": 0.241,
+      "step": 514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972404359213987e-05,
+      "loss": 0.272,
+      "step": 515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972156424930898e-05,
+      "loss": 0.2983,
+      "step": 516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971907383387644e-05,
+      "loss": 0.2651,
+      "step": 517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997165723461188e-05,
+      "loss": 0.3254,
+      "step": 518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971405978631378e-05,
+      "loss": 0.3208,
+      "step": 519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971153615474036e-05,
+      "loss": 0.2534,
+      "step": 520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970900145167877e-05,
+      "loss": 0.2795,
+      "step": 521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997064556774104e-05,
+      "loss": 0.3398,
+      "step": 522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99703898832218e-05,
+      "loss": 0.3257,
+      "step": 523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997013309163854e-05,
+      "loss": 0.2964,
+      "step": 524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969875193019783e-05,
+      "loss": 0.2744,
+      "step": 525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996961618739415e-05,
+      "loss": 0.2754,
+      "step": 526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969356074790412e-05,
+      "loss": 0.2837,
+      "step": 527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969094855237446e-05,
+      "loss": 0.2275,
+      "step": 528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968832528764256e-05,
+      "loss": 0.3066,
+      "step": 529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996856909539997e-05,
+      "loss": 0.3145,
+      "step": 530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996830455517384e-05,
+      "loss": 0.3184,
+      "step": 531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968038908115237e-05,
+      "loss": 0.2512,
+      "step": 532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996777215425366e-05,
+      "loss": 0.3652,
+      "step": 533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967504293618725e-05,
+      "loss": 0.2527,
+      "step": 534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967235326240174e-05,
+      "loss": 0.3572,
+      "step": 535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966965252147873e-05,
+      "loss": 0.2563,
+      "step": 536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996669407137181e-05,
+      "loss": 0.2964,
+      "step": 537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966421783942094e-05,
+      "loss": 0.3169,
+      "step": 538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966148389888964e-05,
+      "loss": 0.2646,
+      "step": 539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965873889242768e-05,
+      "loss": 0.2832,
+      "step": 540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965598282033994e-05,
+      "loss": 0.3411,
+      "step": 541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965321568293233e-05,
+      "loss": 0.3047,
+      "step": 542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965043748051222e-05,
+      "loss": 0.2944,
+      "step": 543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99647648213388e-05,
+      "loss": 0.3145,
+      "step": 544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996448478818694e-05,
+      "loss": 0.3081,
+      "step": 545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996420364862674e-05,
+      "loss": 0.2366,
+      "step": 546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963921402689412e-05,
+      "loss": 0.2786,
+      "step": 547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963638050406297e-05,
+      "loss": 0.3472,
+      "step": 548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963353591808853e-05,
+      "loss": 0.2676,
+      "step": 549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963068026928673e-05,
+      "loss": 0.2747,
+      "step": 550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962781355797456e-05,
+      "loss": 0.3042,
+      "step": 551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962493578447038e-05,
+      "loss": 0.3423,
+      "step": 552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962204694909368e-05,
+      "loss": 0.3005,
+      "step": 553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9961914705216526e-05,
+      "loss": 0.3398,
+      "step": 554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961623609400712e-05,
+      "loss": 0.238,
+      "step": 555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961331407494245e-05,
+      "loss": 0.3018,
+      "step": 556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996103809952957e-05,
+      "loss": 0.2993,
+      "step": 557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996074368553926e-05,
+      "loss": 0.3027,
+      "step": 558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960448165555992e-05,
+      "loss": 0.2502,
+      "step": 559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960151539612593e-05,
+      "loss": 0.2915,
+      "step": 560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959853807741992e-05,
+      "loss": 0.3291,
+      "step": 561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995955496997725e-05,
+      "loss": 0.3042,
+      "step": 562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959255026351548e-05,
+      "loss": 0.2876,
+      "step": 563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995895397689819e-05,
+      "loss": 0.3105,
+      "step": 564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958651821650604e-05,
+      "loss": 0.2842,
+      "step": 565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958348560642336e-05,
+      "loss": 0.269,
+      "step": 566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958044193907063e-05,
+      "loss": 0.2712,
+      "step": 567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995773872147858e-05,
+      "loss": 0.282,
+      "step": 568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99574321433908e-05,
+      "loss": 0.3313,
+      "step": 569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957124459677772e-05,
+      "loss": 0.3315,
+      "step": 570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956815670373655e-05,
+      "loss": 0.291,
+      "step": 571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995650577551274e-05,
+      "loss": 0.2756,
+      "step": 572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956194775129426e-05,
+      "loss": 0.2812,
+      "step": 573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955882669258256e-05,
+      "loss": 0.3098,
+      "step": 574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995556945793388e-05,
+      "loss": 0.2429,
+      "step": 575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955255141191074e-05,
+      "loss": 0.2793,
+      "step": 576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995493971906474e-05,
+      "loss": 0.2368,
+      "step": 577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99546231915899e-05,
+      "loss": 0.3203,
+      "step": 578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954305558801703e-05,
+      "loss": 0.261,
+      "step": 579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953986820735414e-05,
+      "loss": 0.2544,
+      "step": 580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953666977426428e-05,
+      "loss": 0.2876,
+      "step": 581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995334602891025e-05,
+      "loss": 0.2302,
+      "step": 582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953023975222527e-05,
+      "loss": 0.2661,
+      "step": 583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952700816399014e-05,
+      "loss": 0.2578,
+      "step": 584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995237655247559e-05,
+      "loss": 0.2708,
+      "step": 585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952051183488262e-05,
+      "loss": 0.3115,
+      "step": 586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951724709473163e-05,
+      "loss": 0.2651,
+      "step": 587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951397130466535e-05,
+      "loss": 0.311,
+      "step": 588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951068446504753e-05,
+      "loss": 0.2966,
+      "step": 589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950738657624318e-05,
+      "loss": 0.2881,
+      "step": 590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950407763861837e-05,
+      "loss": 0.2654,
+      "step": 591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950075765254063e-05,
+      "loss": 0.2451,
+      "step": 592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994974266183785e-05,
+      "loss": 0.2612,
+      "step": 593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949408453650194e-05,
+      "loss": 0.259,
+      "step": 594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994907314072819e-05,
+      "loss": 0.2791,
+      "step": 595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948736723109082e-05,
+      "loss": 0.2715,
+      "step": 596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994839920083022e-05,
+      "loss": 0.2866,
+      "step": 597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948060573929075e-05,
+      "loss": 0.2998,
+      "step": 598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947720842443255e-05,
+      "loss": 0.2725,
+      "step": 599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947380006410484e-05,
+      "loss": 0.3228,
+      "step": 600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947038065868597e-05,
+      "loss": 0.2817,
+      "step": 601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994669502085557e-05,
+      "loss": 0.2644,
+      "step": 602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946350871409484e-05,
+      "loss": 0.2993,
+      "step": 603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946005617568563e-05,
+      "loss": 0.2786,
+      "step": 604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945659259371133e-05,
+      "loss": 0.2476,
+      "step": 605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945311796855654e-05,
+      "loss": 0.2812,
+      "step": 606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944963230060713e-05,
+      "loss": 0.2764,
+      "step": 607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944613559025005e-05,
+      "loss": 0.3035,
+      "step": 608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994426278378736e-05,
+      "loss": 0.272,
+      "step": 609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943910904386725e-05,
+      "loss": 0.2397,
+      "step": 610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994355792086217e-05,
+      "loss": 0.3169,
+      "step": 611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994320383325289e-05,
+      "loss": 0.3091,
+      "step": 612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942848641598204e-05,
+      "loss": 0.2617,
+      "step": 613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942492345937545e-05,
+      "loss": 0.2114,
+      "step": 614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942134946310477e-05,
+      "loss": 0.3477,
+      "step": 615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941776442756685e-05,
+      "loss": 0.2727,
+      "step": 616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941416835315977e-05,
+      "loss": 0.3047,
+      "step": 617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941056124028277e-05,
+      "loss": 0.2854,
+      "step": 618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940694308933638e-05,
+      "loss": 0.2761,
+      "step": 619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994033139007224e-05,
+      "loss": 0.314,
+      "step": 620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939967367484372e-05,
+      "loss": 0.2483,
+      "step": 621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939602241210457e-05,
+      "loss": 0.2537,
+      "step": 622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993923601129104e-05,
+      "loss": 0.2993,
+      "step": 623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938868677766778e-05,
+      "loss": 0.2844,
+      "step": 624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993850024067846e-05,
+      "loss": 0.2629,
+      "step": 625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938130700067005e-05,
+      "loss": 0.2986,
+      "step": 626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937760055973433e-05,
+      "loss": 0.3022,
+      "step": 627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99373883084389e-05,
+      "loss": 0.2668,
+      "step": 628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993701545750469e-05,
+      "loss": 0.2532,
+      "step": 629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936641503212195e-05,
+      "loss": 0.292,
+      "step": 630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936266445602944e-05,
+      "loss": 0.2686,
+      "step": 631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935890284718574e-05,
+      "loss": 0.2886,
+      "step": 632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993551302060086e-05,
+      "loss": 0.2832,
+      "step": 633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993513465329169e-05,
+      "loss": 0.2666,
+      "step": 634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993475518283307e-05,
+      "loss": 0.2561,
+      "step": 635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993437460926714e-05,
+      "loss": 0.2832,
+      "step": 636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933992932636154e-05,
+      "loss": 0.2607,
+      "step": 637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993361015298249e-05,
+      "loss": 0.2673,
+      "step": 638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993322627034866e-05,
+      "loss": 0.2783,
+      "step": 639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932841284777277e-05,
+      "loss": 0.2434,
+      "step": 640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932455196311093e-05,
+      "loss": 0.2944,
+      "step": 641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993206800499298e-05,
+      "loss": 0.3088,
+      "step": 642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931679710865923e-05,
+      "loss": 0.2703,
+      "step": 643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931290313973043e-05,
+      "loss": 0.2927,
+      "step": 644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993089981435757e-05,
+      "loss": 0.2864,
+      "step": 645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930508212062874e-05,
+      "loss": 0.2219,
+      "step": 646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930115507132424e-05,
+      "loss": 0.2566,
+      "step": 647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929721699609828e-05,
+      "loss": 0.2332,
+      "step": 648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929326789538818e-05,
+      "loss": 0.29,
+      "step": 649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992893077696324e-05,
+      "loss": 0.259,
+      "step": 650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928533661927064e-05,
+      "loss": 0.2893,
+      "step": 651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928135444474382e-05,
+      "loss": 0.3035,
+      "step": 652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927736124649413e-05,
+      "loss": 0.2458,
+      "step": 653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927335702496496e-05,
+      "loss": 0.2959,
+      "step": 654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992693417806009e-05,
+      "loss": 0.2285,
+      "step": 655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992653155138478e-05,
+      "loss": 0.23,
+      "step": 656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926127822515266e-05,
+      "loss": 0.2351,
+      "step": 657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925722991496386e-05,
+      "loss": 0.2937,
+      "step": 658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925317058373086e-05,
+      "loss": 0.2383,
+      "step": 659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924910023190434e-05,
+      "loss": 0.2866,
+      "step": 660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924501885993635e-05,
+      "loss": 0.2747,
+      "step": 661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924092646828e-05,
+      "loss": 0.3057,
+      "step": 662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923682305738966e-05,
+      "loss": 0.2812,
+      "step": 663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923270862772104e-05,
+      "loss": 0.2881,
+      "step": 664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992285831797309e-05,
+      "loss": 0.28,
+      "step": 665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992244467138774e-05,
+      "loss": 0.2771,
+      "step": 666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922029923061973e-05,
+      "loss": 0.301,
+      "step": 667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921614073041847e-05,
+      "loss": 0.249,
+      "step": 668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992119712137354e-05,
+      "loss": 0.2661,
+      "step": 669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920779068103336e-05,
+      "loss": 0.2444,
+      "step": 670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920359913277667e-05,
+      "loss": 0.261,
+      "step": 671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919939656943062e-05,
+      "loss": 0.281,
+      "step": 672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919518299146196e-05,
+      "loss": 0.2463,
+      "step": 673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919095839933846e-05,
+      "loss": 0.2671,
+      "step": 674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918672279352923e-05,
+      "loss": 0.2454,
+      "step": 675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918247617450454e-05,
+      "loss": 0.2383,
+      "step": 676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917821854273597e-05,
+      "loss": 0.2368,
+      "step": 677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991739498986962e-05,
+      "loss": 0.2383,
+      "step": 678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916967024285928e-05,
+      "loss": 0.2837,
+      "step": 679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916537957570035e-05,
+      "loss": 0.2827,
+      "step": 680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916107789769583e-05,
+      "loss": 0.2712,
+      "step": 681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915676520932334e-05,
+      "loss": 0.2297,
+      "step": 682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915244151106177e-05,
+      "loss": 0.2812,
+      "step": 683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991481068033912e-05,
+      "loss": 0.2451,
+      "step": 684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914376108679295e-05,
+      "loss": 0.2786,
+      "step": 685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991394043617495e-05,
+      "loss": 0.322,
+      "step": 686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913503662874462e-05,
+      "loss": 0.3059,
+      "step": 687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991306578882633e-05,
+      "loss": 0.2727,
+      "step": 688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912626814079172e-05,
+      "loss": 0.271,
+      "step": 689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991218673868173e-05,
+      "loss": 0.3032,
+      "step": 690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911745562682866e-05,
+      "loss": 0.2554,
+      "step": 691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911303286131574e-05,
+      "loss": 0.2698,
+      "step": 692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991085990907695e-05,
+      "loss": 0.2966,
+      "step": 693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910415431568233e-05,
+      "loss": 0.3042,
+      "step": 694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990996985365477e-05,
+      "loss": 0.229,
+      "step": 695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990952317538604e-05,
+      "loss": 0.2505,
+      "step": 696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909075396811643e-05,
+      "loss": 0.2512,
+      "step": 697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990862651798129e-05,
+      "loss": 0.2664,
+      "step": 698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990817653894483e-05,
+      "loss": 0.2756,
+      "step": 699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990772545975222e-05,
+      "loss": 0.2983,
+      "step": 700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990727328045355e-05,
+      "loss": 0.3289,
+      "step": 701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906820001099024e-05,
+      "loss": 0.259,
+      "step": 702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906365621738975e-05,
+      "loss": 0.3237,
+      "step": 703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905910142423853e-05,
+      "loss": 0.2896,
+      "step": 704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905453563204237e-05,
+      "loss": 0.2583,
+      "step": 705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990499588413082e-05,
+      "loss": 0.2539,
+      "step": 706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990453710525442e-05,
+      "loss": 0.2578,
+      "step": 707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904077226625978e-05,
+      "loss": 0.2595,
+      "step": 708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990361624829656e-05,
+      "loss": 0.2869,
+      "step": 709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990315417031734e-05,
+      "loss": 0.248,
+      "step": 710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990269099273964e-05,
+      "loss": 0.2607,
+      "step": 711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902226715614876e-05,
+      "loss": 0.2512,
+      "step": 712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901761338994606e-05,
+      "loss": 0.3062,
+      "step": 713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901294862930504e-05,
+      "loss": 0.2498,
+      "step": 714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990082728747436e-05,
+      "loss": 0.2363,
+      "step": 715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.99003586126781e-05,
+      "loss": 0.2446,
+      "step": 716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899888838593754e-05,
+      "loss": 0.2576,
+      "step": 717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899417965273486e-05,
+      "loss": 0.2852,
+      "step": 718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898945992769584e-05,
+      "loss": 0.3184,
+      "step": 719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989847292113445e-05,
+      "loss": 0.2542,
+      "step": 720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989799875042061e-05,
+      "loss": 0.2544,
+      "step": 721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897523480680715e-05,
+      "loss": 0.2461,
+      "step": 722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989704711196754e-05,
+      "loss": 0.2507,
+      "step": 723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989656964433397e-05,
+      "loss": 0.2915,
+      "step": 724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989609107783303e-05,
+      "loss": 0.2708,
+      "step": 725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895611412517856e-05,
+      "loss": 0.2634,
+      "step": 726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895130648441706e-05,
+      "loss": 0.2437,
+      "step": 727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989464878565796e-05,
+      "loss": 0.283,
+      "step": 728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9894165824220125e-05,
+      "loss": 0.2754,
+      "step": 729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893681764181823e-05,
+      "loss": 0.2712,
+      "step": 730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893196605596804e-05,
+      "loss": 0.2883,
+      "step": 731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989271034851894e-05,
+      "loss": 0.3027,
+      "step": 732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989222299300222e-05,
+      "loss": 0.2683,
+      "step": 733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891734539100754e-05,
+      "loss": 0.2842,
+      "step": 734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9891244986868784e-05,
+      "loss": 0.262,
+      "step": 735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890754336360666e-05,
+      "loss": 0.2307,
+      "step": 736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890262587630877e-05,
+      "loss": 0.252,
+      "step": 737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889769740734018e-05,
+      "loss": 0.2559,
+      "step": 738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889275795724815e-05,
+      "loss": 0.2646,
+      "step": 739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888780752658115e-05,
+      "loss": 0.3137,
+      "step": 740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988828461158888e-05,
+      "loss": 0.2354,
+      "step": 741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887787372572203e-05,
+      "loss": 0.269,
+      "step": 742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887289035663294e-05,
+      "loss": 0.2947,
+      "step": 743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886789600917483e-05,
+      "loss": 0.2317,
+      "step": 744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988628906839023e-05,
+      "loss": 0.2844,
+      "step": 745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988578743813711e-05,
+      "loss": 0.2327,
+      "step": 746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885284710213816e-05,
+      "loss": 0.2788,
+      "step": 747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884780884676177e-05,
+      "loss": 0.2444,
+      "step": 748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884275961580134e-05,
+      "loss": 0.2578,
+      "step": 749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988376994098174e-05,
+      "loss": 0.28,
+      "step": 750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98832628229372e-05,
+      "loss": 0.2268,
+      "step": 751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882754607502807e-05,
+      "loss": 0.2288,
+      "step": 752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882245294734998e-05,
+      "loss": 0.2505,
+      "step": 753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881734884690317e-05,
+      "loss": 0.262,
+      "step": 754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988122337742545e-05,
+      "loss": 0.3096,
+      "step": 755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988071077299718e-05,
+      "loss": 0.2808,
+      "step": 756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988019707146243e-05,
+      "loss": 0.2664,
+      "step": 757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879682272878236e-05,
+      "loss": 0.262,
+      "step": 758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879166377301768e-05,
+      "loss": 0.2395,
+      "step": 759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878649384790294e-05,
+      "loss": 0.2473,
+      "step": 760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987813129540123e-05,
+      "loss": 0.2354,
+      "step": 761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877612109192095e-05,
+      "loss": 0.2366,
+      "step": 762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877091826220543e-05,
+      "loss": 0.2681,
+      "step": 763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876570446544335e-05,
+      "loss": 0.2634,
+      "step": 764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876047970221376e-05,
+      "loss": 0.2615,
+      "step": 765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875524397309666e-05,
+      "loss": 0.2385,
+      "step": 766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874999727867347e-05,
+      "loss": 0.2449,
+      "step": 767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874473961952672e-05,
+      "loss": 0.2507,
+      "step": 768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873947099624026e-05,
+      "loss": 0.2207,
+      "step": 769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873419140939903e-05,
+      "loss": 0.2571,
+      "step": 770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872890085958923e-05,
+      "loss": 0.2344,
+      "step": 771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987235993473984e-05,
+      "loss": 0.2664,
+      "step": 772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987182868734151e-05,
+      "loss": 0.3171,
+      "step": 773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987129634382292e-05,
+      "loss": 0.2427,
+      "step": 774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870762904243185e-05,
+      "loss": 0.2344,
+      "step": 775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987022836866153e-05,
+      "loss": 0.2432,
+      "step": 776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869692737137312e-05,
+      "loss": 0.2166,
+      "step": 777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869156009730005e-05,
+      "loss": 0.2588,
+      "step": 778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98686181864992e-05,
+      "loss": 0.2,
+      "step": 779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868079267504615e-05,
+      "loss": 0.3049,
+      "step": 780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867539252806093e-05,
+      "loss": 0.2566,
+      "step": 781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866998142463592e-05,
+      "loss": 0.2688,
+      "step": 782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866455936537195e-05,
+      "loss": 0.26,
+      "step": 783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865912635087105e-05,
+      "loss": 0.2515,
+      "step": 784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865368238173647e-05,
+      "loss": 0.28,
+      "step": 785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986482274585727e-05,
+      "loss": 0.2451,
+      "step": 786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864276158198544e-05,
+      "loss": 0.2129,
+      "step": 787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863728475258156e-05,
+      "loss": 0.2551,
+      "step": 788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986317969709692e-05,
+      "loss": 0.2891,
+      "step": 789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986262982377577e-05,
+      "loss": 0.3198,
+      "step": 790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862078855355755e-05,
+      "loss": 0.23,
+      "step": 791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861526791898062e-05,
+      "loss": 0.2349,
+      "step": 792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860973633463986e-05,
+      "loss": 0.2485,
+      "step": 793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860419380114945e-05,
+      "loss": 0.2295,
+      "step": 794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859864031912482e-05,
+      "loss": 0.2732,
+      "step": 795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859307588918258e-05,
+      "loss": 0.23,
+      "step": 796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858750051194062e-05,
+      "loss": 0.2366,
+      "step": 797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858191418801794e-05,
+      "loss": 0.2207,
+      "step": 798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985763169180349e-05,
+      "loss": 0.2832,
+      "step": 799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857070870261292e-05,
+      "loss": 0.2161,
+      "step": 800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856508954237473e-05,
+      "loss": 0.2424,
+      "step": 801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985594594379443e-05,
+      "loss": 0.2578,
+      "step": 802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855381838994673e-05,
+      "loss": 0.2839,
+      "step": 803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854816639900837e-05,
+      "loss": 0.2913,
+      "step": 804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854250346575677e-05,
+      "loss": 0.2256,
+      "step": 805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853682959082077e-05,
+      "loss": 0.248,
+      "step": 806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9853114477483036e-05,
+      "loss": 0.2361,
+      "step": 807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985254490184167e-05,
+      "loss": 0.2336,
+      "step": 808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851974232221233e-05,
+      "loss": 0.2639,
+      "step": 809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851402468685075e-05,
+      "loss": 0.22,
+      "step": 810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985082961129669e-05,
+      "loss": 0.2502,
+      "step": 811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850255660119683e-05,
+      "loss": 0.2732,
+      "step": 812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984968061521779e-05,
+      "loss": 0.2356,
+      "step": 813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984910447665485e-05,
+      "loss": 0.2632,
+      "step": 814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848527244494843e-05,
+      "loss": 0.2664,
+      "step": 815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847948918801857e-05,
+      "loss": 0.2205,
+      "step": 816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847369499640108e-05,
+      "loss": 0.2705,
+      "step": 817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846788987073933e-05,
+      "loss": 0.2861,
+      "step": 818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984620738116779e-05,
+      "loss": 0.271,
+      "step": 819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845624681986254e-05,
+      "loss": 0.2349,
+      "step": 820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984504088959403e-05,
+      "loss": 0.239,
+      "step": 821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844456004055935e-05,
+      "loss": 0.2256,
+      "step": 822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843870025436914e-05,
+      "loss": 0.2065,
+      "step": 823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984328295380203e-05,
+      "loss": 0.2512,
+      "step": 824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842694789216473e-05,
+      "loss": 0.2109,
+      "step": 825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842105531745547e-05,
+      "loss": 0.2424,
+      "step": 826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984151518145468e-05,
+      "loss": 0.25,
+      "step": 827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984092373840942e-05,
+      "loss": 0.2188,
+      "step": 828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840331202675437e-05,
+      "loss": 0.2432,
+      "step": 829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983973757431853e-05,
+      "loss": 0.2185,
+      "step": 830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9839142853404606e-05,
+      "loss": 0.2949,
+      "step": 831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838547039999706e-05,
+      "loss": 0.2065,
+      "step": 832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983795013416998e-05,
+      "loss": 0.262,
+      "step": 833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983735213598171e-05,
+      "loss": 0.2607,
+      "step": 834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836753045501293e-05,
+      "loss": 0.2446,
+      "step": 835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836152862795245e-05,
+      "loss": 0.2324,
+      "step": 836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835551587930217e-05,
+      "loss": 0.2869,
+      "step": 837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834949220972962e-05,
+      "loss": 0.2466,
+      "step": 838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834345761990376e-05,
+      "loss": 0.24,
+      "step": 839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833741211049448e-05,
+      "loss": 0.2559,
+      "step": 840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833135568217315e-05,
+      "loss": 0.259,
+      "step": 841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832528833561224e-05,
+      "loss": 0.2454,
+      "step": 842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983192100714854e-05,
+      "loss": 0.22,
+      "step": 843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983131208904676e-05,
+      "loss": 0.2502,
+      "step": 844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983070207932349e-05,
+      "loss": 0.2549,
+      "step": 845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830090978046458e-05,
+      "loss": 0.29,
+      "step": 846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829478785283527e-05,
+      "loss": 0.2263,
+      "step": 847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982886550110267e-05,
+      "loss": 0.2488,
+      "step": 848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982825112557198e-05,
+      "loss": 0.2803,
+      "step": 849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827635658759673e-05,
+      "loss": 0.2451,
+      "step": 850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982701910073409e-05,
+      "loss": 0.2476,
+      "step": 851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826401451563693e-05,
+      "loss": 0.2307,
+      "step": 852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982578271131706e-05,
+      "loss": 0.2612,
+      "step": 853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825162880062893e-05,
+      "loss": 0.2507,
+      "step": 854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824541957870016e-05,
+      "loss": 0.2554,
+      "step": 855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982391994480737e-05,
+      "loss": 0.2527,
+      "step": 856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823296840944027e-05,
+      "loss": 0.2939,
+      "step": 857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822672646349167e-05,
+      "loss": 0.2163,
+      "step": 858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822047361092096e-05,
+      "loss": 0.2808,
+      "step": 859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982142098524225e-05,
+      "loss": 0.2971,
+      "step": 860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820793518869177e-05,
+      "loss": 0.2368,
+      "step": 861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820164962042544e-05,
+      "loss": 0.2783,
+      "step": 862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981953531483215e-05,
+      "loss": 0.2302,
+      "step": 863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818904577307897e-05,
+      "loss": 0.282,
+      "step": 864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818272749539827e-05,
+      "loss": 0.2671,
+      "step": 865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981763983159809e-05,
+      "loss": 0.2512,
+      "step": 866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981700582355297e-05,
+      "loss": 0.2734,
+      "step": 867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981637072547486e-05,
+      "loss": 0.2715,
+      "step": 868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815734537434274e-05,
+      "loss": 0.2285,
+      "step": 869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981509725950186e-05,
+      "loss": 0.2773,
+      "step": 870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981445889174837e-05,
+      "loss": 0.2107,
+      "step": 871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813819434244687e-05,
+      "loss": 0.2451,
+      "step": 872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981317888706182e-05,
+      "loss": 0.2268,
+      "step": 873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812537250270882e-05,
+      "loss": 0.2329,
+      "step": 874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811894523943125e-05,
+      "loss": 0.28,
+      "step": 875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981125070814991e-05,
+      "loss": 0.2214,
+      "step": 876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810605802962728e-05,
+      "loss": 0.2576,
+      "step": 877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980995980845318e-05,
+      "loss": 0.2212,
+      "step": 878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809312724692997e-05,
+      "loss": 0.3337,
+      "step": 879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980866455175403e-05,
+      "loss": 0.2456,
+      "step": 880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808015289708248e-05,
+      "loss": 0.2461,
+      "step": 881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807364938627737e-05,
+      "loss": 0.269,
+      "step": 882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806713498584718e-05,
+      "loss": 0.2297,
+      "step": 883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806060969651514e-05,
+      "loss": 0.2268,
+      "step": 884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805407351900588e-05,
+      "loss": 0.2554,
+      "step": 885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804752645404505e-05,
+      "loss": 0.1873,
+      "step": 886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980409685023597e-05,
+      "loss": 0.2356,
+      "step": 887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803439966467792e-05,
+      "loss": 0.2053,
+      "step": 888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980278199417291e-05,
+      "loss": 0.2935,
+      "step": 889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802122933424387e-05,
+      "loss": 0.2744,
+      "step": 890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801462784295395e-05,
+      "loss": 0.218,
+      "step": 891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800801546859238e-05,
+      "loss": 0.2712,
+      "step": 892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800139221189332e-05,
+      "loss": 0.1821,
+      "step": 893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979947580735923e-05,
+      "loss": 0.188,
+      "step": 894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979881130544258e-05,
+      "loss": 0.219,
+      "step": 895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798145715513168e-05,
+      "loss": 0.229,
+      "step": 896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797479037644906e-05,
+      "loss": 0.2329,
+      "step": 897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979681127191181e-05,
+      "loss": 0.2454,
+      "step": 898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796142418388035e-05,
+      "loss": 0.2202,
+      "step": 899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795472477147836e-05,
+      "loss": 0.2749,
+      "step": 900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979480144826561e-05,
+      "loss": 0.2073,
+      "step": 901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979412933181586e-05,
+      "loss": 0.2571,
+      "step": 902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979345612787321e-05,
+      "loss": 0.2568,
+      "step": 903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792781836512418e-05,
+      "loss": 0.2339,
+      "step": 904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792106457808348e-05,
+      "loss": 0.2908,
+      "step": 905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791429991835995e-05,
+      "loss": 0.2542,
+      "step": 906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979075243867047e-05,
+      "loss": 0.2471,
+      "step": 907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790073798387003e-05,
+      "loss": 0.2422,
+      "step": 908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789394071060946e-05,
+      "loss": 0.2336,
+      "step": 909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788713256767777e-05,
+      "loss": 0.2512,
+      "step": 910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788031355583085e-05,
+      "loss": 0.2703,
+      "step": 911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787348367582586e-05,
+      "loss": 0.2393,
+      "step": 912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786664292842122e-05,
+      "loss": 0.2495,
+      "step": 913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785979131437646e-05,
+      "loss": 0.218,
+      "step": 914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978529288344523e-05,
+      "loss": 0.2461,
+      "step": 915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784605548941074e-05,
+      "loss": 0.2761,
+      "step": 916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783917128001503e-05,
+      "loss": 0.2646,
+      "step": 917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783227620702946e-05,
+      "loss": 0.3032,
+      "step": 918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978253702712197e-05,
+      "loss": 0.2119,
+      "step": 919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781845347335253e-05,
+      "loss": 0.2561,
+      "step": 920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781152581419595e-05,
+      "loss": 0.249,
+      "step": 921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780458729451916e-05,
+      "loss": 0.2461,
+      "step": 922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779763791509262e-05,
+      "loss": 0.217,
+      "step": 923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779067767668794e-05,
+      "loss": 0.2507,
+      "step": 924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778370658007792e-05,
+      "loss": 0.2261,
+      "step": 925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977767246260366e-05,
+      "loss": 0.2185,
+      "step": 926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776973181533926e-05,
+      "loss": 0.2996,
+      "step": 927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776272814876235e-05,
+      "loss": 0.2327,
+      "step": 928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775571362708353e-05,
+      "loss": 0.2852,
+      "step": 929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977486882510816e-05,
+      "loss": 0.2434,
+      "step": 930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774165202153665e-05,
+      "loss": 0.2178,
+      "step": 931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773460493922994e-05,
+      "loss": 0.2812,
+      "step": 932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97727547004944e-05,
+      "loss": 0.2178,
+      "step": 933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772047821946242e-05,
+      "loss": 0.2883,
+      "step": 934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977133985835702e-05,
+      "loss": 0.2585,
+      "step": 935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977063080980533e-05,
+      "loss": 0.2126,
+      "step": 936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976992067636991e-05,
+      "loss": 0.2212,
+      "step": 937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976920945812961e-05,
+      "loss": 0.2383,
+      "step": 938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9768497155163392e-05,
+      "loss": 0.2656,
+      "step": 939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767783767550358e-05,
+      "loss": 0.26,
+      "step": 940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767069295369707e-05,
+      "loss": 0.2173,
+      "step": 941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976635373870078e-05,
+      "loss": 0.2744,
+      "step": 942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9765637097623028e-05,
+      "loss": 0.2002,
+      "step": 943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976491937221602e-05,
+      "loss": 0.2117,
+      "step": 944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976420056255945e-05,
+      "loss": 0.2285,
+      "step": 945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763480668733132e-05,
+      "loss": 0.2175,
+      "step": 946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762759690816996e-05,
+      "loss": 0.3159,
+      "step": 947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762037628891097e-05,
+      "loss": 0.2749,
+      "step": 948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761314483035617e-05,
+      "loss": 0.2378,
+      "step": 949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976059025333084e-05,
+      "loss": 0.219,
+      "step": 950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759864939857186e-05,
+      "loss": 0.1965,
+      "step": 951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975913854269519e-05,
+      "loss": 0.2405,
+      "step": 952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975841106192551e-05,
+      "loss": 0.2341,
+      "step": 953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757682497628915e-05,
+      "loss": 0.2427,
+      "step": 954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975695284988631e-05,
+      "loss": 0.2461,
+      "step": 955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756222118778704e-05,
+      "loss": 0.2657,
+      "step": 956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9755490304387236e-05,
+      "loss": 0.2058,
+      "step": 957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754757406793172e-05,
+      "loss": 0.2439,
+      "step": 958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975402342607787e-05,
+      "loss": 0.2461,
+      "step": 959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975328836232285e-05,
+      "loss": 0.2588,
+      "step": 960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752552215609713e-05,
+      "loss": 0.2366,
+      "step": 961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9751814986020203e-05,
+      "loss": 0.2307,
+      "step": 962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975107667363618e-05,
+      "loss": 0.2231,
+      "step": 963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750337278539623e-05,
+      "loss": 0.2288,
+      "step": 964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974959680081263e-05,
+      "loss": 0.2773,
+      "step": 965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748855240537418e-05,
+      "loss": 0.2078,
+      "step": 966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974811259779633e-05,
+      "loss": 0.2566,
+      "step": 967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974736887267182e-05,
+      "loss": 0.1978,
+      "step": 968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974662406524647e-05,
+      "loss": 0.2429,
+      "step": 969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745878175602984e-05,
+      "loss": 0.2375,
+      "step": 970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745131203824177e-05,
+      "loss": 0.2117,
+      "step": 971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974438314999299e-05,
+      "loss": 0.2378,
+      "step": 972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743634014192486e-05,
+      "loss": 0.2517,
+      "step": 973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742883796505843e-05,
+      "loss": 0.2607,
+      "step": 974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974213249701636e-05,
+      "loss": 0.2402,
+      "step": 975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974138011580746e-05,
+      "loss": 0.2402,
+      "step": 976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974062665296269e-05,
+      "loss": 0.2197,
+      "step": 977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739872108565697e-05,
+      "loss": 0.2195,
+      "step": 978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973911648270027e-05,
+      "loss": 0.2593,
+      "step": 979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738359775450313e-05,
+      "loss": 0.2026,
+      "step": 980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973760198689984e-05,
+      "loss": 0.2764,
+      "step": 981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736843117132996e-05,
+      "loss": 0.2168,
+      "step": 982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736083166234047e-05,
+      "loss": 0.2317,
+      "step": 983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735322134287364e-05,
+      "loss": 0.2212,
+      "step": 984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734560021377454e-05,
+      "loss": 0.2363,
+      "step": 985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973379682758894e-05,
+      "loss": 0.2,
+      "step": 986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973303255300656e-05,
+      "loss": 0.2834,
+      "step": 987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732267197715176e-05,
+      "loss": 0.2864,
+      "step": 988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973150076179977e-05,
+      "loss": 0.2493,
+      "step": 989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730733245345445e-05,
+      "loss": 0.25,
+      "step": 990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972996464843742e-05,
+      "loss": 0.2605,
+      "step": 991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972919497116104e-05,
+      "loss": 0.219,
+      "step": 992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728424213601758e-05,
+      "loss": 0.238,
+      "step": 993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727652375845164e-05,
+      "loss": 0.2288,
+      "step": 994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726879457976954e-05,
+      "loss": 0.2815,
+      "step": 995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972610546008295e-05,
+      "loss": 0.2085,
+      "step": 996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725330382249094e-05,
+      "loss": 0.2395,
+      "step": 997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972455422456145e-05,
+      "loss": 0.1921,
+      "step": 998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723776987106193e-05,
+      "loss": 0.2744,
+      "step": 999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722998669969626e-05,
+      "loss": 0.2595,
+      "step": 1000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722219273238166e-05,
+      "loss": 0.2485,
+      "step": 1001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972143879699836e-05,
+      "loss": 0.2441,
+      "step": 1002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720657241336866e-05,
+      "loss": 0.2329,
+      "step": 1003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971987460634046e-05,
+      "loss": 0.2327,
+      "step": 1004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971909089209605e-05,
+      "loss": 0.2605,
+      "step": 1005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971830609869065e-05,
+      "loss": 0.2815,
+      "step": 1006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97175202262114e-05,
+      "loss": 0.3052,
+      "step": 1007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716733274745566e-05,
+      "loss": 0.2788,
+      "step": 1008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715945244380513e-05,
+      "loss": 0.2109,
+      "step": 1009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971515613520376e-05,
+      "loss": 0.2451,
+      "step": 1010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714365947302905e-05,
+      "loss": 0.2612,
+      "step": 1011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97135746807657e-05,
+      "loss": 0.217,
+      "step": 1012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971278233568e-05,
+      "loss": 0.2175,
+      "step": 1013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711988912133783e-05,
+      "loss": 0.2361,
+      "step": 1014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711194410215148e-05,
+      "loss": 0.2693,
+      "step": 1015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710398830012313e-05,
+      "loss": 0.2771,
+      "step": 1016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970960217161361e-05,
+      "loss": 0.2793,
+      "step": 1017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708804435107504e-05,
+      "loss": 0.2805,
+      "step": 1018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708005620582564e-05,
+      "loss": 0.2048,
+      "step": 1019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9707205728127496e-05,
+      "loss": 0.2385,
+      "step": 1020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706404757831104e-05,
+      "loss": 0.2542,
+      "step": 1021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705602709782336e-05,
+      "loss": 0.2195,
+      "step": 1022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970479958407024e-05,
+      "loss": 0.2206,
+      "step": 1023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703995380783993e-05,
+      "loss": 0.2605,
+      "step": 1024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970319010001289e-05,
+      "loss": 0.2534,
+      "step": 1025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702383741846346e-05,
+      "loss": 0.2561,
+      "step": 1026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701576306373896e-05,
+      "loss": 0.2588,
+      "step": 1027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700767793685195e-05,
+      "loss": 0.2305,
+      "step": 1028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969995820387001e-05,
+      "loss": 0.2734,
+      "step": 1029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969914753701824e-05,
+      "loss": 0.2312,
+      "step": 1030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969833579321989e-05,
+      "loss": 0.2534,
+      "step": 1031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697522972565103e-05,
+      "loss": 0.2996,
+      "step": 1032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696709075144123e-05,
+      "loss": 0.2581,
+      "step": 1033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695894101047327e-05,
+      "loss": 0.2498,
+      "step": 1034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96950780503652e-05,
+      "loss": 0.2012,
+      "step": 1035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694260923188354e-05,
+      "loss": 0.2651,
+      "step": 1036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969344271960752e-05,
+      "loss": 0.2263,
+      "step": 1037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692623439713547e-05,
+      "loss": 0.2485,
+      "step": 1038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691803083597403e-05,
+      "loss": 0.2505,
+      "step": 1039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969098165135018e-05,
+      "loss": 0.24,
+      "step": 1040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969015914306308e-05,
+      "loss": 0.2427,
+      "step": 1041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689335558827433e-05,
+      "loss": 0.2283,
+      "step": 1042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688510898734687e-05,
+      "loss": 0.1907,
+      "step": 1043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687685162876406e-05,
+      "loss": 0.2629,
+      "step": 1044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686858351344284e-05,
+      "loss": 0.2026,
+      "step": 1045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968603046423011e-05,
+      "loss": 0.2402,
+      "step": 1046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685201501625822e-05,
+      "loss": 0.2449,
+      "step": 1047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968437146362346e-05,
+      "loss": 0.2058,
+      "step": 1048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968354035031519e-05,
+      "loss": 0.2676,
+      "step": 1049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682708161793287e-05,
+      "loss": 0.24,
+      "step": 1050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968187489815016e-05,
+      "loss": 0.2349,
+      "step": 1051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968104055947833e-05,
+      "loss": 0.2424,
+      "step": 1052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968020514587044e-05,
+      "loss": 0.2385,
+      "step": 1053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967936865741924e-05,
+      "loss": 0.2668,
+      "step": 1054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9678531094217622e-05,
+      "loss": 0.2046,
+      "step": 1055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967769245635858e-05,
+      "loss": 0.2397,
+      "step": 1056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967685274393523e-05,
+      "loss": 0.2356,
+      "step": 1057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676011957040812e-05,
+      "loss": 0.2385,
+      "step": 1058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675170095768685e-05,
+      "loss": 0.2468,
+      "step": 1059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967432716021232e-05,
+      "loss": 0.2417,
+      "step": 1060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673483150465314e-05,
+      "loss": 0.2168,
+      "step": 1061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967263806662139e-05,
+      "loss": 0.2434,
+      "step": 1062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967179190877437e-05,
+      "loss": 0.261,
+      "step": 1063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670944677018214e-05,
+      "loss": 0.2056,
+      "step": 1064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670096371446992e-05,
+      "loss": 0.2563,
+      "step": 1065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96692469921549e-05,
+      "loss": 0.197,
+      "step": 1066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966839653923624e-05,
+      "loss": 0.1797,
+      "step": 1067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667545012785448e-05,
+      "loss": 0.1968,
+      "step": 1068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966669241289708e-05,
+      "loss": 0.2371,
+      "step": 1069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665838739665793e-05,
+      "loss": 0.2451,
+      "step": 1070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966498399318638e-05,
+      "loss": 0.271,
+      "step": 1071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664128173553748e-05,
+      "loss": 0.2217,
+      "step": 1072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663271280862924e-05,
+      "loss": 0.1975,
+      "step": 1073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966241331520905e-05,
+      "loss": 0.2429,
+      "step": 1074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661554276687394e-05,
+      "loss": 0.2354,
+      "step": 1075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660694165393334e-05,
+      "loss": 0.2263,
+      "step": 1076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659832981422383e-05,
+      "loss": 0.2598,
+      "step": 1077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658970724870153e-05,
+      "loss": 0.2422,
+      "step": 1078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9658107395832387e-05,
+      "loss": 0.2627,
+      "step": 1079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657242994404947e-05,
+      "loss": 0.2607,
+      "step": 1080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656377520683808e-05,
+      "loss": 0.2095,
+      "step": 1081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9655510974765074e-05,
+      "loss": 0.2427,
+      "step": 1082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965464335674496e-05,
+      "loss": 0.301,
+      "step": 1083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653774666719796e-05,
+      "loss": 0.2617,
+      "step": 1084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652904904786046e-05,
+      "loss": 0.2319,
+      "step": 1085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652034071040278e-05,
+      "loss": 0.2515,
+      "step": 1086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651162165579188e-05,
+      "loss": 0.2437,
+      "step": 1087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650289188499587e-05,
+      "loss": 0.249,
+      "step": 1088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649415139898407e-05,
+      "loss": 0.2068,
+      "step": 1089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96485400198727e-05,
+      "loss": 0.238,
+      "step": 1090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647663828519633e-05,
+      "loss": 0.2275,
+      "step": 1091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646786565936497e-05,
+      "loss": 0.2163,
+      "step": 1092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645908232220692e-05,
+      "loss": 0.2563,
+      "step": 1093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645028827469748e-05,
+      "loss": 0.2646,
+      "step": 1094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964414835178131e-05,
+      "loss": 0.3184,
+      "step": 1095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964326680525314e-05,
+      "loss": 0.2903,
+      "step": 1096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642384187983126e-05,
+      "loss": 0.2346,
+      "step": 1097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641500500069267e-05,
+      "loss": 0.2285,
+      "step": 1098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964061574160968e-05,
+      "loss": 0.2539,
+      "step": 1099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963972991270261e-05,
+      "loss": 0.2397,
+      "step": 1100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638843013446408e-05,
+      "loss": 0.2278,
+      "step": 1101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637955043939554e-05,
+      "loss": 0.2458,
+      "step": 1102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637066004280646e-05,
+      "loss": 0.217,
+      "step": 1103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636175894568397e-05,
+      "loss": 0.2749,
+      "step": 1104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635284714901646e-05,
+      "loss": 0.2378,
+      "step": 1105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634392465379337e-05,
+      "loss": 0.2263,
+      "step": 1106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963349914610054e-05,
+      "loss": 0.2297,
+      "step": 1107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9632604757164456e-05,
+      "loss": 0.197,
+      "step": 1108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631709298670382e-05,
+      "loss": 0.2144,
+      "step": 1109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630812770717753e-05,
+      "loss": 0.2158,
+      "step": 1110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962991517340611e-05,
+      "loss": 0.2729,
+      "step": 1111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629016506835122e-05,
+      "loss": 0.2126,
+      "step": 1112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962811677110457e-05,
+      "loss": 0.2393,
+      "step": 1113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962721596631436e-05,
+      "loss": 0.3408,
+      "step": 1114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626314092564506e-05,
+      "loss": 0.2146,
+      "step": 1115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625411149955156e-05,
+      "loss": 0.2625,
+      "step": 1116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962450713858656e-05,
+      "loss": 0.1875,
+      "step": 1117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623602058559103e-05,
+      "loss": 0.262,
+      "step": 1118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622695909973276e-05,
+      "loss": 0.2388,
+      "step": 1119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621788692929695e-05,
+      "loss": 0.2954,
+      "step": 1120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620880407529092e-05,
+      "loss": 0.2461,
+      "step": 1121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619971053872318e-05,
+      "loss": 0.2175,
+      "step": 1122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619060632060343e-05,
+      "loss": 0.2083,
+      "step": 1123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618149142194262e-05,
+      "loss": 0.2627,
+      "step": 1124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9617236584375275e-05,
+      "loss": 0.2529,
+      "step": 1125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616322958704708e-05,
+      "loss": 0.2791,
+      "step": 1126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961540826528401e-05,
+      "loss": 0.2769,
+      "step": 1127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614492504214744e-05,
+      "loss": 0.2268,
+      "step": 1128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613575675598588e-05,
+      "loss": 0.2375,
+      "step": 1129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961265777953735e-05,
+      "loss": 0.2798,
+      "step": 1130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9611738816132936e-05,
+      "loss": 0.2148,
+      "step": 1131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610818785487392e-05,
+      "loss": 0.2378,
+      "step": 1132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609897687702874e-05,
+      "loss": 0.2886,
+      "step": 1133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960897552288165e-05,
+      "loss": 0.1899,
+      "step": 1134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608052291126123e-05,
+      "loss": 0.1853,
+      "step": 1135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9607127992538796e-05,
+      "loss": 0.2422,
+      "step": 1136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606202627222298e-05,
+      "loss": 0.2297,
+      "step": 1137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9605276195279385e-05,
+      "loss": 0.2134,
+      "step": 1138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604348696812917e-05,
+      "loss": 0.2585,
+      "step": 1139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960342013192588e-05,
+      "loss": 0.2393,
+      "step": 1140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602490500721375e-05,
+      "loss": 0.2373,
+      "step": 1141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960155980330263e-05,
+      "loss": 0.2024,
+      "step": 1142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960062803977298e-05,
+      "loss": 0.2041,
+      "step": 1143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599695210235886e-05,
+      "loss": 0.2153,
+      "step": 1144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959876131479493e-05,
+      "loss": 0.2173,
+      "step": 1145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597826353553794e-05,
+      "loss": 0.2083,
+      "step": 1146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.95968903266163e-05,
+      "loss": 0.2498,
+      "step": 1147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959595323408638e-05,
+      "loss": 0.2092,
+      "step": 1148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959501507606808e-05,
+      "loss": 0.2554,
+      "step": 1149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959407585266558e-05,
+      "loss": 0.2766,
+      "step": 1150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593135563983152e-05,
+      "loss": 0.2056,
+      "step": 1151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959219421012521e-05,
+      "loss": 0.249,
+      "step": 1152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591251791196274e-05,
+      "loss": 0.2205,
+      "step": 1153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590308307300988e-05,
+      "loss": 0.2981,
+      "step": 1154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589363758544108e-05,
+      "loss": 0.2549,
+      "step": 1155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958841814503052e-05,
+      "loss": 0.2969,
+      "step": 1156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958747146686521e-05,
+      "loss": 0.2122,
+      "step": 1157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95865237241533e-05,
+      "loss": 0.2524,
+      "step": 1158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958557491700002e-05,
+      "loss": 0.207,
+      "step": 1159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584625045510725e-05,
+      "loss": 0.2087,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583674109790878e-05,
+      "loss": 0.2434,
+      "step": 1161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582722109946067e-05,
+      "loss": 0.2422,
+      "step": 1162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581769046082002e-05,
+      "loss": 0.1765,
+      "step": 1163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9580814918304504e-05,
+      "loss": 0.209,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579859726719513e-05,
+      "loss": 0.2498,
+      "step": 1165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957890347143309e-05,
+      "loss": 0.2258,
+      "step": 1166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577946152551417e-05,
+      "loss": 0.2351,
+      "step": 1167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576987770180788e-05,
+      "loss": 0.2212,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576028324427612e-05,
+      "loss": 0.2407,
+      "step": 1169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575067815398423e-05,
+      "loss": 0.2334,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957410624319988e-05,
+      "loss": 0.2258,
+      "step": 1171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957314360793874e-05,
+      "loss": 0.2476,
+      "step": 1172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9572179909721894e-05,
+      "loss": 0.2173,
+      "step": 1173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957121514865635e-05,
+      "loss": 0.2769,
+      "step": 1174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570249324849223e-05,
+      "loss": 0.2375,
+      "step": 1175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569282438407763e-05,
+      "loss": 0.2874,
+      "step": 1176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956831448943932e-05,
+      "loss": 0.2483,
+      "step": 1177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956734547805137e-05,
+      "loss": 0.2278,
+      "step": 1178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956637540435151e-05,
+      "loss": 0.2358,
+      "step": 1179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956540426844746e-05,
+      "loss": 0.262,
+      "step": 1180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564432070447035e-05,
+      "loss": 0.2129,
+      "step": 1181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563458810458195e-05,
+      "loss": 0.2324,
+      "step": 1182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9562484488589005e-05,
+      "loss": 0.2566,
+      "step": 1183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561509104947643e-05,
+      "loss": 0.2275,
+      "step": 1184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560532659642413e-05,
+      "loss": 0.2437,
+      "step": 1185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955955515278174e-05,
+      "loss": 0.2664,
+      "step": 1186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558576584474154e-05,
+      "loss": 0.2749,
+      "step": 1187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557596954828315e-05,
+      "loss": 0.2712,
+      "step": 1188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556616263953e-05,
+      "loss": 0.2092,
+      "step": 1189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955563451195709e-05,
+      "loss": 0.2319,
+      "step": 1190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554651698949603e-05,
+      "loss": 0.2263,
+      "step": 1191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955366782503966e-05,
+      "loss": 0.2222,
+      "step": 1192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552682890336508e-05,
+      "loss": 0.2471,
+      "step": 1193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551696894949513e-05,
+      "loss": 0.2476,
+      "step": 1194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955070983898815e-05,
+      "loss": 0.2429,
+      "step": 1195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954972172256202e-05,
+      "loss": 0.2361,
+      "step": 1196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548732545780833e-05,
+      "loss": 0.2339,
+      "step": 1197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954774230875443e-05,
+      "loss": 0.2168,
+      "step": 1198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954675101159276e-05,
+      "loss": 0.2622,
+      "step": 1199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545758654405888e-05,
+      "loss": 0.229,
+      "step": 1200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544765237304006e-05,
+      "loss": 0.2217,
+      "step": 1201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543770760397413e-05,
+      "loss": 0.2563,
+      "step": 1202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542775223796534e-05,
+      "loss": 0.248,
+      "step": 1203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541778627611908e-05,
+      "loss": 0.2195,
+      "step": 1204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540780971954193e-05,
+      "loss": 0.2236,
+      "step": 1205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539782256934166e-05,
+      "loss": 0.2078,
+      "step": 1206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953878248266271e-05,
+      "loss": 0.2119,
+      "step": 1207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537781649250848e-05,
+      "loss": 0.2917,
+      "step": 1208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95367797568097e-05,
+      "loss": 0.2427,
+      "step": 1209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535776805450512e-05,
+      "loss": 0.2207,
+      "step": 1210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953477279528465e-05,
+      "loss": 0.2175,
+      "step": 1211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533767726423586e-05,
+      "loss": 0.2336,
+      "step": 1212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532761598978932e-05,
+      "loss": 0.2083,
+      "step": 1213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531754413062392e-05,
+      "loss": 0.2126,
+      "step": 1214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95307461687858e-05,
+      "loss": 0.2542,
+      "step": 1215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9529736866261112e-05,
+      "loss": 0.2722,
+      "step": 1216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528726505600396e-05,
+      "loss": 0.1995,
+      "step": 1217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952771508691583e-05,
+      "loss": 0.2366,
+      "step": 1218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526702610319727e-05,
+      "loss": 0.2803,
+      "step": 1219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525689075924498e-05,
+      "loss": 0.2305,
+      "step": 1220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524674483842687e-05,
+      "loss": 0.2627,
+      "step": 1221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952365883418695e-05,
+      "loss": 0.2251,
+      "step": 1222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522642127070057e-05,
+      "loss": 0.3052,
+      "step": 1223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521624362604896e-05,
+      "loss": 0.2512,
+      "step": 1224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952060554090448e-05,
+      "loss": 0.2527,
+      "step": 1225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519585662081932e-05,
+      "loss": 0.2212,
+      "step": 1226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518564726250496e-05,
+      "loss": 0.2061,
+      "step": 1227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517542733523528e-05,
+      "loss": 0.2393,
+      "step": 1228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516519684014505e-05,
+      "loss": 0.2539,
+      "step": 1229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515495577837026e-05,
+      "loss": 0.2288,
+      "step": 1230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514470415104802e-05,
+      "loss": 0.1865,
+      "step": 1231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951344419593166e-05,
+      "loss": 0.2061,
+      "step": 1232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512416920431544e-05,
+      "loss": 0.2039,
+      "step": 1233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9511388588718522e-05,
+      "loss": 0.2463,
+      "step": 1234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510359200906776e-05,
+      "loss": 0.2114,
+      "step": 1235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509328757110598e-05,
+      "loss": 0.2112,
+      "step": 1236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508297257444408e-05,
+      "loss": 0.1853,
+      "step": 1237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950726470202274e-05,
+      "loss": 0.218,
+      "step": 1238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9506231090960244e-05,
+      "loss": 0.2346,
+      "step": 1239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505196424371685e-05,
+      "loss": 0.1841,
+      "step": 1240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504160702371947e-05,
+      "loss": 0.209,
+      "step": 1241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503123925076028e-05,
+      "loss": 0.2629,
+      "step": 1242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502086092599054e-05,
+      "loss": 0.2322,
+      "step": 1243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9501047205056262e-05,
+      "loss": 0.2136,
+      "step": 1244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500007262562994e-05,
+      "loss": 0.2139,
+      "step": 1245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498966265234735e-05,
+      "loss": 0.2434,
+      "step": 1246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497924213187057e-05,
+      "loss": 0.2881,
+      "step": 1247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496881106535675e-05,
+      "loss": 0.2483,
+      "step": 1248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495836945396413e-05,
+      "loss": 0.2698,
+      "step": 1249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494791729885198e-05,
+      "loss": 0.2012,
+      "step": 1250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949374546011809e-05,
+      "loss": 0.1904,
+      "step": 1251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949269813621127e-05,
+      "loss": 0.2266,
+      "step": 1252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491649758281017e-05,
+      "loss": 0.1794,
+      "step": 1253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490600326443743e-05,
+      "loss": 0.2285,
+      "step": 1254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9489549840815974e-05,
+      "loss": 0.2241,
+      "step": 1255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488498301514343e-05,
+      "loss": 0.2073,
+      "step": 1256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9487445708655616e-05,
+      "loss": 0.2065,
+      "step": 1257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486392062356663e-05,
+      "loss": 0.2456,
+      "step": 1258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948533736273448e-05,
+      "loss": 0.2202,
+      "step": 1259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484281609906172e-05,
+      "loss": 0.2505,
+      "step": 1260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483224803988965e-05,
+      "loss": 0.2146,
+      "step": 1261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482166945100204e-05,
+      "loss": 0.1926,
+      "step": 1262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9481108033357344e-05,
+      "loss": 0.2129,
+      "step": 1263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948004806887797e-05,
+      "loss": 0.2319,
+      "step": 1264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478987051779767e-05,
+      "loss": 0.2107,
+      "step": 1265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477924982180548e-05,
+      "loss": 0.2595,
+      "step": 1266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476861860198238e-05,
+      "loss": 0.2258,
+      "step": 1267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475797685950885e-05,
+      "loss": 0.2429,
+      "step": 1268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474732459556647e-05,
+      "loss": 0.2444,
+      "step": 1269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473666181133805e-05,
+      "loss": 0.2671,
+      "step": 1270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947259885080075e-05,
+      "loss": 0.1794,
+      "step": 1271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471530468675995e-05,
+      "loss": 0.189,
+      "step": 1272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470461034878167e-05,
+      "loss": 0.2012,
+      "step": 1273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469390549526007e-05,
+      "loss": 0.2124,
+      "step": 1274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9468319012738383e-05,
+      "loss": 0.2654,
+      "step": 1275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946724642463427e-05,
+      "loss": 0.2202,
+      "step": 1276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466172785332767e-05,
+      "loss": 0.2036,
+      "step": 1277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946509809495308e-05,
+      "loss": 0.1863,
+      "step": 1278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946402235361454e-05,
+      "loss": 0.1841,
+      "step": 1279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946294556143659e-05,
+      "loss": 0.2092,
+      "step": 1280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94618677185388e-05,
+      "loss": 0.2063,
+      "step": 1281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460788825040837e-05,
+      "loss": 0.2168,
+      "step": 1282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459708881062504e-05,
+      "loss": 0.2305,
+      "step": 1283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458627886723706e-05,
+      "loss": 0.2698,
+      "step": 1284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945754584214448e-05,
+      "loss": 0.2322,
+      "step": 1285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456462747444965e-05,
+      "loss": 0.2273,
+      "step": 1286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455378602745426e-05,
+      "loss": 0.1825,
+      "step": 1287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945429340816624e-05,
+      "loss": 0.2302,
+      "step": 1288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94532071638279e-05,
+      "loss": 0.2324,
+      "step": 1289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452119869851014e-05,
+      "loss": 0.1926,
+      "step": 1290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451031526356318e-05,
+      "loss": 0.1997,
+      "step": 1291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449942133464654e-05,
+      "loss": 0.2578,
+      "step": 1292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944885169129698e-05,
+      "loss": 0.2546,
+      "step": 1293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447760199974376e-05,
+      "loss": 0.2249,
+      "step": 1294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944666765961804e-05,
+      "loss": 0.2397,
+      "step": 1295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445574070349272e-05,
+      "loss": 0.26,
+      "step": 1296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444479432289505e-05,
+      "loss": 0.2542,
+      "step": 1297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944338374556028e-05,
+      "loss": 0.2009,
+      "step": 1298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442287010283264e-05,
+      "loss": 0.1862,
+      "step": 1299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441189226580225e-05,
+      "loss": 0.2747,
+      "step": 1300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440090394573056e-05,
+      "loss": 0.2073,
+      "step": 1301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943899051438377e-05,
+      "loss": 0.2434,
+      "step": 1302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437889586134493e-05,
+      "loss": 0.2036,
+      "step": 1303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9436787609947464e-05,
+      "loss": 0.2363,
+      "step": 1304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435684585945037e-05,
+      "loss": 0.2441,
+      "step": 1305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434580514249698e-05,
+      "loss": 0.2124,
+      "step": 1306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433475394984028e-05,
+      "loss": 0.198,
+      "step": 1307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9432369228270733e-05,
+      "loss": 0.2061,
+      "step": 1308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431262014232645e-05,
+      "loss": 0.2671,
+      "step": 1309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94301537529927e-05,
+      "loss": 0.2124,
+      "step": 1310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942904444467395e-05,
+      "loss": 0.2192,
+      "step": 1311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942793408939957e-05,
+      "loss": 0.2209,
+      "step": 1312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426822687292852e-05,
+      "loss": 0.2458,
+      "step": 1313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425710238477197e-05,
+      "loss": 0.2659,
+      "step": 1314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424596743076125e-05,
+      "loss": 0.2527,
+      "step": 1315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423482201213275e-05,
+      "loss": 0.2173,
+      "step": 1316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9422366613012404e-05,
+      "loss": 0.2168,
+      "step": 1317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421249978597375e-05,
+      "loss": 0.2568,
+      "step": 1318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420132298092173e-05,
+      "loss": 0.2393,
+      "step": 1319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9419013571620907e-05,
+      "loss": 0.2637,
+      "step": 1320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941789379930779e-05,
+      "loss": 0.2649,
+      "step": 1321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416772981277156e-05,
+      "loss": 0.2375,
+      "step": 1322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941565111765346e-05,
+      "loss": 0.2163,
+      "step": 1323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414528208561262e-05,
+      "loss": 0.2495,
+      "step": 1324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9413404254125246e-05,
+      "loss": 0.2266,
+      "step": 1325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412279254470215e-05,
+      "loss": 0.2195,
+      "step": 1326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9411153209721078e-05,
+      "loss": 0.2017,
+      "step": 1327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941002612000287e-05,
+      "loss": 0.2314,
+      "step": 1328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408897985440735e-05,
+      "loss": 0.2114,
+      "step": 1329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407768806159935e-05,
+      "loss": 0.1953,
+      "step": 1330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940663858228585e-05,
+      "loss": 0.2039,
+      "step": 1331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405507313943975e-05,
+      "loss": 0.2134,
+      "step": 1332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404375001259916e-05,
+      "loss": 0.196,
+      "step": 1333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940324164435941e-05,
+      "loss": 0.2512,
+      "step": 1334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402107243368288e-05,
+      "loss": 0.23,
+      "step": 1335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400971798412514e-05,
+      "loss": 0.2415,
+      "step": 1336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399835309618165e-05,
+      "loss": 0.2363,
+      "step": 1337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398697777111427e-05,
+      "loss": 0.248,
+      "step": 1338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397559201018604e-05,
+      "loss": 0.2571,
+      "step": 1339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939641958146612e-05,
+      "loss": 0.2051,
+      "step": 1340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939527891858052e-05,
+      "loss": 0.2241,
+      "step": 1341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939413721248845e-05,
+      "loss": 0.2148,
+      "step": 1342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9392994463316677e-05,
+      "loss": 0.1824,
+      "step": 1343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391850671192092e-05,
+      "loss": 0.2021,
+      "step": 1344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9390705836241698e-05,
+      "loss": 0.2131,
+      "step": 1345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389559958592607e-05,
+      "loss": 0.2852,
+      "step": 1346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938841303837205e-05,
+      "loss": 0.238,
+      "step": 1347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938726507570738e-05,
+      "loss": 0.1931,
+      "step": 1348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386116070726063e-05,
+      "loss": 0.2012,
+      "step": 1349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384966023555673e-05,
+      "loss": 0.1938,
+      "step": 1350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938381493432391e-05,
+      "loss": 0.2214,
+      "step": 1351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382662803158585e-05,
+      "loss": 0.2629,
+      "step": 1352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381509630187626e-05,
+      "loss": 0.2603,
+      "step": 1353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938035541553907e-05,
+      "loss": 0.2212,
+      "step": 1354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937920015934108e-05,
+      "loss": 0.2747,
+      "step": 1355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937804386172193e-05,
+      "loss": 0.2192,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937688652281001e-05,
+      "loss": 0.1694,
+      "step": 1357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375728142733825e-05,
+      "loss": 0.2104,
+      "step": 1358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9374568721621996e-05,
+      "loss": 0.2212,
+      "step": 1359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373408259603254e-05,
+      "loss": 0.1938,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9372246756806462e-05,
+      "loss": 0.25,
+      "step": 1361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937108421336058e-05,
+      "loss": 0.2305,
+      "step": 1362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9369920629394693e-05,
+      "loss": 0.2314,
+      "step": 1363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368756005038e-05,
+      "loss": 0.1936,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367590340419814e-05,
+      "loss": 0.2378,
+      "step": 1365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366423635669568e-05,
+      "loss": 0.2549,
+      "step": 1366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365255890916802e-05,
+      "loss": 0.2227,
+      "step": 1367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936408710629118e-05,
+      "loss": 0.2556,
+      "step": 1368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936291728192248e-05,
+      "loss": 0.2493,
+      "step": 1369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9361746417940592e-05,
+      "loss": 0.2329,
+      "step": 1370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360574514475518e-05,
+      "loss": 0.2559,
+      "step": 1371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935940157165739e-05,
+      "loss": 0.2251,
+      "step": 1372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935822758961644e-05,
+      "loss": 0.2024,
+      "step": 1373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357052568483022e-05,
+      "loss": 0.219,
+      "step": 1374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355876508387606e-05,
+      "loss": 0.2559,
+      "step": 1375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9354699409460775e-05,
+      "loss": 0.2087,
+      "step": 1376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935352127183323e-05,
+      "loss": 0.2727,
+      "step": 1377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352342095635782e-05,
+      "loss": 0.2378,
+      "step": 1378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351161880999363e-05,
+      "loss": 0.1877,
+      "step": 1379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9349980628055023e-05,
+      "loss": 0.2004,
+      "step": 1380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348798336933916e-05,
+      "loss": 0.1924,
+      "step": 1381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934761500776732e-05,
+      "loss": 0.1814,
+      "step": 1382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9346430640686625e-05,
+      "loss": 0.24,
+      "step": 1383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345245235823343e-05,
+      "loss": 0.2512,
+      "step": 1384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344058793309085e-05,
+      "loss": 0.23,
+      "step": 1385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93428713132756e-05,
+      "loss": 0.2212,
+      "step": 1386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934168279585473e-05,
+      "loss": 0.2129,
+      "step": 1387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340493241178452e-05,
+      "loss": 0.2456,
+      "step": 1388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933930264937884e-05,
+      "loss": 0.2212,
+      "step": 1389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9338111020588092e-05,
+      "loss": 0.2053,
+      "step": 1390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933691835493853e-05,
+      "loss": 0.1758,
+      "step": 1391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933572465256257e-05,
+      "loss": 0.2065,
+      "step": 1392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933452991359276e-05,
+      "loss": 0.2224,
+      "step": 1393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933333413816176e-05,
+      "loss": 0.2302,
+      "step": 1394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933213732640234e-05,
+      "loss": 0.2268,
+      "step": 1395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9330939478447392e-05,
+      "loss": 0.2266,
+      "step": 1396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329740594429913e-05,
+      "loss": 0.1682,
+      "step": 1397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932854067448303e-05,
+      "loss": 0.2437,
+      "step": 1398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932733971873997e-05,
+      "loss": 0.2161,
+      "step": 1399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932613772733408e-05,
+      "loss": 0.2434,
+      "step": 1400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324934700398827e-05,
+      "loss": 0.2427,
+      "step": 1401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932373063806779e-05,
+      "loss": 0.2087,
+      "step": 1402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932252554047466e-05,
+      "loss": 0.2205,
+      "step": 1403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321319407753244e-05,
+      "loss": 0.269,
+      "step": 1404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320112240037466e-05,
+      "loss": 0.2378,
+      "step": 1405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931890403746137e-05,
+      "loss": 0.2292,
+      "step": 1406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317694800159103e-05,
+      "loss": 0.251,
+      "step": 1407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316484528264932e-05,
+      "loss": 0.2214,
+      "step": 1408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931527322191324e-05,
+      "loss": 0.2195,
+      "step": 1409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9314060881238532e-05,
+      "loss": 0.2356,
+      "step": 1410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9312847506375413e-05,
+      "loss": 0.2053,
+      "step": 1411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311633097458608e-05,
+      "loss": 0.217,
+      "step": 1412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931041765462297e-05,
+      "loss": 0.1843,
+      "step": 1413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309201178003443e-05,
+      "loss": 0.2046,
+      "step": 1414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9307983667735106e-05,
+      "loss": 0.2554,
+      "step": 1415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930676512395315e-05,
+      "loss": 0.1655,
+      "step": 1416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305545546792863e-05,
+      "loss": 0.2053,
+      "step": 1417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304324936389675e-05,
+      "loss": 0.231,
+      "step": 1418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303103292879104e-05,
+      "loss": 0.1924,
+      "step": 1419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301880616396803e-05,
+      "loss": 0.2249,
+      "step": 1420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300656907078533e-05,
+      "loss": 0.2065,
+      "step": 1421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299432165060166e-05,
+      "loss": 0.2075,
+      "step": 1422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298206390477693e-05,
+      "loss": 0.2217,
+      "step": 1423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929697958346722e-05,
+      "loss": 0.248,
+      "step": 1424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9295751744164955e-05,
+      "loss": 0.2261,
+      "step": 1425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929452287270724e-05,
+      "loss": 0.2634,
+      "step": 1426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293292969230527e-05,
+      "loss": 0.228,
+      "step": 1427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292062033871374e-05,
+      "loss": 0.2363,
+      "step": 1428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9290830066766454e-05,
+      "loss": 0.229,
+      "step": 1429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9289597068052563e-05,
+      "loss": 0.2185,
+      "step": 1430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928836303786661e-05,
+      "loss": 0.2173,
+      "step": 1431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928712797634561e-05,
+      "loss": 0.238,
+      "step": 1432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285891883626698e-05,
+      "loss": 0.2537,
+      "step": 1433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284654759847127e-05,
+      "loss": 0.2354,
+      "step": 1434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283416605144264e-05,
+      "loss": 0.2146,
+      "step": 1435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282177419655586e-05,
+      "loss": 0.241,
+      "step": 1436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928093720351868e-05,
+      "loss": 0.2202,
+      "step": 1437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927969595687126e-05,
+      "loss": 0.179,
+      "step": 1438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278453679851147e-05,
+      "loss": 0.2234,
+      "step": 1439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9277210372596278e-05,
+      "loss": 0.2324,
+      "step": 1440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275966035244702e-05,
+      "loss": 0.2207,
+      "step": 1441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274720667934585e-05,
+      "loss": 0.1992,
+      "step": 1442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273474270804206e-05,
+      "loss": 0.2327,
+      "step": 1443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272226843991956e-05,
+      "loss": 0.218,
+      "step": 1444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927097838763635e-05,
+      "loss": 0.1938,
+      "step": 1445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926972890187601e-05,
+      "loss": 0.1885,
+      "step": 1446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268478386849664e-05,
+      "loss": 0.198,
+      "step": 1447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267226842696174e-05,
+      "loss": 0.197,
+      "step": 1448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265974269554497e-05,
+      "loss": 0.183,
+      "step": 1449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264720667563718e-05,
+      "loss": 0.2083,
+      "step": 1450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926346603686303e-05,
+      "loss": 0.271,
+      "step": 1451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926221037759174e-05,
+      "loss": 0.2073,
+      "step": 1452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926095368988927e-05,
+      "loss": 0.2136,
+      "step": 1453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925969597389516e-05,
+      "loss": 0.2383,
+      "step": 1454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258437229749054e-05,
+      "loss": 0.2104,
+      "step": 1455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925717745759072e-05,
+      "loss": 0.2568,
+      "step": 1456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255916657560042e-05,
+      "loss": 0.2188,
+      "step": 1457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254654829797007e-05,
+      "loss": 0.2283,
+      "step": 1458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925339197444173e-05,
+      "loss": 0.2356,
+      "step": 1459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252128091634418e-05,
+      "loss": 0.2371,
+      "step": 1460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925086318151542e-05,
+      "loss": 0.1914,
+      "step": 1461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924959724422518e-05,
+      "loss": 0.2197,
+      "step": 1462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248330279904262e-05,
+      "loss": 0.251,
+      "step": 1463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247062288693342e-05,
+      "loss": 0.2192,
+      "step": 1464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245793270733216e-05,
+      "loss": 0.2097,
+      "step": 1465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244523226164785e-05,
+      "loss": 0.2407,
+      "step": 1466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9243252155129075e-05,
+      "loss": 0.2046,
+      "step": 1467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.924198005776721e-05,
+      "loss": 0.2004,
+      "step": 1468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240706934220447e-05,
+      "loss": 0.2224,
+      "step": 1469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239432784630145e-05,
+      "loss": 0.2373,
+      "step": 1470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238157609137775e-05,
+      "loss": 0.2141,
+      "step": 1471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236881407884928e-05,
+      "loss": 0.1748,
+      "step": 1472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9235604181013306e-05,
+      "loss": 0.2263,
+      "step": 1473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234325928664736e-05,
+      "loss": 0.2268,
+      "step": 1474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9233046650981137e-05,
+      "loss": 0.2427,
+      "step": 1475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231766348104556e-05,
+      "loss": 0.2446,
+      "step": 1476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923048502017716e-05,
+      "loss": 0.2305,
+      "step": 1477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922920266734121e-05,
+      "loss": 0.2212,
+      "step": 1478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92279192897391e-05,
+      "loss": 0.2664,
+      "step": 1479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922663488751333e-05,
+      "loss": 0.2146,
+      "step": 1480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225349460806507e-05,
+      "loss": 0.1904,
+      "step": 1481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224063009761367e-05,
+      "loss": 0.2642,
+      "step": 1482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9222775534520744e-05,
+      "loss": 0.2146,
+      "step": 1483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9221487035227602e-05,
+      "loss": 0.2676,
+      "step": 1484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220197512025e-05,
+      "loss": 0.2344,
+      "step": 1485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218906965056126e-05,
+      "loss": 0.2373,
+      "step": 1486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921761539446427e-05,
+      "loss": 0.248,
+      "step": 1487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9216322800392855e-05,
+      "loss": 0.2126,
+      "step": 1488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215029182985392e-05,
+      "loss": 0.2056,
+      "step": 1489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921373454238552e-05,
+      "loss": 0.2136,
+      "step": 1490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9212438878736997e-05,
+      "loss": 0.2339,
+      "step": 1491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211142192183683e-05,
+      "loss": 0.2271,
+      "step": 1492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920984448286955e-05,
+      "loss": 0.2498,
+      "step": 1493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9208545750938693e-05,
+      "loss": 0.188,
+      "step": 1494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920724599653532e-05,
+      "loss": 0.2158,
+      "step": 1495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920594521980375e-05,
+      "loss": 0.1907,
+      "step": 1496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920464342088841e-05,
+      "loss": 0.1765,
+      "step": 1497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9203340599933852e-05,
+      "loss": 0.2476,
+      "step": 1498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202036757084725e-05,
+      "loss": 0.188,
+      "step": 1499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920073189248581e-05,
+      "loss": 0.1899,
+      "step": 1500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199426006281992e-05,
+      "loss": 0.1941,
+      "step": 1501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198119098618268e-05,
+      "loss": 0.2332,
+      "step": 1502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919681116963975e-05,
+      "loss": 0.207,
+      "step": 1503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195502219491663e-05,
+      "loss": 0.2051,
+      "step": 1504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194192248319355e-05,
+      "loss": 0.2002,
+      "step": 1505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919288125626827e-05,
+      "loss": 0.2063,
+      "step": 1506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191569243483977e-05,
+      "loss": 0.2156,
+      "step": 1507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9190256210112155e-05,
+      "loss": 0.2209,
+      "step": 1508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91889421562986e-05,
+      "loss": 0.2058,
+      "step": 1509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9187627082189212e-05,
+      "loss": 0.2339,
+      "step": 1510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186310987930014e-05,
+      "loss": 0.2449,
+      "step": 1511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918499387366714e-05,
+      "loss": 0.2529,
+      "step": 1512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918367573954684e-05,
+      "loss": 0.28,
+      "step": 1513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9182356585715464e-05,
+      "loss": 0.2212,
+      "step": 1514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918103641231949e-05,
+      "loss": 0.2344,
+      "step": 1515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9179715219505498e-05,
+      "loss": 0.209,
+      "step": 1516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91783930074202e-05,
+      "loss": 0.2366,
+      "step": 1517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917706977621039e-05,
+      "loss": 0.2031,
+      "step": 1518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9175745526023005e-05,
+      "loss": 0.2378,
+      "step": 1519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174420257005085e-05,
+      "loss": 0.2083,
+      "step": 1520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9173093969303775e-05,
+      "loss": 0.1902,
+      "step": 1521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9171766663066345e-05,
+      "loss": 0.2334,
+      "step": 1522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170438338440165e-05,
+      "loss": 0.2095,
+      "step": 1523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9169108995572735e-05,
+      "loss": 0.2451,
+      "step": 1524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9167778634611653e-05,
+      "loss": 0.2634,
+      "step": 1525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166447255704632e-05,
+      "loss": 0.2144,
+      "step": 1526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9165114858999512e-05,
+      "loss": 0.2175,
+      "step": 1527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916378144464423e-05,
+      "loss": 0.2209,
+      "step": 1528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9162447012786843e-05,
+      "loss": 0.2295,
+      "step": 1529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161111563575518e-05,
+      "loss": 0.2004,
+      "step": 1530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159775097158542e-05,
+      "loss": 0.1816,
+      "step": 1531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158437613684305e-05,
+      "loss": 0.2307,
+      "step": 1532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9157099113301316e-05,
+      "loss": 0.2097,
+      "step": 1533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155759596158195e-05,
+      "loss": 0.207,
+      "step": 1534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154419062403675e-05,
+      "loss": 0.2244,
+      "step": 1535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153077512186605e-05,
+      "loss": 0.2314,
+      "step": 1536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151734945655946e-05,
+      "loss": 0.22,
+      "step": 1537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915039136296076e-05,
+      "loss": 0.2202,
+      "step": 1538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149046764250244e-05,
+      "loss": 0.1799,
+      "step": 1539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914770114967369e-05,
+      "loss": 0.1995,
+      "step": 1540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914635451938051e-05,
+      "loss": 0.2251,
+      "step": 1541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145006873520227e-05,
+      "loss": 0.2563,
+      "step": 1542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143658212242475e-05,
+      "loss": 0.23,
+      "step": 1543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142308535697005e-05,
+      "loss": 0.1648,
+      "step": 1544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914095784403368e-05,
+      "loss": 0.1721,
+      "step": 1545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139606137402468e-05,
+      "loss": 0.2374,
+      "step": 1546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138253415953466e-05,
+      "loss": 0.2214,
+      "step": 1547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136899679836863e-05,
+      "loss": 0.2258,
+      "step": 1548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135544929202977e-05,
+      "loss": 0.1755,
+      "step": 1549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134189164202237e-05,
+      "loss": 0.2202,
+      "step": 1550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913283238498517e-05,
+      "loss": 0.2502,
+      "step": 1551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9131474591702438e-05,
+      "loss": 0.2288,
+      "step": 1552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913011578450479e-05,
+      "loss": 0.2119,
+      "step": 1553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128755963543118e-05,
+      "loss": 0.2163,
+      "step": 1554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91273951289684e-05,
+      "loss": 0.1832,
+      "step": 1555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126033280931734e-05,
+      "loss": 0.2263,
+      "step": 1556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9124670419584338e-05,
+      "loss": 0.189,
+      "step": 1557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9123306545077536e-05,
+      "loss": 0.196,
+      "step": 1558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912194165756277e-05,
+      "loss": 0.2117,
+      "step": 1559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9120575757191584e-05,
+      "loss": 0.2751,
+      "step": 1560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119208844115644e-05,
+      "loss": 0.2312,
+      "step": 1561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117840918486727e-05,
+      "loss": 0.2061,
+      "step": 1562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911647198045672e-05,
+      "loss": 0.2378,
+      "step": 1563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911510203017762e-05,
+      "loss": 0.1963,
+      "step": 1564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113731067801543e-05,
+      "loss": 0.176,
+      "step": 1565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112359093480716e-05,
+      "loss": 0.2507,
+      "step": 1566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911098610736747e-05,
+      "loss": 0.2527,
+      "step": 1567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9109612109614263e-05,
+      "loss": 0.1902,
+      "step": 1568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108237100373647e-05,
+      "loss": 0.2305,
+      "step": 1569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106861079798308e-05,
+      "loss": 0.271,
+      "step": 1570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105484048041024e-05,
+      "loss": 0.1853,
+      "step": 1571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104106005254696e-05,
+      "loss": 0.2214,
+      "step": 1572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102726951592338e-05,
+      "loss": 0.2205,
+      "step": 1573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101346887207065e-05,
+      "loss": 0.2083,
+      "step": 1574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9099965812252125e-05,
+      "loss": 0.2166,
+      "step": 1575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909858372688086e-05,
+      "loss": 0.2078,
+      "step": 1576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097200631246727e-05,
+      "loss": 0.2446,
+      "step": 1577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9095816525503304e-05,
+      "loss": 0.2041,
+      "step": 1578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094431409804273e-05,
+      "loss": 0.2053,
+      "step": 1579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909304528430343e-05,
+      "loss": 0.2563,
+      "step": 1580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091658149154683e-05,
+      "loss": 0.2524,
+      "step": 1581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9090270004512053e-05,
+      "loss": 0.2178,
+      "step": 1582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088880850529677e-05,
+      "loss": 0.2075,
+      "step": 1583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9087490687361794e-05,
+      "loss": 0.2126,
+      "step": 1584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9086099515162763e-05,
+      "loss": 0.2109,
+      "step": 1585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084707334087056e-05,
+      "loss": 0.2637,
+      "step": 1586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908331414428925e-05,
+      "loss": 0.2073,
+      "step": 1587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908191994592404e-05,
+      "loss": 0.2539,
+      "step": 1588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080524739146232e-05,
+      "loss": 0.1958,
+      "step": 1589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9079128524110745e-05,
+      "loss": 0.189,
+      "step": 1590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90777313009726e-05,
+      "loss": 0.1958,
+      "step": 1591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076333069886943e-05,
+      "loss": 0.1951,
+      "step": 1592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074933831009028e-05,
+      "loss": 0.1992,
+      "step": 1593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073533584494218e-05,
+      "loss": 0.2092,
+      "step": 1594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072132330497993e-05,
+      "loss": 0.2317,
+      "step": 1595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070730069175936e-05,
+      "loss": 0.2205,
+      "step": 1596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906932680068375e-05,
+      "loss": 0.2673,
+      "step": 1597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906792252517725e-05,
+      "loss": 0.2266,
+      "step": 1598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066517242812353e-05,
+      "loss": 0.2649,
+      "step": 1599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065110953745098e-05,
+      "loss": 0.1868,
+      "step": 1600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063703658131637e-05,
+      "loss": 0.2778,
+      "step": 1601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062295356128225e-05,
+      "loss": 0.2432,
+      "step": 1602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060886047891233e-05,
+      "loss": 0.2346,
+      "step": 1603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9059475733577147e-05,
+      "loss": 0.2444,
+      "step": 1604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058064413342555e-05,
+      "loss": 0.262,
+      "step": 1605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905665208734417e-05,
+      "loss": 0.2378,
+      "step": 1606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055238755738805e-05,
+      "loss": 0.2214,
+      "step": 1607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053824418683395e-05,
+      "loss": 0.1965,
+      "step": 1608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052409076334974e-05,
+      "loss": 0.2146,
+      "step": 1609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90509927288507e-05,
+      "loss": 0.2087,
+      "step": 1610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9049575376387838e-05,
+      "loss": 0.1919,
+      "step": 1611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048157019103758e-05,
+      "loss": 0.2083,
+      "step": 1612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046737657155953e-05,
+      "loss": 0.1986,
+      "step": 1613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904531729070202e-05,
+      "loss": 0.1951,
+      "step": 1614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9043895919899677e-05,
+      "loss": 0.259,
+      "step": 1615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042473544906733e-05,
+      "loss": 0.2253,
+      "step": 1616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041050165881126e-05,
+      "loss": 0.1931,
+      "step": 1617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9039625782980907e-05,
+      "loss": 0.1858,
+      "step": 1618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903820039636423e-05,
+      "loss": 0.2588,
+      "step": 1619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903677400618936e-05,
+      "loss": 0.2563,
+      "step": 1620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903534661261468e-05,
+      "loss": 0.2173,
+      "step": 1621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903391821579868e-05,
+      "loss": 0.1887,
+      "step": 1622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032488815899958e-05,
+      "loss": 0.2009,
+      "step": 1623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9031058413077233e-05,
+      "loss": 0.2073,
+      "step": 1624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902962700748933e-05,
+      "loss": 0.2024,
+      "step": 1625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902819459929518e-05,
+      "loss": 0.2051,
+      "step": 1626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026761188653837e-05,
+      "loss": 0.2278,
+      "step": 1627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902532677572446e-05,
+      "loss": 0.2427,
+      "step": 1628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902389136066631e-05,
+      "loss": 0.2263,
+      "step": 1629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902245494363878e-05,
+      "loss": 0.1707,
+      "step": 1630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021017524801356e-05,
+      "loss": 0.199,
+      "step": 1631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019579104313645e-05,
+      "loss": 0.2019,
+      "step": 1632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018139682335363e-05,
+      "loss": 0.2056,
+      "step": 1633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016699259026332e-05,
+      "loss": 0.2029,
+      "step": 1634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015257834546492e-05,
+      "loss": 0.2002,
+      "step": 1635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9013815409055895e-05,
+      "loss": 0.207,
+      "step": 1636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012371982714698e-05,
+      "loss": 0.22,
+      "step": 1637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010927555683173e-05,
+      "loss": 0.2341,
+      "step": 1638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9009482128121698e-05,
+      "loss": 0.1775,
+      "step": 1639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9008035700190774e-05,
+      "loss": 0.1853,
+      "step": 1640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006588272051e-05,
+      "loss": 0.2354,
+      "step": 1641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900513984386309e-05,
+      "loss": 0.22,
+      "step": 1642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9003690415787882e-05,
+      "loss": 0.2261,
+      "step": 1643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002239987986296e-05,
+      "loss": 0.2075,
+      "step": 1644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900078856061939e-05,
+      "loss": 0.2231,
+      "step": 1645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8999336133848327e-05,
+      "loss": 0.1931,
+      "step": 1646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997882707834372e-05,
+      "loss": 0.302,
+      "step": 1647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996428282738906e-05,
+      "loss": 0.2573,
+      "step": 1648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994972858723425e-05,
+      "loss": 0.2378,
+      "step": 1649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899351643594953e-05,
+      "loss": 0.1863,
+      "step": 1650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8992059014578933e-05,
+      "loss": 0.2305,
+      "step": 1651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899060059477346e-05,
+      "loss": 0.2803,
+      "step": 1652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989141176695054e-05,
+      "loss": 0.2285,
+      "step": 1653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8987680760505753e-05,
+      "loss": 0.2341,
+      "step": 1654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986219346367717e-05,
+      "loss": 0.2129,
+      "step": 1655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984756934443215e-05,
+      "loss": 0.2339,
+      "step": 1656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898329352489463e-05,
+      "loss": 0.2087,
+      "step": 1657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981829117884446e-05,
+      "loss": 0.2173,
+      "step": 1658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8980363713575264e-05,
+      "loss": 0.2185,
+      "step": 1659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.89788973121298e-05,
+      "loss": 0.208,
+      "step": 1660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897742991371087e-05,
+      "loss": 0.2478,
+      "step": 1661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8975961518481412e-05,
+      "loss": 0.1597,
+      "step": 1662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897449212660447e-05,
+      "loss": 0.2192,
+      "step": 1663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973021738243193e-05,
+      "loss": 0.2231,
+      "step": 1664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8971550353560852e-05,
+      "loss": 0.2576,
+      "step": 1665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970077972720816e-05,
+      "loss": 0.1868,
+      "step": 1666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8968604595886578e-05,
+      "loss": 0.2227,
+      "step": 1667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896713022322173e-05,
+      "loss": 0.2056,
+      "step": 1668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896565485488998e-05,
+      "loss": 0.2224,
+      "step": 1669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964178491055144e-05,
+      "loss": 0.2354,
+      "step": 1670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962701131881153e-05,
+      "loss": 0.208,
+      "step": 1671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961222777532048e-05,
+      "loss": 0.2249,
+      "step": 1672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959743428171972e-05,
+      "loss": 0.208,
+      "step": 1673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8958263083965187e-05,
+      "loss": 0.2161,
+      "step": 1674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956781745076068e-05,
+      "loss": 0.187,
+      "step": 1675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955299411669093e-05,
+      "loss": 0.1946,
+      "step": 1676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895381608390885e-05,
+      "loss": 0.187,
+      "step": 1677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8952331761960044e-05,
+      "loss": 0.208,
+      "step": 1678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950846445987486e-05,
+      "loss": 0.2327,
+      "step": 1679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89493601361561e-05,
+      "loss": 0.2168,
+      "step": 1680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947872832630916e-05,
+      "loss": 0.229,
+      "step": 1681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946384535577078e-05,
+      "loss": 0.2412,
+      "step": 1682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8944895245159838e-05,
+      "loss": 0.2073,
+      "step": 1683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943404961544565e-05,
+      "loss": 0.1926,
+      "step": 1684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941913684896724e-05,
+      "loss": 0.2249,
+      "step": 1685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8940421415381908e-05,
+      "loss": 0.2625,
+      "step": 1686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938928153165802e-05,
+      "loss": 0.2393,
+      "step": 1687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937433898414223e-05,
+      "loss": 0.2168,
+      "step": 1688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8935938651293076e-05,
+      "loss": 0.2129,
+      "step": 1689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8934442411968387e-05,
+      "loss": 0.2285,
+      "step": 1690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8932945180606296e-05,
+      "loss": 0.2051,
+      "step": 1691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931446957373045e-05,
+      "loss": 0.2449,
+      "step": 1692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892994774243499e-05,
+      "loss": 0.2129,
+      "step": 1693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928447535958598e-05,
+      "loss": 0.2119,
+      "step": 1694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892694633811044e-05,
+      "loss": 0.1992,
+      "step": 1695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8925444149057206e-05,
+      "loss": 0.2297,
+      "step": 1696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923940968965694e-05,
+      "loss": 0.2351,
+      "step": 1697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8922436798002803e-05,
+      "loss": 0.1816,
+      "step": 1698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920931636335553e-05,
+      "loss": 0.1968,
+      "step": 1699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919425484131072e-05,
+      "loss": 0.2373,
+      "step": 1700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917918341556593e-05,
+      "loss": 0.2405,
+      "step": 1701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891641020877946e-05,
+      "loss": 0.1877,
+      "step": 1702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8914901085967136e-05,
+      "loss": 0.2112,
+      "step": 1703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913390973287176e-05,
+      "loss": 0.2329,
+      "step": 1704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911879870907266e-05,
+      "loss": 0.1677,
+      "step": 1705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910367778995186e-05,
+      "loss": 0.1934,
+      "step": 1706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890885469771883e-05,
+      "loss": 0.2212,
+      "step": 1707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907340627246204e-05,
+      "loss": 0.1978,
+      "step": 1708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890582556774543e-05,
+      "loss": 0.208,
+      "step": 1709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8904309519384726e-05,
+      "loss": 0.2163,
+      "step": 1710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902792482332425e-05,
+      "loss": 0.2539,
+      "step": 1711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890127445675698e-05,
+      "loss": 0.229,
+      "step": 1712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899755442826936e-05,
+      "loss": 0.2056,
+      "step": 1713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8898235440710962e-05,
+      "loss": 0.2188,
+      "step": 1714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889671445057783e-05,
+      "loss": 0.2322,
+      "step": 1715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895192472596425e-05,
+      "loss": 0.1707,
+      "step": 1716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889366950693574e-05,
+      "loss": 0.2256,
+      "step": 1717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892145553764877e-05,
+      "loss": 0.2249,
+      "step": 1718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889062061325305e-05,
+      "loss": 0.2407,
+      "step": 1719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889094685569577e-05,
+      "loss": 0.1907,
+      "step": 1720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888756777088389e-05,
+      "loss": 0.1951,
+      "step": 1721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888603986936554e-05,
+      "loss": 0.2905,
+      "step": 1722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888451098118416e-05,
+      "loss": 0.189,
+      "step": 1723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882981106509528e-05,
+      "loss": 0.2297,
+      "step": 1724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8881450245511502e-05,
+      "loss": 0.2317,
+      "step": 1725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879918398360067e-05,
+      "loss": 0.2415,
+      "step": 1726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878385565225314e-05,
+      "loss": 0.1885,
+      "step": 1727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876851746277434e-05,
+      "loss": 0.2563,
+      "step": 1728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887531694168674e-05,
+      "loss": 0.2312,
+      "step": 1729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8873781151623648e-05,
+      "loss": 0.1687,
+      "step": 1730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887224437625869e-05,
+      "loss": 0.2561,
+      "step": 1731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870706615762492e-05,
+      "loss": 0.1891,
+      "step": 1732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8869167870305806e-05,
+      "loss": 0.2153,
+      "step": 1733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867628140059485e-05,
+      "loss": 0.2012,
+      "step": 1734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8866087425194493e-05,
+      "loss": 0.2144,
+      "step": 1735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864545725881908e-05,
+      "loss": 0.26,
+      "step": 1736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8863003042292904e-05,
+      "loss": 0.2424,
+      "step": 1737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861459374598783e-05,
+      "loss": 0.2305,
+      "step": 1738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.885991472297094e-05,
+      "loss": 0.1819,
+      "step": 1739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8858369087580887e-05,
+      "loss": 0.1897,
+      "step": 1740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856822468600245e-05,
+      "loss": 0.2197,
+      "step": 1741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885527486620074e-05,
+      "loss": 0.2034,
+      "step": 1742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8853726280554215e-05,
+      "loss": 0.2371,
+      "step": 1743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852176711832614e-05,
+      "loss": 0.2086,
+      "step": 1744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850626160207998e-05,
+      "loss": 0.2009,
+      "step": 1745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849074625852527e-05,
+      "loss": 0.3069,
+      "step": 1746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8847522108938482e-05,
+      "loss": 0.2837,
+      "step": 1747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884596860963824e-05,
+      "loss": 0.2302,
+      "step": 1748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8844414128124294e-05,
+      "loss": 0.2512,
+      "step": 1749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8842858664569257e-05,
+      "loss": 0.2107,
+      "step": 1750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884130221914583e-05,
+      "loss": 0.2302,
+      "step": 1751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8839744792026837e-05,
+      "loss": 0.2161,
+      "step": 1752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838186383385205e-05,
+      "loss": 0.2329,
+      "step": 1753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8836626993393972e-05,
+      "loss": 0.177,
+      "step": 1754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883506662222629e-05,
+      "loss": 0.1868,
+      "step": 1755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883350527005541e-05,
+      "loss": 0.2173,
+      "step": 1756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8831942937054697e-05,
+      "loss": 0.2131,
+      "step": 1757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883037962339763e-05,
+      "loss": 0.1802,
+      "step": 1758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882881532925779e-05,
+      "loss": 0.2185,
+      "step": 1759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8827250054808864e-05,
+      "loss": 0.2671,
+      "step": 1760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8825683800224655e-05,
+      "loss": 0.1572,
+      "step": 1761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824116565679074e-05,
+      "loss": 0.1958,
+      "step": 1762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882254835134614e-05,
+      "loss": 0.2351,
+      "step": 1763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820979157399976e-05,
+      "loss": 0.2048,
+      "step": 1764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881940898401482e-05,
+      "loss": 0.241,
+      "step": 1765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817837831365015e-05,
+      "loss": 0.209,
+      "step": 1766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816265699625015e-05,
+      "loss": 0.2131,
+      "step": 1767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814692588969387e-05,
+      "loss": 0.1995,
+      "step": 1768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813118499572796e-05,
+      "loss": 0.2112,
+      "step": 1769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881154343161002e-05,
+      "loss": 0.2175,
+      "step": 1770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8809967385255952e-05,
+      "loss": 0.1823,
+      "step": 1771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808390360685586e-05,
+      "loss": 0.1772,
+      "step": 1772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806812358074024e-05,
+      "loss": 0.1934,
+      "step": 1773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805233377596484e-05,
+      "loss": 0.2065,
+      "step": 1774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880365341942829e-05,
+      "loss": 0.1965,
+      "step": 1775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802072483744867e-05,
+      "loss": 0.2253,
+      "step": 1776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880049057072176e-05,
+      "loss": 0.228,
+      "step": 1777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8798907680534615e-05,
+      "loss": 0.2354,
+      "step": 1778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8797323813359186e-05,
+      "loss": 0.2397,
+      "step": 1779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795738969371343e-05,
+      "loss": 0.2144,
+      "step": 1780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794153148747055e-05,
+      "loss": 0.197,
+      "step": 1781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792566351662405e-05,
+      "loss": 0.1611,
+      "step": 1782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790978578293584e-05,
+      "loss": 0.2012,
+      "step": 1783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8789389828816894e-05,
+      "loss": 0.2405,
+      "step": 1784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787800103408733e-05,
+      "loss": 0.2039,
+      "step": 1785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8786209402245624e-05,
+      "loss": 0.2488,
+      "step": 1786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878461772550419e-05,
+      "loss": 0.2144,
+      "step": 1787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8783025073361162e-05,
+      "loss": 0.1843,
+      "step": 1788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878143144599338e-05,
+      "loss": 0.2034,
+      "step": 1789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8779836843577796e-05,
+      "loss": 0.21,
+      "step": 1790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877824126629146e-05,
+      "loss": 0.2046,
+      "step": 1791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877664471431154e-05,
+      "loss": 0.2048,
+      "step": 1792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775047187815313e-05,
+      "loss": 0.2124,
+      "step": 1793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773448686980156e-05,
+      "loss": 0.207,
+      "step": 1794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771849211983567e-05,
+      "loss": 0.1895,
+      "step": 1795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770248763003135e-05,
+      "loss": 0.2205,
+      "step": 1796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768647340216567e-05,
+      "loss": 0.2085,
+      "step": 1797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767044943801683e-05,
+      "loss": 0.186,
+      "step": 1798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87654415739364e-05,
+      "loss": 0.2107,
+      "step": 1799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876383723079875e-05,
+      "loss": 0.1926,
+      "step": 1800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8762231914566877e-05,
+      "loss": 0.23,
+      "step": 1801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8760625625419014e-05,
+      "loss": 0.2788,
+      "step": 1802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8759018363533528e-05,
+      "loss": 0.2119,
+      "step": 1803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875741012908888e-05,
+      "loss": 0.1865,
+      "step": 1804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755800922263633e-05,
+      "loss": 0.2087,
+      "step": 1805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8754190743236476e-05,
+      "loss": 0.1992,
+      "step": 1806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875257959218619e-05,
+      "loss": 0.1824,
+      "step": 1807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8750967469291666e-05,
+      "loss": 0.2178,
+      "step": 1808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874935437473191e-05,
+      "loss": 0.2068,
+      "step": 1809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874774030868604e-05,
+      "loss": 0.1831,
+      "step": 1810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746125271333257e-05,
+      "loss": 0.2188,
+      "step": 1811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8744509262852902e-05,
+      "loss": 0.2368,
+      "step": 1812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87428922834244e-05,
+      "loss": 0.2178,
+      "step": 1813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.87412743332273e-05,
+      "loss": 0.241,
+      "step": 1814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8739655412441243e-05,
+      "loss": 0.2114,
+      "step": 1815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873803552124599e-05,
+      "loss": 0.2209,
+      "step": 1816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873641465982141e-05,
+      "loss": 0.1998,
+      "step": 1817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734792828347472e-05,
+      "loss": 0.2332,
+      "step": 1818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733170027004254e-05,
+      "loss": 0.2419,
+      "step": 1819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731546255971948e-05,
+      "loss": 0.2158,
+      "step": 1820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872992151543085e-05,
+      "loss": 0.2478,
+      "step": 1821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728295805561355e-05,
+      "loss": 0.2205,
+      "step": 1822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726669126543985e-05,
+      "loss": 0.2031,
+      "step": 1823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725041478559354e-05,
+      "loss": 0.2424,
+      "step": 1824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723412861788187e-05,
+      "loss": 0.1997,
+      "step": 1825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872178327641132e-05,
+      "loss": 0.1929,
+      "step": 1826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720152722609692e-05,
+      "loss": 0.2148,
+      "step": 1827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718521200564352e-05,
+      "loss": 0.1965,
+      "step": 1828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716888710456458e-05,
+      "loss": 0.2092,
+      "step": 1829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715255252467274e-05,
+      "loss": 0.1748,
+      "step": 1830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871362082677817e-05,
+      "loss": 0.2371,
+      "step": 1831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711985433570628e-05,
+      "loss": 0.177,
+      "step": 1832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8710349073026227e-05,
+      "loss": 0.2358,
+      "step": 1833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8708711745326668e-05,
+      "loss": 0.187,
+      "step": 1834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870707345065375e-05,
+      "loss": 0.2087,
+      "step": 1835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705434189189374e-05,
+      "loss": 0.1598,
+      "step": 1836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870379396111557e-05,
+      "loss": 0.1997,
+      "step": 1837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.870215276661445e-05,
+      "loss": 0.1763,
+      "step": 1838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8700510605868246e-05,
+      "loss": 0.1902,
+      "step": 1839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.86988674790593e-05,
+      "loss": 0.2405,
+      "step": 1840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697223386370048e-05,
+      "loss": 0.2034,
+      "step": 1841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695578327983054e-05,
+      "loss": 0.1941,
+      "step": 1842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8693932304080967e-05,
+      "loss": 0.2244,
+      "step": 1843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869228531484656e-05,
+      "loss": 0.2402,
+      "step": 1844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690637360462706e-05,
+      "loss": 0.2056,
+      "step": 1845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868898844111238e-05,
+      "loss": 0.2019,
+      "step": 1846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868733855697868e-05,
+      "loss": 0.2007,
+      "step": 1847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685687708244794e-05,
+      "loss": 0.2058,
+      "step": 1848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684035895094025e-05,
+      "loss": 0.1704,
+      "step": 1849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8682383117709783e-05,
+      "loss": 0.2119,
+      "step": 1850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680729376275584e-05,
+      "loss": 0.2288,
+      "step": 1851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8679074670975056e-05,
+      "loss": 0.2144,
+      "step": 1852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677419001991924e-05,
+      "loss": 0.22,
+      "step": 1853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675762369510027e-05,
+      "loss": 0.2356,
+      "step": 1854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867410477371331e-05,
+      "loss": 0.1891,
+      "step": 1855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8672446214785824e-05,
+      "loss": 0.21,
+      "step": 1856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8670786692911727e-05,
+      "loss": 0.2122,
+      "step": 1857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8669126208275286e-05,
+      "loss": 0.1565,
+      "step": 1858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667464761060874e-05,
+      "loss": 0.238,
+      "step": 1859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665802351452966e-05,
+      "loss": 0.2087,
+      "step": 1860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664138979636152e-05,
+      "loss": 0.209,
+      "step": 1861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866247464579512e-05,
+      "loss": 0.2179,
+      "step": 1862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8660809350114673e-05,
+      "loss": 0.2188,
+      "step": 1863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865914309277972e-05,
+      "loss": 0.1978,
+      "step": 1864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657475873975267e-05,
+      "loss": 0.2292,
+      "step": 1865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865580769388644e-05,
+      "loss": 0.1749,
+      "step": 1866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654138552698463e-05,
+      "loss": 0.1775,
+      "step": 1867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652468450596673e-05,
+      "loss": 0.2197,
+      "step": 1868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8650797387766502e-05,
+      "loss": 0.1809,
+      "step": 1869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864912536439351e-05,
+      "loss": 0.1809,
+      "step": 1870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8647452380663335e-05,
+      "loss": 0.2112,
+      "step": 1871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8645778436761748e-05,
+      "loss": 0.1838,
+      "step": 1872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8644103532874612e-05,
+      "loss": 0.1736,
+      "step": 1873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86424276691879e-05,
+      "loss": 0.2146,
+      "step": 1874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864075084588769e-05,
+      "loss": 0.1863,
+      "step": 1875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8639073063160172e-05,
+      "loss": 0.2034,
+      "step": 1876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863739432119164e-05,
+      "loss": 0.2346,
+      "step": 1877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635714620168488e-05,
+      "loss": 0.2476,
+      "step": 1878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634033960277226e-05,
+      "loss": 0.2383,
+      "step": 1879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863235234170446e-05,
+      "loss": 0.2388,
+      "step": 1880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630669764636922e-05,
+      "loss": 0.2365,
+      "step": 1881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8628986229261426e-05,
+      "loss": 0.2144,
+      "step": 1882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627301735764907e-05,
+      "loss": 0.2273,
+      "step": 1883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625616284334405e-05,
+      "loss": 0.2275,
+      "step": 1884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862392987515706e-05,
+      "loss": 0.2012,
+      "step": 1885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622242508420123e-05,
+      "loss": 0.2073,
+      "step": 1886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8620554184310954e-05,
+      "loss": 0.228,
+      "step": 1887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618864903017018e-05,
+      "loss": 0.1953,
+      "step": 1888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8617174664725877e-05,
+      "loss": 0.1968,
+      "step": 1889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861548346962522e-05,
+      "loss": 0.1971,
+      "step": 1890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613791317902815e-05,
+      "loss": 0.1794,
+      "step": 1891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861209820974656e-05,
+      "loss": 0.2188,
+      "step": 1892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610404145344445e-05,
+      "loss": 0.1853,
+      "step": 1893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860870912488457e-05,
+      "loss": 0.2183,
+      "step": 1894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8607013148555148e-05,
+      "loss": 0.1824,
+      "step": 1895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605316216544485e-05,
+      "loss": 0.2241,
+      "step": 1896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603618329041002e-05,
+      "loss": 0.2161,
+      "step": 1897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601919486233227e-05,
+      "loss": 0.2292,
+      "step": 1898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860021968830979e-05,
+      "loss": 0.2207,
+      "step": 1899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8598518935459424e-05,
+      "loss": 0.2175,
+      "step": 1900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859681722787098e-05,
+      "loss": 0.2173,
+      "step": 1901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85951145657334e-05,
+      "loss": 0.2239,
+      "step": 1902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593410949235747e-05,
+      "loss": 0.1919,
+      "step": 1903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859170637856718e-05,
+      "loss": 0.2485,
+      "step": 1904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859000085391696e-05,
+      "loss": 0.1812,
+      "step": 1905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8588294375474466e-05,
+      "loss": 0.2097,
+      "step": 1906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586586943429177e-05,
+      "loss": 0.1838,
+      "step": 1907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584878557970677e-05,
+      "loss": 0.2278,
+      "step": 1908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583169219288658e-05,
+      "loss": 0.1987,
+      "step": 1909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581458927572912e-05,
+      "loss": 0.1863,
+      "step": 1910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857974768301335e-05,
+      "loss": 0.225,
+      "step": 1911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857803548579997e-05,
+      "loss": 0.2339,
+      "step": 1912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576322336122898e-05,
+      "loss": 0.2595,
+      "step": 1913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574608234172347e-05,
+      "loss": 0.2249,
+      "step": 1914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857289318013864e-05,
+      "loss": 0.1973,
+      "step": 1915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8571177174212214e-05,
+      "loss": 0.2041,
+      "step": 1916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85694602165836e-05,
+      "loss": 0.1722,
+      "step": 1917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856774230744345e-05,
+      "loss": 0.1973,
+      "step": 1918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566023446982503e-05,
+      "loss": 0.2197,
+      "step": 1919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564303635391617e-05,
+      "loss": 0.1926,
+      "step": 1920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562582872861748e-05,
+      "loss": 0.2231,
+      "step": 1921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856086115958397e-05,
+      "loss": 0.1926,
+      "step": 1922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559138495749445e-05,
+      "loss": 0.1973,
+      "step": 1923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557414881549453e-05,
+      "loss": 0.1792,
+      "step": 1924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555690317175375e-05,
+      "loss": 0.1995,
+      "step": 1925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85539648028187e-05,
+      "loss": 0.2001,
+      "step": 1926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855223833867102e-05,
+      "loss": 0.2729,
+      "step": 1927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855051092492403e-05,
+      "loss": 0.2124,
+      "step": 1928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548782561769535e-05,
+      "loss": 0.2271,
+      "step": 1929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547053249399448e-05,
+      "loss": 0.23,
+      "step": 1930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854532298800578e-05,
+      "loss": 0.2153,
+      "step": 1931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543591777780653e-05,
+      "loss": 0.1936,
+      "step": 1932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854185961891629e-05,
+      "loss": 0.2454,
+      "step": 1933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854012651160502e-05,
+      "loss": 0.2144,
+      "step": 1934
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8538392456039286e-05,
+      "loss": 0.1787,
+      "step": 1935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853665745241162e-05,
+      "loss": 0.1901,
+      "step": 1936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534921500914677e-05,
+      "loss": 0.2048,
+      "step": 1937
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8533184601741205e-05,
+      "loss": 0.2131,
+      "step": 1938
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8531446755084057e-05,
+      "loss": 0.2158,
+      "step": 1939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8529707961136202e-05,
+      "loss": 0.2461,
+      "step": 1940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527968220090705e-05,
+      "loss": 0.2083,
+      "step": 1941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526227532140734e-05,
+      "loss": 0.2419,
+      "step": 1942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.852448589747957e-05,
+      "loss": 0.1934,
+      "step": 1943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522743316300597e-05,
+      "loss": 0.1855,
+      "step": 1944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8520999788797303e-05,
+      "loss": 0.1704,
+      "step": 1945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8519255315163278e-05,
+      "loss": 0.2153,
+      "step": 1946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8517509895592223e-05,
+      "loss": 0.2148,
+      "step": 1947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515763530277938e-05,
+      "loss": 0.218,
+      "step": 1948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8514016219414334e-05,
+      "loss": 0.1836,
+      "step": 1949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851226796319542e-05,
+      "loss": 0.1713,
+      "step": 1950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8510518761815318e-05,
+      "loss": 0.1895,
+      "step": 1951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508768615468248e-05,
+      "loss": 0.2065,
+      "step": 1952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850701752434854e-05,
+      "loss": 0.1829,
+      "step": 1953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8505265488650625e-05,
+      "loss": 0.2068,
+      "step": 1954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850351250856904e-05,
+      "loss": 0.2217,
+      "step": 1955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501758584298436e-05,
+      "loss": 0.2288,
+      "step": 1956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8500003716033546e-05,
+      "loss": 0.1614,
+      "step": 1957
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8498247903969232e-05,
+      "loss": 0.2136,
+      "step": 1958
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8496491148300446e-05,
+      "loss": 0.1956,
+      "step": 1959
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494733449222254e-05,
+      "loss": 0.2368,
+      "step": 1960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492974806929816e-05,
+      "loss": 0.2092,
+      "step": 1961
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8491215221618413e-05,
+      "loss": 0.249,
+      "step": 1962
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848945469348341e-05,
+      "loss": 0.208,
+      "step": 1963
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487693222720297e-05,
+      "loss": 0.2114,
+      "step": 1964
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8485930809524655e-05,
+      "loss": 0.217,
+      "step": 1965
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848416745409217e-05,
+      "loss": 0.1753,
+      "step": 1966
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8482403156618644e-05,
+      "loss": 0.1941,
+      "step": 1967
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848063791729997e-05,
+      "loss": 0.23,
+      "step": 1968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8478871736332156e-05,
+      "loss": 0.2097,
+      "step": 1969
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847710461391131e-05,
+      "loss": 0.1892,
+      "step": 1970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847533655023364e-05,
+      "loss": 0.198,
+      "step": 1971
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473567545495464e-05,
+      "loss": 0.2439,
+      "step": 1972
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8471797599893213e-05,
+      "loss": 0.1978,
+      "step": 1973
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.84700267136234e-05,
+      "loss": 0.2148,
+      "step": 1974
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468254886882668e-05,
+      "loss": 0.2026,
+      "step": 1975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846648211986774e-05,
+      "loss": 0.189,
+      "step": 1976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8464708412775464e-05,
+      "loss": 0.1697,
+      "step": 1977
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846293376580278e-05,
+      "loss": 0.1719,
+      "step": 1978
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846115817914674e-05,
+      "loss": 0.176,
+      "step": 1979
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459381653004495e-05,
+      "loss": 0.1793,
+      "step": 1980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.84576041875733e-05,
+      "loss": 0.2109,
+      "step": 1981
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8455825783050517e-05,
+      "loss": 0.2217,
+      "step": 1982
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454046439633612e-05,
+      "loss": 0.2012,
+      "step": 1983
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8452266157520156e-05,
+      "loss": 0.2039,
+      "step": 1984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845048493690782e-05,
+      "loss": 0.1792,
+      "step": 1985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8448702777994386e-05,
+      "loss": 0.2209,
+      "step": 1986
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844691968097773e-05,
+      "loss": 0.1882,
+      "step": 1987
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445135646055844e-05,
+      "loss": 0.2209,
+      "step": 1988
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844335067342682e-05,
+      "loss": 0.1899,
+      "step": 1989
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8441564763288847e-05,
+      "loss": 0.1895,
+      "step": 1990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843977791584023e-05,
+      "loss": 0.1904,
+      "step": 1991
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8437990131279364e-05,
+      "loss": 0.2126,
+      "step": 1992
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8436201409804763e-05,
+      "loss": 0.1946,
+      "step": 1993
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434411751615032e-05,
+      "loss": 0.1658,
+      "step": 1994
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8432621156908894e-05,
+      "loss": 0.2012,
+      "step": 1995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8430829625885166e-05,
+      "loss": 0.1801,
+      "step": 1996
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842903715874276e-05,
+      "loss": 0.1743,
+      "step": 1997
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8427243755680718e-05,
+      "loss": 0.1503,
+      "step": 1998
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425449416898164e-05,
+      "loss": 0.1855,
+      "step": 1999
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8423654142594333e-05,
+      "loss": 0.23,
+      "step": 2000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842185793296856e-05,
+      "loss": 0.1936,
+      "step": 2001
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8420060788220296e-05,
+      "loss": 0.2468,
+      "step": 2002
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418262708549083e-05,
+      "loss": 0.1887,
+      "step": 2003
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8416463694154564e-05,
+      "loss": 0.1985,
+      "step": 2004
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414663745236504e-05,
+      "loss": 0.1702,
+      "step": 2005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8412862861994755e-05,
+      "loss": 0.1722,
+      "step": 2006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411061044629282e-05,
+      "loss": 0.2073,
+      "step": 2007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8409258293340146e-05,
+      "loss": 0.1802,
+      "step": 2008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8407454608327516e-05,
+      "loss": 0.1858,
+      "step": 2009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8405649989791666e-05,
+      "loss": 0.1836,
+      "step": 2010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8403844437932973e-05,
+      "loss": 0.157,
+      "step": 2011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8402037952951916e-05,
+      "loss": 0.2163,
+      "step": 2012
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.840023053504908e-05,
+      "loss": 0.2234,
+      "step": 2013
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398422184425144e-05,
+      "loss": 0.1749,
+      "step": 2014
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396612901280907e-05,
+      "loss": 0.198,
+      "step": 2015
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8394802685817262e-05,
+      "loss": 0.2024,
+      "step": 2016
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392991538235208e-05,
+      "loss": 0.2593,
+      "step": 2017
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839117945873584e-05,
+      "loss": 0.2314,
+      "step": 2018
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838936644752037e-05,
+      "loss": 0.2148,
+      "step": 2019
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8387552504790097e-05,
+      "loss": 0.1787,
+      "step": 2020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838573763074644e-05,
+      "loss": 0.2025,
+      "step": 2021
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838392182559091e-05,
+      "loss": 0.2468,
+      "step": 2022
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382105089525126e-05,
+      "loss": 0.1774,
+      "step": 2023
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8380287422750812e-05,
+      "loss": 0.2012,
+      "step": 2024
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378468825469792e-05,
+      "loss": 0.2068,
+      "step": 2025
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837664929788399e-05,
+      "loss": 0.1984,
+      "step": 2026
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374828840195445e-05,
+      "loss": 0.2332,
+      "step": 2027
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8373007452606283e-05,
+      "loss": 0.1853,
+      "step": 2028
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837118513531875e-05,
+      "loss": 0.1719,
+      "step": 2029
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369361888535187e-05,
+      "loss": 0.2212,
+      "step": 2030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836753771245803e-05,
+      "loss": 0.2349,
+      "step": 2031
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8365712607289835e-05,
+      "loss": 0.2068,
+      "step": 2032
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8363886573233254e-05,
+      "loss": 0.2161,
+      "step": 2033
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836205961049103e-05,
+      "loss": 0.1782,
+      "step": 2034
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8360231719266036e-05,
+      "loss": 0.2144,
+      "step": 2035
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8358402899761218e-05,
+      "loss": 0.23,
+      "step": 2036
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8356573152179646e-05,
+      "loss": 0.2278,
+      "step": 2037
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835474247672449e-05,
+      "loss": 0.2212,
+      "step": 2038
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8352910873599006e-05,
+      "loss": 0.2705,
+      "step": 2039
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8351078343006583e-05,
+      "loss": 0.2444,
+      "step": 2040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8349244885150684e-05,
+      "loss": 0.198,
+      "step": 2041
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834741050023489e-05,
+      "loss": 0.2114,
+      "step": 2042
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834557518846289e-05,
+      "loss": 0.1914,
+      "step": 2043
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834373895003846e-05,
+      "loss": 0.1892,
+      "step": 2044
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834190178516549e-05,
+      "loss": 0.1987,
+      "step": 2045
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8340063694047968e-05,
+      "loss": 0.2046,
+      "step": 2046
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338224676889987e-05,
+      "loss": 0.1963,
+      "step": 2047
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8336384733895748e-05,
+      "loss": 0.2095,
+      "step": 2048
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833454386526954e-05,
+      "loss": 0.186,
+      "step": 2049
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8332702071215773e-05,
+      "loss": 0.1821,
+      "step": 2050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8330859351938946e-05,
+      "loss": 0.2246,
+      "step": 2051
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8329015707643665e-05,
+      "loss": 0.2366,
+      "step": 2052
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8327171138534644e-05,
+      "loss": 0.2026,
+      "step": 2053
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8325325644816686e-05,
+      "loss": 0.1858,
+      "step": 2054
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832347922669472e-05,
+      "loss": 0.2493,
+      "step": 2055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321631884373753e-05,
+      "loss": 0.187,
+      "step": 2056
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319783618058902e-05,
+      "loss": 0.1686,
+      "step": 2057
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317934427955403e-05,
+      "loss": 0.1697,
+      "step": 2058
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316084314268568e-05,
+      "loss": 0.249,
+      "step": 2059
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831423327720383e-05,
+      "loss": 0.1882,
+      "step": 2060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8312381316966718e-05,
+      "loss": 0.252,
+      "step": 2061
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831052843376287e-05,
+      "loss": 0.1951,
+      "step": 2062
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8308674627798014e-05,
+      "loss": 0.2217,
+      "step": 2063
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830681989927799e-05,
+      "loss": 0.1887,
+      "step": 2064
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830496424840874e-05,
+      "loss": 0.1958,
+      "step": 2065
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8303107675396307e-05,
+      "loss": 0.2192,
+      "step": 2066
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830125018044683e-05,
+      "loss": 0.2012,
+      "step": 2067
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829939176376656e-05,
+      "loss": 0.2371,
+      "step": 2068
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829753242556185e-05,
+      "loss": 0.1877,
+      "step": 2069
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829567216603915e-05,
+      "loss": 0.1963,
+      "step": 2070
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8293810985405013e-05,
+      "loss": 0.218,
+      "step": 2071
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8291948883866095e-05,
+      "loss": 0.2134,
+      "step": 2072
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8290085861629153e-05,
+      "loss": 0.2124,
+      "step": 2073
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288221918901053e-05,
+      "loss": 0.2166,
+      "step": 2074
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8286357055888756e-05,
+      "loss": 0.25,
+      "step": 2075
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8284491272799327e-05,
+      "loss": 0.2124,
+      "step": 2076
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8282624569839934e-05,
+      "loss": 0.2166,
+      "step": 2077
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8280756947217845e-05,
+      "loss": 0.2302,
+      "step": 2078
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8278888405140435e-05,
+      "loss": 0.2297,
+      "step": 2079
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277018943815173e-05,
+      "loss": 0.2,
+      "step": 2080
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8275148563449642e-05,
+      "loss": 0.2256,
+      "step": 2081
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8273277264251515e-05,
+      "loss": 0.1824,
+      "step": 2082
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8271405046428572e-05,
+      "loss": 0.203,
+      "step": 2083
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82695319101887e-05,
+      "loss": 0.2046,
+      "step": 2084
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8267657855739874e-05,
+      "loss": 0.1833,
+      "step": 2085
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8265782883290187e-05,
+      "loss": 0.2212,
+      "step": 2086
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263906993047823e-05,
+      "loss": 0.2292,
+      "step": 2087
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826203018522108e-05,
+      "loss": 0.2251,
+      "step": 2088
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8260152460018336e-05,
+      "loss": 0.1836,
+      "step": 2089
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8258273817648097e-05,
+      "loss": 0.2153,
+      "step": 2090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8256394258318954e-05,
+      "loss": 0.208,
+      "step": 2091
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82545137822396e-05,
+      "loss": 0.1838,
+      "step": 2092
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825263238961884e-05,
+      "loss": 0.2432,
+      "step": 2093
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250750080665575e-05,
+      "loss": 0.1982,
+      "step": 2094
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82488668555888e-05,
+      "loss": 0.2178,
+      "step": 2095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824698271459763e-05,
+      "loss": 0.2397,
+      "step": 2096
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8245097657901262e-05,
+      "loss": 0.2068,
+      "step": 2097
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8243211685709002e-05,
+      "loss": 0.2129,
+      "step": 2098
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824132479823027e-05,
+      "loss": 0.1936,
+      "step": 2099
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823943699567457e-05,
+      "loss": 0.2151,
+      "step": 2100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8237548278251517e-05,
+      "loss": 0.1703,
+      "step": 2101
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823565864617082e-05,
+      "loss": 0.2236,
+      "step": 2102
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82337680996423e-05,
+      "loss": 0.1614,
+      "step": 2103
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8231876638875873e-05,
+      "loss": 0.1973,
+      "step": 2104
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8229984264081555e-05,
+      "loss": 0.2021,
+      "step": 2105
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822809097546947e-05,
+      "loss": 0.2339,
+      "step": 2106
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8226196773249836e-05,
+      "loss": 0.2197,
+      "step": 2107
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822430165763298e-05,
+      "loss": 0.1588,
+      "step": 2108
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8222405628829322e-05,
+      "loss": 0.217,
+      "step": 2109
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.822050868704939e-05,
+      "loss": 0.1958,
+      "step": 2110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218610832503814e-05,
+      "loss": 0.1628,
+      "step": 2111
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8216712065403318e-05,
+      "loss": 0.1899,
+      "step": 2112
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8214812385958738e-05,
+      "loss": 0.1973,
+      "step": 2113
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8212911794380996e-05,
+      "loss": 0.1609,
+      "step": 2114
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8211010290881132e-05,
+      "loss": 0.2346,
+      "step": 2115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8209107875670278e-05,
+      "loss": 0.1707,
+      "step": 2116
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820720454895967e-05,
+      "loss": 0.2146,
+      "step": 2117
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205300310960642e-05,
+      "loss": 0.1702,
+      "step": 2118
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820339516188463e-05,
+      "loss": 0.2087,
+      "step": 2119
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201489101943178e-05,
+      "loss": 0.2258,
+      "step": 2120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819958213134792e-05,
+      "loss": 0.2056,
+      "step": 2121
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.81976742503106e-05,
+      "loss": 0.2354,
+      "step": 2122
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819576545904306e-05,
+      "loss": 0.1907,
+      "step": 2123
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8193855757757242e-05,
+      "loss": 0.2051,
+      "step": 2124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191945146665192e-05,
+      "loss": 0.2014,
+      "step": 2125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819003362597905e-05,
+      "loss": 0.231,
+      "step": 2126
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8188121195911067e-05,
+      "loss": 0.241,
+      "step": 2127
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818620785667359e-05,
+      "loss": 0.2102,
+      "step": 2128
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8184293608479066e-05,
+      "loss": 0.1818,
+      "step": 2129
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818237845154004e-05,
+      "loss": 0.1882,
+      "step": 2130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8180462386069163e-05,
+      "loss": 0.1589,
+      "step": 2131
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8178545412279194e-05,
+      "loss": 0.2021,
+      "step": 2132
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8176627530382977e-05,
+      "loss": 0.197,
+      "step": 2133
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817470874059346e-05,
+      "loss": 0.1855,
+      "step": 2134
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8172789043123708e-05,
+      "loss": 0.2146,
+      "step": 2135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170868438186862e-05,
+      "loss": 0.1936,
+      "step": 2136
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168946925996192e-05,
+      "loss": 0.1982,
+      "step": 2137
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816702450676504e-05,
+      "loss": 0.1892,
+      "step": 2138
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8165101180706865e-05,
+      "loss": 0.1843,
+      "step": 2139
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816317694803523e-05,
+      "loss": 0.2483,
+      "step": 2140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8161251808963784e-05,
+      "loss": 0.219,
+      "step": 2141
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8159325763706294e-05,
+      "loss": 0.2256,
+      "step": 2142
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8157398812476613e-05,
+      "loss": 0.1926,
+      "step": 2143
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81554709554887e-05,
+      "loss": 0.2253,
+      "step": 2144
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815354219295662e-05,
+      "loss": 0.2036,
+      "step": 2145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8151612525094525e-05,
+      "loss": 0.1682,
+      "step": 2146
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8149681952116686e-05,
+      "loss": 0.2036,
+      "step": 2147
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8147750474237456e-05,
+      "loss": 0.2029,
+      "step": 2148
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8145818091671304e-05,
+      "loss": 0.1914,
+      "step": 2149
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814388480463279e-05,
+      "loss": 0.2151,
+      "step": 2150
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814195061333657e-05,
+      "loss": 0.219,
+      "step": 2151
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814001551799742e-05,
+      "loss": 0.2295,
+      "step": 2152
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8138079518830192e-05,
+      "loss": 0.2075,
+      "step": 2153
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8136142616049857e-05,
+      "loss": 0.1833,
+      "step": 2154
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8134204809871475e-05,
+      "loss": 0.1772,
+      "step": 2155
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132266100510217e-05,
+      "loss": 0.1877,
+      "step": 2156
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813032648818134e-05,
+      "loss": 0.2419,
+      "step": 2157
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128385973100215e-05,
+      "loss": 0.1904,
+      "step": 2158
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8126444555482306e-05,
+      "loss": 0.1575,
+      "step": 2159
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8124502235543174e-05,
+      "loss": 0.2554,
+      "step": 2160
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8122559013498496e-05,
+      "loss": 0.1917,
+      "step": 2161
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8120614889564026e-05,
+      "loss": 0.2437,
+      "step": 2162
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118669863955637e-05,
+      "loss": 0.1904,
+      "step": 2163
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8116723936889294e-05,
+      "loss": 0.2346,
+      "step": 2164
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811477710858106e-05,
+      "loss": 0.207,
+      "step": 2165
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112829379247106e-05,
+      "loss": 0.1702,
+      "step": 2166
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110880749103696e-05,
+      "loss": 0.2832,
+      "step": 2167
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108931218367198e-05,
+      "loss": 0.1855,
+      "step": 2168
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8106980787254075e-05,
+      "loss": 0.1904,
+      "step": 2169
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.81050294559809e-05,
+      "loss": 0.1899,
+      "step": 2170
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8103077224764333e-05,
+      "loss": 0.2043,
+      "step": 2171
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8101124093821144e-05,
+      "loss": 0.2297,
+      "step": 2172
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8099170063368197e-05,
+      "loss": 0.1672,
+      "step": 2173
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8097215133622464e-05,
+      "loss": 0.1643,
+      "step": 2174
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8095259304801002e-05,
+      "loss": 0.2075,
+      "step": 2175
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809330257712098e-05,
+      "loss": 0.2275,
+      "step": 2176
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091344950799665e-05,
+      "loss": 0.2065,
+      "step": 2177
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8089386426054423e-05,
+      "loss": 0.2083,
+      "step": 2178
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808742700310272e-05,
+      "loss": 0.2043,
+      "step": 2179
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8085466682162116e-05,
+      "loss": 0.1904,
+      "step": 2180
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808350546345028e-05,
+      "loss": 0.1958,
+      "step": 2181
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8081543347184972e-05,
+      "loss": 0.1892,
+      "step": 2182
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8079580333584062e-05,
+      "loss": 0.1912,
+      "step": 2183
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807761642286551e-05,
+      "loss": 0.2227,
+      "step": 2184
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8075651615247382e-05,
+      "loss": 0.1652,
+      "step": 2185
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8073685910947833e-05,
+      "loss": 0.1687,
+      "step": 2186
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8071719310185133e-05,
+      "loss": 0.1602,
+      "step": 2187
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806975181317764e-05,
+      "loss": 0.183,
+      "step": 2188
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806778342014382e-05,
+      "loss": 0.1921,
+      "step": 2189
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806581413130223e-05,
+      "loss": 0.1931,
+      "step": 2190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8063843946871527e-05,
+      "loss": 0.2146,
+      "step": 2191
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806187286707048e-05,
+      "loss": 0.2219,
+      "step": 2192
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8059900892117942e-05,
+      "loss": 0.207,
+      "step": 2193
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8057928022232872e-05,
+      "loss": 0.23,
+      "step": 2194
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805595425763433e-05,
+      "loss": 0.2227,
+      "step": 2195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8053979598541473e-05,
+      "loss": 0.24,
+      "step": 2196
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805200404517355e-05,
+      "loss": 0.2112,
+      "step": 2197
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805002759774993e-05,
+      "loss": 0.1602,
+      "step": 2198
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8048050256490058e-05,
+      "loss": 0.173,
+      "step": 2199
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8046072021613496e-05,
+      "loss": 0.1899,
+      "step": 2200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804409289333989e-05,
+      "loss": 0.1929,
+      "step": 2201
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8042112871888994e-05,
+      "loss": 0.1774,
+      "step": 2202
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804013195748067e-05,
+      "loss": 0.2034,
+      "step": 2203
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038150150334856e-05,
+      "loss": 0.1826,
+      "step": 2204
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803616745067161e-05,
+      "loss": 0.1614,
+      "step": 2205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803418385871108e-05,
+      "loss": 0.1917,
+      "step": 2206
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8032199374673512e-05,
+      "loss": 0.1842,
+      "step": 2207
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803021399877926e-05,
+      "loss": 0.1799,
+      "step": 2208
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8028227731248763e-05,
+      "loss": 0.22,
+      "step": 2209
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8026240572302567e-05,
+      "loss": 0.1904,
+      "step": 2210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024252522161326e-05,
+      "loss": 0.1658,
+      "step": 2211
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8022263581045775e-05,
+      "loss": 0.1931,
+      "step": 2212
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020273749176756e-05,
+      "loss": 0.2258,
+      "step": 2213
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8018283026775213e-05,
+      "loss": 0.2117,
+      "step": 2214
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8016291414062184e-05,
+      "loss": 0.2146,
+      "step": 2215
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8014298911258813e-05,
+      "loss": 0.2361,
+      "step": 2216
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012305518586334e-05,
+      "loss": 0.196,
+      "step": 2217
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8010311236266084e-05,
+      "loss": 0.21,
+      "step": 2218
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8008316064519498e-05,
+      "loss": 0.2078,
+      "step": 2219
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800632000356811e-05,
+      "loss": 0.2314,
+      "step": 2220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800432305363356e-05,
+      "loss": 0.1578,
+      "step": 2221
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800232521493757e-05,
+      "loss": 0.2007,
+      "step": 2222
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800032648770197e-05,
+      "loss": 0.2375,
+      "step": 2223
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.79983268721487e-05,
+      "loss": 0.2109,
+      "step": 2224
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7996326368499776e-05,
+      "loss": 0.1777,
+      "step": 2225
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7994324976977333e-05,
+      "loss": 0.2013,
+      "step": 2226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799232269780359e-05,
+      "loss": 0.1968,
+      "step": 2227
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799031953120087e-05,
+      "loss": 0.1824,
+      "step": 2228
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988315477391604e-05,
+      "loss": 0.2161,
+      "step": 2229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7986310536598304e-05,
+      "loss": 0.2651,
+      "step": 2230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798430470904359e-05,
+      "loss": 0.209,
+      "step": 2231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798229799495018e-05,
+      "loss": 0.1665,
+      "step": 2232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798029039454089e-05,
+      "loss": 0.2634,
+      "step": 2233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7978281908038633e-05,
+      "loss": 0.2046,
+      "step": 2234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976272535666424e-05,
+      "loss": 0.1738,
+      "step": 2235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7974262277647376e-05,
+      "loss": 0.1973,
+      "step": 2236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7972251134204693e-05,
+      "loss": 0.2068,
+      "step": 2237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797023910556168e-05,
+      "loss": 0.2024,
+      "step": 2238
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796822619194176e-05,
+      "loss": 0.1812,
+      "step": 2239
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7966212393568418e-05,
+      "loss": 0.2092,
+      "step": 2240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7964197710665262e-05,
+      "loss": 0.1946,
+      "step": 2241
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7962182143455997e-05,
+      "loss": 0.1914,
+      "step": 2242
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796016569216442e-05,
+      "loss": 0.1697,
+      "step": 2243
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7958148357014424e-05,
+      "loss": 0.2207,
+      "step": 2244
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795613013823001e-05,
+      "loss": 0.198,
+      "step": 2245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795411103603527e-05,
+      "loss": 0.2227,
+      "step": 2246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7952091050654392e-05,
+      "loss": 0.1748,
+      "step": 2247
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795007018231167e-05,
+      "loss": 0.2119,
+      "step": 2248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794804843123149e-05,
+      "loss": 0.2083,
+      "step": 2249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7946025797638332e-05,
+      "loss": 0.1855,
+      "step": 2250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7944002281756784e-05,
+      "loss": 0.2002,
+      "step": 2251
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794197788381153e-05,
+      "loss": 0.2244,
+      "step": 2252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793995260402735e-05,
+      "loss": 0.1737,
+      "step": 2253
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7937926442629116e-05,
+      "loss": 0.2058,
+      "step": 2254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.79358993998418e-05,
+      "loss": 0.1963,
+      "step": 2255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7933871475890484e-05,
+      "loss": 0.2632,
+      "step": 2256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793184267100034e-05,
+      "loss": 0.1608,
+      "step": 2257
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7929812985396625e-05,
+      "loss": 0.197,
+      "step": 2258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7927782419304716e-05,
+      "loss": 0.187,
+      "step": 2259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7925750972950074e-05,
+      "loss": 0.1844,
+      "step": 2260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792371864655826e-05,
+      "loss": 0.1797,
+      "step": 2261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7921685440354933e-05,
+      "loss": 0.2312,
+      "step": 2262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919651354565857e-05,
+      "loss": 0.2502,
+      "step": 2263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791761638941688e-05,
+      "loss": 0.2134,
+      "step": 2264
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7915580545133956e-05,
+      "loss": 0.2031,
+      "step": 2265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7913543821943135e-05,
+      "loss": 0.196,
+      "step": 2266
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911506220070563e-05,
+      "loss": 0.1843,
+      "step": 2267
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7909467739742494e-05,
+      "loss": 0.1772,
+      "step": 2268
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907428381185264e-05,
+      "loss": 0.2046,
+      "step": 2269
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7905388144625316e-05,
+      "loss": 0.2009,
+      "step": 2270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7903347030289184e-05,
+      "loss": 0.1455,
+      "step": 2271
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790130503840351e-05,
+      "loss": 0.197,
+      "step": 2272
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899262169195024e-05,
+      "loss": 0.2305,
+      "step": 2273
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789721842289055e-05,
+      "loss": 0.2202,
+      "step": 2274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895173799717027e-05,
+      "loss": 0.1875,
+      "step": 2275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789312829990147e-05,
+      "loss": 0.1687,
+      "step": 2276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789108192367101e-05,
+      "loss": 0.1868,
+      "step": 2277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7889034671252866e-05,
+      "loss": 0.2344,
+      "step": 2278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7886986542874348e-05,
+      "loss": 0.1799,
+      "step": 2279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884937538762872e-05,
+      "loss": 0.1785,
+      "step": 2280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7882887659145955e-05,
+      "loss": 0.2109,
+      "step": 2281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78808369042512e-05,
+      "loss": 0.2432,
+      "step": 2282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787878527430632e-05,
+      "loss": 0.1748,
+      "step": 2283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7876732769539108e-05,
+      "loss": 0.1797,
+      "step": 2284
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7874679390177476e-05,
+      "loss": 0.196,
+      "step": 2285
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7872625136449412e-05,
+      "loss": 0.2322,
+      "step": 2286
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787057000858301e-05,
+      "loss": 0.1681,
+      "step": 2287
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786851400680647e-05,
+      "loss": 0.2068,
+      "step": 2288
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7866457131348073e-05,
+      "loss": 0.2297,
+      "step": 2289
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7864399382436208e-05,
+      "loss": 0.1743,
+      "step": 2290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7862340760299358e-05,
+      "loss": 0.2549,
+      "step": 2291
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7860281265166097e-05,
+      "loss": 0.2241,
+      "step": 2292
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7858220897265107e-05,
+      "loss": 0.1763,
+      "step": 2293
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785615965682516e-05,
+      "loss": 0.2205,
+      "step": 2294
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785409754407513e-05,
+      "loss": 0.1901,
+      "step": 2295
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7852034559243977e-05,
+      "loss": 0.2202,
+      "step": 2296
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784997070256077e-05,
+      "loss": 0.1777,
+      "step": 2297
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7847905974254663e-05,
+      "loss": 0.2302,
+      "step": 2298
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784584037455492e-05,
+      "loss": 0.1765,
+      "step": 2299
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843773903690896e-05,
+      "loss": 0.178,
+      "step": 2300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7841706561892037e-05,
+      "loss": 0.1824,
+      "step": 2301
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839638349387892e-05,
+      "loss": 0.1846,
+      "step": 2302
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7837569266408107e-05,
+      "loss": 0.1719,
+      "step": 2303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7835499313182426e-05,
+      "loss": 0.2012,
+      "step": 2304
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783342848994068e-05,
+      "loss": 0.1763,
+      "step": 2305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7831356796912805e-05,
+      "loss": 0.1765,
+      "step": 2306
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782928423432883e-05,
+      "loss": 0.2139,
+      "step": 2307
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7827210802418894e-05,
+      "loss": 0.1681,
+      "step": 2308
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7825136501413206e-05,
+      "loss": 0.2051,
+      "step": 2309
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7823061331542094e-05,
+      "loss": 0.2195,
+      "step": 2310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7820985293035974e-05,
+      "loss": 0.2209,
+      "step": 2311
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7818908386125358e-05,
+      "loss": 0.1924,
+      "step": 2312
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7816830611040855e-05,
+      "loss": 0.1865,
+      "step": 2313
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814751968013176e-05,
+      "loss": 0.2253,
+      "step": 2314
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812672457273118e-05,
+      "loss": 0.2123,
+      "step": 2315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7810592079051586e-05,
+      "loss": 0.1843,
+      "step": 2316
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7808510833579565e-05,
+      "loss": 0.1943,
+      "step": 2317
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7806428721088156e-05,
+      "loss": 0.2117,
+      "step": 2318
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804345741808543e-05,
+      "loss": 0.2161,
+      "step": 2319
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780226189597201e-05,
+      "loss": 0.2195,
+      "step": 2320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7800177183809937e-05,
+      "loss": 0.2839,
+      "step": 2321
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.77980916055538e-05,
+      "loss": 0.2366,
+      "step": 2322
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7796005161435173e-05,
+      "loss": 0.167,
+      "step": 2323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7793917851685725e-05,
+      "loss": 0.1727,
+      "step": 2324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791829676537223e-05,
+      "loss": 0.2297,
+      "step": 2325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778974063622152e-05,
+      "loss": 0.2156,
+      "step": 2326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7787650730970578e-05,
+      "loss": 0.1909,
+      "step": 2327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7785559961016452e-05,
+      "loss": 0.1851,
+      "step": 2328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778346832659129e-05,
+      "loss": 0.2014,
+      "step": 2329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7781375827927334e-05,
+      "loss": 0.2146,
+      "step": 2330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779282465256928e-05,
+      "loss": 0.1755,
+      "step": 2331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777188238812503e-05,
+      "loss": 0.1985,
+      "step": 2332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7775093148826602e-05,
+      "loss": 0.1653,
+      "step": 2333
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7772997195531846e-05,
+      "loss": 0.2002,
+      "step": 2334
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777090037916096e-05,
+      "loss": 0.1953,
+      "step": 2335
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7768802699946767e-05,
+      "loss": 0.2064,
+      "step": 2336
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766704158122184e-05,
+      "loss": 0.1846,
+      "step": 2337
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776460475392022e-05,
+      "loss": 0.1689,
+      "step": 2338
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762504487573986e-05,
+      "loss": 0.2805,
+      "step": 2339
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776040335931668e-05,
+      "loss": 0.1907,
+      "step": 2340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7758301369381604e-05,
+      "loss": 0.1619,
+      "step": 2341
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7756198518002158e-05,
+      "loss": 0.207,
+      "step": 2342
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7754094805411825e-05,
+      "loss": 0.2041,
+      "step": 2343
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751990231844195e-05,
+      "loss": 0.1982,
+      "step": 2344
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774988479753295e-05,
+      "loss": 0.1748,
+      "step": 2345
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774777850271186e-05,
+      "loss": 0.2314,
+      "step": 2346
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774567134761481e-05,
+      "loss": 0.2063,
+      "step": 2347
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774356333247576e-05,
+      "loss": 0.1842,
+      "step": 2348
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7741454457528774e-05,
+      "loss": 0.1836,
+      "step": 2349
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7739344723008017e-05,
+      "loss": 0.1643,
+      "step": 2350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7737234129147737e-05,
+      "loss": 0.2039,
+      "step": 2351
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735122676182288e-05,
+      "loss": 0.1936,
+      "step": 2352
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7733010364346115e-05,
+      "loss": 0.1724,
+      "step": 2353
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7730897193873758e-05,
+      "loss": 0.1824,
+      "step": 2354
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7728783164999855e-05,
+      "loss": 0.1691,
+      "step": 2355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726668277959137e-05,
+      "loss": 0.2236,
+      "step": 2356
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772455253298643e-05,
+      "loss": 0.2065,
+      "step": 2357
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7722435930316656e-05,
+      "loss": 0.1638,
+      "step": 2358
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772031847018483e-05,
+      "loss": 0.2083,
+      "step": 2359
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771820015282607e-05,
+      "loss": 0.2275,
+      "step": 2360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7716080978475584e-05,
+      "loss": 0.1868,
+      "step": 2361
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7713960947368666e-05,
+      "loss": 0.1863,
+      "step": 2362
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7711840059740724e-05,
+      "loss": 0.2368,
+      "step": 2363
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7709718315827246e-05,
+      "loss": 0.2124,
+      "step": 2364
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7707595715863823e-05,
+      "loss": 0.2139,
+      "step": 2365
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7705472260086134e-05,
+      "loss": 0.1968,
+      "step": 2366
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7703347948729965e-05,
+      "loss": 0.2229,
+      "step": 2367
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7701222782031177e-05,
+      "loss": 0.2344,
+      "step": 2368
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7699096760225752e-05,
+      "loss": 0.2007,
+      "step": 2369
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696969883549742e-05,
+      "loss": 0.208,
+      "step": 2370
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7694842152239312e-05,
+      "loss": 0.2053,
+      "step": 2371
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7692713566530712e-05,
+      "loss": 0.1864,
+      "step": 2372
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7690584126660292e-05,
+      "loss": 0.1519,
+      "step": 2373
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768845383286449e-05,
+      "loss": 0.2258,
+      "step": 2374
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768632268537985e-05,
+      "loss": 0.1956,
+      "step": 2375
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7684190684443003e-05,
+      "loss": 0.1658,
+      "step": 2376
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7682057830290674e-05,
+      "loss": 0.1735,
+      "step": 2377
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767992412315968e-05,
+      "loss": 0.2046,
+      "step": 2378
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7677789563286948e-05,
+      "loss": 0.1985,
+      "step": 2379
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7675654150909485e-05,
+      "loss": 0.2273,
+      "step": 2380
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7673517886264392e-05,
+      "loss": 0.1592,
+      "step": 2381
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7671380769588878e-05,
+      "loss": 0.1616,
+      "step": 2382
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766924280112023e-05,
+      "loss": 0.2515,
+      "step": 2383
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667103981095844e-05,
+      "loss": 0.2014,
+      "step": 2384
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664964309753202e-05,
+      "loss": 0.1926,
+      "step": 2385
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7662823787329877e-05,
+      "loss": 0.1929,
+      "step": 2386
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766068241406355e-05,
+      "loss": 0.2019,
+      "step": 2387
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7658540190191992e-05,
+      "loss": 0.2087,
+      "step": 2388
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7656397115953055e-05,
+      "loss": 0.1975,
+      "step": 2389
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76542531915847e-05,
+      "loss": 0.2219,
+      "step": 2390
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652108417324976e-05,
+      "loss": 0.2266,
+      "step": 2391
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7649962793412036e-05,
+      "loss": 0.2241,
+      "step": 2392
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7647816320084113e-05,
+      "loss": 0.1953,
+      "step": 2393
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7645668997579544e-05,
+      "loss": 0.1461,
+      "step": 2394
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7643520826136752e-05,
+      "loss": 0.2144,
+      "step": 2395
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7641371805994266e-05,
+      "loss": 0.1941,
+      "step": 2396
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.76392219373907e-05,
+      "loss": 0.1763,
+      "step": 2397
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7637071220564765e-05,
+      "loss": 0.1936,
+      "step": 2398
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7634919655755267e-05,
+      "loss": 0.2148,
+      "step": 2399
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7632767243201104e-05,
+      "loss": 0.2224,
+      "step": 2400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.763061398314127e-05,
+      "loss": 0.1887,
+      "step": 2401
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7628459875814856e-05,
+      "loss": 0.1738,
+      "step": 2402
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7626304921461036e-05,
+      "loss": 0.2493,
+      "step": 2403
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7624149120319092e-05,
+      "loss": 0.2114,
+      "step": 2404
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7621992472628396e-05,
+      "loss": 0.2141,
+      "step": 2405
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619834978628406e-05,
+      "loss": 0.1831,
+      "step": 2406
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761767663855868e-05,
+      "loss": 0.1777,
+      "step": 2407
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7615517452658873e-05,
+      "loss": 0.1887,
+      "step": 2408
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613357421168728e-05,
+      "loss": 0.1851,
+      "step": 2409
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761119654432809e-05,
+      "loss": 0.1951,
+      "step": 2410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609034822376882e-05,
+      "loss": 0.2209,
+      "step": 2411
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760687225555514e-05,
+      "loss": 0.1804,
+      "step": 2412
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760470884410298e-05,
+      "loss": 0.1938,
+      "step": 2413
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760254458826062e-05,
+      "loss": 0.199,
+      "step": 2414
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7600379488268366e-05,
+      "loss": 0.2219,
+      "step": 2415
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598213544366624e-05,
+      "loss": 0.2319,
+      "step": 2416
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7596046756795886e-05,
+      "loss": 0.2021,
+      "step": 2417
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7593879125796746e-05,
+      "loss": 0.186,
+      "step": 2418
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759171065160988e-05,
+      "loss": 0.2019,
+      "step": 2419
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.758954133447607e-05,
+      "loss": 0.1843,
+      "step": 2420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758737117463619e-05,
+      "loss": 0.1716,
+      "step": 2421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7585200172331197e-05,
+      "loss": 0.1714,
+      "step": 2422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758302832780215e-05,
+      "loss": 0.2109,
+      "step": 2423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7580855641290203e-05,
+      "loss": 0.1929,
+      "step": 2424
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7578682113036603e-05,
+      "loss": 0.2109,
+      "step": 2425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7576507743282683e-05,
+      "loss": 0.1956,
+      "step": 2426
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7574332532269873e-05,
+      "loss": 0.2151,
+      "step": 2427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572156480239702e-05,
+      "loss": 0.1603,
+      "step": 2428
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756997958743379e-05,
+      "loss": 0.2063,
+      "step": 2429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7567801854093845e-05,
+      "loss": 0.229,
+      "step": 2430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756562328046167e-05,
+      "loss": 0.1863,
+      "step": 2431
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756344386677917e-05,
+      "loss": 0.2314,
+      "step": 2432
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756126361328833e-05,
+      "loss": 0.2021,
+      "step": 2433
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755908252023124e-05,
+      "loss": 0.2195,
+      "step": 2434
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7556900587850076e-05,
+      "loss": 0.1848,
+      "step": 2435
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755471781638711e-05,
+      "loss": 0.1953,
+      "step": 2436
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7552534206084703e-05,
+      "loss": 0.1807,
+      "step": 2437
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755034975718531e-05,
+      "loss": 0.1814,
+      "step": 2438
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7548164469931495e-05,
+      "loss": 0.2148,
+      "step": 2439
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754597834456589e-05,
+      "loss": 0.2141,
+      "step": 2440
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7543791381331236e-05,
+      "loss": 0.2097,
+      "step": 2441
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7541603580470364e-05,
+      "loss": 0.177,
+      "step": 2442
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753941494222619e-05,
+      "loss": 0.1497,
+      "step": 2443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753722546684174e-05,
+      "loss": 0.2134,
+      "step": 2444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535035154560114e-05,
+      "loss": 0.2327,
+      "step": 2445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532844005624522e-05,
+      "loss": 0.1895,
+      "step": 2446
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7530652020278247e-05,
+      "loss": 0.2041,
+      "step": 2447
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7528459198764688e-05,
+      "loss": 0.2058,
+      "step": 2448
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752626554132732e-05,
+      "loss": 0.1814,
+      "step": 2449
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7524071048209715e-05,
+      "loss": 0.2339,
+      "step": 2450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7521875719655544e-05,
+      "loss": 0.1927,
+      "step": 2451
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751967955590856e-05,
+      "loss": 0.2073,
+      "step": 2452
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751748255721262e-05,
+      "loss": 0.2087,
+      "step": 2453
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7515284723811664e-05,
+      "loss": 0.229,
+      "step": 2454
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751308605594973e-05,
+      "loss": 0.2053,
+      "step": 2455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751088655387094e-05,
+      "loss": 0.1853,
+      "step": 2456
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7508686217819532e-05,
+      "loss": 0.2083,
+      "step": 2457
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750648504803981e-05,
+      "loss": 0.1956,
+      "step": 2458
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7504283044776186e-05,
+      "loss": 0.1807,
+      "step": 2459
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7502080208273152e-05,
+      "loss": 0.2258,
+      "step": 2460
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7499876538775312e-05,
+      "loss": 0.2173,
+      "step": 2461
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749767203652734e-05,
+      "loss": 0.2209,
+      "step": 2462
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749546670177402e-05,
+      "loss": 0.1765,
+      "step": 2463
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749326053476022e-05,
+      "loss": 0.1963,
+      "step": 2464
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7491053535730898e-05,
+      "loss": 0.1978,
+      "step": 2465
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7488845704931113e-05,
+      "loss": 0.1732,
+      "step": 2466
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7486637042606016e-05,
+      "loss": 0.2305,
+      "step": 2467
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7484427549000836e-05,
+      "loss": 0.1707,
+      "step": 2468
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748221722436091e-05,
+      "loss": 0.204,
+      "step": 2469
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748000606893166e-05,
+      "loss": 0.1531,
+      "step": 2470
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747779408295861e-05,
+      "loss": 0.1792,
+      "step": 2471
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475581266687362e-05,
+      "loss": 0.1736,
+      "step": 2472
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747336762036361e-05,
+      "loss": 0.1946,
+      "step": 2473
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7471153144233158e-05,
+      "loss": 0.1854,
+      "step": 2474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468937838541885e-05,
+      "loss": 0.1682,
+      "step": 2475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7466721703535765e-05,
+      "loss": 0.1802,
+      "step": 2476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464504739460874e-05,
+      "loss": 0.178,
+      "step": 2477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7462286946563366e-05,
+      "loss": 0.1899,
+      "step": 2478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7460068325089502e-05,
+      "loss": 0.2249,
+      "step": 2479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7457848875285618e-05,
+      "loss": 0.1917,
+      "step": 2480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745562859739816e-05,
+      "loss": 0.1951,
+      "step": 2481
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745340749167365e-05,
+      "loss": 0.156,
+      "step": 2482
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7451185558358714e-05,
+      "loss": 0.1963,
+      "step": 2483
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744896279770006e-05,
+      "loss": 0.2234,
+      "step": 2484
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7446739209944496e-05,
+      "loss": 0.166,
+      "step": 2485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7444514795338917e-05,
+      "loss": 0.1558,
+      "step": 2486
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7442289554130307e-05,
+      "loss": 0.2002,
+      "step": 2487
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7440063486565755e-05,
+      "loss": 0.1663,
+      "step": 2488
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743783659289243e-05,
+      "loss": 0.1887,
+      "step": 2489
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743560887335759e-05,
+      "loss": 0.2075,
+      "step": 2490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7433380328208594e-05,
+      "loss": 0.1729,
+      "step": 2491
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7431150957692896e-05,
+      "loss": 0.1725,
+      "step": 2492
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7428920762058022e-05,
+      "loss": 0.179,
+      "step": 2493
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742668974155161e-05,
+      "loss": 0.1868,
+      "step": 2494
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7424457896421376e-05,
+      "loss": 0.2043,
+      "step": 2495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7422225226915138e-05,
+      "loss": 0.1746,
+      "step": 2496
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74199917332808e-05,
+      "loss": 0.1906,
+      "step": 2497
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741775741576636e-05,
+      "loss": 0.1929,
+      "step": 2498
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7415522274619902e-05,
+      "loss": 0.2283,
+      "step": 2499
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741328631008961e-05,
+      "loss": 0.175,
+      "step": 2500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741104952242375e-05,
+      "loss": 0.1616,
+      "step": 2501
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7408811911870685e-05,
+      "loss": 0.199,
+      "step": 2502
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740657347867887e-05,
+      "loss": 0.1699,
+      "step": 2503
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404334223096852e-05,
+      "loss": 0.2126,
+      "step": 2504
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740209414537326e-05,
+      "loss": 0.2053,
+      "step": 2505
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739985324575683e-05,
+      "loss": 0.2034,
+      "step": 2506
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7397611524496375e-05,
+      "loss": 0.2246,
+      "step": 2507
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7395368981840804e-05,
+      "loss": 0.2197,
+      "step": 2508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393125618039124e-05,
+      "loss": 0.2061,
+      "step": 2509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7390881433340424e-05,
+      "loss": 0.1995,
+      "step": 2510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7388636427993886e-05,
+      "loss": 0.1709,
+      "step": 2511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7386390602248787e-05,
+      "loss": 0.1726,
+      "step": 2512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738414395635449e-05,
+      "loss": 0.1855,
+      "step": 2513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7381896490560456e-05,
+      "loss": 0.201,
+      "step": 2514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737964820511623e-05,
+      "loss": 0.2163,
+      "step": 2515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737739910027145e-05,
+      "loss": 0.2107,
+      "step": 2516
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7375149176275847e-05,
+      "loss": 0.1797,
+      "step": 2517
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7372898433379243e-05,
+      "loss": 0.2617,
+      "step": 2518
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7370646871831546e-05,
+      "loss": 0.2361,
+      "step": 2519
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7368394491882757e-05,
+      "loss": 0.1687,
+      "step": 2520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7366141293782978e-05,
+      "loss": 0.1953,
+      "step": 2521
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363887277782388e-05,
+      "loss": 0.1853,
+      "step": 2522
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736163244413126e-05,
+      "loss": 0.2068,
+      "step": 2523
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7359376793079963e-05,
+      "loss": 0.2031,
+      "step": 2524
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7357120324878957e-05,
+      "loss": 0.208,
+      "step": 2525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735486303977878e-05,
+      "loss": 0.1921,
+      "step": 2526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7352604938030074e-05,
+      "loss": 0.1707,
+      "step": 2527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735034601988357e-05,
+      "loss": 0.2031,
+      "step": 2528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734808628559009e-05,
+      "loss": 0.1924,
+      "step": 2529
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7345825735400538e-05,
+      "loss": 0.2043,
+      "step": 2530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734356436956592e-05,
+      "loss": 0.1519,
+      "step": 2531
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734130218833732e-05,
+      "loss": 0.207,
+      "step": 2532
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339039191965924e-05,
+      "loss": 0.1995,
+      "step": 2533
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7336775380703005e-05,
+      "loss": 0.1771,
+      "step": 2534
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733451075479992e-05,
+      "loss": 0.2273,
+      "step": 2535
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7332245314508137e-05,
+      "loss": 0.1802,
+      "step": 2536
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7329979060079184e-05,
+      "loss": 0.1692,
+      "step": 2537
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7327711991764698e-05,
+      "loss": 0.1572,
+      "step": 2538
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7325444109816408e-05,
+      "loss": 0.1904,
+      "step": 2539
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7323175414486125e-05,
+      "loss": 0.1951,
+      "step": 2540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7320905906025752e-05,
+      "loss": 0.2,
+      "step": 2541
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7318635584687294e-05,
+      "loss": 0.1801,
+      "step": 2542
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7316364450722827e-05,
+      "loss": 0.241,
+      "step": 2543
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731409250438453e-05,
+      "loss": 0.22,
+      "step": 2544
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7311819745924672e-05,
+      "loss": 0.1664,
+      "step": 2545
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309546175595602e-05,
+      "loss": 0.179,
+      "step": 2546
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730727179364977e-05,
+      "loss": 0.2256,
+      "step": 2547
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7304996600339718e-05,
+      "loss": 0.2275,
+      "step": 2548
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730272059591806e-05,
+      "loss": 0.167,
+      "step": 2549
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7300443780637527e-05,
+      "loss": 0.1936,
+      "step": 2550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7298166154750914e-05,
+      "loss": 0.2119,
+      "step": 2551
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7295887718511123e-05,
+      "loss": 0.189,
+      "step": 2552
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729360847217114e-05,
+      "loss": 0.2195,
+      "step": 2553
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7291328415984038e-05,
+      "loss": 0.2046,
+      "step": 2554
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728904755020299e-05,
+      "loss": 0.1963,
+      "step": 2555
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7286765875081243e-05,
+      "loss": 0.2185,
+      "step": 2556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7284483390872156e-05,
+      "loss": 0.2285,
+      "step": 2557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7282200097829153e-05,
+      "loss": 0.1685,
+      "step": 2558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7279915996205768e-05,
+      "loss": 0.1899,
+      "step": 2559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727763108625561e-05,
+      "loss": 0.1792,
+      "step": 2560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7275345368232392e-05,
+      "loss": 0.207,
+      "step": 2561
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7273058842389906e-05,
+      "loss": 0.2053,
+      "step": 2562
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7270771508982035e-05,
+      "loss": 0.2156,
+      "step": 2563
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.726848336826275e-05,
+      "loss": 0.1737,
+      "step": 2564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7266194420486123e-05,
+      "loss": 0.2166,
+      "step": 2565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7263904665906302e-05,
+      "loss": 0.1921,
+      "step": 2566
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7261614104777534e-05,
+      "loss": 0.1696,
+      "step": 2567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725932273735415e-05,
+      "loss": 0.1725,
+      "step": 2568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7257030563890575e-05,
+      "loss": 0.2004,
+      "step": 2569
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725473758464131e-05,
+      "loss": 0.1851,
+      "step": 2570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7252443799860973e-05,
+      "loss": 0.2681,
+      "step": 2571
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7250149209804247e-05,
+      "loss": 0.1995,
+      "step": 2572
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724785381472591e-05,
+      "loss": 0.248,
+      "step": 2573
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7245557614880836e-05,
+      "loss": 0.1702,
+      "step": 2574
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7243260610523976e-05,
+      "loss": 0.1831,
+      "step": 2575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7240962801910387e-05,
+      "loss": 0.1669,
+      "step": 2576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238664189295204e-05,
+      "loss": 0.22,
+      "step": 2577
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7236364772933653e-05,
+      "loss": 0.1582,
+      "step": 2578
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723406455308105e-05,
+      "loss": 0.2021,
+      "step": 2579
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231763529992796e-05,
+      "loss": 0.2063,
+      "step": 2580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7229461703924398e-05,
+      "loss": 0.1699,
+      "step": 2581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227159075131424e-05,
+      "loss": 0.2075,
+      "step": 2582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7224855643869563e-05,
+      "loss": 0.1892,
+      "step": 2583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222551410394564e-05,
+      "loss": 0.2163,
+      "step": 2584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220246374962283e-05,
+      "loss": 0.1816,
+      "step": 2585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7217940537828658e-05,
+      "loss": 0.1892,
+      "step": 2586
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721563389924972e-05,
+      "loss": 0.163,
+      "step": 2587
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721332645948159e-05,
+      "loss": 0.2217,
+      "step": 2588
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721101821878047e-05,
+      "loss": 0.2212,
+      "step": 2589
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720870917740266e-05,
+      "loss": 0.1987,
+      "step": 2590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7206399335604536e-05,
+      "loss": 0.2368,
+      "step": 2591
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720408869364258e-05,
+      "loss": 0.166,
+      "step": 2592
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7201777251773356e-05,
+      "loss": 0.1914,
+      "step": 2593
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7199465010253508e-05,
+      "loss": 0.2458,
+      "step": 2594
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7197151969339783e-05,
+      "loss": 0.1738,
+      "step": 2595
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194838129289006e-05,
+      "loss": 0.2214,
+      "step": 2596
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7192523490358095e-05,
+      "loss": 0.1919,
+      "step": 2597
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7190208052804056e-05,
+      "loss": 0.1921,
+      "step": 2598
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718789181688399e-05,
+      "loss": 0.1929,
+      "step": 2599
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718557478285507e-05,
+      "loss": 0.2332,
+      "step": 2600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7183256950974578e-05,
+      "loss": 0.2344,
+      "step": 2601
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180938321499867e-05,
+      "loss": 0.1704,
+      "step": 2602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7178618894688394e-05,
+      "loss": 0.2019,
+      "step": 2603
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717629867079769e-05,
+      "loss": 0.1565,
+      "step": 2604
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7173977650085385e-05,
+      "loss": 0.1877,
+      "step": 2605
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7171655832809194e-05,
+      "loss": 0.2039,
+      "step": 2606
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716933321922692e-05,
+      "loss": 0.1934,
+      "step": 2607
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7167009809596457e-05,
+      "loss": 0.1833,
+      "step": 2608
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716468560417578e-05,
+      "loss": 0.2427,
+      "step": 2609
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7162360603222964e-05,
+      "loss": 0.2036,
+      "step": 2610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716003480699616e-05,
+      "loss": 0.2136,
+      "step": 2611
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7157708215753615e-05,
+      "loss": 0.1992,
+      "step": 2612
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715538082975367e-05,
+      "loss": 0.1992,
+      "step": 2613
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7153052649254734e-05,
+      "loss": 0.1897,
+      "step": 2614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7150723674515322e-05,
+      "loss": 0.1809,
+      "step": 2615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148393905794036e-05,
+      "loss": 0.1719,
+      "step": 2616
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146063343349556e-05,
+      "loss": 0.1671,
+      "step": 2617
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7143731987440664e-05,
+      "loss": 0.1732,
+      "step": 2618
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141399838326213e-05,
+      "loss": 0.1714,
+      "step": 2619
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139066896265162e-05,
+      "loss": 0.1931,
+      "step": 2620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7136733161516547e-05,
+      "loss": 0.1826,
+      "step": 2621
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7134398634339492e-05,
+      "loss": 0.2042,
+      "step": 2622
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7132063314993213e-05,
+      "loss": 0.1892,
+      "step": 2623
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7129727203737018e-05,
+      "loss": 0.2065,
+      "step": 2624
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7127390300830288e-05,
+      "loss": 0.1733,
+      "step": 2625
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712505260653251e-05,
+      "loss": 0.1987,
+      "step": 2626
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7122714121103244e-05,
+      "loss": 0.2446,
+      "step": 2627
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712037484480215e-05,
+      "loss": 0.1764,
+      "step": 2628
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7118034777888957e-05,
+      "loss": 0.197,
+      "step": 2629
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7115693920623516e-05,
+      "loss": 0.1836,
+      "step": 2630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7113352273265727e-05,
+      "loss": 0.2035,
+      "step": 2631
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7111009836075604e-05,
+      "loss": 0.1951,
+      "step": 2632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7108666609313233e-05,
+      "loss": 0.2209,
+      "step": 2633
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7106322593238802e-05,
+      "loss": 0.1936,
+      "step": 2634
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103977788112578e-05,
+      "loss": 0.2498,
+      "step": 2635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710163219419491e-05,
+      "loss": 0.2285,
+      "step": 2636
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709928581174625e-05,
+      "loss": 0.1796,
+      "step": 2637
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7096938641027124e-05,
+      "loss": 0.1611,
+      "step": 2638
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709459068229815e-05,
+      "loss": 0.1873,
+      "step": 2639
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092241935820036e-05,
+      "loss": 0.1868,
+      "step": 2640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7089892401853577e-05,
+      "loss": 0.2124,
+      "step": 2641
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7087542080659654e-05,
+      "loss": 0.1943,
+      "step": 2642
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7085190972499228e-05,
+      "loss": 0.1948,
+      "step": 2643
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7082839077633363e-05,
+      "loss": 0.2188,
+      "step": 2644
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70804863963232e-05,
+      "loss": 0.1815,
+      "step": 2645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707813292882997e-05,
+      "loss": 0.2166,
+      "step": 2646
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707577867541499e-05,
+      "loss": 0.1882,
+      "step": 2647
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707342363633966e-05,
+      "loss": 0.1943,
+      "step": 2648
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.2043,
+      "step": 2649
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7068711202254024e-05,
+      "loss": 0.1785,
+      "step": 2650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7066353807766957e-05,
+      "loss": 0.1921,
+      "step": 2651
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706399562866604e-05,
+      "loss": 0.199,
+      "step": 2652
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706163666521311e-05,
+      "loss": 0.1892,
+      "step": 2653
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7059276917670097e-05,
+      "loss": 0.2119,
+      "step": 2654
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705691638629901e-05,
+      "loss": 0.1746,
+      "step": 2655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7054555071361954e-05,
+      "loss": 0.1755,
+      "step": 2656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052192973121124e-05,
+      "loss": 0.2373,
+      "step": 2657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7049830091838788e-05,
+      "loss": 0.1958,
+      "step": 2658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7047466427777313e-05,
+      "loss": 0.2046,
+      "step": 2659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7045101981199144e-05,
+      "loss": 0.2566,
+      "step": 2660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7042736752366828e-05,
+      "loss": 0.1876,
+      "step": 2661
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7040370741542978e-05,
+      "loss": 0.2068,
+      "step": 2662
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703800394899031e-05,
+      "loss": 0.2212,
+      "step": 2663
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7035636374971618e-05,
+      "loss": 0.1659,
+      "step": 2664
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703326801974979e-05,
+      "loss": 0.2034,
+      "step": 2665
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030898883587794e-05,
+      "loss": 0.2024,
+      "step": 2666
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7028528966748686e-05,
+      "loss": 0.2158,
+      "step": 2667
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7026158269495612e-05,
+      "loss": 0.2319,
+      "step": 2668
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7023786792091805e-05,
+      "loss": 0.238,
+      "step": 2669
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702141453480058e-05,
+      "loss": 0.1877,
+      "step": 2670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701904149788534e-05,
+      "loss": 0.2002,
+      "step": 2671
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701666768160958e-05,
+      "loss": 0.1757,
+      "step": 2672
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701429308623687e-05,
+      "loss": 0.1647,
+      "step": 2673
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701191771203088e-05,
+      "loss": 0.2166,
+      "step": 2674
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700954155925536e-05,
+      "loss": 0.1917,
+      "step": 2675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700716462817414e-05,
+      "loss": 0.2224,
+      "step": 2676
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700478691905115e-05,
+      "loss": 0.1772,
+      "step": 2677
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7002408432150396e-05,
+      "loss": 0.2073,
+      "step": 2678
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7000029167735972e-05,
+      "loss": 0.1738,
+      "step": 2679
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6997649126072064e-05,
+      "loss": 0.1875,
+      "step": 2680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699526830742294e-05,
+      "loss": 0.1992,
+      "step": 2681
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699288671205295e-05,
+      "loss": 0.1825,
+      "step": 2682
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6990504340226545e-05,
+      "loss": 0.1729,
+      "step": 2683
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6988121192208237e-05,
+      "loss": 0.1917,
+      "step": 2684
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6985737268262653e-05,
+      "loss": 0.2087,
+      "step": 2685
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6983352568654488e-05,
+      "loss": 0.1663,
+      "step": 2686
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698096709364852e-05,
+      "loss": 0.197,
+      "step": 2687
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6978580843509635e-05,
+      "loss": 0.209,
+      "step": 2688
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6976193818502776e-05,
+      "loss": 0.1978,
+      "step": 2689
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6973806018893e-05,
+      "loss": 0.2356,
+      "step": 2690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6971417444945423e-05,
+      "loss": 0.1736,
+      "step": 2691
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969028096925275e-05,
+      "loss": 0.1821,
+      "step": 2692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6966637975097846e-05,
+      "loss": 0.2124,
+      "step": 2693
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696424707972853e-05,
+      "loss": 0.1775,
+      "step": 2694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69618554110828e-05,
+      "loss": 0.2517,
+      "step": 2695
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6959462969426215e-05,
+      "loss": 0.2102,
+      "step": 2696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6957069755024416e-05,
+      "loss": 0.1877,
+      "step": 2697
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6954675768143138e-05,
+      "loss": 0.2017,
+      "step": 2698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69522810090482e-05,
+      "loss": 0.1938,
+      "step": 2699
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6949885478005497e-05,
+      "loss": 0.1714,
+      "step": 2700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6947489175281027e-05,
+      "loss": 0.2004,
+      "step": 2701
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694509210114086e-05,
+      "loss": 0.1951,
+      "step": 2702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694269425585115e-05,
+      "loss": 0.2009,
+      "step": 2703
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6940295639678147e-05,
+      "loss": 0.2014,
+      "step": 2704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6937896252888183e-05,
+      "loss": 0.1835,
+      "step": 2705
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693549609574767e-05,
+      "loss": 0.1929,
+      "step": 2706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6933095168523116e-05,
+      "loss": 0.208,
+      "step": 2707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69306934714811e-05,
+      "loss": 0.1667,
+      "step": 2708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6928291004888306e-05,
+      "loss": 0.1858,
+      "step": 2709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6925887769011483e-05,
+      "loss": 0.1644,
+      "step": 2710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6923483764117477e-05,
+      "loss": 0.2004,
+      "step": 2711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6921078990473216e-05,
+      "loss": 0.1821,
+      "step": 2712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691867344834572e-05,
+      "loss": 0.2615,
+      "step": 2713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6916267138002086e-05,
+      "loss": 0.207,
+      "step": 2714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.69138600597095e-05,
+      "loss": 0.1852,
+      "step": 2715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6911452213735223e-05,
+      "loss": 0.1812,
+      "step": 2716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690904360034662e-05,
+      "loss": 0.1992,
+      "step": 2717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6906634219811136e-05,
+      "loss": 0.1909,
+      "step": 2718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6904224072396288e-05,
+      "loss": 0.2034,
+      "step": 2719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6901813158369686e-05,
+      "loss": 0.1641,
+      "step": 2720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6899401477999034e-05,
+      "loss": 0.1965,
+      "step": 2721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689698903155211e-05,
+      "loss": 0.1807,
+      "step": 2722
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689457581929678e-05,
+      "loss": 0.1975,
+      "step": 2723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6892161841501e-05,
+      "loss": 0.1938,
+      "step": 2724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6889747098432795e-05,
+      "loss": 0.1552,
+      "step": 2725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.68873315903603e-05,
+      "loss": 0.1814,
+      "step": 2726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688491531755171e-05,
+      "loss": 0.2417,
+      "step": 2727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6882498280275322e-05,
+      "loss": 0.2151,
+      "step": 2728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880080478799512e-05,
+      "loss": 0.1868,
+      "step": 2729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687766191339274e-05,
+      "loss": 0.2251,
+      "step": 2730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6875242584323553e-05,
+      "loss": 0.2056,
+      "step": 2731
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6872822491860583e-05,
+      "loss": 0.1843,
+      "step": 2732
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.687040163627254e-05,
+      "loss": 0.1804,
+      "step": 2733
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6867980017828228e-05,
+      "loss": 0.2432,
+      "step": 2734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6865557636796533e-05,
+      "loss": 0.2078,
+      "step": 2735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686313449344642e-05,
+      "loss": 0.2131,
+      "step": 2736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6860710588046945e-05,
+      "loss": 0.2024,
+      "step": 2737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6858285920867254e-05,
+      "loss": 0.1959,
+      "step": 2738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685586049217656e-05,
+      "loss": 0.2422,
+      "step": 2739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6853434302244175e-05,
+      "loss": 0.2275,
+      "step": 2740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6851007351339493e-05,
+      "loss": 0.1978,
+      "step": 2741
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6848579639731987e-05,
+      "loss": 0.2395,
+      "step": 2742
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684615116769122e-05,
+      "loss": 0.2139,
+      "step": 2743
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684372193548684e-05,
+      "loss": 0.1655,
+      "step": 2744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6841291943388576e-05,
+      "loss": 0.2085,
+      "step": 2745
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683886119166624e-05,
+      "loss": 0.2056,
+      "step": 2746
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683642968058974e-05,
+      "loss": 0.2151,
+      "step": 2747
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6833997410429046e-05,
+      "loss": 0.2195,
+      "step": 2748
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6831564381454235e-05,
+      "loss": 0.2053,
+      "step": 2749
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6829130593935454e-05,
+      "loss": 0.1895,
+      "step": 2750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6826696048142946e-05,
+      "loss": 0.1973,
+      "step": 2751
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682426074434702e-05,
+      "loss": 0.2202,
+      "step": 2752
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682182468281809e-05,
+      "loss": 0.1942,
+      "step": 2753
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681938786382664e-05,
+      "loss": 0.1875,
+      "step": 2754
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6816950287643243e-05,
+      "loss": 0.1831,
+      "step": 2755
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681451195453856e-05,
+      "loss": 0.2056,
+      "step": 2756
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6812072864783324e-05,
+      "loss": 0.1984,
+      "step": 2757
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6809633018648365e-05,
+      "loss": 0.2102,
+      "step": 2758
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680719241640459e-05,
+      "loss": 0.2032,
+      "step": 2759
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6804751058322994e-05,
+      "loss": 0.1831,
+      "step": 2760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680230894467465e-05,
+      "loss": 0.1655,
+      "step": 2761
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6799866075730724e-05,
+      "loss": 0.1919,
+      "step": 2762
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797422451762454e-05,
+      "loss": 0.1685,
+      "step": 2763
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6794978073041176e-05,
+      "loss": 0.2039,
+      "step": 2764
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679253293983829e-05,
+      "loss": 0.2073,
+      "step": 2765
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6790087052425303e-05,
+      "loss": 0.1921,
+      "step": 2766
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678764041107379e-05,
+      "loss": 0.1995,
+      "step": 2767
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6785193016055415e-05,
+      "loss": 0.2031,
+      "step": 2768
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6782744867641924e-05,
+      "loss": 0.1958,
+      "step": 2769
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6780295966105148e-05,
+      "loss": 0.1753,
+      "step": 2770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6777846311717005e-05,
+      "loss": 0.1907,
+      "step": 2771
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6775395904749486e-05,
+      "loss": 0.2007,
+      "step": 2772
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6772944745474676e-05,
+      "loss": 0.1719,
+      "step": 2773
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677049283416474e-05,
+      "loss": 0.1736,
+      "step": 2774
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6768040171091926e-05,
+      "loss": 0.1624,
+      "step": 2775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676558675652857e-05,
+      "loss": 0.1892,
+      "step": 2776
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6763132590747076e-05,
+      "loss": 0.1677,
+      "step": 2777
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6760677674019953e-05,
+      "loss": 0.1677,
+      "step": 2778
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675822200661978e-05,
+      "loss": 0.1699,
+      "step": 2779
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6755765588819226e-05,
+      "loss": 0.2097,
+      "step": 2780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6753308420891034e-05,
+      "loss": 0.1625,
+      "step": 2781
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675085050310804e-05,
+      "loss": 0.1677,
+      "step": 2782
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6748391835743153e-05,
+      "loss": 0.1838,
+      "step": 2783
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674593241906938e-05,
+      "loss": 0.202,
+      "step": 2784
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.67434722533598e-05,
+      "loss": 0.2732,
+      "step": 2785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741011338887573e-05,
+      "loss": 0.2227,
+      "step": 2786
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6738549675925956e-05,
+      "loss": 0.1787,
+      "step": 2787
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736087264748267e-05,
+      "loss": 0.2156,
+      "step": 2788
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6733624105627937e-05,
+      "loss": 0.1943,
+      "step": 2789
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.673116019883845e-05,
+      "loss": 0.2168,
+      "step": 2790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6728695544653397e-05,
+      "loss": 0.2322,
+      "step": 2791
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6726230143346433e-05,
+      "loss": 0.1699,
+      "step": 2792
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6723763995191308e-05,
+      "loss": 0.1858,
+      "step": 2793
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6721297100461845e-05,
+      "loss": 0.1982,
+      "step": 2794
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6718829459431964e-05,
+      "loss": 0.1804,
+      "step": 2795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6716361072375657e-05,
+      "loss": 0.2112,
+      "step": 2796
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6713891939567002e-05,
+      "loss": 0.1835,
+      "step": 2797
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671142206128016e-05,
+      "loss": 0.1858,
+      "step": 2798
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6708951437789373e-05,
+      "loss": 0.2231,
+      "step": 2799
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6706480069368968e-05,
+      "loss": 0.2097,
+      "step": 2800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6704007956293354e-05,
+      "loss": 0.2034,
+      "step": 2801
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6701535098837024e-05,
+      "loss": 0.2241,
+      "step": 2802
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6699061497274548e-05,
+      "loss": 0.2007,
+      "step": 2803
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6696587151880586e-05,
+      "loss": 0.2039,
+      "step": 2804
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669411206292988e-05,
+      "loss": 0.1858,
+      "step": 2805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6691636230697246e-05,
+      "loss": 0.1624,
+      "step": 2806
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668915965545759e-05,
+      "loss": 0.2222,
+      "step": 2807
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6686682337485897e-05,
+      "loss": 0.2649,
+      "step": 2808
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6684204277057246e-05,
+      "loss": 0.2026,
+      "step": 2809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.668172547444678e-05,
+      "loss": 0.1924,
+      "step": 2810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6679245929929735e-05,
+      "loss": 0.1793,
+      "step": 2811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667676564378143e-05,
+      "loss": 0.1931,
+      "step": 2812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6674284616277263e-05,
+      "loss": 0.1445,
+      "step": 2813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667180284769271e-05,
+      "loss": 0.1675,
+      "step": 2814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666932033830334e-05,
+      "loss": 0.1655,
+      "step": 2815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.66668370883848e-05,
+      "loss": 0.1738,
+      "step": 2816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6664353098212817e-05,
+      "loss": 0.2356,
+      "step": 2817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.66618683680632e-05,
+      "loss": 0.2222,
+      "step": 2818
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6659382898211843e-05,
+      "loss": 0.1548,
+      "step": 2819
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6656896688934717e-05,
+      "loss": 0.2002,
+      "step": 2820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6654409740507884e-05,
+      "loss": 0.2368,
+      "step": 2821
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6651922053207478e-05,
+      "loss": 0.165,
+      "step": 2822
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649433627309725e-05,
+      "loss": 0.2039,
+      "step": 2823
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6646944463090922e-05,
+      "loss": 0.2256,
+      "step": 2824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6644454560827457e-05,
+      "loss": 0.1687,
+      "step": 2825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6641963920795795e-05,
+      "loss": 0.1772,
+      "step": 2826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6639472543272488e-05,
+      "loss": 0.197,
+      "step": 2827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6636980428534163e-05,
+      "loss": 0.2095,
+      "step": 2828
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6634487576857534e-05,
+      "loss": 0.1687,
+      "step": 2829
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6631993988519396e-05,
+      "loss": 0.2197,
+      "step": 2830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6629499663796622e-05,
+      "loss": 0.228,
+      "step": 2831
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6627004602966176e-05,
+      "loss": 0.1785,
+      "step": 2832
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6624508806305088e-05,
+      "loss": 0.179,
+      "step": 2833
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622012274090487e-05,
+      "loss": 0.1997,
+      "step": 2834
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6619515006599573e-05,
+      "loss": 0.2156,
+      "step": 2835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661701700410963e-05,
+      "loss": 0.1917,
+      "step": 2836
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661451826689803e-05,
+      "loss": 0.2153,
+      "step": 2837
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6612018795242214e-05,
+      "loss": 0.239,
+      "step": 2838
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6609518589419708e-05,
+      "loss": 0.1743,
+      "step": 2839
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6607017649708133e-05,
+      "loss": 0.2073,
+      "step": 2840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6604515976385176e-05,
+      "loss": 0.2231,
+      "step": 2841
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660201356972861e-05,
+      "loss": 0.1978,
+      "step": 2842
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659951043001629e-05,
+      "loss": 0.1772,
+      "step": 2843
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6597006557526156e-05,
+      "loss": 0.2026,
+      "step": 2844
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6594501952536225e-05,
+      "loss": 0.1772,
+      "step": 2845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6591996615324593e-05,
+      "loss": 0.1782,
+      "step": 2846
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6589490546169443e-05,
+      "loss": 0.1963,
+      "step": 2847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6586983745349033e-05,
+      "loss": 0.2205,
+      "step": 2848
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658447621314171e-05,
+      "loss": 0.1808,
+      "step": 2849
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6581967949825902e-05,
+      "loss": 0.182,
+      "step": 2850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6579458955680106e-05,
+      "loss": 0.1865,
+      "step": 2851
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6576949230982918e-05,
+      "loss": 0.2131,
+      "step": 2852
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6574438776012998e-05,
+      "loss": 0.2366,
+      "step": 2853
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6571927591049094e-05,
+      "loss": 0.2214,
+      "step": 2854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6569415676370044e-05,
+      "loss": 0.1543,
+      "step": 2855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6566903032254754e-05,
+      "loss": 0.2444,
+      "step": 2856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656438965898221e-05,
+      "loss": 0.1743,
+      "step": 2857
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6561875556831497e-05,
+      "loss": 0.2017,
+      "step": 2858
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6559360726081762e-05,
+      "loss": 0.2029,
+      "step": 2859
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6556845167012238e-05,
+      "loss": 0.2278,
+      "step": 2860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6554328879902245e-05,
+      "loss": 0.1509,
+      "step": 2861
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6551811865031174e-05,
+      "loss": 0.1914,
+      "step": 2862
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6549294122678507e-05,
+      "loss": 0.2466,
+      "step": 2863
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6546775653123803e-05,
+      "loss": 0.173,
+      "step": 2864
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6544256456646693e-05,
+      "loss": 0.1854,
+      "step": 2865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6541736533526903e-05,
+      "loss": 0.1442,
+      "step": 2866
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6539215884044235e-05,
+      "loss": 0.2041,
+      "step": 2867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6536694508478565e-05,
+      "loss": 0.2019,
+      "step": 2868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6534172407109857e-05,
+      "loss": 0.1797,
+      "step": 2869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653164958021815e-05,
+      "loss": 0.1458,
+      "step": 2870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6529126028083566e-05,
+      "loss": 0.2107,
+      "step": 2871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6526601750986314e-05,
+      "loss": 0.1953,
+      "step": 2872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6524076749206674e-05,
+      "loss": 0.2061,
+      "step": 2873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6521551023025006e-05,
+      "loss": 0.1956,
+      "step": 2874
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651902457272176e-05,
+      "loss": 0.2009,
+      "step": 2875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651649739857746e-05,
+      "loss": 0.1904,
+      "step": 2876
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6513969500872713e-05,
+      "loss": 0.1833,
+      "step": 2877
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.65114408798882e-05,
+      "loss": 0.2065,
+      "step": 2878
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650891153590469e-05,
+      "loss": 0.2104,
+      "step": 2879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6506381469203025e-05,
+      "loss": 0.1667,
+      "step": 2880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6503850680064135e-05,
+      "loss": 0.2058,
+      "step": 2881
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650131916876903e-05,
+      "loss": 0.2009,
+      "step": 2882
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649878693559879e-05,
+      "loss": 0.2041,
+      "step": 2883
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6496253980834586e-05,
+      "loss": 0.2085,
+      "step": 2884
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6493720304757666e-05,
+      "loss": 0.1628,
+      "step": 2885
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649118590764935e-05,
+      "loss": 0.1892,
+      "step": 2886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6488650789791054e-05,
+      "loss": 0.1826,
+      "step": 2887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648611495146426e-05,
+      "loss": 0.1855,
+      "step": 2888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648357839295054e-05,
+      "loss": 0.1892,
+      "step": 2889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6481041114531535e-05,
+      "loss": 0.1655,
+      "step": 2890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6478503116488975e-05,
+      "loss": 0.1838,
+      "step": 2891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647596439910467e-05,
+      "loss": 0.2129,
+      "step": 2892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6473424962660503e-05,
+      "loss": 0.2107,
+      "step": 2893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647088480743844e-05,
+      "loss": 0.1593,
+      "step": 2894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6468343933720532e-05,
+      "loss": 0.1646,
+      "step": 2895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6465802341788903e-05,
+      "loss": 0.2227,
+      "step": 2896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646326003192576e-05,
+      "loss": 0.1746,
+      "step": 2897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6460717004413383e-05,
+      "loss": 0.1877,
+      "step": 2898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6458173259534148e-05,
+      "loss": 0.1466,
+      "step": 2899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6455628797570494e-05,
+      "loss": 0.1802,
+      "step": 2900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6453083618804944e-05,
+      "loss": 0.2124,
+      "step": 2901
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645053772352011e-05,
+      "loss": 0.1755,
+      "step": 2902
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6447991111998665e-05,
+      "loss": 0.1877,
+      "step": 2903
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644544378452338e-05,
+      "loss": 0.1755,
+      "step": 2904
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6442895741377098e-05,
+      "loss": 0.2124,
+      "step": 2905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6440346982842735e-05,
+      "loss": 0.1443,
+      "step": 2906
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6437797509203305e-05,
+      "loss": 0.2012,
+      "step": 2907
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6435247320741873e-05,
+      "loss": 0.2312,
+      "step": 2908
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6432696417741615e-05,
+      "loss": 0.2014,
+      "step": 2909
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643014480048576e-05,
+      "loss": 0.1897,
+      "step": 2910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6427592469257635e-05,
+      "loss": 0.1875,
+      "step": 2911
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6425039424340633e-05,
+      "loss": 0.1995,
+      "step": 2912
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6422485666018235e-05,
+      "loss": 0.1785,
+      "step": 2913
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6419931194573998e-05,
+      "loss": 0.2026,
+      "step": 2914
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6417376010291556e-05,
+      "loss": 0.1892,
+      "step": 2915
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6414820113454624e-05,
+      "loss": 0.218,
+      "step": 2916
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6412263504347002e-05,
+      "loss": 0.2131,
+      "step": 2917
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6409706183252555e-05,
+      "loss": 0.171,
+      "step": 2918
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6407148150455242e-05,
+      "loss": 0.1704,
+      "step": 2919
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6404589406239094e-05,
+      "loss": 0.1807,
+      "step": 2920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640202995088822e-05,
+      "loss": 0.1887,
+      "step": 2921
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639946978468681e-05,
+      "loss": 0.1833,
+      "step": 2922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639690890791913e-05,
+      "loss": 0.1697,
+      "step": 2923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639434732086953e-05,
+      "loss": 0.2112,
+      "step": 2924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6391785023822436e-05,
+      "loss": 0.2009,
+      "step": 2925
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638922201706236e-05,
+      "loss": 0.1556,
+      "step": 2926
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638665830087387e-05,
+      "loss": 0.2266,
+      "step": 2927
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6384093875541642e-05,
+      "loss": 0.2122,
+      "step": 2928
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6381528741350414e-05,
+      "loss": 0.1748,
+      "step": 2929
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6378962898585005e-05,
+      "loss": 0.1951,
+      "step": 2930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6376396347530314e-05,
+      "loss": 0.177,
+      "step": 2931
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637382908847132e-05,
+      "loss": 0.1804,
+      "step": 2932
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6371261121693075e-05,
+      "loss": 0.197,
+      "step": 2933
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6368692447480716e-05,
+      "loss": 0.1926,
+      "step": 2934
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6366123066119458e-05,
+      "loss": 0.2063,
+      "step": 2935
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636355297789459e-05,
+      "loss": 0.1519,
+      "step": 2936
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6360982183091486e-05,
+      "loss": 0.1583,
+      "step": 2937
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635841068199559e-05,
+      "loss": 0.1946,
+      "step": 2938
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6355838474892435e-05,
+      "loss": 0.2212,
+      "step": 2939
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635326556206762e-05,
+      "loss": 0.1716,
+      "step": 2940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635069194380683e-05,
+      "loss": 0.2087,
+      "step": 2941
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634811762039583e-05,
+      "loss": 0.1714,
+      "step": 2942
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634554259212046e-05,
+      "loss": 0.1982,
+      "step": 2943
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6342966859266637e-05,
+      "loss": 0.2009,
+      "step": 2944
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634039042212036e-05,
+      "loss": 0.207,
+      "step": 2945
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6337813280967703e-05,
+      "loss": 0.1561,
+      "step": 2946
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633523543609482e-05,
+      "loss": 0.1924,
+      "step": 2947
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6332656887787937e-05,
+      "loss": 0.1804,
+      "step": 2948
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633007763633337e-05,
+      "loss": 0.1775,
+      "step": 2949
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6327497682017506e-05,
+      "loss": 0.2212,
+      "step": 2950
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6324917025126816e-05,
+      "loss": 0.1855,
+      "step": 2951
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632233566594783e-05,
+      "loss": 0.1838,
+      "step": 2952
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631975360476718e-05,
+      "loss": 0.2205,
+      "step": 2953
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631717084187156e-05,
+      "loss": 0.2161,
+      "step": 2954
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6314587377547754e-05,
+      "loss": 0.1525,
+      "step": 2955
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631200321208261e-05,
+      "loss": 0.1887,
+      "step": 2956
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6309418345763067e-05,
+      "loss": 0.1909,
+      "step": 2957
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6306832778876135e-05,
+      "loss": 0.1642,
+      "step": 2958
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63042465117089e-05,
+      "loss": 0.1715,
+      "step": 2959
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6301659544548528e-05,
+      "loss": 0.1577,
+      "step": 2960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629907187768227e-05,
+      "loss": 0.2073,
+      "step": 2961
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6296483511397442e-05,
+      "loss": 0.1333,
+      "step": 2962
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6293894445981448e-05,
+      "loss": 0.1943,
+      "step": 2963
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629130468172176e-05,
+      "loss": 0.2249,
+      "step": 2964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6288714218905938e-05,
+      "loss": 0.2485,
+      "step": 2965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628612305782161e-05,
+      "loss": 0.1801,
+      "step": 2966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.628353119875649e-05,
+      "loss": 0.1736,
+      "step": 2967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6280938641998366e-05,
+      "loss": 0.1946,
+      "step": 2968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.62783453878351e-05,
+      "loss": 0.1982,
+      "step": 2969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6275751436554632e-05,
+      "loss": 0.1948,
+      "step": 2970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6273156788444988e-05,
+      "loss": 0.1594,
+      "step": 2971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270561443794262e-05,
+      "loss": 0.1953,
+      "step": 2972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626796540289063e-05,
+      "loss": 0.2,
+      "step": 2973
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6265368666022343e-05,
+      "loss": 0.1926,
+      "step": 2974
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626277123347773e-05,
+      "loss": 0.1746,
+      "step": 2975
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6260173105545198e-05,
+      "loss": 0.2239,
+      "step": 2976
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6257574282513227e-05,
+      "loss": 0.1685,
+      "step": 2977
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6254974764670382e-05,
+      "loss": 0.2051,
+      "step": 2978
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.62523745523053e-05,
+      "loss": 0.2163,
+      "step": 2979
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6249773645706698e-05,
+      "loss": 0.1997,
+      "step": 2980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6247172045163366e-05,
+      "loss": 0.2024,
+      "step": 2981
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624456975096417e-05,
+      "loss": 0.1763,
+      "step": 2982
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624196676339806e-05,
+      "loss": 0.1931,
+      "step": 2983
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.623936308275406e-05,
+      "loss": 0.2014,
+      "step": 2984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6236758709321265e-05,
+      "loss": 0.1654,
+      "step": 2985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.623415364338886e-05,
+      "loss": 0.1721,
+      "step": 2986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6231547885246094e-05,
+      "loss": 0.209,
+      "step": 2987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6228941435182294e-05,
+      "loss": 0.2231,
+      "step": 2988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6226334293486878e-05,
+      "loss": 0.1611,
+      "step": 2989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6223726460449322e-05,
+      "loss": 0.2043,
+      "step": 2990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622111793635919e-05,
+      "loss": 0.1892,
+      "step": 2991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621850872150612e-05,
+      "loss": 0.2131,
+      "step": 2992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6215898816179826e-05,
+      "loss": 0.1921,
+      "step": 2993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6213288220670097e-05,
+      "loss": 0.1987,
+      "step": 2994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6210676935266806e-05,
+      "loss": 0.175,
+      "step": 2995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6208064960259897e-05,
+      "loss": 0.2297,
+      "step": 2996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620545229593939e-05,
+      "loss": 0.2095,
+      "step": 2997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6202838942595378e-05,
+      "loss": 0.188,
+      "step": 2998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620022490051804e-05,
+      "loss": 0.193,
+      "step": 2999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6197610169997625e-05,
+      "loss": 0.1965,
+      "step": 3000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6194994751324462e-05,
+      "loss": 0.2148,
+      "step": 3001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6192378644788955e-05,
+      "loss": 0.2224,
+      "step": 3002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6189761850681578e-05,
+      "loss": 0.2224,
+      "step": 3003
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6187144369292894e-05,
+      "loss": 0.2173,
+      "step": 3004
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6184526200913533e-05,
+      "loss": 0.1841,
+      "step": 3005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6181907345834202e-05,
+      "loss": 0.1478,
+      "step": 3006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6179287804345692e-05,
+      "loss": 0.1733,
+      "step": 3007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6176667576738852e-05,
+      "loss": 0.1929,
+      "step": 3008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6174046663304634e-05,
+      "loss": 0.2512,
+      "step": 3009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617142506433404e-05,
+      "loss": 0.1792,
+      "step": 3010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616880278011817e-05,
+      "loss": 0.1772,
+      "step": 3011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6166179810948182e-05,
+      "loss": 0.1792,
+      "step": 3012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163556157115317e-05,
+      "loss": 0.142,
+      "step": 3013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160931818910902e-05,
+      "loss": 0.2124,
+      "step": 3014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615830679662632e-05,
+      "loss": 0.2222,
+      "step": 3015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615568109055305e-05,
+      "loss": 0.1746,
+      "step": 3016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6153054700982628e-05,
+      "loss": 0.2031,
+      "step": 3017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6150427628206686e-05,
+      "loss": 0.1724,
+      "step": 3018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6147799872516915e-05,
+      "loss": 0.217,
+      "step": 3019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614517143420509e-05,
+      "loss": 0.2148,
+      "step": 3020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614254231356306e-05,
+      "loss": 0.2234,
+      "step": 3021
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6139912510882752e-05,
+      "loss": 0.1711,
+      "step": 3022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6137282026456162e-05,
+      "loss": 0.1792,
+      "step": 3023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613465086057537e-05,
+      "loss": 0.1628,
+      "step": 3024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6132019013532526e-05,
+      "loss": 0.177,
+      "step": 3025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612938648561986e-05,
+      "loss": 0.1768,
+      "step": 3026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6126753277129672e-05,
+      "loss": 0.1978,
+      "step": 3027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6124119388354343e-05,
+      "loss": 0.1754,
+      "step": 3028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6121484819586326e-05,
+      "loss": 0.1626,
+      "step": 3029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6118849571118154e-05,
+      "loss": 0.2041,
+      "step": 3030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611621364324243e-05,
+      "loss": 0.2476,
+      "step": 3031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611357703625183e-05,
+      "loss": 0.2329,
+      "step": 3032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6110939750439118e-05,
+      "loss": 0.2048,
+      "step": 3033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610830178609712e-05,
+      "loss": 0.2214,
+      "step": 3034
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6105663143518748e-05,
+      "loss": 0.1719,
+      "step": 3035
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6103023822996982e-05,
+      "loss": 0.2261,
+      "step": 3036
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6100383824824874e-05,
+      "loss": 0.2139,
+      "step": 3037
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6097743149295565e-05,
+      "loss": 0.2209,
+      "step": 3038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6095101796702257e-05,
+      "loss": 0.2043,
+      "step": 3039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6092459767338236e-05,
+      "loss": 0.2285,
+      "step": 3040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.608981706149686e-05,
+      "loss": 0.1799,
+      "step": 3041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6087173679471565e-05,
+      "loss": 0.1724,
+      "step": 3042
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608452962155585e-05,
+      "loss": 0.2385,
+      "step": 3043
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6081884888043307e-05,
+      "loss": 0.1682,
+      "step": 3044
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6079239479227596e-05,
+      "loss": 0.1924,
+      "step": 3045
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607659339540244e-05,
+      "loss": 0.1633,
+      "step": 3046
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607394663686166e-05,
+      "loss": 0.1816,
+      "step": 3047
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6071299203899127e-05,
+      "loss": 0.1827,
+      "step": 3048
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606865109680881e-05,
+      "loss": 0.2256,
+      "step": 3049
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6066002315884733e-05,
+      "loss": 0.1558,
+      "step": 3050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606335286142101e-05,
+      "loss": 0.1804,
+      "step": 3051
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6060702733711823e-05,
+      "loss": 0.2119,
+      "step": 3052
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6058051933051427e-05,
+      "loss": 0.199,
+      "step": 3053
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6055400459734158e-05,
+      "loss": 0.2031,
+      "step": 3054
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605274831405442e-05,
+      "loss": 0.1873,
+      "step": 3055
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605009549630669e-05,
+      "loss": 0.1688,
+      "step": 3056
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6047442006785533e-05,
+      "loss": 0.1366,
+      "step": 3057
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6044787845785576e-05,
+      "loss": 0.156,
+      "step": 3058
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6042133013601523e-05,
+      "loss": 0.2148,
+      "step": 3059
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6039477510528155e-05,
+      "loss": 0.15,
+      "step": 3060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6036821336860324e-05,
+      "loss": 0.2241,
+      "step": 3061
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603416449289296e-05,
+      "loss": 0.1555,
+      "step": 3062
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6031506978921066e-05,
+      "loss": 0.1738,
+      "step": 3063
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6028848795239725e-05,
+      "loss": 0.1609,
+      "step": 3064
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602618994214408e-05,
+      "loss": 0.1809,
+      "step": 3065
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6023530419929362e-05,
+      "loss": 0.2087,
+      "step": 3066
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6020870228890873e-05,
+      "loss": 0.2085,
+      "step": 3067
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6018209369323983e-05,
+      "loss": 0.2195,
+      "step": 3068
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6015547841524144e-05,
+      "loss": 0.1671,
+      "step": 3069
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6012885645786877e-05,
+      "loss": 0.1633,
+      "step": 3070
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6010222782407784e-05,
+      "loss": 0.2095,
+      "step": 3071
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6007559251682532e-05,
+      "loss": 0.1865,
+      "step": 3072
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600489505390687e-05,
+      "loss": 0.2124,
+      "step": 3073
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6002230189376614e-05,
+      "loss": 0.2207,
+      "step": 3074
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599956465838766e-05,
+      "loss": 0.208,
+      "step": 3075
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5996898461235976e-05,
+      "loss": 0.1587,
+      "step": 3076
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5994231598217607e-05,
+      "loss": 0.1914,
+      "step": 3077
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599156406962866e-05,
+      "loss": 0.1941,
+      "step": 3078
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598889587576533e-05,
+      "loss": 0.2241,
+      "step": 3079
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598622701692388e-05,
+      "loss": 0.2085,
+      "step": 3080
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5983557493400645e-05,
+      "loss": 0.2036,
+      "step": 3081
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.598088730549204e-05,
+      "loss": 0.2087,
+      "step": 3082
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978216453494544e-05,
+      "loss": 0.2278,
+      "step": 3083
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5975544937704724e-05,
+      "loss": 0.1558,
+      "step": 3084
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5972872758419207e-05,
+      "loss": 0.2185,
+      "step": 3085
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970199915934696e-05,
+      "loss": 0.2009,
+      "step": 3086
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596752641054798e-05,
+      "loss": 0.2109,
+      "step": 3087
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59648522425559e-05,
+      "loss": 0.272,
+      "step": 3088
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5962177412255392e-05,
+      "loss": 0.2034,
+      "step": 3089
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595950191994345e-05,
+      "loss": 0.2021,
+      "step": 3090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595682576591715e-05,
+      "loss": 0.1853,
+      "step": 3091
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5954148950473642e-05,
+      "loss": 0.1663,
+      "step": 3092
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5951471473910145e-05,
+      "loss": 0.1936,
+      "step": 3093
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5948793336523953e-05,
+      "loss": 0.2064,
+      "step": 3094
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594611453861243e-05,
+      "loss": 0.1782,
+      "step": 3095
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5943435080473025e-05,
+      "loss": 0.1902,
+      "step": 3096
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5940754962403242e-05,
+      "loss": 0.2141,
+      "step": 3097
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5938074184700675e-05,
+      "loss": 0.1838,
+      "step": 3098
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5935392747662987e-05,
+      "loss": 0.1763,
+      "step": 3099
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5932710651587902e-05,
+      "loss": 0.2009,
+      "step": 3100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5930027896773235e-05,
+      "loss": 0.2346,
+      "step": 3101
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5927344483516865e-05,
+      "loss": 0.1892,
+      "step": 3102
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5924660412116746e-05,
+      "loss": 0.209,
+      "step": 3103
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.59219756828709e-05,
+      "loss": 0.1863,
+      "step": 3104
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591929029607743e-05,
+      "loss": 0.1603,
+      "step": 3105
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5916604252034508e-05,
+      "loss": 0.1748,
+      "step": 3106
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5913917551040377e-05,
+      "loss": 0.1748,
+      "step": 3107
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591123019339336e-05,
+      "loss": 0.1907,
+      "step": 3108
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908542179391844e-05,
+      "loss": 0.1539,
+      "step": 3109
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5905853509334295e-05,
+      "loss": 0.2075,
+      "step": 3110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5903164183519248e-05,
+      "loss": 0.1975,
+      "step": 3111
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5900474202245315e-05,
+      "loss": 0.2395,
+      "step": 3112
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589778356581118e-05,
+      "loss": 0.1836,
+      "step": 3113
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5895092274515597e-05,
+      "loss": 0.1802,
+      "step": 3114
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589240032865739e-05,
+      "loss": 0.1931,
+      "step": 3115
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588970772853546e-05,
+      "loss": 0.2041,
+      "step": 3116
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588701447444879e-05,
+      "loss": 0.1853,
+      "step": 3117
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588432056669641e-05,
+      "loss": 0.2002,
+      "step": 3118
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5881626005577456e-05,
+      "loss": 0.1885,
+      "step": 3119
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5878930791391106e-05,
+      "loss": 0.1975,
+      "step": 3120
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5876234924436633e-05,
+      "loss": 0.1782,
+      "step": 3121
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5873538405013368e-05,
+      "loss": 0.2324,
+      "step": 3122
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5870841233420718e-05,
+      "loss": 0.1636,
+      "step": 3123
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5868143409958162e-05,
+      "loss": 0.1711,
+      "step": 3124
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5865444934925263e-05,
+      "loss": 0.1667,
+      "step": 3125
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586274580862164e-05,
+      "loss": 0.1912,
+      "step": 3126
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5860046031346988e-05,
+      "loss": 0.1777,
+      "step": 3127
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5857345603401082e-05,
+      "loss": 0.1819,
+      "step": 3128
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5854644525083767e-05,
+      "loss": 0.2024,
+      "step": 3129
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585194279669495e-05,
+      "loss": 0.1992,
+      "step": 3130
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5849240418534622e-05,
+      "loss": 0.1987,
+      "step": 3131
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846537390902845e-05,
+      "loss": 0.188,
+      "step": 3132
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5843833714099748e-05,
+      "loss": 0.1846,
+      "step": 3133
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584112938842553e-05,
+      "loss": 0.1636,
+      "step": 3134
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5838424414180473e-05,
+      "loss": 0.1401,
+      "step": 3135
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583571879166492e-05,
+      "loss": 0.1963,
+      "step": 3136
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5833012521179293e-05,
+      "loss": 0.1895,
+      "step": 3137
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583030560302408e-05,
+      "loss": 0.2148,
+      "step": 3138
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582759803749984e-05,
+      "loss": 0.2366,
+      "step": 3139
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5824889824907225e-05,
+      "loss": 0.1681,
+      "step": 3140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5822180965546925e-05,
+      "loss": 0.1771,
+      "step": 3141
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5819471459719723e-05,
+      "loss": 0.196,
+      "step": 3142
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5816761307726474e-05,
+      "loss": 0.1876,
+      "step": 3143
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5814050509868093e-05,
+      "loss": 0.2119,
+      "step": 3144
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581133906644558e-05,
+      "loss": 0.2036,
+      "step": 3145
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5808626977759998e-05,
+      "loss": 0.1934,
+      "step": 3146
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5805914244112485e-05,
+      "loss": 0.2563,
+      "step": 3147
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5803200865804253e-05,
+      "loss": 0.1328,
+      "step": 3148
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5800486843136576e-05,
+      "loss": 0.1877,
+      "step": 3149
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579777217641081e-05,
+      "loss": 0.1819,
+      "step": 3150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5795056865928376e-05,
+      "loss": 0.1804,
+      "step": 3151
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579234091199077e-05,
+      "loss": 0.2107,
+      "step": 3152
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578962431489956e-05,
+      "loss": 0.1699,
+      "step": 3153
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5786907074956384e-05,
+      "loss": 0.2178,
+      "step": 3154
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5784189192462952e-05,
+      "loss": 0.2053,
+      "step": 3155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578147066772104e-05,
+      "loss": 0.1982,
+      "step": 3156
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5778751501032502e-05,
+      "loss": 0.2061,
+      "step": 3157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577603169269926e-05,
+      "loss": 0.1892,
+      "step": 3158
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5773311243023314e-05,
+      "loss": 0.1731,
+      "step": 3159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5770590152306723e-05,
+      "loss": 0.1951,
+      "step": 3160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5767868420851628e-05,
+      "loss": 0.217,
+      "step": 3161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5765146048960234e-05,
+      "loss": 0.209,
+      "step": 3162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576242303693482e-05,
+      "loss": 0.158,
+      "step": 3163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5759699385077744e-05,
+      "loss": 0.2266,
+      "step": 3164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5756975093691415e-05,
+      "loss": 0.1965,
+      "step": 3165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575425016307833e-05,
+      "loss": 0.1766,
+      "step": 3166
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5751524593541057e-05,
+      "loss": 0.2097,
+      "step": 3167
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5748798385382226e-05,
+      "loss": 0.2275,
+      "step": 3168
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746071538904542e-05,
+      "loss": 0.2251,
+      "step": 3169
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5743344054410777e-05,
+      "loss": 0.2178,
+      "step": 3170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5740615932203788e-05,
+      "loss": 0.1555,
+      "step": 3171
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573788717258648e-05,
+      "loss": 0.2139,
+      "step": 3172
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573515777586185e-05,
+      "loss": 0.1777,
+      "step": 3173
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5732427742332955e-05,
+      "loss": 0.1868,
+      "step": 3174
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5729697072302927e-05,
+      "loss": 0.2051,
+      "step": 3175
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5726965766074962e-05,
+      "loss": 0.1946,
+      "step": 3176
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572423382395233e-05,
+      "loss": 0.2083,
+      "step": 3177
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572150124623838e-05,
+      "loss": 0.1768,
+      "step": 3178
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5718768033236518e-05,
+      "loss": 0.1855,
+      "step": 3179
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571603418525023e-05,
+      "loss": 0.2131,
+      "step": 3180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5713299702583067e-05,
+      "loss": 0.1775,
+      "step": 3181
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5710564585538653e-05,
+      "loss": 0.2053,
+      "step": 3182
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5707828834420683e-05,
+      "loss": 0.2383,
+      "step": 3183
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5705092449532922e-05,
+      "loss": 0.1732,
+      "step": 3184
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5702355431179202e-05,
+      "loss": 0.1782,
+      "step": 3185
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5699617779663438e-05,
+      "loss": 0.1677,
+      "step": 3186
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569687949528959e-05,
+      "loss": 0.1598,
+      "step": 3187
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5694140578361722e-05,
+      "loss": 0.2126,
+      "step": 3188
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5691401029183934e-05,
+      "loss": 0.1794,
+      "step": 3189
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568866084806042e-05,
+      "loss": 0.1917,
+      "step": 3190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5685920035295436e-05,
+      "loss": 0.1625,
+      "step": 3191
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5683178591193306e-05,
+      "loss": 0.1719,
+      "step": 3192
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568043651605843e-05,
+      "loss": 0.2131,
+      "step": 3193
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5677693810195272e-05,
+      "loss": 0.218,
+      "step": 3194
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5674950473908373e-05,
+      "loss": 0.1521,
+      "step": 3195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5672206507502337e-05,
+      "loss": 0.2017,
+      "step": 3196
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566946191128184e-05,
+      "loss": 0.1694,
+      "step": 3197
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566671668555163e-05,
+      "loss": 0.2083,
+      "step": 3198
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5663970830616523e-05,
+      "loss": 0.1777,
+      "step": 3199
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566122434678141e-05,
+      "loss": 0.1616,
+      "step": 3200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5658477234351234e-05,
+      "loss": 0.1638,
+      "step": 3201
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5655729493631038e-05,
+      "loss": 0.1951,
+      "step": 3202
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5652981124925907e-05,
+      "loss": 0.1864,
+      "step": 3203
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565023212854101e-05,
+      "loss": 0.2102,
+      "step": 3204
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564748250478158e-05,
+      "loss": 0.209,
+      "step": 3205
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564473225395293e-05,
+      "loss": 0.2283,
+      "step": 3206
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5641981376360423e-05,
+      "loss": 0.2041,
+      "step": 3207
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5639229872309512e-05,
+      "loss": 0.1929,
+      "step": 3208
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5636477742105706e-05,
+      "loss": 0.1794,
+      "step": 3209
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563372498605459e-05,
+      "loss": 0.1807,
+      "step": 3210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630971604461817e-05,
+      "loss": 0.1459,
+      "step": 3211
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5628217597633112e-05,
+      "loss": 0.2202,
+      "step": 3212
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562546296587426e-05,
+      "loss": 0.1444,
+      "step": 3213
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5622707709491125e-05,
+      "loss": 0.1892,
+      "step": 3214
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5619951828789635e-05,
+      "loss": 0.1653,
+      "step": 3215
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56171953240758e-05,
+      "loss": 0.2122,
+      "step": 3216
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5614438195655678e-05,
+      "loss": 0.2019,
+      "step": 3217
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.561168044383541e-05,
+      "loss": 0.2056,
+      "step": 3218
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5608922068921203e-05,
+      "loss": 0.1934,
+      "step": 3219
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560616307121934e-05,
+      "loss": 0.1899,
+      "step": 3220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5603403451036156e-05,
+      "loss": 0.1971,
+      "step": 3221
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5600643208678075e-05,
+      "loss": 0.1829,
+      "step": 3222
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5597882344451575e-05,
+      "loss": 0.2007,
+      "step": 3223
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5595120858663215e-05,
+      "loss": 0.219,
+      "step": 3224
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559235875161961e-05,
+      "loss": 0.1753,
+      "step": 3225
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558959602362746e-05,
+      "loss": 0.1931,
+      "step": 3226
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5586832674993514e-05,
+      "loss": 0.1692,
+      "step": 3227
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5584068706024612e-05,
+      "loss": 0.1892,
+      "step": 3228
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558130411702764e-05,
+      "loss": 0.1743,
+      "step": 3229
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5578538908309578e-05,
+      "loss": 0.1978,
+      "step": 3230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557577308017745e-05,
+      "loss": 0.2351,
+      "step": 3231
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5573006632938364e-05,
+      "loss": 0.1575,
+      "step": 3232
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55702395668995e-05,
+      "loss": 0.1914,
+      "step": 3233
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5567471882368086e-05,
+      "loss": 0.2,
+      "step": 3234
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5564703579651444e-05,
+      "loss": 0.1978,
+      "step": 3235
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5561934659056947e-05,
+      "loss": 0.1699,
+      "step": 3236
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5559165120892048e-05,
+      "loss": 0.1763,
+      "step": 3237
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5556394965464256e-05,
+      "loss": 0.2117,
+      "step": 3238
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.555362419308116e-05,
+      "loss": 0.1527,
+      "step": 3239
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5550852804050412e-05,
+      "loss": 0.1699,
+      "step": 3240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5548080798679732e-05,
+      "loss": 0.2471,
+      "step": 3241
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5545308177276915e-05,
+      "loss": 0.2078,
+      "step": 3242
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5542534940149816e-05,
+      "loss": 0.1699,
+      "step": 3243
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539761087606364e-05,
+      "loss": 0.1493,
+      "step": 3244
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5536986619954553e-05,
+      "loss": 0.2134,
+      "step": 3245
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5534211537502444e-05,
+      "loss": 0.1677,
+      "step": 3246
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553143584055817e-05,
+      "loss": 0.2188,
+      "step": 3247
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5528659529429935e-05,
+      "loss": 0.1899,
+      "step": 3248
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5525882604426005e-05,
+      "loss": 0.1885,
+      "step": 3249
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5523105065854712e-05,
+      "loss": 0.2012,
+      "step": 3250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5520326914024467e-05,
+      "loss": 0.1968,
+      "step": 3251
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551754814924374e-05,
+      "loss": 0.1851,
+      "step": 3252
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551476877182107e-05,
+      "loss": 0.2166,
+      "step": 3253
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5511988782065067e-05,
+      "loss": 0.1702,
+      "step": 3254
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5509208180284406e-05,
+      "loss": 0.1919,
+      "step": 3255
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5506426966787836e-05,
+      "loss": 0.1604,
+      "step": 3256
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5503645141884166e-05,
+      "loss": 0.1768,
+      "step": 3257
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5500862705882278e-05,
+      "loss": 0.1624,
+      "step": 3258
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5498079659091118e-05,
+      "loss": 0.1694,
+      "step": 3259
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5495296001819704e-05,
+      "loss": 0.187,
+      "step": 3260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549251173437712e-05,
+      "loss": 0.1934,
+      "step": 3261
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5489726857072517e-05,
+      "loss": 0.1831,
+      "step": 3262
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5486941370215114e-05,
+      "loss": 0.2068,
+      "step": 3263
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5484155274114197e-05,
+      "loss": 0.1995,
+      "step": 3264
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5481368569079126e-05,
+      "loss": 0.1687,
+      "step": 3265
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547858125541932e-05,
+      "loss": 0.165,
+      "step": 3266
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5475793333444263e-05,
+      "loss": 0.1802,
+      "step": 3267
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547300480346352e-05,
+      "loss": 0.1741,
+      "step": 3268
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5470215665786715e-05,
+      "loss": 0.1858,
+      "step": 3269
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546742592072354e-05,
+      "loss": 0.2192,
+      "step": 3270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546463556858375e-05,
+      "loss": 0.1584,
+      "step": 3271
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546184460967718e-05,
+      "loss": 0.1733,
+      "step": 3272
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5459053044313722e-05,
+      "loss": 0.2554,
+      "step": 3273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5456260872803332e-05,
+      "loss": 0.1995,
+      "step": 3274
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545346809545605e-05,
+      "loss": 0.1851,
+      "step": 3275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.545067471258196e-05,
+      "loss": 0.1663,
+      "step": 3276
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5447880724491234e-05,
+      "loss": 0.1655,
+      "step": 3277
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5445086131494103e-05,
+      "loss": 0.1887,
+      "step": 3278
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5442290933900864e-05,
+      "loss": 0.1953,
+      "step": 3279
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543949513202188e-05,
+      "loss": 0.2041,
+      "step": 3280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5436698726167585e-05,
+      "loss": 0.2302,
+      "step": 3281
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543390171664848e-05,
+      "loss": 0.1858,
+      "step": 3282
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543110410377513e-05,
+      "loss": 0.1965,
+      "step": 3283
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5428305887858166e-05,
+      "loss": 0.2104,
+      "step": 3284
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542550706920829e-05,
+      "loss": 0.1571,
+      "step": 3285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5422707648136275e-05,
+      "loss": 0.1875,
+      "step": 3286
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5419907624952947e-05,
+      "loss": 0.1899,
+      "step": 3287
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541710699996921e-05,
+      "loss": 0.2266,
+      "step": 3288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5414305773496033e-05,
+      "loss": 0.1616,
+      "step": 3289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541150394584445e-05,
+      "loss": 0.1919,
+      "step": 3290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408701517325565e-05,
+      "loss": 0.1775,
+      "step": 3291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540589848825054e-05,
+      "loss": 0.1714,
+      "step": 3292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540309485893061e-05,
+      "loss": 0.2026,
+      "step": 3293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5400290629677085e-05,
+      "loss": 0.165,
+      "step": 3294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539748580080133e-05,
+      "loss": 0.1641,
+      "step": 3295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5394680372614773e-05,
+      "loss": 0.1875,
+      "step": 3296
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5391874345428923e-05,
+      "loss": 0.1929,
+      "step": 3297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538906771955534e-05,
+      "loss": 0.188,
+      "step": 3298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5386260495305666e-05,
+      "loss": 0.2129,
+      "step": 3299
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5383452672991598e-05,
+      "loss": 0.2158,
+      "step": 3300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5380644252924902e-05,
+      "loss": 0.1648,
+      "step": 3301
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5377835235417418e-05,
+      "loss": 0.1661,
+      "step": 3302
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5375025620781034e-05,
+      "loss": 0.1879,
+      "step": 3303
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5372215409327726e-05,
+      "loss": 0.2021,
+      "step": 3304
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5369404601369523e-05,
+      "loss": 0.156,
+      "step": 3305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5366593197218524e-05,
+      "loss": 0.2085,
+      "step": 3306
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5363781197186893e-05,
+      "loss": 0.2073,
+      "step": 3307
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5360968601586858e-05,
+      "loss": 0.1943,
+      "step": 3308
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5358155410730723e-05,
+      "loss": 0.1219,
+      "step": 3309
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5355341624930846e-05,
+      "loss": 0.1973,
+      "step": 3310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352527244499658e-05,
+      "loss": 0.1912,
+      "step": 3311
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5349712269749656e-05,
+      "loss": 0.2197,
+      "step": 3312
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5346896700993398e-05,
+      "loss": 0.1925,
+      "step": 3313
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534408053854351e-05,
+      "loss": 0.1711,
+      "step": 3314
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.534126378271269e-05,
+      "loss": 0.1793,
+      "step": 3315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5338446433813695e-05,
+      "loss": 0.176,
+      "step": 3316
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5335628492159345e-05,
+      "loss": 0.1354,
+      "step": 3317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533280995806254e-05,
+      "loss": 0.1978,
+      "step": 3318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5329990831836227e-05,
+      "loss": 0.1785,
+      "step": 3319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5327171113793437e-05,
+      "loss": 0.198,
+      "step": 3320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5324350804247248e-05,
+      "loss": 0.1713,
+      "step": 3321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5321529903510822e-05,
+      "loss": 0.1812,
+      "step": 3322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5318708411897375e-05,
+      "loss": 0.2114,
+      "step": 3323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531588632972019e-05,
+      "loss": 0.1997,
+      "step": 3324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5313063657292623e-05,
+      "loss": 0.1887,
+      "step": 3325
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5310240394928086e-05,
+      "loss": 0.168,
+      "step": 3326
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5307416542940058e-05,
+      "loss": 0.1953,
+      "step": 3327
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5304592101642087e-05,
+      "loss": 0.1873,
+      "step": 3328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5301767071347794e-05,
+      "loss": 0.166,
+      "step": 3329
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5298941452370846e-05,
+      "loss": 0.1997,
+      "step": 3330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529611524502499e-05,
+      "loss": 0.1865,
+      "step": 3331
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5293288449624033e-05,
+      "loss": 0.1758,
+      "step": 3332
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5290461066481853e-05,
+      "loss": 0.1682,
+      "step": 3333
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5287633095912383e-05,
+      "loss": 0.1792,
+      "step": 3334
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528480453822963e-05,
+      "loss": 0.1917,
+      "step": 3335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528197539374767e-05,
+      "loss": 0.1982,
+      "step": 3336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279145662780622e-05,
+      "loss": 0.2117,
+      "step": 3337
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5276315345642703e-05,
+      "loss": 0.1506,
+      "step": 3338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273484442648165e-05,
+      "loss": 0.1431,
+      "step": 3339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5270652954111344e-05,
+      "loss": 0.218,
+      "step": 3340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5267820880346632e-05,
+      "loss": 0.198,
+      "step": 3341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526498822166849e-05,
+      "loss": 0.1719,
+      "step": 3342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5262154978391442e-05,
+      "loss": 0.176,
+      "step": 3343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5259321150830082e-05,
+      "loss": 0.1851,
+      "step": 3344
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5256486739299061e-05,
+      "loss": 0.1914,
+      "step": 3345
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5253651744113096e-05,
+      "loss": 0.2097,
+      "step": 3346
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5250816165586978e-05,
+      "loss": 0.176,
+      "step": 3347
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247980004035547e-05,
+      "loss": 0.2178,
+      "step": 3348
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5245143259773723e-05,
+      "loss": 0.199,
+      "step": 3349
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.524230593311648e-05,
+      "loss": 0.1829,
+      "step": 3350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5239468024378868e-05,
+      "loss": 0.1997,
+      "step": 3351
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523662953387599e-05,
+      "loss": 0.1804,
+      "step": 3352
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523379046192302e-05,
+      "loss": 0.1454,
+      "step": 3353
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5230950808835187e-05,
+      "loss": 0.188,
+      "step": 3354
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5228110574927806e-05,
+      "loss": 0.2292,
+      "step": 3355
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225269760516233e-05,
+      "loss": 0.2031,
+      "step": 3356
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5222428365915902e-05,
+      "loss": 0.1982,
+      "step": 3357
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.521958639144231e-05,
+      "loss": 0.1685,
+      "step": 3358
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5216743837411008e-05,
+      "loss": 0.1818,
+      "step": 3359
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213900704137627e-05,
+      "loss": 0.218,
+      "step": 3360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5211056991937856e-05,
+      "loss": 0.2644,
+      "step": 3361
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5208212701127437e-05,
+      "loss": 0.2063,
+      "step": 3362
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5205367832022197e-05,
+      "loss": 0.1936,
+      "step": 3363
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5202522384938012e-05,
+      "loss": 0.1802,
+      "step": 3364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5199676360190827e-05,
+      "loss": 0.1851,
+      "step": 3365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5196829758096652e-05,
+      "loss": 0.1887,
+      "step": 3366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193982578971556e-05,
+      "loss": 0.178,
+      "step": 3367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.519113482313168e-05,
+      "loss": 0.1919,
+      "step": 3368
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5188286490893227e-05,
+      "loss": 0.1804,
+      "step": 3369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185437582572458e-05,
+      "loss": 0.2114,
+      "step": 3370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.51825880984857e-05,
+      "loss": 0.1965,
+      "step": 3371
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5179738038949354e-05,
+      "loss": 0.2202,
+      "step": 3372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5176887404279872e-05,
+      "loss": 0.2073,
+      "step": 3373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517403619479377e-05,
+      "loss": 0.2126,
+      "step": 3374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.517118441080764e-05,
+      "loss": 0.2002,
+      "step": 3375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516833205263813e-05,
+      "loss": 0.186,
+      "step": 3376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5165479120601947e-05,
+      "loss": 0.1685,
+      "step": 3377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516262561501587e-05,
+      "loss": 0.2278,
+      "step": 3378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515977153619674e-05,
+      "loss": 0.2017,
+      "step": 3379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5156916884461457e-05,
+      "loss": 0.2104,
+      "step": 3380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5154061660126989e-05,
+      "loss": 0.1892,
+      "step": 3381
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5151205863510367e-05,
+      "loss": 0.15,
+      "step": 3382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5148349494928684e-05,
+      "loss": 0.2336,
+      "step": 3383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51454925546991e-05,
+      "loss": 0.1255,
+      "step": 3384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.514263504313883e-05,
+      "loss": 0.1771,
+      "step": 3385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5139776960565162e-05,
+      "loss": 0.2114,
+      "step": 3386
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5136918307295445e-05,
+      "loss": 0.1926,
+      "step": 3387
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5134059083647086e-05,
+      "loss": 0.2096,
+      "step": 3388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5131199289937566e-05,
+      "loss": 0.1721,
+      "step": 3389
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5128338926484419e-05,
+      "loss": 0.1383,
+      "step": 3390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125477993605243e-05,
+      "loss": 0.1829,
+      "step": 3391
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5122616491617706e-05,
+      "loss": 0.1792,
+      "step": 3392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5119754420839534e-05,
+      "loss": 0.1716,
+      "step": 3393
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5116891781588518e-05,
+      "loss": 0.189,
+      "step": 3394
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5114028574182511e-05,
+      "loss": 0.1841,
+      "step": 3395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5111164798939433e-05,
+      "loss": 0.1909,
+      "step": 3396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5108300456177257e-05,
+      "loss": 0.2158,
+      "step": 3397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5105435546214031e-05,
+      "loss": 0.175,
+      "step": 3398
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5102570069367861e-05,
+      "loss": 0.2065,
+      "step": 3399
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5099704025956911e-05,
+      "loss": 0.1785,
+      "step": 3400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5096837416299421e-05,
+      "loss": 0.196,
+      "step": 3401
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.509397024071368e-05,
+      "loss": 0.1948,
+      "step": 3402
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5091102499518042e-05,
+      "loss": 0.1814,
+      "step": 3403
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5088234193030933e-05,
+      "loss": 0.1836,
+      "step": 3404
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5085365321570833e-05,
+      "loss": 0.1863,
+      "step": 3405
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5082495885456293e-05,
+      "loss": 0.1721,
+      "step": 3406
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5079625885005912e-05,
+      "loss": 0.1802,
+      "step": 3407
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5076755320538368e-05,
+      "loss": 0.1824,
+      "step": 3408
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5073884192372393e-05,
+      "loss": 0.178,
+      "step": 3409
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5071012500826783e-05,
+      "loss": 0.2026,
+      "step": 3410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5068140246220396e-05,
+      "loss": 0.1841,
+      "step": 3411
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5065267428872152e-05,
+      "loss": 0.1899,
+      "step": 3412
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5062394049101038e-05,
+      "loss": 0.2314,
+      "step": 3413
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5059520107226097e-05,
+      "loss": 0.1738,
+      "step": 3414
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5056645603566444e-05,
+      "loss": 0.1898,
+      "step": 3415
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5053770538441246e-05,
+      "loss": 0.1678,
+      "step": 3416
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.505089491216973e-05,
+      "loss": 0.1445,
+      "step": 3417
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5048018725071202e-05,
+      "loss": 0.1656,
+      "step": 3418
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5045141977465016e-05,
+      "loss": 0.1875,
+      "step": 3419
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.504226466967059e-05,
+      "loss": 0.2163,
+      "step": 3420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503938680200741e-05,
+      "loss": 0.2144,
+      "step": 3421
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036508374795018e-05,
+      "loss": 0.2131,
+      "step": 3422
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503362938835302e-05,
+      "loss": 0.1616,
+      "step": 3423
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503074984300109e-05,
+      "loss": 0.1755,
+      "step": 3424
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5027869739058951e-05,
+      "loss": 0.1807,
+      "step": 3425
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5024989076846401e-05,
+      "loss": 0.1938,
+      "step": 3426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5022107856683294e-05,
+      "loss": 0.2092,
+      "step": 3427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5019226078889543e-05,
+      "loss": 0.1829,
+      "step": 3428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5016343743785135e-05,
+      "loss": 0.1772,
+      "step": 3429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5013460851690106e-05,
+      "loss": 0.2017,
+      "step": 3430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5010577402924556e-05,
+      "loss": 0.1687,
+      "step": 3431
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5007693397808652e-05,
+      "loss": 0.1517,
+      "step": 3432
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500480883666262e-05,
+      "loss": 0.1775,
+      "step": 3433
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5001923719806746e-05,
+      "loss": 0.1802,
+      "step": 3434
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499903804756138e-05,
+      "loss": 0.1853,
+      "step": 3435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4996151820246936e-05,
+      "loss": 0.1648,
+      "step": 3436
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4993265038183885e-05,
+      "loss": 0.198,
+      "step": 3437
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499037770169276e-05,
+      "loss": 0.1508,
+      "step": 3438
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4987489811094156e-05,
+      "loss": 0.1642,
+      "step": 3439
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4984601366708735e-05,
+      "loss": 0.1727,
+      "step": 3440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4981712368857213e-05,
+      "loss": 0.2117,
+      "step": 3441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497882281786037e-05,
+      "loss": 0.156,
+      "step": 3442
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4975932714039048e-05,
+      "loss": 0.2168,
+      "step": 3443
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4973042057714154e-05,
+      "loss": 0.178,
+      "step": 3444
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4970150849206643e-05,
+      "loss": 0.2152,
+      "step": 3445
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4967259088837552e-05,
+      "loss": 0.2278,
+      "step": 3446
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4964366776927963e-05,
+      "loss": 0.1671,
+      "step": 3447
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496147391379902e-05,
+      "loss": 0.1903,
+      "step": 3448
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4958580499771943e-05,
+      "loss": 0.2156,
+      "step": 3449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4955686535167994e-05,
+      "loss": 0.1685,
+      "step": 3450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.495279202030851e-05,
+      "loss": 0.1738,
+      "step": 3451
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4949896955514877e-05,
+      "loss": 0.1968,
+      "step": 3452
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4947001341108557e-05,
+      "loss": 0.1929,
+      "step": 3453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4944105177411061e-05,
+      "loss": 0.209,
+      "step": 3454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4941208464743968e-05,
+      "loss": 0.1719,
+      "step": 3455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4938311203428912e-05,
+      "loss": 0.1697,
+      "step": 3456
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493541339378759e-05,
+      "loss": 0.1917,
+      "step": 3457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4932515036141763e-05,
+      "loss": 0.1332,
+      "step": 3458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4929616130813252e-05,
+      "loss": 0.179,
+      "step": 3459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4926716678123933e-05,
+      "loss": 0.2151,
+      "step": 3460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4923816678395753e-05,
+      "loss": 0.187,
+      "step": 3461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492091613195071e-05,
+      "loss": 0.1787,
+      "step": 3462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4918015039110867e-05,
+      "loss": 0.1975,
+      "step": 3463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4915113400198349e-05,
+      "loss": 0.1824,
+      "step": 3464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4912211215535337e-05,
+      "loss": 0.22,
+      "step": 3465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.490930848544408e-05,
+      "loss": 0.1836,
+      "step": 3466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4906405210246881e-05,
+      "loss": 0.2087,
+      "step": 3467
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4903501390266104e-05,
+      "loss": 0.2075,
+      "step": 3468
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4900597025824177e-05,
+      "loss": 0.177,
+      "step": 3469
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4897692117243588e-05,
+      "loss": 0.207,
+      "step": 3470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4894786664846879e-05,
+      "loss": 0.1699,
+      "step": 3471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4891880668956661e-05,
+      "loss": 0.1868,
+      "step": 3472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4888974129895606e-05,
+      "loss": 0.1895,
+      "step": 3473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4886067047986432e-05,
+      "loss": 0.228,
+      "step": 3474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4883159423551937e-05,
+      "loss": 0.2102,
+      "step": 3475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4880251256914964e-05,
+      "loss": 0.1912,
+      "step": 3476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4877342548398421e-05,
+      "loss": 0.194,
+      "step": 3477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4874433298325282e-05,
+      "loss": 0.1826,
+      "step": 3478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487152350701857e-05,
+      "loss": 0.197,
+      "step": 3479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4868613174801382e-05,
+      "loss": 0.2158,
+      "step": 3480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865702301996863e-05,
+      "loss": 0.1892,
+      "step": 3481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4862790888928219e-05,
+      "loss": 0.1982,
+      "step": 3482
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4859878935918724e-05,
+      "loss": 0.1824,
+      "step": 3483
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4856966443291705e-05,
+      "loss": 0.1924,
+      "step": 3484
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4854053411370552e-05,
+      "loss": 0.1726,
+      "step": 3485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4851139840478714e-05,
+      "loss": 0.2031,
+      "step": 3486
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.48482257309397e-05,
+      "loss": 0.1914,
+      "step": 3487
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4845311083077076e-05,
+      "loss": 0.1987,
+      "step": 3488
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4842395897214477e-05,
+      "loss": 0.1964,
+      "step": 3489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4839480173675584e-05,
+      "loss": 0.1826,
+      "step": 3490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4836563912784149e-05,
+      "loss": 0.1664,
+      "step": 3491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4833647114863978e-05,
+      "loss": 0.1716,
+      "step": 3492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4830729780238937e-05,
+      "loss": 0.2039,
+      "step": 3493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4827811909232956e-05,
+      "loss": 0.1917,
+      "step": 3494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4824893502170019e-05,
+      "loss": 0.1937,
+      "step": 3495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4821974559374172e-05,
+      "loss": 0.2024,
+      "step": 3496
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4819055081169523e-05,
+      "loss": 0.1882,
+      "step": 3497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4816135067880232e-05,
+      "loss": 0.1897,
+      "step": 3498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4813214519830527e-05,
+      "loss": 0.1677,
+      "step": 3499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.481029343734469e-05,
+      "loss": 0.1626,
+      "step": 3500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4807371820747066e-05,
+      "loss": 0.2014,
+      "step": 3501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4804449670362053e-05,
+      "loss": 0.2783,
+      "step": 3502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4801526986514117e-05,
+      "loss": 0.2329,
+      "step": 3503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4798603769527775e-05,
+      "loss": 0.1753,
+      "step": 3504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4795680019727607e-05,
+      "loss": 0.1951,
+      "step": 3505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792755737438258e-05,
+      "loss": 0.1559,
+      "step": 3506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.478983092298442e-05,
+      "loss": 0.2007,
+      "step": 3507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4786905576690852e-05,
+      "loss": 0.1846,
+      "step": 3508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4783979698882372e-05,
+      "loss": 0.1886,
+      "step": 3509
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4781053289883851e-05,
+      "loss": 0.1775,
+      "step": 3510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4778126350020227e-05,
+      "loss": 0.1853,
+      "step": 3511
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4775198879616495e-05,
+      "loss": 0.1799,
+      "step": 3512
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4772270878997701e-05,
+      "loss": 0.1499,
+      "step": 3513
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4769342348488964e-05,
+      "loss": 0.1848,
+      "step": 3514
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766413288415447e-05,
+      "loss": 0.1626,
+      "step": 3515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476348369910238e-05,
+      "loss": 0.2053,
+      "step": 3516
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4760553580875057e-05,
+      "loss": 0.1604,
+      "step": 3517
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4757622934058816e-05,
+      "loss": 0.2119,
+      "step": 3518
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4754691758979065e-05,
+      "loss": 0.2058,
+      "step": 3519
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4751760055961267e-05,
+      "loss": 0.2041,
+      "step": 3520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4748827825330947e-05,
+      "loss": 0.1973,
+      "step": 3521
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474589506741368e-05,
+      "loss": 0.1985,
+      "step": 3522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.474296178253511e-05,
+      "loss": 0.1429,
+      "step": 3523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4740027971020936e-05,
+      "loss": 0.175,
+      "step": 3524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4737093633196909e-05,
+      "loss": 0.1531,
+      "step": 3525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.473415876938885e-05,
+      "loss": 0.2104,
+      "step": 3526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4731223379922625e-05,
+      "loss": 0.1543,
+      "step": 3527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4728287465124169e-05,
+      "loss": 0.1907,
+      "step": 3528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4725351025319474e-05,
+      "loss": 0.1422,
+      "step": 3529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4722414060834587e-05,
+      "loss": 0.2085,
+      "step": 3530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.471947657199561e-05,
+      "loss": 0.1742,
+      "step": 3531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4716538559128715e-05,
+      "loss": 0.1901,
+      "step": 3532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4713600022560117e-05,
+      "loss": 0.1576,
+      "step": 3533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4710660962616104e-05,
+      "loss": 0.1805,
+      "step": 3534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4707721379623012e-05,
+      "loss": 0.172,
+      "step": 3535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4704781273907234e-05,
+      "loss": 0.1838,
+      "step": 3536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4701840645795234e-05,
+      "loss": 0.1809,
+      "step": 3537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4698899495613519e-05,
+      "loss": 0.1895,
+      "step": 3538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4695957823688659e-05,
+      "loss": 0.1887,
+      "step": 3539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4693015630347287e-05,
+      "loss": 0.1584,
+      "step": 3540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4690072915916091e-05,
+      "loss": 0.1858,
+      "step": 3541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4687129680721807e-05,
+      "loss": 0.1667,
+      "step": 3542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.468418592509125e-05,
+      "loss": 0.1655,
+      "step": 3543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4681241649351275e-05,
+      "loss": 0.1609,
+      "step": 3544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4678296853828795e-05,
+      "loss": 0.1897,
+      "step": 3545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4675351538850793e-05,
+      "loss": 0.1833,
+      "step": 3546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.46724057047443e-05,
+      "loss": 0.1503,
+      "step": 3547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4669459351836404e-05,
+      "loss": 0.1699,
+      "step": 3548
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4666512480454261e-05,
+      "loss": 0.2131,
+      "step": 3549
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.466356509092507e-05,
+      "loss": 0.2559,
+      "step": 3550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4660617183576102e-05,
+      "loss": 0.2214,
+      "step": 3551
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657668758734673e-05,
+      "loss": 0.2104,
+      "step": 3552
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4654719816728164e-05,
+      "loss": 0.1794,
+      "step": 3553
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.465177035788401e-05,
+      "loss": 0.1709,
+      "step": 3554
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4648820382529707e-05,
+      "loss": 0.2046,
+      "step": 3555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4645869890992804e-05,
+      "loss": 0.2019,
+      "step": 3556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4642918883600909e-05,
+      "loss": 0.2009,
+      "step": 3557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4639967360681691e-05,
+      "loss": 0.2048,
+      "step": 3558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.463701532256287e-05,
+      "loss": 0.1753,
+      "step": 3559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634062769572225e-05,
+      "loss": 0.1643,
+      "step": 3560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4631109702037597e-05,
+      "loss": 0.1963,
+      "step": 3561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4628156120286879e-05,
+      "loss": 0.152,
+      "step": 3562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4625202024648019e-05,
+      "loss": 0.1904,
+      "step": 3563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4622247415449028e-05,
+      "loss": 0.1539,
+      "step": 3564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4619292293017972e-05,
+      "loss": 0.2092,
+      "step": 3565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4616336657682977e-05,
+      "loss": 0.1837,
+      "step": 3566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4613380509772215e-05,
+      "loss": 0.196,
+      "step": 3567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4610423849613926e-05,
+      "loss": 0.2134,
+      "step": 3568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4607466677536405e-05,
+      "loss": 0.2161,
+      "step": 3569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4604508993868e-05,
+      "loss": 0.2249,
+      "step": 3570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601550798937121e-05,
+      "loss": 0.1669,
+      "step": 3571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4598592093072228e-05,
+      "loss": 0.157,
+      "step": 3572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4595632876601841e-05,
+      "loss": 0.1708,
+      "step": 3573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4592673149854541e-05,
+      "loss": 0.1709,
+      "step": 3574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458971291315896e-05,
+      "loss": 0.2056,
+      "step": 3575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4586752166843786e-05,
+      "loss": 0.1829,
+      "step": 3576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.458379091123777e-05,
+      "loss": 0.168,
+      "step": 3577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4580829146669713e-05,
+      "loss": 0.1824,
+      "step": 3578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4577866873468473e-05,
+      "loss": 0.1399,
+      "step": 3579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4574904091962973e-05,
+      "loss": 0.1843,
+      "step": 3580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4571940802482183e-05,
+      "loss": 0.1865,
+      "step": 3581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4568977005355128e-05,
+      "loss": 0.1731,
+      "step": 3582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4566012700910896e-05,
+      "loss": 0.1619,
+      "step": 3583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4563047889478633e-05,
+      "loss": 0.2178,
+      "step": 3584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4560082571387537e-05,
+      "loss": 0.2051,
+      "step": 3585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4557116746966857e-05,
+      "loss": 0.1836,
+      "step": 3586
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4554150416545908e-05,
+      "loss": 0.1611,
+      "step": 3587
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4551183580454056e-05,
+      "loss": 0.1936,
+      "step": 3588
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4548216239020727e-05,
+      "loss": 0.2163,
+      "step": 3589
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4545248392575392e-05,
+      "loss": 0.1802,
+      "step": 3590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4542280041447599e-05,
+      "loss": 0.2234,
+      "step": 3591
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4539311185966929e-05,
+      "loss": 0.1753,
+      "step": 3592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4536341826463032e-05,
+      "loss": 0.2139,
+      "step": 3593
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4533371963265613e-05,
+      "loss": 0.1731,
+      "step": 3594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4530401596704432e-05,
+      "loss": 0.1755,
+      "step": 3595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4527430727109299e-05,
+      "loss": 0.1681,
+      "step": 3596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4524459354810091e-05,
+      "loss": 0.2334,
+      "step": 3597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4521487480136734e-05,
+      "loss": 0.208,
+      "step": 3598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4518515103419207e-05,
+      "loss": 0.1677,
+      "step": 3599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515542224987552e-05,
+      "loss": 0.209,
+      "step": 3600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4512568845171861e-05,
+      "loss": 0.1772,
+      "step": 3601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509594964302283e-05,
+      "loss": 0.1702,
+      "step": 3602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4506620582709026e-05,
+      "loss": 0.1838,
+      "step": 3603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.450364570072235e-05,
+      "loss": 0.2197,
+      "step": 3604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4500670318672571e-05,
+      "loss": 0.1758,
+      "step": 3605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4497694436890062e-05,
+      "loss": 0.2148,
+      "step": 3606
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.449471805570525e-05,
+      "loss": 0.2371,
+      "step": 3607
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4491741175448617e-05,
+      "loss": 0.161,
+      "step": 3608
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4488763796450704e-05,
+      "loss": 0.1731,
+      "step": 3609
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4485785919042101e-05,
+      "loss": 0.1687,
+      "step": 3610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448280754355346e-05,
+      "loss": 0.1982,
+      "step": 3611
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4479828670315486e-05,
+      "loss": 0.1431,
+      "step": 3612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4476849299658938e-05,
+      "loss": 0.1953,
+      "step": 3613
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4473869431914628e-05,
+      "loss": 0.1531,
+      "step": 3614
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4470889067413432e-05,
+      "loss": 0.1934,
+      "step": 3615
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446790820648627e-05,
+      "loss": 0.2053,
+      "step": 3616
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4464926849464122e-05,
+      "loss": 0.1687,
+      "step": 3617
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446194499667803e-05,
+      "loss": 0.1838,
+      "step": 3618
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4458962648459079e-05,
+      "loss": 0.165,
+      "step": 3619
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4455979805138416e-05,
+      "loss": 0.1787,
+      "step": 3620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4452996467047243e-05,
+      "loss": 0.1902,
+      "step": 3621
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445001263451681e-05,
+      "loss": 0.2048,
+      "step": 3622
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4447028307878435e-05,
+      "loss": 0.2002,
+      "step": 3623
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4444043487463479e-05,
+      "loss": 0.1993,
+      "step": 3624
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444105817360336e-05,
+      "loss": 0.1694,
+      "step": 3625
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4438072366629561e-05,
+      "loss": 0.1735,
+      "step": 3626
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4435086066873602e-05,
+      "loss": 0.1553,
+      "step": 3627
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4432099274667073e-05,
+      "loss": 0.157,
+      "step": 3628
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442911199034161e-05,
+      "loss": 0.1772,
+      "step": 3629
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442612421422891e-05,
+      "loss": 0.2021,
+      "step": 3630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442313594666072e-05,
+      "loss": 0.2063,
+      "step": 3631
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442014718796884e-05,
+      "loss": 0.1643,
+      "step": 3632
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4417157938485128e-05,
+      "loss": 0.1838,
+      "step": 3633
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4414168198541503e-05,
+      "loss": 0.1631,
+      "step": 3634
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4411177968469922e-05,
+      "loss": 0.2275,
+      "step": 3635
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440818724860241e-05,
+      "loss": 0.2113,
+      "step": 3636
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440519603927104e-05,
+      "loss": 0.2439,
+      "step": 3637
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4402204340807947e-05,
+      "loss": 0.1646,
+      "step": 3638
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4399212153545305e-05,
+      "loss": 0.1677,
+      "step": 3639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4396219477815363e-05,
+      "loss": 0.1562,
+      "step": 3640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4393226313950406e-05,
+      "loss": 0.1539,
+      "step": 3641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4390232662282784e-05,
+      "loss": 0.2129,
+      "step": 3642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4387238523144894e-05,
+      "loss": 0.1667,
+      "step": 3643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4384243896869194e-05,
+      "loss": 0.1855,
+      "step": 3644
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4381248783788194e-05,
+      "loss": 0.1792,
+      "step": 3645
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4378253184234452e-05,
+      "loss": 0.22,
+      "step": 3646
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4375257098540588e-05,
+      "loss": 0.1929,
+      "step": 3647
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4372260527039275e-05,
+      "loss": 0.1826,
+      "step": 3648
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4369263470063235e-05,
+      "loss": 0.1783,
+      "step": 3649
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4366265927945244e-05,
+      "loss": 0.1829,
+      "step": 3650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4363267901018141e-05,
+      "loss": 0.2224,
+      "step": 3651
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4360269389614809e-05,
+      "loss": 0.1787,
+      "step": 3652
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4357270394068186e-05,
+      "loss": 0.2239,
+      "step": 3653
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4354270914711273e-05,
+      "loss": 0.1926,
+      "step": 3654
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.435127095187711e-05,
+      "loss": 0.1488,
+      "step": 3655
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4348270505898802e-05,
+      "loss": 0.2263,
+      "step": 3656
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345269577109503e-05,
+      "loss": 0.2058,
+      "step": 3657
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4342268165842426e-05,
+      "loss": 0.1738,
+      "step": 3658
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4339266272430828e-05,
+      "loss": 0.1882,
+      "step": 3659
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4336263897208025e-05,
+      "loss": 0.2029,
+      "step": 3660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433326104050739e-05,
+      "loss": 0.1635,
+      "step": 3661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.433025770266234e-05,
+      "loss": 0.1638,
+      "step": 3662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4327253884006358e-05,
+      "loss": 0.1859,
+      "step": 3663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4324249584872969e-05,
+      "loss": 0.2061,
+      "step": 3664
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4321244805595758e-05,
+      "loss": 0.1978,
+      "step": 3665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431823954650836e-05,
+      "loss": 0.186,
+      "step": 3666
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4315233807944462e-05,
+      "loss": 0.1919,
+      "step": 3667
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431222759023781e-05,
+      "loss": 0.1416,
+      "step": 3668
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4309220893722203e-05,
+      "loss": 0.1714,
+      "step": 3669
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.430621371873148e-05,
+      "loss": 0.2,
+      "step": 3670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4303206065599554e-05,
+      "loss": 0.1354,
+      "step": 3671
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4300197934660378e-05,
+      "loss": 0.207,
+      "step": 3672
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4297189326247956e-05,
+      "loss": 0.1411,
+      "step": 3673
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4294180240696351e-05,
+      "loss": 0.1763,
+      "step": 3674
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4291170678339679e-05,
+      "loss": 0.1604,
+      "step": 3675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4288160639512105e-05,
+      "loss": 0.1907,
+      "step": 3676
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4285150124547853e-05,
+      "loss": 0.1713,
+      "step": 3677
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4282139133781192e-05,
+      "loss": 0.1659,
+      "step": 3678
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4279127667546448e-05,
+      "loss": 0.2083,
+      "step": 3679
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276115726178003e-05,
+      "loss": 0.2063,
+      "step": 3680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4273103310010288e-05,
+      "loss": 0.1709,
+      "step": 3681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4270090419377783e-05,
+      "loss": 0.1667,
+      "step": 3682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4267077054615026e-05,
+      "loss": 0.1929,
+      "step": 3683
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4264063216056609e-05,
+      "loss": 0.166,
+      "step": 3684
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4261048904037178e-05,
+      "loss": 0.1816,
+      "step": 3685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425803411889142e-05,
+      "loss": 0.1995,
+      "step": 3686
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4255018860954084e-05,
+      "loss": 0.2449,
+      "step": 3687
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4252003130559971e-05,
+      "loss": 0.2134,
+      "step": 3688
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4248986928043936e-05,
+      "loss": 0.1537,
+      "step": 3689
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4245970253740874e-05,
+      "loss": 0.1968,
+      "step": 3690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4242953107985756e-05,
+      "loss": 0.1863,
+      "step": 3691
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4239935491113579e-05,
+      "loss": 0.1992,
+      "step": 3692
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4236917403459412e-05,
+      "loss": 0.1715,
+      "step": 3693
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4233898845358365e-05,
+      "loss": 0.1858,
+      "step": 3694
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4230879817145607e-05,
+      "loss": 0.1846,
+      "step": 3695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4227860319156353e-05,
+      "loss": 0.2046,
+      "step": 3696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4224840351725875e-05,
+      "loss": 0.1858,
+      "step": 3697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4221819915189496e-05,
+      "loss": 0.1946,
+      "step": 3698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4218799009882593e-05,
+      "loss": 0.1802,
+      "step": 3699
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4215777636140589e-05,
+      "loss": 0.1809,
+      "step": 3700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4212755794298962e-05,
+      "loss": 0.1929,
+      "step": 3701
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.420973348469325e-05,
+      "loss": 0.1383,
+      "step": 3702
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4206710707659025e-05,
+      "loss": 0.1846,
+      "step": 3703
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.420368746353193e-05,
+      "loss": 0.1936,
+      "step": 3704
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4200663752647649e-05,
+      "loss": 0.1895,
+      "step": 3705
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419763957534192e-05,
+      "loss": 0.2014,
+      "step": 3706
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419461493195053e-05,
+      "loss": 0.1725,
+      "step": 3707
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4191589822809326e-05,
+      "loss": 0.1765,
+      "step": 3708
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4188564248254201e-05,
+      "loss": 0.2175,
+      "step": 3709
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4185538208621097e-05,
+      "loss": 0.2073,
+      "step": 3710
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182511704246012e-05,
+      "loss": 0.1858,
+      "step": 3711
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4179484735464995e-05,
+      "loss": 0.2024,
+      "step": 3712
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4176457302614145e-05,
+      "loss": 0.1305,
+      "step": 3713
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4173429406029616e-05,
+      "loss": 0.188,
+      "step": 3714
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4170401046047608e-05,
+      "loss": 0.1807,
+      "step": 3715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4167372223004379e-05,
+      "loss": 0.186,
+      "step": 3716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4164342937236231e-05,
+      "loss": 0.2271,
+      "step": 3717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161313189079523e-05,
+      "loss": 0.1819,
+      "step": 3718
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4158282978870665e-05,
+      "loss": 0.1938,
+      "step": 3719
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4155252306946117e-05,
+      "loss": 0.1777,
+      "step": 3720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4152221173642385e-05,
+      "loss": 0.208,
+      "step": 3721
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.414918957929604e-05,
+      "loss": 0.155,
+      "step": 3722
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4146157524243686e-05,
+      "loss": 0.1733,
+      "step": 3723
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4143125008821996e-05,
+      "loss": 0.1868,
+      "step": 3724
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.414009203336768e-05,
+      "loss": 0.193,
+      "step": 3725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4137058598217511e-05,
+      "loss": 0.1901,
+      "step": 3726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4134024703708301e-05,
+      "loss": 0.1682,
+      "step": 3727
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4130990350176924e-05,
+      "loss": 0.1676,
+      "step": 3728
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4127955537960296e-05,
+      "loss": 0.1721,
+      "step": 3729
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4124920267395388e-05,
+      "loss": 0.1858,
+      "step": 3730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4121884538819224e-05,
+      "loss": 0.1459,
+      "step": 3731
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.411884835256888e-05,
+      "loss": 0.2039,
+      "step": 3732
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4115811708981469e-05,
+      "loss": 0.1792,
+      "step": 3733
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4112774608394174e-05,
+      "loss": 0.17,
+      "step": 3734
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4109737051144221e-05,
+      "loss": 0.1462,
+      "step": 3735
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4106699037568878e-05,
+      "loss": 0.1709,
+      "step": 3736
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4103660568005479e-05,
+      "loss": 0.1838,
+      "step": 3737
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4100621642791395e-05,
+      "loss": 0.2026,
+      "step": 3738
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4097582262264058e-05,
+      "loss": 0.1941,
+      "step": 3739
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4094542426760947e-05,
+      "loss": 0.1473,
+      "step": 3740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4091502136619583e-05,
+      "loss": 0.1646,
+      "step": 3741
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4088461392177555e-05,
+      "loss": 0.1838,
+      "step": 3742
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4085420193772485e-05,
+      "loss": 0.1907,
+      "step": 3743
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4082378541742057e-05,
+      "loss": 0.2068,
+      "step": 3744
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4079336436424e-05,
+      "loss": 0.1597,
+      "step": 3745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4076293878156097e-05,
+      "loss": 0.2083,
+      "step": 3746
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4073250867276174e-05,
+      "loss": 0.1697,
+      "step": 3747
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4070207404122118e-05,
+      "loss": 0.1882,
+      "step": 3748
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4067163489031858e-05,
+      "loss": 0.1978,
+      "step": 3749
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064119122343374e-05,
+      "loss": 0.1882,
+      "step": 3750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.40610743043947e-05,
+      "loss": 0.187,
+      "step": 3751
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058029035523917e-05,
+      "loss": 0.1888,
+      "step": 3752
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4054983316069155e-05,
+      "loss": 0.1953,
+      "step": 3753
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4051937146368602e-05,
+      "loss": 0.1726,
+      "step": 3754
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4048890526760484e-05,
+      "loss": 0.1494,
+      "step": 3755
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4045843457583086e-05,
+      "loss": 0.2002,
+      "step": 3756
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4042795939174738e-05,
+      "loss": 0.1812,
+      "step": 3757
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4039747971873822e-05,
+      "loss": 0.1914,
+      "step": 3758
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403669955601877e-05,
+      "loss": 0.1641,
+      "step": 3759
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4033650691948068e-05,
+      "loss": 0.16,
+      "step": 3760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403060138000024e-05,
+      "loss": 0.197,
+      "step": 3761
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402755162051387e-05,
+      "loss": 0.2009,
+      "step": 3762
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402450141382759e-05,
+      "loss": 0.1812,
+      "step": 3763
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4021450760280076e-05,
+      "loss": 0.1553,
+      "step": 3764
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.401839966021006e-05,
+      "loss": 0.2263,
+      "step": 3765
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4015348113956324e-05,
+      "loss": 0.2075,
+      "step": 3766
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.401229612185769e-05,
+      "loss": 0.1558,
+      "step": 3767
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400924368425305e-05,
+      "loss": 0.1683,
+      "step": 3768
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4006190801481317e-05,
+      "loss": 0.1483,
+      "step": 3769
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4003137473881476e-05,
+      "loss": 0.2068,
+      "step": 3770
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4000083701792554e-05,
+      "loss": 0.1914,
+      "step": 3771
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3997029485553622e-05,
+      "loss": 0.209,
+      "step": 3772
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3993974825503812e-05,
+      "loss": 0.1755,
+      "step": 3773
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3990919721982294e-05,
+      "loss": 0.193,
+      "step": 3774
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3987864175328294e-05,
+      "loss": 0.1926,
+      "step": 3775
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3984808185881086e-05,
+      "loss": 0.2048,
+      "step": 3776
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3981751753979993e-05,
+      "loss": 0.1794,
+      "step": 3777
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3978694879964381e-05,
+      "loss": 0.1598,
+      "step": 3778
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3975637564173676e-05,
+      "loss": 0.196,
+      "step": 3779
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3972579806947347e-05,
+      "loss": 0.2141,
+      "step": 3780
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3969521608624909e-05,
+      "loss": 0.1687,
+      "step": 3781
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3966462969545933e-05,
+      "loss": 0.2241,
+      "step": 3782
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3963403890050036e-05,
+      "loss": 0.1992,
+      "step": 3783
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.396034437047688e-05,
+      "loss": 0.2188,
+      "step": 3784
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3957284411166183e-05,
+      "loss": 0.1777,
+      "step": 3785
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3954224012457707e-05,
+      "loss": 0.1833,
+      "step": 3786
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395116317469126e-05,
+      "loss": 0.1323,
+      "step": 3787
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.394810189820671e-05,
+      "loss": 0.1772,
+      "step": 3788
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3945040183343958e-05,
+      "loss": 0.1826,
+      "step": 3789
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3941978030442968e-05,
+      "loss": 0.153,
+      "step": 3790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3938915439843748e-05,
+      "loss": 0.1829,
+      "step": 3791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3935852411886347e-05,
+      "loss": 0.1838,
+      "step": 3792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3932788946910877e-05,
+      "loss": 0.1992,
+      "step": 3793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392972504525748e-05,
+      "loss": 0.1783,
+      "step": 3794
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3926660707266364e-05,
+      "loss": 0.1948,
+      "step": 3795
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392359593327778e-05,
+      "loss": 0.1541,
+      "step": 3796
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392053072363202e-05,
+      "loss": 0.1534,
+      "step": 3797
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3917465078669433e-05,
+      "loss": 0.1608,
+      "step": 3798
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3914398998730414e-05,
+      "loss": 0.1722,
+      "step": 3799
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3911332484155406e-05,
+      "loss": 0.1804,
+      "step": 3800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3908265535284895e-05,
+      "loss": 0.1982,
+      "step": 3801
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3905198152459427e-05,
+      "loss": 0.2244,
+      "step": 3802
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3902130336019585e-05,
+      "loss": 0.1858,
+      "step": 3803
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3899062086306008e-05,
+      "loss": 0.1702,
+      "step": 3804
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3895993403659377e-05,
+      "loss": 0.2078,
+      "step": 3805
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3892924288420424e-05,
+      "loss": 0.1807,
+      "step": 3806
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3889854740929927e-05,
+      "loss": 0.1578,
+      "step": 3807
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.388678476152872e-05,
+      "loss": 0.1826,
+      "step": 3808
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.388371435055767e-05,
+      "loss": 0.1785,
+      "step": 3809
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3880643508357708e-05,
+      "loss": 0.2046,
+      "step": 3810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3877572235269802e-05,
+      "loss": 0.186,
+      "step": 3811
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3874500531634975e-05,
+      "loss": 0.1666,
+      "step": 3812
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3871428397794286e-05,
+      "loss": 0.2051,
+      "step": 3813
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386835583408886e-05,
+      "loss": 0.1895,
+      "step": 3814
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386528284085985e-05,
+      "loss": 0.1587,
+      "step": 3815
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3862209418448476e-05,
+      "loss": 0.1788,
+      "step": 3816
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3859135567195989e-05,
+      "loss": 0.1654,
+      "step": 3817
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3856061287443696e-05,
+      "loss": 0.1851,
+      "step": 3818
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3852986579532954e-05,
+      "loss": 0.179,
+      "step": 3819
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384991144380516e-05,
+      "loss": 0.178,
+      "step": 3820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3846835880601761e-05,
+      "loss": 0.1605,
+      "step": 3821
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3843759890264256e-05,
+      "loss": 0.1912,
+      "step": 3822
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3840683473134186e-05,
+      "loss": 0.1946,
+      "step": 3823
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837606629553143e-05,
+      "loss": 0.1726,
+      "step": 3824
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3834529359862763e-05,
+      "loss": 0.1702,
+      "step": 3825
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3831451664404737e-05,
+      "loss": 0.1516,
+      "step": 3826
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3828373543520787e-05,
+      "loss": 0.1694,
+      "step": 3827
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3825294997552703e-05,
+      "loss": 0.2058,
+      "step": 3828
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3822216026842306e-05,
+      "loss": 0.1616,
+      "step": 3829
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3819136631731471e-05,
+      "loss": 0.2048,
+      "step": 3830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381605681256212e-05,
+      "loss": 0.249,
+      "step": 3831
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381297656967622e-05,
+      "loss": 0.1707,
+      "step": 3832
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3809895903415788e-05,
+      "loss": 0.1895,
+      "step": 3833
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806814814122888e-05,
+      "loss": 0.2041,
+      "step": 3834
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3803733302139624e-05,
+      "loss": 0.1733,
+      "step": 3835
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3800651367808158e-05,
+      "loss": 0.2156,
+      "step": 3836
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3797569011470688e-05,
+      "loss": 0.2031,
+      "step": 3837
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3794486233469466e-05,
+      "loss": 0.1785,
+      "step": 3838
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3791403034146788e-05,
+      "loss": 0.1926,
+      "step": 3839
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3788319413845004e-05,
+      "loss": 0.199,
+      "step": 3840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3785235372906494e-05,
+      "loss": 0.1851,
+      "step": 3841
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3782150911673702e-05,
+      "loss": 0.1809,
+      "step": 3842
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3779066030489107e-05,
+      "loss": 0.1624,
+      "step": 3843
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3775980729695247e-05,
+      "loss": 0.1611,
+      "step": 3844
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3772895009634693e-05,
+      "loss": 0.1907,
+      "step": 3845
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3769808870650066e-05,
+      "loss": 0.1687,
+      "step": 3846
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3766722313084042e-05,
+      "loss": 0.1863,
+      "step": 3847
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3763635337279335e-05,
+      "loss": 0.1631,
+      "step": 3848
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3760547943578706e-05,
+      "loss": 0.2056,
+      "step": 3849
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3757460132324968e-05,
+      "loss": 0.2026,
+      "step": 3850
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3754371903860976e-05,
+      "loss": 0.1664,
+      "step": 3851
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.375128325852963e-05,
+      "loss": 0.1584,
+      "step": 3852
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3748194196673876e-05,
+      "loss": 0.2068,
+      "step": 3853
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3745104718636718e-05,
+      "loss": 0.1899,
+      "step": 3854
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3742014824761183e-05,
+      "loss": 0.2363,
+      "step": 3855
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373892451539037e-05,
+      "loss": 0.2219,
+      "step": 3856
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3735833790867407e-05,
+      "loss": 0.1978,
+      "step": 3857
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3732742651535471e-05,
+      "loss": 0.2007,
+      "step": 3858
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3729651097737793e-05,
+      "loss": 0.1895,
+      "step": 3859
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3726559129817641e-05,
+      "loss": 0.157,
+      "step": 3860
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3723466748118329e-05,
+      "loss": 0.2078,
+      "step": 3861
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3720373952983226e-05,
+      "loss": 0.1663,
+      "step": 3862
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3717280744755738e-05,
+      "loss": 0.1837,
+      "step": 3863
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3714187123779316e-05,
+      "loss": 0.2119,
+      "step": 3864
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3711093090397471e-05,
+      "loss": 0.1655,
+      "step": 3865
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370799864495374e-05,
+      "loss": 0.196,
+      "step": 3866
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704903787791722e-05,
+      "loss": 0.2117,
+      "step": 3867
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3701808519255051e-05,
+      "loss": 0.2148,
+      "step": 3868
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3698712839687409e-05,
+      "loss": 0.2097,
+      "step": 3869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369561674943253e-05,
+      "loss": 0.2002,
+      "step": 3870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369252024883419e-05,
+      "loss": 0.1689,
+      "step": 3871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3689423338236201e-05,
+      "loss": 0.1794,
+      "step": 3872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3686326017982436e-05,
+      "loss": 0.1565,
+      "step": 3873
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3683228288416807e-05,
+      "loss": 0.15,
+      "step": 3874
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3680130149883266e-05,
+      "loss": 0.1672,
+      "step": 3875
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.367703160272582e-05,
+      "loss": 0.1831,
+      "step": 3876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3673932647288515e-05,
+      "loss": 0.2021,
+      "step": 3877
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3670833283915444e-05,
+      "loss": 0.1409,
+      "step": 3878
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3667733512950742e-05,
+      "loss": 0.1512,
+      "step": 3879
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3664633334738601e-05,
+      "loss": 0.1865,
+      "step": 3880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3661532749623242e-05,
+      "loss": 0.2134,
+      "step": 3881
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3658431757948941e-05,
+      "loss": 0.1816,
+      "step": 3882
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3655330360060025e-05,
+      "loss": 0.1553,
+      "step": 3883
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3652228556300844e-05,
+      "loss": 0.2092,
+      "step": 3884
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3649126347015816e-05,
+      "loss": 0.2029,
+      "step": 3885
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3646023732549396e-05,
+      "loss": 0.186,
+      "step": 3886
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364292071324608e-05,
+      "loss": 0.1987,
+      "step": 3887
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3639817289450412e-05,
+      "loss": 0.1555,
+      "step": 3888
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3636713461506985e-05,
+      "loss": 0.1583,
+      "step": 3889
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3633609229760429e-05,
+      "loss": 0.198,
+      "step": 3890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.363050459455543e-05,
+      "loss": 0.1748,
+      "step": 3891
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.36273995562367e-05,
+      "loss": 0.1614,
+      "step": 3892
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362429411514902e-05,
+      "loss": 0.1895,
+      "step": 3893
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3621188271637193e-05,
+      "loss": 0.1931,
+      "step": 3894
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3618082026046084e-05,
+      "loss": 0.1627,
+      "step": 3895
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3614975378720592e-05,
+      "loss": 0.1411,
+      "step": 3896
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3611868330005663e-05,
+      "loss": 0.199,
+      "step": 3897
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3608760880246293e-05,
+      "loss": 0.1887,
+      "step": 3898
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3605653029787513e-05,
+      "loss": 0.1882,
+      "step": 3899
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3602544778974407e-05,
+      "loss": 0.1692,
+      "step": 3900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3599436128152101e-05,
+      "loss": 0.186,
+      "step": 3901
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3596327077665765e-05,
+      "loss": 0.2001,
+      "step": 3902
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.359321762786061e-05,
+      "loss": 0.1602,
+      "step": 3903
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3590107779081894e-05,
+      "loss": 0.1566,
+      "step": 3904
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3586997531674924e-05,
+      "loss": 0.1722,
+      "step": 3905
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358388688598504e-05,
+      "loss": 0.178,
+      "step": 3906
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3580775842357641e-05,
+      "loss": 0.1951,
+      "step": 3907
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3577664401138158e-05,
+      "loss": 0.1854,
+      "step": 3908
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3574552562672068e-05,
+      "loss": 0.1416,
+      "step": 3909
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.35714403273049e-05,
+      "loss": 0.1772,
+      "step": 3910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3568327695382219e-05,
+      "loss": 0.1514,
+      "step": 3911
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3565214667249633e-05,
+      "loss": 0.1836,
+      "step": 3912
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3562101243252805e-05,
+      "loss": 0.1858,
+      "step": 3913
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3558987423737434e-05,
+      "loss": 0.1819,
+      "step": 3914
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3555873209049252e-05,
+      "loss": 0.2041,
+      "step": 3915
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3552758599534062e-05,
+      "loss": 0.1895,
+      "step": 3916
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3549643595537685e-05,
+      "loss": 0.152,
+      "step": 3917
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3546528197405998e-05,
+      "loss": 0.186,
+      "step": 3918
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3543412405484924e-05,
+      "loss": 0.1816,
+      "step": 3919
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.354029622012042e-05,
+      "loss": 0.1812,
+      "step": 3920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3537179641658495e-05,
+      "loss": 0.1398,
+      "step": 3921
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3534062670445198e-05,
+      "loss": 0.1633,
+      "step": 3922
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3530945306826624e-05,
+      "loss": 0.2224,
+      "step": 3923
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3527827551148906e-05,
+      "loss": 0.1606,
+      "step": 3924
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3524709403758231e-05,
+      "loss": 0.1611,
+      "step": 3925
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3521590865000817e-05,
+      "loss": 0.1868,
+      "step": 3926
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3518471935222936e-05,
+      "loss": 0.1707,
+      "step": 3927
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3515352614770895e-05,
+      "loss": 0.1748,
+      "step": 3928
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3512232903991051e-05,
+      "loss": 0.1392,
+      "step": 3929
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3509112803229803e-05,
+      "loss": 0.1442,
+      "step": 3930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3505992312833586e-05,
+      "loss": 0.1677,
+      "step": 3931
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.350287143314889e-05,
+      "loss": 0.2012,
+      "step": 3932
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.349975016452224e-05,
+      "loss": 0.1753,
+      "step": 3933
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.349662850730021e-05,
+      "loss": 0.1481,
+      "step": 3934
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3493506461829407e-05,
+      "loss": 0.1716,
+      "step": 3935
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3490384028456492e-05,
+      "loss": 0.2024,
+      "step": 3936
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3487261207528167e-05,
+      "loss": 0.1426,
+      "step": 3937
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3484137999391169e-05,
+      "loss": 0.1589,
+      "step": 3938
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.348101440439229e-05,
+      "loss": 0.1858,
+      "step": 3939
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3477890422878357e-05,
+      "loss": 0.176,
+      "step": 3940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3474766055196238e-05,
+      "loss": 0.1484,
+      "step": 3941
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3471641301692855e-05,
+      "loss": 0.1731,
+      "step": 3942
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.346851616271516e-05,
+      "loss": 0.1727,
+      "step": 3943
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3465390638610154e-05,
+      "loss": 0.2097,
+      "step": 3944
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3462264729724882e-05,
+      "loss": 0.2078,
+      "step": 3945
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.345913843640643e-05,
+      "loss": 0.2031,
+      "step": 3946
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3456011759001926e-05,
+      "loss": 0.1819,
+      "step": 3947
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3452884697858538e-05,
+      "loss": 0.1941,
+      "step": 3948
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3449757253323484e-05,
+      "loss": 0.197,
+      "step": 3949
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3446629425744018e-05,
+      "loss": 0.1702,
+      "step": 3950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.344350121546744e-05,
+      "loss": 0.1665,
+      "step": 3951
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3440372622841091e-05,
+      "loss": 0.1926,
+      "step": 3952
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3437243648212358e-05,
+      "loss": 0.1787,
+      "step": 3953
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.343411429192866e-05,
+      "loss": 0.2393,
+      "step": 3954
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3430984554337472e-05,
+      "loss": 0.1765,
+      "step": 3955
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3427854435786303e-05,
+      "loss": 0.1931,
+      "step": 3956
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3424723936622706e-05,
+      "loss": 0.2163,
+      "step": 3957
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3421593057194273e-05,
+      "loss": 0.1464,
+      "step": 3958
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.341846179784865e-05,
+      "loss": 0.2051,
+      "step": 3959
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.341533015893351e-05,
+      "loss": 0.1528,
+      "step": 3960
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3412198140796577e-05,
+      "loss": 0.1995,
+      "step": 3961
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3409065743785619e-05,
+      "loss": 0.1606,
+      "step": 3962
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3405932968248436e-05,
+      "loss": 0.2031,
+      "step": 3963
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3402799814532881e-05,
+      "loss": 0.1714,
+      "step": 3964
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3399666282986843e-05,
+      "loss": 0.2026,
+      "step": 3965
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3396532373958248e-05,
+      "loss": 0.1948,
+      "step": 3966
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3393398087795083e-05,
+      "loss": 0.2148,
+      "step": 3967
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3390263424845353e-05,
+      "loss": 0.1958,
+      "step": 3968
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3387128385457117e-05,
+      "loss": 0.1897,
+      "step": 3969
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3383992969978481e-05,
+      "loss": 0.1736,
+      "step": 3970
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3380857178757584e-05,
+      "loss": 0.1375,
+      "step": 3971
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3377721012142603e-05,
+      "loss": 0.1538,
+      "step": 3972
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3374584470481772e-05,
+      "loss": 0.1995,
+      "step": 3973
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.337144755412335e-05,
+      "loss": 0.1833,
+      "step": 3974
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3368310263415652e-05,
+      "loss": 0.1726,
+      "step": 3975
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3365172598707023e-05,
+      "loss": 0.1904,
+      "step": 3976
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3362034560345854e-05,
+      "loss": 0.1471,
+      "step": 3977
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3358896148680578e-05,
+      "loss": 0.2139,
+      "step": 3978
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3355757364059671e-05,
+      "loss": 0.2361,
+      "step": 3979
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3352618206831647e-05,
+      "loss": 0.1794,
+      "step": 3980
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3349478677345063e-05,
+      "loss": 0.1725,
+      "step": 3981
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3346338775948518e-05,
+      "loss": 0.176,
+      "step": 3982
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3343198502990649e-05,
+      "loss": 0.1661,
+      "step": 3983
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3340057858820141e-05,
+      "loss": 0.1797,
+      "step": 3984
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3336916843785712e-05,
+      "loss": 0.175,
+      "step": 3985
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3333775458236126e-05,
+      "loss": 0.1644,
+      "step": 3986
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333063370252019e-05,
+      "loss": 0.1758,
+      "step": 3987
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3327491576986749e-05,
+      "loss": 0.1764,
+      "step": 3988
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3324349081984684e-05,
+      "loss": 0.1626,
+      "step": 3989
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.332120621786293e-05,
+      "loss": 0.1853,
+      "step": 3990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331806298497045e-05,
+      "loss": 0.1841,
+      "step": 3991
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3314919383656256e-05,
+      "loss": 0.21,
+      "step": 3992
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3311775414269396e-05,
+      "loss": 0.2451,
+      "step": 3993
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3308631077158967e-05,
+      "loss": 0.1621,
+      "step": 3994
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3305486372674092e-05,
+      "loss": 0.1814,
+      "step": 3995
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3302341301163952e-05,
+      "loss": 0.21,
+      "step": 3996
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3299195862977756e-05,
+      "loss": 0.1565,
+      "step": 3997
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3296050058464761e-05,
+      "loss": 0.1379,
+      "step": 3998
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.329290388797426e-05,
+      "loss": 0.1422,
+      "step": 3999
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3289757351855587e-05,
+      "loss": 0.2048,
+      "step": 4000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3286610450458123e-05,
+      "loss": 0.2004,
+      "step": 4001
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3283463184131283e-05,
+      "loss": 0.2297,
+      "step": 4002
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280315553224521e-05,
+      "loss": 0.1479,
+      "step": 4003
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3277167558087338e-05,
+      "loss": 0.1584,
+      "step": 4004
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3274019199069272e-05,
+      "loss": 0.1782,
+      "step": 4005
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32708704765199e-05,
+      "loss": 0.2349,
+      "step": 4006
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267721390788844e-05,
+      "loss": 0.1802,
+      "step": 4007
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326457194222576e-05,
+      "loss": 0.1464,
+      "step": 4008
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3261422131180348e-05,
+      "loss": 0.1597,
+      "step": 4009
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325827195800235e-05,
+      "loss": 0.1508,
+      "step": 4010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3255121423041544e-05,
+      "loss": 0.1559,
+      "step": 4011
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325197052664775e-05,
+      "loss": 0.1443,
+      "step": 4012
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3248819269170831e-05,
+      "loss": 0.1863,
+      "step": 4013
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3245667650960686e-05,
+      "loss": 0.2014,
+      "step": 4014
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3242515672367254e-05,
+      "loss": 0.1797,
+      "step": 4015
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3239363333740517e-05,
+      "loss": 0.1462,
+      "step": 4016
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3236210635430493e-05,
+      "loss": 0.1528,
+      "step": 4017
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3233057577787248e-05,
+      "loss": 0.1755,
+      "step": 4018
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3229904161160877e-05,
+      "loss": 0.1921,
+      "step": 4019
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3226750385901521e-05,
+      "loss": 0.1906,
+      "step": 4020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3223596252359366e-05,
+      "loss": 0.1801,
+      "step": 4021
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3220441760884621e-05,
+      "loss": 0.1765,
+      "step": 4022
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3217286911827554e-05,
+      "loss": 0.1576,
+      "step": 4023
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3214131705538463e-05,
+      "loss": 0.1351,
+      "step": 4024
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3210976142367684e-05,
+      "loss": 0.1426,
+      "step": 4025
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3207820222665594e-05,
+      "loss": 0.2224,
+      "step": 4026
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3204663946782618e-05,
+      "loss": 0.1887,
+      "step": 4027
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.320150731506921e-05,
+      "loss": 0.2012,
+      "step": 4028
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3198350327875866e-05,
+      "loss": 0.168,
+      "step": 4029
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3195192985553128e-05,
+      "loss": 0.1777,
+      "step": 4030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3192035288451562e-05,
+      "loss": 0.1255,
+      "step": 4031
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3188877236921789e-05,
+      "loss": 0.1539,
+      "step": 4032
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3185718831314466e-05,
+      "loss": 0.1992,
+      "step": 4033
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3182560071980284e-05,
+      "loss": 0.2029,
+      "step": 4034
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.317940095926998e-05,
+      "loss": 0.1848,
+      "step": 4035
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3176241493534321e-05,
+      "loss": 0.2078,
+      "step": 4036
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3173081675124122e-05,
+      "loss": 0.1584,
+      "step": 4037
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3169921504390236e-05,
+      "loss": 0.2039,
+      "step": 4038
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316676098168355e-05,
+      "loss": 0.1443,
+      "step": 4039
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3163600107354992e-05,
+      "loss": 0.1453,
+      "step": 4040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3160438881755536e-05,
+      "loss": 0.1667,
+      "step": 4041
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3157277305236185e-05,
+      "loss": 0.1683,
+      "step": 4042
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3154115378147984e-05,
+      "loss": 0.1458,
+      "step": 4043
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3150953100842023e-05,
+      "loss": 0.1735,
+      "step": 4044
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3147790473669422e-05,
+      "loss": 0.1729,
+      "step": 4045
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3144627496981344e-05,
+      "loss": 0.1658,
+      "step": 4046
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3141464171128993e-05,
+      "loss": 0.198,
+      "step": 4047
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3138300496463606e-05,
+      "loss": 0.1729,
+      "step": 4048
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3135136473336467e-05,
+      "loss": 0.162,
+      "step": 4049
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3131972102098889e-05,
+      "loss": 0.1921,
+      "step": 4050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.312880738310223e-05,
+      "loss": 0.1747,
+      "step": 4051
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3125642316697889e-05,
+      "loss": 0.1592,
+      "step": 4052
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3122476903237296e-05,
+      "loss": 0.2009,
+      "step": 4053
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3119311143071922e-05,
+      "loss": 0.2043,
+      "step": 4054
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3116145036553283e-05,
+      "loss": 0.1768,
+      "step": 4055
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3112978584032923e-05,
+      "loss": 0.1411,
+      "step": 4056
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3109811785862428e-05,
+      "loss": 0.1759,
+      "step": 4057
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.310664464239343e-05,
+      "loss": 0.1898,
+      "step": 4058
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3103477153977593e-05,
+      "loss": 0.2053,
+      "step": 4059
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3100309320966613e-05,
+      "loss": 0.1616,
+      "step": 4060
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3097141143712237e-05,
+      "loss": 0.1462,
+      "step": 4061
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3093972622566245e-05,
+      "loss": 0.1592,
+      "step": 4062
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3090803757880448e-05,
+      "loss": 0.1953,
+      "step": 4063
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3087634550006707e-05,
+      "loss": 0.1753,
+      "step": 4064
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3084464999296916e-05,
+      "loss": 0.1466,
+      "step": 4065
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3081295106102999e-05,
+      "loss": 0.1958,
+      "step": 4066
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3078124870776934e-05,
+      "loss": 0.1636,
+      "step": 4067
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3074954293670726e-05,
+      "loss": 0.1422,
+      "step": 4068
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3071783375136417e-05,
+      "loss": 0.1475,
+      "step": 4069
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3068612115526099e-05,
+      "loss": 0.1875,
+      "step": 4070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3065440515191883e-05,
+      "loss": 0.2075,
+      "step": 4071
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3062268574485937e-05,
+      "loss": 0.1753,
+      "step": 4072
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3059096293760454e-05,
+      "loss": 0.2236,
+      "step": 4073
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3055923673367667e-05,
+      "loss": 0.187,
+      "step": 4074
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3052750713659852e-05,
+      "loss": 0.1768,
+      "step": 4075
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3049577414989318e-05,
+      "loss": 0.1936,
+      "step": 4076
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3046403777708409e-05,
+      "loss": 0.1853,
+      "step": 4077
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3043229802169519e-05,
+      "loss": 0.1779,
+      "step": 4078
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3040055488725064e-05,
+      "loss": 0.1654,
+      "step": 4079
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3036880837727503e-05,
+      "loss": 0.2019,
+      "step": 4080
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3033705849529342e-05,
+      "loss": 0.1519,
+      "step": 4081
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.303053052448311e-05,
+      "loss": 0.1648,
+      "step": 4082
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3027354862941379e-05,
+      "loss": 0.2014,
+      "step": 4083
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3024178865256764e-05,
+      "loss": 0.2156,
+      "step": 4084
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3021002531781909e-05,
+      "loss": 0.1381,
+      "step": 4085
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.30178258628695e-05,
+      "loss": 0.1902,
+      "step": 4086
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301464885887226e-05,
+      "loss": 0.1799,
+      "step": 4087
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3011471520142946e-05,
+      "loss": 0.1581,
+      "step": 4088
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3008293847034355e-05,
+      "loss": 0.1577,
+      "step": 4089
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.300511583989932e-05,
+      "loss": 0.201,
+      "step": 4090
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3001937499090716e-05,
+      "loss": 0.1919,
+      "step": 4091
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2998758824961442e-05,
+      "loss": 0.1711,
+      "step": 4092
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.299557981786445e-05,
+      "loss": 0.1787,
+      "step": 4093
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2992400478152717e-05,
+      "loss": 0.1639,
+      "step": 4094
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2989220806179268e-05,
+      "loss": 0.165,
+      "step": 4095
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2986040802297153e-05,
+      "loss": 0.1667,
+      "step": 4096
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2982860466859464e-05,
+      "loss": 0.1674,
+      "step": 4097
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2979679800219333e-05,
+      "loss": 0.1958,
+      "step": 4098
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2976498802729927e-05,
+      "loss": 0.1412,
+      "step": 4099
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.297331747474444e-05,
+      "loss": 0.178,
+      "step": 4100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2970135816616123e-05,
+      "loss": 0.2144,
+      "step": 4101
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2966953828698243e-05,
+      "loss": 0.2107,
+      "step": 4102
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2963771511344119e-05,
+      "loss": 0.2024,
+      "step": 4103
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2960588864907096e-05,
+      "loss": 0.2295,
+      "step": 4104
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.295740588974056e-05,
+      "loss": 0.1814,
+      "step": 4105
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2954222586197935e-05,
+      "loss": 0.1802,
+      "step": 4106
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.295103895463268e-05,
+      "loss": 0.1698,
+      "step": 4107
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2947854995398286e-05,
+      "loss": 0.1794,
+      "step": 4108
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294467070884829e-05,
+      "loss": 0.1726,
+      "step": 4109
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2941486095336258e-05,
+      "loss": 0.1675,
+      "step": 4110
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.293830115521579e-05,
+      "loss": 0.1677,
+      "step": 4111
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2935115888840532e-05,
+      "loss": 0.1753,
+      "step": 4112
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2931930296564159e-05,
+      "loss": 0.1687,
+      "step": 4113
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2928744378740382e-05,
+      "loss": 0.1655,
+      "step": 4114
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2925558135722951e-05,
+      "loss": 0.1843,
+      "step": 4115
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2922371567865652e-05,
+      "loss": 0.197,
+      "step": 4116
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2919184675522305e-05,
+      "loss": 0.1746,
+      "step": 4117
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2915997459046768e-05,
+      "loss": 0.1543,
+      "step": 4118
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2912809918792934e-05,
+      "loss": 0.1636,
+      "step": 4119
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.290962205511473e-05,
+      "loss": 0.1523,
+      "step": 4120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2906433868366123e-05,
+      "loss": 0.1674,
+      "step": 4121
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2903245358901118e-05,
+      "loss": 0.1848,
+      "step": 4122
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2900056527073743e-05,
+      "loss": 0.1628,
+      "step": 4123
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2896867373238075e-05,
+      "loss": 0.1809,
+      "step": 4124
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2893677897748222e-05,
+      "loss": 0.1917,
+      "step": 4125
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.289048810095833e-05,
+      "loss": 0.1846,
+      "step": 4126
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2887297983222577e-05,
+      "loss": 0.1677,
+      "step": 4127
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2884107544895175e-05,
+      "loss": 0.1731,
+      "step": 4128
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2880916786330381e-05,
+      "loss": 0.1652,
+      "step": 4129
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2877725707882478e-05,
+      "loss": 0.2075,
+      "step": 4130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2874534309905789e-05,
+      "loss": 0.1958,
+      "step": 4131
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.287134259275467e-05,
+      "loss": 0.1588,
+      "step": 4132
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2868150556783514e-05,
+      "loss": 0.2085,
+      "step": 4133
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2864958202346748e-05,
+      "loss": 0.1818,
+      "step": 4134
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2861765529798843e-05,
+      "loss": 0.2078,
+      "step": 4135
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2858572539494289e-05,
+      "loss": 0.1364,
+      "step": 4136
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2855379231787623e-05,
+      "loss": 0.1785,
+      "step": 4137
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2852185607033418e-05,
+      "loss": 0.1678,
+      "step": 4138
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2848991665586276e-05,
+      "loss": 0.1597,
+      "step": 4139
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2845797407800834e-05,
+      "loss": 0.189,
+      "step": 4140
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2842602834031773e-05,
+      "loss": 0.1746,
+      "step": 4141
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.28394079446338e-05,
+      "loss": 0.1654,
+      "step": 4142
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2836212739961657e-05,
+      "loss": 0.1917,
+      "step": 4143
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.283301722037013e-05,
+      "loss": 0.1885,
+      "step": 4144
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.282982138621403e-05,
+      "loss": 0.2095,
+      "step": 4145
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2826625237848207e-05,
+      "loss": 0.1853,
+      "step": 4146
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2823428775627551e-05,
+      "loss": 0.1462,
+      "step": 4147
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2820231999906974e-05,
+      "loss": 0.1973,
+      "step": 4148
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2817034911041436e-05,
+      "loss": 0.2122,
+      "step": 4149
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2813837509385926e-05,
+      "loss": 0.1667,
+      "step": 4150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2810639795295466e-05,
+      "loss": 0.1802,
+      "step": 4151
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2807441769125116e-05,
+      "loss": 0.1633,
+      "step": 4152
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2804243431229968e-05,
+      "loss": 0.1963,
+      "step": 4153
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280104478196515e-05,
+      "loss": 0.187,
+      "step": 4154
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2797845821685828e-05,
+      "loss": 0.1686,
+      "step": 4155
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2794646550747196e-05,
+      "loss": 0.1821,
+      "step": 4156
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2791446969504486e-05,
+      "loss": 0.2051,
+      "step": 4157
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2788247078312966e-05,
+      "loss": 0.1646,
+      "step": 4158
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2785046877527934e-05,
+      "loss": 0.1943,
+      "step": 4159
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2781846367504725e-05,
+      "loss": 0.1868,
+      "step": 4160
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277864554859871e-05,
+      "loss": 0.1987,
+      "step": 4161
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.277544442116529e-05,
+      "loss": 0.1741,
+      "step": 4162
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2772242985559906e-05,
+      "loss": 0.2024,
+      "step": 4163
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2769041242138027e-05,
+      "loss": 0.2036,
+      "step": 4164
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2765839191255163e-05,
+      "loss": 0.1921,
+      "step": 4165
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.276263683326685e-05,
+      "loss": 0.1641,
+      "step": 4166
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2759434168528663e-05,
+      "loss": 0.1595,
+      "step": 4167
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2756231197396214e-05,
+      "loss": 0.2239,
+      "step": 4168
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.275302792022514e-05,
+      "loss": 0.1887,
+      "step": 4169
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2749824337371124e-05,
+      "loss": 0.1519,
+      "step": 4170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.274662044918987e-05,
+      "loss": 0.1605,
+      "step": 4171
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2743416256037126e-05,
+      "loss": 0.2029,
+      "step": 4172
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.274021175826867e-05,
+      "loss": 0.1542,
+      "step": 4173
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2737006956240312e-05,
+      "loss": 0.1921,
+      "step": 4174
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2733801850307898e-05,
+      "loss": 0.1956,
+      "step": 4175
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2730596440827312e-05,
+      "loss": 0.2039,
+      "step": 4176
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272739072815446e-05,
+      "loss": 0.1454,
+      "step": 4177
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2724184712645296e-05,
+      "loss": 0.1775,
+      "step": 4178
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2720978394655793e-05,
+      "loss": 0.1567,
+      "step": 4179
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2717771774541973e-05,
+      "loss": 0.1694,
+      "step": 4180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.271456485265988e-05,
+      "loss": 0.1509,
+      "step": 4181
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2711357629365592e-05,
+      "loss": 0.1635,
+      "step": 4182
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2708150105015229e-05,
+      "loss": 0.1621,
+      "step": 4183
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2704942279964936e-05,
+      "loss": 0.1608,
+      "step": 4184
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2701734154570894e-05,
+      "loss": 0.1548,
+      "step": 4185
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2698525729189321e-05,
+      "loss": 0.1536,
+      "step": 4186
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2695317004176464e-05,
+      "loss": 0.1782,
+      "step": 4187
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2692107979888599e-05,
+      "loss": 0.1702,
+      "step": 4188
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2688898656682048e-05,
+      "loss": 0.1652,
+      "step": 4189
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2685689034913156e-05,
+      "loss": 0.1702,
+      "step": 4190
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.26824791149383e-05,
+      "loss": 0.189,
+      "step": 4191
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.26792688971139e-05,
+      "loss": 0.1855,
+      "step": 4192
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2676058381796402e-05,
+      "loss": 0.1689,
+      "step": 4193
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2672847569342284e-05,
+      "loss": 0.1713,
+      "step": 4194
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.266963646010806e-05,
+      "loss": 0.2148,
+      "step": 4195
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2666425054450276e-05,
+      "loss": 0.1997,
+      "step": 4196
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2663213352725507e-05,
+      "loss": 0.188,
+      "step": 4197
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2660001355290374e-05,
+      "loss": 0.1982,
+      "step": 4198
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2656789062501515e-05,
+      "loss": 0.1721,
+      "step": 4199
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265357647471561e-05,
+      "loss": 0.198,
+      "step": 4200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2650363592289368e-05,
+      "loss": 0.1975,
+      "step": 4201
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2647150415579532e-05,
+      "loss": 0.1868,
+      "step": 4202
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2643936944942878e-05,
+      "loss": 0.1819,
+      "step": 4203
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2640723180736216e-05,
+      "loss": 0.1766,
+      "step": 4204
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2637509123316383e-05,
+      "loss": 0.1838,
+      "step": 4205
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2634294773040256e-05,
+      "loss": 0.1517,
+      "step": 4206
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2631080130264742e-05,
+      "loss": 0.1763,
+      "step": 4207
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2627865195346771e-05,
+      "loss": 0.1616,
+      "step": 4208
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2624649968643323e-05,
+      "loss": 0.1742,
+      "step": 4209
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.26214344505114e-05,
+      "loss": 0.1818,
+      "step": 4210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2618218641308033e-05,
+      "loss": 0.139,
+      "step": 4211
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2615002541390293e-05,
+      "loss": 0.2,
+      "step": 4212
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2611786151115282e-05,
+      "loss": 0.209,
+      "step": 4213
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2608569470840128e-05,
+      "loss": 0.1641,
+      "step": 4214
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2605352500921998e-05,
+      "loss": 0.1941,
+      "step": 4215
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2602135241718094e-05,
+      "loss": 0.1501,
+      "step": 4216
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2598917693585635e-05,
+      "loss": 0.1865,
+      "step": 4217
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2595699856881888e-05,
+      "loss": 0.1707,
+      "step": 4218
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2592481731964146e-05,
+      "loss": 0.166,
+      "step": 4219
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2589263319189731e-05,
+      "loss": 0.1631,
+      "step": 4220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2586044618916002e-05,
+      "loss": 0.1968,
+      "step": 4221
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2582825631500351e-05,
+      "loss": 0.1875,
+      "step": 4222
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2579606357300193e-05,
+      "loss": 0.1782,
+      "step": 4223
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2576386796672985e-05,
+      "loss": 0.1606,
+      "step": 4224
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257316694997621e-05,
+      "loss": 0.156,
+      "step": 4225
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2569946817567384e-05,
+      "loss": 0.1799,
+      "step": 4226
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2566726399804056e-05,
+      "loss": 0.1497,
+      "step": 4227
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2563505697043805e-05,
+      "loss": 0.1628,
+      "step": 4228
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.256028470964424e-05,
+      "loss": 0.1824,
+      "step": 4229
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2557063437963011e-05,
+      "loss": 0.2031,
+      "step": 4230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2553841882357783e-05,
+      "loss": 0.1971,
+      "step": 4231
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255062004318627e-05,
+      "loss": 0.1548,
+      "step": 4232
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2547397920806203e-05,
+      "loss": 0.1628,
+      "step": 4233
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2544175515575356e-05,
+      "loss": 0.2017,
+      "step": 4234
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.254095282785153e-05,
+      "loss": 0.1521,
+      "step": 4235
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2537729857992549e-05,
+      "loss": 0.1472,
+      "step": 4236
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2534506606356282e-05,
+      "loss": 0.1403,
+      "step": 4237
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2531283073300627e-05,
+      "loss": 0.1434,
+      "step": 4238
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2528059259183499e-05,
+      "loss": 0.1902,
+      "step": 4239
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2524835164362866e-05,
+      "loss": 0.1846,
+      "step": 4240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2521610789196707e-05,
+      "loss": 0.1819,
+      "step": 4241
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2518386134043046e-05,
+      "loss": 0.2021,
+      "step": 4242
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.251516119925993e-05,
+      "loss": 0.177,
+      "step": 4243
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2511935985205446e-05,
+      "loss": 0.1821,
+      "step": 4244
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2508710492237698e-05,
+      "loss": 0.1699,
+      "step": 4245
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2505484720714838e-05,
+      "loss": 0.1689,
+      "step": 4246
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2502258670995036e-05,
+      "loss": 0.1899,
+      "step": 4247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2499032343436492e-05,
+      "loss": 0.196,
+      "step": 4248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.249580573839745e-05,
+      "loss": 0.1807,
+      "step": 4249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2492578856236173e-05,
+      "loss": 0.189,
+      "step": 4250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2489351697310957e-05,
+      "loss": 0.1904,
+      "step": 4251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2486124261980134e-05,
+      "loss": 0.179,
+      "step": 4252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2482896550602057e-05,
+      "loss": 0.1517,
+      "step": 4253
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2479668563535124e-05,
+      "loss": 0.1799,
+      "step": 4254
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2476440301137749e-05,
+      "loss": 0.1992,
+      "step": 4255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.247321176376838e-05,
+      "loss": 0.1627,
+      "step": 4256
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2469982951785507e-05,
+      "loss": 0.188,
+      "step": 4257
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2466753865547636e-05,
+      "loss": 0.1555,
+      "step": 4258
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2463524505413308e-05,
+      "loss": 0.1985,
+      "step": 4259
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2460294871741102e-05,
+      "loss": 0.1733,
+      "step": 4260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2457064964889614e-05,
+      "loss": 0.1664,
+      "step": 4261
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2453834785217478e-05,
+      "loss": 0.198,
+      "step": 4262
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245060433308336e-05,
+      "loss": 0.228,
+      "step": 4263
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2447373608845956e-05,
+      "loss": 0.1785,
+      "step": 4264
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2444142612863985e-05,
+      "loss": 0.197,
+      "step": 4265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2440911345496204e-05,
+      "loss": 0.1646,
+      "step": 4266
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2437679807101399e-05,
+      "loss": 0.2,
+      "step": 4267
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2434447998038379e-05,
+      "loss": 0.1797,
+      "step": 4268
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2431215918665995e-05,
+      "loss": 0.1863,
+      "step": 4269
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2427983569343118e-05,
+      "loss": 0.1663,
+      "step": 4270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.242475095042865e-05,
+      "loss": 0.2031,
+      "step": 4271
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2421518062281532e-05,
+      "loss": 0.2019,
+      "step": 4272
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2418284905260722e-05,
+      "loss": 0.1605,
+      "step": 4273
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2415051479725218e-05,
+      "loss": 0.1846,
+      "step": 4274
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241181778603404e-05,
+      "loss": 0.1716,
+      "step": 4275
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2408583824546247e-05,
+      "loss": 0.2046,
+      "step": 4276
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2405349595620921e-05,
+      "loss": 0.1509,
+      "step": 4277
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2402115099617174e-05,
+      "loss": 0.2231,
+      "step": 4278
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2398880336894146e-05,
+      "loss": 0.1892,
+      "step": 4279
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2395645307811014e-05,
+      "loss": 0.1566,
+      "step": 4280
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2392410012726981e-05,
+      "loss": 0.2173,
+      "step": 4281
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.238917445200127e-05,
+      "loss": 0.1807,
+      "step": 4282
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2385938625993154e-05,
+      "loss": 0.1577,
+      "step": 4283
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2382702535061917e-05,
+      "loss": 0.1941,
+      "step": 4284
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2379466179566878e-05,
+      "loss": 0.1737,
+      "step": 4285
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2376229559867388e-05,
+      "loss": 0.177,
+      "step": 4286
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2372992676322826e-05,
+      "loss": 0.1404,
+      "step": 4287
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2369755529292598e-05,
+      "loss": 0.1971,
+      "step": 4288
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2366518119136143e-05,
+      "loss": 0.2029,
+      "step": 4289
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2363280446212929e-05,
+      "loss": 0.1636,
+      "step": 4290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2360042510882449e-05,
+      "loss": 0.2332,
+      "step": 4291
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2356804313504227e-05,
+      "loss": 0.1794,
+      "step": 4292
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2353565854437818e-05,
+      "loss": 0.179,
+      "step": 4293
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2350327134042806e-05,
+      "loss": 0.1697,
+      "step": 4294
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2347088152678803e-05,
+      "loss": 0.1541,
+      "step": 4295
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2343848910705449e-05,
+      "loss": 0.1707,
+      "step": 4296
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2340609408482411e-05,
+      "loss": 0.1829,
+      "step": 4297
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2337369646369392e-05,
+      "loss": 0.175,
+      "step": 4298
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2334129624726118e-05,
+      "loss": 0.1934,
+      "step": 4299
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2330889343912347e-05,
+      "loss": 0.1865,
+      "step": 4300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2327648804287862e-05,
+      "loss": 0.187,
+      "step": 4301
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2324408006212475e-05,
+      "loss": 0.2024,
+      "step": 4302
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2321166950046033e-05,
+      "loss": 0.1689,
+      "step": 4303
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2317925636148404e-05,
+      "loss": 0.1611,
+      "step": 4304
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.231468406487949e-05,
+      "loss": 0.1974,
+      "step": 4305
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2311442236599218e-05,
+      "loss": 0.158,
+      "step": 4306
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2308200151667545e-05,
+      "loss": 0.2051,
+      "step": 4307
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2304957810444455e-05,
+      "loss": 0.1794,
+      "step": 4308
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.230171521328997e-05,
+      "loss": 0.198,
+      "step": 4309
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229847236056412e-05,
+      "loss": 0.1682,
+      "step": 4310
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2295229252626985e-05,
+      "loss": 0.1497,
+      "step": 4311
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229198588983866e-05,
+      "loss": 0.1997,
+      "step": 4312
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2288742272559273e-05,
+      "loss": 0.1665,
+      "step": 4313
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2285498401148984e-05,
+      "loss": 0.1953,
+      "step": 4314
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.228225427596797e-05,
+      "loss": 0.1835,
+      "step": 4315
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2279009897376444e-05,
+      "loss": 0.1676,
+      "step": 4316
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2275765265734649e-05,
+      "loss": 0.1721,
+      "step": 4317
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2272520381402856e-05,
+      "loss": 0.1863,
+      "step": 4318
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2269275244741352e-05,
+      "loss": 0.1603,
+      "step": 4319
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.226602985611047e-05,
+      "loss": 0.1605,
+      "step": 4320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2262784215870562e-05,
+      "loss": 0.1899,
+      "step": 4321
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2259538324382001e-05,
+      "loss": 0.1816,
+      "step": 4322
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2256292182005201e-05,
+      "loss": 0.1479,
+      "step": 4323
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2253045789100597e-05,
+      "loss": 0.1875,
+      "step": 4324
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2249799146028651e-05,
+      "loss": 0.1611,
+      "step": 4325
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224655225314986e-05,
+      "loss": 0.1887,
+      "step": 4326
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2243305110824738e-05,
+      "loss": 0.1702,
+      "step": 4327
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224005771941383e-05,
+      "loss": 0.1541,
+      "step": 4328
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2236810079277717e-05,
+      "loss": 0.2087,
+      "step": 4329
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2233562190776996e-05,
+      "loss": 0.1617,
+      "step": 4330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2230314054272302e-05,
+      "loss": 0.1907,
+      "step": 4331
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2227065670124288e-05,
+      "loss": 0.1746,
+      "step": 4332
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2223817038693639e-05,
+      "loss": 0.1423,
+      "step": 4333
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2220568160341072e-05,
+      "loss": 0.1714,
+      "step": 4334
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2217319035427323e-05,
+      "loss": 0.1921,
+      "step": 4335
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2214069664313159e-05,
+      "loss": 0.2109,
+      "step": 4336
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2210820047359377e-05,
+      "loss": 0.1704,
+      "step": 4337
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2207570184926797e-05,
+      "loss": 0.1643,
+      "step": 4338
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2204320077376266e-05,
+      "loss": 0.157,
+      "step": 4339
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2201069725068666e-05,
+      "loss": 0.2041,
+      "step": 4340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2197819128364894e-05,
+      "loss": 0.1821,
+      "step": 4341
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2194568287625886e-05,
+      "loss": 0.1658,
+      "step": 4342
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2191317203212595e-05,
+      "loss": 0.1797,
+      "step": 4343
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.218806587548601e-05,
+      "loss": 0.1895,
+      "step": 4344
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2184814304807138e-05,
+      "loss": 0.1555,
+      "step": 4345
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2181562491537022e-05,
+      "loss": 0.1512,
+      "step": 4346
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2178310436036726e-05,
+      "loss": 0.196,
+      "step": 4347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2175058138667343e-05,
+      "loss": 0.1401,
+      "step": 4348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2171805599789993e-05,
+      "loss": 0.2151,
+      "step": 4349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216855281976582e-05,
+      "loss": 0.1776,
+      "step": 4350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2165299798955998e-05,
+      "loss": 0.1475,
+      "step": 4351
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2162046537721728e-05,
+      "loss": 0.1646,
+      "step": 4352
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2158793036424235e-05,
+      "loss": 0.1831,
+      "step": 4353
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2155539295424774e-05,
+      "loss": 0.1711,
+      "step": 4354
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2152285315084623e-05,
+      "loss": 0.1488,
+      "step": 4355
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2149031095765087e-05,
+      "loss": 0.1707,
+      "step": 4356
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2145776637827503e-05,
+      "loss": 0.1938,
+      "step": 4357
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2142521941633226e-05,
+      "loss": 0.2146,
+      "step": 4358
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2139267007543642e-05,
+      "loss": 0.1812,
+      "step": 4359
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2136011835920167e-05,
+      "loss": 0.1626,
+      "step": 4360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2132756427124237e-05,
+      "loss": 0.1797,
+      "step": 4361
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2129500781517317e-05,
+      "loss": 0.2219,
+      "step": 4362
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2126244899460896e-05,
+      "loss": 0.1707,
+      "step": 4363
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2122988781316499e-05,
+      "loss": 0.1406,
+      "step": 4364
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.211973242744566e-05,
+      "loss": 0.1787,
+      "step": 4365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2116475838209956e-05,
+      "loss": 0.1626,
+      "step": 4366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2113219013970978e-05,
+      "loss": 0.2122,
+      "step": 4367
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2109961955090354e-05,
+      "loss": 0.1917,
+      "step": 4368
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2106704661929727e-05,
+      "loss": 0.1416,
+      "step": 4369
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.210344713485077e-05,
+      "loss": 0.218,
+      "step": 4370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2100189374215188e-05,
+      "loss": 0.2097,
+      "step": 4371
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2096931380384705e-05,
+      "loss": 0.2085,
+      "step": 4372
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2093673153721071e-05,
+      "loss": 0.1914,
+      "step": 4373
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2090414694586065e-05,
+      "loss": 0.1051,
+      "step": 4374
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2087156003341494e-05,
+      "loss": 0.1636,
+      "step": 4375
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2083897080349181e-05,
+      "loss": 0.1582,
+      "step": 4376
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2080637925970985e-05,
+      "loss": 0.199,
+      "step": 4377
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2077378540568789e-05,
+      "loss": 0.2263,
+      "step": 4378
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2074118924504493e-05,
+      "loss": 0.1797,
+      "step": 4379
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2070859078140034e-05,
+      "loss": 0.1826,
+      "step": 4380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2067599001837369e-05,
+      "loss": 0.1616,
+      "step": 4381
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.206433869595848e-05,
+      "loss": 0.1422,
+      "step": 4382
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2061078160865379e-05,
+      "loss": 0.1782,
+      "step": 4383
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2057817396920091e-05,
+      "loss": 0.1741,
+      "step": 4384
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2054556404484688e-05,
+      "loss": 0.1609,
+      "step": 4385
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2051295183921245e-05,
+      "loss": 0.1758,
+      "step": 4386
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2048033735591878e-05,
+      "loss": 0.167,
+      "step": 4387
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2044772059858718e-05,
+      "loss": 0.1641,
+      "step": 4388
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2041510157083931e-05,
+      "loss": 0.2004,
+      "step": 4389
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2038248027629695e-05,
+      "loss": 0.1694,
+      "step": 4390
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2034985671858233e-05,
+      "loss": 0.1616,
+      "step": 4391
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2031723090131771e-05,
+      "loss": 0.1772,
+      "step": 4392
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.202846028281257e-05,
+      "loss": 0.1613,
+      "step": 4393
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2025197250262926e-05,
+      "loss": 0.1506,
+      "step": 4394
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2021933992845142e-05,
+      "loss": 0.1716,
+      "step": 4395
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2018670510921558e-05,
+      "loss": 0.1982,
+      "step": 4396
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2015406804854535e-05,
+      "loss": 0.1367,
+      "step": 4397
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2012142875006456e-05,
+      "loss": 0.1807,
+      "step": 4398
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2008878721739735e-05,
+      "loss": 0.2014,
+      "step": 4399
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2005614345416808e-05,
+      "loss": 0.2227,
+      "step": 4400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2002349746400135e-05,
+      "loss": 0.2324,
+      "step": 4401
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.19990849250522e-05,
+      "loss": 0.1888,
+      "step": 4402
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1995819881735516e-05,
+      "loss": 0.1672,
+      "step": 4403
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1992554616812613e-05,
+      "loss": 0.1738,
+      "step": 4404
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1989289130646056e-05,
+      "loss": 0.1995,
+      "step": 4405
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1986023423598425e-05,
+      "loss": 0.1541,
+      "step": 4406
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1982757496032329e-05,
+      "loss": 0.1951,
+      "step": 4407
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1979491348310402e-05,
+      "loss": 0.1663,
+      "step": 4408
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.19762249807953e-05,
+      "loss": 0.1489,
+      "step": 4409
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1972958393849706e-05,
+      "loss": 0.1978,
+      "step": 4410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1969691587836322e-05,
+      "loss": 0.1943,
+      "step": 4411
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1966424563117886e-05,
+      "loss": 0.151,
+      "step": 4412
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1963157320057145e-05,
+      "loss": 0.188,
+      "step": 4413
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1959889859016885e-05,
+      "loss": 0.189,
+      "step": 4414
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1956622180359905e-05,
+      "loss": 0.1486,
+      "step": 4415
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1953354284449028e-05,
+      "loss": 0.189,
+      "step": 4416
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1950086171647115e-05,
+      "loss": 0.1653,
+      "step": 4417
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1946817842317036e-05,
+      "loss": 0.179,
+      "step": 4418
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1943549296821686e-05,
+      "loss": 0.1403,
+      "step": 4419
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1940280535523998e-05,
+      "loss": 0.1599,
+      "step": 4420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1937011558786915e-05,
+      "loss": 0.1636,
+      "step": 4421
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1933742366973406e-05,
+      "loss": 0.1763,
+      "step": 4422
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1930472960446473e-05,
+      "loss": 0.1855,
+      "step": 4423
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1927203339569128e-05,
+      "loss": 0.1636,
+      "step": 4424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1923933504704417e-05,
+      "loss": 0.1528,
+      "step": 4425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.192066345621541e-05,
+      "loss": 0.2188,
+      "step": 4426
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1917393194465188e-05,
+      "loss": 0.1799,
+      "step": 4427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1914122719816874e-05,
+      "loss": 0.1874,
+      "step": 4428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1910852032633605e-05,
+      "loss": 0.1738,
+      "step": 4429
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1907581133278536e-05,
+      "loss": 0.1707,
+      "step": 4430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1904310022114857e-05,
+      "loss": 0.1924,
+      "step": 4431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1901038699505779e-05,
+      "loss": 0.1593,
+      "step": 4432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1897767165814527e-05,
+      "loss": 0.2012,
+      "step": 4433
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1894495421404364e-05,
+      "loss": 0.1658,
+      "step": 4434
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1891223466638557e-05,
+      "loss": 0.1873,
+      "step": 4435
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1887951301880421e-05,
+      "loss": 0.2212,
+      "step": 4436
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1884678927493276e-05,
+      "loss": 0.1658,
+      "step": 4437
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1881406343840468e-05,
+      "loss": 0.1799,
+      "step": 4438
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.187813355128537e-05,
+      "loss": 0.1714,
+      "step": 4439
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1874860550191385e-05,
+      "loss": 0.1589,
+      "step": 4440
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1871587340921918e-05,
+      "loss": 0.2148,
+      "step": 4441
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1868313923840423e-05,
+      "loss": 0.1809,
+      "step": 4442
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1865040299310356e-05,
+      "loss": 0.1868,
+      "step": 4443
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1861766467695204e-05,
+      "loss": 0.1591,
+      "step": 4444
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1858492429358483e-05,
+      "loss": 0.2197,
+      "step": 4445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1855218184663725e-05,
+      "loss": 0.1843,
+      "step": 4446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1851943733974484e-05,
+      "loss": 0.1628,
+      "step": 4447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1848669077654342e-05,
+      "loss": 0.1561,
+      "step": 4448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.18453942160669e-05,
+      "loss": 0.165,
+      "step": 4449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1842119149575781e-05,
+      "loss": 0.1406,
+      "step": 4450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1838843878544635e-05,
+      "loss": 0.1635,
+      "step": 4451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1835568403337131e-05,
+      "loss": 0.1716,
+      "step": 4452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.183229272431696e-05,
+      "loss": 0.2031,
+      "step": 4453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1829016841847845e-05,
+      "loss": 0.1768,
+      "step": 4454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1825740756293515e-05,
+      "loss": 0.1879,
+      "step": 4455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1822464468017735e-05,
+      "loss": 0.1748,
+      "step": 4456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1819187977384292e-05,
+      "loss": 0.1362,
+      "step": 4457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1815911284756983e-05,
+      "loss": 0.1786,
+      "step": 4458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1812634390499645e-05,
+      "loss": 0.176,
+      "step": 4459
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1809357294976122e-05,
+      "loss": 0.1506,
+      "step": 4460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.180607999855029e-05,
+      "loss": 0.1479,
+      "step": 4461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1802802501586044e-05,
+      "loss": 0.1675,
+      "step": 4462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1799524804447302e-05,
+      "loss": 0.2209,
+      "step": 4463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1796246907498001e-05,
+      "loss": 0.1357,
+      "step": 4464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1792968811102106e-05,
+      "loss": 0.1794,
+      "step": 4465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.17896905156236e-05,
+      "loss": 0.1704,
+      "step": 4466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1786412021426485e-05,
+      "loss": 0.1802,
+      "step": 4467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1783133328874797e-05,
+      "loss": 0.1714,
+      "step": 4468
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1779854438332584e-05,
+      "loss": 0.1899,
+      "step": 4469
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177657535016391e-05,
+      "loss": 0.1699,
+      "step": 4470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177329606473288e-05,
+      "loss": 0.1699,
+      "step": 4471
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1770016582403605e-05,
+      "loss": 0.1442,
+      "step": 4472
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1766736903540222e-05,
+      "loss": 0.1736,
+      "step": 4473
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1763457028506893e-05,
+      "loss": 0.144,
+      "step": 4474
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1760176957667799e-05,
+      "loss": 0.1702,
+      "step": 4475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1756896691387141e-05,
+      "loss": 0.1575,
+      "step": 4476
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175361623002915e-05,
+      "loss": 0.1853,
+      "step": 4477
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1750335573958066e-05,
+      "loss": 0.1597,
+      "step": 4478
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1747054723538158e-05,
+      "loss": 0.1493,
+      "step": 4479
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1743773679133721e-05,
+      "loss": 0.181,
+      "step": 4480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1740492441109059e-05,
+      "loss": 0.1909,
+      "step": 4481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1737211009828513e-05,
+      "loss": 0.1589,
+      "step": 4482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173392938565643e-05,
+      "loss": 0.1661,
+      "step": 4483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.173064756895719e-05,
+      "loss": 0.1604,
+      "step": 4484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.172736556009519e-05,
+      "loss": 0.2329,
+      "step": 4485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1724083359434845e-05,
+      "loss": 0.1853,
+      "step": 4486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1720800967340597e-05,
+      "loss": 0.1805,
+      "step": 4487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.171751838417691e-05,
+      "loss": 0.15,
+      "step": 4488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1714235610308261e-05,
+      "loss": 0.1992,
+      "step": 4489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1710952646099157e-05,
+      "loss": 0.1936,
+      "step": 4490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.170766949191412e-05,
+      "loss": 0.1895,
+      "step": 4491
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.17043861481177e-05,
+      "loss": 0.1746,
+      "step": 4492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1701102615074453e-05,
+      "loss": 0.1807,
+      "step": 4493
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.169781889314898e-05,
+      "loss": 0.1938,
+      "step": 4494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1694534982705883e-05,
+      "loss": 0.158,
+      "step": 4495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1691250884109791e-05,
+      "loss": 0.1848,
+      "step": 4496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1687966597725358e-05,
+      "loss": 0.1816,
+      "step": 4497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.168468212391725e-05,
+      "loss": 0.1738,
+      "step": 4498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1681397463050164e-05,
+      "loss": 0.2073,
+      "step": 4499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.167811261548881e-05,
+      "loss": 0.1748,
+      "step": 4500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1674827581597925e-05,
+      "loss": 0.1537,
+      "step": 4501
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.167154236174226e-05,
+      "loss": 0.1691,
+      "step": 4502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1668256956286595e-05,
+      "loss": 0.2146,
+      "step": 4503
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1664971365595714e-05,
+      "loss": 0.1924,
+      "step": 4504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1661685590034446e-05,
+      "loss": 0.2153,
+      "step": 4505
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1658399629967621e-05,
+      "loss": 0.2136,
+      "step": 4506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1655113485760097e-05,
+      "loss": 0.1487,
+      "step": 4507
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1651827157776754e-05,
+      "loss": 0.1748,
+      "step": 4508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1648540646382487e-05,
+      "loss": 0.1697,
+      "step": 4509
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1645253951942215e-05,
+      "loss": 0.1682,
+      "step": 4510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.164196707482088e-05,
+      "loss": 0.178,
+      "step": 4511
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1638680015383434e-05,
+      "loss": 0.1664,
+      "step": 4512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1635392773994865e-05,
+      "loss": 0.1906,
+      "step": 4513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1632105351020166e-05,
+      "loss": 0.167,
+      "step": 4514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1628817746824356e-05,
+      "loss": 0.2024,
+      "step": 4515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1625529961772481e-05,
+      "loss": 0.1667,
+      "step": 4516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1622241996229597e-05,
+      "loss": 0.2148,
+      "step": 4517
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1618953850560783e-05,
+      "loss": 0.198,
+      "step": 4518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.161566552513114e-05,
+      "loss": 0.1812,
+      "step": 4519
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.161237702030579e-05,
+      "loss": 0.1691,
+      "step": 4520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1609088336449865e-05,
+      "loss": 0.1498,
+      "step": 4521
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1605799473928534e-05,
+      "loss": 0.1843,
+      "step": 4522
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1602510433106972e-05,
+      "loss": 0.1663,
+      "step": 4523
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1599221214350376e-05,
+      "loss": 0.1648,
+      "step": 4524
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1595931818023971e-05,
+      "loss": 0.1836,
+      "step": 4525
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.159264224449299e-05,
+      "loss": 0.2422,
+      "step": 4526
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1589352494122694e-05,
+      "loss": 0.1558,
+      "step": 4527
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1586062567278361e-05,
+      "loss": 0.1639,
+      "step": 4528
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1582772464325292e-05,
+      "loss": 0.1685,
+      "step": 4529
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1579482185628794e-05,
+      "loss": 0.189,
+      "step": 4530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1576191731554214e-05,
+      "loss": 0.2004,
+      "step": 4531
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1572901102466902e-05,
+      "loss": 0.1951,
+      "step": 4532
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1569610298732236e-05,
+      "loss": 0.1564,
+      "step": 4533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1566319320715614e-05,
+      "loss": 0.166,
+      "step": 4534
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1563028168782444e-05,
+      "loss": 0.1812,
+      "step": 4535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1559736843298164e-05,
+      "loss": 0.1562,
+      "step": 4536
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1556445344628226e-05,
+      "loss": 0.1985,
+      "step": 4537
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1553153673138103e-05,
+      "loss": 0.1497,
+      "step": 4538
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1549861829193282e-05,
+      "loss": 0.1868,
+      "step": 4539
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.154656981315928e-05,
+      "loss": 0.1594,
+      "step": 4540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.154327762540162e-05,
+      "loss": 0.1799,
+      "step": 4541
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1539985266285858e-05,
+      "loss": 0.191,
+      "step": 4542
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1536692736177555e-05,
+      "loss": 0.207,
+      "step": 4543
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.15334000354423e-05,
+      "loss": 0.197,
+      "step": 4544
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1530107164445701e-05,
+      "loss": 0.1753,
+      "step": 4545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.152681412355338e-05,
+      "loss": 0.1985,
+      "step": 4546
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1523520913130979e-05,
+      "loss": 0.1841,
+      "step": 4547
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1520227533544164e-05,
+      "loss": 0.167,
+      "step": 4548
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1516933985158612e-05,
+      "loss": 0.1775,
+      "step": 4549
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1513640268340025e-05,
+      "loss": 0.197,
+      "step": 4550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1510346383454122e-05,
+      "loss": 0.2212,
+      "step": 4551
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.150705233086664e-05,
+      "loss": 0.1738,
+      "step": 4552
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.150375811094333e-05,
+      "loss": 0.2273,
+      "step": 4553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1500463724049975e-05,
+      "loss": 0.1655,
+      "step": 4554
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.149716917055236e-05,
+      "loss": 0.1738,
+      "step": 4555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1493874450816302e-05,
+      "loss": 0.175,
+      "step": 4556
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1490579565207629e-05,
+      "loss": 0.1877,
+      "step": 4557
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1487284514092186e-05,
+      "loss": 0.2207,
+      "step": 4558
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1483989297835843e-05,
+      "loss": 0.1899,
+      "step": 4559
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1480693916804484e-05,
+      "loss": 0.1775,
+      "step": 4560
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147739837136401e-05,
+      "loss": 0.2017,
+      "step": 4561
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1474102661880347e-05,
+      "loss": 0.1534,
+      "step": 4562
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1470806788719433e-05,
+      "loss": 0.1743,
+      "step": 4563
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1467510752247222e-05,
+      "loss": 0.1897,
+      "step": 4564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1464214552829695e-05,
+      "loss": 0.1714,
+      "step": 4565
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1460918190832847e-05,
+      "loss": 0.1802,
+      "step": 4566
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1457621666622683e-05,
+      "loss": 0.1649,
+      "step": 4567
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1454324980565239e-05,
+      "loss": 0.186,
+      "step": 4568
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1451028133026563e-05,
+      "loss": 0.1626,
+      "step": 4569
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1447731124372721e-05,
+      "loss": 0.2009,
+      "step": 4570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1444433954969794e-05,
+      "loss": 0.1638,
+      "step": 4571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1441136625183885e-05,
+      "loss": 0.1359,
+      "step": 4572
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1437839135381112e-05,
+      "loss": 0.188,
+      "step": 4573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1434541485927619e-05,
+      "loss": 0.1699,
+      "step": 4574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1431243677189552e-05,
+      "loss": 0.1794,
+      "step": 4575
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1427945709533092e-05,
+      "loss": 0.1643,
+      "step": 4576
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1424647583324424e-05,
+      "loss": 0.1582,
+      "step": 4577
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1421349298929758e-05,
+      "loss": 0.1868,
+      "step": 4578
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.141805085671532e-05,
+      "loss": 0.1846,
+      "step": 4579
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1414752257047352e-05,
+      "loss": 0.1848,
+      "step": 4580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1411453500292115e-05,
+      "loss": 0.1373,
+      "step": 4581
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1408154586815887e-05,
+      "loss": 0.1787,
+      "step": 4582
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1404855516984965e-05,
+      "loss": 0.1567,
+      "step": 4583
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1401556291165659e-05,
+      "loss": 0.1855,
+      "step": 4584
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.13982569097243e-05,
+      "loss": 0.1403,
+      "step": 4585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1394957373027236e-05,
+      "loss": 0.1729,
+      "step": 4586
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1391657681440834e-05,
+      "loss": 0.162,
+      "step": 4587
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1388357835331472e-05,
+      "loss": 0.1562,
+      "step": 4588
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1385057835065549e-05,
+      "loss": 0.1759,
+      "step": 4589
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1381757681009484e-05,
+      "loss": 0.2036,
+      "step": 4590
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.137845737352971e-05,
+      "loss": 0.2087,
+      "step": 4591
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1375156912992674e-05,
+      "loss": 0.199,
+      "step": 4592
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1371856299764845e-05,
+      "loss": 0.1117,
+      "step": 4593
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1368555534212711e-05,
+      "loss": 0.1941,
+      "step": 4594
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1365254616702763e-05,
+      "loss": 0.1823,
+      "step": 4595
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1361953547601532e-05,
+      "loss": 0.1613,
+      "step": 4596
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1358652327275544e-05,
+      "loss": 0.1851,
+      "step": 4597
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.135535095609135e-05,
+      "loss": 0.1968,
+      "step": 4598
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1352049434415524e-05,
+      "loss": 0.1658,
+      "step": 4599
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1348747762614647e-05,
+      "loss": 0.1763,
+      "step": 4600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1345445941055322e-05,
+      "loss": 0.1812,
+      "step": 4601
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1342143970104167e-05,
+      "loss": 0.155,
+      "step": 4602
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.133884185012782e-05,
+      "loss": 0.1669,
+      "step": 4603
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1335539581492925e-05,
+      "loss": 0.1617,
+      "step": 4604
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1332237164566158e-05,
+      "loss": 0.1914,
+      "step": 4605
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1328934599714201e-05,
+      "loss": 0.1626,
+      "step": 4606
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.132563188730375e-05,
+      "loss": 0.1504,
+      "step": 4607
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1322329027701529e-05,
+      "loss": 0.1895,
+      "step": 4608
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1319026021274267e-05,
+      "loss": 0.1851,
+      "step": 4609
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1315722868388719e-05,
+      "loss": 0.1389,
+      "step": 4610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1312419569411645e-05,
+      "loss": 0.1685,
+      "step": 4611
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1309116124709832e-05,
+      "loss": 0.1588,
+      "step": 4612
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1305812534650075e-05,
+      "loss": 0.1636,
+      "step": 4613
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1302508799599193e-05,
+      "loss": 0.1526,
+      "step": 4614
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1299204919924011e-05,
+      "loss": 0.1681,
+      "step": 4615
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1295900895991384e-05,
+      "loss": 0.1685,
+      "step": 4616
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1292596728168168e-05,
+      "loss": 0.2058,
+      "step": 4617
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1289292416821243e-05,
+      "loss": 0.1804,
+      "step": 4618
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1285987962317505e-05,
+      "loss": 0.1575,
+      "step": 4619
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1282683365023866e-05,
+      "loss": 0.1376,
+      "step": 4620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.127937862530725e-05,
+      "loss": 0.166,
+      "step": 4621
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.12760737435346e-05,
+      "loss": 0.1909,
+      "step": 4622
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1272768720072875e-05,
+      "loss": 0.2085,
+      "step": 4623
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1269463555289049e-05,
+      "loss": 0.2029,
+      "step": 4624
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.126615824955011e-05,
+      "loss": 0.176,
+      "step": 4625
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1262852803223065e-05,
+      "loss": 0.177,
+      "step": 4626
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1259547216674932e-05,
+      "loss": 0.1858,
+      "step": 4627
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1256241490272753e-05,
+      "loss": 0.1542,
+      "step": 4628
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1252935624383574e-05,
+      "loss": 0.146,
+      "step": 4629
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1249629619374465e-05,
+      "loss": 0.1775,
+      "step": 4630
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1246323475612509e-05,
+      "loss": 0.1594,
+      "step": 4631
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1243017193464802e-05,
+      "loss": 0.1625,
+      "step": 4632
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1239710773298462e-05,
+      "loss": 0.1487,
+      "step": 4633
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1236404215480617e-05,
+      "loss": 0.1515,
+      "step": 4634
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1233097520378404e-05,
+      "loss": 0.1746,
+      "step": 4635
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1229790688358995e-05,
+      "loss": 0.1477,
+      "step": 4636
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1226483719789556e-05,
+      "loss": 0.1904,
+      "step": 4637
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1223176615037274e-05,
+      "loss": 0.2148,
+      "step": 4638
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1219869374469364e-05,
+      "loss": 0.1802,
+      "step": 4639
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1216561998453038e-05,
+      "loss": 0.1444,
+      "step": 4640
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1213254487355533e-05,
+      "loss": 0.1748,
+      "step": 4641
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1209946841544101e-05,
+      "loss": 0.1875,
+      "step": 4642
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1206639061386006e-05,
+      "loss": 0.1482,
+      "step": 4643
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.120333114724853e-05,
+      "loss": 0.1699,
+      "step": 4644
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1200023099498962e-05,
+      "loss": 0.1593,
+      "step": 4645
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1196714918504616e-05,
+      "loss": 0.1819,
+      "step": 4646
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1193406604632818e-05,
+      "loss": 0.1672,
+      "step": 4647
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1190098158250904e-05,
+      "loss": 0.1484,
+      "step": 4648
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1186789579726226e-05,
+      "loss": 0.1917,
+      "step": 4649
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.118348086942616e-05,
+      "loss": 0.2102,
+      "step": 4650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1180172027718084e-05,
+      "loss": 0.134,
+      "step": 4651
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1176863054969396e-05,
+      "loss": 0.1505,
+      "step": 4652
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1173553951547508e-05,
+      "loss": 0.1924,
+      "step": 4653
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1170244717819849e-05,
+      "loss": 0.2114,
+      "step": 4654
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1166935354153857e-05,
+      "loss": 0.1609,
+      "step": 4655
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1163625860916996e-05,
+      "loss": 0.1676,
+      "step": 4656
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1160316238476727e-05,
+      "loss": 0.1898,
+      "step": 4657
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1157006487200534e-05,
+      "loss": 0.1547,
+      "step": 4658
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1153696607455924e-05,
+      "loss": 0.187,
+      "step": 4659
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1150386599610406e-05,
+      "loss": 0.1816,
+      "step": 4660
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1147076464031503e-05,
+      "loss": 0.1841,
+      "step": 4661
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1143766201086765e-05,
+      "loss": 0.1714,
+      "step": 4662
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.114045581114374e-05,
+      "loss": 0.1731,
+      "step": 4663
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1137145294570002e-05,
+      "loss": 0.1525,
+      "step": 4664
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1133834651733135e-05,
+      "loss": 0.1694,
+      "step": 4665
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1130523883000732e-05,
+      "loss": 0.1899,
+      "step": 4666
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.112721298874041e-05,
+      "loss": 0.1807,
+      "step": 4667
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1123901969319795e-05,
+      "loss": 0.1632,
+      "step": 4668
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1120590825106521e-05,
+      "loss": 0.2134,
+      "step": 4669
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1117279556468248e-05,
+      "loss": 0.1526,
+      "step": 4670
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1113968163772641e-05,
+      "loss": 0.1809,
+      "step": 4671
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1110656647387376e-05,
+      "loss": 0.1707,
+      "step": 4672
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1107345007680157e-05,
+      "loss": 0.1792,
+      "step": 4673
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1104033245018686e-05,
+      "loss": 0.1667,
+      "step": 4674
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1100721359770686e-05,
+      "loss": 0.1882,
+      "step": 4675
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1097409352303896e-05,
+      "loss": 0.1675,
+      "step": 4676
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1094097222986063e-05,
+      "loss": 0.2041,
+      "step": 4677
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1090784972184947e-05,
+      "loss": 0.1603,
+      "step": 4678
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108747260026833e-05,
+      "loss": 0.1793,
+      "step": 4679
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1084160107603998e-05,
+      "loss": 0.1716,
+      "step": 4680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1080847494559754e-05,
+      "loss": 0.1694,
+      "step": 4681
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.107753476150342e-05,
+      "loss": 0.1938,
+      "step": 4682
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1074221908802819e-05,
+      "loss": 0.1926,
+      "step": 4683
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1070908936825795e-05,
+      "loss": 0.1946,
+      "step": 4684
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.106759584594021e-05,
+      "loss": 0.1643,
+      "step": 4685
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1064282636513927e-05,
+      "loss": 0.189,
+      "step": 4686
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1060969308914835e-05,
+      "loss": 0.1851,
+      "step": 4687
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1057655863510826e-05,
+      "loss": 0.1746,
+      "step": 4688
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1054342300669806e-05,
+      "loss": 0.1548,
+      "step": 4689
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1051028620759704e-05,
+      "loss": 0.1593,
+      "step": 4690
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1047714824148454e-05,
+      "loss": 0.209,
+      "step": 4691
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1044400911203998e-05,
+      "loss": 0.2039,
+      "step": 4692
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1041086882294306e-05,
+      "loss": 0.1478,
+      "step": 4693
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1037772737787342e-05,
+      "loss": 0.1934,
+      "step": 4694
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1034458478051098e-05,
+      "loss": 0.1477,
+      "step": 4695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1031144103453572e-05,
+      "loss": 0.1853,
+      "step": 4696
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1027829614362779e-05,
+      "loss": 0.2151,
+      "step": 4697
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1024515011146739e-05,
+      "loss": 0.1643,
+      "step": 4698
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1021200294173493e-05,
+      "loss": 0.1484,
+      "step": 4699
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1017885463811094e-05,
+      "loss": 0.2104,
+      "step": 4700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1014570520427594e-05,
+      "loss": 0.1772,
+      "step": 4701
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1011255464391081e-05,
+      "loss": 0.1538,
+      "step": 4702
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1007940296069638e-05,
+      "loss": 0.197,
+      "step": 4703
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1004625015831362e-05,
+      "loss": 0.1975,
+      "step": 4704
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1001309624044372e-05,
+      "loss": 0.1691,
+      "step": 4705
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0997994121076784e-05,
+      "loss": 0.1631,
+      "step": 4706
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0994678507296743e-05,
+      "loss": 0.198,
+      "step": 4707
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0991362783072399e-05,
+      "loss": 0.1892,
+      "step": 4708
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0988046948771909e-05,
+      "loss": 0.1658,
+      "step": 4709
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0984731004763452e-05,
+      "loss": 0.1851,
+      "step": 4710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0981414951415212e-05,
+      "loss": 0.1626,
+      "step": 4711
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0978098789095386e-05,
+      "loss": 0.1836,
+      "step": 4712
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.097478251817219e-05,
+      "loss": 0.1992,
+      "step": 4713
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0971466139013841e-05,
+      "loss": 0.1633,
+      "step": 4714
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0968149651988578e-05,
+      "loss": 0.1646,
+      "step": 4715
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0964833057464645e-05,
+      "loss": 0.1836,
+      "step": 4716
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09615163558103e-05,
+      "loss": 0.1339,
+      "step": 4717
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0958199547393817e-05,
+      "loss": 0.2087,
+      "step": 4718
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0954882632583477e-05,
+      "loss": 0.2078,
+      "step": 4719
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0951565611747572e-05,
+      "loss": 0.1707,
+      "step": 4720
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.094824848525441e-05,
+      "loss": 0.1709,
+      "step": 4721
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0944931253472313e-05,
+      "loss": 0.2002,
+      "step": 4722
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.09416139167696e-05,
+      "loss": 0.1765,
+      "step": 4723
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093829647551462e-05,
+      "loss": 0.1395,
+      "step": 4724
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0934978930075726e-05,
+      "loss": 0.1855,
+      "step": 4725
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0931661280821276e-05,
+      "loss": 0.1902,
+      "step": 4726
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0928343528119651e-05,
+      "loss": 0.1935,
+      "step": 4727
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.092502567233924e-05,
+      "loss": 0.1653,
+      "step": 4728
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0921707713848434e-05,
+      "loss": 0.2019,
+      "step": 4729
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0918389653015651e-05,
+      "loss": 0.1665,
+      "step": 4730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.091507149020931e-05,
+      "loss": 0.2039,
+      "step": 4731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0911753225797841e-05,
+      "loss": 0.1622,
+      "step": 4732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.090843486014969e-05,
+      "loss": 0.1848,
+      "step": 4733
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0905116393633314e-05,
+      "loss": 0.1729,
+      "step": 4734
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0901797826617176e-05,
+      "loss": 0.1868,
+      "step": 4735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0898479159469756e-05,
+      "loss": 0.1694,
+      "step": 4736
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0895160392559543e-05,
+      "loss": 0.1748,
+      "step": 4737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0891841526255038e-05,
+      "loss": 0.1597,
+      "step": 4738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.088852256092475e-05,
+      "loss": 0.1514,
+      "step": 4739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0885203496937198e-05,
+      "loss": 0.2209,
+      "step": 4740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0881884334660921e-05,
+      "loss": 0.1401,
+      "step": 4741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0878565074464459e-05,
+      "loss": 0.1658,
+      "step": 4742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0875245716716364e-05,
+      "loss": 0.1699,
+      "step": 4743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.087192626178521e-05,
+      "loss": 0.1917,
+      "step": 4744
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0868606710039569e-05,
+      "loss": 0.1533,
+      "step": 4745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0865287061848022e-05,
+      "loss": 0.1544,
+      "step": 4746
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0861967317579176e-05,
+      "loss": 0.1884,
+      "step": 4747
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0858647477601633e-05,
+      "loss": 0.1406,
+      "step": 4748
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0855327542284015e-05,
+      "loss": 0.1887,
+      "step": 4749
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0852007511994953e-05,
+      "loss": 0.1338,
+      "step": 4750
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0848687387103082e-05,
+      "loss": 0.1566,
+      "step": 4751
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0845367167977056e-05,
+      "loss": 0.1687,
+      "step": 4752
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.084204685498554e-05,
+      "loss": 0.1702,
+      "step": 4753
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.08387264484972e-05,
+      "loss": 0.1721,
+      "step": 4754
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0835405948880716e-05,
+      "loss": 0.2092,
+      "step": 4755
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0832085356504787e-05,
+      "loss": 0.2078,
+      "step": 4756
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082876467173811e-05,
+      "loss": 0.189,
+      "step": 4757
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0825443894949398e-05,
+      "loss": 0.1758,
+      "step": 4758
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082212302650738e-05,
+      "loss": 0.1473,
+      "step": 4759
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0818802066780785e-05,
+      "loss": 0.1415,
+      "step": 4760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0815481016138356e-05,
+      "loss": 0.1969,
+      "step": 4761
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0812159874948847e-05,
+      "loss": 0.1921,
+      "step": 4762
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0808838643581022e-05,
+      "loss": 0.1615,
+      "step": 4763
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0805517322403654e-05,
+      "loss": 0.2168,
+      "step": 4764
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0802195911785527e-05,
+      "loss": 0.2014,
+      "step": 4765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.079887441209543e-05,
+      "loss": 0.1665,
+      "step": 4766
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0795552823702176e-05,
+      "loss": 0.1895,
+      "step": 4767
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.079223114697457e-05,
+      "loss": 0.1857,
+      "step": 4768
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0788909382281437e-05,
+      "loss": 0.2069,
+      "step": 4769
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0785587529991612e-05,
+      "loss": 0.1726,
+      "step": 4770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0782265590473934e-05,
+      "loss": 0.153,
+      "step": 4771
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.077894356409726e-05,
+      "loss": 0.1558,
+      "step": 4772
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0775621451230444e-05,
+      "loss": 0.1664,
+      "step": 4773
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0772299252242364e-05,
+      "loss": 0.1887,
+      "step": 4774
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.07689769675019e-05,
+      "loss": 0.17,
+      "step": 4775
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0765654597377941e-05,
+      "loss": 0.1528,
+      "step": 4776
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0762332142239384e-05,
+      "loss": 0.1609,
+      "step": 4777
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0759009602455146e-05,
+      "loss": 0.1863,
+      "step": 4778
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0755686978394142e-05,
+      "loss": 0.186,
+      "step": 4779
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0752364270425294e-05,
+      "loss": 0.1992,
+      "step": 4780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0749041478917552e-05,
+      "loss": 0.1422,
+      "step": 4781
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0745718604239852e-05,
+      "loss": 0.1921,
+      "step": 4782
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0742395646761155e-05,
+      "loss": 0.16,
+      "step": 4783
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0739072606850425e-05,
+      "loss": 0.1315,
+      "step": 4784
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0735749484876638e-05,
+      "loss": 0.1858,
+      "step": 4785
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0732426281208772e-05,
+      "loss": 0.2136,
+      "step": 4786
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0729102996215828e-05,
+      "loss": 0.1301,
+      "step": 4787
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.07257796302668e-05,
+      "loss": 0.1733,
+      "step": 4788
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.07224561837307e-05,
+      "loss": 0.188,
+      "step": 4789
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0719132656976551e-05,
+      "loss": 0.1907,
+      "step": 4790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0715809050373381e-05,
+      "loss": 0.1963,
+      "step": 4791
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0712485364290224e-05,
+      "loss": 0.1577,
+      "step": 4792
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0709161599096128e-05,
+      "loss": 0.2107,
+      "step": 4793
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0705837755160147e-05,
+      "loss": 0.1775,
+      "step": 4794
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0702513832851344e-05,
+      "loss": 0.1772,
+      "step": 4795
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0699189832538795e-05,
+      "loss": 0.1597,
+      "step": 4796
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0695865754591576e-05,
+      "loss": 0.2302,
+      "step": 4797
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069254159937878e-05,
+      "loss": 0.1633,
+      "step": 4798
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0689217367269507e-05,
+      "loss": 0.1951,
+      "step": 4799
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0685893058632855e-05,
+      "loss": 0.1711,
+      "step": 4800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068256867383795e-05,
+      "loss": 0.1763,
+      "step": 4801
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0679244213253907e-05,
+      "loss": 0.1814,
+      "step": 4802
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0675919677249864e-05,
+      "loss": 0.1709,
+      "step": 4803
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0672595066194959e-05,
+      "loss": 0.1841,
+      "step": 4804
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0669270380458342e-05,
+      "loss": 0.1543,
+      "step": 4805
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0665945620409166e-05,
+      "loss": 0.1727,
+      "step": 4806
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0662620786416601e-05,
+      "loss": 0.199,
+      "step": 4807
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0659295878849822e-05,
+      "loss": 0.1708,
+      "step": 4808
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0655970898078003e-05,
+      "loss": 0.1658,
+      "step": 4809
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0652645844470342e-05,
+      "loss": 0.196,
+      "step": 4810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0649320718396032e-05,
+      "loss": 0.2244,
+      "step": 4811
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0645995520224279e-05,
+      "loss": 0.1436,
+      "step": 4812
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0642670250324301e-05,
+      "loss": 0.1873,
+      "step": 4813
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0639344909065312e-05,
+      "loss": 0.1848,
+      "step": 4814
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0636019496816552e-05,
+      "loss": 0.1647,
+      "step": 4815
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0632694013947254e-05,
+      "loss": 0.1609,
+      "step": 4816
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0629368460826662e-05,
+      "loss": 0.1941,
+      "step": 4817
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0626042837824032e-05,
+      "loss": 0.1929,
+      "step": 4818
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0622717145308623e-05,
+      "loss": 0.1456,
+      "step": 4819
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0619391383649701e-05,
+      "loss": 0.2148,
+      "step": 4820
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0616065553216553e-05,
+      "loss": 0.2178,
+      "step": 4821
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0612739654378455e-05,
+      "loss": 0.1956,
+      "step": 4822
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0609413687504698e-05,
+      "loss": 0.1663,
+      "step": 4823
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0606087652964585e-05,
+      "loss": 0.198,
+      "step": 4824
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0602761551127421e-05,
+      "loss": 0.1982,
+      "step": 4825
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.059943538236252e-05,
+      "loss": 0.194,
+      "step": 4826
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0596109147039207e-05,
+      "loss": 0.2017,
+      "step": 4827
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0592782845526808e-05,
+      "loss": 0.1892,
+      "step": 4828
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0589456478194658e-05,
+      "loss": 0.177,
+      "step": 4829
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0586130045412103e-05,
+      "loss": 0.1796,
+      "step": 4830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0582803547548497e-05,
+      "loss": 0.1689,
+      "step": 4831
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0579476984973191e-05,
+      "loss": 0.1543,
+      "step": 4832
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0576150358055559e-05,
+      "loss": 0.1749,
+      "step": 4833
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0572823667164968e-05,
+      "loss": 0.1665,
+      "step": 4834
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0569496912670798e-05,
+      "loss": 0.1663,
+      "step": 4835
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056617009494244e-05,
+      "loss": 0.1836,
+      "step": 4836
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056284321434928e-05,
+      "loss": 0.1719,
+      "step": 4837
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0559516271260727e-05,
+      "loss": 0.1631,
+      "step": 4838
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0556189266046187e-05,
+      "loss": 0.1702,
+      "step": 4839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0552862199075067e-05,
+      "loss": 0.1833,
+      "step": 4840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0549535070716799e-05,
+      "loss": 0.1565,
+      "step": 4841
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0546207881340804e-05,
+      "loss": 0.2375,
+      "step": 4842
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.054288063131652e-05,
+      "loss": 0.1577,
+      "step": 4843
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0539553321013388e-05,
+      "loss": 0.179,
+      "step": 4844
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0536225950800856e-05,
+      "loss": 0.2122,
+      "step": 4845
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0532898521048382e-05,
+      "loss": 0.2146,
+      "step": 4846
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0529571032125426e-05,
+      "loss": 0.1704,
+      "step": 4847
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0526243484401453e-05,
+      "loss": 0.1598,
+      "step": 4848
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0522915878245945e-05,
+      "loss": 0.1506,
+      "step": 4849
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0519588214028377e-05,
+      "loss": 0.1589,
+      "step": 4850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0516260492118238e-05,
+      "loss": 0.1743,
+      "step": 4851
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0512932712885027e-05,
+      "loss": 0.1943,
+      "step": 4852
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0509604876698238e-05,
+      "loss": 0.1694,
+      "step": 4853
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0506276983927382e-05,
+      "loss": 0.153,
+      "step": 4854
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0502949034941969e-05,
+      "loss": 0.1799,
+      "step": 4855
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0499621030111523e-05,
+      "loss": 0.2478,
+      "step": 4856
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0496292969805563e-05,
+      "loss": 0.1733,
+      "step": 4857
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049296485439363e-05,
+      "loss": 0.1558,
+      "step": 4858
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0489636684245255e-05,
+      "loss": 0.158,
+      "step": 4859
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048630845972998e-05,
+      "loss": 0.1665,
+      "step": 4860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048298018121736e-05,
+      "loss": 0.1724,
+      "step": 4861
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0479651849076953e-05,
+      "loss": 0.1636,
+      "step": 4862
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0476323463678313e-05,
+      "loss": 0.1566,
+      "step": 4863
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0472995025391015e-05,
+      "loss": 0.1785,
+      "step": 4864
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.046966653458463e-05,
+      "loss": 0.129,
+      "step": 4865
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0466337991628737e-05,
+      "loss": 0.1917,
+      "step": 4866
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0463009396892923e-05,
+      "loss": 0.1597,
+      "step": 4867
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0459680750746775e-05,
+      "loss": 0.1838,
+      "step": 4868
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0456352053559895e-05,
+      "loss": 0.1953,
+      "step": 4869
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0453023305701886e-05,
+      "loss": 0.168,
+      "step": 4870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044969450754235e-05,
+      "loss": 0.1514,
+      "step": 4871
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0446365659450906e-05,
+      "loss": 0.1758,
+      "step": 4872
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.044303676179717e-05,
+      "loss": 0.1785,
+      "step": 4873
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0439707814950768e-05,
+      "loss": 0.1826,
+      "step": 4874
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0436378819281333e-05,
+      "loss": 0.2097,
+      "step": 4875
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0433049775158497e-05,
+      "loss": 0.1443,
+      "step": 4876
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.04297206829519e-05,
+      "loss": 0.1494,
+      "step": 4877
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426391543031194e-05,
+      "loss": 0.1624,
+      "step": 4878
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0423062355766025e-05,
+      "loss": 0.1877,
+      "step": 4879
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.041973312152605e-05,
+      "loss": 0.1504,
+      "step": 4880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0416403840680938e-05,
+      "loss": 0.1677,
+      "step": 4881
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0413074513600348e-05,
+      "loss": 0.1978,
+      "step": 4882
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0409745140653955e-05,
+      "loss": 0.1597,
+      "step": 4883
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.040641572221144e-05,
+      "loss": 0.1875,
+      "step": 4884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0403086258642482e-05,
+      "loss": 0.1753,
+      "step": 4885
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0399756750316767e-05,
+      "loss": 0.1729,
+      "step": 4886
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0396427197603992e-05,
+      "loss": 0.1962,
+      "step": 4887
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393097600873854e-05,
+      "loss": 0.1663,
+      "step": 4888
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0389767960496051e-05,
+      "loss": 0.1877,
+      "step": 4889
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0386438276840296e-05,
+      "loss": 0.1443,
+      "step": 4890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0383108550276295e-05,
+      "loss": 0.1675,
+      "step": 4891
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.037977878117377e-05,
+      "loss": 0.1826,
+      "step": 4892
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0376448969902441e-05,
+      "loss": 0.1289,
+      "step": 4893
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0373119116832031e-05,
+      "loss": 0.1682,
+      "step": 4894
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0369789222332276e-05,
+      "loss": 0.1798,
+      "step": 4895
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0366459286772907e-05,
+      "loss": 0.1595,
+      "step": 4896
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0363129310523666e-05,
+      "loss": 0.2085,
+      "step": 4897
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0359799293954298e-05,
+      "loss": 0.1716,
+      "step": 4898
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0356469237434552e-05,
+      "loss": 0.1876,
+      "step": 4899
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0353139141334179e-05,
+      "loss": 0.1732,
+      "step": 4900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034980900602294e-05,
+      "loss": 0.1473,
+      "step": 4901
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346478831870596e-05,
+      "loss": 0.1458,
+      "step": 4902
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034314861924691e-05,
+      "loss": 0.1631,
+      "step": 4903
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033981836852166e-05,
+      "loss": 0.1542,
+      "step": 4904
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0336488080064614e-05,
+      "loss": 0.2358,
+      "step": 4905
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0333157754245557e-05,
+      "loss": 0.1831,
+      "step": 4906
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0329827391434268e-05,
+      "loss": 0.1604,
+      "step": 4907
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0326496992000535e-05,
+      "loss": 0.1833,
+      "step": 4908
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.032316655631415e-05,
+      "loss": 0.1519,
+      "step": 4909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0319836084744912e-05,
+      "loss": 0.1602,
+      "step": 4910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0316505577662611e-05,
+      "loss": 0.1946,
+      "step": 4911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0313175035437065e-05,
+      "loss": 0.1847,
+      "step": 4912
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0309844458438068e-05,
+      "loss": 0.1899,
+      "step": 4913
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0306513847035435e-05,
+      "loss": 0.1726,
+      "step": 4914
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0303183201598985e-05,
+      "loss": 0.174,
+      "step": 4915
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0299852522498535e-05,
+      "loss": 0.1716,
+      "step": 4916
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0296521810103906e-05,
+      "loss": 0.2178,
+      "step": 4917
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0293191064784924e-05,
+      "loss": 0.1433,
+      "step": 4918
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028986028691142e-05,
+      "loss": 0.2,
+      "step": 4919
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0286529476853228e-05,
+      "loss": 0.1406,
+      "step": 4920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0283198634980185e-05,
+      "loss": 0.1543,
+      "step": 4921
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0279867761662127e-05,
+      "loss": 0.1829,
+      "step": 4922
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027653685726891e-05,
+      "loss": 0.1699,
+      "step": 4923
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0273205922170369e-05,
+      "loss": 0.1702,
+      "step": 4924
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0269874956736359e-05,
+      "loss": 0.1638,
+      "step": 4925
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0266543961336737e-05,
+      "loss": 0.188,
+      "step": 4926
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0263212936341358e-05,
+      "loss": 0.2046,
+      "step": 4927
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0259881882120082e-05,
+      "loss": 0.156,
+      "step": 4928
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0256550799042779e-05,
+      "loss": 0.1414,
+      "step": 4929
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025321968747931e-05,
+      "loss": 0.1327,
+      "step": 4930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0249888547799547e-05,
+      "loss": 0.1736,
+      "step": 4931
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0246557380373366e-05,
+      "loss": 0.2153,
+      "step": 4932
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0243226185570643e-05,
+      "loss": 0.1862,
+      "step": 4933
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0239894963761254e-05,
+      "loss": 0.2068,
+      "step": 4934
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0236563715315089e-05,
+      "loss": 0.158,
+      "step": 4935
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0233232440602027e-05,
+      "loss": 0.167,
+      "step": 4936
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022990113999196e-05,
+      "loss": 0.2014,
+      "step": 4937
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022656981385478e-05,
+      "loss": 0.2024,
+      "step": 4938
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022323846256038e-05,
+      "loss": 0.166,
+      "step": 4939
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0219907086478655e-05,
+      "loss": 0.1682,
+      "step": 4940
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021657568597951e-05,
+      "loss": 0.1418,
+      "step": 4941
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0213244261432844e-05,
+      "loss": 0.2034,
+      "step": 4942
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0209912813208565e-05,
+      "loss": 0.157,
+      "step": 4943
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.020658134167658e-05,
+      "loss": 0.1667,
+      "step": 4944
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0203249847206796e-05,
+      "loss": 0.1809,
+      "step": 4945
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.019991833016913e-05,
+      "loss": 0.176,
+      "step": 4946
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0196586790933498e-05,
+      "loss": 0.1307,
+      "step": 4947
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0193255229869813e-05,
+      "loss": 0.1956,
+      "step": 4948
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0189923647348003e-05,
+      "loss": 0.1575,
+      "step": 4949
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0186592043737988e-05,
+      "loss": 0.2041,
+      "step": 4950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0183260419409689e-05,
+      "loss": 0.1484,
+      "step": 4951
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.017992877473304e-05,
+      "loss": 0.1694,
+      "step": 4952
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0176597110077964e-05,
+      "loss": 0.1567,
+      "step": 4953
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0173265425814397e-05,
+      "loss": 0.1924,
+      "step": 4954
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0169933722312273e-05,
+      "loss": 0.177,
+      "step": 4955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0166601999941528e-05,
+      "loss": 0.2014,
+      "step": 4956
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.01632702590721e-05,
+      "loss": 0.1608,
+      "step": 4957
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0159938500073928e-05,
+      "loss": 0.175,
+      "step": 4958
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0156606723316962e-05,
+      "loss": 0.2134,
+      "step": 4959
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0153274929171134e-05,
+      "loss": 0.1946,
+      "step": 4960
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.01499431180064e-05,
+      "loss": 0.2014,
+      "step": 4961
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0146611290192706e-05,
+      "loss": 0.1565,
+      "step": 4962
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0143279446099999e-05,
+      "loss": 0.1736,
+      "step": 4963
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0139947586098235e-05,
+      "loss": 0.1641,
+      "step": 4964
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0136615710557362e-05,
+      "loss": 0.2219,
+      "step": 4965
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0133283819847344e-05,
+      "loss": 0.1869,
+      "step": 4966
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0129951914338132e-05,
+      "loss": 0.1765,
+      "step": 4967
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0126619994399684e-05,
+      "loss": 0.1476,
+      "step": 4968
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0123288060401966e-05,
+      "loss": 0.167,
+      "step": 4969
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0119956112714934e-05,
+      "loss": 0.1736,
+      "step": 4970
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0116624151708554e-05,
+      "loss": 0.2219,
+      "step": 4971
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0113292177752791e-05,
+      "loss": 0.1572,
+      "step": 4972
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0109960191217612e-05,
+      "loss": 0.1536,
+      "step": 4973
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0106628192472983e-05,
+      "loss": 0.1675,
+      "step": 4974
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0103296181888873e-05,
+      "loss": 0.146,
+      "step": 4975
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0099964159835249e-05,
+      "loss": 0.1536,
+      "step": 4976
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0096632126682093e-05,
+      "loss": 0.1978,
+      "step": 4977
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0093300082799368e-05,
+      "loss": 0.1631,
+      "step": 4978
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0089968028557054e-05,
+      "loss": 0.1975,
+      "step": 4979
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0086635964325121e-05,
+      "loss": 0.188,
+      "step": 4980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008330389047355e-05,
+      "loss": 0.1594,
+      "step": 4981
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0079971807372318e-05,
+      "loss": 0.1534,
+      "step": 4982
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0076639715391399e-05,
+      "loss": 0.1912,
+      "step": 4983
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0073307614900778e-05,
+      "loss": 0.1804,
+      "step": 4984
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.006997550627043e-05,
+      "loss": 0.1726,
+      "step": 4985
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0066643389870343e-05,
+      "loss": 0.1699,
+      "step": 4986
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0063311266070495e-05,
+      "loss": 0.1792,
+      "step": 4987
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0059979135240868e-05,
+      "loss": 0.1985,
+      "step": 4988
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0056646997751448e-05,
+      "loss": 0.1564,
+      "step": 4989
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0053314853972216e-05,
+      "loss": 0.2048,
+      "step": 4990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.004998270427316e-05,
+      "loss": 0.157,
+      "step": 4991
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0046650549024267e-05,
+      "loss": 0.1848,
+      "step": 4992
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0043318388595523e-05,
+      "loss": 0.1479,
+      "step": 4993
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.003998622335691e-05,
+      "loss": 0.188,
+      "step": 4994
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0036654053678425e-05,
+      "loss": 0.1917,
+      "step": 4995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0033321879930045e-05,
+      "loss": 0.1765,
+      "step": 4996
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0029989702481768e-05,
+      "loss": 0.1577,
+      "step": 4997
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.002665752170358e-05,
+      "loss": 0.1809,
+      "step": 4998
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023325337965466e-05,
+      "loss": 0.2417,
+      "step": 4999
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0019993151637419e-05,
+      "loss": 0.1696,
+      "step": 5000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0016660963089433e-05,
+      "loss": 0.1796,
+      "step": 5001
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0013328772691489e-05,
+      "loss": 0.1736,
+      "step": 5002
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0009996580813585e-05,
+      "loss": 0.1604,
+      "step": 5003
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0006664387825709e-05,
+      "loss": 0.1841,
+      "step": 5004
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.000333219409785e-05,
+      "loss": 0.1782,
+      "step": 5005
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1e-05,
+      "loss": 0.136,
+      "step": 5006
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996667805902154e-06,
+      "loss": 0.1702,
+      "step": 5007
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993335612174295e-06,
+      "loss": 0.1787,
+      "step": 5008
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99000341918642e-06,
+      "loss": 0.1848,
+      "step": 5009
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.986671227308514e-06,
+      "loss": 0.1582,
+      "step": 5010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.983339036910573e-06,
+      "loss": 0.1996,
+      "step": 5011
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.980006848362583e-06,
+      "loss": 0.1936,
+      "step": 5012
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976674662034537e-06,
+      "loss": 0.1587,
+      "step": 5013
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.973342478296423e-06,
+      "loss": 0.2004,
+      "step": 5014
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.970010297518237e-06,
+      "loss": 0.163,
+      "step": 5015
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.966678120069957e-06,
+      "loss": 0.1968,
+      "step": 5016
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.963345946321582e-06,
+      "loss": 0.1632,
+      "step": 5017
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.960013776643091e-06,
+      "loss": 0.1484,
+      "step": 5018
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.956681611404482e-06,
+      "loss": 0.1814,
+      "step": 5019
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.953349450975736e-06,
+      "loss": 0.2207,
+      "step": 5020
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.950017295726845e-06,
+      "loss": 0.1787,
+      "step": 5021
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946685146027787e-06,
+      "loss": 0.1395,
+      "step": 5022
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.94335300224856e-06,
+      "loss": 0.1699,
+      "step": 5023
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.940020864759137e-06,
+      "loss": 0.2156,
+      "step": 5024
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93668873392951e-06,
+      "loss": 0.1332,
+      "step": 5025
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.93335661012966e-06,
+      "loss": 0.1486,
+      "step": 5026
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.930024493729573e-06,
+      "loss": 0.1658,
+      "step": 5027
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.926692385099224e-06,
+      "loss": 0.1951,
+      "step": 5028
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.923360284608606e-06,
+      "loss": 0.1831,
+      "step": 5029
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.920028192627686e-06,
+      "loss": 0.1685,
+      "step": 5030
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.916696109526454e-06,
+      "loss": 0.1785,
+      "step": 5031
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.913364035674882e-06,
+      "loss": 0.1929,
+      "step": 5032
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.91003197144295e-06,
+      "loss": 0.1821,
+      "step": 5033
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.906699917200633e-06,
+      "loss": 0.1635,
+      "step": 5034
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.903367873317913e-06,
+      "loss": 0.1704,
+      "step": 5035
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.900035840164753e-06,
+      "loss": 0.1748,
+      "step": 5036
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.896703818111134e-06,
+      "loss": 0.2117,
+      "step": 5037
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.893371807527022e-06,
+      "loss": 0.1935,
+      "step": 5038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.890039808782393e-06,
+      "loss": 0.1743,
+      "step": 5039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88670782224721e-06,
+      "loss": 0.2166,
+      "step": 5040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88337584829145e-06,
+      "loss": 0.1965,
+      "step": 5041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.880043887285065e-06,
+      "loss": 0.2227,
+      "step": 5042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.876711939598037e-06,
+      "loss": 0.1594,
+      "step": 5043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.873380005600316e-06,
+      "loss": 0.2004,
+      "step": 5044
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.87004808566187e-06,
+      "loss": 0.2297,
+      "step": 5045
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.866716180152656e-06,
+      "loss": 0.1956,
+      "step": 5046
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.86338428944264e-06,
+      "loss": 0.1787,
+      "step": 5047
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.860052413901766e-06,
+      "loss": 0.1624,
+      "step": 5048
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.856720553900004e-06,
+      "loss": 0.1985,
+      "step": 5049
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.853388709807296e-06,
+      "loss": 0.1262,
+      "step": 5050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.850056881993602e-06,
+      "loss": 0.2017,
+      "step": 5051
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.846725070828866e-06,
+      "loss": 0.1753,
+      "step": 5052
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.843393276683042e-06,
+      "loss": 0.156,
+      "step": 5053
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.84006149992607e-06,
+      "loss": 0.1714,
+      "step": 5054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.836729740927904e-06,
+      "loss": 0.1537,
+      "step": 5055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.833398000058473e-06,
+      "loss": 0.176,
+      "step": 5056
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.83006627768773e-06,
+      "loss": 0.1503,
+      "step": 5057
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.826734574185605e-06,
+      "loss": 0.176,
+      "step": 5058
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.823402889922039e-06,
+      "loss": 0.1658,
+      "step": 5059
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.820071225266962e-06,
+      "loss": 0.1667,
+      "step": 5060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.816739580590315e-06,
+      "loss": 0.1407,
+      "step": 5061
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.813407956262016e-06,
+      "loss": 0.1765,
+      "step": 5062
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.810076352651999e-06,
+      "loss": 0.1824,
+      "step": 5063
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.806744770130185e-06,
+      "loss": 0.1387,
+      "step": 5064
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.803413209066504e-06,
+      "loss": 0.1687,
+      "step": 5065
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.80008166983087e-06,
+      "loss": 0.1921,
+      "step": 5066
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.796750152793208e-06,
+      "loss": 0.1936,
+      "step": 5067
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.793418658323422e-06,
+      "loss": 0.1721,
+      "step": 5068
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.790087186791438e-06,
+      "loss": 0.1602,
+      "step": 5069
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.786755738567156e-06,
+      "loss": 0.1821,
+      "step": 5070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.783424314020492e-06,
+      "loss": 0.2004,
+      "step": 5071
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.780092913521345e-06,
+      "loss": 0.177,
+      "step": 5072
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.776761537439623e-06,
+      "loss": 0.1747,
+      "step": 5073
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.773430186145225e-06,
+      "loss": 0.1527,
+      "step": 5074
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770098860008043e-06,
+      "loss": 0.1858,
+      "step": 5075
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.766767559397978e-06,
+      "loss": 0.1495,
+      "step": 5076
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.763436284684915e-06,
+      "loss": 0.1904,
+      "step": 5077
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76010503623875e-06,
+      "loss": 0.1755,
+      "step": 5078
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75677381442936e-06,
+      "loss": 0.1819,
+      "step": 5079
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.753442619626638e-06,
+      "loss": 0.1926,
+      "step": 5080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.750111452200455e-06,
+      "loss": 0.179,
+      "step": 5081
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.746780312520695e-06,
+      "loss": 0.1519,
+      "step": 5082
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.743449200957225e-06,
+      "loss": 0.2119,
+      "step": 5083
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.74011811787992e-06,
+      "loss": 0.2112,
+      "step": 5084
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.736787063658645e-06,
+      "loss": 0.2029,
+      "step": 5085
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.733456038663268e-06,
+      "loss": 0.1381,
+      "step": 5086
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.730125043263645e-06,
+      "loss": 0.1395,
+      "step": 5087
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.726794077829636e-06,
+      "loss": 0.158,
+      "step": 5088
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.723463142731094e-06,
+      "loss": 0.1926,
+      "step": 5089
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.720132238337874e-06,
+      "loss": 0.175,
+      "step": 5090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.716801365019819e-06,
+      "loss": 0.1775,
+      "step": 5091
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.713470523146777e-06,
+      "loss": 0.179,
+      "step": 5092
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.710139713088585e-06,
+      "loss": 0.1794,
+      "step": 5093
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.706808935215081e-06,
+      "loss": 0.1642,
+      "step": 5094
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.703478189896099e-06,
+      "loss": 0.2029,
+      "step": 5095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.700147477501469e-06,
+      "loss": 0.147,
+      "step": 5096
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.696816798401017e-06,
+      "loss": 0.1648,
+      "step": 5097
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.69348615296457e-06,
+      "loss": 0.1995,
+      "step": 5098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.690155541561936e-06,
+      "loss": 0.1661,
+      "step": 5099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.686824964562942e-06,
+      "loss": 0.1753,
+      "step": 5100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68349442233739e-06,
+      "loss": 0.1718,
+      "step": 5101
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.680163915255095e-06,
+      "loss": 0.196,
+      "step": 5102
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.676833443685852e-06,
+      "loss": 0.1687,
+      "step": 5103
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.673503007999469e-06,
+      "loss": 0.198,
+      "step": 5104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.670172608565735e-06,
+      "loss": 0.1711,
+      "step": 5105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.666842245754448e-06,
+      "loss": 0.1689,
+      "step": 5106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.663511919935387e-06,
+      "loss": 0.1716,
+      "step": 5107
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.660181631478343e-06,
+      "loss": 0.2014,
+      "step": 5108
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.65685138075309e-06,
+      "loss": 0.1943,
+      "step": 5109
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.653521168129407e-06,
+      "loss": 0.1565,
+      "step": 5110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.65019099397706e-06,
+      "loss": 0.1519,
+      "step": 5111
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.646860858665825e-06,
+      "loss": 0.1624,
+      "step": 5112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.64353076256545e-06,
+      "loss": 0.1545,
+      "step": 5113
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.640200706045705e-06,
+      "loss": 0.1749,
+      "step": 5114
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.636870689476334e-06,
+      "loss": 0.1729,
+      "step": 5115
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.633540713227095e-06,
+      "loss": 0.177,
+      "step": 5116
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.630210777667726e-06,
+      "loss": 0.1628,
+      "step": 5117
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.626880883167972e-06,
+      "loss": 0.1606,
+      "step": 5118
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.62355103009756e-06,
+      "loss": 0.1521,
+      "step": 5119
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.620221218826233e-06,
+      "loss": 0.1729,
+      "step": 5120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.616891449723705e-06,
+      "loss": 0.1895,
+      "step": 5121
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.613561723159707e-06,
+      "loss": 0.2083,
+      "step": 5122
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.610232039503949e-06,
+      "loss": 0.1597,
+      "step": 5123
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.606902399126148e-06,
+      "loss": 0.1951,
+      "step": 5124
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.603572802396007e-06,
+      "loss": 0.1641,
+      "step": 5125
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.600243249683235e-06,
+      "loss": 0.1658,
+      "step": 5126
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.59691374135752e-06,
+      "loss": 0.1609,
+      "step": 5127
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.593584277788563e-06,
+      "loss": 0.1339,
+      "step": 5128
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.590254859346046e-06,
+      "loss": 0.2136,
+      "step": 5129
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.586925486399656e-06,
+      "loss": 0.2175,
+      "step": 5130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.583596159319064e-06,
+      "loss": 0.1499,
+      "step": 5131
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.580266878473952e-06,
+      "loss": 0.1599,
+      "step": 5132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576937644233977e-06,
+      "loss": 0.2024,
+      "step": 5133
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573608456968811e-06,
+      "loss": 0.1416,
+      "step": 5134
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.5702793170481e-06,
+      "loss": 0.199,
+      "step": 5135
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.566950224841506e-06,
+      "loss": 0.177,
+      "step": 5136
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.563621180718669e-06,
+      "loss": 0.1736,
+      "step": 5137
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.560292185049235e-06,
+      "loss": 0.1665,
+      "step": 5138
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55696323820283e-06,
+      "loss": 0.1624,
+      "step": 5139
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.553634340549098e-06,
+      "loss": 0.1285,
+      "step": 5140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.550305492457655e-06,
+      "loss": 0.1329,
+      "step": 5141
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.546976694298117e-06,
+      "loss": 0.1593,
+      "step": 5142
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.54364794644011e-06,
+      "loss": 0.1594,
+      "step": 5143
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.540319249253229e-06,
+      "loss": 0.1746,
+      "step": 5144
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.536990603107084e-06,
+      "loss": 0.1349,
+      "step": 5145
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.533662008371267e-06,
+      "loss": 0.2334,
+      "step": 5146
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.530333465415375e-06,
+      "loss": 0.1437,
+      "step": 5147
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.527004974608987e-06,
+      "loss": 0.1772,
+      "step": 5148
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.523676536321692e-06,
+      "loss": 0.199,
+      "step": 5149
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52034815092305e-06,
+      "loss": 0.1555,
+      "step": 5150
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.517019818782644e-06,
+      "loss": 0.2004,
+      "step": 5151
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.513691540270023e-06,
+      "loss": 0.1924,
+      "step": 5152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.510363315754752e-06,
+      "loss": 0.1709,
+      "step": 5153
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.507035145606374e-06,
+      "loss": 0.1787,
+      "step": 5154
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.503707030194439e-06,
+      "loss": 0.189,
+      "step": 5155
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.500378969888479e-06,
+      "loss": 0.2029,
+      "step": 5156
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.497050965058036e-06,
+      "loss": 0.1567,
+      "step": 5157
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.493723016072623e-06,
+      "loss": 0.1755,
+      "step": 5158
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.490395123301767e-06,
+      "loss": 0.1863,
+      "step": 5159
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.487067287114978e-06,
+      "loss": 0.2114,
+      "step": 5160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.483739507881764e-06,
+      "loss": 0.1522,
+      "step": 5161
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.480411785971626e-06,
+      "loss": 0.1665,
+      "step": 5162
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.477084121754062e-06,
+      "loss": 0.1963,
+      "step": 5163
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47375651559855e-06,
+      "loss": 0.1721,
+      "step": 5164
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.470428967874579e-06,
+      "loss": 0.1846,
+      "step": 5165
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467101478951621e-06,
+      "loss": 0.1847,
+      "step": 5166
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.463774049199148e-06,
+      "loss": 0.1794,
+      "step": 5167
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.460446678986616e-06,
+      "loss": 0.168,
+      "step": 5168
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.457119368683486e-06,
+      "loss": 0.2061,
+      "step": 5169
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.453792118659198e-06,
+      "loss": 0.1965,
+      "step": 5170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.450464929283208e-06,
+      "loss": 0.1926,
+      "step": 5171
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.447137800924937e-06,
+      "loss": 0.1659,
+      "step": 5172
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.443810733953818e-06,
+      "loss": 0.2039,
+      "step": 5173
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.440483728739274e-06,
+      "loss": 0.1558,
+      "step": 5174
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.437156785650722e-06,
+      "loss": 0.1379,
+      "step": 5175
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.433829905057562e-06,
+      "loss": 0.1968,
+      "step": 5176
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.430503087329204e-06,
+      "loss": 0.1213,
+      "step": 5177
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.427176332835033e-06,
+      "loss": 0.2161,
+      "step": 5178
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.423849641944443e-06,
+      "loss": 0.1833,
+      "step": 5179
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.420523015026809e-06,
+      "loss": 0.1528,
+      "step": 5180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.417196452451506e-06,
+      "loss": 0.1251,
+      "step": 5181
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.413869954587897e-06,
+      "loss": 0.1755,
+      "step": 5182
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.410543521805345e-06,
+      "loss": 0.1741,
+      "step": 5183
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.407217154473195e-06,
+      "loss": 0.2107,
+      "step": 5184
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.403890852960796e-06,
+      "loss": 0.1697,
+      "step": 5185
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.400564617637481e-06,
+      "loss": 0.1892,
+      "step": 5186
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39723844887258e-06,
+      "loss": 0.193,
+      "step": 5187
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.393912347035417e-06,
+      "loss": 0.1624,
+      "step": 5188
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.390586312495306e-06,
+      "loss": 0.1875,
+      "step": 5189
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.387260345621548e-06,
+      "loss": 0.1826,
+      "step": 5190
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.38393444678345e-06,
+      "loss": 0.1619,
+      "step": 5191
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.380608616350297e-06,
+      "loss": 0.2146,
+      "step": 5192
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37728285469138e-06,
+      "loss": 0.1709,
+      "step": 5193
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.37395716217597e-06,
+      "loss": 0.1746,
+      "step": 5194
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.370631539173343e-06,
+      "loss": 0.1743,
+      "step": 5195
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367305986052748e-06,
+      "loss": 0.1846,
+      "step": 5196
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.36398050318345e-06,
+      "loss": 0.1796,
+      "step": 5197
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.360655090934688e-06,
+      "loss": 0.1841,
+      "step": 5198
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.357329749675704e-06,
+      "loss": 0.1829,
+      "step": 5199
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.354004479775722e-06,
+      "loss": 0.1506,
+      "step": 5200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.350679281603972e-06,
+      "loss": 0.1602,
+      "step": 5201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.34735415552966e-06,
+      "loss": 0.1519,
+      "step": 5202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.344029101921999e-06,
+      "loss": 0.1736,
+      "step": 5203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.340704121150182e-06,
+      "loss": 0.173,
+      "step": 5204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.3373792135834e-06,
+      "loss": 0.175,
+      "step": 5205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.334054379590834e-06,
+      "loss": 0.1608,
+      "step": 5206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.330729619541661e-06,
+      "loss": 0.1548,
+      "step": 5207
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.327404933805041e-06,
+      "loss": 0.1958,
+      "step": 5208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.324080322750139e-06,
+      "loss": 0.188,
+      "step": 5209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.320755786746096e-06,
+      "loss": 0.1807,
+      "step": 5210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.317431326162054e-06,
+      "loss": 0.1462,
+      "step": 5211
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.314106941367148e-06,
+      "loss": 0.1582,
+      "step": 5212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.310782632730498e-06,
+      "loss": 0.175,
+      "step": 5213
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.307458400621224e-06,
+      "loss": 0.1316,
+      "step": 5214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.304134245408427e-06,
+      "loss": 0.1616,
+      "step": 5215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.300810167461209e-06,
+      "loss": 0.1755,
+      "step": 5216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297486167148657e-06,
+      "loss": 0.2131,
+      "step": 5217
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.294162244839857e-06,
+      "loss": 0.1958,
+      "step": 5218
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.290838400903874e-06,
+      "loss": 0.1349,
+      "step": 5219
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.28751463570978e-06,
+      "loss": 0.1462,
+      "step": 5220
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.284190949626624e-06,
+      "loss": 0.1902,
+      "step": 5221
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.280867343023452e-06,
+      "loss": 0.1639,
+      "step": 5222
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.277543816269302e-06,
+      "loss": 0.165,
+      "step": 5223
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.274220369733205e-06,
+      "loss": 0.1576,
+      "step": 5224
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.270897003784175e-06,
+      "loss": 0.1792,
+      "step": 5225
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.267573718791233e-06,
+      "loss": 0.1965,
+      "step": 5226
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.264250515123365e-06,
+      "loss": 0.1653,
+      "step": 5227
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.260927393149579e-06,
+      "loss": 0.1373,
+      "step": 5228
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.257604353238848e-06,
+      "loss": 0.178,
+      "step": 5229
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.254281395760151e-06,
+      "loss": 0.1416,
+      "step": 5230
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250958521082452e-06,
+      "loss": 0.188,
+      "step": 5231
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.247635729574707e-06,
+      "loss": 0.2041,
+      "step": 5232
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244313021605862e-06,
+      "loss": 0.2031,
+      "step": 5233
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.240990397544859e-06,
+      "loss": 0.1755,
+      "step": 5234
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.237667857760617e-06,
+      "loss": 0.1921,
+      "step": 5235
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234345402622065e-06,
+      "loss": 0.1692,
+      "step": 5236
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231023032498103e-06,
+      "loss": 0.1442,
+      "step": 5237
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227700747757638e-06,
+      "loss": 0.1768,
+      "step": 5238
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.224378548769557e-06,
+      "loss": 0.2139,
+      "step": 5239
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.221056435902747e-06,
+      "loss": 0.1796,
+      "step": 5240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.217734409526067e-06,
+      "loss": 0.1512,
+      "step": 5241
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.214412470008392e-06,
+      "loss": 0.1743,
+      "step": 5242
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.211090617718563e-06,
+      "loss": 0.165,
+      "step": 5243
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.207768853025432e-06,
+      "loss": 0.1544,
+      "step": 5244
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.204447176297826e-06,
+      "loss": 0.151,
+      "step": 5245
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201125587904572e-06,
+      "loss": 0.1483,
+      "step": 5246
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.197804088214475e-06,
+      "loss": 0.1831,
+      "step": 5247
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19448267759635e-06,
+      "loss": 0.1998,
+      "step": 5248
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19116135641898e-06,
+      "loss": 0.2192,
+      "step": 5249
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.187840125051155e-06,
+      "loss": 0.1472,
+      "step": 5250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.184518983861646e-06,
+      "loss": 0.1465,
+      "step": 5251
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.181197933219218e-06,
+      "loss": 0.1361,
+      "step": 5252
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.177876973492618e-06,
+      "loss": 0.2034,
+      "step": 5253
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.174556105050603e-06,
+      "loss": 0.1917,
+      "step": 5254
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.171235328261893e-06,
+      "loss": 0.1868,
+      "step": 5255
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.167914643495217e-06,
+      "loss": 0.1594,
+      "step": 5256
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.164594051119286e-06,
+      "loss": 0.1892,
+      "step": 5257
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.161273551502804e-06,
+      "loss": 0.21,
+      "step": 5258
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.15795314501446e-06,
+      "loss": 0.2092,
+      "step": 5259
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154632832022945e-06,
+      "loss": 0.1698,
+      "step": 5260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.15131261289692e-06,
+      "loss": 0.1637,
+      "step": 5261
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14799248800505e-06,
+      "loss": 0.1593,
+      "step": 5262
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.144672457715985e-06,
+      "loss": 0.197,
+      "step": 5263
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.141352522398369e-06,
+      "loss": 0.2146,
+      "step": 5264
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.138032682420824e-06,
+      "loss": 0.1792,
+      "step": 5265
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13471293815198e-06,
+      "loss": 0.1573,
+      "step": 5266
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.131393289960435e-06,
+      "loss": 0.1576,
+      "step": 5267
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.128073738214792e-06,
+      "loss": 0.1698,
+      "step": 5268
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.124754283283634e-06,
+      "loss": 0.2034,
+      "step": 5269
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.121434925535544e-06,
+      "loss": 0.2074,
+      "step": 5270
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.11811566533908e-06,
+      "loss": 0.1573,
+      "step": 5271
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.114796503062805e-06,
+      "loss": 0.2229,
+      "step": 5272
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.111477439075252e-06,
+      "loss": 0.1689,
+      "step": 5273
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.108158473744965e-06,
+      "loss": 0.2002,
+      "step": 5274
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.104839607440457e-06,
+      "loss": 0.2185,
+      "step": 5275
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.101520840530245e-06,
+      "loss": 0.199,
+      "step": 5276
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.098202173382829e-06,
+      "loss": 0.1719,
+      "step": 5277
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.094883606366689e-06,
+      "loss": 0.1763,
+      "step": 5278
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.091565139850315e-06,
+      "loss": 0.1726,
+      "step": 5279
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.088246774202162e-06,
+      "loss": 0.1968,
+      "step": 5280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.084928509790696e-06,
+      "loss": 0.1775,
+      "step": 5281
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.08161034698435e-06,
+      "loss": 0.1577,
+      "step": 5282
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.07829228615157e-06,
+      "loss": 0.208,
+      "step": 5283
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.074974327660763e-06,
+      "loss": 0.1741,
+      "step": 5284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.071656471880352e-06,
+      "loss": 0.1641,
+      "step": 5285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.068338719178726e-06,
+      "loss": 0.1726,
+      "step": 5286
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.06502106992428e-06,
+      "loss": 0.1675,
+      "step": 5287
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.061703524485383e-06,
+      "loss": 0.1733,
+      "step": 5288
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.058386083230404e-06,
+      "loss": 0.1354,
+      "step": 5289
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.05506874652769e-06,
+      "loss": 0.1829,
+      "step": 5290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.051751514745594e-06,
+      "loss": 0.1736,
+      "step": 5291
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.048434388252431e-06,
+      "loss": 0.1929,
+      "step": 5292
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.045117367416527e-06,
+      "loss": 0.199,
+      "step": 5293
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.041800452606186e-06,
+      "loss": 0.1511,
+      "step": 5294
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.038483644189704e-06,
+      "loss": 0.1758,
+      "step": 5295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.035166942535358e-06,
+      "loss": 0.1589,
+      "step": 5296
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.031850348011429e-06,
+      "loss": 0.2061,
+      "step": 5297
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.02853386098616e-06,
+      "loss": 0.1415,
+      "step": 5298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.025217481827815e-06,
+      "loss": 0.1588,
+      "step": 5299
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.021901210904616e-06,
+      "loss": 0.178,
+      "step": 5300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.018585048584793e-06,
+      "loss": 0.1714,
+      "step": 5301
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.015268995236551e-06,
+      "loss": 0.1969,
+      "step": 5302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.011953051228096e-06,
+      "loss": 0.145,
+      "step": 5303
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.008637216927604e-06,
+      "loss": 0.1653,
+      "step": 5304
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00532149270326e-06,
+      "loss": 0.1909,
+      "step": 5305
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00200587892322e-06,
+      "loss": 0.1494,
+      "step": 5306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.998690375955635e-06,
+      "loss": 0.116,
+      "step": 5307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.99537498416864e-06,
+      "loss": 0.1624,
+      "step": 5308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.992059703930366e-06,
+      "loss": 0.1367,
+      "step": 5309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.988744535608919e-06,
+      "loss": 0.1758,
+      "step": 5310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.985429479572407e-06,
+      "loss": 0.1636,
+      "step": 5311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.982114536188911e-06,
+      "loss": 0.1886,
+      "step": 5312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.978799705826509e-06,
+      "loss": 0.2118,
+      "step": 5313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.975484988853263e-06,
+      "loss": 0.1667,
+      "step": 5314
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.972170385637224e-06,
+      "loss": 0.1704,
+      "step": 5315
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.96885589654643e-06,
+      "loss": 0.1851,
+      "step": 5316
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.965541521948907e-06,
+      "loss": 0.1759,
+      "step": 5317
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.96222726221266e-06,
+      "loss": 0.1597,
+      "step": 5318
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9589131177057e-06,
+      "loss": 0.1632,
+      "step": 5319
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.955599088796002e-06,
+      "loss": 0.1215,
+      "step": 5320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.952285175851548e-06,
+      "loss": 0.1548,
+      "step": 5321
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.948971379240294e-06,
+      "loss": 0.155,
+      "step": 5322
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.945657699330195e-06,
+      "loss": 0.158,
+      "step": 5323
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.942344136489174e-06,
+      "loss": 0.1444,
+      "step": 5324
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.939030691085168e-06,
+      "loss": 0.1846,
+      "step": 5325
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.935717363486074e-06,
+      "loss": 0.1805,
+      "step": 5326
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.932404154059793e-06,
+      "loss": 0.1826,
+      "step": 5327
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.929091063174205e-06,
+      "loss": 0.2217,
+      "step": 5328
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.925778091197183e-06,
+      "loss": 0.1655,
+      "step": 5329
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.92246523849658e-06,
+      "loss": 0.1699,
+      "step": 5330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.919152505440248e-06,
+      "loss": 0.1504,
+      "step": 5331
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.915839892396004e-06,
+      "loss": 0.1705,
+      "step": 5332
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.912527399731674e-06,
+      "loss": 0.1726,
+      "step": 5333
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.909215027815053e-06,
+      "loss": 0.1442,
+      "step": 5334
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90590277701394e-06,
+      "loss": 0.1787,
+      "step": 5335
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.902590647696104e-06,
+      "loss": 0.1929,
+      "step": 5336
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.899278640229316e-06,
+      "loss": 0.1625,
+      "step": 5337
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.895966754981316e-06,
+      "loss": 0.1394,
+      "step": 5338
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.892654992319846e-06,
+      "loss": 0.1904,
+      "step": 5339
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.889343352612624e-06,
+      "loss": 0.1914,
+      "step": 5340
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.886031836227362e-06,
+      "loss": 0.1863,
+      "step": 5341
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.882720443531752e-06,
+      "loss": 0.1716,
+      "step": 5342
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87940917489348e-06,
+      "loss": 0.1892,
+      "step": 5343
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.87609803068021e-06,
+      "loss": 0.1653,
+      "step": 5344
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.872787011259592e-06,
+      "loss": 0.1415,
+      "step": 5345
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86947611699927e-06,
+      "loss": 0.1678,
+      "step": 5346
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.866165348266869e-06,
+      "loss": 0.2012,
+      "step": 5347
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.862854705430003e-06,
+      "loss": 0.1731,
+      "step": 5348
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.859544188856264e-06,
+      "loss": 0.1748,
+      "step": 5349
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.856233798913241e-06,
+      "loss": 0.1661,
+      "step": 5350
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.852923535968499e-06,
+      "loss": 0.167,
+      "step": 5351
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.849613400389599e-06,
+      "loss": 0.1863,
+      "step": 5352
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.846303392544077e-06,
+      "loss": 0.1293,
+      "step": 5353
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.84299351279947e-06,
+      "loss": 0.2141,
+      "step": 5354
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.839683761523277e-06,
+      "loss": 0.1721,
+      "step": 5355
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.836374139083011e-06,
+      "loss": 0.1882,
+      "step": 5356
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.833064645846144e-06,
+      "loss": 0.1604,
+      "step": 5357
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.829755282180155e-06,
+      "loss": 0.1406,
+      "step": 5358
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.826446048452494e-06,
+      "loss": 0.2031,
+      "step": 5359
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.823136945030607e-06,
+      "loss": 0.1824,
+      "step": 5360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.819827972281917e-06,
+      "loss": 0.1344,
+      "step": 5361
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.816519130573844e-06,
+      "loss": 0.188,
+      "step": 5362
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.813210420273775e-06,
+      "loss": 0.1447,
+      "step": 5363
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.809901841749101e-06,
+      "loss": 0.2178,
+      "step": 5364
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.806593395367186e-06,
+      "loss": 0.1746,
+      "step": 5365
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.803285081495388e-06,
+      "loss": 0.1569,
+      "step": 5366
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.799976900501041e-06,
+      "loss": 0.1766,
+      "step": 5367
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.796668852751476e-06,
+      "loss": 0.175,
+      "step": 5368
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.793360938613997e-06,
+      "loss": 0.2122,
+      "step": 5369
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.790053158455904e-06,
+      "loss": 0.1326,
+      "step": 5370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78674551264447e-06,
+      "loss": 0.1484,
+      "step": 5371
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.783438001546967e-06,
+      "loss": 0.1855,
+      "step": 5372
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.78013062553064e-06,
+      "loss": 0.2124,
+      "step": 5373
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.77682338496273e-06,
+      "loss": 0.1379,
+      "step": 5374
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.773516280210448e-06,
+      "loss": 0.1899,
+      "step": 5375
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.77020931164101e-06,
+      "loss": 0.1987,
+      "step": 5376
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.766902479621596e-06,
+      "loss": 0.1663,
+      "step": 5377
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.763595784519388e-06,
+      "loss": 0.1572,
+      "step": 5378
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.760289226701538e-06,
+      "loss": 0.1825,
+      "step": 5379
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756982806535201e-06,
+      "loss": 0.1356,
+      "step": 5380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.753676524387491e-06,
+      "loss": 0.1855,
+      "step": 5381
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.750370380625539e-06,
+      "loss": 0.1731,
+      "step": 5382
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.747064375616428e-06,
+      "loss": 0.1926,
+      "step": 5383
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.74375850972725e-06,
+      "loss": 0.1558,
+      "step": 5384
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.740452783325069e-06,
+      "loss": 0.1824,
+      "step": 5385
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.737147196776938e-06,
+      "loss": 0.1979,
+      "step": 5386
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.73384175044989e-06,
+      "loss": 0.1919,
+      "step": 5387
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.730536444710955e-06,
+      "loss": 0.1541,
+      "step": 5388
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.727231279927128e-06,
+      "loss": 0.1627,
+      "step": 5389
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.723926256465402e-06,
+      "loss": 0.161,
+      "step": 5390
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.720621374692752e-06,
+      "loss": 0.1738,
+      "step": 5391
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.717316634976137e-06,
+      "loss": 0.1936,
+      "step": 5392
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.714012037682496e-06,
+      "loss": 0.188,
+      "step": 5393
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71070758317876e-06,
+      "loss": 0.1478,
+      "step": 5394
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.707403271831835e-06,
+      "loss": 0.1692,
+      "step": 5395
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.70409910400862e-06,
+      "loss": 0.1489,
+      "step": 5396
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.700795080075989e-06,
+      "loss": 0.1459,
+      "step": 5397
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.697491200400809e-06,
+      "loss": 0.1531,
+      "step": 5398
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.694187465349925e-06,
+      "loss": 0.1821,
+      "step": 5399
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.690883875290173e-06,
+      "loss": 0.158,
+      "step": 5400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.687580430588355e-06,
+      "loss": 0.1892,
+      "step": 5401
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.684277131611284e-06,
+      "loss": 0.1699,
+      "step": 5402
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.680973978725733e-06,
+      "loss": 0.1569,
+      "step": 5403
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.677670972298472e-06,
+      "loss": 0.186,
+      "step": 5404
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.67436811269625e-06,
+      "loss": 0.2004,
+      "step": 5405
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.671065400285802e-06,
+      "loss": 0.1592,
+      "step": 5406
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.667762835433842e-06,
+      "loss": 0.1753,
+      "step": 5407
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.664460418507077e-06,
+      "loss": 0.1816,
+      "step": 5408
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.661158149872183e-06,
+      "loss": 0.1567,
+      "step": 5409
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.657856029895835e-06,
+      "loss": 0.1638,
+      "step": 5410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.654554058944683e-06,
+      "loss": 0.1448,
+      "step": 5411
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.651252237385354e-06,
+      "loss": 0.1851,
+      "step": 5412
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.647950565584483e-06,
+      "loss": 0.1414,
+      "step": 5413
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.644649043908652e-06,
+      "loss": 0.1671,
+      "step": 5414
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.641347672724463e-06,
+      "loss": 0.228,
+      "step": 5415
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.638046452398473e-06,
+      "loss": 0.1688,
+      "step": 5416
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.634745383297239e-06,
+      "loss": 0.1611,
+      "step": 5417
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.631444465787292e-06,
+      "loss": 0.1731,
+      "step": 5418
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.628143700235159e-06,
+      "loss": 0.2349,
+      "step": 5419
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.62484308700733e-06,
+      "loss": 0.16,
+      "step": 5420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.621542626470295e-06,
+      "loss": 0.197,
+      "step": 5421
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.618242318990517e-06,
+      "loss": 0.1749,
+      "step": 5422
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.614942164934453e-06,
+      "loss": 0.1494,
+      "step": 5423
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.61164216466853e-06,
+      "loss": 0.1594,
+      "step": 5424
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.608342318559171e-06,
+      "loss": 0.1855,
+      "step": 5425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.605042626972766e-06,
+      "loss": 0.1423,
+      "step": 5426
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.601743090275705e-06,
+      "loss": 0.2153,
+      "step": 5427
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.598443708834345e-06,
+      "loss": 0.1658,
+      "step": 5428
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.595144483015038e-06,
+      "loss": 0.1503,
+      "step": 5429
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.591845413184114e-06,
+      "loss": 0.152,
+      "step": 5430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58854649970789e-06,
+      "loss": 0.1633,
+      "step": 5431
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.58524774295265e-06,
+      "loss": 0.2031,
+      "step": 5432
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.581949143284685e-06,
+      "loss": 0.1846,
+      "step": 5433
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.578650701070244e-06,
+      "loss": 0.1621,
+      "step": 5434
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.57535241667558e-06,
+      "loss": 0.1555,
+      "step": 5435
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.572054290466911e-06,
+      "loss": 0.1821,
+      "step": 5436
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.56875632281045e-06,
+      "loss": 0.1941,
+      "step": 5437
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.565458514072385e-06,
+      "loss": 0.1396,
+      "step": 5438
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.56216086461889e-06,
+      "loss": 0.1614,
+      "step": 5439
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55886337481612e-06,
+      "loss": 0.177,
+      "step": 5440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.555566045030211e-06,
+      "loss": 0.1973,
+      "step": 5441
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.552268875627284e-06,
+      "loss": 0.1428,
+      "step": 5442
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.548971866973439e-06,
+      "loss": 0.2043,
+      "step": 5443
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.545675019434761e-06,
+      "loss": 0.1273,
+      "step": 5444
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.54237833337732e-06,
+      "loss": 0.1625,
+      "step": 5445
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.539081809167157e-06,
+      "loss": 0.1534,
+      "step": 5446
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.535785447170307e-06,
+      "loss": 0.1554,
+      "step": 5447
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.532489247752778e-06,
+      "loss": 0.1615,
+      "step": 5448
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.529193211280572e-06,
+      "loss": 0.249,
+      "step": 5449
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.525897338119655e-06,
+      "loss": 0.1738,
+      "step": 5450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.522601628635993e-06,
+      "loss": 0.1792,
+      "step": 5451
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.519306083195518e-06,
+      "loss": 0.1372,
+      "step": 5452
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.516010702164162e-06,
+      "loss": 0.1443,
+      "step": 5453
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.512715485907816e-06,
+      "loss": 0.1383,
+      "step": 5454
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.509420434792374e-06,
+      "loss": 0.1689,
+      "step": 5455
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.506125549183698e-06,
+      "loss": 0.2053,
+      "step": 5456
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.502830829447642e-06,
+      "loss": 0.1682,
+      "step": 5457
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.499536275950025e-06,
+      "loss": 0.1497,
+      "step": 5458
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.496241889056671e-06,
+      "loss": 0.1682,
+      "step": 5459
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.492947669133362e-06,
+      "loss": 0.1785,
+      "step": 5460
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.489653616545881e-06,
+      "loss": 0.1721,
+      "step": 5461
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.486359731659977e-06,
+      "loss": 0.1736,
+      "step": 5462
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.48306601484139e-06,
+      "loss": 0.139,
+      "step": 5463
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.479772466455838e-06,
+      "loss": 0.1619,
+      "step": 5464
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.476479086869024e-06,
+      "loss": 0.198,
+      "step": 5465
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.473185876446623e-06,
+      "loss": 0.1351,
+      "step": 5466
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.469892835554302e-06,
+      "loss": 0.1584,
+      "step": 5467
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.4665999645577e-06,
+      "loss": 0.1638,
+      "step": 5468
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.463307263822447e-06,
+      "loss": 0.1665,
+      "step": 5469
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.460014733714144e-06,
+      "loss": 0.1647,
+      "step": 5470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.456722374598381e-06,
+      "loss": 0.1394,
+      "step": 5471
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.453430186840721e-06,
+      "loss": 0.1729,
+      "step": 5472
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.45013817080672e-06,
+      "loss": 0.1488,
+      "step": 5473
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.446846326861899e-06,
+      "loss": 0.1766,
+      "step": 5474
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.443554655371776e-06,
+      "loss": 0.22,
+      "step": 5475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.440263156701835e-06,
+      "loss": 0.1556,
+      "step": 5476
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43697183121756e-06,
+      "loss": 0.1732,
+      "step": 5477
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.43368067928439e-06,
+      "loss": 0.1533,
+      "step": 5478
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.430389701267765e-06,
+      "loss": 0.1868,
+      "step": 5479
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.427098897533101e-06,
+      "loss": 0.1792,
+      "step": 5480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.42380826844579e-06,
+      "loss": 0.1628,
+      "step": 5481
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.420517814371211e-06,
+      "loss": 0.1603,
+      "step": 5482
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.417227535674712e-06,
+      "loss": 0.1353,
+      "step": 5483
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.413937432721644e-06,
+      "loss": 0.2031,
+      "step": 5484
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.410647505877308e-06,
+      "loss": 0.15,
+      "step": 5485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.407357755507013e-06,
+      "loss": 0.1846,
+      "step": 5486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.404068181976032e-06,
+      "loss": 0.1577,
+      "step": 5487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.400778785649626e-06,
+      "loss": 0.1825,
+      "step": 5488
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39748956689303e-06,
+      "loss": 0.2026,
+      "step": 5489
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39420052607147e-06,
+      "loss": 0.1823,
+      "step": 5490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.390911663550137e-06,
+      "loss": 0.1611,
+      "step": 5491
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.387622979694215e-06,
+      "loss": 0.1249,
+      "step": 5492
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38433447486886e-06,
+      "loss": 0.166,
+      "step": 5493
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.38104614943922e-06,
+      "loss": 0.1653,
+      "step": 5494
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.377758003770404e-06,
+      "loss": 0.1785,
+      "step": 5495
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.374470038227522e-06,
+      "loss": 0.1577,
+      "step": 5496
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.371182253175645e-06,
+      "loss": 0.2163,
+      "step": 5497
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36789464897984e-06,
+      "loss": 0.166,
+      "step": 5498
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.36460722600514e-06,
+      "loss": 0.1897,
+      "step": 5499
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.361319984616568e-06,
+      "loss": 0.1755,
+      "step": 5500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.358032925179124e-06,
+      "loss": 0.1543,
+      "step": 5501
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.35474604805779e-06,
+      "loss": 0.1602,
+      "step": 5502
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.351459353617515e-06,
+      "loss": 0.1768,
+      "step": 5503
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348172842223251e-06,
+      "loss": 0.1536,
+      "step": 5504
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.344886514239906e-06,
+      "loss": 0.1692,
+      "step": 5505
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.341600370032384e-06,
+      "loss": 0.1616,
+      "step": 5506
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.338314409965557e-06,
+      "loss": 0.198,
+      "step": 5507
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.335028634404292e-06,
+      "loss": 0.1958,
+      "step": 5508
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.33174304371341e-06,
+      "loss": 0.127,
+      "step": 5509
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.328457638257743e-06,
+      "loss": 0.1614,
+      "step": 5510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.325172418402076e-06,
+      "loss": 0.1864,
+      "step": 5511
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.321887384511191e-06,
+      "loss": 0.1746,
+      "step": 5512
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.318602536949837e-06,
+      "loss": 0.1873,
+      "step": 5513
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.315317876082752e-06,
+      "loss": 0.189,
+      "step": 5514
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.312033402274644e-06,
+      "loss": 0.1671,
+      "step": 5515
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.308749115890212e-06,
+      "loss": 0.1787,
+      "step": 5516
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30546501729412e-06,
+      "loss": 0.179,
+      "step": 5517
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.302181106851022e-06,
+      "loss": 0.2349,
+      "step": 5518
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.298897384925547e-06,
+      "loss": 0.1622,
+      "step": 5519
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.295613851882305e-06,
+      "loss": 0.1934,
+      "step": 5520
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.29233050808588e-06,
+      "loss": 0.1592,
+      "step": 5521
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.289047353900847e-06,
+      "loss": 0.171,
+      "step": 5522
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.28576438969174e-06,
+      "loss": 0.1489,
+      "step": 5523
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.282481615823092e-06,
+      "loss": 0.1434,
+      "step": 5524
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.279199032659403e-06,
+      "loss": 0.1599,
+      "step": 5525
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.275916640565157e-06,
+      "loss": 0.1804,
+      "step": 5526
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.272634439904812e-06,
+      "loss": 0.1575,
+      "step": 5527
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.269352431042813e-06,
+      "loss": 0.1631,
+      "step": 5528
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.266070614343569e-06,
+      "loss": 0.1776,
+      "step": 5529
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.26278899017149e-06,
+      "loss": 0.1227,
+      "step": 5530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.259507558890941e-06,
+      "loss": 0.1897,
+      "step": 5531
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.256226320866282e-06,
+      "loss": 0.1681,
+      "step": 5532
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.252945276461842e-06,
+      "loss": 0.1752,
+      "step": 5533
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.249664426041936e-06,
+      "loss": 0.2007,
+      "step": 5534
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.246383769970852e-06,
+      "loss": 0.1677,
+      "step": 5535
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24310330861286e-06,
+      "loss": 0.1833,
+      "step": 5536
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.239823042332203e-06,
+      "loss": 0.1428,
+      "step": 5537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23654297149311e-06,
+      "loss": 0.1716,
+      "step": 5538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23326309645978e-06,
+      "loss": 0.1686,
+      "step": 5539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.229983417596399e-06,
+      "loss": 0.1344,
+      "step": 5540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.226703935267121e-06,
+      "loss": 0.2041,
+      "step": 5541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.223424649836093e-06,
+      "loss": 0.1486,
+      "step": 5542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.220145561667421e-06,
+      "loss": 0.175,
+      "step": 5543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.216866671125205e-06,
+      "loss": 0.2026,
+      "step": 5544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.213587978573517e-06,
+      "loss": 0.2229,
+      "step": 5545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.210309484376404e-06,
+      "loss": 0.1541,
+      "step": 5546
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.2070311888979e-06,
+      "loss": 0.1824,
+      "step": 5547
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.203753092502002e-06,
+      "loss": 0.1423,
+      "step": 5548
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.200475195552703e-06,
+      "loss": 0.1785,
+      "step": 5549
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.197197498413958e-06,
+      "loss": 0.1692,
+      "step": 5550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.193920001449713e-06,
+      "loss": 0.182,
+      "step": 5551
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.19064270502388e-06,
+      "loss": 0.158,
+      "step": 5552
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18736560950036e-06,
+      "loss": 0.1514,
+      "step": 5553
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.184088715243019e-06,
+      "loss": 0.1699,
+      "step": 5554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.180812022615714e-06,
+      "loss": 0.188,
+      "step": 5555
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.177535531982266e-06,
+      "loss": 0.2012,
+      "step": 5556
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.174259243706488e-06,
+      "loss": 0.1602,
+      "step": 5557
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.17098315815216e-06,
+      "loss": 0.1682,
+      "step": 5558
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.167707275683043e-06,
+      "loss": 0.2078,
+      "step": 5559
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.164431596662872e-06,
+      "loss": 0.1567,
+      "step": 5560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.16115612145537e-06,
+      "loss": 0.1882,
+      "step": 5561
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.157880850424222e-06,
+      "loss": 0.1738,
+      "step": 5562
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.154605783933104e-06,
+      "loss": 0.1763,
+      "step": 5563
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.15133092234566e-06,
+      "loss": 0.1785,
+      "step": 5564
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.148056266025517e-06,
+      "loss": 0.1442,
+      "step": 5565
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.144781815336276e-06,
+      "loss": 0.1666,
+      "step": 5566
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.141507570641522e-06,
+      "loss": 0.1936,
+      "step": 5567
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.1382335323048e-06,
+      "loss": 0.1887,
+      "step": 5568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.13495970068965e-06,
+      "loss": 0.1475,
+      "step": 5569
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.131686076159582e-06,
+      "loss": 0.2268,
+      "step": 5570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.128412659078084e-06,
+      "loss": 0.1442,
+      "step": 5571
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.125139449808618e-06,
+      "loss": 0.2153,
+      "step": 5572
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.121866448714633e-06,
+      "loss": 0.1848,
+      "step": 5573
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.118593656159536e-06,
+      "loss": 0.1791,
+      "step": 5574
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.11532107250673e-06,
+      "loss": 0.1714,
+      "step": 5575
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.11204869811958e-06,
+      "loss": 0.1729,
+      "step": 5576
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.108776533361445e-06,
+      "loss": 0.2173,
+      "step": 5577
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10550457859564e-06,
+      "loss": 0.1509,
+      "step": 5578
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.102232834185478e-06,
+      "loss": 0.1686,
+      "step": 5579
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.098961300494221e-06,
+      "loss": 0.1855,
+      "step": 5580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.095689977885144e-06,
+      "loss": 0.1559,
+      "step": 5581
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.092418866721466e-06,
+      "loss": 0.1704,
+      "step": 5582
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.0891479673664e-06,
+      "loss": 0.1675,
+      "step": 5583
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.085877280183126e-06,
+      "loss": 0.1665,
+      "step": 5584
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.082606805534817e-06,
+      "loss": 0.1622,
+      "step": 5585
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.079336543784592e-06,
+      "loss": 0.1642,
+      "step": 5586
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076066495295586e-06,
+      "loss": 0.1549,
+      "step": 5587
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.072796660430873e-06,
+      "loss": 0.2036,
+      "step": 5588
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.06952703955353e-06,
+      "loss": 0.1736,
+      "step": 5589
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.066257633026594e-06,
+      "loss": 0.166,
+      "step": 5590
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.062988441213087e-06,
+      "loss": 0.1655,
+      "step": 5591
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.059719464476e-06,
+      "loss": 0.1667,
+      "step": 5592
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.056450703178316e-06,
+      "loss": 0.207,
+      "step": 5593
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053182157682968e-06,
+      "loss": 0.166,
+      "step": 5594
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.049913828352889e-06,
+      "loss": 0.1807,
+      "step": 5595
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.046645715550972e-06,
+      "loss": 0.1545,
+      "step": 5596
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.043377819640099e-06,
+      "loss": 0.1499,
+      "step": 5597
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040110140983115e-06,
+      "loss": 0.188,
+      "step": 5598
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.036842679942856e-06,
+      "loss": 0.1892,
+      "step": 5599
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.033575436882115e-06,
+      "loss": 0.2205,
+      "step": 5600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03030841216368e-06,
+      "loss": 0.1428,
+      "step": 5601
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.027041606150296e-06,
+      "loss": 0.1244,
+      "step": 5602
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.023775019204703e-06,
+      "loss": 0.1565,
+      "step": 5603
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.0205086516896e-06,
+      "loss": 0.1685,
+      "step": 5604
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.017242503967675e-06,
+      "loss": 0.1714,
+      "step": 5605
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.013976576401575e-06,
+      "loss": 0.1705,
+      "step": 5606
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.010710869353948e-06,
+      "loss": 0.1393,
+      "step": 5607
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.007445383187387e-06,
+      "loss": 0.1451,
+      "step": 5608
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.004180118264487e-06,
+      "loss": 0.1899,
+      "step": 5609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.000915074947802e-06,
+      "loss": 0.167,
+      "step": 5610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.997650253599868e-06,
+      "loss": 0.1553,
+      "step": 5611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.994385654583192e-06,
+      "loss": 0.1475,
+      "step": 5612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.991121278260267e-06,
+      "loss": 0.1801,
+      "step": 5613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.98785712499355e-06,
+      "loss": 0.1366,
+      "step": 5614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.984593195145468e-06,
+      "loss": 0.1406,
+      "step": 5615
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.981329489078447e-06,
+      "loss": 0.1569,
+      "step": 5616
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.97806600715486e-06,
+      "loss": 0.1305,
+      "step": 5617
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.974802749737079e-06,
+      "loss": 0.1833,
+      "step": 5618
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.971539717187431e-06,
+      "loss": 0.1772,
+      "step": 5619
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.968276909868234e-06,
+      "loss": 0.1594,
+      "step": 5620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96501432814177e-06,
+      "loss": 0.1782,
+      "step": 5621
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.961751972370306e-06,
+      "loss": 0.1931,
+      "step": 5622
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.958489842916072e-06,
+      "loss": 0.2109,
+      "step": 5623
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.955227940141287e-06,
+      "loss": 0.1692,
+      "step": 5624
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.951966264408126e-06,
+      "loss": 0.1676,
+      "step": 5625
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.94870481607876e-06,
+      "loss": 0.1575,
+      "step": 5626
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.945443595515317e-06,
+      "loss": 0.1659,
+      "step": 5627
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.94218260307991e-06,
+      "loss": 0.1797,
+      "step": 5628
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.938921839134626e-06,
+      "loss": 0.1909,
+      "step": 5629
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.935661304041524e-06,
+      "loss": 0.1399,
+      "step": 5630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.932400998162635e-06,
+      "loss": 0.1738,
+      "step": 5631
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92914092185997e-06,
+      "loss": 0.1741,
+      "step": 5632
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92588107549551e-06,
+      "loss": 0.1869,
+      "step": 5633
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.922621459431216e-06,
+      "loss": 0.166,
+      "step": 5634
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.919362074029018e-06,
+      "loss": 0.1948,
+      "step": 5635
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.916102919650826e-06,
+      "loss": 0.2058,
+      "step": 5636
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91284399665851e-06,
+      "loss": 0.1636,
+      "step": 5637
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.90958530541394e-06,
+      "loss": 0.176,
+      "step": 5638
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.906326846278934e-06,
+      "loss": 0.1406,
+      "step": 5639
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.9030686196153e-06,
+      "loss": 0.1577,
+      "step": 5640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.899810625784815e-06,
+      "loss": 0.161,
+      "step": 5641
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.896552865149232e-06,
+      "loss": 0.1899,
+      "step": 5642
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.893295338070278e-06,
+      "loss": 0.1838,
+      "step": 5643
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.890038044909651e-06,
+      "loss": 0.1501,
+      "step": 5644
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.886780986029023e-06,
+      "loss": 0.1704,
+      "step": 5645
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.883524161790048e-06,
+      "loss": 0.1415,
+      "step": 5646
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.880267572554341e-06,
+      "loss": 0.1711,
+      "step": 5647
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.877011218683503e-06,
+      "loss": 0.1414,
+      "step": 5648
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.873755100539102e-06,
+      "loss": 0.1555,
+      "step": 5649
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.870499218482687e-06,
+      "loss": 0.1547,
+      "step": 5650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.867243572875766e-06,
+      "loss": 0.1688,
+      "step": 5651
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.863988164079837e-06,
+      "loss": 0.1721,
+      "step": 5652
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.86073299245636e-06,
+      "loss": 0.1591,
+      "step": 5653
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.857478058366778e-06,
+      "loss": 0.2424,
+      "step": 5654
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.854223362172499e-06,
+      "loss": 0.1829,
+      "step": 5655
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.850968904234918e-06,
+      "loss": 0.1702,
+      "step": 5656
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.847714684915379e-06,
+      "loss": 0.2029,
+      "step": 5657
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.84446070457523e-06,
+      "loss": 0.1207,
+      "step": 5658
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.841206963575767e-06,
+      "loss": 0.176,
+      "step": 5659
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.837953462278273e-06,
+      "loss": 0.1675,
+      "step": 5660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.834700201044002e-06,
+      "loss": 0.1597,
+      "step": 5661
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.831447180234182e-06,
+      "loss": 0.1431,
+      "step": 5662
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.828194400210007e-06,
+      "loss": 0.1492,
+      "step": 5663
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82494186133266e-06,
+      "loss": 0.1843,
+      "step": 5664
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.821689563963276e-06,
+      "loss": 0.1719,
+      "step": 5665
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.818437508462981e-06,
+      "loss": 0.1776,
+      "step": 5666
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.815185695192862e-06,
+      "loss": 0.1694,
+      "step": 5667
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.811934124513994e-06,
+      "loss": 0.1809,
+      "step": 5668
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.808682796787406e-06,
+      "loss": 0.1802,
+      "step": 5669
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.805431712374119e-06,
+      "loss": 0.1953,
+      "step": 5670
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.802180871635107e-06,
+      "loss": 0.1735,
+      "step": 5671
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.798930274931338e-06,
+      "loss": 0.1621,
+      "step": 5672
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.795679922623734e-06,
+      "loss": 0.1407,
+      "step": 5673
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.792429815073206e-06,
+      "loss": 0.1512,
+      "step": 5674
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.789179952640625e-06,
+      "loss": 0.1707,
+      "step": 5675
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.785930335686845e-06,
+      "loss": 0.156,
+      "step": 5676
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.782680964572676e-06,
+      "loss": 0.1858,
+      "step": 5677
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.779431839658931e-06,
+      "loss": 0.1572,
+      "step": 5678
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.776182961306361e-06,
+      "loss": 0.2314,
+      "step": 5679
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.772934329875714e-06,
+      "loss": 0.1768,
+      "step": 5680
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.769685945727703e-06,
+      "loss": 0.2063,
+      "step": 5681
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.766437809223008e-06,
+      "loss": 0.1885,
+      "step": 5682
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76318992072229e-06,
+      "loss": 0.1804,
+      "step": 5683
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.759942280586174e-06,
+      "loss": 0.1321,
+      "step": 5684
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.756694889175269e-06,
+      "loss": 0.1213,
+      "step": 5685
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.753447746850145e-06,
+      "loss": 0.1388,
+      "step": 5686
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.750200853971354e-06,
+      "loss": 0.2148,
+      "step": 5687
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.746954210899404e-06,
+      "loss": 0.1897,
+      "step": 5688
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.743707817994804e-06,
+      "loss": 0.1819,
+      "step": 5689
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.740461675618004e-06,
+      "loss": 0.1826,
+      "step": 5690
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.737215784129443e-06,
+      "loss": 0.1647,
+      "step": 5691
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.733970143889531e-06,
+      "loss": 0.1884,
+      "step": 5692
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.73072475525865e-06,
+      "loss": 0.1941,
+      "step": 5693
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.727479618597148e-06,
+      "loss": 0.167,
+      "step": 5694
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.724234734265355e-06,
+      "loss": 0.1672,
+      "step": 5695
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.72099010262356e-06,
+      "loss": 0.1418,
+      "step": 5696
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.717745724032036e-06,
+      "loss": 0.1755,
+      "step": 5697
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.714501598851021e-06,
+      "loss": 0.2148,
+      "step": 5698
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.711257727440729e-06,
+      "loss": 0.1731,
+      "step": 5699
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.708014110161342e-06,
+      "loss": 0.1187,
+      "step": 5700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.70477074737302e-06,
+      "loss": 0.1545,
+      "step": 5701
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.701527639435883e-06,
+      "loss": 0.1956,
+      "step": 5702
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.698284786710036e-06,
+      "loss": 0.1753,
+      "step": 5703
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.695042189555547e-06,
+      "loss": 0.1875,
+      "step": 5704
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.691799848332459e-06,
+      "loss": 0.1687,
+      "step": 5705
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.688557763400785e-06,
+      "loss": 0.177,
+      "step": 5706
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.685315935120517e-06,
+      "loss": 0.1917,
+      "step": 5707
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.682074363851598e-06,
+      "loss": 0.1727,
+      "step": 5708
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.678833049953972e-06,
+      "loss": 0.1654,
+      "step": 5709
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.675591993787528e-06,
+      "loss": 0.2312,
+      "step": 5710
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.672351195712142e-06,
+      "loss": 0.1633,
+      "step": 5711
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.669110656087655e-06,
+      "loss": 0.2161,
+      "step": 5712
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.665870375273885e-06,
+      "loss": 0.1614,
+      "step": 5713
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.662630353630606e-06,
+      "loss": 0.1741,
+      "step": 5714
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.65939059151759e-06,
+      "loss": 0.1792,
+      "step": 5715
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.656151089294553e-06,
+      "loss": 0.2039,
+      "step": 5716
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.652911847321199e-06,
+      "loss": 0.1523,
+      "step": 5717
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.649672865957194e-06,
+      "loss": 0.1523,
+      "step": 5718
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.646434145562183e-06,
+      "loss": 0.2043,
+      "step": 5719
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.643195686495773e-06,
+      "loss": 0.1689,
+      "step": 5720
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.639957489117555e-06,
+      "loss": 0.1444,
+      "step": 5721
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.636719553787073e-06,
+      "loss": 0.1602,
+      "step": 5722
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.633481880863859e-06,
+      "loss": 0.1533,
+      "step": 5723
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.630244470707404e-06,
+      "loss": 0.1633,
+      "step": 5724
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.627007323677177e-06,
+      "loss": 0.1063,
+      "step": 5725
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.623770440132613e-06,
+      "loss": 0.1378,
+      "step": 5726
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.620533820433126e-06,
+      "loss": 0.2031,
+      "step": 5727
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.617297464938086e-06,
+      "loss": 0.1455,
+      "step": 5728
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.614061374006848e-06,
+      "loss": 0.165,
+      "step": 5729
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.610825547998728e-06,
+      "loss": 0.1674,
+      "step": 5730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.607589987273022e-06,
+      "loss": 0.228,
+      "step": 5731
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.604354692188986e-06,
+      "loss": 0.1799,
+      "step": 5732
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.601119663105857e-06,
+      "loss": 0.2295,
+      "step": 5733
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.597884900382827e-06,
+      "loss": 0.1705,
+      "step": 5734
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.594650404379082e-06,
+      "loss": 0.1753,
+      "step": 5735
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.591416175453753e-06,
+      "loss": 0.1589,
+      "step": 5736
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.58818221396596e-06,
+      "loss": 0.1765,
+      "step": 5737
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.584948520274784e-06,
+      "loss": 0.2139,
+      "step": 5738
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.581715094739279e-06,
+      "loss": 0.1305,
+      "step": 5739
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.578481937718469e-06,
+      "loss": 0.1621,
+      "step": 5740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.575249049571352e-06,
+      "loss": 0.1848,
+      "step": 5741
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.572016430656884e-06,
+      "loss": 0.1768,
+      "step": 5742
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5687840813340076e-06,
+      "loss": 0.1548,
+      "step": 5743
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.56555200196162e-06,
+      "loss": 0.1797,
+      "step": 5744
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.562320192898604e-06,
+      "loss": 0.1763,
+      "step": 5745
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.559088654503796e-06,
+      "loss": 0.1714,
+      "step": 5746
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.555857387136018e-06,
+      "loss": 0.1843,
+      "step": 5747
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.552626391154047e-06,
+      "loss": 0.194,
+      "step": 5748
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5493956669166415e-06,
+      "loss": 0.1685,
+      "step": 5749
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5461652147825255e-06,
+      "loss": 0.1757,
+      "step": 5750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.54293503511039e-06,
+      "loss": 0.1829,
+      "step": 5751
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.539705128258904e-06,
+      "loss": 0.1946,
+      "step": 5752
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.536475494586695e-06,
+      "loss": 0.1843,
+      "step": 5753
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.533246134452368e-06,
+      "loss": 0.1895,
+      "step": 5754
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.530017048214495e-06,
+      "loss": 0.1316,
+      "step": 5755
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.526788236231622e-06,
+      "loss": 0.1738,
+      "step": 5756
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.523559698862254e-06,
+      "loss": 0.1692,
+      "step": 5757
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.520331436464881e-06,
+      "loss": 0.1606,
+      "step": 5758
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.517103449397944e-06,
+      "loss": 0.1785,
+      "step": 5759
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5138757380198714e-06,
+      "loss": 0.1794,
+      "step": 5760
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.510648302689045e-06,
+      "loss": 0.1948,
+      "step": 5761
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.507421143763831e-06,
+      "loss": 0.1703,
+      "step": 5762
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.504194261602553e-06,
+      "loss": 0.1376,
+      "step": 5763
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.500967656563513e-06,
+      "loss": 0.1655,
+      "step": 5764
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.497741329004968e-06,
+      "loss": 0.1997,
+      "step": 5765
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.494515279285166e-06,
+      "loss": 0.1503,
+      "step": 5766
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4912895077623025e-06,
+      "loss": 0.1279,
+      "step": 5767
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.488064014794558e-06,
+      "loss": 0.1565,
+      "step": 5768
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4848388007400705e-06,
+      "loss": 0.2048,
+      "step": 5769
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.481613865956958e-06,
+      "loss": 0.166,
+      "step": 5770
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.478389210803296e-06,
+      "loss": 0.2131,
+      "step": 5771
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.475164835637141e-06,
+      "loss": 0.126,
+      "step": 5772
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.471940740816504e-06,
+      "loss": 0.1802,
+      "step": 5773
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.46871692669938e-06,
+      "loss": 0.1887,
+      "step": 5774
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.465493393643719e-06,
+      "loss": 0.2319,
+      "step": 5775
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.462270142007455e-06,
+      "loss": 0.1276,
+      "step": 5776
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.459047172148474e-06,
+      "loss": 0.1539,
+      "step": 5777
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.455824484424647e-06,
+      "loss": 0.2175,
+      "step": 5778
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4526020791937995e-06,
+      "loss": 0.2207,
+      "step": 5779
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4493799568137335e-06,
+      "loss": 0.1387,
+      "step": 5780
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.446158117642218e-06,
+      "loss": 0.1416,
+      "step": 5781
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.442936562036993e-06,
+      "loss": 0.1902,
+      "step": 5782
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.439715290355759e-06,
+      "loss": 0.1281,
+      "step": 5783
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.436494302956198e-06,
+      "loss": 0.1899,
+      "step": 5784
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.433273600195943e-06,
+      "loss": 0.1841,
+      "step": 5785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.430053182432617e-06,
+      "loss": 0.1599,
+      "step": 5786
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.426833050023791e-06,
+      "loss": 0.1729,
+      "step": 5787
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.423613203327016e-06,
+      "loss": 0.1565,
+      "step": 5788
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4203936426998065e-06,
+      "loss": 0.1539,
+      "step": 5789
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.417174368499653e-06,
+      "loss": 0.1653,
+      "step": 5790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.413955381083997e-06,
+      "loss": 0.1765,
+      "step": 5791
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.410736680810272e-06,
+      "loss": 0.1836,
+      "step": 5792
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.407518268035857e-06,
+      "loss": 0.1738,
+      "step": 5793
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.404300143118115e-06,
+      "loss": 0.175,
+      "step": 5794
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.401082306414367e-06,
+      "loss": 0.1833,
+      "step": 5795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.39786475828191e-06,
+      "loss": 0.1624,
+      "step": 5796
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.394647499078001e-06,
+      "loss": 0.1234,
+      "step": 5797
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.391430529159875e-06,
+      "loss": 0.2085,
+      "step": 5798
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.38821384888472e-06,
+      "loss": 0.1772,
+      "step": 5799
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.384997458609708e-06,
+      "loss": 0.1425,
+      "step": 5800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.381781358691968e-06,
+      "loss": 0.2007,
+      "step": 5801
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.378565549488604e-06,
+      "loss": 0.1877,
+      "step": 5802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.375350031356677e-06,
+      "loss": 0.1748,
+      "step": 5803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.372134804653232e-06,
+      "loss": 0.1487,
+      "step": 5804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3689198697352626e-06,
+      "loss": 0.1978,
+      "step": 5805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.365705226959747e-06,
+      "loss": 0.1858,
+      "step": 5806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.362490876683618e-06,
+      "loss": 0.1479,
+      "step": 5807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.359276819263786e-06,
+      "loss": 0.1345,
+      "step": 5808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.356063055057122e-06,
+      "loss": 0.1787,
+      "step": 5809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.352849584420472e-06,
+      "loss": 0.1626,
+      "step": 5810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.349636407710632e-06,
+      "loss": 0.1538,
+      "step": 5811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.346423525284393e-06,
+      "loss": 0.1902,
+      "step": 5812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.343210937498486e-06,
+      "loss": 0.1729,
+      "step": 5813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.339998644709627e-06,
+      "loss": 0.1847,
+      "step": 5814
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3367866472744964e-06,
+      "loss": 0.1226,
+      "step": 5815
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.333574945549727e-06,
+      "loss": 0.1831,
+      "step": 5816
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.330363539891946e-06,
+      "loss": 0.178,
+      "step": 5817
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.327152430657721e-06,
+      "loss": 0.161,
+      "step": 5818
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3239416182036024e-06,
+      "loss": 0.165,
+      "step": 5819
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.320731102886102e-06,
+      "loss": 0.1653,
+      "step": 5820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.317520885061704e-06,
+      "loss": 0.2056,
+      "step": 5821
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.314310965086848e-06,
+      "loss": 0.1577,
+      "step": 5822
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.311101343317956e-06,
+      "loss": 0.1213,
+      "step": 5823
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3078920201114045e-06,
+      "loss": 0.1476,
+      "step": 5824
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.304682995823542e-06,
+      "loss": 0.1736,
+      "step": 5825
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.301474270810681e-06,
+      "loss": 0.1395,
+      "step": 5826
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.298265845429109e-06,
+      "loss": 0.1704,
+      "step": 5827
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.295057720035066e-06,
+      "loss": 0.1503,
+      "step": 5828
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.291849894984775e-06,
+      "loss": 0.1758,
+      "step": 5829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.288642370634411e-06,
+      "loss": 0.1653,
+      "step": 5830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2854351473401255e-06,
+      "loss": 0.1836,
+      "step": 5831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.28222822545803e-06,
+      "loss": 0.198,
+      "step": 5832
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.27902160534421e-06,
+      "loss": 0.1606,
+      "step": 5833
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.275815287354707e-06,
+      "loss": 0.1555,
+      "step": 5834
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2726092718455455e-06,
+      "loss": 0.1489,
+      "step": 5835
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.269403559172691e-06,
+      "loss": 0.1941,
+      "step": 5836
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2661981496921055e-06,
+      "loss": 0.1788,
+      "step": 5837
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.262993043759692e-06,
+      "loss": 0.1604,
+      "step": 5838
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.259788241731336e-06,
+      "loss": 0.1541,
+      "step": 5839
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2565837439628775e-06,
+      "loss": 0.157,
+      "step": 5840
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.253379550810136e-06,
+      "loss": 0.228,
+      "step": 5841
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.250175662628879e-06,
+      "loss": 0.1296,
+      "step": 5842
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2469720797748635e-06,
+      "loss": 0.1792,
+      "step": 5843
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.24376880260379e-06,
+      "loss": 0.1226,
+      "step": 5844
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.24056583147134e-06,
+      "loss": 0.15,
+      "step": 5845
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.237363166733153e-06,
+      "loss": 0.1403,
+      "step": 5846
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.23416080874484e-06,
+      "loss": 0.1406,
+      "step": 5847
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.230958757861972e-06,
+      "loss": 0.186,
+      "step": 5848
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.227757014440098e-06,
+      "loss": 0.1538,
+      "step": 5849
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.224555578834711e-06,
+      "loss": 0.153,
+      "step": 5850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.221354451401294e-06,
+      "loss": 0.1667,
+      "step": 5851
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.218153632495277e-06,
+      "loss": 0.1646,
+      "step": 5852
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.21495312247207e-06,
+      "loss": 0.1719,
+      "step": 5853
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.211752921687036e-06,
+      "loss": 0.1681,
+      "step": 5854
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.208553030495518e-06,
+      "loss": 0.1802,
+      "step": 5855
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.205353449252807e-06,
+      "loss": 0.1382,
+      "step": 5856
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.202154178314175e-06,
+      "loss": 0.1376,
+      "step": 5857
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.198955218034851e-06,
+      "loss": 0.1877,
+      "step": 5858
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.195756568770036e-06,
+      "loss": 0.2014,
+      "step": 5859
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192558230874887e-06,
+      "loss": 0.1575,
+      "step": 5860
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.1893602047045385e-06,
+      "loss": 0.1882,
+      "step": 5861
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.186162490614075e-06,
+      "loss": 0.1653,
+      "step": 5862
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.182965088958567e-06,
+      "loss": 0.1709,
+      "step": 5863
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.179768000093027e-06,
+      "loss": 0.1443,
+      "step": 5864
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.176571224372453e-06,
+      "loss": 0.1261,
+      "step": 5865
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.173374762151792e-06,
+      "loss": 0.1755,
+      "step": 5866
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.170178613785972e-06,
+      "loss": 0.1719,
+      "step": 5867
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.166982779629871e-06,
+      "loss": 0.1699,
+      "step": 5868
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.163787260038346e-06,
+      "loss": 0.1915,
+      "step": 5869
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.160592055366202e-06,
+      "loss": 0.1587,
+      "step": 5870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.15739716596823e-06,
+      "loss": 0.1584,
+      "step": 5871
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.154202592199166e-06,
+      "loss": 0.1865,
+      "step": 5872
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.151008334413727e-06,
+      "loss": 0.1711,
+      "step": 5873
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.147814392966581e-06,
+      "loss": 0.1407,
+      "step": 5874
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.144620768212378e-06,
+      "loss": 0.1838,
+      "step": 5875
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.141427460505712e-06,
+      "loss": 0.1533,
+      "step": 5876
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.13823447020116e-06,
+      "loss": 0.1763,
+      "step": 5877
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.135041797653251e-06,
+      "loss": 0.1453,
+      "step": 5878
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.131849443216487e-06,
+      "loss": 0.199,
+      "step": 5879
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1286574072453315e-06,
+      "loss": 0.1377,
+      "step": 5880
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.125465690094215e-06,
+      "loss": 0.1599,
+      "step": 5881
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.122274292117526e-06,
+      "loss": 0.179,
+      "step": 5882
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.119083213669621e-06,
+      "loss": 0.1938,
+      "step": 5883
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.115892455104827e-06,
+      "loss": 0.1449,
+      "step": 5884
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1127020167774265e-06,
+      "loss": 0.1711,
+      "step": 5885
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.109511899041674e-06,
+      "loss": 0.1675,
+      "step": 5886
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1063221022517805e-06,
+      "loss": 0.2092,
+      "step": 5887
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.103132626761929e-06,
+      "loss": 0.1744,
+      "step": 5888
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.099943472926261e-06,
+      "loss": 0.1899,
+      "step": 5889
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.096754641098887e-06,
+      "loss": 0.1731,
+      "step": 5890
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.093566131633878e-06,
+      "loss": 0.161,
+      "step": 5891
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.090377944885275e-06,
+      "loss": 0.1792,
+      "step": 5892
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0871900812070695e-06,
+      "loss": 0.239,
+      "step": 5893
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.084002540953236e-06,
+      "loss": 0.1809,
+      "step": 5894
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.080815324477699e-06,
+      "loss": 0.1602,
+      "step": 5895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.077628432134352e-06,
+      "loss": 0.1497,
+      "step": 5896
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0744418642770516e-06,
+      "loss": 0.1838,
+      "step": 5897
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.071255621259622e-06,
+      "loss": 0.1696,
+      "step": 5898
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.068069703435845e-06,
+      "loss": 0.1644,
+      "step": 5899
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.064884111159474e-06,
+      "loss": 0.1436,
+      "step": 5900
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.061698844784213e-06,
+      "loss": 0.2219,
+      "step": 5901
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0585139046637485e-06,
+      "loss": 0.1707,
+      "step": 5902
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0553292911517134e-06,
+      "loss": 0.1587,
+      "step": 5903
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.052145004601717e-06,
+      "loss": 0.1494,
+      "step": 5904
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.048961045367324e-06,
+      "loss": 0.1604,
+      "step": 5905
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.04577741380207e-06,
+      "loss": 0.1733,
+      "step": 5906
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.042594110259444e-06,
+      "loss": 0.1329,
+      "step": 5907
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0394111350929095e-06,
+      "loss": 0.1771,
+      "step": 5908
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.036228488655885e-06,
+      "loss": 0.1443,
+      "step": 5909
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.033046171301761e-06,
+      "loss": 0.1868,
+      "step": 5910
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.02986418338388e-06,
+      "loss": 0.1677,
+      "step": 5911
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.026682525255563e-06,
+      "loss": 0.1963,
+      "step": 5912
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.023501197270077e-06,
+      "loss": 0.1545,
+      "step": 5913
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.02032019978067e-06,
+      "loss": 0.1472,
+      "step": 5914
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017139533140538e-06,
+      "loss": 0.2034,
+      "step": 5915
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.013959197702851e-06,
+      "loss": 0.1681,
+      "step": 5916
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0107791938207335e-06,
+      "loss": 0.1702,
+      "step": 5917
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.007599521847286e-06,
+      "loss": 0.168,
+      "step": 5918
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0044201821355515e-06,
+      "loss": 0.1299,
+      "step": 5919
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.001241175038562e-06,
+      "loss": 0.1958,
+      "step": 5920
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.998062500909288e-06,
+      "loss": 0.1376,
+      "step": 5921
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9948841601006824e-06,
+      "loss": 0.1565,
+      "step": 5922
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.991706152965647e-06,
+      "loss": 0.113,
+      "step": 5923
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.988528479857057e-06,
+      "loss": 0.1956,
+      "step": 5924
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.985351141127741e-06,
+      "loss": 0.1616,
+      "step": 5925
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.982174137130502e-06,
+      "loss": 0.2074,
+      "step": 5926
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.978997468218092e-06,
+      "loss": 0.1807,
+      "step": 5927
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.975821134743238e-06,
+      "loss": 0.1844,
+      "step": 5928
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.972645137058621e-06,
+      "loss": 0.2179,
+      "step": 5929
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.969469475516892e-06,
+      "loss": 0.1621,
+      "step": 5930
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.966294150470658e-06,
+      "loss": 0.2006,
+      "step": 5931
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.963119162272498e-06,
+      "loss": 0.1659,
+      "step": 5932
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.959944511274939e-06,
+      "loss": 0.1882,
+      "step": 5933
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.956770197830485e-06,
+      "loss": 0.137,
+      "step": 5934
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.95359622229159e-06,
+      "loss": 0.136,
+      "step": 5935
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.950422585010686e-06,
+      "loss": 0.1821,
+      "step": 5936
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.947249286340149e-06,
+      "loss": 0.1775,
+      "step": 5937
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.944076326632336e-06,
+      "loss": 0.1501,
+      "step": 5938
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.940903706239548e-06,
+      "loss": 0.1694,
+      "step": 5939
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.937731425514066e-06,
+      "loss": 0.1694,
+      "step": 5940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.934559484808117e-06,
+      "loss": 0.1854,
+      "step": 5941
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.931387884473905e-06,
+      "loss": 0.1708,
+      "step": 5942
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.928216624863582e-06,
+      "loss": 0.1646,
+      "step": 5943
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.925045706329277e-06,
+      "loss": 0.1655,
+      "step": 5944
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.921875129223067e-06,
+      "loss": 0.1631,
+      "step": 5945
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.918704893897004e-06,
+      "loss": 0.2126,
+      "step": 5946
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.915535000703088e-06,
+      "loss": 0.1431,
+      "step": 5947
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.912365449993296e-06,
+      "loss": 0.1716,
+      "step": 5948
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9091962421195566e-06,
+      "loss": 0.1732,
+      "step": 5949
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.906027377433758e-06,
+      "loss": 0.1451,
+      "step": 5950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.902858856287766e-06,
+      "loss": 0.1797,
+      "step": 5951
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.899690679033389e-06,
+      "loss": 0.2026,
+      "step": 5952
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.896522846022413e-06,
+      "loss": 0.156,
+      "step": 5953
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.893355357606573e-06,
+      "loss": 0.1816,
+      "step": 5954
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.890188214137575e-06,
+      "loss": 0.129,
+      "step": 5955
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.887021415967081e-06,
+      "loss": 0.2161,
+      "step": 5956
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883854963446723e-06,
+      "loss": 0.1531,
+      "step": 5957
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.880688856928081e-06,
+      "loss": 0.166,
+      "step": 5958
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.877523096762708e-06,
+      "loss": 0.179,
+      "step": 5959
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8743576833021135e-06,
+      "loss": 0.158,
+      "step": 5960
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.871192616897772e-06,
+      "loss": 0.135,
+      "step": 5961
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.868027897901113e-06,
+      "loss": 0.1672,
+      "step": 5962
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.864863526663539e-06,
+      "loss": 0.1484,
+      "step": 5963
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.861699503536397e-06,
+      "loss": 0.1704,
+      "step": 5964
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.858535828871012e-06,
+      "loss": 0.1498,
+      "step": 5965
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8553725030186605e-06,
+      "loss": 0.1427,
+      "step": 5966
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.852209526330583e-06,
+      "loss": 0.1663,
+      "step": 5967
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.849046899157981e-06,
+      "loss": 0.1512,
+      "step": 5968
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.845884621852021e-06,
+      "loss": 0.1843,
+      "step": 5969
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.842722694763817e-06,
+      "loss": 0.1941,
+      "step": 5970
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8395611182444684e-06,
+      "loss": 0.1497,
+      "step": 5971
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.83639989264501e-06,
+      "loss": 0.1615,
+      "step": 5972
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.833239018316456e-06,
+      "loss": 0.1749,
+      "step": 5973
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.830078495609768e-06,
+      "loss": 0.1714,
+      "step": 5974
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.826918324875882e-06,
+      "loss": 0.189,
+      "step": 5975
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.823758506465682e-06,
+      "loss": 0.1826,
+      "step": 5976
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8205990407300275e-06,
+      "loss": 0.1804,
+      "step": 5977
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.817439928019719e-06,
+      "loss": 0.1941,
+      "step": 5978
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.814281168685538e-06,
+      "loss": 0.1515,
+      "step": 5979
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.811122763078213e-06,
+      "loss": 0.1672,
+      "step": 5980
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.807964711548442e-06,
+      "loss": 0.1812,
+      "step": 5981
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.804807014446877e-06,
+      "loss": 0.2158,
+      "step": 5982
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.801649672124137e-06,
+      "loss": 0.1692,
+      "step": 5983
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7984926849307905e-06,
+      "loss": 0.1616,
+      "step": 5984
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795336053217383e-06,
+      "loss": 0.1532,
+      "step": 5985
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.792179777334404e-06,
+      "loss": 0.183,
+      "step": 5986
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.789023857632319e-06,
+      "loss": 0.1385,
+      "step": 5987
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.785868294461538e-06,
+      "loss": 0.1951,
+      "step": 5988
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.782713088172449e-06,
+      "loss": 0.1841,
+      "step": 5989
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.779558239115378e-06,
+      "loss": 0.2207,
+      "step": 5990
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.776403747640638e-06,
+      "loss": 0.1669,
+      "step": 5991
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.773249614098479e-06,
+      "loss": 0.189,
+      "step": 5992
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.770095838839126e-06,
+      "loss": 0.1665,
+      "step": 5993
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.766942422212755e-06,
+      "loss": 0.1709,
+      "step": 5994
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7637893645695105e-06,
+      "loss": 0.1882,
+      "step": 5995
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.760636666259485e-06,
+      "loss": 0.166,
+      "step": 5996
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.75748432763275e-06,
+      "loss": 0.2109,
+      "step": 5997
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.754332349039316e-06,
+      "loss": 0.1472,
+      "step": 5998
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7511807308291724e-06,
+      "loss": 0.1888,
+      "step": 5999
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7480294733522515e-06,
+      "loss": 0.1476,
+      "step": 6000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.744878576958458e-06,
+      "loss": 0.1477,
+      "step": 6001
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.74172804199765e-06,
+      "loss": 0.1768,
+      "step": 6002
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.738577868819656e-06,
+      "loss": 0.1895,
+      "step": 6003
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.735428057774243e-06,
+      "loss": 0.168,
+      "step": 6004
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732278609211159e-06,
+      "loss": 0.1624,
+      "step": 6005
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7291295234801e-06,
+      "loss": 0.1394,
+      "step": 6006
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.725980800930729e-06,
+      "loss": 0.1462,
+      "step": 6007
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7228324419126616e-06,
+      "loss": 0.1643,
+      "step": 6008
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7196844467754815e-06,
+      "loss": 0.1594,
+      "step": 6009
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.716536815868717e-06,
+      "loss": 0.1572,
+      "step": 6010
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7133895495418775e-06,
+      "loss": 0.1626,
+      "step": 6011
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.710242648144413e-06,
+      "loss": 0.1387,
+      "step": 6012
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.707096112025741e-06,
+      "loss": 0.177,
+      "step": 6013
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.703949941535239e-06,
+      "loss": 0.1672,
+      "step": 6014
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.700804137022247e-06,
+      "loss": 0.1611,
+      "step": 6015
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.697658698836047e-06,
+      "loss": 0.1261,
+      "step": 6016
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.694513627325909e-06,
+      "loss": 0.1531,
+      "step": 6017
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.691368922841037e-06,
+      "loss": 0.1455,
+      "step": 6018
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.688224585730605e-06,
+      "loss": 0.186,
+      "step": 6019
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6850806163437486e-06,
+      "loss": 0.1714,
+      "step": 6020
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.681937015029552e-06,
+      "loss": 0.1499,
+      "step": 6021
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.678793782137075e-06,
+      "loss": 0.1733,
+      "step": 6022
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.675650918015318e-06,
+      "loss": 0.1372,
+      "step": 6023
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.672508423013255e-06,
+      "loss": 0.1562,
+      "step": 6024
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.669366297479811e-06,
+      "loss": 0.1423,
+      "step": 6025
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.666224541763876e-06,
+      "loss": 0.1973,
+      "step": 6026
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.663083156214289e-06,
+      "loss": 0.175,
+      "step": 6027
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.659942141179865e-06,
+      "loss": 0.1575,
+      "step": 6028
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.656801497009353e-06,
+      "loss": 0.1606,
+      "step": 6029
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.653661224051486e-06,
+      "loss": 0.162,
+      "step": 6030
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.65052132265494e-06,
+      "loss": 0.1855,
+      "step": 6031
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.647381793168357e-06,
+      "loss": 0.151,
+      "step": 6032
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6442426359403315e-06,
+      "loss": 0.1614,
+      "step": 6033
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.641103851319426e-06,
+      "loss": 0.1487,
+      "step": 6034
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.63796543965415e-06,
+      "loss": 0.1687,
+      "step": 6035
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.634827401292981e-06,
+      "loss": 0.1948,
+      "step": 6036
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.63168973658435e-06,
+      "loss": 0.1592,
+      "step": 6037
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.62855244587665e-06,
+      "loss": 0.1794,
+      "step": 6038
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.625415529518228e-06,
+      "loss": 0.1595,
+      "step": 6039
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6222789878574e-06,
+      "loss": 0.1799,
+      "step": 6040
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.619142821242419e-06,
+      "loss": 0.158,
+      "step": 6041
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.616007030021523e-06,
+      "loss": 0.1757,
+      "step": 6042
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.612871614542884e-06,
+      "loss": 0.1595,
+      "step": 6043
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.609736575154653e-06,
+      "loss": 0.1733,
+      "step": 6044
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6066019122049216e-06,
+      "loss": 0.2073,
+      "step": 6045
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.603467626041755e-06,
+      "loss": 0.2029,
+      "step": 6046
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.600333717013161e-06,
+      "loss": 0.197,
+      "step": 6047
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5972001854671245e-06,
+      "loss": 0.1741,
+      "step": 6048
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.594067031751565e-06,
+      "loss": 0.1929,
+      "step": 6049
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.590934256214385e-06,
+      "loss": 0.1946,
+      "step": 6050
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.587801859203422e-06,
+      "loss": 0.1302,
+      "step": 6051
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.584669841066491e-06,
+      "loss": 0.2043,
+      "step": 6052
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.581538202151351e-06,
+      "loss": 0.165,
+      "step": 6053
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.578406942805729e-06,
+      "loss": 0.1611,
+      "step": 6054
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.575276063377299e-06,
+      "loss": 0.1315,
+      "step": 6055
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.572145564213701e-06,
+      "loss": 0.1561,
+      "step": 6056
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.56901544566253e-06,
+      "loss": 0.1427,
+      "step": 6057
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5658857080713425e-06,
+      "loss": 0.2097,
+      "step": 6058
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562756351787645e-06,
+      "loss": 0.1843,
+      "step": 6059
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.559627377158911e-06,
+      "loss": 0.166,
+      "step": 6060
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5564987845325614e-06,
+      "loss": 0.1714,
+      "step": 6061
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5533705742559835e-06,
+      "loss": 0.1746,
+      "step": 6062
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.550242746676519e-06,
+      "loss": 0.146,
+      "step": 6063
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.547115302141465e-06,
+      "loss": 0.1428,
+      "step": 6064
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.543988240998076e-06,
+      "loss": 0.1808,
+      "step": 6065
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5408615635935735e-06,
+      "loss": 0.1531,
+      "step": 6066
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5377352702751175e-06,
+      "loss": 0.1709,
+      "step": 6067
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.534609361389847e-06,
+      "loss": 0.1948,
+      "step": 6068
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5314838372848425e-06,
+      "loss": 0.1809,
+      "step": 6069
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.528358698307148e-06,
+      "loss": 0.1252,
+      "step": 6070
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5252339448037615e-06,
+      "loss": 0.1594,
+      "step": 6071
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.522109577121646e-06,
+      "loss": 0.1594,
+      "step": 6072
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.51898559560771e-06,
+      "loss": 0.1716,
+      "step": 6073
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.515862000608834e-06,
+      "loss": 0.1367,
+      "step": 6074
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.512738792471837e-06,
+      "loss": 0.1912,
+      "step": 6075
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5096159715435105e-06,
+      "loss": 0.1836,
+      "step": 6076
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.506493538170595e-06,
+      "loss": 0.1581,
+      "step": 6077
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.503371492699794e-06,
+      "loss": 0.1802,
+      "step": 6078
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.5002498354777584e-06,
+      "loss": 0.144,
+      "step": 6079
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.497128566851112e-06,
+      "loss": 0.1676,
+      "step": 6080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4940076871664145e-06,
+      "loss": 0.1837,
+      "step": 6081
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4908871967702e-06,
+      "loss": 0.1281,
+      "step": 6082
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.487767096008949e-06,
+      "loss": 0.1638,
+      "step": 6083
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4846473852291054e-06,
+      "loss": 0.176,
+      "step": 6084
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.481528064777069e-06,
+      "loss": 0.14,
+      "step": 6085
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4784091349991855e-06,
+      "loss": 0.1807,
+      "step": 6086
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.475290596241772e-06,
+      "loss": 0.1506,
+      "step": 6087
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.472172448851095e-06,
+      "loss": 0.1821,
+      "step": 6088
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.46905469317338e-06,
+      "loss": 0.1715,
+      "step": 6089
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.465937329554804e-06,
+      "loss": 0.1492,
+      "step": 6090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.462820358341511e-06,
+      "loss": 0.1985,
+      "step": 6091
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459703779879585e-06,
+      "loss": 0.1328,
+      "step": 6092
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.456587594515081e-06,
+      "loss": 0.1726,
+      "step": 6093
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.453471802594005e-06,
+      "loss": 0.1694,
+      "step": 6094
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.450356404462319e-06,
+      "loss": 0.1636,
+      "step": 6095
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.447241400465942e-06,
+      "loss": 0.187,
+      "step": 6096
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.444126790950751e-06,
+      "loss": 0.1572,
+      "step": 6097
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.441012576262571e-06,
+      "loss": 0.1561,
+      "step": 6098
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.437898756747198e-06,
+      "loss": 0.198,
+      "step": 6099
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.434785332750368e-06,
+      "loss": 0.186,
+      "step": 6100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.431672304617786e-06,
+      "loss": 0.1798,
+      "step": 6101
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.428559672695102e-06,
+      "loss": 0.166,
+      "step": 6102
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.425447437327936e-06,
+      "loss": 0.1254,
+      "step": 6103
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.422335598861846e-06,
+      "loss": 0.1738,
+      "step": 6104
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.419224157642365e-06,
+      "loss": 0.1536,
+      "step": 6105
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.416113114014964e-06,
+      "loss": 0.1626,
+      "step": 6106
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.413002468325082e-06,
+      "loss": 0.1565,
+      "step": 6107
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.409892220918108e-06,
+      "loss": 0.1904,
+      "step": 6108
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.406782372139395e-06,
+      "loss": 0.1597,
+      "step": 6109
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4036729223342375e-06,
+      "loss": 0.1591,
+      "step": 6110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.400563871847902e-06,
+      "loss": 0.1427,
+      "step": 6111
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.397455221025595e-06,
+      "loss": 0.1835,
+      "step": 6112
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.394346970212491e-06,
+      "loss": 0.1409,
+      "step": 6113
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.39123911975371e-06,
+      "loss": 0.1626,
+      "step": 6114
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.38813166999434e-06,
+      "loss": 0.1506,
+      "step": 6115
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.385024621279411e-06,
+      "loss": 0.168,
+      "step": 6116
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3819179739539214e-06,
+      "loss": 0.1746,
+      "step": 6117
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.378811728362808e-06,
+      "loss": 0.1543,
+      "step": 6118
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.375705884850985e-06,
+      "loss": 0.1777,
+      "step": 6119
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3726004437633e-06,
+      "loss": 0.1351,
+      "step": 6120
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3694954054445745e-06,
+      "loss": 0.1443,
+      "step": 6121
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.36639077023957e-06,
+      "loss": 0.1458,
+      "step": 6122
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.363286538493019e-06,
+      "loss": 0.1619,
+      "step": 6123
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360182710549587e-06,
+      "loss": 0.1609,
+      "step": 6124
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.357079286753925e-06,
+      "loss": 0.199,
+      "step": 6125
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.353976267450608e-06,
+      "loss": 0.1603,
+      "step": 6126
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.350873652984186e-06,
+      "loss": 0.1731,
+      "step": 6127
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.347771443699157e-06,
+      "loss": 0.1361,
+      "step": 6128
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.34466963993998e-06,
+      "loss": 0.1682,
+      "step": 6129
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.341568242051057e-06,
+      "loss": 0.1798,
+      "step": 6130
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.338467250376761e-06,
+      "loss": 0.1592,
+      "step": 6131
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.335366665261401e-06,
+      "loss": 0.1809,
+      "step": 6132
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.332266487049259e-06,
+      "loss": 0.1641,
+      "step": 6133
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.329166716084558e-06,
+      "loss": 0.1417,
+      "step": 6134
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.326067352711487e-06,
+      "loss": 0.1887,
+      "step": 6135
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.322968397274181e-06,
+      "loss": 0.1626,
+      "step": 6136
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.319869850116738e-06,
+      "loss": 0.1677,
+      "step": 6137
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.316771711583196e-06,
+      "loss": 0.2128,
+      "step": 6138
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.313673982017567e-06,
+      "loss": 0.1672,
+      "step": 6139
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.310576661763801e-06,
+      "loss": 0.1313,
+      "step": 6140
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.307479751165814e-06,
+      "loss": 0.1794,
+      "step": 6141
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.30438325056747e-06,
+      "loss": 0.1687,
+      "step": 6142
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.301287160312593e-06,
+      "loss": 0.2058,
+      "step": 6143
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.298191480744951e-06,
+      "loss": 0.183,
+      "step": 6144
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.295096212208281e-06,
+      "loss": 0.2161,
+      "step": 6145
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.29200135504626e-06,
+      "loss": 0.1692,
+      "step": 6146
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.288906909602531e-06,
+      "loss": 0.1732,
+      "step": 6147
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.285812876220682e-06,
+      "loss": 0.1606,
+      "step": 6148
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.282719255244265e-06,
+      "loss": 0.1682,
+      "step": 6149
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.279626047016776e-06,
+      "loss": 0.1572,
+      "step": 6150
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.276533251881674e-06,
+      "loss": 0.1616,
+      "step": 6151
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.273440870182365e-06,
+      "loss": 0.1649,
+      "step": 6152
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.270348902262209e-06,
+      "loss": 0.1907,
+      "step": 6153
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.267257348464532e-06,
+      "loss": 0.1786,
+      "step": 6154
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.264166209132596e-06,
+      "loss": 0.1837,
+      "step": 6155
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.261075484609635e-06,
+      "loss": 0.1613,
+      "step": 6156
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2579851752388186e-06,
+      "loss": 0.1338,
+      "step": 6157
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.254895281363289e-06,
+      "loss": 0.2163,
+      "step": 6158
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2518058033261255e-06,
+      "loss": 0.1816,
+      "step": 6159
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.248716741470376e-06,
+      "loss": 0.1503,
+      "step": 6160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.245628096139027e-06,
+      "loss": 0.1946,
+      "step": 6161
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.242539867675036e-06,
+      "loss": 0.1503,
+      "step": 6162
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2394520564212954e-06,
+      "loss": 0.165,
+      "step": 6163
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.236364662720668e-06,
+      "loss": 0.1764,
+      "step": 6164
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.233277686915961e-06,
+      "loss": 0.1443,
+      "step": 6165
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.230191129349938e-06,
+      "loss": 0.1492,
+      "step": 6166
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2271049903653115e-06,
+      "loss": 0.1963,
+      "step": 6167
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.224019270304757e-06,
+      "loss": 0.1638,
+      "step": 6168
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.220933969510894e-06,
+      "loss": 0.1873,
+      "step": 6169
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.217849088326302e-06,
+      "loss": 0.1652,
+      "step": 6170
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.214764627093509e-06,
+      "loss": 0.1951,
+      "step": 6171
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.211680586155003e-06,
+      "loss": 0.1599,
+      "step": 6172
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.208596965853213e-06,
+      "loss": 0.1814,
+      "step": 6173
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.20551376653054e-06,
+      "loss": 0.162,
+      "step": 6174
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.202430988529315e-06,
+      "loss": 0.1641,
+      "step": 6175
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.199348632191848e-06,
+      "loss": 0.178,
+      "step": 6176
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.196266697860379e-06,
+      "loss": 0.1305,
+      "step": 6177
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.193185185877117e-06,
+      "loss": 0.1533,
+      "step": 6178
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.190104096584213e-06,
+      "loss": 0.179,
+      "step": 6179
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.187023430323782e-06,
+      "loss": 0.2125,
+      "step": 6180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1839431874378816e-06,
+      "loss": 0.1554,
+      "step": 6181
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1808633682685345e-06,
+      "loss": 0.175,
+      "step": 6182
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.177783973157699e-06,
+      "loss": 0.168,
+      "step": 6183
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.174705002447301e-06,
+      "loss": 0.1692,
+      "step": 6184
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.171626456479214e-06,
+      "loss": 0.1593,
+      "step": 6185
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.168548335595268e-06,
+      "loss": 0.1726,
+      "step": 6186
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.165470640137237e-06,
+      "loss": 0.1719,
+      "step": 6187
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.162393370446861e-06,
+      "loss": 0.121,
+      "step": 6188
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.159316526865816e-06,
+      "loss": 0.1765,
+      "step": 6189
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.156240109735747e-06,
+      "loss": 0.1592,
+      "step": 6190
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.153164119398241e-06,
+      "loss": 0.1666,
+      "step": 6191
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.150088556194843e-06,
+      "loss": 0.1729,
+      "step": 6192
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.147013420467047e-06,
+      "loss": 0.1636,
+      "step": 6193
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.143938712556305e-06,
+      "loss": 0.1743,
+      "step": 6194
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.14086443280401e-06,
+      "loss": 0.1554,
+      "step": 6195
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1377905815515255e-06,
+      "loss": 0.1982,
+      "step": 6196
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.134717159140149e-06,
+      "loss": 0.1646,
+      "step": 6197
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.131644165911144e-06,
+      "loss": 0.1924,
+      "step": 6198
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.128571602205713e-06,
+      "loss": 0.1173,
+      "step": 6199
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.125499468365028e-06,
+      "loss": 0.1616,
+      "step": 6200
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.122427764730198e-06,
+      "loss": 0.1392,
+      "step": 6201
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.119356491642294e-06,
+      "loss": 0.1368,
+      "step": 6202
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.116285649442331e-06,
+      "loss": 0.1228,
+      "step": 6203
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.113215238471284e-06,
+      "loss": 0.1553,
+      "step": 6204
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1101452590700735e-06,
+      "loss": 0.1499,
+      "step": 6205
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.10707571157958e-06,
+      "loss": 0.153,
+      "step": 6206
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.104006596340625e-06,
+      "loss": 0.1541,
+      "step": 6207
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.100937913693996e-06,
+      "loss": 0.1554,
+      "step": 6208
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0978696639804155e-06,
+      "loss": 0.1992,
+      "step": 6209
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.094801847540576e-06,
+      "loss": 0.1638,
+      "step": 6210
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.091734464715105e-06,
+      "loss": 0.1643,
+      "step": 6211
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0886675158445976e-06,
+      "loss": 0.1833,
+      "step": 6212
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.085601001269586e-06,
+      "loss": 0.1824,
+      "step": 6213
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.08253492133057e-06,
+      "loss": 0.1824,
+      "step": 6214
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.07946927636798e-06,
+      "loss": 0.1532,
+      "step": 6215
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.076404066722224e-06,
+      "loss": 0.1445,
+      "step": 6216
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.073339292733635e-06,
+      "loss": 0.2183,
+      "step": 6217
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.070274954742521e-06,
+      "loss": 0.1779,
+      "step": 6218
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.06721105308913e-06,
+      "loss": 0.1685,
+      "step": 6219
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.064147588113656e-06,
+      "loss": 0.1731,
+      "step": 6220
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.061084560156257e-06,
+      "loss": 0.1438,
+      "step": 6221
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.058021969557033e-06,
+      "loss": 0.179,
+      "step": 6222
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.054959816656045e-06,
+      "loss": 0.1721,
+      "step": 6223
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.051898101793294e-06,
+      "loss": 0.1575,
+      "step": 6224
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.048836825308745e-06,
+      "loss": 0.1875,
+      "step": 6225
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.045775987542298e-06,
+      "loss": 0.189,
+      "step": 6226
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.042715588833822e-06,
+      "loss": 0.1628,
+      "step": 6227
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.039655629523122e-06,
+      "loss": 0.1376,
+      "step": 6228
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.036596109949968e-06,
+      "loss": 0.1902,
+      "step": 6229
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.033537030454069e-06,
+      "loss": 0.1655,
+      "step": 6230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.030478391375094e-06,
+      "loss": 0.1787,
+      "step": 6231
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.027420193052656e-06,
+      "loss": 0.1753,
+      "step": 6232
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.024362435826328e-06,
+      "loss": 0.1252,
+      "step": 6233
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.021305120035621e-06,
+      "loss": 0.1655,
+      "step": 6234
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018248246020012e-06,
+      "loss": 0.1372,
+      "step": 6235
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0151918141189156e-06,
+      "loss": 0.1573,
+      "step": 6236
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.012135824671707e-06,
+      "loss": 0.158,
+      "step": 6237
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.009080278017707e-06,
+      "loss": 0.1501,
+      "step": 6238
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0060251744961926e-06,
+      "loss": 0.1852,
+      "step": 6239
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.002970514446382e-06,
+      "loss": 0.1421,
+      "step": 6240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.999916298207452e-06,
+      "loss": 0.1616,
+      "step": 6241
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.996862526118528e-06,
+      "loss": 0.1434,
+      "step": 6242
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.993809198518687e-06,
+      "loss": 0.188,
+      "step": 6243
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990756315746955e-06,
+      "loss": 0.144,
+      "step": 6244
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.987703878142313e-06,
+      "loss": 0.1519,
+      "step": 6245
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.98465188604368e-06,
+      "loss": 0.1533,
+      "step": 6246
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.981600339789945e-06,
+      "loss": 0.2122,
+      "step": 6247
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.978549239719929e-06,
+      "loss": 0.1404,
+      "step": 6248
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.975498586172416e-06,
+      "loss": 0.1794,
+      "step": 6249
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.972448379486133e-06,
+      "loss": 0.1703,
+      "step": 6250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.969398619999764e-06,
+      "loss": 0.179,
+      "step": 6251
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.966349308051933e-06,
+      "loss": 0.1895,
+      "step": 6252
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.963300443981231e-06,
+      "loss": 0.1743,
+      "step": 6253
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.960252028126179e-06,
+      "loss": 0.127,
+      "step": 6254
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.957204060825265e-06,
+      "loss": 0.1693,
+      "step": 6255
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.954156542416915e-06,
+      "loss": 0.1465,
+      "step": 6256
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.951109473239518e-06,
+      "loss": 0.1272,
+      "step": 6257
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.948062853631399e-06,
+      "loss": 0.1467,
+      "step": 6258
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.945016683930846e-06,
+      "loss": 0.2063,
+      "step": 6259
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.941970964476085e-06,
+      "loss": 0.1217,
+      "step": 6260
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.938925695605303e-06,
+      "loss": 0.1522,
+      "step": 6261
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.935880877656628e-06,
+      "loss": 0.1425,
+      "step": 6262
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.932836510968145e-06,
+      "loss": 0.1613,
+      "step": 6263
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.929792595877881e-06,
+      "loss": 0.1731,
+      "step": 6264
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.926749132723828e-06,
+      "loss": 0.1555,
+      "step": 6265
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.923706121843905e-06,
+      "loss": 0.1599,
+      "step": 6266
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.920663563576001e-06,
+      "loss": 0.1624,
+      "step": 6267
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.917621458257944e-06,
+      "loss": 0.165,
+      "step": 6268
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.914579806227516e-06,
+      "loss": 0.197,
+      "step": 6269
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.911538607822447e-06,
+      "loss": 0.189,
+      "step": 6270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.90849786338042e-06,
+      "loss": 0.2021,
+      "step": 6271
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.905457573239055e-06,
+      "loss": 0.1619,
+      "step": 6272
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.902417737735944e-06,
+      "loss": 0.1936,
+      "step": 6273
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.899378357208606e-06,
+      "loss": 0.1703,
+      "step": 6274
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8963394319945244e-06,
+      "loss": 0.1731,
+      "step": 6275
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8933009624311235e-06,
+      "loss": 0.1931,
+      "step": 6276
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.890262948855783e-06,
+      "loss": 0.1658,
+      "step": 6277
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8872253916058255e-06,
+      "loss": 0.1467,
+      "step": 6278
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.884188291018533e-06,
+      "loss": 0.1978,
+      "step": 6279
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.881151647431125e-06,
+      "loss": 0.1571,
+      "step": 6280
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.878115461180778e-06,
+      "loss": 0.1851,
+      "step": 6281
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8750797326046134e-06,
+      "loss": 0.1919,
+      "step": 6282
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.872044462039708e-06,
+      "loss": 0.165,
+      "step": 6283
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.869009649823079e-06,
+      "loss": 0.1692,
+      "step": 6284
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.865975296291702e-06,
+      "loss": 0.1746,
+      "step": 6285
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.862941401782494e-06,
+      "loss": 0.1904,
+      "step": 6286
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.859907966632321e-06,
+      "loss": 0.1392,
+      "step": 6287
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.856874991178008e-06,
+      "loss": 0.167,
+      "step": 6288
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.853842475756315e-06,
+      "loss": 0.1443,
+      "step": 6289
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8508104207039674e-06,
+      "loss": 0.1758,
+      "step": 6290
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8477788263576175e-06,
+      "loss": 0.1578,
+      "step": 6291
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.844747693053889e-06,
+      "loss": 0.1417,
+      "step": 6292
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.841717021129337e-06,
+      "loss": 0.1497,
+      "step": 6293
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.83868681092048e-06,
+      "loss": 0.1517,
+      "step": 6294
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.835657062763773e-06,
+      "loss": 0.1918,
+      "step": 6295
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.832627776995625e-06,
+      "loss": 0.2087,
+      "step": 6296
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.829598953952394e-06,
+      "loss": 0.1434,
+      "step": 6297
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.826570593970389e-06,
+      "loss": 0.1914,
+      "step": 6298
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.823542697385858e-06,
+      "loss": 0.1405,
+      "step": 6299
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.820515264535011e-06,
+      "loss": 0.1589,
+      "step": 6300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8174882957539925e-06,
+      "loss": 0.1775,
+      "step": 6301
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.814461791378907e-06,
+      "loss": 0.1987,
+      "step": 6302
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.811435751745802e-06,
+      "loss": 0.1638,
+      "step": 6303
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.808410177190677e-06,
+      "loss": 0.1099,
+      "step": 6304
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.805385068049472e-06,
+      "loss": 0.1508,
+      "step": 6305
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802360424658086e-06,
+      "loss": 0.1753,
+      "step": 6306
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.799336247352356e-06,
+      "loss": 0.1738,
+      "step": 6307
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.796312536468074e-06,
+      "loss": 0.1643,
+      "step": 6308
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.793289292340976e-06,
+      "loss": 0.1658,
+      "step": 6309
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.790266515306757e-06,
+      "loss": 0.1689,
+      "step": 6310
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7872442057010405e-06,
+      "loss": 0.1805,
+      "step": 6311
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.784222363859418e-06,
+      "loss": 0.1604,
+      "step": 6312
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.781200990117408e-06,
+      "loss": 0.1649,
+      "step": 6313
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7781800848105065e-06,
+      "loss": 0.1494,
+      "step": 6314
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7751596482741265e-06,
+      "loss": 0.1743,
+      "step": 6315
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.772139680843652e-06,
+      "loss": 0.1486,
+      "step": 6316
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.769120182854397e-06,
+      "loss": 0.1572,
+      "step": 6317
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.766101154641637e-06,
+      "loss": 0.156,
+      "step": 6318
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.763082596540588e-06,
+      "loss": 0.1658,
+      "step": 6319
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.760064508886423e-06,
+      "loss": 0.1641,
+      "step": 6320
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.757046892014247e-06,
+      "loss": 0.1772,
+      "step": 6321
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7540297462591264e-06,
+      "loss": 0.1733,
+      "step": 6322
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751013071956068e-06,
+      "loss": 0.1394,
+      "step": 6323
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.747996869440029e-06,
+      "loss": 0.1942,
+      "step": 6324
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7449811390459155e-06,
+      "loss": 0.1509,
+      "step": 6325
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.741965881108583e-06,
+      "loss": 0.1719,
+      "step": 6326
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.738951095962824e-06,
+      "loss": 0.1405,
+      "step": 6327
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.735936783943392e-06,
+      "loss": 0.1401,
+      "step": 6328
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.732922945384975e-06,
+      "loss": 0.1851,
+      "step": 6329
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.729909580622219e-06,
+      "loss": 0.1792,
+      "step": 6330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.7268966899897134e-06,
+      "loss": 0.1997,
+      "step": 6331
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723884273822e-06,
+      "loss": 0.144,
+      "step": 6332
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.720872332453553e-06,
+      "loss": 0.1702,
+      "step": 6333
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.717860866218814e-06,
+      "loss": 0.1401,
+      "step": 6334
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.714849875452151e-06,
+      "loss": 0.1782,
+      "step": 6335
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.711839360487897e-06,
+      "loss": 0.1584,
+      "step": 6336
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.708829321660321e-06,
+      "loss": 0.1694,
+      "step": 6337
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.705819759303653e-06,
+      "loss": 0.1877,
+      "step": 6338
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.702810673752046e-06,
+      "loss": 0.1392,
+      "step": 6339
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.699802065339626e-06,
+      "loss": 0.1887,
+      "step": 6340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.696793934400446e-06,
+      "loss": 0.1627,
+      "step": 6341
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.693786281268519e-06,
+      "loss": 0.1906,
+      "step": 6342
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6907791062778e-06,
+      "loss": 0.1838,
+      "step": 6343
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.687772409762193e-06,
+      "loss": 0.1234,
+      "step": 6344
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6847661920555395e-06,
+      "loss": 0.1838,
+      "step": 6345
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.681760453491647e-06,
+      "loss": 0.1592,
+      "step": 6346
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.678755194404246e-06,
+      "loss": 0.1523,
+      "step": 6347
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.675750415127032e-06,
+      "loss": 0.1707,
+      "step": 6348
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.672746115993643e-06,
+      "loss": 0.1814,
+      "step": 6349
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.669742297337662e-06,
+      "loss": 0.1508,
+      "step": 6350
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.666738959492612e-06,
+      "loss": 0.1914,
+      "step": 6351
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.663736102791979e-06,
+      "loss": 0.1675,
+      "step": 6352
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.660733727569176e-06,
+      "loss": 0.1692,
+      "step": 6353
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.657731834157576e-06,
+      "loss": 0.1531,
+      "step": 6354
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6547304228905e-06,
+      "loss": 0.1422,
+      "step": 6355
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.651729494101201e-06,
+      "loss": 0.1602,
+      "step": 6356
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.648729048122896e-06,
+      "loss": 0.1652,
+      "step": 6357
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6457290852887336e-06,
+      "loss": 0.2048,
+      "step": 6358
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.642729605931816e-06,
+      "loss": 0.1415,
+      "step": 6359
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6397306103851944e-06,
+      "loss": 0.1622,
+      "step": 6360
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6367320989818644e-06,
+      "loss": 0.1914,
+      "step": 6361
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.63373407205476e-06,
+      "loss": 0.1616,
+      "step": 6362
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6307365299367735e-06,
+      "loss": 0.1982,
+      "step": 6363
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.627739472960727e-06,
+      "loss": 0.1337,
+      "step": 6364
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624742901459415e-06,
+      "loss": 0.1897,
+      "step": 6365
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.62174681576555e-06,
+      "loss": 0.178,
+      "step": 6366
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.618751216211812e-06,
+      "loss": 0.1626,
+      "step": 6367
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.615756103130809e-06,
+      "loss": 0.1726,
+      "step": 6368
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.612761476855112e-06,
+      "loss": 0.1237,
+      "step": 6369
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.609767337717218e-06,
+      "loss": 0.1481,
+      "step": 6370
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606773686049598e-06,
+      "loss": 0.1619,
+      "step": 6371
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6037805221846395e-06,
+      "loss": 0.1942,
+      "step": 6372
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.600787846454698e-06,
+      "loss": 0.1501,
+      "step": 6373
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5977956591920576e-06,
+      "loss": 0.1708,
+      "step": 6374
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.594803960728962e-06,
+      "loss": 0.1315,
+      "step": 6375
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.591812751397592e-06,
+      "loss": 0.1763,
+      "step": 6376
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.588822031530083e-06,
+      "loss": 0.2002,
+      "step": 6377
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.585831801458501e-06,
+      "loss": 0.1504,
+      "step": 6378
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.582842061514875e-06,
+      "loss": 0.1501,
+      "step": 6379
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.579852812031165e-06,
+      "loss": 0.1755,
+      "step": 6380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.576864053339284e-06,
+      "loss": 0.1479,
+      "step": 6381
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.573875785771091e-06,
+      "loss": 0.177,
+      "step": 6382
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.570888009658393e-06,
+      "loss": 0.2006,
+      "step": 6383
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.56790072533293e-06,
+      "loss": 0.1447,
+      "step": 6384
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.564913933126404e-06,
+      "loss": 0.1455,
+      "step": 6385
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.561927633370443e-06,
+      "loss": 0.158,
+      "step": 6386
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.55894182639664e-06,
+      "loss": 0.1675,
+      "step": 6387
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.555956512536522e-06,
+      "loss": 0.1697,
+      "step": 6388
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.552971692121568e-06,
+      "loss": 0.1708,
+      "step": 6389
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.54998736548319e-06,
+      "loss": 0.1514,
+      "step": 6390
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.547003532952763e-06,
+      "loss": 0.1379,
+      "step": 6391
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.544020194861587e-06,
+      "loss": 0.1846,
+      "step": 6392
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.541037351540923e-06,
+      "loss": 0.1678,
+      "step": 6393
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.538055003321969e-06,
+      "loss": 0.1724,
+      "step": 6394
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.535073150535879e-06,
+      "loss": 0.1715,
+      "step": 6395
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5320917935137315e-06,
+      "loss": 0.2009,
+      "step": 6396
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.529110932586573e-06,
+      "loss": 0.1765,
+      "step": 6397
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5261305680853736e-06,
+      "loss": 0.1707,
+      "step": 6398
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.523150700341065e-06,
+      "loss": 0.167,
+      "step": 6399
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.520171329684514e-06,
+      "loss": 0.1671,
+      "step": 6400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.517192456446543e-06,
+      "loss": 0.1735,
+      "step": 6401
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5142140809579e-06,
+      "loss": 0.1829,
+      "step": 6402
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.511236203549302e-06,
+      "loss": 0.1887,
+      "step": 6403
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.508258824551387e-06,
+      "loss": 0.1362,
+      "step": 6404
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.505281944294752e-06,
+      "loss": 0.1746,
+      "step": 6405
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.502305563109939e-06,
+      "loss": 0.1417,
+      "step": 6406
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.499329681327432e-06,
+      "loss": 0.1941,
+      "step": 6407
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.496354299277652e-06,
+      "loss": 0.1772,
+      "step": 6408
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.493379417290978e-06,
+      "loss": 0.1609,
+      "step": 6409
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.490405035697717e-06,
+      "loss": 0.1432,
+      "step": 6410
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.487431154828142e-06,
+      "loss": 0.1455,
+      "step": 6411
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.484457775012451e-06,
+      "loss": 0.1591,
+      "step": 6412
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.481484896580798e-06,
+      "loss": 0.1697,
+      "step": 6413
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.478512519863269e-06,
+      "loss": 0.1327,
+      "step": 6414
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.475540645189914e-06,
+      "loss": 0.1777,
+      "step": 6415
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4725692728906994e-06,
+      "loss": 0.1765,
+      "step": 6416
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.469598403295572e-06,
+      "loss": 0.1631,
+      "step": 6417
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4666280367343885e-06,
+      "loss": 0.1594,
+      "step": 6418
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.463658173536972e-06,
+      "loss": 0.2249,
+      "step": 6419
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.460688814033075e-06,
+      "loss": 0.1704,
+      "step": 6420
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.457719958552404e-06,
+      "loss": 0.1997,
+      "step": 6421
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4547516074246085e-06,
+      "loss": 0.1582,
+      "step": 6422
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.451783760979276e-06,
+      "loss": 0.1356,
+      "step": 6423
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.448816419545948e-06,
+      "loss": 0.1584,
+      "step": 6424
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.445849583454095e-06,
+      "loss": 0.1882,
+      "step": 6425
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4428832530331445e-06,
+      "loss": 0.1481,
+      "step": 6426
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4399174286124644e-06,
+      "loss": 0.1648,
+      "step": 6427
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.43695211052137e-06,
+      "loss": 0.1573,
+      "step": 6428
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.433987299089105e-06,
+      "loss": 0.1439,
+      "step": 6429
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.431022994644879e-06,
+      "loss": 0.1665,
+      "step": 6430
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.428059197517823e-06,
+      "loss": 0.1506,
+      "step": 6431
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4250959080370305e-06,
+      "loss": 0.1385,
+      "step": 6432
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.422133126531527e-06,
+      "loss": 0.1267,
+      "step": 6433
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.419170853330291e-06,
+      "loss": 0.2073,
+      "step": 6434
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.416209088762233e-06,
+      "loss": 0.1394,
+      "step": 6435
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.413247833156219e-06,
+      "loss": 0.1711,
+      "step": 6436
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.410287086841044e-06,
+      "loss": 0.1816,
+      "step": 6437
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4073268501454615e-06,
+      "loss": 0.1643,
+      "step": 6438
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.404367123398159e-06,
+      "loss": 0.1445,
+      "step": 6439
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.401407906927776e-06,
+      "loss": 0.1726,
+      "step": 6440
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.398449201062881e-06,
+      "loss": 0.144,
+      "step": 6441
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.395491006132003e-06,
+      "loss": 0.1821,
+      "step": 6442
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.392533322463598e-06,
+      "loss": 0.1829,
+      "step": 6443
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.389576150386075e-06,
+      "loss": 0.2026,
+      "step": 6444
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.386619490227787e-06,
+      "loss": 0.2043,
+      "step": 6445
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.383663342317029e-06,
+      "loss": 0.1515,
+      "step": 6446
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3807077069820295e-06,
+      "loss": 0.1841,
+      "step": 6447
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.377752584550977e-06,
+      "loss": 0.1572,
+      "step": 6448
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.374797975351986e-06,
+      "loss": 0.1624,
+      "step": 6449
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.371843879713126e-06,
+      "loss": 0.1885,
+      "step": 6450
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.368890297962403e-06,
+      "loss": 0.1479,
+      "step": 6451
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.365937230427778e-06,
+      "loss": 0.1675,
+      "step": 6452
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.362984677437133e-06,
+      "loss": 0.1499,
+      "step": 6453
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.360032639318313e-06,
+      "loss": 0.1299,
+      "step": 6454
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.357081116399094e-06,
+      "loss": 0.1796,
+      "step": 6455
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.354130109007198e-06,
+      "loss": 0.1587,
+      "step": 6456
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.351179617470294e-06,
+      "loss": 0.1571,
+      "step": 6457
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.348229642115993e-06,
+      "loss": 0.1694,
+      "step": 6458
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.345280183271838e-06,
+      "loss": 0.1995,
+      "step": 6459
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.342331241265332e-06,
+      "loss": 0.1831,
+      "step": 6460
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.339382816423897e-06,
+      "loss": 0.1504,
+      "step": 6461
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.33643490907493e-06,
+      "loss": 0.1648,
+      "step": 6462
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.33348751954574e-06,
+      "loss": 0.1921,
+      "step": 6463
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.330540648163598e-06,
+      "loss": 0.1804,
+      "step": 6464
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.327594295255705e-06,
+      "loss": 0.1794,
+      "step": 6465
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3246484611492135e-06,
+      "loss": 0.1488,
+      "step": 6466
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.321703146171204e-06,
+      "loss": 0.136,
+      "step": 6467
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3187583506487295e-06,
+      "loss": 0.1429,
+      "step": 6468
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.31581407490875e-06,
+      "loss": 0.2069,
+      "step": 6469
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3128703192781936e-06,
+      "loss": 0.1587,
+      "step": 6470
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.309927084083912e-06,
+      "loss": 0.1584,
+      "step": 6471
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.306984369652712e-06,
+      "loss": 0.1573,
+      "step": 6472
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.304042176311339e-06,
+      "loss": 0.1562,
+      "step": 6473
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.301100504386485e-06,
+      "loss": 0.1914,
+      "step": 6474
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.298159354204767e-06,
+      "loss": 0.1912,
+      "step": 6475
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.295218726092768e-06,
+      "loss": 0.114,
+      "step": 6476
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.292278620376992e-06,
+      "loss": 0.1807,
+      "step": 6477
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.289339037383896e-06,
+      "loss": 0.2336,
+      "step": 6478
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.286399977439882e-06,
+      "loss": 0.167,
+      "step": 6479
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.283461440871288e-06,
+      "loss": 0.1355,
+      "step": 6480
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.28052342800439e-06,
+      "loss": 0.1577,
+      "step": 6481
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.277585939165417e-06,
+      "loss": 0.1472,
+      "step": 6482
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2746489746805286e-06,
+      "loss": 0.177,
+      "step": 6483
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.271712534875831e-06,
+      "loss": 0.1521,
+      "step": 6484
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.268776620077376e-06,
+      "loss": 0.167,
+      "step": 6485
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.265841230611155e-06,
+      "loss": 0.1633,
+      "step": 6486
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.262906366803092e-06,
+      "loss": 0.2085,
+      "step": 6487
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.259972028979068e-06,
+      "loss": 0.168,
+      "step": 6488
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.257038217464891e-06,
+      "loss": 0.1543,
+      "step": 6489
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.254104932586321e-06,
+      "loss": 0.2041,
+      "step": 6490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.251172174669058e-06,
+      "loss": 0.1548,
+      "step": 6491
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.248239944038736e-06,
+      "loss": 0.1722,
+      "step": 6492
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.245308241020941e-06,
+      "loss": 0.1588,
+      "step": 6493
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.242377065941188e-06,
+      "loss": 0.162,
+      "step": 6494
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.239446419124948e-06,
+      "loss": 0.1704,
+      "step": 6495
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2365163008976195e-06,
+      "loss": 0.1899,
+      "step": 6496
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233586711584558e-06,
+      "loss": 0.1809,
+      "step": 6497
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.23065765151104e-06,
+      "loss": 0.1469,
+      "step": 6498
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.227729121002303e-06,
+      "loss": 0.1946,
+      "step": 6499
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.224801120383509e-06,
+      "loss": 0.1873,
+      "step": 6500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.221873649979775e-06,
+      "loss": 0.1752,
+      "step": 6501
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.218946710116151e-06,
+      "loss": 0.1682,
+      "step": 6502
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.216020301117634e-06,
+      "loss": 0.1995,
+      "step": 6503
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.213094423309151e-06,
+      "loss": 0.1768,
+      "step": 6504
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2101690770155855e-06,
+      "loss": 0.1501,
+      "step": 6505
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.207244262561746e-06,
+      "loss": 0.1829,
+      "step": 6506
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.204319980272394e-06,
+      "loss": 0.1641,
+      "step": 6507
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.201396230472227e-06,
+      "loss": 0.1638,
+      "step": 6508
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.198473013485888e-06,
+      "loss": 0.1304,
+      "step": 6509
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.195550329637951e-06,
+      "loss": 0.1351,
+      "step": 6510
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.192628179252941e-06,
+      "loss": 0.1682,
+      "step": 6511
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.189706562655314e-06,
+      "loss": 0.1755,
+      "step": 6512
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.186785480169476e-06,
+      "loss": 0.1846,
+      "step": 6513
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.183864932119769e-06,
+      "loss": 0.2124,
+      "step": 6514
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.180944918830481e-06,
+      "loss": 0.1921,
+      "step": 6515
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1780254406258305e-06,
+      "loss": 0.1709,
+      "step": 6516
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.175106497829987e-06,
+      "loss": 0.1799,
+      "step": 6517
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.172188090767044e-06,
+      "loss": 0.1794,
+      "step": 6518
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.169270219761065e-06,
+      "loss": 0.1626,
+      "step": 6519
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1663528851360245e-06,
+      "loss": 0.179,
+      "step": 6520
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.163436087215856e-06,
+      "loss": 0.1351,
+      "step": 6521
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.160519826324419e-06,
+      "loss": 0.1897,
+      "step": 6522
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.157604102785525e-06,
+      "loss": 0.1401,
+      "step": 6523
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.154688916922922e-06,
+      "loss": 0.1482,
+      "step": 6524
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.151774269060302e-06,
+      "loss": 0.1711,
+      "step": 6525
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.148860159521287e-06,
+      "loss": 0.1655,
+      "step": 6526
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1459465886294515e-06,
+      "loss": 0.1561,
+      "step": 6527
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1430335567082965e-06,
+      "loss": 0.139,
+      "step": 6528
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.140121064081277e-06,
+      "loss": 0.1499,
+      "step": 6529
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.13720911107178e-06,
+      "loss": 0.1467,
+      "step": 6530
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134297698003141e-06,
+      "loss": 0.1814,
+      "step": 6531
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.131386825198619e-06,
+      "loss": 0.1848,
+      "step": 6532
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.128476492981432e-06,
+      "loss": 0.1951,
+      "step": 6533
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.12556670167472e-06,
+      "loss": 0.2085,
+      "step": 6534
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122657451601581e-06,
+      "loss": 0.1514,
+      "step": 6535
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.119748743085038e-06,
+      "loss": 0.1581,
+      "step": 6536
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.116840576448067e-06,
+      "loss": 0.1709,
+      "step": 6537
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.113932952013569e-06,
+      "loss": 0.1506,
+      "step": 6538
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1110258701044e-06,
+      "loss": 0.167,
+      "step": 6539
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1081193310433395e-06,
+      "loss": 0.1877,
+      "step": 6540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.105213335153123e-06,
+      "loss": 0.1619,
+      "step": 6541
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.102307882756415e-06,
+      "loss": 0.1575,
+      "step": 6542
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.099402974175825e-06,
+      "loss": 0.152,
+      "step": 6543
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.096498609733898e-06,
+      "loss": 0.1536,
+      "step": 6544
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0935947897531235e-06,
+      "loss": 0.1707,
+      "step": 6545
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.090691514555922e-06,
+      "loss": 0.1455,
+      "step": 6546
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0877887844646626e-06,
+      "loss": 0.1902,
+      "step": 6547
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.08488659980165e-06,
+      "loss": 0.1548,
+      "step": 6548
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.081984960889135e-06,
+      "loss": 0.176,
+      "step": 6549
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.079083868049292e-06,
+      "loss": 0.1436,
+      "step": 6550
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.076183321604251e-06,
+      "loss": 0.1727,
+      "step": 6551
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.073283321876068e-06,
+      "loss": 0.173,
+      "step": 6552
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.07038386918675e-06,
+      "loss": 0.1588,
+      "step": 6553
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.067484963858237e-06,
+      "loss": 0.1453,
+      "step": 6554
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0645866062124135e-06,
+      "loss": 0.1456,
+      "step": 6555
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.061688796571095e-06,
+      "loss": 0.1627,
+      "step": 6556
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.058791535256037e-06,
+      "loss": 0.1611,
+      "step": 6557
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0558948225889405e-06,
+      "loss": 0.1799,
+      "step": 6558
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0529986588914435e-06,
+      "loss": 0.2083,
+      "step": 6559
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.050103044485126e-06,
+      "loss": 0.1741,
+      "step": 6560
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.047207979691496e-06,
+      "loss": 0.1594,
+      "step": 6561
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0443134648320115e-06,
+      "loss": 0.1431,
+      "step": 6562
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.041419500228062e-06,
+      "loss": 0.1586,
+      "step": 6563
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.038526086200981e-06,
+      "loss": 0.1951,
+      "step": 6564
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.035633223072041e-06,
+      "loss": 0.1396,
+      "step": 6565
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.032740911162452e-06,
+      "loss": 0.1748,
+      "step": 6566
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.02984915079336e-06,
+      "loss": 0.1624,
+      "step": 6567
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0269579422858546e-06,
+      "loss": 0.1541,
+      "step": 6568
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.024067285960953e-06,
+      "loss": 0.1538,
+      "step": 6569
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.021177182139634e-06,
+      "loss": 0.1609,
+      "step": 6570
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.01828763114279e-06,
+      "loss": 0.1975,
+      "step": 6571
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.01539863329127e-06,
+      "loss": 0.123,
+      "step": 6572
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.012510188905847e-06,
+      "loss": 0.1478,
+      "step": 6573
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.009622298307247e-06,
+      "loss": 0.1841,
+      "step": 6574
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.006734961816118e-06,
+      "loss": 0.1687,
+      "step": 6575
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.003848179753068e-06,
+      "loss": 0.1738,
+      "step": 6576
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000961952438621e-06,
+      "loss": 0.1868,
+      "step": 6577
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.998076280193259e-06,
+      "loss": 0.155,
+      "step": 6578
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995191163337385e-06,
+      "loss": 0.145,
+      "step": 6579
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.992306602191351e-06,
+      "loss": 0.1353,
+      "step": 6580
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.989422597075446e-06,
+      "loss": 0.1356,
+      "step": 6581
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9865391483098994e-06,
+      "loss": 0.1646,
+      "step": 6582
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983656256214866e-06,
+      "loss": 0.1771,
+      "step": 6583
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.98077392111046e-06,
+      "loss": 0.1586,
+      "step": 6584
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.97789214331671e-06,
+      "loss": 0.1313,
+      "step": 6585
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9750109231536015e-06,
+      "loss": 0.1326,
+      "step": 6586
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9721302609410505e-06,
+      "loss": 0.157,
+      "step": 6587
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9692501569989145e-06,
+      "loss": 0.1663,
+      "step": 6588
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.966370611646981e-06,
+      "loss": 0.1284,
+      "step": 6589
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.963491625204987e-06,
+      "loss": 0.1953,
+      "step": 6590
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.960613197992593e-06,
+      "loss": 0.1321,
+      "step": 6591
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.95773533032941e-06,
+      "loss": 0.1639,
+      "step": 6592
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.954858022534984e-06,
+      "loss": 0.1978,
+      "step": 6593
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9519812749288e-06,
+      "loss": 0.1517,
+      "step": 6594
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9491050878302694e-06,
+      "loss": 0.1538,
+      "step": 6595
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.946229461558759e-06,
+      "loss": 0.1713,
+      "step": 6596
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.943354396433557e-06,
+      "loss": 0.1877,
+      "step": 6597
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.940479892773902e-06,
+      "loss": 0.1931,
+      "step": 6598
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.937605950898962e-06,
+      "loss": 0.1709,
+      "step": 6599
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.93473257112785e-06,
+      "loss": 0.1298,
+      "step": 6600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.931859753779608e-06,
+      "loss": 0.1506,
+      "step": 6601
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.928987499173222e-06,
+      "loss": 0.1545,
+      "step": 6602
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.92611580762761e-06,
+      "loss": 0.2021,
+      "step": 6603
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.923244679461634e-06,
+      "loss": 0.1887,
+      "step": 6604
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9203741149940874e-06,
+      "loss": 0.1635,
+      "step": 6605
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.917504114543711e-06,
+      "loss": 0.1736,
+      "step": 6606
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.914634678429166e-06,
+      "loss": 0.2036,
+      "step": 6607
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.911765806969071e-06,
+      "loss": 0.1521,
+      "step": 6608
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9088975004819604e-06,
+      "loss": 0.1777,
+      "step": 6609
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.906029759286324e-06,
+      "loss": 0.1632,
+      "step": 6610
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9031625837005795e-06,
+      "loss": 0.1584,
+      "step": 6611
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.90029597404309e-06,
+      "loss": 0.1714,
+      "step": 6612
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.897429930632141e-06,
+      "loss": 0.1716,
+      "step": 6613
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.894564453785972e-06,
+      "loss": 0.1575,
+      "step": 6614
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.891699543822743e-06,
+      "loss": 0.1582,
+      "step": 6615
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.888835201060571e-06,
+      "loss": 0.158,
+      "step": 6616
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.88597142581749e-06,
+      "loss": 0.1707,
+      "step": 6617
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.883108218411485e-06,
+      "loss": 0.177,
+      "step": 6618
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8802455791604684e-06,
+      "loss": 0.1819,
+      "step": 6619
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8773835083823e-06,
+      "loss": 0.1669,
+      "step": 6620
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.874522006394757e-06,
+      "loss": 0.1537,
+      "step": 6621
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.871661073515585e-06,
+      "loss": 0.1643,
+      "step": 6622
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.868800710062438e-06,
+      "loss": 0.1859,
+      "step": 6623
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8659409163529155e-06,
+      "loss": 0.1497,
+      "step": 6624
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.863081692704561e-06,
+      "loss": 0.1666,
+      "step": 6625
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8602230394348395e-06,
+      "loss": 0.1597,
+      "step": 6626
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.857364956861175e-06,
+      "loss": 0.1725,
+      "step": 6627
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.854507445300905e-06,
+      "loss": 0.1895,
+      "step": 6628
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.851650505071321e-06,
+      "loss": 0.1541,
+      "step": 6629
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.848794136489637e-06,
+      "loss": 0.1416,
+      "step": 6630
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.845938339873013e-06,
+      "loss": 0.1714,
+      "step": 6631
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.843083115538544e-06,
+      "loss": 0.1387,
+      "step": 6632
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.840228463803264e-06,
+      "loss": 0.167,
+      "step": 6633
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.837374384984132e-06,
+      "loss": 0.1638,
+      "step": 6634
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.834520879398058e-06,
+      "loss": 0.1963,
+      "step": 6635
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.831667947361874e-06,
+      "loss": 0.1539,
+      "step": 6636
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.828815589192362e-06,
+      "loss": 0.1697,
+      "step": 6637
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8259638052062304e-06,
+      "loss": 0.1917,
+      "step": 6638
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.823112595720134e-06,
+      "loss": 0.1707,
+      "step": 6639
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.820261961050648e-06,
+      "loss": 0.1851,
+      "step": 6640
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.817411901514303e-06,
+      "loss": 0.187,
+      "step": 6641
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.814562417427547e-06,
+      "loss": 0.1323,
+      "step": 6642
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.811713509106776e-06,
+      "loss": 0.1863,
+      "step": 6643
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.80886517686832e-06,
+      "loss": 0.1509,
+      "step": 6644
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.806017421028447e-06,
+      "loss": 0.1147,
+      "step": 6645
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.803170241903352e-06,
+      "loss": 0.1442,
+      "step": 6646
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.800323639809179e-06,
+      "loss": 0.1558,
+      "step": 6647
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.797477615061992e-06,
+      "loss": 0.1763,
+      "step": 6648
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.794632167977805e-06,
+      "loss": 0.1776,
+      "step": 6649
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.791787298872563e-06,
+      "loss": 0.1919,
+      "step": 6650
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.78894300806215e-06,
+      "loss": 0.1564,
+      "step": 6651
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.786099295862374e-06,
+      "loss": 0.1628,
+      "step": 6652
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.783256162588996e-06,
+      "loss": 0.1519,
+      "step": 6653
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.780413608557694e-06,
+      "loss": 0.1721,
+      "step": 6654
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.777571634084099e-06,
+      "loss": 0.1487,
+      "step": 6655
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774730239483767e-06,
+      "loss": 0.168,
+      "step": 6656
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.771889425072197e-06,
+      "loss": 0.1873,
+      "step": 6657
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.769049191164813e-06,
+      "loss": 0.1624,
+      "step": 6658
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.766209538076987e-06,
+      "loss": 0.1785,
+      "step": 6659
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7633704661240135e-06,
+      "loss": 0.1737,
+      "step": 6660
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7605319756211335e-06,
+      "loss": 0.167,
+      "step": 6661
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.757694066883518e-06,
+      "loss": 0.1721,
+      "step": 6662
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.754856740226282e-06,
+      "loss": 0.1625,
+      "step": 6663
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.752019995964456e-06,
+      "loss": 0.1458,
+      "step": 6664
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.749183834413028e-06,
+      "loss": 0.1597,
+      "step": 6665
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746348255886902e-06,
+      "loss": 0.1667,
+      "step": 6666
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.743513260700942e-06,
+      "loss": 0.1655,
+      "step": 6667
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.740678849169917e-06,
+      "loss": 0.1487,
+      "step": 6668
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.737845021608558e-06,
+      "loss": 0.1656,
+      "step": 6669
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735011778331511e-06,
+      "loss": 0.1698,
+      "step": 6670
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.732179119653373e-06,
+      "loss": 0.1851,
+      "step": 6671
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.729347045888657e-06,
+      "loss": 0.1414,
+      "step": 6672
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.726515557351837e-06,
+      "loss": 0.1414,
+      "step": 6673
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.723684654357299e-06,
+      "loss": 0.1836,
+      "step": 6674
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.72085433721938e-06,
+      "loss": 0.1465,
+      "step": 6675
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.718024606252335e-06,
+      "loss": 0.188,
+      "step": 6676
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7151954617703686e-06,
+      "loss": 0.1672,
+      "step": 6677
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7123669040876164e-06,
+      "loss": 0.1384,
+      "step": 6678
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.70953893351815e-06,
+      "loss": 0.1625,
+      "step": 6679
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.7067115503759665e-06,
+      "loss": 0.1592,
+      "step": 6680
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.703884754975013e-06,
+      "loss": 0.178,
+      "step": 6681
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.701058547629156e-06,
+      "loss": 0.1829,
+      "step": 6682
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.698232928652206e-06,
+      "loss": 0.1719,
+      "step": 6683
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.695407898357909e-06,
+      "loss": 0.1597,
+      "step": 6684
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.692583457059944e-06,
+      "loss": 0.1465,
+      "step": 6685
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.689759605071916e-06,
+      "loss": 0.1633,
+      "step": 6686
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.68693634270738e-06,
+      "loss": 0.1219,
+      "step": 6687
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.684113670279811e-06,
+      "loss": 0.1885,
+      "step": 6688
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6812915881026265e-06,
+      "loss": 0.1655,
+      "step": 6689
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.678470096489182e-06,
+      "loss": 0.1621,
+      "step": 6690
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6756491957527535e-06,
+      "loss": 0.1569,
+      "step": 6691
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6728288862065705e-06,
+      "loss": 0.1252,
+      "step": 6692
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.670009168163777e-06,
+      "loss": 0.175,
+      "step": 6693
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.667190041937464e-06,
+      "loss": 0.1133,
+      "step": 6694
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.664371507840656e-06,
+      "loss": 0.1561,
+      "step": 6695
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.661553566186311e-06,
+      "loss": 0.174,
+      "step": 6696
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.658736217287313e-06,
+      "loss": 0.1716,
+      "step": 6697
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.655919461456495e-06,
+      "loss": 0.1936,
+      "step": 6698
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6531032990066075e-06,
+      "loss": 0.1534,
+      "step": 6699
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.650287730250347e-06,
+      "loss": 0.1785,
+      "step": 6700
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.647472755500342e-06,
+      "loss": 0.1322,
+      "step": 6701
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.644658375069157e-06,
+      "loss": 0.1814,
+      "step": 6702
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6418445892692794e-06,
+      "loss": 0.1464,
+      "step": 6703
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.639031398413146e-06,
+      "loss": 0.1914,
+      "step": 6704
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.636218802813113e-06,
+      "loss": 0.1415,
+      "step": 6705
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6334068027814805e-06,
+      "loss": 0.1659,
+      "step": 6706
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.630595398630479e-06,
+      "loss": 0.1743,
+      "step": 6707
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.627784590672278e-06,
+      "loss": 0.1562,
+      "step": 6708
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.624974379218968e-06,
+      "loss": 0.1555,
+      "step": 6709
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.622164764582589e-06,
+      "loss": 0.158,
+      "step": 6710
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6193557470751e-06,
+      "loss": 0.1343,
+      "step": 6711
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.616547327008405e-06,
+      "loss": 0.1855,
+      "step": 6712
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6137395046943355e-06,
+      "loss": 0.172,
+      "step": 6713
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.610932280444665e-06,
+      "loss": 0.1783,
+      "step": 6714
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.608125654571083e-06,
+      "loss": 0.1758,
+      "step": 6715
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.605319627385233e-06,
+      "loss": 0.1437,
+      "step": 6716
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.602514199198677e-06,
+      "loss": 0.1598,
+      "step": 6717
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.599709370322917e-06,
+      "loss": 0.1552,
+      "step": 6718
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.59690514106939e-06,
+      "loss": 0.1646,
+      "step": 6719
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.594101511749466e-06,
+      "loss": 0.1743,
+      "step": 6720
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591298482674441e-06,
+      "loss": 0.1794,
+      "step": 6721
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5884960541555556e-06,
+      "loss": 0.1667,
+      "step": 6722
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.585694226503967e-06,
+      "loss": 0.1694,
+      "step": 6723
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.582893000030793e-06,
+      "loss": 0.153,
+      "step": 6724
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.580092375047055e-06,
+      "loss": 0.1908,
+      "step": 6725
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.577292351863731e-06,
+      "loss": 0.1804,
+      "step": 6726
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.574492930791711e-06,
+      "loss": 0.189,
+      "step": 6727
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.571694112141835e-06,
+      "loss": 0.1575,
+      "step": 6728
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.568895896224872e-06,
+      "loss": 0.1687,
+      "step": 6729
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.566098283351523e-06,
+      "loss": 0.1572,
+      "step": 6730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5633012738324156e-06,
+      "loss": 0.1327,
+      "step": 6731
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560504867978124e-06,
+      "loss": 0.1661,
+      "step": 6732
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5577090660991385e-06,
+      "loss": 0.1592,
+      "step": 6733
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5549138685058965e-06,
+      "loss": 0.168,
+      "step": 6734
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.552119275508764e-06,
+      "loss": 0.1554,
+      "step": 6735
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.549325287418042e-06,
+      "loss": 0.1699,
+      "step": 6736
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.546531904543954e-06,
+      "loss": 0.1621,
+      "step": 6737
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.54373912719667e-06,
+      "loss": 0.152,
+      "step": 6738
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5409469556862815e-06,
+      "loss": 0.1882,
+      "step": 6739
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.538155390322819e-06,
+      "loss": 0.1249,
+      "step": 6740
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.535364431416247e-06,
+      "loss": 0.1707,
+      "step": 6741
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.532574079276462e-06,
+      "loss": 0.1628,
+      "step": 6742
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.529784334213284e-06,
+      "loss": 0.1631,
+      "step": 6743
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.526995196536481e-06,
+      "loss": 0.1664,
+      "step": 6744
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.524206666555738e-06,
+      "loss": 0.1611,
+      "step": 6745
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.521418744580683e-06,
+      "loss": 0.2144,
+      "step": 6746
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.518631430920873e-06,
+      "loss": 0.2031,
+      "step": 6747
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.515844725885803e-06,
+      "loss": 0.1782,
+      "step": 6748
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.513058629784887e-06,
+      "loss": 0.1636,
+      "step": 6749
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.510273142927487e-06,
+      "loss": 0.1487,
+      "step": 6750
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.507488265622882e-06,
+      "loss": 0.1882,
+      "step": 6751
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5047039981802975e-06,
+      "loss": 0.1665,
+      "step": 6752
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.501920340908883e-06,
+      "loss": 0.1475,
+      "step": 6753
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.4991372941177255e-06,
+      "loss": 0.1611,
+      "step": 6754
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.496354858115835e-06,
+      "loss": 0.1312,
+      "step": 6755
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.493573033212169e-06,
+      "loss": 0.1555,
+      "step": 6756
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.490791819715597e-06,
+      "loss": 0.1824,
+      "step": 6757
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.488011217934936e-06,
+      "loss": 0.179,
+      "step": 6758
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.485231228178936e-06,
+      "loss": 0.2068,
+      "step": 6759
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.482451850756264e-06,
+      "loss": 0.1851,
+      "step": 6760
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.479673085975539e-06,
+      "loss": 0.1873,
+      "step": 6761
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4768949341452915e-06,
+      "loss": 0.1417,
+      "step": 6762
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474117395573999e-06,
+      "loss": 0.1719,
+      "step": 6763
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.471340470570067e-06,
+      "loss": 0.2,
+      "step": 6764
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.468564159441833e-06,
+      "loss": 0.1575,
+      "step": 6765
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.46578846249756e-06,
+      "loss": 0.1576,
+      "step": 6766
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4630133800454545e-06,
+      "loss": 0.1963,
+      "step": 6767
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4602389123936406e-06,
+      "loss": 0.2115,
+      "step": 6768
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.457465059850185e-06,
+      "loss": 0.1444,
+      "step": 6769
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.454691822723086e-06,
+      "loss": 0.1716,
+      "step": 6770
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.451919201320271e-06,
+      "loss": 0.187,
+      "step": 6771
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.449147195949592e-06,
+      "loss": 0.1608,
+      "step": 6772
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.446375806918847e-06,
+      "loss": 0.1479,
+      "step": 6773
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.443605034535746e-06,
+      "loss": 0.1324,
+      "step": 6774
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.440834879107957e-06,
+      "loss": 0.1641,
+      "step": 6775
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.438065340943054e-06,
+      "loss": 0.1271,
+      "step": 6776
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.435296420348561e-06,
+      "loss": 0.1707,
+      "step": 6777
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.432528117631916e-06,
+      "loss": 0.1373,
+      "step": 6778
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.429760433100508e-06,
+      "loss": 0.1785,
+      "step": 6779
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.426993367061635e-06,
+      "loss": 0.2095,
+      "step": 6780
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.424226919822554e-06,
+      "loss": 0.1388,
+      "step": 6781
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.421461091690425e-06,
+      "loss": 0.1848,
+      "step": 6782
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.418695882972364e-06,
+      "loss": 0.135,
+      "step": 6783
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.415931293975394e-06,
+      "loss": 0.1516,
+      "step": 6784
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.413167325006488e-06,
+      "loss": 0.1577,
+      "step": 6785
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.410403976372544e-06,
+      "loss": 0.1794,
+      "step": 6786
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407641248380392e-06,
+      "loss": 0.1338,
+      "step": 6787
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.404879141336788e-06,
+      "loss": 0.1626,
+      "step": 6788
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.402117655548429e-06,
+      "loss": 0.1704,
+      "step": 6789
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3993567913219295e-06,
+      "loss": 0.1637,
+      "step": 6790
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3965965489638466e-06,
+      "loss": 0.1338,
+      "step": 6791
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.393836928780664e-06,
+      "loss": 0.1711,
+      "step": 6792
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3910779310788e-06,
+      "loss": 0.1375,
+      "step": 6793
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.388319556164593e-06,
+      "loss": 0.1487,
+      "step": 6794
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.385561804344328e-06,
+      "loss": 0.1329,
+      "step": 6795
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3828046759242035e-06,
+      "loss": 0.165,
+      "step": 6796
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.380048171210364e-06,
+      "loss": 0.1417,
+      "step": 6797
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.377292290508876e-06,
+      "loss": 0.1637,
+      "step": 6798
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.374537034125744e-06,
+      "loss": 0.1768,
+      "step": 6799
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.371782402366892e-06,
+      "loss": 0.1387,
+      "step": 6800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.369028395538186e-06,
+      "loss": 0.166,
+      "step": 6801
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3662750139454115e-06,
+      "loss": 0.2058,
+      "step": 6802
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.363522257894295e-06,
+      "loss": 0.1698,
+      "step": 6803
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.360770127690488e-06,
+      "loss": 0.1556,
+      "step": 6804
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.358018623639578e-06,
+      "loss": 0.1733,
+      "step": 6805
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.355267746047072e-06,
+      "loss": 0.1859,
+      "step": 6806
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.352517495218421e-06,
+      "loss": 0.1543,
+      "step": 6807
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349767871458993e-06,
+      "loss": 0.1611,
+      "step": 6808
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.347018875074095e-06,
+      "loss": 0.1897,
+      "step": 6809
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.344270506368964e-06,
+      "loss": 0.1868,
+      "step": 6810
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.341522765648768e-06,
+      "loss": 0.1814,
+      "step": 6811
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.338775653218595e-06,
+      "loss": 0.1664,
+      "step": 6812
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.336029169383481e-06,
+      "loss": 0.1616,
+      "step": 6813
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.333283314448373e-06,
+      "loss": 0.1917,
+      "step": 6814
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.330538088718162e-06,
+      "loss": 0.1558,
+      "step": 6815
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.327793492497664e-06,
+      "loss": 0.1768,
+      "step": 6816
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.325049526091629e-06,
+      "loss": 0.1289,
+      "step": 6817
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.322306189804728e-06,
+      "loss": 0.1241,
+      "step": 6818
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.319563483941574e-06,
+      "loss": 0.1733,
+      "step": 6819
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.316821408806694e-06,
+      "loss": 0.1069,
+      "step": 6820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.314079964704568e-06,
+      "loss": 0.1902,
+      "step": 6821
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.311339151939582e-06,
+      "loss": 0.1499,
+      "step": 6822
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3085989708160705e-06,
+      "loss": 0.175,
+      "step": 6823
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.305859421638282e-06,
+      "loss": 0.1682,
+      "step": 6824
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3031205047104076e-06,
+      "loss": 0.1775,
+      "step": 6825
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.300382220336567e-06,
+      "loss": 0.1473,
+      "step": 6826
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.297644568820797e-06,
+      "loss": 0.1733,
+      "step": 6827
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.294907550467083e-06,
+      "loss": 0.178,
+      "step": 6828
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.292171165579319e-06,
+      "loss": 0.2007,
+      "step": 6829
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2894354144613525e-06,
+      "loss": 0.1777,
+      "step": 6830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.286700297416935e-06,
+      "loss": 0.1276,
+      "step": 6831
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.283965814749774e-06,
+      "loss": 0.1494,
+      "step": 6832
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.281231966763484e-06,
+      "loss": 0.1689,
+      "step": 6833
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.278498753761624e-06,
+      "loss": 0.1761,
+      "step": 6834
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.275766176047672e-06,
+      "loss": 0.1758,
+      "step": 6835
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.273034233925041e-06,
+      "loss": 0.1359,
+      "step": 6836
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.270302927697076e-06,
+      "loss": 0.1561,
+      "step": 6837
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.267572257667048e-06,
+      "loss": 0.1641,
+      "step": 6838
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.264842224138151e-06,
+      "loss": 0.1978,
+      "step": 6839
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.262112827413525e-06,
+      "loss": 0.1399,
+      "step": 6840
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.259384067796219e-06,
+      "loss": 0.1433,
+      "step": 6841
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.256655945589225e-06,
+      "loss": 0.132,
+      "step": 6842
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253928461095461e-06,
+      "loss": 0.1515,
+      "step": 6843
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2512016146177794e-06,
+      "loss": 0.186,
+      "step": 6844
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.248475406458945e-06,
+      "loss": 0.1674,
+      "step": 6845
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.245749836921673e-06,
+      "loss": 0.1853,
+      "step": 6846
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2430249063085896e-06,
+      "loss": 0.108,
+      "step": 6847
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.24030061492226e-06,
+      "loss": 0.1904,
+      "step": 6848
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.237576963065179e-06,
+      "loss": 0.1606,
+      "step": 6849
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23485395103977e-06,
+      "loss": 0.1337,
+      "step": 6850
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.232131579148376e-06,
+      "loss": 0.165,
+      "step": 6851
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2294098476932824e-06,
+      "loss": 0.1454,
+      "step": 6852
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.22668875697669e-06,
+      "loss": 0.1281,
+      "step": 6853
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.223968307300741e-06,
+      "loss": 0.1509,
+      "step": 6854
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2212484989675e-06,
+      "loss": 0.1682,
+      "step": 6855
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.218529332278966e-06,
+      "loss": 0.1787,
+      "step": 6856
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.215810807537052e-06,
+      "loss": 0.1399,
+      "step": 6857
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.213092925043619e-06,
+      "loss": 0.157,
+      "step": 6858
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.210375685100442e-06,
+      "loss": 0.1956,
+      "step": 6859
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.20765908800923e-06,
+      "loss": 0.1545,
+      "step": 6860
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.204943134071625e-06,
+      "loss": 0.1285,
+      "step": 6861
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2022278235891944e-06,
+      "loss": 0.1466,
+      "step": 6862
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1995131568634265e-06,
+      "loss": 0.1445,
+      "step": 6863
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1967991341957525e-06,
+      "loss": 0.1633,
+      "step": 6864
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1940857558875155e-06,
+      "loss": 0.176,
+      "step": 6865
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.191373022240002e-06,
+      "loss": 0.1647,
+      "step": 6866
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.188660933554419e-06,
+      "loss": 0.1444,
+      "step": 6867
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.18594949013191e-06,
+      "loss": 0.1627,
+      "step": 6868
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.183238692273529e-06,
+      "loss": 0.1958,
+      "step": 6869
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1805285402802805e-06,
+      "loss": 0.2031,
+      "step": 6870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.177819034453076e-06,
+      "loss": 0.1448,
+      "step": 6871
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.175110175092778e-06,
+      "loss": 0.1173,
+      "step": 6872
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.172401962500156e-06,
+      "loss": 0.2087,
+      "step": 6873
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.169694396975924e-06,
+      "loss": 0.1743,
+      "step": 6874
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1669874788207095e-06,
+      "loss": 0.1572,
+      "step": 6875
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.164281208335083e-06,
+      "loss": 0.1306,
+      "step": 6876
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.161575585819526e-06,
+      "loss": 0.1685,
+      "step": 6877
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.158870611574471e-06,
+      "loss": 0.1215,
+      "step": 6878
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1561662859002526e-06,
+      "loss": 0.1768,
+      "step": 6879
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1534626090971576e-06,
+      "loss": 0.176,
+      "step": 6880
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.150759581465378e-06,
+      "loss": 0.1541,
+      "step": 6881
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1480572033050505e-06,
+      "loss": 0.1735,
+      "step": 6882
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.145355474916234e-06,
+      "loss": 0.1556,
+      "step": 6883
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1426543965989195e-06,
+      "loss": 0.2002,
+      "step": 6884
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.139953968653013e-06,
+      "loss": 0.1746,
+      "step": 6885
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.137254191378366e-06,
+      "loss": 0.1458,
+      "step": 6886
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.13455506507474e-06,
+      "loss": 0.1931,
+      "step": 6887
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131856590041837e-06,
+      "loss": 0.1469,
+      "step": 6888
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.129158766579284e-06,
+      "loss": 0.1534,
+      "step": 6889
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.126461594986636e-06,
+      "loss": 0.1497,
+      "step": 6890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.123765075563367e-06,
+      "loss": 0.1719,
+      "step": 6891
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.121069208608894e-06,
+      "loss": 0.1613,
+      "step": 6892
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.1183739944225455e-06,
+      "loss": 0.1406,
+      "step": 6893
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.115679433303587e-06,
+      "loss": 0.1891,
+      "step": 6894
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.112985525551216e-06,
+      "loss": 0.1606,
+      "step": 6895
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.110292271464541e-06,
+      "loss": 0.1432,
+      "step": 6896
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.107599671342617e-06,
+      "loss": 0.1222,
+      "step": 6897
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.104907725484409e-06,
+      "loss": 0.1792,
+      "step": 6898
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.102216434188823e-06,
+      "loss": 0.1495,
+      "step": 6899
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.099525797754686e-06,
+      "loss": 0.1594,
+      "step": 6900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.096835816480755e-06,
+      "loss": 0.1212,
+      "step": 6901
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.094146490665708e-06,
+      "loss": 0.1794,
+      "step": 6902
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.091457820608162e-06,
+      "loss": 0.1525,
+      "step": 6903
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.088769806606644e-06,
+      "loss": 0.1479,
+      "step": 6904
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.086082448959624e-06,
+      "loss": 0.175,
+      "step": 6905
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.083395747965494e-06,
+      "loss": 0.1492,
+      "step": 6906
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.080709703922574e-06,
+      "loss": 0.1364,
+      "step": 6907
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.078024317129102e-06,
+      "loss": 0.2109,
+      "step": 6908
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.075339587883259e-06,
+      "loss": 0.1882,
+      "step": 6909
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.072655516483137e-06,
+      "loss": 0.1611,
+      "step": 6910
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.069972103226766e-06,
+      "loss": 0.1333,
+      "step": 6911
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.067289348412099e-06,
+      "loss": 0.1442,
+      "step": 6912
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.06460725233702e-06,
+      "loss": 0.1412,
+      "step": 6913
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.061925815299327e-06,
+      "loss": 0.1727,
+      "step": 6914
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.059245037596763e-06,
+      "loss": 0.1891,
+      "step": 6915
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0565649195269804e-06,
+      "loss": 0.1646,
+      "step": 6916
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0538854613875726e-06,
+      "loss": 0.1838,
+      "step": 6917
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.051206663476049e-06,
+      "loss": 0.2031,
+      "step": 6918
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.048528526089859e-06,
+      "loss": 0.1493,
+      "step": 6919
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.04585104952636e-06,
+      "loss": 0.1578,
+      "step": 6920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.043174234082854e-06,
+      "loss": 0.1635,
+      "step": 6921
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.040498080056555e-06,
+      "loss": 0.1499,
+      "step": 6922
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0378225877446135e-06,
+      "loss": 0.1621,
+      "step": 6923
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.035147757444102e-06,
+      "loss": 0.1447,
+      "step": 6924
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0324735894520265e-06,
+      "loss": 0.1492,
+      "step": 6925
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.029800084065304e-06,
+      "loss": 0.1838,
+      "step": 6926
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.027127241580797e-06,
+      "loss": 0.2041,
+      "step": 6927
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.024455062295274e-06,
+      "loss": 0.1293,
+      "step": 6928
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.021783546505455e-06,
+      "loss": 0.1807,
+      "step": 6929
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0191126945079604e-06,
+      "loss": 0.1484,
+      "step": 6930
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.016442506599357e-06,
+      "loss": 0.1456,
+      "step": 6931
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.013772983076123e-06,
+      "loss": 0.1956,
+      "step": 6932
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.011104124234672e-06,
+      "loss": 0.1689,
+      "step": 6933
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.00843593037134e-06,
+      "loss": 0.1493,
+      "step": 6934
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.005768401782396e-06,
+      "loss": 0.1417,
+      "step": 6935
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.003101538764023e-06,
+      "loss": 0.1943,
+      "step": 6936
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.000435341612341e-06,
+      "loss": 0.1875,
+      "step": 6937
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.997769810623386e-06,
+      "loss": 0.144,
+      "step": 6938
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.995104946093131e-06,
+      "loss": 0.1461,
+      "step": 6939
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.992440748317466e-06,
+      "loss": 0.2031,
+      "step": 6940
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.989777217592218e-06,
+      "loss": 0.1638,
+      "step": 6941
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.987114354213123e-06,
+      "loss": 0.1687,
+      "step": 6942
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98445215847586e-06,
+      "loss": 0.1608,
+      "step": 6943
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.98179063067602e-06,
+      "loss": 0.1819,
+      "step": 6944
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.97912977110913e-06,
+      "loss": 0.1354,
+      "step": 6945
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.976469580070638e-06,
+      "loss": 0.1727,
+      "step": 6946
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.973810057855922e-06,
+      "loss": 0.1631,
+      "step": 6947
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.971151204760277e-06,
+      "loss": 0.1776,
+      "step": 6948
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.968493021078935e-06,
+      "loss": 0.1683,
+      "step": 6949
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.965835507107042e-06,
+      "loss": 0.147,
+      "step": 6950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.963178663139678e-06,
+      "loss": 0.2029,
+      "step": 6951
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.960522489471847e-06,
+      "loss": 0.1619,
+      "step": 6952
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.95786698639848e-06,
+      "loss": 0.1691,
+      "step": 6953
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.955212154214425e-06,
+      "loss": 0.1912,
+      "step": 6954
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.952557993214468e-06,
+      "loss": 0.1658,
+      "step": 6955
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.94990450369331e-06,
+      "loss": 0.1844,
+      "step": 6956
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.947251685945582e-06,
+      "loss": 0.1514,
+      "step": 6957
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.944599540265842e-06,
+      "loss": 0.1641,
+      "step": 6958
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.941948066948574e-06,
+      "loss": 0.1327,
+      "step": 6959
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9392972662881815e-06,
+      "loss": 0.1625,
+      "step": 6960
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.936647138578993e-06,
+      "loss": 0.165,
+      "step": 6961
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.93399768411527e-06,
+      "loss": 0.1672,
+      "step": 6962
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.931348903191193e-06,
+      "loss": 0.1565,
+      "step": 6963
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9287007961008775e-06,
+      "loss": 0.1523,
+      "step": 6964
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.926053363138346e-06,
+      "loss": 0.144,
+      "step": 6965
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.923406604597565e-06,
+      "loss": 0.1383,
+      "step": 6966
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.920760520772411e-06,
+      "loss": 0.1892,
+      "step": 6967
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.918115111956695e-06,
+      "loss": 0.1584,
+      "step": 6968
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.915470378444151e-06,
+      "loss": 0.1349,
+      "step": 6969
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.912826320528441e-06,
+      "loss": 0.1398,
+      "step": 6970
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.910182938503141e-06,
+      "loss": 0.1569,
+      "step": 6971
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.907540232661767e-06,
+      "loss": 0.1876,
+      "step": 6972
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.904898203297746e-06,
+      "loss": 0.1813,
+      "step": 6973
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.902256850704438e-06,
+      "loss": 0.176,
+      "step": 6974
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.899616175175126e-06,
+      "loss": 0.1304,
+      "step": 6975
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8969761770030235e-06,
+      "loss": 0.1624,
+      "step": 6976
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8943368564812545e-06,
+      "loss": 0.1631,
+      "step": 6977
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8916982139028835e-06,
+      "loss": 0.1921,
+      "step": 6978
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.889060249560883e-06,
+      "loss": 0.1956,
+      "step": 6979
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.886422963748173e-06,
+      "loss": 0.1375,
+      "step": 6980
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.883786356757575e-06,
+      "loss": 0.1936,
+      "step": 6981
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8811504288818515e-06,
+      "loss": 0.1765,
+      "step": 6982
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.878515180413676e-06,
+      "loss": 0.1796,
+      "step": 6983
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.875880611645662e-06,
+      "loss": 0.1699,
+      "step": 6984
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.87324672287033e-06,
+      "loss": 0.1743,
+      "step": 6985
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.870613514380144e-06,
+      "loss": 0.1941,
+      "step": 6986
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.867980986467475e-06,
+      "loss": 0.1382,
+      "step": 6987
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.865349139424634e-06,
+      "loss": 0.1238,
+      "step": 6988
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.862717973543841e-06,
+      "loss": 0.1831,
+      "step": 6989
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.860087489117251e-06,
+      "loss": 0.1892,
+      "step": 6990
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.857457686436939e-06,
+      "loss": 0.1871,
+      "step": 6991
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8548285657949126e-06,
+      "loss": 0.1511,
+      "step": 6992
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.852200127483087e-06,
+      "loss": 0.1453,
+      "step": 6993
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.849572371793318e-06,
+      "loss": 0.1606,
+      "step": 6994
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.846945299017373e-06,
+      "loss": 0.129,
+      "step": 6995
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8443189094469535e-06,
+      "loss": 0.1799,
+      "step": 6996
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.84169320337368e-06,
+      "loss": 0.1902,
+      "step": 6997
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.839068181089102e-06,
+      "loss": 0.1428,
+      "step": 6998
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8364438428846825e-06,
+      "loss": 0.1619,
+      "step": 6999
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.833820189051822e-06,
+      "loss": 0.2068,
+      "step": 7000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.831197219881833e-06,
+      "loss": 0.1578,
+      "step": 7001
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.82857493566596e-06,
+      "loss": 0.1448,
+      "step": 7002
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.825953336695366e-06,
+      "loss": 0.1477,
+      "step": 7003
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.823332423261148e-06,
+      "loss": 0.1697,
+      "step": 7004
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.820712195654312e-06,
+      "loss": 0.1799,
+      "step": 7005
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8180926541658005e-06,
+      "loss": 0.1704,
+      "step": 7006
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.815473799086469e-06,
+      "loss": 0.1543,
+      "step": 7007
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8128556307071062e-06,
+      "loss": 0.1906,
+      "step": 7008
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8102381493184206e-06,
+      "loss": 0.1426,
+      "step": 7009
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8076213552110487e-06,
+      "loss": 0.1725,
+      "step": 7010
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8050052486755384e-06,
+      "loss": 0.1396,
+      "step": 7011
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8023898300023776e-06,
+      "loss": 0.1885,
+      "step": 7012
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7997750994819626e-06,
+      "loss": 0.1573,
+      "step": 7013
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7971610574046237e-06,
+      "loss": 0.1837,
+      "step": 7014
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7945477040606125e-06,
+      "loss": 0.1929,
+      "step": 7015
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7919350397401054e-06,
+      "loss": 0.1858,
+      "step": 7016
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7893230647331935e-06,
+      "loss": 0.1985,
+      "step": 7017
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7867117793299047e-06,
+      "loss": 0.1438,
+      "step": 7018
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.784101183820178e-06,
+      "loss": 0.1626,
+      "step": 7019
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.781491278493883e-06,
+      "loss": 0.2109,
+      "step": 7020
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7788820636408107e-06,
+      "loss": 0.1897,
+      "step": 7021
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.776273539550681e-06,
+      "loss": 0.1763,
+      "step": 7022
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7736657065131244e-06,
+      "loss": 0.1592,
+      "step": 7023
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7710585648177076e-06,
+      "loss": 0.1768,
+      "step": 7024
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7684521147539065e-06,
+      "loss": 0.2186,
+      "step": 7025
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7658463566111423e-06,
+      "loss": 0.1914,
+      "step": 7026
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7632412906787384e-06,
+      "loss": 0.1553,
+      "step": 7027
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7606369172459445e-06,
+      "loss": 0.168,
+      "step": 7028
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.758033236601946e-06,
+      "loss": 0.157,
+      "step": 7029
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.755430249035832e-06,
+      "loss": 0.1481,
+      "step": 7030
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7528279548366397e-06,
+      "loss": 0.1921,
+      "step": 7031
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.750226354293305e-06,
+      "loss": 0.1324,
+      "step": 7032
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7476254476947025e-06,
+      "loss": 0.1736,
+      "step": 7033
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7450252353296202e-06,
+      "loss": 0.1403,
+      "step": 7034
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7424257174867784e-06,
+      "loss": 0.1997,
+      "step": 7035
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7398268944548043e-06,
+      "loss": 0.1975,
+      "step": 7036
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.737228766522274e-06,
+      "loss": 0.1672,
+      "step": 7037
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7346313339776597e-06,
+      "loss": 0.1401,
+      "step": 7038
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.732034597109374e-06,
+      "loss": 0.1455,
+      "step": 7039
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.72943855620574e-06,
+      "loss": 0.1586,
+      "step": 7040
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7268432115550135e-06,
+      "loss": 0.186,
+      "step": 7041
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7242485634453686e-06,
+      "loss": 0.1727,
+      "step": 7042
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7216546121649054e-06,
+      "loss": 0.1512,
+      "step": 7043
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7190613580016376e-06,
+      "loss": 0.1373,
+      "step": 7044
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7164688012435136e-06,
+      "loss": 0.1875,
+      "step": 7045
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7138769421783925e-06,
+      "loss": 0.1415,
+      "step": 7046
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.711285781094065e-06,
+      "loss": 0.1562,
+      "step": 7047
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7086953182782413e-06,
+      "loss": 0.2302,
+      "step": 7048
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7061055540185576e-06,
+      "loss": 0.1691,
+      "step": 7049
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.70351648860256e-06,
+      "loss": 0.1897,
+      "step": 7050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.700928122317735e-06,
+      "loss": 0.1331,
+      "step": 7051
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6983404554514746e-06,
+      "loss": 0.147,
+      "step": 7052
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.695753488291105e-06,
+      "loss": 0.1785,
+      "step": 7053
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.693167221123869e-06,
+      "loss": 0.1512,
+      "step": 7054
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6905816542369376e-06,
+      "loss": 0.1775,
+      "step": 7055
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.687996787917393e-06,
+      "loss": 0.1478,
+      "step": 7056
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6854126224522522e-06,
+      "loss": 0.1958,
+      "step": 7057
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6828291581284426e-06,
+      "loss": 0.1794,
+      "step": 7058
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6802463952328237e-06,
+      "loss": 0.1892,
+      "step": 7059
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6776643340521713e-06,
+      "loss": 0.1658,
+      "step": 7060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6750829748731885e-06,
+      "loss": 0.1362,
+      "step": 7061
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6725023179824924e-06,
+      "loss": 0.1493,
+      "step": 7062
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6699223636666316e-06,
+      "loss": 0.1483,
+      "step": 7063
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.667343112212065e-06,
+      "loss": 0.1741,
+      "step": 7064
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6647645639051842e-06,
+      "loss": 0.1469,
+      "step": 7065
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.662186719032299e-06,
+      "loss": 0.1667,
+      "step": 7066
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6596095778796424e-06,
+      "loss": 0.1838,
+      "step": 7067
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6570331407333635e-06,
+      "loss": 0.1788,
+      "step": 7068
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6544574078795434e-06,
+      "loss": 0.1445,
+      "step": 7069
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6518823796041724e-06,
+      "loss": 0.1619,
+      "step": 7070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6493080561931713e-06,
+      "loss": 0.1801,
+      "step": 7071
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.646734437932381e-06,
+      "loss": 0.132,
+      "step": 7072
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6441615251075692e-06,
+      "loss": 0.1553,
+      "step": 7073
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6415893180044105e-06,
+      "loss": 0.155,
+      "step": 7074
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.639017816908518e-06,
+      "loss": 0.1711,
+      "step": 7075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.636447022105407e-06,
+      "loss": 0.145,
+      "step": 7076
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6338769338805434e-06,
+      "loss": 0.1641,
+      "step": 7077
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6313075525192843e-06,
+      "loss": 0.1597,
+      "step": 7078
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.628738878306929e-06,
+      "loss": 0.1499,
+      "step": 7079
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.626170911528684e-06,
+      "loss": 0.1599,
+      "step": 7080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6236036524696904e-06,
+      "loss": 0.1417,
+      "step": 7081
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.621037101414995e-06,
+      "loss": 0.1772,
+      "step": 7082
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6184712586495873e-06,
+      "loss": 0.1519,
+      "step": 7083
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.615906124458358e-06,
+      "loss": 0.1741,
+      "step": 7084
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6133416991261316e-06,
+      "loss": 0.2183,
+      "step": 7085
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6107779829376445e-06,
+      "loss": 0.1331,
+      "step": 7086
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6082149761775632e-06,
+      "loss": 0.1636,
+      "step": 7087
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.605652679130469e-06,
+      "loss": 0.1261,
+      "step": 7088
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.6030910920808727e-06,
+      "loss": 0.1606,
+      "step": 7089
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.600530215313194e-06,
+      "loss": 0.1508,
+      "step": 7090
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5979700491117853e-06,
+      "loss": 0.1621,
+      "step": 7091
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5954105937609084e-06,
+      "loss": 0.1646,
+      "step": 7092
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5928518495447583e-06,
+      "loss": 0.1567,
+      "step": 7093
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.590293816747448e-06,
+      "loss": 0.1665,
+      "step": 7094
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5877364956530013e-06,
+      "loss": 0.1877,
+      "step": 7095
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.585179886545379e-06,
+      "loss": 0.1484,
+      "step": 7096
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.582623989708448e-06,
+      "loss": 0.189,
+      "step": 7097
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5800688054260047e-06,
+      "loss": 0.1477,
+      "step": 7098
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5775143339817652e-06,
+      "loss": 0.1874,
+      "step": 7099
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5749605756593697e-06,
+      "loss": 0.1868,
+      "step": 7100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5724075307423667e-06,
+      "loss": 0.1412,
+      "step": 7101
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5698551995142427e-06,
+      "loss": 0.1536,
+      "step": 7102
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.567303582258389e-06,
+      "loss": 0.168,
+      "step": 7103
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.564752679258128e-06,
+      "loss": 0.1522,
+      "step": 7104
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.562202490796699e-06,
+      "loss": 0.1422,
+      "step": 7105
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5596530171572675e-06,
+      "loss": 0.1702,
+      "step": 7106
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5571042586229053e-06,
+      "loss": 0.1794,
+      "step": 7107
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5545562154766245e-06,
+      "loss": 0.1758,
+      "step": 7108
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5520088880013393e-06,
+      "loss": 0.1476,
+      "step": 7109
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5494622764798946e-06,
+      "loss": 0.1631,
+      "step": 7110
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.546916381195056e-06,
+      "loss": 0.1636,
+      "step": 7111
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5443712024295095e-06,
+      "loss": 0.1591,
+      "step": 7112
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5418267404658534e-06,
+      "loss": 0.1848,
+      "step": 7113
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5392829955866194e-06,
+      "loss": 0.1538,
+      "step": 7114
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5367399680742455e-06,
+      "loss": 0.1213,
+      "step": 7115
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5341976582111003e-06,
+      "loss": 0.1882,
+      "step": 7116
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5316560662794697e-06,
+      "loss": 0.1423,
+      "step": 7117
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.529115192561563e-06,
+      "loss": 0.1639,
+      "step": 7118
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.526575037339501e-06,
+      "loss": 0.156,
+      "step": 7119
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5240356008953367e-06,
+      "loss": 0.1775,
+      "step": 7120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5214968835110284e-06,
+      "loss": 0.1664,
+      "step": 7121
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5189588854684684e-06,
+      "loss": 0.1326,
+      "step": 7122
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.516421607049464e-06,
+      "loss": 0.2004,
+      "step": 7123
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5138850485357436e-06,
+      "loss": 0.1522,
+      "step": 7124
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5113492102089496e-06,
+      "loss": 0.1445,
+      "step": 7125
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5088140923506543e-06,
+      "loss": 0.2061,
+      "step": 7126
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5062796952423407e-06,
+      "loss": 0.178,
+      "step": 7127
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.503746019165416e-06,
+      "loss": 0.1809,
+      "step": 7128
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.501213064401211e-06,
+      "loss": 0.1796,
+      "step": 7129
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.498680831230974e-06,
+      "loss": 0.1418,
+      "step": 7130
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4961493199358653e-06,
+      "loss": 0.1414,
+      "step": 7131
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4936185307969796e-06,
+      "loss": 0.1724,
+      "step": 7132
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4910884640953115e-06,
+      "loss": 0.1348,
+      "step": 7133
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.488559120111803e-06,
+      "loss": 0.1768,
+      "step": 7134
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4860304991272887e-06,
+      "loss": 0.1321,
+      "step": 7135
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4835026014225413e-06,
+      "loss": 0.1616,
+      "step": 7136
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4809754272782403e-06,
+      "loss": 0.1794,
+      "step": 7137
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4784489769749953e-06,
+      "loss": 0.1514,
+      "step": 7138
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4759232507933284e-06,
+      "loss": 0.1456,
+      "step": 7139
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4733982490136884e-06,
+      "loss": 0.1561,
+      "step": 7140
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4708739719164352e-06,
+      "loss": 0.1255,
+      "step": 7141
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.468350419781855e-06,
+      "loss": 0.1724,
+      "step": 7142
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4658275928901474e-06,
+      "loss": 0.1877,
+      "step": 7143
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.463305491521437e-06,
+      "loss": 0.2102,
+      "step": 7144
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4607841159557653e-06,
+      "loss": 0.1514,
+      "step": 7145
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4582634664730974e-06,
+      "loss": 0.1729,
+      "step": 7146
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.455743543353307e-06,
+      "loss": 0.1417,
+      "step": 7147
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.453224346876203e-06,
+      "loss": 0.1451,
+      "step": 7148
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.450705877321495e-06,
+      "loss": 0.1536,
+      "step": 7149
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.448188134968827e-06,
+      "loss": 0.1506,
+      "step": 7150
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.445671120097758e-06,
+      "loss": 0.137,
+      "step": 7151
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.443154832987765e-06,
+      "loss": 0.1425,
+      "step": 7152
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4406392739182402e-06,
+      "loss": 0.1715,
+      "step": 7153
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4381244431685066e-06,
+      "loss": 0.1603,
+      "step": 7154
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4356103410177897e-06,
+      "loss": 0.1448,
+      "step": 7155
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4330969677452496e-06,
+      "loss": 0.178,
+      "step": 7156
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4305843236299564e-06,
+      "loss": 0.1567,
+      "step": 7157
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4280724089509064e-06,
+      "loss": 0.1569,
+      "step": 7158
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4255612239870038e-06,
+      "loss": 0.1378,
+      "step": 7159
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4230507690170854e-06,
+      "loss": 0.1378,
+      "step": 7160
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4205410443198938e-06,
+      "loss": 0.1748,
+      "step": 7161
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.418032050174098e-06,
+      "loss": 0.165,
+      "step": 7162
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.415523786858291e-06,
+      "loss": 0.1763,
+      "step": 7163
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4130162546509695e-06,
+      "loss": 0.1755,
+      "step": 7164
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4105094538305638e-06,
+      "loss": 0.1638,
+      "step": 7165
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4080033846754124e-06,
+      "loss": 0.1719,
+      "step": 7166
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.405498047463779e-06,
+      "loss": 0.1354,
+      "step": 7167
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4029934424738455e-06,
+      "loss": 0.1943,
+      "step": 7168
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4004895699837128e-06,
+      "loss": 0.1522,
+      "step": 7169
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3979864302713928e-06,
+      "loss": 0.1875,
+      "step": 7170
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.395484023614829e-06,
+      "loss": 0.1873,
+      "step": 7171
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3929823502918703e-06,
+      "loss": 0.1512,
+      "step": 7172
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.390481410580293e-06,
+      "loss": 0.1381,
+      "step": 7173
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.38798120475779e-06,
+      "loss": 0.1853,
+      "step": 7174
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3854817331019753e-06,
+      "loss": 0.1484,
+      "step": 7175
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3829829958903704e-06,
+      "loss": 0.1437,
+      "step": 7176
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.380484993400431e-06,
+      "loss": 0.1765,
+      "step": 7177
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3779877259095172e-06,
+      "loss": 0.1677,
+      "step": 7178
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3754911936949154e-06,
+      "loss": 0.1897,
+      "step": 7179
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3729953970338282e-06,
+      "loss": 0.1407,
+      "step": 7180
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3705003362033816e-06,
+      "loss": 0.1334,
+      "step": 7181
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.368006011480608e-06,
+      "loss": 0.1356,
+      "step": 7182
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.365512423142472e-06,
+      "loss": 0.1641,
+      "step": 7183
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3630195714658374e-06,
+      "loss": 0.1603,
+      "step": 7184
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.360527456727516e-06,
+      "loss": 0.1492,
+      "step": 7185
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3580360792042065e-06,
+      "loss": 0.1616,
+      "step": 7186
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3555454391725475e-06,
+      "loss": 0.1582,
+      "step": 7187
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3530555369090813e-06,
+      "loss": 0.167,
+      "step": 7188
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.350566372690278e-06,
+      "loss": 0.16,
+      "step": 7189
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3480779467925217e-06,
+      "loss": 0.1731,
+      "step": 7190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3455902594921186e-06,
+      "loss": 0.1553,
+      "step": 7191
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3431033110652834e-06,
+      "loss": 0.1458,
+      "step": 7192
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3406171017881607e-06,
+      "loss": 0.1837,
+      "step": 7193
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.338131631936802e-06,
+      "loss": 0.1682,
+      "step": 7194
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.335646901787184e-06,
+      "loss": 0.1592,
+      "step": 7195
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3331629116151987e-06,
+      "loss": 0.1833,
+      "step": 7196
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.330679661696661e-06,
+      "loss": 0.1899,
+      "step": 7197
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.328197152307292e-06,
+      "loss": 0.1497,
+      "step": 7198
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.325715383722744e-06,
+      "loss": 0.1554,
+      "step": 7199
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3232343562185742e-06,
+      "loss": 0.187,
+      "step": 7200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3207540700702665e-06,
+      "loss": 0.179,
+      "step": 7201
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3182745255532212e-06,
+      "loss": 0.1821,
+      "step": 7202
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.315795722942756e-06,
+      "loss": 0.1508,
+      "step": 7203
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3133176625141016e-06,
+      "loss": 0.1444,
+      "step": 7204
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.310840344542414e-06,
+      "loss": 0.149,
+      "step": 7205
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.308363769302758e-06,
+      "loss": 0.1512,
+      "step": 7206
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3058879370701226e-06,
+      "loss": 0.1616,
+      "step": 7207
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3034128481194126e-06,
+      "loss": 0.1677,
+      "step": 7208
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.300938502725454e-06,
+      "loss": 0.1777,
+      "step": 7209
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2984649011629775e-06,
+      "loss": 0.1917,
+      "step": 7210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2959920437066486e-06,
+      "loss": 0.139,
+      "step": 7211
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2935199306310327e-06,
+      "loss": 0.1755,
+      "step": 7212
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.291048562210628e-06,
+      "loss": 0.1437,
+      "step": 7213
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2885779387198403e-06,
+      "loss": 0.1736,
+      "step": 7214
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2861080604329996e-06,
+      "loss": 0.1908,
+      "step": 7215
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2836389276243442e-06,
+      "loss": 0.1685,
+      "step": 7216
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2811705405680392e-06,
+      "loss": 0.142,
+      "step": 7217
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2787028995381574e-06,
+      "loss": 0.1771,
+      "step": 7218
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.276236004808696e-06,
+      "loss": 0.2029,
+      "step": 7219
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.273769856653568e-06,
+      "loss": 0.1808,
+      "step": 7220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2713044553466055e-06,
+      "loss": 0.1562,
+      "step": 7221
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2688398011615485e-06,
+      "loss": 0.1843,
+      "step": 7222
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2663758943720658e-06,
+      "loss": 0.1698,
+      "step": 7223
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.263912735251733e-06,
+      "loss": 0.1584,
+      "step": 7224
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.261450324074048e-06,
+      "loss": 0.1554,
+      "step": 7225
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.258988661112428e-06,
+      "loss": 0.1804,
+      "step": 7226
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.256527746640206e-06,
+      "loss": 0.1416,
+      "step": 7227
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2540675809306233e-06,
+      "loss": 0.1456,
+      "step": 7228
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2516081642568508e-06,
+      "loss": 0.1626,
+      "step": 7229
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.249149496891966e-06,
+      "loss": 0.1763,
+      "step": 7230
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2466915791089693e-06,
+      "loss": 0.1875,
+      "step": 7231
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.244234411180779e-06,
+      "loss": 0.1787,
+      "step": 7232
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2417779933802205e-06,
+      "loss": 0.1814,
+      "step": 7233
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.239322325980051e-06,
+      "loss": 0.1631,
+      "step": 7234
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.236867409252924e-06,
+      "loss": 0.1653,
+      "step": 7235
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.234413243471436e-06,
+      "loss": 0.1929,
+      "step": 7236
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2319598289080758e-06,
+      "loss": 0.1743,
+      "step": 7237
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.229507165835264e-06,
+      "loss": 0.1655,
+      "step": 7238
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2270552545253265e-06,
+      "loss": 0.157,
+      "step": 7239
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.22460409525052e-06,
+      "loss": 0.1641,
+      "step": 7240
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2221536882829975e-06,
+      "loss": 0.1658,
+      "step": 7241
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.219704033894854e-06,
+      "loss": 0.1946,
+      "step": 7242
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.217255132358078e-06,
+      "loss": 0.1696,
+      "step": 7243
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2148069839445893e-06,
+      "loss": 0.1592,
+      "step": 7244
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2123595889262126e-06,
+      "loss": 0.1296,
+      "step": 7245
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.209912947574699e-06,
+      "loss": 0.1766,
+      "step": 7246
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.207467060161711e-06,
+      "loss": 0.1559,
+      "step": 7247
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2050219269588302e-06,
+      "loss": 0.146,
+      "step": 7248
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2025775482375478e-06,
+      "loss": 0.1516,
+      "step": 7249
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.200133924269281e-06,
+      "loss": 0.1836,
+      "step": 7250
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1976910553253514e-06,
+      "loss": 0.1948,
+      "step": 7251
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1952489416770083e-06,
+      "loss": 0.1628,
+      "step": 7252
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.192807583595411e-06,
+      "loss": 0.1577,
+      "step": 7253
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1903669813516393e-06,
+      "loss": 0.1816,
+      "step": 7254
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1879271352166785e-06,
+      "loss": 0.1711,
+      "step": 7255
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1854880454614466e-06,
+      "loss": 0.1588,
+      "step": 7256
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1830497123567594e-06,
+      "loss": 0.1604,
+      "step": 7257
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1806121361733623e-06,
+      "loss": 0.1503,
+      "step": 7258
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.178175317181912e-06,
+      "loss": 0.1594,
+      "step": 7259
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1757392556529833e-06,
+      "loss": 0.1481,
+      "step": 7260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1733039518570574e-06,
+      "loss": 0.1422,
+      "step": 7261
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1708694060645483e-06,
+      "loss": 0.1376,
+      "step": 7262
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1684356185457677e-06,
+      "loss": 0.1672,
+      "step": 7263
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1660025895709547e-06,
+      "loss": 0.1678,
+      "step": 7264
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1635703194102618e-06,
+      "loss": 0.1807,
+      "step": 7265
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.161138808333759e-06,
+      "loss": 0.1489,
+      "step": 7266
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1587080566114247e-06,
+      "loss": 0.1885,
+      "step": 7267
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.156278064513163e-06,
+      "loss": 0.1479,
+      "step": 7268
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.153848832308781e-06,
+      "loss": 0.1917,
+      "step": 7269
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.151420360268015e-06,
+      "loss": 0.1863,
+      "step": 7270
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1489926486605094e-06,
+      "loss": 0.1777,
+      "step": 7271
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.146565697755828e-06,
+      "loss": 0.1775,
+      "step": 7272
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1441395078234418e-06,
+      "loss": 0.1361,
+      "step": 7273
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1417140791327507e-06,
+      "loss": 0.1632,
+      "step": 7274
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1392894119530548e-06,
+      "loss": 0.1422,
+      "step": 7275
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1368655065535814e-06,
+      "loss": 0.1569,
+      "step": 7276
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1344423632034692e-06,
+      "loss": 0.1697,
+      "step": 7277
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1320199821717746e-06,
+      "loss": 0.1707,
+      "step": 7278
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1295983637274617e-06,
+      "loss": 0.1626,
+      "step": 7279
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.127177508139423e-06,
+      "loss": 0.1543,
+      "step": 7280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.124757415676446e-06,
+      "loss": 0.1362,
+      "step": 7281
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1223380866072617e-06,
+      "loss": 0.177,
+      "step": 7282
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1199195212004884e-06,
+      "loss": 0.1746,
+      "step": 7283
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1175017197246814e-06,
+      "loss": 0.1707,
+      "step": 7284
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1150846824482928e-06,
+      "loss": 0.1709,
+      "step": 7285
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1126684096397076e-06,
+      "loss": 0.1781,
+      "step": 7286
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1102529015672044e-06,
+      "loss": 0.2178,
+      "step": 7287
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1078381584990047e-06,
+      "loss": 0.1577,
+      "step": 7288
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1054241807032195e-06,
+      "loss": 0.1995,
+      "step": 7289
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1030109684478917e-06,
+      "loss": 0.197,
+      "step": 7290
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1005985220009673e-06,
+      "loss": 0.1602,
+      "step": 7291
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.098186841630314e-06,
+      "loss": 0.1978,
+      "step": 7292
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0957759276037148e-06,
+      "loss": 0.1735,
+      "step": 7293
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0933657801888683e-06,
+      "loss": 0.1763,
+      "step": 7294
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0909563996533797e-06,
+      "loss": 0.1567,
+      "step": 7295
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0885477862647805e-06,
+      "loss": 0.1921,
+      "step": 7296
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.086139940290507e-06,
+      "loss": 0.1809,
+      "step": 7297
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.083732861997917e-06,
+      "loss": 0.1736,
+      "step": 7298
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0813265516542825e-06,
+      "loss": 0.1638,
+      "step": 7299
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.078921009526785e-06,
+      "loss": 0.1433,
+      "step": 7300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0765162358825285e-06,
+      "loss": 0.1655,
+      "step": 7301
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0741122309885218e-06,
+      "loss": 0.1631,
+      "step": 7302
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0717089951116985e-06,
+      "loss": 0.1307,
+      "step": 7303
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0693065285188995e-06,
+      "loss": 0.1431,
+      "step": 7304
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.066904831476889e-06,
+      "loss": 0.1809,
+      "step": 7305
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0645039042523318e-06,
+      "loss": 0.1188,
+      "step": 7306
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.062103747111823e-06,
+      "loss": 0.1305,
+      "step": 7307
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0597043603218567e-06,
+      "loss": 0.1304,
+      "step": 7308
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.057305744148854e-06,
+      "loss": 0.1765,
+      "step": 7309
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054907898859144e-06,
+      "loss": 0.1439,
+      "step": 7310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0525108247189763e-06,
+      "loss": 0.1731,
+      "step": 7311
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.050114521994503e-06,
+      "loss": 0.1621,
+      "step": 7312
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0477189909518047e-06,
+      "loss": 0.1443,
+      "step": 7313
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0453242318568643e-06,
+      "loss": 0.1244,
+      "step": 7314
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0429302449755873e-06,
+      "loss": 0.137,
+      "step": 7315
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.040537030573788e-06,
+      "loss": 0.1758,
+      "step": 7316
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0381445889172047e-06,
+      "loss": 0.1093,
+      "step": 7317
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.035752920271472e-06,
+      "loss": 0.1802,
+      "step": 7318
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.033362024902159e-06,
+      "loss": 0.1663,
+      "step": 7319
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.030971903074731e-06,
+      "loss": 0.1733,
+      "step": 7320
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.028582555054579e-06,
+      "loss": 0.2158,
+      "step": 7321
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.026193981107004e-06,
+      "loss": 0.1692,
+      "step": 7322
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0238061814972274e-06,
+      "loss": 0.1621,
+      "step": 7323
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0214191564903694e-06,
+      "loss": 0.1653,
+      "step": 7324
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0190329063514835e-06,
+      "loss": 0.1552,
+      "step": 7325
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0166474313455186e-06,
+      "loss": 0.1877,
+      "step": 7326
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0142627317373495e-06,
+      "loss": 0.171,
+      "step": 7327
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0118788077917626e-06,
+      "loss": 0.1492,
+      "step": 7328
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0094956597734603e-06,
+      "loss": 0.1398,
+      "step": 7329
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0071132879470497e-06,
+      "loss": 0.1656,
+      "step": 7330
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.004731692577064e-06,
+      "loss": 0.1383,
+      "step": 7331
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0023508739279384e-06,
+      "loss": 0.1638,
+      "step": 7332
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9999708322640296e-06,
+      "loss": 0.2012,
+      "step": 7333
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.997591567849607e-06,
+      "loss": 0.1567,
+      "step": 7334
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9952130809488545e-06,
+      "loss": 0.1748,
+      "step": 7335
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.992835371825862e-06,
+      "loss": 0.1395,
+      "step": 7336
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9904584407446457e-06,
+      "loss": 0.1589,
+      "step": 7337
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9880822879691184e-06,
+      "loss": 0.1536,
+      "step": 7338
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.985706913763131e-06,
+      "loss": 0.1504,
+      "step": 7339
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9833323183904216e-06,
+      "loss": 0.1442,
+      "step": 7340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9809585021146615e-06,
+      "loss": 0.1833,
+      "step": 7341
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.978585465199422e-06,
+      "loss": 0.1633,
+      "step": 7342
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9762132079081952e-06,
+      "loss": 0.1821,
+      "step": 7343
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9738417305043867e-06,
+      "loss": 0.2046,
+      "step": 7344
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9714710332513165e-06,
+      "loss": 0.1335,
+      "step": 7345
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9691011164122085e-06,
+      "loss": 0.129,
+      "step": 7346
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.966731980250214e-06,
+      "loss": 0.157,
+      "step": 7347
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9643636250283837e-06,
+      "loss": 0.1746,
+      "step": 7348
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9619960510096925e-06,
+      "loss": 0.1301,
+      "step": 7349
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9596292584570218e-06,
+      "loss": 0.1833,
+      "step": 7350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.957263247633175e-06,
+      "loss": 0.1421,
+      "step": 7351
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9548980188008545e-06,
+      "loss": 0.1423,
+      "step": 7352
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9525335722226912e-06,
+      "loss": 0.1279,
+      "step": 7353
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9501699081612143e-06,
+      "loss": 0.1588,
+      "step": 7354
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9478070268788774e-06,
+      "loss": 0.1938,
+      "step": 7355
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.945444928638044e-06,
+      "loss": 0.1652,
+      "step": 7356
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9430836137009934e-06,
+      "loss": 0.1202,
+      "step": 7357
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.940723082329906e-06,
+      "loss": 0.1873,
+      "step": 7358
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9383633347868935e-06,
+      "loss": 0.1821,
+      "step": 7359
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9360043713339616e-06,
+      "loss": 0.1449,
+      "step": 7360
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9336461922330438e-06,
+      "loss": 0.152,
+      "step": 7361
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.931288797745979e-06,
+      "loss": 0.1431,
+      "step": 7362
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.1528,
+      "step": 7363
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.926576363660346e-06,
+      "loss": 0.1271,
+      "step": 7364
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.924221324585017e-06,
+      "loss": 0.1353,
+      "step": 7365
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.921867071170034e-06,
+      "loss": 0.1514,
+      "step": 7366
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9195136036768e-06,
+      "loss": 0.14,
+      "step": 7367
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9171609223666396e-06,
+      "loss": 0.1626,
+      "step": 7368
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9148090275007733e-06,
+      "loss": 0.1562,
+      "step": 7369
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9124579193403522e-06,
+      "loss": 0.1724,
+      "step": 7370
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9101075981464267e-06,
+      "loss": 0.2061,
+      "step": 7371
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.907758064179965e-06,
+      "loss": 0.141,
+      "step": 7372
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9054093177018515e-06,
+      "loss": 0.153,
+      "step": 7373
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9030613589728805e-06,
+      "loss": 0.1389,
+      "step": 7374
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9007141882537535e-06,
+      "loss": 0.1466,
+      "step": 7375
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8983678058050934e-06,
+      "loss": 0.1885,
+      "step": 7376
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.896022211887427e-06,
+      "loss": 0.1558,
+      "step": 7377
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8936774067611996e-06,
+      "loss": 0.1895,
+      "step": 7378
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.891333390686767e-06,
+      "loss": 0.1394,
+      "step": 7379
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.888990163924401e-06,
+      "loss": 0.2009,
+      "step": 7380
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8866477267342752e-06,
+      "loss": 0.1667,
+      "step": 7381
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.88430607937649e-06,
+      "loss": 0.187,
+      "step": 7382
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.881965222111043e-06,
+      "loss": 0.1653,
+      "step": 7383
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.879625155197856e-06,
+      "loss": 0.158,
+      "step": 7384
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.877285878896758e-06,
+      "loss": 0.2122,
+      "step": 7385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8749473934674953e-06,
+      "loss": 0.1453,
+      "step": 7386
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.872609699169714e-06,
+      "loss": 0.1431,
+      "step": 7387
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8702727962629883e-06,
+      "loss": 0.1472,
+      "step": 7388
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8679366850067857e-06,
+      "loss": 0.2002,
+      "step": 7389
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8656013656605107e-06,
+      "loss": 0.1583,
+      "step": 7390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8632668384834563e-06,
+      "loss": 0.1324,
+      "step": 7391
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.860933103734842e-06,
+      "loss": 0.1454,
+      "step": 7392
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.858600161673789e-06,
+      "loss": 0.1472,
+      "step": 7393
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.85626801255934e-06,
+      "loss": 0.1548,
+      "step": 7394
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8539366566504445e-06,
+      "loss": 0.1382,
+      "step": 7395
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8516060942059677e-06,
+      "loss": 0.1882,
+      "step": 7396
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.849276325484679e-06,
+      "loss": 0.158,
+      "step": 7397
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8469473507452707e-06,
+      "loss": 0.1831,
+      "step": 7398
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8446191702463343e-06,
+      "loss": 0.1437,
+      "step": 7399
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.842291784246384e-06,
+      "loss": 0.1404,
+      "step": 7400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8399651930038385e-06,
+      "loss": 0.1578,
+      "step": 7401
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.837639396777038e-06,
+      "loss": 0.1572,
+      "step": 7402
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.835314395824219e-06,
+      "loss": 0.1509,
+      "step": 7403
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.832990190403546e-06,
+      "loss": 0.1399,
+      "step": 7404
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8306667807730802e-06,
+      "loss": 0.1482,
+      "step": 7405
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8283441671908064e-06,
+      "loss": 0.1682,
+      "step": 7406
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8260223499146154e-06,
+      "loss": 0.1176,
+      "step": 7407
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.823701329202313e-06,
+      "loss": 0.1462,
+      "step": 7408
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.821381105311609e-06,
+      "loss": 0.1543,
+      "step": 7409
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.819061678500137e-06,
+      "loss": 0.1891,
+      "step": 7410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8167430490254265e-06,
+      "loss": 0.1753,
+      "step": 7411
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8144252171449316e-06,
+      "loss": 0.2209,
+      "step": 7412
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8121081831160125e-06,
+      "loss": 0.1464,
+      "step": 7413
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8097919471959457e-06,
+      "loss": 0.1709,
+      "step": 7414
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8074765096419066e-06,
+      "loss": 0.1577,
+      "step": 7415
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8051618707109984e-06,
+      "loss": 0.1448,
+      "step": 7416
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8028480306602203e-06,
+      "loss": 0.1606,
+      "step": 7417
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.800534989746493e-06,
+      "loss": 0.1729,
+      "step": 7418
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7982227482266454e-06,
+      "loss": 0.1608,
+      "step": 7419
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7959113063574216e-06,
+      "loss": 0.1777,
+      "step": 7420
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7936006643954662e-06,
+      "loss": 0.1804,
+      "step": 7421
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.791290822597347e-06,
+      "loss": 0.1846,
+      "step": 7422
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.788981781219533e-06,
+      "loss": 0.1729,
+      "step": 7423
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7866735405184118e-06,
+      "loss": 0.1548,
+      "step": 7424
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7843661007502787e-06,
+      "loss": 0.1588,
+      "step": 7425
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.782059462171344e-06,
+      "loss": 0.1788,
+      "step": 7426
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7797536250377187e-06,
+      "loss": 0.1581,
+      "step": 7427
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7774485896054404e-06,
+      "loss": 0.1494,
+      "step": 7428
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7751443561304413e-06,
+      "loss": 0.1456,
+      "step": 7429
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.772840924868575e-06,
+      "loss": 0.1797,
+      "step": 7430
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770538296075608e-06,
+      "loss": 0.1744,
+      "step": 7431
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7682364700072053e-06,
+      "loss": 0.144,
+      "step": 7432
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.765935446918957e-06,
+      "loss": 0.1667,
+      "step": 7433
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7636352270663524e-06,
+      "loss": 0.166,
+      "step": 7434
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7613358107048007e-06,
+      "loss": 0.1549,
+      "step": 7435
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.759037198089616e-06,
+      "loss": 0.2126,
+      "step": 7436
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7567393894760273e-06,
+      "loss": 0.199,
+      "step": 7437
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7544423851191705e-06,
+      "loss": 0.145,
+      "step": 7438
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.752146185274095e-06,
+      "loss": 0.2043,
+      "step": 7439
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7498507901957537e-06,
+      "loss": 0.1743,
+      "step": 7440
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.747556200139028e-06,
+      "loss": 0.1729,
+      "step": 7441
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7452624153586883e-06,
+      "loss": 0.1522,
+      "step": 7442
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7429694361094315e-06,
+      "loss": 0.1312,
+      "step": 7443
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7406772626458535e-06,
+      "loss": 0.1748,
+      "step": 7444
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7383858952224708e-06,
+      "loss": 0.1592,
+      "step": 7445
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.736095334093698e-06,
+      "loss": 0.1643,
+      "step": 7446
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.733805579513881e-06,
+      "loss": 0.1914,
+      "step": 7447
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7315166317372523e-06,
+      "loss": 0.1641,
+      "step": 7448
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7292284910179713e-06,
+      "loss": 0.1649,
+      "step": 7449
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7269411576100977e-06,
+      "loss": 0.1884,
+      "step": 7450
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.724654631767608e-06,
+      "loss": 0.1318,
+      "step": 7451
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7223689137443876e-06,
+      "loss": 0.1726,
+      "step": 7452
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7200840037942345e-06,
+      "loss": 0.1721,
+      "step": 7453
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7177999021708477e-06,
+      "loss": 0.1399,
+      "step": 7454
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7155166091278496e-06,
+      "loss": 0.1899,
+      "step": 7455
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7132341249187587e-06,
+      "loss": 0.1571,
+      "step": 7456
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.710952449797014e-06,
+      "loss": 0.14,
+      "step": 7457
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.708671584015964e-06,
+      "loss": 0.1504,
+      "step": 7458
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7063915278288657e-06,
+      "loss": 0.1599,
+      "step": 7459
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.704112281488881e-06,
+      "loss": 0.1705,
+      "step": 7460
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.701833845249091e-06,
+      "loss": 0.1818,
+      "step": 7461
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.699556219362478e-06,
+      "loss": 0.1621,
+      "step": 7462
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6972794040819405e-06,
+      "loss": 0.151,
+      "step": 7463
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6950033996602844e-06,
+      "loss": 0.1763,
+      "step": 7464
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6927282063502313e-06,
+      "loss": 0.1302,
+      "step": 7465
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6904538244044e-06,
+      "loss": 0.1821,
+      "step": 7466
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.688180254075333e-06,
+      "loss": 0.1777,
+      "step": 7467
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6859074956154717e-06,
+      "loss": 0.1672,
+      "step": 7468
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.683635549277174e-06,
+      "loss": 0.1418,
+      "step": 7469
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6813644153127073e-06,
+      "loss": 0.189,
+      "step": 7470
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6790940939742484e-06,
+      "loss": 0.166,
+      "step": 7471
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6768245855138784e-06,
+      "loss": 0.1776,
+      "step": 7472
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.674555890183598e-06,
+      "loss": 0.1619,
+      "step": 7473
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6722880082353065e-06,
+      "loss": 0.1768,
+      "step": 7474
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.670020939920821e-06,
+      "loss": 0.1548,
+      "step": 7475
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.667754685491867e-06,
+      "loss": 0.1841,
+      "step": 7476
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.665489245200079e-06,
+      "loss": 0.1526,
+      "step": 7477
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.663224619296998e-06,
+      "loss": 0.1755,
+      "step": 7478
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6609608080340797e-06,
+      "loss": 0.2253,
+      "step": 7479
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6586978116626837e-06,
+      "loss": 0.1604,
+      "step": 7480
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6564356304340844e-06,
+      "loss": 0.1462,
+      "step": 7481
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.654174264599462e-06,
+      "loss": 0.1697,
+      "step": 7482
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.651913714409912e-06,
+      "loss": 0.1743,
+      "step": 7483
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.649653980116429e-06,
+      "loss": 0.1459,
+      "step": 7484
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6473950619699286e-06,
+      "loss": 0.1304,
+      "step": 7485
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6451369602212206e-06,
+      "loss": 0.1545,
+      "step": 7486
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.642879675121047e-06,
+      "loss": 0.1372,
+      "step": 7487
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6406232069200365e-06,
+      "loss": 0.156,
+      "step": 7488
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6383675558687417e-06,
+      "loss": 0.1951,
+      "step": 7489
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.636112722217614e-06,
+      "loss": 0.1405,
+      "step": 7490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6338587062170253e-06,
+      "loss": 0.2019,
+      "step": 7491
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.631605508117241e-06,
+      "loss": 0.1631,
+      "step": 7492
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.629353128168457e-06,
+      "loss": 0.1442,
+      "step": 7493
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.62710156662076e-06,
+      "loss": 0.166,
+      "step": 7494
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6248508237241556e-06,
+      "loss": 0.1638,
+      "step": 7495
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6226008997285513e-06,
+      "loss": 0.1957,
+      "step": 7496
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.620351794883771e-06,
+      "loss": 0.1589,
+      "step": 7497
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6181035094395456e-06,
+      "loss": 0.1532,
+      "step": 7498
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6158560436455095e-06,
+      "loss": 0.1733,
+      "step": 7499
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6136093977512157e-06,
+      "loss": 0.1763,
+      "step": 7500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.611363572006116e-06,
+      "loss": 0.1919,
+      "step": 7501
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6091185666595777e-06,
+      "loss": 0.1865,
+      "step": 7502
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6068743819608756e-06,
+      "loss": 0.2209,
+      "step": 7503
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.604631018159197e-06,
+      "loss": 0.1587,
+      "step": 7504
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6023884755036277e-06,
+      "loss": 0.166,
+      "step": 7505
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6001467542431746e-06,
+      "loss": 0.1595,
+      "step": 7506
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.597905854626742e-06,
+      "loss": 0.1711,
+      "step": 7507
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.595665776903152e-06,
+      "loss": 0.1649,
+      "step": 7508
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5934265213211305e-06,
+      "loss": 0.131,
+      "step": 7509
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5911880881293185e-06,
+      "loss": 0.1804,
+      "step": 7510
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.588950477576253e-06,
+      "loss": 0.1627,
+      "step": 7511
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.586713689910395e-06,
+      "loss": 0.2024,
+      "step": 7512
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5844777253801e-06,
+      "loss": 0.1436,
+      "step": 7513
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5822425842336417e-06,
+      "loss": 0.141,
+      "step": 7514
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5800082667192004e-06,
+      "loss": 0.1626,
+      "step": 7515
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5777747730848646e-06,
+      "loss": 0.1582,
+      "step": 7516
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5755421035786265e-06,
+      "loss": 0.1702,
+      "step": 7517
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.573310258448397e-06,
+      "loss": 0.1825,
+      "step": 7518
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5710792379419826e-06,
+      "loss": 0.1733,
+      "step": 7519
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5688490423071076e-06,
+      "loss": 0.1573,
+      "step": 7520
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5666196717914036e-06,
+      "loss": 0.1643,
+      "step": 7521
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.564391126642413e-06,
+      "loss": 0.198,
+      "step": 7522
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.562163407107573e-06,
+      "loss": 0.1331,
+      "step": 7523
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.559936513434248e-06,
+      "loss": 0.1721,
+      "step": 7524
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5577104458696942e-06,
+      "loss": 0.1882,
+      "step": 7525
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.555485204661087e-06,
+      "loss": 0.1572,
+      "step": 7526
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.553260790055506e-06,
+      "loss": 0.1538,
+      "step": 7527
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.551037202299944e-06,
+      "loss": 0.1398,
+      "step": 7528
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5488144416412885e-06,
+      "loss": 0.1672,
+      "step": 7529
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.546592508326353e-06,
+      "loss": 0.1555,
+      "step": 7530
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5443714026018427e-06,
+      "loss": 0.1904,
+      "step": 7531
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5421511247143826e-06,
+      "loss": 0.157,
+      "step": 7532
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5399316749105e-06,
+      "loss": 0.147,
+      "step": 7533
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.537713053436636e-06,
+      "loss": 0.1456,
+      "step": 7534
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.535495260539129e-06,
+      "loss": 0.1516,
+      "step": 7535
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.533278296464239e-06,
+      "loss": 0.1163,
+      "step": 7536
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.53106216145812e-06,
+      "loss": 0.1838,
+      "step": 7537
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5288468557668457e-06,
+      "loss": 0.1826,
+      "step": 7538
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.52663237963639e-06,
+      "loss": 0.1526,
+      "step": 7539
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5244187333126437e-06,
+      "loss": 0.1658,
+      "step": 7540
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5222059170413915e-06,
+      "loss": 0.1764,
+      "step": 7541
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5199939310683396e-06,
+      "loss": 0.1667,
+      "step": 7542
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5177827756390895e-06,
+      "loss": 0.1626,
+      "step": 7543
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5155724509991663e-06,
+      "loss": 0.1804,
+      "step": 7544
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5133629573939878e-06,
+      "loss": 0.1729,
+      "step": 7545
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5111542950688893e-06,
+      "loss": 0.153,
+      "step": 7546
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5089464642691043e-06,
+      "loss": 0.1665,
+      "step": 7547
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5067394652397836e-06,
+      "loss": 0.1356,
+      "step": 7548
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5045332982259805e-06,
+      "loss": 0.1809,
+      "step": 7549
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.502327963472662e-06,
+      "loss": 0.1589,
+      "step": 7550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5001234612246893e-06,
+      "loss": 0.1619,
+      "step": 7551
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4979197917268484e-06,
+      "loss": 0.1538,
+      "step": 7552
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.495716955223817e-06,
+      "loss": 0.1317,
+      "step": 7553
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.49351495196019e-06,
+      "loss": 0.167,
+      "step": 7554
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4913137821804667e-06,
+      "loss": 0.1676,
+      "step": 7555
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.489113446129059e-06,
+      "loss": 0.1685,
+      "step": 7556
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4869139440502744e-06,
+      "loss": 0.1699,
+      "step": 7557
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4847152761883408e-06,
+      "loss": 0.1357,
+      "step": 7558
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.482517442787383e-06,
+      "loss": 0.1707,
+      "step": 7559
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4803204440914396e-06,
+      "loss": 0.1323,
+      "step": 7560
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.478124280344456e-06,
+      "loss": 0.1667,
+      "step": 7561
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4759289517902864e-06,
+      "loss": 0.1709,
+      "step": 7562
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4737344586726818e-06,
+      "loss": 0.1912,
+      "step": 7563
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.471540801235316e-06,
+      "loss": 0.1576,
+      "step": 7564
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4693479797217545e-06,
+      "loss": 0.1458,
+      "step": 7565
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4671559943754818e-06,
+      "loss": 0.118,
+      "step": 7566
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4649648454398887e-06,
+      "loss": 0.1801,
+      "step": 7567
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4627745331582632e-06,
+      "loss": 0.2097,
+      "step": 7568
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4605850577738134e-06,
+      "loss": 0.1342,
+      "step": 7569
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4583964195296407e-06,
+      "loss": 0.1614,
+      "step": 7570
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.456208618668766e-06,
+      "loss": 0.1431,
+      "step": 7571
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.454021655434111e-06,
+      "loss": 0.1582,
+      "step": 7572
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4518355300685083e-06,
+      "loss": 0.1638,
+      "step": 7573
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4496502428146896e-06,
+      "loss": 0.1758,
+      "step": 7574
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4474657939153044e-06,
+      "loss": 0.1797,
+      "step": 7575
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.445282183612897e-06,
+      "loss": 0.2102,
+      "step": 7576
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4430994121499273e-06,
+      "loss": 0.1501,
+      "step": 7577
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.440917479768762e-06,
+      "loss": 0.1383,
+      "step": 7578
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.438736386711673e-06,
+      "loss": 0.2253,
+      "step": 7579
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.436556133220833e-06,
+      "loss": 0.1313,
+      "step": 7580
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4343767195383338e-06,
+      "loss": 0.1753,
+      "step": 7581
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4321981459061594e-06,
+      "loss": 0.1936,
+      "step": 7582
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.430020412566213e-06,
+      "loss": 0.1528,
+      "step": 7583
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.427843519760298e-06,
+      "loss": 0.1671,
+      "step": 7584
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4256674677301294e-06,
+      "loss": 0.1232,
+      "step": 7585
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.42349225671732e-06,
+      "loss": 0.1721,
+      "step": 7586
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.421317886963401e-06,
+      "loss": 0.1573,
+      "step": 7587
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4191443587097975e-06,
+      "loss": 0.1489,
+      "step": 7588
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.416971672197851e-06,
+      "loss": 0.1841,
+      "step": 7589
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4147998276688046e-06,
+      "loss": 0.1543,
+      "step": 7590
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.412628825363814e-06,
+      "loss": 0.1741,
+      "step": 7591
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4104586655239305e-06,
+      "loss": 0.1761,
+      "step": 7592
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4082893483901237e-06,
+      "loss": 0.165,
+      "step": 7593
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.406120874203256e-06,
+      "loss": 0.1697,
+      "step": 7594
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4039532432041156e-06,
+      "loss": 0.1763,
+      "step": 7595
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.401786455633377e-06,
+      "loss": 0.1512,
+      "step": 7596
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3996205117316364e-06,
+      "loss": 0.2173,
+      "step": 7597
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3974554117393823e-06,
+      "loss": 0.1456,
+      "step": 7598
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3952911558970214e-06,
+      "loss": 0.168,
+      "step": 7599
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.393127744444862e-06,
+      "loss": 0.178,
+      "step": 7600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.39096517762312e-06,
+      "loss": 0.1556,
+      "step": 7601
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3888034556719142e-06,
+      "loss": 0.1504,
+      "step": 7602
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.386642578831274e-06,
+      "loss": 0.1545,
+      "step": 7603
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3844825473411282e-06,
+      "loss": 0.1943,
+      "step": 7604
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.38232336144132e-06,
+      "loss": 0.1633,
+      "step": 7605
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3801650213715944e-06,
+      "loss": 0.1276,
+      "step": 7606
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3780075273716063e-06,
+      "loss": 0.1895,
+      "step": 7607
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3758508796809067e-06,
+      "loss": 0.1833,
+      "step": 7608
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3736950785389658e-06,
+      "loss": 0.1503,
+      "step": 7609
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.371540124185148e-06,
+      "loss": 0.158,
+      "step": 7610
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.369386016858731e-06,
+      "loss": 0.1523,
+      "step": 7611
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3672327567988963e-06,
+      "loss": 0.1322,
+      "step": 7612
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3650803442447356e-06,
+      "loss": 0.1797,
+      "step": 7613
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.362928779435236e-06,
+      "loss": 0.1875,
+      "step": 7614
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3607780626093036e-06,
+      "loss": 0.1702,
+      "step": 7615
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3586281940057367e-06,
+      "loss": 0.1396,
+      "step": 7616
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3564791738632485e-06,
+      "loss": 0.1237,
+      "step": 7617
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.354331002420458e-06,
+      "loss": 0.1749,
+      "step": 7618
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.352183679915888e-06,
+      "loss": 0.1595,
+      "step": 7619
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.350037206587964e-06,
+      "loss": 0.1801,
+      "step": 7620
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3478915826750236e-06,
+      "loss": 0.1536,
+      "step": 7621
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.345746808415302e-06,
+      "loss": 0.1653,
+      "step": 7622
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.343602884046947e-06,
+      "loss": 0.1582,
+      "step": 7623
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.34145980980801e-06,
+      "loss": 0.2029,
+      "step": 7624
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3393175859364503e-06,
+      "loss": 0.1489,
+      "step": 7625
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3371762126701237e-06,
+      "loss": 0.1387,
+      "step": 7626
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.335035690246804e-06,
+      "loss": 0.1575,
+      "step": 7627
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3328960189041593e-06,
+      "loss": 0.1556,
+      "step": 7628
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3307571988797705e-06,
+      "loss": 0.1255,
+      "step": 7629
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3286192304111232e-06,
+      "loss": 0.197,
+      "step": 7630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3264821137356086e-06,
+      "loss": 0.115,
+      "step": 7631
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.324345849090517e-06,
+      "loss": 0.1643,
+      "step": 7632
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3222104367130536e-06,
+      "loss": 0.187,
+      "step": 7633
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3200758768403196e-06,
+      "loss": 0.1912,
+      "step": 7634
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3179421697093285e-06,
+      "loss": 0.2124,
+      "step": 7635
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3158093155570003e-06,
+      "loss": 0.1471,
+      "step": 7636
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3136773146201506e-06,
+      "loss": 0.1683,
+      "step": 7637
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3115461671355122e-06,
+      "loss": 0.1577,
+      "step": 7638
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.309415873339712e-06,
+      "loss": 0.2007,
+      "step": 7639
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3072864334692903e-06,
+      "loss": 0.1711,
+      "step": 7640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3051578477606907e-06,
+      "loss": 0.1929,
+      "step": 7641
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.303030116450262e-06,
+      "loss": 0.1714,
+      "step": 7642
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3009032397742528e-06,
+      "loss": 0.2107,
+      "step": 7643
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2987772179688263e-06,
+      "loss": 0.1703,
+      "step": 7644
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2966520512700385e-06,
+      "loss": 0.1357,
+      "step": 7645
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.294527739913868e-06,
+      "loss": 0.165,
+      "step": 7646
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2924042841361793e-06,
+      "loss": 0.1558,
+      "step": 7647
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.290281684172757e-06,
+      "loss": 0.1885,
+      "step": 7648
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.288159940259278e-06,
+      "loss": 0.1173,
+      "step": 7649
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.286039052631337e-06,
+      "loss": 0.1875,
+      "step": 7650
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2839190215244177e-06,
+      "loss": 0.1461,
+      "step": 7651
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.281799847173931e-06,
+      "loss": 0.1565,
+      "step": 7652
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.27968152981517e-06,
+      "loss": 0.1321,
+      "step": 7653
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.277564069683349e-06,
+      "loss": 0.1997,
+      "step": 7654
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.275447467013574e-06,
+      "loss": 0.1669,
+      "step": 7655
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2733317220408647e-06,
+      "loss": 0.1473,
+      "step": 7656
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.271216835000145e-06,
+      "loss": 0.1997,
+      "step": 7657
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2691028061262433e-06,
+      "loss": 0.1479,
+      "step": 7658
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2669896356538856e-06,
+      "loss": 0.1677,
+      "step": 7659
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2648773238177147e-06,
+      "loss": 0.1873,
+      "step": 7660
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.262765870852265e-06,
+      "loss": 0.1646,
+      "step": 7661
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2606552769919855e-06,
+      "loss": 0.1711,
+      "step": 7662
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2585455424712255e-06,
+      "loss": 0.1782,
+      "step": 7663
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2564366675242433e-06,
+      "loss": 0.1497,
+      "step": 7664
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.254328652385193e-06,
+      "loss": 0.1785,
+      "step": 7665
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2522214972881417e-06,
+      "loss": 0.1472,
+      "step": 7666
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2501152024670545e-06,
+      "loss": 0.1453,
+      "step": 7667
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2480097681558077e-06,
+      "loss": 0.1436,
+      "step": 7668
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.245905194588176e-06,
+      "loss": 0.1709,
+      "step": 7669
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.243801481997845e-06,
+      "loss": 0.1511,
+      "step": 7670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2416986306183963e-06,
+      "loss": 0.1661,
+      "step": 7671
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2395966406833237e-06,
+      "loss": 0.1794,
+      "step": 7672
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.237495512426019e-06,
+      "loss": 0.1349,
+      "step": 7673
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.235395246079782e-06,
+      "loss": 0.1711,
+      "step": 7674
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2332958418778174e-06,
+      "loss": 0.145,
+      "step": 7675
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.231197300053235e-06,
+      "loss": 0.1105,
+      "step": 7676
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2290996208390423e-06,
+      "loss": 0.1902,
+      "step": 7677
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.227002804468159e-06,
+      "loss": 0.1542,
+      "step": 7678
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.224906851173403e-06,
+      "loss": 0.1499,
+      "step": 7679
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2228117611874987e-06,
+      "loss": 0.1814,
+      "step": 7680
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2207175347430754e-06,
+      "loss": 0.1648,
+      "step": 7681
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2186241720726698e-06,
+      "loss": 0.1763,
+      "step": 7682
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2165316734087128e-06,
+      "loss": 0.2051,
+      "step": 7683
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2144400389835506e-06,
+      "loss": 0.1633,
+      "step": 7684
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2123492690294237e-06,
+      "loss": 0.1738,
+      "step": 7685
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.210259363778482e-06,
+      "loss": 0.1456,
+      "step": 7686
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2081703234627793e-06,
+      "loss": 0.1562,
+      "step": 7687
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.206082148314276e-06,
+      "loss": 0.1399,
+      "step": 7688
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2039948385648267e-06,
+      "loss": 0.1727,
+      "step": 7689
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.2019083944462028e-06,
+      "loss": 0.1539,
+      "step": 7690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1998228161900635e-06,
+      "loss": 0.1656,
+      "step": 7691
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.197738104027992e-06,
+      "loss": 0.1627,
+      "step": 7692
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1956542581914585e-06,
+      "loss": 0.153,
+      "step": 7693
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.193571278911847e-06,
+      "loss": 0.1678,
+      "step": 7694
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1914891664204362e-06,
+      "loss": 0.1567,
+      "step": 7695
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.18940792094842e-06,
+      "loss": 0.1549,
+      "step": 7696
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1873275427268804e-06,
+      "loss": 0.1738,
+      "step": 7697
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1852480319868253e-06,
+      "loss": 0.1577,
+      "step": 7698
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.183169388959144e-06,
+      "loss": 0.1753,
+      "step": 7699
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.181091613874645e-06,
+      "loss": 0.1594,
+      "step": 7700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.179014706964031e-06,
+      "loss": 0.168,
+      "step": 7701
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1769386684579064e-06,
+      "loss": 0.1912,
+      "step": 7702
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1748634985867965e-06,
+      "loss": 0.1897,
+      "step": 7703
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1727891975811098e-06,
+      "loss": 0.1649,
+      "step": 7704
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.170715765671171e-06,
+      "loss": 0.1226,
+      "step": 7705
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1686432030871995e-06,
+      "loss": 0.1484,
+      "step": 7706
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1665715100593244e-06,
+      "loss": 0.1539,
+      "step": 7707
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1645006868175765e-06,
+      "loss": 0.1521,
+      "step": 7708
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.162430733591895e-06,
+      "loss": 0.1346,
+      "step": 7709
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1603616506121093e-06,
+      "loss": 0.2048,
+      "step": 7710
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1582934381079678e-06,
+      "loss": 0.175,
+      "step": 7711
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1562260963091086e-06,
+      "loss": 0.1731,
+      "step": 7712
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1541596254450815e-06,
+      "loss": 0.1353,
+      "step": 7713
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.152094025745338e-06,
+      "loss": 0.1927,
+      "step": 7714
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1500292974392357e-06,
+      "loss": 0.1704,
+      "step": 7715
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1479654407560256e-06,
+      "loss": 0.1931,
+      "step": 7716
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.145902455924874e-06,
+      "loss": 0.1604,
+      "step": 7717
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.14384034317484e-06,
+      "loss": 0.1766,
+      "step": 7718
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.141779102734893e-06,
+      "loss": 0.1755,
+      "step": 7719
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.139718734833903e-06,
+      "loss": 0.1638,
+      "step": 7720
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.137659239700647e-06,
+      "loss": 0.1582,
+      "step": 7721
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1356006175637944e-06,
+      "loss": 0.149,
+      "step": 7722
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1335428686519312e-06,
+      "loss": 0.1661,
+      "step": 7723
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.131485993193534e-06,
+      "loss": 0.1506,
+      "step": 7724
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1294299914169905e-06,
+      "loss": 0.1581,
+      "step": 7725
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.127374863550591e-06,
+      "loss": 0.1526,
+      "step": 7726
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1253206098225286e-06,
+      "loss": 0.1619,
+      "step": 7727
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1232672304608914e-06,
+      "loss": 0.1544,
+      "step": 7728
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1212147256936845e-06,
+      "loss": 0.1776,
+      "step": 7729
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.119163095748801e-06,
+      "loss": 0.1675,
+      "step": 7730
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1171123408540463e-06,
+      "loss": 0.115,
+      "step": 7731
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1150624612371273e-06,
+      "loss": 0.1462,
+      "step": 7732
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1130134571256556e-06,
+      "loss": 0.1523,
+      "step": 7733
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1109653287471375e-06,
+      "loss": 0.1488,
+      "step": 7734
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.108918076328992e-06,
+      "loss": 0.1672,
+      "step": 7735
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1068717000985294e-06,
+      "loss": 0.1589,
+      "step": 7736
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1048262002829754e-06,
+      "loss": 0.1427,
+      "step": 7737
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1027815771094494e-06,
+      "loss": 0.1689,
+      "step": 7738
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.100737830804982e-06,
+      "loss": 0.1671,
+      "step": 7739
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0986949615964926e-06,
+      "loss": 0.1931,
+      "step": 7740
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0966529697108196e-06,
+      "loss": 0.1976,
+      "step": 7741
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.094611855374685e-06,
+      "loss": 0.1482,
+      "step": 7742
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0925716188147384e-06,
+      "loss": 0.1733,
+      "step": 7743
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.090532260257507e-06,
+      "loss": 0.1466,
+      "step": 7744
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0884937799294382e-06,
+      "loss": 0.1069,
+      "step": 7745
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0864561780568693e-06,
+      "loss": 0.1814,
+      "step": 7746
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.08441945486605e-06,
+      "loss": 0.1772,
+      "step": 7747
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.082383610583122e-06,
+      "loss": 0.1985,
+      "step": 7748
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.080348645434146e-06,
+      "loss": 0.1567,
+      "step": 7749
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.078314559645066e-06,
+      "loss": 0.1824,
+      "step": 7750
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0762813534417424e-06,
+      "loss": 0.1777,
+      "step": 7751
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0742490270499284e-06,
+      "loss": 0.1628,
+      "step": 7752
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.072217580695285e-06,
+      "loss": 0.1693,
+      "step": 7753
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0701870146033744e-06,
+      "loss": 0.1812,
+      "step": 7754
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0681573289996646e-06,
+      "loss": 0.1475,
+      "step": 7755
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.066128524109515e-06,
+      "loss": 0.1521,
+      "step": 7756
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.064100600158202e-06,
+      "loss": 0.1414,
+      "step": 7757
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0620735573708893e-06,
+      "loss": 0.1533,
+      "step": 7758
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.060047395972653e-06,
+      "loss": 0.21,
+      "step": 7759
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0580221161884693e-06,
+      "loss": 0.1444,
+      "step": 7760
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.055997718243217e-06,
+      "loss": 0.1622,
+      "step": 7761
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0539742023616703e-06,
+      "loss": 0.1624,
+      "step": 7762
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0519515687685165e-06,
+      "loss": 0.1726,
+      "step": 7763
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.049929817688333e-06,
+      "loss": 0.1448,
+      "step": 7764
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.047908949345608e-06,
+      "loss": 0.1456,
+      "step": 7765
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0458889639647308e-06,
+      "loss": 0.1333,
+      "step": 7766
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0438698617699914e-06,
+      "loss": 0.1953,
+      "step": 7767
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0418516429855796e-06,
+      "loss": 0.1909,
+      "step": 7768
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0398343078355853e-06,
+      "loss": 0.1494,
+      "step": 7769
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0378178565440067e-06,
+      "loss": 0.1865,
+      "step": 7770
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0358022893347396e-06,
+      "loss": 0.1501,
+      "step": 7771
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0337876064315888e-06,
+      "loss": 0.1227,
+      "step": 7772
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0317738080582463e-06,
+      "loss": 0.1582,
+      "step": 7773
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0297608944383208e-06,
+      "loss": 0.167,
+      "step": 7774
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0277488657953125e-06,
+      "loss": 0.1514,
+      "step": 7775
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0257377223526285e-06,
+      "loss": 0.1459,
+      "step": 7776
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.023727464333577e-06,
+      "loss": 0.1824,
+      "step": 7777
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.02171809196137e-06,
+      "loss": 0.168,
+      "step": 7778
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.019709605459114e-06,
+      "loss": 0.1669,
+      "step": 7779
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.017702005049825e-06,
+      "loss": 0.1797,
+      "step": 7780
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0156952909564142e-06,
+      "loss": 0.1887,
+      "step": 7781
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.013689463401699e-06,
+      "loss": 0.1487,
+      "step": 7782
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0116845226083968e-06,
+      "loss": 0.1558,
+      "step": 7783
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.009680468799129e-06,
+      "loss": 0.2073,
+      "step": 7784
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.007677302196411e-06,
+      "loss": 0.1615,
+      "step": 7785
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0056750230226707e-06,
+      "loss": 0.1936,
+      "step": 7786
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0036736315002248e-06,
+      "loss": 0.1398,
+      "step": 7787
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0016731278513023e-06,
+      "loss": 0.1562,
+      "step": 7788
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.999673512298029e-06,
+      "loss": 0.1427,
+      "step": 7789
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9976747850624355e-06,
+      "loss": 0.1636,
+      "step": 7790
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9956769463664436e-06,
+      "loss": 0.1558,
+      "step": 7791
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.993679996431892e-06,
+      "loss": 0.1566,
+      "step": 7792
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.991683935480505e-06,
+      "loss": 0.1111,
+      "step": 7793
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.989688763733919e-06,
+      "loss": 0.1444,
+      "step": 7794
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.987694481413668e-06,
+      "loss": 0.1697,
+      "step": 7795
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.98570108874119e-06,
+      "loss": 0.2141,
+      "step": 7796
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.983708585937817e-06,
+      "loss": 0.1562,
+      "step": 7797
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.981716973224792e-06,
+      "loss": 0.1274,
+      "step": 7798
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9797262508232453e-06,
+      "loss": 0.1346,
+      "step": 7799
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9777364189542283e-06,
+      "loss": 0.1846,
+      "step": 7800
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9757474778386754e-06,
+      "loss": 0.1497,
+      "step": 7801
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9737594276974325e-06,
+      "loss": 0.1366,
+      "step": 7802
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9717722687512387e-06,
+      "loss": 0.1361,
+      "step": 7803
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9697860012207416e-06,
+      "loss": 0.1528,
+      "step": 7804
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.967800625326486e-06,
+      "loss": 0.1797,
+      "step": 7805
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.965816141288922e-06,
+      "loss": 0.1471,
+      "step": 7806
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9638325493283904e-06,
+      "loss": 0.163,
+      "step": 7807
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9618498496651473e-06,
+      "loss": 0.1508,
+      "step": 7808
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.959868042519334e-06,
+      "loss": 0.1315,
+      "step": 7809
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9578871281110056e-06,
+      "loss": 0.1565,
+      "step": 7810
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.955907106660112e-06,
+      "loss": 0.1559,
+      "step": 7811
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9539279783865093e-06,
+      "loss": 0.1973,
+      "step": 7812
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.951949743509943e-06,
+      "loss": 0.1956,
+      "step": 7813
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.949972402250074e-06,
+      "loss": 0.1716,
+      "step": 7814
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9479959548264515e-06,
+      "loss": 0.1887,
+      "step": 7815
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9460204014585314e-06,
+      "loss": 0.1544,
+      "step": 7816
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.944045742365671e-06,
+      "loss": 0.1726,
+      "step": 7817
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9420719777671293e-06,
+      "loss": 0.1958,
+      "step": 7818
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9400991078820587e-06,
+      "loss": 0.1492,
+      "step": 7819
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9381271329295215e-06,
+      "loss": 0.1467,
+      "step": 7820
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.936156053128472e-06,
+      "loss": 0.1833,
+      "step": 7821
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.934185868697771e-06,
+      "loss": 0.1675,
+      "step": 7822
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9322165798561796e-06,
+      "loss": 0.1315,
+      "step": 7823
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.93024818682236e-06,
+      "loss": 0.1779,
+      "step": 7824
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9282806898148677e-06,
+      "loss": 0.1586,
+      "step": 7825
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9263140890521703e-06,
+      "loss": 0.157,
+      "step": 7826
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9243483847526223e-06,
+      "loss": 0.1499,
+      "step": 7827
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.922383577134491e-06,
+      "loss": 0.1505,
+      "step": 7828
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.920419666415938e-06,
+      "loss": 0.1538,
+      "step": 7829
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9184566528150293e-06,
+      "loss": 0.1592,
+      "step": 7830
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.916494536549722e-06,
+      "loss": 0.1323,
+      "step": 7831
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9145333178378876e-06,
+      "loss": 0.1726,
+      "step": 7832
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.912572996897284e-06,
+      "loss": 0.1863,
+      "step": 7833
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.910613573945579e-06,
+      "loss": 0.1372,
+      "step": 7834
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9086550492003385e-06,
+      "loss": 0.142,
+      "step": 7835
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9066974228790235e-06,
+      "loss": 0.1812,
+      "step": 7836
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9047406951990043e-06,
+      "loss": 0.1548,
+      "step": 7837
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9027848663775416e-06,
+      "loss": 0.1237,
+      "step": 7838
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.900829936631804e-06,
+      "loss": 0.1655,
+      "step": 7839
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8988759061788565e-06,
+      "loss": 0.1516,
+      "step": 7840
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8969227752356688e-06,
+      "loss": 0.1453,
+      "step": 7841
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8949705440191024e-06,
+      "loss": 0.1427,
+      "step": 7842
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8930192127459268e-06,
+      "loss": 0.1338,
+      "step": 7843
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8910687816328055e-06,
+      "loss": 0.156,
+      "step": 7844
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8891192508963063e-06,
+      "loss": 0.1984,
+      "step": 7845
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.887170620752896e-06,
+      "loss": 0.1483,
+      "step": 7846
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.885222891418943e-06,
+      "loss": 0.1743,
+      "step": 7847
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8832760631107105e-06,
+      "loss": 0.1364,
+      "step": 7848
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8813301360443682e-06,
+      "loss": 0.1506,
+      "step": 7849
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.879385110435975e-06,
+      "loss": 0.1755,
+      "step": 7850
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8774409865015075e-06,
+      "loss": 0.1417,
+      "step": 7851
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.875497764456825e-06,
+      "loss": 0.167,
+      "step": 7852
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8735554445176984e-06,
+      "loss": 0.1312,
+      "step": 7853
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8716140268997873e-06,
+      "loss": 0.1655,
+      "step": 7854
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8696735118186637e-06,
+      "loss": 0.176,
+      "step": 7855
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8677338994897853e-06,
+      "loss": 0.1704,
+      "step": 7856
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8657951901285266e-06,
+      "loss": 0.1521,
+      "step": 7857
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.863857383950145e-06,
+      "loss": 0.1499,
+      "step": 7858
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.861920481169812e-06,
+      "loss": 0.1488,
+      "step": 7859
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8599844820025847e-06,
+      "loss": 0.1438,
+      "step": 7860
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8580493866634308e-06,
+      "loss": 0.1523,
+      "step": 7861
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.856115195367213e-06,
+      "loss": 0.1487,
+      "step": 7862
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8541819083286983e-06,
+      "loss": 0.2053,
+      "step": 7863
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.852249525762544e-06,
+      "loss": 0.1403,
+      "step": 7864
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8503180478833182e-06,
+      "loss": 0.2439,
+      "step": 7865
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8483874749054764e-06,
+      "loss": 0.1941,
+      "step": 7866
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8464578070433837e-06,
+      "loss": 0.1621,
+      "step": 7867
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8445290445113006e-06,
+      "loss": 0.1805,
+      "step": 7868
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8426011875233907e-06,
+      "loss": 0.1097,
+      "step": 7869
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.840674236293708e-06,
+      "loss": 0.1748,
+      "step": 7870
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8387481910362181e-06,
+      "loss": 0.1526,
+      "step": 7871
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8368230519647735e-06,
+      "loss": 0.189,
+      "step": 7872
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8348988192931361e-06,
+      "loss": 0.1697,
+      "step": 7873
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8329754932349619e-06,
+      "loss": 0.1643,
+      "step": 7874
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8310530740038125e-06,
+      "loss": 0.144,
+      "step": 7875
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8291315618131366e-06,
+      "loss": 0.1746,
+      "step": 7876
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8272109568762975e-06,
+      "loss": 0.1218,
+      "step": 7877
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.825291259406542e-06,
+      "loss": 0.1555,
+      "step": 7878
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8233724696170275e-06,
+      "loss": 0.1414,
+      "step": 7879
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8214545877208078e-06,
+      "loss": 0.1499,
+      "step": 7880
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8195376139308374e-06,
+      "loss": 0.1538,
+      "step": 7881
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8176215484599624e-06,
+      "loss": 0.1694,
+      "step": 7882
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8157063915209395e-06,
+      "loss": 0.1475,
+      "step": 7883
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8137921433264127e-06,
+      "loss": 0.1454,
+      "step": 7884
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8118788040889335e-06,
+      "loss": 0.1658,
+      "step": 7885
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8099663740209495e-06,
+      "loss": 0.1345,
+      "step": 7886
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8080548533348109e-06,
+      "loss": 0.2006,
+      "step": 7887
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.806144242242759e-06,
+      "loss": 0.1707,
+      "step": 7888
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8042345409569418e-06,
+      "loss": 0.1599,
+      "step": 7889
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.802325749689401e-06,
+      "loss": 0.1887,
+      "step": 7890
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8004178686520813e-06,
+      "loss": 0.1631,
+      "step": 7891
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7985108980568234e-06,
+      "loss": 0.2122,
+      "step": 7892
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7966048381153723e-06,
+      "loss": 0.1377,
+      "step": 7893
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.794699689039361e-06,
+      "loss": 0.1603,
+      "step": 7894
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7927954510403346e-06,
+      "loss": 0.174,
+      "step": 7895
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7908921243297216e-06,
+      "loss": 0.1599,
+      "step": 7896
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.788989709118869e-06,
+      "loss": 0.1504,
+      "step": 7897
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7870882056190042e-06,
+      "loss": 0.1907,
+      "step": 7898
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7851876140412673e-06,
+      "loss": 0.1456,
+      "step": 7899
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7832879345966824e-06,
+      "loss": 0.1479,
+      "step": 7900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7813891674961902e-06,
+      "loss": 0.2246,
+      "step": 7901
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7794913129506118e-06,
+      "loss": 0.1594,
+      "step": 7902
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.77759437117068e-06,
+      "loss": 0.1489,
+      "step": 7903
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7756983423670248e-06,
+      "loss": 0.1643,
+      "step": 7904
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.773803226750167e-06,
+      "loss": 0.1627,
+      "step": 7905
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.771909024530536e-06,
+      "loss": 0.1499,
+      "step": 7906
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7700157359184467e-06,
+      "loss": 0.1772,
+      "step": 7907
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.768123361124131e-06,
+      "loss": 0.1489,
+      "step": 7908
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7662319003577022e-06,
+      "loss": 0.1724,
+      "step": 7909
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7643413538291832e-06,
+      "loss": 0.1819,
+      "step": 7910
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7624517217484872e-06,
+      "loss": 0.1548,
+      "step": 7911
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7605630043254307e-06,
+      "loss": 0.1411,
+      "step": 7912
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7586752017697295e-06,
+      "loss": 0.1743,
+      "step": 7913
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.756788314290998e-06,
+      "loss": 0.1782,
+      "step": 7914
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7549023420987422e-06,
+      "loss": 0.1411,
+      "step": 7915
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7530172854023753e-06,
+      "loss": 0.1807,
+      "step": 7916
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7511331444112013e-06,
+      "loss": 0.1587,
+      "step": 7917
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7492499193344281e-06,
+      "loss": 0.1816,
+      "step": 7918
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.74736761038116e-06,
+      "loss": 0.1499,
+      "step": 7919
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7454862177604015e-06,
+      "loss": 0.1838,
+      "step": 7920
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7436057416810493e-06,
+      "loss": 0.1362,
+      "step": 7921
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.741726182351906e-06,
+      "loss": 0.1497,
+      "step": 7922
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.739847539981665e-06,
+      "loss": 0.1633,
+      "step": 7923
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7379698147789238e-06,
+      "loss": 0.1392,
+      "step": 7924
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7360930069521764e-06,
+      "loss": 0.1831,
+      "step": 7925
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7342171167098165e-06,
+      "loss": 0.1573,
+      "step": 7926
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7323421442601284e-06,
+      "loss": 0.1508,
+      "step": 7927
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.730468089811307e-06,
+      "loss": 0.168,
+      "step": 7928
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7285949535714307e-06,
+      "loss": 0.1699,
+      "step": 7929
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7267227357484873e-06,
+      "loss": 0.1697,
+      "step": 7930
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7248514365503588e-06,
+      "loss": 0.176,
+      "step": 7931
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.722981056184828e-06,
+      "loss": 0.1643,
+      "step": 7932
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.721111594859567e-06,
+      "loss": 0.144,
+      "step": 7933
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.719243052782158e-06,
+      "loss": 0.1447,
+      "step": 7934
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7173754301600688e-06,
+      "loss": 0.1425,
+      "step": 7935
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7155087272006743e-06,
+      "loss": 0.1787,
+      "step": 7936
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7136429441112446e-06,
+      "loss": 0.2019,
+      "step": 7937
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7117780810989492e-06,
+      "loss": 0.1343,
+      "step": 7938
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7099141383708473e-06,
+      "loss": 0.1438,
+      "step": 7939
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7080511161339097e-06,
+      "loss": 0.1664,
+      "step": 7940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.70618901459499e-06,
+      "loss": 0.1257,
+      "step": 7941
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7043278339608515e-06,
+      "loss": 0.1349,
+      "step": 7942
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.702467574438149e-06,
+      "loss": 0.1499,
+      "step": 7943
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7006082362334409e-06,
+      "loss": 0.1526,
+      "step": 7944
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6987498195531726e-06,
+      "loss": 0.1447,
+      "step": 7945
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6968923246036996e-06,
+      "loss": 0.1538,
+      "step": 7946
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6950357515912608e-06,
+      "loss": 0.1449,
+      "step": 7947
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6931801007220117e-06,
+      "loss": 0.1453,
+      "step": 7948
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6913253722019885e-06,
+      "loss": 0.1541,
+      "step": 7949
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6894715662371341e-06,
+      "loss": 0.1256,
+      "step": 7950
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.687618683033283e-06,
+      "loss": 0.2124,
+      "step": 7951
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.685766722796174e-06,
+      "loss": 0.1626,
+      "step": 7952
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6839156857314343e-06,
+      "loss": 0.1688,
+      "step": 7953
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6820655720446012e-06,
+      "loss": 0.1732,
+      "step": 7954
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6802163819410977e-06,
+      "loss": 0.1782,
+      "step": 7955
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.678368115626252e-06,
+      "loss": 0.1584,
+      "step": 7956
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6765207733052834e-06,
+      "loss": 0.1873,
+      "step": 7957
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6746743551833133e-06,
+      "loss": 0.1783,
+      "step": 7958
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6728288614653587e-06,
+      "loss": 0.1611,
+      "step": 7959
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.670984292356338e-06,
+      "loss": 0.1533,
+      "step": 7960
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6691406480610572e-06,
+      "loss": 0.1567,
+      "step": 7961
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6672979287842316e-06,
+      "loss": 0.1449,
+      "step": 7962
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.665456134730462e-06,
+      "loss": 0.1965,
+      "step": 7963
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6636152661042549e-06,
+      "loss": 0.1652,
+      "step": 7964
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.661775323110012e-06,
+      "loss": 0.1584,
+      "step": 7965
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6599363059520336e-06,
+      "loss": 0.1768,
+      "step": 7966
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6580982148345114e-06,
+      "loss": 0.1621,
+      "step": 7967
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6562610499615416e-06,
+      "loss": 0.1466,
+      "step": 7968
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6544248115371109e-06,
+      "loss": 0.1931,
+      "step": 7969
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.652589499765107e-06,
+      "loss": 0.157,
+      "step": 7970
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6507551148493184e-06,
+      "loss": 0.1772,
+      "step": 7971
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6489216569934197e-06,
+      "loss": 0.1493,
+      "step": 7972
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6470891264009958e-06,
+      "loss": 0.1337,
+      "step": 7973
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6452575232755152e-06,
+      "loss": 0.1578,
+      "step": 7974
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6434268478203553e-06,
+      "loss": 0.1542,
+      "step": 7975
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6415971002387832e-06,
+      "loss": 0.1289,
+      "step": 7976
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6397682807339677e-06,
+      "loss": 0.1482,
+      "step": 7977
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6379403895089684e-06,
+      "loss": 0.1567,
+      "step": 7978
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.636113426766751e-06,
+      "loss": 0.125,
+      "step": 7979
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6342873927101655e-06,
+      "loss": 0.1711,
+      "step": 7980
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6324622875419704e-06,
+      "loss": 0.1595,
+      "step": 7981
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.630638111464815e-06,
+      "loss": 0.1779,
+      "step": 7982
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6288148646812507e-06,
+      "loss": 0.1533,
+      "step": 7983
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6269925473937177e-06,
+      "loss": 0.1392,
+      "step": 7984
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6251711598045593e-06,
+      "loss": 0.1436,
+      "step": 7985
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6233507021160123e-06,
+      "loss": 0.1819,
+      "step": 7986
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.621531174530211e-06,
+      "loss": 0.1682,
+      "step": 7987
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6197125772491896e-06,
+      "loss": 0.1388,
+      "step": 7988
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6178949104748765e-06,
+      "loss": 0.1633,
+      "step": 7989
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6160781744090925e-06,
+      "loss": 0.1716,
+      "step": 7990
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6142623692535654e-06,
+      "loss": 0.1615,
+      "step": 7991
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6124474952099067e-06,
+      "loss": 0.1588,
+      "step": 7992
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6106335524796346e-06,
+      "loss": 0.1797,
+      "step": 7993
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.608820541264161e-06,
+      "loss": 0.172,
+      "step": 7994
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6070084617647953e-06,
+      "loss": 0.1628,
+      "step": 7995
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6051973141827381e-06,
+      "loss": 0.1489,
+      "step": 7996
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6033870987190947e-06,
+      "loss": 0.136,
+      "step": 7997
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6015778155748585e-06,
+      "loss": 0.1642,
+      "step": 7998
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5997694649509255e-06,
+      "loss": 0.1471,
+      "step": 7999
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5979620470480861e-06,
+      "loss": 0.1414,
+      "step": 8000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.596155562067031e-06,
+      "loss": 0.1674,
+      "step": 8001
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5943500102083365e-06,
+      "loss": 0.196,
+      "step": 8002
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5925453916724887e-06,
+      "loss": 0.1562,
+      "step": 8003
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5907417066598562e-06,
+      "loss": 0.1598,
+      "step": 8004
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5889389553707212e-06,
+      "loss": 0.1663,
+      "step": 8005
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5871371380052457e-06,
+      "loss": 0.1471,
+      "step": 8006
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5853362547634987e-06,
+      "loss": 0.1322,
+      "step": 8007
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5835363058454368e-06,
+      "loss": 0.1593,
+      "step": 8008
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.581737291450921e-06,
+      "loss": 0.1788,
+      "step": 8009
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5799392117797052e-06,
+      "loss": 0.1561,
+      "step": 8010
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5781420670314406e-06,
+      "loss": 0.197,
+      "step": 8011
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.576345857405669e-06,
+      "loss": 0.1482,
+      "step": 8012
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5745505831018393e-06,
+      "loss": 0.1812,
+      "step": 8013
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5727562443192834e-06,
+      "loss": 0.1494,
+      "step": 8014
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5709628412572397e-06,
+      "loss": 0.1833,
+      "step": 8015
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5691703741148378e-06,
+      "loss": 0.1472,
+      "step": 8016
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5673788430911075e-06,
+      "loss": 0.1539,
+      "step": 8017
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5655882483849682e-06,
+      "loss": 0.1553,
+      "step": 8018
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5637985901952412e-06,
+      "loss": 0.1569,
+      "step": 8019
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5620098687206398e-06,
+      "loss": 0.1454,
+      "step": 8020
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5602220841597747e-06,
+      "loss": 0.1559,
+      "step": 8021
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5584352367111544e-06,
+      "loss": 0.1581,
+      "step": 8022
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.556649326573183e-06,
+      "loss": 0.1293,
+      "step": 8023
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5548643539441566e-06,
+      "loss": 0.1486,
+      "step": 8024
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5530803190222722e-06,
+      "loss": 0.1543,
+      "step": 8025
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5512972220056177e-06,
+      "loss": 0.2029,
+      "step": 8026
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5495150630921807e-06,
+      "loss": 0.152,
+      "step": 8027
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.547733842479845e-06,
+      "loss": 0.1699,
+      "step": 8028
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.545953560366389e-06,
+      "loss": 0.1388,
+      "step": 8029
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5441742169494833e-06,
+      "loss": 0.1414,
+      "step": 8030
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5423958124267025e-06,
+      "loss": 0.1826,
+      "step": 8031
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5406183469955072e-06,
+      "loss": 0.1742,
+      "step": 8032
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5388418208532596e-06,
+      "loss": 0.1584,
+      "step": 8033
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5370662341972187e-06,
+      "loss": 0.1411,
+      "step": 8034
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5352915872245377e-06,
+      "loss": 0.1653,
+      "step": 8035
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5335178801322615e-06,
+      "loss": 0.1228,
+      "step": 8036
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5317451131173377e-06,
+      "loss": 0.1586,
+      "step": 8037
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5299732863766015e-06,
+      "loss": 0.1936,
+      "step": 8038
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5282024001067908e-06,
+      "loss": 0.199,
+      "step": 8039
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5264324545045372e-06,
+      "loss": 0.1614,
+      "step": 8040
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5246634497663637e-06,
+      "loss": 0.1189,
+      "step": 8041
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5228953860886963e-06,
+      "loss": 0.1814,
+      "step": 8042
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.521128263667847e-06,
+      "loss": 0.178,
+      "step": 8043
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5193620827000311e-06,
+      "loss": 0.198,
+      "step": 8044
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5175968433813581e-06,
+      "loss": 0.1276,
+      "step": 8045
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5158325459078327e-06,
+      "loss": 0.1338,
+      "step": 8046
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.514069190475349e-06,
+      "loss": 0.1492,
+      "step": 8047
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5123067772797072e-06,
+      "loss": 0.1543,
+      "step": 8048
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.510545306516592e-06,
+      "loss": 0.1216,
+      "step": 8049
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5087847783815912e-06,
+      "loss": 0.1484,
+      "step": 8050
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.507025193070184e-06,
+      "loss": 0.1714,
+      "step": 8051
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.505266550777751e-06,
+      "loss": 0.1367,
+      "step": 8052
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5035088516995566e-06,
+      "loss": 0.1641,
+      "step": 8053
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.501752096030774e-06,
+      "loss": 0.1846,
+      "step": 8054
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4999962839664561e-06,
+      "loss": 0.1697,
+      "step": 8055
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.498241415701569e-06,
+      "loss": 0.1473,
+      "step": 8056
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.496487491430959e-06,
+      "loss": 0.1848,
+      "step": 8057
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4947345113493772e-06,
+      "loss": 0.1504,
+      "step": 8058
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4929824756514621e-06,
+      "loss": 0.156,
+      "step": 8059
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.491231384531756e-06,
+      "loss": 0.146,
+      "step": 8060
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4894812381846835e-06,
+      "loss": 0.1953,
+      "step": 8061
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4877320368045823e-06,
+      "loss": 0.1715,
+      "step": 8062
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4859837805856691e-06,
+      "loss": 0.1599,
+      "step": 8063
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4842364697220657e-06,
+      "loss": 0.136,
+      "step": 8064
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.48249010440778e-06,
+      "loss": 0.1724,
+      "step": 8065
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4807446848367235e-06,
+      "loss": 0.1477,
+      "step": 8066
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.479000211202698e-06,
+      "loss": 0.1909,
+      "step": 8067
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.477256683699404e-06,
+      "loss": 0.1763,
+      "step": 8068
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4755141025204312e-06,
+      "loss": 0.157,
+      "step": 8069
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4737724678592702e-06,
+      "loss": 0.1812,
+      "step": 8070
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4720317799092998e-06,
+      "loss": 0.1677,
+      "step": 8071
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4702920388637998e-06,
+      "loss": 0.1826,
+      "step": 8072
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4685532449159423e-06,
+      "loss": 0.1912,
+      "step": 8073
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4668153982587985e-06,
+      "loss": 0.142,
+      "step": 8074
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.465078499085324e-06,
+      "loss": 0.1547,
+      "step": 8075
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4633425475883812e-06,
+      "loss": 0.1746,
+      "step": 8076
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.461607543960717e-06,
+      "loss": 0.1479,
+      "step": 8077
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4598734883949805e-06,
+      "loss": 0.1486,
+      "step": 8078
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4581403810837124e-06,
+      "loss": 0.1704,
+      "step": 8079
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.456408222219351e-06,
+      "loss": 0.1555,
+      "step": 8080
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4546770119942222e-06,
+      "loss": 0.1333,
+      "step": 8081
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4529467506005557e-06,
+      "loss": 0.1639,
+      "step": 8082
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4512174382304678e-06,
+      "loss": 0.1365,
+      "step": 8083
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.449489075075974e-06,
+      "loss": 0.1604,
+      "step": 8084
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4477616613289834e-06,
+      "loss": 0.1846,
+      "step": 8085
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4460351971813037e-06,
+      "loss": 0.1366,
+      "step": 8086
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4443096828246262e-06,
+      "loss": 0.1445,
+      "step": 8087
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4425851184505501e-06,
+      "loss": 0.1604,
+      "step": 8088
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.440861504250557e-06,
+      "loss": 0.1216,
+      "step": 8089
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4391388404160323e-06,
+      "loss": 0.1655,
+      "step": 8090
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4374171271382508e-06,
+      "loss": 0.1666,
+      "step": 8091
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4356963646083865e-06,
+      "loss": 0.1443,
+      "step": 8092
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4339765530174998e-06,
+      "loss": 0.1398,
+      "step": 8093
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4322576925565545e-06,
+      "loss": 0.187,
+      "step": 8094
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4305397834164015e-06,
+      "loss": 0.12,
+      "step": 8095
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.428822825787789e-06,
+      "loss": 0.1658,
+      "step": 8096
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4271068198613602e-06,
+      "loss": 0.1396,
+      "step": 8097
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4253917658276563e-06,
+      "loss": 0.1664,
+      "step": 8098
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4236776638771022e-06,
+      "loss": 0.1511,
+      "step": 8099
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4219645142000293e-06,
+      "loss": 0.1541,
+      "step": 8100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.42025231698665e-06,
+      "loss": 0.1382,
+      "step": 8101
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4185410724270875e-06,
+      "loss": 0.1317,
+      "step": 8102
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4168307807113435e-06,
+      "loss": 0.1831,
+      "step": 8103
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4151214420293258e-06,
+      "loss": 0.1782,
+      "step": 8104
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.413413056570827e-06,
+      "loss": 0.2065,
+      "step": 8105
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4117056245255344e-06,
+      "loss": 0.1855,
+      "step": 8106
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.409999146083043e-06,
+      "loss": 0.2026,
+      "step": 8107
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4082936214328236e-06,
+      "loss": 0.1434,
+      "step": 8108
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4065890507642555e-06,
+      "loss": 0.1628,
+      "step": 8109
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4048854342666006e-06,
+      "loss": 0.1833,
+      "step": 8110
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4031827721290247e-06,
+      "loss": 0.1892,
+      "step": 8111
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4014810645405764e-06,
+      "loss": 0.1519,
+      "step": 8112
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3997803116902143e-06,
+      "loss": 0.1716,
+      "step": 8113
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3980805137667752e-06,
+      "loss": 0.1921,
+      "step": 8114
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3963816709590017e-06,
+      "loss": 0.1389,
+      "step": 8115
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3946837834555183e-06,
+      "loss": 0.1519,
+      "step": 8116
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3929868514448552e-06,
+      "loss": 0.1729,
+      "step": 8117
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3912908751154298e-06,
+      "loss": 0.1868,
+      "step": 8118
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3895958546555578e-06,
+      "loss": 0.1757,
+      "step": 8119
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.387901790253442e-06,
+      "loss": 0.1425,
+      "step": 8120
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3862086820971866e-06,
+      "loss": 0.1346,
+      "step": 8121
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3845165303747831e-06,
+      "loss": 0.1621,
+      "step": 8122
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3828253352741217e-06,
+      "loss": 0.1914,
+      "step": 8123
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3811350969829828e-06,
+      "loss": 0.1519,
+      "step": 8124
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3794458156890477e-06,
+      "loss": 0.1609,
+      "step": 8125
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3777574915798786e-06,
+      "loss": 0.1809,
+      "step": 8126
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3760701248429453e-06,
+      "loss": 0.1414,
+      "step": 8127
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3743837156655992e-06,
+      "loss": 0.156,
+      "step": 8128
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.372698264235095e-06,
+      "loss": 0.1704,
+      "step": 8129
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.371013770738575e-06,
+      "loss": 0.1422,
+      "step": 8130
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3693302353630799e-06,
+      "loss": 0.1606,
+      "step": 8131
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3676476582955378e-06,
+      "loss": 0.1458,
+      "step": 8132
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3659660397227782e-06,
+      "loss": 0.1509,
+      "step": 8133
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3642853798315147e-06,
+      "loss": 0.1581,
+      "step": 8134
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.362605678808363e-06,
+      "loss": 0.188,
+      "step": 8135
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3609269368398281e-06,
+      "loss": 0.1564,
+      "step": 8136
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.359249154112312e-06,
+      "loss": 0.144,
+      "step": 8137
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3575723308121036e-06,
+      "loss": 0.1528,
+      "step": 8138
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3558964671253926e-06,
+      "loss": 0.1809,
+      "step": 8139
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.354221563238255e-06,
+      "loss": 0.1523,
+      "step": 8140
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.352547619336666e-06,
+      "loss": 0.1617,
+      "step": 8141
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3508746356064939e-06,
+      "loss": 0.1373,
+      "step": 8142
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3492026122334979e-06,
+      "loss": 0.156,
+      "step": 8143
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3475315494033304e-06,
+      "loss": 0.1467,
+      "step": 8144
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3458614473015396e-06,
+      "loss": 0.1761,
+      "step": 8145
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3441923061135632e-06,
+      "loss": 0.1428,
+      "step": 8146
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3425241260247345e-06,
+      "loss": 0.1748,
+      "step": 8147
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3408569072202836e-06,
+      "loss": 0.1473,
+      "step": 8148
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3391906498853302e-06,
+      "loss": 0.1759,
+      "step": 8149
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3375253542048826e-06,
+      "loss": 0.1479,
+      "step": 8150
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3358610203638534e-06,
+      "loss": 0.155,
+      "step": 8151
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3341976485470353e-06,
+      "loss": 0.165,
+      "step": 8152
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3325352389391288e-06,
+      "loss": 0.1371,
+      "step": 8153
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.330873791724714e-06,
+      "loss": 0.1664,
+      "step": 8154
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.329213307088274e-06,
+      "loss": 0.14,
+      "step": 8155
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3275537852141773e-06,
+      "loss": 0.1505,
+      "step": 8156
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.325895226286693e-06,
+      "loss": 0.1677,
+      "step": 8157
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3242376304899729e-06,
+      "loss": 0.1597,
+      "step": 8158
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3225809980080772e-06,
+      "loss": 0.1677,
+      "step": 8159
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3209253290249447e-06,
+      "loss": 0.1488,
+      "step": 8160
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3192706237244167e-06,
+      "loss": 0.1877,
+      "step": 8161
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.317616882290218e-06,
+      "loss": 0.1378,
+      "step": 8162
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.315964104905977e-06,
+      "loss": 0.1774,
+      "step": 8163
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3143122917552077e-06,
+      "loss": 0.1523,
+      "step": 8164
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3126614430213225e-06,
+      "loss": 0.151,
+      "step": 8165
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3110115588876194e-06,
+      "loss": 0.174,
+      "step": 8166
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3093626395372971e-06,
+      "loss": 0.1533,
+      "step": 8167
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3077146851534417e-06,
+      "loss": 0.147,
+      "step": 8168
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3060676959190332e-06,
+      "loss": 0.1454,
+      "step": 8169
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3044216720169479e-06,
+      "loss": 0.1196,
+      "step": 8170
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3027766136299524e-06,
+      "loss": 0.198,
+      "step": 8171
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3011325209407056e-06,
+      "loss": 0.1488,
+      "step": 8172
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2994893941317565e-06,
+      "loss": 0.1982,
+      "step": 8173
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2978472333855529e-06,
+      "loss": 0.2073,
+      "step": 8174
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2962060388844323e-06,
+      "loss": 0.1953,
+      "step": 8175
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2945658108106263e-06,
+      "loss": 0.1514,
+      "step": 8176
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.292926549346254e-06,
+      "loss": 0.151,
+      "step": 8177
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2912882546733353e-06,
+      "loss": 0.1787,
+      "step": 8178
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2896509269737745e-06,
+      "loss": 0.1598,
+      "step": 8179
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2880145664293753e-06,
+      "loss": 0.1525,
+      "step": 8180
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2863791732218301e-06,
+      "loss": 0.1653,
+      "step": 8181
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.284744747532728e-06,
+      "loss": 0.1381,
+      "step": 8182
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2831112895435428e-06,
+      "loss": 0.1892,
+      "step": 8183
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2814787994356515e-06,
+      "loss": 0.1777,
+      "step": 8184
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2798472773903114e-06,
+      "loss": 0.1553,
+      "step": 8185
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2782167235886833e-06,
+      "loss": 0.1542,
+      "step": 8186
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2765871382118145e-06,
+      "loss": 0.157,
+      "step": 8187
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.274958521440649e-06,
+      "loss": 0.1497,
+      "step": 8188
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2733308734560167e-06,
+      "loss": 0.1729,
+      "step": 8189
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2717041944386466e-06,
+      "loss": 0.1683,
+      "step": 8190
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.270078484569155e-06,
+      "loss": 0.1409,
+      "step": 8191
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.268453744028053e-06,
+      "loss": 0.1504,
+      "step": 8192
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2668299729957456e-06,
+      "loss": 0.1787,
+      "step": 8193
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2652071716525305e-06,
+      "loss": 0.1522,
+      "step": 8194
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2635853401785902e-06,
+      "loss": 0.1816,
+      "step": 8195
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2619644787540108e-06,
+      "loss": 0.1821,
+      "step": 8196
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2603445875587594e-06,
+      "loss": 0.1648,
+      "step": 8197
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2587256667727033e-06,
+      "loss": 0.2002,
+      "step": 8198
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2571077165756006e-06,
+      "loss": 0.1498,
+      "step": 8199
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2554907371471015e-06,
+      "loss": 0.1982,
+      "step": 8200
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.253874728666744e-06,
+      "loss": 0.1368,
+      "step": 8201
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2522596913139672e-06,
+      "loss": 0.1687,
+      "step": 8202
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.250645625268091e-06,
+      "loss": 0.1696,
+      "step": 8203
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2490325307083363e-06,
+      "loss": 0.1493,
+      "step": 8204
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2474204078138141e-06,
+      "loss": 0.1753,
+      "step": 8205
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.245809256763527e-06,
+      "loss": 0.1505,
+      "step": 8206
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2441990777363677e-06,
+      "loss": 0.1912,
+      "step": 8207
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.242589870911125e-06,
+      "loss": 0.134,
+      "step": 8208
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2409816364664718e-06,
+      "loss": 0.1462,
+      "step": 8209
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2393743745809883e-06,
+      "loss": 0.1621,
+      "step": 8210
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2377680854331286e-06,
+      "loss": 0.1812,
+      "step": 8211
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2361627692012523e-06,
+      "loss": 0.1512,
+      "step": 8212
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2345584260636024e-06,
+      "loss": 0.1731,
+      "step": 8213
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2329550561983183e-06,
+      "loss": 0.1677,
+      "step": 8214
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2313526597834325e-06,
+      "loss": 0.1495,
+      "step": 8215
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2297512369968678e-06,
+      "loss": 0.1348,
+      "step": 8216
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2281507880164345e-06,
+      "loss": 0.1326,
+      "step": 8217
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2265513130198436e-06,
+      "loss": 0.1597,
+      "step": 8218
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2249528121846888e-06,
+      "loss": 0.1387,
+      "step": 8219
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.223355285688461e-06,
+      "loss": 0.1763,
+      "step": 8220
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2217587337085424e-06,
+      "loss": 0.1532,
+      "step": 8221
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2201631564222095e-06,
+      "loss": 0.2034,
+      "step": 8222
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2185685540066216e-06,
+      "loss": 0.1307,
+      "step": 8223
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2169749266388408e-06,
+      "loss": 0.1797,
+      "step": 8224
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.215382274495811e-06,
+      "loss": 0.149,
+      "step": 8225
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2137905977543762e-06,
+      "loss": 0.1694,
+      "step": 8226
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2121998965912674e-06,
+      "loss": 0.1353,
+      "step": 8227
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2106101711831108e-06,
+      "loss": 0.1406,
+      "step": 8228
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2090214217064177e-06,
+      "loss": 0.1726,
+      "step": 8229
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.207433648337598e-06,
+      "loss": 0.1724,
+      "step": 8230
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2058468512529486e-06,
+      "loss": 0.1208,
+      "step": 8231
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2042610306286595e-06,
+      "loss": 0.1412,
+      "step": 8232
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2026761866408143e-06,
+      "loss": 0.1483,
+      "step": 8233
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.201092319465389e-06,
+      "loss": 0.1748,
+      "step": 8234
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1995094292782416e-06,
+      "loss": 0.1807,
+      "step": 8235
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.197927516255135e-06,
+      "loss": 0.1334,
+      "step": 8236
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1963465805717134e-06,
+      "loss": 0.1597,
+      "step": 8237
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.194766622403517e-06,
+      "loss": 0.1565,
+      "step": 8238
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1931876419259792e-06,
+      "loss": 0.1448,
+      "step": 8239
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1916096393144183e-06,
+      "loss": 0.1729,
+      "step": 8240
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1900326147440521e-06,
+      "loss": 0.2073,
+      "step": 8241
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.188456568389983e-06,
+      "loss": 0.1421,
+      "step": 8242
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1868815004272072e-06,
+      "loss": 0.1677,
+      "step": 8243
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1853074110306139e-06,
+      "loss": 0.1384,
+      "step": 8244
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1837343003749856e-06,
+      "loss": 0.1709,
+      "step": 8245
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1821621686349872e-06,
+      "loss": 0.1503,
+      "step": 8246
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1805910159851852e-06,
+      "loss": 0.1882,
+      "step": 8247
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1790208426000283e-06,
+      "loss": 0.1633,
+      "step": 8248
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1774516486538644e-06,
+      "loss": 0.2087,
+      "step": 8249
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.175883434320928e-06,
+      "loss": 0.1543,
+      "step": 8250
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1743161997753483e-06,
+      "loss": 0.1204,
+      "step": 8251
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1727499451911396e-06,
+      "loss": 0.1897,
+      "step": 8252
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1711846707422158e-06,
+      "loss": 0.1454,
+      "step": 8253
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1696203766023717e-06,
+      "loss": 0.1624,
+      "step": 8254
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1680570629453025e-06,
+      "loss": 0.1555,
+      "step": 8255
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1664947299445917e-06,
+      "loss": 0.1438,
+      "step": 8256
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1649333777737126e-06,
+      "loss": 0.126,
+      "step": 8257
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1633730066060289e-06,
+      "loss": 0.1677,
+      "step": 8258
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1618136166147986e-06,
+      "loss": 0.1488,
+      "step": 8259
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1602552079731644e-06,
+      "loss": 0.1582,
+      "step": 8260
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1586977808541722e-06,
+      "loss": 0.196,
+      "step": 8261
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1571413354307449e-06,
+      "loss": 0.1626,
+      "step": 8262
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1555858718757062e-06,
+      "loss": 0.1359,
+      "step": 8263
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.154031390361764e-06,
+      "loss": 0.1355,
+      "step": 8264
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1524778910615242e-06,
+      "loss": 0.1506,
+      "step": 8265
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1509253741474735e-06,
+      "loss": 0.1359,
+      "step": 8266
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.149373839792004e-06,
+      "loss": 0.1766,
+      "step": 8267
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1478232881673856e-06,
+      "loss": 0.1531,
+      "step": 8268
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1462737194457873e-06,
+      "loss": 0.1985,
+      "step": 8269
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1447251337992616e-06,
+      "loss": 0.1622,
+      "step": 8270
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1431775313997583e-06,
+      "loss": 0.1285,
+      "step": 8271
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.141630912419115e-06,
+      "loss": 0.1462,
+      "step": 8272
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1400852770290638e-06,
+      "loss": 0.1648,
+      "step": 8273
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1385406254012199e-06,
+      "loss": 0.1643,
+      "step": 8274
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1369969577070982e-06,
+      "loss": 0.1731,
+      "step": 8275
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1354542741180962e-06,
+      "loss": 0.1514,
+      "step": 8276
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.133912574805508e-06,
+      "loss": 0.1611,
+      "step": 8277
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1323718599405164e-06,
+      "loss": 0.1362,
+      "step": 8278
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1308321296941981e-06,
+      "loss": 0.1982,
+      "step": 8279
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1292933842375109e-06,
+      "loss": 0.1324,
+      "step": 8280
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1277556237413157e-06,
+      "loss": 0.1951,
+      "step": 8281
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1262188483763525e-06,
+      "loss": 0.2004,
+      "step": 8282
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1246830583132607e-06,
+      "loss": 0.1514,
+      "step": 8283
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1231482537225669e-06,
+      "loss": 0.1726,
+      "step": 8284
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1216144347746904e-06,
+      "loss": 0.1604,
+      "step": 8285
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.120081601639934e-06,
+      "loss": 0.2131,
+      "step": 8286
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1185497544885004e-06,
+      "loss": 0.1492,
+      "step": 8287
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1170188934904758e-06,
+      "loss": 0.1982,
+      "step": 8288
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1154890188158407e-06,
+      "loss": 0.1547,
+      "step": 8289
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.113960130634465e-06,
+      "loss": 0.1638,
+      "step": 8290
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1124322291161105e-06,
+      "loss": 0.1748,
+      "step": 8291
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.110905314430425e-06,
+      "loss": 0.167,
+      "step": 8292
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1093793867469538e-06,
+      "loss": 0.1719,
+      "step": 8293
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1078544462351238e-06,
+      "loss": 0.1498,
+      "step": 8294
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1063304930642604e-06,
+      "loss": 0.1521,
+      "step": 8295
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.104807527403574e-06,
+      "loss": 0.1465,
+      "step": 8296
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1032855494221706e-06,
+      "loss": 0.1831,
+      "step": 8297
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.101764559289039e-06,
+      "loss": 0.1841,
+      "step": 8298
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1002445571730669e-06,
+      "loss": 0.1731,
+      "step": 8299
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0987255432430233e-06,
+      "loss": 0.1647,
+      "step": 8300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0972075176675745e-06,
+      "loss": 0.1742,
+      "step": 8301
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0956904806152758e-06,
+      "loss": 0.1532,
+      "step": 8302
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0941744322545722e-06,
+      "loss": 0.1423,
+      "step": 8303
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.092659372753796e-06,
+      "loss": 0.1455,
+      "step": 8304
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0911453022811735e-06,
+      "loss": 0.1377,
+      "step": 8305
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0896322210048181e-06,
+      "loss": 0.1526,
+      "step": 8306
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0881201290927368e-06,
+      "loss": 0.1816,
+      "step": 8307
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.086609026712826e-06,
+      "loss": 0.1729,
+      "step": 8308
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0850989140328683e-06,
+      "loss": 0.146,
+      "step": 8309
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0835897912205418e-06,
+      "loss": 0.1471,
+      "step": 8310
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.082081658443408e-06,
+      "loss": 0.1672,
+      "step": 8311
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0805745158689297e-06,
+      "loss": 0.1891,
+      "step": 8312
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0790683636644472e-06,
+      "loss": 0.1775,
+      "step": 8313
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0775632019971993e-06,
+      "loss": 0.1318,
+      "step": 8314
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.07605903103431e-06,
+      "loss": 0.1613,
+      "step": 8315
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0745558509427968e-06,
+      "loss": 0.1844,
+      "step": 8316
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0730536618895604e-06,
+      "loss": 0.2014,
+      "step": 8317
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0715524640414055e-06,
+      "loss": 0.147,
+      "step": 8318
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.070052257565012e-06,
+      "loss": 0.1537,
+      "step": 8319
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0685530426269574e-06,
+      "loss": 0.1488,
+      "step": 8320
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0670548193937058e-06,
+      "loss": 0.1575,
+      "step": 8321
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0655575880316127e-06,
+      "loss": 0.1678,
+      "step": 8322
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.064061348706925e-06,
+      "loss": 0.1643,
+      "step": 8323
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0625661015857802e-06,
+      "loss": 0.1395,
+      "step": 8324
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0610718468341975e-06,
+      "loss": 0.1122,
+      "step": 8325
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.059578584618096e-06,
+      "loss": 0.1561,
+      "step": 8326
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0580863151032784e-06,
+      "loss": 0.1543,
+      "step": 8327
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.056595038455438e-06,
+      "loss": 0.1571,
+      "step": 8328
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0551047548401616e-06,
+      "loss": 0.1526,
+      "step": 8329
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0536154644229245e-06,
+      "loss": 0.1423,
+      "step": 8330
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0521271673690857e-06,
+      "loss": 0.1797,
+      "step": 8331
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.050639863843903e-06,
+      "loss": 0.1301,
+      "step": 8332
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0491535540125153e-06,
+      "loss": 0.1354,
+      "step": 8333
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0476682380399572e-06,
+      "loss": 0.1423,
+      "step": 8334
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0461839160911502e-06,
+      "loss": 0.1797,
+      "step": 8335
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0447005883309103e-06,
+      "loss": 0.1844,
+      "step": 8336
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0432182549239333e-06,
+      "loss": 0.147,
+      "step": 8337
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0417369160348145e-06,
+      "loss": 0.1672,
+      "step": 8338
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0402565718280322e-06,
+      "loss": 0.1594,
+      "step": 8339
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0387772224679572e-06,
+      "loss": 0.1849,
+      "step": 8340
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.037298868118849e-06,
+      "loss": 0.1392,
+      "step": 8341
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0358215089448597e-06,
+      "loss": 0.1483,
+      "step": 8342
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.034345145110024e-06,
+      "loss": 0.1633,
+      "step": 8343
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0328697767782748e-06,
+      "loss": 0.1267,
+      "step": 8344
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0313954041134256e-06,
+      "loss": 0.1422,
+      "step": 8345
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0299220272791844e-06,
+      "loss": 0.136,
+      "step": 8346
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0284496464391492e-06,
+      "loss": 0.2087,
+      "step": 8347
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.026978261756808e-06,
+      "loss": 0.2012,
+      "step": 8348
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0255078733955314e-06,
+      "loss": 0.1201,
+      "step": 8349
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.024038481518589e-06,
+      "loss": 0.1614,
+      "step": 8350
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0225700862891308e-06,
+      "loss": 0.1982,
+      "step": 8351
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0211026878702024e-06,
+      "loss": 0.1794,
+      "step": 8352
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0196362864247367e-06,
+      "loss": 0.146,
+      "step": 8353
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0181708821155578e-06,
+      "loss": 0.1567,
+      "step": 8354
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0167064751053724e-06,
+      "loss": 0.1334,
+      "step": 8355
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0152430655567858e-06,
+      "loss": 0.1366,
+      "step": 8356
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0137806536322825e-06,
+      "loss": 0.1594,
+      "step": 8357
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0123192394942493e-06,
+      "loss": 0.1648,
+      "step": 8358
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0108588233049477e-06,
+      "loss": 0.1931,
+      "step": 8359
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.009399405226541e-06,
+      "loss": 0.144,
+      "step": 8360
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0079409854210699e-06,
+      "loss": 0.1754,
+      "step": 8361
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0064835640504754e-06,
+      "loss": 0.1577,
+      "step": 8362
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0050271412765766e-06,
+      "loss": 0.1451,
+      "step": 8363
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.003571717261096e-06,
+      "loss": 0.1794,
+      "step": 8364
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0021172921656296e-06,
+      "loss": 0.1727,
+      "step": 8365
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0006638661516754e-06,
+      "loss": 0.1632,
+      "step": 8366
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.992114393806096e-07,
+      "loss": 0.1191,
+      "step": 8367
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.977600120137054e-07,
+      "loss": 0.1702,
+      "step": 8368
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.963095842121218e-07,
+      "loss": 0.1597,
+      "step": 8369
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.948601561369086e-07,
+      "loss": 0.1309,
+      "step": 8370
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.934117279490009e-07,
+      "loss": 0.1707,
+      "step": 8371
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.919642998092284e-07,
+      "loss": 0.1677,
+      "step": 8372
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.905178718783016e-07,
+      "loss": 0.1765,
+      "step": 8373
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.890724443168286e-07,
+      "loss": 0.1377,
+      "step": 8374
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.876280172853037e-07,
+      "loss": 0.1793,
+      "step": 8375
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.861845909441059e-07,
+      "loss": 0.1503,
+      "step": 8376
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.847421654535095e-07,
+      "loss": 0.113,
+      "step": 8377
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.833007409736706e-07,
+      "loss": 0.1577,
+      "step": 8378
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.8186031766464e-07,
+      "loss": 0.1555,
+      "step": 8379
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.80420895686356e-07,
+      "loss": 0.1785,
+      "step": 8380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.789824751986465e-07,
+      "loss": 0.1929,
+      "step": 8381
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.775450563612233e-07,
+      "loss": 0.1948,
+      "step": 8382
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.761086393336926e-07,
+      "loss": 0.1736,
+      "step": 8383
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.746732242755464e-07,
+      "loss": 0.1404,
+      "step": 8384
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.732388113461656e-07,
+      "loss": 0.1317,
+      "step": 8385
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.718054007048206e-07,
+      "loss": 0.1571,
+      "step": 8386
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.703729925106742e-07,
+      "loss": 0.1736,
+      "step": 8387
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.689415869227692e-07,
+      "loss": 0.2043,
+      "step": 8388
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.675111841000451e-07,
+      "loss": 0.1346,
+      "step": 8389
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.660817842013248e-07,
+      "loss": 0.1578,
+      "step": 8390
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.646533873853225e-07,
+      "loss": 0.175,
+      "step": 8391
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.632259938106403e-07,
+      "loss": 0.1682,
+      "step": 8392
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.617996036357723e-07,
+      "loss": 0.1729,
+      "step": 8393
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.603742170190933e-07,
+      "loss": 0.1613,
+      "step": 8394
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.589498341188751e-07,
+      "loss": 0.1537,
+      "step": 8395
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.575264550932705e-07,
+      "loss": 0.1447,
+      "step": 8396
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.561040801003275e-07,
+      "loss": 0.1689,
+      "step": 8397
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.546827092979781e-07,
+      "loss": 0.1421,
+      "step": 8398
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.532623428440479e-07,
+      "loss": 0.1475,
+      "step": 8399
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.518429808962438e-07,
+      "loss": 0.161,
+      "step": 8400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.504246236121672e-07,
+      "loss": 0.1594,
+      "step": 8401
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.490072711493025e-07,
+      "loss": 0.161,
+      "step": 8402
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.475909236650283e-07,
+      "loss": 0.1481,
+      "step": 8403
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.461755813166085e-07,
+      "loss": 0.1743,
+      "step": 8404
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.447612442611975e-07,
+      "loss": 0.1403,
+      "step": 8405
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.433479126558331e-07,
+      "loss": 0.1897,
+      "step": 8406
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.419355866574487e-07,
+      "loss": 0.1323,
+      "step": 8407
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.405242664228576e-07,
+      "loss": 0.1198,
+      "step": 8408
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.391139521087689e-07,
+      "loss": 0.1786,
+      "step": 8409
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.377046438717763e-07,
+      "loss": 0.1456,
+      "step": 8410
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.362963418683657e-07,
+      "loss": 0.1353,
+      "step": 8411
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.348890462549021e-07,
+      "loss": 0.1472,
+      "step": 8412
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.334827571876504e-07,
+      "loss": 0.1293,
+      "step": 8413
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.320774748227524e-07,
+      "loss": 0.1565,
+      "step": 8414
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.30673199316251e-07,
+      "loss": 0.1848,
+      "step": 8415
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.292699308240649e-07,
+      "loss": 0.1526,
+      "step": 8416
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.278676695020095e-07,
+      "loss": 0.1561,
+      "step": 8417
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.264664155057823e-07,
+      "loss": 0.1826,
+      "step": 8418
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.250661689909724e-07,
+      "loss": 0.1609,
+      "step": 8419
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.236669301130563e-07,
+      "loss": 0.1438,
+      "step": 8420
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.222686990274022e-07,
+      "loss": 0.1653,
+      "step": 8421
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.208714758892579e-07,
+      "loss": 0.1436,
+      "step": 8422
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.194752608537694e-07,
+      "loss": 0.1302,
+      "step": 8423
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.180800540759604e-07,
+      "loss": 0.1775,
+      "step": 8424
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.166858557107505e-07,
+      "loss": 0.1931,
+      "step": 8425
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.152926659129446e-07,
+      "loss": 0.1677,
+      "step": 8426
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.13900484837238e-07,
+      "loss": 0.1294,
+      "step": 8427
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.125093126382078e-07,
+      "loss": 0.1787,
+      "step": 8428
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.111191494703264e-07,
+      "loss": 0.1765,
+      "step": 8429
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.09729995487948e-07,
+      "loss": 0.1396,
+      "step": 8430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.083418508453178e-07,
+      "loss": 0.1461,
+      "step": 8431
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.069547156965708e-07,
+      "loss": 0.1531,
+      "step": 8432
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.055685901957289e-07,
+      "loss": 0.141,
+      "step": 8433
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.041834744966959e-07,
+      "loss": 0.1995,
+      "step": 8434
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.027993687532733e-07,
+      "loss": 0.1268,
+      "step": 8435
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.014162731191411e-07,
+      "loss": 0.1843,
+      "step": 8436
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.000341877478747e-07,
+      "loss": 0.167,
+      "step": 8437
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.986531127929332e-07,
+      "loss": 0.1646,
+      "step": 8438
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.972730484076664e-07,
+      "loss": 0.1426,
+      "step": 8439
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.95893994745306e-07,
+      "loss": 0.1512,
+      "step": 8440
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.945159519589797e-07,
+      "loss": 0.1676,
+      "step": 8441
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.931389202016949e-07,
+      "loss": 0.1436,
+      "step": 8442
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.917628996263528e-07,
+      "loss": 0.1641,
+      "step": 8443
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.90387890385741e-07,
+      "loss": 0.1262,
+      "step": 8444
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.890138926325309e-07,
+      "loss": 0.1821,
+      "step": 8445
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.876409065192882e-07,
+      "loss": 0.1821,
+      "step": 8446
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.862689321984586e-07,
+      "loss": 0.1824,
+      "step": 8447
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.848979698223814e-07,
+      "loss": 0.1595,
+      "step": 8448
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.835280195432827e-07,
+      "loss": 0.1328,
+      "step": 8449
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.821590815132752e-07,
+      "loss": 0.1466,
+      "step": 8450
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.807911558843573e-07,
+      "loss": 0.1702,
+      "step": 8451
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.794242428084198e-07,
+      "loss": 0.1592,
+      "step": 8452
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.780583424372335e-07,
+      "loss": 0.1333,
+      "step": 8453
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.766934549224648e-07,
+      "loss": 0.1492,
+      "step": 8454
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.753295804156636e-07,
+      "loss": 0.156,
+      "step": 8455
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.739667190682699e-07,
+      "loss": 0.1868,
+      "step": 8456
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.726048710316048e-07,
+      "loss": 0.1829,
+      "step": 8457
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.712440364568864e-07,
+      "loss": 0.1765,
+      "step": 8458
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.698842154952103e-07,
+      "loss": 0.1763,
+      "step": 8459
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.685254082975669e-07,
+      "loss": 0.1804,
+      "step": 8460
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.671676150148311e-07,
+      "loss": 0.1265,
+      "step": 8461
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.658108357977679e-07,
+      "loss": 0.1888,
+      "step": 8462
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.644550707970246e-07,
+      "loss": 0.196,
+      "step": 8463
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.631003201631405e-07,
+      "loss": 0.1504,
+      "step": 8464
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.617465840465377e-07,
+      "loss": 0.198,
+      "step": 8465
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.603938625975339e-07,
+      "loss": 0.1554,
+      "step": 8466
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.590421559663243e-07,
+      "loss": 0.1478,
+      "step": 8467
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.576914643029977e-07,
+      "loss": 0.1691,
+      "step": 8468
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.563417877575275e-07,
+      "loss": 0.1442,
+      "step": 8469
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.54993126479775e-07,
+      "loss": 0.1648,
+      "step": 8470
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.536454806194904e-07,
+      "loss": 0.1372,
+      "step": 8471
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.522988503263108e-07,
+      "loss": 0.193,
+      "step": 8472
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.509532357497563e-07,
+      "loss": 0.146,
+      "step": 8473
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.496086370392409e-07,
+      "loss": 0.1716,
+      "step": 8474
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.482650543440585e-07,
+      "loss": 0.1259,
+      "step": 8475
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.469224878133953e-07,
+      "loss": 0.1398,
+      "step": 8476
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.455809375963253e-07,
+      "loss": 0.1851,
+      "step": 8477
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.442404038418084e-07,
+      "loss": 0.1254,
+      "step": 8478
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.429008866986865e-07,
+      "loss": 0.1555,
+      "step": 8479
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.415623863156985e-07,
+      "loss": 0.1775,
+      "step": 8480
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.402249028414611e-07,
+      "loss": 0.1396,
+      "step": 8481
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.38888436424482e-07,
+      "loss": 0.1624,
+      "step": 8482
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.375529872131582e-07,
+      "loss": 0.1543,
+      "step": 8483
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.362185553557723e-07,
+      "loss": 0.1558,
+      "step": 8484
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.348851410004899e-07,
+      "loss": 0.1517,
+      "step": 8485
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.335527442953706e-07,
+      "loss": 0.1708,
+      "step": 8486
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.322213653883526e-07,
+      "loss": 0.142,
+      "step": 8487
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.308910044272689e-07,
+      "loss": 0.1885,
+      "step": 8488
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.295616615598356e-07,
+      "loss": 0.1487,
+      "step": 8489
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.282333369336593e-07,
+      "loss": 0.1548,
+      "step": 8490
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.269060306962262e-07,
+      "loss": 0.1472,
+      "step": 8491
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.255797429949175e-07,
+      "loss": 0.1707,
+      "step": 8492
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.242544739769942e-07,
+      "loss": 0.1561,
+      "step": 8493
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.229302237896108e-07,
+      "loss": 0.1545,
+      "step": 8494
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.216069925798053e-07,
+      "loss": 0.1394,
+      "step": 8495
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.202847804945025e-07,
+      "loss": 0.1455,
+      "step": 8496
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.189635876805135e-07,
+      "loss": 0.1613,
+      "step": 8497
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.176434142845402e-07,
+      "loss": 0.178,
+      "step": 8498
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.16324260453164e-07,
+      "loss": 0.1681,
+      "step": 8499
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.150061263328601e-07,
+      "loss": 0.1565,
+      "step": 8500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.136890120699859e-07,
+      "loss": 0.1536,
+      "step": 8501
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.12372917810792e-07,
+      "loss": 0.1598,
+      "step": 8502
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.110578437014049e-07,
+      "loss": 0.1672,
+      "step": 8503
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.097437898878491e-07,
+      "loss": 0.1458,
+      "step": 8504
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.084307565160266e-07,
+      "loss": 0.2007,
+      "step": 8505
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.071187437317329e-07,
+      "loss": 0.1768,
+      "step": 8506
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.058077516806473e-07,
+      "loss": 0.1506,
+      "step": 8507
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.044977805083376e-07,
+      "loss": 0.1321,
+      "step": 8508
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.031888303602541e-07,
+      "loss": 0.1511,
+      "step": 8509
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.018809013817363e-07,
+      "loss": 0.1648,
+      "step": 8510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.005739937180113e-07,
+      "loss": 0.17,
+      "step": 8511
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.992681075141906e-07,
+      "loss": 0.1685,
+      "step": 8512
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.979632429152773e-07,
+      "loss": 0.1995,
+      "step": 8513
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.96659400066152e-07,
+      "loss": 0.187,
+      "step": 8514
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.953565791115924e-07,
+      "loss": 0.1489,
+      "step": 8515
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.940547801962506e-07,
+      "loss": 0.1449,
+      "step": 8516
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.927540034646808e-07,
+      "loss": 0.1758,
+      "step": 8517
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.914542490613075e-07,
+      "loss": 0.1902,
+      "step": 8518
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.901555171304543e-07,
+      "loss": 0.1184,
+      "step": 8519
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.888578078163211e-07,
+      "loss": 0.139,
+      "step": 8520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.875611212630052e-07,
+      "loss": 0.147,
+      "step": 8521
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.862654576144768e-07,
+      "loss": 0.1753,
+      "step": 8522
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.849708170146087e-07,
+      "loss": 0.176,
+      "step": 8523
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.836771996071457e-07,
+      "loss": 0.1554,
+      "step": 8524
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.823846055357287e-07,
+      "loss": 0.1471,
+      "step": 8525
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.810930349438761e-07,
+      "loss": 0.2014,
+      "step": 8526
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.798024879750021e-07,
+      "loss": 0.1443,
+      "step": 8527
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.78512964772401e-07,
+      "loss": 0.1389,
+      "step": 8528
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.77224465479256e-07,
+      "loss": 0.1434,
+      "step": 8529
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.759369902386349e-07,
+      "loss": 0.1115,
+      "step": 8530
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.746505391934955e-07,
+      "loss": 0.1364,
+      "step": 8531
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.733651124866736e-07,
+      "loss": 0.1687,
+      "step": 8532
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.720807102609007e-07,
+      "loss": 0.1608,
+      "step": 8533
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.707973326587903e-07,
+      "loss": 0.1921,
+      "step": 8534
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.69514979822844e-07,
+      "loss": 0.1643,
+      "step": 8535
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.682336518954447e-07,
+      "loss": 0.1821,
+      "step": 8536
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.669533490188674e-07,
+      "loss": 0.1561,
+      "step": 8537
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.656740713352683e-07,
+      "loss": 0.1628,
+      "step": 8538
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.64395818986694e-07,
+      "loss": 0.1599,
+      "step": 8539
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.631185921150741e-07,
+      "loss": 0.158,
+      "step": 8540
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.618423908622297e-07,
+      "loss": 0.1539,
+      "step": 8541
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.605672153698595e-07,
+      "loss": 0.1592,
+      "step": 8542
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.592930657795549e-07,
+      "loss": 0.1394,
+      "step": 8543
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.580199422327905e-07,
+      "loss": 0.1437,
+      "step": 8544
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.567478448709275e-07,
+      "loss": 0.1674,
+      "step": 8545
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.554767738352142e-07,
+      "loss": 0.1412,
+      "step": 8546
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.542067292667854e-07,
+      "loss": 0.1716,
+      "step": 8547
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.529377113066582e-07,
+      "loss": 0.1558,
+      "step": 8548
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.516697200957412e-07,
+      "loss": 0.1161,
+      "step": 8549
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.504027557748228e-07,
+      "loss": 0.1604,
+      "step": 8550
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.491368184845815e-07,
+      "loss": 0.1796,
+      "step": 8551
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.478719083655827e-07,
+      "loss": 0.1794,
+      "step": 8552
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.466080255582753e-07,
+      "loss": 0.1436,
+      "step": 8553
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.453451702029935e-07,
+      "loss": 0.1836,
+      "step": 8554
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.440833424399596e-07,
+      "loss": 0.1599,
+      "step": 8555
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.428225424092794e-07,
+      "loss": 0.1521,
+      "step": 8556
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.415627702509476e-07,
+      "loss": 0.14,
+      "step": 8557
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.403040261048433e-07,
+      "loss": 0.1814,
+      "step": 8558
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.390463101107326e-07,
+      "loss": 0.1553,
+      "step": 8559
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.377896224082626e-07,
+      "loss": 0.1578,
+      "step": 8560
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.365339631369728e-07,
+      "loss": 0.1685,
+      "step": 8561
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.352793324362828e-07,
+      "loss": 0.1283,
+      "step": 8562
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.340257304455045e-07,
+      "loss": 0.1262,
+      "step": 8563
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.327731573038288e-07,
+      "loss": 0.2383,
+      "step": 8564
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.315216131503377e-07,
+      "loss": 0.1638,
+      "step": 8565
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.302710981239946e-07,
+      "loss": 0.1658,
+      "step": 8566
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.290216123636518e-07,
+      "loss": 0.1807,
+      "step": 8567
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.277731560080436e-07,
+      "loss": 0.1909,
+      "step": 8568
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.265257291957973e-07,
+      "loss": 0.1885,
+      "step": 8569
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.252793320654173e-07,
+      "loss": 0.1853,
+      "step": 8570
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.240339647553007e-07,
+      "loss": 0.1746,
+      "step": 8571
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.227896274037238e-07,
+      "loss": 0.1289,
+      "step": 8572
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.215463201488526e-07,
+      "loss": 0.1809,
+      "step": 8573
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.203040431287389e-07,
+      "loss": 0.1418,
+      "step": 8574
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.190627964813212e-07,
+      "loss": 0.1598,
+      "step": 8575
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.178225803444183e-07,
+      "loss": 0.1372,
+      "step": 8576
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.165833948557366e-07,
+      "loss": 0.1511,
+      "step": 8577
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.153452401528727e-07,
+      "loss": 0.23,
+      "step": 8578
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.141081163733033e-07,
+      "loss": 0.1526,
+      "step": 8579
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.128720236543951e-07,
+      "loss": 0.1511,
+      "step": 8580
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.116369621333941e-07,
+      "loss": 0.1219,
+      "step": 8581
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.104029319474404e-07,
+      "loss": 0.1732,
+      "step": 8582
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.091699332335489e-07,
+      "loss": 0.175,
+      "step": 8583
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.079379661286301e-07,
+      "loss": 0.1753,
+      "step": 8584
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.067070307694745e-07,
+      "loss": 0.1572,
+      "step": 8585
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.054771272927607e-07,
+      "loss": 0.1849,
+      "step": 8586
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.042482558350472e-07,
+      "loss": 0.1841,
+      "step": 8587
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.030204165327869e-07,
+      "loss": 0.1775,
+      "step": 8588
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.0179360952231e-07,
+      "loss": 0.1478,
+      "step": 8589
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.005678349398348e-07,
+      "loss": 0.177,
+      "step": 8590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.993430929214673e-07,
+      "loss": 0.1321,
+      "step": 8591
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.981193836031974e-07,
+      "loss": 0.1489,
+      "step": 8592
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.968967071208976e-07,
+      "loss": 0.1444,
+      "step": 8593
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.956750636103305e-07,
+      "loss": 0.1643,
+      "step": 8594
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.944544532071384e-07,
+      "loss": 0.1348,
+      "step": 8595
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.932348760468543e-07,
+      "loss": 0.156,
+      "step": 8596
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.920163322648943e-07,
+      "loss": 0.1882,
+      "step": 8597
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.907988219965589e-07,
+      "loss": 0.151,
+      "step": 8598
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.895823453770345e-07,
+      "loss": 0.1312,
+      "step": 8599
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.883669025413942e-07,
+      "loss": 0.1553,
+      "step": 8600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.87152493624591e-07,
+      "loss": 0.1846,
+      "step": 8601
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.859391187614705e-07,
+      "loss": 0.1711,
+      "step": 8602
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.847267780867595e-07,
+      "loss": 0.1373,
+      "step": 8603
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.835154717350712e-07,
+      "loss": 0.1899,
+      "step": 8604
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.823051998409002e-07,
+      "loss": 0.1372,
+      "step": 8605
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.810959625386338e-07,
+      "loss": 0.1593,
+      "step": 8606
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.798877599625342e-07,
+      "loss": 0.1322,
+      "step": 8607
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.786805922467588e-07,
+      "loss": 0.1271,
+      "step": 8608
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.774744595253436e-07,
+      "loss": 0.1584,
+      "step": 8609
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.762693619322147e-07,
+      "loss": 0.161,
+      "step": 8610
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.750652996011753e-07,
+      "loss": 0.176,
+      "step": 8611
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.73862272665925e-07,
+      "loss": 0.1851,
+      "step": 8612
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.726602812600358e-07,
+      "loss": 0.1445,
+      "step": 8613
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.714593255169732e-07,
+      "loss": 0.1309,
+      "step": 8614
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.702594055700872e-07,
+      "loss": 0.1727,
+      "step": 8615
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.690605215526114e-07,
+      "loss": 0.1492,
+      "step": 8616
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.678626735976613e-07,
+      "loss": 0.1591,
+      "step": 8617
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.666658618382438e-07,
+      "loss": 0.199,
+      "step": 8618
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.654700864072406e-07,
+      "loss": 0.177,
+      "step": 8619
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.642753474374331e-07,
+      "loss": 0.1306,
+      "step": 8620
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.630816450614741e-07,
+      "loss": 0.1548,
+      "step": 8621
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.618889794119087e-07,
+      "loss": 0.1609,
+      "step": 8622
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.606973506211634e-07,
+      "loss": 0.1636,
+      "step": 8623
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.595067588215509e-07,
+      "loss": 0.1836,
+      "step": 8624
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.58317204145269e-07,
+      "loss": 0.1899,
+      "step": 8625
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.571286867244031e-07,
+      "loss": 0.2019,
+      "step": 8626
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.559412066909155e-07,
+      "loss": 0.1177,
+      "step": 8627
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.547547641766616e-07,
+      "loss": 0.175,
+      "step": 8628
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.535693593133762e-07,
+      "loss": 0.1724,
+      "step": 8629
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.523849922326819e-07,
+      "loss": 0.1641,
+      "step": 8630
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.512016630660855e-07,
+      "loss": 0.1384,
+      "step": 8631
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.500193719449787e-07,
+      "loss": 0.1519,
+      "step": 8632
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.488381190006354e-07,
+      "loss": 0.1836,
+      "step": 8633
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.476579043642194e-07,
+      "loss": 0.1819,
+      "step": 8634
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.464787281667717e-07,
+      "loss": 0.1904,
+      "step": 8635
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.453005905392251e-07,
+      "loss": 0.1621,
+      "step": 8636
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.44123491612394e-07,
+      "loss": 0.1777,
+      "step": 8637
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.429474315169793e-07,
+      "loss": 0.1848,
+      "step": 8638
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.41772410383561e-07,
+      "loss": 0.1865,
+      "step": 8639
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.405984283426125e-07,
+      "loss": 0.1196,
+      "step": 8640
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.39425485524483e-07,
+      "loss": 0.228,
+      "step": 8641
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.382535820594116e-07,
+      "loss": 0.1824,
+      "step": 8642
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.37082718077524e-07,
+      "loss": 0.1616,
+      "step": 8643
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.359128937088222e-07,
+      "loss": 0.1779,
+      "step": 8644
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.347441090832029e-07,
+      "loss": 0.1884,
+      "step": 8645
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.335763643304372e-07,
+      "loss": 0.1422,
+      "step": 8646
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.324096595801887e-07,
+      "loss": 0.1443,
+      "step": 8647
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.312439949620031e-07,
+      "loss": 0.2008,
+      "step": 8648
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.30079370605311e-07,
+      "loss": 0.1456,
+      "step": 8649
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.289157866394224e-07,
+      "loss": 0.1575,
+      "step": 8650
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.277532431935418e-07,
+      "loss": 0.1748,
+      "step": 8651
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.26591740396747e-07,
+      "loss": 0.1526,
+      "step": 8652
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.254312783780081e-07,
+      "loss": 0.1292,
+      "step": 8653
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.242718572661766e-07,
+      "loss": 0.1704,
+      "step": 8654
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.231134771899928e-07,
+      "loss": 0.125,
+      "step": 8655
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.219561382780715e-07,
+      "loss": 0.1541,
+      "step": 8656
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.207998406589233e-07,
+      "loss": 0.1456,
+      "step": 8657
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.196445844609333e-07,
+      "loss": 0.1514,
+      "step": 8658
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.18490369812379e-07,
+      "loss": 0.1342,
+      "step": 8659
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.173371968414165e-07,
+      "loss": 0.173,
+      "step": 8660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.161850656760915e-07,
+      "loss": 0.1532,
+      "step": 8661
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.15033976444328e-07,
+      "loss": 0.1493,
+      "step": 8662
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.138839292739407e-07,
+      "loss": 0.1798,
+      "step": 8663
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.127349242926217e-07,
+      "loss": 0.1639,
+      "step": 8664
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.115869616279523e-07,
+      "loss": 0.1948,
+      "step": 8665
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.104400414073974e-07,
+      "loss": 0.1823,
+      "step": 8666
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.092941637583072e-07,
+      "loss": 0.1528,
+      "step": 8667
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.081493288079099e-07,
+      "loss": 0.1443,
+      "step": 8668
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.070055366833271e-07,
+      "loss": 0.1394,
+      "step": 8669
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.058627875115541e-07,
+      "loss": 0.2007,
+      "step": 8670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.047210814194837e-07,
+      "loss": 0.1267,
+      "step": 8671
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.03580418533879e-07,
+      "loss": 0.1522,
+      "step": 8672
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.024407989813996e-07,
+      "loss": 0.1232,
+      "step": 8673
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.013022228885767e-07,
+      "loss": 0.1638,
+      "step": 8674
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.00164690381837e-07,
+      "loss": 0.1487,
+      "step": 8675
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.99028201587486e-07,
+      "loss": 0.1245,
+      "step": 8676
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.978927566317139e-07,
+      "loss": 0.1799,
+      "step": 8677
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.967583556405932e-07,
+      "loss": 0.1591,
+      "step": 8678
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.956249987400841e-07,
+      "loss": 0.1337,
+      "step": 8679
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.944926860560285e-07,
+      "loss": 0.1853,
+      "step": 8680
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.933614177141522e-07,
+      "loss": 0.1213,
+      "step": 8681
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.922311938400671e-07,
+      "loss": 0.1555,
+      "step": 8682
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.911020145592683e-07,
+      "loss": 0.1475,
+      "step": 8683
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.899738799971321e-07,
+      "loss": 0.1368,
+      "step": 8684
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.888467902789241e-07,
+      "loss": 0.1702,
+      "step": 8685
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.877207455297873e-07,
+      "loss": 0.1562,
+      "step": 8686
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.865957458747551e-07,
+      "loss": 0.1714,
+      "step": 8687
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.854717914387398e-07,
+      "loss": 0.1409,
+      "step": 8688
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.843488823465438e-07,
+      "loss": 0.1619,
+      "step": 8689
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.83227018722845e-07,
+      "loss": 0.1658,
+      "step": 8690
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.82106200692213e-07,
+      "loss": 0.1571,
+      "step": 8691
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.809864283790956e-07,
+      "loss": 0.135,
+      "step": 8692
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.798677019078281e-07,
+      "loss": 0.1447,
+      "step": 8693
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.787500214026287e-07,
+      "loss": 0.145,
+      "step": 8694
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.776333869875994e-07,
+      "loss": 0.1801,
+      "step": 8695
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.765177987867255e-07,
+      "loss": 0.1345,
+      "step": 8696
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.754032569238765e-07,
+      "loss": 0.1473,
+      "step": 8697
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.742897615228049e-07,
+      "loss": 0.1946,
+      "step": 8698
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.731773127071483e-07,
+      "loss": 0.1517,
+      "step": 8699
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.720659106004289e-07,
+      "loss": 0.1575,
+      "step": 8700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.709555553260515e-07,
+      "loss": 0.1547,
+      "step": 8701
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.69846247007303e-07,
+      "loss": 0.1271,
+      "step": 8702
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.68737985767357e-07,
+      "loss": 0.1396,
+      "step": 8703
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.676307717292673e-07,
+      "loss": 0.1516,
+      "step": 8704
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.665246050159756e-07,
+      "loss": 0.1221,
+      "step": 8705
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.654194857503048e-07,
+      "loss": 0.1599,
+      "step": 8706
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.643154140549634e-07,
+      "loss": 0.1373,
+      "step": 8707
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.632123900525388e-07,
+      "loss": 0.1655,
+      "step": 8708
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.621104138655098e-07,
+      "loss": 0.1477,
+      "step": 8709
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.610094856162307e-07,
+      "loss": 0.1646,
+      "step": 8710
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.599096054269449e-07,
+      "loss": 0.1542,
+      "step": 8711
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.58810773419779e-07,
+      "loss": 0.1528,
+      "step": 8712
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.577129897167388e-07,
+      "loss": 0.1526,
+      "step": 8713
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.566162544397214e-07,
+      "loss": 0.1555,
+      "step": 8714
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.55520567710498e-07,
+      "loss": 0.1218,
+      "step": 8715
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.544259296507314e-07,
+      "loss": 0.1449,
+      "step": 8716
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.533323403819635e-07,
+      "loss": 0.1713,
+      "step": 8717
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.522398000256246e-07,
+      "loss": 0.168,
+      "step": 8718
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.511483087030201e-07,
+      "loss": 0.176,
+      "step": 8719
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.500578665353484e-07,
+      "loss": 0.1508,
+      "step": 8720
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.489684736436818e-07,
+      "loss": 0.1582,
+      "step": 8721
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.478801301489866e-07,
+      "loss": 0.1711,
+      "step": 8722
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.467928361721042e-07,
+      "loss": 0.166,
+      "step": 8723
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.457065918337645e-07,
+      "loss": 0.1853,
+      "step": 8724
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.446213972545767e-07,
+      "loss": 0.1414,
+      "step": 8725
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.435372525550375e-07,
+      "loss": 0.1692,
+      "step": 8726
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.424541578555209e-07,
+      "loss": 0.1658,
+      "step": 8727
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.41372113276295e-07,
+      "loss": 0.141,
+      "step": 8728
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.402911189374994e-07,
+      "loss": 0.1235,
+      "step": 8729
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.392111749591666e-07,
+      "loss": 0.1594,
+      "step": 8730
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.381322814612044e-07,
+      "loss": 0.1316,
+      "step": 8731
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.370544385634102e-07,
+      "loss": 0.1597,
+      "step": 8732
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.359776463854616e-07,
+      "loss": 0.1887,
+      "step": 8733
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.349019050469229e-07,
+      "loss": 0.1743,
+      "step": 8734
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.338272146672352e-07,
+      "loss": 0.1268,
+      "step": 8735
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.327535753657309e-07,
+      "loss": 0.1793,
+      "step": 8736
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.316809872616191e-07,
+      "loss": 0.1836,
+      "step": 8737
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.306094504739945e-07,
+      "loss": 0.1736,
+      "step": 8738
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.295389651218374e-07,
+      "loss": 0.1727,
+      "step": 8739
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.284695313240096e-07,
+      "loss": 0.1368,
+      "step": 8740
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.274011491992526e-07,
+      "loss": 0.1558,
+      "step": 8741
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.263338188661981e-07,
+      "loss": 0.1111,
+      "step": 8742
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.252675404433538e-07,
+      "loss": 0.1432,
+      "step": 8743
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.24202314049117e-07,
+      "loss": 0.1285,
+      "step": 8744
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.231381398017631e-07,
+      "loss": 0.1656,
+      "step": 8745
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.220750178194567e-07,
+      "loss": 0.1073,
+      "step": 8746
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.210129482202364e-07,
+      "loss": 0.1758,
+      "step": 8747
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.199519311220347e-07,
+      "loss": 0.17,
+      "step": 8748
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.188919666426573e-07,
+      "loss": 0.1707,
+      "step": 8749
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.17833054899799e-07,
+      "loss": 0.1603,
+      "step": 8750
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.167751960110367e-07,
+      "loss": 0.1364,
+      "step": 8751
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.157183900938311e-07,
+      "loss": 0.1526,
+      "step": 8752
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.146626372655228e-07,
+      "loss": 0.1522,
+      "step": 8753
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.136079376433389e-07,
+      "loss": 0.1467,
+      "step": 8754
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.12554291344386e-07,
+      "loss": 0.1602,
+      "step": 8755
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.11501698485658e-07,
+      "loss": 0.1626,
+      "step": 8756
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.104501591840294e-07,
+      "loss": 0.1377,
+      "step": 8757
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.09399673556259e-07,
+      "loss": 0.1406,
+      "step": 8758
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.083502417189845e-07,
+      "loss": 0.1448,
+      "step": 8759
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.073018637887339e-07,
+      "loss": 0.1312,
+      "step": 8760
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.062545398819108e-07,
+      "loss": 0.1816,
+      "step": 8761
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.052082701148053e-07,
+      "loss": 0.1351,
+      "step": 8762
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.041630546035913e-07,
+      "loss": 0.2002,
+      "step": 8763
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.03118893464325e-07,
+      "loss": 0.1332,
+      "step": 8764
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.020757868129433e-07,
+      "loss": 0.1638,
+      "step": 8765
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.010337347652694e-07,
+      "loss": 0.1627,
+      "step": 8766
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.99992737437005e-07,
+      "loss": 0.1938,
+      "step": 8767
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.989527949437411e-07,
+      "loss": 0.1675,
+      "step": 8768
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.979139074009453e-07,
+      "loss": 0.1716,
+      "step": 8769
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.968760749239732e-07,
+      "loss": 0.1198,
+      "step": 8770
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.958392976280568e-07,
+      "loss": 0.1656,
+      "step": 8771
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.948035756283198e-07,
+      "loss": 0.1549,
+      "step": 8772
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.937689090397568e-07,
+      "loss": 0.1672,
+      "step": 8773
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.927352979772603e-07,
+      "loss": 0.1414,
+      "step": 8774
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.917027425555909e-07,
+      "loss": 0.1743,
+      "step": 8775
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.906712428894033e-07,
+      "loss": 0.1577,
+      "step": 8776
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.896407990932272e-07,
+      "loss": 0.1897,
+      "step": 8777
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.886114112814789e-07,
+      "loss": 0.171,
+      "step": 8778
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.875830795684589e-07,
+      "loss": 0.1694,
+      "step": 8779
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.865558040683438e-07,
+      "loss": 0.1364,
+      "step": 8780
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.855295848952013e-07,
+      "loss": 0.1565,
+      "step": 8781
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.845044221629757e-07,
+      "loss": 0.1121,
+      "step": 8782
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.834803159854961e-07,
+      "loss": 0.176,
+      "step": 8783
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.824572664764748e-07,
+      "loss": 0.1912,
+      "step": 8784
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.814352737495076e-07,
+      "loss": 0.166,
+      "step": 8785
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.804143379180693e-07,
+      "loss": 0.1559,
+      "step": 8786
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.793944590955224e-07,
+      "loss": 0.1619,
+      "step": 8787
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.783756373951054e-07,
+      "loss": 0.1863,
+      "step": 8788
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.773578729299466e-07,
+      "loss": 0.1794,
+      "step": 8789
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7634116581305235e-07,
+      "loss": 0.1426,
+      "step": 8790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.753255161573145e-07,
+      "loss": 0.1523,
+      "step": 8791
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7431092407550397e-07,
+      "loss": 0.1641,
+      "step": 8792
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.732973896802784e-07,
+      "loss": 0.1582,
+      "step": 8793
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7228491308417223e-07,
+      "loss": 0.1709,
+      "step": 8794
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.712734943996078e-07,
+      "loss": 0.1437,
+      "step": 8795
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7026313373888856e-07,
+      "loss": 0.1826,
+      "step": 8796
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.692538312142014e-07,
+      "loss": 0.1261,
+      "step": 8797
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.682455869376123e-07,
+      "loss": 0.1458,
+      "step": 8798
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6723840102107287e-07,
+      "loss": 0.1521,
+      "step": 8799
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6623227357641466e-07,
+      "loss": 0.1527,
+      "step": 8800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6522720471535385e-07,
+      "loss": 0.1818,
+      "step": 8801
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6422319454948907e-07,
+      "loss": 0.1625,
+      "step": 8802
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.632202431903021e-07,
+      "loss": 0.1533,
+      "step": 8803
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.622183507491529e-07,
+      "loss": 0.166,
+      "step": 8804
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6121751733729015e-07,
+      "loss": 0.1566,
+      "step": 8805
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.602177430658372e-07,
+      "loss": 0.1155,
+      "step": 8806
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.592190280458075e-07,
+      "loss": 0.1553,
+      "step": 8807
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5822137238809126e-07,
+      "loss": 0.1327,
+      "step": 8808
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.572247762034676e-07,
+      "loss": 0.1401,
+      "step": 8809
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5622923960258913e-07,
+      "loss": 0.1418,
+      "step": 8810
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5523476269599744e-07,
+      "loss": 0.1665,
+      "step": 8811
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5424134559411413e-07,
+      "loss": 0.1462,
+      "step": 8812
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5324898840724327e-07,
+      "loss": 0.1628,
+      "step": 8813
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.522576912455712e-07,
+      "loss": 0.1715,
+      "step": 8814
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5126745421916863e-07,
+      "loss": 0.1995,
+      "step": 8815
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5027827743798435e-07,
+      "loss": 0.1471,
+      "step": 8816
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4929016101185277e-07,
+      "loss": 0.121,
+      "step": 8817
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4830310505048824e-07,
+      "loss": 0.1635,
+      "step": 8818
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.473171096634921e-07,
+      "loss": 0.1782,
+      "step": 8819
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4633217496034107e-07,
+      "loss": 0.1731,
+      "step": 8820
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.453483010504001e-07,
+      "loss": 0.1232,
+      "step": 8821
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.443654880429116e-07,
+      "loss": 0.1622,
+      "step": 8822
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.43383736047005e-07,
+      "loss": 0.1368,
+      "step": 8823
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.424030451716843e-07,
+      "loss": 0.1263,
+      "step": 8824
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.414234155258468e-07,
+      "loss": 0.1404,
+      "step": 8825
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4044484721826207e-07,
+      "loss": 0.1577,
+      "step": 8826
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3946734035758887e-07,
+      "loss": 0.1528,
+      "step": 8827
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.384908950523603e-07,
+      "loss": 0.1584,
+      "step": 8828
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.375155114109986e-07,
+      "loss": 0.1624,
+      "step": 8829
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.365411895418048e-07,
+      "loss": 0.1536,
+      "step": 8830
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.355679295529658e-07,
+      "loss": 0.2261,
+      "step": 8831
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3459573155254396e-07,
+      "loss": 0.1665,
+      "step": 8832
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.336245956484908e-07,
+      "loss": 0.1442,
+      "step": 8833
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.326545219486333e-07,
+      "loss": 0.1868,
+      "step": 8834
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.316855105606854e-07,
+      "loss": 0.1746,
+      "step": 8835
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3071756159224096e-07,
+      "loss": 0.1403,
+      "step": 8836
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.297506751507785e-07,
+      "loss": 0.1921,
+      "step": 8837
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.287848513436521e-07,
+      "loss": 0.1401,
+      "step": 8838
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.278200902781071e-07,
+      "loss": 0.1711,
+      "step": 8839
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.268563920612623e-07,
+      "loss": 0.1598,
+      "step": 8840
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2589375680012313e-07,
+      "loss": 0.1423,
+      "step": 8841
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2493218460157637e-07,
+      "loss": 0.1592,
+      "step": 8842
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2397167557239214e-07,
+      "loss": 0.155,
+      "step": 8843
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.230122298192163e-07,
+      "loss": 0.1775,
+      "step": 8844
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.220538474485858e-07,
+      "loss": 0.1729,
+      "step": 8845
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2109652856691e-07,
+      "loss": 0.1599,
+      "step": 8846
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.201402732804882e-07,
+      "loss": 0.1548,
+      "step": 8847
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.191850816955001e-07,
+      "loss": 0.1633,
+      "step": 8848
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.182309539180007e-07,
+      "loss": 0.1692,
+      "step": 8849
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1727789005393536e-07,
+      "loss": 0.1143,
+      "step": 8850
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1632589020912607e-07,
+      "loss": 0.1584,
+      "step": 8851
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.153749544892782e-07,
+      "loss": 0.186,
+      "step": 8852
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.144250829999796e-07,
+      "loss": 0.1586,
+      "step": 8853
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1347627584670145e-07,
+      "loss": 0.1729,
+      "step": 8854
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1252853313479055e-07,
+      "loss": 0.1497,
+      "step": 8855
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.115818549694839e-07,
+      "loss": 0.1361,
+      "step": 8856
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1063624145589177e-07,
+      "loss": 0.1591,
+      "step": 8857
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.096916926990135e-07,
+      "loss": 0.1301,
+      "step": 8858
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0874820880372644e-07,
+      "loss": 0.1516,
+      "step": 8859
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0780578987479225e-07,
+      "loss": 0.125,
+      "step": 8860
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.068644360168494e-07,
+      "loss": 0.1432,
+      "step": 8861
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.059241473344244e-07,
+      "loss": 0.1785,
+      "step": 8862
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0498492393191924e-07,
+      "loss": 0.1566,
+      "step": 8863
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.040467659136227e-07,
+      "loss": 0.1619,
+      "step": 8864
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0310967338370253e-07,
+      "loss": 0.1552,
+      "step": 8865
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0217364644621e-07,
+      "loss": 0.1421,
+      "step": 8866
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.012386852050765e-07,
+      "loss": 0.1865,
+      "step": 8867
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.003047897641155e-07,
+      "loss": 0.1373,
+      "step": 8868
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9937196022702185e-07,
+      "loss": 0.1873,
+      "step": 8869
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.984401966973717e-07,
+      "loss": 0.1711,
+      "step": 8870
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9750949927862545e-07,
+      "loss": 0.1467,
+      "step": 8871
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.96579868074124e-07,
+      "loss": 0.2046,
+      "step": 8872
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.956513031870868e-07,
+      "loss": 0.1354,
+      "step": 8873
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.947238047206181e-07,
+      "loss": 0.1211,
+      "step": 8874
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.937973727777011e-07,
+      "loss": 0.1643,
+      "step": 8875
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.928720074612069e-07,
+      "loss": 0.1774,
+      "step": 8876
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9194770887387877e-07,
+      "loss": 0.1582,
+      "step": 8877
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9102447711835025e-07,
+      "loss": 0.1248,
+      "step": 8878
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9010231229712926e-07,
+      "loss": 0.1851,
+      "step": 8879
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8918121451260947e-07,
+      "loss": 0.1656,
+      "step": 8880
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.882611838670658e-07,
+      "loss": 0.1268,
+      "step": 8881
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8734222046265536e-07,
+      "loss": 0.1782,
+      "step": 8882
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.864243244014121e-07,
+      "loss": 0.1343,
+      "step": 8883
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.855074957852578e-07,
+      "loss": 0.1696,
+      "step": 8884
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.845917347159911e-07,
+      "loss": 0.1779,
+      "step": 8885
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.836770412952928e-07,
+      "loss": 0.1858,
+      "step": 8886
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8276341562472843e-07,
+      "loss": 0.1553,
+      "step": 8887
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.818508578057412e-07,
+      "loss": 0.1469,
+      "step": 8888
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.809393679396567e-07,
+      "loss": 0.1956,
+      "step": 8889
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8002894612768517e-07,
+      "loss": 0.1597,
+      "step": 8890
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.791195924709112e-07,
+      "loss": 0.1636,
+      "step": 8891
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.782113070703075e-07,
+      "loss": 0.1627,
+      "step": 8892
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.773040900267255e-07,
+      "loss": 0.1648,
+      "step": 8893
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.76397941440898e-07,
+      "loss": 0.1736,
+      "step": 8894
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.754928614134401e-07,
+      "loss": 0.1951,
+      "step": 8895
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7458885004484693e-07,
+      "loss": 0.1909,
+      "step": 8896
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.736859074354937e-07,
+      "loss": 0.1671,
+      "step": 8897
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.727840336856414e-07,
+      "loss": 0.1697,
+      "step": 8898
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7188322889542884e-07,
+      "loss": 0.209,
+      "step": 8899
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7098349316487816e-07,
+      "loss": 0.2131,
+      "step": 8900
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7008482659389056e-07,
+      "loss": 0.157,
+      "step": 8901
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.691872292822496e-07,
+      "loss": 0.1772,
+      "step": 8902
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.682907013296189e-07,
+      "loss": 0.1265,
+      "step": 8903
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.673952428355465e-07,
+      "loss": 0.1626,
+      "step": 8904
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6650085389945966e-07,
+      "loss": 0.1549,
+      "step": 8905
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.656075346206667e-07,
+      "loss": 0.1616,
+      "step": 8906
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.647152850983571e-07,
+      "loss": 0.1619,
+      "step": 8907
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.638241054316027e-07,
+      "loss": 0.1514,
+      "step": 8908
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.629339957193556e-07,
+      "loss": 0.1777,
+      "step": 8909
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.620449560604478e-07,
+      "loss": 0.1592,
+      "step": 8910
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6115698655359465e-07,
+      "loss": 0.1501,
+      "step": 8911
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.602700872973952e-07,
+      "loss": 0.1931,
+      "step": 8912
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.5938425839032286e-07,
+      "loss": 0.141,
+      "step": 8913
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5849949993073565e-07,
+      "loss": 0.1567,
+      "step": 8914
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.576158120168749e-07,
+      "loss": 0.1334,
+      "step": 8915
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5673319474685885e-07,
+      "loss": 0.1534,
+      "step": 8916
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5585164821869247e-07,
+      "loss": 0.1139,
+      "step": 8917
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5497117253025514e-07,
+      "loss": 0.1599,
+      "step": 8918
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.540917677793132e-07,
+      "loss": 0.155,
+      "step": 8919
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.532134340635085e-07,
+      "loss": 0.171,
+      "step": 8920
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.523361714803686e-07,
+      "loss": 0.1555,
+      "step": 8921
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.514599801273e-07,
+      "loss": 0.1544,
+      "step": 8922
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5058486010159266e-07,
+      "loss": 0.1343,
+      "step": 8923
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.497108115004144e-07,
+      "loss": 0.1797,
+      "step": 8924
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.488378344208154e-07,
+      "loss": 0.1509,
+      "step": 8925
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.479659289597248e-07,
+      "loss": 0.1819,
+      "step": 8926
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4709509521395843e-07,
+      "loss": 0.1658,
+      "step": 8927
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.462253332802068e-07,
+      "loss": 0.1777,
+      "step": 8928
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.453566432550459e-07,
+      "loss": 0.1476,
+      "step": 8929
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4448902523492977e-07,
+      "loss": 0.173,
+      "step": 8930
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4362247931619464e-07,
+      "loss": 0.1564,
+      "step": 8931
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4275700559505687e-07,
+      "loss": 0.1927,
+      "step": 8932
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4189260416761625e-07,
+      "loss": 0.1312,
+      "step": 8933
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.410292751298505e-07,
+      "loss": 0.1511,
+      "step": 8934
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.401670185776207e-07,
+      "loss": 0.1528,
+      "step": 8935
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3930583460666576e-07,
+      "loss": 0.2151,
+      "step": 8936
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3844572331260816e-07,
+      "loss": 0.1765,
+      "step": 8937
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3758668479095146e-07,
+      "loss": 0.144,
+      "step": 8938
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.367287191370794e-07,
+      "loss": 0.1608,
+      "step": 8939
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.358718264462535e-07,
+      "loss": 0.1277,
+      "step": 8940
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.350160068136221e-07,
+      "loss": 0.1421,
+      "step": 8941
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.341612603342104e-07,
+      "loss": 0.1589,
+      "step": 8942
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3330758710292344e-07,
+      "loss": 0.1438,
+      "step": 8943
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3245498721455104e-07,
+      "loss": 0.1548,
+      "step": 8944
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3160346076376303e-07,
+      "loss": 0.1685,
+      "step": 8945
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.307530078451049e-07,
+      "loss": 0.1519,
+      "step": 8946
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2990362855301107e-07,
+      "loss": 0.1638,
+      "step": 8947
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.290553229817894e-07,
+      "loss": 0.2017,
+      "step": 8948
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.282080912256325e-07,
+      "loss": 0.2319,
+      "step": 8949
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.273619333786127e-07,
+      "loss": 0.1555,
+      "step": 8950
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.265168495346849e-07,
+      "loss": 0.1239,
+      "step": 8951
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.256728397876807e-07,
+      "loss": 0.1385,
+      "step": 8952
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2482990423131745e-07,
+      "loss": 0.165,
+      "step": 8953
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.239880429591891e-07,
+      "loss": 0.1721,
+      "step": 8954
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2314725606477084e-07,
+      "loss": 0.1833,
+      "step": 8955
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.223075436414214e-07,
+      "loss": 0.146,
+      "step": 8956
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.214689057823783e-07,
+      "loss": 0.2166,
+      "step": 8957
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2063134258075944e-07,
+      "loss": 0.1636,
+      "step": 8958
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.197948541295637e-07,
+      "loss": 0.1368,
+      "step": 8959
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1895944052167004e-07,
+      "loss": 0.1921,
+      "step": 8960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1812510184983993e-07,
+      "loss": 0.136,
+      "step": 8961
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1729183820671363e-07,
+      "loss": 0.1656,
+      "step": 8962
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.164596496848138e-07,
+      "loss": 0.1501,
+      "step": 8963
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.15628536376541e-07,
+      "loss": 0.1597,
+      "step": 8964
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.147984983741792e-07,
+      "loss": 0.146,
+      "step": 8965
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1396953576989133e-07,
+      "loss": 0.1188,
+      "step": 8966
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.131416486557215e-07,
+      "loss": 0.1465,
+      "step": 8967
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.123148371235929e-07,
+      "loss": 0.1411,
+      "step": 8968
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.114891012653143e-07,
+      "loss": 0.1305,
+      "step": 8969
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.10664441172569e-07,
+      "loss": 0.1653,
+      "step": 8970
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.098408569369238e-07,
+      "loss": 0.1571,
+      "step": 8971
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0901834864982217e-07,
+      "loss": 0.1802,
+      "step": 8972
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0819691640259773e-07,
+      "loss": 0.1162,
+      "step": 8973
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.073765602864542e-07,
+      "loss": 0.1455,
+      "step": 8974
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.06557280392481e-07,
+      "loss": 0.1948,
+      "step": 8975
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.057390768116475e-07,
+      "loss": 0.1335,
+      "step": 8976
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0492194963480215e-07,
+      "loss": 0.1929,
+      "step": 8977
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.041058989526735e-07,
+      "loss": 0.1615,
+      "step": 8978
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0329092485587573e-07,
+      "loss": 0.1851,
+      "step": 8979
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.024770274348976e-07,
+      "loss": 0.1606,
+      "step": 8980
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.01664206780109e-07,
+      "loss": 0.1357,
+      "step": 8981
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.008524629817644e-07,
+      "loss": 0.1342,
+      "step": 8982
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0004179612999175e-07,
+      "loss": 0.1663,
+      "step": 8983
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.99232206314809e-07,
+      "loss": 0.155,
+      "step": 8984
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9842369362610425e-07,
+      "loss": 0.1482,
+      "step": 8985
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.976162581536546e-07,
+      "loss": 0.1699,
+      "step": 8986
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9680989998711163e-07,
+      "loss": 0.1736,
+      "step": 8987
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.960046192160082e-07,
+      "loss": 0.1743,
+      "step": 8988
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9520041592976167e-07,
+      "loss": 0.1711,
+      "step": 8989
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9439729021766725e-07,
+      "loss": 0.1445,
+      "step": 8990
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9359524216889703e-07,
+      "loss": 0.1865,
+      "step": 8991
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9279427187250963e-07,
+      "loss": 0.1678,
+      "step": 8992
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.919943794174374e-07,
+      "loss": 0.1498,
+      "step": 8993
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.911955648925002e-07,
+      "loss": 0.1653,
+      "step": 8994
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.903978283863917e-07,
+      "loss": 0.1686,
+      "step": 8995
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8960116998769103e-07,
+      "loss": 0.1429,
+      "step": 8996
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8880558978485405e-07,
+      "loss": 0.1721,
+      "step": 8997
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.88011087866219e-07,
+      "loss": 0.1381,
+      "step": 8998
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8721766432000195e-07,
+      "loss": 0.1484,
+      "step": 8999
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.864253192343014e-07,
+      "loss": 0.1882,
+      "step": 9000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8563405269709576e-07,
+      "loss": 0.1636,
+      "step": 9001
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8484386479624595e-07,
+      "loss": 0.1892,
+      "step": 9002
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.840547556194861e-07,
+      "loss": 0.14,
+      "step": 9003
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8326672525443853e-07,
+      "loss": 0.1429,
+      "step": 9004
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.824797737885998e-07,
+      "loss": 0.1473,
+      "step": 9005
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8169390130935114e-07,
+      "loss": 0.2126,
+      "step": 9006
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.809091079039505e-07,
+      "loss": 0.1355,
+      "step": 9007
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8012539365953937e-07,
+      "loss": 0.1595,
+      "step": 9008
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7934275866313586e-07,
+      "loss": 0.1509,
+      "step": 9009
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.785612030016427e-07,
+      "loss": 0.1371,
+      "step": 9010
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7778072676183597e-07,
+      "loss": 0.1426,
+      "step": 9011
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7700133003037864e-07,
+      "loss": 0.1545,
+      "step": 9012
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.762230128938115e-07,
+      "loss": 0.1595,
+      "step": 9013
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7544577543855424e-07,
+      "loss": 0.1736,
+      "step": 9014
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.746696177509067e-07,
+      "loss": 0.1824,
+      "step": 9015
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7389453991705226e-07,
+      "loss": 0.1755,
+      "step": 9016
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7312054202304873e-07,
+      "loss": 0.1572,
+      "step": 9017
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.723476241548384e-07,
+      "loss": 0.1646,
+      "step": 9018
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7157578639824267e-07,
+      "loss": 0.1234,
+      "step": 9019
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.70805028838963e-07,
+      "loss": 0.2126,
+      "step": 9020
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.700353515625798e-07,
+      "loss": 0.1509,
+      "step": 9021
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.692667546545558e-07,
+      "loss": 0.1738,
+      "step": 9022
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6849923820022826e-07,
+      "loss": 0.1292,
+      "step": 9023
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.677328022848236e-07,
+      "loss": 0.1616,
+      "step": 9024
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.669674469934402e-07,
+      "loss": 0.1794,
+      "step": 9025
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6620317241106144e-07,
+      "loss": 0.1473,
+      "step": 9026
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6543997862254587e-07,
+      "loss": 0.1824,
+      "step": 9027
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.646778657126381e-07,
+      "loss": 0.1379,
+      "step": 9028
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.639168337659548e-07,
+      "loss": 0.177,
+      "step": 9029
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.631568828670028e-07,
+      "loss": 0.1692,
+      "step": 9030
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6239801310016023e-07,
+      "loss": 0.1427,
+      "step": 9031
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.616402245496896e-07,
+      "loss": 0.1628,
+      "step": 9032
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.608835172997304e-07,
+      "loss": 0.1316,
+      "step": 9033
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.601278914343042e-07,
+      "loss": 0.1753,
+      "step": 9034
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5937334703731386e-07,
+      "loss": 0.1642,
+      "step": 9035
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5861988419253914e-07,
+      "loss": 0.1433,
+      "step": 9036
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5786750298363973e-07,
+      "loss": 0.1968,
+      "step": 9037
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5711620349415876e-07,
+      "loss": 0.1499,
+      "step": 9038
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5636598580751516e-07,
+      "loss": 0.1946,
+      "step": 9039
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5561685000700996e-07,
+      "loss": 0.1652,
+      "step": 9040
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5486879617582337e-07,
+      "loss": 0.197,
+      "step": 9041
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5412182439701784e-07,
+      "loss": 0.1934,
+      "step": 9042
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.533759347535303e-07,
+      "loss": 0.1494,
+      "step": 9043
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.526311273281823e-07,
+      "loss": 0.177,
+      "step": 9044
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5188740220367327e-07,
+      "loss": 0.1523,
+      "step": 9045
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5114475946258265e-07,
+      "loss": 0.1677,
+      "step": 9046
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5040319918737123e-07,
+      "loss": 0.1909,
+      "step": 9047
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.496627214603764e-07,
+      "loss": 0.1589,
+      "step": 9048
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.489233263638202e-07,
+      "loss": 0.1631,
+      "step": 9049
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.48185013979797e-07,
+      "loss": 0.1445,
+      "step": 9050
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.474477843902889e-07,
+      "loss": 0.1444,
+      "step": 9051
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.467116376771528e-07,
+      "loss": 0.1741,
+      "step": 9052
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.459765739221298e-07,
+      "loss": 0.1465,
+      "step": 9053
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.452425932068325e-07,
+      "loss": 0.1887,
+      "step": 9054
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.445096956127646e-07,
+      "loss": 0.176,
+      "step": 9055
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4377788122129765e-07,
+      "loss": 0.1743,
+      "step": 9056
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.430471501136933e-07,
+      "loss": 0.1436,
+      "step": 9057
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.423175023710855e-07,
+      "loss": 0.1765,
+      "step": 9058
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.415889380744929e-07,
+      "loss": 0.1385,
+      "step": 9059
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.408614573048107e-07,
+      "loss": 0.1697,
+      "step": 9060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4013506014281654e-07,
+      "loss": 0.1421,
+      "step": 9061
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.394097466691625e-07,
+      "loss": 0.1592,
+      "step": 9062
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3868551696438647e-07,
+      "loss": 0.1396,
+      "step": 9063
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3796237110890297e-07,
+      "loss": 0.1528,
+      "step": 9064
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3724030918300666e-07,
+      "loss": 0.1868,
+      "step": 9065
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3651933126687233e-07,
+      "loss": 0.2211,
+      "step": 9066
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3579943744055367e-07,
+      "loss": 0.1589,
+      "step": 9067
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3508062778398234e-07,
+      "loss": 0.1321,
+      "step": 9068
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3436290237697334e-07,
+      "loss": 0.1438,
+      "step": 9069
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3364626129921962e-07,
+      "loss": 0.1921,
+      "step": 9070
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.329307046302931e-07,
+      "loss": 0.1268,
+      "step": 9071
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3221623244964576e-07,
+      "loss": 0.1495,
+      "step": 9072
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3150284483660967e-07,
+      "loss": 0.1339,
+      "step": 9073
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3079054187039374e-07,
+      "loss": 0.1709,
+      "step": 9074
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3007932363009133e-07,
+      "loss": 0.1853,
+      "step": 9075
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.293691901946704e-07,
+      "loss": 0.1709,
+      "step": 9076
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2866014164298456e-07,
+      "loss": 0.1678,
+      "step": 9077
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2795217805375857e-07,
+      "loss": 0.1462,
+      "step": 9078
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2724529950560404e-07,
+      "loss": 0.1501,
+      "step": 9079
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2653950607700704e-07,
+      "loss": 0.1796,
+      "step": 9080
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.258347978463382e-07,
+      "loss": 0.1589,
+      "step": 9081
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2513117489184388e-07,
+      "loss": 0.1829,
+      "step": 9082
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2442863729165153e-07,
+      "loss": 0.1378,
+      "step": 9083
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2372718512376545e-07,
+      "loss": 0.1641,
+      "step": 9084
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2302681846607332e-07,
+      "loss": 0.1539,
+      "step": 9085
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.223275373963396e-07,
+      "loss": 0.1321,
+      "step": 9086
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2162934199221108e-07,
+      "loss": 0.1494,
+      "step": 9087
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.209322323312102e-07,
+      "loss": 0.1649,
+      "step": 9088
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2023620849074056e-07,
+      "loss": 0.1805,
+      "step": 9089
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1954127054808484e-07,
+      "loss": 0.1416,
+      "step": 9090
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.188474185804068e-07,
+      "loss": 0.1373,
+      "step": 9091
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1815465266474822e-07,
+      "loss": 0.165,
+      "step": 9092
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1746297287803086e-07,
+      "loss": 0.1527,
+      "step": 9093
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.167723792970544e-07,
+      "loss": 0.1154,
+      "step": 9094
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1608287199850086e-07,
+      "loss": 0.1776,
+      "step": 9095
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1539445105892676e-07,
+      "loss": 0.1738,
+      "step": 9096
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1470711655477316e-07,
+      "loss": 0.1377,
+      "step": 9097
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1402086856235794e-07,
+      "loss": 0.1538,
+      "step": 9098
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.133357071578801e-07,
+      "loss": 0.1224,
+      "step": 9099
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1265163241741437e-07,
+      "loss": 0.189,
+      "step": 9100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.119686444169189e-07,
+      "loss": 0.1575,
+      "step": 9101
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1128674323222742e-07,
+      "loss": 0.157,
+      "step": 9102
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1060592893905607e-07,
+      "loss": 0.1952,
+      "step": 9103
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0992620161299993e-07,
+      "loss": 0.1562,
+      "step": 9104
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0924756132953305e-07,
+      "loss": 0.1704,
+      "step": 9105
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.085700081640052e-07,
+      "loss": 0.1665,
+      "step": 9106
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.078935421916528e-07,
+      "loss": 0.1592,
+      "step": 9107
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0721816348758473e-07,
+      "loss": 0.1479,
+      "step": 9108
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0654387212679094e-07,
+      "loss": 0.1722,
+      "step": 9109
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0587066818414382e-07,
+      "loss": 0.1782,
+      "step": 9110
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.051985517343924e-07,
+      "loss": 0.147,
+      "step": 9111
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.045275228521637e-07,
+      "loss": 0.1707,
+      "step": 9112
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.038575816119681e-07,
+      "loss": 0.1479,
+      "step": 9113
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0318872808818947e-07,
+      "loss": 0.1368,
+      "step": 9114
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.02520962355095e-07,
+      "loss": 0.1505,
+      "step": 9115
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.018542844868332e-07,
+      "loss": 0.1746,
+      "step": 9116
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0118869455742486e-07,
+      "loss": 0.1508,
+      "step": 9117
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0052419264077527e-07,
+      "loss": 0.1418,
+      "step": 9118
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9986077881066769e-07,
+      "loss": 0.187,
+      "step": 9119
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9919845314076426e-07,
+      "loss": 0.1725,
+      "step": 9120
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9853721570460617e-07,
+      "loss": 0.1591,
+      "step": 9121
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.978770665756158e-07,
+      "loss": 0.1538,
+      "step": 9122
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9721800582709006e-07,
+      "loss": 0.1421,
+      "step": 9123
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9656003353221044e-07,
+      "loss": 0.1505,
+      "step": 9124
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9590314976403292e-07,
+      "loss": 0.1824,
+      "step": 9125
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.952473545954958e-07,
+      "loss": 0.1594,
+      "step": 9126
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9459264809941535e-07,
+      "loss": 0.1395,
+      "step": 9127
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9393903034848672e-07,
+      "loss": 0.1729,
+      "step": 9128
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9328650141528516e-07,
+      "loss": 0.1509,
+      "step": 9129
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9263506137226496e-07,
+      "loss": 0.1324,
+      "step": 9130
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9198471029175493e-07,
+      "loss": 0.1796,
+      "step": 9131
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9133544824597172e-07,
+      "loss": 0.1799,
+      "step": 9132
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9068727530700438e-07,
+      "loss": 0.1548,
+      "step": 9133
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9004019154682306e-07,
+      "loss": 0.1727,
+      "step": 9134
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8939419703727592e-07,
+      "loss": 0.1493,
+      "step": 9135
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8874929185009215e-07,
+      "loss": 0.1361,
+      "step": 9136
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.881054760568768e-07,
+      "loss": 0.1447,
+      "step": 9137
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8746274972912038e-07,
+      "loss": 0.1537,
+      "step": 9138
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.868211129381836e-07,
+      "loss": 0.1664,
+      "step": 9139
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8618056575531506e-07,
+      "loss": 0.1646,
+      "step": 9140
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.855411082516334e-07,
+      "loss": 0.1571,
+      "step": 9141
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8490274049814405e-07,
+      "loss": 0.1613,
+      "step": 9142
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8426546256572587e-07,
+      "loss": 0.1582,
+      "step": 9143
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8362927452514335e-07,
+      "loss": 0.1639,
+      "step": 9144
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.829941764470311e-07,
+      "loss": 0.135,
+      "step": 9145
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8236016840191052e-07,
+      "loss": 0.1846,
+      "step": 9146
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8172725046017635e-07,
+      "loss": 0.1459,
+      "step": 9147
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.810954226921058e-07,
+      "loss": 0.189,
+      "step": 9148
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8046468516785487e-07,
+      "loss": 0.2053,
+      "step": 9149
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.798350379574565e-07,
+      "loss": 0.1335,
+      "step": 9150
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7920648113082474e-07,
+      "loss": 0.156,
+      "step": 9151
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.785790147577504e-07,
+      "loss": 0.1077,
+      "step": 9152
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7795263890790448e-07,
+      "loss": 0.1536,
+      "step": 9153
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7732735365083687e-07,
+      "loss": 0.1838,
+      "step": 9154
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7670315905597645e-07,
+      "loss": 0.1641,
+      "step": 9155
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7608005519263116e-07,
+      "loss": 0.1537,
+      "step": 9156
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7545804212998674e-07,
+      "loss": 0.1716,
+      "step": 9157
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7483711993710905e-07,
+      "loss": 0.1554,
+      "step": 9158
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7421728868294185e-07,
+      "loss": 0.1545,
+      "step": 9159
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7359854843630786e-07,
+      "loss": 0.1572,
+      "step": 9160
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7298089926590988e-07,
+      "loss": 0.1729,
+      "step": 9161
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7236434124032865e-07,
+      "loss": 0.1812,
+      "step": 9162
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.717488744280227e-07,
+      "loss": 0.1545,
+      "step": 9163
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.711344988973318e-07,
+      "loss": 0.23,
+      "step": 9164
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7052121471647254e-07,
+      "loss": 0.1521,
+      "step": 9165
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6990902195354153e-07,
+      "loss": 0.1638,
+      "step": 9166
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6929792067651218e-07,
+      "loss": 0.1156,
+      "step": 9167
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6868791095324132e-07,
+      "loss": 0.1614,
+      "step": 9168
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6807899285145812e-07,
+      "loss": 0.198,
+      "step": 9169
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6747116643877736e-07,
+      "loss": 0.1353,
+      "step": 9170
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6686443178268508e-07,
+      "loss": 0.1371,
+      "step": 9171
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6625878895055293e-07,
+      "loss": 0.1401,
+      "step": 9172
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6565423800962822e-07,
+      "loss": 0.1412,
+      "step": 9173
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6505077902703724e-07,
+      "loss": 0.1389,
+      "step": 9174
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6444841206978424e-07,
+      "loss": 0.2314,
+      "step": 9175
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6384713720475454e-07,
+      "loss": 0.1781,
+      "step": 9176
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.632469544987092e-07,
+      "loss": 0.1624,
+      "step": 9177
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6264786401829157e-07,
+      "loss": 0.1423,
+      "step": 9178
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.620498658300207e-07,
+      "loss": 0.1151,
+      "step": 9179
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6145296000029564e-07,
+      "loss": 0.1387,
+      "step": 9180
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6085714659539343e-07,
+      "loss": 0.174,
+      "step": 9181
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.602624256814711e-07,
+      "loss": 0.1598,
+      "step": 9182
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5966879732456365e-07,
+      "loss": 0.1702,
+      "step": 9183
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5907626159058275e-07,
+      "loss": 0.1504,
+      "step": 9184
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.584848185453236e-07,
+      "loss": 0.1638,
+      "step": 9185
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5789446825445476e-07,
+      "loss": 0.1373,
+      "step": 9186
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5730521078352711e-07,
+      "loss": 0.1676,
+      "step": 9187
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5671704619796835e-07,
+      "loss": 0.1387,
+      "step": 9188
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5612997456308733e-07,
+      "loss": 0.1582,
+      "step": 9189
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5554399594406633e-07,
+      "loss": 0.1647,
+      "step": 9190
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5495911040597334e-07,
+      "loss": 0.2069,
+      "step": 9191
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.543753180137475e-07,
+      "loss": 0.1479,
+      "step": 9192
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5379261883221251e-07,
+      "loss": 0.1462,
+      "step": 9193
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5321101292606888e-07,
+      "loss": 0.1572,
+      "step": 9194
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.526305003598949e-07,
+      "loss": 0.1255,
+      "step": 9195
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5205108119814572e-07,
+      "loss": 0.1907,
+      "step": 9196
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5147275550516093e-07,
+      "loss": 0.1875,
+      "step": 9197
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5089552334515255e-07,
+      "loss": 0.1521,
+      "step": 9198
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5031938478221375e-07,
+      "loss": 0.1567,
+      "step": 9199
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4974433988031666e-07,
+      "loss": 0.1582,
+      "step": 9200
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4917038870331246e-07,
+      "loss": 0.1627,
+      "step": 9201
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4859753131492795e-07,
+      "loss": 0.179,
+      "step": 9202
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.480257677787722e-07,
+      "loss": 0.1616,
+      "step": 9203
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4745509815833002e-07,
+      "loss": 0.1267,
+      "step": 9204
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.468855225169652e-07,
+      "loss": 0.1558,
+      "step": 9205
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4631704091792266e-07,
+      "loss": 0.1816,
+      "step": 9206
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.45749653424323e-07,
+      "loss": 0.1819,
+      "step": 9207
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.451833600991659e-07,
+      "loss": 0.1592,
+      "step": 9208
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4461816100532988e-07,
+      "loss": 0.1565,
+      "step": 9209
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4405405620557145e-07,
+      "loss": 0.1414,
+      "step": 9210
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4349104576252715e-07,
+      "loss": 0.1882,
+      "step": 9211
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.429291297387103e-07,
+      "loss": 0.1342,
+      "step": 9212
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4236830819651326e-07,
+      "loss": 0.1368,
+      "step": 9213
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4180858119820728e-07,
+      "loss": 0.1625,
+      "step": 9214
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4124994880594266e-07,
+      "loss": 0.1746,
+      "step": 9215
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4069241108174425e-07,
+      "loss": 0.132,
+      "step": 9216
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.401359680875214e-07,
+      "loss": 0.1576,
+      "step": 9217
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.395806198850569e-07,
+      "loss": 0.1392,
+      "step": 9218
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3902636653601698e-07,
+      "loss": 0.1604,
+      "step": 9219
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.384732081019391e-07,
+      "loss": 0.1517,
+      "step": 9220
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3792114464424633e-07,
+      "loss": 0.1576,
+      "step": 9221
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.373701762242352e-07,
+      "loss": 0.1416,
+      "step": 9222
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3682030290308346e-07,
+      "loss": 0.1836,
+      "step": 9223
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3627152474184669e-07,
+      "loss": 0.1604,
+      "step": 9224
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.357238418014595e-07,
+      "loss": 0.1655,
+      "step": 9225
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3517725414273097e-07,
+      "loss": 0.1527,
+      "step": 9226
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3463176182635485e-07,
+      "loss": 0.1853,
+      "step": 9227
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3408736491289598e-07,
+      "loss": 0.1497,
+      "step": 9228
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3354406346280714e-07,
+      "loss": 0.1388,
+      "step": 9229
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3300185753640892e-07,
+      "loss": 0.1254,
+      "step": 9230
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3246074719390767e-07,
+      "loss": 0.158,
+      "step": 9231
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3192073249538527e-07,
+      "loss": 0.1418,
+      "step": 9232
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.313818135008027e-07,
+      "loss": 0.1755,
+      "step": 9233
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.308439902699965e-07,
+      "loss": 0.1497,
+      "step": 9234
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.303072628626878e-07,
+      "loss": 0.187,
+      "step": 9235
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2977163133847004e-07,
+      "loss": 0.1953,
+      "step": 9236
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2923709575681786e-07,
+      "loss": 0.1586,
+      "step": 9237
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.287036561770816e-07,
+      "loss": 0.1699,
+      "step": 9238
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.281713126584949e-07,
+      "loss": 0.176,
+      "step": 9239
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2764006526016391e-07,
+      "loss": 0.158,
+      "step": 9240
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.271099140410781e-07,
+      "loss": 0.1207,
+      "step": 9241
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.265808590601003e-07,
+      "loss": 0.1461,
+      "step": 9242
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2605290037597694e-07,
+      "loss": 0.1552,
+      "step": 9243
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2552603804732888e-07,
+      "loss": 0.13,
+      "step": 9244
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2500027213265486e-07,
+      "loss": 0.1836,
+      "step": 9245
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2447560269033486e-07,
+      "loss": 0.1482,
+      "step": 9246
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2395202977862675e-07,
+      "loss": 0.1807,
+      "step": 9247
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2342955345566398e-07,
+      "loss": 0.166,
+      "step": 9248
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2290817377946018e-07,
+      "loss": 0.1545,
+      "step": 9249
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.223878908079068e-07,
+      "loss": 0.1232,
+      "step": 9250
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.218687045987732e-07,
+      "loss": 0.1469,
+      "step": 9251
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.213506152097077e-07,
+      "loss": 0.2122,
+      "step": 9252
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.208336226982365e-07,
+      "loss": 0.155,
+      "step": 9253
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.203177271217637e-07,
+      "loss": 0.158,
+      "step": 9254
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1980292853757235e-07,
+      "loss": 0.1724,
+      "step": 9255
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.192892270028223e-07,
+      "loss": 0.1553,
+      "step": 9256
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1877662257455347e-07,
+      "loss": 0.2214,
+      "step": 9257
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1826511530968365e-07,
+      "loss": 0.1298,
+      "step": 9258
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.177547052650052e-07,
+      "loss": 0.1729,
+      "step": 9259
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1724539249719502e-07,
+      "loss": 0.1768,
+      "step": 9260
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.167371770628023e-07,
+      "loss": 0.16,
+      "step": 9261
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1623005901825746e-07,
+      "loss": 0.1536,
+      "step": 9262
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.157240384198688e-07,
+      "loss": 0.1848,
+      "step": 9263
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1521911532382357e-07,
+      "loss": 0.1849,
+      "step": 9264
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1471528978618363e-07,
+      "loss": 0.1777,
+      "step": 9265
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1421256186289308e-07,
+      "loss": 0.1187,
+      "step": 9266
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1371093160977176e-07,
+      "loss": 0.1604,
+      "step": 9267
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.132103990825173e-07,
+      "loss": 0.1437,
+      "step": 9268
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1271096433670748e-07,
+      "loss": 0.1405,
+      "step": 9269
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1221262742779793e-07,
+      "loss": 0.1995,
+      "step": 9270
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1171538841111996e-07,
+      "loss": 0.1559,
+      "step": 9271
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1121924734188716e-07,
+      "loss": 0.1625,
+      "step": 9272
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1072420427518437e-07,
+      "loss": 0.1227,
+      "step": 9273
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1023025926598207e-07,
+      "loss": 0.1472,
+      "step": 9274
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0973741236912527e-07,
+      "loss": 0.1664,
+      "step": 9275
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0924566363933686e-07,
+      "loss": 0.1716,
+      "step": 9276
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.087550131312165e-07,
+      "loss": 0.1368,
+      "step": 9277
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0826546089924727e-07,
+      "loss": 0.1394,
+      "step": 9278
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0777700699778349e-07,
+      "loss": 0.1309,
+      "step": 9279
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0728965148106174e-07,
+      "loss": 0.145,
+      "step": 9280
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0680339440319653e-07,
+      "loss": 0.1503,
+      "step": 9281
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0631823581817913e-07,
+      "loss": 0.1534,
+      "step": 9282
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0583417577987753e-07,
+      "loss": 0.1741,
+      "step": 9283
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0535121434204209e-07,
+      "loss": 0.1412,
+      "step": 9284
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0486935155829658e-07,
+      "loss": 0.2009,
+      "step": 9285
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0438858748214597e-07,
+      "loss": 0.1213,
+      "step": 9286
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.039089221669698e-07,
+      "loss": 0.1772,
+      "step": 9287
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0343035566603099e-07,
+      "loss": 0.145,
+      "step": 9288
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0295288803246484e-07,
+      "loss": 0.1486,
+      "step": 9289
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0247651931928781e-07,
+      "loss": 0.1616,
+      "step": 9290
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0200124957939317e-07,
+      "loss": 0.1493,
+      "step": 9291
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0152707886555423e-07,
+      "loss": 0.1757,
+      "step": 9292
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0105400723041891e-07,
+      "loss": 0.1357,
+      "step": 9293
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0058203472651518e-07,
+      "loss": 0.1846,
+      "step": 9294
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0011116140624888e-07,
+      "loss": 0.1533,
+      "step": 9295
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.964138732190266e-08,
+      "loss": 0.131,
+      "step": 9296
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.91727125256392e-08,
+      "loss": 0.1256,
+      "step": 9297
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.870513706949691e-08,
+      "loss": 0.2061,
+      "step": 9298
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.823866100539425e-08,
+      "loss": 0.1621,
+      "step": 9299
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.777328438512535e-08,
+      "loss": 0.1362,
+      "step": 9300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.73090072603633e-08,
+      "loss": 0.1566,
+      "step": 9301
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.68458296826602e-08,
+      "loss": 0.1907,
+      "step": 9302
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.638375170344494e-08,
+      "loss": 0.1497,
+      "step": 9303
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.592277337402311e-08,
+      "loss": 0.1565,
+      "step": 9304
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.546289474558157e-08,
+      "loss": 0.1533,
+      "step": 9305
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.500411586918168e-08,
+      "loss": 0.125,
+      "step": 9306
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.45464367957638e-08,
+      "loss": 0.1243,
+      "step": 9307
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.408985757614619e-08,
+      "loss": 0.1537,
+      "step": 9308
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.363437826102718e-08,
+      "loss": 0.1553,
+      "step": 9309
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.317999890097961e-08,
+      "loss": 0.1727,
+      "step": 9310
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.272671954645429e-08,
+      "loss": 0.1523,
+      "step": 9311
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.227454024778426e-08,
+      "loss": 0.1266,
+      "step": 9312
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.182346105517492e-08,
+      "loss": 0.1689,
+      "step": 9313
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.137348201871177e-08,
+      "loss": 0.1682,
+      "step": 9314
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.092460318835927e-08,
+      "loss": 0.1682,
+      "step": 9315
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.04768246139598e-08,
+      "loss": 0.1652,
+      "step": 9316
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.003014634523133e-08,
+      "loss": 0.1405,
+      "step": 9317
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.958456843177087e-08,
+      "loss": 0.1678,
+      "step": 9318
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.914009092305221e-08,
+      "loss": 0.1731,
+      "step": 9319
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.869671386842915e-08,
+      "loss": 0.179,
+      "step": 9320
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.825443731713345e-08,
+      "loss": 0.1506,
+      "step": 9321
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.781326131827139e-08,
+      "loss": 0.1482,
+      "step": 9322
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.737318592082932e-08,
+      "loss": 0.1528,
+      "step": 9323
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.693421117367151e-08,
+      "loss": 0.1472,
+      "step": 9324
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.649633712553895e-08,
+      "loss": 0.1775,
+      "step": 9325
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.605956382505166e-08,
+      "loss": 0.1877,
+      "step": 9326
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.562389132070748e-08,
+      "loss": 0.1661,
+      "step": 9327
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.518931966088106e-08,
+      "loss": 0.1763,
+      "step": 9328
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.475584889382383e-08,
+      "loss": 0.1882,
+      "step": 9329
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.432347906766725e-08,
+      "loss": 0.1802,
+      "step": 9330
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.38922102304196e-08,
+      "loss": 0.1775,
+      "step": 9331
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.346204242996703e-08,
+      "loss": 0.1591,
+      "step": 9332
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.303297571407354e-08,
+      "loss": 0.1848,
+      "step": 9333
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.26050101303788e-08,
+      "loss": 0.1649,
+      "step": 9334
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.217814572640481e-08,
+      "loss": 0.188,
+      "step": 9335
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.175238254954587e-08,
+      "loss": 0.186,
+      "step": 9336
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.132772064707972e-08,
+      "loss": 0.156,
+      "step": 9337
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.090416006615643e-08,
+      "loss": 0.155,
+      "step": 9338
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.048170085380613e-08,
+      "loss": 0.1708,
+      "step": 9339
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.006034305693688e-08,
+      "loss": 0.1941,
+      "step": 9340
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.964008672233569e-08,
+      "loss": 0.1537,
+      "step": 9341
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.922093189666413e-08,
+      "loss": 0.1504,
+      "step": 9342
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.880287862646385e-08,
+      "loss": 0.1566,
+      "step": 9343
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.838592695815327e-08,
+      "loss": 0.1794,
+      "step": 9344
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.797007693802872e-08,
+      "loss": 0.1622,
+      "step": 9345
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.755532861226323e-08,
+      "loss": 0.1605,
+      "step": 9346
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.714168202691109e-08,
+      "loss": 0.1807,
+      "step": 9347
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.672913722789888e-08,
+      "loss": 0.1324,
+      "step": 9348
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.631769426103442e-08,
+      "loss": 0.1936,
+      "step": 9349
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.590735317200337e-08,
+      "loss": 0.1747,
+      "step": 9350
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.549811400636709e-08,
+      "loss": 0.1321,
+      "step": 9351
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.508997680956587e-08,
+      "loss": 0.1582,
+      "step": 9352
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.468294162691569e-08,
+      "loss": 0.1506,
+      "step": 9353
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.427700850361375e-08,
+      "loss": 0.1357,
+      "step": 9354
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.387217748473396e-08,
+      "loss": 0.1782,
+      "step": 9355
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.346844861522374e-08,
+      "loss": 0.1694,
+      "step": 9356
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.306582193991274e-08,
+      "loss": 0.1521,
+      "step": 9357
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.266429750350635e-08,
+      "loss": 0.1326,
+      "step": 9358
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.226387535058887e-08,
+      "loss": 0.151,
+      "step": 9359
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.186455552562033e-08,
+      "loss": 0.1168,
+      "step": 9360
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.146633807293968e-08,
+      "loss": 0.1599,
+      "step": 9361
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.106922303676156e-08,
+      "loss": 0.16,
+      "step": 9362
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.067321046118291e-08,
+      "loss": 0.1448,
+      "step": 9363
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.027830039017192e-08,
+      "loss": 0.1578,
+      "step": 9364
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.988449286757792e-08,
+      "loss": 0.1432,
+      "step": 9365
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.94917879371293e-08,
+      "loss": 0.1655,
+      "step": 9366
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.910018564243003e-08,
+      "loss": 0.1549,
+      "step": 9367
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.870968602695871e-08,
+      "loss": 0.137,
+      "step": 9368
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.83202891340784e-08,
+      "loss": 0.1561,
+      "step": 9369
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.79319950070223e-08,
+      "loss": 0.1577,
+      "step": 9370
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.754480368890704e-08,
+      "loss": 0.1461,
+      "step": 9371
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.715871522272376e-08,
+      "loss": 0.1493,
+      "step": 9372
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.677372965134266e-08,
+      "loss": 0.1736,
+      "step": 9373
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.63898470175095e-08,
+      "loss": 0.1704,
+      "step": 9374
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.600706736385021e-08,
+      "loss": 0.1506,
+      "step": 9375
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.562539073286411e-08,
+      "loss": 0.179,
+      "step": 9376
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.524481716693398e-08,
+      "loss": 0.1473,
+      "step": 9377
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.48653467083138e-08,
+      "loss": 0.1888,
+      "step": 9378
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.448697939914095e-08,
+      "loss": 0.1699,
+      "step": 9379
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.41097152814263e-08,
+      "loss": 0.1321,
+      "step": 9380
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.373355439705853e-08,
+      "loss": 0.1523,
+      "step": 9381
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.335849678780536e-08,
+      "loss": 0.1646,
+      "step": 9382
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.298454249531239e-08,
+      "loss": 0.1426,
+      "step": 9383
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.261169156110192e-08,
+      "loss": 0.1785,
+      "step": 9384
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.223994402657086e-08,
+      "loss": 0.1382,
+      "step": 9385
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.18692999329984e-08,
+      "loss": 0.1582,
+      "step": 9386
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.149975932153829e-08,
+      "loss": 0.1536,
+      "step": 9387
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.113132223322327e-08,
+      "loss": 0.1339,
+      "step": 9388
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.076398870896283e-08,
+      "loss": 0.1926,
+      "step": 9389
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.039775878954324e-08,
+      "loss": 0.1946,
+      "step": 9390
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.003263251562863e-08,
+      "loss": 0.1271,
+      "step": 9391
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.966860992776214e-08,
+      "loss": 0.1741,
+      "step": 9392
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.9305691066360306e-08,
+      "loss": 0.1404,
+      "step": 9393
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.894387597172424e-08,
+      "loss": 0.1562,
+      "step": 9394
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.858316468402514e-08,
+      "loss": 0.1259,
+      "step": 9395
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.822355724331541e-08,
+      "loss": 0.2053,
+      "step": 9396
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.786505368952311e-08,
+      "loss": 0.1318,
+      "step": 9397
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7507654062456395e-08,
+      "loss": 0.1372,
+      "step": 9398
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7151358401797975e-08,
+      "loss": 0.1487,
+      "step": 9399
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6796166747110636e-08,
+      "loss": 0.1545,
+      "step": 9400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6442079137830616e-08,
+      "loss": 0.1704,
+      "step": 9401
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6089095613277577e-08,
+      "loss": 0.1395,
+      "step": 9402
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.573721621264239e-08,
+      "loss": 0.145,
+      "step": 9403
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.538644097499601e-08,
+      "loss": 0.135,
+      "step": 9404
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5036769939288415e-08,
+      "loss": 0.1675,
+      "step": 9405
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4688203144345196e-08,
+      "loss": 0.1566,
+      "step": 9406
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.434074062886874e-08,
+      "loss": 0.1412,
+      "step": 9407
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3994382431439285e-08,
+      "loss": 0.1465,
+      "step": 9408
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3649128590516074e-08,
+      "loss": 0.141,
+      "step": 9409
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.330497914443289e-08,
+      "loss": 0.1553,
+      "step": 9410
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.29619341314036e-08,
+      "loss": 0.153,
+      "step": 9411
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2619993589518856e-08,
+      "loss": 0.1819,
+      "step": 9412
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2279157556743846e-08,
+      "loss": 0.1454,
+      "step": 9413
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.193942607092495e-08,
+      "loss": 0.1459,
+      "step": 9414
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.160079916978422e-08,
+      "loss": 0.1577,
+      "step": 9415
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.126327689092048e-08,
+      "loss": 0.1315,
+      "step": 9416
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.092685927181151e-08,
+      "loss": 0.1829,
+      "step": 9417
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.059154634981189e-08,
+      "loss": 0.1321,
+      "step": 9418
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.0257338162150726e-08,
+      "loss": 0.1453,
+      "step": 9419
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.992423474593944e-08,
+      "loss": 0.1342,
+      "step": 9420
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9592236138162887e-08,
+      "loss": 0.1648,
+      "step": 9421
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.926134237568492e-08,
+      "loss": 0.1698,
+      "step": 9422
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.893155349524614e-08,
+      "loss": 0.1394,
+      "step": 9423
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.860286953346616e-08,
+      "loss": 0.1248,
+      "step": 9424
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.827529052683799e-08,
+      "loss": 0.1393,
+      "step": 9425
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7948816511736996e-08,
+      "loss": 0.1865,
+      "step": 9426
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.762344752441084e-08,
+      "loss": 0.1414,
+      "step": 9427
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.72991836009884e-08,
+      "loss": 0.1963,
+      "step": 9428
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.69760247774742e-08,
+      "loss": 0.1233,
+      "step": 9429
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6653971089750674e-08,
+      "loss": 0.1965,
+      "step": 9430
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.633302257357586e-08,
+      "loss": 0.1621,
+      "step": 9431
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6013179264587924e-08,
+      "loss": 0.1407,
+      "step": 9432
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.569444119829847e-08,
+      "loss": 0.1584,
+      "step": 9433
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.537680841010139e-08,
+      "loss": 0.1597,
+      "step": 9434
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.506028093526182e-08,
+      "loss": 0.167,
+      "step": 9435
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4744858808929424e-08,
+      "loss": 0.1714,
+      "step": 9436
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.443054206612396e-08,
+      "loss": 0.1311,
+      "step": 9437
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.411733074174751e-08,
+      "loss": 0.1389,
+      "step": 9438
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.380522487057559e-08,
+      "loss": 0.1599,
+      "step": 9439
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.3494224487264925e-08,
+      "loss": 0.1697,
+      "step": 9440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.318432962634567e-08,
+      "loss": 0.2168,
+      "step": 9441
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.287554032222918e-08,
+      "loss": 0.1381,
+      "step": 9442
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2567856609200265e-08,
+      "loss": 0.1484,
+      "step": 9443
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.226127852142381e-08,
+      "loss": 0.1855,
+      "step": 9444
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1955806092939254e-08,
+      "loss": 0.1851,
+      "step": 9445
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.165143935766613e-08,
+      "loss": 0.1741,
+      "step": 9446
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.134817834939964e-08,
+      "loss": 0.1598,
+      "step": 9447
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1046023101812824e-08,
+      "loss": 0.2043,
+      "step": 9448
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.074497364845331e-08,
+      "loss": 0.1567,
+      "step": 9449
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.044503002275102e-08,
+      "loss": 0.1506,
+      "step": 9450
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.014619225800931e-08,
+      "loss": 0.1504,
+      "step": 9451
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.984846038740831e-08,
+      "loss": 0.1626,
+      "step": 9452
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.955183444400823e-08,
+      "loss": 0.1616,
+      "step": 9453
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.925631446074385e-08,
+      "loss": 0.1249,
+      "step": 9454
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8961900470430025e-08,
+      "loss": 0.1725,
+      "step": 9455
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.866859250575616e-08,
+      "loss": 0.1392,
+      "step": 9456
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.837639059928844e-08,
+      "loss": 0.1193,
+      "step": 9457
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.808529478347422e-08,
+      "loss": 0.1438,
+      "step": 9458
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.779530509063323e-08,
+      "loss": 0.1772,
+      "step": 9459
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.750642155296525e-08,
+      "loss": 0.2029,
+      "step": 9460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.721864420254573e-08,
+      "loss": 0.1392,
+      "step": 9461
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.693197307132912e-08,
+      "loss": 0.1633,
+      "step": 9462
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6646408191146624e-08,
+      "loss": 0.1772,
+      "step": 9463
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.636194959370398e-08,
+      "loss": 0.1389,
+      "step": 9464
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.607859731058816e-08,
+      "loss": 0.187,
+      "step": 9465
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.579635137325954e-08,
+      "loss": 0.1462,
+      "step": 9466
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5515211813058616e-08,
+      "loss": 0.1714,
+      "step": 9467
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.523517866120041e-08,
+      "loss": 0.1895,
+      "step": 9468
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.495625194878005e-08,
+      "loss": 0.14,
+      "step": 9469
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.467843170676721e-08,
+      "loss": 0.1523,
+      "step": 9470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4401717966009443e-08,
+      "loss": 0.1772,
+      "step": 9471
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4126110757232155e-08,
+      "loss": 0.2019,
+      "step": 9472
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.385161011103866e-08,
+      "loss": 0.1555,
+      "step": 9473
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3578216057905674e-08,
+      "loss": 0.1533,
+      "step": 9474
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.330592862819226e-08,
+      "loss": 0.1461,
+      "step": 9475
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.303474785212868e-08,
+      "loss": 0.1428,
+      "step": 9476
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2764673759828617e-08,
+      "loss": 0.1161,
+      "step": 9477
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.249570638127808e-08,
+      "loss": 0.1373,
+      "step": 9478
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.222784574634319e-08,
+      "loss": 0.1477,
+      "step": 9479
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1961091884764596e-08,
+      "loss": 0.1685,
+      "step": 9480
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.169544482616194e-08,
+      "loss": 0.1417,
+      "step": 9481
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.143090460003051e-08,
+      "loss": 0.1406,
+      "step": 9482
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1167471235745705e-08,
+      "loss": 0.1641,
+      "step": 9483
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.090514476255635e-08,
+      "loss": 0.2113,
+      "step": 9484
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.064392520958914e-08,
+      "loss": 0.1956,
+      "step": 9485
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0383812605849774e-08,
+      "loss": 0.1222,
+      "step": 9486
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.012480698022069e-08,
+      "loss": 0.1337,
+      "step": 9487
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.986690836145889e-08,
+      "loss": 0.149,
+      "step": 9488
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.961011677820147e-08,
+      "loss": 0.1804,
+      "step": 9489
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9354432258961174e-08,
+      "loss": 0.1669,
+      "step": 9490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9099854832127516e-08,
+      "loss": 0.1494,
+      "step": 9491
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8846384525966775e-08,
+      "loss": 0.1523,
+      "step": 9492
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8594021368625325e-08,
+      "loss": 0.1659,
+      "step": 9493
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.834276538812186e-08,
+      "loss": 0.1304,
+      "step": 9494
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8092616612356294e-08,
+      "loss": 0.1497,
+      "step": 9495
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7843575069103067e-08,
+      "loss": 0.1562,
+      "step": 9496
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.759564078601562e-08,
+      "loss": 0.1587,
+      "step": 9497
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.734881379062193e-08,
+      "loss": 0.1682,
+      "step": 9498
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7103094110330076e-08,
+      "loss": 0.1738,
+      "step": 9499
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6858481772421563e-08,
+      "loss": 0.1821,
+      "step": 9500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.661497680405911e-08,
+      "loss": 0.1267,
+      "step": 9501
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6372579232278873e-08,
+      "loss": 0.1748,
+      "step": 9502
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.613128908399598e-08,
+      "loss": 0.1674,
+      "step": 9503
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.589110638600234e-08,
+      "loss": 0.1897,
+      "step": 9504
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.565203116496662e-08,
+      "loss": 0.1616,
+      "step": 9505
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5414063447434244e-08,
+      "loss": 0.1565,
+      "step": 9506
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.517720325982853e-08,
+      "loss": 0.1586,
+      "step": 9507
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4941450628448438e-08,
+      "loss": 0.1392,
+      "step": 9508
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4706805579471917e-08,
+      "loss": 0.1353,
+      "step": 9509
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4473268138951457e-08,
+      "loss": 0.187,
+      "step": 9510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.424083833281965e-08,
+      "loss": 0.1768,
+      "step": 9511
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.400951618688252e-08,
+      "loss": 0.1353,
+      "step": 9512
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3779301726826188e-08,
+      "loss": 0.1686,
+      "step": 9513
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3550194978212425e-08,
+      "loss": 0.1411,
+      "step": 9514
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3322195966479778e-08,
+      "loss": 0.1855,
+      "step": 9515
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.3095304716944655e-08,
+      "loss": 0.1486,
+      "step": 9516
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2869521254799133e-08,
+      "loss": 0.1597,
+      "step": 9517
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.264484560511426e-08,
+      "loss": 0.179,
+      "step": 9518
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.242127779283565e-08,
+      "loss": 0.1619,
+      "step": 9519
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2198817842788988e-08,
+      "loss": 0.1198,
+      "step": 9520
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1977465779672303e-08,
+      "loss": 0.1602,
+      "step": 9521
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1757221628067038e-08,
+      "loss": 0.1479,
+      "step": 9522
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1538085412424747e-08,
+      "loss": 0.1697,
+      "step": 9523
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.13200571570793e-08,
+      "loss": 0.1716,
+      "step": 9524
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1103136886239105e-08,
+      "loss": 0.1472,
+      "step": 9525
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0887324623989347e-08,
+      "loss": 0.1611,
+      "step": 9526
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0672620394293073e-08,
+      "loss": 0.1292,
+      "step": 9527
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.045902422099122e-08,
+      "loss": 0.1641,
+      "step": 9528
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0246536127798144e-08,
+      "loss": 0.1509,
+      "step": 9529
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0035156138308308e-08,
+      "loss": 0.1554,
+      "step": 9530
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.982488427599405e-08,
+      "loss": 0.1208,
+      "step": 9531
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9615720564201134e-08,
+      "loss": 0.1804,
+      "step": 9532
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.940766502615432e-08,
+      "loss": 0.1582,
+      "step": 9533
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9200717684956237e-08,
+      "loss": 0.1628,
+      "step": 9534
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.8994878563582953e-08,
+      "loss": 0.1458,
+      "step": 9535
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8790147684892847e-08,
+      "loss": 0.1744,
+      "step": 9536
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8586525071615514e-08,
+      "loss": 0.1194,
+      "step": 9537
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.838401074636176e-08,
+      "loss": 0.1797,
+      "step": 9538
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8182604731618035e-08,
+      "loss": 0.1816,
+      "step": 9539
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.798230704974646e-08,
+      "loss": 0.1329,
+      "step": 9540
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.778311772298813e-08,
+      "loss": 0.1766,
+      "step": 9541
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7585036773458686e-08,
+      "loss": 0.1875,
+      "step": 9542
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7388064223153868e-08,
+      "loss": 0.1693,
+      "step": 9543
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7192200093943955e-08,
+      "loss": 0.1636,
+      "step": 9544
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6997444407574893e-08,
+      "loss": 0.157,
+      "step": 9545
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6803797185673822e-08,
+      "loss": 0.1494,
+      "step": 9546
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.661125844974132e-08,
+      "loss": 0.1436,
+      "step": 9547
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6419828221156953e-08,
+      "loss": 0.1704,
+      "step": 9548
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.622950652117483e-08,
+      "loss": 0.1543,
+      "step": 9549
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.604029337092805e-08,
+      "loss": 0.1022,
+      "step": 9550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.585218879142536e-08,
+      "loss": 0.1541,
+      "step": 9551
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5665192803554498e-08,
+      "loss": 0.1652,
+      "step": 9552
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5479305428076628e-08,
+      "loss": 0.1661,
+      "step": 9553
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5294526685633026e-08,
+      "loss": 0.1577,
+      "step": 9554
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5110856596739498e-08,
+      "loss": 0.156,
+      "step": 9555
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4928295181791953e-08,
+      "loss": 0.1624,
+      "step": 9556
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4746842461059729e-08,
+      "loss": 0.1746,
+      "step": 9557
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4566498454690048e-08,
+      "loss": 0.1686,
+      "step": 9558
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4387263182708e-08,
+      "loss": 0.1655,
+      "step": 9559
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4209136665015444e-08,
+      "loss": 0.1868,
+      "step": 9560
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4032118921391003e-08,
+      "loss": 0.1309,
+      "step": 9561
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3856209971487845e-08,
+      "loss": 0.1584,
+      "step": 9562
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3681409834839233e-08,
+      "loss": 0.1525,
+      "step": 9563
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3507718530855196e-08,
+      "loss": 0.1447,
+      "step": 9564
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.333513607882031e-08,
+      "loss": 0.1395,
+      "step": 9565
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3163662497897024e-08,
+      "loss": 0.1687,
+      "step": 9566
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.299329780712677e-08,
+      "loss": 0.1293,
+      "step": 9567
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2824042025423311e-08,
+      "loss": 0.1687,
+      "step": 9568
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2655895171582722e-08,
+      "loss": 0.1721,
+      "step": 9569
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2488857264273402e-08,
+      "loss": 0.1686,
+      "step": 9570
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2322928322042738e-08,
+      "loss": 0.1421,
+      "step": 9571
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.215810836331488e-08,
+      "loss": 0.1545,
+      "step": 9572
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1994397406391855e-08,
+      "loss": 0.1406,
+      "step": 9573
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1831795469449125e-08,
+      "loss": 0.1226,
+      "step": 9574
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1670302570542247e-08,
+      "loss": 0.1613,
+      "step": 9575
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1509918727602432e-08,
+      "loss": 0.1724,
+      "step": 9576
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1350643958438768e-08,
+      "loss": 0.1814,
+      "step": 9577
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1192478280735996e-08,
+      "loss": 0.152,
+      "step": 9578
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1035421712054518e-08,
+      "loss": 0.1858,
+      "step": 9579
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0879474269835933e-08,
+      "loss": 0.1667,
+      "step": 9580
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.072463597139306e-08,
+      "loss": 0.1663,
+      "step": 9581
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0570906833919925e-08,
+      "loss": 0.1611,
+      "step": 9582
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0418286874486206e-08,
+      "loss": 0.1262,
+      "step": 9583
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0266776110038346e-08,
+      "loss": 0.1481,
+      "step": 9584
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0116374557397334e-08,
+      "loss": 0.1731,
+      "step": 9585
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.967082233264258e-09,
+      "loss": 0.1719,
+      "step": 9586
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.818899154215855e-09,
+      "loss": 0.1892,
+      "step": 9587
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.671825336705631e-09,
+      "loss": 0.1479,
+      "step": 9588
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.525860797064967e-09,
+      "loss": 0.1571,
+      "step": 9589
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.381005551498678e-09,
+      "loss": 0.1427,
+      "step": 9590
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.237259616092786e-09,
+      "loss": 0.1431,
+      "step": 9591
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.094623006807857e-09,
+      "loss": 0.1311,
+      "step": 9592
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.953095739481221e-09,
+      "loss": 0.175,
+      "step": 9593
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.812677829826977e-09,
+      "loss": 0.139,
+      "step": 9594
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.673369293437096e-09,
+      "loss": 0.175,
+      "step": 9595
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.535170145779204e-09,
+      "loss": 0.1401,
+      "step": 9596
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.398080402198804e-09,
+      "loss": 0.135,
+      "step": 9597
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.262100077917057e-09,
+      "loss": 0.1542,
+      "step": 9598
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.127229188032992e-09,
+      "loss": 0.1455,
+      "step": 9599
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.99346774752241e-09,
+      "loss": 0.1412,
+      "step": 9600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.860815771237872e-09,
+      "loss": 0.1503,
+      "step": 9601
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.72927327390649e-09,
+      "loss": 0.1682,
+      "step": 9602
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.598840270135466e-09,
+      "loss": 0.152,
+      "step": 9603
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.469516774406549e-09,
+      "loss": 0.1598,
+      "step": 9604
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.341302801081585e-09,
+      "loss": 0.1559,
+      "step": 9605
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.214198364393632e-09,
+      "loss": 0.1271,
+      "step": 9606
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.088203478458067e-09,
+      "loss": 0.1401,
+      "step": 9607
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.9633181572636985e-09,
+      "loss": 0.1887,
+      "step": 9608
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.839542414677214e-09,
+      "loss": 0.1689,
+      "step": 9609
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.7168762644431725e-09,
+      "loss": 0.1598,
+      "step": 9610
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.595319720180681e-09,
+      "loss": 0.1846,
+      "step": 9611
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.474872795386722e-09,
+      "loss": 0.1431,
+      "step": 9612
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.3555355034350395e-09,
+      "loss": 0.1707,
+      "step": 9613
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.237307857577257e-09,
+      "loss": 0.14,
+      "step": 9614
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.120189870940652e-09,
+      "loss": 0.1733,
+      "step": 9615
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.004181556529265e-09,
+      "loss": 0.1292,
+      "step": 9616
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.889282927222795e-09,
+      "loss": 0.1647,
+      "step": 9617
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.775493995781034e-09,
+      "loss": 0.1598,
+      "step": 9618
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.662814774837211e-09,
+      "loss": 0.2019,
+      "step": 9619
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.551245276903538e-09,
+      "loss": 0.1465,
+      "step": 9620
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.440785514366776e-09,
+      "loss": 0.1451,
+      "step": 9621
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.331435499493776e-09,
+      "loss": 0.1802,
+      "step": 9622
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.2231952444259384e-09,
+      "loss": 0.1417,
+      "step": 9623
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.116064761179207e-09,
+      "loss": 0.1504,
+      "step": 9624
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.010044061651842e-09,
+      "loss": 0.1385,
+      "step": 9625
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.905133157614428e-09,
+      "loss": 0.1779,
+      "step": 9626
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.8013320607165345e-09,
+      "loss": 0.1833,
+      "step": 9627
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.698640782483388e-09,
+      "loss": 0.2124,
+      "step": 9628
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.597059334316978e-09,
+      "loss": 0.1564,
+      "step": 9629
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.49658772749606e-09,
+      "loss": 0.1606,
+      "step": 9630
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.397225973178376e-09,
+      "loss": 0.1505,
+      "step": 9631
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.298974082393992e-09,
+      "loss": 0.1737,
+      "step": 9632
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.201832066054179e-09,
+      "loss": 0.1875,
+      "step": 9633
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.105799934944754e-09,
+      "loss": 0.1477,
+      "step": 9634
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.0108776997282995e-09,
+      "loss": 0.1862,
+      "step": 9635
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.9170653709441616e-09,
+      "loss": 0.1582,
+      "step": 9636
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.824362959010674e-09,
+      "loss": 0.1296,
+      "step": 9637
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.732770474218494e-09,
+      "loss": 0.1904,
+      "step": 9638
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.642287926739485e-09,
+      "loss": 0.1522,
+      "step": 9639
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5529153266200542e-09,
+      "loss": 0.1909,
+      "step": 9640
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.464652683783376e-09,
+      "loss": 0.1609,
+      "step": 9641
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3775000080293884e-09,
+      "loss": 0.1885,
+      "step": 9642
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2914573090347956e-09,
+      "loss": 0.1487,
+      "step": 9643
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.206524596355287e-09,
+      "loss": 0.151,
+      "step": 9644
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1227018794199872e-09,
+      "loss": 0.1165,
+      "step": 9645
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.039989167535895e-09,
+      "loss": 0.1541,
+      "step": 9646
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.958386469887886e-09,
+      "loss": 0.1733,
+      "step": 9647
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.87789379553538e-09,
+      "loss": 0.1566,
+      "step": 9648
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7985111534178934e-09,
+      "loss": 0.1792,
+      "step": 9649
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7202385523483753e-09,
+      "loss": 0.1459,
+      "step": 9650
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6430760010176525e-09,
+      "loss": 0.1373,
+      "step": 9651
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.567023507994426e-09,
+      "loss": 0.1931,
+      "step": 9652
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.492081081721942e-09,
+      "loss": 0.1472,
+      "step": 9653
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4182487305224323e-09,
+      "loss": 0.1405,
+      "step": 9654
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3455264625937834e-09,
+      "loss": 0.1566,
+      "step": 9655
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2739142860106476e-09,
+      "loss": 0.1699,
+      "step": 9656
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2034122087244425e-09,
+      "loss": 0.1853,
+      "step": 9657
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.13402023856335e-09,
+      "loss": 0.1256,
+      "step": 9658
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0657383832323184e-09,
+      "loss": 0.1616,
+      "step": 9659
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.998566650313061e-09,
+      "loss": 0.1543,
+      "step": 9660
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.932505047264055e-09,
+      "loss": 0.1624,
+      "step": 9661
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8675535814205446e-09,
+      "loss": 0.1462,
+      "step": 9662
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8037122599945389e-09,
+      "loss": 0.1538,
+      "step": 9663
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7409810900737012e-09,
+      "loss": 0.1654,
+      "step": 9664
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6793600786246811e-09,
+      "loss": 0.1538,
+      "step": 9665
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6188492324886729e-09,
+      "loss": 0.1704,
+      "step": 9666
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5594485583847462e-09,
+      "loss": 0.1487,
+      "step": 9667
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5011580629087364e-09,
+      "loss": 0.1388,
+      "step": 9668
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4439777525332432e-09,
+      "loss": 0.1605,
+      "step": 9669
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.387907633605412e-09,
+      "loss": 0.1697,
+      "step": 9670
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.332947712353594e-09,
+      "loss": 0.1597,
+      "step": 9671
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2790979948784643e-09,
+      "loss": 0.1548,
+      "step": 9672
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2263584871607948e-09,
+      "loss": 0.1619,
+      "step": 9673
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1747291950547912e-09,
+      "loss": 0.1682,
+      "step": 9674
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1242101242947556e-09,
+      "loss": 0.1472,
+      "step": 9675
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0748012804884245e-09,
+      "loss": 0.1517,
+      "step": 9676
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.02650266912252e-09,
+      "loss": 0.1995,
+      "step": 9677
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.793142955605294e-10,
+      "loss": 0.1863,
+      "step": 9678
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.332361650415956e-10,
+      "loss": 0.1648,
+      "step": 9679
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.882682826816258e-10,
+      "loss": 0.1456,
+      "step": 9680
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.444106534744034e-10,
+      "loss": 0.1387,
+      "step": 9681
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.016632822893666e-10,
+      "loss": 0.1899,
+      "step": 9682
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.600261738727188e-10,
+      "loss": 0.1743,
+      "step": 9683
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.194993328485389e-10,
+      "loss": 0.1965,
+      "step": 9684
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.800827637154506e-10,
+      "loss": 0.1198,
+      "step": 9685
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.41776470849953e-10,
+      "loss": 0.1646,
+      "step": 9686
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.04580458507531e-10,
+      "loss": 0.1552,
+      "step": 9687
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.684947308159939e-10,
+      "loss": 0.1611,
+      "step": 9688
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.335192917821363e-10,
+      "loss": 0.1454,
+      "step": 9689
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.996541452917391e-10,
+      "loss": 0.1677,
+      "step": 9690
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.668992951029072e-10,
+      "loss": 0.1843,
+      "step": 9691
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.352547448527311e-10,
+      "loss": 0.1561,
+      "step": 9692
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.04720498056177e-10,
+      "loss": 0.1626,
+      "step": 9693
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7529655810275566e-10,
+      "loss": 0.1492,
+      "step": 9694
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.4698292825874335e-10,
+      "loss": 0.1295,
+      "step": 9695
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.197796116694019e-10,
+      "loss": 0.1505,
+      "step": 9696
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.936866113545378e-10,
+      "loss": 0.1772,
+      "step": 9697
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.687039302118333e-10,
+      "loss": 0.1619,
+      "step": 9698
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.448315710146254e-10,
+      "loss": 0.1838,
+      "step": 9699
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.2206953641412676e-10,
+      "loss": 0.1677,
+      "step": 9700
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0041782893720497e-10,
+      "loss": 0.1593,
+      "step": 9701
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7987645098860306e-10,
+      "loss": 0.1622,
+      "step": 9702
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.604454048487192e-10,
+      "loss": 0.1582,
+      "step": 9703
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4212469267582684e-10,
+      "loss": 0.1726,
+      "step": 9704
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2491431650274445e-10,
+      "loss": 0.1348,
+      "step": 9705
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0881427824016578e-10,
+      "loss": 0.1201,
+      "step": 9706
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.382457967777037e-11,
+      "loss": 0.16,
+      "step": 9707
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.994522247978254e-11,
+      "loss": 0.1497,
+      "step": 9708
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.717620818497139e-11,
+      "loss": 0.1636,
+      "step": 9709
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.5517538213312186e-11,
+      "loss": 0.1309,
+      "step": 9710
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.4969213858214735e-11,
+      "loss": 0.161,
+      "step": 9711
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.5531236292074555e-11,
+      "loss": 0.1477,
+      "step": 9712
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.7203606561831962e-11,
+      "loss": 0.1252,
+      "step": 9713
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.9986325592302734e-11,
+      "loss": 0.1688,
+      "step": 9714
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3879394185067896e-11,
+      "loss": 0.1436,
+      "step": 9715
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.882813018473713e-12,
+      "loss": 0.156,
+      "step": 9716
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.996582646521475e-12,
+      "loss": 0.1748,
+      "step": 9717
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.2207035010879396e-12,
+      "loss": 0.1689,
+      "step": 9718
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.551758908151073e-13,
+      "loss": 0.165,
+      "step": 9719
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.1747,
+      "step": 9720
+    },
+    {
+      "epoch": 5.0,
+      "step": 9720,
+      "total_flos": 1.5692699693937066e+19,
+      "train_loss": 0.21506105901773084,
+      "train_runtime": 41597.6724,
+      "train_samples_per_second": 14.951,
+      "train_steps_per_second": 0.234
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 9720,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 1.5692699693937066e+19,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..4f7c443088d0d5e8a29d29cf62807f726ca8efb5
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/README.md
@@ -0,0 +1,11 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+- PEFT 0.5.0
+- PEFT 0.5.0
+
+- PEFT 0.5.0
diff --git a/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..6bfa60db5ceb28f2668f804c4b6aa9526a12a551
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "vicuna-v1-3-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "up_proj",
+    "o_proj",
+    "gate_proj",
+    "v_proj",
+    "q_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..38e5a6ae74048162720dd8a7cecff1a03004f76f
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f8383a6ab82cffb87089bc220430c976e4e48233126240829df9097cc036b62
+size 319970957
diff --git a/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8a11c13945ca6691233666c289a41f105fe5499a
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/config.json
@@ -0,0 +1,36 @@
+{
+  "_name_or_path": "vicuna-v1-3-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "mm_graph_tower": "hvqvae2",
+  "mm_hidden_size": 308,
+  "mm_projector_type": "hlinear",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "model_type": "llava_graph",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.33.2",
+  "tune_mm_mlp_adapter": false,
+  "use_cache": true,
+  "use_lap_pe": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}
diff --git a/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ea3cf1f78fae795a3bdc271a2ea7e5419a02e599
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:112ad676a20bd1f3c6359b7d20a0f8b26e0cf2e82651a9fd23b5b325cf9df8bb
+size 11335231
diff --git a/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..5d3218cbbc89c2971cff72f824b152f815475b48
--- /dev/null
+++ b/vicuna-selfies/graph-text-molgen/retrosynthesis-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-5ep16bz/trainer_state.json
@@ -0,0 +1,60358 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 10055,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.622516556291392e-08,
+      "loss": 0.79,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3245033112582784e-07,
+      "loss": 0.665,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9867549668874176e-07,
+      "loss": 0.748,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.649006622516557e-07,
+      "loss": 0.7568,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.311258278145696e-07,
+      "loss": 0.7217,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.973509933774835e-07,
+      "loss": 0.7822,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.635761589403974e-07,
+      "loss": 0.7588,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.298013245033113e-07,
+      "loss": 0.7705,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.960264900662252e-07,
+      "loss": 0.7715,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.622516556291392e-07,
+      "loss": 0.6963,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.28476821192053e-07,
+      "loss": 0.7139,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.94701986754967e-07,
+      "loss": 0.7607,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.609271523178808e-07,
+      "loss": 0.7207,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.271523178807948e-07,
+      "loss": 0.7266,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.933774834437087e-07,
+      "loss": 0.7744,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0596026490066227e-06,
+      "loss": 0.7383,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1258278145695367e-06,
+      "loss": 0.7764,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1920529801324504e-06,
+      "loss": 0.7598,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2582781456953644e-06,
+      "loss": 0.749,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3245033112582784e-06,
+      "loss": 0.7656,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3907284768211921e-06,
+      "loss": 0.7783,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.456953642384106e-06,
+      "loss": 0.8018,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52317880794702e-06,
+      "loss": 0.7705,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.589403973509934e-06,
+      "loss": 0.7324,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.655629139072848e-06,
+      "loss": 0.7363,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7218543046357616e-06,
+      "loss": 0.7324,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7880794701986755e-06,
+      "loss": 0.7773,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8543046357615895e-06,
+      "loss": 0.8271,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9205298013245035e-06,
+      "loss": 0.709,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9867549668874175e-06,
+      "loss": 0.7744,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0529801324503314e-06,
+      "loss": 0.7939,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1192052980132454e-06,
+      "loss": 0.7246,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1854304635761594e-06,
+      "loss": 0.7637,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2516556291390733e-06,
+      "loss": 0.75,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.317880794701987e-06,
+      "loss": 0.7129,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.384105960264901e-06,
+      "loss": 0.7295,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.450331125827815e-06,
+      "loss": 0.7412,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.516556291390729e-06,
+      "loss": 0.748,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5827814569536424e-06,
+      "loss": 0.7422,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6490066225165567e-06,
+      "loss": 0.7734,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7152317880794703e-06,
+      "loss": 0.7168,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7814569536423843e-06,
+      "loss": 0.8047,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8476821192052982e-06,
+      "loss": 0.7637,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.913907284768212e-06,
+      "loss": 0.7695,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.980132450331126e-06,
+      "loss": 0.709,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04635761589404e-06,
+      "loss": 0.792,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1125827814569537e-06,
+      "loss": 0.7793,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.178807947019868e-06,
+      "loss": 0.7793,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2450331125827816e-06,
+      "loss": 0.7637,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.311258278145696e-06,
+      "loss": 0.6514,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3774834437086096e-06,
+      "loss": 0.7803,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.443708609271523e-06,
+      "loss": 0.7529,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5099337748344375e-06,
+      "loss": 0.708,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.576158940397351e-06,
+      "loss": 0.6318,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.642384105960265e-06,
+      "loss": 0.7021,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.708609271523179e-06,
+      "loss": 0.7656,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.774834437086093e-06,
+      "loss": 0.7549,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.841059602649007e-06,
+      "loss": 0.6924,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9072847682119205e-06,
+      "loss": 0.7432,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.973509933774835e-06,
+      "loss": 0.7646,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0397350993377485e-06,
+      "loss": 0.7744,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.105960264900663e-06,
+      "loss": 0.7188,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.172185430463576e-06,
+      "loss": 0.7568,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.238410596026491e-06,
+      "loss": 0.7227,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.304635761589404e-06,
+      "loss": 0.7236,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.370860927152319e-06,
+      "loss": 0.7021,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.437086092715232e-06,
+      "loss": 0.6895,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.503311258278147e-06,
+      "loss": 0.7031,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56953642384106e-06,
+      "loss": 0.6602,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.635761589403974e-06,
+      "loss": 0.6855,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.701986754966888e-06,
+      "loss": 0.6641,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.768211920529802e-06,
+      "loss": 0.7051,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.834437086092716e-06,
+      "loss": 0.709,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.90066225165563e-06,
+      "loss": 0.7266,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.966887417218543e-06,
+      "loss": 0.6777,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.033112582781458e-06,
+      "loss": 0.7275,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.099337748344372e-06,
+      "loss": 0.7373,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.165562913907285e-06,
+      "loss": 0.6338,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.231788079470199e-06,
+      "loss": 0.6738,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2980132450331135e-06,
+      "loss": 0.6143,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.364238410596026e-06,
+      "loss": 0.6816,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.430463576158941e-06,
+      "loss": 0.6924,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.496688741721855e-06,
+      "loss": 0.6426,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.5629139072847685e-06,
+      "loss": 0.7061,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.629139072847682e-06,
+      "loss": 0.6699,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6953642384105965e-06,
+      "loss": 0.6865,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.76158940397351e-06,
+      "loss": 0.6357,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.827814569536424e-06,
+      "loss": 0.6523,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.894039735099338e-06,
+      "loss": 0.6309,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.960264900662252e-06,
+      "loss": 0.6572,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.026490066225166e-06,
+      "loss": 0.6426,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.09271523178808e-06,
+      "loss": 0.6904,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.158940397350994e-06,
+      "loss": 0.5947,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.225165562913907e-06,
+      "loss": 0.5908,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.291390728476822e-06,
+      "loss": 0.5918,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.357615894039736e-06,
+      "loss": 0.6191,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.423841059602649e-06,
+      "loss": 0.6426,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.490066225165563e-06,
+      "loss": 0.5542,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.556291390728478e-06,
+      "loss": 0.6572,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.622516556291392e-06,
+      "loss": 0.584,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.688741721854305e-06,
+      "loss": 0.6074,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.754966887417219e-06,
+      "loss": 0.5571,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.8211920529801336e-06,
+      "loss": 0.6221,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.887417218543046e-06,
+      "loss": 0.6172,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.953642384105961e-06,
+      "loss": 0.6475,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.019867549668875e-06,
+      "loss": 0.5596,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.086092715231789e-06,
+      "loss": 0.5615,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.152317880794702e-06,
+      "loss": 0.5723,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2185430463576166e-06,
+      "loss": 0.5576,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.28476821192053e-06,
+      "loss": 0.5503,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.3509933774834445e-06,
+      "loss": 0.5918,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.417218543046358e-06,
+      "loss": 0.5728,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.4834437086092724e-06,
+      "loss": 0.5693,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.549668874172186e-06,
+      "loss": 0.5825,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6158940397351e-06,
+      "loss": 0.5474,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.682119205298014e-06,
+      "loss": 0.5596,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.748344370860927e-06,
+      "loss": 0.5229,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.814569536423841e-06,
+      "loss": 0.5967,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.880794701986755e-06,
+      "loss": 0.5894,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.94701986754967e-06,
+      "loss": 0.5537,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.013245033112583e-06,
+      "loss": 0.5342,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.079470198675497e-06,
+      "loss": 0.5405,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.145695364238411e-06,
+      "loss": 0.5859,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.211920529801326e-06,
+      "loss": 0.542,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.278145695364238e-06,
+      "loss": 0.5498,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.344370860927153e-06,
+      "loss": 0.4873,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.410596026490067e-06,
+      "loss": 0.5527,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.476821192052982e-06,
+      "loss": 0.5596,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.543046357615894e-06,
+      "loss": 0.5151,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.609271523178809e-06,
+      "loss": 0.5859,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.675496688741723e-06,
+      "loss": 0.5288,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.741721854304637e-06,
+      "loss": 0.5488,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.80794701986755e-06,
+      "loss": 0.5347,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.874172185430465e-06,
+      "loss": 0.4507,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.940397350993379e-06,
+      "loss": 0.5137,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.006622516556293e-06,
+      "loss": 0.5449,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.072847682119206e-06,
+      "loss": 0.5244,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.13907284768212e-06,
+      "loss": 0.5063,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.205298013245035e-06,
+      "loss": 0.543,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.271523178807948e-06,
+      "loss": 0.541,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.337748344370862e-06,
+      "loss": 0.4849,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.403973509933776e-06,
+      "loss": 0.5317,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.470198675496689e-06,
+      "loss": 0.5127,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.536423841059603e-06,
+      "loss": 0.5142,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.602649006622518e-06,
+      "loss": 0.4692,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.668874172185432e-06,
+      "loss": 0.4888,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.735099337748345e-06,
+      "loss": 0.5059,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.80132450331126e-06,
+      "loss": 0.4575,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867549668874174e-06,
+      "loss": 0.4775,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.933774834437086e-06,
+      "loss": 0.5327,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.5156,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0066225165562915e-05,
+      "loss": 0.5088,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.013245033112583e-05,
+      "loss": 0.5049,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0198675496688744e-05,
+      "loss": 0.5024,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0264900662251655e-05,
+      "loss": 0.4937,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.033112582781457e-05,
+      "loss": 0.5039,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0397350993377484e-05,
+      "loss": 0.46,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0463576158940398e-05,
+      "loss": 0.4526,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0529801324503313e-05,
+      "loss": 0.498,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0596026490066227e-05,
+      "loss": 0.501,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.066225165562914e-05,
+      "loss": 0.5049,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0728476821192052e-05,
+      "loss": 0.4614,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0794701986754967e-05,
+      "loss": 0.4648,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0860927152317881e-05,
+      "loss": 0.4756,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0927152317880796e-05,
+      "loss": 0.4634,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.099337748344371e-05,
+      "loss": 0.4951,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1059602649006624e-05,
+      "loss": 0.4795,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1125827814569537e-05,
+      "loss": 0.4761,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1192052980132451e-05,
+      "loss": 0.4302,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1258278145695364e-05,
+      "loss": 0.4038,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1324503311258279e-05,
+      "loss": 0.5205,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1390728476821193e-05,
+      "loss": 0.48,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1456953642384107e-05,
+      "loss": 0.4487,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.152317880794702e-05,
+      "loss": 0.4688,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1589403973509934e-05,
+      "loss": 0.4658,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1655629139072849e-05,
+      "loss": 0.4487,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1721854304635763e-05,
+      "loss": 0.4253,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1788079470198676e-05,
+      "loss": 0.4614,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.185430463576159e-05,
+      "loss": 0.4355,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1920529801324505e-05,
+      "loss": 0.4634,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1986754966887417e-05,
+      "loss": 0.4663,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2052980132450332e-05,
+      "loss": 0.4561,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2119205298013246e-05,
+      "loss": 0.4863,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.218543046357616e-05,
+      "loss": 0.4307,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2251655629139075e-05,
+      "loss": 0.4766,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2317880794701988e-05,
+      "loss": 0.4312,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2384105960264902e-05,
+      "loss": 0.4736,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2450331125827815e-05,
+      "loss": 0.4536,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.251655629139073e-05,
+      "loss": 0.458,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2582781456953644e-05,
+      "loss": 0.4292,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2649006622516558e-05,
+      "loss": 0.4409,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2715231788079472e-05,
+      "loss": 0.478,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2781456953642387e-05,
+      "loss": 0.4414,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2847682119205298e-05,
+      "loss": 0.4077,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2913907284768212e-05,
+      "loss": 0.4917,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2980132450331127e-05,
+      "loss": 0.4419,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3046357615894041e-05,
+      "loss": 0.436,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3112582781456955e-05,
+      "loss": 0.3813,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.317880794701987e-05,
+      "loss": 0.4517,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3245033112582784e-05,
+      "loss": 0.4287,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3311258278145695e-05,
+      "loss": 0.4546,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.337748344370861e-05,
+      "loss": 0.4316,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3443708609271524e-05,
+      "loss": 0.4258,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3509933774834438e-05,
+      "loss": 0.4565,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3576158940397353e-05,
+      "loss": 0.4277,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3642384105960267e-05,
+      "loss": 0.4185,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3708609271523178e-05,
+      "loss": 0.4551,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3774834437086093e-05,
+      "loss": 0.417,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3841059602649007e-05,
+      "loss": 0.439,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3907284768211921e-05,
+      "loss": 0.4209,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3973509933774836e-05,
+      "loss": 0.4009,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.403973509933775e-05,
+      "loss": 0.3872,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4105960264900665e-05,
+      "loss": 0.4009,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4172185430463577e-05,
+      "loss": 0.4048,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.423841059602649e-05,
+      "loss": 0.4199,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4304635761589404e-05,
+      "loss": 0.4258,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4370860927152319e-05,
+      "loss": 0.4541,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4437086092715233e-05,
+      "loss": 0.3979,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4503311258278147e-05,
+      "loss": 0.3892,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.456953642384106e-05,
+      "loss": 0.4243,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4635761589403975e-05,
+      "loss": 0.4282,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4701986754966889e-05,
+      "loss": 0.3789,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4768211920529802e-05,
+      "loss": 0.4316,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4834437086092716e-05,
+      "loss": 0.4038,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.490066225165563e-05,
+      "loss": 0.4185,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4966887417218545e-05,
+      "loss": 0.3789,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5033112582781458e-05,
+      "loss": 0.4146,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5099337748344372e-05,
+      "loss": 0.4082,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5165562913907286e-05,
+      "loss": 0.395,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.52317880794702e-05,
+      "loss": 0.4019,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5298013245033113e-05,
+      "loss": 0.4282,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5364238410596028e-05,
+      "loss": 0.3853,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5430463576158942e-05,
+      "loss": 0.4341,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5496688741721853e-05,
+      "loss": 0.4062,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5562913907284768e-05,
+      "loss": 0.4062,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5629139072847682e-05,
+      "loss": 0.3882,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5695364238410596e-05,
+      "loss": 0.3579,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.576158940397351e-05,
+      "loss": 0.3711,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5827814569536425e-05,
+      "loss": 0.3716,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.589403973509934e-05,
+      "loss": 0.4292,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.596026490066225e-05,
+      "loss": 0.4307,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6026490066225165e-05,
+      "loss": 0.376,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.609271523178808e-05,
+      "loss": 0.3799,
+      "step": 243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6158940397350994e-05,
+      "loss": 0.3955,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6225165562913908e-05,
+      "loss": 0.3848,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6291390728476823e-05,
+      "loss": 0.4106,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6357615894039737e-05,
+      "loss": 0.3604,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.642384105960265e-05,
+      "loss": 0.3413,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6490066225165562e-05,
+      "loss": 0.3843,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6556291390728477e-05,
+      "loss": 0.3853,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.662251655629139e-05,
+      "loss": 0.395,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6688741721854306e-05,
+      "loss": 0.394,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.675496688741722e-05,
+      "loss": 0.3696,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6821192052980134e-05,
+      "loss": 0.3818,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.688741721854305e-05,
+      "loss": 0.3823,
+      "step": 255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6953642384105963e-05,
+      "loss": 0.374,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7019867549668878e-05,
+      "loss": 0.3198,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.708609271523179e-05,
+      "loss": 0.3325,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7152317880794703e-05,
+      "loss": 0.3384,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7218543046357617e-05,
+      "loss": 0.3867,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7284768211920532e-05,
+      "loss": 0.3887,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7350993377483446e-05,
+      "loss": 0.3413,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.741721854304636e-05,
+      "loss": 0.4092,
+      "step": 263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7483443708609275e-05,
+      "loss": 0.3706,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.754966887417219e-05,
+      "loss": 0.374,
+      "step": 265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.76158940397351e-05,
+      "loss": 0.3335,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7682119205298015e-05,
+      "loss": 0.3721,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.774834437086093e-05,
+      "loss": 0.3345,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7814569536423844e-05,
+      "loss": 0.3589,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7880794701986758e-05,
+      "loss": 0.3428,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7947019867549672e-05,
+      "loss": 0.3389,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8013245033112587e-05,
+      "loss": 0.3677,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8079470198675498e-05,
+      "loss": 0.3022,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8145695364238412e-05,
+      "loss": 0.3584,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8211920529801327e-05,
+      "loss": 0.3438,
+      "step": 275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.827814569536424e-05,
+      "loss": 0.3511,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8344370860927155e-05,
+      "loss": 0.3608,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.841059602649007e-05,
+      "loss": 0.3413,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8476821192052984e-05,
+      "loss": 0.3101,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8543046357615895e-05,
+      "loss": 0.3335,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.860927152317881e-05,
+      "loss": 0.3135,
+      "step": 281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8675496688741724e-05,
+      "loss": 0.3599,
+      "step": 282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8741721854304638e-05,
+      "loss": 0.3447,
+      "step": 283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8807947019867553e-05,
+      "loss": 0.3335,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8874172185430467e-05,
+      "loss": 0.3298,
+      "step": 285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8940397350993378e-05,
+      "loss": 0.291,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9006622516556292e-05,
+      "loss": 0.3203,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9072847682119207e-05,
+      "loss": 0.3525,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913907284768212e-05,
+      "loss": 0.2876,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9205298013245036e-05,
+      "loss": 0.3247,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.927152317880795e-05,
+      "loss": 0.3389,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9337748344370864e-05,
+      "loss": 0.3232,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9403973509933775e-05,
+      "loss": 0.3149,
+      "step": 293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.947019867549669e-05,
+      "loss": 0.3154,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9536423841059604e-05,
+      "loss": 0.3232,
+      "step": 295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960264900662252e-05,
+      "loss": 0.3091,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9668874172185433e-05,
+      "loss": 0.3057,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9735099337748347e-05,
+      "loss": 0.3149,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.980132450331126e-05,
+      "loss": 0.2844,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9867549668874173e-05,
+      "loss": 0.3516,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9933774834437087e-05,
+      "loss": 0.3262,
+      "step": 301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.3574,
+      "step": 302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999948120797e-05,
+      "loss": 0.3335,
+      "step": 303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999997924831927e-05,
+      "loss": 0.2708,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999995330872033e-05,
+      "loss": 0.3452,
+      "step": 305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999991699328562e-05,
+      "loss": 0.3276,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999987030201884e-05,
+      "loss": 0.3047,
+      "step": 307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999981323492487e-05,
+      "loss": 0.293,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999974579200967e-05,
+      "loss": 0.3147,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999966797328016e-05,
+      "loss": 0.3032,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999995797787445e-05,
+      "loss": 0.3042,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999948120841176e-05,
+      "loss": 0.3267,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999993722622922e-05,
+      "loss": 0.3657,
+      "step": 313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999992529403971e-05,
+      "loss": 0.3398,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999912324273893e-05,
+      "loss": 0.3013,
+      "step": 315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999898316933108e-05,
+      "loss": 0.3052,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999883272018805e-05,
+      "loss": 0.3276,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999867189532547e-05,
+      "loss": 0.3208,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999985006947601e-05,
+      "loss": 0.2783,
+      "step": 319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999983191185096e-05,
+      "loss": 0.3301,
+      "step": 320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999981271665929e-05,
+      "loss": 0.291,
+      "step": 321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999792483902983e-05,
+      "loss": 0.2986,
+      "step": 322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999771213584147e-05,
+      "loss": 0.2959,
+      "step": 323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999748905704984e-05,
+      "loss": 0.2769,
+      "step": 324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999725560267808e-05,
+      "loss": 0.259,
+      "step": 325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999701177275045e-05,
+      "loss": 0.293,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999967575672922e-05,
+      "loss": 0.3198,
+      "step": 327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999649298632977e-05,
+      "loss": 0.2778,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999962180298905e-05,
+      "loss": 0.3018,
+      "step": 329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999593269800307e-05,
+      "loss": 0.293,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999563699069698e-05,
+      "loss": 0.293,
+      "step": 331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999533090800293e-05,
+      "loss": 0.2805,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999950144499527e-05,
+      "loss": 0.2834,
+      "step": 333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946876165791e-05,
+      "loss": 0.3096,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999435040791612e-05,
+      "loss": 0.2917,
+      "step": 335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999400282399863e-05,
+      "loss": 0.262,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999364486486277e-05,
+      "loss": 0.3022,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999327653054563e-05,
+      "loss": 0.2776,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999928978210855e-05,
+      "loss": 0.2573,
+      "step": 339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999250873652164e-05,
+      "loss": 0.3281,
+      "step": 340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999210927689438e-05,
+      "loss": 0.2959,
+      "step": 341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999169944224518e-05,
+      "loss": 0.3228,
+      "step": 342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999127923261664e-05,
+      "loss": 0.3127,
+      "step": 343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999908486480523e-05,
+      "loss": 0.3008,
+      "step": 344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999040768859682e-05,
+      "loss": 0.2998,
+      "step": 345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998995635429598e-05,
+      "loss": 0.2722,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999894946451966e-05,
+      "loss": 0.2705,
+      "step": 347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999890225613466e-05,
+      "loss": 0.2476,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998854010279497e-05,
+      "loss": 0.2822,
+      "step": 349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998804726959173e-05,
+      "loss": 0.2417,
+      "step": 350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998754406178803e-05,
+      "loss": 0.3042,
+      "step": 351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998703047943614e-05,
+      "loss": 0.2759,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998650652258926e-05,
+      "loss": 0.2793,
+      "step": 353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998597219130182e-05,
+      "loss": 0.2976,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999854274856292e-05,
+      "loss": 0.269,
+      "step": 355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998487240562798e-05,
+      "loss": 0.2571,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998430695135573e-05,
+      "loss": 0.2375,
+      "step": 357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999837311228711e-05,
+      "loss": 0.2839,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998314492023387e-05,
+      "loss": 0.2842,
+      "step": 359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998254834350484e-05,
+      "loss": 0.2791,
+      "step": 360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998194139274593e-05,
+      "loss": 0.2693,
+      "step": 361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998132406802008e-05,
+      "loss": 0.3237,
+      "step": 362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999806963693914e-05,
+      "loss": 0.2385,
+      "step": 363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99980058296925e-05,
+      "loss": 0.2668,
+      "step": 364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997940985068702e-05,
+      "loss": 0.2959,
+      "step": 365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997875103074483e-05,
+      "loss": 0.2666,
+      "step": 366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997808183716674e-05,
+      "loss": 0.2849,
+      "step": 367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997740227002217e-05,
+      "loss": 0.2451,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999767123293817e-05,
+      "loss": 0.3101,
+      "step": 369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997601201531685e-05,
+      "loss": 0.2773,
+      "step": 370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997530132790034e-05,
+      "loss": 0.2642,
+      "step": 371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997458026720587e-05,
+      "loss": 0.2671,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997384883330825e-05,
+      "loss": 0.3125,
+      "step": 373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997310702628338e-05,
+      "loss": 0.2673,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997235484620825e-05,
+      "loss": 0.2627,
+      "step": 375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997159229316088e-05,
+      "loss": 0.2588,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997081936722037e-05,
+      "loss": 0.2581,
+      "step": 377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997003606846702e-05,
+      "loss": 0.2759,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99969242396982e-05,
+      "loss": 0.2578,
+      "step": 379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996843835284765e-05,
+      "loss": 0.241,
+      "step": 380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996762393614748e-05,
+      "loss": 0.2678,
+      "step": 381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996679914696596e-05,
+      "loss": 0.2388,
+      "step": 382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996596398538865e-05,
+      "loss": 0.2314,
+      "step": 383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999651184515022e-05,
+      "loss": 0.2832,
+      "step": 384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999642625453944e-05,
+      "loss": 0.2659,
+      "step": 385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99963396267154e-05,
+      "loss": 0.2434,
+      "step": 386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996251961687086e-05,
+      "loss": 0.217,
+      "step": 387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99961632594636e-05,
+      "loss": 0.2483,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996073520054143e-05,
+      "loss": 0.27,
+      "step": 389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995982743468025e-05,
+      "loss": 0.2776,
+      "step": 390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999589092971467e-05,
+      "loss": 0.2678,
+      "step": 391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99957980788036e-05,
+      "loss": 0.2683,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999570419074445e-05,
+      "loss": 0.2512,
+      "step": 393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999560926554696e-05,
+      "loss": 0.2261,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999551330322098e-05,
+      "loss": 0.2424,
+      "step": 395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999541630377647e-05,
+      "loss": 0.249,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999531826722349e-05,
+      "loss": 0.2009,
+      "step": 397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995219193572216e-05,
+      "loss": 0.2781,
+      "step": 398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995119082832927e-05,
+      "loss": 0.2822,
+      "step": 399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999501793501601e-05,
+      "loss": 0.2693,
+      "step": 400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999491575013196e-05,
+      "loss": 0.2734,
+      "step": 401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994812528191375e-05,
+      "loss": 0.2798,
+      "step": 402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994708269204972e-05,
+      "loss": 0.2852,
+      "step": 403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999460297318357e-05,
+      "loss": 0.2756,
+      "step": 404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994496640138084e-05,
+      "loss": 0.251,
+      "step": 405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999438927007955e-05,
+      "loss": 0.2612,
+      "step": 406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999428086301912e-05,
+      "loss": 0.2729,
+      "step": 407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999417141896803e-05,
+      "loss": 0.2451,
+      "step": 408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994060937937637e-05,
+      "loss": 0.229,
+      "step": 409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993949419939412e-05,
+      "loss": 0.2649,
+      "step": 410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999383686498492e-05,
+      "loss": 0.2639,
+      "step": 411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993723273085835e-05,
+      "loss": 0.2449,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993608644253954e-05,
+      "loss": 0.2495,
+      "step": 413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993492978501164e-05,
+      "loss": 0.2588,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993376275839466e-05,
+      "loss": 0.2361,
+      "step": 415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999325853628097e-05,
+      "loss": 0.27,
+      "step": 416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993139759837895e-05,
+      "loss": 0.27,
+      "step": 417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993019946522563e-05,
+      "loss": 0.2273,
+      "step": 418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992899096347403e-05,
+      "loss": 0.2449,
+      "step": 419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999277720932496e-05,
+      "loss": 0.2378,
+      "step": 420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992654285467874e-05,
+      "loss": 0.2634,
+      "step": 421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992530324788903e-05,
+      "loss": 0.2629,
+      "step": 422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992405327300912e-05,
+      "loss": 0.2368,
+      "step": 423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992279293016866e-05,
+      "loss": 0.2078,
+      "step": 424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992152221949842e-05,
+      "loss": 0.2656,
+      "step": 425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992024114113027e-05,
+      "loss": 0.2156,
+      "step": 426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991894969519716e-05,
+      "loss": 0.2361,
+      "step": 427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991764788183303e-05,
+      "loss": 0.2749,
+      "step": 428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991633570117298e-05,
+      "loss": 0.2646,
+      "step": 429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991501315335316e-05,
+      "loss": 0.2388,
+      "step": 430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991368023851078e-05,
+      "loss": 0.2302,
+      "step": 431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991233695678415e-05,
+      "loss": 0.2468,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991098330831266e-05,
+      "loss": 0.2449,
+      "step": 433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990961929323674e-05,
+      "loss": 0.2375,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990824491169792e-05,
+      "loss": 0.2449,
+      "step": 435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990686016383884e-05,
+      "loss": 0.2502,
+      "step": 436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990546504980318e-05,
+      "loss": 0.2415,
+      "step": 437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990405956973563e-05,
+      "loss": 0.2085,
+      "step": 438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990264372378207e-05,
+      "loss": 0.2593,
+      "step": 439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999012175120894e-05,
+      "loss": 0.2295,
+      "step": 440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989978093480558e-05,
+      "loss": 0.2876,
+      "step": 441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998983339920797e-05,
+      "loss": 0.2466,
+      "step": 442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989687668406184e-05,
+      "loss": 0.2097,
+      "step": 443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989540901090327e-05,
+      "loss": 0.2427,
+      "step": 444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989393097275628e-05,
+      "loss": 0.2214,
+      "step": 445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989244256977415e-05,
+      "loss": 0.2566,
+      "step": 446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989094380211137e-05,
+      "loss": 0.25,
+      "step": 447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988943466992346e-05,
+      "loss": 0.2458,
+      "step": 448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99887915173367e-05,
+      "loss": 0.2207,
+      "step": 449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998863853125996e-05,
+      "loss": 0.2148,
+      "step": 450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988484508778003e-05,
+      "loss": 0.2375,
+      "step": 451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998832944990681e-05,
+      "loss": 0.2427,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988173354662472e-05,
+      "loss": 0.238,
+      "step": 453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988016223061183e-05,
+      "loss": 0.2568,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987858055119243e-05,
+      "loss": 0.241,
+      "step": 455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987698850853072e-05,
+      "loss": 0.2063,
+      "step": 456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987538610279183e-05,
+      "loss": 0.218,
+      "step": 457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987377333414203e-05,
+      "loss": 0.2292,
+      "step": 458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987215020274867e-05,
+      "loss": 0.217,
+      "step": 459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987051670878012e-05,
+      "loss": 0.2268,
+      "step": 460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986887285240592e-05,
+      "loss": 0.2053,
+      "step": 461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998672186337966e-05,
+      "loss": 0.2083,
+      "step": 462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986555405312383e-05,
+      "loss": 0.2227,
+      "step": 463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986387911056034e-05,
+      "loss": 0.2217,
+      "step": 464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986219380627987e-05,
+      "loss": 0.2583,
+      "step": 465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986049814045732e-05,
+      "loss": 0.2202,
+      "step": 466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985879211326857e-05,
+      "loss": 0.2036,
+      "step": 467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998570757248907e-05,
+      "loss": 0.2581,
+      "step": 468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998553489755018e-05,
+      "loss": 0.2302,
+      "step": 469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985361186528097e-05,
+      "loss": 0.231,
+      "step": 470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998518643944085e-05,
+      "loss": 0.219,
+      "step": 471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985010656306572e-05,
+      "loss": 0.2478,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.99848338371435e-05,
+      "loss": 0.2402,
+      "step": 473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984655981969977e-05,
+      "loss": 0.1921,
+      "step": 474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984477090804465e-05,
+      "loss": 0.2124,
+      "step": 475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984297163665518e-05,
+      "loss": 0.2429,
+      "step": 476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998411620057181e-05,
+      "loss": 0.2295,
+      "step": 477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983934201542108e-05,
+      "loss": 0.2236,
+      "step": 478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998375116659531e-05,
+      "loss": 0.2273,
+      "step": 479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983567095750396e-05,
+      "loss": 0.2139,
+      "step": 480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998338198902647e-05,
+      "loss": 0.2205,
+      "step": 481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998319584644274e-05,
+      "loss": 0.2305,
+      "step": 482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983008668018514e-05,
+      "loss": 0.2097,
+      "step": 483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998282045377322e-05,
+      "loss": 0.2566,
+      "step": 484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982631203726385e-05,
+      "loss": 0.2275,
+      "step": 485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998244091789764e-05,
+      "loss": 0.2402,
+      "step": 486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982249596306733e-05,
+      "loss": 0.2456,
+      "step": 487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982057238973516e-05,
+      "loss": 0.2163,
+      "step": 488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981863845917945e-05,
+      "loss": 0.2324,
+      "step": 489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981669417160092e-05,
+      "loss": 0.2209,
+      "step": 490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981473952720122e-05,
+      "loss": 0.2148,
+      "step": 491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981277452618322e-05,
+      "loss": 0.2197,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998107991687508e-05,
+      "loss": 0.2053,
+      "step": 493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998088134551089e-05,
+      "loss": 0.2524,
+      "step": 494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980681738546358e-05,
+      "loss": 0.2375,
+      "step": 495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998048109600219e-05,
+      "loss": 0.2332,
+      "step": 496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998027941789921e-05,
+      "loss": 0.2051,
+      "step": 497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998007670425834e-05,
+      "loss": 0.228,
+      "step": 498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979872955100618e-05,
+      "loss": 0.2126,
+      "step": 499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979668170447176e-05,
+      "loss": 0.2634,
+      "step": 500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997946235031927e-05,
+      "loss": 0.2373,
+      "step": 501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979255494738257e-05,
+      "loss": 0.2305,
+      "step": 502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997904760372559e-05,
+      "loss": 0.2288,
+      "step": 503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997883867730285e-05,
+      "loss": 0.2068,
+      "step": 504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978628715491707e-05,
+      "loss": 0.2107,
+      "step": 505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978417718313953e-05,
+      "loss": 0.207,
+      "step": 506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978205685791472e-05,
+      "loss": 0.2229,
+      "step": 507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977992617946275e-05,
+      "loss": 0.2197,
+      "step": 508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977778514800462e-05,
+      "loss": 0.2202,
+      "step": 509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977563376376253e-05,
+      "loss": 0.2395,
+      "step": 510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977347202695965e-05,
+      "loss": 0.2263,
+      "step": 511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997712999378203e-05,
+      "loss": 0.186,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976911749656988e-05,
+      "loss": 0.2124,
+      "step": 513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997669247034348e-05,
+      "loss": 0.2012,
+      "step": 514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976472155864258e-05,
+      "loss": 0.2104,
+      "step": 515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976250806242185e-05,
+      "loss": 0.2083,
+      "step": 516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976028421500223e-05,
+      "loss": 0.1768,
+      "step": 517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997580500166145e-05,
+      "loss": 0.2209,
+      "step": 518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975580546749052e-05,
+      "loss": 0.199,
+      "step": 519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975355056786307e-05,
+      "loss": 0.2002,
+      "step": 520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997512853179662e-05,
+      "loss": 0.2065,
+      "step": 521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974900971803493e-05,
+      "loss": 0.2112,
+      "step": 522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974672376830532e-05,
+      "loss": 0.1934,
+      "step": 523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974442746901464e-05,
+      "loss": 0.1941,
+      "step": 524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997421208204011e-05,
+      "loss": 0.2046,
+      "step": 525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973980382270405e-05,
+      "loss": 0.2334,
+      "step": 526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973747647616387e-05,
+      "loss": 0.2136,
+      "step": 527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997351387810221e-05,
+      "loss": 0.1924,
+      "step": 528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973279073752124e-05,
+      "loss": 0.2002,
+      "step": 529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973043234590495e-05,
+      "loss": 0.2339,
+      "step": 530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972806360641793e-05,
+      "loss": 0.2134,
+      "step": 531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972568451930592e-05,
+      "loss": 0.2292,
+      "step": 532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997232950848158e-05,
+      "loss": 0.1768,
+      "step": 533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972089530319554e-05,
+      "loss": 0.2473,
+      "step": 534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971848517469404e-05,
+      "loss": 0.249,
+      "step": 535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971606469956146e-05,
+      "loss": 0.2358,
+      "step": 536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997136338780489e-05,
+      "loss": 0.1965,
+      "step": 537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971119271040854e-05,
+      "loss": 0.2388,
+      "step": 538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997087411968938e-05,
+      "loss": 0.2095,
+      "step": 539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997062793377589e-05,
+      "loss": 0.2092,
+      "step": 540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970380713325937e-05,
+      "loss": 0.2014,
+      "step": 541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970132458365165e-05,
+      "loss": 0.2009,
+      "step": 542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969883168919342e-05,
+      "loss": 0.2285,
+      "step": 543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996963284501433e-05,
+      "loss": 0.2305,
+      "step": 544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969381486676092e-05,
+      "loss": 0.2324,
+      "step": 545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969129093930727e-05,
+      "loss": 0.2041,
+      "step": 546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968875666804407e-05,
+      "loss": 0.2261,
+      "step": 547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968621205323434e-05,
+      "loss": 0.2109,
+      "step": 548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968365709514215e-05,
+      "loss": 0.2192,
+      "step": 549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968109179403253e-05,
+      "loss": 0.1899,
+      "step": 550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967851615017164e-05,
+      "loss": 0.1963,
+      "step": 551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967593016382677e-05,
+      "loss": 0.207,
+      "step": 552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967333383526623e-05,
+      "loss": 0.2107,
+      "step": 553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967072716475938e-05,
+      "loss": 0.1917,
+      "step": 554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966811015257672e-05,
+      "loss": 0.1995,
+      "step": 555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966548279898982e-05,
+      "loss": 0.2195,
+      "step": 556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966284510427118e-05,
+      "loss": 0.2109,
+      "step": 557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966019706869456e-05,
+      "loss": 0.2051,
+      "step": 558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965753869253474e-05,
+      "loss": 0.1924,
+      "step": 559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965486997606747e-05,
+      "loss": 0.2266,
+      "step": 560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996521909195697e-05,
+      "loss": 0.1985,
+      "step": 561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964950152331943e-05,
+      "loss": 0.2314,
+      "step": 562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964680178759565e-05,
+      "loss": 0.1956,
+      "step": 563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964409171267852e-05,
+      "loss": 0.1592,
+      "step": 564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964137129884922e-05,
+      "loss": 0.2163,
+      "step": 565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963864054639e-05,
+      "loss": 0.1929,
+      "step": 566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963589945558423e-05,
+      "loss": 0.2327,
+      "step": 567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996331480267163e-05,
+      "loss": 0.2344,
+      "step": 568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996303862600717e-05,
+      "loss": 0.2058,
+      "step": 569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962761415593698e-05,
+      "loss": 0.2124,
+      "step": 570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962483171459978e-05,
+      "loss": 0.2332,
+      "step": 571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996220389363488e-05,
+      "loss": 0.2068,
+      "step": 572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996192358214738e-05,
+      "loss": 0.2014,
+      "step": 573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961642237026565e-05,
+      "loss": 0.1892,
+      "step": 574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961359858301622e-05,
+      "loss": 0.2,
+      "step": 575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996107644600186e-05,
+      "loss": 0.2119,
+      "step": 576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960792000156676e-05,
+      "loss": 0.2219,
+      "step": 577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960506520795585e-05,
+      "loss": 0.1704,
+      "step": 578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960220007948212e-05,
+      "loss": 0.2322,
+      "step": 579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959932461644282e-05,
+      "loss": 0.2236,
+      "step": 580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995964388191363e-05,
+      "loss": 0.2043,
+      "step": 581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959354268786205e-05,
+      "loss": 0.1951,
+      "step": 582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995906362229205e-05,
+      "loss": 0.2041,
+      "step": 583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995877194246132e-05,
+      "loss": 0.1802,
+      "step": 584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958479229324282e-05,
+      "loss": 0.2202,
+      "step": 585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958185482911312e-05,
+      "loss": 0.1968,
+      "step": 586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957890703252882e-05,
+      "loss": 0.1975,
+      "step": 587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957594890379584e-05,
+      "loss": 0.1958,
+      "step": 588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957298044322108e-05,
+      "loss": 0.1943,
+      "step": 589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995700016511125e-05,
+      "loss": 0.1897,
+      "step": 590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956701252777922e-05,
+      "loss": 0.2017,
+      "step": 591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956401307353142e-05,
+      "loss": 0.2014,
+      "step": 592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995610032886803e-05,
+      "loss": 0.2029,
+      "step": 593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995579831735381e-05,
+      "loss": 0.2244,
+      "step": 594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995549527284182e-05,
+      "loss": 0.2009,
+      "step": 595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955191195363505e-05,
+      "loss": 0.2168,
+      "step": 596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954886084950418e-05,
+      "loss": 0.1997,
+      "step": 597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995457994163421e-05,
+      "loss": 0.1836,
+      "step": 598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954272765446656e-05,
+      "loss": 0.1902,
+      "step": 599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995396455641962e-05,
+      "loss": 0.1995,
+      "step": 600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995365531458508e-05,
+      "loss": 0.23,
+      "step": 601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995334503997513e-05,
+      "loss": 0.1995,
+      "step": 602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953033732621958e-05,
+      "loss": 0.1941,
+      "step": 603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952721392557867e-05,
+      "loss": 0.2297,
+      "step": 604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952408019815266e-05,
+      "loss": 0.1797,
+      "step": 605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952093614426667e-05,
+      "loss": 0.2063,
+      "step": 606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995177817642469e-05,
+      "loss": 0.2161,
+      "step": 607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951461705842073e-05,
+      "loss": 0.2019,
+      "step": 608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951144202711644e-05,
+      "loss": 0.2012,
+      "step": 609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995082566706635e-05,
+      "loss": 0.1963,
+      "step": 610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950506098939243e-05,
+      "loss": 0.2009,
+      "step": 611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950185498363483e-05,
+      "loss": 0.2021,
+      "step": 612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9949863865372324e-05,
+      "loss": 0.2,
+      "step": 613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994954119999915e-05,
+      "loss": 0.1655,
+      "step": 614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949217502277438e-05,
+      "loss": 0.1775,
+      "step": 615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948892772240768e-05,
+      "loss": 0.2014,
+      "step": 616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948567009922842e-05,
+      "loss": 0.2058,
+      "step": 617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948240215357452e-05,
+      "loss": 0.1956,
+      "step": 618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947912388578515e-05,
+      "loss": 0.1987,
+      "step": 619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947583529620038e-05,
+      "loss": 0.2085,
+      "step": 620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947253638516146e-05,
+      "loss": 0.1931,
+      "step": 621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946922715301066e-05,
+      "loss": 0.1687,
+      "step": 622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946590760009137e-05,
+      "loss": 0.2339,
+      "step": 623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946257772674803e-05,
+      "loss": 0.2073,
+      "step": 624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994592375333261e-05,
+      "loss": 0.1853,
+      "step": 625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945588702017215e-05,
+      "loss": 0.1848,
+      "step": 626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945252618763388e-05,
+      "loss": 0.1809,
+      "step": 627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944915503606e-05,
+      "loss": 0.1898,
+      "step": 628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9944577356580023e-05,
+      "loss": 0.1929,
+      "step": 629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994423817772055e-05,
+      "loss": 0.1917,
+      "step": 630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943897967062767e-05,
+      "loss": 0.1794,
+      "step": 631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943556724641975e-05,
+      "loss": 0.1951,
+      "step": 632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943214450493586e-05,
+      "loss": 0.1899,
+      "step": 633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942871144653108e-05,
+      "loss": 0.1731,
+      "step": 634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942526807156166e-05,
+      "loss": 0.1888,
+      "step": 635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942181438038486e-05,
+      "loss": 0.1968,
+      "step": 636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941835037335903e-05,
+      "loss": 0.1713,
+      "step": 637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994148760508436e-05,
+      "loss": 0.1687,
+      "step": 638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941139141319903e-05,
+      "loss": 0.1726,
+      "step": 639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940789646078694e-05,
+      "loss": 0.2026,
+      "step": 640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940439119396985e-05,
+      "loss": 0.217,
+      "step": 641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994008756131116e-05,
+      "loss": 0.2173,
+      "step": 642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939734971857687e-05,
+      "loss": 0.2009,
+      "step": 643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939381351073153e-05,
+      "loss": 0.1913,
+      "step": 644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993902669899425e-05,
+      "loss": 0.2051,
+      "step": 645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938671015657773e-05,
+      "loss": 0.2014,
+      "step": 646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993831430110063e-05,
+      "loss": 0.198,
+      "step": 647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937956555359833e-05,
+      "loss": 0.1907,
+      "step": 648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937597778472497e-05,
+      "loss": 0.189,
+      "step": 649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937237970475857e-05,
+      "loss": 0.175,
+      "step": 650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936877131407234e-05,
+      "loss": 0.1809,
+      "step": 651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936515261304076e-05,
+      "loss": 0.2095,
+      "step": 652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993615236020393e-05,
+      "loss": 0.1951,
+      "step": 653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993578842814445e-05,
+      "loss": 0.1992,
+      "step": 654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935423465163395e-05,
+      "loss": 0.1584,
+      "step": 655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935057471298633e-05,
+      "loss": 0.1768,
+      "step": 656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993469044658814e-05,
+      "loss": 0.1548,
+      "step": 657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934322391069996e-05,
+      "loss": 0.2188,
+      "step": 658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993395330478239e-05,
+      "loss": 0.1931,
+      "step": 659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933583187763625e-05,
+      "loss": 0.2007,
+      "step": 660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933212040052092e-05,
+      "loss": 0.1907,
+      "step": 661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993283986168631e-05,
+      "loss": 0.2163,
+      "step": 662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932466652704893e-05,
+      "loss": 0.1975,
+      "step": 663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993209241314656e-05,
+      "loss": 0.1697,
+      "step": 664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931717143050147e-05,
+      "loss": 0.1716,
+      "step": 665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931340842454588e-05,
+      "loss": 0.1604,
+      "step": 666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930963511398932e-05,
+      "loss": 0.1912,
+      "step": 667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930585149922325e-05,
+      "loss": 0.199,
+      "step": 668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930205758064033e-05,
+      "loss": 0.2104,
+      "step": 669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992982533586341e-05,
+      "loss": 0.1646,
+      "step": 670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929443883359934e-05,
+      "loss": 0.1862,
+      "step": 671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929061400593186e-05,
+      "loss": 0.1958,
+      "step": 672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992867788760285e-05,
+      "loss": 0.2039,
+      "step": 673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928293344428714e-05,
+      "loss": 0.1858,
+      "step": 674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927907771110682e-05,
+      "loss": 0.1956,
+      "step": 675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927521167688762e-05,
+      "loss": 0.1924,
+      "step": 676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927133534203064e-05,
+      "loss": 0.1775,
+      "step": 677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992674487069381e-05,
+      "loss": 0.1813,
+      "step": 678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992635517720133e-05,
+      "loss": 0.1694,
+      "step": 679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992596445376605e-05,
+      "loss": 0.1855,
+      "step": 680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992557270042852e-05,
+      "loss": 0.1965,
+      "step": 681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992517991722938e-05,
+      "loss": 0.218,
+      "step": 682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992478610420939e-05,
+      "loss": 0.1631,
+      "step": 683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924391261409405e-05,
+      "loss": 0.1943,
+      "step": 684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923995388870404e-05,
+      "loss": 0.1848,
+      "step": 685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992359848663345e-05,
+      "loss": 0.2017,
+      "step": 686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923200554739735e-05,
+      "loss": 0.2239,
+      "step": 687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992280159323054e-05,
+      "loss": 0.1721,
+      "step": 688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922401602147266e-05,
+      "loss": 0.1919,
+      "step": 689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922000581531412e-05,
+      "loss": 0.1873,
+      "step": 690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921598531424592e-05,
+      "loss": 0.1736,
+      "step": 691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921195451868514e-05,
+      "loss": 0.1858,
+      "step": 692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9920791342905005e-05,
+      "loss": 0.1799,
+      "step": 693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920386204576002e-05,
+      "loss": 0.1892,
+      "step": 694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991998003692353e-05,
+      "loss": 0.1648,
+      "step": 695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919572839989734e-05,
+      "loss": 0.2007,
+      "step": 696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919164613816874e-05,
+      "loss": 0.196,
+      "step": 697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918755358447298e-05,
+      "loss": 0.1689,
+      "step": 698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991834507392347e-05,
+      "loss": 0.2102,
+      "step": 699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917933760287966e-05,
+      "loss": 0.2029,
+      "step": 700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917521417583456e-05,
+      "loss": 0.1951,
+      "step": 701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917108045852727e-05,
+      "loss": 0.2229,
+      "step": 702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916693645138673e-05,
+      "loss": 0.1611,
+      "step": 703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916278215484288e-05,
+      "loss": 0.1746,
+      "step": 704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915861756932677e-05,
+      "loss": 0.1897,
+      "step": 705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915444269527052e-05,
+      "loss": 0.1741,
+      "step": 706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915025753310727e-05,
+      "loss": 0.1843,
+      "step": 707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914606208327134e-05,
+      "loss": 0.1985,
+      "step": 708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914185634619797e-05,
+      "loss": 0.1868,
+      "step": 709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913764032232362e-05,
+      "loss": 0.187,
+      "step": 710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913341401208563e-05,
+      "loss": 0.1948,
+      "step": 711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991291774159226e-05,
+      "loss": 0.1711,
+      "step": 712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991249305342741e-05,
+      "loss": 0.2014,
+      "step": 713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9912067336758075e-05,
+      "loss": 0.1772,
+      "step": 714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911640591628428e-05,
+      "loss": 0.2036,
+      "step": 715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911212818082746e-05,
+      "loss": 0.1992,
+      "step": 716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991078401616542e-05,
+      "loss": 0.1613,
+      "step": 717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910354185920936e-05,
+      "loss": 0.1731,
+      "step": 718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990992332739389e-05,
+      "loss": 0.1941,
+      "step": 719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909491440628997e-05,
+      "loss": 0.1692,
+      "step": 720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990905852567106e-05,
+      "loss": 0.1665,
+      "step": 721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908624582565002e-05,
+      "loss": 0.175,
+      "step": 722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908189611355845e-05,
+      "loss": 0.1743,
+      "step": 723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907753612088726e-05,
+      "loss": 0.1719,
+      "step": 724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990731658480888e-05,
+      "loss": 0.1537,
+      "step": 725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906878529561652e-05,
+      "loss": 0.1733,
+      "step": 726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906439446392495e-05,
+      "loss": 0.1755,
+      "step": 727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905999335346967e-05,
+      "loss": 0.1829,
+      "step": 728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905558196470732e-05,
+      "loss": 0.1887,
+      "step": 729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905116029809567e-05,
+      "loss": 0.1698,
+      "step": 730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990467283540934e-05,
+      "loss": 0.1748,
+      "step": 731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990422861331605e-05,
+      "loss": 0.1816,
+      "step": 732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903783363575778e-05,
+      "loss": 0.1777,
+      "step": 733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990333708623473e-05,
+      "loss": 0.1743,
+      "step": 734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902889781339202e-05,
+      "loss": 0.2,
+      "step": 735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902441448935614e-05,
+      "loss": 0.1677,
+      "step": 736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901992089070483e-05,
+      "loss": 0.163,
+      "step": 737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901541701790427e-05,
+      "loss": 0.1658,
+      "step": 738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901090287142185e-05,
+      "loss": 0.1833,
+      "step": 739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900637845172594e-05,
+      "loss": 0.1936,
+      "step": 740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900184375928597e-05,
+      "loss": 0.2058,
+      "step": 741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989972987945724e-05,
+      "loss": 0.1936,
+      "step": 742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989927435580569e-05,
+      "loss": 0.2046,
+      "step": 743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898817805021207e-05,
+      "loss": 0.2134,
+      "step": 744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898360227151166e-05,
+      "loss": 0.2014,
+      "step": 745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897901622243038e-05,
+      "loss": 0.209,
+      "step": 746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989744199034441e-05,
+      "loss": 0.1711,
+      "step": 747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896981331502974e-05,
+      "loss": 0.1733,
+      "step": 748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989651964576653e-05,
+      "loss": 0.175,
+      "step": 749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989605693318297e-05,
+      "loss": 0.1741,
+      "step": 750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895593193800316e-05,
+      "loss": 0.2134,
+      "step": 751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989512842766668e-05,
+      "loss": 0.1907,
+      "step": 752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989466263483029e-05,
+      "loss": 0.1493,
+      "step": 753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894195815339468e-05,
+      "loss": 0.1953,
+      "step": 754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893727969242657e-05,
+      "loss": 0.1694,
+      "step": 755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98932590965884e-05,
+      "loss": 0.1631,
+      "step": 756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989278919742534e-05,
+      "loss": 0.1797,
+      "step": 757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989231827180224e-05,
+      "loss": 0.1807,
+      "step": 758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989184631976796e-05,
+      "loss": 0.1821,
+      "step": 759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989137334137147e-05,
+      "loss": 0.2312,
+      "step": 760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989089933666184e-05,
+      "loss": 0.199,
+      "step": 761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890424305688262e-05,
+      "loss": 0.1582,
+      "step": 762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889948248500014e-05,
+      "loss": 0.1794,
+      "step": 763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9889471165146495e-05,
+      "loss": 0.1958,
+      "step": 764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988899305567721e-05,
+      "loss": 0.1807,
+      "step": 765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888513920141764e-05,
+      "loss": 0.1763,
+      "step": 766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988803375858987e-05,
+      "loss": 0.1924,
+      "step": 767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887552571071352e-05,
+      "loss": 0.1892,
+      "step": 768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988707035763613e-05,
+      "loss": 0.1777,
+      "step": 769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886587118334248e-05,
+      "loss": 0.1721,
+      "step": 770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886102853215842e-05,
+      "loss": 0.1863,
+      "step": 771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885617562331155e-05,
+      "loss": 0.1924,
+      "step": 772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988513124573054e-05,
+      "loss": 0.1907,
+      "step": 773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884643903464464e-05,
+      "loss": 0.1726,
+      "step": 774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884155535583488e-05,
+      "loss": 0.1538,
+      "step": 775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883666142138282e-05,
+      "loss": 0.1785,
+      "step": 776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988317572317963e-05,
+      "loss": 0.1625,
+      "step": 777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882684278758412e-05,
+      "loss": 0.177,
+      "step": 778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988219180892562e-05,
+      "loss": 0.1687,
+      "step": 779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881698313732355e-05,
+      "loss": 0.1951,
+      "step": 780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988120379322982e-05,
+      "loss": 0.1887,
+      "step": 781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880708247469328e-05,
+      "loss": 0.1677,
+      "step": 782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880211676502287e-05,
+      "loss": 0.1702,
+      "step": 783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987971408038023e-05,
+      "loss": 0.1931,
+      "step": 784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879215459154787e-05,
+      "loss": 0.1768,
+      "step": 785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878715812877686e-05,
+      "loss": 0.2,
+      "step": 786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878215141600778e-05,
+      "loss": 0.1443,
+      "step": 787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877713445376005e-05,
+      "loss": 0.1816,
+      "step": 788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877210724255427e-05,
+      "loss": 0.1907,
+      "step": 789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876706978291204e-05,
+      "loss": 0.1667,
+      "step": 790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.98762022075356e-05,
+      "loss": 0.1843,
+      "step": 791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875696412041e-05,
+      "loss": 0.1682,
+      "step": 792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875189591859873e-05,
+      "loss": 0.1687,
+      "step": 793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987468174704481e-05,
+      "loss": 0.1707,
+      "step": 794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9874172877648506e-05,
+      "loss": 0.1846,
+      "step": 795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873662983723762e-05,
+      "loss": 0.1909,
+      "step": 796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873152065323476e-05,
+      "loss": 0.1719,
+      "step": 797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872640122500666e-05,
+      "loss": 0.1919,
+      "step": 798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987212715530845e-05,
+      "loss": 0.1476,
+      "step": 799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987161316380005e-05,
+      "loss": 0.1614,
+      "step": 800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.98710981480288e-05,
+      "loss": 0.1855,
+      "step": 801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870582108048136e-05,
+      "loss": 0.1587,
+      "step": 802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870065043911603e-05,
+      "loss": 0.1973,
+      "step": 803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869546955672847e-05,
+      "loss": 0.1504,
+      "step": 804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986902784338563e-05,
+      "loss": 0.1663,
+      "step": 805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868507707103806e-05,
+      "loss": 0.1702,
+      "step": 806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986798654688135e-05,
+      "loss": 0.1748,
+      "step": 807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867464362772333e-05,
+      "loss": 0.189,
+      "step": 808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986694115483094e-05,
+      "loss": 0.1597,
+      "step": 809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866416923111455e-05,
+      "loss": 0.1738,
+      "step": 810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865891667668277e-05,
+      "loss": 0.1885,
+      "step": 811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865365388555896e-05,
+      "loss": 0.1677,
+      "step": 812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864838085828924e-05,
+      "loss": 0.1418,
+      "step": 813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864309759542074e-05,
+      "loss": 0.1802,
+      "step": 814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986378040975016e-05,
+      "loss": 0.1775,
+      "step": 815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986325003650811e-05,
+      "loss": 0.1816,
+      "step": 816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862718639870953e-05,
+      "loss": 0.1426,
+      "step": 817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862186219893825e-05,
+      "loss": 0.1663,
+      "step": 818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861652776631972e-05,
+      "loss": 0.1975,
+      "step": 819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861118310140742e-05,
+      "loss": 0.1584,
+      "step": 820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9860582820475593e-05,
+      "loss": 0.1812,
+      "step": 821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986004630769208e-05,
+      "loss": 0.1453,
+      "step": 822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985950877184587e-05,
+      "loss": 0.1843,
+      "step": 823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985897021299275e-05,
+      "loss": 0.1548,
+      "step": 824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858430631188588e-05,
+      "loss": 0.1646,
+      "step": 825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857890026489374e-05,
+      "loss": 0.1746,
+      "step": 826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98573483989512e-05,
+      "loss": 0.208,
+      "step": 827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985680574863026e-05,
+      "loss": 0.1964,
+      "step": 828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9856262075582865e-05,
+      "loss": 0.1802,
+      "step": 829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855717379865424e-05,
+      "loss": 0.1602,
+      "step": 830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855171661534453e-05,
+      "loss": 0.1517,
+      "step": 831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854624920646575e-05,
+      "loss": 0.1565,
+      "step": 832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985407715725852e-05,
+      "loss": 0.179,
+      "step": 833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985352837142712e-05,
+      "loss": 0.1743,
+      "step": 834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852978563209318e-05,
+      "loss": 0.1726,
+      "step": 835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985242773266216e-05,
+      "loss": 0.1819,
+      "step": 836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851875879842803e-05,
+      "loss": 0.1641,
+      "step": 837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851323004808505e-05,
+      "loss": 0.1719,
+      "step": 838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985076910761663e-05,
+      "loss": 0.1838,
+      "step": 839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9850214188324647e-05,
+      "loss": 0.1611,
+      "step": 840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849658246990138e-05,
+      "loss": 0.1665,
+      "step": 841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9849101283670787e-05,
+      "loss": 0.1829,
+      "step": 842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848543298424377e-05,
+      "loss": 0.1172,
+      "step": 843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847984291308813e-05,
+      "loss": 0.1743,
+      "step": 844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847424262382087e-05,
+      "loss": 0.1554,
+      "step": 845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846863211702316e-05,
+      "loss": 0.1523,
+      "step": 846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984630113932771e-05,
+      "loss": 0.1702,
+      "step": 847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845738045316584e-05,
+      "loss": 0.1619,
+      "step": 848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984517392972737e-05,
+      "loss": 0.1722,
+      "step": 849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844608792618597e-05,
+      "loss": 0.1794,
+      "step": 850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844042634048905e-05,
+      "loss": 0.1581,
+      "step": 851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984347545407704e-05,
+      "loss": 0.1733,
+      "step": 852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984290725276184e-05,
+      "loss": 0.1697,
+      "step": 853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842338030162273e-05,
+      "loss": 0.1404,
+      "step": 854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841767786337392e-05,
+      "loss": 0.162,
+      "step": 855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984119652134637e-05,
+      "loss": 0.1699,
+      "step": 856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984062423524848e-05,
+      "loss": 0.1694,
+      "step": 857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.98400509281031e-05,
+      "loss": 0.1592,
+      "step": 858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983947659996972e-05,
+      "loss": 0.197,
+      "step": 859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838901250907924e-05,
+      "loss": 0.1807,
+      "step": 860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983832488097741e-05,
+      "loss": 0.1606,
+      "step": 861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983774749023799e-05,
+      "loss": 0.1736,
+      "step": 862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837169078749567e-05,
+      "loss": 0.1775,
+      "step": 863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983658964657215e-05,
+      "loss": 0.1472,
+      "step": 864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983600919376587e-05,
+      "loss": 0.1677,
+      "step": 865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983542772039095e-05,
+      "loss": 0.1575,
+      "step": 866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834845226507725e-05,
+      "loss": 0.1921,
+      "step": 867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834261712176626e-05,
+      "loss": 0.1589,
+      "step": 868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833677177458207e-05,
+      "loss": 0.1792,
+      "step": 869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9833091622413115e-05,
+      "loss": 0.1609,
+      "step": 870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832505047102104e-05,
+      "loss": 0.1611,
+      "step": 871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831917451586036e-05,
+      "loss": 0.1577,
+      "step": 872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983132883592588e-05,
+      "loss": 0.1992,
+      "step": 873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830739200182715e-05,
+      "loss": 0.1567,
+      "step": 874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.983014854441771e-05,
+      "loss": 0.1829,
+      "step": 875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829556868692163e-05,
+      "loss": 0.1731,
+      "step": 876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828964173067457e-05,
+      "loss": 0.1636,
+      "step": 877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982837045760509e-05,
+      "loss": 0.1941,
+      "step": 878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827775722366667e-05,
+      "loss": 0.1755,
+      "step": 879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827179967413898e-05,
+      "loss": 0.1785,
+      "step": 880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982658319280859e-05,
+      "loss": 0.1616,
+      "step": 881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825985398612674e-05,
+      "loss": 0.1593,
+      "step": 882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982538658488817e-05,
+      "loss": 0.1533,
+      "step": 883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824786751697206e-05,
+      "loss": 0.1641,
+      "step": 884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9824185899102032e-05,
+      "loss": 0.1499,
+      "step": 885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823584027164977e-05,
+      "loss": 0.1819,
+      "step": 886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98229811359485e-05,
+      "loss": 0.189,
+      "step": 887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822377225515155e-05,
+      "loss": 0.1873,
+      "step": 888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821772295927603e-05,
+      "loss": 0.1743,
+      "step": 889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821166347248607e-05,
+      "loss": 0.1678,
+      "step": 890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820559379541043e-05,
+      "loss": 0.1569,
+      "step": 891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9819951392867883e-05,
+      "loss": 0.2046,
+      "step": 892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981934238729222e-05,
+      "loss": 0.1685,
+      "step": 893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818732362877234e-05,
+      "loss": 0.1566,
+      "step": 894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9818121319686226e-05,
+      "loss": 0.1562,
+      "step": 895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817509257782595e-05,
+      "loss": 0.1394,
+      "step": 896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981689617722985e-05,
+      "loss": 0.1545,
+      "step": 897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98162820780916e-05,
+      "loss": 0.1724,
+      "step": 898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981566696043156e-05,
+      "loss": 0.187,
+      "step": 899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815050824313565e-05,
+      "loss": 0.1731,
+      "step": 900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981443366980153e-05,
+      "loss": 0.1532,
+      "step": 901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.98138154969595e-05,
+      "loss": 0.1534,
+      "step": 902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813196305851618e-05,
+      "loss": 0.158,
+      "step": 903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812576096542117e-05,
+      "loss": 0.1821,
+      "step": 904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981195486909536e-05,
+      "loss": 0.1597,
+      "step": 905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811332623575802e-05,
+      "loss": 0.1426,
+      "step": 906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810709360048005e-05,
+      "loss": 0.1799,
+      "step": 907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981008507857664e-05,
+      "loss": 0.1526,
+      "step": 908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809459779226474e-05,
+      "loss": 0.1721,
+      "step": 909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9808833462062398e-05,
+      "loss": 0.1646,
+      "step": 910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980820612714939e-05,
+      "loss": 0.1719,
+      "step": 911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9807577774552545e-05,
+      "loss": 0.1587,
+      "step": 912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806948404337062e-05,
+      "loss": 0.1711,
+      "step": 913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806318016568235e-05,
+      "loss": 0.1501,
+      "step": 914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805686611311482e-05,
+      "loss": 0.1572,
+      "step": 915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9805054188632308e-05,
+      "loss": 0.1592,
+      "step": 916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980442074859634e-05,
+      "loss": 0.1831,
+      "step": 917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803786291269294e-05,
+      "loss": 0.1609,
+      "step": 918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803150816717008e-05,
+      "loss": 0.156,
+      "step": 919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802514325005414e-05,
+      "loss": 0.1587,
+      "step": 920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9801876816200557e-05,
+      "loss": 0.1848,
+      "step": 921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980123829036858e-05,
+      "loss": 0.1819,
+      "step": 922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9800598747575734e-05,
+      "loss": 0.1975,
+      "step": 923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979995818788838e-05,
+      "loss": 0.158,
+      "step": 924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979931661137298e-05,
+      "loss": 0.1439,
+      "step": 925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798674018096106e-05,
+      "loss": 0.179,
+      "step": 926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979803040812443e-05,
+      "loss": 0.1562,
+      "step": 927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797385781524734e-05,
+      "loss": 0.1572,
+      "step": 928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.97967401383639e-05,
+      "loss": 0.1821,
+      "step": 929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796093478708916e-05,
+      "loss": 0.1746,
+      "step": 930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979544580262689e-05,
+      "loss": 0.1738,
+      "step": 931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794797110185015e-05,
+      "loss": 0.1489,
+      "step": 932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794147401450597e-05,
+      "loss": 0.167,
+      "step": 933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793496676491058e-05,
+      "loss": 0.1653,
+      "step": 934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9792844935373905e-05,
+      "loss": 0.1625,
+      "step": 935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792192178166768e-05,
+      "loss": 0.1663,
+      "step": 936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791538404937378e-05,
+      "loss": 0.156,
+      "step": 937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.979088361575356e-05,
+      "loss": 0.1423,
+      "step": 938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790227810683263e-05,
+      "loss": 0.1804,
+      "step": 939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789570989794532e-05,
+      "loss": 0.1765,
+      "step": 940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788913153155512e-05,
+      "loss": 0.1687,
+      "step": 941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9788254300834464e-05,
+      "loss": 0.1423,
+      "step": 942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9787594432899742e-05,
+      "loss": 0.1593,
+      "step": 943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786933549419826e-05,
+      "loss": 0.1665,
+      "step": 944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786271650463276e-05,
+      "loss": 0.155,
+      "step": 945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9785608736098778e-05,
+      "loss": 0.1743,
+      "step": 946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978494480639511e-05,
+      "loss": 0.1777,
+      "step": 947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978427986142116e-05,
+      "loss": 0.1506,
+      "step": 948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783613901245923e-05,
+      "loss": 0.1602,
+      "step": 949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97829469259385e-05,
+      "loss": 0.1721,
+      "step": 950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9782278935568092e-05,
+      "loss": 0.1709,
+      "step": 951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978160993020401e-05,
+      "loss": 0.1541,
+      "step": 952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780939909915666e-05,
+      "loss": 0.1843,
+      "step": 953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978026887477259e-05,
+      "loss": 0.158,
+      "step": 954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9779596824844396e-05,
+      "loss": 0.1229,
+      "step": 955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977892376020082e-05,
+      "loss": 0.1523,
+      "step": 956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778249680911698e-05,
+      "loss": 0.1753,
+      "step": 957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977757458704697e-05,
+      "loss": 0.1873,
+      "step": 958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776898478676684e-05,
+      "loss": 0.1477,
+      "step": 959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776221355870992e-05,
+      "loss": 0.1621,
+      "step": 960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977554321870015e-05,
+      "loss": 0.1543,
+      "step": 961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774864067234525e-05,
+      "loss": 0.1711,
+      "step": 962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977418390154458e-05,
+      "loss": 0.1648,
+      "step": 963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773502721700886e-05,
+      "loss": 0.1758,
+      "step": 964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772820527774127e-05,
+      "loss": 0.1389,
+      "step": 965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772137319835078e-05,
+      "loss": 0.1597,
+      "step": 966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771453097954637e-05,
+      "loss": 0.1602,
+      "step": 967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770767862203795e-05,
+      "loss": 0.1533,
+      "step": 968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770081612653646e-05,
+      "loss": 0.1582,
+      "step": 969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.97693943493754e-05,
+      "loss": 0.1677,
+      "step": 970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976870607244036e-05,
+      "loss": 0.1436,
+      "step": 971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976801678191995e-05,
+      "loss": 0.1655,
+      "step": 972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767326477885683e-05,
+      "loss": 0.1504,
+      "step": 973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766635160409186e-05,
+      "loss": 0.1479,
+      "step": 974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9765942829562186e-05,
+      "loss": 0.1501,
+      "step": 975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976524948541652e-05,
+      "loss": 0.1648,
+      "step": 976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764555128044128e-05,
+      "loss": 0.1417,
+      "step": 977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976385975751706e-05,
+      "loss": 0.141,
+      "step": 978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763163373907458e-05,
+      "loss": 0.1611,
+      "step": 979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762465977287587e-05,
+      "loss": 0.1531,
+      "step": 980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.97617675677298e-05,
+      "loss": 0.1646,
+      "step": 981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761068145306566e-05,
+      "loss": 0.1489,
+      "step": 982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976036771009046e-05,
+      "loss": 0.1621,
+      "step": 983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9759666262154146e-05,
+      "loss": 0.1812,
+      "step": 984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758963801570416e-05,
+      "loss": 0.1526,
+      "step": 985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758260328412154e-05,
+      "loss": 0.1564,
+      "step": 986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757555842752353e-05,
+      "loss": 0.1516,
+      "step": 987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9756850344664103e-05,
+      "loss": 0.1609,
+      "step": 988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975614383422061e-05,
+      "loss": 0.1594,
+      "step": 989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975543631149518e-05,
+      "loss": 0.1627,
+      "step": 990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754727776561222e-05,
+      "loss": 0.1588,
+      "step": 991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754018229492254e-05,
+      "loss": 0.1566,
+      "step": 992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753307670361904e-05,
+      "loss": 0.1743,
+      "step": 993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975259609924389e-05,
+      "loss": 0.1445,
+      "step": 994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975188351621204e-05,
+      "loss": 0.1567,
+      "step": 995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9751169921340302e-05,
+      "loss": 0.147,
+      "step": 996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750455314702708e-05,
+      "loss": 0.1565,
+      "step": 997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974973969637341e-05,
+      "loss": 0.1792,
+      "step": 998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749023066426658e-05,
+      "loss": 0.187,
+      "step": 999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748305424936808e-05,
+      "loss": 0.1283,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974758677197832e-05,
+      "loss": 0.1667,
+      "step": 1001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746867107625763e-05,
+      "loss": 0.1772,
+      "step": 1002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746146431953806e-05,
+      "loss": 0.1501,
+      "step": 1003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9745424745037226e-05,
+      "loss": 0.1536,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744702046950904e-05,
+      "loss": 0.1504,
+      "step": 1005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743978337769827e-05,
+      "loss": 0.1279,
+      "step": 1006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743253617569085e-05,
+      "loss": 0.1697,
+      "step": 1007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742527886423872e-05,
+      "loss": 0.1487,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741801144409493e-05,
+      "loss": 0.1736,
+      "step": 1009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974107339160135e-05,
+      "loss": 0.1613,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740344628074952e-05,
+      "loss": 0.1573,
+      "step": 1011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973961485390592e-05,
+      "loss": 0.1444,
+      "step": 1012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738884069169972e-05,
+      "loss": 0.186,
+      "step": 1013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973815227394293e-05,
+      "loss": 0.1567,
+      "step": 1014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737419468300726e-05,
+      "loss": 0.1677,
+      "step": 1015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9736685652319398e-05,
+      "loss": 0.155,
+      "step": 1016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735950826075077e-05,
+      "loss": 0.1589,
+      "step": 1017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735214989644017e-05,
+      "loss": 0.1644,
+      "step": 1018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973447814310256e-05,
+      "loss": 0.177,
+      "step": 1019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733740286527166e-05,
+      "loss": 0.1312,
+      "step": 1020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973300141999439e-05,
+      "loss": 0.1582,
+      "step": 1021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732261543580894e-05,
+      "loss": 0.1731,
+      "step": 1022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9731520657363453e-05,
+      "loss": 0.1549,
+      "step": 1023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730778761418934e-05,
+      "loss": 0.1355,
+      "step": 1024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730035855824317e-05,
+      "loss": 0.1838,
+      "step": 1025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729291940656684e-05,
+      "loss": 0.1602,
+      "step": 1026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9728547015993223e-05,
+      "loss": 0.1525,
+      "step": 1027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727801081911227e-05,
+      "loss": 0.1504,
+      "step": 1028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727054138488095e-05,
+      "loss": 0.1804,
+      "step": 1029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9726306185801323e-05,
+      "loss": 0.1721,
+      "step": 1030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972555722392852e-05,
+      "loss": 0.1454,
+      "step": 1031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724807252947396e-05,
+      "loss": 0.1484,
+      "step": 1032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9724056272935774e-05,
+      "loss": 0.1609,
+      "step": 1033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723304283971566e-05,
+      "loss": 0.1848,
+      "step": 1034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722551286132797e-05,
+      "loss": 0.1528,
+      "step": 1035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97217972794976e-05,
+      "loss": 0.1467,
+      "step": 1036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721042264144214e-05,
+      "loss": 0.146,
+      "step": 1037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.972028624015097e-05,
+      "loss": 0.1553,
+      "step": 1038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719529207596312e-05,
+      "loss": 0.1627,
+      "step": 1039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718771166558796e-05,
+      "loss": 0.1462,
+      "step": 1040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971801211711707e-05,
+      "loss": 0.1355,
+      "step": 1041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717252059349895e-05,
+      "loss": 0.1467,
+      "step": 1042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971649099333613e-05,
+      "loss": 0.1448,
+      "step": 1043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971572891915474e-05,
+      "loss": 0.1387,
+      "step": 1044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97149658368848e-05,
+      "loss": 0.1654,
+      "step": 1045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971420174660549e-05,
+      "loss": 0.1631,
+      "step": 1046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9713436648396086e-05,
+      "loss": 0.1809,
+      "step": 1047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712670542335968e-05,
+      "loss": 0.1638,
+      "step": 1048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971190342850464e-05,
+      "loss": 0.1542,
+      "step": 1049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971113530698168e-05,
+      "loss": 0.1807,
+      "step": 1050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9710366177846802e-05,
+      "loss": 0.1587,
+      "step": 1051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709596041179802e-05,
+      "loss": 0.1462,
+      "step": 1052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970882489706059e-05,
+      "loss": 0.1748,
+      "step": 1053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970805274556917e-05,
+      "loss": 0.1729,
+      "step": 1054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970727958678568e-05,
+      "loss": 0.1488,
+      "step": 1055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970650542079032e-05,
+      "loss": 0.1609,
+      "step": 1056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9705730247663427e-05,
+      "loss": 0.1587,
+      "step": 1057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704954067485432e-05,
+      "loss": 0.1853,
+      "step": 1058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704176880336864e-05,
+      "loss": 0.1707,
+      "step": 1059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703398686298372e-05,
+      "loss": 0.1467,
+      "step": 1060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970261948545069e-05,
+      "loss": 0.1719,
+      "step": 1061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701839277874676e-05,
+      "loss": 0.1454,
+      "step": 1062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970105806365128e-05,
+      "loss": 0.1619,
+      "step": 1063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970027584286155e-05,
+      "loss": 0.1565,
+      "step": 1064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969949261558666e-05,
+      "loss": 0.1506,
+      "step": 1065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698708381907878e-05,
+      "loss": 0.126,
+      "step": 1066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697923141906563e-05,
+      "loss": 0.1639,
+      "step": 1067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96971368956642e-05,
+      "loss": 0.14,
+      "step": 1068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696349643262364e-05,
+      "loss": 0.1621,
+      "step": 1069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695561384782743e-05,
+      "loss": 0.155,
+      "step": 1070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969477212030712e-05,
+      "loss": 0.1438,
+      "step": 1071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969398184991739e-05,
+      "loss": 0.1543,
+      "step": 1072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969319057369555e-05,
+      "loss": 0.1692,
+      "step": 1073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692398291723706e-05,
+      "loss": 0.1711,
+      "step": 1074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969160500408406e-05,
+      "loss": 0.1322,
+      "step": 1075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.969081071085892e-05,
+      "loss": 0.1846,
+      "step": 1076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690015412130705e-05,
+      "loss": 0.1445,
+      "step": 1077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689219107981935e-05,
+      "loss": 0.1481,
+      "step": 1078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688421798495225e-05,
+      "loss": 0.1602,
+      "step": 1079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968762348375331e-05,
+      "loss": 0.1545,
+      "step": 1080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968682416383902e-05,
+      "loss": 0.1521,
+      "step": 1081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686023838835292e-05,
+      "loss": 0.152,
+      "step": 1082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9685222508825162e-05,
+      "loss": 0.1558,
+      "step": 1083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684420173891784e-05,
+      "loss": 0.1403,
+      "step": 1084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683616834118398e-05,
+      "loss": 0.1375,
+      "step": 1085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968281248958836e-05,
+      "loss": 0.1448,
+      "step": 1086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682007140385128e-05,
+      "loss": 0.1632,
+      "step": 1087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681200786592265e-05,
+      "loss": 0.151,
+      "step": 1088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680393428293434e-05,
+      "loss": 0.1492,
+      "step": 1089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679585065572408e-05,
+      "loss": 0.1505,
+      "step": 1090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967877569851306e-05,
+      "loss": 0.1718,
+      "step": 1091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967796532719937e-05,
+      "loss": 0.1676,
+      "step": 1092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677153951715418e-05,
+      "loss": 0.1425,
+      "step": 1093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967634157214539e-05,
+      "loss": 0.1479,
+      "step": 1094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675528188573585e-05,
+      "loss": 0.1626,
+      "step": 1095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9674713801084394e-05,
+      "loss": 0.1467,
+      "step": 1096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673898409762315e-05,
+      "loss": 0.1721,
+      "step": 1097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967308201469195e-05,
+      "loss": 0.1267,
+      "step": 1098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967226461595801e-05,
+      "loss": 0.1748,
+      "step": 1099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671446213645306e-05,
+      "loss": 0.1636,
+      "step": 1100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670626807838756e-05,
+      "loss": 0.1605,
+      "step": 1101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966980639862338e-05,
+      "loss": 0.1614,
+      "step": 1102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.96689849860843e-05,
+      "loss": 0.1461,
+      "step": 1103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668162570306743e-05,
+      "loss": 0.1439,
+      "step": 1104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667339151376047e-05,
+      "loss": 0.1461,
+      "step": 1105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966651472937765e-05,
+      "loss": 0.1758,
+      "step": 1106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665689304397082e-05,
+      "loss": 0.1498,
+      "step": 1107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664862876519995e-05,
+      "loss": 0.1611,
+      "step": 1108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966403544583214e-05,
+      "loss": 0.1499,
+      "step": 1109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663207012419364e-05,
+      "loss": 0.1616,
+      "step": 1110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966237757636763e-05,
+      "loss": 0.1628,
+      "step": 1111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661547137762994e-05,
+      "loss": 0.1567,
+      "step": 1112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9660715696691627e-05,
+      "loss": 0.1577,
+      "step": 1113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965988325323979e-05,
+      "loss": 0.1704,
+      "step": 1114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965904980749386e-05,
+      "loss": 0.1499,
+      "step": 1115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965821535954031e-05,
+      "loss": 0.1589,
+      "step": 1116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965737990946573e-05,
+      "loss": 0.1335,
+      "step": 1117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.96565434573568e-05,
+      "loss": 0.1578,
+      "step": 1118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965570600330031e-05,
+      "loss": 0.1442,
+      "step": 1119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965486754738315e-05,
+      "loss": 0.1489,
+      "step": 1120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654028089692317e-05,
+      "loss": 0.1215,
+      "step": 1121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653187630314917e-05,
+      "loss": 0.1699,
+      "step": 1122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652346169338147e-05,
+      "loss": 0.1558,
+      "step": 1123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965150370684932e-05,
+      "loss": 0.1689,
+      "step": 1124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965066024293585e-05,
+      "loss": 0.142,
+      "step": 1125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9649815777685256e-05,
+      "loss": 0.1416,
+      "step": 1126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964897031118515e-05,
+      "loss": 0.1404,
+      "step": 1127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648123843523258e-05,
+      "loss": 0.1472,
+      "step": 1128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647276374787415e-05,
+      "loss": 0.168,
+      "step": 1129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646427905065545e-05,
+      "loss": 0.1587,
+      "step": 1130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645578434445693e-05,
+      "loss": 0.1406,
+      "step": 1131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9644727963015985e-05,
+      "loss": 0.1655,
+      "step": 1132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643876490864678e-05,
+      "loss": 0.1609,
+      "step": 1133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643024018080115e-05,
+      "loss": 0.1628,
+      "step": 1134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642170544750742e-05,
+      "loss": 0.1531,
+      "step": 1135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641316070965123e-05,
+      "loss": 0.1249,
+      "step": 1136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640460596811907e-05,
+      "loss": 0.1372,
+      "step": 1137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9639604122379867e-05,
+      "loss": 0.1355,
+      "step": 1138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963874664775786e-05,
+      "loss": 0.136,
+      "step": 1139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637888173034866e-05,
+      "loss": 0.1383,
+      "step": 1140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637028698299947e-05,
+      "loss": 0.1697,
+      "step": 1141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9636168223642288e-05,
+      "loss": 0.1672,
+      "step": 1142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635306749151173e-05,
+      "loss": 0.1639,
+      "step": 1143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963444427491598e-05,
+      "loss": 0.1493,
+      "step": 1144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633580801026207e-05,
+      "loss": 0.1531,
+      "step": 1145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963271632757144e-05,
+      "loss": 0.1241,
+      "step": 1146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631850854641374e-05,
+      "loss": 0.1423,
+      "step": 1147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630984382325816e-05,
+      "loss": 0.156,
+      "step": 1148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630116910714664e-05,
+      "loss": 0.1641,
+      "step": 1149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9629248439897924e-05,
+      "loss": 0.1726,
+      "step": 1150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9628378969965712e-05,
+      "loss": 0.1322,
+      "step": 1151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627508501008245e-05,
+      "loss": 0.1492,
+      "step": 1152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626637033115835e-05,
+      "loss": 0.1631,
+      "step": 1153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625764566378903e-05,
+      "loss": 0.1456,
+      "step": 1154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624891100887985e-05,
+      "loss": 0.1483,
+      "step": 1155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624016636733702e-05,
+      "loss": 0.1299,
+      "step": 1156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9623141174006785e-05,
+      "loss": 0.134,
+      "step": 1157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622264712798078e-05,
+      "loss": 0.1724,
+      "step": 1158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621387253198517e-05,
+      "loss": 0.1461,
+      "step": 1159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620508795299148e-05,
+      "loss": 0.1553,
+      "step": 1160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9619629339191112e-05,
+      "loss": 0.1589,
+      "step": 1161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961874888496567e-05,
+      "loss": 0.1355,
+      "step": 1162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961786743271417e-05,
+      "loss": 0.146,
+      "step": 1163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616984982528073e-05,
+      "loss": 0.1819,
+      "step": 1164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9616101534498938e-05,
+      "loss": 0.1501,
+      "step": 1165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961521708871843e-05,
+      "loss": 0.1428,
+      "step": 1166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961433164527832e-05,
+      "loss": 0.1702,
+      "step": 1167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961344520427048e-05,
+      "loss": 0.1525,
+      "step": 1168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612557765786884e-05,
+      "loss": 0.1578,
+      "step": 1169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961166932991961e-05,
+      "loss": 0.1526,
+      "step": 1170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610779896760847e-05,
+      "loss": 0.1334,
+      "step": 1171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609889466402877e-05,
+      "loss": 0.1643,
+      "step": 1172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608998038938084e-05,
+      "loss": 0.1709,
+      "step": 1173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960810561445897e-05,
+      "loss": 0.1332,
+      "step": 1174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960721219305813e-05,
+      "loss": 0.1448,
+      "step": 1175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960631777482826e-05,
+      "loss": 0.1562,
+      "step": 1176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960542235986217e-05,
+      "loss": 0.1443,
+      "step": 1177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604525948252758e-05,
+      "loss": 0.1567,
+      "step": 1178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960362854009304e-05,
+      "loss": 0.1467,
+      "step": 1179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602730135476127e-05,
+      "loss": 0.1304,
+      "step": 1180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601830734495236e-05,
+      "loss": 0.1594,
+      "step": 1181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600930337243694e-05,
+      "loss": 0.158,
+      "step": 1182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9600028943814913e-05,
+      "loss": 0.1633,
+      "step": 1183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959912655430243e-05,
+      "loss": 0.1755,
+      "step": 1184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9598223168799878e-05,
+      "loss": 0.1599,
+      "step": 1185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959731878740098e-05,
+      "loss": 0.1465,
+      "step": 1186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596413410199574e-05,
+      "loss": 0.1584,
+      "step": 1187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959550703728961e-05,
+      "loss": 0.1453,
+      "step": 1188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9594599668765127e-05,
+      "loss": 0.1506,
+      "step": 1189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959369130472027e-05,
+      "loss": 0.1307,
+      "step": 1190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959278194524929e-05,
+      "loss": 0.1244,
+      "step": 1191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959187159044654e-05,
+      "loss": 0.1388,
+      "step": 1192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590960240406483e-05,
+      "loss": 0.1589,
+      "step": 1193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9590047895223673e-05,
+      "loss": 0.1553,
+      "step": 1194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589134554992773e-05,
+      "loss": 0.1302,
+      "step": 1195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588220219808554e-05,
+      "loss": 0.1288,
+      "step": 1196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9587304889765887e-05,
+      "loss": 0.1362,
+      "step": 1197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958638856495974e-05,
+      "loss": 0.1191,
+      "step": 1198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585471245485193e-05,
+      "loss": 0.1515,
+      "step": 1199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9584552931437423e-05,
+      "loss": 0.1351,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583633622911713e-05,
+      "loss": 0.1511,
+      "step": 1201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582713320003454e-05,
+      "loss": 0.168,
+      "step": 1202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581792022808125e-05,
+      "loss": 0.1738,
+      "step": 1203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958086973142133e-05,
+      "loss": 0.1379,
+      "step": 1204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579946445938755e-05,
+      "loss": 0.1711,
+      "step": 1205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579022166456208e-05,
+      "loss": 0.1323,
+      "step": 1206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957809689306958e-05,
+      "loss": 0.1323,
+      "step": 1207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577170625874885e-05,
+      "loss": 0.1403,
+      "step": 1208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576243364968226e-05,
+      "loss": 0.1293,
+      "step": 1209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575315110445815e-05,
+      "loss": 0.1646,
+      "step": 1210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9574385862403965e-05,
+      "loss": 0.1394,
+      "step": 1211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.95734556209391e-05,
+      "loss": 0.1459,
+      "step": 1212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957252438614773e-05,
+      "loss": 0.1467,
+      "step": 1213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571592158126488e-05,
+      "loss": 0.1328,
+      "step": 1214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9570658936972095e-05,
+      "loss": 0.1526,
+      "step": 1215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.956972472278138e-05,
+      "loss": 0.1517,
+      "step": 1216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956878951565128e-05,
+      "loss": 0.141,
+      "step": 1217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567853315678826e-05,
+      "loss": 0.16,
+      "step": 1218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956691612296116e-05,
+      "loss": 0.1511,
+      "step": 1219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565977937595524e-05,
+      "loss": 0.1621,
+      "step": 1220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565038759679256e-05,
+      "loss": 0.1345,
+      "step": 1221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956409858930981e-05,
+      "loss": 0.158,
+      "step": 1222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563157426584737e-05,
+      "loss": 0.1261,
+      "step": 1223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956221527160169e-05,
+      "loss": 0.136,
+      "step": 1224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956127212445842e-05,
+      "loss": 0.139,
+      "step": 1225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9560327985252794e-05,
+      "loss": 0.1355,
+      "step": 1226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955938285408277e-05,
+      "loss": 0.1495,
+      "step": 1227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955843673104641e-05,
+      "loss": 0.1707,
+      "step": 1228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955748961624189e-05,
+      "loss": 0.1414,
+      "step": 1229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556541509767477e-05,
+      "loss": 0.1323,
+      "step": 1230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9555592411721548e-05,
+      "loss": 0.1552,
+      "step": 1231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554642322202574e-05,
+      "loss": 0.1329,
+      "step": 1232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553691241309137e-05,
+      "loss": 0.1436,
+      "step": 1233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9552739169139927e-05,
+      "loss": 0.142,
+      "step": 1234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955178610579372e-05,
+      "loss": 0.1372,
+      "step": 1235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550832051369408e-05,
+      "loss": 0.1429,
+      "step": 1236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954987700596598e-05,
+      "loss": 0.132,
+      "step": 1237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9548920969682535e-05,
+      "loss": 0.142,
+      "step": 1238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547963942618266e-05,
+      "loss": 0.1405,
+      "step": 1239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547005924872468e-05,
+      "loss": 0.1255,
+      "step": 1240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546046916544555e-05,
+      "loss": 0.1543,
+      "step": 1241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545086917734024e-05,
+      "loss": 0.1478,
+      "step": 1242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9544125928540485e-05,
+      "loss": 0.1357,
+      "step": 1243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543163949063648e-05,
+      "loss": 0.1521,
+      "step": 1244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542200979403327e-05,
+      "loss": 0.1401,
+      "step": 1245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541237019659438e-05,
+      "loss": 0.1536,
+      "step": 1246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9540272069932e-05,
+      "loss": 0.1479,
+      "step": 1247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539306130321132e-05,
+      "loss": 0.14,
+      "step": 1248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9538339200927066e-05,
+      "loss": 0.1389,
+      "step": 1249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537371281850123e-05,
+      "loss": 0.1602,
+      "step": 1250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9536402373190736e-05,
+      "loss": 0.1465,
+      "step": 1251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953543247504943e-05,
+      "loss": 0.1399,
+      "step": 1252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9534461587526847e-05,
+      "loss": 0.165,
+      "step": 1253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533489710723725e-05,
+      "loss": 0.1416,
+      "step": 1254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.95325168447409e-05,
+      "loss": 0.1503,
+      "step": 1255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953154298967932e-05,
+      "loss": 0.146,
+      "step": 1256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.953056814564003e-05,
+      "loss": 0.1366,
+      "step": 1257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952959231272417e-05,
+      "loss": 0.1503,
+      "step": 1258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528615491033e-05,
+      "loss": 0.1448,
+      "step": 1259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9527637680667874e-05,
+      "loss": 0.1488,
+      "step": 1260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952665888173024e-05,
+      "loss": 0.1383,
+      "step": 1261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9525679094321667e-05,
+      "loss": 0.1409,
+      "step": 1262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9524698318543803e-05,
+      "loss": 0.1249,
+      "step": 1263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523716554498428e-05,
+      "loss": 0.1523,
+      "step": 1264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522733802287394e-05,
+      "loss": 0.1277,
+      "step": 1265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9521750062012678e-05,
+      "loss": 0.16,
+      "step": 1266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520765333776347e-05,
+      "loss": 0.1602,
+      "step": 1267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9519779617680577e-05,
+      "loss": 0.136,
+      "step": 1268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518792913827643e-05,
+      "loss": 0.1235,
+      "step": 1269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9517805222319926e-05,
+      "loss": 0.1382,
+      "step": 1270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516816543259908e-05,
+      "loss": 0.1516,
+      "step": 1271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9515826876750168e-05,
+      "loss": 0.1423,
+      "step": 1272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514836222893393e-05,
+      "loss": 0.1436,
+      "step": 1273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951384458179238e-05,
+      "loss": 0.139,
+      "step": 1274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951285195355001e-05,
+      "loss": 0.1338,
+      "step": 1275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951185833826928e-05,
+      "loss": 0.1626,
+      "step": 1276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510863736053286e-05,
+      "loss": 0.1653,
+      "step": 1277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509868147005228e-05,
+      "loss": 0.136,
+      "step": 1278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9508871571228404e-05,
+      "loss": 0.1431,
+      "step": 1279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950787400882622e-05,
+      "loss": 0.1353,
+      "step": 1280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950687545990218e-05,
+      "loss": 0.135,
+      "step": 1281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9505875924559888e-05,
+      "loss": 0.1455,
+      "step": 1282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950487540290306e-05,
+      "loss": 0.1449,
+      "step": 1283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950387389503551e-05,
+      "loss": 0.1376,
+      "step": 1284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502871401061146e-05,
+      "loss": 0.1365,
+      "step": 1285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950186792108399e-05,
+      "loss": 0.1444,
+      "step": 1286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500863455208158e-05,
+      "loss": 0.1455,
+      "step": 1287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499858003537875e-05,
+      "loss": 0.1392,
+      "step": 1288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9498851566177462e-05,
+      "loss": 0.1433,
+      "step": 1289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949784414323135e-05,
+      "loss": 0.1448,
+      "step": 1290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496835734804062e-05,
+      "loss": 0.157,
+      "step": 1291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495826341000237e-05,
+      "loss": 0.1338,
+      "step": 1292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494815961924597e-05,
+      "loss": 0.1438,
+      "step": 1293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493804597681986e-05,
+      "loss": 0.1562,
+      "step": 1294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9492792248377337e-05,
+      "loss": 0.1379,
+      "step": 1295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9491778914115692e-05,
+      "loss": 0.149,
+      "step": 1296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949076459500219e-05,
+      "loss": 0.1304,
+      "step": 1297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948974929114208e-05,
+      "loss": 0.1248,
+      "step": 1298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948873300264071e-05,
+      "loss": 0.1611,
+      "step": 1299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948771572960352e-05,
+      "loss": 0.1328,
+      "step": 1300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486697472136063e-05,
+      "loss": 0.1276,
+      "step": 1301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485678230343995e-05,
+      "loss": 0.1592,
+      "step": 1302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948465800433307e-05,
+      "loss": 0.134,
+      "step": 1303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9483636794209143e-05,
+      "loss": 0.1407,
+      "step": 1304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482614600078178e-05,
+      "loss": 0.1313,
+      "step": 1305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948159142204623e-05,
+      "loss": 0.152,
+      "step": 1306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480567260219466e-05,
+      "loss": 0.1279,
+      "step": 1307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9479542114704152e-05,
+      "loss": 0.1477,
+      "step": 1308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947851598560665e-05,
+      "loss": 0.1222,
+      "step": 1309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9477488873033435e-05,
+      "loss": 0.1565,
+      "step": 1310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947646077709108e-05,
+      "loss": 0.1506,
+      "step": 1311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475431697886256e-05,
+      "loss": 0.1609,
+      "step": 1312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9474401635525738e-05,
+      "loss": 0.1134,
+      "step": 1313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9473370590116403e-05,
+      "loss": 0.1432,
+      "step": 1314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947233856176523e-05,
+      "loss": 0.1523,
+      "step": 1315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471305550579305e-05,
+      "loss": 0.1244,
+      "step": 1316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470271556665807e-05,
+      "loss": 0.1318,
+      "step": 1317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469236580132024e-05,
+      "loss": 0.1433,
+      "step": 1318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946820062108534e-05,
+      "loss": 0.1399,
+      "step": 1319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467163679633254e-05,
+      "loss": 0.1172,
+      "step": 1320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466125755883347e-05,
+      "loss": 0.1583,
+      "step": 1321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465086849943318e-05,
+      "loss": 0.1398,
+      "step": 1322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946404696192096e-05,
+      "loss": 0.1426,
+      "step": 1323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463006091924165e-05,
+      "loss": 0.146,
+      "step": 1324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461964240060944e-05,
+      "loss": 0.1327,
+      "step": 1325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946092140643939e-05,
+      "loss": 0.1382,
+      "step": 1326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9459877591167706e-05,
+      "loss": 0.1538,
+      "step": 1327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458832794354198e-05,
+      "loss": 0.1384,
+      "step": 1328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457787016107273e-05,
+      "loss": 0.1694,
+      "step": 1329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9456740256535437e-05,
+      "loss": 0.1285,
+      "step": 1330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455692515747298e-05,
+      "loss": 0.1528,
+      "step": 1331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945464379385158e-05,
+      "loss": 0.1375,
+      "step": 1332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945359409095708e-05,
+      "loss": 0.1672,
+      "step": 1333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452543407172727e-05,
+      "loss": 0.1418,
+      "step": 1334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451491742607534e-05,
+      "loss": 0.1317,
+      "step": 1335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450439097370614e-05,
+      "loss": 0.1519,
+      "step": 1336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449385471571197e-05,
+      "loss": 0.1416,
+      "step": 1337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94483308653186e-05,
+      "loss": 0.1528,
+      "step": 1338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9447275278722247e-05,
+      "loss": 0.1521,
+      "step": 1339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446218711891666e-05,
+      "loss": 0.1378,
+      "step": 1340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445161164936486e-05,
+      "loss": 0.1602,
+      "step": 1341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9444102637966433e-05,
+      "loss": 0.1323,
+      "step": 1342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443043131091343e-05,
+      "loss": 0.1155,
+      "step": 1343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9441982644421144e-05,
+      "loss": 0.1255,
+      "step": 1344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440921178065868e-05,
+      "loss": 0.1587,
+      "step": 1345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9439858732135657e-05,
+      "loss": 0.1359,
+      "step": 1346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943879530674075e-05,
+      "loss": 0.1482,
+      "step": 1347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943773090199148e-05,
+      "loss": 0.1106,
+      "step": 1348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943666551799829e-05,
+      "loss": 0.1707,
+      "step": 1349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435599154871726e-05,
+      "loss": 0.1292,
+      "step": 1350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434531812722427e-05,
+      "loss": 0.1338,
+      "step": 1351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433463491661143e-05,
+      "loss": 0.1403,
+      "step": 1352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943239419179872e-05,
+      "loss": 0.1499,
+      "step": 1353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431323913246107e-05,
+      "loss": 0.1394,
+      "step": 1354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943025265611435e-05,
+      "loss": 0.1189,
+      "step": 1355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429180420514608e-05,
+      "loss": 0.1384,
+      "step": 1356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428107206558133e-05,
+      "loss": 0.121,
+      "step": 1357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9427033014356276e-05,
+      "loss": 0.1353,
+      "step": 1358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9425957844020494e-05,
+      "loss": 0.1329,
+      "step": 1359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942488169566235e-05,
+      "loss": 0.1326,
+      "step": 1360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423804569393497e-05,
+      "loss": 0.139,
+      "step": 1361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94227264653257e-05,
+      "loss": 0.1249,
+      "step": 1362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421647383570822e-05,
+      "loss": 0.1528,
+      "step": 1363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420567324240822e-05,
+      "loss": 0.1389,
+      "step": 1364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941948628744777e-05,
+      "loss": 0.1318,
+      "step": 1365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941840427330383e-05,
+      "loss": 0.1582,
+      "step": 1366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417321281921275e-05,
+      "loss": 0.1381,
+      "step": 1367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9416237313412467e-05,
+      "loss": 0.1151,
+      "step": 1368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941515236788988e-05,
+      "loss": 0.1537,
+      "step": 1369
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941406644546609e-05,
+      "loss": 0.1626,
+      "step": 1370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941297954625376e-05,
+      "loss": 0.1345,
+      "step": 1371
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941189167036568e-05,
+      "loss": 0.1351,
+      "step": 1372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410802817914715e-05,
+      "loss": 0.1423,
+      "step": 1373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9409712989013848e-05,
+      "loss": 0.1545,
+      "step": 1374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9408622183776154e-05,
+      "loss": 0.139,
+      "step": 1375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407530402314818e-05,
+      "loss": 0.1482,
+      "step": 1376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9406437644743115e-05,
+      "loss": 0.1224,
+      "step": 1377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9405343911174432e-05,
+      "loss": 0.1112,
+      "step": 1378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9404249201722255e-05,
+      "loss": 0.1569,
+      "step": 1379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403153516500165e-05,
+      "loss": 0.142,
+      "step": 1380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940205685562185e-05,
+      "loss": 0.1672,
+      "step": 1381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400959219201096e-05,
+      "loss": 0.1534,
+      "step": 1382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.93998606073518e-05,
+      "loss": 0.1511,
+      "step": 1383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939876102018794e-05,
+      "loss": 0.1499,
+      "step": 1384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397660457823618e-05,
+      "loss": 0.1484,
+      "step": 1385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396558920373022e-05,
+      "loss": 0.1349,
+      "step": 1386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9395456407950445e-05,
+      "loss": 0.1311,
+      "step": 1387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394352920670282e-05,
+      "loss": 0.1296,
+      "step": 1388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939324845864703e-05,
+      "loss": 0.1371,
+      "step": 1389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.939214302199529e-05,
+      "loss": 0.1489,
+      "step": 1390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391036610829753e-05,
+      "loss": 0.1388,
+      "step": 1391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389929225265222e-05,
+      "loss": 0.1011,
+      "step": 1392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.93888208654166e-05,
+      "loss": 0.1184,
+      "step": 1393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9387711531398883e-05,
+      "loss": 0.1458,
+      "step": 1394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386601223327183e-05,
+      "loss": 0.1356,
+      "step": 1395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385489941316692e-05,
+      "loss": 0.1459,
+      "step": 1396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9384377685482725e-05,
+      "loss": 0.129,
+      "step": 1397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938326445594068e-05,
+      "loss": 0.1453,
+      "step": 1398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9382150252806072e-05,
+      "loss": 0.1339,
+      "step": 1399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381035076194502e-05,
+      "loss": 0.1429,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937991892622168e-05,
+      "loss": 0.1313,
+      "step": 1401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9378801803003417e-05,
+      "loss": 0.1421,
+      "step": 1402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377683706655626e-05,
+      "loss": 0.1294,
+      "step": 1403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937656463729432e-05,
+      "loss": 0.1276,
+      "step": 1404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9375444595035605e-05,
+      "loss": 0.1405,
+      "step": 1405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.93743235799957e-05,
+      "loss": 0.1316,
+      "step": 1406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937320159229092e-05,
+      "loss": 0.1383,
+      "step": 1407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9372078632037676e-05,
+      "loss": 0.1365,
+      "step": 1408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937095469935249e-05,
+      "loss": 0.1638,
+      "step": 1409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936982979435198e-05,
+      "loss": 0.1487,
+      "step": 1410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9368703917152857e-05,
+      "loss": 0.1448,
+      "step": 1411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367577067871948e-05,
+      "loss": 0.1719,
+      "step": 1412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366449246626167e-05,
+      "loss": 0.1185,
+      "step": 1413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936532045353254e-05,
+      "loss": 0.135,
+      "step": 1414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9364190688708184e-05,
+      "loss": 0.1523,
+      "step": 1415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936305995227033e-05,
+      "loss": 0.1378,
+      "step": 1416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936192824433629e-05,
+      "loss": 0.1492,
+      "step": 1417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360795565023494e-05,
+      "loss": 0.1488,
+      "step": 1418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935966191444947e-05,
+      "loss": 0.1499,
+      "step": 1419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935852729273184e-05,
+      "loss": 0.1587,
+      "step": 1420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935739169998833e-05,
+      "loss": 0.124,
+      "step": 1421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935625513633677e-05,
+      "loss": 0.1401,
+      "step": 1422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355117601895084e-05,
+      "loss": 0.1626,
+      "step": 1423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353979096781304e-05,
+      "loss": 0.1401,
+      "step": 1424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352839621113562e-05,
+      "loss": 0.1665,
+      "step": 1425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9351699175010082e-05,
+      "loss": 0.1433,
+      "step": 1426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350557758589195e-05,
+      "loss": 0.1477,
+      "step": 1427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934941537196934e-05,
+      "loss": 0.1492,
+      "step": 1428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348272015269045e-05,
+      "loss": 0.1641,
+      "step": 1429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934712768860694e-05,
+      "loss": 0.1284,
+      "step": 1430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934598239210176e-05,
+      "loss": 0.1504,
+      "step": 1431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9344836125872345e-05,
+      "loss": 0.1545,
+      "step": 1432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934368889003762e-05,
+      "loss": 0.1359,
+      "step": 1433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934254068471663e-05,
+      "loss": 0.1206,
+      "step": 1434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93413915100285e-05,
+      "loss": 0.1202,
+      "step": 1435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340241366092475e-05,
+      "loss": 0.1334,
+      "step": 1436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9339090253027893e-05,
+      "loss": 0.1493,
+      "step": 1437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337938170954184e-05,
+      "loss": 0.1299,
+      "step": 1438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9336785119990894e-05,
+      "loss": 0.1836,
+      "step": 1439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9335631100257655e-05,
+      "loss": 0.1289,
+      "step": 1440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933447611187421e-05,
+      "loss": 0.1379,
+      "step": 1441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333320154960403e-05,
+      "loss": 0.13,
+      "step": 1442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9332163229636168e-05,
+      "loss": 0.1389,
+      "step": 1443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331005336021543e-05,
+      "loss": 0.1256,
+      "step": 1444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9329846474236676e-05,
+      "loss": 0.1367,
+      "step": 1445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328686644401807e-05,
+      "loss": 0.1343,
+      "step": 1446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327525846637277e-05,
+      "loss": 0.1195,
+      "step": 1447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932636408106353e-05,
+      "loss": 0.1555,
+      "step": 1448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325201347801105e-05,
+      "loss": 0.1338,
+      "step": 1449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932403764697065e-05,
+      "loss": 0.1528,
+      "step": 1450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322872978692907e-05,
+      "loss": 0.1486,
+      "step": 1451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932170734308872e-05,
+      "loss": 0.1188,
+      "step": 1452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320540740279035e-05,
+      "loss": 0.1392,
+      "step": 1453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9319373170384895e-05,
+      "loss": 0.1229,
+      "step": 1454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318204633527442e-05,
+      "loss": 0.1191,
+      "step": 1455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317035129827925e-05,
+      "loss": 0.1155,
+      "step": 1456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315864659407696e-05,
+      "loss": 0.141,
+      "step": 1457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931469322238819e-05,
+      "loss": 0.1167,
+      "step": 1458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313520818890957e-05,
+      "loss": 0.1263,
+      "step": 1459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931234744903765e-05,
+      "loss": 0.1201,
+      "step": 1460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931117311295001e-05,
+      "loss": 0.1311,
+      "step": 1461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9309997810749883e-05,
+      "loss": 0.132,
+      "step": 1462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930882154255922e-05,
+      "loss": 0.1382,
+      "step": 1463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930764430850007e-05,
+      "loss": 0.1448,
+      "step": 1464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306466108694574e-05,
+      "loss": 0.1538,
+      "step": 1465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930528694326499e-05,
+      "loss": 0.1484,
+      "step": 1466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9304106812333657e-05,
+      "loss": 0.1266,
+      "step": 1467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9302925716023025e-05,
+      "loss": 0.1477,
+      "step": 1468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301743654455652e-05,
+      "loss": 0.1636,
+      "step": 1469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300560627754176e-05,
+      "loss": 0.1321,
+      "step": 1470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929937663604135e-05,
+      "loss": 0.142,
+      "step": 1471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298191679440024e-05,
+      "loss": 0.1543,
+      "step": 1472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9297005758073145e-05,
+      "loss": 0.1472,
+      "step": 1473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929581887206377e-05,
+      "loss": 0.1355,
+      "step": 1474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929463102153503e-05,
+      "loss": 0.152,
+      "step": 1475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293442206610196e-05,
+      "loss": 0.1384,
+      "step": 1476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292252427412605e-05,
+      "loss": 0.1193,
+      "step": 1477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929106168406571e-05,
+      "loss": 0.1287,
+      "step": 1478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928986997669306e-05,
+      "loss": 0.1304,
+      "step": 1479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9288677305418306e-05,
+      "loss": 0.1327,
+      "step": 1480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287483670365193e-05,
+      "loss": 0.145,
+      "step": 1481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286289071657576e-05,
+      "loss": 0.1353,
+      "step": 1482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285093509419404e-05,
+      "loss": 0.1216,
+      "step": 1483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283896983774727e-05,
+      "loss": 0.1506,
+      "step": 1484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928269949484769e-05,
+      "loss": 0.1212,
+      "step": 1485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928150104276255e-05,
+      "loss": 0.1436,
+      "step": 1486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280301627643647e-05,
+      "loss": 0.1329,
+      "step": 1487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927910124961544e-05,
+      "loss": 0.1289,
+      "step": 1488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927789990880247e-05,
+      "loss": 0.1323,
+      "step": 1489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9276697605329392e-05,
+      "loss": 0.1017,
+      "step": 1490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275494339320954e-05,
+      "loss": 0.1473,
+      "step": 1491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9274290110902002e-05,
+      "loss": 0.1494,
+      "step": 1492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273084920197488e-05,
+      "loss": 0.1504,
+      "step": 1493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271878767332457e-05,
+      "loss": 0.1316,
+      "step": 1494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9270671652432063e-05,
+      "loss": 0.155,
+      "step": 1495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9269463575621552e-05,
+      "loss": 0.1333,
+      "step": 1496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926825453702627e-05,
+      "loss": 0.1371,
+      "step": 1497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9267044536771667e-05,
+      "loss": 0.1254,
+      "step": 1498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926583357498329e-05,
+      "loss": 0.1068,
+      "step": 1499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9264621651786787e-05,
+      "loss": 0.1274,
+      "step": 1500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263408767307903e-05,
+      "loss": 0.1477,
+      "step": 1501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926219492167249e-05,
+      "loss": 0.1663,
+      "step": 1502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926098011500649e-05,
+      "loss": 0.1606,
+      "step": 1503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925976434743595e-05,
+      "loss": 0.1389,
+      "step": 1504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9258547619087017e-05,
+      "loss": 0.1376,
+      "step": 1505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925732993008594e-05,
+      "loss": 0.1323,
+      "step": 1506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925611128055906e-05,
+      "loss": 0.1609,
+      "step": 1507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254891670632823e-05,
+      "loss": 0.1523,
+      "step": 1508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9253671100433773e-05,
+      "loss": 0.1055,
+      "step": 1509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252449570088555e-05,
+      "loss": 0.114,
+      "step": 1510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9251227079723917e-05,
+      "loss": 0.1406,
+      "step": 1511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250003629466696e-05,
+      "loss": 0.1389,
+      "step": 1512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248779219443838e-05,
+      "loss": 0.1211,
+      "step": 1513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924755384978239e-05,
+      "loss": 0.1422,
+      "step": 1514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9246327520609488e-05,
+      "loss": 0.1273,
+      "step": 1515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9245100232052377e-05,
+      "loss": 0.1305,
+      "step": 1516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.92438719842384e-05,
+      "loss": 0.1417,
+      "step": 1517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242642777294996e-05,
+      "loss": 0.1276,
+      "step": 1518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241412611349704e-05,
+      "loss": 0.1592,
+      "step": 1519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9240181486530166e-05,
+      "loss": 0.141,
+      "step": 1520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923894940296412e-05,
+      "loss": 0.1626,
+      "step": 1521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923771636077941e-05,
+      "loss": 0.135,
+      "step": 1522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923648236010397e-05,
+      "loss": 0.1465,
+      "step": 1523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923524740106584e-05,
+      "loss": 0.1326,
+      "step": 1524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234011483793153e-05,
+      "loss": 0.1309,
+      "step": 1525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232774608414153e-05,
+      "loss": 0.1326,
+      "step": 1526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231536775057172e-05,
+      "loss": 0.1278,
+      "step": 1527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923029798385064e-05,
+      "loss": 0.1337,
+      "step": 1528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9229058234923104e-05,
+      "loss": 0.1263,
+      "step": 1529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9227817528403194e-05,
+      "loss": 0.1082,
+      "step": 1530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226575864419635e-05,
+      "loss": 0.1431,
+      "step": 1531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9225333243101275e-05,
+      "loss": 0.1372,
+      "step": 1532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9224089664577034e-05,
+      "loss": 0.1499,
+      "step": 1533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9222845128975947e-05,
+      "loss": 0.1237,
+      "step": 1534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922159963642715e-05,
+      "loss": 0.145,
+      "step": 1535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220353187059865e-05,
+      "loss": 0.1433,
+      "step": 1536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921910578100343e-05,
+      "loss": 0.1541,
+      "step": 1537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921785741838727e-05,
+      "loss": 0.1459,
+      "step": 1538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921660809934091e-05,
+      "loss": 0.1472,
+      "step": 1539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215357823993985e-05,
+      "loss": 0.1448,
+      "step": 1540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214106592476215e-05,
+      "loss": 0.132,
+      "step": 1541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9212854404917424e-05,
+      "loss": 0.1675,
+      "step": 1542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211601261447544e-05,
+      "loss": 0.147,
+      "step": 1543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210347162196598e-05,
+      "loss": 0.1377,
+      "step": 1544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9209092107294704e-05,
+      "loss": 0.1506,
+      "step": 1545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207836096872093e-05,
+      "loss": 0.1235,
+      "step": 1546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9206579131059076e-05,
+      "loss": 0.1497,
+      "step": 1547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205321209986083e-05,
+      "loss": 0.1521,
+      "step": 1548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920406233378363e-05,
+      "loss": 0.123,
+      "step": 1549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202802502582334e-05,
+      "loss": 0.1449,
+      "step": 1550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9201541716512917e-05,
+      "loss": 0.1334,
+      "step": 1551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200279975706192e-05,
+      "loss": 0.1393,
+      "step": 1552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919901728029308e-05,
+      "loss": 0.1251,
+      "step": 1553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9197753630404595e-05,
+      "loss": 0.1417,
+      "step": 1554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196489026171846e-05,
+      "loss": 0.1558,
+      "step": 1555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9195223467726056e-05,
+      "loss": 0.1301,
+      "step": 1556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919395695519853e-05,
+      "loss": 0.1547,
+      "step": 1557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919268948872068e-05,
+      "loss": 0.1359,
+      "step": 1558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191421068424017e-05,
+      "loss": 0.1333,
+      "step": 1559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.919015169444015e-05,
+      "loss": 0.1511,
+      "step": 1560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9188881366900788e-05,
+      "loss": 0.1345,
+      "step": 1561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918761008593774e-05,
+      "loss": 0.1307,
+      "step": 1562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918633785168291e-05,
+      "loss": 0.1345,
+      "step": 1563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91850646642683e-05,
+      "loss": 0.1335,
+      "step": 1564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183790523826022e-05,
+      "loss": 0.1443,
+      "step": 1565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9182515430488272e-05,
+      "loss": 0.1281,
+      "step": 1566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9181239384387355e-05,
+      "loss": 0.1415,
+      "step": 1567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9179962385655665e-05,
+      "loss": 0.1107,
+      "step": 1568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9178684434425712e-05,
+      "loss": 0.1232,
+      "step": 1569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9177405530830087e-05,
+      "loss": 0.134,
+      "step": 1570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176125675001487e-05,
+      "loss": 0.142,
+      "step": 1571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9174844867072712e-05,
+      "loss": 0.1409,
+      "step": 1572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917356310717665e-05,
+      "loss": 0.1479,
+      "step": 1573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91722803954463e-05,
+      "loss": 0.1357,
+      "step": 1574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170996732014756e-05,
+      "loss": 0.1471,
+      "step": 1575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91697121170152e-05,
+      "loss": 0.1255,
+      "step": 1576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916842655058093e-05,
+      "loss": 0.1364,
+      "step": 1577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916714003284533e-05,
+      "loss": 0.1263,
+      "step": 1578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916585256394189e-05,
+      "loss": 0.1162,
+      "step": 1579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916456414400419e-05,
+      "loss": 0.1101,
+      "step": 1580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9163274773165923e-05,
+      "loss": 0.1301,
+      "step": 1581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161984451560867e-05,
+      "loss": 0.1312,
+      "step": 1582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.91606931793229e-05,
+      "loss": 0.1466,
+      "step": 1583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915940095658601e-05,
+      "loss": 0.1425,
+      "step": 1584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158107783484275e-05,
+      "loss": 0.1537,
+      "step": 1585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9156813660151866e-05,
+      "loss": 0.1364,
+      "step": 1586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155518586723066e-05,
+      "loss": 0.145,
+      "step": 1587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9154222563332245e-05,
+      "loss": 0.1479,
+      "step": 1588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9152925590113878e-05,
+      "loss": 0.1406,
+      "step": 1589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915162766720254e-05,
+      "loss": 0.1511,
+      "step": 1590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9150328794732896e-05,
+      "loss": 0.1215,
+      "step": 1591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149028972839718e-05,
+      "loss": 0.1479,
+      "step": 1592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914772820165787e-05,
+      "loss": 0.1438,
+      "step": 1593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9146426481322327e-05,
+      "loss": 0.1506,
+      "step": 1594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9145123811968145e-05,
+      "loss": 0.1274,
+      "step": 1595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9143820193730488e-05,
+      "loss": 0.1388,
+      "step": 1596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914251562674462e-05,
+      "loss": 0.1453,
+      "step": 1597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9141210111145898e-05,
+      "loss": 0.1161,
+      "step": 1598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913990364706978e-05,
+      "loss": 0.14,
+      "step": 1599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138596234651832e-05,
+      "loss": 0.142,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9137287874027696e-05,
+      "loss": 0.1385,
+      "step": 1601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913597856533313e-05,
+      "loss": 0.1412,
+      "step": 1602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9134668308703984e-05,
+      "loss": 0.1267,
+      "step": 1603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9133357104276218e-05,
+      "loss": 0.1356,
+      "step": 1604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913204495218587e-05,
+      "loss": 0.1561,
+      "step": 1605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9130731852569088e-05,
+      "loss": 0.1172,
+      "step": 1606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912941780556212e-05,
+      "loss": 0.1328,
+      "step": 1607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912810281130131e-05,
+      "loss": 0.1155,
+      "step": 1608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126786869923094e-05,
+      "loss": 0.1357,
+      "step": 1609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912546998156402e-05,
+      "loss": 0.1248,
+      "step": 1610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912415214636072e-05,
+      "loss": 0.1602,
+      "step": 1611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912283336444994e-05,
+      "loss": 0.1486,
+      "step": 1612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9121513635968497e-05,
+      "loss": 0.1362,
+      "step": 1613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912019296105334e-05,
+      "loss": 0.11,
+      "step": 1614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9118871339841495e-05,
+      "loss": 0.1405,
+      "step": 1615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117548772470093e-05,
+      "loss": 0.1338,
+      "step": 1616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116225259076354e-05,
+      "loss": 0.1365,
+      "step": 1617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114900799797612e-05,
+      "loss": 0.1373,
+      "step": 1618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113575394771287e-05,
+      "loss": 0.1317,
+      "step": 1619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9112249044134903e-05,
+      "loss": 0.1334,
+      "step": 1620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9110921748026076e-05,
+      "loss": 0.135,
+      "step": 1621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910959350658253e-05,
+      "loss": 0.1056,
+      "step": 1622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108264319942077e-05,
+      "loss": 0.1237,
+      "step": 1623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106934188242635e-05,
+      "loss": 0.1326,
+      "step": 1624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105603111622212e-05,
+      "loss": 0.1277,
+      "step": 1625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910427109021892e-05,
+      "loss": 0.1475,
+      "step": 1626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102938124170968e-05,
+      "loss": 0.1372,
+      "step": 1627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910160421361666e-05,
+      "loss": 0.1295,
+      "step": 1628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.91002693586944e-05,
+      "loss": 0.1169,
+      "step": 1629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098933559542702e-05,
+      "loss": 0.1398,
+      "step": 1630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909759681630015e-05,
+      "loss": 0.1376,
+      "step": 1631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096259129105453e-05,
+      "loss": 0.1359,
+      "step": 1632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90949204980974e-05,
+      "loss": 0.1022,
+      "step": 1633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909358092341489e-05,
+      "loss": 0.1255,
+      "step": 1634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909224040519692e-05,
+      "loss": 0.129,
+      "step": 1635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909089894358257e-05,
+      "loss": 0.1379,
+      "step": 1636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908955653871103e-05,
+      "loss": 0.1405,
+      "step": 1637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908821319072159e-05,
+      "loss": 0.1152,
+      "step": 1638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908686889975363e-05,
+      "loss": 0.1268,
+      "step": 1639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908552366594664e-05,
+      "loss": 0.1334,
+      "step": 1640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9084177489440187e-05,
+      "loss": 0.1323,
+      "step": 1641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082830370373954e-05,
+      "loss": 0.1454,
+      "step": 1642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081482308887716e-05,
+      "loss": 0.1273,
+      "step": 1643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080133305121343e-05,
+      "loss": 0.1185,
+      "step": 1644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9078783359214812e-05,
+      "loss": 0.1276,
+      "step": 1645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9077432471308182e-05,
+      "loss": 0.1339,
+      "step": 1646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9076080641541626e-05,
+      "loss": 0.1262,
+      "step": 1647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074727870055407e-05,
+      "loss": 0.1346,
+      "step": 1648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9073374156989888e-05,
+      "loss": 0.14,
+      "step": 1649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907201950248552e-05,
+      "loss": 0.1284,
+      "step": 1650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070663906682866e-05,
+      "loss": 0.1494,
+      "step": 1651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906930736972258e-05,
+      "loss": 0.1182,
+      "step": 1652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9067949891745413e-05,
+      "loss": 0.1431,
+      "step": 1653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066591472892216e-05,
+      "loss": 0.1179,
+      "step": 1654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065232113303934e-05,
+      "loss": 0.1541,
+      "step": 1655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9063871813121613e-05,
+      "loss": 0.1426,
+      "step": 1656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.90625105724864e-05,
+      "loss": 0.1298,
+      "step": 1657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9061148391539534e-05,
+      "loss": 0.1346,
+      "step": 1658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9059785270422342e-05,
+      "loss": 0.125,
+      "step": 1659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9058421209276272e-05,
+      "loss": 0.1266,
+      "step": 1660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905705620824285e-05,
+      "loss": 0.1454,
+      "step": 1661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9055690267463708e-05,
+      "loss": 0.1326,
+      "step": 1662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905432338708058e-05,
+      "loss": 0.1123,
+      "step": 1663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905295556723528e-05,
+      "loss": 0.1267,
+      "step": 1664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9051586808069737e-05,
+      "loss": 0.1255,
+      "step": 1665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9050217109725975e-05,
+      "loss": 0.1218,
+      "step": 1666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048846472346102e-05,
+      "loss": 0.1202,
+      "step": 1667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9047474896072342e-05,
+      "loss": 0.1284,
+      "step": 1668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046102381047003e-05,
+      "loss": 0.1343,
+      "step": 1669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9044728927412495e-05,
+      "loss": 0.1393,
+      "step": 1670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904335453531133e-05,
+      "loss": 0.1149,
+      "step": 1671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9041979204886107e-05,
+      "loss": 0.1207,
+      "step": 1672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904060293627953e-05,
+      "loss": 0.1331,
+      "step": 1673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90392257296344e-05,
+      "loss": 0.1169,
+      "step": 1674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903784758509361e-05,
+      "loss": 0.1215,
+      "step": 1675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903646850280016e-05,
+      "loss": 0.1149,
+      "step": 1676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9035088482897136e-05,
+      "loss": 0.1538,
+      "step": 1677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903370752552773e-05,
+      "loss": 0.1238,
+      "step": 1678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9032325630835227e-05,
+      "loss": 0.1387,
+      "step": 1679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030942798963007e-05,
+      "loss": 0.1385,
+      "step": 1680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029559030054558e-05,
+      "loss": 0.1471,
+      "step": 1681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902817432425345e-05,
+      "loss": 0.1345,
+      "step": 1682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9026788681703357e-05,
+      "loss": 0.1241,
+      "step": 1683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902540210254806e-05,
+      "loss": 0.1497,
+      "step": 1684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902401458693142e-05,
+      "loss": 0.1255,
+      "step": 1685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022626134997412e-05,
+      "loss": 0.1238,
+      "step": 1686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021236746890087e-05,
+      "loss": 0.1166,
+      "step": 1687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9019846422753615e-05,
+      "loss": 0.1268,
+      "step": 1688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018455162732256e-05,
+      "loss": 0.1278,
+      "step": 1689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9017062966970353e-05,
+      "loss": 0.1254,
+      "step": 1690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9015669835612375e-05,
+      "loss": 0.1393,
+      "step": 1691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9014275768802855e-05,
+      "loss": 0.1311,
+      "step": 1692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012880766686445e-05,
+      "loss": 0.1321,
+      "step": 1693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901148482940789e-05,
+      "loss": 0.1216,
+      "step": 1694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9010087957112032e-05,
+      "loss": 0.1548,
+      "step": 1695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.90086901499438e-05,
+      "loss": 0.1263,
+      "step": 1696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9007291408048238e-05,
+      "loss": 0.1141,
+      "step": 1697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900589173157047e-05,
+      "loss": 0.1278,
+      "step": 1698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900449112065573e-05,
+      "loss": 0.1257,
+      "step": 1699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900308957544934e-05,
+      "loss": 0.144,
+      "step": 1700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9001687096096724e-05,
+      "loss": 0.1265,
+      "step": 1701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000283682743394e-05,
+      "loss": 0.1256,
+      "step": 1702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8998879335534973e-05,
+      "loss": 0.111,
+      "step": 1703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997474054617177e-05,
+      "loss": 0.1416,
+      "step": 1704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996067840135804e-05,
+      "loss": 0.1311,
+      "step": 1705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899466069223677e-05,
+      "loss": 0.1211,
+      "step": 1706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8993252611066077e-05,
+      "loss": 0.1166,
+      "step": 1707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991843596769827e-05,
+      "loss": 0.1497,
+      "step": 1708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899043364949421e-05,
+      "loss": 0.1222,
+      "step": 1709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989022769385523e-05,
+      "loss": 0.1144,
+      "step": 1710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898761095659016e-05,
+      "loss": 0.1526,
+      "step": 1711
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8986198211254604e-05,
+      "loss": 0.1271,
+      "step": 1712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8984784533525445e-05,
+      "loss": 0.1101,
+      "step": 1713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8983369923549357e-05,
+      "loss": 0.1178,
+      "step": 1714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981954381473122e-05,
+      "loss": 0.1489,
+      "step": 1715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898053790744361e-05,
+      "loss": 0.1078,
+      "step": 1716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979120501607803e-05,
+      "loss": 0.1045,
+      "step": 1717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977702164112757e-05,
+      "loss": 0.1395,
+      "step": 1718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976282895105642e-05,
+      "loss": 0.1166,
+      "step": 1719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8974862694733716e-05,
+      "loss": 0.1379,
+      "step": 1720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973441563144338e-05,
+      "loss": 0.1119,
+      "step": 1721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8972019500484964e-05,
+      "loss": 0.1187,
+      "step": 1722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970596506903144e-05,
+      "loss": 0.1305,
+      "step": 1723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8969172582546528e-05,
+      "loss": 0.1338,
+      "step": 1724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896774772756285e-05,
+      "loss": 0.1417,
+      "step": 1725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8966321942099967e-05,
+      "loss": 0.11,
+      "step": 1726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964895226305802e-05,
+      "loss": 0.1379,
+      "step": 1727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963467580328397e-05,
+      "loss": 0.1301,
+      "step": 1728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8962039004315876e-05,
+      "loss": 0.1161,
+      "step": 1729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896060949841647e-05,
+      "loss": 0.1288,
+      "step": 1730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959179062778503e-05,
+      "loss": 0.1313,
+      "step": 1731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957747697550394e-05,
+      "loss": 0.1193,
+      "step": 1732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8956315402880655e-05,
+      "loss": 0.127,
+      "step": 1733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8954882178917903e-05,
+      "loss": 0.1267,
+      "step": 1734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8953448025810847e-05,
+      "loss": 0.118,
+      "step": 1735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895201294370829e-05,
+      "loss": 0.1371,
+      "step": 1736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950576932759138e-05,
+      "loss": 0.1379,
+      "step": 1737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8949139993112386e-05,
+      "loss": 0.1326,
+      "step": 1738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8947702124917126e-05,
+      "loss": 0.1176,
+      "step": 1739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8946263328322555e-05,
+      "loss": 0.1315,
+      "step": 1740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894482360347796e-05,
+      "loss": 0.1437,
+      "step": 1741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943382950532713e-05,
+      "loss": 0.1338,
+      "step": 1742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894194136963631e-05,
+      "loss": 0.1316,
+      "step": 1743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8940498860938315e-05,
+      "loss": 0.1145,
+      "step": 1744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8939055424588407e-05,
+      "loss": 0.1212,
+      "step": 1745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937611060736355e-05,
+      "loss": 0.1296,
+      "step": 1746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893616576953202e-05,
+      "loss": 0.1353,
+      "step": 1747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8934719551125364e-05,
+      "loss": 0.1016,
+      "step": 1748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8933272405666444e-05,
+      "loss": 0.1337,
+      "step": 1749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931824333305418e-05,
+      "loss": 0.1219,
+      "step": 1750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893037533419253e-05,
+      "loss": 0.1157,
+      "step": 1751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8928925408478128e-05,
+      "loss": 0.1063,
+      "step": 1752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927474556312656e-05,
+      "loss": 0.1289,
+      "step": 1753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8926022777846647e-05,
+      "loss": 0.1205,
+      "step": 1754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892457007323074e-05,
+      "loss": 0.1201,
+      "step": 1755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923116442615666e-05,
+      "loss": 0.1211,
+      "step": 1756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921661886152248e-05,
+      "loss": 0.1288,
+      "step": 1757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8920206403991407e-05,
+      "loss": 0.1171,
+      "step": 1758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918749996284167e-05,
+      "loss": 0.1273,
+      "step": 1759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917292663181638e-05,
+      "loss": 0.1239,
+      "step": 1760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915834404835036e-05,
+      "loss": 0.1458,
+      "step": 1761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891437522139566e-05,
+      "loss": 0.1356,
+      "step": 1762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912915113014918e-05,
+      "loss": 0.1455,
+      "step": 1763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8911454079844305e-05,
+      "loss": 0.1241,
+      "step": 1764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890999212203542e-05,
+      "loss": 0.1526,
+      "step": 1765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908529239739946e-05,
+      "loss": 0.1372,
+      "step": 1766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8907065433109676e-05,
+      "loss": 0.1256,
+      "step": 1767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8905600702296495e-05,
+      "loss": 0.1317,
+      "step": 1768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890413504745237e-05,
+      "loss": 0.1329,
+      "step": 1769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902668468729385e-05,
+      "loss": 0.1472,
+      "step": 1770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8901200966279707e-05,
+      "loss": 0.1157,
+      "step": 1771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.88997325402556e-05,
+      "loss": 0.1193,
+      "step": 1772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889826319080943e-05,
+      "loss": 0.1292,
+      "step": 1773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8896792918093645e-05,
+      "loss": 0.1182,
+      "step": 1774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895321722260806e-05,
+      "loss": 0.1226,
+      "step": 1775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889384960346356e-05,
+      "loss": 0.1331,
+      "step": 1776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892376561854653e-05,
+      "loss": 0.1219,
+      "step": 1777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890902597586926e-05,
+      "loss": 0.1,
+      "step": 1778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8889427710813308e-05,
+      "loss": 0.1193,
+      "step": 1779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8887951901686842e-05,
+      "loss": 0.1228,
+      "step": 1780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8886475170360644e-05,
+      "loss": 0.1382,
+      "step": 1781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888499751698795e-05,
+      "loss": 0.1232,
+      "step": 1782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8883518941722065e-05,
+      "loss": 0.1453,
+      "step": 1783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882039444716417e-05,
+      "loss": 0.1558,
+      "step": 1784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880559026124507e-05,
+      "loss": 0.1283,
+      "step": 1785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8879077686099944e-05,
+      "loss": 0.1246,
+      "step": 1786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8877595424796425e-05,
+      "loss": 0.1107,
+      "step": 1787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876112242367758e-05,
+      "loss": 0.1453,
+      "step": 1788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8874628138967827e-05,
+      "loss": 0.12,
+      "step": 1789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887314311475062e-05,
+      "loss": 0.1085,
+      "step": 1790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871657169870228e-05,
+      "loss": 0.1406,
+      "step": 1791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8870170304480825e-05,
+      "loss": 0.1193,
+      "step": 1792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886868251873668e-05,
+      "loss": 0.1383,
+      "step": 1793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8867193812792174e-05,
+      "loss": 0.1375,
+      "step": 1794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886570418680177e-05,
+      "loss": 0.1393,
+      "step": 1795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864213640920023e-05,
+      "loss": 0.1235,
+      "step": 1796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8862722175301595e-05,
+      "loss": 0.1229,
+      "step": 1797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861229790101238e-05,
+      "loss": 0.137,
+      "step": 1798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.88597364854738e-05,
+      "loss": 0.1244,
+      "step": 1799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8858242261574216e-05,
+      "loss": 0.1294,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885674711855754e-05,
+      "loss": 0.1263,
+      "step": 1801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885525105657889e-05,
+      "loss": 0.1362,
+      "step": 1802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88537540757935e-05,
+      "loss": 0.1117,
+      "step": 1803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852256176356704e-05,
+      "loss": 0.1388,
+      "step": 1804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8850757358423907e-05,
+      "loss": 0.1455,
+      "step": 1805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884925762215063e-05,
+      "loss": 0.108,
+      "step": 1806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8847756967692488e-05,
+      "loss": 0.1268,
+      "step": 1807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884625539520518e-05,
+      "loss": 0.1035,
+      "step": 1808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884475290484451e-05,
+      "loss": 0.1472,
+      "step": 1809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884324949676637e-05,
+      "loss": 0.1158,
+      "step": 1810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8841745171126757e-05,
+      "loss": 0.1321,
+      "step": 1811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8840239928081755e-05,
+      "loss": 0.1222,
+      "step": 1812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8838733767787543e-05,
+      "loss": 0.1176,
+      "step": 1813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88372266904004e-05,
+      "loss": 0.1196,
+      "step": 1814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88357186960767e-05,
+      "loss": 0.1224,
+      "step": 1815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8834209784972905e-05,
+      "loss": 0.1215,
+      "step": 1816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8832699957245585e-05,
+      "loss": 0.1356,
+      "step": 1817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883118921305139e-05,
+      "loss": 0.1304,
+      "step": 1818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8829677552547073e-05,
+      "loss": 0.1046,
+      "step": 1819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8828164975889486e-05,
+      "loss": 0.1498,
+      "step": 1820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882665148323557e-05,
+      "loss": 0.1254,
+      "step": 1821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8825137074742358e-05,
+      "loss": 0.1088,
+      "step": 1822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882362175056699e-05,
+      "loss": 0.1506,
+      "step": 1823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8822105510866686e-05,
+      "loss": 0.1298,
+      "step": 1824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8820588355798776e-05,
+      "loss": 0.1094,
+      "step": 1825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819070285520673e-05,
+      "loss": 0.111,
+      "step": 1826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817551300189893e-05,
+      "loss": 0.1351,
+      "step": 1827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8816031399964037e-05,
+      "loss": 0.1078,
+      "step": 1828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8814510585000813e-05,
+      "loss": 0.1124,
+      "step": 1829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8812988855458013e-05,
+      "loss": 0.1182,
+      "step": 1830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881146621149354e-05,
+      "loss": 0.1235,
+      "step": 1831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880994265326537e-05,
+      "loss": 0.118,
+      "step": 1832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880841818093159e-05,
+      "loss": 0.1182,
+      "step": 1833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880689279465038e-05,
+      "loss": 0.1182,
+      "step": 1834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8805366494580002e-05,
+      "loss": 0.141,
+      "step": 1835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8803839280878827e-05,
+      "loss": 0.0977,
+      "step": 1836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8802311153705324e-05,
+      "loss": 0.1223,
+      "step": 1837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800782113218038e-05,
+      "loss": 0.1221,
+      "step": 1838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799252159575627e-05,
+      "loss": 0.1196,
+      "step": 1839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879772129293683e-05,
+      "loss": 0.1239,
+      "step": 1840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796189513460495e-05,
+      "loss": 0.1415,
+      "step": 1841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879465682130555e-05,
+      "loss": 0.109,
+      "step": 1842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8793123216631032e-05,
+      "loss": 0.097,
+      "step": 1843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8791588699596057e-05,
+      "loss": 0.1003,
+      "step": 1844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879005327035985e-05,
+      "loss": 0.1307,
+      "step": 1845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878851692908172e-05,
+      "loss": 0.1163,
+      "step": 1846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878697967592108e-05,
+      "loss": 0.1007,
+      "step": 1847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8785441511037434e-05,
+      "loss": 0.1266,
+      "step": 1848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878390243459037e-05,
+      "loss": 0.1172,
+      "step": 1849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782362446739594e-05,
+      "loss": 0.1206,
+      "step": 1850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8780821547644882e-05,
+      "loss": 0.114,
+      "step": 1851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877927973746612e-05,
+      "loss": 0.1086,
+      "step": 1852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877773701636328e-05,
+      "loss": 0.1233,
+      "step": 1853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8776193384496436e-05,
+      "loss": 0.1296,
+      "step": 1854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8774648842025752e-05,
+      "loss": 0.1234,
+      "step": 1855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773103389111486e-05,
+      "loss": 0.1443,
+      "step": 1856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771557025913995e-05,
+      "loss": 0.1364,
+      "step": 1857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770009752593723e-05,
+      "loss": 0.1346,
+      "step": 1858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8768461569311215e-05,
+      "loss": 0.1094,
+      "step": 1859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8766912476227105e-05,
+      "loss": 0.1042,
+      "step": 1860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876536247350213e-05,
+      "loss": 0.1348,
+      "step": 1861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876381156129711e-05,
+      "loss": 0.0995,
+      "step": 1862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876225973977297e-05,
+      "loss": 0.1418,
+      "step": 1863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876070700909072e-05,
+      "loss": 0.1344,
+      "step": 1864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875915336941147e-05,
+      "loss": 0.1227,
+      "step": 1865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8757598820896427e-05,
+      "loss": 0.1404,
+      "step": 1866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875604336370689e-05,
+      "loss": 0.1436,
+      "step": 1867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875448699800424e-05,
+      "loss": 0.1439,
+      "step": 1868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.875292972394997e-05,
+      "loss": 0.1156,
+      "step": 1869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8751371541705663e-05,
+      "loss": 0.1263,
+      "step": 1870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874981245143299e-05,
+      "loss": 0.118,
+      "step": 1871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8748252453293717e-05,
+      "loss": 0.1281,
+      "step": 1872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8746691547449713e-05,
+      "loss": 0.1287,
+      "step": 1873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8745129734062934e-05,
+      "loss": 0.139,
+      "step": 1874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8743567013295427e-05,
+      "loss": 0.1372,
+      "step": 1875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8742003385309337e-05,
+      "loss": 0.1619,
+      "step": 1876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8740438850266907e-05,
+      "loss": 0.1279,
+      "step": 1877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738873408330475e-05,
+      "loss": 0.108,
+      "step": 1878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8737307059662463e-05,
+      "loss": 0.1138,
+      "step": 1879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873573980442539e-05,
+      "loss": 0.1154,
+      "step": 1880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8734171642781877e-05,
+      "loss": 0.1312,
+      "step": 1881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873260257489463e-05,
+      "loss": 0.1082,
+      "step": 1882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873103260092646e-05,
+      "loss": 0.1404,
+      "step": 1883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872946172104026e-05,
+      "loss": 0.1198,
+      "step": 1884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872788993539902e-05,
+      "loss": 0.1211,
+      "step": 1885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872631724416583e-05,
+      "loss": 0.116,
+      "step": 1886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8724743647503865e-05,
+      "loss": 0.1254,
+      "step": 1887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8723169145576404e-05,
+      "loss": 0.126,
+      "step": 1888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721593738546815e-05,
+      "loss": 0.1115,
+      "step": 1889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872001742657856e-05,
+      "loss": 0.1365,
+      "step": 1890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8718440209835187e-05,
+      "loss": 0.1404,
+      "step": 1891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8716862088480353e-05,
+      "loss": 0.1144,
+      "step": 1892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.87152830626778e-05,
+      "loss": 0.1273,
+      "step": 1893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713703132591365e-05,
+      "loss": 0.1373,
+      "step": 1894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712122298384977e-05,
+      "loss": 0.1184,
+      "step": 1895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871054056022266e-05,
+      "loss": 0.1277,
+      "step": 1896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870895791826854e-05,
+      "loss": 0.1365,
+      "step": 1897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707374372686825e-05,
+      "loss": 0.1193,
+      "step": 1898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8705789923641815e-05,
+      "loss": 0.1228,
+      "step": 1899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870420457129792e-05,
+      "loss": 0.1299,
+      "step": 1900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702618315819628e-05,
+      "loss": 0.1365,
+      "step": 1901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8701031157371523e-05,
+      "loss": 0.1241,
+      "step": 1902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699443096118294e-05,
+      "loss": 0.1316,
+      "step": 1903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697854132224713e-05,
+      "loss": 0.1196,
+      "step": 1904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8696264265855647e-05,
+      "loss": 0.1176,
+      "step": 1905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869467349717606e-05,
+      "loss": 0.1233,
+      "step": 1906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8693081826351002e-05,
+      "loss": 0.1166,
+      "step": 1907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869148925354563e-05,
+      "loss": 0.1135,
+      "step": 1908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8689895778925185e-05,
+      "loss": 0.1432,
+      "step": 1909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8688301402654995e-05,
+      "loss": 0.1188,
+      "step": 1910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8686706124900502e-05,
+      "loss": 0.1212,
+      "step": 1911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868510994582722e-05,
+      "loss": 0.126,
+      "step": 1912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868351286560077e-05,
+      "loss": 0.1294,
+      "step": 1913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868191488438687e-05,
+      "loss": 0.1273,
+      "step": 1914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680316002351308e-05,
+      "loss": 0.1036,
+      "step": 1915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678716219659992e-05,
+      "loss": 0.1285,
+      "step": 1916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8677115536478917e-05,
+      "loss": 0.1119,
+      "step": 1917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8675513952974156e-05,
+      "loss": 0.1282,
+      "step": 1918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867391146931189e-05,
+      "loss": 0.109,
+      "step": 1919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8672308085658395e-05,
+      "loss": 0.1274,
+      "step": 1920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.867070380218003e-05,
+      "loss": 0.1278,
+      "step": 1921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866909861904326e-05,
+      "loss": 0.1176,
+      "step": 1922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667492536414627e-05,
+      "loss": 0.1166,
+      "step": 1923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8665885554460784e-05,
+      "loss": 0.1069,
+      "step": 1924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664277673348463e-05,
+      "loss": 0.1384,
+      "step": 1925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.86626688932445e-05,
+      "loss": 0.1406,
+      "step": 1926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866105921431581e-05,
+      "loss": 0.1245,
+      "step": 1927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8659448636729426e-05,
+      "loss": 0.1154,
+      "step": 1928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657837160652447e-05,
+      "loss": 0.1172,
+      "step": 1929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656224786252077e-05,
+      "loss": 0.1191,
+      "step": 1930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8654611513695622e-05,
+      "loss": 0.1213,
+      "step": 1931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8652997343150466e-05,
+      "loss": 0.1309,
+      "step": 1932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8651382274784095e-05,
+      "loss": 0.1173,
+      "step": 1933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8649766308764085e-05,
+      "loss": 0.1219,
+      "step": 1934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8648149445258104e-05,
+      "loss": 0.1172,
+      "step": 1935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646531684433924e-05,
+      "loss": 0.1083,
+      "step": 1936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864491302645939e-05,
+      "loss": 0.1247,
+      "step": 1937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8643293471502458e-05,
+      "loss": 0.1118,
+      "step": 1938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864167301973117e-05,
+      "loss": 0.1097,
+      "step": 1939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640051671313656e-05,
+      "loss": 0.1345,
+      "step": 1940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863842942641815e-05,
+      "loss": 0.137,
+      "step": 1941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8636806285212975e-05,
+      "loss": 0.1331,
+      "step": 1942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8635182247866545e-05,
+      "loss": 0.118,
+      "step": 1943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863355731454736e-05,
+      "loss": 0.1224,
+      "step": 1944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863193148542403e-05,
+      "loss": 0.1299,
+      "step": 1945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8630304760665237e-05,
+      "loss": 0.1289,
+      "step": 1946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8628677140439784e-05,
+      "loss": 0.1212,
+      "step": 1947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627048624916532e-05,
+      "loss": 0.1389,
+      "step": 1948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862541921426447e-05,
+      "loss": 0.116,
+      "step": 1949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8623788908652653e-05,
+      "loss": 0.1036,
+      "step": 1950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862215770825024e-05,
+      "loss": 0.092,
+      "step": 1951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862052561322648e-05,
+      "loss": 0.1438,
+      "step": 1952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618892623750723e-05,
+      "loss": 0.13,
+      "step": 1953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86172587399924e-05,
+      "loss": 0.1255,
+      "step": 1954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8615623962121043e-05,
+      "loss": 0.1262,
+      "step": 1955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8613988290306274e-05,
+      "loss": 0.1184,
+      "step": 1956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861235172471781e-05,
+      "loss": 0.1108,
+      "step": 1957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861071426552545e-05,
+      "loss": 0.108,
+      "step": 1958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8609075912899096e-05,
+      "loss": 0.1357,
+      "step": 1959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8607436667008748e-05,
+      "loss": 0.1316,
+      "step": 1960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860579652802449e-05,
+      "loss": 0.1204,
+      "step": 1961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8604155496116495e-05,
+      "loss": 0.1373,
+      "step": 1962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860251357145504e-05,
+      "loss": 0.1392,
+      "step": 1963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8600870754210477e-05,
+      "loss": 0.1248,
+      "step": 1964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8599227044553276e-05,
+      "loss": 0.1204,
+      "step": 1965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859758244265398e-05,
+      "loss": 0.1227,
+      "step": 1966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8595936948683234e-05,
+      "loss": 0.1276,
+      "step": 1967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8594290562811762e-05,
+      "loss": 0.1144,
+      "step": 1968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85926432852104e-05,
+      "loss": 0.1199,
+      "step": 1969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859099511605006e-05,
+      "loss": 0.1113,
+      "step": 1970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589346055501757e-05,
+      "loss": 0.1229,
+      "step": 1971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85876961037366e-05,
+      "loss": 0.1437,
+      "step": 1972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586045260925773e-05,
+      "loss": 0.1108,
+      "step": 1973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8584393527240576e-05,
+      "loss": 0.1144,
+      "step": 1974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8582740902852385e-05,
+      "loss": 0.136,
+      "step": 1975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581087387932676e-05,
+      "loss": 0.1113,
+      "step": 1976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8579432982653013e-05,
+      "loss": 0.105,
+      "step": 1977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577777687185054e-05,
+      "loss": 0.1008,
+      "step": 1978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576121501700553e-05,
+      "loss": 0.1272,
+      "step": 1979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857446442637135e-05,
+      "loss": 0.1019,
+      "step": 1980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8572806461369383e-05,
+      "loss": 0.1173,
+      "step": 1981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8571147606866677e-05,
+      "loss": 0.1484,
+      "step": 1982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569487863035355e-05,
+      "loss": 0.1112,
+      "step": 1983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856782723004763e-05,
+      "loss": 0.1191,
+      "step": 1984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85661657080758e-05,
+      "loss": 0.1338,
+      "step": 1985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564503297292267e-05,
+      "loss": 0.1196,
+      "step": 1986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562839997869523e-05,
+      "loss": 0.1176,
+      "step": 1987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8561175809980144e-05,
+      "loss": 0.1265,
+      "step": 1988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559510733796807e-05,
+      "loss": 0.1132,
+      "step": 1989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557844769492272e-05,
+      "loss": 0.1464,
+      "step": 1990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8556177917239406e-05,
+      "loss": 0.1213,
+      "step": 1991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8554510177211155e-05,
+      "loss": 0.1271,
+      "step": 1992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8552841549580555e-05,
+      "loss": 0.1248,
+      "step": 1993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8551172034520746e-05,
+      "loss": 0.1068,
+      "step": 1994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8549501632204953e-05,
+      "loss": 0.1287,
+      "step": 1995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8547830342806493e-05,
+      "loss": 0.1161,
+      "step": 1996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8546158166498783e-05,
+      "loss": 0.1296,
+      "step": 1997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8544485103455317e-05,
+      "loss": 0.1066,
+      "step": 1998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8542811153849692e-05,
+      "loss": 0.1061,
+      "step": 1999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8541136317855598e-05,
+      "loss": 0.1008,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8539460595646804e-05,
+      "loss": 0.1133,
+      "step": 2001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537783987397193e-05,
+      "loss": 0.1375,
+      "step": 2002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853610649328072e-05,
+      "loss": 0.1333,
+      "step": 2003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8534428113471437e-05,
+      "loss": 0.1143,
+      "step": 2004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8532748848143493e-05,
+      "loss": 0.1266,
+      "step": 2005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8531068697471125e-05,
+      "loss": 0.1121,
+      "step": 2006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8529387661628667e-05,
+      "loss": 0.1382,
+      "step": 2007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8527705740790532e-05,
+      "loss": 0.1278,
+      "step": 2008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526022935131244e-05,
+      "loss": 0.1298,
+      "step": 2009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85243392448254e-05,
+      "loss": 0.1254,
+      "step": 2010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522654670047702e-05,
+      "loss": 0.1212,
+      "step": 2011
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8520969210972932e-05,
+      "loss": 0.1085,
+      "step": 2012
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851928286777598e-05,
+      "loss": 0.1239,
+      "step": 2013
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851759564063181e-05,
+      "loss": 0.1121,
+      "step": 2014
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8515907529715492e-05,
+      "loss": 0.1255,
+      "step": 2015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8514218535202175e-05,
+      "loss": 0.1221,
+      "step": 2016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512528657267114e-05,
+      "loss": 0.1167,
+      "step": 2017
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8510837896085642e-05,
+      "loss": 0.1138,
+      "step": 2018
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8509146251833193e-05,
+      "loss": 0.1084,
+      "step": 2019
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850745372468529e-05,
+      "loss": 0.1222,
+      "step": 2020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8505760314817544e-05,
+      "loss": 0.1125,
+      "step": 2021
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8504066022405663e-05,
+      "loss": 0.1195,
+      "step": 2022
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8502370847625442e-05,
+      "loss": 0.1029,
+      "step": 2023
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.850067479065277e-05,
+      "loss": 0.1033,
+      "step": 2024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849897785166363e-05,
+      "loss": 0.1241,
+      "step": 2025
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849728003083409e-05,
+      "loss": 0.1278,
+      "step": 2026
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8495581328340315e-05,
+      "loss": 0.1117,
+      "step": 2027
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8493881744358558e-05,
+      "loss": 0.1182,
+      "step": 2028
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8492181279065166e-05,
+      "loss": 0.1389,
+      "step": 2029
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849047993263658e-05,
+      "loss": 0.1079,
+      "step": 2030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8488777705249324e-05,
+      "loss": 0.1042,
+      "step": 2031
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8487074597080023e-05,
+      "loss": 0.0966,
+      "step": 2032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8485370608305384e-05,
+      "loss": 0.1121,
+      "step": 2033
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848366573910221e-05,
+      "loss": 0.1235,
+      "step": 2034
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84819599896474e-05,
+      "loss": 0.1013,
+      "step": 2035
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848025336011794e-05,
+      "loss": 0.1145,
+      "step": 2036
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8478545850690902e-05,
+      "loss": 0.1324,
+      "step": 2037
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847683746154346e-05,
+      "loss": 0.1381,
+      "step": 2038
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847512819285287e-05,
+      "loss": 0.135,
+      "step": 2039
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8473418044796484e-05,
+      "loss": 0.1144,
+      "step": 2040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8471707017551743e-05,
+      "loss": 0.1135,
+      "step": 2041
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469995111296183e-05,
+      "loss": 0.111,
+      "step": 2042
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8468282326207426e-05,
+      "loss": 0.1283,
+      "step": 2043
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846656866246319e-05,
+      "loss": 0.1259,
+      "step": 2044
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846485412024128e-05,
+      "loss": 0.1133,
+      "step": 2045
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.84631386997196e-05,
+      "loss": 0.1068,
+      "step": 2046
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846142240107613e-05,
+      "loss": 0.1217,
+      "step": 2047
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8459705224488958e-05,
+      "loss": 0.1302,
+      "step": 2048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8457987170136254e-05,
+      "loss": 0.1342,
+      "step": 2049
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8456268238196277e-05,
+      "loss": 0.1362,
+      "step": 2050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454548428847383e-05,
+      "loss": 0.1112,
+      "step": 2051
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8452827742268017e-05,
+      "loss": 0.1194,
+      "step": 2052
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8451106178636714e-05,
+      "loss": 0.1187,
+      "step": 2053
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.84493837381321e-05,
+      "loss": 0.0948,
+      "step": 2054
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8447660420932895e-05,
+      "loss": 0.1367,
+      "step": 2055
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8445936227217904e-05,
+      "loss": 0.0923,
+      "step": 2056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844421115716603e-05,
+      "loss": 0.1334,
+      "step": 2057
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8442485210956263e-05,
+      "loss": 0.1279,
+      "step": 2058
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8440758388767685e-05,
+      "loss": 0.0961,
+      "step": 2059
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439030690779468e-05,
+      "loss": 0.0958,
+      "step": 2060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843730211717087e-05,
+      "loss": 0.1411,
+      "step": 2061
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8435572668121252e-05,
+      "loss": 0.1261,
+      "step": 2062
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8433842343810058e-05,
+      "loss": 0.119,
+      "step": 2063
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.843211114441682e-05,
+      "loss": 0.1036,
+      "step": 2064
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8430379070121163e-05,
+      "loss": 0.1132,
+      "step": 2065
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8428646121102815e-05,
+      "loss": 0.1426,
+      "step": 2066
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8426912297541574e-05,
+      "loss": 0.1377,
+      "step": 2067
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8425177599617342e-05,
+      "loss": 0.1245,
+      "step": 2068
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8423442027510104e-05,
+      "loss": 0.0919,
+      "step": 2069
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421705581399946e-05,
+      "loss": 0.1136,
+      "step": 2070
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8419968261467042e-05,
+      "loss": 0.1139,
+      "step": 2071
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418230067891644e-05,
+      "loss": 0.1172,
+      "step": 2072
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841649100085411e-05,
+      "loss": 0.1102,
+      "step": 2073
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8414751060534878e-05,
+      "loss": 0.1107,
+      "step": 2074
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413010247114492e-05,
+      "loss": 0.1127,
+      "step": 2075
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411268560773562e-05,
+      "loss": 0.0917,
+      "step": 2076
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8409526001692817e-05,
+      "loss": 0.1196,
+      "step": 2077
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840778257005305e-05,
+      "loss": 0.1053,
+      "step": 2078
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8406038266035163e-05,
+      "loss": 0.1088,
+      "step": 2079
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840429308982014e-05,
+      "loss": 0.1376,
+      "step": 2080
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8402547041589057e-05,
+      "loss": 0.1172,
+      "step": 2081
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8400800121523087e-05,
+      "loss": 0.1195,
+      "step": 2082
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.839905232980348e-05,
+      "loss": 0.1173,
+      "step": 2083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8397303666611588e-05,
+      "loss": 0.1084,
+      "step": 2084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8395554132128854e-05,
+      "loss": 0.1162,
+      "step": 2085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393803726536793e-05,
+      "loss": 0.1051,
+      "step": 2086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8392052450017036e-05,
+      "loss": 0.1281,
+      "step": 2087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390300302751292e-05,
+      "loss": 0.1183,
+      "step": 2088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838854728492136e-05,
+      "loss": 0.1198,
+      "step": 2089
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8386793396709123e-05,
+      "loss": 0.116,
+      "step": 2090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385038638296577e-05,
+      "loss": 0.1228,
+      "step": 2091
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8383283009865773e-05,
+      "loss": 0.131,
+      "step": 2092
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838152651159889e-05,
+      "loss": 0.144,
+      "step": 2093
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837976914367817e-05,
+      "loss": 0.1261,
+      "step": 2094
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378010906285958e-05,
+      "loss": 0.1172,
+      "step": 2095
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8376251799604684e-05,
+      "loss": 0.1229,
+      "step": 2096
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8374491823816872e-05,
+      "loss": 0.1152,
+      "step": 2097
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8372730979105133e-05,
+      "loss": 0.1213,
+      "step": 2098
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837096926565217e-05,
+      "loss": 0.1189,
+      "step": 2099
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369206683640776e-05,
+      "loss": 0.1079,
+      "step": 2100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8367443233253833e-05,
+      "loss": 0.1085,
+      "step": 2101
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836567891467431e-05,
+      "loss": 0.1129,
+      "step": 2102
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8363913728085277e-05,
+      "loss": 0.1244,
+      "step": 2103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8362147673669884e-05,
+      "loss": 0.1216,
+      "step": 2104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8360380751611375e-05,
+      "loss": 0.1088,
+      "step": 2105
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835861296209308e-05,
+      "loss": 0.1254,
+      "step": 2106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8356844305298424e-05,
+      "loss": 0.1306,
+      "step": 2107
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355074781410918e-05,
+      "loss": 0.1237,
+      "step": 2108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835330439061417e-05,
+      "loss": 0.1119,
+      "step": 2109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8351533133091867e-05,
+      "loss": 0.1095,
+      "step": 2110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8349761009027794e-05,
+      "loss": 0.1274,
+      "step": 2111
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8347988018605826e-05,
+      "loss": 0.1227,
+      "step": 2112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8346214162009924e-05,
+      "loss": 0.1259,
+      "step": 2113
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8344439439424142e-05,
+      "loss": 0.099,
+      "step": 2114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834266385103262e-05,
+      "loss": 0.1224,
+      "step": 2115
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834088739701959e-05,
+      "loss": 0.1375,
+      "step": 2116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8339110077569376e-05,
+      "loss": 0.1379,
+      "step": 2117
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337331892866384e-05,
+      "loss": 0.1077,
+      "step": 2118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8335552843095128e-05,
+      "loss": 0.1166,
+      "step": 2119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8333772928440187e-05,
+      "loss": 0.1364,
+      "step": 2120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833199214908625e-05,
+      "loss": 0.0975,
+      "step": 2121
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8330210505218084e-05,
+      "loss": 0.1125,
+      "step": 2122
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832842799702055e-05,
+      "loss": 0.1228,
+      "step": 2123
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83266446246786e-05,
+      "loss": 0.0974,
+      "step": 2124
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832486038837727e-05,
+      "loss": 0.0933,
+      "step": 2125
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8323075288301693e-05,
+      "loss": 0.114,
+      "step": 2126
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321289324637087e-05,
+      "loss": 0.1288,
+      "step": 2127
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831950249756876e-05,
+      "loss": 0.1287,
+      "step": 2128
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8317714807282115e-05,
+      "loss": 0.1179,
+      "step": 2129
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8315926253962636e-05,
+      "loss": 0.1194,
+      "step": 2130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8314136837795894e-05,
+      "loss": 0.118,
+      "step": 2131
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831234655896757e-05,
+      "loss": 0.1163,
+      "step": 2132
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831055541766341e-05,
+      "loss": 0.1124,
+      "step": 2133
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830876341406926e-05,
+      "loss": 0.126,
+      "step": 2134
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8306970548371062e-05,
+      "loss": 0.1119,
+      "step": 2135
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8305176820754838e-05,
+      "loss": 0.1215,
+      "step": 2136
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.83033822314067e-05,
+      "loss": 0.1199,
+      "step": 2137
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830158678051285e-05,
+      "loss": 0.1224,
+      "step": 2138
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829979046825959e-05,
+      "loss": 0.1306,
+      "step": 2139
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8297993294833294e-05,
+      "loss": 0.1057,
+      "step": 2140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8296195260420438e-05,
+      "loss": 0.1097,
+      "step": 2141
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8294396365207584e-05,
+      "loss": 0.1254,
+      "step": 2142
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8292596609381378e-05,
+      "loss": 0.1078,
+      "step": 2143
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829079599312856e-05,
+      "loss": 0.0977,
+      "step": 2144
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288994516635963e-05,
+      "loss": 0.1074,
+      "step": 2145
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8287192180090505e-05,
+      "loss": 0.1196,
+      "step": 2146
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8285388983679192e-05,
+      "loss": 0.0986,
+      "step": 2147
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8283584927589123e-05,
+      "loss": 0.1151,
+      "step": 2148
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828178001200748e-05,
+      "loss": 0.1044,
+      "step": 2149
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279974237121537e-05,
+      "loss": 0.1211,
+      "step": 2150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8278167603118667e-05,
+      "loss": 0.1176,
+      "step": 2151
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8276360110186315e-05,
+      "loss": 0.1155,
+      "step": 2152
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8274551758512026e-05,
+      "loss": 0.0986,
+      "step": 2153
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8272742548283434e-05,
+      "loss": 0.103,
+      "step": 2154
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8270932479688255e-05,
+      "loss": 0.1193,
+      "step": 2155
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8269121552914307e-05,
+      "loss": 0.1016,
+      "step": 2156
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8267309768149482e-05,
+      "loss": 0.114,
+      "step": 2157
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826549712558177e-05,
+      "loss": 0.1271,
+      "step": 2158
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263683625399244e-05,
+      "loss": 0.1145,
+      "step": 2159
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8261869267790077e-05,
+      "loss": 0.1077,
+      "step": 2160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826005405294252e-05,
+      "loss": 0.1287,
+      "step": 2161
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8258237981044915e-05,
+      "loss": 0.0964,
+      "step": 2162
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8256421052285698e-05,
+      "loss": 0.1051,
+      "step": 2163
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825460326685339e-05,
+      "loss": 0.0979,
+      "step": 2164
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82527846249366e-05,
+      "loss": 0.1045,
+      "step": 2165
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825096512672403e-05,
+      "loss": 0.1213,
+      "step": 2166
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8249144772404467e-05,
+      "loss": 0.1003,
+      "step": 2167
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8247323562166785e-05,
+      "loss": 0.1189,
+      "step": 2168
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8245501496199954e-05,
+      "loss": 0.1198,
+      "step": 2169
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8243678574693026e-05,
+      "loss": 0.1259,
+      "step": 2170
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824185479783515e-05,
+      "loss": 0.1108,
+      "step": 2171
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8240030165815546e-05,
+      "loss": 0.1349,
+      "step": 2172
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823820467882355e-05,
+      "loss": 0.1277,
+      "step": 2173
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8236378337048562e-05,
+      "loss": 0.1205,
+      "step": 2174
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234551140680083e-05,
+      "loss": 0.1145,
+      "step": 2175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8232723089907694e-05,
+      "loss": 0.1061,
+      "step": 2176
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823089418492108e-05,
+      "loss": 0.1067,
+      "step": 2177
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8229064425910005e-05,
+      "loss": 0.1233,
+      "step": 2178
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8227233813064318e-05,
+      "loss": 0.1195,
+      "step": 2179
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8225402346573958e-05,
+      "loss": 0.1165,
+      "step": 2180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8223570026628958e-05,
+      "loss": 0.1323,
+      "step": 2181
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8221736853419437e-05,
+      "loss": 0.0938,
+      "step": 2182
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82199028271356e-05,
+      "loss": 0.1299,
+      "step": 2183
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821806794796775e-05,
+      "loss": 0.1145,
+      "step": 2184
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821623221610626e-05,
+      "loss": 0.1039,
+      "step": 2185
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821439563174161e-05,
+      "loss": 0.1085,
+      "step": 2186
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821255819506436e-05,
+      "loss": 0.1367,
+      "step": 2187
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8210719906265155e-05,
+      "loss": 0.1056,
+      "step": 2188
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820888076553474e-05,
+      "loss": 0.1252,
+      "step": 2189
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8207040773063938e-05,
+      "loss": 0.1245,
+      "step": 2190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205199929043664e-05,
+      "loss": 0.123,
+      "step": 2191
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8203358233664915e-05,
+      "loss": 0.1006,
+      "step": 2192
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201515687118795e-05,
+      "loss": 0.115,
+      "step": 2193
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8199672289596473e-05,
+      "loss": 0.1097,
+      "step": 2194
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819782804128922e-05,
+      "loss": 0.113,
+      "step": 2195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8195982942388397e-05,
+      "loss": 0.1202,
+      "step": 2196
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819413699308544e-05,
+      "loss": 0.1022,
+      "step": 2197
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819229019357189e-05,
+      "loss": 0.1294,
+      "step": 2198
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819044254403936e-05,
+      "loss": 0.1078,
+      "step": 2199
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8188594044679566e-05,
+      "loss": 0.1243,
+      "step": 2200
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.81867446956843e-05,
+      "loss": 0.1196,
+      "step": 2201
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8184894497245457e-05,
+      "loss": 0.1106,
+      "step": 2202
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8183043449554998e-05,
+      "loss": 0.1174,
+      "step": 2203
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818119155280499e-05,
+      "loss": 0.1243,
+      "step": 2204
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8179338807187587e-05,
+      "loss": 0.1206,
+      "step": 2205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8177485212895022e-05,
+      "loss": 0.1052,
+      "step": 2206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817563077011962e-05,
+      "loss": 0.1138,
+      "step": 2207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.81737754790538e-05,
+      "loss": 0.1215,
+      "step": 2208
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817191933989006e-05,
+      "loss": 0.1017,
+      "step": 2209
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170062352820993e-05,
+      "loss": 0.1479,
+      "step": 2210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168204518039273e-05,
+      "loss": 0.1104,
+      "step": 2211
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8166345835737674e-05,
+      "loss": 0.132,
+      "step": 2212
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816448630610904e-05,
+      "loss": 0.1321,
+      "step": 2213
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8162625929346314e-05,
+      "loss": 0.1099,
+      "step": 2214
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8160764705642534e-05,
+      "loss": 0.1104,
+      "step": 2215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8158902635190812e-05,
+      "loss": 0.1245,
+      "step": 2216
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815703971818435e-05,
+      "loss": 0.12,
+      "step": 2217
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8155175954816453e-05,
+      "loss": 0.119,
+      "step": 2218
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815331134528049e-05,
+      "loss": 0.1141,
+      "step": 2219
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8151445889769934e-05,
+      "loss": 0.1139,
+      "step": 2220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8149579588478345e-05,
+      "loss": 0.1221,
+      "step": 2221
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.814771244159936e-05,
+      "loss": 0.1072,
+      "step": 2222
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8145844449326718e-05,
+      "loss": 0.1093,
+      "step": 2223
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8143975611854235e-05,
+      "loss": 0.1013,
+      "step": 2224
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8142105929375823e-05,
+      "loss": 0.116,
+      "step": 2225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814023540208547e-05,
+      "loss": 0.1467,
+      "step": 2226
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813836403017727e-05,
+      "loss": 0.1194,
+      "step": 2227
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813649181384538e-05,
+      "loss": 0.0844,
+      "step": 2228
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813461875328407e-05,
+      "loss": 0.0885,
+      "step": 2229
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132744848687677e-05,
+      "loss": 0.1196,
+      "step": 2230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8130870100250643e-05,
+      "loss": 0.1068,
+      "step": 2231
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812899450816748e-05,
+      "loss": 0.1205,
+      "step": 2232
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8127118072632805e-05,
+      "loss": 0.1072,
+      "step": 2233
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8125240793841304e-05,
+      "loss": 0.137,
+      "step": 2234
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812336267198777e-05,
+      "loss": 0.1041,
+      "step": 2235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812148370726707e-05,
+      "loss": 0.1067,
+      "step": 2236
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8119603899874163e-05,
+      "loss": 0.1354,
+      "step": 2237
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811772325000409e-05,
+      "loss": 0.1002,
+      "step": 2238
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.811584175785199e-05,
+      "loss": 0.105,
+      "step": 2239
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113959423613084e-05,
+      "loss": 0.1229,
+      "step": 2240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112076247482678e-05,
+      "loss": 0.1005,
+      "step": 2241
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110192229656168e-05,
+      "loss": 0.114,
+      "step": 2242
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8108307370329032e-05,
+      "loss": 0.1041,
+      "step": 2243
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8106421669696846e-05,
+      "loss": 0.1343,
+      "step": 2244
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8104535127955265e-05,
+      "loss": 0.1006,
+      "step": 2245
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810264774530004e-05,
+      "loss": 0.1021,
+      "step": 2246
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8100759521926987e-05,
+      "loss": 0.1345,
+      "step": 2247
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809887045803204e-05,
+      "loss": 0.118,
+      "step": 2248
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.80969805538112e-05,
+      "loss": 0.1021,
+      "step": 2249
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809508980946056e-05,
+      "loss": 0.1123,
+      "step": 2250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8093198225176302e-05,
+      "loss": 0.106,
+      "step": 2251
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809130580115469e-05,
+      "loss": 0.1149,
+      "step": 2252
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8089412537592085e-05,
+      "loss": 0.1394,
+      "step": 2253
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8087518434684927e-05,
+      "loss": 0.106,
+      "step": 2254
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808562349262974e-05,
+      "loss": 0.1299,
+      "step": 2255
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083727711623145e-05,
+      "loss": 0.137,
+      "step": 2256
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8081831091861843e-05,
+      "loss": 0.1313,
+      "step": 2257
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807993363354263e-05,
+      "loss": 0.1152,
+      "step": 2258
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8078035336862375e-05,
+      "loss": 0.1208,
+      "step": 2259
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8076136202018045e-05,
+      "loss": 0.1257,
+      "step": 2260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8074236229206694e-05,
+      "loss": 0.1254,
+      "step": 2261
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807233541862546e-05,
+      "loss": 0.1206,
+      "step": 2262
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8070433770471566e-05,
+      "loss": 0.1161,
+      "step": 2263
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068531284942324e-05,
+      "loss": 0.109,
+      "step": 2264
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806662796223513e-05,
+      "loss": 0.1196,
+      "step": 2265
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8064723802547478e-05,
+      "loss": 0.0825,
+      "step": 2266
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8062818806076934e-05,
+      "loss": 0.1372,
+      "step": 2267
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8060912973021158e-05,
+      "loss": 0.1127,
+      "step": 2268
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80590063035779e-05,
+      "loss": 0.1071,
+      "step": 2269
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8057098797944987e-05,
+      "loss": 0.1272,
+      "step": 2270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8055190456320344e-05,
+      "loss": 0.121,
+      "step": 2271
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8053281278901976e-05,
+      "loss": 0.0969,
+      "step": 2272
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805137126588797e-05,
+      "loss": 0.1145,
+      "step": 2273
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804946041747652e-05,
+      "loss": 0.1099,
+      "step": 2274
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8047548733865877e-05,
+      "loss": 0.0969,
+      "step": 2275
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8045636215254407e-05,
+      "loss": 0.1123,
+      "step": 2276
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804372286184054e-05,
+      "loss": 0.1229,
+      "step": 2277
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8041808673822806e-05,
+      "loss": 0.1274,
+      "step": 2278
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8039893651399823e-05,
+      "loss": 0.1119,
+      "step": 2279
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8037977794770285e-05,
+      "loss": 0.1254,
+      "step": 2280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803606110413298e-05,
+      "loss": 0.0975,
+      "step": 2281
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803414357968678e-05,
+      "loss": 0.0984,
+      "step": 2282
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8032225221630645e-05,
+      "loss": 0.1326,
+      "step": 2283
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803030603016362e-05,
+      "loss": 0.1208,
+      "step": 2284
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8028386005484837e-05,
+      "loss": 0.1044,
+      "step": 2285
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8026465147793516e-05,
+      "loss": 0.1212,
+      "step": 2286
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024543457288962e-05,
+      "loss": 0.1234,
+      "step": 2287
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8022620934170568e-05,
+      "loss": 0.1008,
+      "step": 2288
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802069757863781e-05,
+      "loss": 0.1158,
+      "step": 2289
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801877339089025e-05,
+      "loss": 0.1134,
+      "step": 2290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801684837112754e-05,
+      "loss": 0.0889,
+      "step": 2291
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8014922519549423e-05,
+      "loss": 0.1331,
+      "step": 2292
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8012995836355712e-05,
+      "loss": 0.1198,
+      "step": 2293
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801106832174633e-05,
+      "loss": 0.1021,
+      "step": 2294
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800913997592126e-05,
+      "loss": 0.1023,
+      "step": 2295
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8007210799080586e-05,
+      "loss": 0.094,
+      "step": 2296
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8005280791424483e-05,
+      "loss": 0.1178,
+      "step": 2297
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.80033499531532e-05,
+      "loss": 0.0985,
+      "step": 2298
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8001418284467077e-05,
+      "loss": 0.1161,
+      "step": 2299
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799948578556655e-05,
+      "loss": 0.1099,
+      "step": 2300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799755245665212e-05,
+      "loss": 0.1205,
+      "step": 2301
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799561829792439e-05,
+      "loss": 0.1113,
+      "step": 2302
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799368330958405e-05,
+      "loss": 0.1006,
+      "step": 2303
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.799174749183187e-05,
+      "loss": 0.1185,
+      "step": 2304
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79898108448687e-05,
+      "loss": 0.1187,
+      "step": 2305
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7987873368895494e-05,
+      "loss": 0.1151,
+      "step": 2306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798593506411327e-05,
+      "loss": 0.113,
+      "step": 2307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7983995930723156e-05,
+      "loss": 0.1,
+      "step": 2308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7982055968926344e-05,
+      "loss": 0.0957,
+      "step": 2309
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7980115178924125e-05,
+      "loss": 0.1112,
+      "step": 2310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7978173560917872e-05,
+      "loss": 0.109,
+      "step": 2311
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797623111510904e-05,
+      "loss": 0.0958,
+      "step": 2312
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797428784169918e-05,
+      "loss": 0.1094,
+      "step": 2313
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7972343740889922e-05,
+      "loss": 0.1025,
+      "step": 2314
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7970398812882982e-05,
+      "loss": 0.1005,
+      "step": 2315
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796845305788016e-05,
+      "loss": 0.1014,
+      "step": 2316
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7966506476083353e-05,
+      "loss": 0.1158,
+      "step": 2317
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796455906769452e-05,
+      "loss": 0.1085,
+      "step": 2318
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7962610832915738e-05,
+      "loss": 0.1056,
+      "step": 2319
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7960661771949137e-05,
+      "loss": 0.1036,
+      "step": 2320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.795871188499696e-05,
+      "loss": 0.0999,
+      "step": 2321
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7956761172261518e-05,
+      "loss": 0.1113,
+      "step": 2322
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7954809633945222e-05,
+      "loss": 0.1287,
+      "step": 2323
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.795285727025055e-05,
+      "loss": 0.1112,
+      "step": 2324
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7950904081380082e-05,
+      "loss": 0.1411,
+      "step": 2325
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7948950067536475e-05,
+      "loss": 0.091,
+      "step": 2326
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7946995228922474e-05,
+      "loss": 0.1019,
+      "step": 2327
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7945039565740914e-05,
+      "loss": 0.1188,
+      "step": 2328
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7943083078194712e-05,
+      "loss": 0.0966,
+      "step": 2329
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7941125766486865e-05,
+      "loss": 0.108,
+      "step": 2330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7939167630820465e-05,
+      "loss": 0.0968,
+      "step": 2331
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7937208671398677e-05,
+      "loss": 0.1082,
+      "step": 2332
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793524888842477e-05,
+      "loss": 0.1206,
+      "step": 2333
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7933288282102084e-05,
+      "loss": 0.1045,
+      "step": 2334
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7931326852634044e-05,
+      "loss": 0.1162,
+      "step": 2335
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792936460022417e-05,
+      "loss": 0.1094,
+      "step": 2336
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7927401525076066e-05,
+      "loss": 0.1252,
+      "step": 2337
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792543762739341e-05,
+      "loss": 0.1002,
+      "step": 2338
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7923472907379968e-05,
+      "loss": 0.106,
+      "step": 2339
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.792150736523961e-05,
+      "loss": 0.1223,
+      "step": 2340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791954100117627e-05,
+      "loss": 0.1139,
+      "step": 2341
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7917573815393975e-05,
+      "loss": 0.105,
+      "step": 2342
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791560580809684e-05,
+      "loss": 0.1072,
+      "step": 2343
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.791363697948906e-05,
+      "loss": 0.1147,
+      "step": 2344
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911667329774914e-05,
+      "loss": 0.1271,
+      "step": 2345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7909696859158778e-05,
+      "loss": 0.1179,
+      "step": 2346
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907725567845094e-05,
+      "loss": 0.1318,
+      "step": 2347
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790575345603841e-05,
+      "loss": 0.0885,
+      "step": 2348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7903780523943344e-05,
+      "loss": 0.1312,
+      "step": 2349
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790180677176461e-05,
+      "loss": 0.1039,
+      "step": 2350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899832199706993e-05,
+      "loss": 0.1244,
+      "step": 2351
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7897856807975377e-05,
+      "loss": 0.1143,
+      "step": 2352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7895880596774723e-05,
+      "loss": 0.1082,
+      "step": 2353
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789390356631008e-05,
+      "loss": 0.1147,
+      "step": 2354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7891925716786584e-05,
+      "loss": 0.1039,
+      "step": 2355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788994704840945e-05,
+      "loss": 0.1064,
+      "step": 2356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7887967561383986e-05,
+      "loss": 0.0959,
+      "step": 2357
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885987255915574e-05,
+      "loss": 0.124,
+      "step": 2358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7884006132209693e-05,
+      "loss": 0.0914,
+      "step": 2359
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.78820241904719e-05,
+      "loss": 0.111,
+      "step": 2360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7880041430907836e-05,
+      "loss": 0.1229,
+      "step": 2361
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787805785372323e-05,
+      "loss": 0.1016,
+      "step": 2362
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7876073459123895e-05,
+      "loss": 0.1233,
+      "step": 2363
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7874088247315727e-05,
+      "loss": 0.1157,
+      "step": 2364
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7872102218504715e-05,
+      "loss": 0.1056,
+      "step": 2365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7870115372896915e-05,
+      "loss": 0.1158,
+      "step": 2366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7868127710698488e-05,
+      "loss": 0.0977,
+      "step": 2367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786613923211567e-05,
+      "loss": 0.1078,
+      "step": 2368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786414993735478e-05,
+      "loss": 0.1083,
+      "step": 2369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786215982662222e-05,
+      "loss": 0.0968,
+      "step": 2370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786016890012449e-05,
+      "loss": 0.1041,
+      "step": 2371
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7858177158068154e-05,
+      "loss": 0.0906,
+      "step": 2372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856184600659886e-05,
+      "loss": 0.1075,
+      "step": 2373
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7854191228106414e-05,
+      "loss": 0.106,
+      "step": 2374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7852197040614583e-05,
+      "loss": 0.108,
+      "step": 2375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7850202038391297e-05,
+      "loss": 0.1072,
+      "step": 2376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784820622164356e-05,
+      "loss": 0.1041,
+      "step": 2377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784620959057845e-05,
+      "loss": 0.1036,
+      "step": 2378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7844212145403133e-05,
+      "loss": 0.1078,
+      "step": 2379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784221388632487e-05,
+      "loss": 0.1187,
+      "step": 2380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7840214813550986e-05,
+      "loss": 0.1078,
+      "step": 2381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783821492728891e-05,
+      "loss": 0.1116,
+      "step": 2382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7836214227746138e-05,
+      "loss": 0.1093,
+      "step": 2383
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783421271513027e-05,
+      "loss": 0.1162,
+      "step": 2384
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.783221038964897e-05,
+      "loss": 0.1208,
+      "step": 2385
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830207251510008e-05,
+      "loss": 0.1276,
+      "step": 2386
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7828203300921216e-05,
+      "loss": 0.1102,
+      "step": 2387
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.782619853809052e-05,
+      "loss": 0.1138,
+      "step": 2388
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824192963225938e-05,
+      "loss": 0.1146,
+      "step": 2389
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7822186576535566e-05,
+      "loss": 0.1327,
+      "step": 2390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7820179378227572e-05,
+      "loss": 0.116,
+      "step": 2391
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7818171368510234e-05,
+      "loss": 0.1288,
+      "step": 2392
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781616254759189e-05,
+      "loss": 0.131,
+      "step": 2393
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814152915680975e-05,
+      "loss": 0.1219,
+      "step": 2394
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812142472986007e-05,
+      "loss": 0.1052,
+      "step": 2395
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7810131219715585e-05,
+      "loss": 0.0913,
+      "step": 2396
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7808119156078392e-05,
+      "loss": 0.1019,
+      "step": 2397
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78061062822832e-05,
+      "loss": 0.0986,
+      "step": 2398
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804092598538857e-05,
+      "loss": 0.1387,
+      "step": 2399
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7802078105054305e-05,
+      "loss": 0.0995,
+      "step": 2400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780006280203856e-05,
+      "loss": 0.1237,
+      "step": 2401
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7798046689700728e-05,
+      "loss": 0.1621,
+      "step": 2402
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779602976825e-05,
+      "loss": 0.0994,
+      "step": 2403
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779401203789564e-05,
+      "loss": 0.1228,
+      "step": 2404
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7791993498847016e-05,
+      "loss": 0.0991,
+      "step": 2405
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7789974151313566e-05,
+      "loss": 0.1187,
+      "step": 2406
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7787953995504807e-05,
+      "loss": 0.1204,
+      "step": 2407
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778593303163035e-05,
+      "loss": 0.116,
+      "step": 2408
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7783911259899894e-05,
+      "loss": 0.1216,
+      "step": 2409
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7781888680523208e-05,
+      "loss": 0.1157,
+      "step": 2410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777986529371015e-05,
+      "loss": 0.1238,
+      "step": 2411
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777841099670672e-05,
+      "loss": 0.1387,
+      "step": 2412
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777581609861479e-05,
+      "loss": 0.1166,
+      "step": 2413
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7773790290752626e-05,
+      "loss": 0.116,
+      "step": 2414
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7771763676294368e-05,
+      "loss": 0.1187,
+      "step": 2415
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7769736255450292e-05,
+      "loss": 0.0999,
+      "step": 2416
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7767708028430767e-05,
+      "loss": 0.1089,
+      "step": 2417
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7765678995446233e-05,
+      "loss": 0.1226,
+      "step": 2418
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7763649156707222e-05,
+      "loss": 0.104,
+      "step": 2419
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7761618512424347e-05,
+      "loss": 0.1132,
+      "step": 2420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7759587062808302e-05,
+      "loss": 0.1151,
+      "step": 2421
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7757554808069867e-05,
+      "loss": 0.1211,
+      "step": 2422
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7755521748419912e-05,
+      "loss": 0.1071,
+      "step": 2423
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7753487884069375e-05,
+      "loss": 0.0957,
+      "step": 2424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751453215229292e-05,
+      "loss": 0.1351,
+      "step": 2425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7749417742110772e-05,
+      "loss": 0.1074,
+      "step": 2426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774738146492502e-05,
+      "loss": 0.119,
+      "step": 2427
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745344383883312e-05,
+      "loss": 0.1277,
+      "step": 2428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7743306499197014e-05,
+      "loss": 0.1088,
+      "step": 2429
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7741267811077573e-05,
+      "loss": 0.1182,
+      "step": 2430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7739228319736517e-05,
+      "loss": 0.1188,
+      "step": 2431
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7737188025385466e-05,
+      "loss": 0.1038,
+      "step": 2432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773514692823611e-05,
+      "loss": 0.1146,
+      "step": 2433
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773310502850024e-05,
+      "loss": 0.1027,
+      "step": 2434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7731062326389716e-05,
+      "loss": 0.1102,
+      "step": 2435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7729018822116482e-05,
+      "loss": 0.1074,
+      "step": 2436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726974515892573e-05,
+      "loss": 0.1157,
+      "step": 2437
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.77249294079301e-05,
+      "loss": 0.0945,
+      "step": 2438
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772288349844126e-05,
+      "loss": 0.0996,
+      "step": 2439
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772083678763834e-05,
+      "loss": 0.1195,
+      "step": 2440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718789275733694e-05,
+      "loss": 0.1044,
+      "step": 2441
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7716740962939772e-05,
+      "loss": 0.1265,
+      "step": 2442
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.771469184946911e-05,
+      "loss": 0.113,
+      "step": 2443
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.771264193553431e-05,
+      "loss": 0.1233,
+      "step": 2444
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7710591221348074e-05,
+      "loss": 0.0948,
+      "step": 2445
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770853970712318e-05,
+      "loss": 0.1287,
+      "step": 2446
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7706487393072492e-05,
+      "loss": 0.1262,
+      "step": 2447
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770443427940895e-05,
+      "loss": 0.1206,
+      "step": 2448
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7702380366345585e-05,
+      "loss": 0.0957,
+      "step": 2449
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770032565409551e-05,
+      "loss": 0.1028,
+      "step": 2450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7698270142871914e-05,
+      "loss": 0.0957,
+      "step": 2451
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696213832888074e-05,
+      "loss": 0.099,
+      "step": 2452
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7694156724357352e-05,
+      "loss": 0.1157,
+      "step": 2453
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7692098817493192e-05,
+      "loss": 0.1215,
+      "step": 2454
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7690040112509114e-05,
+      "loss": 0.118,
+      "step": 2455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7687980609618726e-05,
+      "loss": 0.1251,
+      "step": 2456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7685920309035723e-05,
+      "loss": 0.1174,
+      "step": 2457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7683859210973874e-05,
+      "loss": 0.1024,
+      "step": 2458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768179731564704e-05,
+      "loss": 0.1052,
+      "step": 2459
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679734623269156e-05,
+      "loss": 0.1115,
+      "step": 2460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767767113405425e-05,
+      "loss": 0.1163,
+      "step": 2461
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767560684821642e-05,
+      "loss": 0.1062,
+      "step": 2462
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767354176596985e-05,
+      "loss": 0.1201,
+      "step": 2463
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7671475887528817e-05,
+      "loss": 0.1127,
+      "step": 2464
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7669409213107674e-05,
+      "loss": 0.1079,
+      "step": 2465
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667341742920846e-05,
+      "loss": 0.1075,
+      "step": 2466
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7665273477182863e-05,
+      "loss": 0.1183,
+      "step": 2467
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7663204416108315e-05,
+      "loss": 0.1199,
+      "step": 2468
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766113455991189e-05,
+      "loss": 0.1262,
+      "step": 2469
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765906390880835e-05,
+      "loss": 0.1078,
+      "step": 2470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7656992463012548e-05,
+      "loss": 0.0933,
+      "step": 2471
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7654920222739403e-05,
+      "loss": 0.109,
+      "step": 2472
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652847188203938e-05,
+      "loss": 0.1174,
+      "step": 2473
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765077335962124e-05,
+      "loss": 0.1128,
+      "step": 2474
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7648698737206497e-05,
+      "loss": 0.1031,
+      "step": 2475
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764662332117496e-05,
+      "loss": 0.1151,
+      "step": 2476
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7644547111741968e-05,
+      "loss": 0.1051,
+      "step": 2477
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7642470109122954e-05,
+      "loss": 0.1146,
+      "step": 2478
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7640392313533416e-05,
+      "loss": 0.1042,
+      "step": 2479
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7638313725188948e-05,
+      "loss": 0.1111,
+      "step": 2480
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7636234344305217e-05,
+      "loss": 0.11,
+      "step": 2481
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7634154171097984e-05,
+      "loss": 0.127,
+      "step": 2482
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7632073205783076e-05,
+      "loss": 0.0896,
+      "step": 2483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762999144857642e-05,
+      "loss": 0.1172,
+      "step": 2484
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7627908899694005e-05,
+      "loss": 0.1058,
+      "step": 2485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7625825559351917e-05,
+      "loss": 0.1077,
+      "step": 2486
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762374142776632e-05,
+      "loss": 0.1082,
+      "step": 2487
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7621656505153466e-05,
+      "loss": 0.1302,
+      "step": 2488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619570791729676e-05,
+      "loss": 0.1177,
+      "step": 2489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761748428771136e-05,
+      "loss": 0.1261,
+      "step": 2490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761539699331502e-05,
+      "loss": 0.1169,
+      "step": 2491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613308908757215e-05,
+      "loss": 0.1001,
+      "step": 2492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7611220034254612e-05,
+      "loss": 0.1101,
+      "step": 2493
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760913037002395e-05,
+      "loss": 0.1128,
+      "step": 2494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7607039916282044e-05,
+      "loss": 0.1149,
+      "step": 2495
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7604948673245798e-05,
+      "loss": 0.0848,
+      "step": 2496
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7602856641132197e-05,
+      "loss": 0.1073,
+      "step": 2497
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7600763820158308e-05,
+      "loss": 0.1088,
+      "step": 2498
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7598670210541273e-05,
+      "loss": 0.1074,
+      "step": 2499
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759657581249833e-05,
+      "loss": 0.0929,
+      "step": 2500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7594480626246784e-05,
+      "loss": 0.114,
+      "step": 2501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7592384652004032e-05,
+      "loss": 0.1097,
+      "step": 2502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7590287889987544e-05,
+      "loss": 0.129,
+      "step": 2503
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7588190340414882e-05,
+      "loss": 0.111,
+      "step": 2504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758609200350368e-05,
+      "loss": 0.1176,
+      "step": 2505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7583992879471664e-05,
+      "loss": 0.1024,
+      "step": 2506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.758189296853663e-05,
+      "loss": 0.1129,
+      "step": 2507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7579792270916463e-05,
+      "loss": 0.111,
+      "step": 2508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757769078682913e-05,
+      "loss": 0.1135,
+      "step": 2509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7575588516492677e-05,
+      "loss": 0.108,
+      "step": 2510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7573485460125227e-05,
+      "loss": 0.1349,
+      "step": 2511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7571381617945e-05,
+      "loss": 0.0956,
+      "step": 2512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7569276990170276e-05,
+      "loss": 0.1041,
+      "step": 2513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7567171577019438e-05,
+      "loss": 0.0828,
+      "step": 2514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7565065378710934e-05,
+      "loss": 0.1012,
+      "step": 2515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75629583954633e-05,
+      "loss": 0.1091,
+      "step": 2516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7560850627495156e-05,
+      "loss": 0.1084,
+      "step": 2517
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75587420750252e-05,
+      "loss": 0.12,
+      "step": 2518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755663273827221e-05,
+      "loss": 0.1099,
+      "step": 2519
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755452261745505e-05,
+      "loss": 0.1176,
+      "step": 2520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7552411712792664e-05,
+      "loss": 0.0955,
+      "step": 2521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550300024504067e-05,
+      "loss": 0.097,
+      "step": 2522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754818755280838e-05,
+      "loss": 0.1003,
+      "step": 2523
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7546074297924774e-05,
+      "loss": 0.0985,
+      "step": 2524
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7543960260072522e-05,
+      "loss": 0.1348,
+      "step": 2525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754184543947098e-05,
+      "loss": 0.1021,
+      "step": 2526
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7539729836339573e-05,
+      "loss": 0.1067,
+      "step": 2527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753761345089781e-05,
+      "loss": 0.0895,
+      "step": 2528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7535496283365288e-05,
+      "loss": 0.1185,
+      "step": 2529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753337833396168e-05,
+      "loss": 0.1017,
+      "step": 2530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753125960290674e-05,
+      "loss": 0.1156,
+      "step": 2531
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7529140090420307e-05,
+      "loss": 0.1346,
+      "step": 2532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7527019796722296e-05,
+      "loss": 0.1365,
+      "step": 2533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7524898722032704e-05,
+      "loss": 0.1055,
+      "step": 2534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752277686657161e-05,
+      "loss": 0.1149,
+      "step": 2535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752065423055918e-05,
+      "loss": 0.127,
+      "step": 2536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751853081421565e-05,
+      "loss": 0.1086,
+      "step": 2537
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7516406617761342e-05,
+      "loss": 0.1196,
+      "step": 2538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7514281641416662e-05,
+      "loss": 0.1117,
+      "step": 2539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7512155885402095e-05,
+      "loss": 0.1044,
+      "step": 2540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75100293499382e-05,
+      "loss": 0.0997,
+      "step": 2541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7507902035245628e-05,
+      "loss": 0.1035,
+      "step": 2542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7505773941545108e-05,
+      "loss": 0.1128,
+      "step": 2543
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7503645069057444e-05,
+      "loss": 0.1226,
+      "step": 2544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7501515418003522e-05,
+      "loss": 0.1166,
+      "step": 2545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7499384988604316e-05,
+      "loss": 0.0897,
+      "step": 2546
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7497253781080876e-05,
+      "loss": 0.1125,
+      "step": 2547
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7495121795654326e-05,
+      "loss": 0.119,
+      "step": 2548
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7492989032545886e-05,
+      "loss": 0.0959,
+      "step": 2549
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7490855491976843e-05,
+      "loss": 0.0946,
+      "step": 2550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7488721174168573e-05,
+      "loss": 0.1224,
+      "step": 2551
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7486586079342523e-05,
+      "loss": 0.1235,
+      "step": 2552
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7484450207720236e-05,
+      "loss": 0.1166,
+      "step": 2553
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748231355952332e-05,
+      "loss": 0.1128,
+      "step": 2554
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7480176134973474e-05,
+      "loss": 0.1044,
+      "step": 2555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7478037934292468e-05,
+      "loss": 0.1202,
+      "step": 2556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475898957702167e-05,
+      "loss": 0.0962,
+      "step": 2557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.74737592054245e-05,
+      "loss": 0.1147,
+      "step": 2558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747161867768149e-05,
+      "loss": 0.1294,
+      "step": 2559
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746947737469523e-05,
+      "loss": 0.1024,
+      "step": 2560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7467335296687903e-05,
+      "loss": 0.0959,
+      "step": 2561
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7465192443881763e-05,
+      "loss": 0.1344,
+      "step": 2562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746304881649915e-05,
+      "loss": 0.1173,
+      "step": 2563
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7460904414762488e-05,
+      "loss": 0.1145,
+      "step": 2564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7458759238894273e-05,
+      "loss": 0.1051,
+      "step": 2565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745661328911708e-05,
+      "loss": 0.1152,
+      "step": 2566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745446656565358e-05,
+      "loss": 0.1072,
+      "step": 2567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7452319068726503e-05,
+      "loss": 0.1266,
+      "step": 2568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745017079855868e-05,
+      "loss": 0.1097,
+      "step": 2569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7448021755373005e-05,
+      "loss": 0.1255,
+      "step": 2570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7445871939392457e-05,
+      "loss": 0.111,
+      "step": 2571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443721350840106e-05,
+      "loss": 0.1057,
+      "step": 2572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7441569989939092e-05,
+      "loss": 0.1165,
+      "step": 2573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7439417856912632e-05,
+      "loss": 0.1123,
+      "step": 2574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7437264951984027e-05,
+      "loss": 0.1211,
+      "step": 2575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7435111275376668e-05,
+      "loss": 0.1089,
+      "step": 2576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432956827314004e-05,
+      "loss": 0.1184,
+      "step": 2577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.743080160801959e-05,
+      "loss": 0.111,
+      "step": 2578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742864561771704e-05,
+      "loss": 0.1115,
+      "step": 2579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7426488856630058e-05,
+      "loss": 0.1093,
+      "step": 2580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7424331324982425e-05,
+      "loss": 0.1089,
+      "step": 2581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742217302299801e-05,
+      "loss": 0.113,
+      "step": 2582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7420013950900743e-05,
+      "loss": 0.106,
+      "step": 2583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7417854108914656e-05,
+      "loss": 0.1093,
+      "step": 2584
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741569349726385e-05,
+      "loss": 0.1023,
+      "step": 2585
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.74135321161725e-05,
+      "loss": 0.1105,
+      "step": 2586
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7411369965864872e-05,
+      "loss": 0.1144,
+      "step": 2587
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409207046565306e-05,
+      "loss": 0.0919,
+      "step": 2588
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7407043358498227e-05,
+      "loss": 0.1228,
+      "step": 2589
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740487890188813e-05,
+      "loss": 0.114,
+      "step": 2590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7402713676959598e-05,
+      "loss": 0.1121,
+      "step": 2591
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740054768393729e-05,
+      "loss": 0.1167,
+      "step": 2592
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739838092304595e-05,
+      "loss": 0.089,
+      "step": 2593
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7396213394510393e-05,
+      "loss": 0.1331,
+      "step": 2594
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7394045098555522e-05,
+      "loss": 0.1187,
+      "step": 2595
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7391876035406312e-05,
+      "loss": 0.1113,
+      "step": 2596
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389706205287824e-05,
+      "loss": 0.1144,
+      "step": 2597
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7387535608425197e-05,
+      "loss": 0.1234,
+      "step": 2598
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7385364245043646e-05,
+      "loss": 0.0968,
+      "step": 2599
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738319211536847e-05,
+      "loss": 0.1011,
+      "step": 2600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738101921962505e-05,
+      "loss": 0.1039,
+      "step": 2601
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.737884555803883e-05,
+      "loss": 0.1008,
+      "step": 2602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7376671130835362e-05,
+      "loss": 0.0967,
+      "step": 2603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7374495938240246e-05,
+      "loss": 0.0891,
+      "step": 2604
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7372319980479185e-05,
+      "loss": 0.0988,
+      "step": 2605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.737014325777795e-05,
+      "loss": 0.1171,
+      "step": 2606
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7367965770362393e-05,
+      "loss": 0.1044,
+      "step": 2607
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7365787518458452e-05,
+      "loss": 0.1173,
+      "step": 2608
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363608502292136e-05,
+      "loss": 0.1173,
+      "step": 2609
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7361428722089532e-05,
+      "loss": 0.1169,
+      "step": 2610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7359248178076818e-05,
+      "loss": 0.1163,
+      "step": 2611
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735706687048024e-05,
+      "loss": 0.1106,
+      "step": 2612
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7354884799526127e-05,
+      "loss": 0.0996,
+      "step": 2613
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735270196544089e-05,
+      "loss": 0.1135,
+      "step": 2614
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735051836845101e-05,
+      "loss": 0.1036,
+      "step": 2615
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734833400878306e-05,
+      "loss": 0.118,
+      "step": 2616
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734614888666368e-05,
+      "loss": 0.1166,
+      "step": 2617
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7343963002319597e-05,
+      "loss": 0.1184,
+      "step": 2618
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734177635597762e-05,
+      "loss": 0.1105,
+      "step": 2619
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339588947864626e-05,
+      "loss": 0.1128,
+      "step": 2620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7337400778207578e-05,
+      "loss": 0.0952,
+      "step": 2621
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335211847233514e-05,
+      "loss": 0.1089,
+      "step": 2622
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7333022155169563e-05,
+      "loss": 0.1152,
+      "step": 2623
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.733083170224292e-05,
+      "loss": 0.1122,
+      "step": 2624
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7328640488680854e-05,
+      "loss": 0.1096,
+      "step": 2625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7326448514710733e-05,
+      "loss": 0.082,
+      "step": 2626
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7324255780559993e-05,
+      "loss": 0.1104,
+      "step": 2627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7322062286456142e-05,
+      "loss": 0.1139,
+      "step": 2628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7319868032626773e-05,
+      "loss": 0.1208,
+      "step": 2629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7317673019299566e-05,
+      "loss": 0.1134,
+      "step": 2630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7315477246702263e-05,
+      "loss": 0.0942,
+      "step": 2631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.73132807150627e-05,
+      "loss": 0.0989,
+      "step": 2632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7311083424608785e-05,
+      "loss": 0.0945,
+      "step": 2633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7308885375568505e-05,
+      "loss": 0.1177,
+      "step": 2634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7306686568169924e-05,
+      "loss": 0.1218,
+      "step": 2635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730448700264119e-05,
+      "loss": 0.0992,
+      "step": 2636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730228667921052e-05,
+      "loss": 0.1029,
+      "step": 2637
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7300085598106223e-05,
+      "loss": 0.1268,
+      "step": 2638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7297883759556676e-05,
+      "loss": 0.0973,
+      "step": 2639
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7295681163790343e-05,
+      "loss": 0.1238,
+      "step": 2640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7293477811035758e-05,
+      "loss": 0.105,
+      "step": 2641
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7291273701521534e-05,
+      "loss": 0.1178,
+      "step": 2642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728906883547637e-05,
+      "loss": 0.1013,
+      "step": 2643
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7286863213129045e-05,
+      "loss": 0.1108,
+      "step": 2644
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72846568347084e-05,
+      "loss": 0.1033,
+      "step": 2645
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.728244970044337e-05,
+      "loss": 0.1016,
+      "step": 2646
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7280241810562964e-05,
+      "loss": 0.1027,
+      "step": 2647
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7278033165296267e-05,
+      "loss": 0.0903,
+      "step": 2648
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727582376487245e-05,
+      "loss": 0.1344,
+      "step": 2649
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727361360952075e-05,
+      "loss": 0.0988,
+      "step": 2650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7271402699470498e-05,
+      "loss": 0.1239,
+      "step": 2651
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7269191034951086e-05,
+      "loss": 0.1167,
+      "step": 2652
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7266978616191996e-05,
+      "loss": 0.1091,
+      "step": 2653
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264765443422783e-05,
+      "loss": 0.118,
+      "step": 2654
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7262551516873086e-05,
+      "loss": 0.1134,
+      "step": 2655
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7260336836772617e-05,
+      "loss": 0.1117,
+      "step": 2656
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7258121403351168e-05,
+      "loss": 0.1125,
+      "step": 2657
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7255905216838607e-05,
+      "loss": 0.105,
+      "step": 2658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7253688277464884e-05,
+      "loss": 0.0991,
+      "step": 2659
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7251470585460026e-05,
+      "loss": 0.1077,
+      "step": 2660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7249252141054133e-05,
+      "loss": 0.1115,
+      "step": 2661
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724703294447739e-05,
+      "loss": 0.1063,
+      "step": 2662
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7244812995960056e-05,
+      "loss": 0.1044,
+      "step": 2663
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724259229573247e-05,
+      "loss": 0.1235,
+      "step": 2664
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.724037084402505e-05,
+      "loss": 0.1221,
+      "step": 2665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7238148641068292e-05,
+      "loss": 0.1035,
+      "step": 2666
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723592568709276e-05,
+      "loss": 0.1151,
+      "step": 2667
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7233701982329113e-05,
+      "loss": 0.1215,
+      "step": 2668
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231477527008074e-05,
+      "loss": 0.1095,
+      "step": 2669
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722925232136045e-05,
+      "loss": 0.105,
+      "step": 2670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227026365617124e-05,
+      "loss": 0.1086,
+      "step": 2671
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7224799660009064e-05,
+      "loss": 0.1074,
+      "step": 2672
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222572204767298e-05,
+      "loss": 0.1095,
+      "step": 2673
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220344000122954e-05,
+      "loss": 0.1172,
+      "step": 2674
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721811504630722e-05,
+      "loss": 0.1116,
+      "step": 2675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721588534355137e-05,
+      "loss": 0.1067,
+      "step": 2676
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721365489208676e-05,
+      "loss": 0.1115,
+      "step": 2677
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721142369214481e-05,
+      "loss": 0.1168,
+      "step": 2678
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7209191743957027e-05,
+      "loss": 0.1202,
+      "step": 2679
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7206959047755e-05,
+      "loss": 0.1189,
+      "step": 2680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7204725603770387e-05,
+      "loss": 0.1201,
+      "step": 2681
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7202491412234925e-05,
+      "loss": 0.1169,
+      "step": 2682
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720025647338043e-05,
+      "loss": 0.1208,
+      "step": 2683
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.71980207874388e-05,
+      "loss": 0.1221,
+      "step": 2684
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7195784354642004e-05,
+      "loss": 0.0802,
+      "step": 2685
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719354717522209e-05,
+      "loss": 0.1083,
+      "step": 2686
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.719130924941118e-05,
+      "loss": 0.1055,
+      "step": 2687
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7189070577441485e-05,
+      "loss": 0.1187,
+      "step": 2688
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7186831159545284e-05,
+      "loss": 0.1104,
+      "step": 2689
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718459099595493e-05,
+      "loss": 0.0931,
+      "step": 2690
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718235008690287e-05,
+      "loss": 0.0974,
+      "step": 2691
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7180108432621604e-05,
+      "loss": 0.0923,
+      "step": 2692
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717786603334373e-05,
+      "loss": 0.0953,
+      "step": 2693
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7175622889301916e-05,
+      "loss": 0.1267,
+      "step": 2694
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7173379000728906e-05,
+      "loss": 0.1431,
+      "step": 2695
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717113436785752e-05,
+      "loss": 0.0973,
+      "step": 2696
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716888899092066e-05,
+      "loss": 0.0918,
+      "step": 2697
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7166642870151303e-05,
+      "loss": 0.1125,
+      "step": 2698
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71643960057825e-05,
+      "loss": 0.1097,
+      "step": 2699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7162148398047386e-05,
+      "loss": 0.1107,
+      "step": 2700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7159900047179167e-05,
+      "loss": 0.0956,
+      "step": 2701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715765095341113e-05,
+      "loss": 0.1223,
+      "step": 2702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7155401116976634e-05,
+      "loss": 0.1113,
+      "step": 2703
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715315053810912e-05,
+      "loss": 0.1234,
+      "step": 2704
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.715089921704211e-05,
+      "loss": 0.1003,
+      "step": 2705
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7148647154009183e-05,
+      "loss": 0.1281,
+      "step": 2706
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7146394349244023e-05,
+      "loss": 0.1195,
+      "step": 2707
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7144140802980377e-05,
+      "loss": 0.0985,
+      "step": 2708
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141886515452065e-05,
+      "loss": 0.084,
+      "step": 2709
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7139631486892986e-05,
+      "loss": 0.1221,
+      "step": 2710
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7137375717537122e-05,
+      "loss": 0.0969,
+      "step": 2711
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7135119207618526e-05,
+      "loss": 0.1211,
+      "step": 2712
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7132861957371335e-05,
+      "loss": 0.1265,
+      "step": 2713
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713060396702975e-05,
+      "loss": 0.0806,
+      "step": 2714
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712834523682806e-05,
+      "loss": 0.1023,
+      "step": 2715
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712608576700063e-05,
+      "loss": 0.1069,
+      "step": 2716
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7123825557781894e-05,
+      "loss": 0.0831,
+      "step": 2717
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7121564609406372e-05,
+      "loss": 0.1035,
+      "step": 2718
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7119302922108655e-05,
+      "loss": 0.092,
+      "step": 2719
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7117040496123408e-05,
+      "loss": 0.1021,
+      "step": 2720
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7114777331685384e-05,
+      "loss": 0.0894,
+      "step": 2721
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7112513429029403e-05,
+      "loss": 0.1089,
+      "step": 2722
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7110248788390358e-05,
+      "loss": 0.1144,
+      "step": 2723
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.710798341000323e-05,
+      "loss": 0.1007,
+      "step": 2724
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710571729410307e-05,
+      "loss": 0.0958,
+      "step": 2725
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7103450440925013e-05,
+      "loss": 0.12,
+      "step": 2726
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710118285070425e-05,
+      "loss": 0.1075,
+      "step": 2727
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7098914523676075e-05,
+      "loss": 0.1262,
+      "step": 2728
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7096645460075837e-05,
+      "loss": 0.1073,
+      "step": 2729
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7094375660138978e-05,
+      "loss": 0.1134,
+      "step": 2730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092105124101005e-05,
+      "loss": 0.0966,
+      "step": 2731
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7089833852197508e-05,
+      "loss": 0.1024,
+      "step": 2732
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7087561844664144e-05,
+      "loss": 0.1169,
+      "step": 2733
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708528910173666e-05,
+      "loss": 0.1202,
+      "step": 2734
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7083015623650867e-05,
+      "loss": 0.1075,
+      "step": 2735
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7080741410642667e-05,
+      "loss": 0.0988,
+      "step": 2736
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7078466462948015e-05,
+      "loss": 0.1155,
+      "step": 2737
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707619078080296e-05,
+      "loss": 0.098,
+      "step": 2738
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.707391436444363e-05,
+      "loss": 0.1023,
+      "step": 2739
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7071637214106218e-05,
+      "loss": 0.1072,
+      "step": 2740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7069359330027e-05,
+      "loss": 0.1207,
+      "step": 2741
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706708071244232e-05,
+      "loss": 0.1072,
+      "step": 2742
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7064801361588603e-05,
+      "loss": 0.1167,
+      "step": 2743
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706252127770236e-05,
+      "loss": 0.1224,
+      "step": 2744
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7060240461020164e-05,
+      "loss": 0.1042,
+      "step": 2745
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7057958911778665e-05,
+      "loss": 0.0978,
+      "step": 2746
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7055676630214598e-05,
+      "loss": 0.0907,
+      "step": 2747
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705339361656477e-05,
+      "loss": 0.1165,
+      "step": 2748
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7051109871066055e-05,
+      "loss": 0.0875,
+      "step": 2749
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704882539395542e-05,
+      "loss": 0.1064,
+      "step": 2750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7046540185469895e-05,
+      "loss": 0.1224,
+      "step": 2751
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7044254245846586e-05,
+      "loss": 0.1234,
+      "step": 2752
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704196757532268e-05,
+      "loss": 0.1,
+      "step": 2753
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7039680174135446e-05,
+      "loss": 0.104,
+      "step": 2754
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703739204252221e-05,
+      "loss": 0.0817,
+      "step": 2755
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7035103180720392e-05,
+      "loss": 0.1021,
+      "step": 2756
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703281358896748e-05,
+      "loss": 0.1106,
+      "step": 2757
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030523267501037e-05,
+      "loss": 0.1051,
+      "step": 2758
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70282322165587e-05,
+      "loss": 0.0963,
+      "step": 2759
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702594043637819e-05,
+      "loss": 0.115,
+      "step": 2760
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7023647927197297e-05,
+      "loss": 0.1047,
+      "step": 2761
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7021354689253888e-05,
+      "loss": 0.1029,
+      "step": 2762
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70190607227859e-05,
+      "loss": 0.1022,
+      "step": 2763
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7016766028031363e-05,
+      "loss": 0.1072,
+      "step": 2764
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701447060522836e-05,
+      "loss": 0.1047,
+      "step": 2765
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7012174454615066e-05,
+      "loss": 0.0902,
+      "step": 2766
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7009877576429724e-05,
+      "loss": 0.0983,
+      "step": 2767
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7007579970910657e-05,
+      "loss": 0.1167,
+      "step": 2768
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700528163829626e-05,
+      "loss": 0.1156,
+      "step": 2769
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7002982578825e-05,
+      "loss": 0.0983,
+      "step": 2770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7000682792735427e-05,
+      "loss": 0.1039,
+      "step": 2771
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6998382280266167e-05,
+      "loss": 0.0898,
+      "step": 2772
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.699608104165591e-05,
+      "loss": 0.1152,
+      "step": 2773
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6993779077143437e-05,
+      "loss": 0.0977,
+      "step": 2774
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6991476386967584e-05,
+      "loss": 0.1136,
+      "step": 2775
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698917297136729e-05,
+      "loss": 0.085,
+      "step": 2776
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6986868830581542e-05,
+      "loss": 0.1267,
+      "step": 2777
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698456396484942e-05,
+      "loss": 0.0927,
+      "step": 2778
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6982258374410067e-05,
+      "loss": 0.1219,
+      "step": 2779
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6979952059502715e-05,
+      "loss": 0.0979,
+      "step": 2780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697764502036666e-05,
+      "loss": 0.1049,
+      "step": 2781
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6975337257241275e-05,
+      "loss": 0.093,
+      "step": 2782
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6973028770366015e-05,
+      "loss": 0.1014,
+      "step": 2783
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.69707195599804e-05,
+      "loss": 0.1129,
+      "step": 2784
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6968409626324028e-05,
+      "loss": 0.1111,
+      "step": 2785
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6966098969636583e-05,
+      "loss": 0.1095,
+      "step": 2786
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6963787590157805e-05,
+      "loss": 0.1116,
+      "step": 2787
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6961475488127525e-05,
+      "loss": 0.111,
+      "step": 2788
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695916266378564e-05,
+      "loss": 0.0901,
+      "step": 2789
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6956849117372133e-05,
+      "loss": 0.1014,
+      "step": 2790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6954534849127045e-05,
+      "loss": 0.092,
+      "step": 2791
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69522198592905e-05,
+      "loss": 0.0939,
+      "step": 2792
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.69499041481027e-05,
+      "loss": 0.1029,
+      "step": 2793
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6947587715803923e-05,
+      "loss": 0.1042,
+      "step": 2794
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694527056263452e-05,
+      "loss": 0.1148,
+      "step": 2795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6942952688834907e-05,
+      "loss": 0.1023,
+      "step": 2796
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6940634094645586e-05,
+      "loss": 0.0994,
+      "step": 2797
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693831478030713e-05,
+      "loss": 0.1068,
+      "step": 2798
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693599474606019e-05,
+      "loss": 0.0986,
+      "step": 2799
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693367399214549e-05,
+      "loss": 0.1106,
+      "step": 2800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6931352518803825e-05,
+      "loss": 0.1111,
+      "step": 2801
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6929030326276067e-05,
+      "loss": 0.1318,
+      "step": 2802
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6926707414803165e-05,
+      "loss": 0.1232,
+      "step": 2803
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692438378462614e-05,
+      "loss": 0.1205,
+      "step": 2804
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6922059435986082e-05,
+      "loss": 0.1196,
+      "step": 2805
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6919734369124175e-05,
+      "loss": 0.1361,
+      "step": 2806
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6917408584281654e-05,
+      "loss": 0.1046,
+      "step": 2807
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6915082081699842e-05,
+      "loss": 0.1074,
+      "step": 2808
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691275486162013e-05,
+      "loss": 0.1014,
+      "step": 2809
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6910426924283993e-05,
+      "loss": 0.1025,
+      "step": 2810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6908098269932967e-05,
+      "loss": 0.1213,
+      "step": 2811
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6905768898808676e-05,
+      "loss": 0.1213,
+      "step": 2812
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6903438811152803e-05,
+      "loss": 0.1035,
+      "step": 2813
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6901108007207124e-05,
+      "loss": 0.1071,
+      "step": 2814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6898776487213472e-05,
+      "loss": 0.0795,
+      "step": 2815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6896444251413768e-05,
+      "loss": 0.1147,
+      "step": 2816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6894111300050002e-05,
+      "loss": 0.1149,
+      "step": 2817
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6891777633364224e-05,
+      "loss": 0.0967,
+      "step": 2818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688944325159859e-05,
+      "loss": 0.0936,
+      "step": 2819
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6887108154995298e-05,
+      "loss": 0.0986,
+      "step": 2820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688477234379664e-05,
+      "loss": 0.1035,
+      "step": 2821
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6882435818244976e-05,
+      "loss": 0.0957,
+      "step": 2822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6880098578582737e-05,
+      "loss": 0.1042,
+      "step": 2823
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6877760625052432e-05,
+      "loss": 0.1157,
+      "step": 2824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6875421957896646e-05,
+      "loss": 0.0995,
+      "step": 2825
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6873082577358033e-05,
+      "loss": 0.1176,
+      "step": 2826
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6870742483679326e-05,
+      "loss": 0.1024,
+      "step": 2827
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6868401677103324e-05,
+      "loss": 0.0975,
+      "step": 2828
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6866060157872913e-05,
+      "loss": 0.0961,
+      "step": 2829
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6863717926231042e-05,
+      "loss": 0.1104,
+      "step": 2830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686137498242073e-05,
+      "loss": 0.1001,
+      "step": 2831
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685903132668509e-05,
+      "loss": 0.1125,
+      "step": 2832
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685668695926729e-05,
+      "loss": 0.1021,
+      "step": 2833
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6854341880410573e-05,
+      "loss": 0.1215,
+      "step": 2834
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6851996090358266e-05,
+      "loss": 0.114,
+      "step": 2835
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6849649589353764e-05,
+      "loss": 0.1067,
+      "step": 2836
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6847302377640538e-05,
+      "loss": 0.101,
+      "step": 2837
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6844954455462124e-05,
+      "loss": 0.1201,
+      "step": 2838
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684260582306215e-05,
+      "loss": 0.0966,
+      "step": 2839
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6840256480684294e-05,
+      "loss": 0.1102,
+      "step": 2840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683790642857233e-05,
+      "loss": 0.1248,
+      "step": 2841
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6835555666970086e-05,
+      "loss": 0.0979,
+      "step": 2842
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683320419612148e-05,
+      "loss": 0.0913,
+      "step": 2843
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68308520162705e-05,
+      "loss": 0.1083,
+      "step": 2844
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6828499127661198e-05,
+      "loss": 0.1266,
+      "step": 2845
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6826145530537705e-05,
+      "loss": 0.0967,
+      "step": 2846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6823791225144232e-05,
+      "loss": 0.1128,
+      "step": 2847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6821436211725057e-05,
+      "loss": 0.1106,
+      "step": 2848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6819080490524527e-05,
+      "loss": 0.1001,
+      "step": 2849
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6816724061787076e-05,
+      "loss": 0.0953,
+      "step": 2850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6814366925757192e-05,
+      "loss": 0.0787,
+      "step": 2851
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681200908267946e-05,
+      "loss": 0.1044,
+      "step": 2852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680965053279852e-05,
+      "loss": 0.1019,
+      "step": 2853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680729127635909e-05,
+      "loss": 0.1252,
+      "step": 2854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680493131360597e-05,
+      "loss": 0.1053,
+      "step": 2855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6802570644784016e-05,
+      "loss": 0.105,
+      "step": 2856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6800209270138174e-05,
+      "loss": 0.085,
+      "step": 2857
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6797847189913456e-05,
+      "loss": 0.1116,
+      "step": 2858
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679548440435494e-05,
+      "loss": 0.1146,
+      "step": 2859
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6793120913707798e-05,
+      "loss": 0.1174,
+      "step": 2860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6790756718217252e-05,
+      "loss": 0.1035,
+      "step": 2861
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678839181812861e-05,
+      "loss": 0.1062,
+      "step": 2862
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6786026213687252e-05,
+      "loss": 0.1091,
+      "step": 2863
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6783659905138626e-05,
+      "loss": 0.1136,
+      "step": 2864
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6781292892728257e-05,
+      "loss": 0.103,
+      "step": 2865
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6778925176701747e-05,
+      "loss": 0.0862,
+      "step": 2866
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677655675730476e-05,
+      "loss": 0.121,
+      "step": 2867
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6774187634783046e-05,
+      "loss": 0.0879,
+      "step": 2868
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6771817809382415e-05,
+      "loss": 0.1246,
+      "step": 2869
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6769447281348757e-05,
+      "loss": 0.1053,
+      "step": 2870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676707605092804e-05,
+      "loss": 0.1158,
+      "step": 2871
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6764704118366298e-05,
+      "loss": 0.0835,
+      "step": 2872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676233148390963e-05,
+      "loss": 0.0811,
+      "step": 2873
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6759958147804228e-05,
+      "loss": 0.0978,
+      "step": 2874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675758411029634e-05,
+      "loss": 0.1136,
+      "step": 2875
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675520937163229e-05,
+      "loss": 0.0884,
+      "step": 2876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6752833932058484e-05,
+      "loss": 0.1027,
+      "step": 2877
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675045779182139e-05,
+      "loss": 0.1272,
+      "step": 2878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6748080951167552e-05,
+      "loss": 0.1074,
+      "step": 2879
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6745703410343586e-05,
+      "loss": 0.1045,
+      "step": 2880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6743325169596184e-05,
+      "loss": 0.0956,
+      "step": 2881
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.674094622917211e-05,
+      "loss": 0.1285,
+      "step": 2882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6738566589318196e-05,
+      "loss": 0.0972,
+      "step": 2883
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673618625028135e-05,
+      "loss": 0.1274,
+      "step": 2884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6733805212308553e-05,
+      "loss": 0.0894,
+      "step": 2885
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.673142347564686e-05,
+      "loss": 0.0964,
+      "step": 2886
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672904104054339e-05,
+      "loss": 0.1177,
+      "step": 2887
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6726657907245348e-05,
+      "loss": 0.1046,
+      "step": 2888
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6724274076e-05,
+      "loss": 0.1228,
+      "step": 2889
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6721889547054687e-05,
+      "loss": 0.1067,
+      "step": 2890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6719504320656827e-05,
+      "loss": 0.1064,
+      "step": 2891
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671711839705391e-05,
+      "loss": 0.1022,
+      "step": 2892
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6714731776493486e-05,
+      "loss": 0.1183,
+      "step": 2893
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6712344459223198e-05,
+      "loss": 0.1195,
+      "step": 2894
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670995644549074e-05,
+      "loss": 0.0813,
+      "step": 2895
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6707567735543898e-05,
+      "loss": 0.0823,
+      "step": 2896
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670517832963052e-05,
+      "loss": 0.0886,
+      "step": 2897
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6702788227998517e-05,
+      "loss": 0.1105,
+      "step": 2898
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6700397430895888e-05,
+      "loss": 0.1188,
+      "step": 2899
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6698005938570702e-05,
+      "loss": 0.1245,
+      "step": 2900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695613751271094e-05,
+      "loss": 0.0933,
+      "step": 2901
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669322086924527e-05,
+      "loss": 0.0931,
+      "step": 2902
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669082729274152e-05,
+      "loss": 0.0927,
+      "step": 2903
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6688433022008187e-05,
+      "loss": 0.1055,
+      "step": 2904
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6686038057293705e-05,
+      "loss": 0.0978,
+      "step": 2905
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6683642398846563e-05,
+      "loss": 0.0917,
+      "step": 2906
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.668124604691534e-05,
+      "loss": 0.1064,
+      "step": 2907
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6678849001748677e-05,
+      "loss": 0.1127,
+      "step": 2908
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6676451263595276e-05,
+      "loss": 0.0884,
+      "step": 2909
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667405283270394e-05,
+      "loss": 0.1278,
+      "step": 2910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6671653709323513e-05,
+      "loss": 0.1104,
+      "step": 2911
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666925389370293e-05,
+      "loss": 0.1047,
+      "step": 2912
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666685338609119e-05,
+      "loss": 0.0919,
+      "step": 2913
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6664452186737366e-05,
+      "loss": 0.1022,
+      "step": 2914
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6662050295890605e-05,
+      "loss": 0.0897,
+      "step": 2915
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6659647713800117e-05,
+      "loss": 0.1011,
+      "step": 2916
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6657244440715197e-05,
+      "loss": 0.1077,
+      "step": 2917
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6654840476885205e-05,
+      "loss": 0.0874,
+      "step": 2918
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6652435822559566e-05,
+      "loss": 0.1039,
+      "step": 2919
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6650030477987787e-05,
+      "loss": 0.1052,
+      "step": 2920
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6647624443419446e-05,
+      "loss": 0.099,
+      "step": 2921
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6645217719104187e-05,
+      "loss": 0.0967,
+      "step": 2922
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.664281030529172e-05,
+      "loss": 0.1161,
+      "step": 2923
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6640402202231847e-05,
+      "loss": 0.0948,
+      "step": 2924
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.663799341017442e-05,
+      "loss": 0.1017,
+      "step": 2925
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6635583929369375e-05,
+      "loss": 0.1071,
+      "step": 2926
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6633173760066717e-05,
+      "loss": 0.1144,
+      "step": 2927
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.663076290251652e-05,
+      "loss": 0.0912,
+      "step": 2928
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662835135696893e-05,
+      "loss": 0.1045,
+      "step": 2929
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6625939123674165e-05,
+      "loss": 0.0934,
+      "step": 2930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6623526202882516e-05,
+      "loss": 0.0908,
+      "step": 2931
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6621112594844342e-05,
+      "loss": 0.1154,
+      "step": 2932
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6618698299810078e-05,
+      "loss": 0.1069,
+      "step": 2933
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6616283318030225e-05,
+      "loss": 0.1198,
+      "step": 2934
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661386764975536e-05,
+      "loss": 0.097,
+      "step": 2935
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661145129523612e-05,
+      "loss": 0.1036,
+      "step": 2936
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6609034254723234e-05,
+      "loss": 0.1127,
+      "step": 2937
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6606616528467486e-05,
+      "loss": 0.0867,
+      "step": 2938
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6604198116719735e-05,
+      "loss": 0.1107,
+      "step": 2939
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6601779019730908e-05,
+      "loss": 0.1178,
+      "step": 2940
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6599359237752015e-05,
+      "loss": 0.0912,
+      "step": 2941
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6596938771034116e-05,
+      "loss": 0.1106,
+      "step": 2942
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659451761982837e-05,
+      "loss": 0.0831,
+      "step": 2943
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6592095784385983e-05,
+      "loss": 0.1284,
+      "step": 2944
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658967326495824e-05,
+      "loss": 0.1099,
+      "step": 2945
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6587250061796498e-05,
+      "loss": 0.1024,
+      "step": 2946
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6584826175152192e-05,
+      "loss": 0.09,
+      "step": 2947
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6582401605276813e-05,
+      "loss": 0.093,
+      "step": 2948
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657997635242193e-05,
+      "loss": 0.1046,
+      "step": 2949
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.657755041683919e-05,
+      "loss": 0.1108,
+      "step": 2950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65751237987803e-05,
+      "loss": 0.1093,
+      "step": 2951
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572696498497037e-05,
+      "loss": 0.0927,
+      "step": 2952
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6570268516241264e-05,
+      "loss": 0.1201,
+      "step": 2953
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6567839852264898e-05,
+      "loss": 0.0968,
+      "step": 2954
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6565410506819938e-05,
+      "loss": 0.1182,
+      "step": 2955
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6562980480158443e-05,
+      "loss": 0.1088,
+      "step": 2956
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656054977253255e-05,
+      "loss": 0.1035,
+      "step": 2957
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655811838419447e-05,
+      "loss": 0.1113,
+      "step": 2958
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6555686315396475e-05,
+      "loss": 0.0955,
+      "step": 2959
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6553253566390916e-05,
+      "loss": 0.0978,
+      "step": 2960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6550820137430214e-05,
+      "loss": 0.1178,
+      "step": 2961
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6548386028766846e-05,
+      "loss": 0.1132,
+      "step": 2962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6545951240653383e-05,
+      "loss": 0.1035,
+      "step": 2963
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6543515773342446e-05,
+      "loss": 0.0989,
+      "step": 2964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6541079627086744e-05,
+      "loss": 0.1067,
+      "step": 2965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6538642802139042e-05,
+      "loss": 0.0925,
+      "step": 2966
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6536205298752183e-05,
+      "loss": 0.108,
+      "step": 2967
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6533767117179077e-05,
+      "loss": 0.1028,
+      "step": 2968
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6531328257672707e-05,
+      "loss": 0.1104,
+      "step": 2969
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6528888720486124e-05,
+      "loss": 0.0995,
+      "step": 2970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6526448505872453e-05,
+      "loss": 0.1064,
+      "step": 2971
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6524007614084886e-05,
+      "loss": 0.0911,
+      "step": 2972
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.652156604537668e-05,
+      "loss": 0.1002,
+      "step": 2973
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651912380000118e-05,
+      "loss": 0.0989,
+      "step": 2974
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651668087821178e-05,
+      "loss": 0.1018,
+      "step": 2975
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651423728026195e-05,
+      "loss": 0.0988,
+      "step": 2976
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6511793006405254e-05,
+      "loss": 0.1066,
+      "step": 2977
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6509348056895284e-05,
+      "loss": 0.1083,
+      "step": 2978
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6506902431985734e-05,
+      "loss": 0.1003,
+      "step": 2979
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6504456131930356e-05,
+      "loss": 0.1034,
+      "step": 2980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6502009156982974e-05,
+      "loss": 0.1067,
+      "step": 2981
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6499561507397483e-05,
+      "loss": 0.1223,
+      "step": 2982
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.649711318342785e-05,
+      "loss": 0.1123,
+      "step": 2983
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6494664185328103e-05,
+      "loss": 0.0974,
+      "step": 2984
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6492214513352354e-05,
+      "loss": 0.108,
+      "step": 2985
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6489764167754768e-05,
+      "loss": 0.1154,
+      "step": 2986
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6487313148789597e-05,
+      "loss": 0.1121,
+      "step": 2987
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648486145671115e-05,
+      "loss": 0.0972,
+      "step": 2988
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648240909177381e-05,
+      "loss": 0.1266,
+      "step": 2989
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6479956054232034e-05,
+      "loss": 0.1179,
+      "step": 2990
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6477502344340345e-05,
+      "loss": 0.0991,
+      "step": 2991
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6475047962353335e-05,
+      "loss": 0.0771,
+      "step": 2992
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6472592908525666e-05,
+      "loss": 0.1205,
+      "step": 2993
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647013718311207e-05,
+      "loss": 0.1056,
+      "step": 2994
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6467680786367353e-05,
+      "loss": 0.1038,
+      "step": 2995
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6465223718546383e-05,
+      "loss": 0.0837,
+      "step": 2996
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.64627659799041e-05,
+      "loss": 0.0945,
+      "step": 2997
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6460307570695517e-05,
+      "loss": 0.1069,
+      "step": 2998
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645784849117572e-05,
+      "loss": 0.1041,
+      "step": 2999
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645538874159985e-05,
+      "loss": 0.0996,
+      "step": 3000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6452928322223134e-05,
+      "loss": 0.0859,
+      "step": 3001
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6450467233300854e-05,
+      "loss": 0.1127,
+      "step": 3002
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6448005475088376e-05,
+      "loss": 0.1124,
+      "step": 3003
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6445543047841127e-05,
+      "loss": 0.0885,
+      "step": 3004
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.64430799518146e-05,
+      "loss": 0.123,
+      "step": 3005
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6440616187264365e-05,
+      "loss": 0.1033,
+      "step": 3006
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6438151754446057e-05,
+      "loss": 0.0985,
+      "step": 3007
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643568665361538e-05,
+      "loss": 0.0948,
+      "step": 3008
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6433220885028116e-05,
+      "loss": 0.1149,
+      "step": 3009
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6430754448940103e-05,
+      "loss": 0.1141,
+      "step": 3010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6428287345607255e-05,
+      "loss": 0.1239,
+      "step": 3011
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6425819575285557e-05,
+      "loss": 0.0886,
+      "step": 3012
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.642335113823106e-05,
+      "loss": 0.0862,
+      "step": 3013
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6420882034699882e-05,
+      "loss": 0.0974,
+      "step": 3014
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6418412264948214e-05,
+      "loss": 0.098,
+      "step": 3015
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6415941829232325e-05,
+      "loss": 0.099,
+      "step": 3016
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6413470727808533e-05,
+      "loss": 0.1097,
+      "step": 3017
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6410998960933234e-05,
+      "loss": 0.1133,
+      "step": 3018
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.64085265288629e-05,
+      "loss": 0.1029,
+      "step": 3019
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6406053431854066e-05,
+      "loss": 0.1135,
+      "step": 3020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640357967016334e-05,
+      "loss": 0.1254,
+      "step": 3021
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640110524404739e-05,
+      "loss": 0.1034,
+      "step": 3022
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639863015376296e-05,
+      "loss": 0.1007,
+      "step": 3023
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639615439956686e-05,
+      "loss": 0.1085,
+      "step": 3024
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6393677981715976e-05,
+      "loss": 0.1184,
+      "step": 3025
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6391200900467245e-05,
+      "loss": 0.0898,
+      "step": 3026
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63887231560777e-05,
+      "loss": 0.1014,
+      "step": 3027
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638624474880442e-05,
+      "loss": 0.0985,
+      "step": 3028
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6383765678904563e-05,
+      "loss": 0.109,
+      "step": 3029
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6381285946635346e-05,
+      "loss": 0.0797,
+      "step": 3030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.637880555225407e-05,
+      "loss": 0.1055,
+      "step": 3031
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6376324496018096e-05,
+      "loss": 0.0861,
+      "step": 3032
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6373842778184856e-05,
+      "loss": 0.114,
+      "step": 3033
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6371360399011842e-05,
+      "loss": 0.1049,
+      "step": 3034
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636887735875663e-05,
+      "loss": 0.0946,
+      "step": 3035
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636639365767685e-05,
+      "loss": 0.0902,
+      "step": 3036
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6363909296030208e-05,
+      "loss": 0.1068,
+      "step": 3037
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.636142427407448e-05,
+      "loss": 0.1208,
+      "step": 3038
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6358938592067506e-05,
+      "loss": 0.0929,
+      "step": 3039
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6356452250267198e-05,
+      "loss": 0.1172,
+      "step": 3040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635396524893153e-05,
+      "loss": 0.0934,
+      "step": 3041
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6351477588318556e-05,
+      "loss": 0.1124,
+      "step": 3042
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6348989268686385e-05,
+      "loss": 0.0927,
+      "step": 3043
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634650029029321e-05,
+      "loss": 0.095,
+      "step": 3044
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6344010653397274e-05,
+      "loss": 0.0967,
+      "step": 3045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634152035825691e-05,
+      "loss": 0.0999,
+      "step": 3046
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633902940513049e-05,
+      "loss": 0.1044,
+      "step": 3047
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6336537794276482e-05,
+      "loss": 0.1021,
+      "step": 3048
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6334045525953412e-05,
+      "loss": 0.1005,
+      "step": 3049
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633155260041987e-05,
+      "loss": 0.0885,
+      "step": 3050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632905901793452e-05,
+      "loss": 0.1055,
+      "step": 3051
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6326564778756097e-05,
+      "loss": 0.1075,
+      "step": 3052
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632406988314339e-05,
+      "loss": 0.0942,
+      "step": 3053
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6321574331355266e-05,
+      "loss": 0.1141,
+      "step": 3054
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631907812365067e-05,
+      "loss": 0.0923,
+      "step": 3055
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631658126028859e-05,
+      "loss": 0.1252,
+      "step": 3056
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631408374152811e-05,
+      "loss": 0.0931,
+      "step": 3057
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631158556762836e-05,
+      "loss": 0.072,
+      "step": 3058
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630908673884855e-05,
+      "loss": 0.1146,
+      "step": 3059
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6306587255447954e-05,
+      "loss": 0.1165,
+      "step": 3060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630408711768591e-05,
+      "loss": 0.0873,
+      "step": 3061
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630158632582184e-05,
+      "loss": 0.0967,
+      "step": 3062
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629908488011521e-05,
+      "loss": 0.0956,
+      "step": 3063
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.629658278082557e-05,
+      "loss": 0.1198,
+      "step": 3064
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6294080028212532e-05,
+      "loss": 0.1117,
+      "step": 3065
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6291576622535788e-05,
+      "loss": 0.1138,
+      "step": 3066
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6289072564055075e-05,
+      "loss": 0.0891,
+      "step": 3067
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6286567853030212e-05,
+      "loss": 0.0985,
+      "step": 3068
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6284062489721092e-05,
+      "loss": 0.1073,
+      "step": 3069
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6281556474387656e-05,
+      "loss": 0.1167,
+      "step": 3070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6279049807289936e-05,
+      "loss": 0.1118,
+      "step": 3071
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627654248868801e-05,
+      "loss": 0.1017,
+      "step": 3072
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6274034518842036e-05,
+      "loss": 0.1213,
+      "step": 3073
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6271525898012242e-05,
+      "loss": 0.1021,
+      "step": 3074
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6269016626458914e-05,
+      "loss": 0.1167,
+      "step": 3075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626650670444241e-05,
+      "loss": 0.0875,
+      "step": 3076
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6263996132223155e-05,
+      "loss": 0.1251,
+      "step": 3077
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6261484910061644e-05,
+      "loss": 0.1018,
+      "step": 3078
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6258973038218436e-05,
+      "loss": 0.1106,
+      "step": 3079
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625646051695416e-05,
+      "loss": 0.076,
+      "step": 3080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625394734652951e-05,
+      "loss": 0.0966,
+      "step": 3081
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625143352720525e-05,
+      "loss": 0.1189,
+      "step": 3082
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624891905924221e-05,
+      "loss": 0.0951,
+      "step": 3083
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6246403942901284e-05,
+      "loss": 0.1011,
+      "step": 3084
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624388817844344e-05,
+      "loss": 0.0985,
+      "step": 3085
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6241371766129707e-05,
+      "loss": 0.1007,
+      "step": 3086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6238854706221186e-05,
+      "loss": 0.0863,
+      "step": 3087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6236336998979044e-05,
+      "loss": 0.0916,
+      "step": 3088
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6233818644664514e-05,
+      "loss": 0.1243,
+      "step": 3089
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6231299643538895e-05,
+      "loss": 0.0902,
+      "step": 3090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622877999586355e-05,
+      "loss": 0.0958,
+      "step": 3091
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6226259701899922e-05,
+      "loss": 0.1022,
+      "step": 3092
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622373876190951e-05,
+      "loss": 0.1116,
+      "step": 3093
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6221217176153877e-05,
+      "loss": 0.1029,
+      "step": 3094
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6218694944894666e-05,
+      "loss": 0.1079,
+      "step": 3095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621617206839358e-05,
+      "loss": 0.1078,
+      "step": 3096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621364854691238e-05,
+      "loss": 0.1122,
+      "step": 3097
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6211124380712914e-05,
+      "loss": 0.1135,
+      "step": 3098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620859957005708e-05,
+      "loss": 0.1002,
+      "step": 3099
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6206074115206845e-05,
+      "loss": 0.1248,
+      "step": 3100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620354801642425e-05,
+      "loss": 0.1075,
+      "step": 3101
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6201021273971398e-05,
+      "loss": 0.0914,
+      "step": 3102
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619849388811046e-05,
+      "loss": 0.1108,
+      "step": 3103
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6195965859103675e-05,
+      "loss": 0.104,
+      "step": 3104
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6193437187213344e-05,
+      "loss": 0.0938,
+      "step": 3105
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619090787270184e-05,
+      "loss": 0.1105,
+      "step": 3106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6188377915831605e-05,
+      "loss": 0.1083,
+      "step": 3107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6185847316865134e-05,
+      "loss": 0.0851,
+      "step": 3108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6183316076065004e-05,
+      "loss": 0.125,
+      "step": 3109
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6180784193693852e-05,
+      "loss": 0.124,
+      "step": 3110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617825167001438e-05,
+      "loss": 0.0841,
+      "step": 3111
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617571850528936e-05,
+      "loss": 0.1013,
+      "step": 3112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6173184699781632e-05,
+      "loss": 0.1021,
+      "step": 3113
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6170650253754097e-05,
+      "loss": 0.1021,
+      "step": 3114
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616811516746972e-05,
+      "loss": 0.1116,
+      "step": 3115
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6165579441191546e-05,
+      "loss": 0.103,
+      "step": 3116
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6163043075182673e-05,
+      "loss": 0.1139,
+      "step": 3117
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160506069706273e-05,
+      "loss": 0.1039,
+      "step": 3118
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6157968425025577e-05,
+      "loss": 0.108,
+      "step": 3119
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615543014140389e-05,
+      "loss": 0.1073,
+      "step": 3120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6152891219104585e-05,
+      "loss": 0.0853,
+      "step": 3121
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6150351658391086e-05,
+      "loss": 0.1135,
+      "step": 3122
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6147811459526902e-05,
+      "loss": 0.091,
+      "step": 3123
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6145270622775598e-05,
+      "loss": 0.0955,
+      "step": 3124
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614272914840081e-05,
+      "loss": 0.1173,
+      "step": 3125
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614018703666623e-05,
+      "loss": 0.1139,
+      "step": 3126
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6137644287835632e-05,
+      "loss": 0.1099,
+      "step": 3127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6135100902172838e-05,
+      "loss": 0.0931,
+      "step": 3128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613255687994175e-05,
+      "loss": 0.1067,
+      "step": 3129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6130012221406338e-05,
+      "loss": 0.1252,
+      "step": 3130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6127466926830625e-05,
+      "loss": 0.141,
+      "step": 3131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6124920996478707e-05,
+      "loss": 0.0896,
+      "step": 3132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6122374430614747e-05,
+      "loss": 0.1206,
+      "step": 3133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6119827229502972e-05,
+      "loss": 0.1017,
+      "step": 3134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6117279393407673e-05,
+      "loss": 0.1094,
+      "step": 3135
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6114730922593215e-05,
+      "loss": 0.1034,
+      "step": 3136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611218181732402e-05,
+      "loss": 0.1067,
+      "step": 3137
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610963207786458e-05,
+      "loss": 0.104,
+      "step": 3138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6107081704479452e-05,
+      "loss": 0.1167,
+      "step": 3139
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6104530697433258e-05,
+      "loss": 0.0927,
+      "step": 3140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101979056990686e-05,
+      "loss": 0.1021,
+      "step": 3141
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609942678341649e-05,
+      "loss": 0.1105,
+      "step": 3142
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6096873876975492e-05,
+      "loss": 0.095,
+      "step": 3143
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6094320337932577e-05,
+      "loss": 0.1139,
+      "step": 3144
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6091766166552693e-05,
+      "loss": 0.0907,
+      "step": 3145
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089211363100858e-05,
+      "loss": 0.119,
+      "step": 3146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6086655927842157e-05,
+      "loss": 0.1049,
+      "step": 3147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6084099861041735e-05,
+      "loss": 0.0962,
+      "step": 3148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.608154316296481e-05,
+      "loss": 0.098,
+      "step": 3149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6078985833876654e-05,
+      "loss": 0.0974,
+      "step": 3150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6076427874042617e-05,
+      "loss": 0.1096,
+      "step": 3151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6073869283728103e-05,
+      "loss": 0.0974,
+      "step": 3152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6071310063198596e-05,
+      "loss": 0.0876,
+      "step": 3153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606875021271963e-05,
+      "loss": 0.0891,
+      "step": 3154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6066189732556812e-05,
+      "loss": 0.1056,
+      "step": 3155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6063628622975812e-05,
+      "loss": 0.1208,
+      "step": 3156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6061066884242374e-05,
+      "loss": 0.1105,
+      "step": 3157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6058504516622288e-05,
+      "loss": 0.1125,
+      "step": 3158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6055941520381432e-05,
+      "loss": 0.0913,
+      "step": 3159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6053377895785733e-05,
+      "loss": 0.0865,
+      "step": 3160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6050813643101194e-05,
+      "loss": 0.0911,
+      "step": 3161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6048248762593867e-05,
+      "loss": 0.111,
+      "step": 3162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604568325452989e-05,
+      "loss": 0.0861,
+      "step": 3163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604311711917545e-05,
+      "loss": 0.0884,
+      "step": 3164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604055035679681e-05,
+      "loss": 0.1116,
+      "step": 3165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603798296766029e-05,
+      "loss": 0.104,
+      "step": 3166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6035414952032277e-05,
+      "loss": 0.1072,
+      "step": 3167
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.603284631017923e-05,
+      "loss": 0.1072,
+      "step": 3168
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030277042367658e-05,
+      "loss": 0.094,
+      "step": 3169
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6027707148864155e-05,
+      "loss": 0.1051,
+      "step": 3170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6025136629935362e-05,
+      "loss": 0.1042,
+      "step": 3171
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602256548584799e-05,
+      "loss": 0.1111,
+      "step": 3172
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601999371686883e-05,
+      "loss": 0.1042,
+      "step": 3173
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6017421323264704e-05,
+      "loss": 0.0828,
+      "step": 3174
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6014848305302535e-05,
+      "loss": 0.1143,
+      "step": 3175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6012274663249293e-05,
+      "loss": 0.1119,
+      "step": 3176
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600970039737201e-05,
+      "loss": 0.1033,
+      "step": 3177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6007125507937787e-05,
+      "loss": 0.0835,
+      "step": 3178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.60045499952138e-05,
+      "loss": 0.1051,
+      "step": 3179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600197385946727e-05,
+      "loss": 0.0931,
+      "step": 3180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5999397100965497e-05,
+      "loss": 0.1021,
+      "step": 3181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599681971997584e-05,
+      "loss": 0.1036,
+      "step": 3182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599424171676573e-05,
+      "loss": 0.1128,
+      "step": 3183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599166309160265e-05,
+      "loss": 0.0903,
+      "step": 3184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5989083844754153e-05,
+      "loss": 0.0957,
+      "step": 3185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598650397648786e-05,
+      "loss": 0.1084,
+      "step": 3186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5983923487071457e-05,
+      "loss": 0.1099,
+      "step": 3187
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5981342376772687e-05,
+      "loss": 0.1044,
+      "step": 3188
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597876064585936e-05,
+      "loss": 0.0957,
+      "step": 3189
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597617829459936e-05,
+      "loss": 0.1024,
+      "step": 3190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597359532326062e-05,
+      "loss": 0.088,
+      "step": 3191
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5971011732111153e-05,
+      "loss": 0.1067,
+      "step": 3192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596842752141902e-05,
+      "loss": 0.0983,
+      "step": 3193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596584269145236e-05,
+      "loss": 0.0897,
+      "step": 3194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5963257242479365e-05,
+      "loss": 0.0905,
+      "step": 3195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5960671174768306e-05,
+      "loss": 0.0839,
+      "step": 3196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59580844885875e-05,
+      "loss": 0.0884,
+      "step": 3197
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5955497184205348e-05,
+      "loss": 0.1077,
+      "step": 3198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5952909261890294e-05,
+      "loss": 0.0891,
+      "step": 3199
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5950320721910863e-05,
+      "loss": 0.1091,
+      "step": 3200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5947731564535636e-05,
+      "loss": 0.0867,
+      "step": 3201
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5945141790033257e-05,
+      "loss": 0.1047,
+      "step": 3202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5942551398672443e-05,
+      "loss": 0.1127,
+      "step": 3203
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5939960390721964e-05,
+      "loss": 0.1307,
+      "step": 3204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593736876645066e-05,
+      "loss": 0.1033,
+      "step": 3205
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5934776526127437e-05,
+      "loss": 0.0873,
+      "step": 3206
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5932183670021255e-05,
+      "loss": 0.0914,
+      "step": 3207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592959019840115e-05,
+      "loss": 0.0769,
+      "step": 3208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5926996111536212e-05,
+      "loss": 0.1052,
+      "step": 3209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592440140969561e-05,
+      "loss": 0.1104,
+      "step": 3210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5921806093148552e-05,
+      "loss": 0.093,
+      "step": 3211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591921016216433e-05,
+      "loss": 0.0944,
+      "step": 3212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.59166136170123e-05,
+      "loss": 0.0911,
+      "step": 3213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5914016457961862e-05,
+      "loss": 0.1019,
+      "step": 3214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5911418685282506e-05,
+      "loss": 0.1185,
+      "step": 3215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5908820299243764e-05,
+      "loss": 0.1062,
+      "step": 3216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5906221300115246e-05,
+      "loss": 0.104,
+      "step": 3217
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5903621688166614e-05,
+      "loss": 0.1237,
+      "step": 3218
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5901021463667604e-05,
+      "loss": 0.1038,
+      "step": 3219
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5898420626888015e-05,
+      "loss": 0.1027,
+      "step": 3220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.58958191780977e-05,
+      "loss": 0.1162,
+      "step": 3221
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5893217117566576e-05,
+      "loss": 0.1013,
+      "step": 3222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589061444556464e-05,
+      "loss": 0.0826,
+      "step": 3223
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588801116236194e-05,
+      "loss": 0.1071,
+      "step": 3224
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5885407268228576e-05,
+      "loss": 0.1167,
+      "step": 3225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5882802763434738e-05,
+      "loss": 0.1088,
+      "step": 3226
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5880197648250658e-05,
+      "loss": 0.0889,
+      "step": 3227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587759192294664e-05,
+      "loss": 0.0995,
+      "step": 3228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5874985587793047e-05,
+      "loss": 0.1119,
+      "step": 3229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.587237864306032e-05,
+      "loss": 0.1016,
+      "step": 3230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5869771089018933e-05,
+      "loss": 0.0939,
+      "step": 3231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5867162925939456e-05,
+      "loss": 0.0913,
+      "step": 3232
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5864554154092503e-05,
+      "loss": 0.1073,
+      "step": 3233
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586194477374876e-05,
+      "loss": 0.0922,
+      "step": 3234
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5859334785178965e-05,
+      "loss": 0.1191,
+      "step": 3235
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5856724188653928e-05,
+      "loss": 0.0948,
+      "step": 3236
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5854112984444525e-05,
+      "loss": 0.1139,
+      "step": 3237
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5851501172821686e-05,
+      "loss": 0.1095,
+      "step": 3238
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5848888754056408e-05,
+      "loss": 0.1129,
+      "step": 3239
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846275728419755e-05,
+      "loss": 0.1064,
+      "step": 3240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584366209618285e-05,
+      "loss": 0.1044,
+      "step": 3241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5841047857616876e-05,
+      "loss": 0.1132,
+      "step": 3242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5838433012993087e-05,
+      "loss": 0.0923,
+      "step": 3243
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583581756258279e-05,
+      "loss": 0.1069,
+      "step": 3244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583320150665736e-05,
+      "loss": 0.1051,
+      "step": 3245
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583058484548824e-05,
+      "loss": 0.1028,
+      "step": 3246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582796757934693e-05,
+      "loss": 0.0911,
+      "step": 3247
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5825349708504988e-05,
+      "loss": 0.1039,
+      "step": 3248
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5822731233234044e-05,
+      "loss": 0.1064,
+      "step": 3249
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5820112153805785e-05,
+      "loss": 0.1038,
+      "step": 3250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5817492470491962e-05,
+      "loss": 0.1008,
+      "step": 3251
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5814872183564393e-05,
+      "loss": 0.1151,
+      "step": 3252
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.581225129329495e-05,
+      "loss": 0.1045,
+      "step": 3253
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5809629799955576e-05,
+      "loss": 0.1055,
+      "step": 3254
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5807007703818275e-05,
+      "loss": 0.101,
+      "step": 3255
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5804385005155107e-05,
+      "loss": 0.1062,
+      "step": 3256
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5801761704238197e-05,
+      "loss": 0.1094,
+      "step": 3257
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579913780133974e-05,
+      "loss": 0.1096,
+      "step": 3258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796513296731985e-05,
+      "loss": 0.0895,
+      "step": 3259
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5793888190687247e-05,
+      "loss": 0.0928,
+      "step": 3260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5791262483477904e-05,
+      "loss": 0.1221,
+      "step": 3261
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5788636175376395e-05,
+      "loss": 0.1091,
+      "step": 3262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578600926665522e-05,
+      "loss": 0.1128,
+      "step": 3263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5783381757586946e-05,
+      "loss": 0.1112,
+      "step": 3264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5780753648444194e-05,
+      "loss": 0.099,
+      "step": 3265
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5778124939499654e-05,
+      "loss": 0.0879,
+      "step": 3266
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577549563102608e-05,
+      "loss": 0.0922,
+      "step": 3267
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.577286572329628e-05,
+      "loss": 0.1057,
+      "step": 3268
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5770235216583136e-05,
+      "loss": 0.093,
+      "step": 3269
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5767604111159578e-05,
+      "loss": 0.0845,
+      "step": 3270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5764972407298607e-05,
+      "loss": 0.1031,
+      "step": 3271
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576234010527329e-05,
+      "loss": 0.1265,
+      "step": 3272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5759707205356745e-05,
+      "loss": 0.0884,
+      "step": 3273
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5757073707822155e-05,
+      "loss": 0.0883,
+      "step": 3274
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5754439612942774e-05,
+      "loss": 0.1013,
+      "step": 3275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5751804920991912e-05,
+      "loss": 0.1261,
+      "step": 3276
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5749169632242932e-05,
+      "loss": 0.1044,
+      "step": 3277
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746533746969275e-05,
+      "loss": 0.0869,
+      "step": 3278
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.574389726544443e-05,
+      "loss": 0.1014,
+      "step": 3279
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5741260187941963e-05,
+      "loss": 0.1084,
+      "step": 3280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5738622514735483e-05,
+      "loss": 0.1084,
+      "step": 3281
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573598424609868e-05,
+      "loss": 0.0891,
+      "step": 3282
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.573334538230529e-05,
+      "loss": 0.1307,
+      "step": 3283
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5730705923629116e-05,
+      "loss": 0.0975,
+      "step": 3284
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5728065870344033e-05,
+      "loss": 0.085,
+      "step": 3285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572542522272396e-05,
+      "loss": 0.0916,
+      "step": 3286
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5722783981042892e-05,
+      "loss": 0.1218,
+      "step": 3287
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5720142145574877e-05,
+      "loss": 0.1038,
+      "step": 3288
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5717499716594024e-05,
+      "loss": 0.093,
+      "step": 3289
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5714856694374514e-05,
+      "loss": 0.0796,
+      "step": 3290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5712213079190585e-05,
+      "loss": 0.0942,
+      "step": 3291
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570956887131652e-05,
+      "loss": 0.0787,
+      "step": 3292
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5706924071026693e-05,
+      "loss": 0.0961,
+      "step": 3293
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570427867859552e-05,
+      "loss": 0.0963,
+      "step": 3294
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570163269429748e-05,
+      "loss": 0.1089,
+      "step": 3295
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5698986118407113e-05,
+      "loss": 0.109,
+      "step": 3296
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569633895119903e-05,
+      "loss": 0.0768,
+      "step": 3297
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5693691192947902e-05,
+      "loss": 0.126,
+      "step": 3298
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.569104284392844e-05,
+      "loss": 0.1013,
+      "step": 3299
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5688393904415446e-05,
+      "loss": 0.1014,
+      "step": 3300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5685744374683762e-05,
+      "loss": 0.1013,
+      "step": 3301
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5683094255008304e-05,
+      "loss": 0.0953,
+      "step": 3302
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5680443545664043e-05,
+      "loss": 0.0957,
+      "step": 3303
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567779224692601e-05,
+      "loss": 0.0898,
+      "step": 3304
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5675140359069302e-05,
+      "loss": 0.0914,
+      "step": 3305
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5672487882369072e-05,
+      "loss": 0.1127,
+      "step": 3306
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566983481710054e-05,
+      "loss": 0.1033,
+      "step": 3307
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566718116353898e-05,
+      "loss": 0.076,
+      "step": 3308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.566452692195974e-05,
+      "loss": 0.0958,
+      "step": 3309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5661872092638208e-05,
+      "loss": 0.1171,
+      "step": 3310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565921667584985e-05,
+      "loss": 0.0867,
+      "step": 3311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565656067187019e-05,
+      "loss": 0.0972,
+      "step": 3312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5653904080974805e-05,
+      "loss": 0.0731,
+      "step": 3313
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5651246903439344e-05,
+      "loss": 0.0959,
+      "step": 3314
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5648589139539514e-05,
+      "loss": 0.0793,
+      "step": 3315
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5645930789551073e-05,
+      "loss": 0.0823,
+      "step": 3316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5643271853749848e-05,
+      "loss": 0.0835,
+      "step": 3317
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5640612332411728e-05,
+      "loss": 0.095,
+      "step": 3318
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5637952225812662e-05,
+      "loss": 0.0928,
+      "step": 3319
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563529153422866e-05,
+      "loss": 0.1095,
+      "step": 3320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5632630257935785e-05,
+      "loss": 0.1051,
+      "step": 3321
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5629968397210175e-05,
+      "loss": 0.0997,
+      "step": 3322
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562730595232801e-05,
+      "loss": 0.1072,
+      "step": 3323
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562464292356555e-05,
+      "loss": 0.1025,
+      "step": 3324
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56219793111991e-05,
+      "loss": 0.1106,
+      "step": 3325
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5619315115505037e-05,
+      "loss": 0.0946,
+      "step": 3326
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5616650336759794e-05,
+      "loss": 0.116,
+      "step": 3327
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561398497523986e-05,
+      "loss": 0.0963,
+      "step": 3328
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5611319031221793e-05,
+      "loss": 0.11,
+      "step": 3329
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5608652504982206e-05,
+      "loss": 0.1044,
+      "step": 3330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5605985396797773e-05,
+      "loss": 0.0983,
+      "step": 3331
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5603317706945224e-05,
+      "loss": 0.0868,
+      "step": 3332
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560064943570136e-05,
+      "loss": 0.1044,
+      "step": 3333
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5597980583343036e-05,
+      "loss": 0.1234,
+      "step": 3334
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5595311150147167e-05,
+      "loss": 0.1007,
+      "step": 3335
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5592641136390732e-05,
+      "loss": 0.0857,
+      "step": 3336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5589970542350764e-05,
+      "loss": 0.0995,
+      "step": 3337
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5587299368304362e-05,
+      "loss": 0.1055,
+      "step": 3338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5584627614528675e-05,
+      "loss": 0.1022,
+      "step": 3339
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.558195528130093e-05,
+      "loss": 0.0972,
+      "step": 3340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55792823688984e-05,
+      "loss": 0.0807,
+      "step": 3341
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557660887759842e-05,
+      "loss": 0.1104,
+      "step": 3342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557393480767839e-05,
+      "loss": 0.0975,
+      "step": 3343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557126015941577e-05,
+      "loss": 0.1094,
+      "step": 3344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5568584933088067e-05,
+      "loss": 0.1024,
+      "step": 3345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5565909128972873e-05,
+      "loss": 0.0909,
+      "step": 3346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5563232747347813e-05,
+      "loss": 0.0972,
+      "step": 3347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556055578849059e-05,
+      "loss": 0.0994,
+      "step": 3348
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5557878252678956e-05,
+      "loss": 0.1077,
+      "step": 3349
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5555200140190732e-05,
+      "loss": 0.104,
+      "step": 3350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5552521451303794e-05,
+      "loss": 0.0989,
+      "step": 3351
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554984218629608e-05,
+      "loss": 0.0916,
+      "step": 3352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5547162345445584e-05,
+      "loss": 0.1206,
+      "step": 3353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5544481929030363e-05,
+      "loss": 0.0928,
+      "step": 3354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5541800937328534e-05,
+      "loss": 0.0952,
+      "step": 3355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5539119370618267e-05,
+      "loss": 0.1021,
+      "step": 3356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5536437229177802e-05,
+      "loss": 0.0781,
+      "step": 3357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553375451328543e-05,
+      "loss": 0.0929,
+      "step": 3358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5531071223219513e-05,
+      "loss": 0.0961,
+      "step": 3359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5528387359258457e-05,
+      "loss": 0.1105,
+      "step": 3360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552570292168074e-05,
+      "loss": 0.0895,
+      "step": 3361
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5523017910764892e-05,
+      "loss": 0.0978,
+      "step": 3362
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5520332326789508e-05,
+      "loss": 0.1237,
+      "step": 3363
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5517646170033236e-05,
+      "loss": 0.1079,
+      "step": 3364
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.551495944077479e-05,
+      "loss": 0.1028,
+      "step": 3365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5512272139292942e-05,
+      "loss": 0.1022,
+      "step": 3366
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5509584265866523e-05,
+      "loss": 0.1185,
+      "step": 3367
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5506895820774416e-05,
+      "loss": 0.1049,
+      "step": 3368
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5504206804295577e-05,
+      "loss": 0.0991,
+      "step": 3369
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.550151721670901e-05,
+      "loss": 0.1044,
+      "step": 3370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5498827058293785e-05,
+      "loss": 0.1039,
+      "step": 3371
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5496136329329026e-05,
+      "loss": 0.0977,
+      "step": 3372
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5493445030093918e-05,
+      "loss": 0.0912,
+      "step": 3373
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549075316086771e-05,
+      "loss": 0.1311,
+      "step": 3374
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.54880607219297e-05,
+      "loss": 0.0866,
+      "step": 3375
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548536771355926e-05,
+      "loss": 0.1069,
+      "step": 3376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5482674136035804e-05,
+      "loss": 0.0996,
+      "step": 3377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5479979989638815e-05,
+      "loss": 0.0814,
+      "step": 3378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547728527464784e-05,
+      "loss": 0.1094,
+      "step": 3379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5474589991342468e-05,
+      "loss": 0.1156,
+      "step": 3380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5471894140002366e-05,
+      "loss": 0.1018,
+      "step": 3381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5469197720907244e-05,
+      "loss": 0.0978,
+      "step": 3382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5466500734336886e-05,
+      "loss": 0.1125,
+      "step": 3383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5463803180571118e-05,
+      "loss": 0.0944,
+      "step": 3384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546110505988984e-05,
+      "loss": 0.0952,
+      "step": 3385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5458406372573006e-05,
+      "loss": 0.0829,
+      "step": 3386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5455707118900628e-05,
+      "loss": 0.0898,
+      "step": 3387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5453007299152767e-05,
+      "loss": 0.0834,
+      "step": 3388
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5450306913609557e-05,
+      "loss": 0.1106,
+      "step": 3389
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5447605962551194e-05,
+      "loss": 0.0944,
+      "step": 3390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5444904446257912e-05,
+      "loss": 0.0913,
+      "step": 3391
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5442202365010022e-05,
+      "loss": 0.084,
+      "step": 3392
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5439499719087886e-05,
+      "loss": 0.1001,
+      "step": 3393
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543679650877193e-05,
+      "loss": 0.089,
+      "step": 3394
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543409273434263e-05,
+      "loss": 0.0887,
+      "step": 3395
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5431388396080527e-05,
+      "loss": 0.1063,
+      "step": 3396
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542868349426622e-05,
+      "loss": 0.094,
+      "step": 3397
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5425978029180367e-05,
+      "loss": 0.0983,
+      "step": 3398
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5423272001103676e-05,
+      "loss": 0.1078,
+      "step": 3399
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5420565410316926e-05,
+      "loss": 0.0903,
+      "step": 3400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5417858257100946e-05,
+      "loss": 0.0906,
+      "step": 3401
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5415150541736626e-05,
+      "loss": 0.1063,
+      "step": 3402
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541244226450492e-05,
+      "loss": 0.0952,
+      "step": 3403
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5409733425686822e-05,
+      "loss": 0.1017,
+      "step": 3404
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.540702402556341e-05,
+      "loss": 0.0962,
+      "step": 3405
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54043140644158e-05,
+      "loss": 0.1106,
+      "step": 3406
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5401603542525172e-05,
+      "loss": 0.093,
+      "step": 3407
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539889246017277e-05,
+      "loss": 0.1122,
+      "step": 3408
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.539618081763989e-05,
+      "loss": 0.0977,
+      "step": 3409
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5393468615207887e-05,
+      "loss": 0.0985,
+      "step": 3410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5390755853158177e-05,
+      "loss": 0.099,
+      "step": 3411
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5388042531772228e-05,
+      "loss": 0.1007,
+      "step": 3412
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.538532865133157e-05,
+      "loss": 0.088,
+      "step": 3413
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5382614212117798e-05,
+      "loss": 0.0986,
+      "step": 3414
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537989921441255e-05,
+      "loss": 0.1058,
+      "step": 3415
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537718365849753e-05,
+      "loss": 0.1193,
+      "step": 3416
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5374467544654504e-05,
+      "loss": 0.0885,
+      "step": 3417
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537175087316529e-05,
+      "loss": 0.0969,
+      "step": 3418
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5369033644311768e-05,
+      "loss": 0.105,
+      "step": 3419
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536631585837587e-05,
+      "loss": 0.11,
+      "step": 3420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536359751563959e-05,
+      "loss": 0.1038,
+      "step": 3421
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5360878616384975e-05,
+      "loss": 0.0951,
+      "step": 3422
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5358159160894138e-05,
+      "loss": 0.1095,
+      "step": 3423
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5355439149449245e-05,
+      "loss": 0.1069,
+      "step": 3424
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352718582332524e-05,
+      "loss": 0.1107,
+      "step": 3425
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534999745982625e-05,
+      "loss": 0.0961,
+      "step": 3426
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5347275782212765e-05,
+      "loss": 0.101,
+      "step": 3427
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5344553549774466e-05,
+      "loss": 0.0942,
+      "step": 3428
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5341830762793808e-05,
+      "loss": 0.1,
+      "step": 3429
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.53391074215533e-05,
+      "loss": 0.1086,
+      "step": 3430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5336383526335517e-05,
+      "loss": 0.1134,
+      "step": 3431
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5333659077423084e-05,
+      "loss": 0.0996,
+      "step": 3432
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533093407509868e-05,
+      "loss": 0.0907,
+      "step": 3433
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5328208519645052e-05,
+      "loss": 0.1012,
+      "step": 3434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5325482411345e-05,
+      "loss": 0.097,
+      "step": 3435
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532275575048138e-05,
+      "loss": 0.1072,
+      "step": 3436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5320028537337108e-05,
+      "loss": 0.0984,
+      "step": 3437
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317300772195147e-05,
+      "loss": 0.1125,
+      "step": 3438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5314572455338533e-05,
+      "loss": 0.0958,
+      "step": 3439
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5311843587050352e-05,
+      "loss": 0.1136,
+      "step": 3440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530911416761374e-05,
+      "loss": 0.1138,
+      "step": 3441
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530638419731191e-05,
+      "loss": 0.0863,
+      "step": 3442
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5303653676428106e-05,
+      "loss": 0.0795,
+      "step": 3443
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530092260524565e-05,
+      "loss": 0.0972,
+      "step": 3444
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529819098404791e-05,
+      "loss": 0.0872,
+      "step": 3445
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529545881311832e-05,
+      "loss": 0.1052,
+      "step": 3446
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5292726092740358e-05,
+      "loss": 0.1204,
+      "step": 3447
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.528999282319757e-05,
+      "loss": 0.1196,
+      "step": 3448
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.528725900477356e-05,
+      "loss": 0.0983,
+      "step": 3449
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5284524637751982e-05,
+      "loss": 0.1232,
+      "step": 3450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5281789722416542e-05,
+      "loss": 0.1055,
+      "step": 3451
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5279054259051022e-05,
+      "loss": 0.0968,
+      "step": 3452
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5276318247939246e-05,
+      "loss": 0.1007,
+      "step": 3453
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273581689365093e-05,
+      "loss": 0.0902,
+      "step": 3454
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5270844583612507e-05,
+      "loss": 0.1001,
+      "step": 3455
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526810693096549e-05,
+      "loss": 0.0958,
+      "step": 3456
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526536873170809e-05,
+      "loss": 0.0936,
+      "step": 3457
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5262629986124422e-05,
+      "loss": 0.1108,
+      "step": 3458
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.525989069449865e-05,
+      "loss": 0.1108,
+      "step": 3459
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5257150857115006e-05,
+      "loss": 0.1157,
+      "step": 3460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5254410474257765e-05,
+      "loss": 0.108,
+      "step": 3461
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5251669546211265e-05,
+      "loss": 0.0928,
+      "step": 3462
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.52489280732599e-05,
+      "loss": 0.095,
+      "step": 3463
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5246186055688128e-05,
+      "loss": 0.1078,
+      "step": 3464
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5243443493780445e-05,
+      "loss": 0.1027,
+      "step": 3465
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5240700387821426e-05,
+      "loss": 0.1066,
+      "step": 3466
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5237956738095681e-05,
+      "loss": 0.1086,
+      "step": 3467
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5235212544887891e-05,
+      "loss": 0.0911,
+      "step": 3468
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5232467808482794e-05,
+      "loss": 0.1073,
+      "step": 3469
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5229722529165175e-05,
+      "loss": 0.1146,
+      "step": 3470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5226976707219877e-05,
+      "loss": 0.0822,
+      "step": 3471
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5224230342931807e-05,
+      "loss": 0.0977,
+      "step": 3472
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5221483436585923e-05,
+      "loss": 0.104,
+      "step": 3473
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5218735988467237e-05,
+      "loss": 0.1056,
+      "step": 3474
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5215987998860824e-05,
+      "loss": 0.1018,
+      "step": 3475
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213239468051801e-05,
+      "loss": 0.1177,
+      "step": 3476
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5210490396325361e-05,
+      "loss": 0.1017,
+      "step": 3477
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5207740783966744e-05,
+      "loss": 0.0909,
+      "step": 3478
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520499063126124e-05,
+      "loss": 0.088,
+      "step": 3479
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5202239938494201e-05,
+      "loss": 0.1204,
+      "step": 3480
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5199488705951037e-05,
+      "loss": 0.1066,
+      "step": 3481
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5196736933917211e-05,
+      "loss": 0.1113,
+      "step": 3482
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193984622678241e-05,
+      "loss": 0.1136,
+      "step": 3483
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5191231772519706e-05,
+      "loss": 0.0939,
+      "step": 3484
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.518847838372723e-05,
+      "loss": 0.0956,
+      "step": 3485
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185724456586508e-05,
+      "loss": 0.1022,
+      "step": 3486
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.518296999138328e-05,
+      "loss": 0.1135,
+      "step": 3487
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5180214988403343e-05,
+      "loss": 0.0817,
+      "step": 3488
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5177459447932554e-05,
+      "loss": 0.0991,
+      "step": 3489
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5174703370256823e-05,
+      "loss": 0.1017,
+      "step": 3490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5171946755662116e-05,
+      "loss": 0.0922,
+      "step": 3491
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5169189604434458e-05,
+      "loss": 0.0977,
+      "step": 3492
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5166431916859923e-05,
+      "loss": 0.099,
+      "step": 3493
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5163673693224644e-05,
+      "loss": 0.1035,
+      "step": 3494
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5160914933814809e-05,
+      "loss": 0.0787,
+      "step": 3495
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515815563891667e-05,
+      "loss": 0.09,
+      "step": 3496
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5155395808816518e-05,
+      "loss": 0.1151,
+      "step": 3497
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5152635443800714e-05,
+      "loss": 0.0981,
+      "step": 3498
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149874544155666e-05,
+      "loss": 0.0913,
+      "step": 3499
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5147113110167841e-05,
+      "loss": 0.1023,
+      "step": 3500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5144351142123763e-05,
+      "loss": 0.0981,
+      "step": 3501
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141588640310006e-05,
+      "loss": 0.071,
+      "step": 3502
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5138825605013208e-05,
+      "loss": 0.0912,
+      "step": 3503
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5136062036520054e-05,
+      "loss": 0.0847,
+      "step": 3504
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5133297935117284e-05,
+      "loss": 0.0914,
+      "step": 3505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51305333010917e-05,
+      "loss": 0.0934,
+      "step": 3506
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.512776813473016e-05,
+      "loss": 0.1023,
+      "step": 3507
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125002436319572e-05,
+      "loss": 0.0819,
+      "step": 3508
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5122236206146892e-05,
+      "loss": 0.1039,
+      "step": 3509
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5119469444499148e-05,
+      "loss": 0.1096,
+      "step": 3510
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5116702151663413e-05,
+      "loss": 0.1067,
+      "step": 3511
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5113934327926817e-05,
+      "loss": 0.1005,
+      "step": 3512
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5111165973576545e-05,
+      "loss": 0.0935,
+      "step": 3513
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5108397088899834e-05,
+      "loss": 0.1041,
+      "step": 3514
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510562767418398e-05,
+      "loss": 0.0907,
+      "step": 3515
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5102857729716338e-05,
+      "loss": 0.0938,
+      "step": 3516
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5100087255784308e-05,
+      "loss": 0.099,
+      "step": 3517
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5097316252675352e-05,
+      "loss": 0.097,
+      "step": 3518
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5094544720676985e-05,
+      "loss": 0.1021,
+      "step": 3519
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5091772660076775e-05,
+      "loss": 0.1085,
+      "step": 3520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5089000071162347e-05,
+      "loss": 0.0919,
+      "step": 3521
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5086226954221381e-05,
+      "loss": 0.1083,
+      "step": 3522
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5083453309541613e-05,
+      "loss": 0.1042,
+      "step": 3523
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.508067913741083e-05,
+      "loss": 0.0903,
+      "step": 3524
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5077904438116875e-05,
+      "loss": 0.0923,
+      "step": 3525
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5075129211947647e-05,
+      "loss": 0.1101,
+      "step": 3526
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.50723534591911e-05,
+      "loss": 0.1195,
+      "step": 3527
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5069577180135242e-05,
+      "loss": 0.1106,
+      "step": 3528
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066800375068132e-05,
+      "loss": 0.0996,
+      "step": 3529
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5064023044277891e-05,
+      "loss": 0.0891,
+      "step": 3530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5061245188052689e-05,
+      "loss": 0.083,
+      "step": 3531
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5058466806680749e-05,
+      "loss": 0.1189,
+      "step": 3532
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5055687900450355e-05,
+      "loss": 0.1084,
+      "step": 3533
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5052908469649843e-05,
+      "loss": 0.1166,
+      "step": 3534
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5050128514567598e-05,
+      "loss": 0.0961,
+      "step": 3535
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5047348035492067e-05,
+      "loss": 0.0959,
+      "step": 3536
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5044567032711746e-05,
+      "loss": 0.0867,
+      "step": 3537
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5041785506515187e-05,
+      "loss": 0.0991,
+      "step": 3538
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5039003457191e-05,
+      "loss": 0.1036,
+      "step": 3539
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036220885027843e-05,
+      "loss": 0.0857,
+      "step": 3540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5033437790314436e-05,
+      "loss": 0.107,
+      "step": 3541
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.503065417333954e-05,
+      "loss": 0.0929,
+      "step": 3542
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5027870034391984e-05,
+      "loss": 0.0913,
+      "step": 3543
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5025085373760649e-05,
+      "loss": 0.0997,
+      "step": 3544
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.502230019173446e-05,
+      "loss": 0.1073,
+      "step": 3545
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5019514488602406e-05,
+      "loss": 0.1007,
+      "step": 3546
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5016728264653531e-05,
+      "loss": 0.0908,
+      "step": 3547
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5013941520176922e-05,
+      "loss": 0.1084,
+      "step": 3548
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5011154255461732e-05,
+      "loss": 0.0946,
+      "step": 3549
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5008366470797162e-05,
+      "loss": 0.1036,
+      "step": 3550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500557816647247e-05,
+      "loss": 0.0829,
+      "step": 3551
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.500278934277696e-05,
+      "loss": 0.0892,
+      "step": 3552
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1281,
+      "step": 3553
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4997210138431011e-05,
+      "loss": 0.097,
+      "step": 3554
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.499441975835946e-05,
+      "loss": 0.0973,
+      "step": 3555
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991628860074872e-05,
+      "loss": 0.11,
+      "step": 3556
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4988837443866829e-05,
+      "loss": 0.1023,
+      "step": 3557
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4986045510024965e-05,
+      "loss": 0.1068,
+      "step": 3558
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.498325305883896e-05,
+      "loss": 0.0989,
+      "step": 3559
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4980460090598562e-05,
+      "loss": 0.0884,
+      "step": 3560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4977666605593557e-05,
+      "loss": 0.1013,
+      "step": 3561
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4974872604113801e-05,
+      "loss": 0.1016,
+      "step": 3562
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.497207808644919e-05,
+      "loss": 0.1095,
+      "step": 3563
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496928305288968e-05,
+      "loss": 0.1016,
+      "step": 3564
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496648750372528e-05,
+      "loss": 0.103,
+      "step": 3565
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496369143924605e-05,
+      "loss": 0.0999,
+      "step": 3566
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4960894859742105e-05,
+      "loss": 0.1011,
+      "step": 3567
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4958097765503614e-05,
+      "loss": 0.0812,
+      "step": 3568
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4955300156820805e-05,
+      "loss": 0.0898,
+      "step": 3569
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4952502033983944e-05,
+      "loss": 0.0984,
+      "step": 3570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4949703397283365e-05,
+      "loss": 0.0795,
+      "step": 3571
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4946904247009446e-05,
+      "loss": 0.0953,
+      "step": 3572
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.494410458345263e-05,
+      "loss": 0.1013,
+      "step": 3573
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.49413044069034e-05,
+      "loss": 0.0922,
+      "step": 3574
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.49385037176523e-05,
+      "loss": 0.103,
+      "step": 3575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4935702515989919e-05,
+      "loss": 0.0853,
+      "step": 3576
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493290080220691e-05,
+      "loss": 0.1154,
+      "step": 3577
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4930098576593978e-05,
+      "loss": 0.1023,
+      "step": 3578
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4927295839441875e-05,
+      "loss": 0.0886,
+      "step": 3579
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4924492591041405e-05,
+      "loss": 0.1062,
+      "step": 3580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4921688831683433e-05,
+      "loss": 0.098,
+      "step": 3581
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4918884561658869e-05,
+      "loss": 0.1077,
+      "step": 3582
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4916079781258681e-05,
+      "loss": 0.106,
+      "step": 3583
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.491327449077389e-05,
+      "loss": 0.1116,
+      "step": 3584
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910468690495565e-05,
+      "loss": 0.0951,
+      "step": 3585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4907662380714837e-05,
+      "loss": 0.0933,
+      "step": 3586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4904855561722881e-05,
+      "loss": 0.1066,
+      "step": 3587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4902048233810926e-05,
+      "loss": 0.1085,
+      "step": 3588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4899240397270256e-05,
+      "loss": 0.094,
+      "step": 3589
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4896432052392213e-05,
+      "loss": 0.0974,
+      "step": 3590
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4893623199468184e-05,
+      "loss": 0.1016,
+      "step": 3591
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4890813838789606e-05,
+      "loss": 0.1099,
+      "step": 3592
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4888003970647979e-05,
+      "loss": 0.1055,
+      "step": 3593
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4885193595334847e-05,
+      "loss": 0.0989,
+      "step": 3594
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4882382713141816e-05,
+      "loss": 0.0905,
+      "step": 3595
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4879571324360533e-05,
+      "loss": 0.0917,
+      "step": 3596
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4876759429282705e-05,
+      "loss": 0.0924,
+      "step": 3597
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4873947028200094e-05,
+      "loss": 0.113,
+      "step": 3598
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4871134121404503e-05,
+      "loss": 0.0999,
+      "step": 3599
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4868320709187796e-05,
+      "loss": 0.0692,
+      "step": 3600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865506791841893e-05,
+      "loss": 0.105,
+      "step": 3601
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4862692369658755e-05,
+      "loss": 0.089,
+      "step": 3602
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4859877442930408e-05,
+      "loss": 0.1132,
+      "step": 3603
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4857062011948923e-05,
+      "loss": 0.0978,
+      "step": 3604
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485424607700642e-05,
+      "loss": 0.0897,
+      "step": 3605
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485142963839508e-05,
+      "loss": 0.1017,
+      "step": 3606
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4848612696407135e-05,
+      "loss": 0.1019,
+      "step": 3607
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4845795251334863e-05,
+      "loss": 0.0997,
+      "step": 3608
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4842977303470596e-05,
+      "loss": 0.0934,
+      "step": 3609
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4840158853106718e-05,
+      "loss": 0.0974,
+      "step": 3610
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4837339900535674e-05,
+      "loss": 0.0853,
+      "step": 3611
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4834520446049952e-05,
+      "loss": 0.0884,
+      "step": 3612
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483170048994209e-05,
+      "loss": 0.0931,
+      "step": 3613
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4828880032504684e-05,
+      "loss": 0.0864,
+      "step": 3614
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826059074030381e-05,
+      "loss": 0.0938,
+      "step": 3615
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482323761481188e-05,
+      "loss": 0.0864,
+      "step": 3616
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4820415655141932e-05,
+      "loss": 0.1116,
+      "step": 3617
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4817593195313334e-05,
+      "loss": 0.1083,
+      "step": 3618
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4814770235618942e-05,
+      "loss": 0.088,
+      "step": 3619
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4811946776351667e-05,
+      "loss": 0.0791,
+      "step": 3620
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480912281780446e-05,
+      "loss": 0.0864,
+      "step": 3621
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4806298360270333e-05,
+      "loss": 0.0844,
+      "step": 3622
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4803473404042345e-05,
+      "loss": 0.0897,
+      "step": 3623
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4800647949413615e-05,
+      "loss": 0.1016,
+      "step": 3624
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4797821996677303e-05,
+      "loss": 0.0907,
+      "step": 3625
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4794995546126625e-05,
+      "loss": 0.0983,
+      "step": 3626
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4792168598054847e-05,
+      "loss": 0.0887,
+      "step": 3627
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4789341152755297e-05,
+      "loss": 0.0861,
+      "step": 3628
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4786513210521339e-05,
+      "loss": 0.1047,
+      "step": 3629
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4783684771646397e-05,
+      "loss": 0.0857,
+      "step": 3630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4780855836423946e-05,
+      "loss": 0.0907,
+      "step": 3631
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4778026405147515e-05,
+      "loss": 0.0961,
+      "step": 3632
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4775196478110674e-05,
+      "loss": 0.0972,
+      "step": 3633
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4772366055607057e-05,
+      "loss": 0.1028,
+      "step": 3634
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4769535137930343e-05,
+      "loss": 0.1012,
+      "step": 3635
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766703725374264e-05,
+      "loss": 0.093,
+      "step": 3636
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4763871818232604e-05,
+      "loss": 0.0869,
+      "step": 3637
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4761039416799192e-05,
+      "loss": 0.1106,
+      "step": 3638
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4758206521367919e-05,
+      "loss": 0.0885,
+      "step": 3639
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.475537313223272e-05,
+      "loss": 0.0978,
+      "step": 3640
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4752539249687583e-05,
+      "loss": 0.0923,
+      "step": 3641
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4749704874026547e-05,
+      "loss": 0.0934,
+      "step": 3642
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.47468700055437e-05,
+      "loss": 0.0736,
+      "step": 3643
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4744034644533185e-05,
+      "loss": 0.1132,
+      "step": 3644
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4741198791289196e-05,
+      "loss": 0.1123,
+      "step": 3645
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4738362446105975e-05,
+      "loss": 0.1049,
+      "step": 3646
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4735525609277819e-05,
+      "loss": 0.1113,
+      "step": 3647
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4732688281099072e-05,
+      "loss": 0.1045,
+      "step": 3648
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.472985046186413e-05,
+      "loss": 0.1117,
+      "step": 3649
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4727012151867442e-05,
+      "loss": 0.0786,
+      "step": 3650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4724173351403504e-05,
+      "loss": 0.0919,
+      "step": 3651
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.472133406076687e-05,
+      "loss": 0.0907,
+      "step": 3652
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4718494280252133e-05,
+      "loss": 0.0847,
+      "step": 3653
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4715654010153953e-05,
+      "loss": 0.1012,
+      "step": 3654
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4712813250767024e-05,
+      "loss": 0.0906,
+      "step": 3655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4709972002386104e-05,
+      "loss": 0.0997,
+      "step": 3656
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4707130265305993e-05,
+      "loss": 0.1106,
+      "step": 3657
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4704288039821551e-05,
+      "loss": 0.0911,
+      "step": 3658
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4701445326227675e-05,
+      "loss": 0.0911,
+      "step": 3659
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4698602124819321e-05,
+      "loss": 0.0829,
+      "step": 3660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.46957584358915e-05,
+      "loss": 0.1066,
+      "step": 3661
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4692914259739268e-05,
+      "loss": 0.088,
+      "step": 3662
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4690069596657732e-05,
+      "loss": 0.1116,
+      "step": 3663
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4687224446942045e-05,
+      "loss": 0.1034,
+      "step": 3664
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684378810887422e-05,
+      "loss": 0.0846,
+      "step": 3665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4681532688789114e-05,
+      "loss": 0.0874,
+      "step": 3666
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4678686080942439e-05,
+      "loss": 0.0889,
+      "step": 3667
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.467583898764275e-05,
+      "loss": 0.089,
+      "step": 3668
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4672991409185457e-05,
+      "loss": 0.0836,
+      "step": 3669
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4670143345866024e-05,
+      "loss": 0.1217,
+      "step": 3670
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4667294797979958e-05,
+      "loss": 0.1018,
+      "step": 3671
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4664445765822823e-05,
+      "loss": 0.1051,
+      "step": 3672
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4661596249690227e-05,
+      "loss": 0.121,
+      "step": 3673
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4658746249877833e-05,
+      "loss": 0.1005,
+      "step": 3674
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4655895766681351e-05,
+      "loss": 0.0991,
+      "step": 3675
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4653044800396545e-05,
+      "loss": 0.1072,
+      "step": 3676
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4650193351319224e-05,
+      "loss": 0.0903,
+      "step": 3677
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4647341419745251e-05,
+      "loss": 0.0933,
+      "step": 3678
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4644489005970539e-05,
+      "loss": 0.1199,
+      "step": 3679
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4641636110291051e-05,
+      "loss": 0.0997,
+      "step": 3680
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4638782733002792e-05,
+      "loss": 0.1029,
+      "step": 3681
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.463592887440183e-05,
+      "loss": 0.0967,
+      "step": 3682
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4633074534784278e-05,
+      "loss": 0.0814,
+      "step": 3683
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4630219714446292e-05,
+      "loss": 0.1146,
+      "step": 3684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4627364413684091e-05,
+      "loss": 0.1003,
+      "step": 3685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4624508632793928e-05,
+      "loss": 0.1053,
+      "step": 3686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621652372072122e-05,
+      "loss": 0.0906,
+      "step": 3687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4618795631815027e-05,
+      "loss": 0.1069,
+      "step": 3688
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.461593841231906e-05,
+      "loss": 0.0836,
+      "step": 3689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4613080713880677e-05,
+      "loss": 0.0829,
+      "step": 3690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4610222536796393e-05,
+      "loss": 0.0834,
+      "step": 3691
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4607363881362765e-05,
+      "loss": 0.1018,
+      "step": 3692
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4604504747876397e-05,
+      "loss": 0.0903,
+      "step": 3693
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601645136633959e-05,
+      "loss": 0.0922,
+      "step": 3694
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4598785047932153e-05,
+      "loss": 0.0984,
+      "step": 3695
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4595924482067742e-05,
+      "loss": 0.1039,
+      "step": 3696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4593063439337524e-05,
+      "loss": 0.1066,
+      "step": 3697
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4590201920038367e-05,
+      "loss": 0.1024,
+      "step": 3698
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4587339924467175e-05,
+      "loss": 0.0896,
+      "step": 3699
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.45844774529209e-05,
+      "loss": 0.078,
+      "step": 3700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4581614505696551e-05,
+      "loss": 0.0967,
+      "step": 3701
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457875108309118e-05,
+      "loss": 0.0728,
+      "step": 3702
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4575887185401893e-05,
+      "loss": 0.115,
+      "step": 3703
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4573022812925845e-05,
+      "loss": 0.0876,
+      "step": 3704
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4570157965960236e-05,
+      "loss": 0.0986,
+      "step": 3705
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.456729264480232e-05,
+      "loss": 0.0853,
+      "step": 3706
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.45644268497494e-05,
+      "loss": 0.0997,
+      "step": 3707
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4561560581098819e-05,
+      "loss": 0.0966,
+      "step": 3708
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4558693839147985e-05,
+      "loss": 0.1,
+      "step": 3709
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4555826624194339e-05,
+      "loss": 0.0935,
+      "step": 3710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4552958936535381e-05,
+      "loss": 0.1123,
+      "step": 3711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4550090776468664e-05,
+      "loss": 0.0862,
+      "step": 3712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547222144291777e-05,
+      "loss": 0.0773,
+      "step": 3713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4544353040302364e-05,
+      "loss": 0.1095,
+      "step": 3714
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4541483464798125e-05,
+      "loss": 0.0851,
+      "step": 3715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4538613418076795e-05,
+      "loss": 0.1018,
+      "step": 3716
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4535742900436171e-05,
+      "loss": 0.0837,
+      "step": 3717
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.453287191217409e-05,
+      "loss": 0.0852,
+      "step": 3718
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4530000453588447e-05,
+      "loss": 0.092,
+      "step": 3719
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4527128524977172e-05,
+      "loss": 0.0977,
+      "step": 3720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4524256126638257e-05,
+      "loss": 0.0917,
+      "step": 3721
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4521383258869735e-05,
+      "loss": 0.0961,
+      "step": 3722
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4518509921969687e-05,
+      "loss": 0.1072,
+      "step": 3723
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4515636116236258e-05,
+      "loss": 0.0851,
+      "step": 3724
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4512761841967615e-05,
+      "loss": 0.1042,
+      "step": 3725
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509887099462e-05,
+      "loss": 0.1073,
+      "step": 3726
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4507011889017679e-05,
+      "loss": 0.097,
+      "step": 3727
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.450413621093299e-05,
+      "loss": 0.0811,
+      "step": 3728
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4501260065506306e-05,
+      "loss": 0.1068,
+      "step": 3729
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4498383453036051e-05,
+      "loss": 0.0889,
+      "step": 3730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4495506373820695e-05,
+      "loss": 0.1196,
+      "step": 3731
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4492628828158764e-05,
+      "loss": 0.113,
+      "step": 3732
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4489750816348824e-05,
+      "loss": 0.0786,
+      "step": 3733
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4486872338689492e-05,
+      "loss": 0.0908,
+      "step": 3734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4483993395479439e-05,
+      "loss": 0.0869,
+      "step": 3735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4481113987017375e-05,
+      "loss": 0.0853,
+      "step": 3736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4478234113602063e-05,
+      "loss": 0.0991,
+      "step": 3737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4475353775532316e-05,
+      "loss": 0.1046,
+      "step": 3738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4472472973106997e-05,
+      "loss": 0.1019,
+      "step": 3739
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4469591706625003e-05,
+      "loss": 0.118,
+      "step": 3740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4466709976385303e-05,
+      "loss": 0.1104,
+      "step": 3741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446382778268689e-05,
+      "loss": 0.0897,
+      "step": 3742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.446094512582882e-05,
+      "loss": 0.1077,
+      "step": 3743
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445806200611019e-05,
+      "loss": 0.0897,
+      "step": 3744
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4455178423830152e-05,
+      "loss": 0.0963,
+      "step": 3745
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.44522943792879e-05,
+      "loss": 0.0897,
+      "step": 3746
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444940987278268e-05,
+      "loss": 0.0957,
+      "step": 3747
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444652490461378e-05,
+      "loss": 0.0931,
+      "step": 3748
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444363947508054e-05,
+      "loss": 0.1051,
+      "step": 3749
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4440753584482351e-05,
+      "loss": 0.1082,
+      "step": 3750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4437867233118647e-05,
+      "loss": 0.098,
+      "step": 3751
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4434980421288911e-05,
+      "loss": 0.0756,
+      "step": 3752
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4432093149292672e-05,
+      "loss": 0.0973,
+      "step": 3753
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4429205417429513e-05,
+      "loss": 0.1107,
+      "step": 3754
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4426317225999055e-05,
+      "loss": 0.1072,
+      "step": 3755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4423428575300978e-05,
+      "loss": 0.1061,
+      "step": 3756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4420539465635e-05,
+      "loss": 0.0683,
+      "step": 3757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4417649897300891e-05,
+      "loss": 0.0814,
+      "step": 3758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4414759870598467e-05,
+      "loss": 0.0944,
+      "step": 3759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4411869385827592e-05,
+      "loss": 0.0947,
+      "step": 3760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4408978443288186e-05,
+      "loss": 0.0891,
+      "step": 3761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4406087043280199e-05,
+      "loss": 0.0704,
+      "step": 3762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4403195186103644e-05,
+      "loss": 0.0958,
+      "step": 3763
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4400302872058568e-05,
+      "loss": 0.095,
+      "step": 3764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4397410101445082e-05,
+      "loss": 0.0974,
+      "step": 3765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4394516874563332e-05,
+      "loss": 0.1235,
+      "step": 3766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4391623191713513e-05,
+      "loss": 0.083,
+      "step": 3767
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4388729053195869e-05,
+      "loss": 0.0974,
+      "step": 3768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438583445931069e-05,
+      "loss": 0.0909,
+      "step": 3769
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438293941035832e-05,
+      "loss": 0.1007,
+      "step": 3770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.438004390663914e-05,
+      "loss": 0.0985,
+      "step": 3771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4377147948453584e-05,
+      "loss": 0.1025,
+      "step": 3772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4374251536102131e-05,
+      "loss": 0.1038,
+      "step": 3773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4371354669885312e-05,
+      "loss": 0.0853,
+      "step": 3774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4368457350103695e-05,
+      "loss": 0.0881,
+      "step": 3775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4365559577057905e-05,
+      "loss": 0.0907,
+      "step": 3776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.436266135104861e-05,
+      "loss": 0.0939,
+      "step": 3777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4359762672376528e-05,
+      "loss": 0.0981,
+      "step": 3778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4356863541342416e-05,
+      "loss": 0.0814,
+      "step": 3779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4353963958247086e-05,
+      "loss": 0.0837,
+      "step": 3780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4351063923391393e-05,
+      "loss": 0.0852,
+      "step": 3781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4348163437076243e-05,
+      "loss": 0.1038,
+      "step": 3782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345262499602581e-05,
+      "loss": 0.0988,
+      "step": 3783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4342361111271408e-05,
+      "loss": 0.0831,
+      "step": 3784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4339459272383766e-05,
+      "loss": 0.0908,
+      "step": 3785
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4336556983240747e-05,
+      "loss": 0.0876,
+      "step": 3786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4333654244143482e-05,
+      "loss": 0.0923,
+      "step": 3787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4330751055393162e-05,
+      "loss": 0.0977,
+      "step": 3788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4327847417291009e-05,
+      "loss": 0.0859,
+      "step": 3789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.432494333013831e-05,
+      "loss": 0.1045,
+      "step": 3790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4322038794236379e-05,
+      "loss": 0.0835,
+      "step": 3791
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4319133809886592e-05,
+      "loss": 0.1052,
+      "step": 3792
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4316228377390363e-05,
+      "loss": 0.0908,
+      "step": 3793
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4313322497049153e-05,
+      "loss": 0.1128,
+      "step": 3794
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4310416169164477e-05,
+      "loss": 0.0764,
+      "step": 3795
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307509394037888e-05,
+      "loss": 0.095,
+      "step": 3796
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.430460217197099e-05,
+      "loss": 0.1014,
+      "step": 3797
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4301694503265426e-05,
+      "loss": 0.1044,
+      "step": 3798
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4298786388222895e-05,
+      "loss": 0.0946,
+      "step": 3799
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4295877827145144e-05,
+      "loss": 0.0855,
+      "step": 3800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4292968820333953e-05,
+      "loss": 0.1039,
+      "step": 3801
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290059368091156e-05,
+      "loss": 0.1112,
+      "step": 3802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4287149470718635e-05,
+      "loss": 0.073,
+      "step": 3803
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428423912851832e-05,
+      "loss": 0.0995,
+      "step": 3804
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4281328341792178e-05,
+      "loss": 0.0807,
+      "step": 3805
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427841711084223e-05,
+      "loss": 0.0814,
+      "step": 3806
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427550543597054e-05,
+      "loss": 0.0875,
+      "step": 3807
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427259331747922e-05,
+      "loss": 0.0866,
+      "step": 3808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4269680755670425e-05,
+      "loss": 0.09,
+      "step": 3809
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.426676775084636e-05,
+      "loss": 0.1008,
+      "step": 3810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4263854303309268e-05,
+      "loss": 0.0773,
+      "step": 3811
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4260940413361452e-05,
+      "loss": 0.1022,
+      "step": 3812
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4258026081305252e-05,
+      "loss": 0.1128,
+      "step": 3813
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4255111307443046e-05,
+      "loss": 0.1134,
+      "step": 3814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.425219609207727e-05,
+      "loss": 0.1066,
+      "step": 3815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4249280435510407e-05,
+      "loss": 0.101,
+      "step": 3816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4246364338044977e-05,
+      "loss": 0.0929,
+      "step": 3817
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424344779998355e-05,
+      "loss": 0.0958,
+      "step": 3818
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424053082162874e-05,
+      "loss": 0.102,
+      "step": 3819
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423761340328321e-05,
+      "loss": 0.0894,
+      "step": 3820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4234695545249666e-05,
+      "loss": 0.1017,
+      "step": 3821
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423177724783086e-05,
+      "loss": 0.0857,
+      "step": 3822
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4228858511329591e-05,
+      "loss": 0.0895,
+      "step": 3823
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4225939336048703e-05,
+      "loss": 0.0898,
+      "step": 3824
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.422301972229108e-05,
+      "loss": 0.084,
+      "step": 3825
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220099670359664e-05,
+      "loss": 0.0958,
+      "step": 3826
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4217179180557428e-05,
+      "loss": 0.0859,
+      "step": 3827
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4214258253187401e-05,
+      "loss": 0.0885,
+      "step": 3828
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4211336888552657e-05,
+      "loss": 0.0933,
+      "step": 3829
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4208415086956305e-05,
+      "loss": 0.0771,
+      "step": 3830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4205492848701507e-05,
+      "loss": 0.0819,
+      "step": 3831
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4202570174091474e-05,
+      "loss": 0.1173,
+      "step": 3832
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.419964706342946e-05,
+      "loss": 0.0842,
+      "step": 3833
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4196723517018757e-05,
+      "loss": 0.0854,
+      "step": 3834
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4193799535162711e-05,
+      "loss": 0.0977,
+      "step": 3835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4190875118164706e-05,
+      "loss": 0.0729,
+      "step": 3836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4187950266328179e-05,
+      "loss": 0.1138,
+      "step": 3837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4185024979956602e-05,
+      "loss": 0.0829,
+      "step": 3838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4182099259353508e-05,
+      "loss": 0.0946,
+      "step": 3839
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4179173104822454e-05,
+      "loss": 0.1031,
+      "step": 3840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4176246516667061e-05,
+      "loss": 0.0818,
+      "step": 3841
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4173319495190984e-05,
+      "loss": 0.0968,
+      "step": 3842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4170392040697926e-05,
+      "loss": 0.079,
+      "step": 3843
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4167464153491634e-05,
+      "loss": 0.0956,
+      "step": 3844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4164535833875905e-05,
+      "loss": 0.0914,
+      "step": 3845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161607082154575e-05,
+      "loss": 0.0789,
+      "step": 3846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4158677898631524e-05,
+      "loss": 0.1035,
+      "step": 3847
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.415574828361068e-05,
+      "loss": 0.0925,
+      "step": 3848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4152818237396017e-05,
+      "loss": 0.1,
+      "step": 3849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4149887760291552e-05,
+      "loss": 0.0996,
+      "step": 3850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4146956852601349e-05,
+      "loss": 0.1008,
+      "step": 3851
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4144025514629504e-05,
+      "loss": 0.0879,
+      "step": 3852
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4141093746680182e-05,
+      "loss": 0.104,
+      "step": 3853
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.413816154905757e-05,
+      "loss": 0.0973,
+      "step": 3854
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4135228922065909e-05,
+      "loss": 0.0889,
+      "step": 3855
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4132295866009482e-05,
+      "loss": 0.0944,
+      "step": 3856
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4129362381192626e-05,
+      "loss": 0.1027,
+      "step": 3857
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4126428467919707e-05,
+      "loss": 0.1062,
+      "step": 3858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4123494126495145e-05,
+      "loss": 0.0995,
+      "step": 3859
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4120559357223407e-05,
+      "loss": 0.098,
+      "step": 3860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4117624160408991e-05,
+      "loss": 0.0968,
+      "step": 3861
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4114688536356457e-05,
+      "loss": 0.0936,
+      "step": 3862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4111752485370399e-05,
+      "loss": 0.0898,
+      "step": 3863
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4108816007755452e-05,
+      "loss": 0.0942,
+      "step": 3864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4105879103816303e-05,
+      "loss": 0.0844,
+      "step": 3865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4102941773857683e-05,
+      "loss": 0.1034,
+      "step": 3866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410000401818436e-05,
+      "loss": 0.0986,
+      "step": 3867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4097065837101161e-05,
+      "loss": 0.0956,
+      "step": 3868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4094127230912931e-05,
+      "loss": 0.0941,
+      "step": 3869
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4091188199924589e-05,
+      "loss": 0.0991,
+      "step": 3870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4088248744441075e-05,
+      "loss": 0.0977,
+      "step": 3871
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4085308864767389e-05,
+      "loss": 0.0956,
+      "step": 3872
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4082368561208564e-05,
+      "loss": 0.1023,
+      "step": 3873
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407942783406968e-05,
+      "loss": 0.0973,
+      "step": 3874
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407648668365587e-05,
+      "loss": 0.0924,
+      "step": 3875
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4073545110272295e-05,
+      "loss": 0.0868,
+      "step": 3876
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.407060311422417e-05,
+      "loss": 0.0875,
+      "step": 3877
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4067660695816751e-05,
+      "loss": 0.0914,
+      "step": 3878
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064717855355345e-05,
+      "loss": 0.0956,
+      "step": 3879
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4061774593145288e-05,
+      "loss": 0.0973,
+      "step": 3880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058830909491971e-05,
+      "loss": 0.0978,
+      "step": 3881
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.405588680470083e-05,
+      "loss": 0.1034,
+      "step": 3882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4052942279077334e-05,
+      "loss": 0.0891,
+      "step": 3883
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4049997332927007e-05,
+      "loss": 0.0985,
+      "step": 3884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4047051966555412e-05,
+      "loss": 0.1073,
+      "step": 3885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4044106180268152e-05,
+      "loss": 0.0848,
+      "step": 3886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4041159974370881e-05,
+      "loss": 0.0864,
+      "step": 3887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403821334916929e-05,
+      "loss": 0.0897,
+      "step": 3888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4035266304969115e-05,
+      "loss": 0.0864,
+      "step": 3889
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403231884207614e-05,
+      "loss": 0.0892,
+      "step": 3890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4029370960796189e-05,
+      "loss": 0.0896,
+      "step": 3891
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4026422661435127e-05,
+      "loss": 0.0875,
+      "step": 3892
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4023473944298864e-05,
+      "loss": 0.1028,
+      "step": 3893
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4020524809693356e-05,
+      "loss": 0.1,
+      "step": 3894
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4017575257924603e-05,
+      "loss": 0.1084,
+      "step": 3895
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4014625289298645e-05,
+      "loss": 0.0989,
+      "step": 3896
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4011674904121562e-05,
+      "loss": 0.0983,
+      "step": 3897
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.400872410269948e-05,
+      "loss": 0.1057,
+      "step": 3898
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4005772885338578e-05,
+      "loss": 0.0909,
+      "step": 3899
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4002821252345062e-05,
+      "loss": 0.0981,
+      "step": 3900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3999869204025197e-05,
+      "loss": 0.105,
+      "step": 3901
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.399691674068527e-05,
+      "loss": 0.1041,
+      "step": 3902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3993963862631637e-05,
+      "loss": 0.1162,
+      "step": 3903
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3991010570170673e-05,
+      "loss": 0.0952,
+      "step": 3904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3988056863608815e-05,
+      "loss": 0.11,
+      "step": 3905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3985102743252532e-05,
+      "loss": 0.0905,
+      "step": 3906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.398214820940834e-05,
+      "loss": 0.0844,
+      "step": 3907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3979193262382791e-05,
+      "loss": 0.1058,
+      "step": 3908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3976237902482495e-05,
+      "loss": 0.0917,
+      "step": 3909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3973282130014087e-05,
+      "loss": 0.106,
+      "step": 3910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970325945284255e-05,
+      "loss": 0.0881,
+      "step": 3911
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3967369348599738e-05,
+      "loss": 0.1084,
+      "step": 3912
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3964412340267293e-05,
+      "loss": 0.094,
+      "step": 3913
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3961454920593743e-05,
+      "loss": 0.0952,
+      "step": 3914
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3958497089885939e-05,
+      "loss": 0.098,
+      "step": 3915
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3955538848450787e-05,
+      "loss": 0.1047,
+      "step": 3916
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3952580196595232e-05,
+      "loss": 0.1288,
+      "step": 3917
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3949621134626253e-05,
+      "loss": 0.0942,
+      "step": 3918
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3946661662850874e-05,
+      "loss": 0.0856,
+      "step": 3919
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3943701781576172e-05,
+      "loss": 0.08,
+      "step": 3920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3940741491109258e-05,
+      "loss": 0.0908,
+      "step": 3921
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3937780791757287e-05,
+      "loss": 0.1075,
+      "step": 3922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3934819683827457e-05,
+      "loss": 0.0981,
+      "step": 3923
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3931858167627007e-05,
+      "loss": 0.0979,
+      "step": 3924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3928896243463218e-05,
+      "loss": 0.0807,
+      "step": 3925
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3925933911643415e-05,
+      "loss": 0.0986,
+      "step": 3926
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3922971172474964e-05,
+      "loss": 0.1046,
+      "step": 3927
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3920008026265278e-05,
+      "loss": 0.0809,
+      "step": 3928
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3917044473321805e-05,
+      "loss": 0.0786,
+      "step": 3929
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.391408051395204e-05,
+      "loss": 0.0968,
+      "step": 3930
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3911116148463517e-05,
+      "loss": 0.0875,
+      "step": 3931
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3908151377163815e-05,
+      "loss": 0.092,
+      "step": 3932
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3905186200360555e-05,
+      "loss": 0.1042,
+      "step": 3933
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3902220618361399e-05,
+      "loss": 0.0917,
+      "step": 3934
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3899254631474048e-05,
+      "loss": 0.0875,
+      "step": 3935
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3896288240006249e-05,
+      "loss": 0.0898,
+      "step": 3936
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3893321444265793e-05,
+      "loss": 0.0906,
+      "step": 3937
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3890354244560507e-05,
+      "loss": 0.0828,
+      "step": 3938
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3887386641198265e-05,
+      "loss": 0.1007,
+      "step": 3939
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3884418634486978e-05,
+      "loss": 0.1053,
+      "step": 3940
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3881450224734604e-05,
+      "loss": 0.113,
+      "step": 3941
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.387848141224914e-05,
+      "loss": 0.0867,
+      "step": 3942
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3875512197338628e-05,
+      "loss": 0.085,
+      "step": 3943
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3872542580311144e-05,
+      "loss": 0.0907,
+      "step": 3944
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386957256147481e-05,
+      "loss": 0.0931,
+      "step": 3945
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3866602141137797e-05,
+      "loss": 0.1078,
+      "step": 3946
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3863631319608306e-05,
+      "loss": 0.0785,
+      "step": 3947
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3860660097194584e-05,
+      "loss": 0.0779,
+      "step": 3948
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3857688474204926e-05,
+      "loss": 0.0958,
+      "step": 3949
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3854716450947658e-05,
+      "loss": 0.1007,
+      "step": 3950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3851744027731156e-05,
+      "loss": 0.0957,
+      "step": 3951
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3848771204863827e-05,
+      "loss": 0.0745,
+      "step": 3952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3845797982654134e-05,
+      "loss": 0.0958,
+      "step": 3953
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.384282436141057e-05,
+      "loss": 0.0836,
+      "step": 3954
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3839850341441674e-05,
+      "loss": 0.0924,
+      "step": 3955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3836875923056026e-05,
+      "loss": 0.0967,
+      "step": 3956
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3833901106562245e-05,
+      "loss": 0.0876,
+      "step": 3957
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3830925892268994e-05,
+      "loss": 0.108,
+      "step": 3958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3827950280484981e-05,
+      "loss": 0.1096,
+      "step": 3959
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3824974271518943e-05,
+      "loss": 0.0989,
+      "step": 3960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3821997865679669e-05,
+      "loss": 0.0945,
+      "step": 3961
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381902106327599e-05,
+      "loss": 0.0941,
+      "step": 3962
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381604386461677e-05,
+      "loss": 0.104,
+      "step": 3963
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3813066270010919e-05,
+      "loss": 0.1014,
+      "step": 3964
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3810088279767389e-05,
+      "loss": 0.0919,
+      "step": 3965
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3807109894195169e-05,
+      "loss": 0.0924,
+      "step": 3966
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3804131113603299e-05,
+      "loss": 0.1028,
+      "step": 3967
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.380115193830084e-05,
+      "loss": 0.0956,
+      "step": 3968
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3798172368596913e-05,
+      "loss": 0.1023,
+      "step": 3969
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3795192404800677e-05,
+      "loss": 0.0928,
+      "step": 3970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3792212047221326e-05,
+      "loss": 0.0885,
+      "step": 3971
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.378923129616809e-05,
+      "loss": 0.0826,
+      "step": 3972
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3786250151950257e-05,
+      "loss": 0.0951,
+      "step": 3973
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3783268614877144e-05,
+      "loss": 0.0924,
+      "step": 3974
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3780286685258104e-05,
+      "loss": 0.092,
+      "step": 3975
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3777304363402544e-05,
+      "loss": 0.0947,
+      "step": 3976
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3774321649619902e-05,
+      "loss": 0.0914,
+      "step": 3977
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3771338544219657e-05,
+      "loss": 0.1167,
+      "step": 3978
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3768355047511339e-05,
+      "loss": 0.0963,
+      "step": 3979
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3765371159804503e-05,
+      "loss": 0.0967,
+      "step": 3980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3762386881408759e-05,
+      "loss": 0.0872,
+      "step": 3981
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3759402212633743e-05,
+      "loss": 0.0775,
+      "step": 3982
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3756417153789148e-05,
+      "loss": 0.1058,
+      "step": 3983
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3753431705184694e-05,
+      "loss": 0.0918,
+      "step": 3984
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3750445867130148e-05,
+      "loss": 0.0989,
+      "step": 3985
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3747459639935312e-05,
+      "loss": 0.0833,
+      "step": 3986
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3744473023910039e-05,
+      "loss": 0.1023,
+      "step": 3987
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3741486019364212e-05,
+      "loss": 0.0767,
+      "step": 3988
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3738498626607758e-05,
+      "loss": 0.0983,
+      "step": 3989
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373551084595064e-05,
+      "loss": 0.0934,
+      "step": 3990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3732522677702873e-05,
+      "loss": 0.0883,
+      "step": 3991
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.37295341221745e-05,
+      "loss": 0.0864,
+      "step": 3992
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372654517967561e-05,
+      "loss": 0.1039,
+      "step": 3993
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372355585051633e-05,
+      "loss": 0.0895,
+      "step": 3994
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.372056613500683e-05,
+      "loss": 0.1002,
+      "step": 3995
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3717576033457313e-05,
+      "loss": 0.1055,
+      "step": 3996
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3714585546178033e-05,
+      "loss": 0.0878,
+      "step": 3997
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3711594673479279e-05,
+      "loss": 0.103,
+      "step": 3998
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3708603415671369e-05,
+      "loss": 0.1117,
+      "step": 3999
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3705611773064684e-05,
+      "loss": 0.0762,
+      "step": 4000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3702619745969628e-05,
+      "loss": 0.0897,
+      "step": 4001
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3699627334696643e-05,
+      "loss": 0.0873,
+      "step": 4002
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3696634539556221e-05,
+      "loss": 0.09,
+      "step": 4003
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3693641360858891e-05,
+      "loss": 0.0918,
+      "step": 4004
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.369064779891522e-05,
+      "loss": 0.0845,
+      "step": 4005
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3687653854035813e-05,
+      "loss": 0.084,
+      "step": 4006
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368465952653132e-05,
+      "loss": 0.1,
+      "step": 4007
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3681664816712428e-05,
+      "loss": 0.1221,
+      "step": 4008
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.367866972488986e-05,
+      "loss": 0.0917,
+      "step": 4009
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3675674251374382e-05,
+      "loss": 0.0818,
+      "step": 4010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.36726783964768e-05,
+      "loss": 0.093,
+      "step": 4011
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3669682160507964e-05,
+      "loss": 0.0885,
+      "step": 4012
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3666685543778755e-05,
+      "loss": 0.0958,
+      "step": 4013
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3663688546600093e-05,
+      "loss": 0.0948,
+      "step": 4014
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3660691169282946e-05,
+      "loss": 0.0811,
+      "step": 4015
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3657693412138318e-05,
+      "loss": 0.0927,
+      "step": 4016
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3654695275477252e-05,
+      "loss": 0.1008,
+      "step": 4017
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3651696759610827e-05,
+      "loss": 0.098,
+      "step": 4018
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3648697864850162e-05,
+      "loss": 0.0809,
+      "step": 4019
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3645698591506423e-05,
+      "loss": 0.0884,
+      "step": 4020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3642698939890808e-05,
+      "loss": 0.0911,
+      "step": 4021
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3639698910314556e-05,
+      "loss": 0.0936,
+      "step": 4022
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.622516556291392e-08,
+      "loss": 0.0814,
+      "step": 4023
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3245033112582784e-07,
+      "loss": 0.0945,
+      "step": 4024
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9867549668874176e-07,
+      "loss": 0.089,
+      "step": 4025
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.649006622516557e-07,
+      "loss": 0.0983,
+      "step": 4026
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.311258278145696e-07,
+      "loss": 0.0941,
+      "step": 4027
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.973509933774835e-07,
+      "loss": 0.0897,
+      "step": 4028
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.635761589403974e-07,
+      "loss": 0.0826,
+      "step": 4029
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298013245033113e-07,
+      "loss": 0.0822,
+      "step": 4030
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.960264900662252e-07,
+      "loss": 0.0925,
+      "step": 4031
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.622516556291392e-07,
+      "loss": 0.0868,
+      "step": 4032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.28476821192053e-07,
+      "loss": 0.0945,
+      "step": 4033
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.94701986754967e-07,
+      "loss": 0.0798,
+      "step": 4034
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 8.609271523178808e-07,
+      "loss": 0.08,
+      "step": 4035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.271523178807948e-07,
+      "loss": 0.0903,
+      "step": 4036
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.933774834437087e-07,
+      "loss": 0.1016,
+      "step": 4037
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0596026490066227e-06,
+      "loss": 0.0831,
+      "step": 4038
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1258278145695367e-06,
+      "loss": 0.0924,
+      "step": 4039
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1920529801324504e-06,
+      "loss": 0.1115,
+      "step": 4040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2582781456953644e-06,
+      "loss": 0.0804,
+      "step": 4041
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3245033112582784e-06,
+      "loss": 0.075,
+      "step": 4042
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3907284768211921e-06,
+      "loss": 0.0687,
+      "step": 4043
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.456953642384106e-06,
+      "loss": 0.0877,
+      "step": 4044
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.52317880794702e-06,
+      "loss": 0.0967,
+      "step": 4045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.589403973509934e-06,
+      "loss": 0.0747,
+      "step": 4046
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.655629139072848e-06,
+      "loss": 0.0858,
+      "step": 4047
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.7218543046357616e-06,
+      "loss": 0.0983,
+      "step": 4048
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.7880794701986755e-06,
+      "loss": 0.1022,
+      "step": 4049
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8543046357615895e-06,
+      "loss": 0.1002,
+      "step": 4050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9205298013245035e-06,
+      "loss": 0.0878,
+      "step": 4051
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9867549668874175e-06,
+      "loss": 0.0854,
+      "step": 4052
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.0529801324503314e-06,
+      "loss": 0.0809,
+      "step": 4053
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.1192052980132454e-06,
+      "loss": 0.0974,
+      "step": 4054
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.1854304635761594e-06,
+      "loss": 0.0929,
+      "step": 4055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2516556291390733e-06,
+      "loss": 0.0819,
+      "step": 4056
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.317880794701987e-06,
+      "loss": 0.0756,
+      "step": 4057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.384105960264901e-06,
+      "loss": 0.097,
+      "step": 4058
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.450331125827815e-06,
+      "loss": 0.1016,
+      "step": 4059
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.516556291390729e-06,
+      "loss": 0.1075,
+      "step": 4060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.5827814569536424e-06,
+      "loss": 0.1027,
+      "step": 4061
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.6490066225165567e-06,
+      "loss": 0.0823,
+      "step": 4062
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7152317880794703e-06,
+      "loss": 0.0861,
+      "step": 4063
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7814569536423843e-06,
+      "loss": 0.0848,
+      "step": 4064
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8476821192052982e-06,
+      "loss": 0.0693,
+      "step": 4065
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.913907284768212e-06,
+      "loss": 0.1013,
+      "step": 4066
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.980132450331126e-06,
+      "loss": 0.0699,
+      "step": 4067
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.04635761589404e-06,
+      "loss": 0.0995,
+      "step": 4068
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.1125827814569537e-06,
+      "loss": 0.0999,
+      "step": 4069
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.178807947019868e-06,
+      "loss": 0.067,
+      "step": 4070
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2450331125827816e-06,
+      "loss": 0.0697,
+      "step": 4071
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.311258278145696e-06,
+      "loss": 0.1117,
+      "step": 4072
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3774834437086096e-06,
+      "loss": 0.0973,
+      "step": 4073
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.443708609271523e-06,
+      "loss": 0.0875,
+      "step": 4074
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.5099337748344375e-06,
+      "loss": 0.0752,
+      "step": 4075
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.576158940397351e-06,
+      "loss": 0.0879,
+      "step": 4076
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.642384105960265e-06,
+      "loss": 0.1086,
+      "step": 4077
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.708609271523179e-06,
+      "loss": 0.1035,
+      "step": 4078
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.774834437086093e-06,
+      "loss": 0.098,
+      "step": 4079
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.841059602649007e-06,
+      "loss": 0.0752,
+      "step": 4080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9072847682119205e-06,
+      "loss": 0.0896,
+      "step": 4081
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.973509933774835e-06,
+      "loss": 0.0877,
+      "step": 4082
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.0397350993377485e-06,
+      "loss": 0.0903,
+      "step": 4083
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.105960264900663e-06,
+      "loss": 0.0832,
+      "step": 4084
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.172185430463576e-06,
+      "loss": 0.0798,
+      "step": 4085
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.238410596026491e-06,
+      "loss": 0.0839,
+      "step": 4086
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.304635761589404e-06,
+      "loss": 0.0717,
+      "step": 4087
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.370860927152319e-06,
+      "loss": 0.0891,
+      "step": 4088
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.437086092715232e-06,
+      "loss": 0.0765,
+      "step": 4089
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.503311258278147e-06,
+      "loss": 0.0807,
+      "step": 4090
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.56953642384106e-06,
+      "loss": 0.106,
+      "step": 4091
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.635761589403974e-06,
+      "loss": 0.0886,
+      "step": 4092
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.701986754966888e-06,
+      "loss": 0.0881,
+      "step": 4093
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.768211920529802e-06,
+      "loss": 0.0944,
+      "step": 4094
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.834437086092716e-06,
+      "loss": 0.0825,
+      "step": 4095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.90066225165563e-06,
+      "loss": 0.0881,
+      "step": 4096
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.966887417218543e-06,
+      "loss": 0.0817,
+      "step": 4097
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.033112582781458e-06,
+      "loss": 0.0977,
+      "step": 4098
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.099337748344372e-06,
+      "loss": 0.0885,
+      "step": 4099
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.165562913907285e-06,
+      "loss": 0.0962,
+      "step": 4100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.231788079470199e-06,
+      "loss": 0.085,
+      "step": 4101
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.2980132450331135e-06,
+      "loss": 0.0925,
+      "step": 4102
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.364238410596026e-06,
+      "loss": 0.0941,
+      "step": 4103
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.430463576158941e-06,
+      "loss": 0.1117,
+      "step": 4104
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.496688741721855e-06,
+      "loss": 0.0953,
+      "step": 4105
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.5629139072847685e-06,
+      "loss": 0.0875,
+      "step": 4106
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.629139072847682e-06,
+      "loss": 0.0931,
+      "step": 4107
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.6953642384105965e-06,
+      "loss": 0.0864,
+      "step": 4108
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.76158940397351e-06,
+      "loss": 0.0908,
+      "step": 4109
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.827814569536424e-06,
+      "loss": 0.0865,
+      "step": 4110
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.894039735099338e-06,
+      "loss": 0.0792,
+      "step": 4111
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.960264900662252e-06,
+      "loss": 0.0789,
+      "step": 4112
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.026490066225166e-06,
+      "loss": 0.0815,
+      "step": 4113
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.09271523178808e-06,
+      "loss": 0.0923,
+      "step": 4114
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.158940397350994e-06,
+      "loss": 0.095,
+      "step": 4115
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.225165562913907e-06,
+      "loss": 0.0778,
+      "step": 4116
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.291390728476822e-06,
+      "loss": 0.0987,
+      "step": 4117
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.357615894039736e-06,
+      "loss": 0.0985,
+      "step": 4118
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.423841059602649e-06,
+      "loss": 0.093,
+      "step": 4119
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.490066225165563e-06,
+      "loss": 0.0812,
+      "step": 4120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.556291390728478e-06,
+      "loss": 0.0845,
+      "step": 4121
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.622516556291392e-06,
+      "loss": 0.1029,
+      "step": 4122
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.688741721854305e-06,
+      "loss": 0.0934,
+      "step": 4123
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.754966887417219e-06,
+      "loss": 0.0994,
+      "step": 4124
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.8211920529801336e-06,
+      "loss": 0.071,
+      "step": 4125
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.887417218543046e-06,
+      "loss": 0.093,
+      "step": 4126
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.953642384105961e-06,
+      "loss": 0.1094,
+      "step": 4127
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.019867549668875e-06,
+      "loss": 0.1061,
+      "step": 4128
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.086092715231789e-06,
+      "loss": 0.0803,
+      "step": 4129
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.152317880794702e-06,
+      "loss": 0.0908,
+      "step": 4130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.2185430463576166e-06,
+      "loss": 0.1034,
+      "step": 4131
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.28476821192053e-06,
+      "loss": 0.0743,
+      "step": 4132
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.3509933774834445e-06,
+      "loss": 0.0853,
+      "step": 4133
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.417218543046358e-06,
+      "loss": 0.0912,
+      "step": 4134
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.4834437086092724e-06,
+      "loss": 0.076,
+      "step": 4135
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.549668874172186e-06,
+      "loss": 0.073,
+      "step": 4136
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.6158940397351e-06,
+      "loss": 0.0895,
+      "step": 4137
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.682119205298014e-06,
+      "loss": 0.0966,
+      "step": 4138
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.748344370860927e-06,
+      "loss": 0.097,
+      "step": 4139
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.814569536423841e-06,
+      "loss": 0.0872,
+      "step": 4140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.880794701986755e-06,
+      "loss": 0.0928,
+      "step": 4141
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.94701986754967e-06,
+      "loss": 0.0938,
+      "step": 4142
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.013245033112583e-06,
+      "loss": 0.0875,
+      "step": 4143
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.079470198675497e-06,
+      "loss": 0.0867,
+      "step": 4144
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.145695364238411e-06,
+      "loss": 0.094,
+      "step": 4145
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.211920529801326e-06,
+      "loss": 0.0889,
+      "step": 4146
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.278145695364238e-06,
+      "loss": 0.0933,
+      "step": 4147
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.344370860927153e-06,
+      "loss": 0.0929,
+      "step": 4148
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.410596026490067e-06,
+      "loss": 0.0909,
+      "step": 4149
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.476821192052982e-06,
+      "loss": 0.1036,
+      "step": 4150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.543046357615894e-06,
+      "loss": 0.0835,
+      "step": 4151
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 8.609271523178809e-06,
+      "loss": 0.0814,
+      "step": 4152
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.675496688741723e-06,
+      "loss": 0.0935,
+      "step": 4153
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.741721854304637e-06,
+      "loss": 0.0833,
+      "step": 4154
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.80794701986755e-06,
+      "loss": 0.072,
+      "step": 4155
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.874172185430465e-06,
+      "loss": 0.0826,
+      "step": 4156
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 8.940397350993379e-06,
+      "loss": 0.0905,
+      "step": 4157
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.006622516556293e-06,
+      "loss": 0.0752,
+      "step": 4158
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.072847682119206e-06,
+      "loss": 0.0874,
+      "step": 4159
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.13907284768212e-06,
+      "loss": 0.076,
+      "step": 4160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.205298013245035e-06,
+      "loss": 0.093,
+      "step": 4161
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.271523178807948e-06,
+      "loss": 0.0885,
+      "step": 4162
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.337748344370862e-06,
+      "loss": 0.0936,
+      "step": 4163
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.403973509933776e-06,
+      "loss": 0.0762,
+      "step": 4164
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.470198675496689e-06,
+      "loss": 0.0758,
+      "step": 4165
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.536423841059603e-06,
+      "loss": 0.0907,
+      "step": 4166
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.602649006622518e-06,
+      "loss": 0.0773,
+      "step": 4167
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.668874172185432e-06,
+      "loss": 0.0858,
+      "step": 4168
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.735099337748345e-06,
+      "loss": 0.097,
+      "step": 4169
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.80132450331126e-06,
+      "loss": 0.0909,
+      "step": 4170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.867549668874174e-06,
+      "loss": 0.0821,
+      "step": 4171
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.933774834437086e-06,
+      "loss": 0.0998,
+      "step": 4172
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1e-05,
+      "loss": 0.0756,
+      "step": 4173
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0066225165562915e-05,
+      "loss": 0.0797,
+      "step": 4174
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.013245033112583e-05,
+      "loss": 0.0763,
+      "step": 4175
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0198675496688744e-05,
+      "loss": 0.0801,
+      "step": 4176
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0264900662251655e-05,
+      "loss": 0.0917,
+      "step": 4177
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.033112582781457e-05,
+      "loss": 0.0757,
+      "step": 4178
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0397350993377484e-05,
+      "loss": 0.0889,
+      "step": 4179
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0463576158940398e-05,
+      "loss": 0.0948,
+      "step": 4180
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0529801324503313e-05,
+      "loss": 0.0981,
+      "step": 4181
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0596026490066227e-05,
+      "loss": 0.0859,
+      "step": 4182
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.066225165562914e-05,
+      "loss": 0.1033,
+      "step": 4183
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0728476821192052e-05,
+      "loss": 0.0989,
+      "step": 4184
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0794701986754967e-05,
+      "loss": 0.0953,
+      "step": 4185
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0860927152317881e-05,
+      "loss": 0.0848,
+      "step": 4186
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0927152317880796e-05,
+      "loss": 0.077,
+      "step": 4187
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.099337748344371e-05,
+      "loss": 0.0817,
+      "step": 4188
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1059602649006624e-05,
+      "loss": 0.0974,
+      "step": 4189
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1125827814569537e-05,
+      "loss": 0.0947,
+      "step": 4190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1192052980132451e-05,
+      "loss": 0.0935,
+      "step": 4191
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1258278145695364e-05,
+      "loss": 0.1005,
+      "step": 4192
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1324503311258279e-05,
+      "loss": 0.0724,
+      "step": 4193
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1390728476821193e-05,
+      "loss": 0.0981,
+      "step": 4194
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1456953642384107e-05,
+      "loss": 0.0892,
+      "step": 4195
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.152317880794702e-05,
+      "loss": 0.079,
+      "step": 4196
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1589403973509934e-05,
+      "loss": 0.0833,
+      "step": 4197
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1655629139072849e-05,
+      "loss": 0.1069,
+      "step": 4198
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1721854304635763e-05,
+      "loss": 0.0857,
+      "step": 4199
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1788079470198676e-05,
+      "loss": 0.0964,
+      "step": 4200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.185430463576159e-05,
+      "loss": 0.0952,
+      "step": 4201
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1920529801324505e-05,
+      "loss": 0.0912,
+      "step": 4202
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1986754966887417e-05,
+      "loss": 0.0747,
+      "step": 4203
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2052980132450332e-05,
+      "loss": 0.0891,
+      "step": 4204
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2119205298013246e-05,
+      "loss": 0.0859,
+      "step": 4205
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.218543046357616e-05,
+      "loss": 0.0874,
+      "step": 4206
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2251655629139075e-05,
+      "loss": 0.0933,
+      "step": 4207
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2317880794701988e-05,
+      "loss": 0.0793,
+      "step": 4208
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2384105960264902e-05,
+      "loss": 0.1018,
+      "step": 4209
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2450331125827815e-05,
+      "loss": 0.0796,
+      "step": 4210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.251655629139073e-05,
+      "loss": 0.095,
+      "step": 4211
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2582781456953644e-05,
+      "loss": 0.0935,
+      "step": 4212
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.2649006622516558e-05,
+      "loss": 0.0807,
+      "step": 4213
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2715231788079472e-05,
+      "loss": 0.0942,
+      "step": 4214
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2781456953642387e-05,
+      "loss": 0.0944,
+      "step": 4215
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2847682119205298e-05,
+      "loss": 0.0956,
+      "step": 4216
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2913907284768212e-05,
+      "loss": 0.0844,
+      "step": 4217
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2980132450331127e-05,
+      "loss": 0.0875,
+      "step": 4218
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3046357615894041e-05,
+      "loss": 0.0931,
+      "step": 4219
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3112582781456955e-05,
+      "loss": 0.0783,
+      "step": 4220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.317880794701987e-05,
+      "loss": 0.1207,
+      "step": 4221
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3245033112582784e-05,
+      "loss": 0.0837,
+      "step": 4222
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3311258278145695e-05,
+      "loss": 0.1033,
+      "step": 4223
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.337748344370861e-05,
+      "loss": 0.1037,
+      "step": 4224
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3443708609271524e-05,
+      "loss": 0.0849,
+      "step": 4225
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3509933774834438e-05,
+      "loss": 0.0852,
+      "step": 4226
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3576158940397353e-05,
+      "loss": 0.0957,
+      "step": 4227
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3642384105960267e-05,
+      "loss": 0.0924,
+      "step": 4228
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3708609271523178e-05,
+      "loss": 0.0944,
+      "step": 4229
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3774834437086093e-05,
+      "loss": 0.0854,
+      "step": 4230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3841059602649007e-05,
+      "loss": 0.0912,
+      "step": 4231
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3907284768211921e-05,
+      "loss": 0.0969,
+      "step": 4232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3973509933774836e-05,
+      "loss": 0.0826,
+      "step": 4233
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.403973509933775e-05,
+      "loss": 0.0858,
+      "step": 4234
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4105960264900665e-05,
+      "loss": 0.0753,
+      "step": 4235
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4172185430463577e-05,
+      "loss": 0.0901,
+      "step": 4236
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.423841059602649e-05,
+      "loss": 0.1147,
+      "step": 4237
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4304635761589404e-05,
+      "loss": 0.0923,
+      "step": 4238
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4370860927152319e-05,
+      "loss": 0.0645,
+      "step": 4239
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4437086092715233e-05,
+      "loss": 0.0688,
+      "step": 4240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4503311258278147e-05,
+      "loss": 0.092,
+      "step": 4241
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.456953642384106e-05,
+      "loss": 0.084,
+      "step": 4242
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4635761589403975e-05,
+      "loss": 0.0934,
+      "step": 4243
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4701986754966889e-05,
+      "loss": 0.0802,
+      "step": 4244
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4768211920529802e-05,
+      "loss": 0.1102,
+      "step": 4245
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4834437086092716e-05,
+      "loss": 0.0846,
+      "step": 4246
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.490066225165563e-05,
+      "loss": 0.0816,
+      "step": 4247
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4966887417218545e-05,
+      "loss": 0.1082,
+      "step": 4248
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5033112582781458e-05,
+      "loss": 0.0801,
+      "step": 4249
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5099337748344372e-05,
+      "loss": 0.0775,
+      "step": 4250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5165562913907286e-05,
+      "loss": 0.1008,
+      "step": 4251
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.52317880794702e-05,
+      "loss": 0.0777,
+      "step": 4252
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.5298013245033113e-05,
+      "loss": 0.0886,
+      "step": 4253
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5364238410596028e-05,
+      "loss": 0.0832,
+      "step": 4254
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5430463576158942e-05,
+      "loss": 0.1061,
+      "step": 4255
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5496688741721853e-05,
+      "loss": 0.0786,
+      "step": 4256
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5562913907284768e-05,
+      "loss": 0.0817,
+      "step": 4257
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5629139072847682e-05,
+      "loss": 0.1073,
+      "step": 4258
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5695364238410596e-05,
+      "loss": 0.0924,
+      "step": 4259
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.576158940397351e-05,
+      "loss": 0.0823,
+      "step": 4260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.5827814569536425e-05,
+      "loss": 0.0917,
+      "step": 4261
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.589403973509934e-05,
+      "loss": 0.0825,
+      "step": 4262
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.596026490066225e-05,
+      "loss": 0.0908,
+      "step": 4263
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6026490066225165e-05,
+      "loss": 0.1161,
+      "step": 4264
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.609271523178808e-05,
+      "loss": 0.0801,
+      "step": 4265
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6158940397350994e-05,
+      "loss": 0.1064,
+      "step": 4266
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6225165562913908e-05,
+      "loss": 0.1052,
+      "step": 4267
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6291390728476823e-05,
+      "loss": 0.1064,
+      "step": 4268
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6357615894039737e-05,
+      "loss": 0.093,
+      "step": 4269
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.642384105960265e-05,
+      "loss": 0.0991,
+      "step": 4270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6490066225165562e-05,
+      "loss": 0.0997,
+      "step": 4271
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.6556291390728477e-05,
+      "loss": 0.1011,
+      "step": 4272
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.662251655629139e-05,
+      "loss": 0.0968,
+      "step": 4273
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6688741721854306e-05,
+      "loss": 0.0927,
+      "step": 4274
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.675496688741722e-05,
+      "loss": 0.0873,
+      "step": 4275
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6821192052980134e-05,
+      "loss": 0.0919,
+      "step": 4276
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.688741721854305e-05,
+      "loss": 0.0648,
+      "step": 4277
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.6953642384105963e-05,
+      "loss": 0.1108,
+      "step": 4278
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7019867549668878e-05,
+      "loss": 0.089,
+      "step": 4279
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.708609271523179e-05,
+      "loss": 0.0861,
+      "step": 4280
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7152317880794703e-05,
+      "loss": 0.1014,
+      "step": 4281
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7218543046357617e-05,
+      "loss": 0.0996,
+      "step": 4282
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7284768211920532e-05,
+      "loss": 0.0754,
+      "step": 4283
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7350993377483446e-05,
+      "loss": 0.0927,
+      "step": 4284
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.741721854304636e-05,
+      "loss": 0.0842,
+      "step": 4285
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7483443708609275e-05,
+      "loss": 0.076,
+      "step": 4286
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.754966887417219e-05,
+      "loss": 0.0908,
+      "step": 4287
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.76158940397351e-05,
+      "loss": 0.0954,
+      "step": 4288
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7682119205298015e-05,
+      "loss": 0.1041,
+      "step": 4289
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.774834437086093e-05,
+      "loss": 0.09,
+      "step": 4290
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7814569536423844e-05,
+      "loss": 0.0985,
+      "step": 4291
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7880794701986758e-05,
+      "loss": 0.0782,
+      "step": 4292
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.7947019867549672e-05,
+      "loss": 0.0768,
+      "step": 4293
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8013245033112587e-05,
+      "loss": 0.1104,
+      "step": 4294
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8079470198675498e-05,
+      "loss": 0.0975,
+      "step": 4295
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8145695364238412e-05,
+      "loss": 0.0862,
+      "step": 4296
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8211920529801327e-05,
+      "loss": 0.0978,
+      "step": 4297
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.827814569536424e-05,
+      "loss": 0.1001,
+      "step": 4298
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8344370860927155e-05,
+      "loss": 0.0806,
+      "step": 4299
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.841059602649007e-05,
+      "loss": 0.0928,
+      "step": 4300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8476821192052984e-05,
+      "loss": 0.0935,
+      "step": 4301
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8543046357615895e-05,
+      "loss": 0.0688,
+      "step": 4302
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.860927152317881e-05,
+      "loss": 0.1075,
+      "step": 4303
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8675496688741724e-05,
+      "loss": 0.094,
+      "step": 4304
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8741721854304638e-05,
+      "loss": 0.0812,
+      "step": 4305
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8807947019867553e-05,
+      "loss": 0.0812,
+      "step": 4306
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8874172185430467e-05,
+      "loss": 0.0731,
+      "step": 4307
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8940397350993378e-05,
+      "loss": 0.0957,
+      "step": 4308
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9006622516556292e-05,
+      "loss": 0.0815,
+      "step": 4309
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9072847682119207e-05,
+      "loss": 0.0964,
+      "step": 4310
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.913907284768212e-05,
+      "loss": 0.0893,
+      "step": 4311
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9205298013245036e-05,
+      "loss": 0.0979,
+      "step": 4312
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.927152317880795e-05,
+      "loss": 0.0883,
+      "step": 4313
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9337748344370864e-05,
+      "loss": 0.0778,
+      "step": 4314
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9403973509933775e-05,
+      "loss": 0.0933,
+      "step": 4315
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.947019867549669e-05,
+      "loss": 0.0995,
+      "step": 4316
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9536423841059604e-05,
+      "loss": 0.0884,
+      "step": 4317
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.960264900662252e-05,
+      "loss": 0.0901,
+      "step": 4318
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9668874172185433e-05,
+      "loss": 0.0836,
+      "step": 4319
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9735099337748347e-05,
+      "loss": 0.0731,
+      "step": 4320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.980132450331126e-05,
+      "loss": 0.0938,
+      "step": 4321
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9867549668874173e-05,
+      "loss": 0.0892,
+      "step": 4322
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9933774834437087e-05,
+      "loss": 0.0751,
+      "step": 4323
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2e-05,
+      "loss": 0.0929,
+      "step": 4324
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.999999948120797e-05,
+      "loss": 0.0813,
+      "step": 4325
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999997924831927e-05,
+      "loss": 0.0837,
+      "step": 4326
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999995330872033e-05,
+      "loss": 0.0819,
+      "step": 4327
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999991699328562e-05,
+      "loss": 0.0936,
+      "step": 4328
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999987030201884e-05,
+      "loss": 0.0888,
+      "step": 4329
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999981323492487e-05,
+      "loss": 0.0865,
+      "step": 4330
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999974579200967e-05,
+      "loss": 0.0798,
+      "step": 4331
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9999966797328016e-05,
+      "loss": 0.0797,
+      "step": 4332
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.999995797787445e-05,
+      "loss": 0.0894,
+      "step": 4333
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999948120841176e-05,
+      "loss": 0.1044,
+      "step": 4334
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999993722622922e-05,
+      "loss": 0.0914,
+      "step": 4335
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999992529403971e-05,
+      "loss": 0.1151,
+      "step": 4336
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999912324273893e-05,
+      "loss": 0.0714,
+      "step": 4337
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999898316933108e-05,
+      "loss": 0.0822,
+      "step": 4338
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999883272018805e-05,
+      "loss": 0.0992,
+      "step": 4339
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999867189532547e-05,
+      "loss": 0.0786,
+      "step": 4340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999985006947601e-05,
+      "loss": 0.088,
+      "step": 4341
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999983191185096e-05,
+      "loss": 0.0787,
+      "step": 4342
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999981271665929e-05,
+      "loss": 0.0913,
+      "step": 4343
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999792483902983e-05,
+      "loss": 0.1022,
+      "step": 4344
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999771213584147e-05,
+      "loss": 0.0887,
+      "step": 4345
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999748905704984e-05,
+      "loss": 0.0938,
+      "step": 4346
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999725560267808e-05,
+      "loss": 0.0889,
+      "step": 4347
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999701177275045e-05,
+      "loss": 0.1046,
+      "step": 4348
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999967575672922e-05,
+      "loss": 0.0814,
+      "step": 4349
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999649298632977e-05,
+      "loss": 0.0842,
+      "step": 4350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.999962180298905e-05,
+      "loss": 0.0981,
+      "step": 4351
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999593269800307e-05,
+      "loss": 0.0959,
+      "step": 4352
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9999563699069698e-05,
+      "loss": 0.0873,
+      "step": 4353
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999533090800293e-05,
+      "loss": 0.0891,
+      "step": 4354
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.999950144499527e-05,
+      "loss": 0.0944,
+      "step": 4355
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.999946876165791e-05,
+      "loss": 0.1089,
+      "step": 4356
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999435040791612e-05,
+      "loss": 0.0948,
+      "step": 4357
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999400282399863e-05,
+      "loss": 0.1106,
+      "step": 4358
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999364486486277e-05,
+      "loss": 0.0711,
+      "step": 4359
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999327653054563e-05,
+      "loss": 0.1063,
+      "step": 4360
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.999928978210855e-05,
+      "loss": 0.0839,
+      "step": 4361
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999250873652164e-05,
+      "loss": 0.1042,
+      "step": 4362
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999210927689438e-05,
+      "loss": 0.0951,
+      "step": 4363
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999169944224518e-05,
+      "loss": 0.0867,
+      "step": 4364
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999127923261664e-05,
+      "loss": 0.0951,
+      "step": 4365
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.999908486480523e-05,
+      "loss": 0.084,
+      "step": 4366
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9999040768859682e-05,
+      "loss": 0.0858,
+      "step": 4367
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9998995635429598e-05,
+      "loss": 0.079,
+      "step": 4368
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.999894946451966e-05,
+      "loss": 0.0988,
+      "step": 4369
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.999890225613466e-05,
+      "loss": 0.0753,
+      "step": 4370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9998854010279497e-05,
+      "loss": 0.0924,
+      "step": 4371
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9998804726959173e-05,
+      "loss": 0.1007,
+      "step": 4372
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9998754406178803e-05,
+      "loss": 0.0815,
+      "step": 4373
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998703047943614e-05,
+      "loss": 0.1002,
+      "step": 4374
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998650652258926e-05,
+      "loss": 0.0947,
+      "step": 4375
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998597219130182e-05,
+      "loss": 0.0822,
+      "step": 4376
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.999854274856292e-05,
+      "loss": 0.0947,
+      "step": 4377
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998487240562798e-05,
+      "loss": 0.0811,
+      "step": 4378
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998430695135573e-05,
+      "loss": 0.0881,
+      "step": 4379
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.999837311228711e-05,
+      "loss": 0.0902,
+      "step": 4380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998314492023387e-05,
+      "loss": 0.078,
+      "step": 4381
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998254834350484e-05,
+      "loss": 0.0856,
+      "step": 4382
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998194139274593e-05,
+      "loss": 0.0703,
+      "step": 4383
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9998132406802008e-05,
+      "loss": 0.0875,
+      "step": 4384
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.999806963693914e-05,
+      "loss": 0.0867,
+      "step": 4385
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.99980058296925e-05,
+      "loss": 0.0909,
+      "step": 4386
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997940985068702e-05,
+      "loss": 0.0854,
+      "step": 4387
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997875103074483e-05,
+      "loss": 0.0829,
+      "step": 4388
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997808183716674e-05,
+      "loss": 0.0827,
+      "step": 4389
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997740227002217e-05,
+      "loss": 0.0865,
+      "step": 4390
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.999767123293817e-05,
+      "loss": 0.0964,
+      "step": 4391
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997601201531685e-05,
+      "loss": 0.0857,
+      "step": 4392
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997530132790034e-05,
+      "loss": 0.0896,
+      "step": 4393
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9997458026720587e-05,
+      "loss": 0.0885,
+      "step": 4394
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9997384883330825e-05,
+      "loss": 0.0941,
+      "step": 4395
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9997310702628338e-05,
+      "loss": 0.1021,
+      "step": 4396
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9997235484620825e-05,
+      "loss": 0.1033,
+      "step": 4397
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9997159229316088e-05,
+      "loss": 0.0866,
+      "step": 4398
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9997081936722037e-05,
+      "loss": 0.0916,
+      "step": 4399
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9997003606846702e-05,
+      "loss": 0.0924,
+      "step": 4400
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.99969242396982e-05,
+      "loss": 0.111,
+      "step": 4401
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9996843835284765e-05,
+      "loss": 0.0942,
+      "step": 4402
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9996762393614748e-05,
+      "loss": 0.1049,
+      "step": 4403
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9996679914696596e-05,
+      "loss": 0.1093,
+      "step": 4404
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9996596398538865e-05,
+      "loss": 0.1007,
+      "step": 4405
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.999651184515022e-05,
+      "loss": 0.0892,
+      "step": 4406
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.999642625453944e-05,
+      "loss": 0.0745,
+      "step": 4407
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.99963396267154e-05,
+      "loss": 0.0833,
+      "step": 4408
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9996251961687086e-05,
+      "loss": 0.0801,
+      "step": 4409
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.99961632594636e-05,
+      "loss": 0.116,
+      "step": 4410
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9996073520054143e-05,
+      "loss": 0.0796,
+      "step": 4411
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9995982743468025e-05,
+      "loss": 0.1,
+      "step": 4412
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.999589092971467e-05,
+      "loss": 0.1388,
+      "step": 4413
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.99957980788036e-05,
+      "loss": 0.0793,
+      "step": 4414
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999570419074445e-05,
+      "loss": 0.1011,
+      "step": 4415
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999560926554696e-05,
+      "loss": 0.08,
+      "step": 4416
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999551330322098e-05,
+      "loss": 0.1003,
+      "step": 4417
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999541630377647e-05,
+      "loss": 0.101,
+      "step": 4418
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999531826722349e-05,
+      "loss": 0.0951,
+      "step": 4419
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9995219193572216e-05,
+      "loss": 0.1024,
+      "step": 4420
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9995119082832927e-05,
+      "loss": 0.0955,
+      "step": 4421
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999501793501601e-05,
+      "loss": 0.1003,
+      "step": 4422
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999491575013196e-05,
+      "loss": 0.1139,
+      "step": 4423
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9994812528191375e-05,
+      "loss": 0.0946,
+      "step": 4424
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9994708269204972e-05,
+      "loss": 0.0952,
+      "step": 4425
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999460297318357e-05,
+      "loss": 0.0958,
+      "step": 4426
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9994496640138084e-05,
+      "loss": 0.08,
+      "step": 4427
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999438927007955e-05,
+      "loss": 0.0867,
+      "step": 4428
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999428086301912e-05,
+      "loss": 0.0985,
+      "step": 4429
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999417141896803e-05,
+      "loss": 0.082,
+      "step": 4430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9994060937937637e-05,
+      "loss": 0.093,
+      "step": 4431
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9993949419939412e-05,
+      "loss": 0.0944,
+      "step": 4432
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.999383686498492e-05,
+      "loss": 0.0969,
+      "step": 4433
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9993723273085835e-05,
+      "loss": 0.0845,
+      "step": 4434
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9993608644253954e-05,
+      "loss": 0.0786,
+      "step": 4435
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9993492978501164e-05,
+      "loss": 0.1122,
+      "step": 4436
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9993376275839466e-05,
+      "loss": 0.087,
+      "step": 4437
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.999325853628097e-05,
+      "loss": 0.0968,
+      "step": 4438
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9993139759837895e-05,
+      "loss": 0.1022,
+      "step": 4439
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9993019946522563e-05,
+      "loss": 0.0874,
+      "step": 4440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992899096347403e-05,
+      "loss": 0.0989,
+      "step": 4441
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.999277720932496e-05,
+      "loss": 0.0961,
+      "step": 4442
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992654285467874e-05,
+      "loss": 0.083,
+      "step": 4443
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992530324788903e-05,
+      "loss": 0.0967,
+      "step": 4444
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992405327300912e-05,
+      "loss": 0.0858,
+      "step": 4445
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992279293016866e-05,
+      "loss": 0.0884,
+      "step": 4446
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992152221949842e-05,
+      "loss": 0.0868,
+      "step": 4447
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9992024114113027e-05,
+      "loss": 0.0925,
+      "step": 4448
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9991894969519716e-05,
+      "loss": 0.0759,
+      "step": 4449
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9991764788183303e-05,
+      "loss": 0.0812,
+      "step": 4450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9991633570117298e-05,
+      "loss": 0.0973,
+      "step": 4451
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9991501315335316e-05,
+      "loss": 0.0817,
+      "step": 4452
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9991368023851078e-05,
+      "loss": 0.1031,
+      "step": 4453
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9991233695678415e-05,
+      "loss": 0.0907,
+      "step": 4454
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9991098330831266e-05,
+      "loss": 0.1023,
+      "step": 4455
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9990961929323674e-05,
+      "loss": 0.0749,
+      "step": 4456
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9990824491169792e-05,
+      "loss": 0.1062,
+      "step": 4457
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9990686016383884e-05,
+      "loss": 0.1044,
+      "step": 4458
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9990546504980318e-05,
+      "loss": 0.0969,
+      "step": 4459
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9990405956973563e-05,
+      "loss": 0.0756,
+      "step": 4460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9990264372378207e-05,
+      "loss": 0.084,
+      "step": 4461
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.999012175120894e-05,
+      "loss": 0.078,
+      "step": 4462
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9989978093480558e-05,
+      "loss": 0.0797,
+      "step": 4463
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.998983339920797e-05,
+      "loss": 0.0925,
+      "step": 4464
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9989687668406184e-05,
+      "loss": 0.1012,
+      "step": 4465
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9989540901090327e-05,
+      "loss": 0.0979,
+      "step": 4466
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9989393097275628e-05,
+      "loss": 0.1024,
+      "step": 4467
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9989244256977415e-05,
+      "loss": 0.0942,
+      "step": 4468
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9989094380211137e-05,
+      "loss": 0.0804,
+      "step": 4469
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9988943466992346e-05,
+      "loss": 0.0845,
+      "step": 4470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.99887915173367e-05,
+      "loss": 0.092,
+      "step": 4471
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.998863853125996e-05,
+      "loss": 0.0962,
+      "step": 4472
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9988484508778003e-05,
+      "loss": 0.0889,
+      "step": 4473
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.998832944990681e-05,
+      "loss": 0.0958,
+      "step": 4474
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9988173354662472e-05,
+      "loss": 0.0889,
+      "step": 4475
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9988016223061183e-05,
+      "loss": 0.0851,
+      "step": 4476
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9987858055119243e-05,
+      "loss": 0.0884,
+      "step": 4477
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9987698850853072e-05,
+      "loss": 0.1005,
+      "step": 4478
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9987538610279183e-05,
+      "loss": 0.0989,
+      "step": 4479
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9987377333414203e-05,
+      "loss": 0.1046,
+      "step": 4480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9987215020274867e-05,
+      "loss": 0.0876,
+      "step": 4481
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9987051670878012e-05,
+      "loss": 0.0751,
+      "step": 4482
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9986887285240592e-05,
+      "loss": 0.0872,
+      "step": 4483
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.998672186337966e-05,
+      "loss": 0.0947,
+      "step": 4484
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9986555405312383e-05,
+      "loss": 0.0958,
+      "step": 4485
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9986387911056034e-05,
+      "loss": 0.0846,
+      "step": 4486
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9986219380627987e-05,
+      "loss": 0.0936,
+      "step": 4487
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9986049814045732e-05,
+      "loss": 0.0885,
+      "step": 4488
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9985879211326857e-05,
+      "loss": 0.0957,
+      "step": 4489
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.998570757248907e-05,
+      "loss": 0.0835,
+      "step": 4490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.998553489755018e-05,
+      "loss": 0.0862,
+      "step": 4491
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9985361186528097e-05,
+      "loss": 0.0891,
+      "step": 4492
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.998518643944085e-05,
+      "loss": 0.1028,
+      "step": 4493
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9985010656306572e-05,
+      "loss": 0.0724,
+      "step": 4494
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.99848338371435e-05,
+      "loss": 0.0967,
+      "step": 4495
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9984655981969977e-05,
+      "loss": 0.0844,
+      "step": 4496
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9984477090804465e-05,
+      "loss": 0.0878,
+      "step": 4497
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9984297163665518e-05,
+      "loss": 0.0883,
+      "step": 4498
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.998411620057181e-05,
+      "loss": 0.1041,
+      "step": 4499
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9983934201542108e-05,
+      "loss": 0.0972,
+      "step": 4500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.998375116659531e-05,
+      "loss": 0.105,
+      "step": 4501
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9983567095750396e-05,
+      "loss": 0.0939,
+      "step": 4502
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.998338198902647e-05,
+      "loss": 0.0815,
+      "step": 4503
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.998319584644274e-05,
+      "loss": 0.0911,
+      "step": 4504
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9983008668018514e-05,
+      "loss": 0.0918,
+      "step": 4505
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.998282045377322e-05,
+      "loss": 0.0947,
+      "step": 4506
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9982631203726385e-05,
+      "loss": 0.0716,
+      "step": 4507
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.998244091789764e-05,
+      "loss": 0.0856,
+      "step": 4508
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9982249596306733e-05,
+      "loss": 0.087,
+      "step": 4509
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9982057238973516e-05,
+      "loss": 0.0851,
+      "step": 4510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9981863845917945e-05,
+      "loss": 0.0732,
+      "step": 4511
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9981669417160092e-05,
+      "loss": 0.0945,
+      "step": 4512
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9981473952720122e-05,
+      "loss": 0.0873,
+      "step": 4513
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9981277452618322e-05,
+      "loss": 0.1058,
+      "step": 4514
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.998107991687508e-05,
+      "loss": 0.0909,
+      "step": 4515
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.998088134551089e-05,
+      "loss": 0.0981,
+      "step": 4516
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9980681738546358e-05,
+      "loss": 0.0812,
+      "step": 4517
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.998048109600219e-05,
+      "loss": 0.0969,
+      "step": 4518
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.998027941789921e-05,
+      "loss": 0.0905,
+      "step": 4519
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.998007670425834e-05,
+      "loss": 0.0967,
+      "step": 4520
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9979872955100618e-05,
+      "loss": 0.0892,
+      "step": 4521
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9979668170447176e-05,
+      "loss": 0.1111,
+      "step": 4522
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.997946235031927e-05,
+      "loss": 0.0789,
+      "step": 4523
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9979255494738257e-05,
+      "loss": 0.0834,
+      "step": 4524
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.997904760372559e-05,
+      "loss": 0.0677,
+      "step": 4525
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.997883867730285e-05,
+      "loss": 0.0826,
+      "step": 4526
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9978628715491707e-05,
+      "loss": 0.0913,
+      "step": 4527
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9978417718313953e-05,
+      "loss": 0.0887,
+      "step": 4528
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9978205685791472e-05,
+      "loss": 0.0991,
+      "step": 4529
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9977992617946275e-05,
+      "loss": 0.0901,
+      "step": 4530
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9977778514800462e-05,
+      "loss": 0.098,
+      "step": 4531
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9977563376376253e-05,
+      "loss": 0.075,
+      "step": 4532
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9977347202695965e-05,
+      "loss": 0.08,
+      "step": 4533
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.997712999378203e-05,
+      "loss": 0.0835,
+      "step": 4534
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9976911749656988e-05,
+      "loss": 0.0803,
+      "step": 4535
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.997669247034348e-05,
+      "loss": 0.1136,
+      "step": 4536
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9976472155864258e-05,
+      "loss": 0.0853,
+      "step": 4537
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9976250806242185e-05,
+      "loss": 0.0868,
+      "step": 4538
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9976028421500223e-05,
+      "loss": 0.0714,
+      "step": 4539
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.997580500166145e-05,
+      "loss": 0.0997,
+      "step": 4540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9975580546749052e-05,
+      "loss": 0.0826,
+      "step": 4541
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9975355056786307e-05,
+      "loss": 0.0975,
+      "step": 4542
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.997512853179662e-05,
+      "loss": 0.1108,
+      "step": 4543
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9974900971803493e-05,
+      "loss": 0.113,
+      "step": 4544
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9974672376830532e-05,
+      "loss": 0.0861,
+      "step": 4545
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9974442746901464e-05,
+      "loss": 0.0958,
+      "step": 4546
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.997421208204011e-05,
+      "loss": 0.1046,
+      "step": 4547
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9973980382270405e-05,
+      "loss": 0.0889,
+      "step": 4548
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9973747647616387e-05,
+      "loss": 0.1007,
+      "step": 4549
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.997351387810221e-05,
+      "loss": 0.0941,
+      "step": 4550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9973279073752124e-05,
+      "loss": 0.0898,
+      "step": 4551
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9973043234590495e-05,
+      "loss": 0.0813,
+      "step": 4552
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9972806360641793e-05,
+      "loss": 0.0845,
+      "step": 4553
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9972568451930592e-05,
+      "loss": 0.0924,
+      "step": 4554
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.997232950848158e-05,
+      "loss": 0.0997,
+      "step": 4555
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9972089530319554e-05,
+      "loss": 0.0984,
+      "step": 4556
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9971848517469404e-05,
+      "loss": 0.0735,
+      "step": 4557
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9971606469956146e-05,
+      "loss": 0.0947,
+      "step": 4558
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.997136338780489e-05,
+      "loss": 0.0983,
+      "step": 4559
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9971119271040854e-05,
+      "loss": 0.0806,
+      "step": 4560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.997087411968938e-05,
+      "loss": 0.0762,
+      "step": 4561
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.997062793377589e-05,
+      "loss": 0.1034,
+      "step": 4562
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9970380713325937e-05,
+      "loss": 0.104,
+      "step": 4563
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9970132458365165e-05,
+      "loss": 0.099,
+      "step": 4564
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9969883168919342e-05,
+      "loss": 0.0909,
+      "step": 4565
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.996963284501433e-05,
+      "loss": 0.0834,
+      "step": 4566
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9969381486676092e-05,
+      "loss": 0.0968,
+      "step": 4567
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9969129093930727e-05,
+      "loss": 0.0728,
+      "step": 4568
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9968875666804407e-05,
+      "loss": 0.0968,
+      "step": 4569
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9968621205323434e-05,
+      "loss": 0.1082,
+      "step": 4570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9968365709514215e-05,
+      "loss": 0.0836,
+      "step": 4571
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9968109179403253e-05,
+      "loss": 0.0791,
+      "step": 4572
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9967851615017164e-05,
+      "loss": 0.1125,
+      "step": 4573
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9967593016382677e-05,
+      "loss": 0.0942,
+      "step": 4574
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9967333383526623e-05,
+      "loss": 0.0938,
+      "step": 4575
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9967072716475938e-05,
+      "loss": 0.0876,
+      "step": 4576
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9966811015257672e-05,
+      "loss": 0.097,
+      "step": 4577
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9966548279898982e-05,
+      "loss": 0.0862,
+      "step": 4578
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9966284510427118e-05,
+      "loss": 0.1044,
+      "step": 4579
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9966019706869456e-05,
+      "loss": 0.0919,
+      "step": 4580
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9965753869253474e-05,
+      "loss": 0.1058,
+      "step": 4581
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9965486997606747e-05,
+      "loss": 0.0903,
+      "step": 4582
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.996521909195697e-05,
+      "loss": 0.0894,
+      "step": 4583
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9964950152331943e-05,
+      "loss": 0.0968,
+      "step": 4584
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9964680178759565e-05,
+      "loss": 0.0928,
+      "step": 4585
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9964409171267852e-05,
+      "loss": 0.105,
+      "step": 4586
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9964137129884922e-05,
+      "loss": 0.092,
+      "step": 4587
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9963864054639e-05,
+      "loss": 0.097,
+      "step": 4588
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9963589945558423e-05,
+      "loss": 0.0891,
+      "step": 4589
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.996331480267163e-05,
+      "loss": 0.0923,
+      "step": 4590
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.996303862600717e-05,
+      "loss": 0.0876,
+      "step": 4591
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9962761415593698e-05,
+      "loss": 0.0878,
+      "step": 4592
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9962483171459978e-05,
+      "loss": 0.0941,
+      "step": 4593
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.996220389363488e-05,
+      "loss": 0.0879,
+      "step": 4594
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.996192358214738e-05,
+      "loss": 0.0889,
+      "step": 4595
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9961642237026565e-05,
+      "loss": 0.0854,
+      "step": 4596
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9961359858301622e-05,
+      "loss": 0.0908,
+      "step": 4597
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.996107644600186e-05,
+      "loss": 0.0927,
+      "step": 4598
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9960792000156676e-05,
+      "loss": 0.0754,
+      "step": 4599
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9960506520795585e-05,
+      "loss": 0.1006,
+      "step": 4600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9960220007948212e-05,
+      "loss": 0.0948,
+      "step": 4601
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9959932461644282e-05,
+      "loss": 0.0945,
+      "step": 4602
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.995964388191363e-05,
+      "loss": 0.0962,
+      "step": 4603
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9959354268786205e-05,
+      "loss": 0.0737,
+      "step": 4604
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.995906362229205e-05,
+      "loss": 0.1068,
+      "step": 4605
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.995877194246132e-05,
+      "loss": 0.1005,
+      "step": 4606
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9958479229324282e-05,
+      "loss": 0.0884,
+      "step": 4607
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9958185482911312e-05,
+      "loss": 0.0956,
+      "step": 4608
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9957890703252882e-05,
+      "loss": 0.1041,
+      "step": 4609
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9957594890379584e-05,
+      "loss": 0.0756,
+      "step": 4610
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9957298044322108e-05,
+      "loss": 0.0779,
+      "step": 4611
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.995700016511125e-05,
+      "loss": 0.0846,
+      "step": 4612
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9956701252777922e-05,
+      "loss": 0.0834,
+      "step": 4613
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9956401307353142e-05,
+      "loss": 0.0803,
+      "step": 4614
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.995610032886803e-05,
+      "loss": 0.0712,
+      "step": 4615
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995579831735381e-05,
+      "loss": 0.0822,
+      "step": 4616
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995549527284182e-05,
+      "loss": 0.0988,
+      "step": 4617
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9955191195363505e-05,
+      "loss": 0.0851,
+      "step": 4618
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9954886084950418e-05,
+      "loss": 0.0981,
+      "step": 4619
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995457994163421e-05,
+      "loss": 0.0975,
+      "step": 4620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9954272765446656e-05,
+      "loss": 0.0989,
+      "step": 4621
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995396455641962e-05,
+      "loss": 0.0941,
+      "step": 4622
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995365531458508e-05,
+      "loss": 0.0924,
+      "step": 4623
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995334503997513e-05,
+      "loss": 0.081,
+      "step": 4624
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9953033732621958e-05,
+      "loss": 0.0896,
+      "step": 4625
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9952721392557867e-05,
+      "loss": 0.0851,
+      "step": 4626
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9952408019815266e-05,
+      "loss": 0.098,
+      "step": 4627
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9952093614426667e-05,
+      "loss": 0.0959,
+      "step": 4628
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995177817642469e-05,
+      "loss": 0.0972,
+      "step": 4629
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9951461705842073e-05,
+      "loss": 0.0879,
+      "step": 4630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9951144202711644e-05,
+      "loss": 0.0914,
+      "step": 4631
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.995082566706635e-05,
+      "loss": 0.0793,
+      "step": 4632
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9950506098939243e-05,
+      "loss": 0.0896,
+      "step": 4633
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9950185498363483e-05,
+      "loss": 0.0956,
+      "step": 4634
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9949863865372324e-05,
+      "loss": 0.0968,
+      "step": 4635
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.994954119999915e-05,
+      "loss": 0.0898,
+      "step": 4636
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9949217502277438e-05,
+      "loss": 0.0654,
+      "step": 4637
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9948892772240768e-05,
+      "loss": 0.0948,
+      "step": 4638
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9948567009922842e-05,
+      "loss": 0.0957,
+      "step": 4639
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9948240215357452e-05,
+      "loss": 0.1013,
+      "step": 4640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9947912388578515e-05,
+      "loss": 0.0936,
+      "step": 4641
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9947583529620038e-05,
+      "loss": 0.0782,
+      "step": 4642
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9947253638516146e-05,
+      "loss": 0.0844,
+      "step": 4643
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9946922715301066e-05,
+      "loss": 0.078,
+      "step": 4644
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9946590760009137e-05,
+      "loss": 0.0963,
+      "step": 4645
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9946257772674803e-05,
+      "loss": 0.1057,
+      "step": 4646
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.994592375333261e-05,
+      "loss": 0.079,
+      "step": 4647
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9945588702017215e-05,
+      "loss": 0.0841,
+      "step": 4648
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9945252618763388e-05,
+      "loss": 0.1053,
+      "step": 4649
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9944915503606e-05,
+      "loss": 0.0787,
+      "step": 4650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9944577356580023e-05,
+      "loss": 0.099,
+      "step": 4651
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.994423817772055e-05,
+      "loss": 0.0841,
+      "step": 4652
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9943897967062767e-05,
+      "loss": 0.1,
+      "step": 4653
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9943556724641975e-05,
+      "loss": 0.0848,
+      "step": 4654
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9943214450493586e-05,
+      "loss": 0.0919,
+      "step": 4655
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9942871144653108e-05,
+      "loss": 0.083,
+      "step": 4656
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9942526807156166e-05,
+      "loss": 0.0844,
+      "step": 4657
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9942181438038486e-05,
+      "loss": 0.083,
+      "step": 4658
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9941835037335903e-05,
+      "loss": 0.0706,
+      "step": 4659
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.994148760508436e-05,
+      "loss": 0.1122,
+      "step": 4660
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9941139141319903e-05,
+      "loss": 0.0804,
+      "step": 4661
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9940789646078694e-05,
+      "loss": 0.1028,
+      "step": 4662
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9940439119396985e-05,
+      "loss": 0.1002,
+      "step": 4663
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.994008756131116e-05,
+      "loss": 0.0873,
+      "step": 4664
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9939734971857687e-05,
+      "loss": 0.0981,
+      "step": 4665
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9939381351073153e-05,
+      "loss": 0.0907,
+      "step": 4666
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.993902669899425e-05,
+      "loss": 0.0934,
+      "step": 4667
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9938671015657773e-05,
+      "loss": 0.0948,
+      "step": 4668
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.993831430110063e-05,
+      "loss": 0.0867,
+      "step": 4669
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9937956555359833e-05,
+      "loss": 0.0817,
+      "step": 4670
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9937597778472497e-05,
+      "loss": 0.0916,
+      "step": 4671
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9937237970475857e-05,
+      "loss": 0.0953,
+      "step": 4672
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9936877131407234e-05,
+      "loss": 0.0912,
+      "step": 4673
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9936515261304076e-05,
+      "loss": 0.0868,
+      "step": 4674
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.993615236020393e-05,
+      "loss": 0.1024,
+      "step": 4675
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.993578842814445e-05,
+      "loss": 0.1014,
+      "step": 4676
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9935423465163395e-05,
+      "loss": 0.0845,
+      "step": 4677
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9935057471298633e-05,
+      "loss": 0.094,
+      "step": 4678
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.993469044658814e-05,
+      "loss": 0.0996,
+      "step": 4679
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9934322391069996e-05,
+      "loss": 0.089,
+      "step": 4680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.993395330478239e-05,
+      "loss": 0.0858,
+      "step": 4681
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9933583187763625e-05,
+      "loss": 0.0876,
+      "step": 4682
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9933212040052092e-05,
+      "loss": 0.0884,
+      "step": 4683
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.993283986168631e-05,
+      "loss": 0.0891,
+      "step": 4684
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9932466652704893e-05,
+      "loss": 0.0967,
+      "step": 4685
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.993209241314656e-05,
+      "loss": 0.0938,
+      "step": 4686
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9931717143050147e-05,
+      "loss": 0.0853,
+      "step": 4687
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9931340842454588e-05,
+      "loss": 0.0923,
+      "step": 4688
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9930963511398932e-05,
+      "loss": 0.0967,
+      "step": 4689
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9930585149922325e-05,
+      "loss": 0.1002,
+      "step": 4690
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9930205758064033e-05,
+      "loss": 0.0995,
+      "step": 4691
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.992982533586341e-05,
+      "loss": 0.1027,
+      "step": 4692
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9929443883359934e-05,
+      "loss": 0.0974,
+      "step": 4693
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9929061400593186e-05,
+      "loss": 0.1001,
+      "step": 4694
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.992867788760285e-05,
+      "loss": 0.1031,
+      "step": 4695
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9928293344428714e-05,
+      "loss": 0.0651,
+      "step": 4696
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9927907771110682e-05,
+      "loss": 0.0908,
+      "step": 4697
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9927521167688762e-05,
+      "loss": 0.0875,
+      "step": 4698
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9927133534203064e-05,
+      "loss": 0.1021,
+      "step": 4699
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992674487069381e-05,
+      "loss": 0.0907,
+      "step": 4700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992635517720133e-05,
+      "loss": 0.0789,
+      "step": 4701
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992596445376605e-05,
+      "loss": 0.0777,
+      "step": 4702
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992557270042852e-05,
+      "loss": 0.075,
+      "step": 4703
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992517991722938e-05,
+      "loss": 0.0798,
+      "step": 4704
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992478610420939e-05,
+      "loss": 0.104,
+      "step": 4705
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9924391261409405e-05,
+      "loss": 0.1193,
+      "step": 4706
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9923995388870404e-05,
+      "loss": 0.08,
+      "step": 4707
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992359848663345e-05,
+      "loss": 0.0751,
+      "step": 4708
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9923200554739735e-05,
+      "loss": 0.0923,
+      "step": 4709
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.992280159323054e-05,
+      "loss": 0.0874,
+      "step": 4710
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9922401602147266e-05,
+      "loss": 0.0898,
+      "step": 4711
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9922000581531412e-05,
+      "loss": 0.0786,
+      "step": 4712
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9921598531424592e-05,
+      "loss": 0.1011,
+      "step": 4713
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9921195451868514e-05,
+      "loss": 0.0969,
+      "step": 4714
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9920791342905005e-05,
+      "loss": 0.1017,
+      "step": 4715
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9920386204576002e-05,
+      "loss": 0.083,
+      "step": 4716
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.991998003692353e-05,
+      "loss": 0.1063,
+      "step": 4717
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9919572839989734e-05,
+      "loss": 0.1004,
+      "step": 4718
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9919164613816874e-05,
+      "loss": 0.0833,
+      "step": 4719
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9918755358447298e-05,
+      "loss": 0.0684,
+      "step": 4720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.991834507392347e-05,
+      "loss": 0.0989,
+      "step": 4721
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9917933760287966e-05,
+      "loss": 0.0802,
+      "step": 4722
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9917521417583456e-05,
+      "loss": 0.1047,
+      "step": 4723
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9917108045852727e-05,
+      "loss": 0.1085,
+      "step": 4724
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9916693645138673e-05,
+      "loss": 0.0656,
+      "step": 4725
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9916278215484288e-05,
+      "loss": 0.085,
+      "step": 4726
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9915861756932677e-05,
+      "loss": 0.0873,
+      "step": 4727
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9915444269527052e-05,
+      "loss": 0.0672,
+      "step": 4728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9915025753310727e-05,
+      "loss": 0.0872,
+      "step": 4729
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9914606208327134e-05,
+      "loss": 0.0763,
+      "step": 4730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9914185634619797e-05,
+      "loss": 0.0847,
+      "step": 4731
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9913764032232362e-05,
+      "loss": 0.0753,
+      "step": 4732
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9913341401208563e-05,
+      "loss": 0.0878,
+      "step": 4733
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.991291774159226e-05,
+      "loss": 0.0955,
+      "step": 4734
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.991249305342741e-05,
+      "loss": 0.0801,
+      "step": 4735
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9912067336758075e-05,
+      "loss": 0.0792,
+      "step": 4736
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9911640591628428e-05,
+      "loss": 0.1005,
+      "step": 4737
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9911212818082746e-05,
+      "loss": 0.0928,
+      "step": 4738
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.991078401616542e-05,
+      "loss": 0.1067,
+      "step": 4739
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9910354185920936e-05,
+      "loss": 0.0905,
+      "step": 4740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.990992332739389e-05,
+      "loss": 0.0939,
+      "step": 4741
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9909491440628997e-05,
+      "loss": 0.0795,
+      "step": 4742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.990905852567106e-05,
+      "loss": 0.0833,
+      "step": 4743
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9908624582565002e-05,
+      "loss": 0.0978,
+      "step": 4744
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9908189611355845e-05,
+      "loss": 0.103,
+      "step": 4745
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9907753612088726e-05,
+      "loss": 0.0909,
+      "step": 4746
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.990731658480888e-05,
+      "loss": 0.0825,
+      "step": 4747
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9906878529561652e-05,
+      "loss": 0.0977,
+      "step": 4748
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9906439446392495e-05,
+      "loss": 0.0813,
+      "step": 4749
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9905999335346967e-05,
+      "loss": 0.0851,
+      "step": 4750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9905558196470732e-05,
+      "loss": 0.0884,
+      "step": 4751
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9905116029809567e-05,
+      "loss": 0.0991,
+      "step": 4752
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.990467283540934e-05,
+      "loss": 0.0852,
+      "step": 4753
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.990422861331605e-05,
+      "loss": 0.097,
+      "step": 4754
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9903783363575778e-05,
+      "loss": 0.1058,
+      "step": 4755
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.990333708623473e-05,
+      "loss": 0.0857,
+      "step": 4756
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9902889781339202e-05,
+      "loss": 0.0803,
+      "step": 4757
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9902441448935614e-05,
+      "loss": 0.0731,
+      "step": 4758
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9901992089070483e-05,
+      "loss": 0.0962,
+      "step": 4759
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9901541701790427e-05,
+      "loss": 0.0733,
+      "step": 4760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9901090287142185e-05,
+      "loss": 0.0889,
+      "step": 4761
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9900637845172594e-05,
+      "loss": 0.1014,
+      "step": 4762
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9900184375928597e-05,
+      "loss": 0.1039,
+      "step": 4763
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989972987945724e-05,
+      "loss": 0.0837,
+      "step": 4764
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989927435580569e-05,
+      "loss": 0.0862,
+      "step": 4765
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9898817805021207e-05,
+      "loss": 0.0662,
+      "step": 4766
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9898360227151166e-05,
+      "loss": 0.0839,
+      "step": 4767
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9897901622243038e-05,
+      "loss": 0.0903,
+      "step": 4768
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989744199034441e-05,
+      "loss": 0.0878,
+      "step": 4769
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9896981331502974e-05,
+      "loss": 0.0825,
+      "step": 4770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989651964576653e-05,
+      "loss": 0.0978,
+      "step": 4771
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989605693318297e-05,
+      "loss": 0.0913,
+      "step": 4772
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9895593193800316e-05,
+      "loss": 0.0854,
+      "step": 4773
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989512842766668e-05,
+      "loss": 0.0826,
+      "step": 4774
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.989466263483029e-05,
+      "loss": 0.0892,
+      "step": 4775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9894195815339468e-05,
+      "loss": 0.0895,
+      "step": 4776
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9893727969242657e-05,
+      "loss": 0.0747,
+      "step": 4777
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.98932590965884e-05,
+      "loss": 0.0822,
+      "step": 4778
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.989278919742534e-05,
+      "loss": 0.0953,
+      "step": 4779
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.989231827180224e-05,
+      "loss": 0.0967,
+      "step": 4780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.989184631976796e-05,
+      "loss": 0.0813,
+      "step": 4781
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.989137334137147e-05,
+      "loss": 0.0827,
+      "step": 4782
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.989089933666184e-05,
+      "loss": 0.0732,
+      "step": 4783
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9890424305688262e-05,
+      "loss": 0.0946,
+      "step": 4784
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9889948248500014e-05,
+      "loss": 0.08,
+      "step": 4785
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9889471165146495e-05,
+      "loss": 0.098,
+      "step": 4786
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.988899305567721e-05,
+      "loss": 0.068,
+      "step": 4787
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9888513920141764e-05,
+      "loss": 0.1022,
+      "step": 4788
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.988803375858987e-05,
+      "loss": 0.0747,
+      "step": 4789
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9887552571071352e-05,
+      "loss": 0.0997,
+      "step": 4790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.988707035763613e-05,
+      "loss": 0.0795,
+      "step": 4791
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9886587118334248e-05,
+      "loss": 0.089,
+      "step": 4792
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9886102853215842e-05,
+      "loss": 0.0764,
+      "step": 4793
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9885617562331155e-05,
+      "loss": 0.0852,
+      "step": 4794
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.988513124573054e-05,
+      "loss": 0.0936,
+      "step": 4795
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9884643903464464e-05,
+      "loss": 0.0924,
+      "step": 4796
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9884155535583488e-05,
+      "loss": 0.0858,
+      "step": 4797
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9883666142138282e-05,
+      "loss": 0.0904,
+      "step": 4798
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.988317572317963e-05,
+      "loss": 0.0953,
+      "step": 4799
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9882684278758412e-05,
+      "loss": 0.0738,
+      "step": 4800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.988219180892562e-05,
+      "loss": 0.0827,
+      "step": 4801
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9881698313732355e-05,
+      "loss": 0.0785,
+      "step": 4802
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.988120379322982e-05,
+      "loss": 0.0753,
+      "step": 4803
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9880708247469328e-05,
+      "loss": 0.0835,
+      "step": 4804
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9880211676502287e-05,
+      "loss": 0.0852,
+      "step": 4805
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.987971408038023e-05,
+      "loss": 0.0975,
+      "step": 4806
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9879215459154787e-05,
+      "loss": 0.0862,
+      "step": 4807
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9878715812877686e-05,
+      "loss": 0.0817,
+      "step": 4808
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9878215141600778e-05,
+      "loss": 0.0858,
+      "step": 4809
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9877713445376005e-05,
+      "loss": 0.0826,
+      "step": 4810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9877210724255427e-05,
+      "loss": 0.0934,
+      "step": 4811
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9876706978291204e-05,
+      "loss": 0.0941,
+      "step": 4812
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.98762022075356e-05,
+      "loss": 0.1129,
+      "step": 4813
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9875696412041e-05,
+      "loss": 0.1027,
+      "step": 4814
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9875189591859873e-05,
+      "loss": 0.1042,
+      "step": 4815
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.987468174704481e-05,
+      "loss": 0.1022,
+      "step": 4816
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9874172877648506e-05,
+      "loss": 0.1167,
+      "step": 4817
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9873662983723762e-05,
+      "loss": 0.0858,
+      "step": 4818
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9873152065323476e-05,
+      "loss": 0.0907,
+      "step": 4819
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9872640122500666e-05,
+      "loss": 0.0853,
+      "step": 4820
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.987212715530845e-05,
+      "loss": 0.0847,
+      "step": 4821
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.987161316380005e-05,
+      "loss": 0.099,
+      "step": 4822
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.98710981480288e-05,
+      "loss": 0.1001,
+      "step": 4823
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9870582108048136e-05,
+      "loss": 0.0867,
+      "step": 4824
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9870065043911603e-05,
+      "loss": 0.0876,
+      "step": 4825
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9869546955672847e-05,
+      "loss": 0.0682,
+      "step": 4826
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.986902784338563e-05,
+      "loss": 0.0922,
+      "step": 4827
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9868507707103806e-05,
+      "loss": 0.0962,
+      "step": 4828
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.986798654688135e-05,
+      "loss": 0.0754,
+      "step": 4829
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9867464362772333e-05,
+      "loss": 0.078,
+      "step": 4830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.986694115483094e-05,
+      "loss": 0.0782,
+      "step": 4831
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9866416923111455e-05,
+      "loss": 0.0835,
+      "step": 4832
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9865891667668277e-05,
+      "loss": 0.0763,
+      "step": 4833
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9865365388555896e-05,
+      "loss": 0.0872,
+      "step": 4834
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9864838085828924e-05,
+      "loss": 0.0953,
+      "step": 4835
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9864309759542074e-05,
+      "loss": 0.0809,
+      "step": 4836
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986378040975016e-05,
+      "loss": 0.0958,
+      "step": 4837
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986325003650811e-05,
+      "loss": 0.0847,
+      "step": 4838
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9862718639870953e-05,
+      "loss": 0.0804,
+      "step": 4839
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9862186219893825e-05,
+      "loss": 0.082,
+      "step": 4840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9861652776631972e-05,
+      "loss": 0.0911,
+      "step": 4841
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9861118310140742e-05,
+      "loss": 0.0824,
+      "step": 4842
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9860582820475593e-05,
+      "loss": 0.0967,
+      "step": 4843
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986004630769208e-05,
+      "loss": 0.082,
+      "step": 4844
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985950877184587e-05,
+      "loss": 0.1036,
+      "step": 4845
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985897021299275e-05,
+      "loss": 0.0959,
+      "step": 4846
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9858430631188588e-05,
+      "loss": 0.0911,
+      "step": 4847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9857890026489374e-05,
+      "loss": 0.085,
+      "step": 4848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.98573483989512e-05,
+      "loss": 0.1003,
+      "step": 4849
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985680574863026e-05,
+      "loss": 0.0806,
+      "step": 4850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9856262075582865e-05,
+      "loss": 0.0956,
+      "step": 4851
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9855717379865424e-05,
+      "loss": 0.1044,
+      "step": 4852
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9855171661534453e-05,
+      "loss": 0.0785,
+      "step": 4853
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9854624920646575e-05,
+      "loss": 0.0769,
+      "step": 4854
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985407715725852e-05,
+      "loss": 0.0903,
+      "step": 4855
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985352837142712e-05,
+      "loss": 0.104,
+      "step": 4856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9852978563209318e-05,
+      "loss": 0.0813,
+      "step": 4857
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.985242773266216e-05,
+      "loss": 0.0933,
+      "step": 4858
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9851875879842803e-05,
+      "loss": 0.0911,
+      "step": 4859
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9851323004808505e-05,
+      "loss": 0.0817,
+      "step": 4860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.985076910761663e-05,
+      "loss": 0.0778,
+      "step": 4861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9850214188324647e-05,
+      "loss": 0.0646,
+      "step": 4862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9849658246990138e-05,
+      "loss": 0.0875,
+      "step": 4863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9849101283670787e-05,
+      "loss": 0.0854,
+      "step": 4864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9848543298424377e-05,
+      "loss": 0.1073,
+      "step": 4865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9847984291308813e-05,
+      "loss": 0.0865,
+      "step": 4866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9847424262382087e-05,
+      "loss": 0.0846,
+      "step": 4867
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9846863211702316e-05,
+      "loss": 0.07,
+      "step": 4868
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.984630113932771e-05,
+      "loss": 0.0919,
+      "step": 4869
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9845738045316584e-05,
+      "loss": 0.0951,
+      "step": 4870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.984517392972737e-05,
+      "loss": 0.0963,
+      "step": 4871
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9844608792618597e-05,
+      "loss": 0.0876,
+      "step": 4872
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9844042634048905e-05,
+      "loss": 0.0875,
+      "step": 4873
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.984347545407704e-05,
+      "loss": 0.0917,
+      "step": 4874
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.984290725276184e-05,
+      "loss": 0.0968,
+      "step": 4875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9842338030162273e-05,
+      "loss": 0.0862,
+      "step": 4876
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9841767786337392e-05,
+      "loss": 0.073,
+      "step": 4877
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.984119652134637e-05,
+      "loss": 0.1049,
+      "step": 4878
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.984062423524848e-05,
+      "loss": 0.0722,
+      "step": 4879
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.98400509281031e-05,
+      "loss": 0.1019,
+      "step": 4880
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983947659996972e-05,
+      "loss": 0.0889,
+      "step": 4881
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9838901250907924e-05,
+      "loss": 0.0984,
+      "step": 4882
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983832488097741e-05,
+      "loss": 0.0662,
+      "step": 4883
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983774749023799e-05,
+      "loss": 0.0694,
+      "step": 4884
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9837169078749567e-05,
+      "loss": 0.0801,
+      "step": 4885
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983658964657215e-05,
+      "loss": 0.0929,
+      "step": 4886
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983600919376587e-05,
+      "loss": 0.0733,
+      "step": 4887
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983542772039095e-05,
+      "loss": 0.0849,
+      "step": 4888
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9834845226507725e-05,
+      "loss": 0.1091,
+      "step": 4889
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9834261712176626e-05,
+      "loss": 0.0854,
+      "step": 4890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9833677177458207e-05,
+      "loss": 0.0844,
+      "step": 4891
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9833091622413115e-05,
+      "loss": 0.0776,
+      "step": 4892
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9832505047102104e-05,
+      "loss": 0.1062,
+      "step": 4893
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9831917451586036e-05,
+      "loss": 0.0781,
+      "step": 4894
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983132883592588e-05,
+      "loss": 0.1072,
+      "step": 4895
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9830739200182715e-05,
+      "loss": 0.0724,
+      "step": 4896
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.983014854441771e-05,
+      "loss": 0.078,
+      "step": 4897
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9829556868692163e-05,
+      "loss": 0.0986,
+      "step": 4898
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9828964173067457e-05,
+      "loss": 0.0848,
+      "step": 4899
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.982837045760509e-05,
+      "loss": 0.1024,
+      "step": 4900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9827775722366667e-05,
+      "loss": 0.0894,
+      "step": 4901
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9827179967413898e-05,
+      "loss": 0.0873,
+      "step": 4902
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.982658319280859e-05,
+      "loss": 0.0839,
+      "step": 4903
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9825985398612674e-05,
+      "loss": 0.0997,
+      "step": 4904
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.982538658488817e-05,
+      "loss": 0.0977,
+      "step": 4905
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9824786751697206e-05,
+      "loss": 0.0681,
+      "step": 4906
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9824185899102032e-05,
+      "loss": 0.0674,
+      "step": 4907
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9823584027164977e-05,
+      "loss": 0.0729,
+      "step": 4908
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.98229811359485e-05,
+      "loss": 0.0918,
+      "step": 4909
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9822377225515155e-05,
+      "loss": 0.0992,
+      "step": 4910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9821772295927603e-05,
+      "loss": 0.1036,
+      "step": 4911
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9821166347248607e-05,
+      "loss": 0.076,
+      "step": 4912
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9820559379541043e-05,
+      "loss": 0.0778,
+      "step": 4913
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9819951392867883e-05,
+      "loss": 0.079,
+      "step": 4914
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.981934238729222e-05,
+      "loss": 0.0863,
+      "step": 4915
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9818732362877234e-05,
+      "loss": 0.0823,
+      "step": 4916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9818121319686226e-05,
+      "loss": 0.0771,
+      "step": 4917
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9817509257782595e-05,
+      "loss": 0.0882,
+      "step": 4918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.981689617722985e-05,
+      "loss": 0.0947,
+      "step": 4919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.98162820780916e-05,
+      "loss": 0.0759,
+      "step": 4920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.981566696043156e-05,
+      "loss": 0.1085,
+      "step": 4921
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9815050824313565e-05,
+      "loss": 0.0927,
+      "step": 4922
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.981443366980153e-05,
+      "loss": 0.0833,
+      "step": 4923
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.98138154969595e-05,
+      "loss": 0.0757,
+      "step": 4924
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9813196305851618e-05,
+      "loss": 0.0873,
+      "step": 4925
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9812576096542117e-05,
+      "loss": 0.0746,
+      "step": 4926
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.981195486909536e-05,
+      "loss": 0.0844,
+      "step": 4927
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9811332623575802e-05,
+      "loss": 0.0925,
+      "step": 4928
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9810709360048005e-05,
+      "loss": 0.0735,
+      "step": 4929
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.981008507857664e-05,
+      "loss": 0.089,
+      "step": 4930
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9809459779226474e-05,
+      "loss": 0.0869,
+      "step": 4931
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9808833462062398e-05,
+      "loss": 0.0813,
+      "step": 4932
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.980820612714939e-05,
+      "loss": 0.0831,
+      "step": 4933
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9807577774552545e-05,
+      "loss": 0.0959,
+      "step": 4934
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9806948404337062e-05,
+      "loss": 0.0784,
+      "step": 4935
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9806318016568235e-05,
+      "loss": 0.0865,
+      "step": 4936
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9805686611311482e-05,
+      "loss": 0.0894,
+      "step": 4937
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9805054188632308e-05,
+      "loss": 0.0995,
+      "step": 4938
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.980442074859634e-05,
+      "loss": 0.0747,
+      "step": 4939
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9803786291269294e-05,
+      "loss": 0.0887,
+      "step": 4940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9803150816717008e-05,
+      "loss": 0.0731,
+      "step": 4941
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9802514325005414e-05,
+      "loss": 0.0756,
+      "step": 4942
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9801876816200557e-05,
+      "loss": 0.0962,
+      "step": 4943
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.980123829036858e-05,
+      "loss": 0.0902,
+      "step": 4944
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9800598747575734e-05,
+      "loss": 0.1036,
+      "step": 4945
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.979995818788838e-05,
+      "loss": 0.0806,
+      "step": 4946
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.979931661137298e-05,
+      "loss": 0.0859,
+      "step": 4947
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9798674018096106e-05,
+      "loss": 0.0929,
+      "step": 4948
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.979803040812443e-05,
+      "loss": 0.0753,
+      "step": 4949
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9797385781524734e-05,
+      "loss": 0.0941,
+      "step": 4950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.97967401383639e-05,
+      "loss": 0.1013,
+      "step": 4951
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9796093478708916e-05,
+      "loss": 0.0771,
+      "step": 4952
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.979544580262689e-05,
+      "loss": 0.0944,
+      "step": 4953
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9794797110185015e-05,
+      "loss": 0.0687,
+      "step": 4954
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9794147401450597e-05,
+      "loss": 0.1077,
+      "step": 4955
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9793496676491058e-05,
+      "loss": 0.0929,
+      "step": 4956
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9792844935373905e-05,
+      "loss": 0.0825,
+      "step": 4957
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9792192178166768e-05,
+      "loss": 0.076,
+      "step": 4958
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9791538404937378e-05,
+      "loss": 0.0789,
+      "step": 4959
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.979088361575356e-05,
+      "loss": 0.0883,
+      "step": 4960
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9790227810683263e-05,
+      "loss": 0.0936,
+      "step": 4961
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9789570989794532e-05,
+      "loss": 0.0934,
+      "step": 4962
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9788913153155512e-05,
+      "loss": 0.0782,
+      "step": 4963
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9788254300834464e-05,
+      "loss": 0.1024,
+      "step": 4964
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9787594432899742e-05,
+      "loss": 0.0811,
+      "step": 4965
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9786933549419826e-05,
+      "loss": 0.0999,
+      "step": 4966
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9786271650463276e-05,
+      "loss": 0.0956,
+      "step": 4967
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9785608736098778e-05,
+      "loss": 0.0903,
+      "step": 4968
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.978494480639511e-05,
+      "loss": 0.0953,
+      "step": 4969
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.978427986142116e-05,
+      "loss": 0.0807,
+      "step": 4970
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9783613901245923e-05,
+      "loss": 0.0814,
+      "step": 4971
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.97829469259385e-05,
+      "loss": 0.0964,
+      "step": 4972
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9782278935568092e-05,
+      "loss": 0.0944,
+      "step": 4973
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.978160993020401e-05,
+      "loss": 0.0859,
+      "step": 4974
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9780939909915666e-05,
+      "loss": 0.0845,
+      "step": 4975
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.978026887477259e-05,
+      "loss": 0.0922,
+      "step": 4976
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9779596824844396e-05,
+      "loss": 0.0757,
+      "step": 4977
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.977892376020082e-05,
+      "loss": 0.094,
+      "step": 4978
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9778249680911698e-05,
+      "loss": 0.0868,
+      "step": 4979
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.977757458704697e-05,
+      "loss": 0.0901,
+      "step": 4980
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9776898478676684e-05,
+      "loss": 0.0785,
+      "step": 4981
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9776221355870992e-05,
+      "loss": 0.0879,
+      "step": 4982
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.977554321870015e-05,
+      "loss": 0.0743,
+      "step": 4983
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9774864067234525e-05,
+      "loss": 0.0825,
+      "step": 4984
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.977418390154458e-05,
+      "loss": 0.0805,
+      "step": 4985
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9773502721700886e-05,
+      "loss": 0.0865,
+      "step": 4986
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9772820527774127e-05,
+      "loss": 0.0861,
+      "step": 4987
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9772137319835078e-05,
+      "loss": 0.0909,
+      "step": 4988
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9771453097954637e-05,
+      "loss": 0.0892,
+      "step": 4989
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9770767862203795e-05,
+      "loss": 0.0836,
+      "step": 4990
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9770081612653646e-05,
+      "loss": 0.0848,
+      "step": 4991
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.97693943493754e-05,
+      "loss": 0.088,
+      "step": 4992
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.976870607244036e-05,
+      "loss": 0.1029,
+      "step": 4993
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.976801678191995e-05,
+      "loss": 0.0953,
+      "step": 4994
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9767326477885683e-05,
+      "loss": 0.0837,
+      "step": 4995
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9766635160409186e-05,
+      "loss": 0.0891,
+      "step": 4996
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9765942829562186e-05,
+      "loss": 0.1002,
+      "step": 4997
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.976524948541652e-05,
+      "loss": 0.0975,
+      "step": 4998
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9764555128044128e-05,
+      "loss": 0.0839,
+      "step": 4999
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.976385975751706e-05,
+      "loss": 0.1095,
+      "step": 5000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9763163373907458e-05,
+      "loss": 0.0984,
+      "step": 5001
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9762465977287587e-05,
+      "loss": 0.0865,
+      "step": 5002
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.97617675677298e-05,
+      "loss": 0.0654,
+      "step": 5003
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9761068145306566e-05,
+      "loss": 0.1058,
+      "step": 5004
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.976036771009046e-05,
+      "loss": 0.0874,
+      "step": 5005
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9759666262154146e-05,
+      "loss": 0.0872,
+      "step": 5006
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9758963801570416e-05,
+      "loss": 0.0692,
+      "step": 5007
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9758260328412154e-05,
+      "loss": 0.0791,
+      "step": 5008
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9757555842752353e-05,
+      "loss": 0.0924,
+      "step": 5009
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9756850344664103e-05,
+      "loss": 0.0898,
+      "step": 5010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.975614383422061e-05,
+      "loss": 0.0844,
+      "step": 5011
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.975543631149518e-05,
+      "loss": 0.0702,
+      "step": 5012
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9754727776561222e-05,
+      "loss": 0.0963,
+      "step": 5013
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9754018229492254e-05,
+      "loss": 0.0941,
+      "step": 5014
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9753307670361904e-05,
+      "loss": 0.0746,
+      "step": 5015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.975259609924389e-05,
+      "loss": 0.1031,
+      "step": 5016
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.975188351621204e-05,
+      "loss": 0.0867,
+      "step": 5017
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9751169921340302e-05,
+      "loss": 0.0818,
+      "step": 5018
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9750455314702708e-05,
+      "loss": 0.0792,
+      "step": 5019
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.974973969637341e-05,
+      "loss": 0.0956,
+      "step": 5020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9749023066426658e-05,
+      "loss": 0.097,
+      "step": 5021
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9748305424936808e-05,
+      "loss": 0.1006,
+      "step": 5022
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.974758677197832e-05,
+      "loss": 0.0745,
+      "step": 5023
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9746867107625763e-05,
+      "loss": 0.0717,
+      "step": 5024
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9746146431953806e-05,
+      "loss": 0.0835,
+      "step": 5025
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9745424745037226e-05,
+      "loss": 0.0817,
+      "step": 5026
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9744702046950904e-05,
+      "loss": 0.082,
+      "step": 5027
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9743978337769827e-05,
+      "loss": 0.0906,
+      "step": 5028
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9743253617569085e-05,
+      "loss": 0.0989,
+      "step": 5029
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9742527886423872e-05,
+      "loss": 0.0878,
+      "step": 5030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9741801144409493e-05,
+      "loss": 0.1017,
+      "step": 5031
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.974107339160135e-05,
+      "loss": 0.1093,
+      "step": 5032
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9740344628074952e-05,
+      "loss": 0.0839,
+      "step": 5033
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.973961485390592e-05,
+      "loss": 0.0834,
+      "step": 5034
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9738884069169972e-05,
+      "loss": 0.0911,
+      "step": 5035
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.973815227394293e-05,
+      "loss": 0.1008,
+      "step": 5036
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9737419468300726e-05,
+      "loss": 0.0756,
+      "step": 5037
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9736685652319398e-05,
+      "loss": 0.0848,
+      "step": 5038
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9735950826075077e-05,
+      "loss": 0.0831,
+      "step": 5039
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9735214989644017e-05,
+      "loss": 0.0958,
+      "step": 5040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.973447814310256e-05,
+      "loss": 0.0662,
+      "step": 5041
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9733740286527166e-05,
+      "loss": 0.089,
+      "step": 5042
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.973300141999439e-05,
+      "loss": 0.0728,
+      "step": 5043
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9732261543580894e-05,
+      "loss": 0.0963,
+      "step": 5044
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9731520657363453e-05,
+      "loss": 0.088,
+      "step": 5045
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9730778761418934e-05,
+      "loss": 0.0813,
+      "step": 5046
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9730035855824317e-05,
+      "loss": 0.0727,
+      "step": 5047
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9729291940656684e-05,
+      "loss": 0.0901,
+      "step": 5048
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9728547015993223e-05,
+      "loss": 0.1025,
+      "step": 5049
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9727801081911227e-05,
+      "loss": 0.0764,
+      "step": 5050
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9727054138488095e-05,
+      "loss": 0.0999,
+      "step": 5051
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9726306185801323e-05,
+      "loss": 0.0751,
+      "step": 5052
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.972555722392852e-05,
+      "loss": 0.0973,
+      "step": 5053
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9724807252947396e-05,
+      "loss": 0.0778,
+      "step": 5054
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9724056272935774e-05,
+      "loss": 0.0801,
+      "step": 5055
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9723304283971566e-05,
+      "loss": 0.0811,
+      "step": 5056
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9722551286132797e-05,
+      "loss": 0.083,
+      "step": 5057
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.97217972794976e-05,
+      "loss": 0.0861,
+      "step": 5058
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9721042264144214e-05,
+      "loss": 0.087,
+      "step": 5059
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.972028624015097e-05,
+      "loss": 0.0822,
+      "step": 5060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9719529207596312e-05,
+      "loss": 0.0735,
+      "step": 5061
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9718771166558796e-05,
+      "loss": 0.0848,
+      "step": 5062
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.971801211711707e-05,
+      "loss": 0.0884,
+      "step": 5063
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9717252059349895e-05,
+      "loss": 0.0803,
+      "step": 5064
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.971649099333613e-05,
+      "loss": 0.0938,
+      "step": 5065
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.971572891915474e-05,
+      "loss": 0.077,
+      "step": 5066
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.97149658368848e-05,
+      "loss": 0.1057,
+      "step": 5067
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.971420174660549e-05,
+      "loss": 0.0754,
+      "step": 5068
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9713436648396086e-05,
+      "loss": 0.0593,
+      "step": 5069
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9712670542335968e-05,
+      "loss": 0.0984,
+      "step": 5070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.971190342850464e-05,
+      "loss": 0.1005,
+      "step": 5071
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.971113530698168e-05,
+      "loss": 0.0732,
+      "step": 5072
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9710366177846802e-05,
+      "loss": 0.0786,
+      "step": 5073
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9709596041179802e-05,
+      "loss": 0.0801,
+      "step": 5074
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.970882489706059e-05,
+      "loss": 0.101,
+      "step": 5075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.970805274556917e-05,
+      "loss": 0.0944,
+      "step": 5076
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.970727958678568e-05,
+      "loss": 0.0995,
+      "step": 5077
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.970650542079032e-05,
+      "loss": 0.0746,
+      "step": 5078
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9705730247663427e-05,
+      "loss": 0.0848,
+      "step": 5079
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9704954067485432e-05,
+      "loss": 0.0867,
+      "step": 5080
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9704176880336864e-05,
+      "loss": 0.0978,
+      "step": 5081
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9703398686298372e-05,
+      "loss": 0.0936,
+      "step": 5082
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.970261948545069e-05,
+      "loss": 0.085,
+      "step": 5083
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9701839277874676e-05,
+      "loss": 0.1053,
+      "step": 5084
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.970105806365128e-05,
+      "loss": 0.0872,
+      "step": 5085
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.970027584286155e-05,
+      "loss": 0.0959,
+      "step": 5086
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.969949261558666e-05,
+      "loss": 0.0748,
+      "step": 5087
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9698708381907878e-05,
+      "loss": 0.1058,
+      "step": 5088
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9697923141906563e-05,
+      "loss": 0.0879,
+      "step": 5089
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.96971368956642e-05,
+      "loss": 0.0913,
+      "step": 5090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9696349643262364e-05,
+      "loss": 0.0641,
+      "step": 5091
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9695561384782743e-05,
+      "loss": 0.0839,
+      "step": 5092
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.969477212030712e-05,
+      "loss": 0.1005,
+      "step": 5093
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.969398184991739e-05,
+      "loss": 0.0811,
+      "step": 5094
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.969319057369555e-05,
+      "loss": 0.0848,
+      "step": 5095
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9692398291723706e-05,
+      "loss": 0.0802,
+      "step": 5096
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.969160500408406e-05,
+      "loss": 0.0885,
+      "step": 5097
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.969081071085892e-05,
+      "loss": 0.0723,
+      "step": 5098
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9690015412130705e-05,
+      "loss": 0.0753,
+      "step": 5099
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9689219107981935e-05,
+      "loss": 0.1064,
+      "step": 5100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9688421798495225e-05,
+      "loss": 0.0748,
+      "step": 5101
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.968762348375331e-05,
+      "loss": 0.08,
+      "step": 5102
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.968682416383902e-05,
+      "loss": 0.088,
+      "step": 5103
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9686023838835292e-05,
+      "loss": 0.0936,
+      "step": 5104
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9685222508825162e-05,
+      "loss": 0.0847,
+      "step": 5105
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9684420173891784e-05,
+      "loss": 0.0885,
+      "step": 5106
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9683616834118398e-05,
+      "loss": 0.0912,
+      "step": 5107
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.968281248958836e-05,
+      "loss": 0.0919,
+      "step": 5108
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9682007140385128e-05,
+      "loss": 0.0963,
+      "step": 5109
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9681200786592265e-05,
+      "loss": 0.0846,
+      "step": 5110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9680393428293434e-05,
+      "loss": 0.1095,
+      "step": 5111
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9679585065572408e-05,
+      "loss": 0.0914,
+      "step": 5112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.967877569851306e-05,
+      "loss": 0.0762,
+      "step": 5113
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.967796532719937e-05,
+      "loss": 0.0919,
+      "step": 5114
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9677153951715418e-05,
+      "loss": 0.0897,
+      "step": 5115
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.967634157214539e-05,
+      "loss": 0.0792,
+      "step": 5116
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9675528188573585e-05,
+      "loss": 0.0947,
+      "step": 5117
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9674713801084394e-05,
+      "loss": 0.0927,
+      "step": 5118
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9673898409762315e-05,
+      "loss": 0.0713,
+      "step": 5119
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.967308201469195e-05,
+      "loss": 0.108,
+      "step": 5120
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.967226461595801e-05,
+      "loss": 0.1074,
+      "step": 5121
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9671446213645306e-05,
+      "loss": 0.0721,
+      "step": 5122
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9670626807838756e-05,
+      "loss": 0.0842,
+      "step": 5123
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.966980639862338e-05,
+      "loss": 0.0844,
+      "step": 5124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.96689849860843e-05,
+      "loss": 0.0865,
+      "step": 5125
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9668162570306743e-05,
+      "loss": 0.0939,
+      "step": 5126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9667339151376047e-05,
+      "loss": 0.0867,
+      "step": 5127
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.966651472937765e-05,
+      "loss": 0.0964,
+      "step": 5128
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9665689304397082e-05,
+      "loss": 0.086,
+      "step": 5129
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9664862876519995e-05,
+      "loss": 0.0896,
+      "step": 5130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.966403544583214e-05,
+      "loss": 0.0868,
+      "step": 5131
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9663207012419364e-05,
+      "loss": 0.0708,
+      "step": 5132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.966237757636763e-05,
+      "loss": 0.0934,
+      "step": 5133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9661547137762994e-05,
+      "loss": 0.0753,
+      "step": 5134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.9660715696691627e-05,
+      "loss": 0.0817,
+      "step": 5135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.965988325323979e-05,
+      "loss": 0.0977,
+      "step": 5136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.965904980749386e-05,
+      "loss": 0.0991,
+      "step": 5137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.965821535954031e-05,
+      "loss": 0.0941,
+      "step": 5138
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.965737990946573e-05,
+      "loss": 0.0765,
+      "step": 5139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.96565434573568e-05,
+      "loss": 0.0925,
+      "step": 5140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.965570600330031e-05,
+      "loss": 0.1046,
+      "step": 5141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.965486754738315e-05,
+      "loss": 0.1267,
+      "step": 5142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9654028089692317e-05,
+      "loss": 0.0764,
+      "step": 5143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9653187630314917e-05,
+      "loss": 0.1024,
+      "step": 5144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9652346169338147e-05,
+      "loss": 0.0865,
+      "step": 5145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.965150370684932e-05,
+      "loss": 0.0939,
+      "step": 5146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.965066024293585e-05,
+      "loss": 0.0853,
+      "step": 5147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9649815777685256e-05,
+      "loss": 0.0923,
+      "step": 5148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.964897031118515e-05,
+      "loss": 0.0845,
+      "step": 5149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9648123843523258e-05,
+      "loss": 0.1019,
+      "step": 5150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9647276374787415e-05,
+      "loss": 0.0768,
+      "step": 5151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9646427905065545e-05,
+      "loss": 0.0861,
+      "step": 5152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9645578434445693e-05,
+      "loss": 0.094,
+      "step": 5153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9644727963015985e-05,
+      "loss": 0.0834,
+      "step": 5154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9643876490864678e-05,
+      "loss": 0.0969,
+      "step": 5155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9643024018080115e-05,
+      "loss": 0.0756,
+      "step": 5156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9642170544750742e-05,
+      "loss": 0.0973,
+      "step": 5157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9641316070965123e-05,
+      "loss": 0.0934,
+      "step": 5158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9640460596811907e-05,
+      "loss": 0.08,
+      "step": 5159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9639604122379867e-05,
+      "loss": 0.0826,
+      "step": 5160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.963874664775786e-05,
+      "loss": 0.0826,
+      "step": 5161
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9637888173034866e-05,
+      "loss": 0.0935,
+      "step": 5162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9637028698299947e-05,
+      "loss": 0.0829,
+      "step": 5163
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9636168223642288e-05,
+      "loss": 0.0712,
+      "step": 5164
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9635306749151173e-05,
+      "loss": 0.0751,
+      "step": 5165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.963444427491598e-05,
+      "loss": 0.0891,
+      "step": 5166
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9633580801026207e-05,
+      "loss": 0.1024,
+      "step": 5167
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.963271632757144e-05,
+      "loss": 0.0895,
+      "step": 5168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9631850854641374e-05,
+      "loss": 0.0958,
+      "step": 5169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9630984382325816e-05,
+      "loss": 0.0739,
+      "step": 5170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9630116910714664e-05,
+      "loss": 0.0706,
+      "step": 5171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9629248439897924e-05,
+      "loss": 0.0763,
+      "step": 5172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9628378969965712e-05,
+      "loss": 0.0902,
+      "step": 5173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9627508501008245e-05,
+      "loss": 0.0737,
+      "step": 5174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9626637033115835e-05,
+      "loss": 0.0744,
+      "step": 5175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9625764566378903e-05,
+      "loss": 0.0948,
+      "step": 5176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9624891100887985e-05,
+      "loss": 0.0858,
+      "step": 5177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9624016636733702e-05,
+      "loss": 0.0919,
+      "step": 5178
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9623141174006785e-05,
+      "loss": 0.0875,
+      "step": 5179
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9622264712798078e-05,
+      "loss": 0.0782,
+      "step": 5180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9621387253198517e-05,
+      "loss": 0.088,
+      "step": 5181
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9620508795299148e-05,
+      "loss": 0.0857,
+      "step": 5182
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9619629339191112e-05,
+      "loss": 0.0922,
+      "step": 5183
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.961874888496567e-05,
+      "loss": 0.0876,
+      "step": 5184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.961786743271417e-05,
+      "loss": 0.0668,
+      "step": 5185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9616984982528073e-05,
+      "loss": 0.0984,
+      "step": 5186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9616101534498938e-05,
+      "loss": 0.0955,
+      "step": 5187
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.961521708871843e-05,
+      "loss": 0.0887,
+      "step": 5188
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.961433164527832e-05,
+      "loss": 0.0673,
+      "step": 5189
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.961344520427048e-05,
+      "loss": 0.0902,
+      "step": 5190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9612557765786884e-05,
+      "loss": 0.0778,
+      "step": 5191
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.961166932991961e-05,
+      "loss": 0.0856,
+      "step": 5192
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9610779896760847e-05,
+      "loss": 0.088,
+      "step": 5193
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9609889466402877e-05,
+      "loss": 0.0944,
+      "step": 5194
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9608998038938084e-05,
+      "loss": 0.076,
+      "step": 5195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.960810561445897e-05,
+      "loss": 0.0787,
+      "step": 5196
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.960721219305813e-05,
+      "loss": 0.0901,
+      "step": 5197
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.960631777482826e-05,
+      "loss": 0.0923,
+      "step": 5198
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.960542235986217e-05,
+      "loss": 0.0896,
+      "step": 5199
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9604525948252758e-05,
+      "loss": 0.0804,
+      "step": 5200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.960362854009304e-05,
+      "loss": 0.0856,
+      "step": 5201
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9602730135476127e-05,
+      "loss": 0.0731,
+      "step": 5202
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9601830734495236e-05,
+      "loss": 0.0868,
+      "step": 5203
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9600930337243694e-05,
+      "loss": 0.0822,
+      "step": 5204
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9600028943814913e-05,
+      "loss": 0.0764,
+      "step": 5205
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959912655430243e-05,
+      "loss": 0.0747,
+      "step": 5206
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9598223168799878e-05,
+      "loss": 0.0713,
+      "step": 5207
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959731878740098e-05,
+      "loss": 0.076,
+      "step": 5208
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9596413410199574e-05,
+      "loss": 0.0897,
+      "step": 5209
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959550703728961e-05,
+      "loss": 0.0743,
+      "step": 5210
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9594599668765127e-05,
+      "loss": 0.0944,
+      "step": 5211
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959369130472027e-05,
+      "loss": 0.0723,
+      "step": 5212
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959278194524929e-05,
+      "loss": 0.0881,
+      "step": 5213
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959187159044654e-05,
+      "loss": 0.0992,
+      "step": 5214
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9590960240406483e-05,
+      "loss": 0.1099,
+      "step": 5215
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9590047895223673e-05,
+      "loss": 0.0852,
+      "step": 5216
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9589134554992773e-05,
+      "loss": 0.072,
+      "step": 5217
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9588220219808554e-05,
+      "loss": 0.0765,
+      "step": 5218
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9587304889765887e-05,
+      "loss": 0.066,
+      "step": 5219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.958638856495974e-05,
+      "loss": 0.0904,
+      "step": 5220
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9585471245485193e-05,
+      "loss": 0.0936,
+      "step": 5221
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9584552931437423e-05,
+      "loss": 0.0782,
+      "step": 5222
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9583633622911713e-05,
+      "loss": 0.0787,
+      "step": 5223
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9582713320003454e-05,
+      "loss": 0.0802,
+      "step": 5224
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9581792022808125e-05,
+      "loss": 0.0853,
+      "step": 5225
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.958086973142133e-05,
+      "loss": 0.1012,
+      "step": 5226
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9579946445938755e-05,
+      "loss": 0.0911,
+      "step": 5227
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9579022166456208e-05,
+      "loss": 0.0885,
+      "step": 5228
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.957809689306958e-05,
+      "loss": 0.1056,
+      "step": 5229
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9577170625874885e-05,
+      "loss": 0.0876,
+      "step": 5230
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9576243364968226e-05,
+      "loss": 0.0848,
+      "step": 5231
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9575315110445815e-05,
+      "loss": 0.0966,
+      "step": 5232
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9574385862403965e-05,
+      "loss": 0.085,
+      "step": 5233
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.95734556209391e-05,
+      "loss": 0.0669,
+      "step": 5234
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.957252438614773e-05,
+      "loss": 0.0906,
+      "step": 5235
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9571592158126488e-05,
+      "loss": 0.0991,
+      "step": 5236
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9570658936972095e-05,
+      "loss": 0.0916,
+      "step": 5237
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.956972472278138e-05,
+      "loss": 0.0751,
+      "step": 5238
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.956878951565128e-05,
+      "loss": 0.085,
+      "step": 5239
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9567853315678826e-05,
+      "loss": 0.0927,
+      "step": 5240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.956691612296116e-05,
+      "loss": 0.0826,
+      "step": 5241
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9565977937595524e-05,
+      "loss": 0.0763,
+      "step": 5242
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9565038759679256e-05,
+      "loss": 0.0773,
+      "step": 5243
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.956409858930981e-05,
+      "loss": 0.0941,
+      "step": 5244
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9563157426584737e-05,
+      "loss": 0.0822,
+      "step": 5245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.956221527160169e-05,
+      "loss": 0.1073,
+      "step": 5246
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.956127212445842e-05,
+      "loss": 0.0802,
+      "step": 5247
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9560327985252794e-05,
+      "loss": 0.0979,
+      "step": 5248
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.955938285408277e-05,
+      "loss": 0.0957,
+      "step": 5249
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.955843673104641e-05,
+      "loss": 0.0986,
+      "step": 5250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.955748961624189e-05,
+      "loss": 0.09,
+      "step": 5251
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9556541509767477e-05,
+      "loss": 0.0878,
+      "step": 5252
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9555592411721548e-05,
+      "loss": 0.0975,
+      "step": 5253
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9554642322202574e-05,
+      "loss": 0.0776,
+      "step": 5254
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9553691241309137e-05,
+      "loss": 0.0923,
+      "step": 5255
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9552739169139927e-05,
+      "loss": 0.0884,
+      "step": 5256
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.955178610579372e-05,
+      "loss": 0.0858,
+      "step": 5257
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9550832051369408e-05,
+      "loss": 0.0768,
+      "step": 5258
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.954987700596598e-05,
+      "loss": 0.0874,
+      "step": 5259
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9548920969682535e-05,
+      "loss": 0.0883,
+      "step": 5260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9547963942618266e-05,
+      "loss": 0.0881,
+      "step": 5261
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9547005924872468e-05,
+      "loss": 0.0858,
+      "step": 5262
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9546046916544555e-05,
+      "loss": 0.0966,
+      "step": 5263
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9545086917734024e-05,
+      "loss": 0.0892,
+      "step": 5264
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9544125928540485e-05,
+      "loss": 0.0895,
+      "step": 5265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9543163949063648e-05,
+      "loss": 0.0852,
+      "step": 5266
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9542200979403327e-05,
+      "loss": 0.0903,
+      "step": 5267
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9541237019659438e-05,
+      "loss": 0.0919,
+      "step": 5268
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9540272069932e-05,
+      "loss": 0.094,
+      "step": 5269
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9539306130321132e-05,
+      "loss": 0.0743,
+      "step": 5270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9538339200927066e-05,
+      "loss": 0.08,
+      "step": 5271
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9537371281850123e-05,
+      "loss": 0.104,
+      "step": 5272
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9536402373190736e-05,
+      "loss": 0.0911,
+      "step": 5273
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.953543247504943e-05,
+      "loss": 0.0963,
+      "step": 5274
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9534461587526847e-05,
+      "loss": 0.0912,
+      "step": 5275
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9533489710723725e-05,
+      "loss": 0.0818,
+      "step": 5276
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.95325168447409e-05,
+      "loss": 0.08,
+      "step": 5277
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.953154298967932e-05,
+      "loss": 0.077,
+      "step": 5278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.953056814564003e-05,
+      "loss": 0.0881,
+      "step": 5279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.952959231272417e-05,
+      "loss": 0.077,
+      "step": 5280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9528615491033e-05,
+      "loss": 0.0709,
+      "step": 5281
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9527637680667874e-05,
+      "loss": 0.0873,
+      "step": 5282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.952665888173024e-05,
+      "loss": 0.1085,
+      "step": 5283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9525679094321667e-05,
+      "loss": 0.0741,
+      "step": 5284
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9524698318543803e-05,
+      "loss": 0.0747,
+      "step": 5285
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9523716554498428e-05,
+      "loss": 0.0829,
+      "step": 5286
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9522733802287394e-05,
+      "loss": 0.1072,
+      "step": 5287
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9521750062012678e-05,
+      "loss": 0.0872,
+      "step": 5288
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9520765333776347e-05,
+      "loss": 0.0728,
+      "step": 5289
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9519779617680577e-05,
+      "loss": 0.0818,
+      "step": 5290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9518792913827643e-05,
+      "loss": 0.0969,
+      "step": 5291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9517805222319926e-05,
+      "loss": 0.0935,
+      "step": 5292
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9516816543259908e-05,
+      "loss": 0.0745,
+      "step": 5293
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9515826876750168e-05,
+      "loss": 0.1091,
+      "step": 5294
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9514836222893393e-05,
+      "loss": 0.0832,
+      "step": 5295
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.951384458179238e-05,
+      "loss": 0.0714,
+      "step": 5296
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.951285195355001e-05,
+      "loss": 0.0774,
+      "step": 5297
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.951185833826928e-05,
+      "loss": 0.1014,
+      "step": 5298
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9510863736053286e-05,
+      "loss": 0.0883,
+      "step": 5299
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9509868147005228e-05,
+      "loss": 0.0791,
+      "step": 5300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9508871571228404e-05,
+      "loss": 0.0671,
+      "step": 5301
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.950787400882622e-05,
+      "loss": 0.0797,
+      "step": 5302
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.950687545990218e-05,
+      "loss": 0.0678,
+      "step": 5303
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9505875924559888e-05,
+      "loss": 0.0809,
+      "step": 5304
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.950487540290306e-05,
+      "loss": 0.0822,
+      "step": 5305
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.950387389503551e-05,
+      "loss": 0.0956,
+      "step": 5306
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9502871401061146e-05,
+      "loss": 0.0925,
+      "step": 5307
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.950186792108399e-05,
+      "loss": 0.0653,
+      "step": 5308
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9500863455208158e-05,
+      "loss": 0.1085,
+      "step": 5309
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9499858003537875e-05,
+      "loss": 0.0876,
+      "step": 5310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9498851566177462e-05,
+      "loss": 0.087,
+      "step": 5311
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.949784414323135e-05,
+      "loss": 0.0889,
+      "step": 5312
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9496835734804062e-05,
+      "loss": 0.0798,
+      "step": 5313
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9495826341000237e-05,
+      "loss": 0.0834,
+      "step": 5314
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9494815961924597e-05,
+      "loss": 0.0744,
+      "step": 5315
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9493804597681986e-05,
+      "loss": 0.0808,
+      "step": 5316
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9492792248377337e-05,
+      "loss": 0.0917,
+      "step": 5317
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.9491778914115692e-05,
+      "loss": 0.0881,
+      "step": 5318
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.949076459500219e-05,
+      "loss": 0.0634,
+      "step": 5319
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.948974929114208e-05,
+      "loss": 0.0797,
+      "step": 5320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.948873300264071e-05,
+      "loss": 0.0989,
+      "step": 5321
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.948771572960352e-05,
+      "loss": 0.0747,
+      "step": 5322
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9486697472136063e-05,
+      "loss": 0.083,
+      "step": 5323
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9485678230343995e-05,
+      "loss": 0.0616,
+      "step": 5324
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.948465800433307e-05,
+      "loss": 0.079,
+      "step": 5325
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9483636794209143e-05,
+      "loss": 0.0671,
+      "step": 5326
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9482614600078178e-05,
+      "loss": 0.0664,
+      "step": 5327
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.948159142204623e-05,
+      "loss": 0.0687,
+      "step": 5328
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9480567260219466e-05,
+      "loss": 0.079,
+      "step": 5329
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9479542114704152e-05,
+      "loss": 0.0782,
+      "step": 5330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.947851598560665e-05,
+      "loss": 0.0933,
+      "step": 5331
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9477488873033435e-05,
+      "loss": 0.0883,
+      "step": 5332
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.947646077709108e-05,
+      "loss": 0.0823,
+      "step": 5333
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9475431697886256e-05,
+      "loss": 0.0898,
+      "step": 5334
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9474401635525738e-05,
+      "loss": 0.0863,
+      "step": 5335
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9473370590116403e-05,
+      "loss": 0.0938,
+      "step": 5336
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.947233856176523e-05,
+      "loss": 0.079,
+      "step": 5337
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9471305550579305e-05,
+      "loss": 0.101,
+      "step": 5338
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.9470271556665807e-05,
+      "loss": 0.083,
+      "step": 5339
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9469236580132024e-05,
+      "loss": 0.0925,
+      "step": 5340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.946820062108534e-05,
+      "loss": 0.0882,
+      "step": 5341
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9467163679633254e-05,
+      "loss": 0.0817,
+      "step": 5342
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9466125755883347e-05,
+      "loss": 0.0698,
+      "step": 5343
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9465086849943318e-05,
+      "loss": 0.0908,
+      "step": 5344
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.946404696192096e-05,
+      "loss": 0.1038,
+      "step": 5345
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9463006091924165e-05,
+      "loss": 0.0845,
+      "step": 5346
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9461964240060944e-05,
+      "loss": 0.0712,
+      "step": 5347
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.946092140643939e-05,
+      "loss": 0.0856,
+      "step": 5348
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9459877591167706e-05,
+      "loss": 0.0885,
+      "step": 5349
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9458832794354198e-05,
+      "loss": 0.0883,
+      "step": 5350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9457787016107273e-05,
+      "loss": 0.0811,
+      "step": 5351
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9456740256535437e-05,
+      "loss": 0.0675,
+      "step": 5352
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9455692515747298e-05,
+      "loss": 0.0922,
+      "step": 5353
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945464379385158e-05,
+      "loss": 0.0808,
+      "step": 5354
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.945359409095708e-05,
+      "loss": 0.0951,
+      "step": 5355
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9452543407172727e-05,
+      "loss": 0.0856,
+      "step": 5356
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9451491742607534e-05,
+      "loss": 0.0756,
+      "step": 5357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9450439097370614e-05,
+      "loss": 0.0822,
+      "step": 5358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9449385471571197e-05,
+      "loss": 0.0844,
+      "step": 5359
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.94483308653186e-05,
+      "loss": 0.0898,
+      "step": 5360
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9447275278722247e-05,
+      "loss": 0.0903,
+      "step": 5361
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9446218711891666e-05,
+      "loss": 0.0823,
+      "step": 5362
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9445161164936486e-05,
+      "loss": 0.0773,
+      "step": 5363
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9444102637966433e-05,
+      "loss": 0.105,
+      "step": 5364
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9443043131091343e-05,
+      "loss": 0.0791,
+      "step": 5365
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9441982644421144e-05,
+      "loss": 0.0764,
+      "step": 5366
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9440921178065868e-05,
+      "loss": 0.0853,
+      "step": 5367
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9439858732135657e-05,
+      "loss": 0.0649,
+      "step": 5368
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.943879530674075e-05,
+      "loss": 0.0786,
+      "step": 5369
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.943773090199148e-05,
+      "loss": 0.0814,
+      "step": 5370
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.943666551799829e-05,
+      "loss": 0.094,
+      "step": 5371
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9435599154871726e-05,
+      "loss": 0.0748,
+      "step": 5372
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9434531812722427e-05,
+      "loss": 0.0846,
+      "step": 5373
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9433463491661143e-05,
+      "loss": 0.1053,
+      "step": 5374
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.943239419179872e-05,
+      "loss": 0.0907,
+      "step": 5375
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9431323913246107e-05,
+      "loss": 0.0837,
+      "step": 5376
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.943025265611435e-05,
+      "loss": 0.0841,
+      "step": 5377
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9429180420514608e-05,
+      "loss": 0.1005,
+      "step": 5378
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.9428107206558133e-05,
+      "loss": 0.0897,
+      "step": 5379
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9427033014356276e-05,
+      "loss": 0.0867,
+      "step": 5380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9425957844020494e-05,
+      "loss": 0.0878,
+      "step": 5381
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.942488169566235e-05,
+      "loss": 0.087,
+      "step": 5382
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9423804569393497e-05,
+      "loss": 0.082,
+      "step": 5383
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.94227264653257e-05,
+      "loss": 0.076,
+      "step": 5384
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9421647383570822e-05,
+      "loss": 0.1106,
+      "step": 5385
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9420567324240822e-05,
+      "loss": 0.0741,
+      "step": 5386
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.941948628744777e-05,
+      "loss": 0.0905,
+      "step": 5387
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.941840427330383e-05,
+      "loss": 0.0833,
+      "step": 5388
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9417321281921275e-05,
+      "loss": 0.0667,
+      "step": 5389
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9416237313412467e-05,
+      "loss": 0.0942,
+      "step": 5390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.941515236788988e-05,
+      "loss": 0.0989,
+      "step": 5391
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.941406644546609e-05,
+      "loss": 0.088,
+      "step": 5392
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.941297954625376e-05,
+      "loss": 0.0825,
+      "step": 5393
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.941189167036568e-05,
+      "loss": 0.0964,
+      "step": 5394
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9410802817914715e-05,
+      "loss": 0.0803,
+      "step": 5395
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9409712989013848e-05,
+      "loss": 0.0812,
+      "step": 5396
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9408622183776154e-05,
+      "loss": 0.0685,
+      "step": 5397
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9407530402314818e-05,
+      "loss": 0.0753,
+      "step": 5398
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.9406437644743115e-05,
+      "loss": 0.069,
+      "step": 5399
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9405343911174432e-05,
+      "loss": 0.0914,
+      "step": 5400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9404249201722255e-05,
+      "loss": 0.0804,
+      "step": 5401
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9403153516500165e-05,
+      "loss": 0.0765,
+      "step": 5402
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.940205685562185e-05,
+      "loss": 0.0706,
+      "step": 5403
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9400959219201096e-05,
+      "loss": 0.0837,
+      "step": 5404
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.93998606073518e-05,
+      "loss": 0.0737,
+      "step": 5405
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.939876102018794e-05,
+      "loss": 0.074,
+      "step": 5406
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9397660457823618e-05,
+      "loss": 0.0908,
+      "step": 5407
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9396558920373022e-05,
+      "loss": 0.0817,
+      "step": 5408
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9395456407950445e-05,
+      "loss": 0.0857,
+      "step": 5409
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9394352920670282e-05,
+      "loss": 0.0897,
+      "step": 5410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.939324845864703e-05,
+      "loss": 0.0759,
+      "step": 5411
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.939214302199529e-05,
+      "loss": 0.075,
+      "step": 5412
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9391036610829753e-05,
+      "loss": 0.0887,
+      "step": 5413
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9389929225265222e-05,
+      "loss": 0.082,
+      "step": 5414
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.93888208654166e-05,
+      "loss": 0.0825,
+      "step": 5415
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9387711531398883e-05,
+      "loss": 0.0824,
+      "step": 5416
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9386601223327183e-05,
+      "loss": 0.0917,
+      "step": 5417
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9385489941316692e-05,
+      "loss": 0.0822,
+      "step": 5418
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.9384377685482725e-05,
+      "loss": 0.095,
+      "step": 5419
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.938326445594068e-05,
+      "loss": 0.0818,
+      "step": 5420
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9382150252806072e-05,
+      "loss": 0.0853,
+      "step": 5421
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9381035076194502e-05,
+      "loss": 0.0826,
+      "step": 5422
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.937991892622168e-05,
+      "loss": 0.0856,
+      "step": 5423
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9378801803003417e-05,
+      "loss": 0.0768,
+      "step": 5424
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9377683706655626e-05,
+      "loss": 0.0822,
+      "step": 5425
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.937656463729432e-05,
+      "loss": 0.0878,
+      "step": 5426
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9375444595035605e-05,
+      "loss": 0.1047,
+      "step": 5427
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.93743235799957e-05,
+      "loss": 0.0782,
+      "step": 5428
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.937320159229092e-05,
+      "loss": 0.0824,
+      "step": 5429
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9372078632037676e-05,
+      "loss": 0.0894,
+      "step": 5430
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.937095469935249e-05,
+      "loss": 0.095,
+      "step": 5431
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.936982979435198e-05,
+      "loss": 0.0874,
+      "step": 5432
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9368703917152857e-05,
+      "loss": 0.0793,
+      "step": 5433
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9367577067871948e-05,
+      "loss": 0.0912,
+      "step": 5434
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9366449246626167e-05,
+      "loss": 0.0886,
+      "step": 5435
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.936532045353254e-05,
+      "loss": 0.0968,
+      "step": 5436
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9364190688708184e-05,
+      "loss": 0.0802,
+      "step": 5437
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.936305995227033e-05,
+      "loss": 0.0873,
+      "step": 5438
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.936192824433629e-05,
+      "loss": 0.0796,
+      "step": 5439
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9360795565023494e-05,
+      "loss": 0.0851,
+      "step": 5440
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.935966191444947e-05,
+      "loss": 0.0939,
+      "step": 5441
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.935852729273184e-05,
+      "loss": 0.0983,
+      "step": 5442
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.935739169998833e-05,
+      "loss": 0.0835,
+      "step": 5443
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.935625513633677e-05,
+      "loss": 0.0793,
+      "step": 5444
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9355117601895084e-05,
+      "loss": 0.0848,
+      "step": 5445
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9353979096781304e-05,
+      "loss": 0.0812,
+      "step": 5446
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9352839621113562e-05,
+      "loss": 0.0922,
+      "step": 5447
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9351699175010082e-05,
+      "loss": 0.0851,
+      "step": 5448
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9350557758589195e-05,
+      "loss": 0.0969,
+      "step": 5449
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.934941537196934e-05,
+      "loss": 0.0828,
+      "step": 5450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9348272015269045e-05,
+      "loss": 0.0969,
+      "step": 5451
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.934712768860694e-05,
+      "loss": 0.0985,
+      "step": 5452
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.934598239210176e-05,
+      "loss": 0.0752,
+      "step": 5453
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9344836125872345e-05,
+      "loss": 0.0682,
+      "step": 5454
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.934368889003762e-05,
+      "loss": 0.0828,
+      "step": 5455
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.934254068471663e-05,
+      "loss": 0.0723,
+      "step": 5456
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.93413915100285e-05,
+      "loss": 0.0931,
+      "step": 5457
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9340241366092475e-05,
+      "loss": 0.1045,
+      "step": 5458
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9339090253027893e-05,
+      "loss": 0.1033,
+      "step": 5459
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9337938170954184e-05,
+      "loss": 0.0856,
+      "step": 5460
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9336785119990894e-05,
+      "loss": 0.1068,
+      "step": 5461
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9335631100257655e-05,
+      "loss": 0.0881,
+      "step": 5462
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.933447611187421e-05,
+      "loss": 0.0848,
+      "step": 5463
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9333320154960403e-05,
+      "loss": 0.0873,
+      "step": 5464
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9332163229636168e-05,
+      "loss": 0.0769,
+      "step": 5465
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9331005336021543e-05,
+      "loss": 0.0857,
+      "step": 5466
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9329846474236676e-05,
+      "loss": 0.085,
+      "step": 5467
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9328686644401807e-05,
+      "loss": 0.0778,
+      "step": 5468
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9327525846637277e-05,
+      "loss": 0.0919,
+      "step": 5469
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.932636408106353e-05,
+      "loss": 0.0929,
+      "step": 5470
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9325201347801105e-05,
+      "loss": 0.101,
+      "step": 5471
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.932403764697065e-05,
+      "loss": 0.0902,
+      "step": 5472
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9322872978692907e-05,
+      "loss": 0.0764,
+      "step": 5473
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.932170734308872e-05,
+      "loss": 0.0797,
+      "step": 5474
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9320540740279035e-05,
+      "loss": 0.0895,
+      "step": 5475
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9319373170384895e-05,
+      "loss": 0.0841,
+      "step": 5476
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9318204633527442e-05,
+      "loss": 0.0903,
+      "step": 5477
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9317035129827925e-05,
+      "loss": 0.0923,
+      "step": 5478
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9315864659407696e-05,
+      "loss": 0.0778,
+      "step": 5479
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.931469322238819e-05,
+      "loss": 0.0898,
+      "step": 5480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9313520818890957e-05,
+      "loss": 0.0999,
+      "step": 5481
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.931234744903765e-05,
+      "loss": 0.0719,
+      "step": 5482
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.931117311295001e-05,
+      "loss": 0.0841,
+      "step": 5483
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9309997810749883e-05,
+      "loss": 0.0883,
+      "step": 5484
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.930882154255922e-05,
+      "loss": 0.0908,
+      "step": 5485
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.930764430850007e-05,
+      "loss": 0.0844,
+      "step": 5486
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9306466108694574e-05,
+      "loss": 0.1008,
+      "step": 5487
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.930528694326499e-05,
+      "loss": 0.0889,
+      "step": 5488
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9304106812333657e-05,
+      "loss": 0.0777,
+      "step": 5489
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9302925716023025e-05,
+      "loss": 0.0761,
+      "step": 5490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9301743654455652e-05,
+      "loss": 0.1042,
+      "step": 5491
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9300560627754176e-05,
+      "loss": 0.0921,
+      "step": 5492
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.929937663604135e-05,
+      "loss": 0.0933,
+      "step": 5493
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9298191679440024e-05,
+      "loss": 0.0981,
+      "step": 5494
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9297005758073145e-05,
+      "loss": 0.0798,
+      "step": 5495
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.929581887206377e-05,
+      "loss": 0.0814,
+      "step": 5496
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.929463102153503e-05,
+      "loss": 0.0892,
+      "step": 5497
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9293442206610196e-05,
+      "loss": 0.0978,
+      "step": 5498
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.9292252427412605e-05,
+      "loss": 0.067,
+      "step": 5499
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.929106168406571e-05,
+      "loss": 0.0853,
+      "step": 5500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.928986997669306e-05,
+      "loss": 0.0867,
+      "step": 5501
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9288677305418306e-05,
+      "loss": 0.0819,
+      "step": 5502
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9287483670365193e-05,
+      "loss": 0.0841,
+      "step": 5503
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9286289071657576e-05,
+      "loss": 0.0856,
+      "step": 5504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9285093509419404e-05,
+      "loss": 0.0876,
+      "step": 5505
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9283896983774727e-05,
+      "loss": 0.0672,
+      "step": 5506
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.928269949484769e-05,
+      "loss": 0.0737,
+      "step": 5507
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.928150104276255e-05,
+      "loss": 0.1001,
+      "step": 5508
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9280301627643647e-05,
+      "loss": 0.0869,
+      "step": 5509
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.927910124961544e-05,
+      "loss": 0.0781,
+      "step": 5510
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.927789990880247e-05,
+      "loss": 0.0881,
+      "step": 5511
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9276697605329392e-05,
+      "loss": 0.0847,
+      "step": 5512
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9275494339320954e-05,
+      "loss": 0.0587,
+      "step": 5513
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9274290110902002e-05,
+      "loss": 0.0781,
+      "step": 5514
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9273084920197488e-05,
+      "loss": 0.0725,
+      "step": 5515
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9271878767332457e-05,
+      "loss": 0.0789,
+      "step": 5516
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9270671652432063e-05,
+      "loss": 0.0792,
+      "step": 5517
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9269463575621552e-05,
+      "loss": 0.0876,
+      "step": 5518
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.926825453702627e-05,
+      "loss": 0.0673,
+      "step": 5519
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.9267044536771667e-05,
+      "loss": 0.088,
+      "step": 5520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.926583357498329e-05,
+      "loss": 0.0927,
+      "step": 5521
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9264621651786787e-05,
+      "loss": 0.093,
+      "step": 5522
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9263408767307903e-05,
+      "loss": 0.0817,
+      "step": 5523
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.926219492167249e-05,
+      "loss": 0.0795,
+      "step": 5524
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.926098011500649e-05,
+      "loss": 0.0911,
+      "step": 5525
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.925976434743595e-05,
+      "loss": 0.0759,
+      "step": 5526
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9258547619087017e-05,
+      "loss": 0.0774,
+      "step": 5527
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.925732993008594e-05,
+      "loss": 0.0836,
+      "step": 5528
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.925611128055906e-05,
+      "loss": 0.0823,
+      "step": 5529
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9254891670632823e-05,
+      "loss": 0.0851,
+      "step": 5530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9253671100433773e-05,
+      "loss": 0.0928,
+      "step": 5531
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9252449570088555e-05,
+      "loss": 0.0785,
+      "step": 5532
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9251227079723917e-05,
+      "loss": 0.0922,
+      "step": 5533
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9250003629466696e-05,
+      "loss": 0.0876,
+      "step": 5534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9248779219443838e-05,
+      "loss": 0.0765,
+      "step": 5535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.924755384978239e-05,
+      "loss": 0.0817,
+      "step": 5536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9246327520609488e-05,
+      "loss": 0.0942,
+      "step": 5537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9245100232052377e-05,
+      "loss": 0.1044,
+      "step": 5538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.92438719842384e-05,
+      "loss": 0.0935,
+      "step": 5539
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.9242642777294996e-05,
+      "loss": 0.0845,
+      "step": 5540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9241412611349704e-05,
+      "loss": 0.0732,
+      "step": 5541
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9240181486530166e-05,
+      "loss": 0.0729,
+      "step": 5542
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.923894940296412e-05,
+      "loss": 0.1016,
+      "step": 5543
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.923771636077941e-05,
+      "loss": 0.0958,
+      "step": 5544
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.923648236010397e-05,
+      "loss": 0.1002,
+      "step": 5545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.923524740106584e-05,
+      "loss": 0.0813,
+      "step": 5546
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9234011483793153e-05,
+      "loss": 0.0818,
+      "step": 5547
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9232774608414153e-05,
+      "loss": 0.0738,
+      "step": 5548
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9231536775057172e-05,
+      "loss": 0.0847,
+      "step": 5549
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.923029798385064e-05,
+      "loss": 0.0876,
+      "step": 5550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9229058234923104e-05,
+      "loss": 0.0713,
+      "step": 5551
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9227817528403194e-05,
+      "loss": 0.0883,
+      "step": 5552
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9226575864419635e-05,
+      "loss": 0.0796,
+      "step": 5553
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9225333243101275e-05,
+      "loss": 0.0771,
+      "step": 5554
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9224089664577034e-05,
+      "loss": 0.0828,
+      "step": 5555
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9222845128975947e-05,
+      "loss": 0.0923,
+      "step": 5556
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.922159963642715e-05,
+      "loss": 0.0884,
+      "step": 5557
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9220353187059865e-05,
+      "loss": 0.0766,
+      "step": 5558
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.921910578100343e-05,
+      "loss": 0.0922,
+      "step": 5559
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.921785741838727e-05,
+      "loss": 0.0824,
+      "step": 5560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.921660809934091e-05,
+      "loss": 0.0884,
+      "step": 5561
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9215357823993985e-05,
+      "loss": 0.0714,
+      "step": 5562
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9214106592476215e-05,
+      "loss": 0.0767,
+      "step": 5563
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9212854404917424e-05,
+      "loss": 0.1121,
+      "step": 5564
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9211601261447544e-05,
+      "loss": 0.0833,
+      "step": 5565
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9210347162196598e-05,
+      "loss": 0.0824,
+      "step": 5566
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9209092107294704e-05,
+      "loss": 0.0931,
+      "step": 5567
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9207836096872093e-05,
+      "loss": 0.0867,
+      "step": 5568
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9206579131059076e-05,
+      "loss": 0.0923,
+      "step": 5569
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9205321209986083e-05,
+      "loss": 0.0852,
+      "step": 5570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.920406233378363e-05,
+      "loss": 0.0748,
+      "step": 5571
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9202802502582334e-05,
+      "loss": 0.0856,
+      "step": 5572
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9201541716512917e-05,
+      "loss": 0.0862,
+      "step": 5573
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9200279975706192e-05,
+      "loss": 0.0909,
+      "step": 5574
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.919901728029308e-05,
+      "loss": 0.0846,
+      "step": 5575
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9197753630404595e-05,
+      "loss": 0.0875,
+      "step": 5576
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9196489026171846e-05,
+      "loss": 0.0871,
+      "step": 5577
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9195223467726056e-05,
+      "loss": 0.084,
+      "step": 5578
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.919395695519853e-05,
+      "loss": 0.069,
+      "step": 5579
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.919268948872068e-05,
+      "loss": 0.0776,
+      "step": 5580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9191421068424017e-05,
+      "loss": 0.0834,
+      "step": 5581
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.919015169444015e-05,
+      "loss": 0.0673,
+      "step": 5582
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9188881366900788e-05,
+      "loss": 0.0823,
+      "step": 5583
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.918761008593774e-05,
+      "loss": 0.0851,
+      "step": 5584
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.918633785168291e-05,
+      "loss": 0.0796,
+      "step": 5585
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.91850646642683e-05,
+      "loss": 0.0868,
+      "step": 5586
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9183790523826022e-05,
+      "loss": 0.0734,
+      "step": 5587
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9182515430488272e-05,
+      "loss": 0.0991,
+      "step": 5588
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9181239384387355e-05,
+      "loss": 0.0868,
+      "step": 5589
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9179962385655665e-05,
+      "loss": 0.075,
+      "step": 5590
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9178684434425712e-05,
+      "loss": 0.0859,
+      "step": 5591
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9177405530830087e-05,
+      "loss": 0.0856,
+      "step": 5592
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9176125675001487e-05,
+      "loss": 0.094,
+      "step": 5593
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9174844867072712e-05,
+      "loss": 0.0884,
+      "step": 5594
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.917356310717665e-05,
+      "loss": 0.0924,
+      "step": 5595
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.91722803954463e-05,
+      "loss": 0.0811,
+      "step": 5596
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.9170996732014756e-05,
+      "loss": 0.0804,
+      "step": 5597
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.91697121170152e-05,
+      "loss": 0.0923,
+      "step": 5598
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.916842655058093e-05,
+      "loss": 0.094,
+      "step": 5599
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.916714003284533e-05,
+      "loss": 0.0809,
+      "step": 5600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.916585256394189e-05,
+      "loss": 0.0832,
+      "step": 5601
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.916456414400419e-05,
+      "loss": 0.0854,
+      "step": 5602
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9163274773165923e-05,
+      "loss": 0.0928,
+      "step": 5603
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9161984451560867e-05,
+      "loss": 0.0889,
+      "step": 5604
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.91606931793229e-05,
+      "loss": 0.0835,
+      "step": 5605
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.915940095658601e-05,
+      "loss": 0.0754,
+      "step": 5606
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9158107783484275e-05,
+      "loss": 0.0771,
+      "step": 5607
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9156813660151866e-05,
+      "loss": 0.0803,
+      "step": 5608
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9155518586723066e-05,
+      "loss": 0.0969,
+      "step": 5609
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9154222563332245e-05,
+      "loss": 0.0853,
+      "step": 5610
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9152925590113878e-05,
+      "loss": 0.0591,
+      "step": 5611
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.915162766720254e-05,
+      "loss": 0.0883,
+      "step": 5612
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9150328794732896e-05,
+      "loss": 0.0767,
+      "step": 5613
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9149028972839718e-05,
+      "loss": 0.0955,
+      "step": 5614
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.914772820165787e-05,
+      "loss": 0.0824,
+      "step": 5615
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9146426481322327e-05,
+      "loss": 0.078,
+      "step": 5616
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9145123811968145e-05,
+      "loss": 0.0882,
+      "step": 5617
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9143820193730488e-05,
+      "loss": 0.0861,
+      "step": 5618
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.914251562674462e-05,
+      "loss": 0.0818,
+      "step": 5619
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9141210111145898e-05,
+      "loss": 0.0794,
+      "step": 5620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.913990364706978e-05,
+      "loss": 0.0828,
+      "step": 5621
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9138596234651832e-05,
+      "loss": 0.0731,
+      "step": 5622
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9137287874027696e-05,
+      "loss": 0.0732,
+      "step": 5623
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.913597856533313e-05,
+      "loss": 0.0793,
+      "step": 5624
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9134668308703984e-05,
+      "loss": 0.0736,
+      "step": 5625
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9133357104276218e-05,
+      "loss": 0.0799,
+      "step": 5626
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.913204495218587e-05,
+      "loss": 0.069,
+      "step": 5627
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9130731852569088e-05,
+      "loss": 0.0978,
+      "step": 5628
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.912941780556212e-05,
+      "loss": 0.0939,
+      "step": 5629
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.912810281130131e-05,
+      "loss": 0.0754,
+      "step": 5630
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9126786869923094e-05,
+      "loss": 0.0666,
+      "step": 5631
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.912546998156402e-05,
+      "loss": 0.074,
+      "step": 5632
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.912415214636072e-05,
+      "loss": 0.0721,
+      "step": 5633
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.912283336444994e-05,
+      "loss": 0.0734,
+      "step": 5634
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9121513635968497e-05,
+      "loss": 0.0864,
+      "step": 5635
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.912019296105334e-05,
+      "loss": 0.0781,
+      "step": 5636
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9118871339841495e-05,
+      "loss": 0.0865,
+      "step": 5637
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9117548772470093e-05,
+      "loss": 0.0756,
+      "step": 5638
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9116225259076354e-05,
+      "loss": 0.0708,
+      "step": 5639
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9114900799797612e-05,
+      "loss": 0.0934,
+      "step": 5640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9113575394771287e-05,
+      "loss": 0.0732,
+      "step": 5641
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9112249044134903e-05,
+      "loss": 0.0752,
+      "step": 5642
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9110921748026076e-05,
+      "loss": 0.0799,
+      "step": 5643
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.910959350658253e-05,
+      "loss": 0.0835,
+      "step": 5644
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9108264319942077e-05,
+      "loss": 0.0901,
+      "step": 5645
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9106934188242635e-05,
+      "loss": 0.0846,
+      "step": 5646
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9105603111622212e-05,
+      "loss": 0.0784,
+      "step": 5647
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.910427109021892e-05,
+      "loss": 0.072,
+      "step": 5648
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9102938124170968e-05,
+      "loss": 0.0933,
+      "step": 5649
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.910160421361666e-05,
+      "loss": 0.0762,
+      "step": 5650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.91002693586944e-05,
+      "loss": 0.0847,
+      "step": 5651
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9098933559542702e-05,
+      "loss": 0.0811,
+      "step": 5652
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.909759681630015e-05,
+      "loss": 0.0786,
+      "step": 5653
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9096259129105453e-05,
+      "loss": 0.0628,
+      "step": 5654
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.90949204980974e-05,
+      "loss": 0.0994,
+      "step": 5655
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.909358092341489e-05,
+      "loss": 0.099,
+      "step": 5656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.909224040519692e-05,
+      "loss": 0.0914,
+      "step": 5657
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.909089894358257e-05,
+      "loss": 0.0953,
+      "step": 5658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.908955653871103e-05,
+      "loss": 0.0876,
+      "step": 5659
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.908821319072159e-05,
+      "loss": 0.094,
+      "step": 5660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.908686889975363e-05,
+      "loss": 0.0685,
+      "step": 5661
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.908552366594664e-05,
+      "loss": 0.0793,
+      "step": 5662
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9084177489440187e-05,
+      "loss": 0.0765,
+      "step": 5663
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9082830370373954e-05,
+      "loss": 0.0737,
+      "step": 5664
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9081482308887716e-05,
+      "loss": 0.0829,
+      "step": 5665
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9080133305121343e-05,
+      "loss": 0.0783,
+      "step": 5666
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9078783359214812e-05,
+      "loss": 0.0836,
+      "step": 5667
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9077432471308182e-05,
+      "loss": 0.0942,
+      "step": 5668
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9076080641541626e-05,
+      "loss": 0.0785,
+      "step": 5669
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9074727870055407e-05,
+      "loss": 0.0776,
+      "step": 5670
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9073374156989888e-05,
+      "loss": 0.0706,
+      "step": 5671
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.907201950248552e-05,
+      "loss": 0.0909,
+      "step": 5672
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9070663906682866e-05,
+      "loss": 0.0743,
+      "step": 5673
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.906930736972258e-05,
+      "loss": 0.0925,
+      "step": 5674
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9067949891745413e-05,
+      "loss": 0.0885,
+      "step": 5675
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9066591472892216e-05,
+      "loss": 0.072,
+      "step": 5676
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9065232113303934e-05,
+      "loss": 0.0736,
+      "step": 5677
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9063871813121613e-05,
+      "loss": 0.0789,
+      "step": 5678
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.90625105724864e-05,
+      "loss": 0.0716,
+      "step": 5679
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061148391539534e-05,
+      "loss": 0.0731,
+      "step": 5680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9059785270422342e-05,
+      "loss": 0.1046,
+      "step": 5681
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9058421209276272e-05,
+      "loss": 0.0885,
+      "step": 5682
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905705620824285e-05,
+      "loss": 0.0886,
+      "step": 5683
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9055690267463708e-05,
+      "loss": 0.1036,
+      "step": 5684
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905432338708058e-05,
+      "loss": 0.0854,
+      "step": 5685
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.905295556723528e-05,
+      "loss": 0.0863,
+      "step": 5686
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9051586808069737e-05,
+      "loss": 0.0889,
+      "step": 5687
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9050217109725975e-05,
+      "loss": 0.0788,
+      "step": 5688
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9048846472346102e-05,
+      "loss": 0.0797,
+      "step": 5689
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9047474896072342e-05,
+      "loss": 0.1011,
+      "step": 5690
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9046102381047003e-05,
+      "loss": 0.0839,
+      "step": 5691
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9044728927412495e-05,
+      "loss": 0.0873,
+      "step": 5692
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.904335453531133e-05,
+      "loss": 0.0848,
+      "step": 5693
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9041979204886107e-05,
+      "loss": 0.0702,
+      "step": 5694
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.904060293627953e-05,
+      "loss": 0.0961,
+      "step": 5695
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.90392257296344e-05,
+      "loss": 0.0845,
+      "step": 5696
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.903784758509361e-05,
+      "loss": 0.0918,
+      "step": 5697
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.903646850280016e-05,
+      "loss": 0.08,
+      "step": 5698
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9035088482897136e-05,
+      "loss": 0.0912,
+      "step": 5699
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.903370752552773e-05,
+      "loss": 0.0691,
+      "step": 5700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9032325630835227e-05,
+      "loss": 0.0732,
+      "step": 5701
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9030942798963007e-05,
+      "loss": 0.0671,
+      "step": 5702
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9029559030054558e-05,
+      "loss": 0.0876,
+      "step": 5703
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.902817432425345e-05,
+      "loss": 0.0785,
+      "step": 5704
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9026788681703357e-05,
+      "loss": 0.079,
+      "step": 5705
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.902540210254806e-05,
+      "loss": 0.0842,
+      "step": 5706
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.902401458693142e-05,
+      "loss": 0.089,
+      "step": 5707
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9022626134997412e-05,
+      "loss": 0.0876,
+      "step": 5708
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9021236746890087e-05,
+      "loss": 0.0845,
+      "step": 5709
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9019846422753615e-05,
+      "loss": 0.0774,
+      "step": 5710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9018455162732256e-05,
+      "loss": 0.0668,
+      "step": 5711
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9017062966970353e-05,
+      "loss": 0.0806,
+      "step": 5712
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9015669835612375e-05,
+      "loss": 0.0605,
+      "step": 5713
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9014275768802855e-05,
+      "loss": 0.0973,
+      "step": 5714
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9012880766686445e-05,
+      "loss": 0.0737,
+      "step": 5715
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.901148482940789e-05,
+      "loss": 0.0848,
+      "step": 5716
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9010087957112032e-05,
+      "loss": 0.0696,
+      "step": 5717
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.90086901499438e-05,
+      "loss": 0.0863,
+      "step": 5718
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9007291408048238e-05,
+      "loss": 0.085,
+      "step": 5719
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.900589173157047e-05,
+      "loss": 0.0848,
+      "step": 5720
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.900449112065573e-05,
+      "loss": 0.0774,
+      "step": 5721
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.900308957544934e-05,
+      "loss": 0.0988,
+      "step": 5722
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9001687096096724e-05,
+      "loss": 0.0721,
+      "step": 5723
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.9000283682743394e-05,
+      "loss": 0.0643,
+      "step": 5724
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8998879335534973e-05,
+      "loss": 0.0891,
+      "step": 5725
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8997474054617177e-05,
+      "loss": 0.0704,
+      "step": 5726
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8996067840135804e-05,
+      "loss": 0.0867,
+      "step": 5727
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.899466069223677e-05,
+      "loss": 0.0709,
+      "step": 5728
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8993252611066077e-05,
+      "loss": 0.0729,
+      "step": 5729
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8991843596769827e-05,
+      "loss": 0.0784,
+      "step": 5730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.899043364949421e-05,
+      "loss": 0.0854,
+      "step": 5731
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8989022769385523e-05,
+      "loss": 0.0775,
+      "step": 5732
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.898761095659016e-05,
+      "loss": 0.0813,
+      "step": 5733
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8986198211254604e-05,
+      "loss": 0.0936,
+      "step": 5734
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8984784533525445e-05,
+      "loss": 0.0728,
+      "step": 5735
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8983369923549357e-05,
+      "loss": 0.0875,
+      "step": 5736
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8981954381473122e-05,
+      "loss": 0.0932,
+      "step": 5737
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.898053790744361e-05,
+      "loss": 0.0853,
+      "step": 5738
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8979120501607803e-05,
+      "loss": 0.0675,
+      "step": 5739
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8977702164112757e-05,
+      "loss": 0.0929,
+      "step": 5740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8976282895105642e-05,
+      "loss": 0.0752,
+      "step": 5741
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8974862694733716e-05,
+      "loss": 0.1067,
+      "step": 5742
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8973441563144338e-05,
+      "loss": 0.0985,
+      "step": 5743
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8972019500484964e-05,
+      "loss": 0.0649,
+      "step": 5744
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8970596506903144e-05,
+      "loss": 0.0773,
+      "step": 5745
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8969172582546528e-05,
+      "loss": 0.0738,
+      "step": 5746
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.896774772756285e-05,
+      "loss": 0.0742,
+      "step": 5747
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8966321942099967e-05,
+      "loss": 0.0861,
+      "step": 5748
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8964895226305802e-05,
+      "loss": 0.0891,
+      "step": 5749
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8963467580328397e-05,
+      "loss": 0.0858,
+      "step": 5750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8962039004315876e-05,
+      "loss": 0.0991,
+      "step": 5751
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.896060949841647e-05,
+      "loss": 0.0961,
+      "step": 5752
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8959179062778503e-05,
+      "loss": 0.0778,
+      "step": 5753
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8957747697550394e-05,
+      "loss": 0.0895,
+      "step": 5754
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8956315402880655e-05,
+      "loss": 0.0775,
+      "step": 5755
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8954882178917903e-05,
+      "loss": 0.0853,
+      "step": 5756
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8953448025810847e-05,
+      "loss": 0.0772,
+      "step": 5757
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.895201294370829e-05,
+      "loss": 0.0818,
+      "step": 5758
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8950576932759138e-05,
+      "loss": 0.0799,
+      "step": 5759
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8949139993112386e-05,
+      "loss": 0.0884,
+      "step": 5760
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8947702124917126e-05,
+      "loss": 0.0945,
+      "step": 5761
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8946263328322555e-05,
+      "loss": 0.0862,
+      "step": 5762
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.894482360347796e-05,
+      "loss": 0.0646,
+      "step": 5763
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8943382950532713e-05,
+      "loss": 0.0823,
+      "step": 5764
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.894194136963631e-05,
+      "loss": 0.0997,
+      "step": 5765
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8940498860938315e-05,
+      "loss": 0.0933,
+      "step": 5766
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8939055424588407e-05,
+      "loss": 0.0892,
+      "step": 5767
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8937611060736355e-05,
+      "loss": 0.0576,
+      "step": 5768
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.893616576953202e-05,
+      "loss": 0.0689,
+      "step": 5769
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8934719551125364e-05,
+      "loss": 0.0803,
+      "step": 5770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8933272405666444e-05,
+      "loss": 0.0812,
+      "step": 5771
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8931824333305418e-05,
+      "loss": 0.0732,
+      "step": 5772
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.893037533419253e-05,
+      "loss": 0.0597,
+      "step": 5773
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8928925408478128e-05,
+      "loss": 0.0804,
+      "step": 5774
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8927474556312656e-05,
+      "loss": 0.0811,
+      "step": 5775
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8926022777846647e-05,
+      "loss": 0.0839,
+      "step": 5776
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.892457007323074e-05,
+      "loss": 0.1095,
+      "step": 5777
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8923116442615666e-05,
+      "loss": 0.071,
+      "step": 5778
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8921661886152248e-05,
+      "loss": 0.0829,
+      "step": 5779
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8920206403991407e-05,
+      "loss": 0.0786,
+      "step": 5780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8918749996284167e-05,
+      "loss": 0.0881,
+      "step": 5781
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8917292663181638e-05,
+      "loss": 0.0839,
+      "step": 5782
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8915834404835036e-05,
+      "loss": 0.0924,
+      "step": 5783
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.891437522139566e-05,
+      "loss": 0.0896,
+      "step": 5784
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8912915113014918e-05,
+      "loss": 0.074,
+      "step": 5785
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8911454079844305e-05,
+      "loss": 0.0754,
+      "step": 5786
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.890999212203542e-05,
+      "loss": 0.0776,
+      "step": 5787
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8908529239739946e-05,
+      "loss": 0.0828,
+      "step": 5788
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8907065433109676e-05,
+      "loss": 0.084,
+      "step": 5789
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8905600702296495e-05,
+      "loss": 0.0696,
+      "step": 5790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.890413504745237e-05,
+      "loss": 0.0717,
+      "step": 5791
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8902668468729385e-05,
+      "loss": 0.0726,
+      "step": 5792
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8901200966279707e-05,
+      "loss": 0.0914,
+      "step": 5793
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.88997325402556e-05,
+      "loss": 0.0902,
+      "step": 5794
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.889826319080943e-05,
+      "loss": 0.0728,
+      "step": 5795
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8896792918093645e-05,
+      "loss": 0.0769,
+      "step": 5796
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8895321722260806e-05,
+      "loss": 0.0759,
+      "step": 5797
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.889384960346356e-05,
+      "loss": 0.079,
+      "step": 5798
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8892376561854653e-05,
+      "loss": 0.0842,
+      "step": 5799
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8890902597586926e-05,
+      "loss": 0.0718,
+      "step": 5800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8889427710813308e-05,
+      "loss": 0.088,
+      "step": 5801
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8887951901686842e-05,
+      "loss": 0.072,
+      "step": 5802
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8886475170360644e-05,
+      "loss": 0.0923,
+      "step": 5803
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.888499751698795e-05,
+      "loss": 0.0765,
+      "step": 5804
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8883518941722065e-05,
+      "loss": 0.0984,
+      "step": 5805
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8882039444716417e-05,
+      "loss": 0.067,
+      "step": 5806
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8880559026124507e-05,
+      "loss": 0.0821,
+      "step": 5807
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8879077686099944e-05,
+      "loss": 0.0895,
+      "step": 5808
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8877595424796425e-05,
+      "loss": 0.0892,
+      "step": 5809
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8876112242367758e-05,
+      "loss": 0.08,
+      "step": 5810
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8874628138967827e-05,
+      "loss": 0.0714,
+      "step": 5811
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.887314311475062e-05,
+      "loss": 0.0901,
+      "step": 5812
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8871657169870228e-05,
+      "loss": 0.0956,
+      "step": 5813
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8870170304480825e-05,
+      "loss": 0.0627,
+      "step": 5814
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.886868251873668e-05,
+      "loss": 0.0857,
+      "step": 5815
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8867193812792174e-05,
+      "loss": 0.0682,
+      "step": 5816
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.886570418680177e-05,
+      "loss": 0.0692,
+      "step": 5817
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8864213640920023e-05,
+      "loss": 0.0742,
+      "step": 5818
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8862722175301595e-05,
+      "loss": 0.071,
+      "step": 5819
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8861229790101238e-05,
+      "loss": 0.0766,
+      "step": 5820
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.88597364854738e-05,
+      "loss": 0.0852,
+      "step": 5821
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8858242261574216e-05,
+      "loss": 0.0626,
+      "step": 5822
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.885674711855754e-05,
+      "loss": 0.0873,
+      "step": 5823
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.885525105657889e-05,
+      "loss": 0.0962,
+      "step": 5824
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.88537540757935e-05,
+      "loss": 0.095,
+      "step": 5825
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8852256176356704e-05,
+      "loss": 0.093,
+      "step": 5826
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8850757358423907e-05,
+      "loss": 0.0868,
+      "step": 5827
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.884925762215063e-05,
+      "loss": 0.0787,
+      "step": 5828
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8847756967692488e-05,
+      "loss": 0.0803,
+      "step": 5829
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.884625539520518e-05,
+      "loss": 0.0888,
+      "step": 5830
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.884475290484451e-05,
+      "loss": 0.0763,
+      "step": 5831
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.884324949676637e-05,
+      "loss": 0.0895,
+      "step": 5832
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8841745171126757e-05,
+      "loss": 0.073,
+      "step": 5833
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8840239928081755e-05,
+      "loss": 0.078,
+      "step": 5834
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8838733767787543e-05,
+      "loss": 0.0768,
+      "step": 5835
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.88372266904004e-05,
+      "loss": 0.0696,
+      "step": 5836
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.88357186960767e-05,
+      "loss": 0.0831,
+      "step": 5837
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8834209784972905e-05,
+      "loss": 0.0735,
+      "step": 5838
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8832699957245585e-05,
+      "loss": 0.0764,
+      "step": 5839
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.883118921305139e-05,
+      "loss": 0.0836,
+      "step": 5840
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8829677552547073e-05,
+      "loss": 0.0656,
+      "step": 5841
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8828164975889486e-05,
+      "loss": 0.0739,
+      "step": 5842
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.882665148323557e-05,
+      "loss": 0.1042,
+      "step": 5843
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8825137074742358e-05,
+      "loss": 0.0702,
+      "step": 5844
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.882362175056699e-05,
+      "loss": 0.0729,
+      "step": 5845
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8822105510866686e-05,
+      "loss": 0.0836,
+      "step": 5846
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8820588355798776e-05,
+      "loss": 0.0626,
+      "step": 5847
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8819070285520673e-05,
+      "loss": 0.0992,
+      "step": 5848
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8817551300189893e-05,
+      "loss": 0.0714,
+      "step": 5849
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8816031399964037e-05,
+      "loss": 0.0801,
+      "step": 5850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8814510585000813e-05,
+      "loss": 0.0928,
+      "step": 5851
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8812988855458013e-05,
+      "loss": 0.0696,
+      "step": 5852
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.881146621149354e-05,
+      "loss": 0.0863,
+      "step": 5853
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.880994265326537e-05,
+      "loss": 0.0698,
+      "step": 5854
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.880841818093159e-05,
+      "loss": 0.0815,
+      "step": 5855
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.880689279465038e-05,
+      "loss": 0.076,
+      "step": 5856
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8805366494580002e-05,
+      "loss": 0.0684,
+      "step": 5857
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8803839280878827e-05,
+      "loss": 0.0874,
+      "step": 5858
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8802311153705324e-05,
+      "loss": 0.0794,
+      "step": 5859
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8800782113218038e-05,
+      "loss": 0.0892,
+      "step": 5860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8799252159575627e-05,
+      "loss": 0.087,
+      "step": 5861
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.879772129293683e-05,
+      "loss": 0.0883,
+      "step": 5862
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8796189513460495e-05,
+      "loss": 0.0762,
+      "step": 5863
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.879465682130555e-05,
+      "loss": 0.0885,
+      "step": 5864
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8793123216631032e-05,
+      "loss": 0.0831,
+      "step": 5865
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8791588699596057e-05,
+      "loss": 0.0767,
+      "step": 5866
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.879005327035985e-05,
+      "loss": 0.08,
+      "step": 5867
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.878851692908172e-05,
+      "loss": 0.0874,
+      "step": 5868
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.878697967592108e-05,
+      "loss": 0.0912,
+      "step": 5869
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8785441511037434e-05,
+      "loss": 0.0869,
+      "step": 5870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.878390243459037e-05,
+      "loss": 0.0847,
+      "step": 5871
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8782362446739594e-05,
+      "loss": 0.0826,
+      "step": 5872
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8780821547644882e-05,
+      "loss": 0.0828,
+      "step": 5873
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.877927973746612e-05,
+      "loss": 0.0754,
+      "step": 5874
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.877773701636328e-05,
+      "loss": 0.0834,
+      "step": 5875
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8776193384496436e-05,
+      "loss": 0.0735,
+      "step": 5876
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8774648842025752e-05,
+      "loss": 0.0912,
+      "step": 5877
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8773103389111486e-05,
+      "loss": 0.0859,
+      "step": 5878
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8771557025913995e-05,
+      "loss": 0.0802,
+      "step": 5879
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8770009752593723e-05,
+      "loss": 0.0794,
+      "step": 5880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8768461569311215e-05,
+      "loss": 0.0862,
+      "step": 5881
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8766912476227105e-05,
+      "loss": 0.0838,
+      "step": 5882
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.876536247350213e-05,
+      "loss": 0.0822,
+      "step": 5883
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.876381156129711e-05,
+      "loss": 0.089,
+      "step": 5884
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.876225973977297e-05,
+      "loss": 0.0851,
+      "step": 5885
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.876070700909072e-05,
+      "loss": 0.0807,
+      "step": 5886
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.875915336941147e-05,
+      "loss": 0.0731,
+      "step": 5887
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8757598820896427e-05,
+      "loss": 0.0748,
+      "step": 5888
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.875604336370689e-05,
+      "loss": 0.0789,
+      "step": 5889
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.875448699800424e-05,
+      "loss": 0.0806,
+      "step": 5890
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.875292972394997e-05,
+      "loss": 0.0815,
+      "step": 5891
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8751371541705663e-05,
+      "loss": 0.0851,
+      "step": 5892
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.874981245143299e-05,
+      "loss": 0.0875,
+      "step": 5893
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8748252453293717e-05,
+      "loss": 0.0768,
+      "step": 5894
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8746691547449713e-05,
+      "loss": 0.0862,
+      "step": 5895
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8745129734062934e-05,
+      "loss": 0.0927,
+      "step": 5896
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8743567013295427e-05,
+      "loss": 0.0701,
+      "step": 5897
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8742003385309337e-05,
+      "loss": 0.0737,
+      "step": 5898
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8740438850266907e-05,
+      "loss": 0.0775,
+      "step": 5899
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8738873408330475e-05,
+      "loss": 0.0739,
+      "step": 5900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8737307059662463e-05,
+      "loss": 0.0763,
+      "step": 5901
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.873573980442539e-05,
+      "loss": 0.0792,
+      "step": 5902
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8734171642781877e-05,
+      "loss": 0.0736,
+      "step": 5903
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.873260257489463e-05,
+      "loss": 0.0901,
+      "step": 5904
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.873103260092646e-05,
+      "loss": 0.0857,
+      "step": 5905
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.872946172104026e-05,
+      "loss": 0.0923,
+      "step": 5906
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.872788993539902e-05,
+      "loss": 0.0852,
+      "step": 5907
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.872631724416583e-05,
+      "loss": 0.0806,
+      "step": 5908
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8724743647503865e-05,
+      "loss": 0.0901,
+      "step": 5909
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8723169145576404e-05,
+      "loss": 0.078,
+      "step": 5910
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8721593738546815e-05,
+      "loss": 0.0876,
+      "step": 5911
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.872001742657856e-05,
+      "loss": 0.0898,
+      "step": 5912
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8718440209835187e-05,
+      "loss": 0.0892,
+      "step": 5913
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8716862088480353e-05,
+      "loss": 0.1005,
+      "step": 5914
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.87152830626778e-05,
+      "loss": 0.08,
+      "step": 5915
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8713703132591365e-05,
+      "loss": 0.0968,
+      "step": 5916
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8712122298384977e-05,
+      "loss": 0.078,
+      "step": 5917
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.871054056022266e-05,
+      "loss": 0.0714,
+      "step": 5918
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.870895791826854e-05,
+      "loss": 0.0912,
+      "step": 5919
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8707374372686825e-05,
+      "loss": 0.078,
+      "step": 5920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8705789923641815e-05,
+      "loss": 0.0903,
+      "step": 5921
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.870420457129792e-05,
+      "loss": 0.075,
+      "step": 5922
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8702618315819628e-05,
+      "loss": 0.093,
+      "step": 5923
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8701031157371523e-05,
+      "loss": 0.0802,
+      "step": 5924
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8699443096118294e-05,
+      "loss": 0.0811,
+      "step": 5925
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8697854132224713e-05,
+      "loss": 0.08,
+      "step": 5926
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8696264265855647e-05,
+      "loss": 0.0944,
+      "step": 5927
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.869467349717606e-05,
+      "loss": 0.1105,
+      "step": 5928
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8693081826351002e-05,
+      "loss": 0.0811,
+      "step": 5929
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.869148925354563e-05,
+      "loss": 0.0714,
+      "step": 5930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8689895778925185e-05,
+      "loss": 0.0679,
+      "step": 5931
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8688301402654995e-05,
+      "loss": 0.077,
+      "step": 5932
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8686706124900502e-05,
+      "loss": 0.0905,
+      "step": 5933
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.868510994582722e-05,
+      "loss": 0.0825,
+      "step": 5934
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.868351286560077e-05,
+      "loss": 0.0844,
+      "step": 5935
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.868191488438687e-05,
+      "loss": 0.0681,
+      "step": 5936
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8680316002351308e-05,
+      "loss": 0.0837,
+      "step": 5937
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8678716219659992e-05,
+      "loss": 0.0924,
+      "step": 5938
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8677115536478917e-05,
+      "loss": 0.067,
+      "step": 5939
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8675513952974156e-05,
+      "loss": 0.0639,
+      "step": 5940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.867391146931189e-05,
+      "loss": 0.0862,
+      "step": 5941
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8672308085658395e-05,
+      "loss": 0.0766,
+      "step": 5942
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.867070380218003e-05,
+      "loss": 0.077,
+      "step": 5943
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.866909861904326e-05,
+      "loss": 0.093,
+      "step": 5944
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8667492536414627e-05,
+      "loss": 0.0808,
+      "step": 5945
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8665885554460784e-05,
+      "loss": 0.0745,
+      "step": 5946
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8664277673348463e-05,
+      "loss": 0.0781,
+      "step": 5947
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.86626688932445e-05,
+      "loss": 0.0807,
+      "step": 5948
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.866105921431581e-05,
+      "loss": 0.0698,
+      "step": 5949
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8659448636729426e-05,
+      "loss": 0.0875,
+      "step": 5950
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8657837160652447e-05,
+      "loss": 0.0915,
+      "step": 5951
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8656224786252077e-05,
+      "loss": 0.0979,
+      "step": 5952
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8654611513695622e-05,
+      "loss": 0.0743,
+      "step": 5953
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8652997343150466e-05,
+      "loss": 0.0754,
+      "step": 5954
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8651382274784095e-05,
+      "loss": 0.0784,
+      "step": 5955
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8649766308764085e-05,
+      "loss": 0.081,
+      "step": 5956
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8648149445258104e-05,
+      "loss": 0.0918,
+      "step": 5957
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8646531684433924e-05,
+      "loss": 0.0676,
+      "step": 5958
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.864491302645939e-05,
+      "loss": 0.0648,
+      "step": 5959
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8643293471502458e-05,
+      "loss": 0.0865,
+      "step": 5960
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.864167301973117e-05,
+      "loss": 0.0889,
+      "step": 5961
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8640051671313656e-05,
+      "loss": 0.0848,
+      "step": 5962
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.863842942641815e-05,
+      "loss": 0.0663,
+      "step": 5963
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8636806285212975e-05,
+      "loss": 0.0787,
+      "step": 5964
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8635182247866545e-05,
+      "loss": 0.0712,
+      "step": 5965
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.863355731454736e-05,
+      "loss": 0.0815,
+      "step": 5966
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.863193148542403e-05,
+      "loss": 0.0856,
+      "step": 5967
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8630304760665237e-05,
+      "loss": 0.075,
+      "step": 5968
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8628677140439784e-05,
+      "loss": 0.0914,
+      "step": 5969
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8627048624916532e-05,
+      "loss": 0.0948,
+      "step": 5970
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.862541921426447e-05,
+      "loss": 0.0813,
+      "step": 5971
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8623788908652653e-05,
+      "loss": 0.0813,
+      "step": 5972
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.862215770825024e-05,
+      "loss": 0.0857,
+      "step": 5973
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.862052561322648e-05,
+      "loss": 0.0916,
+      "step": 5974
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8618892623750723e-05,
+      "loss": 0.0869,
+      "step": 5975
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.86172587399924e-05,
+      "loss": 0.0802,
+      "step": 5976
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8615623962121043e-05,
+      "loss": 0.0815,
+      "step": 5977
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8613988290306274e-05,
+      "loss": 0.0896,
+      "step": 5978
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.861235172471781e-05,
+      "loss": 0.0826,
+      "step": 5979
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.861071426552545e-05,
+      "loss": 0.0878,
+      "step": 5980
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8609075912899096e-05,
+      "loss": 0.0775,
+      "step": 5981
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8607436667008748e-05,
+      "loss": 0.0767,
+      "step": 5982
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.860579652802449e-05,
+      "loss": 0.0703,
+      "step": 5983
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8604155496116495e-05,
+      "loss": 0.0821,
+      "step": 5984
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.860251357145504e-05,
+      "loss": 0.0787,
+      "step": 5985
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8600870754210477e-05,
+      "loss": 0.0802,
+      "step": 5986
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8599227044553276e-05,
+      "loss": 0.0814,
+      "step": 5987
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.859758244265398e-05,
+      "loss": 0.078,
+      "step": 5988
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8595936948683234e-05,
+      "loss": 0.1007,
+      "step": 5989
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8594290562811762e-05,
+      "loss": 0.084,
+      "step": 5990
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.85926432852104e-05,
+      "loss": 0.0846,
+      "step": 5991
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.859099511605006e-05,
+      "loss": 0.0773,
+      "step": 5992
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8589346055501757e-05,
+      "loss": 0.0656,
+      "step": 5993
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.85876961037366e-05,
+      "loss": 0.0909,
+      "step": 5994
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8586045260925773e-05,
+      "loss": 0.0813,
+      "step": 5995
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8584393527240576e-05,
+      "loss": 0.0843,
+      "step": 5996
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8582740902852385e-05,
+      "loss": 0.0687,
+      "step": 5997
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8581087387932676e-05,
+      "loss": 0.0875,
+      "step": 5998
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8579432982653013e-05,
+      "loss": 0.0682,
+      "step": 5999
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8577777687185054e-05,
+      "loss": 0.0837,
+      "step": 6000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8576121501700553e-05,
+      "loss": 0.0827,
+      "step": 6001
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.857446442637135e-05,
+      "loss": 0.0768,
+      "step": 6002
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8572806461369383e-05,
+      "loss": 0.0777,
+      "step": 6003
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8571147606866677e-05,
+      "loss": 0.0875,
+      "step": 6004
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8569487863035355e-05,
+      "loss": 0.0784,
+      "step": 6005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.856782723004763e-05,
+      "loss": 0.0861,
+      "step": 6006
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.85661657080758e-05,
+      "loss": 0.0947,
+      "step": 6007
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8564503297292267e-05,
+      "loss": 0.0774,
+      "step": 6008
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8562839997869523e-05,
+      "loss": 0.0911,
+      "step": 6009
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8561175809980144e-05,
+      "loss": 0.0973,
+      "step": 6010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8559510733796807e-05,
+      "loss": 0.0651,
+      "step": 6011
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8557844769492272e-05,
+      "loss": 0.0759,
+      "step": 6012
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8556177917239406e-05,
+      "loss": 0.0732,
+      "step": 6013
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8554510177211155e-05,
+      "loss": 0.0773,
+      "step": 6014
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8552841549580555e-05,
+      "loss": 0.0778,
+      "step": 6015
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8551172034520746e-05,
+      "loss": 0.0748,
+      "step": 6016
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8549501632204953e-05,
+      "loss": 0.0712,
+      "step": 6017
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8547830342806493e-05,
+      "loss": 0.085,
+      "step": 6018
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8546158166498783e-05,
+      "loss": 0.108,
+      "step": 6019
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8544485103455317e-05,
+      "loss": 0.0795,
+      "step": 6020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8542811153849692e-05,
+      "loss": 0.0713,
+      "step": 6021
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8541136317855598e-05,
+      "loss": 0.08,
+      "step": 6022
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8539460595646804e-05,
+      "loss": 0.0768,
+      "step": 6023
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8537783987397193e-05,
+      "loss": 0.0867,
+      "step": 6024
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.853610649328072e-05,
+      "loss": 0.0846,
+      "step": 6025
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8534428113471437e-05,
+      "loss": 0.0695,
+      "step": 6026
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8532748848143493e-05,
+      "loss": 0.0798,
+      "step": 6027
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8531068697471125e-05,
+      "loss": 0.088,
+      "step": 6028
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8529387661628667e-05,
+      "loss": 0.0875,
+      "step": 6029
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8527705740790532e-05,
+      "loss": 0.0713,
+      "step": 6030
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8526022935131244e-05,
+      "loss": 0.0767,
+      "step": 6031
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.85243392448254e-05,
+      "loss": 0.0786,
+      "step": 6032
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8522654670047702e-05,
+      "loss": 0.0823,
+      "step": 6033
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8520969210972932e-05,
+      "loss": 0.0754,
+      "step": 6034
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.851928286777598e-05,
+      "loss": 0.0762,
+      "step": 6035
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.851759564063181e-05,
+      "loss": 0.0878,
+      "step": 6036
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8515907529715492e-05,
+      "loss": 0.0779,
+      "step": 6037
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8514218535202175e-05,
+      "loss": 0.0719,
+      "step": 6038
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8512528657267114e-05,
+      "loss": 0.0705,
+      "step": 6039
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8510837896085642e-05,
+      "loss": 0.0712,
+      "step": 6040
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8509146251833193e-05,
+      "loss": 0.0888,
+      "step": 6041
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.850745372468529e-05,
+      "loss": 0.0667,
+      "step": 6042
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8505760314817544e-05,
+      "loss": 0.064,
+      "step": 6043
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8504066022405663e-05,
+      "loss": 0.0668,
+      "step": 6044
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8502370847625442e-05,
+      "loss": 0.0837,
+      "step": 6045
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.850067479065277e-05,
+      "loss": 0.0668,
+      "step": 6046
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.849897785166363e-05,
+      "loss": 0.0781,
+      "step": 6047
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.849728003083409e-05,
+      "loss": 0.077,
+      "step": 6048
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8495581328340315e-05,
+      "loss": 0.0853,
+      "step": 6049
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8493881744358558e-05,
+      "loss": 0.0911,
+      "step": 6050
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8492181279065166e-05,
+      "loss": 0.0944,
+      "step": 6051
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.849047993263658e-05,
+      "loss": 0.0748,
+      "step": 6052
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8488777705249324e-05,
+      "loss": 0.093,
+      "step": 6053
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8487074597080023e-05,
+      "loss": 0.0661,
+      "step": 6054
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8485370608305384e-05,
+      "loss": 0.074,
+      "step": 6055
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.848366573910221e-05,
+      "loss": 0.0876,
+      "step": 6056
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.84819599896474e-05,
+      "loss": 0.0779,
+      "step": 6057
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.848025336011794e-05,
+      "loss": 0.074,
+      "step": 6058
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8478545850690902e-05,
+      "loss": 0.0928,
+      "step": 6059
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.847683746154346e-05,
+      "loss": 0.0795,
+      "step": 6060
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.847512819285287e-05,
+      "loss": 0.075,
+      "step": 6061
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8473418044796484e-05,
+      "loss": 0.0772,
+      "step": 6062
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8471707017551743e-05,
+      "loss": 0.0654,
+      "step": 6063
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8469995111296183e-05,
+      "loss": 0.079,
+      "step": 6064
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8468282326207426e-05,
+      "loss": 0.0706,
+      "step": 6065
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.846656866246319e-05,
+      "loss": 0.0586,
+      "step": 6066
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.846485412024128e-05,
+      "loss": 0.0851,
+      "step": 6067
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.84631386997196e-05,
+      "loss": 0.0699,
+      "step": 6068
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.846142240107613e-05,
+      "loss": 0.0853,
+      "step": 6069
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8459705224488958e-05,
+      "loss": 0.0906,
+      "step": 6070
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8457987170136254e-05,
+      "loss": 0.0664,
+      "step": 6071
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8456268238196277e-05,
+      "loss": 0.0785,
+      "step": 6072
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8454548428847383e-05,
+      "loss": 0.0674,
+      "step": 6073
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8452827742268017e-05,
+      "loss": 0.0825,
+      "step": 6074
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8451106178636714e-05,
+      "loss": 0.0927,
+      "step": 6075
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.84493837381321e-05,
+      "loss": 0.0716,
+      "step": 6076
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8447660420932895e-05,
+      "loss": 0.0859,
+      "step": 6077
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8445936227217904e-05,
+      "loss": 0.0623,
+      "step": 6078
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.844421115716603e-05,
+      "loss": 0.072,
+      "step": 6079
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8442485210956263e-05,
+      "loss": 0.087,
+      "step": 6080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8440758388767685e-05,
+      "loss": 0.0852,
+      "step": 6081
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8439030690779468e-05,
+      "loss": 0.0809,
+      "step": 6082
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.843730211717087e-05,
+      "loss": 0.0914,
+      "step": 6083
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8435572668121252e-05,
+      "loss": 0.0768,
+      "step": 6084
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8433842343810058e-05,
+      "loss": 0.0884,
+      "step": 6085
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.843211114441682e-05,
+      "loss": 0.072,
+      "step": 6086
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8430379070121163e-05,
+      "loss": 0.0643,
+      "step": 6087
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8428646121102815e-05,
+      "loss": 0.0856,
+      "step": 6088
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8426912297541574e-05,
+      "loss": 0.0677,
+      "step": 6089
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8425177599617342e-05,
+      "loss": 0.0852,
+      "step": 6090
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8423442027510104e-05,
+      "loss": 0.0615,
+      "step": 6091
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8421705581399946e-05,
+      "loss": 0.0845,
+      "step": 6092
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8419968261467042e-05,
+      "loss": 0.0604,
+      "step": 6093
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8418230067891644e-05,
+      "loss": 0.0844,
+      "step": 6094
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.841649100085411e-05,
+      "loss": 0.1025,
+      "step": 6095
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8414751060534878e-05,
+      "loss": 0.0673,
+      "step": 6096
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8413010247114492e-05,
+      "loss": 0.0709,
+      "step": 6097
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8411268560773562e-05,
+      "loss": 0.0797,
+      "step": 6098
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8409526001692817e-05,
+      "loss": 0.069,
+      "step": 6099
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.840778257005305e-05,
+      "loss": 0.0822,
+      "step": 6100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8406038266035163e-05,
+      "loss": 0.0781,
+      "step": 6101
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.840429308982014e-05,
+      "loss": 0.0867,
+      "step": 6102
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8402547041589057e-05,
+      "loss": 0.0817,
+      "step": 6103
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8400800121523087e-05,
+      "loss": 0.0671,
+      "step": 6104
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.839905232980348e-05,
+      "loss": 0.0666,
+      "step": 6105
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8397303666611588e-05,
+      "loss": 0.0793,
+      "step": 6106
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8395554132128854e-05,
+      "loss": 0.0688,
+      "step": 6107
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8393803726536793e-05,
+      "loss": 0.0647,
+      "step": 6108
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8392052450017036e-05,
+      "loss": 0.0859,
+      "step": 6109
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8390300302751292e-05,
+      "loss": 0.09,
+      "step": 6110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.838854728492136e-05,
+      "loss": 0.0925,
+      "step": 6111
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8386793396709123e-05,
+      "loss": 0.0738,
+      "step": 6112
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8385038638296577e-05,
+      "loss": 0.0851,
+      "step": 6113
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8383283009865773e-05,
+      "loss": 0.0734,
+      "step": 6114
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.838152651159889e-05,
+      "loss": 0.0682,
+      "step": 6115
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.837976914367817e-05,
+      "loss": 0.0722,
+      "step": 6116
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8378010906285958e-05,
+      "loss": 0.0785,
+      "step": 6117
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8376251799604684e-05,
+      "loss": 0.0558,
+      "step": 6118
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8374491823816872e-05,
+      "loss": 0.0883,
+      "step": 6119
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8372730979105133e-05,
+      "loss": 0.0732,
+      "step": 6120
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.837096926565217e-05,
+      "loss": 0.076,
+      "step": 6121
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8369206683640776e-05,
+      "loss": 0.0773,
+      "step": 6122
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8367443233253833e-05,
+      "loss": 0.0828,
+      "step": 6123
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.836567891467431e-05,
+      "loss": 0.071,
+      "step": 6124
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8363913728085277e-05,
+      "loss": 0.0753,
+      "step": 6125
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8362147673669884e-05,
+      "loss": 0.0677,
+      "step": 6126
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8360380751611375e-05,
+      "loss": 0.0725,
+      "step": 6127
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.835861296209308e-05,
+      "loss": 0.0787,
+      "step": 6128
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8356844305298424e-05,
+      "loss": 0.0767,
+      "step": 6129
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8355074781410918e-05,
+      "loss": 0.077,
+      "step": 6130
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.835330439061417e-05,
+      "loss": 0.0815,
+      "step": 6131
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8351533133091867e-05,
+      "loss": 0.0787,
+      "step": 6132
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8349761009027794e-05,
+      "loss": 0.0719,
+      "step": 6133
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8347988018605826e-05,
+      "loss": 0.0803,
+      "step": 6134
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8346214162009924e-05,
+      "loss": 0.0746,
+      "step": 6135
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8344439439424142e-05,
+      "loss": 0.088,
+      "step": 6136
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.834266385103262e-05,
+      "loss": 0.0935,
+      "step": 6137
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.834088739701959e-05,
+      "loss": 0.0657,
+      "step": 6138
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8339110077569376e-05,
+      "loss": 0.0935,
+      "step": 6139
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8337331892866384e-05,
+      "loss": 0.1019,
+      "step": 6140
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8335552843095128e-05,
+      "loss": 0.0778,
+      "step": 6141
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8333772928440187e-05,
+      "loss": 0.0878,
+      "step": 6142
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.833199214908625e-05,
+      "loss": 0.082,
+      "step": 6143
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8330210505218084e-05,
+      "loss": 0.0806,
+      "step": 6144
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.832842799702055e-05,
+      "loss": 0.0867,
+      "step": 6145
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.83266446246786e-05,
+      "loss": 0.0761,
+      "step": 6146
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.832486038837727e-05,
+      "loss": 0.0762,
+      "step": 6147
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8323075288301693e-05,
+      "loss": 0.0776,
+      "step": 6148
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8321289324637087e-05,
+      "loss": 0.0823,
+      "step": 6149
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.831950249756876e-05,
+      "loss": 0.0632,
+      "step": 6150
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8317714807282115e-05,
+      "loss": 0.0812,
+      "step": 6151
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8315926253962636e-05,
+      "loss": 0.0753,
+      "step": 6152
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8314136837795894e-05,
+      "loss": 0.0679,
+      "step": 6153
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.831234655896757e-05,
+      "loss": 0.0797,
+      "step": 6154
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.831055541766341e-05,
+      "loss": 0.0757,
+      "step": 6155
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.830876341406926e-05,
+      "loss": 0.085,
+      "step": 6156
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8306970548371062e-05,
+      "loss": 0.0714,
+      "step": 6157
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8305176820754838e-05,
+      "loss": 0.0836,
+      "step": 6158
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.83033822314067e-05,
+      "loss": 0.0825,
+      "step": 6159
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.830158678051285e-05,
+      "loss": 0.0778,
+      "step": 6160
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.829979046825959e-05,
+      "loss": 0.0903,
+      "step": 6161
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8297993294833294e-05,
+      "loss": 0.0605,
+      "step": 6162
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8296195260420438e-05,
+      "loss": 0.0715,
+      "step": 6163
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8294396365207584e-05,
+      "loss": 0.0701,
+      "step": 6164
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8292596609381378e-05,
+      "loss": 0.0709,
+      "step": 6165
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.829079599312856e-05,
+      "loss": 0.0864,
+      "step": 6166
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8288994516635963e-05,
+      "loss": 0.0845,
+      "step": 6167
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8287192180090505e-05,
+      "loss": 0.0745,
+      "step": 6168
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8285388983679192e-05,
+      "loss": 0.0755,
+      "step": 6169
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8283584927589123e-05,
+      "loss": 0.0676,
+      "step": 6170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.828178001200748e-05,
+      "loss": 0.0781,
+      "step": 6171
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8279974237121537e-05,
+      "loss": 0.0804,
+      "step": 6172
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8278167603118667e-05,
+      "loss": 0.0711,
+      "step": 6173
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8276360110186315e-05,
+      "loss": 0.0721,
+      "step": 6174
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8274551758512026e-05,
+      "loss": 0.0923,
+      "step": 6175
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8272742548283434e-05,
+      "loss": 0.0721,
+      "step": 6176
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8270932479688255e-05,
+      "loss": 0.092,
+      "step": 6177
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8269121552914307e-05,
+      "loss": 0.0812,
+      "step": 6178
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8267309768149482e-05,
+      "loss": 0.0842,
+      "step": 6179
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.826549712558177e-05,
+      "loss": 0.0911,
+      "step": 6180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8263683625399244e-05,
+      "loss": 0.074,
+      "step": 6181
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8261869267790077e-05,
+      "loss": 0.0829,
+      "step": 6182
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.826005405294252e-05,
+      "loss": 0.0707,
+      "step": 6183
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8258237981044915e-05,
+      "loss": 0.0858,
+      "step": 6184
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8256421052285698e-05,
+      "loss": 0.075,
+      "step": 6185
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.825460326685339e-05,
+      "loss": 0.0793,
+      "step": 6186
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.82527846249366e-05,
+      "loss": 0.0709,
+      "step": 6187
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.825096512672403e-05,
+      "loss": 0.0814,
+      "step": 6188
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8249144772404467e-05,
+      "loss": 0.0776,
+      "step": 6189
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8247323562166785e-05,
+      "loss": 0.0662,
+      "step": 6190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8245501496199954e-05,
+      "loss": 0.0809,
+      "step": 6191
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8243678574693026e-05,
+      "loss": 0.0631,
+      "step": 6192
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.824185479783515e-05,
+      "loss": 0.0756,
+      "step": 6193
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8240030165815546e-05,
+      "loss": 0.0858,
+      "step": 6194
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.823820467882355e-05,
+      "loss": 0.0747,
+      "step": 6195
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8236378337048562e-05,
+      "loss": 0.077,
+      "step": 6196
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8234551140680083e-05,
+      "loss": 0.0628,
+      "step": 6197
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8232723089907694e-05,
+      "loss": 0.0768,
+      "step": 6198
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.823089418492108e-05,
+      "loss": 0.0731,
+      "step": 6199
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8229064425910005e-05,
+      "loss": 0.0761,
+      "step": 6200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8227233813064318e-05,
+      "loss": 0.084,
+      "step": 6201
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8225402346573958e-05,
+      "loss": 0.0798,
+      "step": 6202
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8223570026628958e-05,
+      "loss": 0.0654,
+      "step": 6203
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8221736853419437e-05,
+      "loss": 0.0754,
+      "step": 6204
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.82199028271356e-05,
+      "loss": 0.0662,
+      "step": 6205
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.821806794796775e-05,
+      "loss": 0.0806,
+      "step": 6206
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.821623221610626e-05,
+      "loss": 0.071,
+      "step": 6207
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.821439563174161e-05,
+      "loss": 0.075,
+      "step": 6208
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.821255819506436e-05,
+      "loss": 0.0709,
+      "step": 6209
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8210719906265155e-05,
+      "loss": 0.079,
+      "step": 6210
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.820888076553474e-05,
+      "loss": 0.0704,
+      "step": 6211
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8207040773063938e-05,
+      "loss": 0.08,
+      "step": 6212
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8205199929043664e-05,
+      "loss": 0.0918,
+      "step": 6213
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8203358233664915e-05,
+      "loss": 0.0887,
+      "step": 6214
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8201515687118795e-05,
+      "loss": 0.0891,
+      "step": 6215
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8199672289596473e-05,
+      "loss": 0.0677,
+      "step": 6216
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.819782804128922e-05,
+      "loss": 0.0876,
+      "step": 6217
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8195982942388397e-05,
+      "loss": 0.0759,
+      "step": 6218
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.819413699308544e-05,
+      "loss": 0.0677,
+      "step": 6219
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.819229019357189e-05,
+      "loss": 0.0895,
+      "step": 6220
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.819044254403936e-05,
+      "loss": 0.0716,
+      "step": 6221
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8188594044679566e-05,
+      "loss": 0.0677,
+      "step": 6222
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.81867446956843e-05,
+      "loss": 0.0791,
+      "step": 6223
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8184894497245457e-05,
+      "loss": 0.0945,
+      "step": 6224
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8183043449554998e-05,
+      "loss": 0.0959,
+      "step": 6225
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.818119155280499e-05,
+      "loss": 0.0652,
+      "step": 6226
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8179338807187587e-05,
+      "loss": 0.0707,
+      "step": 6227
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8177485212895022e-05,
+      "loss": 0.0646,
+      "step": 6228
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.817563077011962e-05,
+      "loss": 0.0799,
+      "step": 6229
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.81737754790538e-05,
+      "loss": 0.0673,
+      "step": 6230
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.817191933989006e-05,
+      "loss": 0.0758,
+      "step": 6231
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8170062352820993e-05,
+      "loss": 0.0778,
+      "step": 6232
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8168204518039273e-05,
+      "loss": 0.0676,
+      "step": 6233
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8166345835737674e-05,
+      "loss": 0.079,
+      "step": 6234
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.816448630610904e-05,
+      "loss": 0.079,
+      "step": 6235
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8162625929346314e-05,
+      "loss": 0.0697,
+      "step": 6236
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8160764705642534e-05,
+      "loss": 0.0737,
+      "step": 6237
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8158902635190812e-05,
+      "loss": 0.0713,
+      "step": 6238
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.815703971818435e-05,
+      "loss": 0.0876,
+      "step": 6239
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8155175954816453e-05,
+      "loss": 0.0931,
+      "step": 6240
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.815331134528049e-05,
+      "loss": 0.0758,
+      "step": 6241
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8151445889769934e-05,
+      "loss": 0.0702,
+      "step": 6242
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8149579588478345e-05,
+      "loss": 0.0645,
+      "step": 6243
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.814771244159936e-05,
+      "loss": 0.0646,
+      "step": 6244
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8145844449326718e-05,
+      "loss": 0.0742,
+      "step": 6245
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8143975611854235e-05,
+      "loss": 0.0833,
+      "step": 6246
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8142105929375823e-05,
+      "loss": 0.0787,
+      "step": 6247
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.814023540208547e-05,
+      "loss": 0.0789,
+      "step": 6248
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.813836403017727e-05,
+      "loss": 0.0734,
+      "step": 6249
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.813649181384538e-05,
+      "loss": 0.0712,
+      "step": 6250
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.813461875328407e-05,
+      "loss": 0.0705,
+      "step": 6251
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8132744848687677e-05,
+      "loss": 0.067,
+      "step": 6252
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8130870100250643e-05,
+      "loss": 0.084,
+      "step": 6253
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.812899450816748e-05,
+      "loss": 0.0732,
+      "step": 6254
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8127118072632805e-05,
+      "loss": 0.0925,
+      "step": 6255
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8125240793841304e-05,
+      "loss": 0.0707,
+      "step": 6256
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.812336267198777e-05,
+      "loss": 0.0693,
+      "step": 6257
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.812148370726707e-05,
+      "loss": 0.0865,
+      "step": 6258
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8119603899874163e-05,
+      "loss": 0.0922,
+      "step": 6259
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.811772325000409e-05,
+      "loss": 0.0653,
+      "step": 6260
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.811584175785199e-05,
+      "loss": 0.0986,
+      "step": 6261
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8113959423613084e-05,
+      "loss": 0.0778,
+      "step": 6262
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8112076247482678e-05,
+      "loss": 0.1035,
+      "step": 6263
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8110192229656168e-05,
+      "loss": 0.0679,
+      "step": 6264
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8108307370329032e-05,
+      "loss": 0.0804,
+      "step": 6265
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8106421669696846e-05,
+      "loss": 0.0734,
+      "step": 6266
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8104535127955265e-05,
+      "loss": 0.0831,
+      "step": 6267
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.810264774530004e-05,
+      "loss": 0.076,
+      "step": 6268
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8100759521926987e-05,
+      "loss": 0.0842,
+      "step": 6269
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.809887045803204e-05,
+      "loss": 0.0752,
+      "step": 6270
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.80969805538112e-05,
+      "loss": 0.0688,
+      "step": 6271
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.809508980946056e-05,
+      "loss": 0.0916,
+      "step": 6272
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8093198225176302e-05,
+      "loss": 0.0797,
+      "step": 6273
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.809130580115469e-05,
+      "loss": 0.0759,
+      "step": 6274
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8089412537592085e-05,
+      "loss": 0.0734,
+      "step": 6275
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8087518434684927e-05,
+      "loss": 0.0701,
+      "step": 6276
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.808562349262974e-05,
+      "loss": 0.0778,
+      "step": 6277
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8083727711623145e-05,
+      "loss": 0.0773,
+      "step": 6278
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8081831091861843e-05,
+      "loss": 0.0897,
+      "step": 6279
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.807993363354263e-05,
+      "loss": 0.0814,
+      "step": 6280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8078035336862375e-05,
+      "loss": 0.0677,
+      "step": 6281
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8076136202018045e-05,
+      "loss": 0.0756,
+      "step": 6282
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8074236229206694e-05,
+      "loss": 0.0861,
+      "step": 6283
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.807233541862546e-05,
+      "loss": 0.0758,
+      "step": 6284
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8070433770471566e-05,
+      "loss": 0.0823,
+      "step": 6285
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8068531284942324e-05,
+      "loss": 0.0732,
+      "step": 6286
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.806662796223513e-05,
+      "loss": 0.0819,
+      "step": 6287
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8064723802547478e-05,
+      "loss": 0.0707,
+      "step": 6288
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8062818806076934e-05,
+      "loss": 0.0701,
+      "step": 6289
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8060912973021158e-05,
+      "loss": 0.0653,
+      "step": 6290
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.80590063035779e-05,
+      "loss": 0.0759,
+      "step": 6291
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8057098797944987e-05,
+      "loss": 0.076,
+      "step": 6292
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8055190456320344e-05,
+      "loss": 0.071,
+      "step": 6293
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8053281278901976e-05,
+      "loss": 0.0712,
+      "step": 6294
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.805137126588797e-05,
+      "loss": 0.0736,
+      "step": 6295
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.804946041747652e-05,
+      "loss": 0.0712,
+      "step": 6296
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8047548733865877e-05,
+      "loss": 0.0784,
+      "step": 6297
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8045636215254407e-05,
+      "loss": 0.0749,
+      "step": 6298
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.804372286184054e-05,
+      "loss": 0.0757,
+      "step": 6299
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8041808673822806e-05,
+      "loss": 0.0898,
+      "step": 6300
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8039893651399823e-05,
+      "loss": 0.0627,
+      "step": 6301
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8037977794770285e-05,
+      "loss": 0.0774,
+      "step": 6302
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.803606110413298e-05,
+      "loss": 0.0951,
+      "step": 6303
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.803414357968678e-05,
+      "loss": 0.0858,
+      "step": 6304
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8032225221630645e-05,
+      "loss": 0.0824,
+      "step": 6305
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.803030603016362e-05,
+      "loss": 0.0687,
+      "step": 6306
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8028386005484837e-05,
+      "loss": 0.0789,
+      "step": 6307
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8026465147793516e-05,
+      "loss": 0.0655,
+      "step": 6308
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8024543457288962e-05,
+      "loss": 0.0689,
+      "step": 6309
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8022620934170568e-05,
+      "loss": 0.0683,
+      "step": 6310
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.802069757863781e-05,
+      "loss": 0.081,
+      "step": 6311
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.801877339089025e-05,
+      "loss": 0.0734,
+      "step": 6312
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.801684837112754e-05,
+      "loss": 0.0953,
+      "step": 6313
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8014922519549423e-05,
+      "loss": 0.0713,
+      "step": 6314
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8012995836355712e-05,
+      "loss": 0.0859,
+      "step": 6315
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.801106832174633e-05,
+      "loss": 0.0749,
+      "step": 6316
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.800913997592126e-05,
+      "loss": 0.0771,
+      "step": 6317
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8007210799080586e-05,
+      "loss": 0.0896,
+      "step": 6318
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8005280791424483e-05,
+      "loss": 0.0891,
+      "step": 6319
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.80033499531532e-05,
+      "loss": 0.067,
+      "step": 6320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8001418284467077e-05,
+      "loss": 0.0714,
+      "step": 6321
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.799948578556655e-05,
+      "loss": 0.0643,
+      "step": 6322
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.799755245665212e-05,
+      "loss": 0.0717,
+      "step": 6323
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.799561829792439e-05,
+      "loss": 0.082,
+      "step": 6324
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.799368330958405e-05,
+      "loss": 0.0724,
+      "step": 6325
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.799174749183187e-05,
+      "loss": 0.0746,
+      "step": 6326
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.79898108448687e-05,
+      "loss": 0.0811,
+      "step": 6327
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7987873368895494e-05,
+      "loss": 0.0879,
+      "step": 6328
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.798593506411327e-05,
+      "loss": 0.0873,
+      "step": 6329
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7983995930723156e-05,
+      "loss": 0.0742,
+      "step": 6330
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7982055968926344e-05,
+      "loss": 0.0792,
+      "step": 6331
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7980115178924125e-05,
+      "loss": 0.0874,
+      "step": 6332
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7978173560917872e-05,
+      "loss": 0.0825,
+      "step": 6333
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.797623111510904e-05,
+      "loss": 0.0978,
+      "step": 6334
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.797428784169918e-05,
+      "loss": 0.071,
+      "step": 6335
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7972343740889922e-05,
+      "loss": 0.0687,
+      "step": 6336
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7970398812882982e-05,
+      "loss": 0.0784,
+      "step": 6337
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.796845305788016e-05,
+      "loss": 0.084,
+      "step": 6338
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7966506476083353e-05,
+      "loss": 0.0648,
+      "step": 6339
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.796455906769452e-05,
+      "loss": 0.0842,
+      "step": 6340
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7962610832915738e-05,
+      "loss": 0.097,
+      "step": 6341
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7960661771949137e-05,
+      "loss": 0.0784,
+      "step": 6342
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.795871188499696e-05,
+      "loss": 0.0883,
+      "step": 6343
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7956761172261518e-05,
+      "loss": 0.0724,
+      "step": 6344
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7954809633945222e-05,
+      "loss": 0.0608,
+      "step": 6345
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.795285727025055e-05,
+      "loss": 0.0774,
+      "step": 6346
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7950904081380082e-05,
+      "loss": 0.0726,
+      "step": 6347
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7948950067536475e-05,
+      "loss": 0.0978,
+      "step": 6348
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7946995228922474e-05,
+      "loss": 0.0844,
+      "step": 6349
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7945039565740914e-05,
+      "loss": 0.0829,
+      "step": 6350
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7943083078194712e-05,
+      "loss": 0.0813,
+      "step": 6351
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7941125766486865e-05,
+      "loss": 0.0717,
+      "step": 6352
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7939167630820465e-05,
+      "loss": 0.0779,
+      "step": 6353
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7937208671398677e-05,
+      "loss": 0.0793,
+      "step": 6354
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.793524888842477e-05,
+      "loss": 0.0598,
+      "step": 6355
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7933288282102084e-05,
+      "loss": 0.0832,
+      "step": 6356
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7931326852634044e-05,
+      "loss": 0.063,
+      "step": 6357
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.792936460022417e-05,
+      "loss": 0.0639,
+      "step": 6358
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7927401525076066e-05,
+      "loss": 0.0785,
+      "step": 6359
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.792543762739341e-05,
+      "loss": 0.0918,
+      "step": 6360
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7923472907379968e-05,
+      "loss": 0.0699,
+      "step": 6361
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.792150736523961e-05,
+      "loss": 0.0721,
+      "step": 6362
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.791954100117627e-05,
+      "loss": 0.0844,
+      "step": 6363
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7917573815393975e-05,
+      "loss": 0.0724,
+      "step": 6364
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.791560580809684e-05,
+      "loss": 0.0548,
+      "step": 6365
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.791363697948906e-05,
+      "loss": 0.069,
+      "step": 6366
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7911667329774914e-05,
+      "loss": 0.0797,
+      "step": 6367
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7909696859158778e-05,
+      "loss": 0.0734,
+      "step": 6368
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7907725567845094e-05,
+      "loss": 0.0796,
+      "step": 6369
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.790575345603841e-05,
+      "loss": 0.0853,
+      "step": 6370
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7903780523943344e-05,
+      "loss": 0.0784,
+      "step": 6371
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.790180677176461e-05,
+      "loss": 0.0668,
+      "step": 6372
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7899832199706993e-05,
+      "loss": 0.0813,
+      "step": 6373
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7897856807975377e-05,
+      "loss": 0.0793,
+      "step": 6374
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7895880596774723e-05,
+      "loss": 0.0685,
+      "step": 6375
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.789390356631008e-05,
+      "loss": 0.084,
+      "step": 6376
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7891925716786584e-05,
+      "loss": 0.0822,
+      "step": 6377
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.788994704840945e-05,
+      "loss": 0.0806,
+      "step": 6378
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7887967561383986e-05,
+      "loss": 0.0762,
+      "step": 6379
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7885987255915574e-05,
+      "loss": 0.0681,
+      "step": 6380
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7884006132209693e-05,
+      "loss": 0.0649,
+      "step": 6381
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.78820241904719e-05,
+      "loss": 0.0685,
+      "step": 6382
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7880041430907836e-05,
+      "loss": 0.0793,
+      "step": 6383
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.787805785372323e-05,
+      "loss": 0.0701,
+      "step": 6384
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7876073459123895e-05,
+      "loss": 0.0841,
+      "step": 6385
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7874088247315727e-05,
+      "loss": 0.0784,
+      "step": 6386
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7872102218504715e-05,
+      "loss": 0.0786,
+      "step": 6387
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7870115372896915e-05,
+      "loss": 0.0734,
+      "step": 6388
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7868127710698488e-05,
+      "loss": 0.0648,
+      "step": 6389
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.786613923211567e-05,
+      "loss": 0.0753,
+      "step": 6390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.786414993735478e-05,
+      "loss": 0.0655,
+      "step": 6391
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.786215982662222e-05,
+      "loss": 0.0732,
+      "step": 6392
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.786016890012449e-05,
+      "loss": 0.0696,
+      "step": 6393
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7858177158068154e-05,
+      "loss": 0.0651,
+      "step": 6394
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7856184600659886e-05,
+      "loss": 0.0817,
+      "step": 6395
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7854191228106414e-05,
+      "loss": 0.0804,
+      "step": 6396
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7852197040614583e-05,
+      "loss": 0.0858,
+      "step": 6397
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7850202038391297e-05,
+      "loss": 0.0713,
+      "step": 6398
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.784820622164356e-05,
+      "loss": 0.0762,
+      "step": 6399
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.784620959057845e-05,
+      "loss": 0.0712,
+      "step": 6400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7844212145403133e-05,
+      "loss": 0.0802,
+      "step": 6401
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.784221388632487e-05,
+      "loss": 0.08,
+      "step": 6402
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7840214813550986e-05,
+      "loss": 0.0714,
+      "step": 6403
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.783821492728891e-05,
+      "loss": 0.079,
+      "step": 6404
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.7836214227746138e-05,
+      "loss": 0.0701,
+      "step": 6405
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.783421271513027e-05,
+      "loss": 0.0925,
+      "step": 6406
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.783221038964897e-05,
+      "loss": 0.0787,
+      "step": 6407
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7830207251510008e-05,
+      "loss": 0.0765,
+      "step": 6408
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7828203300921216e-05,
+      "loss": 0.0842,
+      "step": 6409
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.782619853809052e-05,
+      "loss": 0.0839,
+      "step": 6410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7824192963225938e-05,
+      "loss": 0.0731,
+      "step": 6411
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7822186576535566e-05,
+      "loss": 0.0735,
+      "step": 6412
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7820179378227572e-05,
+      "loss": 0.0654,
+      "step": 6413
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7818171368510234e-05,
+      "loss": 0.0792,
+      "step": 6414
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.781616254759189e-05,
+      "loss": 0.085,
+      "step": 6415
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7814152915680975e-05,
+      "loss": 0.0828,
+      "step": 6416
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7812142472986007e-05,
+      "loss": 0.0939,
+      "step": 6417
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7810131219715585e-05,
+      "loss": 0.0629,
+      "step": 6418
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7808119156078392e-05,
+      "loss": 0.0717,
+      "step": 6419
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.78061062822832e-05,
+      "loss": 0.081,
+      "step": 6420
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7804092598538857e-05,
+      "loss": 0.0756,
+      "step": 6421
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7802078105054305e-05,
+      "loss": 0.0731,
+      "step": 6422
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.780006280203856e-05,
+      "loss": 0.0786,
+      "step": 6423
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7798046689700728e-05,
+      "loss": 0.0857,
+      "step": 6424
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.779602976825e-05,
+      "loss": 0.0729,
+      "step": 6425
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.779401203789564e-05,
+      "loss": 0.062,
+      "step": 6426
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7791993498847016e-05,
+      "loss": 0.0858,
+      "step": 6427
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7789974151313566e-05,
+      "loss": 0.0757,
+      "step": 6428
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7787953995504807e-05,
+      "loss": 0.0773,
+      "step": 6429
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.778593303163035e-05,
+      "loss": 0.0717,
+      "step": 6430
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7783911259899894e-05,
+      "loss": 0.0721,
+      "step": 6431
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7781888680523208e-05,
+      "loss": 0.0851,
+      "step": 6432
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.777986529371015e-05,
+      "loss": 0.0698,
+      "step": 6433
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7777841099670672e-05,
+      "loss": 0.08,
+      "step": 6434
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.777581609861479e-05,
+      "loss": 0.092,
+      "step": 6435
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7773790290752626e-05,
+      "loss": 0.0768,
+      "step": 6436
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7771763676294368e-05,
+      "loss": 0.0854,
+      "step": 6437
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7769736255450292e-05,
+      "loss": 0.0726,
+      "step": 6438
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7767708028430767e-05,
+      "loss": 0.0826,
+      "step": 6439
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7765678995446233e-05,
+      "loss": 0.0807,
+      "step": 6440
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7763649156707222e-05,
+      "loss": 0.0789,
+      "step": 6441
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7761618512424347e-05,
+      "loss": 0.071,
+      "step": 6442
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7759587062808302e-05,
+      "loss": 0.0812,
+      "step": 6443
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7757554808069867e-05,
+      "loss": 0.0918,
+      "step": 6444
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7755521748419912e-05,
+      "loss": 0.066,
+      "step": 6445
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7753487884069375e-05,
+      "loss": 0.0685,
+      "step": 6446
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7751453215229292e-05,
+      "loss": 0.0547,
+      "step": 6447
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7749417742110772e-05,
+      "loss": 0.0715,
+      "step": 6448
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.774738146492502e-05,
+      "loss": 0.0677,
+      "step": 6449
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7745344383883312e-05,
+      "loss": 0.061,
+      "step": 6450
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7743306499197014e-05,
+      "loss": 0.0705,
+      "step": 6451
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7741267811077573e-05,
+      "loss": 0.0706,
+      "step": 6452
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7739228319736517e-05,
+      "loss": 0.0724,
+      "step": 6453
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7737188025385466e-05,
+      "loss": 0.0774,
+      "step": 6454
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.773514692823611e-05,
+      "loss": 0.1021,
+      "step": 6455
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.773310502850024e-05,
+      "loss": 0.0868,
+      "step": 6456
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7731062326389716e-05,
+      "loss": 0.0884,
+      "step": 6457
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7729018822116482e-05,
+      "loss": 0.0812,
+      "step": 6458
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7726974515892573e-05,
+      "loss": 0.0885,
+      "step": 6459
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.77249294079301e-05,
+      "loss": 0.079,
+      "step": 6460
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.772288349844126e-05,
+      "loss": 0.0688,
+      "step": 6461
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.772083678763834e-05,
+      "loss": 0.0906,
+      "step": 6462
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7718789275733694e-05,
+      "loss": 0.0718,
+      "step": 6463
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7716740962939772e-05,
+      "loss": 0.0763,
+      "step": 6464
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.771469184946911e-05,
+      "loss": 0.0649,
+      "step": 6465
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.771264193553431e-05,
+      "loss": 0.079,
+      "step": 6466
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7710591221348074e-05,
+      "loss": 0.0918,
+      "step": 6467
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.770853970712318e-05,
+      "loss": 0.0654,
+      "step": 6468
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7706487393072492e-05,
+      "loss": 0.0814,
+      "step": 6469
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.770443427940895e-05,
+      "loss": 0.094,
+      "step": 6470
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7702380366345585e-05,
+      "loss": 0.0779,
+      "step": 6471
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.770032565409551e-05,
+      "loss": 0.0703,
+      "step": 6472
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7698270142871914e-05,
+      "loss": 0.0712,
+      "step": 6473
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7696213832888074e-05,
+      "loss": 0.082,
+      "step": 6474
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7694156724357352e-05,
+      "loss": 0.0649,
+      "step": 6475
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7692098817493192e-05,
+      "loss": 0.0632,
+      "step": 6476
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7690040112509114e-05,
+      "loss": 0.0809,
+      "step": 6477
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7687980609618726e-05,
+      "loss": 0.0831,
+      "step": 6478
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7685920309035723e-05,
+      "loss": 0.0715,
+      "step": 6479
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7683859210973874e-05,
+      "loss": 0.0995,
+      "step": 6480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.768179731564704e-05,
+      "loss": 0.0701,
+      "step": 6481
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7679734623269156e-05,
+      "loss": 0.0869,
+      "step": 6482
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.767767113405425e-05,
+      "loss": 0.0767,
+      "step": 6483
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.767560684821642e-05,
+      "loss": 0.0813,
+      "step": 6484
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.767354176596985e-05,
+      "loss": 0.0684,
+      "step": 6485
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7671475887528817e-05,
+      "loss": 0.0703,
+      "step": 6486
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7669409213107674e-05,
+      "loss": 0.068,
+      "step": 6487
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7667341742920846e-05,
+      "loss": 0.0891,
+      "step": 6488
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7665273477182863e-05,
+      "loss": 0.0704,
+      "step": 6489
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7663204416108315e-05,
+      "loss": 0.0746,
+      "step": 6490
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.766113455991189e-05,
+      "loss": 0.0726,
+      "step": 6491
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.765906390880835e-05,
+      "loss": 0.0688,
+      "step": 6492
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7656992463012548e-05,
+      "loss": 0.0812,
+      "step": 6493
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7654920222739403e-05,
+      "loss": 0.0671,
+      "step": 6494
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7652847188203938e-05,
+      "loss": 0.0773,
+      "step": 6495
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.765077335962124e-05,
+      "loss": 0.0525,
+      "step": 6496
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7648698737206497e-05,
+      "loss": 0.0723,
+      "step": 6497
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.764662332117496e-05,
+      "loss": 0.0804,
+      "step": 6498
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7644547111741968e-05,
+      "loss": 0.0793,
+      "step": 6499
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7642470109122954e-05,
+      "loss": 0.0852,
+      "step": 6500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7640392313533416e-05,
+      "loss": 0.0828,
+      "step": 6501
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7638313725188948e-05,
+      "loss": 0.0825,
+      "step": 6502
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7636234344305217e-05,
+      "loss": 0.0756,
+      "step": 6503
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7634154171097984e-05,
+      "loss": 0.0688,
+      "step": 6504
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7632073205783076e-05,
+      "loss": 0.0875,
+      "step": 6505
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.762999144857642e-05,
+      "loss": 0.0694,
+      "step": 6506
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7627908899694005e-05,
+      "loss": 0.0858,
+      "step": 6507
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7625825559351917e-05,
+      "loss": 0.0579,
+      "step": 6508
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.762374142776632e-05,
+      "loss": 0.0742,
+      "step": 6509
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7621656505153466e-05,
+      "loss": 0.09,
+      "step": 6510
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7619570791729676e-05,
+      "loss": 0.0684,
+      "step": 6511
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.761748428771136e-05,
+      "loss": 0.0955,
+      "step": 6512
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.761539699331502e-05,
+      "loss": 0.0892,
+      "step": 6513
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7613308908757215e-05,
+      "loss": 0.0748,
+      "step": 6514
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7611220034254612e-05,
+      "loss": 0.0688,
+      "step": 6515
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.760913037002395e-05,
+      "loss": 0.0735,
+      "step": 6516
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7607039916282044e-05,
+      "loss": 0.0655,
+      "step": 6517
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7604948673245798e-05,
+      "loss": 0.0814,
+      "step": 6518
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7602856641132197e-05,
+      "loss": 0.0718,
+      "step": 6519
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7600763820158308e-05,
+      "loss": 0.0883,
+      "step": 6520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7598670210541273e-05,
+      "loss": 0.096,
+      "step": 6521
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.759657581249833e-05,
+      "loss": 0.0688,
+      "step": 6522
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7594480626246784e-05,
+      "loss": 0.0855,
+      "step": 6523
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7592384652004032e-05,
+      "loss": 0.0951,
+      "step": 6524
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7590287889987544e-05,
+      "loss": 0.08,
+      "step": 6525
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7588190340414882e-05,
+      "loss": 0.081,
+      "step": 6526
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.758609200350368e-05,
+      "loss": 0.0691,
+      "step": 6527
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7583992879471664e-05,
+      "loss": 0.09,
+      "step": 6528
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.758189296853663e-05,
+      "loss": 0.0814,
+      "step": 6529
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7579792270916463e-05,
+      "loss": 0.0708,
+      "step": 6530
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.757769078682913e-05,
+      "loss": 0.0717,
+      "step": 6531
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7575588516492677e-05,
+      "loss": 0.0706,
+      "step": 6532
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7573485460125227e-05,
+      "loss": 0.1007,
+      "step": 6533
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7571381617945e-05,
+      "loss": 0.0834,
+      "step": 6534
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7569276990170276e-05,
+      "loss": 0.0746,
+      "step": 6535
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7567171577019438e-05,
+      "loss": 0.0779,
+      "step": 6536
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7565065378710934e-05,
+      "loss": 0.0667,
+      "step": 6537
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.75629583954633e-05,
+      "loss": 0.0782,
+      "step": 6538
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7560850627495156e-05,
+      "loss": 0.0649,
+      "step": 6539
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.75587420750252e-05,
+      "loss": 0.0786,
+      "step": 6540
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.755663273827221e-05,
+      "loss": 0.082,
+      "step": 6541
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.755452261745505e-05,
+      "loss": 0.0718,
+      "step": 6542
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7552411712792664e-05,
+      "loss": 0.0673,
+      "step": 6543
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7550300024504067e-05,
+      "loss": 0.0924,
+      "step": 6544
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.754818755280838e-05,
+      "loss": 0.0726,
+      "step": 6545
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7546074297924774e-05,
+      "loss": 0.075,
+      "step": 6546
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7543960260072522e-05,
+      "loss": 0.0756,
+      "step": 6547
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.754184543947098e-05,
+      "loss": 0.0643,
+      "step": 6548
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7539729836339573e-05,
+      "loss": 0.0623,
+      "step": 6549
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.753761345089781e-05,
+      "loss": 0.0635,
+      "step": 6550
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7535496283365288e-05,
+      "loss": 0.0754,
+      "step": 6551
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.753337833396168e-05,
+      "loss": 0.1031,
+      "step": 6552
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.753125960290674e-05,
+      "loss": 0.0868,
+      "step": 6553
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7529140090420307e-05,
+      "loss": 0.0802,
+      "step": 6554
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7527019796722296e-05,
+      "loss": 0.0715,
+      "step": 6555
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7524898722032704e-05,
+      "loss": 0.0753,
+      "step": 6556
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.752277686657161e-05,
+      "loss": 0.0834,
+      "step": 6557
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.752065423055918e-05,
+      "loss": 0.0844,
+      "step": 6558
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.751853081421565e-05,
+      "loss": 0.0735,
+      "step": 6559
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7516406617761342e-05,
+      "loss": 0.0771,
+      "step": 6560
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7514281641416662e-05,
+      "loss": 0.0802,
+      "step": 6561
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7512155885402095e-05,
+      "loss": 0.0706,
+      "step": 6562
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.75100293499382e-05,
+      "loss": 0.0746,
+      "step": 6563
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7507902035245628e-05,
+      "loss": 0.0814,
+      "step": 6564
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7505773941545108e-05,
+      "loss": 0.0803,
+      "step": 6565
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7503645069057444e-05,
+      "loss": 0.0746,
+      "step": 6566
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7501515418003522e-05,
+      "loss": 0.0867,
+      "step": 6567
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7499384988604316e-05,
+      "loss": 0.0966,
+      "step": 6568
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7497253781080876e-05,
+      "loss": 0.0862,
+      "step": 6569
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7495121795654326e-05,
+      "loss": 0.0757,
+      "step": 6570
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7492989032545886e-05,
+      "loss": 0.0776,
+      "step": 6571
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7490855491976843e-05,
+      "loss": 0.0663,
+      "step": 6572
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7488721174168573e-05,
+      "loss": 0.0683,
+      "step": 6573
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7486586079342523e-05,
+      "loss": 0.0938,
+      "step": 6574
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7484450207720236e-05,
+      "loss": 0.0671,
+      "step": 6575
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.748231355952332e-05,
+      "loss": 0.0782,
+      "step": 6576
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7480176134973474e-05,
+      "loss": 0.0752,
+      "step": 6577
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7478037934292468e-05,
+      "loss": 0.0757,
+      "step": 6578
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7475898957702167e-05,
+      "loss": 0.07,
+      "step": 6579
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.74737592054245e-05,
+      "loss": 0.0802,
+      "step": 6580
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.747161867768149e-05,
+      "loss": 0.0847,
+      "step": 6581
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.746947737469523e-05,
+      "loss": 0.0748,
+      "step": 6582
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7467335296687903e-05,
+      "loss": 0.0552,
+      "step": 6583
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7465192443881763e-05,
+      "loss": 0.0758,
+      "step": 6584
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.746304881649915e-05,
+      "loss": 0.0837,
+      "step": 6585
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7460904414762488e-05,
+      "loss": 0.0691,
+      "step": 6586
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7458759238894273e-05,
+      "loss": 0.0824,
+      "step": 6587
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.745661328911708e-05,
+      "loss": 0.0824,
+      "step": 6588
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.745446656565358e-05,
+      "loss": 0.0747,
+      "step": 6589
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7452319068726503e-05,
+      "loss": 0.0793,
+      "step": 6590
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.745017079855868e-05,
+      "loss": 0.0767,
+      "step": 6591
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7448021755373005e-05,
+      "loss": 0.0889,
+      "step": 6592
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7445871939392457e-05,
+      "loss": 0.0887,
+      "step": 6593
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7443721350840106e-05,
+      "loss": 0.0771,
+      "step": 6594
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7441569989939092e-05,
+      "loss": 0.061,
+      "step": 6595
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7439417856912632e-05,
+      "loss": 0.0583,
+      "step": 6596
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7437264951984027e-05,
+      "loss": 0.0704,
+      "step": 6597
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7435111275376668e-05,
+      "loss": 0.0634,
+      "step": 6598
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7432956827314004e-05,
+      "loss": 0.0645,
+      "step": 6599
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.743080160801959e-05,
+      "loss": 0.0824,
+      "step": 6600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.742864561771704e-05,
+      "loss": 0.0734,
+      "step": 6601
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7426488856630058e-05,
+      "loss": 0.0777,
+      "step": 6602
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7424331324982425e-05,
+      "loss": 0.069,
+      "step": 6603
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.742217302299801e-05,
+      "loss": 0.074,
+      "step": 6604
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7420013950900743e-05,
+      "loss": 0.0857,
+      "step": 6605
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7417854108914656e-05,
+      "loss": 0.0776,
+      "step": 6606
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.741569349726385e-05,
+      "loss": 0.0574,
+      "step": 6607
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.74135321161725e-05,
+      "loss": 0.0707,
+      "step": 6608
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7411369965864872e-05,
+      "loss": 0.0718,
+      "step": 6609
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7409207046565306e-05,
+      "loss": 0.0797,
+      "step": 6610
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7407043358498227e-05,
+      "loss": 0.0664,
+      "step": 6611
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.740487890188813e-05,
+      "loss": 0.0703,
+      "step": 6612
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7402713676959598e-05,
+      "loss": 0.0809,
+      "step": 6613
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.740054768393729e-05,
+      "loss": 0.079,
+      "step": 6614
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.739838092304595e-05,
+      "loss": 0.0851,
+      "step": 6615
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7396213394510393e-05,
+      "loss": 0.0825,
+      "step": 6616
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7394045098555522e-05,
+      "loss": 0.0806,
+      "step": 6617
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7391876035406312e-05,
+      "loss": 0.0865,
+      "step": 6618
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7389706205287824e-05,
+      "loss": 0.0886,
+      "step": 6619
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7387535608425197e-05,
+      "loss": 0.0697,
+      "step": 6620
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7385364245043646e-05,
+      "loss": 0.0862,
+      "step": 6621
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.738319211536847e-05,
+      "loss": 0.066,
+      "step": 6622
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.738101921962505e-05,
+      "loss": 0.0833,
+      "step": 6623
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.737884555803883e-05,
+      "loss": 0.0729,
+      "step": 6624
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7376671130835362e-05,
+      "loss": 0.0625,
+      "step": 6625
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7374495938240246e-05,
+      "loss": 0.0659,
+      "step": 6626
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7372319980479185e-05,
+      "loss": 0.063,
+      "step": 6627
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.737014325777795e-05,
+      "loss": 0.0737,
+      "step": 6628
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7367965770362393e-05,
+      "loss": 0.0808,
+      "step": 6629
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7365787518458452e-05,
+      "loss": 0.0851,
+      "step": 6630
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7363608502292136e-05,
+      "loss": 0.0825,
+      "step": 6631
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7361428722089532e-05,
+      "loss": 0.0729,
+      "step": 6632
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7359248178076818e-05,
+      "loss": 0.0695,
+      "step": 6633
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.735706687048024e-05,
+      "loss": 0.0883,
+      "step": 6634
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7354884799526127e-05,
+      "loss": 0.0736,
+      "step": 6635
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.735270196544089e-05,
+      "loss": 0.0789,
+      "step": 6636
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.735051836845101e-05,
+      "loss": 0.0868,
+      "step": 6637
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.734833400878306e-05,
+      "loss": 0.0665,
+      "step": 6638
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.734614888666368e-05,
+      "loss": 0.0818,
+      "step": 6639
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7343963002319597e-05,
+      "loss": 0.0741,
+      "step": 6640
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.734177635597762e-05,
+      "loss": 0.0764,
+      "step": 6641
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7339588947864626e-05,
+      "loss": 0.0784,
+      "step": 6642
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7337400778207578e-05,
+      "loss": 0.0726,
+      "step": 6643
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7335211847233514e-05,
+      "loss": 0.0729,
+      "step": 6644
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7333022155169563e-05,
+      "loss": 0.0787,
+      "step": 6645
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.733083170224292e-05,
+      "loss": 0.062,
+      "step": 6646
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7328640488680854e-05,
+      "loss": 0.1082,
+      "step": 6647
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7326448514710733e-05,
+      "loss": 0.0766,
+      "step": 6648
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7324255780559993e-05,
+      "loss": 0.0903,
+      "step": 6649
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7322062286456142e-05,
+      "loss": 0.0753,
+      "step": 6650
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7319868032626773e-05,
+      "loss": 0.0747,
+      "step": 6651
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7317673019299566e-05,
+      "loss": 0.0706,
+      "step": 6652
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7315477246702263e-05,
+      "loss": 0.0639,
+      "step": 6653
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.73132807150627e-05,
+      "loss": 0.0908,
+      "step": 6654
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7311083424608785e-05,
+      "loss": 0.0784,
+      "step": 6655
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7308885375568505e-05,
+      "loss": 0.0636,
+      "step": 6656
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7306686568169924e-05,
+      "loss": 0.0731,
+      "step": 6657
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.730448700264119e-05,
+      "loss": 0.0697,
+      "step": 6658
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.730228667921052e-05,
+      "loss": 0.0746,
+      "step": 6659
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7300085598106223e-05,
+      "loss": 0.0775,
+      "step": 6660
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7297883759556676e-05,
+      "loss": 0.0854,
+      "step": 6661
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7295681163790343e-05,
+      "loss": 0.0733,
+      "step": 6662
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7293477811035758e-05,
+      "loss": 0.0723,
+      "step": 6663
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7291273701521534e-05,
+      "loss": 0.0785,
+      "step": 6664
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.728906883547637e-05,
+      "loss": 0.0576,
+      "step": 6665
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7286863213129045e-05,
+      "loss": 0.0664,
+      "step": 6666
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.72846568347084e-05,
+      "loss": 0.0652,
+      "step": 6667
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.728244970044337e-05,
+      "loss": 0.0574,
+      "step": 6668
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7280241810562964e-05,
+      "loss": 0.077,
+      "step": 6669
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7278033165296267e-05,
+      "loss": 0.0722,
+      "step": 6670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.727582376487245e-05,
+      "loss": 0.0889,
+      "step": 6671
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.727361360952075e-05,
+      "loss": 0.0914,
+      "step": 6672
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7271402699470498e-05,
+      "loss": 0.0845,
+      "step": 6673
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7269191034951086e-05,
+      "loss": 0.0605,
+      "step": 6674
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7266978616191996e-05,
+      "loss": 0.0829,
+      "step": 6675
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7264765443422783e-05,
+      "loss": 0.0681,
+      "step": 6676
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7262551516873086e-05,
+      "loss": 0.0798,
+      "step": 6677
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7260336836772617e-05,
+      "loss": 0.0877,
+      "step": 6678
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7258121403351168e-05,
+      "loss": 0.0707,
+      "step": 6679
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7255905216838607e-05,
+      "loss": 0.0784,
+      "step": 6680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7253688277464884e-05,
+      "loss": 0.073,
+      "step": 6681
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7251470585460026e-05,
+      "loss": 0.075,
+      "step": 6682
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7249252141054133e-05,
+      "loss": 0.0849,
+      "step": 6683
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.724703294447739e-05,
+      "loss": 0.0865,
+      "step": 6684
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.7244812995960056e-05,
+      "loss": 0.0785,
+      "step": 6685
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.724259229573247e-05,
+      "loss": 0.0613,
+      "step": 6686
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.724037084402505e-05,
+      "loss": 0.0652,
+      "step": 6687
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7238148641068292e-05,
+      "loss": 0.0775,
+      "step": 6688
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.723592568709276e-05,
+      "loss": 0.0769,
+      "step": 6689
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7233701982329113e-05,
+      "loss": 0.085,
+      "step": 6690
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7231477527008074e-05,
+      "loss": 0.064,
+      "step": 6691
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.722925232136045e-05,
+      "loss": 0.0688,
+      "step": 6692
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7227026365617124e-05,
+      "loss": 0.0708,
+      "step": 6693
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7224799660009064e-05,
+      "loss": 0.082,
+      "step": 6694
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7222572204767298e-05,
+      "loss": 0.0722,
+      "step": 6695
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7220344000122954e-05,
+      "loss": 0.0981,
+      "step": 6696
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.721811504630722e-05,
+      "loss": 0.0589,
+      "step": 6697
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.721588534355137e-05,
+      "loss": 0.0869,
+      "step": 6698
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.721365489208676e-05,
+      "loss": 0.0785,
+      "step": 6699
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.721142369214481e-05,
+      "loss": 0.0792,
+      "step": 6700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7209191743957027e-05,
+      "loss": 0.0677,
+      "step": 6701
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7206959047755e-05,
+      "loss": 0.0869,
+      "step": 6702
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7204725603770387e-05,
+      "loss": 0.0933,
+      "step": 6703
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.7202491412234925e-05,
+      "loss": 0.0734,
+      "step": 6704
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.720025647338043e-05,
+      "loss": 0.0817,
+      "step": 6705
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.71980207874388e-05,
+      "loss": 0.0823,
+      "step": 6706
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7195784354642004e-05,
+      "loss": 0.0801,
+      "step": 6707
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.719354717522209e-05,
+      "loss": 0.0822,
+      "step": 6708
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.719130924941118e-05,
+      "loss": 0.0688,
+      "step": 6709
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7189070577441485e-05,
+      "loss": 0.0734,
+      "step": 6710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7186831159545284e-05,
+      "loss": 0.0685,
+      "step": 6711
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.718459099595493e-05,
+      "loss": 0.0723,
+      "step": 6712
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.718235008690287e-05,
+      "loss": 0.0704,
+      "step": 6713
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7180108432621604e-05,
+      "loss": 0.0742,
+      "step": 6714
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.717786603334373e-05,
+      "loss": 0.0859,
+      "step": 6715
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7175622889301916e-05,
+      "loss": 0.0775,
+      "step": 6716
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7173379000728906e-05,
+      "loss": 0.0601,
+      "step": 6717
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.717113436785752e-05,
+      "loss": 0.0733,
+      "step": 6718
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.716888899092066e-05,
+      "loss": 0.0815,
+      "step": 6719
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7166642870151303e-05,
+      "loss": 0.0717,
+      "step": 6720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.71643960057825e-05,
+      "loss": 0.0735,
+      "step": 6721
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7162148398047386e-05,
+      "loss": 0.0878,
+      "step": 6722
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7159900047179167e-05,
+      "loss": 0.0769,
+      "step": 6723
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.715765095341113e-05,
+      "loss": 0.0766,
+      "step": 6724
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.7155401116976634e-05,
+      "loss": 0.0766,
+      "step": 6725
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.715315053810912e-05,
+      "loss": 0.0745,
+      "step": 6726
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.715089921704211e-05,
+      "loss": 0.0725,
+      "step": 6727
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7148647154009183e-05,
+      "loss": 0.0682,
+      "step": 6728
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7146394349244023e-05,
+      "loss": 0.0787,
+      "step": 6729
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7144140802980377e-05,
+      "loss": 0.0715,
+      "step": 6730
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7141886515452065e-05,
+      "loss": 0.067,
+      "step": 6731
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7139631486892986e-05,
+      "loss": 0.0839,
+      "step": 6732
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7137375717537122e-05,
+      "loss": 0.0739,
+      "step": 6733
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7135119207618526e-05,
+      "loss": 0.0728,
+      "step": 6734
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7132861957371335e-05,
+      "loss": 0.0726,
+      "step": 6735
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.713060396702975e-05,
+      "loss": 0.0676,
+      "step": 6736
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.712834523682806e-05,
+      "loss": 0.0729,
+      "step": 6737
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.712608576700063e-05,
+      "loss": 0.0814,
+      "step": 6738
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7123825557781894e-05,
+      "loss": 0.0796,
+      "step": 6739
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7121564609406372e-05,
+      "loss": 0.0655,
+      "step": 6740
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7119302922108655e-05,
+      "loss": 0.0817,
+      "step": 6741
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7117040496123408e-05,
+      "loss": 0.0862,
+      "step": 6742
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7114777331685384e-05,
+      "loss": 0.0689,
+      "step": 6743
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7112513429029403e-05,
+      "loss": 0.0782,
+      "step": 6744
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.7110248788390358e-05,
+      "loss": 0.0635,
+      "step": 6745
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.710798341000323e-05,
+      "loss": 0.0792,
+      "step": 6746
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.710571729410307e-05,
+      "loss": 0.0638,
+      "step": 6747
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7103450440925013e-05,
+      "loss": 0.071,
+      "step": 6748
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.710118285070425e-05,
+      "loss": 0.0748,
+      "step": 6749
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7098914523676075e-05,
+      "loss": 0.0765,
+      "step": 6750
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7096645460075837e-05,
+      "loss": 0.0613,
+      "step": 6751
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7094375660138978e-05,
+      "loss": 0.0715,
+      "step": 6752
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7092105124101005e-05,
+      "loss": 0.0726,
+      "step": 6753
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7089833852197508e-05,
+      "loss": 0.081,
+      "step": 6754
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7087561844664144e-05,
+      "loss": 0.0747,
+      "step": 6755
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.708528910173666e-05,
+      "loss": 0.0703,
+      "step": 6756
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7083015623650867e-05,
+      "loss": 0.0657,
+      "step": 6757
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7080741410642667e-05,
+      "loss": 0.0668,
+      "step": 6758
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7078466462948015e-05,
+      "loss": 0.0675,
+      "step": 6759
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.707619078080296e-05,
+      "loss": 0.0673,
+      "step": 6760
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.707391436444363e-05,
+      "loss": 0.0775,
+      "step": 6761
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7071637214106218e-05,
+      "loss": 0.0739,
+      "step": 6762
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7069359330027e-05,
+      "loss": 0.0781,
+      "step": 6763
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.706708071244232e-05,
+      "loss": 0.0638,
+      "step": 6764
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7064801361588603e-05,
+      "loss": 0.0752,
+      "step": 6765
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.706252127770236e-05,
+      "loss": 0.0693,
+      "step": 6766
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.7060240461020164e-05,
+      "loss": 0.0673,
+      "step": 6767
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7057958911778665e-05,
+      "loss": 0.0861,
+      "step": 6768
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7055676630214598e-05,
+      "loss": 0.069,
+      "step": 6769
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.705339361656477e-05,
+      "loss": 0.0801,
+      "step": 6770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7051109871066055e-05,
+      "loss": 0.0753,
+      "step": 6771
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.704882539395542e-05,
+      "loss": 0.097,
+      "step": 6772
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7046540185469895e-05,
+      "loss": 0.0768,
+      "step": 6773
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7044254245846586e-05,
+      "loss": 0.0683,
+      "step": 6774
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.704196757532268e-05,
+      "loss": 0.0704,
+      "step": 6775
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7039680174135446e-05,
+      "loss": 0.0753,
+      "step": 6776
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.703739204252221e-05,
+      "loss": 0.0715,
+      "step": 6777
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7035103180720392e-05,
+      "loss": 0.0721,
+      "step": 6778
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.703281358896748e-05,
+      "loss": 0.0854,
+      "step": 6779
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7030523267501037e-05,
+      "loss": 0.0785,
+      "step": 6780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.70282322165587e-05,
+      "loss": 0.0887,
+      "step": 6781
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.702594043637819e-05,
+      "loss": 0.0834,
+      "step": 6782
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7023647927197297e-05,
+      "loss": 0.0727,
+      "step": 6783
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7021354689253888e-05,
+      "loss": 0.078,
+      "step": 6784
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.70190607227859e-05,
+      "loss": 0.0696,
+      "step": 6785
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.7016766028031363e-05,
+      "loss": 0.0936,
+      "step": 6786
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.701447060522836e-05,
+      "loss": 0.0659,
+      "step": 6787
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7012174454615066e-05,
+      "loss": 0.0773,
+      "step": 6788
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7009877576429724e-05,
+      "loss": 0.0696,
+      "step": 6789
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7007579970910657e-05,
+      "loss": 0.0784,
+      "step": 6790
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.700528163829626e-05,
+      "loss": 0.0905,
+      "step": 6791
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7002982578825e-05,
+      "loss": 0.0759,
+      "step": 6792
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.7000682792735427e-05,
+      "loss": 0.0644,
+      "step": 6793
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6998382280266167e-05,
+      "loss": 0.0563,
+      "step": 6794
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.699608104165591e-05,
+      "loss": 0.0864,
+      "step": 6795
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6993779077143437e-05,
+      "loss": 0.0933,
+      "step": 6796
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6991476386967584e-05,
+      "loss": 0.0923,
+      "step": 6797
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.698917297136729e-05,
+      "loss": 0.0776,
+      "step": 6798
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6986868830581542e-05,
+      "loss": 0.0869,
+      "step": 6799
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.698456396484942e-05,
+      "loss": 0.0848,
+      "step": 6800
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6982258374410067e-05,
+      "loss": 0.0803,
+      "step": 6801
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6979952059502715e-05,
+      "loss": 0.0793,
+      "step": 6802
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.697764502036666e-05,
+      "loss": 0.0739,
+      "step": 6803
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6975337257241275e-05,
+      "loss": 0.0612,
+      "step": 6804
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6973028770366015e-05,
+      "loss": 0.0603,
+      "step": 6805
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.69707195599804e-05,
+      "loss": 0.0672,
+      "step": 6806
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6968409626324028e-05,
+      "loss": 0.0697,
+      "step": 6807
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6966098969636583e-05,
+      "loss": 0.0942,
+      "step": 6808
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6963787590157805e-05,
+      "loss": 0.0714,
+      "step": 6809
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6961475488127525e-05,
+      "loss": 0.0739,
+      "step": 6810
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.695916266378564e-05,
+      "loss": 0.0919,
+      "step": 6811
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6956849117372133e-05,
+      "loss": 0.0641,
+      "step": 6812
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6954534849127045e-05,
+      "loss": 0.0677,
+      "step": 6813
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.69522198592905e-05,
+      "loss": 0.0652,
+      "step": 6814
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.69499041481027e-05,
+      "loss": 0.0679,
+      "step": 6815
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6947587715803923e-05,
+      "loss": 0.0685,
+      "step": 6816
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.694527056263452e-05,
+      "loss": 0.0797,
+      "step": 6817
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6942952688834907e-05,
+      "loss": 0.0768,
+      "step": 6818
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6940634094645586e-05,
+      "loss": 0.0782,
+      "step": 6819
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.693831478030713e-05,
+      "loss": 0.0774,
+      "step": 6820
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.693599474606019e-05,
+      "loss": 0.078,
+      "step": 6821
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.693367399214549e-05,
+      "loss": 0.0722,
+      "step": 6822
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6931352518803825e-05,
+      "loss": 0.0766,
+      "step": 6823
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6929030326276067e-05,
+      "loss": 0.0714,
+      "step": 6824
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6926707414803165e-05,
+      "loss": 0.0609,
+      "step": 6825
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.692438378462614e-05,
+      "loss": 0.0572,
+      "step": 6826
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6922059435986082e-05,
+      "loss": 0.0835,
+      "step": 6827
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6919734369124175e-05,
+      "loss": 0.076,
+      "step": 6828
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6917408584281654e-05,
+      "loss": 0.0703,
+      "step": 6829
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6915082081699842e-05,
+      "loss": 0.08,
+      "step": 6830
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.691275486162013e-05,
+      "loss": 0.0736,
+      "step": 6831
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6910426924283993e-05,
+      "loss": 0.0864,
+      "step": 6832
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6908098269932967e-05,
+      "loss": 0.0709,
+      "step": 6833
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6905768898808676e-05,
+      "loss": 0.082,
+      "step": 6834
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6903438811152803e-05,
+      "loss": 0.0863,
+      "step": 6835
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6901108007207124e-05,
+      "loss": 0.0789,
+      "step": 6836
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6898776487213472e-05,
+      "loss": 0.0687,
+      "step": 6837
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6896444251413768e-05,
+      "loss": 0.0701,
+      "step": 6838
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6894111300050002e-05,
+      "loss": 0.0852,
+      "step": 6839
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6891777633364224e-05,
+      "loss": 0.0841,
+      "step": 6840
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.688944325159859e-05,
+      "loss": 0.0773,
+      "step": 6841
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6887108154995298e-05,
+      "loss": 0.0671,
+      "step": 6842
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.688477234379664e-05,
+      "loss": 0.0603,
+      "step": 6843
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6882435818244976e-05,
+      "loss": 0.078,
+      "step": 6844
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6880098578582737e-05,
+      "loss": 0.073,
+      "step": 6845
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6877760625052432e-05,
+      "loss": 0.0562,
+      "step": 6846
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6875421957896646e-05,
+      "loss": 0.0781,
+      "step": 6847
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6873082577358033e-05,
+      "loss": 0.0568,
+      "step": 6848
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6870742483679326e-05,
+      "loss": 0.0773,
+      "step": 6849
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6868401677103324e-05,
+      "loss": 0.0774,
+      "step": 6850
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6866060157872913e-05,
+      "loss": 0.0858,
+      "step": 6851
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6863717926231042e-05,
+      "loss": 0.0793,
+      "step": 6852
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.686137498242073e-05,
+      "loss": 0.0721,
+      "step": 6853
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.685903132668509e-05,
+      "loss": 0.0613,
+      "step": 6854
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.685668695926729e-05,
+      "loss": 0.0529,
+      "step": 6855
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6854341880410573e-05,
+      "loss": 0.0963,
+      "step": 6856
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6851996090358266e-05,
+      "loss": 0.0681,
+      "step": 6857
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6849649589353764e-05,
+      "loss": 0.0768,
+      "step": 6858
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6847302377640538e-05,
+      "loss": 0.0757,
+      "step": 6859
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6844954455462124e-05,
+      "loss": 0.0776,
+      "step": 6860
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.684260582306215e-05,
+      "loss": 0.0788,
+      "step": 6861
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6840256480684294e-05,
+      "loss": 0.0895,
+      "step": 6862
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.683790642857233e-05,
+      "loss": 0.0791,
+      "step": 6863
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6835555666970086e-05,
+      "loss": 0.0638,
+      "step": 6864
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.683320419612148e-05,
+      "loss": 0.0676,
+      "step": 6865
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.68308520162705e-05,
+      "loss": 0.0957,
+      "step": 6866
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6828499127661198e-05,
+      "loss": 0.0816,
+      "step": 6867
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6826145530537705e-05,
+      "loss": 0.0639,
+      "step": 6868
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6823791225144232e-05,
+      "loss": 0.0763,
+      "step": 6869
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6821436211725057e-05,
+      "loss": 0.0745,
+      "step": 6870
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6819080490524527e-05,
+      "loss": 0.0756,
+      "step": 6871
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6816724061787076e-05,
+      "loss": 0.0734,
+      "step": 6872
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6814366925757192e-05,
+      "loss": 0.078,
+      "step": 6873
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.681200908267946e-05,
+      "loss": 0.0851,
+      "step": 6874
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.680965053279852e-05,
+      "loss": 0.0825,
+      "step": 6875
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.680729127635909e-05,
+      "loss": 0.0854,
+      "step": 6876
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.680493131360597e-05,
+      "loss": 0.0679,
+      "step": 6877
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6802570644784016e-05,
+      "loss": 0.0853,
+      "step": 6878
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6800209270138174e-05,
+      "loss": 0.0746,
+      "step": 6879
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6797847189913456e-05,
+      "loss": 0.0679,
+      "step": 6880
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.679548440435494e-05,
+      "loss": 0.0688,
+      "step": 6881
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6793120913707798e-05,
+      "loss": 0.082,
+      "step": 6882
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6790756718217252e-05,
+      "loss": 0.085,
+      "step": 6883
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.678839181812861e-05,
+      "loss": 0.0833,
+      "step": 6884
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6786026213687252e-05,
+      "loss": 0.0742,
+      "step": 6885
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6783659905138626e-05,
+      "loss": 0.0795,
+      "step": 6886
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6781292892728257e-05,
+      "loss": 0.0716,
+      "step": 6887
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6778925176701747e-05,
+      "loss": 0.0638,
+      "step": 6888
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.677655675730476e-05,
+      "loss": 0.0636,
+      "step": 6889
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6774187634783046e-05,
+      "loss": 0.0566,
+      "step": 6890
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6771817809382415e-05,
+      "loss": 0.0523,
+      "step": 6891
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6769447281348757e-05,
+      "loss": 0.078,
+      "step": 6892
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.676707605092804e-05,
+      "loss": 0.0847,
+      "step": 6893
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6764704118366298e-05,
+      "loss": 0.0705,
+      "step": 6894
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.676233148390963e-05,
+      "loss": 0.076,
+      "step": 6895
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6759958147804228e-05,
+      "loss": 0.0634,
+      "step": 6896
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.675758411029634e-05,
+      "loss": 0.0802,
+      "step": 6897
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.675520937163229e-05,
+      "loss": 0.0775,
+      "step": 6898
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6752833932058484e-05,
+      "loss": 0.0728,
+      "step": 6899
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.675045779182139e-05,
+      "loss": 0.0716,
+      "step": 6900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6748080951167552e-05,
+      "loss": 0.0844,
+      "step": 6901
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6745703410343586e-05,
+      "loss": 0.0734,
+      "step": 6902
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6743325169596184e-05,
+      "loss": 0.0819,
+      "step": 6903
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.674094622917211e-05,
+      "loss": 0.0557,
+      "step": 6904
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6738566589318196e-05,
+      "loss": 0.0723,
+      "step": 6905
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.673618625028135e-05,
+      "loss": 0.0646,
+      "step": 6906
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6733805212308553e-05,
+      "loss": 0.0648,
+      "step": 6907
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.673142347564686e-05,
+      "loss": 0.0764,
+      "step": 6908
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.672904104054339e-05,
+      "loss": 0.0705,
+      "step": 6909
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6726657907245348e-05,
+      "loss": 0.0776,
+      "step": 6910
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6724274076e-05,
+      "loss": 0.0773,
+      "step": 6911
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6721889547054687e-05,
+      "loss": 0.0646,
+      "step": 6912
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6719504320656827e-05,
+      "loss": 0.0898,
+      "step": 6913
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.671711839705391e-05,
+      "loss": 0.0884,
+      "step": 6914
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6714731776493486e-05,
+      "loss": 0.0807,
+      "step": 6915
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6712344459223198e-05,
+      "loss": 0.062,
+      "step": 6916
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.670995644549074e-05,
+      "loss": 0.0671,
+      "step": 6917
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6707567735543898e-05,
+      "loss": 0.0748,
+      "step": 6918
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.670517832963052e-05,
+      "loss": 0.0823,
+      "step": 6919
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6702788227998517e-05,
+      "loss": 0.0762,
+      "step": 6920
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6700397430895888e-05,
+      "loss": 0.0824,
+      "step": 6921
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6698005938570702e-05,
+      "loss": 0.0672,
+      "step": 6922
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6695613751271094e-05,
+      "loss": 0.0822,
+      "step": 6923
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.669322086924527e-05,
+      "loss": 0.0917,
+      "step": 6924
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.669082729274152e-05,
+      "loss": 0.0808,
+      "step": 6925
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6688433022008187e-05,
+      "loss": 0.0661,
+      "step": 6926
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6686038057293705e-05,
+      "loss": 0.0779,
+      "step": 6927
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6683642398846563e-05,
+      "loss": 0.0824,
+      "step": 6928
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.668124604691534e-05,
+      "loss": 0.0721,
+      "step": 6929
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6678849001748677e-05,
+      "loss": 0.0657,
+      "step": 6930
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6676451263595276e-05,
+      "loss": 0.0752,
+      "step": 6931
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.667405283270394e-05,
+      "loss": 0.0663,
+      "step": 6932
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6671653709323513e-05,
+      "loss": 0.0654,
+      "step": 6933
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.666925389370293e-05,
+      "loss": 0.0757,
+      "step": 6934
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.666685338609119e-05,
+      "loss": 0.0696,
+      "step": 6935
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6664452186737366e-05,
+      "loss": 0.0708,
+      "step": 6936
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6662050295890605e-05,
+      "loss": 0.0874,
+      "step": 6937
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6659647713800117e-05,
+      "loss": 0.067,
+      "step": 6938
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6657244440715197e-05,
+      "loss": 0.0727,
+      "step": 6939
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6654840476885205e-05,
+      "loss": 0.074,
+      "step": 6940
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6652435822559566e-05,
+      "loss": 0.0734,
+      "step": 6941
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6650030477987787e-05,
+      "loss": 0.0623,
+      "step": 6942
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6647624443419446e-05,
+      "loss": 0.0814,
+      "step": 6943
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6645217719104187e-05,
+      "loss": 0.0739,
+      "step": 6944
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.664281030529172e-05,
+      "loss": 0.0618,
+      "step": 6945
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6640402202231847e-05,
+      "loss": 0.08,
+      "step": 6946
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.663799341017442e-05,
+      "loss": 0.0779,
+      "step": 6947
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6635583929369375e-05,
+      "loss": 0.0736,
+      "step": 6948
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6633173760066717e-05,
+      "loss": 0.0723,
+      "step": 6949
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.663076290251652e-05,
+      "loss": 0.0744,
+      "step": 6950
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.662835135696893e-05,
+      "loss": 0.064,
+      "step": 6951
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6625939123674165e-05,
+      "loss": 0.0722,
+      "step": 6952
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6623526202882516e-05,
+      "loss": 0.0761,
+      "step": 6953
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6621112594844342e-05,
+      "loss": 0.0692,
+      "step": 6954
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6618698299810078e-05,
+      "loss": 0.0814,
+      "step": 6955
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6616283318030225e-05,
+      "loss": 0.0607,
+      "step": 6956
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.661386764975536e-05,
+      "loss": 0.0787,
+      "step": 6957
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.661145129523612e-05,
+      "loss": 0.0667,
+      "step": 6958
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6609034254723234e-05,
+      "loss": 0.0643,
+      "step": 6959
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6606616528467486e-05,
+      "loss": 0.0806,
+      "step": 6960
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6604198116719735e-05,
+      "loss": 0.0812,
+      "step": 6961
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6601779019730908e-05,
+      "loss": 0.0807,
+      "step": 6962
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6599359237752015e-05,
+      "loss": 0.0916,
+      "step": 6963
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6596938771034116e-05,
+      "loss": 0.0684,
+      "step": 6964
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.659451761982837e-05,
+      "loss": 0.0666,
+      "step": 6965
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6592095784385983e-05,
+      "loss": 0.0826,
+      "step": 6966
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.658967326495824e-05,
+      "loss": 0.0793,
+      "step": 6967
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6587250061796498e-05,
+      "loss": 0.0736,
+      "step": 6968
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6584826175152192e-05,
+      "loss": 0.0701,
+      "step": 6969
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6582401605276813e-05,
+      "loss": 0.0709,
+      "step": 6970
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.657997635242193e-05,
+      "loss": 0.0677,
+      "step": 6971
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.657755041683919e-05,
+      "loss": 0.0699,
+      "step": 6972
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.65751237987803e-05,
+      "loss": 0.077,
+      "step": 6973
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6572696498497037e-05,
+      "loss": 0.0779,
+      "step": 6974
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6570268516241264e-05,
+      "loss": 0.0665,
+      "step": 6975
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6567839852264898e-05,
+      "loss": 0.0757,
+      "step": 6976
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6565410506819938e-05,
+      "loss": 0.0734,
+      "step": 6977
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6562980480158443e-05,
+      "loss": 0.079,
+      "step": 6978
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.656054977253255e-05,
+      "loss": 0.0741,
+      "step": 6979
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.655811838419447e-05,
+      "loss": 0.0823,
+      "step": 6980
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6555686315396475e-05,
+      "loss": 0.0624,
+      "step": 6981
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6553253566390916e-05,
+      "loss": 0.076,
+      "step": 6982
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6550820137430214e-05,
+      "loss": 0.0707,
+      "step": 6983
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6548386028766846e-05,
+      "loss": 0.0801,
+      "step": 6984
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6545951240653383e-05,
+      "loss": 0.0812,
+      "step": 6985
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6543515773342446e-05,
+      "loss": 0.0746,
+      "step": 6986
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6541079627086744e-05,
+      "loss": 0.0708,
+      "step": 6987
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6538642802139042e-05,
+      "loss": 0.0613,
+      "step": 6988
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6536205298752183e-05,
+      "loss": 0.0779,
+      "step": 6989
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6533767117179077e-05,
+      "loss": 0.0716,
+      "step": 6990
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6531328257672707e-05,
+      "loss": 0.0748,
+      "step": 6991
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6528888720486124e-05,
+      "loss": 0.0673,
+      "step": 6992
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6526448505872453e-05,
+      "loss": 0.0711,
+      "step": 6993
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6524007614084886e-05,
+      "loss": 0.09,
+      "step": 6994
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.652156604537668e-05,
+      "loss": 0.0795,
+      "step": 6995
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.651912380000118e-05,
+      "loss": 0.0721,
+      "step": 6996
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.651668087821178e-05,
+      "loss": 0.0577,
+      "step": 6997
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.651423728026195e-05,
+      "loss": 0.0834,
+      "step": 6998
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6511793006405254e-05,
+      "loss": 0.0596,
+      "step": 6999
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6509348056895284e-05,
+      "loss": 0.0751,
+      "step": 7000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6506902431985734e-05,
+      "loss": 0.0889,
+      "step": 7001
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6504456131930356e-05,
+      "loss": 0.0644,
+      "step": 7002
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6502009156982974e-05,
+      "loss": 0.0684,
+      "step": 7003
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6499561507397483e-05,
+      "loss": 0.0645,
+      "step": 7004
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.649711318342785e-05,
+      "loss": 0.0656,
+      "step": 7005
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6494664185328103e-05,
+      "loss": 0.0655,
+      "step": 7006
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6492214513352354e-05,
+      "loss": 0.0786,
+      "step": 7007
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6489764167754768e-05,
+      "loss": 0.0724,
+      "step": 7008
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6487313148789597e-05,
+      "loss": 0.0767,
+      "step": 7009
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.648486145671115e-05,
+      "loss": 0.0674,
+      "step": 7010
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.648240909177381e-05,
+      "loss": 0.0717,
+      "step": 7011
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6479956054232034e-05,
+      "loss": 0.0669,
+      "step": 7012
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6477502344340345e-05,
+      "loss": 0.0583,
+      "step": 7013
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6475047962353335e-05,
+      "loss": 0.0771,
+      "step": 7014
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6472592908525666e-05,
+      "loss": 0.079,
+      "step": 7015
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.647013718311207e-05,
+      "loss": 0.0737,
+      "step": 7016
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6467680786367353e-05,
+      "loss": 0.0664,
+      "step": 7017
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6465223718546383e-05,
+      "loss": 0.074,
+      "step": 7018
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.64627659799041e-05,
+      "loss": 0.0664,
+      "step": 7019
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6460307570695517e-05,
+      "loss": 0.0647,
+      "step": 7020
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.645784849117572e-05,
+      "loss": 0.0723,
+      "step": 7021
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.645538874159985e-05,
+      "loss": 0.0709,
+      "step": 7022
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6452928322223134e-05,
+      "loss": 0.0736,
+      "step": 7023
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6450467233300854e-05,
+      "loss": 0.0756,
+      "step": 7024
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6448005475088376e-05,
+      "loss": 0.0786,
+      "step": 7025
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6445543047841127e-05,
+      "loss": 0.0671,
+      "step": 7026
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.64430799518146e-05,
+      "loss": 0.0665,
+      "step": 7027
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6440616187264365e-05,
+      "loss": 0.0736,
+      "step": 7028
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6438151754446057e-05,
+      "loss": 0.0652,
+      "step": 7029
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.643568665361538e-05,
+      "loss": 0.0745,
+      "step": 7030
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6433220885028116e-05,
+      "loss": 0.0734,
+      "step": 7031
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6430754448940103e-05,
+      "loss": 0.0759,
+      "step": 7032
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6428287345607255e-05,
+      "loss": 0.0835,
+      "step": 7033
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6425819575285557e-05,
+      "loss": 0.0682,
+      "step": 7034
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.642335113823106e-05,
+      "loss": 0.0872,
+      "step": 7035
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6420882034699882e-05,
+      "loss": 0.072,
+      "step": 7036
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6418412264948214e-05,
+      "loss": 0.0863,
+      "step": 7037
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6415941829232325e-05,
+      "loss": 0.0847,
+      "step": 7038
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6413470727808533e-05,
+      "loss": 0.0752,
+      "step": 7039
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6410998960933234e-05,
+      "loss": 0.0651,
+      "step": 7040
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.64085265288629e-05,
+      "loss": 0.0791,
+      "step": 7041
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6406053431854066e-05,
+      "loss": 0.0811,
+      "step": 7042
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.640357967016334e-05,
+      "loss": 0.062,
+      "step": 7043
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.640110524404739e-05,
+      "loss": 0.0653,
+      "step": 7044
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.639863015376296e-05,
+      "loss": 0.0731,
+      "step": 7045
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.639615439956686e-05,
+      "loss": 0.0691,
+      "step": 7046
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6393677981715976e-05,
+      "loss": 0.0808,
+      "step": 7047
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6391200900467245e-05,
+      "loss": 0.0778,
+      "step": 7048
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.63887231560777e-05,
+      "loss": 0.0801,
+      "step": 7049
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.638624474880442e-05,
+      "loss": 0.0662,
+      "step": 7050
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6383765678904563e-05,
+      "loss": 0.0789,
+      "step": 7051
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6381285946635346e-05,
+      "loss": 0.0793,
+      "step": 7052
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.637880555225407e-05,
+      "loss": 0.0828,
+      "step": 7053
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6376324496018096e-05,
+      "loss": 0.0661,
+      "step": 7054
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6373842778184856e-05,
+      "loss": 0.0592,
+      "step": 7055
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6371360399011842e-05,
+      "loss": 0.0803,
+      "step": 7056
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.636887735875663e-05,
+      "loss": 0.0704,
+      "step": 7057
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.636639365767685e-05,
+      "loss": 0.0918,
+      "step": 7058
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6363909296030208e-05,
+      "loss": 0.0636,
+      "step": 7059
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.636142427407448e-05,
+      "loss": 0.0639,
+      "step": 7060
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6358938592067506e-05,
+      "loss": 0.0682,
+      "step": 7061
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6356452250267198e-05,
+      "loss": 0.074,
+      "step": 7062
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.635396524893153e-05,
+      "loss": 0.0754,
+      "step": 7063
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6351477588318556e-05,
+      "loss": 0.0619,
+      "step": 7064
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6348989268686385e-05,
+      "loss": 0.0677,
+      "step": 7065
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.634650029029321e-05,
+      "loss": 0.0624,
+      "step": 7066
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6344010653397274e-05,
+      "loss": 0.0693,
+      "step": 7067
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.634152035825691e-05,
+      "loss": 0.0829,
+      "step": 7068
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.633902940513049e-05,
+      "loss": 0.0775,
+      "step": 7069
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6336537794276482e-05,
+      "loss": 0.0703,
+      "step": 7070
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6334045525953412e-05,
+      "loss": 0.0742,
+      "step": 7071
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.633155260041987e-05,
+      "loss": 0.0762,
+      "step": 7072
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.632905901793452e-05,
+      "loss": 0.058,
+      "step": 7073
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6326564778756097e-05,
+      "loss": 0.0742,
+      "step": 7074
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.632406988314339e-05,
+      "loss": 0.0813,
+      "step": 7075
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6321574331355266e-05,
+      "loss": 0.082,
+      "step": 7076
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.631907812365067e-05,
+      "loss": 0.0854,
+      "step": 7077
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.631658126028859e-05,
+      "loss": 0.0754,
+      "step": 7078
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.631408374152811e-05,
+      "loss": 0.0793,
+      "step": 7079
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.631158556762836e-05,
+      "loss": 0.0732,
+      "step": 7080
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.630908673884855e-05,
+      "loss": 0.0735,
+      "step": 7081
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6306587255447954e-05,
+      "loss": 0.0776,
+      "step": 7082
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.630408711768591e-05,
+      "loss": 0.071,
+      "step": 7083
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.630158632582184e-05,
+      "loss": 0.0683,
+      "step": 7084
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.629908488011521e-05,
+      "loss": 0.0819,
+      "step": 7085
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.629658278082557e-05,
+      "loss": 0.0801,
+      "step": 7086
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6294080028212532e-05,
+      "loss": 0.0815,
+      "step": 7087
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6291576622535788e-05,
+      "loss": 0.0748,
+      "step": 7088
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6289072564055075e-05,
+      "loss": 0.0769,
+      "step": 7089
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6286567853030212e-05,
+      "loss": 0.0607,
+      "step": 7090
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6284062489721092e-05,
+      "loss": 0.0703,
+      "step": 7091
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6281556474387656e-05,
+      "loss": 0.0673,
+      "step": 7092
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6279049807289936e-05,
+      "loss": 0.0857,
+      "step": 7093
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.627654248868801e-05,
+      "loss": 0.0779,
+      "step": 7094
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6274034518842036e-05,
+      "loss": 0.0923,
+      "step": 7095
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6271525898012242e-05,
+      "loss": 0.0631,
+      "step": 7096
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6269016626458914e-05,
+      "loss": 0.0698,
+      "step": 7097
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.626650670444241e-05,
+      "loss": 0.075,
+      "step": 7098
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6263996132223155e-05,
+      "loss": 0.0802,
+      "step": 7099
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6261484910061644e-05,
+      "loss": 0.0736,
+      "step": 7100
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6258973038218436e-05,
+      "loss": 0.088,
+      "step": 7101
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.625646051695416e-05,
+      "loss": 0.0804,
+      "step": 7102
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.625394734652951e-05,
+      "loss": 0.0702,
+      "step": 7103
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.625143352720525e-05,
+      "loss": 0.0889,
+      "step": 7104
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.624891905924221e-05,
+      "loss": 0.0906,
+      "step": 7105
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6246403942901284e-05,
+      "loss": 0.0718,
+      "step": 7106
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.624388817844344e-05,
+      "loss": 0.0962,
+      "step": 7107
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6241371766129707e-05,
+      "loss": 0.0735,
+      "step": 7108
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6238854706221186e-05,
+      "loss": 0.0746,
+      "step": 7109
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6236336998979044e-05,
+      "loss": 0.0678,
+      "step": 7110
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6233818644664514e-05,
+      "loss": 0.0813,
+      "step": 7111
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6231299643538895e-05,
+      "loss": 0.0704,
+      "step": 7112
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.622877999586355e-05,
+      "loss": 0.0756,
+      "step": 7113
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6226259701899922e-05,
+      "loss": 0.0776,
+      "step": 7114
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.622373876190951e-05,
+      "loss": 0.0604,
+      "step": 7115
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6221217176153877e-05,
+      "loss": 0.0804,
+      "step": 7116
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6218694944894666e-05,
+      "loss": 0.0769,
+      "step": 7117
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.621617206839358e-05,
+      "loss": 0.0675,
+      "step": 7118
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.621364854691238e-05,
+      "loss": 0.0745,
+      "step": 7119
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6211124380712914e-05,
+      "loss": 0.0765,
+      "step": 7120
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.620859957005708e-05,
+      "loss": 0.0879,
+      "step": 7121
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6206074115206845e-05,
+      "loss": 0.083,
+      "step": 7122
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.620354801642425e-05,
+      "loss": 0.075,
+      "step": 7123
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6201021273971398e-05,
+      "loss": 0.0738,
+      "step": 7124
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.619849388811046e-05,
+      "loss": 0.0698,
+      "step": 7125
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6195965859103675e-05,
+      "loss": 0.0873,
+      "step": 7126
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6193437187213344e-05,
+      "loss": 0.0641,
+      "step": 7127
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.619090787270184e-05,
+      "loss": 0.0731,
+      "step": 7128
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6188377915831605e-05,
+      "loss": 0.0708,
+      "step": 7129
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6185847316865134e-05,
+      "loss": 0.0763,
+      "step": 7130
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6183316076065004e-05,
+      "loss": 0.0613,
+      "step": 7131
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6180784193693852e-05,
+      "loss": 0.0726,
+      "step": 7132
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.617825167001438e-05,
+      "loss": 0.0762,
+      "step": 7133
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.617571850528936e-05,
+      "loss": 0.0757,
+      "step": 7134
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6173184699781632e-05,
+      "loss": 0.0684,
+      "step": 7135
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6170650253754097e-05,
+      "loss": 0.0529,
+      "step": 7136
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.616811516746972e-05,
+      "loss": 0.0698,
+      "step": 7137
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6165579441191546e-05,
+      "loss": 0.0815,
+      "step": 7138
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6163043075182673e-05,
+      "loss": 0.0626,
+      "step": 7139
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6160506069706273e-05,
+      "loss": 0.0759,
+      "step": 7140
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6157968425025577e-05,
+      "loss": 0.0801,
+      "step": 7141
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.615543014140389e-05,
+      "loss": 0.0648,
+      "step": 7142
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6152891219104585e-05,
+      "loss": 0.0612,
+      "step": 7143
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6150351658391086e-05,
+      "loss": 0.0822,
+      "step": 7144
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6147811459526902e-05,
+      "loss": 0.0687,
+      "step": 7145
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6145270622775598e-05,
+      "loss": 0.0689,
+      "step": 7146
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.614272914840081e-05,
+      "loss": 0.0643,
+      "step": 7147
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.614018703666623e-05,
+      "loss": 0.082,
+      "step": 7148
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6137644287835632e-05,
+      "loss": 0.0723,
+      "step": 7149
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6135100902172838e-05,
+      "loss": 0.0756,
+      "step": 7150
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.613255687994175e-05,
+      "loss": 0.0726,
+      "step": 7151
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6130012221406338e-05,
+      "loss": 0.0786,
+      "step": 7152
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6127466926830625e-05,
+      "loss": 0.0703,
+      "step": 7153
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6124920996478707e-05,
+      "loss": 0.0715,
+      "step": 7154
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6122374430614747e-05,
+      "loss": 0.072,
+      "step": 7155
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6119827229502972e-05,
+      "loss": 0.0774,
+      "step": 7156
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6117279393407673e-05,
+      "loss": 0.074,
+      "step": 7157
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6114730922593215e-05,
+      "loss": 0.0635,
+      "step": 7158
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.611218181732402e-05,
+      "loss": 0.0767,
+      "step": 7159
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.610963207786458e-05,
+      "loss": 0.0817,
+      "step": 7160
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6107081704479452e-05,
+      "loss": 0.0721,
+      "step": 7161
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6104530697433258e-05,
+      "loss": 0.0728,
+      "step": 7162
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6101979056990686e-05,
+      "loss": 0.0809,
+      "step": 7163
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.609942678341649e-05,
+      "loss": 0.07,
+      "step": 7164
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6096873876975492e-05,
+      "loss": 0.0768,
+      "step": 7165
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6094320337932577e-05,
+      "loss": 0.0723,
+      "step": 7166
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6091766166552693e-05,
+      "loss": 0.069,
+      "step": 7167
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6089211363100858e-05,
+      "loss": 0.0741,
+      "step": 7168
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.6086655927842157e-05,
+      "loss": 0.074,
+      "step": 7169
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6084099861041735e-05,
+      "loss": 0.0643,
+      "step": 7170
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.608154316296481e-05,
+      "loss": 0.0618,
+      "step": 7171
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6078985833876654e-05,
+      "loss": 0.0729,
+      "step": 7172
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6076427874042617e-05,
+      "loss": 0.0894,
+      "step": 7173
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6073869283728103e-05,
+      "loss": 0.0887,
+      "step": 7174
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6071310063198596e-05,
+      "loss": 0.0693,
+      "step": 7175
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.606875021271963e-05,
+      "loss": 0.0679,
+      "step": 7176
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6066189732556812e-05,
+      "loss": 0.0729,
+      "step": 7177
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6063628622975812e-05,
+      "loss": 0.0819,
+      "step": 7178
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6061066884242374e-05,
+      "loss": 0.0691,
+      "step": 7179
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6058504516622288e-05,
+      "loss": 0.0947,
+      "step": 7180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6055941520381432e-05,
+      "loss": 0.0662,
+      "step": 7181
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6053377895785733e-05,
+      "loss": 0.0737,
+      "step": 7182
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6050813643101194e-05,
+      "loss": 0.0716,
+      "step": 7183
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6048248762593867e-05,
+      "loss": 0.0856,
+      "step": 7184
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.604568325452989e-05,
+      "loss": 0.0796,
+      "step": 7185
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.604311711917545e-05,
+      "loss": 0.0601,
+      "step": 7186
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.604055035679681e-05,
+      "loss": 0.0812,
+      "step": 7187
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.603798296766029e-05,
+      "loss": 0.0747,
+      "step": 7188
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6035414952032277e-05,
+      "loss": 0.0724,
+      "step": 7189
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.603284631017923e-05,
+      "loss": 0.0917,
+      "step": 7190
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6030277042367658e-05,
+      "loss": 0.0607,
+      "step": 7191
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6027707148864155e-05,
+      "loss": 0.0784,
+      "step": 7192
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6025136629935362e-05,
+      "loss": 0.0753,
+      "step": 7193
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.602256548584799e-05,
+      "loss": 0.0682,
+      "step": 7194
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.601999371686883e-05,
+      "loss": 0.0739,
+      "step": 7195
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6017421323264704e-05,
+      "loss": 0.0718,
+      "step": 7196
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6014848305302535e-05,
+      "loss": 0.088,
+      "step": 7197
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6012274663249293e-05,
+      "loss": 0.0704,
+      "step": 7198
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.600970039737201e-05,
+      "loss": 0.0628,
+      "step": 7199
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6007125507937787e-05,
+      "loss": 0.0715,
+      "step": 7200
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.60045499952138e-05,
+      "loss": 0.0634,
+      "step": 7201
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.600197385946727e-05,
+      "loss": 0.0591,
+      "step": 7202
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.5999397100965497e-05,
+      "loss": 0.0681,
+      "step": 7203
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.599681971997584e-05,
+      "loss": 0.0569,
+      "step": 7204
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.599424171676573e-05,
+      "loss": 0.0822,
+      "step": 7205
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.599166309160265e-05,
+      "loss": 0.0695,
+      "step": 7206
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.5989083844754153e-05,
+      "loss": 0.0762,
+      "step": 7207
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.598650397648786e-05,
+      "loss": 0.0649,
+      "step": 7208
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.5983923487071457e-05,
+      "loss": 0.0629,
+      "step": 7209
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5981342376772687e-05,
+      "loss": 0.0955,
+      "step": 7210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.597876064585936e-05,
+      "loss": 0.0732,
+      "step": 7211
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.597617829459936e-05,
+      "loss": 0.0721,
+      "step": 7212
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.597359532326062e-05,
+      "loss": 0.0742,
+      "step": 7213
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5971011732111153e-05,
+      "loss": 0.068,
+      "step": 7214
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.596842752141902e-05,
+      "loss": 0.0839,
+      "step": 7215
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.596584269145236e-05,
+      "loss": 0.0658,
+      "step": 7216
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5963257242479365e-05,
+      "loss": 0.0742,
+      "step": 7217
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5960671174768306e-05,
+      "loss": 0.0657,
+      "step": 7218
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.59580844885875e-05,
+      "loss": 0.0745,
+      "step": 7219
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5955497184205348e-05,
+      "loss": 0.0646,
+      "step": 7220
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5952909261890294e-05,
+      "loss": 0.0573,
+      "step": 7221
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5950320721910863e-05,
+      "loss": 0.0887,
+      "step": 7222
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5947731564535636e-05,
+      "loss": 0.0762,
+      "step": 7223
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5945141790033257e-05,
+      "loss": 0.0742,
+      "step": 7224
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5942551398672443e-05,
+      "loss": 0.0729,
+      "step": 7225
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5939960390721964e-05,
+      "loss": 0.0823,
+      "step": 7226
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.593736876645066e-05,
+      "loss": 0.0706,
+      "step": 7227
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5934776526127437e-05,
+      "loss": 0.0718,
+      "step": 7228
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.5932183670021255e-05,
+      "loss": 0.0565,
+      "step": 7229
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.592959019840115e-05,
+      "loss": 0.0648,
+      "step": 7230
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5926996111536212e-05,
+      "loss": 0.0629,
+      "step": 7231
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.592440140969561e-05,
+      "loss": 0.08,
+      "step": 7232
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5921806093148552e-05,
+      "loss": 0.0682,
+      "step": 7233
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.591921016216433e-05,
+      "loss": 0.0626,
+      "step": 7234
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.59166136170123e-05,
+      "loss": 0.0604,
+      "step": 7235
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5914016457961862e-05,
+      "loss": 0.0651,
+      "step": 7236
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5911418685282506e-05,
+      "loss": 0.0756,
+      "step": 7237
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5908820299243764e-05,
+      "loss": 0.0726,
+      "step": 7238
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5906221300115246e-05,
+      "loss": 0.0663,
+      "step": 7239
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5903621688166614e-05,
+      "loss": 0.0855,
+      "step": 7240
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5901021463667604e-05,
+      "loss": 0.0688,
+      "step": 7241
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5898420626888015e-05,
+      "loss": 0.08,
+      "step": 7242
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.58958191780977e-05,
+      "loss": 0.0715,
+      "step": 7243
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5893217117566576e-05,
+      "loss": 0.0889,
+      "step": 7244
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.589061444556464e-05,
+      "loss": 0.078,
+      "step": 7245
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.588801116236194e-05,
+      "loss": 0.0781,
+      "step": 7246
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5885407268228576e-05,
+      "loss": 0.0683,
+      "step": 7247
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5882802763434738e-05,
+      "loss": 0.0689,
+      "step": 7248
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.5880197648250658e-05,
+      "loss": 0.0638,
+      "step": 7249
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.587759192294664e-05,
+      "loss": 0.0808,
+      "step": 7250
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5874985587793047e-05,
+      "loss": 0.0622,
+      "step": 7251
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.587237864306032e-05,
+      "loss": 0.0751,
+      "step": 7252
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5869771089018933e-05,
+      "loss": 0.076,
+      "step": 7253
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5867162925939456e-05,
+      "loss": 0.0618,
+      "step": 7254
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5864554154092503e-05,
+      "loss": 0.0563,
+      "step": 7255
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.586194477374876e-05,
+      "loss": 0.0975,
+      "step": 7256
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5859334785178965e-05,
+      "loss": 0.0851,
+      "step": 7257
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5856724188653928e-05,
+      "loss": 0.0718,
+      "step": 7258
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5854112984444525e-05,
+      "loss": 0.071,
+      "step": 7259
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5851501172821686e-05,
+      "loss": 0.0776,
+      "step": 7260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5848888754056408e-05,
+      "loss": 0.0635,
+      "step": 7261
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5846275728419755e-05,
+      "loss": 0.0582,
+      "step": 7262
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.584366209618285e-05,
+      "loss": 0.0817,
+      "step": 7263
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5841047857616876e-05,
+      "loss": 0.0814,
+      "step": 7264
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.5838433012993087e-05,
+      "loss": 0.0875,
+      "step": 7265
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.583581756258279e-05,
+      "loss": 0.0607,
+      "step": 7266
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.583320150665736e-05,
+      "loss": 0.0559,
+      "step": 7267
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.583058484548824e-05,
+      "loss": 0.0753,
+      "step": 7268
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.582796757934693e-05,
+      "loss": 0.0607,
+      "step": 7269
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5825349708504988e-05,
+      "loss": 0.0627,
+      "step": 7270
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5822731233234044e-05,
+      "loss": 0.0757,
+      "step": 7271
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5820112153805785e-05,
+      "loss": 0.0902,
+      "step": 7272
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5817492470491962e-05,
+      "loss": 0.06,
+      "step": 7273
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5814872183564393e-05,
+      "loss": 0.0696,
+      "step": 7274
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.581225129329495e-05,
+      "loss": 0.0646,
+      "step": 7275
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5809629799955576e-05,
+      "loss": 0.0903,
+      "step": 7276
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5807007703818275e-05,
+      "loss": 0.0801,
+      "step": 7277
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5804385005155107e-05,
+      "loss": 0.0607,
+      "step": 7278
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5801761704238197e-05,
+      "loss": 0.0944,
+      "step": 7279
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.579913780133974e-05,
+      "loss": 0.0775,
+      "step": 7280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5796513296731985e-05,
+      "loss": 0.0618,
+      "step": 7281
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5793888190687247e-05,
+      "loss": 0.0624,
+      "step": 7282
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5791262483477904e-05,
+      "loss": 0.0789,
+      "step": 7283
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5788636175376395e-05,
+      "loss": 0.0685,
+      "step": 7284
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.578600926665522e-05,
+      "loss": 0.0704,
+      "step": 7285
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5783381757586946e-05,
+      "loss": 0.0732,
+      "step": 7286
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5780753648444194e-05,
+      "loss": 0.0684,
+      "step": 7287
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.5778124939499654e-05,
+      "loss": 0.0763,
+      "step": 7288
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.577549563102608e-05,
+      "loss": 0.0735,
+      "step": 7289
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.577286572329628e-05,
+      "loss": 0.0856,
+      "step": 7290
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5770235216583136e-05,
+      "loss": 0.0554,
+      "step": 7291
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5767604111159578e-05,
+      "loss": 0.0682,
+      "step": 7292
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5764972407298607e-05,
+      "loss": 0.0784,
+      "step": 7293
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.576234010527329e-05,
+      "loss": 0.0717,
+      "step": 7294
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5759707205356745e-05,
+      "loss": 0.0641,
+      "step": 7295
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5757073707822155e-05,
+      "loss": 0.0665,
+      "step": 7296
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5754439612942774e-05,
+      "loss": 0.0676,
+      "step": 7297
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5751804920991912e-05,
+      "loss": 0.076,
+      "step": 7298
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5749169632242932e-05,
+      "loss": 0.0723,
+      "step": 7299
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5746533746969275e-05,
+      "loss": 0.0861,
+      "step": 7300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.574389726544443e-05,
+      "loss": 0.0767,
+      "step": 7301
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5741260187941963e-05,
+      "loss": 0.0696,
+      "step": 7302
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5738622514735483e-05,
+      "loss": 0.0696,
+      "step": 7303
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.573598424609868e-05,
+      "loss": 0.0756,
+      "step": 7304
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.573334538230529e-05,
+      "loss": 0.0795,
+      "step": 7305
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5730705923629116e-05,
+      "loss": 0.0746,
+      "step": 7306
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5728065870344033e-05,
+      "loss": 0.0599,
+      "step": 7307
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.572542522272396e-05,
+      "loss": 0.0656,
+      "step": 7308
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.5722783981042892e-05,
+      "loss": 0.083,
+      "step": 7309
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5720142145574877e-05,
+      "loss": 0.0569,
+      "step": 7310
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5717499716594024e-05,
+      "loss": 0.0865,
+      "step": 7311
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5714856694374514e-05,
+      "loss": 0.0734,
+      "step": 7312
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5712213079190585e-05,
+      "loss": 0.0543,
+      "step": 7313
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.570956887131652e-05,
+      "loss": 0.0576,
+      "step": 7314
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5706924071026693e-05,
+      "loss": 0.0801,
+      "step": 7315
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.570427867859552e-05,
+      "loss": 0.0659,
+      "step": 7316
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.570163269429748e-05,
+      "loss": 0.0824,
+      "step": 7317
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5698986118407113e-05,
+      "loss": 0.0762,
+      "step": 7318
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.569633895119903e-05,
+      "loss": 0.0657,
+      "step": 7319
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5693691192947902e-05,
+      "loss": 0.0634,
+      "step": 7320
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.569104284392844e-05,
+      "loss": 0.0763,
+      "step": 7321
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5688393904415446e-05,
+      "loss": 0.0662,
+      "step": 7322
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5685744374683762e-05,
+      "loss": 0.079,
+      "step": 7323
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5683094255008304e-05,
+      "loss": 0.0684,
+      "step": 7324
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5680443545664043e-05,
+      "loss": 0.0711,
+      "step": 7325
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.567779224692601e-05,
+      "loss": 0.0657,
+      "step": 7326
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5675140359069302e-05,
+      "loss": 0.0747,
+      "step": 7327
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.5672487882369072e-05,
+      "loss": 0.0806,
+      "step": 7328
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.566983481710054e-05,
+      "loss": 0.0786,
+      "step": 7329
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.566718116353898e-05,
+      "loss": 0.0812,
+      "step": 7330
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.566452692195974e-05,
+      "loss": 0.0652,
+      "step": 7331
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5661872092638208e-05,
+      "loss": 0.0833,
+      "step": 7332
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.565921667584985e-05,
+      "loss": 0.0809,
+      "step": 7333
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.565656067187019e-05,
+      "loss": 0.0746,
+      "step": 7334
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5653904080974805e-05,
+      "loss": 0.0734,
+      "step": 7335
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5651246903439344e-05,
+      "loss": 0.075,
+      "step": 7336
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5648589139539514e-05,
+      "loss": 0.0722,
+      "step": 7337
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5645930789551073e-05,
+      "loss": 0.0779,
+      "step": 7338
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5643271853749848e-05,
+      "loss": 0.0811,
+      "step": 7339
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5640612332411728e-05,
+      "loss": 0.0646,
+      "step": 7340
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5637952225812662e-05,
+      "loss": 0.0642,
+      "step": 7341
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.563529153422866e-05,
+      "loss": 0.0657,
+      "step": 7342
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5632630257935785e-05,
+      "loss": 0.0598,
+      "step": 7343
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5629968397210175e-05,
+      "loss": 0.0789,
+      "step": 7344
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.562730595232801e-05,
+      "loss": 0.0698,
+      "step": 7345
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.562464292356555e-05,
+      "loss": 0.0709,
+      "step": 7346
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.56219793111991e-05,
+      "loss": 0.0671,
+      "step": 7347
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5619315115505037e-05,
+      "loss": 0.0681,
+      "step": 7348
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.5616650336759794e-05,
+      "loss": 0.0736,
+      "step": 7349
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.561398497523986e-05,
+      "loss": 0.0697,
+      "step": 7350
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5611319031221793e-05,
+      "loss": 0.0742,
+      "step": 7351
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5608652504982206e-05,
+      "loss": 0.0818,
+      "step": 7352
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5605985396797773e-05,
+      "loss": 0.0752,
+      "step": 7353
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5603317706945224e-05,
+      "loss": 0.0648,
+      "step": 7354
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.560064943570136e-05,
+      "loss": 0.0662,
+      "step": 7355
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5597980583343036e-05,
+      "loss": 0.0694,
+      "step": 7356
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5595311150147167e-05,
+      "loss": 0.0745,
+      "step": 7357
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5592641136390732e-05,
+      "loss": 0.0692,
+      "step": 7358
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5589970542350764e-05,
+      "loss": 0.0651,
+      "step": 7359
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5587299368304362e-05,
+      "loss": 0.0787,
+      "step": 7360
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5584627614528675e-05,
+      "loss": 0.0566,
+      "step": 7361
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.558195528130093e-05,
+      "loss": 0.0721,
+      "step": 7362
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.55792823688984e-05,
+      "loss": 0.0685,
+      "step": 7363
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.557660887759842e-05,
+      "loss": 0.0629,
+      "step": 7364
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.557393480767839e-05,
+      "loss": 0.063,
+      "step": 7365
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.557126015941577e-05,
+      "loss": 0.0792,
+      "step": 7366
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5568584933088067e-05,
+      "loss": 0.0962,
+      "step": 7367
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5565909128972873e-05,
+      "loss": 0.0607,
+      "step": 7368
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.5563232747347813e-05,
+      "loss": 0.0736,
+      "step": 7369
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.556055578849059e-05,
+      "loss": 0.0673,
+      "step": 7370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5557878252678956e-05,
+      "loss": 0.0576,
+      "step": 7371
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5555200140190732e-05,
+      "loss": 0.0588,
+      "step": 7372
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5552521451303794e-05,
+      "loss": 0.0808,
+      "step": 7373
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.554984218629608e-05,
+      "loss": 0.0739,
+      "step": 7374
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5547162345445584e-05,
+      "loss": 0.0654,
+      "step": 7375
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5544481929030363e-05,
+      "loss": 0.0792,
+      "step": 7376
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5541800937328534e-05,
+      "loss": 0.063,
+      "step": 7377
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5539119370618267e-05,
+      "loss": 0.0764,
+      "step": 7378
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5536437229177802e-05,
+      "loss": 0.0754,
+      "step": 7379
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.553375451328543e-05,
+      "loss": 0.0704,
+      "step": 7380
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5531071223219513e-05,
+      "loss": 0.0754,
+      "step": 7381
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5528387359258457e-05,
+      "loss": 0.0768,
+      "step": 7382
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.552570292168074e-05,
+      "loss": 0.0856,
+      "step": 7383
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5523017910764892e-05,
+      "loss": 0.0773,
+      "step": 7384
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5520332326789508e-05,
+      "loss": 0.0759,
+      "step": 7385
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5517646170033236e-05,
+      "loss": 0.064,
+      "step": 7386
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.551495944077479e-05,
+      "loss": 0.0855,
+      "step": 7387
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5512272139292942e-05,
+      "loss": 0.0657,
+      "step": 7388
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5509584265866523e-05,
+      "loss": 0.0696,
+      "step": 7389
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.5506895820774416e-05,
+      "loss": 0.0667,
+      "step": 7390
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5504206804295577e-05,
+      "loss": 0.0637,
+      "step": 7391
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.550151721670901e-05,
+      "loss": 0.0615,
+      "step": 7392
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5498827058293785e-05,
+      "loss": 0.0811,
+      "step": 7393
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5496136329329026e-05,
+      "loss": 0.0706,
+      "step": 7394
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5493445030093918e-05,
+      "loss": 0.0765,
+      "step": 7395
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.549075316086771e-05,
+      "loss": 0.0732,
+      "step": 7396
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.54880607219297e-05,
+      "loss": 0.0795,
+      "step": 7397
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.548536771355926e-05,
+      "loss": 0.0719,
+      "step": 7398
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5482674136035804e-05,
+      "loss": 0.0793,
+      "step": 7399
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5479979989638815e-05,
+      "loss": 0.0724,
+      "step": 7400
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.547728527464784e-05,
+      "loss": 0.0635,
+      "step": 7401
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5474589991342468e-05,
+      "loss": 0.0859,
+      "step": 7402
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5471894140002366e-05,
+      "loss": 0.0676,
+      "step": 7403
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5469197720907244e-05,
+      "loss": 0.0707,
+      "step": 7404
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5466500734336886e-05,
+      "loss": 0.0745,
+      "step": 7405
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5463803180571118e-05,
+      "loss": 0.0751,
+      "step": 7406
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.546110505988984e-05,
+      "loss": 0.0799,
+      "step": 7407
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5458406372573006e-05,
+      "loss": 0.0756,
+      "step": 7408
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5455707118900628e-05,
+      "loss": 0.0717,
+      "step": 7409
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.5453007299152767e-05,
+      "loss": 0.0757,
+      "step": 7410
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5450306913609557e-05,
+      "loss": 0.0715,
+      "step": 7411
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5447605962551194e-05,
+      "loss": 0.0679,
+      "step": 7412
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5444904446257912e-05,
+      "loss": 0.0731,
+      "step": 7413
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5442202365010022e-05,
+      "loss": 0.083,
+      "step": 7414
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5439499719087886e-05,
+      "loss": 0.0798,
+      "step": 7415
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.543679650877193e-05,
+      "loss": 0.0679,
+      "step": 7416
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.543409273434263e-05,
+      "loss": 0.0829,
+      "step": 7417
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5431388396080527e-05,
+      "loss": 0.065,
+      "step": 7418
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.542868349426622e-05,
+      "loss": 0.0715,
+      "step": 7419
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5425978029180367e-05,
+      "loss": 0.0577,
+      "step": 7420
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5423272001103676e-05,
+      "loss": 0.0691,
+      "step": 7421
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5420565410316926e-05,
+      "loss": 0.0715,
+      "step": 7422
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5417858257100946e-05,
+      "loss": 0.0708,
+      "step": 7423
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5415150541736626e-05,
+      "loss": 0.0736,
+      "step": 7424
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.541244226450492e-05,
+      "loss": 0.0637,
+      "step": 7425
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5409733425686822e-05,
+      "loss": 0.0706,
+      "step": 7426
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.540702402556341e-05,
+      "loss": 0.0523,
+      "step": 7427
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.54043140644158e-05,
+      "loss": 0.0751,
+      "step": 7428
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.5401603542525172e-05,
+      "loss": 0.0608,
+      "step": 7429
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.539889246017277e-05,
+      "loss": 0.0723,
+      "step": 7430
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.539618081763989e-05,
+      "loss": 0.0886,
+      "step": 7431
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5393468615207887e-05,
+      "loss": 0.0856,
+      "step": 7432
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5390755853158177e-05,
+      "loss": 0.0646,
+      "step": 7433
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5388042531772228e-05,
+      "loss": 0.0759,
+      "step": 7434
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.538532865133157e-05,
+      "loss": 0.0721,
+      "step": 7435
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5382614212117798e-05,
+      "loss": 0.0707,
+      "step": 7436
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.537989921441255e-05,
+      "loss": 0.0726,
+      "step": 7437
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.537718365849753e-05,
+      "loss": 0.0548,
+      "step": 7438
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5374467544654504e-05,
+      "loss": 0.0811,
+      "step": 7439
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.537175087316529e-05,
+      "loss": 0.0577,
+      "step": 7440
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5369033644311768e-05,
+      "loss": 0.0632,
+      "step": 7441
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.536631585837587e-05,
+      "loss": 0.0774,
+      "step": 7442
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.536359751563959e-05,
+      "loss": 0.0836,
+      "step": 7443
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5360878616384975e-05,
+      "loss": 0.0923,
+      "step": 7444
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5358159160894138e-05,
+      "loss": 0.0718,
+      "step": 7445
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5355439149449245e-05,
+      "loss": 0.078,
+      "step": 7446
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5352718582332524e-05,
+      "loss": 0.0885,
+      "step": 7447
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.534999745982625e-05,
+      "loss": 0.0714,
+      "step": 7448
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5347275782212765e-05,
+      "loss": 0.0733,
+      "step": 7449
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.5344553549774466e-05,
+      "loss": 0.0715,
+      "step": 7450
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5341830762793808e-05,
+      "loss": 0.0763,
+      "step": 7451
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.53391074215533e-05,
+      "loss": 0.0621,
+      "step": 7452
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5336383526335517e-05,
+      "loss": 0.0691,
+      "step": 7453
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5333659077423084e-05,
+      "loss": 0.0768,
+      "step": 7454
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.533093407509868e-05,
+      "loss": 0.0741,
+      "step": 7455
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5328208519645052e-05,
+      "loss": 0.0862,
+      "step": 7456
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5325482411345e-05,
+      "loss": 0.0833,
+      "step": 7457
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.532275575048138e-05,
+      "loss": 0.0706,
+      "step": 7458
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5320028537337108e-05,
+      "loss": 0.0813,
+      "step": 7459
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5317300772195147e-05,
+      "loss": 0.0718,
+      "step": 7460
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5314572455338533e-05,
+      "loss": 0.0641,
+      "step": 7461
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5311843587050352e-05,
+      "loss": 0.0653,
+      "step": 7462
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.530911416761374e-05,
+      "loss": 0.0674,
+      "step": 7463
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.530638419731191e-05,
+      "loss": 0.0664,
+      "step": 7464
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5303653676428106e-05,
+      "loss": 0.0764,
+      "step": 7465
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.530092260524565e-05,
+      "loss": 0.0613,
+      "step": 7466
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.529819098404791e-05,
+      "loss": 0.0886,
+      "step": 7467
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.529545881311832e-05,
+      "loss": 0.0756,
+      "step": 7468
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.5292726092740358e-05,
+      "loss": 0.0588,
+      "step": 7469
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.528999282319757e-05,
+      "loss": 0.0807,
+      "step": 7470
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.528725900477356e-05,
+      "loss": 0.0679,
+      "step": 7471
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5284524637751982e-05,
+      "loss": 0.0845,
+      "step": 7472
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5281789722416542e-05,
+      "loss": 0.0829,
+      "step": 7473
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5279054259051022e-05,
+      "loss": 0.0571,
+      "step": 7474
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5276318247939246e-05,
+      "loss": 0.0742,
+      "step": 7475
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5273581689365093e-05,
+      "loss": 0.0637,
+      "step": 7476
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5270844583612507e-05,
+      "loss": 0.065,
+      "step": 7477
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.526810693096549e-05,
+      "loss": 0.0692,
+      "step": 7478
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.526536873170809e-05,
+      "loss": 0.08,
+      "step": 7479
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5262629986124422e-05,
+      "loss": 0.0746,
+      "step": 7480
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.525989069449865e-05,
+      "loss": 0.0818,
+      "step": 7481
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5257150857115006e-05,
+      "loss": 0.0762,
+      "step": 7482
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5254410474257765e-05,
+      "loss": 0.0918,
+      "step": 7483
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5251669546211265e-05,
+      "loss": 0.0563,
+      "step": 7484
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.52489280732599e-05,
+      "loss": 0.0794,
+      "step": 7485
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5246186055688128e-05,
+      "loss": 0.0615,
+      "step": 7486
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5243443493780445e-05,
+      "loss": 0.0776,
+      "step": 7487
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5240700387821426e-05,
+      "loss": 0.083,
+      "step": 7488
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5237956738095681e-05,
+      "loss": 0.0739,
+      "step": 7489
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.5235212544887891e-05,
+      "loss": 0.09,
+      "step": 7490
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5232467808482794e-05,
+      "loss": 0.0727,
+      "step": 7491
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5229722529165175e-05,
+      "loss": 0.0763,
+      "step": 7492
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5226976707219877e-05,
+      "loss": 0.0724,
+      "step": 7493
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5224230342931807e-05,
+      "loss": 0.0757,
+      "step": 7494
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5221483436585923e-05,
+      "loss": 0.0576,
+      "step": 7495
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5218735988467237e-05,
+      "loss": 0.0795,
+      "step": 7496
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5215987998860824e-05,
+      "loss": 0.0748,
+      "step": 7497
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5213239468051801e-05,
+      "loss": 0.0745,
+      "step": 7498
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5210490396325361e-05,
+      "loss": 0.0804,
+      "step": 7499
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5207740783966744e-05,
+      "loss": 0.0715,
+      "step": 7500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.520499063126124e-05,
+      "loss": 0.0645,
+      "step": 7501
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5202239938494201e-05,
+      "loss": 0.0806,
+      "step": 7502
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5199488705951037e-05,
+      "loss": 0.0596,
+      "step": 7503
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5196736933917211e-05,
+      "loss": 0.0822,
+      "step": 7504
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5193984622678241e-05,
+      "loss": 0.0604,
+      "step": 7505
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5191231772519706e-05,
+      "loss": 0.0612,
+      "step": 7506
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.518847838372723e-05,
+      "loss": 0.0759,
+      "step": 7507
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5185724456586508e-05,
+      "loss": 0.0715,
+      "step": 7508
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.518296999138328e-05,
+      "loss": 0.0681,
+      "step": 7509
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5180214988403343e-05,
+      "loss": 0.079,
+      "step": 7510
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5177459447932554e-05,
+      "loss": 0.087,
+      "step": 7511
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5174703370256823e-05,
+      "loss": 0.0936,
+      "step": 7512
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5171946755662116e-05,
+      "loss": 0.0837,
+      "step": 7513
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5169189604434458e-05,
+      "loss": 0.0771,
+      "step": 7514
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5166431916859923e-05,
+      "loss": 0.0663,
+      "step": 7515
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5163673693224644e-05,
+      "loss": 0.0692,
+      "step": 7516
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5160914933814809e-05,
+      "loss": 0.0726,
+      "step": 7517
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.515815563891667e-05,
+      "loss": 0.0747,
+      "step": 7518
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5155395808816518e-05,
+      "loss": 0.0574,
+      "step": 7519
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5152635443800714e-05,
+      "loss": 0.061,
+      "step": 7520
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5149874544155666e-05,
+      "loss": 0.0746,
+      "step": 7521
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5147113110167841e-05,
+      "loss": 0.0692,
+      "step": 7522
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5144351142123763e-05,
+      "loss": 0.0733,
+      "step": 7523
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5141588640310006e-05,
+      "loss": 0.0823,
+      "step": 7524
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5138825605013208e-05,
+      "loss": 0.071,
+      "step": 7525
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5136062036520054e-05,
+      "loss": 0.1025,
+      "step": 7526
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5133297935117284e-05,
+      "loss": 0.0695,
+      "step": 7527
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.51305333010917e-05,
+      "loss": 0.0851,
+      "step": 7528
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.512776813473016e-05,
+      "loss": 0.0694,
+      "step": 7529
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5125002436319572e-05,
+      "loss": 0.0571,
+      "step": 7530
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5122236206146892e-05,
+      "loss": 0.0709,
+      "step": 7531
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5119469444499148e-05,
+      "loss": 0.0733,
+      "step": 7532
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5116702151663413e-05,
+      "loss": 0.0706,
+      "step": 7533
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5113934327926817e-05,
+      "loss": 0.0601,
+      "step": 7534
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5111165973576545e-05,
+      "loss": 0.0629,
+      "step": 7535
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5108397088899834e-05,
+      "loss": 0.0594,
+      "step": 7536
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.510562767418398e-05,
+      "loss": 0.066,
+      "step": 7537
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5102857729716338e-05,
+      "loss": 0.0558,
+      "step": 7538
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5100087255784308e-05,
+      "loss": 0.0659,
+      "step": 7539
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5097316252675352e-05,
+      "loss": 0.0701,
+      "step": 7540
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5094544720676985e-05,
+      "loss": 0.0856,
+      "step": 7541
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5091772660076775e-05,
+      "loss": 0.0787,
+      "step": 7542
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5089000071162347e-05,
+      "loss": 0.0701,
+      "step": 7543
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5086226954221381e-05,
+      "loss": 0.0773,
+      "step": 7544
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5083453309541613e-05,
+      "loss": 0.0729,
+      "step": 7545
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.508067913741083e-05,
+      "loss": 0.0564,
+      "step": 7546
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5077904438116875e-05,
+      "loss": 0.0753,
+      "step": 7547
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5075129211947647e-05,
+      "loss": 0.0706,
+      "step": 7548
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.50723534591911e-05,
+      "loss": 0.0789,
+      "step": 7549
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5069577180135242e-05,
+      "loss": 0.0789,
+      "step": 7550
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5066800375068132e-05,
+      "loss": 0.06,
+      "step": 7551
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5064023044277891e-05,
+      "loss": 0.0673,
+      "step": 7552
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5061245188052689e-05,
+      "loss": 0.07,
+      "step": 7553
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5058466806680749e-05,
+      "loss": 0.0668,
+      "step": 7554
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5055687900450355e-05,
+      "loss": 0.0731,
+      "step": 7555
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5052908469649843e-05,
+      "loss": 0.0748,
+      "step": 7556
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5050128514567598e-05,
+      "loss": 0.0793,
+      "step": 7557
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5047348035492067e-05,
+      "loss": 0.0826,
+      "step": 7558
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5044567032711746e-05,
+      "loss": 0.0823,
+      "step": 7559
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5041785506515187e-05,
+      "loss": 0.0649,
+      "step": 7560
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5039003457191e-05,
+      "loss": 0.0807,
+      "step": 7561
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5036220885027843e-05,
+      "loss": 0.0667,
+      "step": 7562
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5033437790314436e-05,
+      "loss": 0.0796,
+      "step": 7563
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.503065417333954e-05,
+      "loss": 0.0609,
+      "step": 7564
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5027870034391984e-05,
+      "loss": 0.0618,
+      "step": 7565
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5025085373760649e-05,
+      "loss": 0.0676,
+      "step": 7566
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.502230019173446e-05,
+      "loss": 0.0582,
+      "step": 7567
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5019514488602406e-05,
+      "loss": 0.0793,
+      "step": 7568
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5016728264653531e-05,
+      "loss": 0.0806,
+      "step": 7569
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5013941520176922e-05,
+      "loss": 0.0662,
+      "step": 7570
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5011154255461732e-05,
+      "loss": 0.0879,
+      "step": 7571
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5008366470797162e-05,
+      "loss": 0.0748,
+      "step": 7572
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.500557816647247e-05,
+      "loss": 0.0626,
+      "step": 7573
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.500278934277696e-05,
+      "loss": 0.0762,
+      "step": 7574
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0685,
+      "step": 7575
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4997210138431011e-05,
+      "loss": 0.0717,
+      "step": 7576
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.499441975835946e-05,
+      "loss": 0.0648,
+      "step": 7577
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4991628860074872e-05,
+      "loss": 0.0757,
+      "step": 7578
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4988837443866829e-05,
+      "loss": 0.0697,
+      "step": 7579
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4986045510024965e-05,
+      "loss": 0.0763,
+      "step": 7580
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.498325305883896e-05,
+      "loss": 0.0515,
+      "step": 7581
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4980460090598562e-05,
+      "loss": 0.0645,
+      "step": 7582
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4977666605593557e-05,
+      "loss": 0.0809,
+      "step": 7583
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4974872604113801e-05,
+      "loss": 0.0867,
+      "step": 7584
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.497207808644919e-05,
+      "loss": 0.0801,
+      "step": 7585
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.496928305288968e-05,
+      "loss": 0.0726,
+      "step": 7586
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.496648750372528e-05,
+      "loss": 0.0681,
+      "step": 7587
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.496369143924605e-05,
+      "loss": 0.074,
+      "step": 7588
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4960894859742105e-05,
+      "loss": 0.0798,
+      "step": 7589
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4958097765503614e-05,
+      "loss": 0.051,
+      "step": 7590
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4955300156820805e-05,
+      "loss": 0.0709,
+      "step": 7591
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4952502033983944e-05,
+      "loss": 0.0723,
+      "step": 7592
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4949703397283365e-05,
+      "loss": 0.0698,
+      "step": 7593
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4946904247009446e-05,
+      "loss": 0.0596,
+      "step": 7594
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.494410458345263e-05,
+      "loss": 0.0571,
+      "step": 7595
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.49413044069034e-05,
+      "loss": 0.0786,
+      "step": 7596
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.49385037176523e-05,
+      "loss": 0.0963,
+      "step": 7597
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4935702515989919e-05,
+      "loss": 0.0748,
+      "step": 7598
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.493290080220691e-05,
+      "loss": 0.0835,
+      "step": 7599
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4930098576593978e-05,
+      "loss": 0.0665,
+      "step": 7600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4927295839441875e-05,
+      "loss": 0.0744,
+      "step": 7601
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4924492591041405e-05,
+      "loss": 0.0759,
+      "step": 7602
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4921688831683433e-05,
+      "loss": 0.0687,
+      "step": 7603
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4918884561658869e-05,
+      "loss": 0.0688,
+      "step": 7604
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4916079781258681e-05,
+      "loss": 0.0847,
+      "step": 7605
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.491327449077389e-05,
+      "loss": 0.0723,
+      "step": 7606
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4910468690495565e-05,
+      "loss": 0.0725,
+      "step": 7607
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4907662380714837e-05,
+      "loss": 0.0775,
+      "step": 7608
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4904855561722881e-05,
+      "loss": 0.0718,
+      "step": 7609
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4902048233810926e-05,
+      "loss": 0.0555,
+      "step": 7610
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4899240397270256e-05,
+      "loss": 0.0896,
+      "step": 7611
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4896432052392213e-05,
+      "loss": 0.0735,
+      "step": 7612
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4893623199468184e-05,
+      "loss": 0.081,
+      "step": 7613
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4890813838789606e-05,
+      "loss": 0.0845,
+      "step": 7614
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4888003970647979e-05,
+      "loss": 0.0895,
+      "step": 7615
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4885193595334847e-05,
+      "loss": 0.06,
+      "step": 7616
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4882382713141816e-05,
+      "loss": 0.0687,
+      "step": 7617
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4879571324360533e-05,
+      "loss": 0.0724,
+      "step": 7618
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4876759429282705e-05,
+      "loss": 0.0709,
+      "step": 7619
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4873947028200094e-05,
+      "loss": 0.0812,
+      "step": 7620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4871134121404503e-05,
+      "loss": 0.0837,
+      "step": 7621
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4868320709187796e-05,
+      "loss": 0.0597,
+      "step": 7622
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4865506791841893e-05,
+      "loss": 0.0754,
+      "step": 7623
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4862692369658755e-05,
+      "loss": 0.0667,
+      "step": 7624
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4859877442930408e-05,
+      "loss": 0.0634,
+      "step": 7625
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4857062011948923e-05,
+      "loss": 0.064,
+      "step": 7626
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.485424607700642e-05,
+      "loss": 0.0654,
+      "step": 7627
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.485142963839508e-05,
+      "loss": 0.082,
+      "step": 7628
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4848612696407135e-05,
+      "loss": 0.0717,
+      "step": 7629
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4845795251334863e-05,
+      "loss": 0.0734,
+      "step": 7630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4842977303470596e-05,
+      "loss": 0.0909,
+      "step": 7631
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4840158853106718e-05,
+      "loss": 0.0841,
+      "step": 7632
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4837339900535674e-05,
+      "loss": 0.0599,
+      "step": 7633
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4834520446049952e-05,
+      "loss": 0.0683,
+      "step": 7634
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.483170048994209e-05,
+      "loss": 0.0711,
+      "step": 7635
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4828880032504684e-05,
+      "loss": 0.0748,
+      "step": 7636
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4826059074030381e-05,
+      "loss": 0.0738,
+      "step": 7637
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.482323761481188e-05,
+      "loss": 0.0742,
+      "step": 7638
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4820415655141932e-05,
+      "loss": 0.0764,
+      "step": 7639
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4817593195313334e-05,
+      "loss": 0.0754,
+      "step": 7640
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4814770235618942e-05,
+      "loss": 0.0782,
+      "step": 7641
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4811946776351667e-05,
+      "loss": 0.0693,
+      "step": 7642
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.480912281780446e-05,
+      "loss": 0.0764,
+      "step": 7643
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4806298360270333e-05,
+      "loss": 0.0908,
+      "step": 7644
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4803473404042345e-05,
+      "loss": 0.0717,
+      "step": 7645
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4800647949413615e-05,
+      "loss": 0.0661,
+      "step": 7646
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4797821996677303e-05,
+      "loss": 0.0723,
+      "step": 7647
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4794995546126625e-05,
+      "loss": 0.0782,
+      "step": 7648
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4792168598054847e-05,
+      "loss": 0.0812,
+      "step": 7649
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4789341152755297e-05,
+      "loss": 0.0825,
+      "step": 7650
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4786513210521339e-05,
+      "loss": 0.0703,
+      "step": 7651
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4783684771646397e-05,
+      "loss": 0.0657,
+      "step": 7652
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4780855836423946e-05,
+      "loss": 0.0708,
+      "step": 7653
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4778026405147515e-05,
+      "loss": 0.07,
+      "step": 7654
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4775196478110674e-05,
+      "loss": 0.0748,
+      "step": 7655
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4772366055607057e-05,
+      "loss": 0.0749,
+      "step": 7656
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4769535137930343e-05,
+      "loss": 0.0593,
+      "step": 7657
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4766703725374264e-05,
+      "loss": 0.0715,
+      "step": 7658
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4763871818232604e-05,
+      "loss": 0.0729,
+      "step": 7659
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4761039416799192e-05,
+      "loss": 0.0682,
+      "step": 7660
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4758206521367919e-05,
+      "loss": 0.0579,
+      "step": 7661
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.475537313223272e-05,
+      "loss": 0.0576,
+      "step": 7662
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4752539249687583e-05,
+      "loss": 0.0684,
+      "step": 7663
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4749704874026547e-05,
+      "loss": 0.0684,
+      "step": 7664
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.47468700055437e-05,
+      "loss": 0.0842,
+      "step": 7665
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4744034644533185e-05,
+      "loss": 0.0658,
+      "step": 7666
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4741198791289196e-05,
+      "loss": 0.0659,
+      "step": 7667
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4738362446105975e-05,
+      "loss": 0.0502,
+      "step": 7668
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4735525609277819e-05,
+      "loss": 0.0632,
+      "step": 7669
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4732688281099072e-05,
+      "loss": 0.0601,
+      "step": 7670
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.472985046186413e-05,
+      "loss": 0.0716,
+      "step": 7671
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4727012151867442e-05,
+      "loss": 0.0677,
+      "step": 7672
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4724173351403504e-05,
+      "loss": 0.0635,
+      "step": 7673
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.472133406076687e-05,
+      "loss": 0.0911,
+      "step": 7674
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4718494280252133e-05,
+      "loss": 0.0784,
+      "step": 7675
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4715654010153953e-05,
+      "loss": 0.067,
+      "step": 7676
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4712813250767024e-05,
+      "loss": 0.0615,
+      "step": 7677
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4709972002386104e-05,
+      "loss": 0.0667,
+      "step": 7678
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4707130265305993e-05,
+      "loss": 0.0737,
+      "step": 7679
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4704288039821551e-05,
+      "loss": 0.0695,
+      "step": 7680
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4701445326227675e-05,
+      "loss": 0.0726,
+      "step": 7681
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4698602124819321e-05,
+      "loss": 0.0722,
+      "step": 7682
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.46957584358915e-05,
+      "loss": 0.0751,
+      "step": 7683
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4692914259739268e-05,
+      "loss": 0.0901,
+      "step": 7684
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4690069596657732e-05,
+      "loss": 0.0739,
+      "step": 7685
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4687224446942045e-05,
+      "loss": 0.0933,
+      "step": 7686
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4684378810887422e-05,
+      "loss": 0.0762,
+      "step": 7687
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4681532688789114e-05,
+      "loss": 0.0672,
+      "step": 7688
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4678686080942439e-05,
+      "loss": 0.0698,
+      "step": 7689
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.467583898764275e-05,
+      "loss": 0.0742,
+      "step": 7690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4672991409185457e-05,
+      "loss": 0.0668,
+      "step": 7691
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4670143345866024e-05,
+      "loss": 0.0735,
+      "step": 7692
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4667294797979958e-05,
+      "loss": 0.0817,
+      "step": 7693
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4664445765822823e-05,
+      "loss": 0.0811,
+      "step": 7694
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4661596249690227e-05,
+      "loss": 0.0685,
+      "step": 7695
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4658746249877833e-05,
+      "loss": 0.0587,
+      "step": 7696
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4655895766681351e-05,
+      "loss": 0.0708,
+      "step": 7697
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4653044800396545e-05,
+      "loss": 0.0564,
+      "step": 7698
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4650193351319224e-05,
+      "loss": 0.075,
+      "step": 7699
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4647341419745251e-05,
+      "loss": 0.0633,
+      "step": 7700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4644489005970539e-05,
+      "loss": 0.0805,
+      "step": 7701
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4641636110291051e-05,
+      "loss": 0.0688,
+      "step": 7702
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4638782733002792e-05,
+      "loss": 0.0859,
+      "step": 7703
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.463592887440183e-05,
+      "loss": 0.083,
+      "step": 7704
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4633074534784278e-05,
+      "loss": 0.0767,
+      "step": 7705
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4630219714446292e-05,
+      "loss": 0.069,
+      "step": 7706
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4627364413684091e-05,
+      "loss": 0.0814,
+      "step": 7707
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4624508632793928e-05,
+      "loss": 0.0792,
+      "step": 7708
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4621652372072122e-05,
+      "loss": 0.0621,
+      "step": 7709
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4618795631815027e-05,
+      "loss": 0.0663,
+      "step": 7710
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.461593841231906e-05,
+      "loss": 0.0715,
+      "step": 7711
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4613080713880677e-05,
+      "loss": 0.0659,
+      "step": 7712
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4610222536796393e-05,
+      "loss": 0.0865,
+      "step": 7713
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4607363881362765e-05,
+      "loss": 0.0656,
+      "step": 7714
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4604504747876397e-05,
+      "loss": 0.0829,
+      "step": 7715
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4601645136633959e-05,
+      "loss": 0.0593,
+      "step": 7716
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4598785047932153e-05,
+      "loss": 0.0663,
+      "step": 7717
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4595924482067742e-05,
+      "loss": 0.0703,
+      "step": 7718
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4593063439337524e-05,
+      "loss": 0.0784,
+      "step": 7719
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4590201920038367e-05,
+      "loss": 0.0564,
+      "step": 7720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4587339924467175e-05,
+      "loss": 0.0699,
+      "step": 7721
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.45844774529209e-05,
+      "loss": 0.0825,
+      "step": 7722
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4581614505696551e-05,
+      "loss": 0.0694,
+      "step": 7723
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.457875108309118e-05,
+      "loss": 0.0869,
+      "step": 7724
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4575887185401893e-05,
+      "loss": 0.0656,
+      "step": 7725
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4573022812925845e-05,
+      "loss": 0.0748,
+      "step": 7726
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4570157965960236e-05,
+      "loss": 0.0759,
+      "step": 7727
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.456729264480232e-05,
+      "loss": 0.0697,
+      "step": 7728
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.45644268497494e-05,
+      "loss": 0.103,
+      "step": 7729
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4561560581098819e-05,
+      "loss": 0.0745,
+      "step": 7730
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4558693839147985e-05,
+      "loss": 0.0757,
+      "step": 7731
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4555826624194339e-05,
+      "loss": 0.088,
+      "step": 7732
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4552958936535381e-05,
+      "loss": 0.0695,
+      "step": 7733
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4550090776468664e-05,
+      "loss": 0.0605,
+      "step": 7734
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4547222144291777e-05,
+      "loss": 0.0693,
+      "step": 7735
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4544353040302364e-05,
+      "loss": 0.0776,
+      "step": 7736
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4541483464798125e-05,
+      "loss": 0.0836,
+      "step": 7737
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4538613418076795e-05,
+      "loss": 0.0628,
+      "step": 7738
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4535742900436171e-05,
+      "loss": 0.0866,
+      "step": 7739
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.453287191217409e-05,
+      "loss": 0.0523,
+      "step": 7740
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4530000453588447e-05,
+      "loss": 0.074,
+      "step": 7741
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4527128524977172e-05,
+      "loss": 0.0706,
+      "step": 7742
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4524256126638257e-05,
+      "loss": 0.0651,
+      "step": 7743
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4521383258869735e-05,
+      "loss": 0.0711,
+      "step": 7744
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4518509921969687e-05,
+      "loss": 0.0898,
+      "step": 7745
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4515636116236258e-05,
+      "loss": 0.0916,
+      "step": 7746
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4512761841967615e-05,
+      "loss": 0.0488,
+      "step": 7747
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4509887099462e-05,
+      "loss": 0.0774,
+      "step": 7748
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4507011889017679e-05,
+      "loss": 0.0598,
+      "step": 7749
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.450413621093299e-05,
+      "loss": 0.0751,
+      "step": 7750
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4501260065506306e-05,
+      "loss": 0.0739,
+      "step": 7751
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4498383453036051e-05,
+      "loss": 0.0674,
+      "step": 7752
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4495506373820695e-05,
+      "loss": 0.0537,
+      "step": 7753
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4492628828158764e-05,
+      "loss": 0.0778,
+      "step": 7754
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4489750816348824e-05,
+      "loss": 0.0709,
+      "step": 7755
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4486872338689492e-05,
+      "loss": 0.0742,
+      "step": 7756
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4483993395479439e-05,
+      "loss": 0.0809,
+      "step": 7757
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4481113987017375e-05,
+      "loss": 0.0833,
+      "step": 7758
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4478234113602063e-05,
+      "loss": 0.072,
+      "step": 7759
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4475353775532316e-05,
+      "loss": 0.064,
+      "step": 7760
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4472472973106997e-05,
+      "loss": 0.0738,
+      "step": 7761
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4469591706625003e-05,
+      "loss": 0.0618,
+      "step": 7762
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4466709976385303e-05,
+      "loss": 0.066,
+      "step": 7763
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.446382778268689e-05,
+      "loss": 0.0941,
+      "step": 7764
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.446094512582882e-05,
+      "loss": 0.0681,
+      "step": 7765
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.445806200611019e-05,
+      "loss": 0.0728,
+      "step": 7766
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4455178423830152e-05,
+      "loss": 0.0773,
+      "step": 7767
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.44522943792879e-05,
+      "loss": 0.084,
+      "step": 7768
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.444940987278268e-05,
+      "loss": 0.0829,
+      "step": 7769
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.444652490461378e-05,
+      "loss": 0.0575,
+      "step": 7770
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.444363947508054e-05,
+      "loss": 0.0942,
+      "step": 7771
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4440753584482351e-05,
+      "loss": 0.0623,
+      "step": 7772
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4437867233118647e-05,
+      "loss": 0.0669,
+      "step": 7773
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4434980421288911e-05,
+      "loss": 0.0739,
+      "step": 7774
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4432093149292672e-05,
+      "loss": 0.0593,
+      "step": 7775
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4429205417429513e-05,
+      "loss": 0.0704,
+      "step": 7776
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4426317225999055e-05,
+      "loss": 0.0833,
+      "step": 7777
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4423428575300978e-05,
+      "loss": 0.0726,
+      "step": 7778
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4420539465635e-05,
+      "loss": 0.0641,
+      "step": 7779
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4417649897300891e-05,
+      "loss": 0.0768,
+      "step": 7780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4414759870598467e-05,
+      "loss": 0.0867,
+      "step": 7781
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4411869385827592e-05,
+      "loss": 0.0548,
+      "step": 7782
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4408978443288186e-05,
+      "loss": 0.0679,
+      "step": 7783
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4406087043280199e-05,
+      "loss": 0.0567,
+      "step": 7784
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4403195186103644e-05,
+      "loss": 0.0657,
+      "step": 7785
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4400302872058568e-05,
+      "loss": 0.0644,
+      "step": 7786
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4397410101445082e-05,
+      "loss": 0.0817,
+      "step": 7787
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4394516874563332e-05,
+      "loss": 0.0665,
+      "step": 7788
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4391623191713513e-05,
+      "loss": 0.078,
+      "step": 7789
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4388729053195869e-05,
+      "loss": 0.0645,
+      "step": 7790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.438583445931069e-05,
+      "loss": 0.0822,
+      "step": 7791
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.438293941035832e-05,
+      "loss": 0.0767,
+      "step": 7792
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.438004390663914e-05,
+      "loss": 0.0851,
+      "step": 7793
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4377147948453584e-05,
+      "loss": 0.0884,
+      "step": 7794
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4374251536102131e-05,
+      "loss": 0.0582,
+      "step": 7795
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4371354669885312e-05,
+      "loss": 0.0701,
+      "step": 7796
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4368457350103695e-05,
+      "loss": 0.0753,
+      "step": 7797
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4365559577057905e-05,
+      "loss": 0.086,
+      "step": 7798
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.436266135104861e-05,
+      "loss": 0.0815,
+      "step": 7799
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4359762672376528e-05,
+      "loss": 0.0953,
+      "step": 7800
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4356863541342416e-05,
+      "loss": 0.0667,
+      "step": 7801
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4353963958247086e-05,
+      "loss": 0.0627,
+      "step": 7802
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4351063923391393e-05,
+      "loss": 0.0812,
+      "step": 7803
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4348163437076243e-05,
+      "loss": 0.0752,
+      "step": 7804
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4345262499602581e-05,
+      "loss": 0.075,
+      "step": 7805
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4342361111271408e-05,
+      "loss": 0.0718,
+      "step": 7806
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4339459272383766e-05,
+      "loss": 0.0684,
+      "step": 7807
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4336556983240747e-05,
+      "loss": 0.0754,
+      "step": 7808
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4333654244143482e-05,
+      "loss": 0.0719,
+      "step": 7809
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4330751055393162e-05,
+      "loss": 0.0663,
+      "step": 7810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4327847417291009e-05,
+      "loss": 0.0658,
+      "step": 7811
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.432494333013831e-05,
+      "loss": 0.0785,
+      "step": 7812
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4322038794236379e-05,
+      "loss": 0.0859,
+      "step": 7813
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4319133809886592e-05,
+      "loss": 0.0792,
+      "step": 7814
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4316228377390363e-05,
+      "loss": 0.0814,
+      "step": 7815
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4313322497049153e-05,
+      "loss": 0.0825,
+      "step": 7816
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4310416169164477e-05,
+      "loss": 0.0626,
+      "step": 7817
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4307509394037888e-05,
+      "loss": 0.0674,
+      "step": 7818
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.430460217197099e-05,
+      "loss": 0.066,
+      "step": 7819
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4301694503265426e-05,
+      "loss": 0.0661,
+      "step": 7820
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4298786388222895e-05,
+      "loss": 0.0709,
+      "step": 7821
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4295877827145144e-05,
+      "loss": 0.0743,
+      "step": 7822
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4292968820333953e-05,
+      "loss": 0.0594,
+      "step": 7823
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4290059368091156e-05,
+      "loss": 0.0674,
+      "step": 7824
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4287149470718635e-05,
+      "loss": 0.0868,
+      "step": 7825
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.428423912851832e-05,
+      "loss": 0.0756,
+      "step": 7826
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4281328341792178e-05,
+      "loss": 0.0795,
+      "step": 7827
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.427841711084223e-05,
+      "loss": 0.0649,
+      "step": 7828
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.427550543597054e-05,
+      "loss": 0.0565,
+      "step": 7829
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.427259331747922e-05,
+      "loss": 0.0587,
+      "step": 7830
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4269680755670425e-05,
+      "loss": 0.0637,
+      "step": 7831
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.426676775084636e-05,
+      "loss": 0.0671,
+      "step": 7832
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4263854303309268e-05,
+      "loss": 0.0571,
+      "step": 7833
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4260940413361452e-05,
+      "loss": 0.0847,
+      "step": 7834
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4258026081305252e-05,
+      "loss": 0.0848,
+      "step": 7835
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4255111307443046e-05,
+      "loss": 0.0538,
+      "step": 7836
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.425219609207727e-05,
+      "loss": 0.0635,
+      "step": 7837
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4249280435510407e-05,
+      "loss": 0.0641,
+      "step": 7838
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4246364338044977e-05,
+      "loss": 0.0632,
+      "step": 7839
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.424344779998355e-05,
+      "loss": 0.0692,
+      "step": 7840
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.424053082162874e-05,
+      "loss": 0.0728,
+      "step": 7841
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.423761340328321e-05,
+      "loss": 0.0594,
+      "step": 7842
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4234695545249666e-05,
+      "loss": 0.0683,
+      "step": 7843
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.423177724783086e-05,
+      "loss": 0.0704,
+      "step": 7844
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4228858511329591e-05,
+      "loss": 0.08,
+      "step": 7845
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4225939336048703e-05,
+      "loss": 0.0641,
+      "step": 7846
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.422301972229108e-05,
+      "loss": 0.0691,
+      "step": 7847
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4220099670359664e-05,
+      "loss": 0.0851,
+      "step": 7848
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4217179180557428e-05,
+      "loss": 0.0759,
+      "step": 7849
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4214258253187401e-05,
+      "loss": 0.0801,
+      "step": 7850
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4211336888552657e-05,
+      "loss": 0.0813,
+      "step": 7851
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4208415086956305e-05,
+      "loss": 0.0857,
+      "step": 7852
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4205492848701507e-05,
+      "loss": 0.0717,
+      "step": 7853
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4202570174091474e-05,
+      "loss": 0.0602,
+      "step": 7854
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.419964706342946e-05,
+      "loss": 0.0688,
+      "step": 7855
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4196723517018757e-05,
+      "loss": 0.069,
+      "step": 7856
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4193799535162711e-05,
+      "loss": 0.077,
+      "step": 7857
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4190875118164706e-05,
+      "loss": 0.0788,
+      "step": 7858
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4187950266328179e-05,
+      "loss": 0.0795,
+      "step": 7859
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4185024979956602e-05,
+      "loss": 0.0693,
+      "step": 7860
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4182099259353508e-05,
+      "loss": 0.0748,
+      "step": 7861
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4179173104822454e-05,
+      "loss": 0.07,
+      "step": 7862
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4176246516667061e-05,
+      "loss": 0.058,
+      "step": 7863
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4173319495190984e-05,
+      "loss": 0.0825,
+      "step": 7864
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4170392040697926e-05,
+      "loss": 0.0646,
+      "step": 7865
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4167464153491634e-05,
+      "loss": 0.0538,
+      "step": 7866
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4164535833875905e-05,
+      "loss": 0.0713,
+      "step": 7867
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4161607082154575e-05,
+      "loss": 0.0769,
+      "step": 7868
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4158677898631524e-05,
+      "loss": 0.0817,
+      "step": 7869
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.415574828361068e-05,
+      "loss": 0.0732,
+      "step": 7870
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4152818237396017e-05,
+      "loss": 0.083,
+      "step": 7871
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4149887760291552e-05,
+      "loss": 0.0759,
+      "step": 7872
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.4146956852601349e-05,
+      "loss": 0.0669,
+      "step": 7873
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4144025514629504e-05,
+      "loss": 0.0918,
+      "step": 7874
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4141093746680182e-05,
+      "loss": 0.0676,
+      "step": 7875
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.413816154905757e-05,
+      "loss": 0.0924,
+      "step": 7876
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4135228922065909e-05,
+      "loss": 0.0763,
+      "step": 7877
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4132295866009482e-05,
+      "loss": 0.0653,
+      "step": 7878
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4129362381192626e-05,
+      "loss": 0.0706,
+      "step": 7879
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4126428467919707e-05,
+      "loss": 0.0683,
+      "step": 7880
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4123494126495145e-05,
+      "loss": 0.0597,
+      "step": 7881
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4120559357223407e-05,
+      "loss": 0.0885,
+      "step": 7882
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4117624160408991e-05,
+      "loss": 0.0724,
+      "step": 7883
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4114688536356457e-05,
+      "loss": 0.079,
+      "step": 7884
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4111752485370399e-05,
+      "loss": 0.0674,
+      "step": 7885
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4108816007755452e-05,
+      "loss": 0.0781,
+      "step": 7886
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4105879103816303e-05,
+      "loss": 0.0833,
+      "step": 7887
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4102941773857683e-05,
+      "loss": 0.069,
+      "step": 7888
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.410000401818436e-05,
+      "loss": 0.0617,
+      "step": 7889
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4097065837101161e-05,
+      "loss": 0.0671,
+      "step": 7890
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4094127230912931e-05,
+      "loss": 0.0756,
+      "step": 7891
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4091188199924589e-05,
+      "loss": 0.0645,
+      "step": 7892
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.4088248744441075e-05,
+      "loss": 0.0669,
+      "step": 7893
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4085308864767389e-05,
+      "loss": 0.0543,
+      "step": 7894
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4082368561208564e-05,
+      "loss": 0.0709,
+      "step": 7895
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.407942783406968e-05,
+      "loss": 0.0637,
+      "step": 7896
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.407648668365587e-05,
+      "loss": 0.0696,
+      "step": 7897
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4073545110272295e-05,
+      "loss": 0.0657,
+      "step": 7898
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.407060311422417e-05,
+      "loss": 0.0872,
+      "step": 7899
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4067660695816751e-05,
+      "loss": 0.0654,
+      "step": 7900
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4064717855355345e-05,
+      "loss": 0.0692,
+      "step": 7901
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4061774593145288e-05,
+      "loss": 0.0737,
+      "step": 7902
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4058830909491971e-05,
+      "loss": 0.074,
+      "step": 7903
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.405588680470083e-05,
+      "loss": 0.0807,
+      "step": 7904
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4052942279077334e-05,
+      "loss": 0.0562,
+      "step": 7905
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4049997332927007e-05,
+      "loss": 0.0675,
+      "step": 7906
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4047051966555412e-05,
+      "loss": 0.0801,
+      "step": 7907
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4044106180268152e-05,
+      "loss": 0.0678,
+      "step": 7908
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4041159974370881e-05,
+      "loss": 0.0627,
+      "step": 7909
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.403821334916929e-05,
+      "loss": 0.0686,
+      "step": 7910
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4035266304969115e-05,
+      "loss": 0.0709,
+      "step": 7911
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.403231884207614e-05,
+      "loss": 0.0634,
+      "step": 7912
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4029370960796189e-05,
+      "loss": 0.0654,
+      "step": 7913
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4026422661435127e-05,
+      "loss": 0.0718,
+      "step": 7914
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4023473944298864e-05,
+      "loss": 0.0778,
+      "step": 7915
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4020524809693356e-05,
+      "loss": 0.0583,
+      "step": 7916
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4017575257924603e-05,
+      "loss": 0.0706,
+      "step": 7917
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4014625289298645e-05,
+      "loss": 0.0734,
+      "step": 7918
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4011674904121562e-05,
+      "loss": 0.0687,
+      "step": 7919
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.400872410269948e-05,
+      "loss": 0.0614,
+      "step": 7920
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4005772885338578e-05,
+      "loss": 0.0743,
+      "step": 7921
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4002821252345062e-05,
+      "loss": 0.0543,
+      "step": 7922
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3999869204025197e-05,
+      "loss": 0.0715,
+      "step": 7923
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.399691674068527e-05,
+      "loss": 0.0659,
+      "step": 7924
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3993963862631637e-05,
+      "loss": 0.0657,
+      "step": 7925
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3991010570170673e-05,
+      "loss": 0.0974,
+      "step": 7926
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3988056863608815e-05,
+      "loss": 0.0671,
+      "step": 7927
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3985102743252532e-05,
+      "loss": 0.0573,
+      "step": 7928
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.398214820940834e-05,
+      "loss": 0.0697,
+      "step": 7929
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3979193262382791e-05,
+      "loss": 0.0678,
+      "step": 7930
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3976237902482495e-05,
+      "loss": 0.0759,
+      "step": 7931
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3973282130014087e-05,
+      "loss": 0.0704,
+      "step": 7932
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3970325945284255e-05,
+      "loss": 0.0705,
+      "step": 7933
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3967369348599738e-05,
+      "loss": 0.0717,
+      "step": 7934
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3964412340267293e-05,
+      "loss": 0.0711,
+      "step": 7935
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3961454920593743e-05,
+      "loss": 0.076,
+      "step": 7936
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3958497089885939e-05,
+      "loss": 0.0798,
+      "step": 7937
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3955538848450787e-05,
+      "loss": 0.061,
+      "step": 7938
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3952580196595232e-05,
+      "loss": 0.0709,
+      "step": 7939
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3949621134626253e-05,
+      "loss": 0.0763,
+      "step": 7940
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3946661662850874e-05,
+      "loss": 0.077,
+      "step": 7941
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3943701781576172e-05,
+      "loss": 0.0683,
+      "step": 7942
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3940741491109258e-05,
+      "loss": 0.0602,
+      "step": 7943
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3937780791757287e-05,
+      "loss": 0.0775,
+      "step": 7944
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3934819683827457e-05,
+      "loss": 0.0867,
+      "step": 7945
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3931858167627007e-05,
+      "loss": 0.0697,
+      "step": 7946
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3928896243463218e-05,
+      "loss": 0.0627,
+      "step": 7947
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3925933911643415e-05,
+      "loss": 0.0795,
+      "step": 7948
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3922971172474964e-05,
+      "loss": 0.0625,
+      "step": 7949
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3920008026265278e-05,
+      "loss": 0.0603,
+      "step": 7950
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3917044473321805e-05,
+      "loss": 0.0646,
+      "step": 7951
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.391408051395204e-05,
+      "loss": 0.0741,
+      "step": 7952
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3911116148463517e-05,
+      "loss": 0.0769,
+      "step": 7953
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3908151377163815e-05,
+      "loss": 0.0576,
+      "step": 7954
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3905186200360555e-05,
+      "loss": 0.0711,
+      "step": 7955
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3902220618361399e-05,
+      "loss": 0.0751,
+      "step": 7956
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3899254631474048e-05,
+      "loss": 0.0704,
+      "step": 7957
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3896288240006249e-05,
+      "loss": 0.0865,
+      "step": 7958
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3893321444265793e-05,
+      "loss": 0.0627,
+      "step": 7959
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3890354244560507e-05,
+      "loss": 0.075,
+      "step": 7960
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3887386641198265e-05,
+      "loss": 0.0839,
+      "step": 7961
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3884418634486978e-05,
+      "loss": 0.0471,
+      "step": 7962
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3881450224734604e-05,
+      "loss": 0.0742,
+      "step": 7963
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.387848141224914e-05,
+      "loss": 0.0682,
+      "step": 7964
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3875512197338628e-05,
+      "loss": 0.0585,
+      "step": 7965
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3872542580311144e-05,
+      "loss": 0.0647,
+      "step": 7966
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.386957256147481e-05,
+      "loss": 0.0613,
+      "step": 7967
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3866602141137797e-05,
+      "loss": 0.0692,
+      "step": 7968
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3863631319608306e-05,
+      "loss": 0.0666,
+      "step": 7969
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3860660097194584e-05,
+      "loss": 0.0724,
+      "step": 7970
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3857688474204926e-05,
+      "loss": 0.0808,
+      "step": 7971
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3854716450947658e-05,
+      "loss": 0.076,
+      "step": 7972
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3851744027731156e-05,
+      "loss": 0.0699,
+      "step": 7973
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3848771204863827e-05,
+      "loss": 0.0632,
+      "step": 7974
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3845797982654134e-05,
+      "loss": 0.072,
+      "step": 7975
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.384282436141057e-05,
+      "loss": 0.0633,
+      "step": 7976
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3839850341441674e-05,
+      "loss": 0.0602,
+      "step": 7977
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3836875923056026e-05,
+      "loss": 0.0612,
+      "step": 7978
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3833901106562245e-05,
+      "loss": 0.0661,
+      "step": 7979
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3830925892268994e-05,
+      "loss": 0.0728,
+      "step": 7980
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3827950280484981e-05,
+      "loss": 0.0622,
+      "step": 7981
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3824974271518943e-05,
+      "loss": 0.0706,
+      "step": 7982
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3821997865679669e-05,
+      "loss": 0.0562,
+      "step": 7983
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.381902106327599e-05,
+      "loss": 0.0684,
+      "step": 7984
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.381604386461677e-05,
+      "loss": 0.0569,
+      "step": 7985
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3813066270010919e-05,
+      "loss": 0.0616,
+      "step": 7986
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3810088279767389e-05,
+      "loss": 0.0834,
+      "step": 7987
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3807109894195169e-05,
+      "loss": 0.079,
+      "step": 7988
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3804131113603299e-05,
+      "loss": 0.0795,
+      "step": 7989
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.380115193830084e-05,
+      "loss": 0.0681,
+      "step": 7990
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3798172368596913e-05,
+      "loss": 0.0811,
+      "step": 7991
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3795192404800677e-05,
+      "loss": 0.0703,
+      "step": 7992
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3792212047221326e-05,
+      "loss": 0.0654,
+      "step": 7993
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.378923129616809e-05,
+      "loss": 0.0467,
+      "step": 7994
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3786250151950257e-05,
+      "loss": 0.0559,
+      "step": 7995
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3783268614877144e-05,
+      "loss": 0.0709,
+      "step": 7996
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3780286685258104e-05,
+      "loss": 0.0702,
+      "step": 7997
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3777304363402544e-05,
+      "loss": 0.0742,
+      "step": 7998
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3774321649619902e-05,
+      "loss": 0.0632,
+      "step": 7999
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3771338544219657e-05,
+      "loss": 0.0527,
+      "step": 8000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3768355047511339e-05,
+      "loss": 0.0857,
+      "step": 8001
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3765371159804503e-05,
+      "loss": 0.0714,
+      "step": 8002
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3762386881408759e-05,
+      "loss": 0.0633,
+      "step": 8003
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3759402212633743e-05,
+      "loss": 0.0547,
+      "step": 8004
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3756417153789148e-05,
+      "loss": 0.0695,
+      "step": 8005
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3753431705184694e-05,
+      "loss": 0.0654,
+      "step": 8006
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3750445867130148e-05,
+      "loss": 0.0701,
+      "step": 8007
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3747459639935312e-05,
+      "loss": 0.0652,
+      "step": 8008
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3744473023910039e-05,
+      "loss": 0.0622,
+      "step": 8009
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3741486019364212e-05,
+      "loss": 0.0858,
+      "step": 8010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3738498626607758e-05,
+      "loss": 0.0703,
+      "step": 8011
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.373551084595064e-05,
+      "loss": 0.0684,
+      "step": 8012
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3732522677702873e-05,
+      "loss": 0.0775,
+      "step": 8013
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.37295341221745e-05,
+      "loss": 0.0614,
+      "step": 8014
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.372654517967561e-05,
+      "loss": 0.0696,
+      "step": 8015
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.372355585051633e-05,
+      "loss": 0.0857,
+      "step": 8016
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.372056613500683e-05,
+      "loss": 0.0671,
+      "step": 8017
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3717576033457313e-05,
+      "loss": 0.0698,
+      "step": 8018
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3714585546178033e-05,
+      "loss": 0.0708,
+      "step": 8019
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3711594673479279e-05,
+      "loss": 0.0624,
+      "step": 8020
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3708603415671369e-05,
+      "loss": 0.0759,
+      "step": 8021
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3705611773064684e-05,
+      "loss": 0.0656,
+      "step": 8022
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3702619745969628e-05,
+      "loss": 0.0698,
+      "step": 8023
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3699627334696643e-05,
+      "loss": 0.063,
+      "step": 8024
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3696634539556221e-05,
+      "loss": 0.0736,
+      "step": 8025
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3693641360858891e-05,
+      "loss": 0.0594,
+      "step": 8026
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.369064779891522e-05,
+      "loss": 0.0704,
+      "step": 8027
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3687653854035813e-05,
+      "loss": 0.0613,
+      "step": 8028
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.368465952653132e-05,
+      "loss": 0.0631,
+      "step": 8029
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3681664816712428e-05,
+      "loss": 0.0735,
+      "step": 8030
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.367866972488986e-05,
+      "loss": 0.075,
+      "step": 8031
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3675674251374382e-05,
+      "loss": 0.0663,
+      "step": 8032
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.36726783964768e-05,
+      "loss": 0.0786,
+      "step": 8033
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3669682160507964e-05,
+      "loss": 0.0623,
+      "step": 8034
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3666685543778755e-05,
+      "loss": 0.0782,
+      "step": 8035
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3663688546600093e-05,
+      "loss": 0.0693,
+      "step": 8036
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3660691169282946e-05,
+      "loss": 0.075,
+      "step": 8037
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3657693412138318e-05,
+      "loss": 0.0615,
+      "step": 8038
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3654695275477252e-05,
+      "loss": 0.082,
+      "step": 8039
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3651696759610827e-05,
+      "loss": 0.0748,
+      "step": 8040
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3648697864850162e-05,
+      "loss": 0.0864,
+      "step": 8041
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3645698591506423e-05,
+      "loss": 0.0764,
+      "step": 8042
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3642698939890808e-05,
+      "loss": 0.0719,
+      "step": 8043
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3639698910314556e-05,
+      "loss": 0.0734,
+      "step": 8044
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3636698503088945e-05,
+      "loss": 0.0823,
+      "step": 8045
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3633697718525291e-05,
+      "loss": 0.0605,
+      "step": 8046
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3630696556934955e-05,
+      "loss": 0.0723,
+      "step": 8047
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3627695018629327e-05,
+      "loss": 0.0596,
+      "step": 8048
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3624693103919848e-05,
+      "loss": 0.0803,
+      "step": 8049
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3621690813117987e-05,
+      "loss": 0.097,
+      "step": 8050
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.361868814653526e-05,
+      "loss": 0.0578,
+      "step": 8051
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3615685104483215e-05,
+      "loss": 0.0703,
+      "step": 8052
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3612681687273445e-05,
+      "loss": 0.066,
+      "step": 8053
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3609677895217578e-05,
+      "loss": 0.0714,
+      "step": 8054
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3606673728627288e-05,
+      "loss": 0.0671,
+      "step": 8055
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.360366918781428e-05,
+      "loss": 0.0531,
+      "step": 8056
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.36006642730903e-05,
+      "loss": 0.0802,
+      "step": 8057
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.359765898476713e-05,
+      "loss": 0.0689,
+      "step": 8058
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3594653323156597e-05,
+      "loss": 0.0712,
+      "step": 8059
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3591647288570565e-05,
+      "loss": 0.0551,
+      "step": 8060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3588640881320934e-05,
+      "loss": 0.0736,
+      "step": 8061
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3585634101719642e-05,
+      "loss": 0.0768,
+      "step": 8062
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.358262695007867e-05,
+      "loss": 0.0652,
+      "step": 8063
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3579619426710035e-05,
+      "loss": 0.0618,
+      "step": 8064
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3576611531925791e-05,
+      "loss": 0.067,
+      "step": 8065
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3573603266038036e-05,
+      "loss": 0.0731,
+      "step": 8066
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3570594629358902e-05,
+      "loss": 0.0687,
+      "step": 8067
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3567585622200556e-05,
+      "loss": 0.0722,
+      "step": 8068
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3564576244875212e-05,
+      "loss": 0.0671,
+      "step": 8069
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3561566497695117e-05,
+      "loss": 0.0497,
+      "step": 8070
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3558556380972555e-05,
+      "loss": 0.0585,
+      "step": 8071
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3555545895019861e-05,
+      "loss": 0.0782,
+      "step": 8072
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3552535040149386e-05,
+      "loss": 0.0688,
+      "step": 8073
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3549523816673536e-05,
+      "loss": 0.058,
+      "step": 8074
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3546512224904752e-05,
+      "loss": 0.0541,
+      "step": 8075
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3543500265155509e-05,
+      "loss": 0.0649,
+      "step": 8076
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3540487937738327e-05,
+      "loss": 0.0773,
+      "step": 8077
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3537475242965758e-05,
+      "loss": 0.0786,
+      "step": 8078
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3534462181150395e-05,
+      "loss": 0.062,
+      "step": 8079
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3531448752604867e-05,
+      "loss": 0.0837,
+      "step": 8080
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3528434957641846e-05,
+      "loss": 0.0615,
+      "step": 8081
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3525420796574037e-05,
+      "loss": 0.0632,
+      "step": 8082
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3522406269714182e-05,
+      "loss": 0.0721,
+      "step": 8083
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3519391377375065e-05,
+      "loss": 0.0841,
+      "step": 8084
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.351637611986951e-05,
+      "loss": 0.064,
+      "step": 8085
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.351336049751037e-05,
+      "loss": 0.0562,
+      "step": 8086
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3510344510610545e-05,
+      "loss": 0.0657,
+      "step": 8087
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3507328159482963e-05,
+      "loss": 0.0791,
+      "step": 8088
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3504311444440605e-05,
+      "loss": 0.0756,
+      "step": 8089
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3501294365796475e-05,
+      "loss": 0.0873,
+      "step": 8090
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.349827692386362e-05,
+      "loss": 0.0742,
+      "step": 8091
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3495259118955124e-05,
+      "loss": 0.0629,
+      "step": 8092
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3492240951384115e-05,
+      "loss": 0.0837,
+      "step": 8093
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3489222421463748e-05,
+      "loss": 0.0662,
+      "step": 8094
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3486203529507225e-05,
+      "loss": 0.0762,
+      "step": 8095
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3483184275827776e-05,
+      "loss": 0.0604,
+      "step": 8096
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3480164660738679e-05,
+      "loss": 0.0711,
+      "step": 8097
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3477144684553243e-05,
+      "loss": 0.0692,
+      "step": 8098
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3474124347584814e-05,
+      "loss": 0.0656,
+      "step": 8099
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.347110365014678e-05,
+      "loss": 0.0638,
+      "step": 8100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3468082592552562e-05,
+      "loss": 0.0573,
+      "step": 8101
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3465061175115621e-05,
+      "loss": 0.0863,
+      "step": 8102
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3462039398149454e-05,
+      "loss": 0.0736,
+      "step": 8103
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3459017261967593e-05,
+      "loss": 0.066,
+      "step": 8104
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3455994766883618e-05,
+      "loss": 0.066,
+      "step": 8105
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3452971913211132e-05,
+      "loss": 0.0654,
+      "step": 8106
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3449948701263782e-05,
+      "loss": 0.0685,
+      "step": 8107
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.344692513135525e-05,
+      "loss": 0.0779,
+      "step": 8108
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.344390120379926e-05,
+      "loss": 0.062,
+      "step": 8109
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3440876918909571e-05,
+      "loss": 0.0674,
+      "step": 8110
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3437852276999976e-05,
+      "loss": 0.0528,
+      "step": 8111
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3434827278384306e-05,
+      "loss": 0.0721,
+      "step": 8112
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.343180192337643e-05,
+      "loss": 0.0789,
+      "step": 8113
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3428776212290258e-05,
+      "loss": 0.0565,
+      "step": 8114
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3425750145439729e-05,
+      "loss": 0.0823,
+      "step": 8115
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3422723723138824e-05,
+      "loss": 0.0687,
+      "step": 8116
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3419696945701559e-05,
+      "loss": 0.0708,
+      "step": 8117
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3416669813441989e-05,
+      "loss": 0.0696,
+      "step": 8118
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.34136423266742e-05,
+      "loss": 0.0651,
+      "step": 8119
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.341061448571233e-05,
+      "loss": 0.0541,
+      "step": 8120
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3407586290870528e-05,
+      "loss": 0.0667,
+      "step": 8121
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3404557742463009e-05,
+      "loss": 0.0677,
+      "step": 8122
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3401528840804001e-05,
+      "loss": 0.0626,
+      "step": 8123
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3398499586207782e-05,
+      "loss": 0.0557,
+      "step": 8124
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.339546997898866e-05,
+      "loss": 0.0903,
+      "step": 8125
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3392440019460984e-05,
+      "loss": 0.0675,
+      "step": 8126
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3389409707939138e-05,
+      "loss": 0.0693,
+      "step": 8127
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3386379044737545e-05,
+      "loss": 0.0629,
+      "step": 8128
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3383348030170654e-05,
+      "loss": 0.0554,
+      "step": 8129
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3380316664552963e-05,
+      "loss": 0.0863,
+      "step": 8130
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3377284948199006e-05,
+      "loss": 0.0678,
+      "step": 8131
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3374252881423339e-05,
+      "loss": 0.0652,
+      "step": 8132
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3371220464540573e-05,
+      "loss": 0.0793,
+      "step": 8133
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3368187697865342e-05,
+      "loss": 0.0704,
+      "step": 8134
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3365154581712326e-05,
+      "loss": 0.0546,
+      "step": 8135
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3362121116396231e-05,
+      "loss": 0.0678,
+      "step": 8136
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3359087302231806e-05,
+      "loss": 0.0477,
+      "step": 8137
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3356053139533835e-05,
+      "loss": 0.0809,
+      "step": 8138
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.335301862861714e-05,
+      "loss": 0.0723,
+      "step": 8139
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3349983769796574e-05,
+      "loss": 0.0646,
+      "step": 8140
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3346948563387035e-05,
+      "loss": 0.0689,
+      "step": 8141
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.334391300970344e-05,
+      "loss": 0.0693,
+      "step": 8142
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3340877109060762e-05,
+      "loss": 0.0692,
+      "step": 8143
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3337840861774001e-05,
+      "loss": 0.0577,
+      "step": 8144
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3334804268158187e-05,
+      "loss": 0.0579,
+      "step": 8145
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3331767328528398e-05,
+      "loss": 0.0773,
+      "step": 8146
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.332873004319974e-05,
+      "loss": 0.0624,
+      "step": 8147
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.332569241248736e-05,
+      "loss": 0.0615,
+      "step": 8148
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.332265443670643e-05,
+      "loss": 0.0751,
+      "step": 8149
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3319616116172175e-05,
+      "loss": 0.0553,
+      "step": 8150
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3316577451199839e-05,
+      "loss": 0.066,
+      "step": 8151
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3313538442104714e-05,
+      "loss": 0.0699,
+      "step": 8152
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3310499089202118e-05,
+      "loss": 0.0725,
+      "step": 8153
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3307459392807413e-05,
+      "loss": 0.0679,
+      "step": 8154
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3304419353235991e-05,
+      "loss": 0.0828,
+      "step": 8155
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3301378970803285e-05,
+      "loss": 0.0685,
+      "step": 8156
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3298338245824755e-05,
+      "loss": 0.0774,
+      "step": 8157
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3295297178615904e-05,
+      "loss": 0.0515,
+      "step": 8158
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3292255769492273e-05,
+      "loss": 0.0756,
+      "step": 8159
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3289214018769428e-05,
+      "loss": 0.0737,
+      "step": 8160
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3286171926762977e-05,
+      "loss": 0.0615,
+      "step": 8161
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3283129493788562e-05,
+      "loss": 0.0748,
+      "step": 8162
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3280086720161864e-05,
+      "loss": 0.0694,
+      "step": 8163
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3277043606198596e-05,
+      "loss": 0.0723,
+      "step": 8164
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3274000152214503e-05,
+      "loss": 0.0553,
+      "step": 8165
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3270956358525376e-05,
+      "loss": 0.0658,
+      "step": 8166
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3267912225447026e-05,
+      "loss": 0.0562,
+      "step": 8167
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3264867753295311e-05,
+      "loss": 0.0596,
+      "step": 8168
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3261822942386122e-05,
+      "loss": 0.0478,
+      "step": 8169
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.325877779303538e-05,
+      "loss": 0.0692,
+      "step": 8170
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3255732305559048e-05,
+      "loss": 0.0649,
+      "step": 8171
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3252686480273122e-05,
+      "loss": 0.0577,
+      "step": 8172
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3249640317493628e-05,
+      "loss": 0.0787,
+      "step": 8173
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3246593817536635e-05,
+      "loss": 0.0708,
+      "step": 8174
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3243546980718238e-05,
+      "loss": 0.058,
+      "step": 8175
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3240499807354577e-05,
+      "loss": 0.0758,
+      "step": 8176
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3237452297761819e-05,
+      "loss": 0.0621,
+      "step": 8177
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3234404452256174e-05,
+      "loss": 0.0681,
+      "step": 8178
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.323135627115387e-05,
+      "loss": 0.0629,
+      "step": 8179
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3228307754771192e-05,
+      "loss": 0.0718,
+      "step": 8180
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3225258903424447e-05,
+      "loss": 0.0665,
+      "step": 8181
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3222209717429974e-05,
+      "loss": 0.0592,
+      "step": 8182
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.321916019710416e-05,
+      "loss": 0.0577,
+      "step": 8183
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3216110342763413e-05,
+      "loss": 0.0691,
+      "step": 8184
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3213060154724179e-05,
+      "loss": 0.0775,
+      "step": 8185
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3210009633302945e-05,
+      "loss": 0.0659,
+      "step": 8186
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3206958778816228e-05,
+      "loss": 0.054,
+      "step": 8187
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3203907591580573e-05,
+      "loss": 0.0623,
+      "step": 8188
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.320085607191258e-05,
+      "loss": 0.0708,
+      "step": 8189
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3197804220128855e-05,
+      "loss": 0.0684,
+      "step": 8190
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3194752036546063e-05,
+      "loss": 0.0691,
+      "step": 8191
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3191699521480889e-05,
+      "loss": 0.066,
+      "step": 8192
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3188646675250061e-05,
+      "loss": 0.0577,
+      "step": 8193
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3185593498170334e-05,
+      "loss": 0.0696,
+      "step": 8194
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3182539990558502e-05,
+      "loss": 0.0753,
+      "step": 8195
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.317948615273139e-05,
+      "loss": 0.058,
+      "step": 8196
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3176431985005864e-05,
+      "loss": 0.0628,
+      "step": 8197
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3173377487698815e-05,
+      "loss": 0.0605,
+      "step": 8198
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3170322661127178e-05,
+      "loss": 0.0668,
+      "step": 8199
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.316726750560791e-05,
+      "loss": 0.0748,
+      "step": 8200
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3164212021458016e-05,
+      "loss": 0.0572,
+      "step": 8201
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3161156208994524e-05,
+      "loss": 0.0663,
+      "step": 8202
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.31581000685345e-05,
+      "loss": 0.0739,
+      "step": 8203
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3155043600395045e-05,
+      "loss": 0.0715,
+      "step": 8204
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3151986804893296e-05,
+      "loss": 0.0468,
+      "step": 8205
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3148929682346418e-05,
+      "loss": 0.0743,
+      "step": 8206
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3145872233071613e-05,
+      "loss": 0.0811,
+      "step": 8207
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.314281445738612e-05,
+      "loss": 0.0681,
+      "step": 8208
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3139756355607203e-05,
+      "loss": 0.0739,
+      "step": 8209
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3136697928052171e-05,
+      "loss": 0.0813,
+      "step": 8210
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3133639175038367e-05,
+      "loss": 0.082,
+      "step": 8211
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.313058009688315e-05,
+      "loss": 0.0497,
+      "step": 8212
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3127520693903928e-05,
+      "loss": 0.0788,
+      "step": 8213
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3124460966418147e-05,
+      "loss": 0.087,
+      "step": 8214
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3121400914743275e-05,
+      "loss": 0.0757,
+      "step": 8215
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3118340539196819e-05,
+      "loss": 0.0723,
+      "step": 8216
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3115279840096315e-05,
+      "loss": 0.0724,
+      "step": 8217
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3112218817759338e-05,
+      "loss": 0.0635,
+      "step": 8218
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.31091574725035e-05,
+      "loss": 0.0602,
+      "step": 8219
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3106095804646435e-05,
+      "loss": 0.0594,
+      "step": 8220
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3103033814505817e-05,
+      "loss": 0.0594,
+      "step": 8221
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.309997150239936e-05,
+      "loss": 0.0753,
+      "step": 8222
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3096908868644798e-05,
+      "loss": 0.0593,
+      "step": 8223
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3093845913559906e-05,
+      "loss": 0.0692,
+      "step": 8224
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3090782637462489e-05,
+      "loss": 0.0611,
+      "step": 8225
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3087719040670397e-05,
+      "loss": 0.0774,
+      "step": 8226
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3084655123501495e-05,
+      "loss": 0.0665,
+      "step": 8227
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3081590886273695e-05,
+      "loss": 0.0691,
+      "step": 8228
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.307852632930493e-05,
+      "loss": 0.0688,
+      "step": 8229
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3075461452913181e-05,
+      "loss": 0.0717,
+      "step": 8230
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3072396257416453e-05,
+      "loss": 0.0684,
+      "step": 8231
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3069330743132787e-05,
+      "loss": 0.0594,
+      "step": 8232
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3066264910380251e-05,
+      "loss": 0.0597,
+      "step": 8233
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3063198759476956e-05,
+      "loss": 0.0797,
+      "step": 8234
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3060132290741038e-05,
+      "loss": 0.0579,
+      "step": 8235
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3057065504490672e-05,
+      "loss": 0.0535,
+      "step": 8236
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.305399840104406e-05,
+      "loss": 0.061,
+      "step": 8237
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3050930980719438e-05,
+      "loss": 0.0648,
+      "step": 8238
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3047863243835081e-05,
+      "loss": 0.0752,
+      "step": 8239
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3044795190709292e-05,
+      "loss": 0.0761,
+      "step": 8240
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3041726821660406e-05,
+      "loss": 0.0671,
+      "step": 8241
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3038658137006788e-05,
+      "loss": 0.0673,
+      "step": 8242
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3035589137066847e-05,
+      "loss": 0.0752,
+      "step": 8243
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3032519822159015e-05,
+      "loss": 0.0665,
+      "step": 8244
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3029450192601758e-05,
+      "loss": 0.0552,
+      "step": 8245
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3026380248713574e-05,
+      "loss": 0.0781,
+      "step": 8246
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3023309990813e-05,
+      "loss": 0.0774,
+      "step": 8247
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.30202394192186e-05,
+      "loss": 0.071,
+      "step": 8248
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3017168534248974e-05,
+      "loss": 0.069,
+      "step": 8249
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3014097336222742e-05,
+      "loss": 0.0574,
+      "step": 8250
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3011025825458576e-05,
+      "loss": 0.0746,
+      "step": 8251
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.300795400227517e-05,
+      "loss": 0.0589,
+      "step": 8252
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3004881866991249e-05,
+      "loss": 0.0706,
+      "step": 8253
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3001809419925575e-05,
+      "loss": 0.0781,
+      "step": 8254
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.2998736661396935e-05,
+      "loss": 0.0677,
+      "step": 8255
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2995663591724161e-05,
+      "loss": 0.065,
+      "step": 8256
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2992590211226106e-05,
+      "loss": 0.0726,
+      "step": 8257
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2989516520221661e-05,
+      "loss": 0.0525,
+      "step": 8258
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.298644251902974e-05,
+      "loss": 0.0633,
+      "step": 8259
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2983368207969309e-05,
+      "loss": 0.0629,
+      "step": 8260
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2980293587359345e-05,
+      "loss": 0.0709,
+      "step": 8261
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2977218657518866e-05,
+      "loss": 0.0594,
+      "step": 8262
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2974143418766922e-05,
+      "loss": 0.0702,
+      "step": 8263
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2971067871422599e-05,
+      "loss": 0.0805,
+      "step": 8264
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2967992015805008e-05,
+      "loss": 0.0637,
+      "step": 8265
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2964915852233295e-05,
+      "loss": 0.0708,
+      "step": 8266
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2961839381026638e-05,
+      "loss": 0.0617,
+      "step": 8267
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2958762602504244e-05,
+      "loss": 0.0632,
+      "step": 8268
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.295568551698536e-05,
+      "loss": 0.0643,
+      "step": 8269
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2952608124789255e-05,
+      "loss": 0.0563,
+      "step": 8270
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2949530426235237e-05,
+      "loss": 0.0685,
+      "step": 8271
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2946452421642643e-05,
+      "loss": 0.0745,
+      "step": 8272
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.294337411133084e-05,
+      "loss": 0.0689,
+      "step": 8273
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2940295495619228e-05,
+      "loss": 0.0549,
+      "step": 8274
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.2937216574827245e-05,
+      "loss": 0.0602,
+      "step": 8275
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2934137349274346e-05,
+      "loss": 0.0624,
+      "step": 8276
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2931057819280034e-05,
+      "loss": 0.0764,
+      "step": 8277
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2927977985163834e-05,
+      "loss": 0.0695,
+      "step": 8278
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.29248978472453e-05,
+      "loss": 0.0612,
+      "step": 8279
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2921817405844031e-05,
+      "loss": 0.0502,
+      "step": 8280
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.291873666127964e-05,
+      "loss": 0.0753,
+      "step": 8281
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2915655613871786e-05,
+      "loss": 0.0625,
+      "step": 8282
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2912574263940152e-05,
+      "loss": 0.0586,
+      "step": 8283
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2909492611804455e-05,
+      "loss": 0.0497,
+      "step": 8284
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2906410657784438e-05,
+      "loss": 0.0756,
+      "step": 8285
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2903328402199885e-05,
+      "loss": 0.0628,
+      "step": 8286
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2900245845370603e-05,
+      "loss": 0.078,
+      "step": 8287
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2897162987616434e-05,
+      "loss": 0.0667,
+      "step": 8288
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.289407982925725e-05,
+      "loss": 0.0661,
+      "step": 8289
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2890996370612954e-05,
+      "loss": 0.0704,
+      "step": 8290
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2887912612003482e-05,
+      "loss": 0.0557,
+      "step": 8291
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.2884828553748796e-05,
+      "loss": 0.0695,
+      "step": 8292
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.28817441961689e-05,
+      "loss": 0.0593,
+      "step": 8293
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.287865953958382e-05,
+      "loss": 0.0598,
+      "step": 8294
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.287557458431361e-05,
+      "loss": 0.0629,
+      "step": 8295
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2872489330678363e-05,
+      "loss": 0.0876,
+      "step": 8296
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2869403778998201e-05,
+      "loss": 0.0728,
+      "step": 8297
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2866317929593276e-05,
+      "loss": 0.0565,
+      "step": 8298
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2863231782783774e-05,
+      "loss": 0.0744,
+      "step": 8299
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.28601453388899e-05,
+      "loss": 0.0632,
+      "step": 8300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2857058598231904e-05,
+      "loss": 0.0641,
+      "step": 8301
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2853971561130062e-05,
+      "loss": 0.0704,
+      "step": 8302
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.285088422790468e-05,
+      "loss": 0.0587,
+      "step": 8303
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.284779659887609e-05,
+      "loss": 0.0554,
+      "step": 8304
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2844708674364665e-05,
+      "loss": 0.0875,
+      "step": 8305
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.28416204546908e-05,
+      "loss": 0.0641,
+      "step": 8306
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2838531940174925e-05,
+      "loss": 0.0747,
+      "step": 8307
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2835443131137502e-05,
+      "loss": 0.0738,
+      "step": 8308
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2832354027899012e-05,
+      "loss": 0.0845,
+      "step": 8309
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2829264630779987e-05,
+      "loss": 0.058,
+      "step": 8310
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.282617494010097e-05,
+      "loss": 0.0683,
+      "step": 8311
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2823084956182545e-05,
+      "loss": 0.0627,
+      "step": 8312
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2819994679345324e-05,
+      "loss": 0.0643,
+      "step": 8313
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.2816904109909948e-05,
+      "loss": 0.0829,
+      "step": 8314
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.281381324819709e-05,
+      "loss": 0.0602,
+      "step": 8315
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2810722094527458e-05,
+      "loss": 0.0682,
+      "step": 8316
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2807630649221777e-05,
+      "loss": 0.0648,
+      "step": 8317
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2804538912600812e-05,
+      "loss": 0.0634,
+      "step": 8318
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.280144688498536e-05,
+      "loss": 0.0699,
+      "step": 8319
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2798354566696245e-05,
+      "loss": 0.0698,
+      "step": 8320
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2795261958054317e-05,
+      "loss": 0.0657,
+      "step": 8321
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2792169059380463e-05,
+      "loss": 0.0645,
+      "step": 8322
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.27890758709956e-05,
+      "loss": 0.0617,
+      "step": 8323
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2785982393220667e-05,
+      "loss": 0.068,
+      "step": 8324
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.278288862637664e-05,
+      "loss": 0.0626,
+      "step": 8325
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.277979457078452e-05,
+      "loss": 0.0727,
+      "step": 8326
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.277670022676535e-05,
+      "loss": 0.076,
+      "step": 8327
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.277360559464019e-05,
+      "loss": 0.0695,
+      "step": 8328
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2770510674730132e-05,
+      "loss": 0.0612,
+      "step": 8329
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2767415467356297e-05,
+      "loss": 0.0763,
+      "step": 8330
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2764319972839847e-05,
+      "loss": 0.048,
+      "step": 8331
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2761224191501964e-05,
+      "loss": 0.0583,
+      "step": 8332
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2758128123663855e-05,
+      "loss": 0.0593,
+      "step": 8333
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.275503176964677e-05,
+      "loss": 0.0589,
+      "step": 8334
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.2751935129771974e-05,
+      "loss": 0.0782,
+      "step": 8335
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.274883820436078e-05,
+      "loss": 0.0741,
+      "step": 8336
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.274574099373451e-05,
+      "loss": 0.0692,
+      "step": 8337
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2742643498214534e-05,
+      "loss": 0.0474,
+      "step": 8338
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2739545718122234e-05,
+      "loss": 0.0703,
+      "step": 8339
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2736447653779039e-05,
+      "loss": 0.0591,
+      "step": 8340
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2733349305506395e-05,
+      "loss": 0.0659,
+      "step": 8341
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2730250673625783e-05,
+      "loss": 0.0643,
+      "step": 8342
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2727151758458712e-05,
+      "loss": 0.084,
+      "step": 8343
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2724052560326722e-05,
+      "loss": 0.0721,
+      "step": 8344
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2720953079551377e-05,
+      "loss": 0.0558,
+      "step": 8345
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2717853316454276e-05,
+      "loss": 0.0743,
+      "step": 8346
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2714753271357047e-05,
+      "loss": 0.0548,
+      "step": 8347
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2711652944581345e-05,
+      "loss": 0.0676,
+      "step": 8348
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2708552336448857e-05,
+      "loss": 0.069,
+      "step": 8349
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2705451447281289e-05,
+      "loss": 0.0609,
+      "step": 8350
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2702350277400392e-05,
+      "loss": 0.0803,
+      "step": 8351
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2699248827127937e-05,
+      "loss": 0.0661,
+      "step": 8352
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2696147096785727e-05,
+      "loss": 0.0785,
+      "step": 8353
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2693045086695587e-05,
+      "loss": 0.0598,
+      "step": 8354
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.2689942797179385e-05,
+      "loss": 0.0681,
+      "step": 8355
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2686840228559001e-05,
+      "loss": 0.0599,
+      "step": 8356
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2683737381156357e-05,
+      "loss": 0.0807,
+      "step": 8357
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.26806342552934e-05,
+      "loss": 0.0664,
+      "step": 8358
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.26775308512921e-05,
+      "loss": 0.0643,
+      "step": 8359
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2674427169474473e-05,
+      "loss": 0.0655,
+      "step": 8360
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2671323210162543e-05,
+      "loss": 0.0717,
+      "step": 8361
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.266821897367837e-05,
+      "loss": 0.0675,
+      "step": 8362
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2665114460344048e-05,
+      "loss": 0.0714,
+      "step": 8363
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2662009670481699e-05,
+      "loss": 0.0788,
+      "step": 8364
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2658904604413468e-05,
+      "loss": 0.0754,
+      "step": 8365
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2655799262461535e-05,
+      "loss": 0.0643,
+      "step": 8366
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2652693644948097e-05,
+      "loss": 0.0686,
+      "step": 8367
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2649587752195397e-05,
+      "loss": 0.0591,
+      "step": 8368
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2646481584525693e-05,
+      "loss": 0.0817,
+      "step": 8369
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2643375142261279e-05,
+      "loss": 0.0585,
+      "step": 8370
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2640268425724469e-05,
+      "loss": 0.0732,
+      "step": 8371
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2637161435237617e-05,
+      "loss": 0.0695,
+      "step": 8372
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2634054171123097e-05,
+      "loss": 0.0734,
+      "step": 8373
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2630946633703314e-05,
+      "loss": 0.0562,
+      "step": 8374
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.2627838823300697e-05,
+      "loss": 0.0657,
+      "step": 8375
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2624730740237714e-05,
+      "loss": 0.0787,
+      "step": 8376
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2621622384836853e-05,
+      "loss": 0.0579,
+      "step": 8377
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2618513757420633e-05,
+      "loss": 0.0662,
+      "step": 8378
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2615404858311593e-05,
+      "loss": 0.0608,
+      "step": 8379
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2612295687832315e-05,
+      "loss": 0.0726,
+      "step": 8380
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2609186246305397e-05,
+      "loss": 0.0557,
+      "step": 8381
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2606076534053476e-05,
+      "loss": 0.0685,
+      "step": 8382
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2602966551399206e-05,
+      "loss": 0.0591,
+      "step": 8383
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.259985629866527e-05,
+      "loss": 0.0809,
+      "step": 8384
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.259674577617439e-05,
+      "loss": 0.0668,
+      "step": 8385
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2593634984249307e-05,
+      "loss": 0.067,
+      "step": 8386
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2590523923212792e-05,
+      "loss": 0.0785,
+      "step": 8387
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.258741259338764e-05,
+      "loss": 0.0758,
+      "step": 8388
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2584300995096684e-05,
+      "loss": 0.0612,
+      "step": 8389
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2581189128662772e-05,
+      "loss": 0.06,
+      "step": 8390
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.257807699440879e-05,
+      "loss": 0.0631,
+      "step": 8391
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2574964592657648e-05,
+      "loss": 0.0781,
+      "step": 8392
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2571851923732284e-05,
+      "loss": 0.0597,
+      "step": 8393
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.256873898795566e-05,
+      "loss": 0.0841,
+      "step": 8394
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.2565625785650774e-05,
+      "loss": 0.0487,
+      "step": 8395
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2562512317140643e-05,
+      "loss": 0.075,
+      "step": 8396
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2559398582748317e-05,
+      "loss": 0.0546,
+      "step": 8397
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2556284582796874e-05,
+      "loss": 0.0612,
+      "step": 8398
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.255317031760942e-05,
+      "loss": 0.0588,
+      "step": 8399
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.255005578750908e-05,
+      "loss": 0.0626,
+      "step": 8400
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.254694099281901e-05,
+      "loss": 0.0813,
+      "step": 8401
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2543825933862403e-05,
+      "loss": 0.0887,
+      "step": 8402
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2540710610962474e-05,
+      "loss": 0.057,
+      "step": 8403
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2537595024442462e-05,
+      "loss": 0.0605,
+      "step": 8404
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.253447917462563e-05,
+      "loss": 0.0636,
+      "step": 8405
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.253136306183528e-05,
+      "loss": 0.059,
+      "step": 8406
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2528246686394732e-05,
+      "loss": 0.0669,
+      "step": 8407
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2525130048627335e-05,
+      "loss": 0.072,
+      "step": 8408
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.252201314885647e-05,
+      "loss": 0.0612,
+      "step": 8409
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2518895987405539e-05,
+      "loss": 0.0444,
+      "step": 8410
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2515778564597974e-05,
+      "loss": 0.061,
+      "step": 8411
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2512660880757234e-05,
+      "loss": 0.0672,
+      "step": 8412
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.2509542936206802e-05,
+      "loss": 0.0677,
+      "step": 8413
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.25064247312702e-05,
+      "loss": 0.0689,
+      "step": 8414
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.250330626627096e-05,
+      "loss": 0.0743,
+      "step": 8415
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.250018754153265e-05,
+      "loss": 0.0622,
+      "step": 8416
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2497068557378862e-05,
+      "loss": 0.0676,
+      "step": 8417
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2493949314133222e-05,
+      "loss": 0.0695,
+      "step": 8418
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2490829812119376e-05,
+      "loss": 0.0651,
+      "step": 8419
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2487710051661e-05,
+      "loss": 0.0693,
+      "step": 8420
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.248459003308179e-05,
+      "loss": 0.0745,
+      "step": 8421
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2481469756705478e-05,
+      "loss": 0.0713,
+      "step": 8422
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.247834922285582e-05,
+      "loss": 0.0714,
+      "step": 8423
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2475228431856594e-05,
+      "loss": 0.0687,
+      "step": 8424
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.247210738403161e-05,
+      "loss": 0.0886,
+      "step": 8425
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2468986079704703e-05,
+      "loss": 0.0635,
+      "step": 8426
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.246586451919974e-05,
+      "loss": 0.0577,
+      "step": 8427
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2462742702840597e-05,
+      "loss": 0.0804,
+      "step": 8428
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2459620630951198e-05,
+      "loss": 0.0659,
+      "step": 8429
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2456498303855482e-05,
+      "loss": 0.0563,
+      "step": 8430
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2453375721877417e-05,
+      "loss": 0.0608,
+      "step": 8431
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2450252885340999e-05,
+      "loss": 0.0644,
+      "step": 8432
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2447129794570243e-05,
+      "loss": 0.0631,
+      "step": 8433
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2444006449889198e-05,
+      "loss": 0.0685,
+      "step": 8434
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.244088285162194e-05,
+      "loss": 0.0707,
+      "step": 8435
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.2437759000092568e-05,
+      "loss": 0.0719,
+      "step": 8436
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2434634895625206e-05,
+      "loss": 0.0656,
+      "step": 8437
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2431510538544007e-05,
+      "loss": 0.0728,
+      "step": 8438
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2428385929173147e-05,
+      "loss": 0.0674,
+      "step": 8439
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2425261067836835e-05,
+      "loss": 0.0627,
+      "step": 8440
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.24221359548593e-05,
+      "loss": 0.0716,
+      "step": 8441
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2419010590564798e-05,
+      "loss": 0.0687,
+      "step": 8442
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.241588497527761e-05,
+      "loss": 0.0702,
+      "step": 8443
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2412759109322048e-05,
+      "loss": 0.0632,
+      "step": 8444
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2409632993022445e-05,
+      "loss": 0.0596,
+      "step": 8445
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2406506626703163e-05,
+      "loss": 0.0645,
+      "step": 8446
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2403380010688586e-05,
+      "loss": 0.0562,
+      "step": 8447
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2400253145303129e-05,
+      "loss": 0.0704,
+      "step": 8448
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2397126030871235e-05,
+      "loss": 0.0589,
+      "step": 8449
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2393998667717361e-05,
+      "loss": 0.057,
+      "step": 8450
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2390871056166e-05,
+      "loss": 0.0598,
+      "step": 8451
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2387743196541669e-05,
+      "loss": 0.0596,
+      "step": 8452
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2384615089168908e-05,
+      "loss": 0.0723,
+      "step": 8453
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2381486734372288e-05,
+      "loss": 0.0846,
+      "step": 8454
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2378358132476395e-05,
+      "loss": 0.0644,
+      "step": 8455
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2375229283805854e-05,
+      "loss": 0.0786,
+      "step": 8456
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2372100188685308e-05,
+      "loss": 0.0564,
+      "step": 8457
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2368970847439426e-05,
+      "loss": 0.0696,
+      "step": 8458
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2365841260392905e-05,
+      "loss": 0.0652,
+      "step": 8459
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2362711427870465e-05,
+      "loss": 0.0519,
+      "step": 8460
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.235958135019685e-05,
+      "loss": 0.0673,
+      "step": 8461
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2356451027696836e-05,
+      "loss": 0.0632,
+      "step": 8462
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2353320460695217e-05,
+      "loss": 0.063,
+      "step": 8463
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2350189649516818e-05,
+      "loss": 0.0497,
+      "step": 8464
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2347058594486487e-05,
+      "loss": 0.0737,
+      "step": 8465
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2343927295929094e-05,
+      "loss": 0.0682,
+      "step": 8466
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2340795754169544e-05,
+      "loss": 0.0823,
+      "step": 8467
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2337663969532752e-05,
+      "loss": 0.0674,
+      "step": 8468
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2334531942343673e-05,
+      "loss": 0.0566,
+      "step": 8469
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.233139967292728e-05,
+      "loss": 0.0684,
+      "step": 8470
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2328267161608574e-05,
+      "loss": 0.0651,
+      "step": 8471
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2325134408712575e-05,
+      "loss": 0.0616,
+      "step": 8472
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2322001414564336e-05,
+      "loss": 0.0644,
+      "step": 8473
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.231886817948893e-05,
+      "loss": 0.0616,
+      "step": 8474
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2315734703811457e-05,
+      "loss": 0.0593,
+      "step": 8475
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2312600987857041e-05,
+      "loss": 0.0632,
+      "step": 8476
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2309467031950833e-05,
+      "loss": 0.067,
+      "step": 8477
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2306332836418005e-05,
+      "loss": 0.0828,
+      "step": 8478
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2303198401583759e-05,
+      "loss": 0.0699,
+      "step": 8479
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2300063727773312e-05,
+      "loss": 0.0618,
+      "step": 8480
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2296928815311922e-05,
+      "loss": 0.0613,
+      "step": 8481
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.229379366452486e-05,
+      "loss": 0.0658,
+      "step": 8482
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.229065827573742e-05,
+      "loss": 0.0632,
+      "step": 8483
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2287522649274926e-05,
+      "loss": 0.0669,
+      "step": 8484
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2284386785462728e-05,
+      "loss": 0.0694,
+      "step": 8485
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2281250684626198e-05,
+      "loss": 0.0759,
+      "step": 8486
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2278114347090736e-05,
+      "loss": 0.0601,
+      "step": 8487
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2274977773181753e-05,
+      "loss": 0.0661,
+      "step": 8488
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2271840963224703e-05,
+      "loss": 0.0613,
+      "step": 8489
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2268703917545057e-05,
+      "loss": 0.0578,
+      "step": 8490
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2265566636468309e-05,
+      "loss": 0.0655,
+      "step": 8491
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2262429120319974e-05,
+      "loss": 0.0623,
+      "step": 8492
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2259291369425602e-05,
+      "loss": 0.08,
+      "step": 8493
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2256153384110754e-05,
+      "loss": 0.0616,
+      "step": 8494
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2253015164701027e-05,
+      "loss": 0.069,
+      "step": 8495
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2249876711522037e-05,
+      "loss": 0.0646,
+      "step": 8496
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2246738024899424e-05,
+      "loss": 0.0627,
+      "step": 8497
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2243599105158857e-05,
+      "loss": 0.0557,
+      "step": 8498
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.224045995262602e-05,
+      "loss": 0.0724,
+      "step": 8499
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.223732056762663e-05,
+      "loss": 0.0605,
+      "step": 8500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2234180950486418e-05,
+      "loss": 0.0699,
+      "step": 8501
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2231041101531155e-05,
+      "loss": 0.0557,
+      "step": 8502
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2227901021086624e-05,
+      "loss": 0.0613,
+      "step": 8503
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2224760709478631e-05,
+      "loss": 0.0653,
+      "step": 8504
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2221620167033013e-05,
+      "loss": 0.0644,
+      "step": 8505
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2218479394075624e-05,
+      "loss": 0.0823,
+      "step": 8506
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2215338390932351e-05,
+      "loss": 0.0701,
+      "step": 8507
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2212197157929095e-05,
+      "loss": 0.0679,
+      "step": 8508
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.220905569539179e-05,
+      "loss": 0.0672,
+      "step": 8509
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2205914003646385e-05,
+      "loss": 0.056,
+      "step": 8510
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2202772083018857e-05,
+      "loss": 0.0634,
+      "step": 8511
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2199629933835208e-05,
+      "loss": 0.0671,
+      "step": 8512
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2196487556421464e-05,
+      "loss": 0.0754,
+      "step": 8513
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2193344951103667e-05,
+      "loss": 0.0594,
+      "step": 8514
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.21902021182079e-05,
+      "loss": 0.0646,
+      "step": 8515
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2187059058060246e-05,
+      "loss": 0.066,
+      "step": 8516
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2183915770986832e-05,
+      "loss": 0.0603,
+      "step": 8517
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2180772257313793e-05,
+      "loss": 0.0463,
+      "step": 8518
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2177628517367304e-05,
+      "loss": 0.0612,
+      "step": 8519
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.217448455147355e-05,
+      "loss": 0.0605,
+      "step": 8520
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2171340359958742e-05,
+      "loss": 0.067,
+      "step": 8521
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.216819594314912e-05,
+      "loss": 0.0767,
+      "step": 8522
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.216505130137094e-05,
+      "loss": 0.0641,
+      "step": 8523
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.216190643495049e-05,
+      "loss": 0.067,
+      "step": 8524
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2158761344214072e-05,
+      "loss": 0.0767,
+      "step": 8525
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2155616029488017e-05,
+      "loss": 0.089,
+      "step": 8526
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2152470491098678e-05,
+      "loss": 0.0552,
+      "step": 8527
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2149324729372432e-05,
+      "loss": 0.0575,
+      "step": 8528
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2146178744635675e-05,
+      "loss": 0.0729,
+      "step": 8529
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2143032537214832e-05,
+      "loss": 0.0645,
+      "step": 8530
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2139886107436347e-05,
+      "loss": 0.0523,
+      "step": 8531
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2136739455626692e-05,
+      "loss": 0.0547,
+      "step": 8532
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2133592582112354e-05,
+      "loss": 0.0728,
+      "step": 8533
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2130445487219851e-05,
+      "loss": 0.0649,
+      "step": 8534
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2127298171275716e-05,
+      "loss": 0.055,
+      "step": 8535
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2124150634606515e-05,
+      "loss": 0.0703,
+      "step": 8536
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2121002877538832e-05,
+      "loss": 0.0529,
+      "step": 8537
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2117854900399266e-05,
+      "loss": 0.0545,
+      "step": 8538
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2114706703514452e-05,
+      "loss": 0.078,
+      "step": 8539
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.211155828721104e-05,
+      "loss": 0.0754,
+      "step": 8540
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2108409651815704e-05,
+      "loss": 0.0738,
+      "step": 8541
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2105260797655144e-05,
+      "loss": 0.0745,
+      "step": 8542
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2102111725056078e-05,
+      "loss": 0.0683,
+      "step": 8543
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2098962434345251e-05,
+      "loss": 0.0602,
+      "step": 8544
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2095812925849424e-05,
+      "loss": 0.0548,
+      "step": 8545
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2092663199895389e-05,
+      "loss": 0.0745,
+      "step": 8546
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2089513256809953e-05,
+      "loss": 0.0652,
+      "step": 8547
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2086363096919953e-05,
+      "loss": 0.0571,
+      "step": 8548
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2083212720552242e-05,
+      "loss": 0.0756,
+      "step": 8549
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.20800621280337e-05,
+      "loss": 0.0605,
+      "step": 8550
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2076911319691222e-05,
+      "loss": 0.0597,
+      "step": 8551
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2073760295851739e-05,
+      "loss": 0.0584,
+      "step": 8552
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2070609056842193e-05,
+      "loss": 0.058,
+      "step": 8553
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2067457602989552e-05,
+      "loss": 0.0587,
+      "step": 8554
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2064305934620798e-05,
+      "loss": 0.0668,
+      "step": 8555
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.2061154052062956e-05,
+      "loss": 0.0658,
+      "step": 8556
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.205800195564305e-05,
+      "loss": 0.0617,
+      "step": 8557
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2054849645688145e-05,
+      "loss": 0.061,
+      "step": 8558
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2051697122525312e-05,
+      "loss": 0.0715,
+      "step": 8559
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2048544386481656e-05,
+      "loss": 0.068,
+      "step": 8560
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2045391437884297e-05,
+      "loss": 0.0452,
+      "step": 8561
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2042238277060384e-05,
+      "loss": 0.0656,
+      "step": 8562
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2039084904337082e-05,
+      "loss": 0.0514,
+      "step": 8563
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2035931320041576e-05,
+      "loss": 0.0701,
+      "step": 8564
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2032777524501086e-05,
+      "loss": 0.0721,
+      "step": 8565
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2029623518042837e-05,
+      "loss": 0.0605,
+      "step": 8566
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2026469300994085e-05,
+      "loss": 0.0537,
+      "step": 8567
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2023314873682109e-05,
+      "loss": 0.0667,
+      "step": 8568
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2020160236434203e-05,
+      "loss": 0.0706,
+      "step": 8569
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2017005389577695e-05,
+      "loss": 0.0704,
+      "step": 8570
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.201385033343992e-05,
+      "loss": 0.0607,
+      "step": 8571
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2010695068348238e-05,
+      "loss": 0.0756,
+      "step": 8572
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2007539594630046e-05,
+      "loss": 0.0643,
+      "step": 8573
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2004383912612745e-05,
+      "loss": 0.0604,
+      "step": 8574
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.2001228022623762e-05,
+      "loss": 0.0578,
+      "step": 8575
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1998071924990546e-05,
+      "loss": 0.073,
+      "step": 8576
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1994915620040574e-05,
+      "loss": 0.0612,
+      "step": 8577
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1991759108101335e-05,
+      "loss": 0.0685,
+      "step": 8578
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1988602389500345e-05,
+      "loss": 0.0704,
+      "step": 8579
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1985445464565139e-05,
+      "loss": 0.0784,
+      "step": 8580
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1982288333623277e-05,
+      "loss": 0.0761,
+      "step": 8581
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1979130997002337e-05,
+      "loss": 0.0727,
+      "step": 8582
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1975973455029917e-05,
+      "loss": 0.0634,
+      "step": 8583
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.197281570803364e-05,
+      "loss": 0.0756,
+      "step": 8584
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1969657756341149e-05,
+      "loss": 0.0734,
+      "step": 8585
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1966499600280109e-05,
+      "loss": 0.0623,
+      "step": 8586
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1963341240178206e-05,
+      "loss": 0.0811,
+      "step": 8587
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.196018267636314e-05,
+      "loss": 0.0675,
+      "step": 8588
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1957023909162643e-05,
+      "loss": 0.0688,
+      "step": 8589
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1953864938904467e-05,
+      "loss": 0.0719,
+      "step": 8590
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1950705765916377e-05,
+      "loss": 0.0612,
+      "step": 8591
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1947546390526168e-05,
+      "loss": 0.0635,
+      "step": 8592
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1944386813061644e-05,
+      "loss": 0.07,
+      "step": 8593
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1941227033850646e-05,
+      "loss": 0.0647,
+      "step": 8594
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1938067053221023e-05,
+      "loss": 0.0654,
+      "step": 8595
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1934906871500654e-05,
+      "loss": 0.0818,
+      "step": 8596
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1931746489017427e-05,
+      "loss": 0.0624,
+      "step": 8597
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1928585906099265e-05,
+      "loss": 0.0712,
+      "step": 8598
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1925425123074102e-05,
+      "loss": 0.074,
+      "step": 8599
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1922264140269897e-05,
+      "loss": 0.0647,
+      "step": 8600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1919102958014626e-05,
+      "loss": 0.0756,
+      "step": 8601
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1915941576636293e-05,
+      "loss": 0.0723,
+      "step": 8602
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1912779996462913e-05,
+      "loss": 0.0745,
+      "step": 8603
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1909618217822531e-05,
+      "loss": 0.0583,
+      "step": 8604
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1906456241043203e-05,
+      "loss": 0.0664,
+      "step": 8605
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1903294066453015e-05,
+      "loss": 0.0587,
+      "step": 8606
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1900131694380066e-05,
+      "loss": 0.0579,
+      "step": 8607
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1896969125152482e-05,
+      "loss": 0.0584,
+      "step": 8608
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1893806359098406e-05,
+      "loss": 0.0468,
+      "step": 8609
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1890643396545998e-05,
+      "loss": 0.0656,
+      "step": 8610
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1887480237823443e-05,
+      "loss": 0.0607,
+      "step": 8611
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1884316883258948e-05,
+      "loss": 0.0891,
+      "step": 8612
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1881153333180736e-05,
+      "loss": 0.0707,
+      "step": 8613
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1877989587917046e-05,
+      "loss": 0.0679,
+      "step": 8614
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1874825647796158e-05,
+      "loss": 0.073,
+      "step": 8615
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1871661513146344e-05,
+      "loss": 0.0764,
+      "step": 8616
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.1868497184295916e-05,
+      "loss": 0.0663,
+      "step": 8617
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1865332661573192e-05,
+      "loss": 0.0712,
+      "step": 8618
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1862167945306529e-05,
+      "loss": 0.058,
+      "step": 8619
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1859003035824289e-05,
+      "loss": 0.0706,
+      "step": 8620
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1855837933454852e-05,
+      "loss": 0.0688,
+      "step": 8621
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1852672638526628e-05,
+      "loss": 0.0575,
+      "step": 8622
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1849507151368045e-05,
+      "loss": 0.0605,
+      "step": 8623
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1846341472307548e-05,
+      "loss": 0.0637,
+      "step": 8624
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1843175601673602e-05,
+      "loss": 0.0623,
+      "step": 8625
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.184000953979469e-05,
+      "loss": 0.067,
+      "step": 8626
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1836843286999324e-05,
+      "loss": 0.0622,
+      "step": 8627
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1833676843616024e-05,
+      "loss": 0.0514,
+      "step": 8628
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1830510209973335e-05,
+      "loss": 0.0681,
+      "step": 8629
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1827343386399826e-05,
+      "loss": 0.0541,
+      "step": 8630
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1824176373224079e-05,
+      "loss": 0.0731,
+      "step": 8631
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1821009170774697e-05,
+      "loss": 0.0708,
+      "step": 8632
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1817841779380307e-05,
+      "loss": 0.0595,
+      "step": 8633
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.181467419936955e-05,
+      "loss": 0.0549,
+      "step": 8634
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1811506431071088e-05,
+      "loss": 0.0762,
+      "step": 8635
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1808338474813606e-05,
+      "loss": 0.0668,
+      "step": 8636
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.1805170330925808e-05,
+      "loss": 0.0692,
+      "step": 8637
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1802001999736412e-05,
+      "loss": 0.0661,
+      "step": 8638
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1798833481574156e-05,
+      "loss": 0.0662,
+      "step": 8639
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.179566477676781e-05,
+      "loss": 0.0638,
+      "step": 8640
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1792495885646148e-05,
+      "loss": 0.0894,
+      "step": 8641
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1789326808537968e-05,
+      "loss": 0.0748,
+      "step": 8642
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1786157545772088e-05,
+      "loss": 0.0652,
+      "step": 8643
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1782988097677349e-05,
+      "loss": 0.0613,
+      "step": 8644
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1779818464582604e-05,
+      "loss": 0.0648,
+      "step": 8645
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1776648646816733e-05,
+      "loss": 0.0702,
+      "step": 8646
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1773478644708631e-05,
+      "loss": 0.079,
+      "step": 8647
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.177030845858721e-05,
+      "loss": 0.0632,
+      "step": 8648
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1767138088781403e-05,
+      "loss": 0.0742,
+      "step": 8649
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1763967535620164e-05,
+      "loss": 0.06,
+      "step": 8650
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1760796799432465e-05,
+      "loss": 0.0673,
+      "step": 8651
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1757625880547296e-05,
+      "loss": 0.0661,
+      "step": 8652
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1754454779293669e-05,
+      "loss": 0.0564,
+      "step": 8653
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1751283496000607e-05,
+      "loss": 0.0732,
+      "step": 8654
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.174811203099716e-05,
+      "loss": 0.0632,
+      "step": 8655
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1744940384612394e-05,
+      "loss": 0.0663,
+      "step": 8656
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.1741768557175395e-05,
+      "loss": 0.0664,
+      "step": 8657
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1738596549015266e-05,
+      "loss": 0.0573,
+      "step": 8658
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1735424360461134e-05,
+      "loss": 0.062,
+      "step": 8659
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1732251991842129e-05,
+      "loss": 0.0644,
+      "step": 8660
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1729079443487422e-05,
+      "loss": 0.0601,
+      "step": 8661
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1725906715726185e-05,
+      "loss": 0.0881,
+      "step": 8662
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1722733808887619e-05,
+      "loss": 0.0674,
+      "step": 8663
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1719560723300935e-05,
+      "loss": 0.0473,
+      "step": 8664
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1716387459295375e-05,
+      "loss": 0.0512,
+      "step": 8665
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1713214017200185e-05,
+      "loss": 0.0652,
+      "step": 8666
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.171004039734464e-05,
+      "loss": 0.0689,
+      "step": 8667
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1706866600058025e-05,
+      "loss": 0.0677,
+      "step": 8668
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1703692625669654e-05,
+      "loss": 0.0736,
+      "step": 8669
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1700518474508853e-05,
+      "loss": 0.0683,
+      "step": 8670
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1697344146904964e-05,
+      "loss": 0.0826,
+      "step": 8671
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1694169643187346e-05,
+      "loss": 0.0764,
+      "step": 8672
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1690994963685389e-05,
+      "loss": 0.067,
+      "step": 8673
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1687820108728491e-05,
+      "loss": 0.0637,
+      "step": 8674
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1684645078646069e-05,
+      "loss": 0.0758,
+      "step": 8675
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1681469873767552e-05,
+      "loss": 0.0742,
+      "step": 8676
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.1678294494422406e-05,
+      "loss": 0.0781,
+      "step": 8677
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1675118940940096e-05,
+      "loss": 0.0618,
+      "step": 8678
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1671943213650114e-05,
+      "loss": 0.0678,
+      "step": 8679
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1668767312881967e-05,
+      "loss": 0.062,
+      "step": 8680
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1665591238965187e-05,
+      "loss": 0.0611,
+      "step": 8681
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.166241499222931e-05,
+      "loss": 0.0582,
+      "step": 8682
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1659238573003903e-05,
+      "loss": 0.0679,
+      "step": 8683
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.165606198161855e-05,
+      "loss": 0.0746,
+      "step": 8684
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1652885218402838e-05,
+      "loss": 0.0629,
+      "step": 8685
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1649708283686394e-05,
+      "loss": 0.054,
+      "step": 8686
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1646531177798847e-05,
+      "loss": 0.0669,
+      "step": 8687
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1643353901069849e-05,
+      "loss": 0.0524,
+      "step": 8688
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1640176453829066e-05,
+      "loss": 0.0479,
+      "step": 8689
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.163699883640619e-05,
+      "loss": 0.0801,
+      "step": 8690
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1633821049130924e-05,
+      "loss": 0.0673,
+      "step": 8691
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.163064309233299e-05,
+      "loss": 0.0579,
+      "step": 8692
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1627464966342125e-05,
+      "loss": 0.0645,
+      "step": 8693
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.162428667148809e-05,
+      "loss": 0.0558,
+      "step": 8694
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1621108208100657e-05,
+      "loss": 0.0764,
+      "step": 8695
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.1617929576509622e-05,
+      "loss": 0.0707,
+      "step": 8696
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.161475077704479e-05,
+      "loss": 0.0737,
+      "step": 8697
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1611571810035991e-05,
+      "loss": 0.0587,
+      "step": 8698
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1608392675813072e-05,
+      "loss": 0.0751,
+      "step": 8699
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1605213374705887e-05,
+      "loss": 0.0736,
+      "step": 8700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1602033907044324e-05,
+      "loss": 0.0612,
+      "step": 8701
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1598854273158274e-05,
+      "loss": 0.0643,
+      "step": 8702
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1595674473377655e-05,
+      "loss": 0.0611,
+      "step": 8703
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1592494508032393e-05,
+      "loss": 0.0628,
+      "step": 8704
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1589314377452436e-05,
+      "loss": 0.0791,
+      "step": 8705
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1586134081967755e-05,
+      "loss": 0.0685,
+      "step": 8706
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1582953621908328e-05,
+      "loss": 0.0504,
+      "step": 8707
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1579772997604157e-05,
+      "loss": 0.0714,
+      "step": 8708
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1576592209385257e-05,
+      "loss": 0.0823,
+      "step": 8709
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1573411257581659e-05,
+      "loss": 0.0653,
+      "step": 8710
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1570230142523419e-05,
+      "loss": 0.0742,
+      "step": 8711
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1567048864540602e-05,
+      "loss": 0.0752,
+      "step": 8712
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1563867423963291e-05,
+      "loss": 0.0539,
+      "step": 8713
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.156068582112159e-05,
+      "loss": 0.0756,
+      "step": 8714
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1557504056345613e-05,
+      "loss": 0.0573,
+      "step": 8715
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1554322129965495e-05,
+      "loss": 0.0779,
+      "step": 8716
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.1551140042311393e-05,
+      "loss": 0.065,
+      "step": 8717
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1547957793713471e-05,
+      "loss": 0.062,
+      "step": 8718
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1544775384501914e-05,
+      "loss": 0.0694,
+      "step": 8719
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1541592815006928e-05,
+      "loss": 0.0616,
+      "step": 8720
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1538410085558727e-05,
+      "loss": 0.053,
+      "step": 8721
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1535227196487545e-05,
+      "loss": 0.0695,
+      "step": 8722
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1532044148123636e-05,
+      "loss": 0.0643,
+      "step": 8723
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1528860940797268e-05,
+      "loss": 0.0919,
+      "step": 8724
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1525677574838728e-05,
+      "loss": 0.0598,
+      "step": 8725
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1522494050578308e-05,
+      "loss": 0.0765,
+      "step": 8726
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1519310368346335e-05,
+      "loss": 0.0672,
+      "step": 8727
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.151612652847314e-05,
+      "loss": 0.0634,
+      "step": 8728
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1512942531289071e-05,
+      "loss": 0.066,
+      "step": 8729
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1509758377124496e-05,
+      "loss": 0.0671,
+      "step": 8730
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1506574066309796e-05,
+      "loss": 0.0689,
+      "step": 8731
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1503389599175375e-05,
+      "loss": 0.0437,
+      "step": 8732
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1500204976051643e-05,
+      "loss": 0.0731,
+      "step": 8733
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1497020197269033e-05,
+      "loss": 0.0511,
+      "step": 8734
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1493835263157992e-05,
+      "loss": 0.0704,
+      "step": 8735
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1490650174048989e-05,
+      "loss": 0.0668,
+      "step": 8736
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.1487464930272496e-05,
+      "loss": 0.0759,
+      "step": 8737
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1484279532159012e-05,
+      "loss": 0.0635,
+      "step": 8738
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1481093980039048e-05,
+      "loss": 0.0692,
+      "step": 8739
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1477908274243135e-05,
+      "loss": 0.0898,
+      "step": 8740
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1474722415101813e-05,
+      "loss": 0.0507,
+      "step": 8741
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1471536402945647e-05,
+      "loss": 0.0541,
+      "step": 8742
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1468350238105203e-05,
+      "loss": 0.0505,
+      "step": 8743
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1465163920911082e-05,
+      "loss": 0.0573,
+      "step": 8744
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1461977451693887e-05,
+      "loss": 0.079,
+      "step": 8745
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1458790830784242e-05,
+      "loss": 0.0635,
+      "step": 8746
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1455604058512782e-05,
+      "loss": 0.0494,
+      "step": 8747
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1452417135210166e-05,
+      "loss": 0.0621,
+      "step": 8748
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1449230061207064e-05,
+      "loss": 0.0709,
+      "step": 8749
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1446042836834159e-05,
+      "loss": 0.0659,
+      "step": 8750
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1442855462422153e-05,
+      "loss": 0.0618,
+      "step": 8751
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1439667938301762e-05,
+      "loss": 0.0618,
+      "step": 8752
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1436480264803724e-05,
+      "loss": 0.0674,
+      "step": 8753
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1433292442258781e-05,
+      "loss": 0.0551,
+      "step": 8754
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1430104470997698e-05,
+      "loss": 0.0668,
+      "step": 8755
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.1426916351351251e-05,
+      "loss": 0.0671,
+      "step": 8756
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.142372808365024e-05,
+      "loss": 0.0524,
+      "step": 8757
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1420539668225475e-05,
+      "loss": 0.0655,
+      "step": 8758
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1417351105407775e-05,
+      "loss": 0.0468,
+      "step": 8759
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1414162395527978e-05,
+      "loss": 0.0598,
+      "step": 8760
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.141097353891695e-05,
+      "loss": 0.0761,
+      "step": 8761
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1407784535905554e-05,
+      "loss": 0.0809,
+      "step": 8762
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1404595386824679e-05,
+      "loss": 0.0681,
+      "step": 8763
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1401406092005226e-05,
+      "loss": 0.0589,
+      "step": 8764
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.139821665177811e-05,
+      "loss": 0.0567,
+      "step": 8765
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.139502706647426e-05,
+      "loss": 0.0608,
+      "step": 8766
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1391837336424625e-05,
+      "loss": 0.072,
+      "step": 8767
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1388647461960167e-05,
+      "loss": 0.0544,
+      "step": 8768
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1385457443411858e-05,
+      "loss": 0.0632,
+      "step": 8769
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1382267281110697e-05,
+      "loss": 0.0617,
+      "step": 8770
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1379076975387684e-05,
+      "loss": 0.0508,
+      "step": 8771
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1375886526573843e-05,
+      "loss": 0.0626,
+      "step": 8772
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1372695935000204e-05,
+      "loss": 0.0665,
+      "step": 8773
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1369505200997826e-05,
+      "loss": 0.0674,
+      "step": 8774
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1366314324897771e-05,
+      "loss": 0.0671,
+      "step": 8775
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1363123307031118e-05,
+      "loss": 0.0532,
+      "step": 8776
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.135993214772896e-05,
+      "loss": 0.0682,
+      "step": 8777
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1356740847322411e-05,
+      "loss": 0.0577,
+      "step": 8778
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1353549406142596e-05,
+      "loss": 0.0522,
+      "step": 8779
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1350357824520651e-05,
+      "loss": 0.0513,
+      "step": 8780
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1347166102787728e-05,
+      "loss": 0.0709,
+      "step": 8781
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1343974241274998e-05,
+      "loss": 0.0641,
+      "step": 8782
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1340782240313644e-05,
+      "loss": 0.0577,
+      "step": 8783
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1337590100234859e-05,
+      "loss": 0.0708,
+      "step": 8784
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1334397821369858e-05,
+      "loss": 0.0642,
+      "step": 8785
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1331205404049865e-05,
+      "loss": 0.0656,
+      "step": 8786
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.132801284860612e-05,
+      "loss": 0.0666,
+      "step": 8787
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1324820155369878e-05,
+      "loss": 0.0768,
+      "step": 8788
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1321627324672408e-05,
+      "loss": 0.0522,
+      "step": 8789
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1318434356844992e-05,
+      "loss": 0.0745,
+      "step": 8790
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1315241252218929e-05,
+      "loss": 0.0797,
+      "step": 8791
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1312048011125535e-05,
+      "loss": 0.0657,
+      "step": 8792
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.130885463389612e-05,
+      "loss": 0.064,
+      "step": 8793
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.130566112086204e-05,
+      "loss": 0.0721,
+      "step": 8794
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1302467472354642e-05,
+      "loss": 0.0696,
+      "step": 8795
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1299273688705294e-05,
+      "loss": 0.0741,
+      "step": 8796
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1296079770245378e-05,
+      "loss": 0.0754,
+      "step": 8797
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1292885717306293e-05,
+      "loss": 0.0699,
+      "step": 8798
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1289691530219445e-05,
+      "loss": 0.0682,
+      "step": 8799
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1286497209316256e-05,
+      "loss": 0.0692,
+      "step": 8800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.128330275492817e-05,
+      "loss": 0.0562,
+      "step": 8801
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1280108167386632e-05,
+      "loss": 0.0657,
+      "step": 8802
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1276913447023114e-05,
+      "loss": 0.0588,
+      "step": 8803
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.127371859416909e-05,
+      "loss": 0.0624,
+      "step": 8804
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1270523609156054e-05,
+      "loss": 0.0622,
+      "step": 8805
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1267328492315513e-05,
+      "loss": 0.0579,
+      "step": 8806
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1264133243978987e-05,
+      "loss": 0.0494,
+      "step": 8807
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1260937864478011e-05,
+      "loss": 0.054,
+      "step": 8808
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1257742354144132e-05,
+      "loss": 0.0727,
+      "step": 8809
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.125454671330891e-05,
+      "loss": 0.0532,
+      "step": 8810
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1251350942303918e-05,
+      "loss": 0.0633,
+      "step": 8811
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1248155041460749e-05,
+      "loss": 0.0583,
+      "step": 8812
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1244959011111002e-05,
+      "loss": 0.0524,
+      "step": 8813
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.124176285158629e-05,
+      "loss": 0.0682,
+      "step": 8814
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1238566563218244e-05,
+      "loss": 0.0729,
+      "step": 8815
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1235370146338505e-05,
+      "loss": 0.0616,
+      "step": 8816
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1232173601278728e-05,
+      "loss": 0.0648,
+      "step": 8817
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1228976928370583e-05,
+      "loss": 0.0676,
+      "step": 8818
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1225780127945748e-05,
+      "loss": 0.0552,
+      "step": 8819
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1222583200335923e-05,
+      "loss": 0.0603,
+      "step": 8820
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1219386145872812e-05,
+      "loss": 0.053,
+      "step": 8821
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1216188964888137e-05,
+      "loss": 0.0624,
+      "step": 8822
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1212991657713633e-05,
+      "loss": 0.085,
+      "step": 8823
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1209794224681048e-05,
+      "loss": 0.0614,
+      "step": 8824
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1206596666122146e-05,
+      "loss": 0.0752,
+      "step": 8825
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1203398982368691e-05,
+      "loss": 0.073,
+      "step": 8826
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1200201173752476e-05,
+      "loss": 0.0634,
+      "step": 8827
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1197003240605302e-05,
+      "loss": 0.0557,
+      "step": 8828
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1193805183258976e-05,
+      "loss": 0.0673,
+      "step": 8829
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1190607002045332e-05,
+      "loss": 0.0798,
+      "step": 8830
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1187408697296198e-05,
+      "loss": 0.066,
+      "step": 8831
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1184210269343432e-05,
+      "loss": 0.0684,
+      "step": 8832
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1181011718518895e-05,
+      "loss": 0.0696,
+      "step": 8833
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1177813045154465e-05,
+      "loss": 0.0659,
+      "step": 8834
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1174614249582027e-05,
+      "loss": 0.0654,
+      "step": 8835
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1171415332133488e-05,
+      "loss": 0.0824,
+      "step": 8836
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1168216293140764e-05,
+      "loss": 0.0789,
+      "step": 8837
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1165017132935774e-05,
+      "loss": 0.051,
+      "step": 8838
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1161817851850464e-05,
+      "loss": 0.0646,
+      "step": 8839
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1158618450216785e-05,
+      "loss": 0.0681,
+      "step": 8840
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1155418928366705e-05,
+      "loss": 0.0676,
+      "step": 8841
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1152219286632197e-05,
+      "loss": 0.0713,
+      "step": 8842
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.114901952534525e-05,
+      "loss": 0.0684,
+      "step": 8843
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1145819644837868e-05,
+      "loss": 0.0702,
+      "step": 8844
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1142619645442068e-05,
+      "loss": 0.0839,
+      "step": 8845
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1139419527489872e-05,
+      "loss": 0.0662,
+      "step": 8846
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1136219291313327e-05,
+      "loss": 0.0447,
+      "step": 8847
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1133018937244471e-05,
+      "loss": 0.0591,
+      "step": 8848
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1129818465615379e-05,
+      "loss": 0.0824,
+      "step": 8849
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1126617876758123e-05,
+      "loss": 0.0578,
+      "step": 8850
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1123417171004794e-05,
+      "loss": 0.0606,
+      "step": 8851
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1120216348687487e-05,
+      "loss": 0.0555,
+      "step": 8852
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1117015410138316e-05,
+      "loss": 0.0729,
+      "step": 8853
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1113814355689408e-05,
+      "loss": 0.0682,
+      "step": 8854
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1110613185672896e-05,
+      "loss": 0.0568,
+      "step": 8855
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1107411900420927e-05,
+      "loss": 0.0534,
+      "step": 8856
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.1104210500265668e-05,
+      "loss": 0.0718,
+      "step": 8857
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.110100898553929e-05,
+      "loss": 0.0571,
+      "step": 8858
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1097807356573969e-05,
+      "loss": 0.054,
+      "step": 8859
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1094605613701905e-05,
+      "loss": 0.0622,
+      "step": 8860
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.109140375725531e-05,
+      "loss": 0.0588,
+      "step": 8861
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1088201787566399e-05,
+      "loss": 0.0584,
+      "step": 8862
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1084999704967406e-05,
+      "loss": 0.0579,
+      "step": 8863
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1081797509790572e-05,
+      "loss": 0.0615,
+      "step": 8864
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1078595202368152e-05,
+      "loss": 0.0712,
+      "step": 8865
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1075392783032412e-05,
+      "loss": 0.0626,
+      "step": 8866
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1072190252115631e-05,
+      "loss": 0.054,
+      "step": 8867
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1068987609950099e-05,
+      "loss": 0.0515,
+      "step": 8868
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1065784856868116e-05,
+      "loss": 0.0522,
+      "step": 8869
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1062581993201994e-05,
+      "loss": 0.0568,
+      "step": 8870
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1059379019284057e-05,
+      "loss": 0.0762,
+      "step": 8871
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1056175935446642e-05,
+      "loss": 0.0721,
+      "step": 8872
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1052972742022093e-05,
+      "loss": 0.0519,
+      "step": 8873
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.104976943934277e-05,
+      "loss": 0.0619,
+      "step": 8874
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1046566027741048e-05,
+      "loss": 0.0584,
+      "step": 8875
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1043362507549297e-05,
+      "loss": 0.056,
+      "step": 8876
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1040158879099915e-05,
+      "loss": 0.0683,
+      "step": 8877
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.1036955142725309e-05,
+      "loss": 0.0795,
+      "step": 8878
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1033751298757889e-05,
+      "loss": 0.08,
+      "step": 8879
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1030547347530082e-05,
+      "loss": 0.0717,
+      "step": 8880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1027343289374322e-05,
+      "loss": 0.067,
+      "step": 8881
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1024139124623061e-05,
+      "loss": 0.0695,
+      "step": 8882
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1020934853608755e-05,
+      "loss": 0.0708,
+      "step": 8883
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1017730476663878e-05,
+      "loss": 0.0598,
+      "step": 8884
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1014525994120906e-05,
+      "loss": 0.071,
+      "step": 8885
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1011321406312335e-05,
+      "loss": 0.0656,
+      "step": 8886
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1008116713570664e-05,
+      "loss": 0.0605,
+      "step": 8887
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1004911916228414e-05,
+      "loss": 0.0565,
+      "step": 8888
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.1001707014618102e-05,
+      "loss": 0.0452,
+      "step": 8889
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0998502009072264e-05,
+      "loss": 0.0756,
+      "step": 8890
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0995296899923453e-05,
+      "loss": 0.0761,
+      "step": 8891
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.099209168750422e-05,
+      "loss": 0.0609,
+      "step": 8892
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0988886372147135e-05,
+      "loss": 0.0654,
+      "step": 8893
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0985680954184772e-05,
+      "loss": 0.0524,
+      "step": 8894
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0982475433949726e-05,
+      "loss": 0.0709,
+      "step": 8895
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0979269811774598e-05,
+      "loss": 0.0672,
+      "step": 8896
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0976064087991993e-05,
+      "loss": 0.0675,
+      "step": 8897
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0972858262934532e-05,
+      "loss": 0.0811,
+      "step": 8898
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.096965233693485e-05,
+      "loss": 0.0662,
+      "step": 8899
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0966446310325585e-05,
+      "loss": 0.0648,
+      "step": 8900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0963240183439393e-05,
+      "loss": 0.0589,
+      "step": 8901
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0960033956608931e-05,
+      "loss": 0.0654,
+      "step": 8902
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0956827630166877e-05,
+      "loss": 0.0684,
+      "step": 8903
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0953621204445915e-05,
+      "loss": 0.0599,
+      "step": 8904
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0950414679778736e-05,
+      "loss": 0.066,
+      "step": 8905
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0947208056498042e-05,
+      "loss": 0.0549,
+      "step": 8906
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.094400133493655e-05,
+      "loss": 0.061,
+      "step": 8907
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0940794515426986e-05,
+      "loss": 0.0604,
+      "step": 8908
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.093758759830208e-05,
+      "loss": 0.0723,
+      "step": 8909
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.093438058389458e-05,
+      "loss": 0.0589,
+      "step": 8910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0931173472537237e-05,
+      "loss": 0.0646,
+      "step": 8911
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0927966264562821e-05,
+      "loss": 0.0754,
+      "step": 8912
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0924758960304108e-05,
+      "loss": 0.0627,
+      "step": 8913
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0921551560093872e-05,
+      "loss": 0.0646,
+      "step": 8914
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0918344064264917e-05,
+      "loss": 0.0557,
+      "step": 8915
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0915136473150045e-05,
+      "loss": 0.062,
+      "step": 8916
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.091192878708207e-05,
+      "loss": 0.077,
+      "step": 8917
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.0908721006393818e-05,
+      "loss": 0.0583,
+      "step": 8918
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0905513131418124e-05,
+      "loss": 0.0659,
+      "step": 8919
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0902305162487829e-05,
+      "loss": 0.0692,
+      "step": 8920
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0899097099935789e-05,
+      "loss": 0.0566,
+      "step": 8921
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0895888944094868e-05,
+      "loss": 0.0667,
+      "step": 8922
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0892680695297932e-05,
+      "loss": 0.0748,
+      "step": 8923
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0889472353877876e-05,
+      "loss": 0.0583,
+      "step": 8924
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0886263920167586e-05,
+      "loss": 0.0635,
+      "step": 8925
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0883055394499962e-05,
+      "loss": 0.0582,
+      "step": 8926
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0879846777207915e-05,
+      "loss": 0.0663,
+      "step": 8927
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.087663806862437e-05,
+      "loss": 0.0585,
+      "step": 8928
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0873429269082256e-05,
+      "loss": 0.0617,
+      "step": 8929
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0870220378914515e-05,
+      "loss": 0.0718,
+      "step": 8930
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.086701139845409e-05,
+      "loss": 0.0579,
+      "step": 8931
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0863802328033947e-05,
+      "loss": 0.0674,
+      "step": 8932
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.086059316798705e-05,
+      "loss": 0.0664,
+      "step": 8933
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0857383918646378e-05,
+      "loss": 0.065,
+      "step": 8934
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0854174580344918e-05,
+      "loss": 0.0624,
+      "step": 8935
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0850965153415662e-05,
+      "loss": 0.0618,
+      "step": 8936
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0847755638191619e-05,
+      "loss": 0.0739,
+      "step": 8937
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.0844546035005803e-05,
+      "loss": 0.0512,
+      "step": 8938
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0841336344191236e-05,
+      "loss": 0.0677,
+      "step": 8939
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0838126566080946e-05,
+      "loss": 0.0559,
+      "step": 8940
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0834916701007985e-05,
+      "loss": 0.0826,
+      "step": 8941
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0831706749305396e-05,
+      "loss": 0.0643,
+      "step": 8942
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.082849671130624e-05,
+      "loss": 0.0538,
+      "step": 8943
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0825286587343582e-05,
+      "loss": 0.0715,
+      "step": 8944
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0822076377750506e-05,
+      "loss": 0.0606,
+      "step": 8945
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0818866082860093e-05,
+      "loss": 0.0542,
+      "step": 8946
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0815655703005446e-05,
+      "loss": 0.0668,
+      "step": 8947
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0812445238519655e-05,
+      "loss": 0.0645,
+      "step": 8948
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0809234689735845e-05,
+      "loss": 0.0529,
+      "step": 8949
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0806024056987132e-05,
+      "loss": 0.0635,
+      "step": 8950
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0802813340606647e-05,
+      "loss": 0.0646,
+      "step": 8951
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0799602540927532e-05,
+      "loss": 0.0671,
+      "step": 8952
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.079639165828293e-05,
+      "loss": 0.0642,
+      "step": 8953
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0793180693005998e-05,
+      "loss": 0.0646,
+      "step": 8954
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.07899696454299e-05,
+      "loss": 0.0593,
+      "step": 8955
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0786758515887814e-05,
+      "loss": 0.0508,
+      "step": 8956
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0783547304712913e-05,
+      "loss": 0.0651,
+      "step": 8957
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.0780336012238397e-05,
+      "loss": 0.0671,
+      "step": 8958
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.077712463879746e-05,
+      "loss": 0.0633,
+      "step": 8959
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0773913184723306e-05,
+      "loss": 0.0636,
+      "step": 8960
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0770701650349152e-05,
+      "loss": 0.0695,
+      "step": 8961
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0767490036008225e-05,
+      "loss": 0.075,
+      "step": 8962
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0764278342033759e-05,
+      "loss": 0.0706,
+      "step": 8963
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0761066568758985e-05,
+      "loss": 0.0776,
+      "step": 8964
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0757854716517156e-05,
+      "loss": 0.0875,
+      "step": 8965
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0754642785641531e-05,
+      "loss": 0.0637,
+      "step": 8966
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0751430776465373e-05,
+      "loss": 0.0607,
+      "step": 8967
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0748218689321954e-05,
+      "loss": 0.0633,
+      "step": 8968
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.074500652454456e-05,
+      "loss": 0.0659,
+      "step": 8969
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0741794282466472e-05,
+      "loss": 0.0692,
+      "step": 8970
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0738581963420994e-05,
+      "loss": 0.0626,
+      "step": 8971
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0735369567741425e-05,
+      "loss": 0.0706,
+      "step": 8972
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0732157095761084e-05,
+      "loss": 0.0505,
+      "step": 8973
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0728944547813289e-05,
+      "loss": 0.0669,
+      "step": 8974
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.072573192423137e-05,
+      "loss": 0.0655,
+      "step": 8975
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0722519225348662e-05,
+      "loss": 0.0483,
+      "step": 8976
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0719306451498513e-05,
+      "loss": 0.0692,
+      "step": 8977
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0716093603014268e-05,
+      "loss": 0.0637,
+      "step": 8978
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.0712880680229294e-05,
+      "loss": 0.0603,
+      "step": 8979
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0709667683476962e-05,
+      "loss": 0.0543,
+      "step": 8980
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0706454613090634e-05,
+      "loss": 0.067,
+      "step": 8981
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0703241469403704e-05,
+      "loss": 0.0708,
+      "step": 8982
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0700028252749559e-05,
+      "loss": 0.0616,
+      "step": 8983
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0696814963461599e-05,
+      "loss": 0.0645,
+      "step": 8984
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.069360160187323e-05,
+      "loss": 0.0676,
+      "step": 8985
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0690388168317863e-05,
+      "loss": 0.064,
+      "step": 8986
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.068717466312892e-05,
+      "loss": 0.0741,
+      "step": 8987
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0683961086639829e-05,
+      "loss": 0.0754,
+      "step": 8988
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0680747439184025e-05,
+      "loss": 0.0688,
+      "step": 8989
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.067753372109495e-05,
+      "loss": 0.0834,
+      "step": 8990
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0674319932706062e-05,
+      "loss": 0.0692,
+      "step": 8991
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0671106074350805e-05,
+      "loss": 0.059,
+      "step": 8992
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0667892146362655e-05,
+      "loss": 0.0695,
+      "step": 8993
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0664678149075078e-05,
+      "loss": 0.0614,
+      "step": 8994
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0661464082821558e-05,
+      "loss": 0.0605,
+      "step": 8995
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0658249947935579e-05,
+      "loss": 0.0717,
+      "step": 8996
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0655035744750632e-05,
+      "loss": 0.0652,
+      "step": 8997
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0651821473600218e-05,
+      "loss": 0.0754,
+      "step": 8998
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.0648607134817852e-05,
+      "loss": 0.0715,
+      "step": 8999
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.064539272873704e-05,
+      "loss": 0.0721,
+      "step": 9000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.064217825569131e-05,
+      "loss": 0.0543,
+      "step": 9001
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0638963716014184e-05,
+      "loss": 0.0602,
+      "step": 9002
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0635749110039202e-05,
+      "loss": 0.0773,
+      "step": 9003
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0632534438099906e-05,
+      "loss": 0.0859,
+      "step": 9004
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0629319700529843e-05,
+      "loss": 0.0659,
+      "step": 9005
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0626104897662572e-05,
+      "loss": 0.0645,
+      "step": 9006
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0622890029831656e-05,
+      "loss": 0.049,
+      "step": 9007
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0619675097370662e-05,
+      "loss": 0.0641,
+      "step": 9008
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0616460100613167e-05,
+      "loss": 0.0723,
+      "step": 9009
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0613245039892755e-05,
+      "loss": 0.0664,
+      "step": 9010
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0610029915543013e-05,
+      "loss": 0.061,
+      "step": 9011
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.060681472789754e-05,
+      "loss": 0.0737,
+      "step": 9012
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0603599477289939e-05,
+      "loss": 0.0525,
+      "step": 9013
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0600384164053816e-05,
+      "loss": 0.0608,
+      "step": 9014
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.059716878852279e-05,
+      "loss": 0.0538,
+      "step": 9015
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0593953351030481e-05,
+      "loss": 0.0591,
+      "step": 9016
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.059073785191052e-05,
+      "loss": 0.0704,
+      "step": 9017
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0587522291496543e-05,
+      "loss": 0.0737,
+      "step": 9018
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.0584306670122186e-05,
+      "loss": 0.0578,
+      "step": 9019
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0581090988121101e-05,
+      "loss": 0.0835,
+      "step": 9020
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0577875245826941e-05,
+      "loss": 0.0587,
+      "step": 9021
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0574659443573367e-05,
+      "loss": 0.0621,
+      "step": 9022
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0571443581694043e-05,
+      "loss": 0.078,
+      "step": 9023
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0568227660522645e-05,
+      "loss": 0.055,
+      "step": 9024
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0565011680392852e-05,
+      "loss": 0.061,
+      "step": 9025
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0561795641638346e-05,
+      "loss": 0.0743,
+      "step": 9026
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.055857954459282e-05,
+      "loss": 0.0665,
+      "step": 9027
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0555363389589966e-05,
+      "loss": 0.0789,
+      "step": 9028
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0552147176963496e-05,
+      "loss": 0.0757,
+      "step": 9029
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0548930907047117e-05,
+      "loss": 0.0662,
+      "step": 9030
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.054571458017454e-05,
+      "loss": 0.0565,
+      "step": 9031
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0542498196679482e-05,
+      "loss": 0.0519,
+      "step": 9032
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0539281756895683e-05,
+      "loss": 0.0591,
+      "step": 9033
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0536065261156864e-05,
+      "loss": 0.056,
+      "step": 9034
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0532848709796772e-05,
+      "loss": 0.0704,
+      "step": 9035
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0529632103149144e-05,
+      "loss": 0.0557,
+      "step": 9036
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0526415441547732e-05,
+      "loss": 0.0635,
+      "step": 9037
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.0523198725326296e-05,
+      "loss": 0.0638,
+      "step": 9038
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.051998195481859e-05,
+      "loss": 0.0603,
+      "step": 9039
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0516765130358389e-05,
+      "loss": 0.0634,
+      "step": 9040
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.051354825227946e-05,
+      "loss": 0.0814,
+      "step": 9041
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0510331320915581e-05,
+      "loss": 0.0613,
+      "step": 9042
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0507114336600539e-05,
+      "loss": 0.0618,
+      "step": 9043
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0503897299668118e-05,
+      "loss": 0.0605,
+      "step": 9044
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0500680210452122e-05,
+      "loss": 0.0651,
+      "step": 9045
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0497463069286343e-05,
+      "loss": 0.0771,
+      "step": 9046
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0494245876504588e-05,
+      "loss": 0.0634,
+      "step": 9047
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0491028632440667e-05,
+      "loss": 0.064,
+      "step": 9048
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.04878113374284e-05,
+      "loss": 0.0474,
+      "step": 9049
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0484593991801603e-05,
+      "loss": 0.069,
+      "step": 9050
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0481376595894108e-05,
+      "loss": 0.068,
+      "step": 9051
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0478159150039745e-05,
+      "loss": 0.06,
+      "step": 9052
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0474941654572351e-05,
+      "loss": 0.0635,
+      "step": 9053
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0471724109825767e-05,
+      "loss": 0.0777,
+      "step": 9054
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.046850651613384e-05,
+      "loss": 0.0524,
+      "step": 9055
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0465288873830424e-05,
+      "loss": 0.0507,
+      "step": 9056
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.0462071183249376e-05,
+      "loss": 0.0565,
+      "step": 9057
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.045885344472456e-05,
+      "loss": 0.0574,
+      "step": 9058
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.045563565858984e-05,
+      "loss": 0.0545,
+      "step": 9059
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.045241782517909e-05,
+      "loss": 0.0555,
+      "step": 9060
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0449199944826185e-05,
+      "loss": 0.0563,
+      "step": 9061
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0445982017865013e-05,
+      "loss": 0.0647,
+      "step": 9062
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0442764044629459e-05,
+      "loss": 0.0508,
+      "step": 9063
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0439546025453411e-05,
+      "loss": 0.0682,
+      "step": 9064
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0436327960670765e-05,
+      "loss": 0.0583,
+      "step": 9065
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0433109850615429e-05,
+      "loss": 0.0881,
+      "step": 9066
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0429891695621304e-05,
+      "loss": 0.0666,
+      "step": 9067
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0426673496022303e-05,
+      "loss": 0.0494,
+      "step": 9068
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0423455252152336e-05,
+      "loss": 0.074,
+      "step": 9069
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0420236964345332e-05,
+      "loss": 0.0715,
+      "step": 9070
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0417018632935207e-05,
+      "loss": 0.0674,
+      "step": 9071
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0413800258255895e-05,
+      "loss": 0.0627,
+      "step": 9072
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0410581840641324e-05,
+      "loss": 0.062,
+      "step": 9073
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.040736338042544e-05,
+      "loss": 0.0646,
+      "step": 9074
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.040414487794218e-05,
+      "loss": 0.0671,
+      "step": 9075
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.040092633352549e-05,
+      "loss": 0.075,
+      "step": 9076
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0397707747509323e-05,
+      "loss": 0.0584,
+      "step": 9077
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0394489120227631e-05,
+      "loss": 0.0671,
+      "step": 9078
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.0391270452014382e-05,
+      "loss": 0.0568,
+      "step": 9079
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.038805174320353e-05,
+      "loss": 0.0506,
+      "step": 9080
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.038483299412905e-05,
+      "loss": 0.0786,
+      "step": 9081
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.038161420512491e-05,
+      "loss": 0.0556,
+      "step": 9082
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0378395376525088e-05,
+      "loss": 0.0596,
+      "step": 9083
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0375176508663567e-05,
+      "loss": 0.0711,
+      "step": 9084
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.037195760187433e-05,
+      "loss": 0.0531,
+      "step": 9085
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0368738656491361e-05,
+      "loss": 0.062,
+      "step": 9086
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0365519672848658e-05,
+      "loss": 0.0709,
+      "step": 9087
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0362300651280217e-05,
+      "loss": 0.0677,
+      "step": 9088
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0359081592120039e-05,
+      "loss": 0.0674,
+      "step": 9089
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0355862495702125e-05,
+      "loss": 0.0595,
+      "step": 9090
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0352643362360486e-05,
+      "loss": 0.0646,
+      "step": 9091
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0349424192429137e-05,
+      "loss": 0.0718,
+      "step": 9092
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0346204986242086e-05,
+      "loss": 0.0561,
+      "step": 9093
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0342985744133358e-05,
+      "loss": 0.058,
+      "step": 9094
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0339766466436976e-05,
+      "loss": 0.068,
+      "step": 9095
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0336547153486968e-05,
+      "loss": 0.0604,
+      "step": 9096
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0333327805617367e-05,
+      "loss": 0.0594,
+      "step": 9097
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0330108423162197e-05,
+      "loss": 0.0602,
+      "step": 9098
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.0326889006455505e-05,
+      "loss": 0.0742,
+      "step": 9099
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0323669555831332e-05,
+      "loss": 0.0673,
+      "step": 9100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0320450071623724e-05,
+      "loss": 0.0787,
+      "step": 9101
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0317230554166725e-05,
+      "loss": 0.0544,
+      "step": 9102
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0314011003794386e-05,
+      "loss": 0.0595,
+      "step": 9103
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0310791420840771e-05,
+      "loss": 0.0649,
+      "step": 9104
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0307571805639931e-05,
+      "loss": 0.0596,
+      "step": 9105
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.030435215852593e-05,
+      "loss": 0.0596,
+      "step": 9106
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0301132479832836e-05,
+      "loss": 0.0647,
+      "step": 9107
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0297912769894713e-05,
+      "loss": 0.0646,
+      "step": 9108
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0294693029045636e-05,
+      "loss": 0.0581,
+      "step": 9109
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0291473257619685e-05,
+      "loss": 0.0554,
+      "step": 9110
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0288253455950925e-05,
+      "loss": 0.0654,
+      "step": 9111
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0285033624373453e-05,
+      "loss": 0.0588,
+      "step": 9112
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0281813763221344e-05,
+      "loss": 0.0643,
+      "step": 9113
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0278593872828686e-05,
+      "loss": 0.0629,
+      "step": 9114
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0275373953529572e-05,
+      "loss": 0.0549,
+      "step": 9115
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0272154005658098e-05,
+      "loss": 0.073,
+      "step": 9116
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0268934029548357e-05,
+      "loss": 0.0604,
+      "step": 9117
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0265714025534451e-05,
+      "loss": 0.0595,
+      "step": 9118
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.0262493993950476e-05,
+      "loss": 0.0568,
+      "step": 9119
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0259273935130547e-05,
+      "loss": 0.0641,
+      "step": 9120
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0256053849408768e-05,
+      "loss": 0.072,
+      "step": 9121
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0252833737119245e-05,
+      "loss": 0.0903,
+      "step": 9122
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.02496135985961e-05,
+      "loss": 0.0574,
+      "step": 9123
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0246393434173446e-05,
+      "loss": 0.0643,
+      "step": 9124
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.02431732441854e-05,
+      "loss": 0.0735,
+      "step": 9125
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0239953028966087e-05,
+      "loss": 0.072,
+      "step": 9126
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.023673278884963e-05,
+      "loss": 0.0763,
+      "step": 9127
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0233512524170155e-05,
+      "loss": 0.064,
+      "step": 9128
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0230292235261796e-05,
+      "loss": 0.0729,
+      "step": 9129
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.022707192245868e-05,
+      "loss": 0.0506,
+      "step": 9130
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0223851586094942e-05,
+      "loss": 0.0772,
+      "step": 9131
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0220631226504718e-05,
+      "loss": 0.0669,
+      "step": 9132
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0217410844022154e-05,
+      "loss": 0.0613,
+      "step": 9133
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0214190438981383e-05,
+      "loss": 0.0632,
+      "step": 9134
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0210970011716558e-05,
+      "loss": 0.0605,
+      "step": 9135
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0207749562561817e-05,
+      "loss": 0.0647,
+      "step": 9136
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0204529091851314e-05,
+      "loss": 0.0589,
+      "step": 9137
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.02013085999192e-05,
+      "loss": 0.0717,
+      "step": 9138
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0198088087099624e-05,
+      "loss": 0.064,
+      "step": 9139
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0194867553726743e-05,
+      "loss": 0.0566,
+      "step": 9140
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0191647000134718e-05,
+      "loss": 0.0594,
+      "step": 9141
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0188426426657705e-05,
+      "loss": 0.0728,
+      "step": 9142
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0185205833629866e-05,
+      "loss": 0.0753,
+      "step": 9143
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0181985221385365e-05,
+      "loss": 0.0541,
+      "step": 9144
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0178764590258363e-05,
+      "loss": 0.0646,
+      "step": 9145
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0175543940583039e-05,
+      "loss": 0.0756,
+      "step": 9146
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0172323272693552e-05,
+      "loss": 0.0654,
+      "step": 9147
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0169102586924077e-05,
+      "loss": 0.0703,
+      "step": 9148
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0165881883608785e-05,
+      "loss": 0.0574,
+      "step": 9149
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0162661163081856e-05,
+      "loss": 0.0629,
+      "step": 9150
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0159440425677466e-05,
+      "loss": 0.0576,
+      "step": 9151
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0156219671729788e-05,
+      "loss": 0.0587,
+      "step": 9152
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0152998901573003e-05,
+      "loss": 0.0575,
+      "step": 9153
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.01497781155413e-05,
+      "loss": 0.0592,
+      "step": 9154
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0146557313968858e-05,
+      "loss": 0.0544,
+      "step": 9155
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0143336497189861e-05,
+      "loss": 0.0604,
+      "step": 9156
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0140115665538502e-05,
+      "loss": 0.0704,
+      "step": 9157
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.013689481934896e-05,
+      "loss": 0.0543,
+      "step": 9158
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0133673958955433e-05,
+      "loss": 0.0692,
+      "step": 9159
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0130453084692108e-05,
+      "loss": 0.0585,
+      "step": 9160
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0127232196893176e-05,
+      "loss": 0.067,
+      "step": 9161
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0124011295892837e-05,
+      "loss": 0.0526,
+      "step": 9162
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0120790382025282e-05,
+      "loss": 0.0588,
+      "step": 9163
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0117569455624709e-05,
+      "loss": 0.0538,
+      "step": 9164
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0114348517025318e-05,
+      "loss": 0.0714,
+      "step": 9165
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0111127566561305e-05,
+      "loss": 0.0645,
+      "step": 9166
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0107906604566874e-05,
+      "loss": 0.068,
+      "step": 9167
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.010468563137623e-05,
+      "loss": 0.0627,
+      "step": 9168
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0101464647323567e-05,
+      "loss": 0.0471,
+      "step": 9169
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0098243652743093e-05,
+      "loss": 0.0688,
+      "step": 9170
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0095022647969018e-05,
+      "loss": 0.0707,
+      "step": 9171
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0091801633335544e-05,
+      "loss": 0.059,
+      "step": 9172
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0088580609176874e-05,
+      "loss": 0.0719,
+      "step": 9173
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0085359575827226e-05,
+      "loss": 0.0602,
+      "step": 9174
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0082138533620803e-05,
+      "loss": 0.0662,
+      "step": 9175
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0078917482891817e-05,
+      "loss": 0.06,
+      "step": 9176
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0075696423974478e-05,
+      "loss": 0.0651,
+      "step": 9177
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0072475357202998e-05,
+      "loss": 0.0629,
+      "step": 9178
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.006925428291159e-05,
+      "loss": 0.0645,
+      "step": 9179
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.006603320143447e-05,
+      "loss": 0.0553,
+      "step": 9180
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.006281211310585e-05,
+      "loss": 0.077,
+      "step": 9181
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0059591018259945e-05,
+      "loss": 0.0519,
+      "step": 9182
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.005636991723097e-05,
+      "loss": 0.0596,
+      "step": 9183
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0053148810353146e-05,
+      "loss": 0.0696,
+      "step": 9184
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0049927697960682e-05,
+      "loss": 0.0779,
+      "step": 9185
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.00467065803878e-05,
+      "loss": 0.0779,
+      "step": 9186
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0043485457968717e-05,
+      "loss": 0.0624,
+      "step": 9187
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0040264331037655e-05,
+      "loss": 0.0566,
+      "step": 9188
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0037043199928827e-05,
+      "loss": 0.0567,
+      "step": 9189
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0033822064976457e-05,
+      "loss": 0.0719,
+      "step": 9190
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0030600926514763e-05,
+      "loss": 0.0615,
+      "step": 9191
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0027379784877969e-05,
+      "loss": 0.0605,
+      "step": 9192
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0024158640400288e-05,
+      "loss": 0.0582,
+      "step": 9193
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0020937493415947e-05,
+      "loss": 0.0687,
+      "step": 9194
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0017716344259165e-05,
+      "loss": 0.0642,
+      "step": 9195
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0014495193264162e-05,
+      "loss": 0.0662,
+      "step": 9196
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0011274040765162e-05,
+      "loss": 0.0557,
+      "step": 9197
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0008052887096386e-05,
+      "loss": 0.0809,
+      "step": 9198
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0004831732592053e-05,
+      "loss": 0.0553,
+      "step": 9199
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.0001610577586386e-05,
+      "loss": 0.0519,
+      "step": 9200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.998389422413612e-06,
+      "loss": 0.054,
+      "step": 9201
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.99516826740795e-06,
+      "loss": 0.0657,
+      "step": 9202
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.991947112903617e-06,
+      "loss": 0.0657,
+      "step": 9203
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.988725959234841e-06,
+      "loss": 0.066,
+      "step": 9204
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.985504806735841e-06,
+      "loss": 0.0443,
+      "step": 9205
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.982283655740837e-06,
+      "loss": 0.0743,
+      "step": 9206
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.979062506584055e-06,
+      "loss": 0.0645,
+      "step": 9207
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.975841359599712e-06,
+      "loss": 0.0623,
+      "step": 9208
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.972620215122035e-06,
+      "loss": 0.0558,
+      "step": 9209
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.969399073485242e-06,
+      "loss": 0.0734,
+      "step": 9210
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.966177935023545e-06,
+      "loss": 0.0601,
+      "step": 9211
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.962956800071175e-06,
+      "loss": 0.0645,
+      "step": 9212
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.959735668962347e-06,
+      "loss": 0.0656,
+      "step": 9213
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.956514542031286e-06,
+      "loss": 0.0603,
+      "step": 9214
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.953293419612205e-06,
+      "loss": 0.0569,
+      "step": 9215
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.950072302039321e-06,
+      "loss": 0.0638,
+      "step": 9216
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.94685118964686e-06,
+      "loss": 0.0668,
+      "step": 9217
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.943630082769032e-06,
+      "loss": 0.0715,
+      "step": 9218
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.940408981740059e-06,
+      "loss": 0.0515,
+      "step": 9219
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.937187886894153e-06,
+      "loss": 0.06,
+      "step": 9220
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.933966798565531e-06,
+      "loss": 0.0714,
+      "step": 9221
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.930745717088413e-06,
+      "loss": 0.0641,
+      "step": 9222
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.927524642797003e-06,
+      "loss": 0.0721,
+      "step": 9223
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.924303576025524e-06,
+      "loss": 0.0626,
+      "step": 9224
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.921082517108185e-06,
+      "loss": 0.063,
+      "step": 9225
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.9178614663792e-06,
+      "loss": 0.0625,
+      "step": 9226
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.91464042417278e-06,
+      "loss": 0.0568,
+      "step": 9227
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.911419390823128e-06,
+      "loss": 0.0756,
+      "step": 9228
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.908198366664461e-06,
+      "loss": 0.0552,
+      "step": 9229
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.904977352030984e-06,
+      "loss": 0.0623,
+      "step": 9230
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.901756347256908e-06,
+      "loss": 0.056,
+      "step": 9231
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.898535352676438e-06,
+      "loss": 0.0735,
+      "step": 9232
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.895314368623773e-06,
+      "loss": 0.0615,
+      "step": 9233
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.892093395433127e-06,
+      "loss": 0.0557,
+      "step": 9234
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.888872433438695e-06,
+      "loss": 0.0625,
+      "step": 9235
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.885651482974683e-06,
+      "loss": 0.0613,
+      "step": 9236
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.882430544375294e-06,
+      "loss": 0.0667,
+      "step": 9237
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.879209617974721e-06,
+      "loss": 0.0576,
+      "step": 9238
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.87598870410717e-06,
+      "loss": 0.0569,
+      "step": 9239
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.872767803106826e-06,
+      "loss": 0.0548,
+      "step": 9240
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.869546915307897e-06,
+      "loss": 0.0559,
+      "step": 9241
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.866326041044574e-06,
+      "loss": 0.0704,
+      "step": 9242
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.863105180651042e-06,
+      "loss": 0.0733,
+      "step": 9243
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.859884334461503e-06,
+      "loss": 0.0665,
+      "step": 9244
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.856663502810137e-06,
+      "loss": 0.0673,
+      "step": 9245
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.853442686031145e-06,
+      "loss": 0.0715,
+      "step": 9246
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.8502218844587e-06,
+      "loss": 0.0527,
+      "step": 9247
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.847001098426999e-06,
+      "loss": 0.0577,
+      "step": 9248
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.843780328270217e-06,
+      "loss": 0.0684,
+      "step": 9249
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.840559574322538e-06,
+      "loss": 0.0585,
+      "step": 9250
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.837338836918148e-06,
+      "loss": 0.0613,
+      "step": 9251
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.834118116391216e-06,
+      "loss": 0.0512,
+      "step": 9252
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.830897413075926e-06,
+      "loss": 0.0815,
+      "step": 9253
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.827676727306453e-06,
+      "loss": 0.0621,
+      "step": 9254
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.824456059416964e-06,
+      "loss": 0.0579,
+      "step": 9255
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.82123540974164e-06,
+      "loss": 0.0634,
+      "step": 9256
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.818014778614638e-06,
+      "loss": 0.0637,
+      "step": 9257
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.814794166370137e-06,
+      "loss": 0.0605,
+      "step": 9258
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.8115735733423e-06,
+      "loss": 0.0566,
+      "step": 9259
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.808352999865285e-06,
+      "loss": 0.0604,
+      "step": 9260
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.805132446273258e-06,
+      "loss": 0.058,
+      "step": 9261
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.801911912900378e-06,
+      "loss": 0.0491,
+      "step": 9262
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.798691400080804e-06,
+      "loss": 0.0638,
+      "step": 9263
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.795470908148686e-06,
+      "loss": 0.0659,
+      "step": 9264
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.792250437438186e-06,
+      "loss": 0.0486,
+      "step": 9265
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.789029988283447e-06,
+      "loss": 0.064,
+      "step": 9266
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.785809561018617e-06,
+      "loss": 0.0674,
+      "step": 9267
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.78258915597785e-06,
+      "loss": 0.0714,
+      "step": 9268
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.779368773495282e-06,
+      "loss": 0.0702,
+      "step": 9269
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.776148413905061e-06,
+      "loss": 0.0813,
+      "step": 9270
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.772928077541325e-06,
+      "loss": 0.0673,
+      "step": 9271
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.769707764738206e-06,
+      "loss": 0.0633,
+      "step": 9272
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.766487475829848e-06,
+      "loss": 0.0544,
+      "step": 9273
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.763267211150372e-06,
+      "loss": 0.0771,
+      "step": 9274
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.760046971033914e-06,
+      "loss": 0.054,
+      "step": 9275
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.756826755814603e-06,
+      "loss": 0.0579,
+      "step": 9276
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.753606565826556e-06,
+      "loss": 0.0593,
+      "step": 9277
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.750386401403902e-06,
+      "loss": 0.0489,
+      "step": 9278
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.747166262880754e-06,
+      "loss": 0.067,
+      "step": 9279
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.743946150591237e-06,
+      "loss": 0.0573,
+      "step": 9280
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.740726064869458e-06,
+      "loss": 0.0491,
+      "step": 9281
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.737506006049526e-06,
+      "loss": 0.0644,
+      "step": 9282
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.734285974465554e-06,
+      "loss": 0.0497,
+      "step": 9283
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.731065970451645e-06,
+      "loss": 0.0707,
+      "step": 9284
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.727845994341907e-06,
+      "loss": 0.0582,
+      "step": 9285
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.72462604647043e-06,
+      "loss": 0.0625,
+      "step": 9286
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.721406127171315e-06,
+      "loss": 0.0655,
+      "step": 9287
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.718186236778661e-06,
+      "loss": 0.0582,
+      "step": 9288
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.71496637562655e-06,
+      "loss": 0.0638,
+      "step": 9289
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.711746544049077e-06,
+      "loss": 0.0532,
+      "step": 9290
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.70852674238032e-06,
+      "loss": 0.0497,
+      "step": 9291
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.705306970954365e-06,
+      "loss": 0.0679,
+      "step": 9292
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.702087230105292e-06,
+      "loss": 0.0591,
+      "step": 9293
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.698867520167167e-06,
+      "loss": 0.067,
+      "step": 9294
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.695647841474073e-06,
+      "loss": 0.075,
+      "step": 9295
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.69242819436007e-06,
+      "loss": 0.0677,
+      "step": 9296
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.689208579159232e-06,
+      "loss": 0.0578,
+      "step": 9297
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.685988996205616e-06,
+      "loss": 0.0558,
+      "step": 9298
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.682769445833278e-06,
+      "loss": 0.0575,
+      "step": 9299
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.679549928376281e-06,
+      "loss": 0.0553,
+      "step": 9300
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.676330444168668e-06,
+      "loss": 0.0596,
+      "step": 9301
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.673110993544498e-06,
+      "loss": 0.0707,
+      "step": 9302
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.669891576837807e-06,
+      "loss": 0.0645,
+      "step": 9303
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.666672194382639e-06,
+      "loss": 0.0585,
+      "step": 9304
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.663452846513035e-06,
+      "loss": 0.0717,
+      "step": 9305
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.660233533563024e-06,
+      "loss": 0.0906,
+      "step": 9306
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.657014255866643e-06,
+      "loss": 0.0525,
+      "step": 9307
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.653795013757914e-06,
+      "loss": 0.0642,
+      "step": 9308
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.650575807570868e-06,
+      "loss": 0.0674,
+      "step": 9309
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.647356637639518e-06,
+      "loss": 0.0601,
+      "step": 9310
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.644137504297876e-06,
+      "loss": 0.0575,
+      "step": 9311
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.640918407879965e-06,
+      "loss": 0.0588,
+      "step": 9312
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.637699348719783e-06,
+      "loss": 0.0562,
+      "step": 9313
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.634480327151344e-06,
+      "loss": 0.0614,
+      "step": 9314
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.631261343508642e-06,
+      "loss": 0.0552,
+      "step": 9315
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.628042398125673e-06,
+      "loss": 0.0574,
+      "step": 9316
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.624823491336438e-06,
+      "loss": 0.0626,
+      "step": 9317
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.621604623474912e-06,
+      "loss": 0.0615,
+      "step": 9318
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.618385794875094e-06,
+      "loss": 0.0608,
+      "step": 9319
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.615167005870955e-06,
+      "loss": 0.0613,
+      "step": 9320
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.611948256796471e-06,
+      "loss": 0.0628,
+      "step": 9321
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.608729547985623e-06,
+      "loss": 0.0761,
+      "step": 9322
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.60551087977237e-06,
+      "loss": 0.0638,
+      "step": 9323
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.60229225249068e-06,
+      "loss": 0.0696,
+      "step": 9324
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.599073666474516e-06,
+      "loss": 0.0783,
+      "step": 9325
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.595855122057825e-06,
+      "loss": 0.0627,
+      "step": 9326
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.592636619574564e-06,
+      "loss": 0.0784,
+      "step": 9327
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.589418159358677e-06,
+      "loss": 0.0499,
+      "step": 9328
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.58619974174411e-06,
+      "loss": 0.0632,
+      "step": 9329
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.582981367064793e-06,
+      "loss": 0.0684,
+      "step": 9330
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.579763035654671e-06,
+      "loss": 0.0676,
+      "step": 9331
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.576544747847665e-06,
+      "loss": 0.0711,
+      "step": 9332
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.573326503977698e-06,
+      "loss": 0.062,
+      "step": 9333
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.5701083043787e-06,
+      "loss": 0.073,
+      "step": 9334
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.566890149384573e-06,
+      "loss": 0.0551,
+      "step": 9335
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.563672039329237e-06,
+      "loss": 0.0626,
+      "step": 9336
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.560453974546594e-06,
+      "loss": 0.0574,
+      "step": 9337
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.557235955370545e-06,
+      "loss": 0.0767,
+      "step": 9338
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.55401798213499e-06,
+      "loss": 0.0731,
+      "step": 9339
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.550800055173815e-06,
+      "loss": 0.0675,
+      "step": 9340
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.547582174820913e-06,
+      "loss": 0.0659,
+      "step": 9341
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.544364341410166e-06,
+      "loss": 0.0658,
+      "step": 9342
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.541146555275444e-06,
+      "loss": 0.0601,
+      "step": 9343
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.537928816750629e-06,
+      "loss": 0.069,
+      "step": 9344
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.534711126169578e-06,
+      "loss": 0.0582,
+      "step": 9345
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.531493483866163e-06,
+      "loss": 0.0568,
+      "step": 9346
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.528275890174234e-06,
+      "loss": 0.056,
+      "step": 9347
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.525058345427652e-06,
+      "loss": 0.0571,
+      "step": 9348
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.521840849960256e-06,
+      "loss": 0.0635,
+      "step": 9349
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.51862340410589e-06,
+      "loss": 0.0707,
+      "step": 9350
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.515406008198398e-06,
+      "loss": 0.0706,
+      "step": 9351
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.512188662571601e-06,
+      "loss": 0.0753,
+      "step": 9352
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.508971367559336e-06,
+      "loss": 0.0707,
+      "step": 9353
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.505754123495417e-06,
+      "loss": 0.0566,
+      "step": 9354
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.502536930713659e-06,
+      "loss": 0.0686,
+      "step": 9355
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.499319789547883e-06,
+      "loss": 0.0734,
+      "step": 9356
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.496102700331882e-06,
+      "loss": 0.0642,
+      "step": 9357
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.492885663399465e-06,
+      "loss": 0.0699,
+      "step": 9358
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.489668679084424e-06,
+      "loss": 0.0627,
+      "step": 9359
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.486451747720544e-06,
+      "loss": 0.0721,
+      "step": 9360
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.483234869641616e-06,
+      "loss": 0.0697,
+      "step": 9361
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.480018045181412e-06,
+      "loss": 0.0716,
+      "step": 9362
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.476801274673708e-06,
+      "loss": 0.0699,
+      "step": 9363
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.473584558452273e-06,
+      "loss": 0.074,
+      "step": 9364
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.47036789685086e-06,
+      "loss": 0.0613,
+      "step": 9365
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.467151290203233e-06,
+      "loss": 0.0667,
+      "step": 9366
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.463934738843135e-06,
+      "loss": 0.0721,
+      "step": 9367
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.46071824310432e-06,
+      "loss": 0.072,
+      "step": 9368
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.457501803320518e-06,
+      "loss": 0.0442,
+      "step": 9369
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.454285419825464e-06,
+      "loss": 0.0589,
+      "step": 9370
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.451069092952888e-06,
+      "loss": 0.0587,
+      "step": 9371
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.447852823036506e-06,
+      "loss": 0.0776,
+      "step": 9372
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.444636610410036e-06,
+      "loss": 0.0525,
+      "step": 9373
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.441420455407185e-06,
+      "loss": 0.0774,
+      "step": 9374
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.438204358361657e-06,
+      "loss": 0.0568,
+      "step": 9375
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.434988319607153e-06,
+      "loss": 0.0708,
+      "step": 9376
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.431772339477356e-06,
+      "loss": 0.0676,
+      "step": 9377
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.428556418305958e-06,
+      "loss": 0.0724,
+      "step": 9378
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.425340556426635e-06,
+      "loss": 0.0818,
+      "step": 9379
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.422124754173062e-06,
+      "loss": 0.0652,
+      "step": 9380
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.418909011878904e-06,
+      "loss": 0.055,
+      "step": 9381
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.415693329877818e-06,
+      "loss": 0.0735,
+      "step": 9382
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.412477708503462e-06,
+      "loss": 0.062,
+      "step": 9383
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.40926214808948e-06,
+      "loss": 0.0626,
+      "step": 9384
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.40604664896952e-06,
+      "loss": 0.0807,
+      "step": 9385
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.402831211477214e-06,
+      "loss": 0.0688,
+      "step": 9386
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.399615835946185e-06,
+      "loss": 0.0711,
+      "step": 9387
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.396400522710066e-06,
+      "loss": 0.053,
+      "step": 9388
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.393185272102462e-06,
+      "loss": 0.0648,
+      "step": 9389
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.389970084456992e-06,
+      "loss": 0.0496,
+      "step": 9390
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.38675496010725e-06,
+      "loss": 0.0667,
+      "step": 9391
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.383539899386837e-06,
+      "loss": 0.0577,
+      "step": 9392
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.380324902629343e-06,
+      "loss": 0.0562,
+      "step": 9393
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.377109970168348e-06,
+      "loss": 0.0639,
+      "step": 9394
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.37389510233743e-06,
+      "loss": 0.0623,
+      "step": 9395
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.370680299470156e-06,
+      "loss": 0.0683,
+      "step": 9396
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.367465561900097e-06,
+      "loss": 0.067,
+      "step": 9397
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.364250889960803e-06,
+      "loss": 0.0573,
+      "step": 9398
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.36103628398582e-06,
+      "loss": 0.0567,
+      "step": 9399
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.357821744308696e-06,
+      "loss": 0.0796,
+      "step": 9400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.354607271262961e-06,
+      "loss": 0.0657,
+      "step": 9401
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.351392865182153e-06,
+      "loss": 0.0604,
+      "step": 9402
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.348178526399783e-06,
+      "loss": 0.0797,
+      "step": 9403
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.34496425524937e-06,
+      "loss": 0.0655,
+      "step": 9404
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.341750052064426e-06,
+      "loss": 0.0568,
+      "step": 9405
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.338535917178444e-06,
+      "loss": 0.0657,
+      "step": 9406
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.335321850924924e-06,
+      "loss": 0.0595,
+      "step": 9407
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.332107853637349e-06,
+      "loss": 0.0532,
+      "step": 9408
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.328893925649196e-06,
+      "loss": 0.0718,
+      "step": 9409
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.325680067293945e-06,
+      "loss": 0.0621,
+      "step": 9410
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.32246627890505e-06,
+      "loss": 0.0573,
+      "step": 9411
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.319252560815977e-06,
+      "loss": 0.0636,
+      "step": 9412
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.316038913360171e-06,
+      "loss": 0.0599,
+      "step": 9413
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.312825336871083e-06,
+      "loss": 0.056,
+      "step": 9414
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.309611831682142e-06,
+      "loss": 0.0571,
+      "step": 9415
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.306398398126772e-06,
+      "loss": 0.0607,
+      "step": 9416
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.303185036538403e-06,
+      "loss": 0.0662,
+      "step": 9417
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.29997174725044e-06,
+      "loss": 0.0771,
+      "step": 9418
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.296758530596299e-06,
+      "loss": 0.0721,
+      "step": 9419
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.29354538690937e-06,
+      "loss": 0.0712,
+      "step": 9420
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.290332316523043e-06,
+      "loss": 0.065,
+      "step": 9421
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.28711931977071e-06,
+      "loss": 0.0527,
+      "step": 9422
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.283906396985734e-06,
+      "loss": 0.0556,
+      "step": 9423
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.28069354850149e-06,
+      "loss": 0.0701,
+      "step": 9424
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.277480774651342e-06,
+      "loss": 0.065,
+      "step": 9425
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.274268075768633e-06,
+      "loss": 0.0646,
+      "step": 9426
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.271055452186716e-06,
+      "loss": 0.06,
+      "step": 9427
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.267842904238918e-06,
+      "loss": 0.0634,
+      "step": 9428
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.264630432258577e-06,
+      "loss": 0.0656,
+      "step": 9429
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.261418036579008e-06,
+      "loss": 0.0497,
+      "step": 9430
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.258205717533532e-06,
+      "loss": 0.0533,
+      "step": 9431
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.254993475455444e-06,
+      "loss": 0.0521,
+      "step": 9432
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.251781310678046e-06,
+      "loss": 0.0593,
+      "step": 9433
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.24856922353463e-06,
+      "loss": 0.0696,
+      "step": 9434
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.24535721435847e-06,
+      "loss": 0.0532,
+      "step": 9435
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.242145283482848e-06,
+      "loss": 0.0607,
+      "step": 9436
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.23893343124102e-06,
+      "loss": 0.0661,
+      "step": 9437
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.235721657966245e-06,
+      "loss": 0.0637,
+      "step": 9438
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.232509963991776e-06,
+      "loss": 0.0536,
+      "step": 9439
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.229298349650848e-06,
+      "loss": 0.0589,
+      "step": 9440
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.226086815276697e-06,
+      "loss": 0.062,
+      "step": 9441
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.222875361202546e-06,
+      "loss": 0.0497,
+      "step": 9442
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.219663987761605e-06,
+      "loss": 0.0512,
+      "step": 9443
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.216452695287089e-06,
+      "loss": 0.0609,
+      "step": 9444
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.213241484112188e-06,
+      "loss": 0.0787,
+      "step": 9445
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.210030354570101e-06,
+      "loss": 0.0676,
+      "step": 9446
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.206819306994007e-06,
+      "loss": 0.0626,
+      "step": 9447
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.203608341717073e-06,
+      "loss": 0.0611,
+      "step": 9448
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.20039745907247e-06,
+      "loss": 0.0676,
+      "step": 9449
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.197186659393351e-06,
+      "loss": 0.0477,
+      "step": 9450
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.19397594301287e-06,
+      "loss": 0.0737,
+      "step": 9451
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.190765310264155e-06,
+      "loss": 0.0646,
+      "step": 9452
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.187554761480347e-06,
+      "loss": 0.0672,
+      "step": 9453
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.184344296994559e-06,
+      "loss": 0.0656,
+      "step": 9454
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.181133917139905e-06,
+      "loss": 0.0679,
+      "step": 9455
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.177923622249497e-06,
+      "loss": 0.0621,
+      "step": 9456
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.174713412656418e-06,
+      "loss": 0.0643,
+      "step": 9457
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.171503288693763e-06,
+      "loss": 0.0649,
+      "step": 9458
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.16829325069461e-06,
+      "loss": 0.0735,
+      "step": 9459
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.165083298992019e-06,
+      "loss": 0.0601,
+      "step": 9460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.161873433919057e-06,
+      "loss": 0.0674,
+      "step": 9461
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.158663655808767e-06,
+      "loss": 0.0559,
+      "step": 9462
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.155453964994202e-06,
+      "loss": 0.0661,
+      "step": 9463
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.152244361808386e-06,
+      "loss": 0.0721,
+      "step": 9464
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.14903484658434e-06,
+      "loss": 0.062,
+      "step": 9465
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.145825419655086e-06,
+      "loss": 0.0651,
+      "step": 9466
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.142616081353623e-06,
+      "loss": 0.0776,
+      "step": 9467
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.139406832012951e-06,
+      "loss": 0.0704,
+      "step": 9468
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.136197671966058e-06,
+      "loss": 0.0524,
+      "step": 9469
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.132988601545912e-06,
+      "loss": 0.0699,
+      "step": 9470
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.12977962108549e-06,
+      "loss": 0.0723,
+      "step": 9471
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.126570730917744e-06,
+      "loss": 0.0724,
+      "step": 9472
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.123361931375634e-06,
+      "loss": 0.0582,
+      "step": 9473
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.120153222792087e-06,
+      "loss": 0.0709,
+      "step": 9474
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.116944605500041e-06,
+      "loss": 0.0844,
+      "step": 9475
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.11373607983242e-06,
+      "loss": 0.0649,
+      "step": 9476
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.110527646122125e-06,
+      "loss": 0.0798,
+      "step": 9477
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.10731930470207e-06,
+      "loss": 0.0726,
+      "step": 9478
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.104111055905135e-06,
+      "loss": 0.0614,
+      "step": 9479
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.100902900064213e-06,
+      "loss": 0.0557,
+      "step": 9480
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.097694837512175e-06,
+      "loss": 0.0686,
+      "step": 9481
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.094486868581878e-06,
+      "loss": 0.0598,
+      "step": 9482
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.091278993606183e-06,
+      "loss": 0.0569,
+      "step": 9483
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.08807121291793e-06,
+      "loss": 0.0694,
+      "step": 9484
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.084863526849959e-06,
+      "loss": 0.0636,
+      "step": 9485
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.081655935735088e-06,
+      "loss": 0.062,
+      "step": 9486
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.07844843990613e-06,
+      "loss": 0.0591,
+      "step": 9487
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.075241039695899e-06,
+      "loss": 0.053,
+      "step": 9488
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.07203373543718e-06,
+      "loss": 0.0551,
+      "step": 9489
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.068826527462766e-06,
+      "loss": 0.0555,
+      "step": 9490
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.065619416105425e-06,
+      "loss": 0.0541,
+      "step": 9491
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.062412401697922e-06,
+      "loss": 0.0674,
+      "step": 9492
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.059205484573019e-06,
+      "loss": 0.057,
+      "step": 9493
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.055998665063452e-06,
+      "loss": 0.0693,
+      "step": 9494
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.052791943501961e-06,
+      "loss": 0.0568,
+      "step": 9495
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.049585320221266e-06,
+      "loss": 0.0684,
+      "step": 9496
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.046378795554088e-06,
+      "loss": 0.0577,
+      "step": 9497
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.043172369833126e-06,
+      "loss": 0.0518,
+      "step": 9498
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.03996604339107e-06,
+      "loss": 0.062,
+      "step": 9499
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.036759816560613e-06,
+      "loss": 0.0463,
+      "step": 9500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.033553689674416e-06,
+      "loss": 0.0556,
+      "step": 9501
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.030347663065152e-06,
+      "loss": 0.0653,
+      "step": 9502
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.027141737065471e-06,
+      "loss": 0.0587,
+      "step": 9503
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.023935912008009e-06,
+      "loss": 0.068,
+      "step": 9504
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.020730188225405e-06,
+      "loss": 0.0612,
+      "step": 9505
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.017524566050272e-06,
+      "loss": 0.0571,
+      "step": 9506
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.014319045815231e-06,
+      "loss": 0.0602,
+      "step": 9507
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.01111362785287e-06,
+      "loss": 0.077,
+      "step": 9508
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.007908312495783e-06,
+      "loss": 0.0706,
+      "step": 9509
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.004703100076553e-06,
+      "loss": 0.0659,
+      "step": 9510
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.001497990927738e-06,
+      "loss": 0.0765,
+      "step": 9511
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.998292985381903e-06,
+      "loss": 0.0528,
+      "step": 9512
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.995088083771587e-06,
+      "loss": 0.0624,
+      "step": 9513
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.991883286429337e-06,
+      "loss": 0.0583,
+      "step": 9514
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.98867859368767e-06,
+      "loss": 0.0756,
+      "step": 9515
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.985474005879098e-06,
+      "loss": 0.06,
+      "step": 9516
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.982269523336126e-06,
+      "loss": 0.0591,
+      "step": 9517
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.979065146391247e-06,
+      "loss": 0.0532,
+      "step": 9518
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.975860875376942e-06,
+      "loss": 0.0688,
+      "step": 9519
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.972656710625682e-06,
+      "loss": 0.0663,
+      "step": 9520
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.96945265246992e-06,
+      "loss": 0.0648,
+      "step": 9521
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 8.966248701242114e-06,
+      "loss": 0.055,
+      "step": 9522
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.963044857274691e-06,
+      "loss": 0.0754,
+      "step": 9523
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.959841120900086e-06,
+      "loss": 0.0609,
+      "step": 9524
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.956637492450706e-06,
+      "loss": 0.0526,
+      "step": 9525
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.953433972258955e-06,
+      "loss": 0.0682,
+      "step": 9526
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.950230560657232e-06,
+      "loss": 0.0637,
+      "step": 9527
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.947027257977909e-06,
+      "loss": 0.053,
+      "step": 9528
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.943824064553361e-06,
+      "loss": 0.0643,
+      "step": 9529
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.940620980715948e-06,
+      "loss": 0.0504,
+      "step": 9530
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.93741800679801e-06,
+      "loss": 0.0717,
+      "step": 9531
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.934215143131891e-06,
+      "loss": 0.0713,
+      "step": 9532
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.931012390049903e-06,
+      "loss": 0.0651,
+      "step": 9533
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.92780974788437e-06,
+      "loss": 0.0496,
+      "step": 9534
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.924607216967588e-06,
+      "loss": 0.0666,
+      "step": 9535
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.921404797631851e-06,
+      "loss": 0.0641,
+      "step": 9536
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.918202490209431e-06,
+      "loss": 0.0578,
+      "step": 9537
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.915000295032594e-06,
+      "loss": 0.0493,
+      "step": 9538
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.911798212433603e-06,
+      "loss": 0.062,
+      "step": 9539
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.90859624274469e-06,
+      "loss": 0.0509,
+      "step": 9540
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.905394386298098e-06,
+      "loss": 0.0594,
+      "step": 9541
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 8.902192643426034e-06,
+      "loss": 0.0676,
+      "step": 9542
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.898991014460714e-06,
+      "loss": 0.056,
+      "step": 9543
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.895789499734335e-06,
+      "loss": 0.054,
+      "step": 9544
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.892588099579073e-06,
+      "loss": 0.0635,
+      "step": 9545
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.889386814327106e-06,
+      "loss": 0.0673,
+      "step": 9546
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.886185644310597e-06,
+      "loss": 0.0563,
+      "step": 9547
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.882984589861687e-06,
+      "loss": 0.0592,
+      "step": 9548
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.879783651312516e-06,
+      "loss": 0.069,
+      "step": 9549
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.876582828995211e-06,
+      "loss": 0.0782,
+      "step": 9550
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.873382123241879e-06,
+      "loss": 0.0599,
+      "step": 9551
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.870181534384626e-06,
+      "loss": 0.0628,
+      "step": 9552
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.866981062755532e-06,
+      "loss": 0.0781,
+      "step": 9553
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.86378070868668e-06,
+      "loss": 0.0728,
+      "step": 9554
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.860580472510128e-06,
+      "loss": 0.0552,
+      "step": 9555
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.857380354557937e-06,
+      "loss": 0.066,
+      "step": 9556
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.854180355162132e-06,
+      "loss": 0.0721,
+      "step": 9557
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.850980474654752e-06,
+      "loss": 0.0803,
+      "step": 9558
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.847780713367808e-06,
+      "loss": 0.0568,
+      "step": 9559
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.844581071633297e-06,
+      "loss": 0.0562,
+      "step": 9560
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.841381549783217e-06,
+      "loss": 0.0659,
+      "step": 9561
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 8.838182148149537e-06,
+      "loss": 0.0564,
+      "step": 9562
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.834982867064228e-06,
+      "loss": 0.0684,
+      "step": 9563
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.831783706859243e-06,
+      "loss": 0.0547,
+      "step": 9564
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.828584667866514e-06,
+      "loss": 0.0753,
+      "step": 9565
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.825385750417975e-06,
+      "loss": 0.0493,
+      "step": 9566
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.822186954845537e-06,
+      "loss": 0.0572,
+      "step": 9567
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.818988281481109e-06,
+      "loss": 0.0674,
+      "step": 9568
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.815789730656573e-06,
+      "loss": 0.0665,
+      "step": 9569
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.812591302703805e-06,
+      "loss": 0.0737,
+      "step": 9570
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.809392997954673e-06,
+      "loss": 0.0634,
+      "step": 9571
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.806194816741022e-06,
+      "loss": 0.0599,
+      "step": 9572
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.802996759394701e-06,
+      "loss": 0.0682,
+      "step": 9573
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.799798826247526e-06,
+      "loss": 0.0618,
+      "step": 9574
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.796601017631312e-06,
+      "loss": 0.049,
+      "step": 9575
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.79340333387786e-06,
+      "loss": 0.0679,
+      "step": 9576
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.790205775318952e-06,
+      "loss": 0.0668,
+      "step": 9577
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.787008342286369e-06,
+      "loss": 0.0645,
+      "step": 9578
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.783811035111864e-06,
+      "loss": 0.0652,
+      "step": 9579
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.78061385412719e-06,
+      "loss": 0.0602,
+      "step": 9580
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.777416799664082e-06,
+      "loss": 0.0576,
+      "step": 9581
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 8.774219872054253e-06,
+      "loss": 0.0773,
+      "step": 9582
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.77102307162942e-06,
+      "loss": 0.054,
+      "step": 9583
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.767826398721272e-06,
+      "loss": 0.0571,
+      "step": 9584
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.764629853661497e-06,
+      "loss": 0.0594,
+      "step": 9585
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.76143343678176e-06,
+      "loss": 0.0479,
+      "step": 9586
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.758237148413713e-06,
+      "loss": 0.0656,
+      "step": 9587
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.755040988889002e-06,
+      "loss": 0.0596,
+      "step": 9588
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.751844958539251e-06,
+      "loss": 0.0679,
+      "step": 9589
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.748649057696085e-06,
+      "loss": 0.0566,
+      "step": 9590
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.745453286691094e-06,
+      "loss": 0.0773,
+      "step": 9591
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.74225764585587e-06,
+      "loss": 0.0624,
+      "step": 9592
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.739062135521992e-06,
+      "loss": 0.0648,
+      "step": 9593
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.735866756021015e-06,
+      "loss": 0.0797,
+      "step": 9594
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.73267150768449e-06,
+      "loss": 0.0688,
+      "step": 9595
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.729476390843947e-06,
+      "loss": 0.0488,
+      "step": 9596
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.726281405830911e-06,
+      "loss": 0.0695,
+      "step": 9597
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.72308655297689e-06,
+      "loss": 0.0699,
+      "step": 9598
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.719891832613368e-06,
+      "loss": 0.056,
+      "step": 9599
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.716697245071832e-06,
+      "loss": 0.0515,
+      "step": 9600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.713502790683743e-06,
+      "loss": 0.0586,
+      "step": 9601
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 8.71030846978056e-06,
+      "loss": 0.059,
+      "step": 9602
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.707114282693712e-06,
+      "loss": 0.07,
+      "step": 9603
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.703920229754624e-06,
+      "loss": 0.0785,
+      "step": 9604
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.700726311294709e-06,
+      "loss": 0.0759,
+      "step": 9605
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.69753252764536e-06,
+      "loss": 0.0675,
+      "step": 9606
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.694338879137962e-06,
+      "loss": 0.059,
+      "step": 9607
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.691145366103881e-06,
+      "loss": 0.0551,
+      "step": 9608
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.68795198887447e-06,
+      "loss": 0.0564,
+      "step": 9609
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.684758747781073e-06,
+      "loss": 0.072,
+      "step": 9610
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.681565643155008e-06,
+      "loss": 0.0771,
+      "step": 9611
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.678372675327593e-06,
+      "loss": 0.0602,
+      "step": 9612
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.675179844630125e-06,
+      "loss": 0.063,
+      "step": 9613
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.671987151393882e-06,
+      "loss": 0.0626,
+      "step": 9614
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.668794595950141e-06,
+      "loss": 0.0746,
+      "step": 9615
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.665602178630146e-06,
+      "loss": 0.0532,
+      "step": 9616
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.662409899765144e-06,
+      "loss": 0.056,
+      "step": 9617
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.659217759686357e-06,
+      "loss": 0.0609,
+      "step": 9618
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.656025758725004e-06,
+      "loss": 0.0654,
+      "step": 9619
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.652833897212275e-06,
+      "loss": 0.0598,
+      "step": 9620
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.64964217547935e-06,
+      "loss": 0.0564,
+      "step": 9621
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.646450593857407e-06,
+      "loss": 0.0646,
+      "step": 9622
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.643259152677589e-06,
+      "loss": 0.0837,
+      "step": 9623
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.640067852271043e-06,
+      "loss": 0.0619,
+      "step": 9624
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.636876692968887e-06,
+      "loss": 0.0603,
+      "step": 9625
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.633685675102232e-06,
+      "loss": 0.0654,
+      "step": 9626
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.630494799002178e-06,
+      "loss": 0.0591,
+      "step": 9627
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.627304064999798e-06,
+      "loss": 0.0638,
+      "step": 9628
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.62411347342616e-06,
+      "loss": 0.0598,
+      "step": 9629
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.620923024612321e-06,
+      "loss": 0.0499,
+      "step": 9630
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.617732718889305e-06,
+      "loss": 0.0616,
+      "step": 9631
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.614542556588145e-06,
+      "loss": 0.056,
+      "step": 9632
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.611352538039837e-06,
+      "loss": 0.0495,
+      "step": 9633
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.608162663575378e-06,
+      "loss": 0.0593,
+      "step": 9634
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.604972933525746e-06,
+      "loss": 0.0555,
+      "step": 9635
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.601783348221895e-06,
+      "loss": 0.0559,
+      "step": 9636
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.598593907994778e-06,
+      "loss": 0.0624,
+      "step": 9637
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.595404613175321e-06,
+      "loss": 0.0591,
+      "step": 9638
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.592215464094447e-06,
+      "loss": 0.0701,
+      "step": 9639
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.58902646108305e-06,
+      "loss": 0.0626,
+      "step": 9640
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.585837604472023e-06,
+      "loss": 0.0699,
+      "step": 9641
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.582648894592232e-06,
+      "loss": 0.0616,
+      "step": 9642
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.579460331774529e-06,
+      "loss": 0.0624,
+      "step": 9643
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.576271916349763e-06,
+      "loss": 0.0647,
+      "step": 9644
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.57308364864875e-06,
+      "loss": 0.0679,
+      "step": 9645
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.569895529002305e-06,
+      "loss": 0.0662,
+      "step": 9646
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.566707557741226e-06,
+      "loss": 0.058,
+      "step": 9647
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.563519735196279e-06,
+      "loss": 0.0518,
+      "step": 9648
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.560332061698242e-06,
+      "loss": 0.0607,
+      "step": 9649
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.55714453757785e-06,
+      "loss": 0.0674,
+      "step": 9650
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.553957163165844e-06,
+      "loss": 0.0545,
+      "step": 9651
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.550769938792943e-06,
+      "loss": 0.0592,
+      "step": 9652
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.547582864789836e-06,
+      "loss": 0.0439,
+      "step": 9653
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.54439594148722e-06,
+      "loss": 0.0682,
+      "step": 9654
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.54120916921576e-06,
+      "loss": 0.0641,
+      "step": 9655
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.538022548306116e-06,
+      "loss": 0.0793,
+      "step": 9656
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.534836079088923e-06,
+      "loss": 0.0621,
+      "step": 9657
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.5316497618948e-06,
+      "loss": 0.0483,
+      "step": 9658
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.528463597054358e-06,
+      "loss": 0.0534,
+      "step": 9659
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.525277584898186e-06,
+      "loss": 0.0509,
+      "step": 9660
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.522091725756868e-06,
+      "loss": 0.0603,
+      "step": 9661
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.518906019960954e-06,
+      "loss": 0.0573,
+      "step": 9662
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.51572046784099e-06,
+      "loss": 0.0633,
+      "step": 9663
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.51253506972751e-06,
+      "loss": 0.0602,
+      "step": 9664
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.509349825951015e-06,
+      "loss": 0.0663,
+      "step": 9665
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.50616473684201e-06,
+      "loss": 0.071,
+      "step": 9666
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.502979802730968e-06,
+      "loss": 0.0564,
+      "step": 9667
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.49979502394836e-06,
+      "loss": 0.0544,
+      "step": 9668
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.49661040082463e-06,
+      "loss": 0.06,
+      "step": 9669
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.493425933690205e-06,
+      "loss": 0.0695,
+      "step": 9670
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.490241622875508e-06,
+      "loss": 0.0712,
+      "step": 9671
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.48705746871093e-06,
+      "loss": 0.0598,
+      "step": 9672
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.483873471526865e-06,
+      "loss": 0.059,
+      "step": 9673
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.480689631653668e-06,
+      "loss": 0.0599,
+      "step": 9674
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.477505949421695e-06,
+      "loss": 0.065,
+      "step": 9675
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.474322425161279e-06,
+      "loss": 0.0624,
+      "step": 9676
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.471139059202734e-06,
+      "loss": 0.058,
+      "step": 9677
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.467955851876368e-06,
+      "loss": 0.0624,
+      "step": 9678
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.464772803512458e-06,
+      "loss": 0.0638,
+      "step": 9679
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.461589914441276e-06,
+      "loss": 0.0452,
+      "step": 9680
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.458407184993076e-06,
+      "loss": 0.066,
+      "step": 9681
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 8.455224615498086e-06,
+      "loss": 0.0492,
+      "step": 9682
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.45204220628653e-06,
+      "loss": 0.0617,
+      "step": 9683
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.448859957688607e-06,
+      "loss": 0.0685,
+      "step": 9684
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.445677870034506e-06,
+      "loss": 0.0687,
+      "step": 9685
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.442495943654393e-06,
+      "loss": 0.0695,
+      "step": 9686
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.439314178878414e-06,
+      "loss": 0.0604,
+      "step": 9687
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.43613257603671e-06,
+      "loss": 0.0556,
+      "step": 9688
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.4329511354594e-06,
+      "loss": 0.0583,
+      "step": 9689
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.429769857476583e-06,
+      "loss": 0.0786,
+      "step": 9690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.426588742418343e-06,
+      "loss": 0.0631,
+      "step": 9691
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.423407790614745e-06,
+      "loss": 0.0598,
+      "step": 9692
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.420227002395846e-06,
+      "loss": 0.0558,
+      "step": 9693
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.417046378091674e-06,
+      "loss": 0.0652,
+      "step": 9694
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.413865918032249e-06,
+      "loss": 0.063,
+      "step": 9695
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.410685622547567e-06,
+      "loss": 0.0798,
+      "step": 9696
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.40750549196761e-06,
+      "loss": 0.0486,
+      "step": 9697
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.404325526622352e-06,
+      "loss": 0.0728,
+      "step": 9698
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.401145726841728e-06,
+      "loss": 0.0726,
+      "step": 9699
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.397966092955678e-06,
+      "loss": 0.0498,
+      "step": 9700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.394786625294112e-06,
+      "loss": 0.0607,
+      "step": 9701
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.391607324186933e-06,
+      "loss": 0.0693,
+      "step": 9702
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 8.388428189964014e-06,
+      "loss": 0.0565,
+      "step": 9703
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.385249222955212e-06,
+      "loss": 0.0623,
+      "step": 9704
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.382070423490382e-06,
+      "loss": 0.0547,
+      "step": 9705
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.378891791899343e-06,
+      "loss": 0.0615,
+      "step": 9706
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.375713328511912e-06,
+      "loss": 0.0555,
+      "step": 9707
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.372535033657878e-06,
+      "loss": 0.0513,
+      "step": 9708
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.369356907667013e-06,
+      "loss": 0.0582,
+      "step": 9709
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.36617895086908e-06,
+      "loss": 0.063,
+      "step": 9710
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.36300116359381e-06,
+      "loss": 0.0654,
+      "step": 9711
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.359823546170936e-06,
+      "loss": 0.0607,
+      "step": 9712
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.356646098930156e-06,
+      "loss": 0.0712,
+      "step": 9713
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.353468822201155e-06,
+      "loss": 0.0591,
+      "step": 9714
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.35029171631361e-06,
+      "loss": 0.061,
+      "step": 9715
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.347114781597164e-06,
+      "loss": 0.0662,
+      "step": 9716
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.343938018381454e-06,
+      "loss": 0.0692,
+      "step": 9717
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.3407614269961e-06,
+      "loss": 0.0533,
+      "step": 9718
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.337585007770692e-06,
+      "loss": 0.0698,
+      "step": 9719
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.334408761034818e-06,
+      "loss": 0.0641,
+      "step": 9720
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.331232687118035e-06,
+      "loss": 0.0679,
+      "step": 9721
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.32805678634989e-06,
+      "loss": 0.0651,
+      "step": 9722
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.324881059059904e-06,
+      "loss": 0.0642,
+      "step": 9723
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.321705505577597e-06,
+      "loss": 0.0634,
+      "step": 9724
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.31853012623245e-06,
+      "loss": 0.0798,
+      "step": 9725
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.315354921353935e-06,
+      "loss": 0.0544,
+      "step": 9726
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.312179891271512e-06,
+      "loss": 0.0654,
+      "step": 9727
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.309005036314611e-06,
+      "loss": 0.0596,
+      "step": 9728
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.305830356812656e-06,
+      "loss": 0.0587,
+      "step": 9729
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.302655853095043e-06,
+      "loss": 0.0556,
+      "step": 9730
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.299481525491152e-06,
+      "loss": 0.0766,
+      "step": 9731
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.29630737433035e-06,
+      "loss": 0.0775,
+      "step": 9732
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.293133399941977e-06,
+      "loss": 0.0624,
+      "step": 9733
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.289959602655365e-06,
+      "loss": 0.0648,
+      "step": 9734
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.28678598279982e-06,
+      "loss": 0.0666,
+      "step": 9735
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.283612540704628e-06,
+      "loss": 0.0613,
+      "step": 9736
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.280439276699067e-06,
+      "loss": 0.0641,
+      "step": 9737
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.277266191112385e-06,
+      "loss": 0.0618,
+      "step": 9738
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.274093284273819e-06,
+      "loss": 0.0616,
+      "step": 9739
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.270920556512585e-06,
+      "loss": 0.0646,
+      "step": 9740
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.267748008157874e-06,
+      "loss": 0.06,
+      "step": 9741
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.264575639538873e-06,
+      "loss": 0.0703,
+      "step": 9742
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.261403450984734e-06,
+      "loss": 0.067,
+      "step": 9743
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.258231442824606e-06,
+      "loss": 0.062,
+      "step": 9744
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.255059615387606e-06,
+      "loss": 0.0608,
+      "step": 9745
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.251887969002841e-06,
+      "loss": 0.0647,
+      "step": 9746
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.248716503999398e-06,
+      "loss": 0.0615,
+      "step": 9747
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.245545220706334e-06,
+      "loss": 0.0677,
+      "step": 9748
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.242374119452708e-06,
+      "loss": 0.054,
+      "step": 9749
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.239203200567536e-06,
+      "loss": 0.0701,
+      "step": 9750
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.236032464379838e-06,
+      "loss": 0.0598,
+      "step": 9751
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.232861911218602e-06,
+      "loss": 0.0576,
+      "step": 9752
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.229691541412795e-06,
+      "loss": 0.0457,
+      "step": 9753
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.226521355291372e-06,
+      "loss": 0.0573,
+      "step": 9754
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.223351353183267e-06,
+      "loss": 0.0723,
+      "step": 9755
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.2201815354174e-06,
+      "loss": 0.0714,
+      "step": 9756
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.217011902322656e-06,
+      "loss": 0.0728,
+      "step": 9757
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.213842454227915e-06,
+      "loss": 0.0591,
+      "step": 9758
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.210673191462037e-06,
+      "loss": 0.056,
+      "step": 9759
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.207504114353854e-06,
+      "loss": 0.0601,
+      "step": 9760
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.204335223232192e-06,
+      "loss": 0.0556,
+      "step": 9761
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.201166518425842e-06,
+      "loss": 0.0527,
+      "step": 9762
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 8.197998000263591e-06,
+      "loss": 0.0733,
+      "step": 9763
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.194829669074196e-06,
+      "loss": 0.0604,
+      "step": 9764
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.191661525186394e-06,
+      "loss": 0.0586,
+      "step": 9765
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.188493568928916e-06,
+      "loss": 0.0576,
+      "step": 9766
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.185325800630454e-06,
+      "loss": 0.0693,
+      "step": 9767
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.182158220619695e-06,
+      "loss": 0.048,
+      "step": 9768
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.178990829225308e-06,
+      "loss": 0.0621,
+      "step": 9769
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.175823626775924e-06,
+      "loss": 0.057,
+      "step": 9770
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.172656613600176e-06,
+      "loss": 0.0587,
+      "step": 9771
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.169489790026664e-06,
+      "loss": 0.0653,
+      "step": 9772
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.16632315638398e-06,
+      "loss": 0.0597,
+      "step": 9773
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.163156713000681e-06,
+      "loss": 0.0507,
+      "step": 9774
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.159990460205312e-06,
+      "loss": 0.0604,
+      "step": 9775
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.156824398326402e-06,
+      "loss": 0.0676,
+      "step": 9776
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.153658527692452e-06,
+      "loss": 0.0692,
+      "step": 9777
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.150492848631958e-06,
+      "loss": 0.0566,
+      "step": 9778
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.147327361473375e-06,
+      "loss": 0.0579,
+      "step": 9779
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.14416206654515e-06,
+      "loss": 0.0549,
+      "step": 9780
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.140996964175716e-06,
+      "loss": 0.0633,
+      "step": 9781
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.137832054693472e-06,
+      "loss": 0.0668,
+      "step": 9782
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 8.13466733842681e-06,
+      "loss": 0.0611,
+      "step": 9783
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.131502815704087e-06,
+      "loss": 0.0602,
+      "step": 9784
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.128338486853657e-06,
+      "loss": 0.076,
+      "step": 9785
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.125174352203847e-06,
+      "loss": 0.0672,
+      "step": 9786
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.122010412082952e-06,
+      "loss": 0.0622,
+      "step": 9787
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.118846666819268e-06,
+      "loss": 0.0543,
+      "step": 9788
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.115683116741053e-06,
+      "loss": 0.0658,
+      "step": 9789
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.112519762176559e-06,
+      "loss": 0.0604,
+      "step": 9790
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.109356603454009e-06,
+      "loss": 0.0618,
+      "step": 9791
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.106193640901599e-06,
+      "loss": 0.0694,
+      "step": 9792
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.103030874847521e-06,
+      "loss": 0.0634,
+      "step": 9793
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.099868305619934e-06,
+      "loss": 0.0547,
+      "step": 9794
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.096705933546989e-06,
+      "loss": 0.0634,
+      "step": 9795
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.093543758956802e-06,
+      "loss": 0.0779,
+      "step": 9796
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.090381782177472e-06,
+      "loss": 0.0558,
+      "step": 9797
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.08722000353709e-06,
+      "loss": 0.0521,
+      "step": 9798
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.084058423363709e-06,
+      "loss": 0.0637,
+      "step": 9799
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.080897041985378e-06,
+      "loss": 0.0599,
+      "step": 9800
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.077735859730108e-06,
+      "loss": 0.0804,
+      "step": 9801
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.0745748769259e-06,
+      "loss": 0.0526,
+      "step": 9802
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.07141409390074e-06,
+      "loss": 0.0482,
+      "step": 9803
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.068253510982575e-06,
+      "loss": 0.0668,
+      "step": 9804
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.065093128499351e-06,
+      "loss": 0.0784,
+      "step": 9805
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.061932946778977e-06,
+      "loss": 0.0634,
+      "step": 9806
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.058772966149358e-06,
+      "loss": 0.0596,
+      "step": 9807
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.055613186938357e-06,
+      "loss": 0.0674,
+      "step": 9808
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.052453609473835e-06,
+      "loss": 0.0574,
+      "step": 9809
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.049294234083626e-06,
+      "loss": 0.075,
+      "step": 9810
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.046135061095534e-06,
+      "loss": 0.0505,
+      "step": 9811
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.042976090837358e-06,
+      "loss": 0.077,
+      "step": 9812
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.039817323636865e-06,
+      "loss": 0.0644,
+      "step": 9813
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.036658759821799e-06,
+      "loss": 0.064,
+      "step": 9814
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.033500399719895e-06,
+      "loss": 0.067,
+      "step": 9815
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.030342243658851e-06,
+      "loss": 0.0571,
+      "step": 9816
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.027184291966361e-06,
+      "loss": 0.0817,
+      "step": 9817
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.024026544970088e-06,
+      "loss": 0.0602,
+      "step": 9818
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.020869002997665e-06,
+      "loss": 0.0552,
+      "step": 9819
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.017711666376726e-06,
+      "loss": 0.0486,
+      "step": 9820
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.014554535434863e-06,
+      "loss": 0.0672,
+      "step": 9821
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.011397610499657e-06,
+      "loss": 0.0668,
+      "step": 9822
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.00824089189867e-06,
+      "loss": 0.0761,
+      "step": 9823
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.005084379959429e-06,
+      "loss": 0.0809,
+      "step": 9824
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.001928075009455e-06,
+      "loss": 0.0687,
+      "step": 9825
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.99877197737624e-06,
+      "loss": 0.0554,
+      "step": 9826
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.99561608738726e-06,
+      "loss": 0.0471,
+      "step": 9827
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.992460405369954e-06,
+      "loss": 0.056,
+      "step": 9828
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.989304931651763e-06,
+      "loss": 0.0543,
+      "step": 9829
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.986149666560086e-06,
+      "loss": 0.0723,
+      "step": 9830
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.982994610422308e-06,
+      "loss": 0.0449,
+      "step": 9831
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.9798397635658e-06,
+      "loss": 0.0549,
+      "step": 9832
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.976685126317893e-06,
+      "loss": 0.0721,
+      "step": 9833
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.973530699005916e-06,
+      "loss": 0.0663,
+      "step": 9834
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.970376481957166e-06,
+      "loss": 0.0679,
+      "step": 9835
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.967222475498917e-06,
+      "loss": 0.0565,
+      "step": 9836
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.964068679958426e-06,
+      "loss": 0.0687,
+      "step": 9837
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.960915095662922e-06,
+      "loss": 0.0594,
+      "step": 9838
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.957761722939619e-06,
+      "loss": 0.0596,
+      "step": 9839
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.954608562115706e-06,
+      "loss": 0.0676,
+      "step": 9840
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.951455613518348e-06,
+      "loss": 0.0533,
+      "step": 9841
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.948302877474692e-06,
+      "loss": 0.0687,
+      "step": 9842
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.945150354311857e-06,
+      "loss": 0.0698,
+      "step": 9843
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.941998044356951e-06,
+      "loss": 0.0665,
+      "step": 9844
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.93884594793705e-06,
+      "loss": 0.0527,
+      "step": 9845
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.935694065379203e-06,
+      "loss": 0.06,
+      "step": 9846
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.932542397010453e-06,
+      "loss": 0.0656,
+      "step": 9847
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.929390943157809e-06,
+      "loss": 0.0641,
+      "step": 9848
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.926239704148263e-06,
+      "loss": 0.0708,
+      "step": 9849
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.923088680308777e-06,
+      "loss": 0.0632,
+      "step": 9850
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.919937871966304e-06,
+      "loss": 0.0584,
+      "step": 9851
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.916787279447761e-06,
+      "loss": 0.0595,
+      "step": 9852
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.91363690308005e-06,
+      "loss": 0.0611,
+      "step": 9853
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.910486743190052e-06,
+      "loss": 0.0543,
+      "step": 9854
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.907336800104615e-06,
+      "loss": 0.0801,
+      "step": 9855
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.90418707415058e-06,
+      "loss": 0.0612,
+      "step": 9856
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.901037565654754e-06,
+      "loss": 0.062,
+      "step": 9857
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.897888274943923e-06,
+      "loss": 0.0668,
+      "step": 9858
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.894739202344857e-06,
+      "loss": 0.0525,
+      "step": 9859
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.891590348184294e-06,
+      "loss": 0.056,
+      "step": 9860
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.888441712788961e-06,
+      "loss": 0.069,
+      "step": 9861
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.885293296485551e-06,
+      "loss": 0.0545,
+      "step": 9862
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.882145099600736e-06,
+      "loss": 0.0703,
+      "step": 9863
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.878997122461171e-06,
+      "loss": 0.0643,
+      "step": 9864
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.875849365393484e-06,
+      "loss": 0.049,
+      "step": 9865
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.872701828724285e-06,
+      "loss": 0.059,
+      "step": 9866
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.86955451278015e-06,
+      "loss": 0.0509,
+      "step": 9867
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.866407417887647e-06,
+      "loss": 0.0536,
+      "step": 9868
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.863260544373312e-06,
+      "loss": 0.059,
+      "step": 9869
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.860113892563655e-06,
+      "loss": 0.0602,
+      "step": 9870
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.85696746278517e-06,
+      "loss": 0.0541,
+      "step": 9871
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.853821255364325e-06,
+      "loss": 0.0605,
+      "step": 9872
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.850675270627573e-06,
+      "loss": 0.0707,
+      "step": 9873
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.847529508901327e-06,
+      "loss": 0.0659,
+      "step": 9874
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.844383970511986e-06,
+      "loss": 0.0524,
+      "step": 9875
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.841238655785931e-06,
+      "loss": 0.0604,
+      "step": 9876
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.83809356504951e-06,
+      "loss": 0.065,
+      "step": 9877
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.834948698629061e-06,
+      "loss": 0.0768,
+      "step": 9878
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.831804056850884e-06,
+      "loss": 0.0529,
+      "step": 9879
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.82865964004126e-06,
+      "loss": 0.0516,
+      "step": 9880
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.825515448526456e-06,
+      "loss": 0.0689,
+      "step": 9881
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.822371482632698e-06,
+      "loss": 0.0604,
+      "step": 9882
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.81922774268621e-06,
+      "loss": 0.0592,
+      "step": 9883
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.816084229013175e-06,
+      "loss": 0.0598,
+      "step": 9884
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.812940941939757e-06,
+      "loss": 0.0562,
+      "step": 9885
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.809797881792108e-06,
+      "loss": 0.066,
+      "step": 9886
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.806655048896333e-06,
+      "loss": 0.0573,
+      "step": 9887
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.80351244357854e-06,
+      "loss": 0.0543,
+      "step": 9888
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.800370066164793e-06,
+      "loss": 0.066,
+      "step": 9889
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.797227916981146e-06,
+      "loss": 0.0599,
+      "step": 9890
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.79408599635362e-06,
+      "loss": 0.0562,
+      "step": 9891
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.790944304608214e-06,
+      "loss": 0.0638,
+      "step": 9892
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.787802842070907e-06,
+      "loss": 0.0609,
+      "step": 9893
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.78466160906765e-06,
+      "loss": 0.0604,
+      "step": 9894
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.781520605924378e-06,
+      "loss": 0.0579,
+      "step": 9895
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.778379832966992e-06,
+      "loss": 0.0612,
+      "step": 9896
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.77523929052137e-06,
+      "loss": 0.0524,
+      "step": 9897
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.772098978913381e-06,
+      "loss": 0.0511,
+      "step": 9898
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.768958898468845e-06,
+      "loss": 0.0569,
+      "step": 9899
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.765819049513583e-06,
+      "loss": 0.0597,
+      "step": 9900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.762679432373376e-06,
+      "loss": 0.0615,
+      "step": 9901
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.759540047373982e-06,
+      "loss": 0.074,
+      "step": 9902
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.756400894841148e-06,
+      "loss": 0.0567,
+      "step": 9903
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.753261975100577e-06,
+      "loss": 0.0743,
+      "step": 9904
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.750123288477965e-06,
+      "loss": 0.0491,
+      "step": 9905
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.746984835298978e-06,
+      "loss": 0.0744,
+      "step": 9906
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.74384661588925e-06,
+      "loss": 0.0513,
+      "step": 9907
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.740708630574405e-06,
+      "loss": 0.0687,
+      "step": 9908
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.737570879680028e-06,
+      "loss": 0.0592,
+      "step": 9909
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.734433363531694e-06,
+      "loss": 0.0835,
+      "step": 9910
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.731296082454943e-06,
+      "loss": 0.0625,
+      "step": 9911
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.728159036775299e-06,
+      "loss": 0.0622,
+      "step": 9912
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.72502222681825e-06,
+      "loss": 0.0614,
+      "step": 9913
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.721885652909268e-06,
+      "loss": 0.0749,
+      "step": 9914
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.718749315373804e-06,
+      "loss": 0.0549,
+      "step": 9915
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.715613214537272e-06,
+      "loss": 0.0638,
+      "step": 9916
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.712477350725078e-06,
+      "loss": 0.0583,
+      "step": 9917
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.709341724262585e-06,
+      "loss": 0.066,
+      "step": 9918
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.706206335475143e-06,
+      "loss": 0.0651,
+      "step": 9919
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.703071184688082e-06,
+      "loss": 0.0646,
+      "step": 9920
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.69993627222669e-06,
+      "loss": 0.0673,
+      "step": 9921
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.696801598416245e-06,
+      "loss": 0.0723,
+      "step": 9922
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.693667163582e-06,
+      "loss": 0.0759,
+      "step": 9923
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 7.690532968049169e-06,
+      "loss": 0.0566,
+      "step": 9924
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.687399012142964e-06,
+      "loss": 0.0671,
+      "step": 9925
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.684265296188546e-06,
+      "loss": 0.0671,
+      "step": 9926
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.681131820511073e-06,
+      "loss": 0.0582,
+      "step": 9927
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.677998585435669e-06,
+      "loss": 0.076,
+      "step": 9928
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.674865591287428e-06,
+      "loss": 0.0505,
+      "step": 9929
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.67173283839143e-06,
+      "loss": 0.0561,
+      "step": 9930
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.668600327072721e-06,
+      "loss": 0.0541,
+      "step": 9931
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.66546805765633e-06,
+      "loss": 0.0634,
+      "step": 9932
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.66233603046725e-06,
+      "loss": 0.0677,
+      "step": 9933
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.65920424583046e-06,
+      "loss": 0.056,
+      "step": 9934
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.656072704070909e-06,
+      "loss": 0.0611,
+      "step": 9935
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.652941405513515e-06,
+      "loss": 0.0612,
+      "step": 9936
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.649810350483187e-06,
+      "loss": 0.0537,
+      "step": 9937
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.646679539304785e-06,
+      "loss": 0.0731,
+      "step": 9938
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.643548972303167e-06,
+      "loss": 0.0549,
+      "step": 9939
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.640418649803155e-06,
+      "loss": 0.0483,
+      "step": 9940
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.637288572129539e-06,
+      "loss": 0.0619,
+      "step": 9941
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.634158739607098e-06,
+      "loss": 0.0594,
+      "step": 9942
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.631029152560574e-06,
+      "loss": 0.0601,
+      "step": 9943
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.6278998113146945e-06,
+      "loss": 0.0568,
+      "step": 9944
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.624770716194151e-06,
+      "loss": 0.0597,
+      "step": 9945
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.621641867523608e-06,
+      "loss": 0.0637,
+      "step": 9946
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.618513265627718e-06,
+      "loss": 0.0616,
+      "step": 9947
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.615384910831093e-06,
+      "loss": 0.0591,
+      "step": 9948
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.612256803458335e-06,
+      "loss": 0.077,
+      "step": 9949
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.609128943834001e-06,
+      "loss": 0.0717,
+      "step": 9950
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.606001332282641e-06,
+      "loss": 0.0555,
+      "step": 9951
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.602873969128769e-06,
+      "loss": 0.0612,
+      "step": 9952
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.59974685469687e-06,
+      "loss": 0.0635,
+      "step": 9953
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.596619989311418e-06,
+      "loss": 0.0554,
+      "step": 9954
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.593493373296841e-06,
+      "loss": 0.059,
+      "step": 9955
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.590367006977558e-06,
+      "loss": 0.0817,
+      "step": 9956
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.587240890677958e-06,
+      "loss": 0.0568,
+      "step": 9957
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.584115024722392e-06,
+      "loss": 0.0569,
+      "step": 9958
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.580989409435206e-06,
+      "loss": 0.0564,
+      "step": 9959
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.577864045140701e-06,
+      "loss": 0.0699,
+      "step": 9960
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.574738932163167e-06,
+      "loss": 0.0672,
+      "step": 9961
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.571614070826857e-06,
+      "loss": 0.063,
+      "step": 9962
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.568489461455997e-06,
+      "loss": 0.06,
+      "step": 9963
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.565365104374798e-06,
+      "loss": 0.0557,
+      "step": 9964
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.562240999907433e-06,
+      "loss": 0.0626,
+      "step": 9965
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.559117148378062e-06,
+      "loss": 0.0698,
+      "step": 9966
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.555993550110805e-06,
+      "loss": 0.083,
+      "step": 9967
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.552870205429759e-06,
+      "loss": 0.0681,
+      "step": 9968
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.549747114659006e-06,
+      "loss": 0.0643,
+      "step": 9969
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.546624278122583e-06,
+      "loss": 0.0573,
+      "step": 9970
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.543501696144521e-06,
+      "loss": 0.0663,
+      "step": 9971
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.5403793690488025e-06,
+      "loss": 0.064,
+      "step": 9972
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.537257297159404e-06,
+      "loss": 0.0504,
+      "step": 9973
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.534135480800266e-06,
+      "loss": 0.067,
+      "step": 9974
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.531013920295297e-06,
+      "loss": 0.0629,
+      "step": 9975
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.527892615968392e-06,
+      "loss": 0.0612,
+      "step": 9976
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.524771568143407e-06,
+      "loss": 0.058,
+      "step": 9977
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.521650777144184e-06,
+      "loss": 0.0639,
+      "step": 9978
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.518530243294526e-06,
+      "loss": 0.067,
+      "step": 9979
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.5154099669182125e-06,
+      "loss": 0.0649,
+      "step": 9980
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.512289948339004e-06,
+      "loss": 0.051,
+      "step": 9981
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.509170187880623e-06,
+      "loss": 0.0654,
+      "step": 9982
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.5060506858667796e-06,
+      "loss": 0.0748,
+      "step": 9983
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.5029314426211395e-06,
+      "loss": 0.0543,
+      "step": 9984
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.499812458467353e-06,
+      "loss": 0.0731,
+      "step": 9985
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.496693733729046e-06,
+      "loss": 0.048,
+      "step": 9986
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.493575268729802e-06,
+      "loss": 0.053,
+      "step": 9987
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.490457063793199e-06,
+      "loss": 0.062,
+      "step": 9988
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.487339119242771e-06,
+      "loss": 0.0537,
+      "step": 9989
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.484221435402029e-06,
+      "loss": 0.0485,
+      "step": 9990
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.481104012594466e-06,
+      "loss": 0.0684,
+      "step": 9991
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.477986851143532e-06,
+      "loss": 0.0607,
+      "step": 9992
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.4748699513726675e-06,
+      "loss": 0.0814,
+      "step": 9993
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.47175331360527e-06,
+      "loss": 0.0622,
+      "step": 9994
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.468636938164724e-06,
+      "loss": 0.0566,
+      "step": 9995
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.465520825374373e-06,
+      "loss": 0.0585,
+      "step": 9996
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.46240497555754e-06,
+      "loss": 0.0572,
+      "step": 9997
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.459289389037527e-06,
+      "loss": 0.0551,
+      "step": 9998
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.456174066137596e-06,
+      "loss": 0.0562,
+      "step": 9999
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.453059007180994e-06,
+      "loss": 0.0577,
+      "step": 10000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.449944212490927e-06,
+      "loss": 0.0554,
+      "step": 10001
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.446829682390584e-06,
+      "loss": 0.065,
+      "step": 10002
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.443715417203128e-06,
+      "loss": 0.0637,
+      "step": 10003
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.440601417251683e-06,
+      "loss": 0.0665,
+      "step": 10004
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.437487682859359e-06,
+      "loss": 0.076,
+      "step": 10005
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.434374214349232e-06,
+      "loss": 0.0774,
+      "step": 10006
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.431261012044342e-06,
+      "loss": 0.0555,
+      "step": 10007
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.428148076267723e-06,
+      "loss": 0.0535,
+      "step": 10008
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.425035407342355e-06,
+      "loss": 0.0565,
+      "step": 10009
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.421923005591212e-06,
+      "loss": 0.0557,
+      "step": 10010
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.418810871337232e-06,
+      "loss": 0.0518,
+      "step": 10011
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.415699004903319e-06,
+      "loss": 0.0503,
+      "step": 10012
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.412587406612362e-06,
+      "loss": 0.0744,
+      "step": 10013
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.4094760767872095e-06,
+      "loss": 0.057,
+      "step": 10014
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.406365015750696e-06,
+      "loss": 0.0634,
+      "step": 10015
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.40325422382561e-06,
+      "loss": 0.0723,
+      "step": 10016
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.400143701334732e-06,
+      "loss": 0.0729,
+      "step": 10017
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.3970334486008e-06,
+      "loss": 0.0624,
+      "step": 10018
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.393923465946527e-06,
+      "loss": 0.0679,
+      "step": 10019
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.390813753694606e-06,
+      "loss": 0.0601,
+      "step": 10020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.387704312167687e-06,
+      "loss": 0.0489,
+      "step": 10021
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.384595141688409e-06,
+      "loss": 0.0583,
+      "step": 10022
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.381486242579374e-06,
+      "loss": 0.0544,
+      "step": 10023
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.378377615163148e-06,
+      "loss": 0.0725,
+      "step": 10024
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.375269259762289e-06,
+      "loss": 0.0646,
+      "step": 10025
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.372161176699304e-06,
+      "loss": 0.0546,
+      "step": 10026
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.36905336629669e-06,
+      "loss": 0.0573,
+      "step": 10027
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3659458288769085e-06,
+      "loss": 0.0633,
+      "step": 10028
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3628385647623845e-06,
+      "loss": 0.0632,
+      "step": 10029
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.359731574275533e-06,
+      "loss": 0.0754,
+      "step": 10030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3566248577387235e-06,
+      "loss": 0.058,
+      "step": 10031
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.35351841547431e-06,
+      "loss": 0.0576,
+      "step": 10032
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.350412247804603e-06,
+      "loss": 0.0587,
+      "step": 10033
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.347306355051906e-06,
+      "loss": 0.0618,
+      "step": 10034
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.34420073753847e-06,
+      "loss": 0.0662,
+      "step": 10035
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3410953955865324e-06,
+      "loss": 0.0616,
+      "step": 10036
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.337990329518303e-06,
+      "loss": 0.0626,
+      "step": 10037
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.334885539655952e-06,
+      "loss": 0.0625,
+      "step": 10038
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.331781026321631e-06,
+      "loss": 0.0831,
+      "step": 10039
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.328676789837463e-06,
+      "loss": 0.0609,
+      "step": 10040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3255728305255294e-06,
+      "loss": 0.077,
+      "step": 10041
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3224691487079e-06,
+      "loss": 0.0677,
+      "step": 10042
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.3193657447066025e-06,
+      "loss": 0.0711,
+      "step": 10043
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.316262618843645e-06,
+      "loss": 0.0568,
+      "step": 10044
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.313159771441003e-06,
+      "loss": 0.0541,
+      "step": 10045
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.31005720282062e-06,
+      "loss": 0.0651,
+      "step": 10046
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.306954913304413e-06,
+      "loss": 0.0594,
+      "step": 10047
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.303852903214274e-06,
+      "loss": 0.0585,
+      "step": 10048
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.300751172872064e-06,
+      "loss": 0.0668,
+      "step": 10049
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.297649722599612e-06,
+      "loss": 0.0667,
+      "step": 10050
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.294548552718714e-06,
+      "loss": 0.0627,
+      "step": 10051
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.291447663551149e-06,
+      "loss": 0.057,
+      "step": 10052
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.2883470554186565e-06,
+      "loss": 0.0428,
+      "step": 10053
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.285246728642956e-06,
+      "loss": 0.0743,
+      "step": 10054
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.282146683545724e-06,
+      "loss": 0.0607,
+      "step": 10055
+    },
+    {
+      "epoch": 5.0,
+      "step": 10055,
+      "total_flos": 1.3325997360124264e+19,
+      "train_loss": 0.0,
+      "train_runtime": 6.3775,
+      "train_samples_per_second": 100889.53,
+      "train_steps_per_second": 1576.644
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 10055,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 1.3325997360124264e+19,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/config.json b/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1e8479218d42cfa03fd0ec5769aaa810ba56862a
--- /dev/null
+++ b/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/config.json
@@ -0,0 +1,36 @@
+{
+  "_name_or_path": "vicuna-v1-3-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "mm_graph_tower": "hvqvae2",
+  "mm_hidden_size": 308,
+  "mm_projector_type": "hlinear",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "model_type": "llava_graph",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.33.2",
+  "tune_mm_mlp_adapter": true,
+  "use_cache": true,
+  "use_lap_pe": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}
diff --git a/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/mm_projector.bin b/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/mm_projector.bin
new file mode 100644
index 0000000000000000000000000000000000000000..df31a841f31926f2b5adc8f3425fc62cea3a4162
--- /dev/null
+++ b/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/mm_projector.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:012ebe5b3092982b8285d8dd747f16ba82045b800c4e83085dfbb33e04306ef2
+size 7596223
diff --git a/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/trainer_state.json b/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..038c3cb1f9b98e177bb06e7c75d7d5ae2dd6483a
--- /dev/null
+++ b/vicuna-selfies/llava-hvqvae2-vicuna-v1-3-7b-pretrain/trainer_state.json
@@ -0,0 +1,138418 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 23065,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.890173410404624e-06,
+      "loss": 2.2656,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.780346820809248e-06,
+      "loss": 2.3867,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.670520231213873e-06,
+      "loss": 2.4922,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1560693641618496e-05,
+      "loss": 2.4805,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4450867052023121e-05,
+      "loss": 2.5312,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7341040462427746e-05,
+      "loss": 2.2109,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.023121387283237e-05,
+      "loss": 2.4453,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3121387283236992e-05,
+      "loss": 2.3887,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.601156069364162e-05,
+      "loss": 2.0449,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8901734104046242e-05,
+      "loss": 2.2539,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.179190751445086e-05,
+      "loss": 2.0723,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.468208092485549e-05,
+      "loss": 1.9062,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.757225433526012e-05,
+      "loss": 1.9219,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.046242774566474e-05,
+      "loss": 1.9258,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.335260115606936e-05,
+      "loss": 1.8145,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6242774566473984e-05,
+      "loss": 1.6797,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.913294797687862e-05,
+      "loss": 1.8047,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.202312138728324e-05,
+      "loss": 1.7969,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.491329479768786e-05,
+      "loss": 1.7891,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7803468208092484e-05,
+      "loss": 1.7871,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.0693641618497105e-05,
+      "loss": 1.6914,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.358381502890173e-05,
+      "loss": 1.6484,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.647398843930635e-05,
+      "loss": 1.6914,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.936416184971098e-05,
+      "loss": 1.6016,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.225433526011561e-05,
+      "loss": 1.7734,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.514450867052024e-05,
+      "loss": 1.6152,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.803468208092485e-05,
+      "loss": 1.6777,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.092485549132948e-05,
+      "loss": 1.5625,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.381502890173411e-05,
+      "loss": 1.582,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.670520231213873e-05,
+      "loss": 1.6543,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.959537572254335e-05,
+      "loss": 1.6152,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.248554913294797e-05,
+      "loss": 1.5508,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.53757225433526e-05,
+      "loss": 1.5742,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.826589595375724e-05,
+      "loss": 1.5254,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010115606936416185,
+      "loss": 1.459,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010404624277456648,
+      "loss": 1.5723,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001069364161849711,
+      "loss": 1.5156,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010982658959537572,
+      "loss": 1.6074,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011271676300578035,
+      "loss": 1.5859,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011560693641618497,
+      "loss": 1.5117,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001184971098265896,
+      "loss": 1.4883,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012138728323699421,
+      "loss": 1.5059,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012427745664739885,
+      "loss": 1.5059,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012716763005780345,
+      "loss": 1.5,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013005780346820808,
+      "loss": 1.5254,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001329479768786127,
+      "loss": 1.4785,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013583815028901734,
+      "loss": 1.4297,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013872832369942197,
+      "loss": 1.4492,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001416184971098266,
+      "loss": 1.4238,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014450867052023122,
+      "loss": 1.5039,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014739884393063585,
+      "loss": 1.4336,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015028901734104048,
+      "loss": 1.4219,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001531791907514451,
+      "loss": 1.4648,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001560693641618497,
+      "loss": 1.3574,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015895953757225434,
+      "loss": 1.3262,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016184971098265897,
+      "loss": 1.3672,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001647398843930636,
+      "loss": 1.3945,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016763005780346822,
+      "loss": 1.332,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017052023121387282,
+      "loss": 1.3535,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017341040462427745,
+      "loss": 1.3418,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017630057803468208,
+      "loss": 1.3164,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001791907514450867,
+      "loss": 1.3105,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018208092485549134,
+      "loss": 1.3125,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018497109826589594,
+      "loss": 1.3203,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018786127167630057,
+      "loss": 1.2559,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001907514450867052,
+      "loss": 1.2734,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019364161849710982,
+      "loss": 1.2988,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019653179190751448,
+      "loss": 1.3281,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019942196531791908,
+      "loss": 1.2871,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002023121387283237,
+      "loss": 1.293,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00020520231213872834,
+      "loss": 1.3086,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00020809248554913296,
+      "loss": 1.2461,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002109826589595376,
+      "loss": 1.2363,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002138728323699422,
+      "loss": 1.2461,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00021676300578034682,
+      "loss": 1.2266,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00021965317919075145,
+      "loss": 1.1973,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00022254335260115608,
+      "loss": 1.1621,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002254335260115607,
+      "loss": 1.1309,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002283236994219653,
+      "loss": 1.209,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00023121387283236994,
+      "loss": 1.1992,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00023410404624277456,
+      "loss": 1.2285,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002369942196531792,
+      "loss": 1.1543,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00023988439306358382,
+      "loss": 1.1426,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00024277456647398842,
+      "loss": 1.1309,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00024566473988439305,
+      "loss": 1.1406,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002485549132947977,
+      "loss": 1.1738,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002514450867052023,
+      "loss": 1.0625,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002543352601156069,
+      "loss": 1.082,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00025722543352601156,
+      "loss": 1.0518,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00026011560693641616,
+      "loss": 1.0742,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002630057803468208,
+      "loss": 1.0547,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002658959537572254,
+      "loss": 1.0439,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002687861271676301,
+      "loss": 1.0117,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002716763005780347,
+      "loss": 1.0469,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00027456647398843933,
+      "loss": 1.001,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00027745664739884393,
+      "loss": 0.9512,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028034682080924853,
+      "loss": 1.0088,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002832369942196532,
+      "loss": 0.9971,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002861271676300578,
+      "loss": 1.002,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028901734104046245,
+      "loss": 0.9268,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00029190751445086705,
+      "loss": 0.9492,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002947976878612717,
+      "loss": 0.9453,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002976878612716763,
+      "loss": 0.8457,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030057803468208096,
+      "loss": 0.96,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030346820809248556,
+      "loss": 0.9746,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003063583815028902,
+      "loss": 0.8799,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030924855491329476,
+      "loss": 0.9541,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003121387283236994,
+      "loss": 0.8467,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000315028901734104,
+      "loss": 0.8887,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003179190751445087,
+      "loss": 0.7959,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003208092485549133,
+      "loss": 0.8125,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032369942196531793,
+      "loss": 0.7529,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032658959537572253,
+      "loss": 0.7686,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003294797687861272,
+      "loss": 0.6621,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003323699421965318,
+      "loss": 0.793,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00033526011560693644,
+      "loss": 0.7422,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003381502890173411,
+      "loss": 0.7686,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00034104046242774565,
+      "loss": 0.6396,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003439306358381503,
+      "loss": 0.7354,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003468208092485549,
+      "loss": 0.7822,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00034971098265895956,
+      "loss": 0.7031,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00035260115606936416,
+      "loss": 0.7588,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003554913294797688,
+      "loss": 0.7852,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003583815028901734,
+      "loss": 0.7832,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036127167630057807,
+      "loss": 0.6172,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036416184971098267,
+      "loss": 0.5469,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00036705202312138733,
+      "loss": 0.6162,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003699421965317919,
+      "loss": 0.6885,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00037283236994219653,
+      "loss": 0.7139,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00037572254335260113,
+      "loss": 0.7539,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003786127167630058,
+      "loss": 0.5479,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003815028901734104,
+      "loss": 0.623,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00038439306358381504,
+      "loss": 0.6069,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00038728323699421964,
+      "loss": 0.5542,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003901734104046243,
+      "loss": 0.6543,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039306358381502895,
+      "loss": 0.4277,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039595375722543356,
+      "loss": 0.3853,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039884393063583816,
+      "loss": 0.5264,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00040173410404624276,
+      "loss": 0.6895,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004046242774566474,
+      "loss": 0.5737,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000407514450867052,
+      "loss": 0.4351,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00041040462427745667,
+      "loss": 0.4644,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00041329479768786127,
+      "loss": 0.4585,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004161849710982659,
+      "loss": 0.7627,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00041907514450867053,
+      "loss": 0.6787,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004219653179190752,
+      "loss": 0.458,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004248554913294798,
+      "loss": 0.5791,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004277456647398844,
+      "loss": 0.3623,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000430635838150289,
+      "loss": 0.522,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00043352601156069364,
+      "loss": 0.6387,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00043641618497109824,
+      "loss": 0.5645,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004393063583815029,
+      "loss": 0.5347,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004421965317919075,
+      "loss": 0.4604,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00044508670520231215,
+      "loss": 0.4756,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004479768786127168,
+      "loss": 0.6602,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004508670520231214,
+      "loss": 0.5137,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00045375722543352607,
+      "loss": 0.4995,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004566473988439306,
+      "loss": 0.4165,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00045953757225433527,
+      "loss": 0.4624,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00046242774566473987,
+      "loss": 0.4722,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004653179190751445,
+      "loss": 0.4531,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004682080924855491,
+      "loss": 0.4463,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004710982658959538,
+      "loss": 0.3828,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004739884393063584,
+      "loss": 0.4233,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00047687861271676304,
+      "loss": 0.3306,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00047976878612716764,
+      "loss": 0.3789,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004826589595375723,
+      "loss": 0.543,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00048554913294797684,
+      "loss": 0.3955,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004884393063583815,
+      "loss": 0.3496,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004913294797687861,
+      "loss": 0.3706,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004942196531791908,
+      "loss": 0.3955,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0004971098265895954,
+      "loss": 0.3687,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005,
+      "loss": 0.3481,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005028901734104046,
+      "loss": 0.4102,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005057803468208093,
+      "loss": 0.4336,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005086705202312138,
+      "loss": 0.4365,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005115606936416185,
+      "loss": 0.321,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005144508670520231,
+      "loss": 0.6855,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005173410404624278,
+      "loss": 0.5234,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005202312138728323,
+      "loss": 0.417,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000523121387283237,
+      "loss": 0.4351,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005260115606936416,
+      "loss": 0.3188,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005289017341040463,
+      "loss": 0.5068,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005317919075144508,
+      "loss": 0.3965,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005346820809248554,
+      "loss": 0.4385,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005375722543352602,
+      "loss": 0.2427,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005404624277456648,
+      "loss": 0.3123,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005433526011560694,
+      "loss": 0.3545,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000546242774566474,
+      "loss": 0.4043,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005491329479768787,
+      "loss": 0.3564,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005520231213872833,
+      "loss": 0.311,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005549132947976879,
+      "loss": 0.2915,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005578034682080925,
+      "loss": 0.4065,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005606936416184971,
+      "loss": 0.2979,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005635838150289018,
+      "loss": 0.3921,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005664739884393064,
+      "loss": 0.3633,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000569364161849711,
+      "loss": 0.3867,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005722543352601156,
+      "loss": 0.3823,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005751445086705203,
+      "loss": 0.3584,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005780346820809249,
+      "loss": 0.3926,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005809248554913295,
+      "loss": 0.324,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005838150289017341,
+      "loss": 0.3696,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005867052023121388,
+      "loss": 0.5176,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005895953757225434,
+      "loss": 0.3198,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005924855491329479,
+      "loss": 0.3818,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005953757225433526,
+      "loss": 0.5122,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0005982658959537572,
+      "loss": 0.2715,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006011560693641619,
+      "loss": 0.3833,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006040462427745664,
+      "loss": 0.3408,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006069364161849711,
+      "loss": 0.4521,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006098265895953757,
+      "loss": 0.4116,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006127167630057804,
+      "loss": 0.3545,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000615606936416185,
+      "loss": 0.6973,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006184971098265895,
+      "loss": 0.3926,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006213872832369942,
+      "loss": 0.1868,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006242774566473988,
+      "loss": 0.271,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006271676300578035,
+      "loss": 0.437,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000630057803468208,
+      "loss": 0.4438,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006329479768786127,
+      "loss": 0.3093,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006358381502890173,
+      "loss": 0.3982,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006387283236994221,
+      "loss": 0.3733,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006416184971098266,
+      "loss": 0.3784,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006445086705202313,
+      "loss": 0.3394,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006473988439306359,
+      "loss": 0.3301,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006502890173410405,
+      "loss": 0.2874,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006531791907514451,
+      "loss": 0.3291,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006560693641618497,
+      "loss": 0.3574,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006589595375722544,
+      "loss": 0.314,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000661849710982659,
+      "loss": 0.3325,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006647398843930636,
+      "loss": 0.3813,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006676300578034682,
+      "loss": 0.4248,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006705202312138729,
+      "loss": 0.3083,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006734104046242775,
+      "loss": 0.2561,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006763005780346822,
+      "loss": 0.4316,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006791907514450867,
+      "loss": 0.2366,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006820809248554913,
+      "loss": 0.4141,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000684971098265896,
+      "loss": 0.3154,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006878612716763006,
+      "loss": 0.321,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006907514450867052,
+      "loss": 0.4106,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006936416184971098,
+      "loss": 0.2983,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006965317919075145,
+      "loss": 0.3413,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0006994219653179191,
+      "loss": 0.3596,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007023121387283237,
+      "loss": 0.1841,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007052023121387283,
+      "loss": 0.2974,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007080924855491329,
+      "loss": 0.2705,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007109826589595376,
+      "loss": 0.4072,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007138728323699421,
+      "loss": 0.2988,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007167630057803468,
+      "loss": 0.2981,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007196531791907514,
+      "loss": 0.3594,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007225433526011561,
+      "loss": 0.2993,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007254335260115607,
+      "loss": 0.3833,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007283236994219653,
+      "loss": 0.1775,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0007312138728323699,
+      "loss": 0.3926,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007341040462427747,
+      "loss": 0.2979,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007369942196531793,
+      "loss": 0.2615,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007398843930635837,
+      "loss": 0.27,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007427745664739885,
+      "loss": 0.3171,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007456647398843931,
+      "loss": 0.2766,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007485549132947978,
+      "loss": 0.3003,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007514450867052023,
+      "loss": 0.356,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000754335260115607,
+      "loss": 0.1946,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007572254335260116,
+      "loss": 0.2417,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007601156069364163,
+      "loss": 0.5176,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007630057803468208,
+      "loss": 0.3894,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007658959537572254,
+      "loss": 0.2578,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007687861271676301,
+      "loss": 0.3311,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007716763005780347,
+      "loss": 0.4297,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007745664739884393,
+      "loss": 0.438,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007774566473988439,
+      "loss": 0.2712,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007803468208092486,
+      "loss": 0.3008,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007832369942196532,
+      "loss": 0.3616,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007861271676300579,
+      "loss": 0.3528,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007890173410404624,
+      "loss": 0.3218,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007919075144508671,
+      "loss": 0.2534,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007947976878612717,
+      "loss": 0.3096,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0007976878612716763,
+      "loss": 0.314,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008005780346820809,
+      "loss": 0.2021,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008034682080924855,
+      "loss": 0.2568,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008063583815028902,
+      "loss": 0.1853,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008092485549132948,
+      "loss": 0.2434,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008121387283236994,
+      "loss": 0.2583,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000815028901734104,
+      "loss": 0.2192,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008179190751445087,
+      "loss": 0.3052,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008208092485549133,
+      "loss": 0.2832,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008236994219653178,
+      "loss": 0.2576,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008265895953757225,
+      "loss": 0.325,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008294797687861271,
+      "loss": 0.3035,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008323699421965319,
+      "loss": 0.2554,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008352601156069365,
+      "loss": 0.2559,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008381502890173411,
+      "loss": 0.314,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008410404624277457,
+      "loss": 0.3442,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008439306358381504,
+      "loss": 0.2607,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000846820809248555,
+      "loss": 0.2188,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008497109826589596,
+      "loss": 0.2151,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008526011560693642,
+      "loss": 0.2788,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008554913294797688,
+      "loss": 0.3008,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008583815028901735,
+      "loss": 0.2583,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000861271676300578,
+      "loss": 0.2485,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0008641618497109827,
+      "loss": 0.3521,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008670520231213873,
+      "loss": 0.3052,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000869942196531792,
+      "loss": 0.2212,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008728323699421965,
+      "loss": 0.1578,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008757225433526012,
+      "loss": 0.3369,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008786127167630058,
+      "loss": 0.2554,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008815028901734105,
+      "loss": 0.2319,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000884393063583815,
+      "loss": 0.2288,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008872832369942196,
+      "loss": 0.3176,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008901734104046243,
+      "loss": 0.2759,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008930635838150289,
+      "loss": 0.3672,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008959537572254336,
+      "loss": 0.3486,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0008988439306358381,
+      "loss": 0.2198,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009017341040462428,
+      "loss": 0.2437,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009046242774566474,
+      "loss": 0.2163,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009075144508670521,
+      "loss": 0.2157,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009104046242774566,
+      "loss": 0.2971,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009132947976878612,
+      "loss": 0.2371,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009161849710982659,
+      "loss": 0.2368,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009190751445086705,
+      "loss": 0.3281,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009219653179190751,
+      "loss": 0.4932,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009248554913294797,
+      "loss": 0.304,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009277456647398844,
+      "loss": 0.2634,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000930635838150289,
+      "loss": 0.1794,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009335260115606937,
+      "loss": 0.2502,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009364161849710983,
+      "loss": 0.2151,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000939306358381503,
+      "loss": 0.2131,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009421965317919076,
+      "loss": 0.3075,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009450867052023122,
+      "loss": 0.1748,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009479768786127168,
+      "loss": 0.2231,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009508670520231214,
+      "loss": 0.2366,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009537572254335261,
+      "loss": 0.1495,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009566473988439307,
+      "loss": 0.2219,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009595375722543353,
+      "loss": 0.1823,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009624277456647399,
+      "loss": 0.3223,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009653179190751446,
+      "loss": 0.3025,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009682080924855492,
+      "loss": 0.2356,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009710982658959537,
+      "loss": 0.2285,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009739884393063584,
+      "loss": 0.2173,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000976878612716763,
+      "loss": 0.2229,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009797687861271677,
+      "loss": 0.1819,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009826589595375722,
+      "loss": 0.2148,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000985549132947977,
+      "loss": 0.2744,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009884393063583816,
+      "loss": 0.2305,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009913294797687861,
+      "loss": 0.2117,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009942196531791908,
+      "loss": 0.2195,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009971098265895953,
+      "loss": 0.1545,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001,
+      "loss": 0.2712,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010028901734104045,
+      "loss": 0.1715,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010057803468208092,
+      "loss": 0.2373,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001008670520231214,
+      "loss": 0.2681,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010115606936416186,
+      "loss": 0.2629,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010144508670520231,
+      "loss": 0.2048,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010173410404624276,
+      "loss": 0.2632,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010202312138728326,
+      "loss": 0.2412,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001023121387283237,
+      "loss": 0.3032,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010260115606936415,
+      "loss": 0.2236,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010289017341040462,
+      "loss": 0.1768,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010317919075144507,
+      "loss": 0.2078,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010346820809248557,
+      "loss": 0.2534,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010375722543352602,
+      "loss": 0.2771,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010404624277456647,
+      "loss": 0.1964,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010433526011560694,
+      "loss": 0.2083,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001046242774566474,
+      "loss": 0.2664,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010491329479768786,
+      "loss": 0.2852,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010520231213872833,
+      "loss": 0.1931,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010549132947976878,
+      "loss": 0.2201,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010578034682080927,
+      "loss": 0.3154,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010606936416184972,
+      "loss": 0.1748,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010635838150289017,
+      "loss": 0.3491,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010664739884393064,
+      "loss": 0.2361,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010693641618497109,
+      "loss": 0.3701,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010722543352601158,
+      "loss": 0.2974,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010751445086705203,
+      "loss": 0.2581,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010780346820809248,
+      "loss": 0.1567,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010809248554913295,
+      "loss": 0.3057,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010838150289017342,
+      "loss": 0.2806,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010867052023121387,
+      "loss": 0.2454,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010895953757225434,
+      "loss": 0.1545,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001092485549132948,
+      "loss": 0.2822,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010953757225433526,
+      "loss": 0.2434,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0010982658959537573,
+      "loss": 0.2061,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011011560693641618,
+      "loss": 0.1895,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011040462427745665,
+      "loss": 0.2292,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001106936416184971,
+      "loss": 0.2827,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011098265895953757,
+      "loss": 0.2048,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011127167630057804,
+      "loss": 0.2554,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001115606936416185,
+      "loss": 0.2168,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011184971098265896,
+      "loss": 0.2723,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011213872832369941,
+      "loss": 0.2795,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011242774566473988,
+      "loss": 0.2478,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011271676300578036,
+      "loss": 0.2241,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.001130057803468208,
+      "loss": 0.3032,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0011329479768786128,
+      "loss": 0.1948,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011358381502890175,
+      "loss": 0.2593,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001138728323699422,
+      "loss": 0.2291,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011416184971098267,
+      "loss": 0.2817,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011445086705202312,
+      "loss": 0.1888,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011473988439306357,
+      "loss": 0.2803,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011502890173410406,
+      "loss": 0.2498,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001153179190751445,
+      "loss": 0.2212,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011560693641618498,
+      "loss": 0.2896,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011589595375722543,
+      "loss": 0.1749,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001161849710982659,
+      "loss": 0.202,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011647398843930637,
+      "loss": 0.1628,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011676300578034682,
+      "loss": 0.2214,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001170520231213873,
+      "loss": 0.3105,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011734104046242776,
+      "loss": 0.2678,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001176300578034682,
+      "loss": 0.1842,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011791907514450868,
+      "loss": 0.159,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011820809248554913,
+      "loss": 0.2393,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011849710982658958,
+      "loss": 0.2017,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011878612716763007,
+      "loss": 0.2913,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011907514450867052,
+      "loss": 0.1957,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00119364161849711,
+      "loss": 0.2512,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011965317919075144,
+      "loss": 0.1816,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0011994219653179191,
+      "loss": 0.2771,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012023121387283238,
+      "loss": 0.2198,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012052023121387283,
+      "loss": 0.2925,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012080924855491328,
+      "loss": 0.1478,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012109826589595375,
+      "loss": 0.2654,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012138728323699422,
+      "loss": 0.1809,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001216763005780347,
+      "loss": 0.3252,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012196531791907514,
+      "loss": 0.2239,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001222543352601156,
+      "loss": 0.2029,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012254335260115609,
+      "loss": 0.2473,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012283236994219654,
+      "loss": 0.2161,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00123121387283237,
+      "loss": 0.1323,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012341040462427746,
+      "loss": 0.217,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001236994219653179,
+      "loss": 0.2754,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001239884393063584,
+      "loss": 0.2078,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012427745664739885,
+      "loss": 0.1981,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001245664739884393,
+      "loss": 0.292,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012485549132947977,
+      "loss": 0.2393,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012514450867052024,
+      "loss": 0.2269,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001254335260115607,
+      "loss": 0.2063,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012572254335260116,
+      "loss": 0.2603,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001260115606936416,
+      "loss": 0.2725,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.001263005780346821,
+      "loss": 0.2014,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0012658959537572255,
+      "loss": 0.2087,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00126878612716763,
+      "loss": 0.2202,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012716763005780347,
+      "loss": 0.2056,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012745664739884392,
+      "loss": 0.2959,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012774566473988441,
+      "loss": 0.1857,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012803468208092486,
+      "loss": 0.2207,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001283236994219653,
+      "loss": 0.2563,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012861271676300578,
+      "loss": 0.1989,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012890173410404625,
+      "loss": 0.1771,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012919075144508672,
+      "loss": 0.2018,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012947976878612717,
+      "loss": 0.2112,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0012976878612716762,
+      "loss": 0.2009,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001300578034682081,
+      "loss": 0.2111,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013034682080924856,
+      "loss": 0.2065,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013063583815028901,
+      "loss": 0.155,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013092485549132948,
+      "loss": 0.2231,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013121387283236993,
+      "loss": 0.2378,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013150289017341043,
+      "loss": 0.2593,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013179190751445087,
+      "loss": 0.1906,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013208092485549132,
+      "loss": 0.1887,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001323699421965318,
+      "loss": 0.2307,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013265895953757224,
+      "loss": 0.2554,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013294797687861272,
+      "loss": 0.1422,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013323699421965319,
+      "loss": 0.2134,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013352601156069364,
+      "loss": 0.2805,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001338150289017341,
+      "loss": 0.2363,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013410404624277458,
+      "loss": 0.1302,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013439306358381503,
+      "loss": 0.2095,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001346820809248555,
+      "loss": 0.2161,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013497109826589595,
+      "loss": 0.2117,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013526011560693644,
+      "loss": 0.1731,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013554913294797689,
+      "loss": 0.1947,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013583815028901734,
+      "loss": 0.2866,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001361271676300578,
+      "loss": 0.2574,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013641618497109826,
+      "loss": 0.1414,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013670520231213873,
+      "loss": 0.2705,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001369942196531792,
+      "loss": 0.2463,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013728323699421965,
+      "loss": 0.2004,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013757225433526012,
+      "loss": 0.1191,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001378612716763006,
+      "loss": 0.2073,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013815028901734104,
+      "loss": 0.1705,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013843930635838151,
+      "loss": 0.2632,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013872832369942196,
+      "loss": 0.2117,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013901734104046243,
+      "loss": 0.1887,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.001393063583815029,
+      "loss": 0.178,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013959537572254335,
+      "loss": 0.3367,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0013988439306358382,
+      "loss": 0.2654,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014017341040462427,
+      "loss": 0.2592,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014046242774566474,
+      "loss": 0.198,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014075144508670521,
+      "loss": 0.2034,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014104046242774566,
+      "loss": 0.198,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014132947976878613,
+      "loss": 0.2334,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014161849710982658,
+      "loss": 0.1501,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014190751445086705,
+      "loss": 0.197,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014219653179190753,
+      "loss": 0.2183,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014248554913294798,
+      "loss": 0.2119,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014277456647398842,
+      "loss": 0.1088,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014306358381502892,
+      "loss": 0.3496,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014335260115606937,
+      "loss": 0.1273,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014364161849710984,
+      "loss": 0.2129,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014393063583815029,
+      "loss": 0.1471,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014421965317919074,
+      "loss": 0.2502,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014450867052023123,
+      "loss": 0.1891,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014479768786127168,
+      "loss": 0.1344,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014508670520231215,
+      "loss": 0.1738,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001453757225433526,
+      "loss": 0.2363,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014566473988439307,
+      "loss": 0.2073,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014595375722543354,
+      "loss": 0.179,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014624277456647399,
+      "loss": 0.1543,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014653179190751444,
+      "loss": 0.1738,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014682080924855493,
+      "loss": 0.2378,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014710982658959538,
+      "loss": 0.1499,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014739884393063585,
+      "loss": 0.2117,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001476878612716763,
+      "loss": 0.2075,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014797687861271675,
+      "loss": 0.1583,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014826589595375724,
+      "loss": 0.2106,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001485549132947977,
+      "loss": 0.2156,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014884393063583814,
+      "loss": 0.1981,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014913294797687861,
+      "loss": 0.157,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014942196531791908,
+      "loss": 0.1987,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0014971098265895955,
+      "loss": 0.2234,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015,
+      "loss": 0.2732,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015028901734104045,
+      "loss": 0.1956,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015057803468208092,
+      "loss": 0.2188,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001508670520231214,
+      "loss": 0.2297,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015115606936416187,
+      "loss": 0.1357,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015144508670520231,
+      "loss": 0.3003,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015173410404624276,
+      "loss": 0.1879,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015202312138728326,
+      "loss": 0.219,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.001523121387283237,
+      "loss": 0.1963,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015260115606936415,
+      "loss": 0.2133,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015289017341040463,
+      "loss": 0.2617,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0015317919075144508,
+      "loss": 0.1355,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015346820809248557,
+      "loss": 0.186,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015375722543352602,
+      "loss": 0.2368,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015404624277456647,
+      "loss": 0.2578,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015433526011560694,
+      "loss": 0.2424,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001546242774566474,
+      "loss": 0.2944,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015491329479768786,
+      "loss": 0.1313,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015520231213872833,
+      "loss": 0.1598,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015549132947976878,
+      "loss": 0.2659,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015578034682080927,
+      "loss": 0.2524,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015606936416184972,
+      "loss": 0.1844,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015635838150289017,
+      "loss": 0.2654,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015664739884393064,
+      "loss": 0.1826,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001569364161849711,
+      "loss": 0.2512,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015722543352601158,
+      "loss": 0.1587,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015751445086705203,
+      "loss": 0.1434,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015780346820809248,
+      "loss": 0.2366,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015809248554913295,
+      "loss": 0.2485,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015838150289017342,
+      "loss": 0.2252,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015867052023121387,
+      "loss": 0.231,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015895953757225434,
+      "loss": 0.1936,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001592485549132948,
+      "loss": 0.2207,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015953757225433526,
+      "loss": 0.2136,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0015982658959537573,
+      "loss": 0.1931,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016011560693641618,
+      "loss": 0.1689,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016040462427745665,
+      "loss": 0.2871,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001606936416184971,
+      "loss": 0.2644,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016098265895953757,
+      "loss": 0.2128,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016127167630057805,
+      "loss": 0.2136,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001615606936416185,
+      "loss": 0.1823,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016184971098265897,
+      "loss": 0.2561,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016213872832369941,
+      "loss": 0.2163,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016242774566473989,
+      "loss": 0.2406,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016271676300578036,
+      "loss": 0.1816,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001630057803468208,
+      "loss": 0.2561,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016329479768786128,
+      "loss": 0.2544,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016358381502890175,
+      "loss": 0.2126,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001638728323699422,
+      "loss": 0.2285,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016416184971098267,
+      "loss": 0.2031,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016445086705202312,
+      "loss": 0.1848,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016473988439306357,
+      "loss": 0.1383,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016502890173410406,
+      "loss": 0.1996,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001653179190751445,
+      "loss": 0.1533,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016560693641618498,
+      "loss": 0.1538,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016589595375722543,
+      "loss": 0.1309,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.001661849710982659,
+      "loss": 0.1433,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0016647398843930637,
+      "loss": 0.2529,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016676300578034682,
+      "loss": 0.2368,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001670520231213873,
+      "loss": 0.1267,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016734104046242776,
+      "loss": 0.2596,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016763005780346821,
+      "loss": 0.1555,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016791907514450868,
+      "loss": 0.1451,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016820809248554913,
+      "loss": 0.17,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016849710982658958,
+      "loss": 0.1586,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016878612716763007,
+      "loss": 0.1697,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016907514450867052,
+      "loss": 0.208,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00169364161849711,
+      "loss": 0.229,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016965317919075144,
+      "loss": 0.2072,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0016994219653179191,
+      "loss": 0.1405,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017023121387283238,
+      "loss": 0.2245,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017052023121387283,
+      "loss": 0.2693,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017080924855491328,
+      "loss": 0.1794,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017109826589595375,
+      "loss": 0.1282,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017138728323699423,
+      "loss": 0.1932,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001716763005780347,
+      "loss": 0.1543,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017196531791907515,
+      "loss": 0.1573,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001722543352601156,
+      "loss": 0.1266,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017254335260115609,
+      "loss": 0.1823,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017283236994219654,
+      "loss": 0.186,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00173121387283237,
+      "loss": 0.1995,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017341040462427746,
+      "loss": 0.161,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001736994219653179,
+      "loss": 0.1667,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001739884393063584,
+      "loss": 0.1622,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017427745664739885,
+      "loss": 0.1809,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001745664739884393,
+      "loss": 0.1123,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017485549132947977,
+      "loss": 0.2004,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017514450867052024,
+      "loss": 0.2214,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001754335260115607,
+      "loss": 0.2078,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017572254335260116,
+      "loss": 0.1869,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001760115606936416,
+      "loss": 0.1201,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.001763005780346821,
+      "loss": 0.2837,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017658959537572255,
+      "loss": 0.1555,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00176878612716763,
+      "loss": 0.1793,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017716763005780347,
+      "loss": 0.1562,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017745664739884392,
+      "loss": 0.2195,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017774566473988441,
+      "loss": 0.1681,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017803468208092486,
+      "loss": 0.175,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017832369942196531,
+      "loss": 0.4395,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017861271676300578,
+      "loss": 0.4031,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017890173410404625,
+      "loss": 0.8701,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017919075144508672,
+      "loss": 0.374,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017947976878612717,
+      "loss": 0.3457,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0017976878612716762,
+      "loss": 0.4771,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001800578034682081,
+      "loss": 0.3965,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018034682080924856,
+      "loss": 0.3774,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018063583815028901,
+      "loss": 0.2891,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018092485549132948,
+      "loss": 0.3516,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018121387283236993,
+      "loss": 0.2463,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018150289017341043,
+      "loss": 0.3682,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018179190751445088,
+      "loss": 0.3176,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018208092485549133,
+      "loss": 0.4678,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001823699421965318,
+      "loss": 0.3008,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018265895953757225,
+      "loss": 0.3005,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018294797687861272,
+      "loss": 0.3281,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018323699421965319,
+      "loss": 0.3428,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018352601156069364,
+      "loss": 0.3337,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001838150289017341,
+      "loss": 0.3701,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018410404624277458,
+      "loss": 0.2788,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018439306358381503,
+      "loss": 0.3994,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001846820809248555,
+      "loss": 0.3481,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018497109826589595,
+      "loss": 0.4199,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018526011560693644,
+      "loss": 0.2324,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001855491329479769,
+      "loss": 0.3691,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018583815028901734,
+      "loss": 0.259,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001861271676300578,
+      "loss": 0.3562,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018641618497109826,
+      "loss": 0.3706,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018670520231213873,
+      "loss": 0.2817,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001869942196531792,
+      "loss": 0.4092,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018728323699421965,
+      "loss": 0.394,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018757225433526012,
+      "loss": 0.321,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001878612716763006,
+      "loss": 0.4038,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018815028901734104,
+      "loss": 0.4043,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018843930635838151,
+      "loss": 0.375,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018872832369942196,
+      "loss": 0.2334,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018901734104046243,
+      "loss": 0.3066,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.001893063583815029,
+      "loss": 0.3716,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018959537572254335,
+      "loss": 0.3584,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0018988439306358382,
+      "loss": 0.1808,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019017341040462427,
+      "loss": 0.2937,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019046242774566474,
+      "loss": 0.366,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019075144508670522,
+      "loss": 0.2815,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019104046242774566,
+      "loss": 0.2649,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019132947976878614,
+      "loss": 0.2732,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019161849710982658,
+      "loss": 0.3926,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019190751445086706,
+      "loss": 0.3196,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019219653179190753,
+      "loss": 0.2922,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019248554913294798,
+      "loss": 0.3105,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019277456647398843,
+      "loss": 0.2847,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0019306358381502892,
+      "loss": 0.3672,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019335260115606937,
+      "loss": 0.2158,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019364161849710984,
+      "loss": 0.2046,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019393063583815029,
+      "loss": 0.2659,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019421965317919074,
+      "loss": 0.2461,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019450867052023123,
+      "loss": 0.3008,
+      "step": 673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019479768786127168,
+      "loss": 0.2347,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019508670520231215,
+      "loss": 0.2573,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001953757225433526,
+      "loss": 0.2542,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019566473988439307,
+      "loss": 0.2834,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019595375722543354,
+      "loss": 0.2852,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00196242774566474,
+      "loss": 0.2371,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019653179190751444,
+      "loss": 0.3413,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001968208092485549,
+      "loss": 0.2881,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001971098265895954,
+      "loss": 0.1945,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019739884393063585,
+      "loss": 0.2717,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019768786127167632,
+      "loss": 0.2603,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019797687861271675,
+      "loss": 0.2415,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019826589595375722,
+      "loss": 0.3638,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001985549132947977,
+      "loss": 0.3398,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019884393063583816,
+      "loss": 0.2517,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019913294797687863,
+      "loss": 0.2295,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019942196531791906,
+      "loss": 0.2703,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019971098265895953,
+      "loss": 0.2427,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.002,
+      "loss": 0.2893,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999999901412625,
+      "loss": 0.3086,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00199999996056505,
+      "loss": 0.3579,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999999112713636,
+      "loss": 0.2339,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999998422602036,
+      "loss": 0.2136,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001999999753531572,
+      "loss": 0.2065,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00199999964508547,
+      "loss": 0.2332,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999995169219,
+      "loss": 0.2512,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999993690408644,
+      "loss": 0.1733,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999992014423664,
+      "loss": 0.2139,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999990141264094,
+      "loss": 0.2197,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999988070929964,
+      "loss": 0.2288,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999985803421318,
+      "loss": 0.2537,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999983338738203,
+      "loss": 0.2715,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999980676880668,
+      "loss": 0.2271,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001999997781784876,
+      "loss": 0.1854,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001999997476164254,
+      "loss": 0.2217,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999971508262073,
+      "loss": 0.301,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999968057707413,
+      "loss": 0.2312,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001999996440997863,
+      "loss": 0.2417,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999960565075803,
+      "loss": 0.1995,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999956522999004,
+      "loss": 0.1941,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.001999995228374831,
+      "loss": 0.1816,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0019999947847323804,
+      "loss": 0.2766,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999994321372558,
+      "loss": 0.2563,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999993838295372,
+      "loss": 0.1973,
+      "step": 717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999993335500833,
+      "loss": 0.2616,
+      "step": 718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999928129889504,
+      "loss": 0.4575,
+      "step": 719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999922707597344,
+      "loss": 0.15,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999917088131956,
+      "loss": 0.2179,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999911271493453,
+      "loss": 0.2589,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999990525768195,
+      "loss": 0.2456,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999989904669757,
+      "loss": 0.2263,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999989263854042,
+      "loss": 0.2507,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999886033210644,
+      "loss": 0.2019,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999879230708363,
+      "loss": 0.1963,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999987223103371,
+      "loss": 0.1719,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999865034186826,
+      "loss": 0.2029,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999857640167857,
+      "loss": 0.1848,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999850048976944,
+      "loss": 0.2001,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999842260614237,
+      "loss": 0.219,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999983427507989,
+      "loss": 0.2991,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999982609237406,
+      "loss": 0.3103,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999981771249691,
+      "loss": 0.2605,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00199998091354486,
+      "loss": 0.2129,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999800361229308,
+      "loss": 0.2793,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00199997913898392,
+      "loss": 0.3206,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999782221278455,
+      "loss": 0.2231,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999772855547253,
+      "loss": 0.1792,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999976329264578,
+      "loss": 0.2483,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999753532574226,
+      "loss": 0.2661,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999974357533278,
+      "loss": 0.1849,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999733420921636,
+      "loss": 0.249,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999723069341,
+      "loss": 0.2079,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999712520591077,
+      "loss": 0.3098,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999701774672074,
+      "loss": 0.1772,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999690831584197,
+      "loss": 0.2269,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999679691327667,
+      "loss": 0.261,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00199996683539027,
+      "loss": 0.1467,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999656819309525,
+      "loss": 0.2378,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999645087548363,
+      "loss": 0.2898,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999963315861945,
+      "loss": 0.3143,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999962103252302,
+      "loss": 0.1379,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999608709259315,
+      "loss": 0.3276,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999596188828577,
+      "loss": 0.199,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999583471231046,
+      "loss": 0.23,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.001999957055646698,
+      "loss": 0.1654,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999557444536627,
+      "loss": 0.2455,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999544135440254,
+      "loss": 0.2285,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0019999530629178116,
+      "loss": 0.1761,
+      "step": 761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999516925750488,
+      "loss": 0.1921,
+      "step": 762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999503025157628,
+      "loss": 0.1722,
+      "step": 763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999488927399823,
+      "loss": 0.2024,
+      "step": 764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999947463247734,
+      "loss": 0.1917,
+      "step": 765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999460140390467,
+      "loss": 0.1552,
+      "step": 766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999944545113949,
+      "loss": 0.1823,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999430564724696,
+      "loss": 0.1769,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999415481146378,
+      "loss": 0.325,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999940020040484,
+      "loss": 0.2623,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999384722500374,
+      "loss": 0.1744,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999369047433288,
+      "loss": 0.1963,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999353175203897,
+      "loss": 0.2271,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999337105812507,
+      "loss": 0.2053,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999320839259437,
+      "loss": 0.259,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999930437554501,
+      "loss": 0.1919,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999287714669543,
+      "loss": 0.178,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999927085663338,
+      "loss": 0.2773,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999253801436835,
+      "loss": 0.29,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999923654908026,
+      "loss": 0.1729,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999219099563984,
+      "loss": 0.2234,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999201452888353,
+      "loss": 0.2036,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999918360905372,
+      "loss": 0.2061,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999916556806043,
+      "loss": 0.1876,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999147329908848,
+      "loss": 0.3013,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999912889459933,
+      "loss": 0.2683,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999911026213223,
+      "loss": 0.2087,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999909143250793,
+      "loss": 0.2429,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999907240572679,
+      "loss": 0.2178,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999905318178919,
+      "loss": 0.2603,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019999033760695513,
+      "loss": 0.25,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999901414244613,
+      "loss": 0.3196,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999899432704144,
+      "loss": 0.1948,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999897431448183,
+      "loss": 0.2251,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019998954104767696,
+      "loss": 0.2954,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999893369789943,
+      "loss": 0.175,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999891309387744,
+      "loss": 0.144,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999889229270213,
+      "loss": 0.1925,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999887129437391,
+      "loss": 0.2297,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019998850098893197,
+      "loss": 0.2617,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019998828706260406,
+      "loss": 0.2192,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019998807116475958,
+      "loss": 0.1733,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999878532954028,
+      "loss": 0.207,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019998763345453804,
+      "loss": 0.1903,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999874116421696,
+      "loss": 0.1721,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.001999871878583019,
+      "loss": 0.2788,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0019998696210293933,
+      "loss": 0.1582,
+      "step": 807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999867343760863,
+      "loss": 0.1877,
+      "step": 808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998650467774734,
+      "loss": 0.2168,
+      "step": 809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998627300792694,
+      "loss": 0.1088,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998603936662975,
+      "loss": 0.2124,
+      "step": 811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999858037538603,
+      "loss": 0.1654,
+      "step": 812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998556616962326,
+      "loss": 0.2214,
+      "step": 813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998532661392333,
+      "loss": 0.2321,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999850850867652,
+      "loss": 0.2031,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998484158815365,
+      "loss": 0.1578,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999845961180935,
+      "loss": 0.2341,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998434867658957,
+      "loss": 0.2056,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998409926364675,
+      "loss": 0.2971,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998384787926994,
+      "loss": 0.1663,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998359452346407,
+      "loss": 0.2698,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998333919623423,
+      "loss": 0.209,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998308189758535,
+      "loss": 0.2742,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999828226275226,
+      "loss": 0.1589,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00199982561386051,
+      "loss": 0.2651,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998229817317576,
+      "loss": 0.2727,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999820329889021,
+      "loss": 0.2795,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999817658332351,
+      "loss": 0.2133,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999814967061802,
+      "loss": 0.2129,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998122560774263,
+      "loss": 0.2249,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998095253792774,
+      "loss": 0.1976,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999806774967409,
+      "loss": 0.1356,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019998040048418756,
+      "loss": 0.1917,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999801215002731,
+      "loss": 0.1697,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999798405450032,
+      "loss": 0.1636,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997955761838324,
+      "loss": 0.2319,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997927272041886,
+      "loss": 0.2224,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997898585111568,
+      "loss": 0.2263,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999786970104793,
+      "loss": 0.1412,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999784061985155,
+      "loss": 0.1682,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997811341522995,
+      "loss": 0.1646,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997781866062845,
+      "loss": 0.1876,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999775219347168,
+      "loss": 0.1884,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999772232375009,
+      "loss": 0.1439,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997692256898654,
+      "loss": 0.2043,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997661992917975,
+      "loss": 0.1798,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999763153180864,
+      "loss": 0.1741,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999760087357126,
+      "loss": 0.17,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997570018206435,
+      "loss": 0.1663,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.001999753896571477,
+      "loss": 0.2142,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997507716096883,
+      "loss": 0.1897,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0019997476269353383,
+      "loss": 0.1956,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00199974446254849,
+      "loss": 0.2107,
+      "step": 853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999741278449205,
+      "loss": 0.2645,
+      "step": 854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997380746375457,
+      "loss": 0.3147,
+      "step": 855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997348511135767,
+      "loss": 0.1399,
+      "step": 856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997316078773602,
+      "loss": 0.2148,
+      "step": 857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997283449289614,
+      "loss": 0.1721,
+      "step": 858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999725062268443,
+      "loss": 0.2708,
+      "step": 859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997217598958714,
+      "loss": 0.1522,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997184378113104,
+      "loss": 0.2095,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997150960148263,
+      "loss": 0.1807,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999711734506485,
+      "loss": 0.3347,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999708353286352,
+      "loss": 0.1759,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997049523544953,
+      "loss": 0.2351,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019997015317109804,
+      "loss": 0.1876,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999698091355876,
+      "loss": 0.1837,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996946312892494,
+      "loss": 0.1655,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996911515111687,
+      "loss": 0.2139,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996876520217027,
+      "loss": 0.1609,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996841328209203,
+      "loss": 0.1631,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999680593908891,
+      "loss": 0.1958,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996770352856846,
+      "loss": 0.1819,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996734569513717,
+      "loss": 0.2556,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996698589060218,
+      "loss": 0.2349,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996662411497067,
+      "loss": 0.1737,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996626036824974,
+      "loss": 0.2241,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996589465044656,
+      "loss": 0.1527,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996552696156834,
+      "loss": 0.2136,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996515730162233,
+      "loss": 0.1726,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999647856706159,
+      "loss": 0.2034,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996441206855623,
+      "loss": 0.2764,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999640364954508,
+      "loss": 0.2173,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996365895130694,
+      "loss": 0.1892,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999632794361322,
+      "loss": 0.1992,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999628979499339,
+      "loss": 0.1987,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999625144927197,
+      "loss": 0.1437,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996212906449712,
+      "loss": 0.2319,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999617416652738,
+      "loss": 0.248,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999613522950573,
+      "loss": 0.1792,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999609609538553,
+      "loss": 0.2637,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996056764167557,
+      "loss": 0.2461,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019996017235852587,
+      "loss": 0.178,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019995977510441393,
+      "loss": 0.2051,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019995937587934764,
+      "loss": 0.1707,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999589746833348,
+      "loss": 0.1892,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019995857151638347,
+      "loss": 0.1549,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0019995816637850146,
+      "loss": 0.1622,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.001999577592696968,
+      "loss": 0.1263,
+      "step": 899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999573501899775,
+      "loss": 0.1926,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995693913935166,
+      "loss": 0.2729,
+      "step": 901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999565261178274,
+      "loss": 0.2136,
+      "step": 902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999561111254128,
+      "loss": 0.2218,
+      "step": 903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995569416211604,
+      "loss": 0.1816,
+      "step": 904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995527522794544,
+      "loss": 0.241,
+      "step": 905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995485432290913,
+      "loss": 0.1658,
+      "step": 906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995443144701554,
+      "loss": 0.1229,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999540066002729,
+      "loss": 0.2192,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995357978268968,
+      "loss": 0.2004,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999531509942742,
+      "loss": 0.2131,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00199952720235035,
+      "loss": 0.1078,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999522875049805,
+      "loss": 0.1287,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999518528041193,
+      "loss": 0.1422,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999514161324599,
+      "loss": 0.1537,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00199950977490011,
+      "loss": 0.1591,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019995053687678114,
+      "loss": 0.1449,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999500942927791,
+      "loss": 0.119,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994964973801357,
+      "loss": 0.1458,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999492032124933,
+      "loss": 0.2476,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994875471622715,
+      "loss": 0.1746,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994830424922387,
+      "loss": 0.1527,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999478518114924,
+      "loss": 0.1295,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999473974030417,
+      "loss": 0.2803,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994694102388067,
+      "loss": 0.1633,
+      "step": 924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994648267401835,
+      "loss": 0.1593,
+      "step": 925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999460223534637,
+      "loss": 0.1816,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999455600622259,
+      "loss": 0.2158,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00199945095800314,
+      "loss": 0.1615,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994462956773717,
+      "loss": 0.1731,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999441613645046,
+      "loss": 0.186,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999436911906255,
+      "loss": 0.1559,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994321904610924,
+      "loss": 0.1832,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994274493096496,
+      "loss": 0.1383,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999422688452022,
+      "loss": 0.2109,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994179078883017,
+      "loss": 0.1525,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999413107618584,
+      "loss": 0.1855,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019994082876429635,
+      "loss": 0.1859,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999403447961535,
+      "loss": 0.1931,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019993985885743935,
+      "loss": 0.193,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999393709481636,
+      "loss": 0.229,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019993888106833576,
+      "loss": 0.2454,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019993838921796557,
+      "loss": 0.1833,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.001999378953970626,
+      "loss": 0.1666,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019993739960563676,
+      "loss": 0.1539,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0019993690184369767,
+      "loss": 0.2065,
+      "step": 945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993640211125527,
+      "loss": 0.2286,
+      "step": 946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999359004083193,
+      "loss": 0.1545,
+      "step": 947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993539673489975,
+      "loss": 0.1864,
+      "step": 948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999348910910065,
+      "loss": 0.1373,
+      "step": 949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993438347664957,
+      "loss": 0.209,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999338738918389,
+      "loss": 0.1378,
+      "step": 951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993336233658453,
+      "loss": 0.2275,
+      "step": 952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993284881089663,
+      "loss": 0.14,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993233331478525,
+      "loss": 0.1937,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993181584826063,
+      "loss": 0.199,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993129641133287,
+      "loss": 0.2126,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999307750040123,
+      "loss": 0.1405,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019993025162630914,
+      "loss": 0.204,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999297262782338,
+      "loss": 0.197,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019992919895979655,
+      "loss": 0.1489,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999286696710078,
+      "loss": 0.1219,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00199928138411878,
+      "loss": 0.2334,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019992760518241764,
+      "loss": 0.2117,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019992706998263722,
+      "loss": 0.1685,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999265328125473,
+      "loss": 0.0988,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999259936721584,
+      "loss": 0.1586,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999254525614813,
+      "loss": 0.1841,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999249094805266,
+      "loss": 0.1938,
+      "step": 968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999243644293049,
+      "loss": 0.209,
+      "step": 969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999238174078271,
+      "loss": 0.2007,
+      "step": 970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999232684161039,
+      "loss": 0.0779,
+      "step": 971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019992271745414617,
+      "loss": 0.1278,
+      "step": 972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019992216452196475,
+      "loss": 0.2363,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019992160961957053,
+      "loss": 0.1801,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999210527469745,
+      "loss": 0.1927,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999204939041876,
+      "loss": 0.1328,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991993309122083,
+      "loss": 0.1461,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999193703080853,
+      "loss": 0.2136,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991880555479207,
+      "loss": 0.1672,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999182388313523,
+      "loss": 0.1423,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991767013777715,
+      "loss": 0.1985,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991709947407782,
+      "loss": 0.1586,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999165268402656,
+      "loss": 0.1487,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999159522363517,
+      "loss": 0.2351,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991537566234753,
+      "loss": 0.1062,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.001999147971182645,
+      "loss": 0.1907,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991421660411385,
+      "loss": 0.0905,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991363411990723,
+      "loss": 0.1355,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991304966565596,
+      "loss": 0.1326,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991246324137166,
+      "loss": 0.1829,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0019991187484706585,
+      "loss": 0.1088,
+      "step": 991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019991128448275013,
+      "loss": 0.1461,
+      "step": 992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019991069214843615,
+      "loss": 0.23,
+      "step": 993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019991009784413563,
+      "loss": 0.1768,
+      "step": 994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990950156986024,
+      "loss": 0.1506,
+      "step": 995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001999089033256217,
+      "loss": 0.188,
+      "step": 996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990830311143195,
+      "loss": 0.2832,
+      "step": 997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990770092730265,
+      "loss": 0.179,
+      "step": 998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990709677324575,
+      "loss": 0.1636,
+      "step": 999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001999064906492732,
+      "loss": 0.2966,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001999058825553969,
+      "loss": 0.1943,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990527249162887,
+      "loss": 0.1412,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001999046604579811,
+      "loss": 0.2192,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990404645446568,
+      "loss": 0.2556,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990343048109475,
+      "loss": 0.2053,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001999028125378804,
+      "loss": 0.1439,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990219262483486,
+      "loss": 0.2979,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990157074197034,
+      "loss": 0.1746,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990094688929903,
+      "loss": 0.118,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019990032106683335,
+      "loss": 0.1093,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998996932745856,
+      "loss": 0.2307,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998990635125681,
+      "loss": 0.1937,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989843178079333,
+      "loss": 0.1359,
+      "step": 1013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989779807927373,
+      "loss": 0.1586,
+      "step": 1014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989716240802176,
+      "loss": 0.1785,
+      "step": 1015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989652476705,
+      "loss": 0.2017,
+      "step": 1016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989588515637105,
+      "loss": 0.2285,
+      "step": 1017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989524357599746,
+      "loss": 0.2087,
+      "step": 1018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989460002594185,
+      "loss": 0.1912,
+      "step": 1019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989395450621702,
+      "loss": 0.21,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998933070168356,
+      "loss": 0.2925,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998926575578104,
+      "loss": 0.1515,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989200612915425,
+      "loss": 0.1716,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989135273087992,
+      "loss": 0.16,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998906973630004,
+      "loss": 0.2332,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019989004002552847,
+      "loss": 0.2842,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998893807184772,
+      "loss": 0.1149,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019988871944185958,
+      "loss": 0.2446,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998880561956886,
+      "loss": 0.2019,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998873909799774,
+      "loss": 0.2915,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019988672379473906,
+      "loss": 0.1384,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019988605463998666,
+      "loss": 0.166,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019988538351573357,
+      "loss": 0.1423,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998847104219929,
+      "loss": 0.1882,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0019988403535877793,
+      "loss": 0.1017,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00199883358326102,
+      "loss": 0.1581,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.001998826793239784,
+      "loss": 0.1846,
+      "step": 1037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019988199835242063,
+      "loss": 0.1931,
+      "step": 1038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019988131541144203,
+      "loss": 0.2241,
+      "step": 1039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019988063050105608,
+      "loss": 0.197,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987994362127633,
+      "loss": 0.2524,
+      "step": 1041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998792547721163,
+      "loss": 0.1533,
+      "step": 1042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987856395358946,
+      "loss": 0.2197,
+      "step": 1043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987787116570958,
+      "loss": 0.1797,
+      "step": 1044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987717640849024,
+      "loss": 0.1465,
+      "step": 1045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998764796819452,
+      "loss": 0.151,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987578098608822,
+      "loss": 0.1606,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987508032093294,
+      "loss": 0.2043,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998743776864933,
+      "loss": 0.2097,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998736730827831,
+      "loss": 0.1927,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987296650981626,
+      "loss": 0.1123,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987225796760668,
+      "loss": 0.1553,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019987154745616837,
+      "loss": 0.1506,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998708349755153,
+      "loss": 0.1345,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998701205256615,
+      "loss": 0.1973,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986940410662113,
+      "loss": 0.1941,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986868571840828,
+      "loss": 0.1619,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986796536103713,
+      "loss": 0.1824,
+      "step": 1058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986724303452186,
+      "loss": 0.1254,
+      "step": 1059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986651873887666,
+      "loss": 0.1421,
+      "step": 1060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986579247411596,
+      "loss": 0.1499,
+      "step": 1061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986506424025394,
+      "loss": 0.124,
+      "step": 1062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00199864334037305,
+      "loss": 0.2036,
+      "step": 1063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986360186528356,
+      "loss": 0.1698,
+      "step": 1064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986286772420408,
+      "loss": 0.1519,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986213161408096,
+      "loss": 0.1215,
+      "step": 1066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986139353492874,
+      "loss": 0.1896,
+      "step": 1067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019986065348676202,
+      "loss": 0.167,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985991146959535,
+      "loss": 0.1333,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998591674834434,
+      "loss": 0.13,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985842152832074,
+      "loss": 0.1582,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998576736042422,
+      "loss": 0.2263,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985692371122248,
+      "loss": 0.2786,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998561718492763,
+      "loss": 0.2125,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998554180184186,
+      "loss": 0.125,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.001998546622186642,
+      "loss": 0.1716,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985390445002795,
+      "loss": 0.1838,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985314471252487,
+      "loss": 0.1149,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985238300616985,
+      "loss": 0.1464,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00199851619330978,
+      "loss": 0.1802,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985085368696438,
+      "loss": 0.2541,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019985008607414398,
+      "loss": 0.1868,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00199849316492532,
+      "loss": 0.1289,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0019984854494214366,
+      "loss": 0.1342,
+      "step": 1084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998477714229941,
+      "loss": 0.1378,
+      "step": 1085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998469959350986,
+      "loss": 0.2207,
+      "step": 1086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019984621847847247,
+      "loss": 0.2266,
+      "step": 1087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00199845439053131,
+      "loss": 0.2087,
+      "step": 1088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019984465765908958,
+      "loss": 0.1157,
+      "step": 1089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019984387429636367,
+      "loss": 0.1239,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019984308896496858,
+      "loss": 0.1295,
+      "step": 1091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019984230166491994,
+      "loss": 0.1788,
+      "step": 1092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019984151239623316,
+      "loss": 0.1135,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998407211589239,
+      "loss": 0.1658,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998399279530077,
+      "loss": 0.1686,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998391327785002,
+      "loss": 0.1219,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998383356354171,
+      "loss": 0.1882,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998375365237741,
+      "loss": 0.1607,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983673544358696,
+      "loss": 0.1912,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983593239487154,
+      "loss": 0.2061,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983512737764353,
+      "loss": 0.1086,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983432039191893,
+      "loss": 0.1261,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998335114377136,
+      "loss": 0.1616,
+      "step": 1103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983270051504355,
+      "loss": 0.1743,
+      "step": 1104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983188762392468,
+      "loss": 0.1257,
+      "step": 1105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983107276437305,
+      "loss": 0.1774,
+      "step": 1106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019983025593640477,
+      "loss": 0.1341,
+      "step": 1107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998294371400359,
+      "loss": 0.0994,
+      "step": 1108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982861637528257,
+      "loss": 0.1174,
+      "step": 1109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982779364216105,
+      "loss": 0.124,
+      "step": 1110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982696894068745,
+      "loss": 0.141,
+      "step": 1111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998261422708781,
+      "loss": 0.1444,
+      "step": 1112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998253136327493,
+      "loss": 0.1909,
+      "step": 1113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982448302631735,
+      "loss": 0.1554,
+      "step": 1114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982365045159863,
+      "loss": 0.1672,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998228159086096,
+      "loss": 0.1504,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982197939736666,
+      "loss": 0.1943,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982114091788637,
+      "loss": 0.1489,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019982030047018522,
+      "loss": 0.2394,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998194580542798,
+      "loss": 0.1257,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019981861367018667,
+      "loss": 0.1232,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019981776731792253,
+      "loss": 0.1385,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998169189975041,
+      "loss": 0.2776,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00199816068708948,
+      "loss": 0.1292,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998152164522711,
+      "loss": 0.1649,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019981436222749017,
+      "loss": 0.1526,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00199813506034622,
+      "loss": 0.1625,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019981264787368356,
+      "loss": 0.147,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.001998117877446917,
+      "loss": 0.1586,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0019981092564766343,
+      "loss": 0.1459,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001998100615826157,
+      "loss": 0.1294,
+      "step": 1131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001998091955495656,
+      "loss": 0.1731,
+      "step": 1132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980832754853016,
+      "loss": 0.1794,
+      "step": 1133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980745757952655,
+      "loss": 0.1171,
+      "step": 1134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980658564257184,
+      "loss": 0.2178,
+      "step": 1135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980571173768328,
+      "loss": 0.1682,
+      "step": 1136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980483586487813,
+      "loss": 0.1771,
+      "step": 1137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001998039580241736,
+      "loss": 0.1442,
+      "step": 1138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00199803078215587,
+      "loss": 0.2454,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001998021964391357,
+      "loss": 0.1665,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980131269483714,
+      "loss": 0.2349,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019980042698270863,
+      "loss": 0.194,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997995393027677,
+      "loss": 0.174,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997986496550319,
+      "loss": 0.2073,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997977580395186,
+      "loss": 0.2122,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997968644562456,
+      "loss": 0.2128,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997959689052304,
+      "loss": 0.1561,
+      "step": 1147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019979507138649065,
+      "loss": 0.1812,
+      "step": 1148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997941719000441,
+      "loss": 0.2523,
+      "step": 1149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019979327044590844,
+      "loss": 0.1685,
+      "step": 1150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997923670241015,
+      "loss": 0.2275,
+      "step": 1151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00199791461634641,
+      "loss": 0.1597,
+      "step": 1152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019979055427754486,
+      "loss": 0.2544,
+      "step": 1153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978964495283094,
+      "loss": 0.1829,
+      "step": 1154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997887336605173,
+      "loss": 0.2407,
+      "step": 1155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978782040062167,
+      "loss": 0.1887,
+      "step": 1156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978690517316222,
+      "loss": 0.2715,
+      "step": 1157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978598797815695,
+      "loss": 0.1385,
+      "step": 1158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00199785068815624,
+      "loss": 0.2244,
+      "step": 1159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997841476855814,
+      "loss": 0.1613,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978322458804737,
+      "loss": 0.2054,
+      "step": 1161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978229952304014,
+      "loss": 0.1888,
+      "step": 1162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997813724905779,
+      "loss": 0.1897,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019978044349067887,
+      "loss": 0.1583,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019977951252336153,
+      "loss": 0.2229,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019977857958864407,
+      "loss": 0.2136,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00199777644686545,
+      "loss": 0.1508,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997767078170827,
+      "loss": 0.1128,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019977576898027568,
+      "loss": 0.1511,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997748281761424,
+      "loss": 0.2568,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019977388540470143,
+      "loss": 0.267,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997729406659714,
+      "loss": 0.2524,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019977199395997086,
+      "loss": 0.2881,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0019977104528671854,
+      "loss": 0.2356,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997700946462331,
+      "loss": 0.3035,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.001997691420385334,
+      "loss": 0.1425,
+      "step": 1176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019976818746363805,
+      "loss": 0.1498,
+      "step": 1177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00199767230921566,
+      "loss": 0.1917,
+      "step": 1178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019976627241233605,
+      "loss": 0.167,
+      "step": 1179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997653119359671,
+      "loss": 0.1765,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997643494924781,
+      "loss": 0.1135,
+      "step": 1181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019976338508188806,
+      "loss": 0.1804,
+      "step": 1182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019976241870421595,
+      "loss": 0.1014,
+      "step": 1183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019976145035948086,
+      "loss": 0.1844,
+      "step": 1184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019976048004770185,
+      "loss": 0.171,
+      "step": 1185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975950776889803,
+      "loss": 0.1636,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975853352308865,
+      "loss": 0.1429,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975755731029285,
+      "loss": 0.1427,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975657913052993,
+      "loss": 0.1484,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975559898381917,
+      "loss": 0.1438,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975461687017984,
+      "loss": 0.1937,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975363278963132,
+      "loss": 0.1836,
+      "step": 1192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997526467421931,
+      "loss": 0.1671,
+      "step": 1193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975165872788453,
+      "loss": 0.1793,
+      "step": 1194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019975066874672515,
+      "loss": 0.1824,
+      "step": 1195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997496767987344,
+      "loss": 0.1167,
+      "step": 1196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974868288393194,
+      "loss": 0.2019,
+      "step": 1197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974768700233726,
+      "loss": 0.207,
+      "step": 1198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997466891539701,
+      "loss": 0.1368,
+      "step": 1199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974568933885006,
+      "loss": 0.2139,
+      "step": 1200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997446875569969,
+      "loss": 0.2148,
+      "step": 1201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974368380843033,
+      "loss": 0.1356,
+      "step": 1202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974267809317017,
+      "loss": 0.2178,
+      "step": 1203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974167041123627,
+      "loss": 0.1677,
+      "step": 1204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019974066076264845,
+      "loss": 0.1775,
+      "step": 1205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973964914742665,
+      "loss": 0.217,
+      "step": 1206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973863556559082,
+      "loss": 0.2957,
+      "step": 1207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973762001716092,
+      "loss": 0.1675,
+      "step": 1208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973660250215703,
+      "loss": 0.1533,
+      "step": 1209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997355830205991,
+      "loss": 0.198,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973456157250736,
+      "loss": 0.1348,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973353815790186,
+      "loss": 0.27,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973251277680282,
+      "loss": 0.1322,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973148542923045,
+      "loss": 0.2048,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019973045611520503,
+      "loss": 0.1389,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997294248347468,
+      "loss": 0.2473,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997283915878761,
+      "loss": 0.2251,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997273563746134,
+      "loss": 0.1443,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019972631919497903,
+      "loss": 0.3184,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.001997252800489934,
+      "loss": 0.1626,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019972423893667712,
+      "loss": 0.1819,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0019972319585805057,
+      "loss": 0.1233,
+      "step": 1222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019972215081313447,
+      "loss": 0.2161,
+      "step": 1223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997211038019493,
+      "loss": 0.1869,
+      "step": 1224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019972005482451582,
+      "loss": 0.1936,
+      "step": 1225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019971900388085456,
+      "loss": 0.2028,
+      "step": 1226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997179509709864,
+      "loss": 0.1448,
+      "step": 1227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00199716896094932,
+      "loss": 0.1768,
+      "step": 1228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997158392527122,
+      "loss": 0.1926,
+      "step": 1229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019971478044434787,
+      "loss": 0.1268,
+      "step": 1230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997137196698598,
+      "loss": 0.1653,
+      "step": 1231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019971265692926895,
+      "loss": 0.151,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019971159222259625,
+      "loss": 0.1649,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997105255498628,
+      "loss": 0.1428,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997094569110895,
+      "loss": 0.1499,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997083863062975,
+      "loss": 0.158,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019970731373550785,
+      "loss": 0.1752,
+      "step": 1237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019970623919874178,
+      "loss": 0.1117,
+      "step": 1238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019970516269602035,
+      "loss": 0.2075,
+      "step": 1239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997040842273649,
+      "loss": 0.1416,
+      "step": 1240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001997030037927967,
+      "loss": 0.165,
+      "step": 1241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00199701921392337,
+      "loss": 0.1429,
+      "step": 1242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019970083702600715,
+      "loss": 0.1531,
+      "step": 1243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996997506938285,
+      "loss": 0.1602,
+      "step": 1244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019969866239582254,
+      "loss": 0.1523,
+      "step": 1245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996975721320107,
+      "loss": 0.2,
+      "step": 1246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019969647990241445,
+      "loss": 0.1304,
+      "step": 1247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019969538570705535,
+      "loss": 0.1658,
+      "step": 1248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00199694289545955,
+      "loss": 0.1924,
+      "step": 1249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019969319141913493,
+      "loss": 0.2224,
+      "step": 1250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996920913266169,
+      "loss": 0.1045,
+      "step": 1251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019969098926842254,
+      "loss": 0.1417,
+      "step": 1252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019968988524457355,
+      "loss": 0.1637,
+      "step": 1253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996887792550918,
+      "loss": 0.2471,
+      "step": 1254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00199687671299999,
+      "loss": 0.1381,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019968656137931704,
+      "loss": 0.2146,
+      "step": 1256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019968544949306778,
+      "loss": 0.2045,
+      "step": 1257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019968433564127317,
+      "loss": 0.2349,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019968321982395515,
+      "loss": 0.252,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019968210204113573,
+      "loss": 0.1378,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00199680982292837,
+      "loss": 0.2064,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996798605790809,
+      "loss": 0.2402,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019967873689988973,
+      "loss": 0.2064,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996776112552855,
+      "loss": 0.2419,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996764836452905,
+      "loss": 0.1166,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0019967535406992688,
+      "loss": 0.2147,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00199674222529217,
+      "loss": 0.207,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.001996730890231831,
+      "loss": 0.1743,
+      "step": 1268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019967195355184756,
+      "loss": 0.1772,
+      "step": 1269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019967081611523274,
+      "loss": 0.1949,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019966967671336115,
+      "loss": 0.1865,
+      "step": 1271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019966853534625515,
+      "loss": 0.1663,
+      "step": 1272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996673920139373,
+      "loss": 0.1456,
+      "step": 1273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019966624671643018,
+      "loss": 0.1123,
+      "step": 1274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996650994537563,
+      "loss": 0.2202,
+      "step": 1275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019966395022593827,
+      "loss": 0.1398,
+      "step": 1276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019966279903299885,
+      "loss": 0.1979,
+      "step": 1277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996616458749606,
+      "loss": 0.1101,
+      "step": 1278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019966049075184644,
+      "loss": 0.1549,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019965933366367897,
+      "loss": 0.1392,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996581746104811,
+      "loss": 0.2225,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996570135922757,
+      "loss": 0.1458,
+      "step": 1282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019965585060908557,
+      "loss": 0.1516,
+      "step": 1283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019965468566093374,
+      "loss": 0.1506,
+      "step": 1284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019965351874784306,
+      "loss": 0.1825,
+      "step": 1285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996523498698367,
+      "loss": 0.142,
+      "step": 1286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996511790269376,
+      "loss": 0.1246,
+      "step": 1287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996500062191688,
+      "loss": 0.182,
+      "step": 1288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964883144655355,
+      "loss": 0.2036,
+      "step": 1289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964765470911494,
+      "loss": 0.2046,
+      "step": 1290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964647600687617,
+      "loss": 0.145,
+      "step": 1291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964529533986054,
+      "loss": 0.1919,
+      "step": 1292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964411270809125,
+      "loss": 0.2429,
+      "step": 1293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964292811159167,
+      "loss": 0.1639,
+      "step": 1294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019964174155038513,
+      "loss": 0.1259,
+      "step": 1295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996405530244951,
+      "loss": 0.1361,
+      "step": 1296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019963936253394486,
+      "loss": 0.1403,
+      "step": 1297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019963817007875805,
+      "loss": 0.1831,
+      "step": 1298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996369756589581,
+      "loss": 0.1528,
+      "step": 1299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019963577927456853,
+      "loss": 0.1866,
+      "step": 1300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00199634580925613,
+      "loss": 0.259,
+      "step": 1301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996333806121151,
+      "loss": 0.1466,
+      "step": 1302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996321783340985,
+      "loss": 0.1902,
+      "step": 1303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019963097409158694,
+      "loss": 0.1249,
+      "step": 1304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019962976788460416,
+      "loss": 0.1135,
+      "step": 1305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019962855971317386,
+      "loss": 0.1229,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996273495773199,
+      "loss": 0.158,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019962613747706625,
+      "loss": 0.1659,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996249234124367,
+      "loss": 0.1847,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996237073834552,
+      "loss": 0.1549,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019962248939014568,
+      "loss": 0.2346,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019962126943253223,
+      "loss": 0.2173,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001996200475106389,
+      "loss": 0.1487,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0019961882362448985,
+      "loss": 0.1025,
+      "step": 1314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019961759777410906,
+      "loss": 0.0989,
+      "step": 1315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019961636995952077,
+      "loss": 0.1304,
+      "step": 1316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019961514018074925,
+      "loss": 0.2065,
+      "step": 1317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001996139084378186,
+      "loss": 0.0992,
+      "step": 1318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019961267473075327,
+      "loss": 0.1875,
+      "step": 1319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001996114390595775,
+      "loss": 0.1313,
+      "step": 1320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019961020142431566,
+      "loss": 0.2061,
+      "step": 1321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019960896182499217,
+      "loss": 0.1194,
+      "step": 1322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019960772026163145,
+      "loss": 0.1257,
+      "step": 1323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00199606476734258,
+      "loss": 0.2128,
+      "step": 1324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019960523124289632,
+      "loss": 0.1771,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00199603983787571,
+      "loss": 0.1218,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019960273436830664,
+      "loss": 0.1122,
+      "step": 1327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001996014829851278,
+      "loss": 0.187,
+      "step": 1328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001996002296380592,
+      "loss": 0.1561,
+      "step": 1329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995989743271256,
+      "loss": 0.1393,
+      "step": 1330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995977170523517,
+      "loss": 0.1156,
+      "step": 1331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995964578137623,
+      "loss": 0.2944,
+      "step": 1332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019959519661138223,
+      "loss": 0.1078,
+      "step": 1333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019959393344523633,
+      "loss": 0.1194,
+      "step": 1334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995926683153496,
+      "loss": 0.1543,
+      "step": 1335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995914012217469,
+      "loss": 0.1251,
+      "step": 1336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995901321644532,
+      "loss": 0.2002,
+      "step": 1337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019958886114349356,
+      "loss": 0.1586,
+      "step": 1338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019958758815889306,
+      "loss": 0.1455,
+      "step": 1339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995863132106768,
+      "loss": 0.163,
+      "step": 1340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019958503629886985,
+      "loss": 0.1268,
+      "step": 1341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019958375742349745,
+      "loss": 0.1434,
+      "step": 1342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995824765845848,
+      "loss": 0.191,
+      "step": 1343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019958119378215714,
+      "loss": 0.16,
+      "step": 1344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995799090162398,
+      "loss": 0.1661,
+      "step": 1345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019957862228685806,
+      "loss": 0.156,
+      "step": 1346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019957733359403738,
+      "loss": 0.1515,
+      "step": 1347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019957604293780306,
+      "loss": 0.1189,
+      "step": 1348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995747503181806,
+      "loss": 0.1315,
+      "step": 1349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019957345573519553,
+      "loss": 0.1711,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995721591888733,
+      "loss": 0.1555,
+      "step": 1351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995708606792395,
+      "loss": 0.1581,
+      "step": 1352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995695602063198,
+      "loss": 0.1271,
+      "step": 1353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019956825777013975,
+      "loss": 0.1522,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019956695337072505,
+      "loss": 0.1393,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.001995656470081014,
+      "loss": 0.2124,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019956433868229466,
+      "loss": 0.1119,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019956302839333052,
+      "loss": 0.186,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019956171614123486,
+      "loss": 0.2051,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0019956040192603357,
+      "loss": 0.1805,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995590857477525,
+      "loss": 0.1458,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995577676064177,
+      "loss": 0.2063,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019955644750205503,
+      "loss": 0.1361,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019955512543469067,
+      "loss": 0.2025,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019955380140435053,
+      "loss": 0.103,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995524754110609,
+      "loss": 0.1885,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019955114745484775,
+      "loss": 0.1664,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019954981753573732,
+      "loss": 0.1832,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995484856537559,
+      "loss": 0.1201,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019954715180892967,
+      "loss": 0.2017,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00199545816001285,
+      "loss": 0.1018,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019954447823084813,
+      "loss": 0.129,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995431384976455,
+      "loss": 0.1122,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019954179680170357,
+      "loss": 0.2288,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019954045314304874,
+      "loss": 0.1208,
+      "step": 1375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019953910752170748,
+      "loss": 0.1643,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019953775993770636,
+      "loss": 0.1782,
+      "step": 1377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019953641039107198,
+      "loss": 0.1719,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019953505888183087,
+      "loss": 0.1655,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019953370541000975,
+      "loss": 0.1992,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995323499756353,
+      "loss": 0.1469,
+      "step": 1381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019953099257873416,
+      "loss": 0.1647,
+      "step": 1382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995296332193332,
+      "loss": 0.1498,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995282718974592,
+      "loss": 0.2283,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019952690861313894,
+      "loss": 0.2112,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995255433663994,
+      "loss": 0.0992,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995241761572674,
+      "loss": 0.1416,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019952280698576994,
+      "loss": 0.1449,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019952143585193404,
+      "loss": 0.1538,
+      "step": 1389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995200627557867,
+      "loss": 0.1068,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00199518687697355,
+      "loss": 0.1232,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995173106766661,
+      "loss": 0.2218,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019951593169374704,
+      "loss": 0.1099,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019951455074862516,
+      "loss": 0.1665,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995131678413276,
+      "loss": 0.1663,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995117829718816,
+      "loss": 0.2532,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019951039614031453,
+      "loss": 0.1455,
+      "step": 1397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019950900734665373,
+      "loss": 0.1637,
+      "step": 1398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995076165909265,
+      "loss": 0.1462,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019950622387316036,
+      "loss": 0.1324,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019950482919338275,
+      "loss": 0.1615,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019950343255162113,
+      "loss": 0.1078,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.001995020339479031,
+      "loss": 0.2317,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019950063338225617,
+      "loss": 0.1687,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00199499230854708,
+      "loss": 0.1738,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0019949782636528622,
+      "loss": 0.1514,
+      "step": 1406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019949641991401854,
+      "loss": 0.1526,
+      "step": 1407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994950115009327,
+      "loss": 0.1841,
+      "step": 1408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994936011260564,
+      "loss": 0.1599,
+      "step": 1409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994921887894176,
+      "loss": 0.2322,
+      "step": 1410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019949077449104395,
+      "loss": 0.1799,
+      "step": 1411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994893582309635,
+      "loss": 0.1482,
+      "step": 1412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994879400092041,
+      "loss": 0.1323,
+      "step": 1413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994865198257937,
+      "loss": 0.172,
+      "step": 1414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994850976807604,
+      "loss": 0.1224,
+      "step": 1415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994836735741321,
+      "loss": 0.1837,
+      "step": 1416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019948224750593697,
+      "loss": 0.1686,
+      "step": 1417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019948081947620314,
+      "loss": 0.1454,
+      "step": 1418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994793894849587,
+      "loss": 0.1443,
+      "step": 1419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994779575322319,
+      "loss": 0.163,
+      "step": 1420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019947652361805095,
+      "loss": 0.1411,
+      "step": 1421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019947508774244413,
+      "loss": 0.2301,
+      "step": 1422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019947364990543976,
+      "loss": 0.1138,
+      "step": 1423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994722101070662,
+      "loss": 0.1621,
+      "step": 1424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994707683473518,
+      "loss": 0.187,
+      "step": 1425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00199469324626325,
+      "loss": 0.1696,
+      "step": 1426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019946787894401428,
+      "loss": 0.1317,
+      "step": 1427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019946643130044814,
+      "loss": 0.153,
+      "step": 1428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019946498169565513,
+      "loss": 0.1313,
+      "step": 1429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994635301296638,
+      "loss": 0.1061,
+      "step": 1430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019946207660250285,
+      "loss": 0.21,
+      "step": 1431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019946062111420083,
+      "loss": 0.1409,
+      "step": 1432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994591636647866,
+      "loss": 0.1704,
+      "step": 1433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019945770425428867,
+      "loss": 0.1418,
+      "step": 1434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00199456242882736,
+      "loss": 0.1687,
+      "step": 1435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019945477955015736,
+      "loss": 0.1505,
+      "step": 1436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019945331425658156,
+      "loss": 0.2072,
+      "step": 1437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019945184700203753,
+      "loss": 0.1595,
+      "step": 1438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019945037778655423,
+      "loss": 0.2034,
+      "step": 1439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019944890661016054,
+      "loss": 0.1663,
+      "step": 1440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019944743347288553,
+      "loss": 0.1782,
+      "step": 1441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994459583747582,
+      "loss": 0.1636,
+      "step": 1442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994444813158077,
+      "loss": 0.1566,
+      "step": 1443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019944300229606314,
+      "loss": 0.2488,
+      "step": 1444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019944152131555364,
+      "loss": 0.1763,
+      "step": 1445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019944003837430845,
+      "loss": 0.1072,
+      "step": 1446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019943855347235675,
+      "loss": 0.1448,
+      "step": 1447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994370666097279,
+      "loss": 0.1211,
+      "step": 1448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019943557778645117,
+      "loss": 0.197,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994340870025559,
+      "loss": 0.1396,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994325942580715,
+      "loss": 0.1636,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0019943109955302737,
+      "loss": 0.1462,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.001994296028874531,
+      "loss": 0.1711,
+      "step": 1453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019942810426137807,
+      "loss": 0.1558,
+      "step": 1454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019942660367483185,
+      "loss": 0.1353,
+      "step": 1455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994251011278441,
+      "loss": 0.1046,
+      "step": 1456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019942359662044437,
+      "loss": 0.1226,
+      "step": 1457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994220901526624,
+      "loss": 0.1205,
+      "step": 1458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994205817245278,
+      "loss": 0.1632,
+      "step": 1459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994190713360704,
+      "loss": 0.1649,
+      "step": 1460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019941755898731993,
+      "loss": 0.1302,
+      "step": 1461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019941604467830624,
+      "loss": 0.2046,
+      "step": 1462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019941452840905914,
+      "loss": 0.1711,
+      "step": 1463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994130101796086,
+      "loss": 0.1466,
+      "step": 1464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994114899899845,
+      "loss": 0.1483,
+      "step": 1465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019940996784021683,
+      "loss": 0.1609,
+      "step": 1466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019940844373033556,
+      "loss": 0.1196,
+      "step": 1467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019940691766037084,
+      "loss": 0.1323,
+      "step": 1468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994053896303527,
+      "loss": 0.09,
+      "step": 1469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019940385964031123,
+      "loss": 0.1208,
+      "step": 1470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994023276902767,
+      "loss": 0.1566,
+      "step": 1471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001994007937802792,
+      "loss": 0.1643,
+      "step": 1472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019939925791034904,
+      "loss": 0.2294,
+      "step": 1473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993977200805165,
+      "loss": 0.1746,
+      "step": 1474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993961802908119,
+      "loss": 0.1877,
+      "step": 1475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993946385412656,
+      "loss": 0.1477,
+      "step": 1476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00199393094831908,
+      "loss": 0.2466,
+      "step": 1477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019939154916276953,
+      "loss": 0.2094,
+      "step": 1478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993900015338807,
+      "loss": 0.2218,
+      "step": 1479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019938845194527196,
+      "loss": 0.1085,
+      "step": 1480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019938690039697387,
+      "loss": 0.1533,
+      "step": 1481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019938534688901713,
+      "loss": 0.1555,
+      "step": 1482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019938379142143228,
+      "loss": 0.1556,
+      "step": 1483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019938223399424996,
+      "loss": 0.188,
+      "step": 1484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019938067460750094,
+      "loss": 0.1721,
+      "step": 1485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00199379113261216,
+      "loss": 0.1406,
+      "step": 1486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993775499554258,
+      "loss": 0.1396,
+      "step": 1487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019937598469016125,
+      "loss": 0.1278,
+      "step": 1488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019937441746545326,
+      "loss": 0.1161,
+      "step": 1489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019937284828133264,
+      "loss": 0.2102,
+      "step": 1490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993712771378304,
+      "loss": 0.197,
+      "step": 1491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019936970403497743,
+      "loss": 0.1497,
+      "step": 1492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993681289728048,
+      "loss": 0.1135,
+      "step": 1493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019936655195134365,
+      "loss": 0.2112,
+      "step": 1494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993649729706249,
+      "loss": 0.1525,
+      "step": 1495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019936339203067985,
+      "loss": 0.1349,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019936180913153953,
+      "loss": 0.1859,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.001993602242732353,
+      "loss": 0.1609,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0019935863745579826,
+      "loss": 0.0818,
+      "step": 1499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993570486792598,
+      "loss": 0.1417,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993554579436512,
+      "loss": 0.2197,
+      "step": 1501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019935386524900386,
+      "loss": 0.125,
+      "step": 1502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993522705953491,
+      "loss": 0.0933,
+      "step": 1503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993506739827185,
+      "loss": 0.141,
+      "step": 1504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993490754111435,
+      "loss": 0.1327,
+      "step": 1505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993474748806555,
+      "loss": 0.1488,
+      "step": 1506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019934587239128615,
+      "loss": 0.1375,
+      "step": 1507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993442679430671,
+      "loss": 0.2368,
+      "step": 1508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019934266153602987,
+      "loss": 0.1666,
+      "step": 1509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019934105317020623,
+      "loss": 0.1687,
+      "step": 1510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019933944284562786,
+      "loss": 0.1791,
+      "step": 1511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019933783056232646,
+      "loss": 0.1692,
+      "step": 1512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993362163203339,
+      "loss": 0.1086,
+      "step": 1513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019933460011968196,
+      "loss": 0.1549,
+      "step": 1514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993329819604025,
+      "loss": 0.1107,
+      "step": 1515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993313618425275,
+      "loss": 0.1616,
+      "step": 1516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019932973976608884,
+      "loss": 0.1635,
+      "step": 1517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019932811573111857,
+      "loss": 0.2227,
+      "step": 1518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019932648973764854,
+      "loss": 0.155,
+      "step": 1519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019932486178571105,
+      "loss": 0.1525,
+      "step": 1520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00199323231875338,
+      "loss": 0.3545,
+      "step": 1521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019932160000656163,
+      "loss": 0.2219,
+      "step": 1522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993199661794141,
+      "loss": 0.1667,
+      "step": 1523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019931833039392756,
+      "loss": 0.124,
+      "step": 1524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019931669265013437,
+      "loss": 0.1189,
+      "step": 1525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019931505294806676,
+      "loss": 0.2139,
+      "step": 1526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993134112877571,
+      "loss": 0.1615,
+      "step": 1527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019931176766923775,
+      "loss": 0.1028,
+      "step": 1528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019931012209254105,
+      "loss": 0.1428,
+      "step": 1529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993084745576995,
+      "loss": 0.1548,
+      "step": 1530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993068250647456,
+      "loss": 0.1067,
+      "step": 1531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019930517361371187,
+      "loss": 0.2043,
+      "step": 1532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019930352020463087,
+      "loss": 0.167,
+      "step": 1533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019930186483753513,
+      "loss": 0.1436,
+      "step": 1534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001993002075124574,
+      "loss": 0.2195,
+      "step": 1535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001992985482294303,
+      "loss": 0.1674,
+      "step": 1536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019929688698848654,
+      "loss": 0.1185,
+      "step": 1537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001992952237896589,
+      "loss": 0.1481,
+      "step": 1538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001992935586329802,
+      "loss": 0.1221,
+      "step": 1539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019929189151848318,
+      "loss": 0.2156,
+      "step": 1540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001992902224462008,
+      "loss": 0.1261,
+      "step": 1541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019928855141616592,
+      "loss": 0.188,
+      "step": 1542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001992868784284115,
+      "loss": 0.0881,
+      "step": 1543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0019928520348297058,
+      "loss": 0.182,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.001992835265798761,
+      "loss": 0.2195,
+      "step": 1545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019928184771916117,
+      "loss": 0.1057,
+      "step": 1546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992801669008589,
+      "loss": 0.1953,
+      "step": 1547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992784841250024,
+      "loss": 0.1257,
+      "step": 1548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992767993916249,
+      "loss": 0.1078,
+      "step": 1549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992751127007596,
+      "loss": 0.1208,
+      "step": 1550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019927342405243968,
+      "loss": 0.0979,
+      "step": 1551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019927173344669855,
+      "loss": 0.1388,
+      "step": 1552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019927004088356945,
+      "loss": 0.2039,
+      "step": 1553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019926834636308584,
+      "loss": 0.1489,
+      "step": 1554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019926664988528108,
+      "loss": 0.1484,
+      "step": 1555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992649514501886,
+      "loss": 0.1423,
+      "step": 1556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019926325105784196,
+      "loss": 0.1735,
+      "step": 1557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019926154870827466,
+      "loss": 0.17,
+      "step": 1558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992598444015202,
+      "loss": 0.1232,
+      "step": 1559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019925813813761227,
+      "loss": 0.1222,
+      "step": 1560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992564299165845,
+      "loss": 0.1374,
+      "step": 1561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992547197384705,
+      "loss": 0.2324,
+      "step": 1562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019925300760330407,
+      "loss": 0.1449,
+      "step": 1563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992512935111189,
+      "loss": 0.1208,
+      "step": 1564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992495774619489,
+      "loss": 0.1364,
+      "step": 1565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992478594558278,
+      "loss": 0.1664,
+      "step": 1566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019924613949278947,
+      "loss": 0.1504,
+      "step": 1567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019924441757286793,
+      "loss": 0.1174,
+      "step": 1568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019924269369609704,
+      "loss": 0.2126,
+      "step": 1569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992409678625108,
+      "loss": 0.1658,
+      "step": 1570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992392400721433,
+      "loss": 0.1511,
+      "step": 1571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019923751032502853,
+      "loss": 0.14,
+      "step": 1572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019923577862120062,
+      "loss": 0.0883,
+      "step": 1573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019923404496069376,
+      "loss": 0.1788,
+      "step": 1574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019923230934354207,
+      "loss": 0.152,
+      "step": 1575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019923057176977977,
+      "loss": 0.1073,
+      "step": 1576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992288322394412,
+      "loss": 0.1504,
+      "step": 1577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992270907525606,
+      "loss": 0.1333,
+      "step": 1578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992253473091723,
+      "loss": 0.1467,
+      "step": 1579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992236019093107,
+      "loss": 0.101,
+      "step": 1580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992218545530102,
+      "loss": 0.1201,
+      "step": 1581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019922010524030527,
+      "loss": 0.1635,
+      "step": 1582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019921835397123037,
+      "loss": 0.1273,
+      "step": 1583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992166007458201,
+      "loss": 0.1705,
+      "step": 1584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992148455641089,
+      "loss": 0.2112,
+      "step": 1585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992130884261315,
+      "loss": 0.085,
+      "step": 1586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019921132933192255,
+      "loss": 0.1481,
+      "step": 1587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019920956828151665,
+      "loss": 0.1156,
+      "step": 1588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0019920780527494858,
+      "loss": 0.1278,
+      "step": 1589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.001992060403122531,
+      "loss": 0.1577,
+      "step": 1590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00199204273393465,
+      "loss": 0.1034,
+      "step": 1591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001992025045186191,
+      "loss": 0.1428,
+      "step": 1592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001992007336877503,
+      "loss": 0.157,
+      "step": 1593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019919896090089353,
+      "loss": 0.1526,
+      "step": 1594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019919718615808366,
+      "loss": 0.1521,
+      "step": 1595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991954094593558,
+      "loss": 0.1442,
+      "step": 1596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019919363080474492,
+      "loss": 0.149,
+      "step": 1597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991918501942861,
+      "loss": 0.1113,
+      "step": 1598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019919006762801447,
+      "loss": 0.147,
+      "step": 1599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991882831059651,
+      "loss": 0.1334,
+      "step": 1600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019918649662817333,
+      "loss": 0.1372,
+      "step": 1601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991847081946742,
+      "loss": 0.1323,
+      "step": 1602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019918291780550313,
+      "loss": 0.1506,
+      "step": 1603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991811254606953,
+      "loss": 0.142,
+      "step": 1604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019917933116028615,
+      "loss": 0.1625,
+      "step": 1605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019917753490431103,
+      "loss": 0.106,
+      "step": 1606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019917573669280533,
+      "loss": 0.1689,
+      "step": 1607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019917393652580445,
+      "loss": 0.1241,
+      "step": 1608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00199172134403344,
+      "loss": 0.1035,
+      "step": 1609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991703303254595,
+      "loss": 0.1597,
+      "step": 1610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019916852429218645,
+      "loss": 0.1101,
+      "step": 1611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019916671630356053,
+      "loss": 0.1101,
+      "step": 1612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991649063596173,
+      "loss": 0.1287,
+      "step": 1613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019916309446039258,
+      "loss": 0.1127,
+      "step": 1614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019916128060592196,
+      "loss": 0.1647,
+      "step": 1615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019915946479624126,
+      "loss": 0.151,
+      "step": 1616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991576470313863,
+      "loss": 0.1488,
+      "step": 1617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019915582731139295,
+      "loss": 0.0989,
+      "step": 1618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019915400563629703,
+      "loss": 0.181,
+      "step": 1619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991521820061345,
+      "loss": 0.1765,
+      "step": 1620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019915035642094126,
+      "loss": 0.166,
+      "step": 1621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019914852888075337,
+      "loss": 0.1489,
+      "step": 1622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019914669938560685,
+      "loss": 0.1467,
+      "step": 1623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019914486793553773,
+      "loss": 0.118,
+      "step": 1624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019914303453058215,
+      "loss": 0.2026,
+      "step": 1625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991411991707763,
+      "loss": 0.1951,
+      "step": 1626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991393618561563,
+      "loss": 0.1252,
+      "step": 1627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019913752258675843,
+      "loss": 0.1371,
+      "step": 1628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991356813626189,
+      "loss": 0.1559,
+      "step": 1629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019913383818377407,
+      "loss": 0.1958,
+      "step": 1630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.001991319930502603,
+      "loss": 0.1412,
+      "step": 1631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019913014596211385,
+      "loss": 0.1549,
+      "step": 1632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019912829691937127,
+      "loss": 0.1006,
+      "step": 1633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019912644592206897,
+      "loss": 0.2103,
+      "step": 1634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019912459297024344,
+      "loss": 0.1467,
+      "step": 1635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019912273806393125,
+      "loss": 0.1096,
+      "step": 1636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0019912088120316892,
+      "loss": 0.0999,
+      "step": 1637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001991190223879931,
+      "loss": 0.0963,
+      "step": 1638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001991171616184404,
+      "loss": 0.162,
+      "step": 1639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019911529889454762,
+      "loss": 0.101,
+      "step": 1640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001991134342163513,
+      "loss": 0.1176,
+      "step": 1641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019911156758388843,
+      "loss": 0.1486,
+      "step": 1642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019910969899719563,
+      "loss": 0.1927,
+      "step": 1643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001991078284563098,
+      "loss": 0.165,
+      "step": 1644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001991059559612679,
+      "loss": 0.1356,
+      "step": 1645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019910408151210673,
+      "loss": 0.2139,
+      "step": 1646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001991022051088634,
+      "loss": 0.2378,
+      "step": 1647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019910032675157473,
+      "loss": 0.1086,
+      "step": 1648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019909844644027787,
+      "loss": 0.1028,
+      "step": 1649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019909656417500987,
+      "loss": 0.1804,
+      "step": 1650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990946799558078,
+      "loss": 0.1877,
+      "step": 1651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990927937827089,
+      "loss": 0.1467,
+      "step": 1652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990909056557503,
+      "loss": 0.1264,
+      "step": 1653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019908901557496926,
+      "loss": 0.2327,
+      "step": 1654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00199087123540403,
+      "loss": 0.1949,
+      "step": 1655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019908522955208886,
+      "loss": 0.1479,
+      "step": 1656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019908333361006417,
+      "loss": 0.1532,
+      "step": 1657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019908143571436636,
+      "loss": 0.132,
+      "step": 1658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990795358650328,
+      "loss": 0.1107,
+      "step": 1659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019907763406210094,
+      "loss": 0.1699,
+      "step": 1660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990757303056084,
+      "loss": 0.1119,
+      "step": 1661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990738245955925,
+      "loss": 0.1101,
+      "step": 1662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019907191693209103,
+      "loss": 0.1194,
+      "step": 1663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019907000731514144,
+      "loss": 0.2524,
+      "step": 1664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990680957447815,
+      "loss": 0.1104,
+      "step": 1665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019906618222104884,
+      "loss": 0.2432,
+      "step": 1666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990642667439812,
+      "loss": 0.1221,
+      "step": 1667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990623493136164,
+      "loss": 0.1516,
+      "step": 1668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990604299299921,
+      "loss": 0.1246,
+      "step": 1669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019905850859314638,
+      "loss": 0.1709,
+      "step": 1670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990565853031169,
+      "loss": 0.1359,
+      "step": 1671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990546600599417,
+      "loss": 0.1765,
+      "step": 1672
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990527328636587,
+      "loss": 0.118,
+      "step": 1673
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019905080371430592,
+      "loss": 0.1906,
+      "step": 1674
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990488726119214,
+      "loss": 0.0809,
+      "step": 1675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990469395565432,
+      "loss": 0.175,
+      "step": 1676
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019904500454820946,
+      "loss": 0.1449,
+      "step": 1677
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.001990430675869583,
+      "loss": 0.0846,
+      "step": 1678
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019904112867282792,
+      "loss": 0.1349,
+      "step": 1679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019903918780585657,
+      "loss": 0.1472,
+      "step": 1680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019903724498608255,
+      "loss": 0.1721,
+      "step": 1681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019903530021354405,
+      "loss": 0.167,
+      "step": 1682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0019903335348827956,
+      "loss": 0.1169,
+      "step": 1683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019903140481032735,
+      "loss": 0.2085,
+      "step": 1684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019902945417972586,
+      "loss": 0.1565,
+      "step": 1685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019902750159651364,
+      "loss": 0.1758,
+      "step": 1686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019902554706072912,
+      "loss": 0.1764,
+      "step": 1687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019902359057241085,
+      "loss": 0.1654,
+      "step": 1688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019902163213159738,
+      "loss": 0.1675,
+      "step": 1689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001990196717383274,
+      "loss": 0.1899,
+      "step": 1690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019901770939263945,
+      "loss": 0.2107,
+      "step": 1691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019901574509457232,
+      "loss": 0.1605,
+      "step": 1692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019901377884416474,
+      "loss": 0.2092,
+      "step": 1693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019901181064145537,
+      "loss": 0.1736,
+      "step": 1694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001990098404864832,
+      "loss": 0.1444,
+      "step": 1695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001990078683792869,
+      "loss": 0.1831,
+      "step": 1696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019900589431990543,
+      "loss": 0.1541,
+      "step": 1697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019900391830837776,
+      "loss": 0.1011,
+      "step": 1698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019900194034474274,
+      "loss": 0.1814,
+      "step": 1699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019899996042903947,
+      "loss": 0.1711,
+      "step": 1700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019899797856130696,
+      "loss": 0.2407,
+      "step": 1701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019899599474158423,
+      "loss": 0.1284,
+      "step": 1702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019899400896991053,
+      "loss": 0.1973,
+      "step": 1703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019899202124632484,
+      "loss": 0.1449,
+      "step": 1704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989900315708665,
+      "loss": 0.2141,
+      "step": 1705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989880399435747,
+      "loss": 0.2502,
+      "step": 1706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019898604636448864,
+      "loss": 0.1584,
+      "step": 1707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989840508336477,
+      "loss": 0.1788,
+      "step": 1708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019898205335109123,
+      "loss": 0.1482,
+      "step": 1709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989800539168586,
+      "loss": 0.1643,
+      "step": 1710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989780525309892,
+      "loss": 0.1655,
+      "step": 1711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019897604919352253,
+      "loss": 0.168,
+      "step": 1712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989740439044981,
+      "loss": 0.1077,
+      "step": 1713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989720366639554,
+      "loss": 0.1132,
+      "step": 1714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019897002747193407,
+      "loss": 0.1519,
+      "step": 1715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019896801632847366,
+      "loss": 0.2161,
+      "step": 1716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019896600323361393,
+      "loss": 0.113,
+      "step": 1717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019896398818739446,
+      "loss": 0.1616,
+      "step": 1718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00198961971189855,
+      "loss": 0.0863,
+      "step": 1719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019895995224103536,
+      "loss": 0.1724,
+      "step": 1720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019895793134097534,
+      "loss": 0.1058,
+      "step": 1721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989559084897148,
+      "loss": 0.0731,
+      "step": 1722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019895388368729354,
+      "loss": 0.0894,
+      "step": 1723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019895185693375164,
+      "loss": 0.1702,
+      "step": 1724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989498282291289,
+      "loss": 0.1035,
+      "step": 1725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989477975734654,
+      "loss": 0.103,
+      "step": 1726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.001989457649668012,
+      "loss": 0.1655,
+      "step": 1727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019894373040917636,
+      "loss": 0.1655,
+      "step": 1728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0019894169390063097,
+      "loss": 0.1415,
+      "step": 1729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989396554412052,
+      "loss": 0.1166,
+      "step": 1730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019893761503093923,
+      "loss": 0.1687,
+      "step": 1731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019893557266987332,
+      "loss": 0.1503,
+      "step": 1732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019893352835804773,
+      "loss": 0.1472,
+      "step": 1733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019893148209550278,
+      "loss": 0.1318,
+      "step": 1734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989294338822788,
+      "loss": 0.1943,
+      "step": 1735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019892738371841616,
+      "loss": 0.1189,
+      "step": 1736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989253316039553,
+      "loss": 0.1451,
+      "step": 1737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019892327753893665,
+      "loss": 0.093,
+      "step": 1738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019892122152340083,
+      "loss": 0.1271,
+      "step": 1739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019891916355738825,
+      "loss": 0.1572,
+      "step": 1740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989171036409395,
+      "loss": 0.1266,
+      "step": 1741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989150417740953,
+      "loss": 0.1677,
+      "step": 1742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989129779568962,
+      "loss": 0.1357,
+      "step": 1743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989109121893829,
+      "loss": 0.1858,
+      "step": 1744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989088444715962,
+      "loss": 0.1464,
+      "step": 1745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989067748035768,
+      "loss": 0.1113,
+      "step": 1746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019890470318536554,
+      "loss": 0.129,
+      "step": 1747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989026296170033,
+      "loss": 0.1359,
+      "step": 1748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001989005540985309,
+      "loss": 0.1206,
+      "step": 1749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988984766299893,
+      "loss": 0.099,
+      "step": 1750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019889639721141945,
+      "loss": 0.1555,
+      "step": 1751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988943158428624,
+      "loss": 0.1113,
+      "step": 1752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988922325243591,
+      "loss": 0.2303,
+      "step": 1753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988901472559507,
+      "loss": 0.1685,
+      "step": 1754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019888806003767827,
+      "loss": 0.1591,
+      "step": 1755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00198885970869583,
+      "loss": 0.1324,
+      "step": 1756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988838797517061,
+      "loss": 0.1917,
+      "step": 1757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019888178668408868,
+      "loss": 0.1385,
+      "step": 1758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988796916667722,
+      "loss": 0.1249,
+      "step": 1759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988775946997978,
+      "loss": 0.158,
+      "step": 1760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019887549578320694,
+      "loss": 0.2391,
+      "step": 1761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019887339491704093,
+      "loss": 0.1176,
+      "step": 1762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019887129210134124,
+      "loss": 0.1467,
+      "step": 1763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019886918733614933,
+      "loss": 0.2065,
+      "step": 1764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019886708062150666,
+      "loss": 0.2178,
+      "step": 1765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019886497195745483,
+      "loss": 0.1362,
+      "step": 1766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019886286134403533,
+      "loss": 0.1119,
+      "step": 1767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988607487812899,
+      "loss": 0.1874,
+      "step": 1768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988586342692601,
+      "loss": 0.2074,
+      "step": 1769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019885651780798762,
+      "loss": 0.2144,
+      "step": 1770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019885439939751424,
+      "loss": 0.1437,
+      "step": 1771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988522790378817,
+      "loss": 0.127,
+      "step": 1772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019885015672913182,
+      "loss": 0.1384,
+      "step": 1773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019884803247130645,
+      "loss": 0.1821,
+      "step": 1774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.001988459062644475,
+      "loss": 0.1595,
+      "step": 1775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0019884377810859676,
+      "loss": 0.1925,
+      "step": 1776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988416480037964,
+      "loss": 0.1602,
+      "step": 1777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019883951595008826,
+      "loss": 0.2141,
+      "step": 1778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019883738194751443,
+      "loss": 0.1255,
+      "step": 1779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019883524599611704,
+      "loss": 0.1224,
+      "step": 1780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988331080959381,
+      "loss": 0.1088,
+      "step": 1781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019883096824701986,
+      "loss": 0.1089,
+      "step": 1782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019882882644940447,
+      "loss": 0.1804,
+      "step": 1783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019882668270313412,
+      "loss": 0.1816,
+      "step": 1784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019882453700825115,
+      "loss": 0.1139,
+      "step": 1785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019882238936479784,
+      "loss": 0.1447,
+      "step": 1786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988202397728165,
+      "loss": 0.1794,
+      "step": 1787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019881808823234964,
+      "loss": 0.2318,
+      "step": 1788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019881593474343952,
+      "loss": 0.0922,
+      "step": 1789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988137793061287,
+      "loss": 0.1356,
+      "step": 1790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019881162192045966,
+      "loss": 0.1743,
+      "step": 1791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988094625864749,
+      "loss": 0.1423,
+      "step": 1792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019880730130421706,
+      "loss": 0.0887,
+      "step": 1793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988051380737287,
+      "loss": 0.1543,
+      "step": 1794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019880297289505255,
+      "loss": 0.1627,
+      "step": 1795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001988008057682312,
+      "loss": 0.1382,
+      "step": 1796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019879863669330744,
+      "loss": 0.1266,
+      "step": 1797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00198796465670324,
+      "loss": 0.2644,
+      "step": 1798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019879429269932377,
+      "loss": 0.1791,
+      "step": 1799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987921177803495,
+      "loss": 0.1718,
+      "step": 1800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987899409134441,
+      "loss": 0.1665,
+      "step": 1801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019878776209865056,
+      "loss": 0.2061,
+      "step": 1802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019878558133601177,
+      "loss": 0.1691,
+      "step": 1803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019878339862557074,
+      "loss": 0.1534,
+      "step": 1804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987812139673705,
+      "loss": 0.1277,
+      "step": 1805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019877902736145417,
+      "loss": 0.1398,
+      "step": 1806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987768388078648,
+      "loss": 0.1378,
+      "step": 1807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987746483066456,
+      "loss": 0.1523,
+      "step": 1808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019877245585783977,
+      "loss": 0.1626,
+      "step": 1809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987702614614905,
+      "loss": 0.2032,
+      "step": 1810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019876806511764105,
+      "loss": 0.1361,
+      "step": 1811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019876586682633472,
+      "loss": 0.1006,
+      "step": 1812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987636665876149,
+      "loss": 0.2034,
+      "step": 1813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019876146440152495,
+      "loss": 0.178,
+      "step": 1814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987592602681083,
+      "loss": 0.1689,
+      "step": 1815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019875705418740835,
+      "loss": 0.2128,
+      "step": 1816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987548461594687,
+      "loss": 0.147,
+      "step": 1817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987526361843328,
+      "loss": 0.1188,
+      "step": 1818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019875042426204434,
+      "loss": 0.1973,
+      "step": 1819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0019874821039264677,
+      "loss": 0.231,
+      "step": 1820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987459945761839,
+      "loss": 0.2715,
+      "step": 1821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.001987437768126993,
+      "loss": 0.1385,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019874155710223676,
+      "loss": 0.1926,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019873933544484004,
+      "loss": 0.1672,
+      "step": 1824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019873711184055292,
+      "loss": 0.1586,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001987348862894193,
+      "loss": 0.2037,
+      "step": 1826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00198732658791483,
+      "loss": 0.1351,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019873042934678794,
+      "loss": 0.1801,
+      "step": 1828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019872819795537814,
+      "loss": 0.1538,
+      "step": 1829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019872596461729757,
+      "loss": 0.1465,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019872372933259024,
+      "loss": 0.2244,
+      "step": 1831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001987214921013002,
+      "loss": 0.2332,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001987192529234717,
+      "loss": 0.1031,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001987170117991487,
+      "loss": 0.1564,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019871476872837554,
+      "loss": 0.1228,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001987125237111964,
+      "loss": 0.1852,
+      "step": 1836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019871027674765555,
+      "loss": 0.135,
+      "step": 1837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019870802783779726,
+      "loss": 0.1398,
+      "step": 1838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019870577698166587,
+      "loss": 0.1057,
+      "step": 1839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001987035241793058,
+      "loss": 0.1564,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019870126943076147,
+      "loss": 0.1199,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019869901273607735,
+      "loss": 0.1777,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019869675409529787,
+      "loss": 0.1443,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019869449350846764,
+      "loss": 0.1201,
+      "step": 1844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019869223097563117,
+      "loss": 0.0956,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019868996649683307,
+      "loss": 0.1495,
+      "step": 1846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986877000721181,
+      "loss": 0.1088,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019868543170153082,
+      "loss": 0.1206,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00198683161385116,
+      "loss": 0.1763,
+      "step": 1849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019868088912291843,
+      "loss": 0.1299,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019867861491498287,
+      "loss": 0.0681,
+      "step": 1851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019867633876135423,
+      "loss": 0.161,
+      "step": 1852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986740606620773,
+      "loss": 0.1399,
+      "step": 1853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019867178061719706,
+      "loss": 0.1163,
+      "step": 1854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019866949862675847,
+      "loss": 0.2073,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986672146908065,
+      "loss": 0.1221,
+      "step": 1856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019866492880938618,
+      "loss": 0.0674,
+      "step": 1857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019866264098254258,
+      "loss": 0.117,
+      "step": 1858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986603512103208,
+      "loss": 0.1404,
+      "step": 1859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019865805949276605,
+      "loss": 0.1392,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019865576582992345,
+      "loss": 0.1316,
+      "step": 1861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019865347022183827,
+      "loss": 0.1198,
+      "step": 1862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986511726685557,
+      "loss": 0.234,
+      "step": 1863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019864887317012116,
+      "loss": 0.0885,
+      "step": 1864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986465717265799,
+      "loss": 0.0999,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019864426833797735,
+      "loss": 0.1804,
+      "step": 1866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.001986419630043589,
+      "loss": 0.1436,
+      "step": 1867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0019863965572577,
+      "loss": 0.124,
+      "step": 1868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019863734650225613,
+      "loss": 0.176,
+      "step": 1869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019863503533386288,
+      "loss": 0.1415,
+      "step": 1870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019863272222063575,
+      "loss": 0.1122,
+      "step": 1871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986304071626204,
+      "loss": 0.1387,
+      "step": 1872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019862809015986247,
+      "loss": 0.1326,
+      "step": 1873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019862577121240762,
+      "loss": 0.1215,
+      "step": 1874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986234503203016,
+      "loss": 0.132,
+      "step": 1875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019862112748359017,
+      "loss": 0.1663,
+      "step": 1876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986188027023191,
+      "loss": 0.1477,
+      "step": 1877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019861647597653426,
+      "loss": 0.1567,
+      "step": 1878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986141473062815,
+      "loss": 0.1608,
+      "step": 1879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019861181669160676,
+      "loss": 0.1646,
+      "step": 1880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00198609484132556,
+      "loss": 0.1586,
+      "step": 1881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986071496291752,
+      "loss": 0.093,
+      "step": 1882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986048131815104,
+      "loss": 0.171,
+      "step": 1883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001986024747896076,
+      "loss": 0.1282,
+      "step": 1884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019860013445351307,
+      "loss": 0.0927,
+      "step": 1885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985977921732728,
+      "loss": 0.1279,
+      "step": 1886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00198595447948933,
+      "loss": 0.193,
+      "step": 1887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019859310178053993,
+      "loss": 0.134,
+      "step": 1888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019859075366813984,
+      "loss": 0.1736,
+      "step": 1889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00198588403611779,
+      "loss": 0.1165,
+      "step": 1890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985860516115038,
+      "loss": 0.1332,
+      "step": 1891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985836976673606,
+      "loss": 0.1353,
+      "step": 1892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019858134177939578,
+      "loss": 0.1429,
+      "step": 1893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019857898394765585,
+      "loss": 0.129,
+      "step": 1894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985766241721873,
+      "loss": 0.1782,
+      "step": 1895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019857426245303653,
+      "loss": 0.1521,
+      "step": 1896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019857189879025025,
+      "loss": 0.1692,
+      "step": 1897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00198569533183875,
+      "loss": 0.2109,
+      "step": 1898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019856716563395748,
+      "loss": 0.1395,
+      "step": 1899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985647961405443,
+      "loss": 0.1492,
+      "step": 1900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985624247036822,
+      "loss": 0.147,
+      "step": 1901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00198560051323418,
+      "loss": 0.1061,
+      "step": 1902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985576759997984,
+      "loss": 0.1664,
+      "step": 1903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985552987328703,
+      "loss": 0.1501,
+      "step": 1904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985529195226806,
+      "loss": 0.239,
+      "step": 1905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985505383692761,
+      "loss": 0.1172,
+      "step": 1906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019854815527270385,
+      "loss": 0.1837,
+      "step": 1907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019854577023301085,
+      "loss": 0.1472,
+      "step": 1908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019854338325024402,
+      "loss": 0.1163,
+      "step": 1909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019854099432445056,
+      "loss": 0.1205,
+      "step": 1910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019853860345567746,
+      "loss": 0.1202,
+      "step": 1911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985362106439719,
+      "loss": 0.198,
+      "step": 1912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.001985338158893811,
+      "loss": 0.1532,
+      "step": 1913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0019853141919195226,
+      "loss": 0.1536,
+      "step": 1914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019852902055173256,
+      "loss": 0.1494,
+      "step": 1915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019852661996876938,
+      "loss": 0.1658,
+      "step": 1916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019852421744311005,
+      "loss": 0.1176,
+      "step": 1917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001985218129748019,
+      "loss": 0.116,
+      "step": 1918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001985194065638924,
+      "loss": 0.1608,
+      "step": 1919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019851699821042896,
+      "loss": 0.1892,
+      "step": 1920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019851458791445903,
+      "loss": 0.1924,
+      "step": 1921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001985121756760302,
+      "loss": 0.1428,
+      "step": 1922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019850976149519,
+      "loss": 0.1467,
+      "step": 1923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019850734537198603,
+      "loss": 0.1259,
+      "step": 1924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019850492730646594,
+      "loss": 0.1366,
+      "step": 1925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019850250729867742,
+      "loss": 0.1274,
+      "step": 1926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019850008534866817,
+      "loss": 0.1624,
+      "step": 1927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019849766145648593,
+      "loss": 0.1316,
+      "step": 1928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019849523562217855,
+      "loss": 0.1606,
+      "step": 1929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019849280784579376,
+      "loss": 0.0698,
+      "step": 1930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019849037812737958,
+      "loss": 0.1606,
+      "step": 1931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019848794646698375,
+      "loss": 0.1492,
+      "step": 1932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984855128646543,
+      "loss": 0.136,
+      "step": 1933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019848307732043924,
+      "loss": 0.0923,
+      "step": 1934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019848063983438653,
+      "loss": 0.115,
+      "step": 1935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984782004065443,
+      "loss": 0.1542,
+      "step": 1936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984757590369606,
+      "loss": 0.1503,
+      "step": 1937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984733157256836,
+      "loss": 0.0984,
+      "step": 1938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984708704727614,
+      "loss": 0.1548,
+      "step": 1939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984684232782423,
+      "loss": 0.144,
+      "step": 1940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019846597414217453,
+      "loss": 0.1742,
+      "step": 1941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019846352306460637,
+      "loss": 0.0775,
+      "step": 1942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019846107004558616,
+      "loss": 0.1718,
+      "step": 1943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019845861508516224,
+      "loss": 0.1765,
+      "step": 1944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984561581833831,
+      "loss": 0.1589,
+      "step": 1945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019845369934029704,
+      "loss": 0.0952,
+      "step": 1946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019845123855595265,
+      "loss": 0.103,
+      "step": 1947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019844877583039843,
+      "loss": 0.0949,
+      "step": 1948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019844631116368297,
+      "loss": 0.1626,
+      "step": 1949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019844384455585473,
+      "loss": 0.1399,
+      "step": 1950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984413760069625,
+      "loss": 0.0742,
+      "step": 1951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984389055170549,
+      "loss": 0.1702,
+      "step": 1952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984364330861807,
+      "loss": 0.1576,
+      "step": 1953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019843395871438846,
+      "loss": 0.1042,
+      "step": 1954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019843148240172716,
+      "loss": 0.1237,
+      "step": 1955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.001984290041482456,
+      "loss": 0.1338,
+      "step": 1956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019842652395399255,
+      "loss": 0.1162,
+      "step": 1957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00198424041819017,
+      "loss": 0.0756,
+      "step": 1958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019842155774336787,
+      "loss": 0.1436,
+      "step": 1959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0019841907172709413,
+      "loss": 0.1346,
+      "step": 1960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019841658377024477,
+      "loss": 0.1104,
+      "step": 1961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001984140938728689,
+      "loss": 0.0963,
+      "step": 1962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001984116020350156,
+      "loss": 0.1155,
+      "step": 1963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00198409108256734,
+      "loss": 0.1653,
+      "step": 1964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019840661253807327,
+      "loss": 0.134,
+      "step": 1965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019840411487908258,
+      "loss": 0.1471,
+      "step": 1966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019840161527981126,
+      "loss": 0.1636,
+      "step": 1967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983991137403085,
+      "loss": 0.0856,
+      "step": 1968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983966102606237,
+      "loss": 0.173,
+      "step": 1969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019839410484080618,
+      "loss": 0.1093,
+      "step": 1970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019839159748090535,
+      "loss": 0.1375,
+      "step": 1971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019838908818097067,
+      "loss": 0.1311,
+      "step": 1972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019838657694105162,
+      "loss": 0.1013,
+      "step": 1973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983840637611977,
+      "loss": 0.1851,
+      "step": 1974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019838154864145843,
+      "loss": 0.1364,
+      "step": 1975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019837903158188342,
+      "loss": 0.1221,
+      "step": 1976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019837651258252233,
+      "loss": 0.1678,
+      "step": 1977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983739916434248,
+      "loss": 0.2234,
+      "step": 1978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019837146876464057,
+      "loss": 0.1296,
+      "step": 1979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019836894394621935,
+      "loss": 0.1371,
+      "step": 1980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019836641718821092,
+      "loss": 0.1443,
+      "step": 1981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019836388849066512,
+      "loss": 0.1315,
+      "step": 1982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983613578536318,
+      "loss": 0.1536,
+      "step": 1983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983588252771609,
+      "loss": 0.1106,
+      "step": 1984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983562907613023,
+      "loss": 0.1136,
+      "step": 1985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019835375430610597,
+      "loss": 0.0804,
+      "step": 1986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019835121591162195,
+      "loss": 0.0884,
+      "step": 1987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983486755779003,
+      "loss": 0.1567,
+      "step": 1988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019834613330499107,
+      "loss": 0.1577,
+      "step": 1989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019834358909294448,
+      "loss": 0.1136,
+      "step": 1990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983410429418106,
+      "loss": 0.1243,
+      "step": 1991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983384948516396,
+      "loss": 0.1594,
+      "step": 1992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983359448224819,
+      "loss": 0.1592,
+      "step": 1993
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983333928543876,
+      "loss": 0.0768,
+      "step": 1994
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019833083894740706,
+      "loss": 0.1433,
+      "step": 1995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019832828310159065,
+      "loss": 0.1365,
+      "step": 1996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983257253169888,
+      "loss": 0.129,
+      "step": 1997
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019832316559365193,
+      "loss": 0.1123,
+      "step": 1998
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983206039316305,
+      "loss": 0.1453,
+      "step": 1999
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019831804033097495,
+      "loss": 0.142,
+      "step": 2000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019831547479173595,
+      "loss": 0.1997,
+      "step": 2001
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019831290731396403,
+      "loss": 0.1543,
+      "step": 2002
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019831033789770984,
+      "loss": 0.1715,
+      "step": 2003
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00198307766543024,
+      "loss": 0.1571,
+      "step": 2004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.001983051932499572,
+      "loss": 0.1251,
+      "step": 2005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0019830261801856023,
+      "loss": 0.132,
+      "step": 2006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019830004084888386,
+      "loss": 0.0939,
+      "step": 2007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982974617409789,
+      "loss": 0.1515,
+      "step": 2008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982948806948961,
+      "loss": 0.1454,
+      "step": 2009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019829229771068653,
+      "loss": 0.2258,
+      "step": 2010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00198289712788401,
+      "loss": 0.1514,
+      "step": 2011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019828712592809055,
+      "loss": 0.1605,
+      "step": 2012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019828453712980613,
+      "loss": 0.1428,
+      "step": 2013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982819463935988,
+      "loss": 0.1237,
+      "step": 2014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019827935371951964,
+      "loss": 0.1998,
+      "step": 2015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982767591076198,
+      "loss": 0.1801,
+      "step": 2016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019827416255795037,
+      "loss": 0.1476,
+      "step": 2017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019827156407056265,
+      "loss": 0.1208,
+      "step": 2018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019826896364550776,
+      "loss": 0.1377,
+      "step": 2019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982663612828371,
+      "loss": 0.1473,
+      "step": 2020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982637569826019,
+      "loss": 0.1196,
+      "step": 2021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982611507448535,
+      "loss": 0.1449,
+      "step": 2022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982585425696433,
+      "loss": 0.1299,
+      "step": 2023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982559324570228,
+      "loss": 0.1166,
+      "step": 2024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019825332040704336,
+      "loss": 0.0892,
+      "step": 2025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019825070641975653,
+      "loss": 0.1022,
+      "step": 2026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019824809049521384,
+      "loss": 0.1592,
+      "step": 2027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982454726334669,
+      "loss": 0.118,
+      "step": 2028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019824285283456733,
+      "loss": 0.0784,
+      "step": 2029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019824023109856676,
+      "loss": 0.1327,
+      "step": 2030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019823760742551685,
+      "loss": 0.1256,
+      "step": 2031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019823498181546937,
+      "loss": 0.1384,
+      "step": 2032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982323542684761,
+      "loss": 0.0975,
+      "step": 2033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019822972478458884,
+      "loss": 0.1584,
+      "step": 2034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019822709336385943,
+      "loss": 0.162,
+      "step": 2035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019822446000633976,
+      "loss": 0.0842,
+      "step": 2036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019822182471208177,
+      "loss": 0.0879,
+      "step": 2037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019821918748113743,
+      "loss": 0.1078,
+      "step": 2038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019821654831355864,
+      "loss": 0.1793,
+      "step": 2039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019821390720939757,
+      "loss": 0.1101,
+      "step": 2040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019821126416870623,
+      "loss": 0.1388,
+      "step": 2041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982086191915367,
+      "loss": 0.0876,
+      "step": 2042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019820597227794123,
+      "loss": 0.1647,
+      "step": 2043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001982033234279719,
+      "loss": 0.1446,
+      "step": 2044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019820067264168105,
+      "loss": 0.1539,
+      "step": 2045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019819801991912086,
+      "loss": 0.1831,
+      "step": 2046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019819536526034364,
+      "loss": 0.1263,
+      "step": 2047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.001981927086654018,
+      "loss": 0.0817,
+      "step": 2048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019819005013434767,
+      "loss": 0.2009,
+      "step": 2049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019818738966723365,
+      "loss": 0.0882,
+      "step": 2050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019818472726411225,
+      "loss": 0.1033,
+      "step": 2051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0019818206292503593,
+      "loss": 0.0907,
+      "step": 2052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019817939665005726,
+      "loss": 0.1899,
+      "step": 2053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019817672843922877,
+      "loss": 0.1264,
+      "step": 2054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981740582926031,
+      "loss": 0.1049,
+      "step": 2055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019817138621023287,
+      "loss": 0.1055,
+      "step": 2056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019816871219217077,
+      "loss": 0.1335,
+      "step": 2057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019816603623846957,
+      "loss": 0.0792,
+      "step": 2058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019816335834918204,
+      "loss": 0.1501,
+      "step": 2059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019816067852436088,
+      "loss": 0.0932,
+      "step": 2060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00198157996764059,
+      "loss": 0.1365,
+      "step": 2061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981553130683293,
+      "loss": 0.1243,
+      "step": 2062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019815262743722462,
+      "loss": 0.115,
+      "step": 2063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00198149939870798,
+      "loss": 0.114,
+      "step": 2064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019814725036910235,
+      "loss": 0.0906,
+      "step": 2065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981445589321908,
+      "loss": 0.1387,
+      "step": 2066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981418655601163,
+      "loss": 0.1332,
+      "step": 2067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981391702529321,
+      "loss": 0.1678,
+      "step": 2068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019813647301069118,
+      "loss": 0.1252,
+      "step": 2069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019813377383344688,
+      "loss": 0.1547,
+      "step": 2070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981310727212523,
+      "loss": 0.1207,
+      "step": 2071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981283696741607,
+      "loss": 0.1221,
+      "step": 2072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019812566469222547,
+      "loss": 0.132,
+      "step": 2073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981229577754999,
+      "loss": 0.2102,
+      "step": 2074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019812024892403735,
+      "loss": 0.1711,
+      "step": 2075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019811753813789123,
+      "loss": 0.0933,
+      "step": 2076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019811482541711506,
+      "loss": 0.1243,
+      "step": 2077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981121107617622,
+      "loss": 0.1396,
+      "step": 2078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001981093941718863,
+      "loss": 0.1287,
+      "step": 2079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019810667564754084,
+      "loss": 0.1001,
+      "step": 2080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019810395518877946,
+      "loss": 0.1578,
+      "step": 2081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019810123279565576,
+      "loss": 0.2297,
+      "step": 2082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980985084682235,
+      "loss": 0.0909,
+      "step": 2083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019809578220653635,
+      "loss": 0.1018,
+      "step": 2084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019809305401064805,
+      "loss": 0.1743,
+      "step": 2085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019809032388061238,
+      "loss": 0.1373,
+      "step": 2086
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019808759181648324,
+      "loss": 0.0989,
+      "step": 2087
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019808485781831445,
+      "loss": 0.1289,
+      "step": 2088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980821218861599,
+      "loss": 0.1646,
+      "step": 2089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980793840200736,
+      "loss": 0.1404,
+      "step": 2090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980766442201095,
+      "loss": 0.168,
+      "step": 2091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019807390248632157,
+      "loss": 0.1115,
+      "step": 2092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019807115881876392,
+      "loss": 0.1526,
+      "step": 2093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980684132174907,
+      "loss": 0.1283,
+      "step": 2094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980656656825559,
+      "loss": 0.2019,
+      "step": 2095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.001980629162140139,
+      "loss": 0.1024,
+      "step": 2096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019806016481191875,
+      "loss": 0.1511,
+      "step": 2097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0019805741147632472,
+      "loss": 0.1465,
+      "step": 2098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019805465620728615,
+      "loss": 0.1942,
+      "step": 2099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980518990048573,
+      "loss": 0.1005,
+      "step": 2100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019804913986909266,
+      "loss": 0.1943,
+      "step": 2101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019804637880004652,
+      "loss": 0.1379,
+      "step": 2102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019804361579777337,
+      "loss": 0.199,
+      "step": 2103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980408508623276,
+      "loss": 0.1852,
+      "step": 2104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980380839937639,
+      "loss": 0.2139,
+      "step": 2105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019803531519213668,
+      "loss": 0.1649,
+      "step": 2106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980325444575006,
+      "loss": 0.0688,
+      "step": 2107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019802977178991024,
+      "loss": 0.1915,
+      "step": 2108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019802699718942034,
+      "loss": 0.1425,
+      "step": 2109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019802422065608553,
+      "loss": 0.1643,
+      "step": 2110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980214421899606,
+      "loss": 0.1152,
+      "step": 2111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019801866179110037,
+      "loss": 0.0969,
+      "step": 2112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980158794595596,
+      "loss": 0.1345,
+      "step": 2113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980130951953932,
+      "loss": 0.1875,
+      "step": 2114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019801030899865603,
+      "loss": 0.1266,
+      "step": 2115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019800752086940304,
+      "loss": 0.2009,
+      "step": 2116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980047308076892,
+      "loss": 0.1595,
+      "step": 2117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001980019388135695,
+      "loss": 0.1525,
+      "step": 2118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019799914488709907,
+      "loss": 0.0881,
+      "step": 2119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979963490283329,
+      "loss": 0.1444,
+      "step": 2120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979935512373262,
+      "loss": 0.1797,
+      "step": 2121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979907515141341,
+      "loss": 0.1156,
+      "step": 2122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019798794985881178,
+      "loss": 0.1052,
+      "step": 2123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979851462714145,
+      "loss": 0.1519,
+      "step": 2124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979823407519976,
+      "loss": 0.1179,
+      "step": 2125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979795333006163,
+      "loss": 0.125,
+      "step": 2126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00197976723917326,
+      "loss": 0.2036,
+      "step": 2127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019797391260218213,
+      "loss": 0.082,
+      "step": 2128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019797109935524005,
+      "loss": 0.2474,
+      "step": 2129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019796828417655526,
+      "loss": 0.1311,
+      "step": 2130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979654670661833,
+      "loss": 0.187,
+      "step": 2131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979626480241797,
+      "loss": 0.1586,
+      "step": 2132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979598270506,
+      "loss": 0.1998,
+      "step": 2133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019795700414549982,
+      "loss": 0.1741,
+      "step": 2134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019795417930893494,
+      "loss": 0.1217,
+      "step": 2135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979513525409609,
+      "loss": 0.1664,
+      "step": 2136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979485238416336,
+      "loss": 0.1604,
+      "step": 2137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019794569321100867,
+      "loss": 0.1107,
+      "step": 2138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019794286064914197,
+      "loss": 0.201,
+      "step": 2139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979400261560894,
+      "loss": 0.1434,
+      "step": 2140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979371897319068,
+      "loss": 0.1213,
+      "step": 2141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979343513766501,
+      "loss": 0.1772,
+      "step": 2142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019793151109037525,
+      "loss": 0.2007,
+      "step": 2143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0019792866887313834,
+      "loss": 0.1479,
+      "step": 2144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.001979258247249953,
+      "loss": 0.1301,
+      "step": 2145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019792297864600224,
+      "loss": 0.099,
+      "step": 2146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019792013063621534,
+      "loss": 0.1868,
+      "step": 2147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019791728069569066,
+      "loss": 0.1381,
+      "step": 2148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019791442882448446,
+      "loss": 0.1135,
+      "step": 2149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019791157502265294,
+      "loss": 0.126,
+      "step": 2150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019790871929025243,
+      "loss": 0.0929,
+      "step": 2151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001979058616273392,
+      "loss": 0.1158,
+      "step": 2152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019790300203396953,
+      "loss": 0.1316,
+      "step": 2153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001979001405101999,
+      "loss": 0.1139,
+      "step": 2154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019789727705608667,
+      "loss": 0.1216,
+      "step": 2155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019789441167168634,
+      "loss": 0.1199,
+      "step": 2156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019789154435705543,
+      "loss": 0.1805,
+      "step": 2157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978886751122504,
+      "loss": 0.1401,
+      "step": 2158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019788580393732785,
+      "loss": 0.2017,
+      "step": 2159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978829308323444,
+      "loss": 0.1428,
+      "step": 2160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019788005579735677,
+      "loss": 0.1533,
+      "step": 2161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978771788324215,
+      "loss": 0.1351,
+      "step": 2162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019787429993759547,
+      "loss": 0.1859,
+      "step": 2163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019787141911293536,
+      "loss": 0.1072,
+      "step": 2164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019786853635849794,
+      "loss": 0.1583,
+      "step": 2165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019786565167434016,
+      "loss": 0.1077,
+      "step": 2166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978627650605188,
+      "loss": 0.1936,
+      "step": 2167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019785987651709082,
+      "loss": 0.1561,
+      "step": 2168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978569860441132,
+      "loss": 0.1365,
+      "step": 2169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019785409364164288,
+      "loss": 0.1215,
+      "step": 2170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978511993097369,
+      "loss": 0.2047,
+      "step": 2171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019784830304845234,
+      "loss": 0.1509,
+      "step": 2172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978454048578463,
+      "loss": 0.0927,
+      "step": 2173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019784250473797596,
+      "loss": 0.0985,
+      "step": 2174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978396026888985,
+      "loss": 0.1339,
+      "step": 2175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978366987106711,
+      "loss": 0.0969,
+      "step": 2176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00197833792803351,
+      "loss": 0.1494,
+      "step": 2177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019783088496699553,
+      "loss": 0.1218,
+      "step": 2178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019782797520166207,
+      "loss": 0.1941,
+      "step": 2179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019782506350740797,
+      "loss": 0.1257,
+      "step": 2180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978221498842906,
+      "loss": 0.1256,
+      "step": 2181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001978192343323674,
+      "loss": 0.1278,
+      "step": 2182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019781631685169593,
+      "loss": 0.1345,
+      "step": 2183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019781339744233363,
+      "loss": 0.0552,
+      "step": 2184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019781047610433816,
+      "loss": 0.1288,
+      "step": 2185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019780755283776708,
+      "loss": 0.1207,
+      "step": 2186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019780462764267797,
+      "loss": 0.1429,
+      "step": 2187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0019780170051912857,
+      "loss": 0.1163,
+      "step": 2188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001977987714671766,
+      "loss": 0.1763,
+      "step": 2189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001977958404868798,
+      "loss": 0.1484,
+      "step": 2190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.001977929075782959,
+      "loss": 0.1174,
+      "step": 2191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019778997274148284,
+      "loss": 0.1017,
+      "step": 2192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019778703597649844,
+      "loss": 0.1361,
+      "step": 2193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977840972834006,
+      "loss": 0.1752,
+      "step": 2194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977811566622472,
+      "loss": 0.1211,
+      "step": 2195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019777821411309632,
+      "loss": 0.1174,
+      "step": 2196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00197775269636006,
+      "loss": 0.1458,
+      "step": 2197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019777232323103418,
+      "loss": 0.1152,
+      "step": 2198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019776937489823904,
+      "loss": 0.0762,
+      "step": 2199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019776642463767867,
+      "loss": 0.1625,
+      "step": 2200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019776347244941134,
+      "loss": 0.1541,
+      "step": 2201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019776051833349513,
+      "loss": 0.1526,
+      "step": 2202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019775756228998832,
+      "loss": 0.1165,
+      "step": 2203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977546043189492,
+      "loss": 0.1082,
+      "step": 2204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977516444204362,
+      "loss": 0.2163,
+      "step": 2205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019774868259450754,
+      "loss": 0.1705,
+      "step": 2206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977457188412217,
+      "loss": 0.1145,
+      "step": 2207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019774275316063704,
+      "loss": 0.1005,
+      "step": 2208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977397855528121,
+      "loss": 0.0947,
+      "step": 2209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019773681601780537,
+      "loss": 0.1389,
+      "step": 2210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019773384455567546,
+      "loss": 0.1544,
+      "step": 2211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977308711664809,
+      "loss": 0.111,
+      "step": 2212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977278958502803,
+      "loss": 0.0635,
+      "step": 2213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977249186071323,
+      "loss": 0.1685,
+      "step": 2214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019772193943709574,
+      "loss": 0.1147,
+      "step": 2215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977189583402293,
+      "loss": 0.0929,
+      "step": 2216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977159753165917,
+      "loss": 0.188,
+      "step": 2217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977129903662418,
+      "loss": 0.1324,
+      "step": 2218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019771000348923844,
+      "loss": 0.1377,
+      "step": 2219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001977070146856405,
+      "loss": 0.1259,
+      "step": 2220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00197704023955507,
+      "loss": 0.1648,
+      "step": 2221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019770103129889685,
+      "loss": 0.1031,
+      "step": 2222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019769803671586903,
+      "loss": 0.1792,
+      "step": 2223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019769504020648258,
+      "loss": 0.1582,
+      "step": 2224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001976920417707967,
+      "loss": 0.1636,
+      "step": 2225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019768904140887032,
+      "loss": 0.0841,
+      "step": 2226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019768603912076275,
+      "loss": 0.1738,
+      "step": 2227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019768303490653315,
+      "loss": 0.1235,
+      "step": 2228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019768002876624077,
+      "loss": 0.1614,
+      "step": 2229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019767702069994485,
+      "loss": 0.1025,
+      "step": 2230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019767401070770467,
+      "loss": 0.1594,
+      "step": 2231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001976709987895797,
+      "loss": 0.1671,
+      "step": 2232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001976679849456292,
+      "loss": 0.1158,
+      "step": 2233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0019766496917591266,
+      "loss": 0.1013,
+      "step": 2234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001976619514804895,
+      "loss": 0.2104,
+      "step": 2235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001976589318594193,
+      "loss": 0.1281,
+      "step": 2236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.001976559103127615,
+      "loss": 0.1658,
+      "step": 2237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019765288684057574,
+      "loss": 0.1256,
+      "step": 2238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019764986144292164,
+      "loss": 0.1515,
+      "step": 2239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019764683411985884,
+      "loss": 0.0848,
+      "step": 2240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00197643804871447,
+      "loss": 0.1931,
+      "step": 2241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001976407736977459,
+      "loss": 0.1469,
+      "step": 2242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001976377405988153,
+      "loss": 0.166,
+      "step": 2243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019763470557471498,
+      "loss": 0.1014,
+      "step": 2244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019763166862550477,
+      "loss": 0.1442,
+      "step": 2245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001976286297512446,
+      "loss": 0.2068,
+      "step": 2246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001976255889519943,
+      "loss": 0.1843,
+      "step": 2247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019762254622781392,
+      "loss": 0.1154,
+      "step": 2248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019761950157876347,
+      "loss": 0.1789,
+      "step": 2249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019761645500490288,
+      "loss": 0.1819,
+      "step": 2250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001976134065062923,
+      "loss": 0.1102,
+      "step": 2251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001976103560829918,
+      "loss": 0.1356,
+      "step": 2252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019760730373506153,
+      "loss": 0.1462,
+      "step": 2253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019760424946256173,
+      "loss": 0.1426,
+      "step": 2254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019760119326555255,
+      "loss": 0.1041,
+      "step": 2255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019759813514409427,
+      "loss": 0.1456,
+      "step": 2256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975950750982472,
+      "loss": 0.1969,
+      "step": 2257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975920131280717,
+      "loss": 0.1328,
+      "step": 2258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019758894923362813,
+      "loss": 0.1258,
+      "step": 2259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019758588341497684,
+      "loss": 0.1606,
+      "step": 2260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019758281567217838,
+      "loss": 0.0954,
+      "step": 2261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019757974600529316,
+      "loss": 0.1927,
+      "step": 2262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975766744143817,
+      "loss": 0.1617,
+      "step": 2263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975736008995047,
+      "loss": 0.1437,
+      "step": 2264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019757052546072254,
+      "loss": 0.1013,
+      "step": 2265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975674480980961,
+      "loss": 0.1733,
+      "step": 2266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019756436881168586,
+      "loss": 0.1439,
+      "step": 2267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019756128760155265,
+      "loss": 0.2056,
+      "step": 2268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019755820446775716,
+      "loss": 0.1388,
+      "step": 2269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975551194103602,
+      "loss": 0.1876,
+      "step": 2270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019755203242942266,
+      "loss": 0.1187,
+      "step": 2271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019754894352500534,
+      "loss": 0.1055,
+      "step": 2272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019754585269716914,
+      "loss": 0.1146,
+      "step": 2273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00197542759945975,
+      "loss": 0.1208,
+      "step": 2274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00197539665271484,
+      "loss": 0.1436,
+      "step": 2275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019753656867375706,
+      "loss": 0.1182,
+      "step": 2276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019753347015285527,
+      "loss": 0.1072,
+      "step": 2277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975303697088397,
+      "loss": 0.1654,
+      "step": 2278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975272673417715,
+      "loss": 0.1749,
+      "step": 2279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0019752416305171185,
+      "loss": 0.1108,
+      "step": 2280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975210568387219,
+      "loss": 0.1665,
+      "step": 2281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00197517948702863,
+      "loss": 0.134,
+      "step": 2282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.001975148386441964,
+      "loss": 0.1307,
+      "step": 2283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019751172666278337,
+      "loss": 0.1279,
+      "step": 2284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019750861275868534,
+      "loss": 0.088,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019750549693196362,
+      "loss": 0.1422,
+      "step": 2286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001975023791826797,
+      "loss": 0.1147,
+      "step": 2287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019749925951089513,
+      "loss": 0.1329,
+      "step": 2288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019749613791667127,
+      "loss": 0.1086,
+      "step": 2289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974930144000698,
+      "loss": 0.1205,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974898889611522,
+      "loss": 0.0809,
+      "step": 2291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974867615999802,
+      "loss": 0.1859,
+      "step": 2292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019748363231661538,
+      "loss": 0.1351,
+      "step": 2293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974805011111195,
+      "loss": 0.1516,
+      "step": 2294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019747736798355424,
+      "loss": 0.1586,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974742329339814,
+      "loss": 0.1329,
+      "step": 2296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019747109596246286,
+      "loss": 0.1379,
+      "step": 2297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019746795706906037,
+      "loss": 0.1887,
+      "step": 2298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974648162538359,
+      "loss": 0.0986,
+      "step": 2299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019746167351685135,
+      "loss": 0.1237,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974585288581687,
+      "loss": 0.0883,
+      "step": 2301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974553822778499,
+      "loss": 0.1163,
+      "step": 2302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019745223377595706,
+      "loss": 0.0955,
+      "step": 2303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019744908335255222,
+      "loss": 0.1382,
+      "step": 2304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019744593100769754,
+      "loss": 0.1027,
+      "step": 2305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019744277674145516,
+      "loss": 0.1562,
+      "step": 2306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974396205538872,
+      "loss": 0.15,
+      "step": 2307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00197436462445056,
+      "loss": 0.1516,
+      "step": 2308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019743330241502374,
+      "loss": 0.1487,
+      "step": 2309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019743014046385283,
+      "loss": 0.1932,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019742697659160557,
+      "loss": 0.0809,
+      "step": 2311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019742381079834426,
+      "loss": 0.1387,
+      "step": 2312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019742064308413145,
+      "loss": 0.0848,
+      "step": 2313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019741747344902954,
+      "loss": 0.0864,
+      "step": 2314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00197414301893101,
+      "loss": 0.1315,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974111284164084,
+      "loss": 0.1143,
+      "step": 2316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974079530190143,
+      "loss": 0.0868,
+      "step": 2317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974047757009814,
+      "loss": 0.2029,
+      "step": 2318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.001974015964623722,
+      "loss": 0.1619,
+      "step": 2319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019739841530324945,
+      "loss": 0.2346,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019739523222367587,
+      "loss": 0.108,
+      "step": 2321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019739204722371427,
+      "loss": 0.1362,
+      "step": 2322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019738886030342736,
+      "loss": 0.1289,
+      "step": 2323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019738567146287806,
+      "loss": 0.1228,
+      "step": 2324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019738248070212926,
+      "loss": 0.1753,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019737928802124375,
+      "loss": 0.1051,
+      "step": 2326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019737609342028463,
+      "loss": 0.1288,
+      "step": 2327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019737289689931476,
+      "loss": 0.1559,
+      "step": 2328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0019736969845839726,
+      "loss": 0.1595,
+      "step": 2329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019736649809759515,
+      "loss": 0.1274,
+      "step": 2330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019736329581697153,
+      "loss": 0.162,
+      "step": 2331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973600916165896,
+      "loss": 0.1267,
+      "step": 2332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973568854965125,
+      "loss": 0.0708,
+      "step": 2333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973536774568034,
+      "loss": 0.1545,
+      "step": 2334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973504674975256,
+      "loss": 0.1492,
+      "step": 2335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019734725561874242,
+      "loss": 0.1926,
+      "step": 2336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973440418205171,
+      "loss": 0.1599,
+      "step": 2337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019734082610291313,
+      "loss": 0.106,
+      "step": 2338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973376084659938,
+      "loss": 0.1156,
+      "step": 2339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019733438890982267,
+      "loss": 0.1425,
+      "step": 2340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973311674344631,
+      "loss": 0.1576,
+      "step": 2341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973279440399787,
+      "loss": 0.0927,
+      "step": 2342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019732471872643295,
+      "loss": 0.1306,
+      "step": 2343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019732149149388955,
+      "loss": 0.1438,
+      "step": 2344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019731826234241205,
+      "loss": 0.106,
+      "step": 2345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973150312720641,
+      "loss": 0.0756,
+      "step": 2346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973117982829095,
+      "loss": 0.1053,
+      "step": 2347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973085633750119,
+      "loss": 0.2278,
+      "step": 2348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001973053265484352,
+      "loss": 0.1836,
+      "step": 2349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019730208780324313,
+      "loss": 0.0981,
+      "step": 2350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019729884713949957,
+      "loss": 0.0908,
+      "step": 2351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972956045572684,
+      "loss": 0.1394,
+      "step": 2352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972923600566136,
+      "loss": 0.1058,
+      "step": 2353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019728911363759913,
+      "loss": 0.1851,
+      "step": 2354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019728586530028897,
+      "loss": 0.1309,
+      "step": 2355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019728261504474718,
+      "loss": 0.0892,
+      "step": 2356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019727936287103785,
+      "loss": 0.0957,
+      "step": 2357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019727610877922513,
+      "loss": 0.1388,
+      "step": 2358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972728527693732,
+      "loss": 0.121,
+      "step": 2359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972695948415462,
+      "loss": 0.1328,
+      "step": 2360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019726633499580837,
+      "loss": 0.196,
+      "step": 2361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00197263073232224,
+      "loss": 0.1971,
+      "step": 2362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972598095508575,
+      "loss": 0.0833,
+      "step": 2363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019725654395177307,
+      "loss": 0.1494,
+      "step": 2364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019725327643503516,
+      "loss": 0.1089,
+      "step": 2365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972500070007082,
+      "loss": 0.1467,
+      "step": 2366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972467356488567,
+      "loss": 0.0941,
+      "step": 2367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00197243462379545,
+      "loss": 0.1091,
+      "step": 2368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972401871928379,
+      "loss": 0.0972,
+      "step": 2369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019723691008879975,
+      "loss": 0.1515,
+      "step": 2370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019723363106749527,
+      "loss": 0.146,
+      "step": 2371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019723035012898914,
+      "loss": 0.1506,
+      "step": 2372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019722706727334596,
+      "loss": 0.0872,
+      "step": 2373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0019722378250063054,
+      "loss": 0.0845,
+      "step": 2374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.001972204958109076,
+      "loss": 0.2003,
+      "step": 2375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019721720720424197,
+      "loss": 0.1053,
+      "step": 2376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001972139166806985,
+      "loss": 0.0682,
+      "step": 2377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00197210624240342,
+      "loss": 0.1375,
+      "step": 2378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001972073298832375,
+      "loss": 0.079,
+      "step": 2379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001972040336094499,
+      "loss": 0.2102,
+      "step": 2380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019720073541904417,
+      "loss": 0.09,
+      "step": 2381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019719743531208538,
+      "loss": 0.0996,
+      "step": 2382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019719413328863856,
+      "loss": 0.1177,
+      "step": 2383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019719082934876887,
+      "loss": 0.1329,
+      "step": 2384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019718752349254147,
+      "loss": 0.1166,
+      "step": 2385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019718421572002143,
+      "loss": 0.1392,
+      "step": 2386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019718090603127406,
+      "loss": 0.0779,
+      "step": 2387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019717759442636463,
+      "loss": 0.1162,
+      "step": 2388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001971742809053584,
+      "loss": 0.2012,
+      "step": 2389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001971709654683207,
+      "loss": 0.1423,
+      "step": 2390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019716764811531693,
+      "loss": 0.0752,
+      "step": 2391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019716432884641245,
+      "loss": 0.104,
+      "step": 2392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019716100766167277,
+      "loss": 0.1943,
+      "step": 2393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019715768456116337,
+      "loss": 0.2083,
+      "step": 2394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019715435954494975,
+      "loss": 0.1357,
+      "step": 2395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019715103261309743,
+      "loss": 0.1556,
+      "step": 2396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019714770376567207,
+      "loss": 0.1238,
+      "step": 2397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001971443730027393,
+      "loss": 0.162,
+      "step": 2398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019714104032436475,
+      "loss": 0.1368,
+      "step": 2399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001971377057306142,
+      "loss": 0.2003,
+      "step": 2400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019713436922155336,
+      "loss": 0.1754,
+      "step": 2401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00197131030797248,
+      "loss": 0.1628,
+      "step": 2402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019712769045776395,
+      "loss": 0.1584,
+      "step": 2403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001971243482031671,
+      "loss": 0.1458,
+      "step": 2404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019712100403352336,
+      "loss": 0.1444,
+      "step": 2405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019711765794889866,
+      "loss": 0.1555,
+      "step": 2406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019711430994935894,
+      "loss": 0.1685,
+      "step": 2407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019711096003497026,
+      "loss": 0.1505,
+      "step": 2408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019710760820579863,
+      "loss": 0.1425,
+      "step": 2409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019710425446191017,
+      "loss": 0.1486,
+      "step": 2410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019710089880337098,
+      "loss": 0.1231,
+      "step": 2411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001970975412302473,
+      "loss": 0.1716,
+      "step": 2412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001970941817426052,
+      "loss": 0.118,
+      "step": 2413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019709082034051104,
+      "loss": 0.0759,
+      "step": 2414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019708745702403103,
+      "loss": 0.1765,
+      "step": 2415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019708409179323154,
+      "loss": 0.1146,
+      "step": 2416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019708072464817887,
+      "loss": 0.1653,
+      "step": 2417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019707735558893943,
+      "loss": 0.1267,
+      "step": 2418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001970739846155797,
+      "loss": 0.1241,
+      "step": 2419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0019707061172816605,
+      "loss": 0.1097,
+      "step": 2420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.001970672369267651,
+      "loss": 0.162,
+      "step": 2421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019706386021144328,
+      "loss": 0.1405,
+      "step": 2422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019706048158226724,
+      "loss": 0.1029,
+      "step": 2423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019705710103930358,
+      "loss": 0.2047,
+      "step": 2424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970537185826189,
+      "loss": 0.1315,
+      "step": 2425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019705033421228003,
+      "loss": 0.1372,
+      "step": 2426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970469479283536,
+      "loss": 0.147,
+      "step": 2427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019704355973090637,
+      "loss": 0.1041,
+      "step": 2428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019704016962000517,
+      "loss": 0.1919,
+      "step": 2429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970367775957169,
+      "loss": 0.1339,
+      "step": 2430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019703338365810834,
+      "loss": 0.1469,
+      "step": 2431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019702998780724646,
+      "loss": 0.11,
+      "step": 2432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970265900431982,
+      "loss": 0.1547,
+      "step": 2433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019702319036603065,
+      "loss": 0.1538,
+      "step": 2434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019701978877581074,
+      "loss": 0.1199,
+      "step": 2435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019701638527260556,
+      "loss": 0.1268,
+      "step": 2436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970129798564822,
+      "loss": 0.1699,
+      "step": 2437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019700957252750785,
+      "loss": 0.1877,
+      "step": 2438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970061632857497,
+      "loss": 0.0953,
+      "step": 2439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001970027521312749,
+      "loss": 0.1091,
+      "step": 2440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969993390641508,
+      "loss": 0.2007,
+      "step": 2441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019699592408444462,
+      "loss": 0.101,
+      "step": 2442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019699250719222377,
+      "loss": 0.157,
+      "step": 2443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019698908838755554,
+      "loss": 0.1407,
+      "step": 2444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019698566767050742,
+      "loss": 0.1179,
+      "step": 2445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969822450411468,
+      "loss": 0.1104,
+      "step": 2446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969788204995412,
+      "loss": 0.092,
+      "step": 2447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969753940457581,
+      "loss": 0.1261,
+      "step": 2448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969719656798651,
+      "loss": 0.1084,
+      "step": 2449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969685354019298,
+      "loss": 0.1167,
+      "step": 2450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969651032120198,
+      "loss": 0.2052,
+      "step": 2451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019696166911020285,
+      "loss": 0.1282,
+      "step": 2452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969582330965466,
+      "loss": 0.0836,
+      "step": 2453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019695479517111876,
+      "loss": 0.113,
+      "step": 2454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969513553339872,
+      "loss": 0.135,
+      "step": 2455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019694791358521972,
+      "loss": 0.1628,
+      "step": 2456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969444699248842,
+      "loss": 0.1357,
+      "step": 2457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019694102435304844,
+      "loss": 0.1444,
+      "step": 2458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019693757686978053,
+      "loss": 0.0967,
+      "step": 2459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969341274751483,
+      "loss": 0.1262,
+      "step": 2460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019693067616921986,
+      "loss": 0.1696,
+      "step": 2461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019692722295206327,
+      "loss": 0.1046,
+      "step": 2462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019692376782374653,
+      "loss": 0.188,
+      "step": 2463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019692031078433784,
+      "loss": 0.0963,
+      "step": 2464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.001969168518339054,
+      "loss": 0.1141,
+      "step": 2465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019691339097251724,
+      "loss": 0.0929,
+      "step": 2466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0019690992820024177,
+      "loss": 0.132,
+      "step": 2467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019690646351714722,
+      "loss": 0.1169,
+      "step": 2468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001969029969233019,
+      "loss": 0.1112,
+      "step": 2469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019689952841877413,
+      "loss": 0.1028,
+      "step": 2470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019689605800363237,
+      "loss": 0.1263,
+      "step": 2471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019689258567794497,
+      "loss": 0.1211,
+      "step": 2472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019688911144178046,
+      "loss": 0.1162,
+      "step": 2473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968856352952073,
+      "loss": 0.0863,
+      "step": 2474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019688215723829405,
+      "loss": 0.1045,
+      "step": 2475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019687867727110927,
+      "loss": 0.0865,
+      "step": 2476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968751953937216,
+      "loss": 0.118,
+      "step": 2477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968717116061997,
+      "loss": 0.1183,
+      "step": 2478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968682259086122,
+      "loss": 0.0939,
+      "step": 2479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019686473830102793,
+      "loss": 0.089,
+      "step": 2480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019686124878351557,
+      "loss": 0.1461,
+      "step": 2481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019685775735614397,
+      "loss": 0.1182,
+      "step": 2482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019685426401898193,
+      "loss": 0.1134,
+      "step": 2483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968507687720984,
+      "loss": 0.1123,
+      "step": 2484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968472716155622,
+      "loss": 0.1152,
+      "step": 2485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968437725494424,
+      "loss": 0.1028,
+      "step": 2486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019684027157380792,
+      "loss": 0.1548,
+      "step": 2487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968367686887278,
+      "loss": 0.1079,
+      "step": 2488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968332638942711,
+      "loss": 0.1941,
+      "step": 2489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019682975719050692,
+      "loss": 0.0953,
+      "step": 2490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019682624857750445,
+      "loss": 0.1483,
+      "step": 2491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019682273805533283,
+      "loss": 0.1069,
+      "step": 2492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968192256240613,
+      "loss": 0.1135,
+      "step": 2493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968157112837591,
+      "loss": 0.0977,
+      "step": 2494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019681219503449553,
+      "loss": 0.1855,
+      "step": 2495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019680867687633993,
+      "loss": 0.1481,
+      "step": 2496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019680515680936166,
+      "loss": 0.1642,
+      "step": 2497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001968016348336301,
+      "loss": 0.0985,
+      "step": 2498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019679811094921477,
+      "loss": 0.1151,
+      "step": 2499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019679458515618506,
+      "loss": 0.1161,
+      "step": 2500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001967910574546106,
+      "loss": 0.1273,
+      "step": 2501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001967875278445608,
+      "loss": 0.0946,
+      "step": 2502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019678399632610537,
+      "loss": 0.1133,
+      "step": 2503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019678046289931393,
+      "loss": 0.1003,
+      "step": 2504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001967769275642561,
+      "loss": 0.1604,
+      "step": 2505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019677339032100165,
+      "loss": 0.0985,
+      "step": 2506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019676985116962027,
+      "loss": 0.1365,
+      "step": 2507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019676631011018176,
+      "loss": 0.092,
+      "step": 2508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00196762767142756,
+      "loss": 0.083,
+      "step": 2509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019675922226741276,
+      "loss": 0.1515,
+      "step": 2510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019675567548422196,
+      "loss": 0.1577,
+      "step": 2511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0019675212679325354,
+      "loss": 0.1644,
+      "step": 2512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001967485761945775,
+      "loss": 0.1029,
+      "step": 2513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.001967450236882638,
+      "loss": 0.1208,
+      "step": 2514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967414692743825,
+      "loss": 0.1046,
+      "step": 2515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019673791295300373,
+      "loss": 0.1198,
+      "step": 2516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967343547241976,
+      "loss": 0.1227,
+      "step": 2517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967307945880342,
+      "loss": 0.0933,
+      "step": 2518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019672723254458378,
+      "loss": 0.115,
+      "step": 2519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019672366859391655,
+      "loss": 0.0977,
+      "step": 2520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019672010273610284,
+      "loss": 0.123,
+      "step": 2521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967165349712129,
+      "loss": 0.2184,
+      "step": 2522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967129652993171,
+      "loss": 0.1562,
+      "step": 2523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019670939372048585,
+      "loss": 0.0862,
+      "step": 2524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967058202347895,
+      "loss": 0.1721,
+      "step": 2525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001967022448422986,
+      "loss": 0.0972,
+      "step": 2526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966986675430836,
+      "loss": 0.1187,
+      "step": 2527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00196695088337215,
+      "loss": 0.0715,
+      "step": 2528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966915072247634,
+      "loss": 0.1912,
+      "step": 2529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019668792420579947,
+      "loss": 0.0956,
+      "step": 2530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019668433928039376,
+      "loss": 0.1222,
+      "step": 2531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019668075244861703,
+      "loss": 0.1243,
+      "step": 2532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019667716371053997,
+      "loss": 0.1639,
+      "step": 2533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966735730662334,
+      "loss": 0.1725,
+      "step": 2534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019666998051576802,
+      "loss": 0.1292,
+      "step": 2535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966663860592147,
+      "loss": 0.0958,
+      "step": 2536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019666278969664437,
+      "loss": 0.1542,
+      "step": 2537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019665919142812785,
+      "loss": 0.1027,
+      "step": 2538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019665559125373617,
+      "loss": 0.1375,
+      "step": 2539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966519891735403,
+      "loss": 0.1131,
+      "step": 2540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966483851876112,
+      "loss": 0.0986,
+      "step": 2541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019664477929601997,
+      "loss": 0.079,
+      "step": 2542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966411714988378,
+      "loss": 0.0859,
+      "step": 2543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019663756179613564,
+      "loss": 0.1946,
+      "step": 2544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019663395018798486,
+      "loss": 0.1362,
+      "step": 2545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019663033667445652,
+      "loss": 0.1132,
+      "step": 2546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019662672125562194,
+      "loss": 0.1339,
+      "step": 2547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001966231039315524,
+      "loss": 0.0825,
+      "step": 2548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019661948470231926,
+      "loss": 0.1268,
+      "step": 2549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019661586356799383,
+      "loss": 0.1017,
+      "step": 2550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019661224052864752,
+      "loss": 0.1587,
+      "step": 2551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019660861558435175,
+      "loss": 0.0853,
+      "step": 2552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00196604988735178,
+      "loss": 0.1736,
+      "step": 2553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019660135998119783,
+      "loss": 0.1145,
+      "step": 2554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019659772932248275,
+      "loss": 0.0836,
+      "step": 2555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019659409675910435,
+      "loss": 0.1195,
+      "step": 2556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001965904622911343,
+      "loss": 0.098,
+      "step": 2557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001965868259186442,
+      "loss": 0.1168,
+      "step": 2558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.001965831876417057,
+      "loss": 0.1191,
+      "step": 2559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0019657954746039067,
+      "loss": 0.0948,
+      "step": 2560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965759053747708,
+      "loss": 0.1526,
+      "step": 2561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00196572261384918,
+      "loss": 0.1498,
+      "step": 2562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019656861549090395,
+      "loss": 0.1422,
+      "step": 2563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965649676928007,
+      "loss": 0.1413,
+      "step": 2564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965613179906801,
+      "loss": 0.1414,
+      "step": 2565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019655766638461414,
+      "loss": 0.1687,
+      "step": 2566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965540128746747,
+      "loss": 0.1301,
+      "step": 2567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019655035746093405,
+      "loss": 0.1638,
+      "step": 2568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965467001434641,
+      "loss": 0.1189,
+      "step": 2569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00196543040922337,
+      "loss": 0.0999,
+      "step": 2570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965393797976249,
+      "loss": 0.1741,
+      "step": 2571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965357167694,
+      "loss": 0.1116,
+      "step": 2572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965320518377345,
+      "loss": 0.0641,
+      "step": 2573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019652838500270066,
+      "loss": 0.0737,
+      "step": 2574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019652471626437082,
+      "loss": 0.0925,
+      "step": 2575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965210456228173,
+      "loss": 0.127,
+      "step": 2576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019651737307811246,
+      "loss": 0.1492,
+      "step": 2577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019651369863032874,
+      "loss": 0.1566,
+      "step": 2578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965100222795386,
+      "loss": 0.1554,
+      "step": 2579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019650634402581446,
+      "loss": 0.1239,
+      "step": 2580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001965026638692289,
+      "loss": 0.0973,
+      "step": 2581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964989818098545,
+      "loss": 0.1611,
+      "step": 2582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964952978477638,
+      "loss": 0.1958,
+      "step": 2583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964916119830295,
+      "loss": 0.0952,
+      "step": 2584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019648792421572427,
+      "loss": 0.1196,
+      "step": 2585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019648423454592077,
+      "loss": 0.0917,
+      "step": 2586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019648054297369184,
+      "loss": 0.1173,
+      "step": 2587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019647684949911015,
+      "loss": 0.1248,
+      "step": 2588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964731541222486,
+      "loss": 0.12,
+      "step": 2589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019646945684318007,
+      "loss": 0.0705,
+      "step": 2590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964657576619774,
+      "loss": 0.1094,
+      "step": 2591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019646205657871357,
+      "loss": 0.1201,
+      "step": 2592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964583535934616,
+      "loss": 0.1803,
+      "step": 2593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964546487062944,
+      "loss": 0.1878,
+      "step": 2594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019645094191728506,
+      "loss": 0.1096,
+      "step": 2595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964472332265067,
+      "loss": 0.1583,
+      "step": 2596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019644352263403246,
+      "loss": 0.1244,
+      "step": 2597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019643981013993546,
+      "loss": 0.1277,
+      "step": 2598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964360957442889,
+      "loss": 0.1025,
+      "step": 2599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019643237944716604,
+      "loss": 0.0875,
+      "step": 2600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964286612486401,
+      "loss": 0.092,
+      "step": 2601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964249411487845,
+      "loss": 0.0853,
+      "step": 2602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964212191476725,
+      "loss": 0.1006,
+      "step": 2603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0019641749524537755,
+      "loss": 0.1952,
+      "step": 2604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00196413769441973,
+      "loss": 0.1107,
+      "step": 2605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.001964100417375324,
+      "loss": 0.1604,
+      "step": 2606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001964063121321292,
+      "loss": 0.1349,
+      "step": 2607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001964025806258369,
+      "loss": 0.1344,
+      "step": 2608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019639884721872917,
+      "loss": 0.105,
+      "step": 2609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019639511191087955,
+      "loss": 0.1675,
+      "step": 2610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019639137470236173,
+      "loss": 0.1239,
+      "step": 2611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019638763559324942,
+      "loss": 0.1351,
+      "step": 2612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963838945836163,
+      "loss": 0.1381,
+      "step": 2613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019638015167353614,
+      "loss": 0.1238,
+      "step": 2614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963764068630827,
+      "loss": 0.1675,
+      "step": 2615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019637266015232993,
+      "loss": 0.1,
+      "step": 2616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963689115413516,
+      "loss": 0.0869,
+      "step": 2617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963651610302217,
+      "loss": 0.1667,
+      "step": 2618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963614086190141,
+      "loss": 0.0665,
+      "step": 2619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019635765430780286,
+      "loss": 0.1323,
+      "step": 2620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00196353898096662,
+      "loss": 0.1332,
+      "step": 2621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963501399856655,
+      "loss": 0.1215,
+      "step": 2622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019634637997488756,
+      "loss": 0.1434,
+      "step": 2623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019634261806440226,
+      "loss": 0.1547,
+      "step": 2624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019633885425428383,
+      "loss": 0.1632,
+      "step": 2625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963350885446064,
+      "loss": 0.1122,
+      "step": 2626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963313209354443,
+      "loss": 0.1733,
+      "step": 2627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963275514268718,
+      "loss": 0.1464,
+      "step": 2628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019632378001896316,
+      "loss": 0.0983,
+      "step": 2629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019632000671179284,
+      "loss": 0.1802,
+      "step": 2630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963162315054352,
+      "loss": 0.1868,
+      "step": 2631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019631245439996462,
+      "loss": 0.1233,
+      "step": 2632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019630867539545565,
+      "loss": 0.1071,
+      "step": 2633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001963048944919828,
+      "loss": 0.1379,
+      "step": 2634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019630111168962056,
+      "loss": 0.2078,
+      "step": 2635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019629732698844358,
+      "loss": 0.1407,
+      "step": 2636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019629354038852645,
+      "loss": 0.0969,
+      "step": 2637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019628975188994385,
+      "loss": 0.0807,
+      "step": 2638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019628596149277046,
+      "loss": 0.111,
+      "step": 2639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019628216919708105,
+      "loss": 0.1785,
+      "step": 2640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019627837500295034,
+      "loss": 0.1683,
+      "step": 2641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001962745789104532,
+      "loss": 0.1908,
+      "step": 2642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019627078091966443,
+      "loss": 0.1038,
+      "step": 2643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00196266981030659,
+      "loss": 0.136,
+      "step": 2644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001962631792435117,
+      "loss": 0.1464,
+      "step": 2645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019625937555829763,
+      "loss": 0.1185,
+      "step": 2646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001962555699750917,
+      "loss": 0.1337,
+      "step": 2647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00196251762493969,
+      "loss": 0.1188,
+      "step": 2648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001962479531150045,
+      "loss": 0.1187,
+      "step": 2649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.001962441418382735,
+      "loss": 0.1171,
+      "step": 2650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019624032866385095,
+      "loss": 0.1316,
+      "step": 2651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0019623651359181216,
+      "loss": 0.1791,
+      "step": 2652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001962326966222323,
+      "loss": 0.1091,
+      "step": 2653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019622887775518666,
+      "loss": 0.207,
+      "step": 2654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001962250569907505,
+      "loss": 0.0886,
+      "step": 2655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001962212343289992,
+      "loss": 0.1582,
+      "step": 2656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019621740977000812,
+      "loss": 0.155,
+      "step": 2657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019621358331385268,
+      "loss": 0.1227,
+      "step": 2658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019620975496060828,
+      "loss": 0.1525,
+      "step": 2659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019620592471035047,
+      "loss": 0.1322,
+      "step": 2660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001962020925631547,
+      "loss": 0.0935,
+      "step": 2661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961982585190966,
+      "loss": 0.2378,
+      "step": 2662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019619442257825174,
+      "loss": 0.0895,
+      "step": 2663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961905847406958,
+      "loss": 0.1367,
+      "step": 2664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019618674500650433,
+      "loss": 0.1323,
+      "step": 2665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019618290337575315,
+      "loss": 0.0864,
+      "step": 2666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00196179059848518,
+      "loss": 0.0861,
+      "step": 2667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961752144248746,
+      "loss": 0.0968,
+      "step": 2668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019617136710489884,
+      "loss": 0.1295,
+      "step": 2669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961675178886665,
+      "loss": 0.1039,
+      "step": 2670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961636667762536,
+      "loss": 0.1108,
+      "step": 2671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019615981376773597,
+      "loss": 0.1116,
+      "step": 2672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961559588631896,
+      "loss": 0.1307,
+      "step": 2673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019615210206269055,
+      "loss": 0.1315,
+      "step": 2674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961482433663148,
+      "loss": 0.1807,
+      "step": 2675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019614438277413843,
+      "loss": 0.1143,
+      "step": 2676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019614052028623765,
+      "loss": 0.1586,
+      "step": 2677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019613665590268855,
+      "loss": 0.1216,
+      "step": 2678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961327896235673,
+      "loss": 0.0897,
+      "step": 2679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019612892144895023,
+      "loss": 0.2104,
+      "step": 2680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019612505137891353,
+      "loss": 0.2231,
+      "step": 2681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001961211794135335,
+      "loss": 0.0996,
+      "step": 2682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019611730555288654,
+      "loss": 0.0966,
+      "step": 2683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00196113429797049,
+      "loss": 0.0709,
+      "step": 2684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019610955214609733,
+      "loss": 0.1406,
+      "step": 2685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019610567260010797,
+      "loss": 0.1208,
+      "step": 2686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019610179115915737,
+      "loss": 0.1095,
+      "step": 2687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019609790782332214,
+      "loss": 0.1423,
+      "step": 2688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019609402259267877,
+      "loss": 0.0935,
+      "step": 2689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019609013546730398,
+      "loss": 0.0589,
+      "step": 2690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001960862464472743,
+      "loss": 0.113,
+      "step": 2691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019608235553266644,
+      "loss": 0.1897,
+      "step": 2692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019607846272355718,
+      "loss": 0.064,
+      "step": 2693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001960745680200232,
+      "loss": 0.0811,
+      "step": 2694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019607067142214133,
+      "loss": 0.1215,
+      "step": 2695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019606677292998836,
+      "loss": 0.1362,
+      "step": 2696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0019606287254364126,
+      "loss": 0.1017,
+      "step": 2697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.001960589702631768,
+      "loss": 0.0472,
+      "step": 2698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019605506608867205,
+      "loss": 0.1366,
+      "step": 2699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960511600202039,
+      "loss": 0.1068,
+      "step": 2700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019604725205784944,
+      "loss": 0.1649,
+      "step": 2701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019604334220168565,
+      "loss": 0.131,
+      "step": 2702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960394304517897,
+      "loss": 0.1818,
+      "step": 2703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960355168082386,
+      "loss": 0.1423,
+      "step": 2704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019603160127110966,
+      "loss": 0.0888,
+      "step": 2705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019602768384048,
+      "loss": 0.0892,
+      "step": 2706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960237645164269,
+      "loss": 0.1633,
+      "step": 2707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960198432990276,
+      "loss": 0.1511,
+      "step": 2708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019601592018835944,
+      "loss": 0.2141,
+      "step": 2709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019601199518449975,
+      "loss": 0.1413,
+      "step": 2710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00196008068287526,
+      "loss": 0.1187,
+      "step": 2711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960041394975155,
+      "loss": 0.1471,
+      "step": 2712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001960002088145458,
+      "loss": 0.1234,
+      "step": 2713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019599627623869444,
+      "loss": 0.1786,
+      "step": 2714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019599234177003883,
+      "loss": 0.181,
+      "step": 2715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019598840540865667,
+      "loss": 0.1603,
+      "step": 2716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019598446715462545,
+      "loss": 0.1125,
+      "step": 2717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019598052700802297,
+      "loss": 0.1049,
+      "step": 2718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019597658496892678,
+      "loss": 0.0859,
+      "step": 2719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959726410374147,
+      "loss": 0.1539,
+      "step": 2720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959686952135645,
+      "loss": 0.1351,
+      "step": 2721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019596474749745394,
+      "loss": 0.2141,
+      "step": 2722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019596079788916083,
+      "loss": 0.116,
+      "step": 2723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959568463887631,
+      "loss": 0.1565,
+      "step": 2724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019595289299633866,
+      "loss": 0.1045,
+      "step": 2725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959489377119654,
+      "loss": 0.106,
+      "step": 2726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019594498053572143,
+      "loss": 0.1729,
+      "step": 2727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019594102146768466,
+      "loss": 0.1843,
+      "step": 2728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019593706050793323,
+      "loss": 0.1843,
+      "step": 2729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019593309765654518,
+      "loss": 0.1478,
+      "step": 2730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019592913291359864,
+      "loss": 0.1263,
+      "step": 2731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959251662791719,
+      "loss": 0.1433,
+      "step": 2732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019592119775334303,
+      "loss": 0.1411,
+      "step": 2733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019591722733619033,
+      "loss": 0.1274,
+      "step": 2734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019591325502779215,
+      "loss": 0.1033,
+      "step": 2735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959092808282267,
+      "loss": 0.1292,
+      "step": 2736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001959053047375724,
+      "loss": 0.1226,
+      "step": 2737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019590132675590772,
+      "loss": 0.1174,
+      "step": 2738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019589734688331096,
+      "loss": 0.1184,
+      "step": 2739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019589336511986067,
+      "loss": 0.1411,
+      "step": 2740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001958893814656354,
+      "loss": 0.1536,
+      "step": 2741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019588539592071356,
+      "loss": 0.1143,
+      "step": 2742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.001958814084851739,
+      "loss": 0.1727,
+      "step": 2743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0019587741915909488,
+      "loss": 0.1379,
+      "step": 2744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001958734279425553,
+      "loss": 0.1511,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001958694348356338,
+      "loss": 0.1249,
+      "step": 2746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019586543983840907,
+      "loss": 0.1375,
+      "step": 2747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019586144295095998,
+      "loss": 0.1254,
+      "step": 2748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019585744417336525,
+      "loss": 0.1417,
+      "step": 2749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019585344350570378,
+      "loss": 0.1332,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001958494409480544,
+      "loss": 0.1804,
+      "step": 2751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001958454365004961,
+      "loss": 0.1323,
+      "step": 2752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019584143016310777,
+      "loss": 0.1377,
+      "step": 2753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001958374219359684,
+      "loss": 0.1115,
+      "step": 2754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001958334118191571,
+      "loss": 0.2189,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019582939981275293,
+      "loss": 0.129,
+      "step": 2756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019582538591683494,
+      "loss": 0.1232,
+      "step": 2757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019582137013148225,
+      "loss": 0.088,
+      "step": 2758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019581735245677414,
+      "loss": 0.1329,
+      "step": 2759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019581333289278974,
+      "loss": 0.1262,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019580931143960835,
+      "loss": 0.083,
+      "step": 2761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019580528809730927,
+      "loss": 0.16,
+      "step": 2762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019580126286597183,
+      "loss": 0.1565,
+      "step": 2763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019579723574567535,
+      "loss": 0.0955,
+      "step": 2764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957932067364993,
+      "loss": 0.0536,
+      "step": 2765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019578917583852303,
+      "loss": 0.1548,
+      "step": 2766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019578514305182613,
+      "loss": 0.2397,
+      "step": 2767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019578110837648803,
+      "loss": 0.1763,
+      "step": 2768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019577707181258836,
+      "loss": 0.0814,
+      "step": 2769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957730333602066,
+      "loss": 0.1752,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957689930194225,
+      "loss": 0.1201,
+      "step": 2771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019576495079031567,
+      "loss": 0.1482,
+      "step": 2772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957609066729658,
+      "loss": 0.0996,
+      "step": 2773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957568606674527,
+      "loss": 0.1611,
+      "step": 2774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019575281277385602,
+      "loss": 0.1078,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019574876299225568,
+      "loss": 0.1458,
+      "step": 2776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019574471132273144,
+      "loss": 0.1384,
+      "step": 2777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019574065776536333,
+      "loss": 0.0715,
+      "step": 2778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957366023202311,
+      "loss": 0.152,
+      "step": 2779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957325449874149,
+      "loss": 0.1127,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019572848576699456,
+      "loss": 0.1272,
+      "step": 2781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019572442465905027,
+      "loss": 0.1759,
+      "step": 2782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019572036166366196,
+      "loss": 0.1542,
+      "step": 2783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001957162967809098,
+      "loss": 0.1215,
+      "step": 2784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019571223001087403,
+      "loss": 0.0795,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019570816135363466,
+      "loss": 0.1714,
+      "step": 2786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019570409080927205,
+      "loss": 0.1068,
+      "step": 2787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019570001837786646,
+      "loss": 0.1624,
+      "step": 2788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.001956959440594981,
+      "loss": 0.1522,
+      "step": 2789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0019569186785424737,
+      "loss": 0.131,
+      "step": 2790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956877897621946,
+      "loss": 0.1074,
+      "step": 2791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956837097834203,
+      "loss": 0.1536,
+      "step": 2792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019567962791800474,
+      "loss": 0.0715,
+      "step": 2793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019567554416602857,
+      "loss": 0.0658,
+      "step": 2794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956714585275722,
+      "loss": 0.1078,
+      "step": 2795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956673710027163,
+      "loss": 0.1189,
+      "step": 2796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019566328159154136,
+      "loss": 0.1279,
+      "step": 2797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019565919029412805,
+      "loss": 0.1689,
+      "step": 2798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019565509711055706,
+      "loss": 0.059,
+      "step": 2799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956510020409091,
+      "loss": 0.1202,
+      "step": 2800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019564690508526487,
+      "loss": 0.1381,
+      "step": 2801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019564280624370517,
+      "loss": 0.0886,
+      "step": 2802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019563870551631084,
+      "loss": 0.1044,
+      "step": 2803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019563460290316272,
+      "loss": 0.1088,
+      "step": 2804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956304984043417,
+      "loss": 0.1642,
+      "step": 2805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956263920199287,
+      "loss": 0.0775,
+      "step": 2806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019562228375000477,
+      "loss": 0.1335,
+      "step": 2807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956181735946508,
+      "loss": 0.161,
+      "step": 2808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956140615539479,
+      "loss": 0.1199,
+      "step": 2809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019560994762797706,
+      "loss": 0.1199,
+      "step": 2810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019560583181681955,
+      "loss": 0.1256,
+      "step": 2811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001956017141205564,
+      "loss": 0.0997,
+      "step": 2812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019559759453926886,
+      "loss": 0.1501,
+      "step": 2813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955934730730381,
+      "loss": 0.1482,
+      "step": 2814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019558934972194546,
+      "loss": 0.14,
+      "step": 2815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955852244860722,
+      "loss": 0.1995,
+      "step": 2816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019558109736549966,
+      "loss": 0.1101,
+      "step": 2817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955769683603092,
+      "loss": 0.064,
+      "step": 2818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955728374705823,
+      "loss": 0.0906,
+      "step": 2819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019556870469640032,
+      "loss": 0.1381,
+      "step": 2820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955645700378448,
+      "loss": 0.1772,
+      "step": 2821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955604334949973,
+      "loss": 0.1334,
+      "step": 2822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019555629506793925,
+      "loss": 0.1619,
+      "step": 2823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019555215475675243,
+      "loss": 0.113,
+      "step": 2824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955480125615183,
+      "loss": 0.0933,
+      "step": 2825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019554386848231868,
+      "loss": 0.1116,
+      "step": 2826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955397225192352,
+      "loss": 0.1909,
+      "step": 2827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955355746723496,
+      "loss": 0.1315,
+      "step": 2828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019553142494174378,
+      "loss": 0.1649,
+      "step": 2829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955272733274994,
+      "loss": 0.13,
+      "step": 2830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019552311982969843,
+      "loss": 0.1951,
+      "step": 2831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955189644484227,
+      "loss": 0.1312,
+      "step": 2832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019551480718375414,
+      "loss": 0.1207,
+      "step": 2833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019551064803577482,
+      "loss": 0.0919,
+      "step": 2834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.001955064870045667,
+      "loss": 0.1486,
+      "step": 2835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0019550232409021174,
+      "loss": 0.1714,
+      "step": 2836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954981592927921,
+      "loss": 0.1149,
+      "step": 2837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019549399261238995,
+      "loss": 0.1671,
+      "step": 2838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019548982404908734,
+      "loss": 0.1277,
+      "step": 2839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954856536029665,
+      "loss": 0.1055,
+      "step": 2840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954814812741097,
+      "loss": 0.0645,
+      "step": 2841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019547730706259915,
+      "loss": 0.1473,
+      "step": 2842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954731309685172,
+      "loss": 0.0941,
+      "step": 2843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019546895299194615,
+      "loss": 0.1069,
+      "step": 2844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019546477313296846,
+      "loss": 0.0756,
+      "step": 2845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954605913916664,
+      "loss": 0.1091,
+      "step": 2846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954564077681226,
+      "loss": 0.1736,
+      "step": 2847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954522222624194,
+      "loss": 0.0925,
+      "step": 2848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019544803487463944,
+      "loss": 0.1377,
+      "step": 2849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954438456048652,
+      "loss": 0.1588,
+      "step": 2850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019543965445317935,
+      "loss": 0.1024,
+      "step": 2851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019543546141966447,
+      "loss": 0.0949,
+      "step": 2852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954312665044033,
+      "loss": 0.0746,
+      "step": 2853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019542706970747848,
+      "loss": 0.1504,
+      "step": 2854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954228710289728,
+      "loss": 0.0903,
+      "step": 2855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954186704689691,
+      "loss": 0.1057,
+      "step": 2856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019541446802755005,
+      "loss": 0.0854,
+      "step": 2857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019541026370479865,
+      "loss": 0.0677,
+      "step": 2858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001954060575007978,
+      "loss": 0.1108,
+      "step": 2859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019540184941563036,
+      "loss": 0.1418,
+      "step": 2860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953976394493793,
+      "loss": 0.1365,
+      "step": 2861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953934276021277,
+      "loss": 0.1493,
+      "step": 2862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953892138739586,
+      "loss": 0.0969,
+      "step": 2863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019538499826495504,
+      "loss": 0.1433,
+      "step": 2864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019538078077520017,
+      "loss": 0.1331,
+      "step": 2865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953765614047771,
+      "loss": 0.1495,
+      "step": 2866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019537234015376908,
+      "loss": 0.1508,
+      "step": 2867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019536811702225935,
+      "loss": 0.1115,
+      "step": 2868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953638920103311,
+      "loss": 0.1301,
+      "step": 2869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019535966511806777,
+      "loss": 0.0976,
+      "step": 2870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953554363455526,
+      "loss": 0.1138,
+      "step": 2871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019535120569286893,
+      "loss": 0.1018,
+      "step": 2872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953469731601003,
+      "loss": 0.1014,
+      "step": 2873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953427387473301,
+      "loss": 0.0778,
+      "step": 2874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953385024546418,
+      "loss": 0.1807,
+      "step": 2875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00195334264282119,
+      "loss": 0.1056,
+      "step": 2876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953300242298452,
+      "loss": 0.2209,
+      "step": 2877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019532578229790407,
+      "loss": 0.3452,
+      "step": 2878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019532153848637918,
+      "loss": 0.1442,
+      "step": 2879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953172927953542,
+      "loss": 0.0955,
+      "step": 2880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.001953130452249129,
+      "loss": 0.097,
+      "step": 2881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019530879577513902,
+      "loss": 0.1254,
+      "step": 2882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0019530454444611637,
+      "loss": 0.1281,
+      "step": 2883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001953002912379287,
+      "loss": 0.1368,
+      "step": 2884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019529603615065993,
+      "loss": 0.1245,
+      "step": 2885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019529177918439396,
+      "loss": 0.0988,
+      "step": 2886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019528752033921473,
+      "loss": 0.1318,
+      "step": 2887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019528325961520614,
+      "loss": 0.1207,
+      "step": 2888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019527899701245232,
+      "loss": 0.2124,
+      "step": 2889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001952747325310372,
+      "loss": 0.1954,
+      "step": 2890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019527046617104499,
+      "loss": 0.1415,
+      "step": 2891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019526619793255966,
+      "loss": 0.1903,
+      "step": 2892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019526192781566549,
+      "loss": 0.1494,
+      "step": 2893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019525765582044665,
+      "loss": 0.1433,
+      "step": 2894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019525338194698737,
+      "loss": 0.1628,
+      "step": 2895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019524910619537188,
+      "loss": 0.1365,
+      "step": 2896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019524482856568453,
+      "loss": 0.1576,
+      "step": 2897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019524054905800965,
+      "loss": 0.0952,
+      "step": 2898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019523626767243161,
+      "loss": 0.151,
+      "step": 2899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019523198440903487,
+      "loss": 0.1407,
+      "step": 2900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019522769926790385,
+      "loss": 0.1906,
+      "step": 2901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019522341224912304,
+      "loss": 0.1882,
+      "step": 2902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019521912335277696,
+      "loss": 0.1235,
+      "step": 2903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001952148325789502,
+      "loss": 0.1176,
+      "step": 2904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019521053992772738,
+      "loss": 0.1521,
+      "step": 2905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001952062453991931,
+      "loss": 0.1125,
+      "step": 2906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019520194899343205,
+      "loss": 0.1525,
+      "step": 2907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019519765071052894,
+      "loss": 0.1002,
+      "step": 2908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019519335055056856,
+      "loss": 0.1415,
+      "step": 2909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019518904851363566,
+      "loss": 0.0688,
+      "step": 2910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019518474459981503,
+      "loss": 0.1157,
+      "step": 2911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019518043880919164,
+      "loss": 0.1671,
+      "step": 2912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001951761311418503,
+      "loss": 0.0814,
+      "step": 2913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019517182159787595,
+      "loss": 0.1542,
+      "step": 2914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019516751017735362,
+      "loss": 0.1277,
+      "step": 2915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019516319688036826,
+      "loss": 0.1122,
+      "step": 2916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019515888170700494,
+      "loss": 0.1377,
+      "step": 2917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019515456465734875,
+      "loss": 0.1609,
+      "step": 2918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001951502457314848,
+      "loss": 0.1481,
+      "step": 2919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019514592492949827,
+      "loss": 0.1536,
+      "step": 2920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019514160225147434,
+      "loss": 0.1785,
+      "step": 2921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019513727769749824,
+      "loss": 0.1413,
+      "step": 2922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019513295126765526,
+      "loss": 0.0622,
+      "step": 2923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019512862296203067,
+      "loss": 0.156,
+      "step": 2924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019512429278070985,
+      "loss": 0.1599,
+      "step": 2925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019511996072377814,
+      "loss": 0.2039,
+      "step": 2926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00195115626791321,
+      "loss": 0.0995,
+      "step": 2927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0019511129098342384,
+      "loss": 0.1244,
+      "step": 2928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.001951069533001722,
+      "loss": 0.1622,
+      "step": 2929
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019510261374165157,
+      "loss": 0.125,
+      "step": 2930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019509827230794752,
+      "loss": 0.1283,
+      "step": 2931
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.001950939289991457,
+      "loss": 0.1632,
+      "step": 2932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019508958381533165,
+      "loss": 0.1018,
+      "step": 2933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019508523675659118,
+      "loss": 0.1063,
+      "step": 2934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019508088782300987,
+      "loss": 0.1274,
+      "step": 2935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019507653701467357,
+      "loss": 0.146,
+      "step": 2936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019507218433166798,
+      "loss": 0.1339,
+      "step": 2937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019506782977407899,
+      "loss": 0.1196,
+      "step": 2938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019506347334199243,
+      "loss": 0.0969,
+      "step": 2939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019505911503549425,
+      "loss": 0.122,
+      "step": 2940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.001950547548546703,
+      "loss": 0.108,
+      "step": 2941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019505039279960662,
+      "loss": 0.0926,
+      "step": 2942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.001950460288703892,
+      "loss": 0.1134,
+      "step": 2943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019504166306710405,
+      "loss": 0.1366,
+      "step": 2944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.001950372953898373,
+      "loss": 0.092,
+      "step": 2945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019503292583867504,
+      "loss": 0.0957,
+      "step": 2946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019502855441370346,
+      "loss": 0.1388,
+      "step": 2947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019502418111500873,
+      "loss": 0.1207,
+      "step": 2948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019501980594267706,
+      "loss": 0.0919,
+      "step": 2949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019501542889679474,
+      "loss": 0.104,
+      "step": 2950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.001950110499774481,
+      "loss": 0.0764,
+      "step": 2951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019500666918472343,
+      "loss": 0.1218,
+      "step": 2952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019500228651870715,
+      "loss": 0.0989,
+      "step": 2953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019499790197948564,
+      "loss": 0.2483,
+      "step": 2954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019499351556714537,
+      "loss": 0.1152,
+      "step": 2955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019498912728177285,
+      "loss": 0.1158,
+      "step": 2956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019498473712345457,
+      "loss": 0.1094,
+      "step": 2957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.001949803450922771,
+      "loss": 0.1194,
+      "step": 2958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019497595118832707,
+      "loss": 0.1267,
+      "step": 2959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019497155541169109,
+      "loss": 0.0894,
+      "step": 2960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019496715776245585,
+      "loss": 0.1025,
+      "step": 2961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019496275824070802,
+      "loss": 0.1265,
+      "step": 2962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019495835684653438,
+      "loss": 0.1514,
+      "step": 2963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019495395358002173,
+      "loss": 0.1177,
+      "step": 2964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019494954844125685,
+      "loss": 0.1274,
+      "step": 2965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019494514143032662,
+      "loss": 0.1228,
+      "step": 2966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019494073254731794,
+      "loss": 0.1206,
+      "step": 2967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019493632179231777,
+      "loss": 0.1384,
+      "step": 2968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019493190916541302,
+      "loss": 0.1332,
+      "step": 2969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019492749466669072,
+      "loss": 0.1399,
+      "step": 2970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019492307829623789,
+      "loss": 0.0839,
+      "step": 2971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019491866005414166,
+      "loss": 0.1273,
+      "step": 2972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019491423994048912,
+      "loss": 0.1117,
+      "step": 2973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019490981795536745,
+      "loss": 0.1763,
+      "step": 2974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0019490539409886378,
+      "loss": 0.1153,
+      "step": 2975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019490096837106537,
+      "loss": 0.115,
+      "step": 2976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001948965407720595,
+      "loss": 0.1383,
+      "step": 2977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019489211130193347,
+      "loss": 0.0924,
+      "step": 2978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019488767996077457,
+      "loss": 0.092,
+      "step": 2979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019488324674867023,
+      "loss": 0.1543,
+      "step": 2980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019487881166570787,
+      "loss": 0.144,
+      "step": 2981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019487437471197486,
+      "loss": 0.0864,
+      "step": 2982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019486993588755876,
+      "loss": 0.2546,
+      "step": 2983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001948654951925471,
+      "loss": 0.1133,
+      "step": 2984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019486105262702736,
+      "loss": 0.1085,
+      "step": 2985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019485660819108718,
+      "loss": 0.1731,
+      "step": 2986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019485216188481424,
+      "loss": 0.1541,
+      "step": 2987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019484771370829612,
+      "loss": 0.1644,
+      "step": 2988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019484326366162059,
+      "loss": 0.1072,
+      "step": 2989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001948388117448754,
+      "loss": 0.1531,
+      "step": 2990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019483435795814828,
+      "loss": 0.151,
+      "step": 2991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001948299023015271,
+      "loss": 0.0732,
+      "step": 2992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019482544477509966,
+      "loss": 0.121,
+      "step": 2993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019482098537895393,
+      "loss": 0.1013,
+      "step": 2994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019481652411317775,
+      "loss": 0.1019,
+      "step": 2995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019481206097785914,
+      "loss": 0.1312,
+      "step": 2996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019480759597308606,
+      "loss": 0.1763,
+      "step": 2997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001948031290989466,
+      "loss": 0.1556,
+      "step": 2998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001947986603555288,
+      "loss": 0.2117,
+      "step": 2999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001947941897429208,
+      "loss": 0.1437,
+      "step": 3000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001947897172612107,
+      "loss": 0.1186,
+      "step": 3001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019478524291048674,
+      "loss": 0.1676,
+      "step": 3002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001947807666908371,
+      "loss": 0.131,
+      "step": 3003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.001947762886023501,
+      "loss": 0.1161,
+      "step": 3004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019477180864511392,
+      "loss": 0.0994,
+      "step": 3005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019476732681921703,
+      "loss": 0.1118,
+      "step": 3006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019476284312474773,
+      "loss": 0.131,
+      "step": 3007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019475835756179443,
+      "loss": 0.1115,
+      "step": 3008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019475387013044559,
+      "loss": 0.1857,
+      "step": 3009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019474938083078968,
+      "loss": 0.1296,
+      "step": 3010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019474488966291523,
+      "loss": 0.1062,
+      "step": 3011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019474039662691076,
+      "loss": 0.1317,
+      "step": 3012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019473590172286491,
+      "loss": 0.1383,
+      "step": 3013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019473140495086625,
+      "loss": 0.136,
+      "step": 3014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019472690631100352,
+      "loss": 0.0975,
+      "step": 3015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019472240580336536,
+      "loss": 0.1559,
+      "step": 3016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019471790342804054,
+      "loss": 0.1078,
+      "step": 3017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019471339918511782,
+      "loss": 0.1348,
+      "step": 3018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00194708893074686,
+      "loss": 0.1385,
+      "step": 3019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019470438509683397,
+      "loss": 0.1383,
+      "step": 3020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0019469987525165058,
+      "loss": 0.0966,
+      "step": 3021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019469536353922475,
+      "loss": 0.1288,
+      "step": 3022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019469084995964547,
+      "loss": 0.1571,
+      "step": 3023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019468633451300173,
+      "loss": 0.0989,
+      "step": 3024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019468181719938253,
+      "loss": 0.1309,
+      "step": 3025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019467729801887698,
+      "loss": 0.1646,
+      "step": 3026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001946727769715742,
+      "loss": 0.1383,
+      "step": 3027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019466825405756326,
+      "loss": 0.118,
+      "step": 3028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019466372927693341,
+      "loss": 0.1476,
+      "step": 3029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001946592026297738,
+      "loss": 0.0765,
+      "step": 3030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019465467411617378,
+      "loss": 0.105,
+      "step": 3031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019465014373622256,
+      "loss": 0.1084,
+      "step": 3032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001946456114900095,
+      "loss": 0.0857,
+      "step": 3033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019464107737762397,
+      "loss": 0.1364,
+      "step": 3034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019463654139915535,
+      "loss": 0.1433,
+      "step": 3035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019463200355469307,
+      "loss": 0.155,
+      "step": 3036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019462746384432666,
+      "loss": 0.1577,
+      "step": 3037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019462292226814558,
+      "loss": 0.151,
+      "step": 3038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019461837882623937,
+      "loss": 0.109,
+      "step": 3039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019461383351869767,
+      "loss": 0.0671,
+      "step": 3040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019460928634561006,
+      "loss": 0.065,
+      "step": 3041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001946047373070662,
+      "loss": 0.136,
+      "step": 3042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019460018640315583,
+      "loss": 0.0875,
+      "step": 3043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001945956336339686,
+      "loss": 0.1412,
+      "step": 3044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019459107899959434,
+      "loss": 0.111,
+      "step": 3045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019458652250012286,
+      "loss": 0.1112,
+      "step": 3046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019458196413564396,
+      "loss": 0.1105,
+      "step": 3047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019457740390624756,
+      "loss": 0.1329,
+      "step": 3048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019457284181202358,
+      "loss": 0.1465,
+      "step": 3049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019456827785306193,
+      "loss": 0.1444,
+      "step": 3050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001945637120294526,
+      "loss": 0.2026,
+      "step": 3051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019455914434128568,
+      "loss": 0.08,
+      "step": 3052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019455457478865117,
+      "loss": 0.0758,
+      "step": 3053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001945500033716392,
+      "loss": 0.0929,
+      "step": 3054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019454543009033989,
+      "loss": 0.2057,
+      "step": 3055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019454085494484343,
+      "loss": 0.1124,
+      "step": 3056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019453627793524005,
+      "loss": 0.0951,
+      "step": 3057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019453169906161994,
+      "loss": 0.1122,
+      "step": 3058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019452711832407344,
+      "loss": 0.1886,
+      "step": 3059
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019452253572269083,
+      "loss": 0.1542,
+      "step": 3060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019451795125756247,
+      "loss": 0.092,
+      "step": 3061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019451336492877877,
+      "loss": 0.1331,
+      "step": 3062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.001945087767364302,
+      "loss": 0.1124,
+      "step": 3063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019450418668060711,
+      "loss": 0.1533,
+      "step": 3064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019449959476140013,
+      "loss": 0.1011,
+      "step": 3065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019449500097889977,
+      "loss": 0.1057,
+      "step": 3066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0019449040533319653,
+      "loss": 0.1348,
+      "step": 3067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019448580782438112,
+      "loss": 0.1061,
+      "step": 3068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019448120845254415,
+      "loss": 0.1212,
+      "step": 3069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001944766072177763,
+      "loss": 0.1319,
+      "step": 3070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019447200412016832,
+      "loss": 0.1606,
+      "step": 3071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019446739915981095,
+      "loss": 0.1619,
+      "step": 3072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00194462792336795,
+      "loss": 0.2441,
+      "step": 3073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019445818365121127,
+      "loss": 0.1733,
+      "step": 3074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001944535731031507,
+      "loss": 0.1312,
+      "step": 3075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019444896069270418,
+      "loss": 0.1487,
+      "step": 3076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019444434641996261,
+      "loss": 0.2373,
+      "step": 3077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00194439730285017,
+      "loss": 0.1458,
+      "step": 3078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019443511228795837,
+      "loss": 0.0833,
+      "step": 3079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019443049242887777,
+      "loss": 0.1105,
+      "step": 3080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001944258707078663,
+      "loss": 0.1259,
+      "step": 3081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019442124712501505,
+      "loss": 0.0784,
+      "step": 3082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019441662168041525,
+      "loss": 0.175,
+      "step": 3083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019441199437415806,
+      "loss": 0.1459,
+      "step": 3084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019440736520633473,
+      "loss": 0.0861,
+      "step": 3085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019440273417703653,
+      "loss": 0.0845,
+      "step": 3086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001943981012863548,
+      "loss": 0.1448,
+      "step": 3087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019439346653438083,
+      "loss": 0.1023,
+      "step": 3088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019438882992120607,
+      "loss": 0.1091,
+      "step": 3089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001943841914469219,
+      "loss": 0.1331,
+      "step": 3090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001943795511116198,
+      "loss": 0.1221,
+      "step": 3091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019437490891539124,
+      "loss": 0.1127,
+      "step": 3092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019437026485832779,
+      "loss": 0.1732,
+      "step": 3093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019436561894052098,
+      "loss": 0.1235,
+      "step": 3094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019436097116206246,
+      "loss": 0.1381,
+      "step": 3095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019435632152304384,
+      "loss": 0.1469,
+      "step": 3096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001943516700235568,
+      "loss": 0.1561,
+      "step": 3097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019434701666369305,
+      "loss": 0.0782,
+      "step": 3098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019434236144354438,
+      "loss": 0.1038,
+      "step": 3099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019433770436320255,
+      "loss": 0.1019,
+      "step": 3100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019433304542275937,
+      "loss": 0.1261,
+      "step": 3101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019432838462230674,
+      "loss": 0.0908,
+      "step": 3102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019432372196193652,
+      "loss": 0.0756,
+      "step": 3103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019431905744174069,
+      "loss": 0.1768,
+      "step": 3104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019431439106181122,
+      "loss": 0.2222,
+      "step": 3105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019430972282224008,
+      "loss": 0.1095,
+      "step": 3106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019430505272311934,
+      "loss": 0.1053,
+      "step": 3107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019430038076454106,
+      "loss": 0.1287,
+      "step": 3108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001942957069465974,
+      "loss": 0.1653,
+      "step": 3109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019429103126938048,
+      "loss": 0.0818,
+      "step": 3110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019428635373298251,
+      "loss": 0.1469,
+      "step": 3111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.001942816743374957,
+      "loss": 0.1533,
+      "step": 3112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0019427699308301235,
+      "loss": 0.1296,
+      "step": 3113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019427230996962473,
+      "loss": 0.0812,
+      "step": 3114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001942676249974252,
+      "loss": 0.0783,
+      "step": 3115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019426293816650615,
+      "loss": 0.134,
+      "step": 3116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019425824947695994,
+      "loss": 0.0902,
+      "step": 3117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019425355892887905,
+      "loss": 0.1228,
+      "step": 3118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019424886652235599,
+      "loss": 0.1168,
+      "step": 3119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019424417225748325,
+      "loss": 0.0931,
+      "step": 3120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019423947613435339,
+      "loss": 0.1403,
+      "step": 3121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00194234778153059,
+      "loss": 0.1124,
+      "step": 3122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019423007831369272,
+      "loss": 0.132,
+      "step": 3123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019422537661634726,
+      "loss": 0.1283,
+      "step": 3124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019422067306111529,
+      "loss": 0.2528,
+      "step": 3125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019421596764808951,
+      "loss": 0.1013,
+      "step": 3126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019421126037736278,
+      "loss": 0.1611,
+      "step": 3127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019420655124902787,
+      "loss": 0.1344,
+      "step": 3128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019420184026317762,
+      "loss": 0.1074,
+      "step": 3129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019419712741990493,
+      "loss": 0.1151,
+      "step": 3130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019419241271930276,
+      "loss": 0.1385,
+      "step": 3131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019418769616146405,
+      "loss": 0.1096,
+      "step": 3132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019418297774648177,
+      "loss": 0.177,
+      "step": 3133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00194178257474449,
+      "loss": 0.1091,
+      "step": 3134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019417353534545879,
+      "loss": 0.1193,
+      "step": 3135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019416881135960423,
+      "loss": 0.0985,
+      "step": 3136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001941640855169785,
+      "loss": 0.12,
+      "step": 3137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019415935781767473,
+      "loss": 0.1359,
+      "step": 3138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001941546282617862,
+      "loss": 0.1424,
+      "step": 3139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019414989684940617,
+      "loss": 0.2086,
+      "step": 3140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019414516358062786,
+      "loss": 0.1226,
+      "step": 3141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019414042845554464,
+      "loss": 0.1528,
+      "step": 3142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019413569147424987,
+      "loss": 0.0886,
+      "step": 3143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019413095263683697,
+      "loss": 0.1512,
+      "step": 3144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019412621194339935,
+      "loss": 0.0739,
+      "step": 3145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001941214693940305,
+      "loss": 0.1635,
+      "step": 3146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019411672498882394,
+      "loss": 0.1248,
+      "step": 3147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001941119787278732,
+      "loss": 0.1185,
+      "step": 3148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019410723061127186,
+      "loss": 0.1584,
+      "step": 3149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019410248063911357,
+      "loss": 0.1658,
+      "step": 3150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019409772881149195,
+      "loss": 0.0858,
+      "step": 3151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019409297512850072,
+      "loss": 0.1208,
+      "step": 3152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001940882195902336,
+      "loss": 0.1802,
+      "step": 3153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019408346219678436,
+      "loss": 0.1069,
+      "step": 3154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001940787029482468,
+      "loss": 0.1365,
+      "step": 3155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.001940739418447148,
+      "loss": 0.145,
+      "step": 3156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019406917888628215,
+      "loss": 0.1942,
+      "step": 3157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019406441407304282,
+      "loss": 0.1324,
+      "step": 3158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0019405964740509076,
+      "loss": 0.0791,
+      "step": 3159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019405487888251999,
+      "loss": 0.1382,
+      "step": 3160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019405010850542445,
+      "loss": 0.131,
+      "step": 3161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019404533627389827,
+      "loss": 0.1278,
+      "step": 3162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001940405621880355,
+      "loss": 0.1127,
+      "step": 3163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019403578624793032,
+      "loss": 0.1002,
+      "step": 3164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019403100845367684,
+      "loss": 0.1238,
+      "step": 3165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019402622880536934,
+      "loss": 0.1232,
+      "step": 3166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019402144730310197,
+      "loss": 0.1007,
+      "step": 3167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001940166639469691,
+      "loss": 0.1318,
+      "step": 3168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00194011878737065,
+      "loss": 0.1699,
+      "step": 3169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019400709167348403,
+      "loss": 0.1327,
+      "step": 3170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019400230275632055,
+      "loss": 0.1206,
+      "step": 3171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019399751198566906,
+      "loss": 0.1411,
+      "step": 3172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019399271936162395,
+      "loss": 0.1068,
+      "step": 3173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019398792488427975,
+      "loss": 0.142,
+      "step": 3174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019398312855373097,
+      "loss": 0.1274,
+      "step": 3175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019397833037007222,
+      "loss": 0.1267,
+      "step": 3176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939735303333981,
+      "loss": 0.1185,
+      "step": 3177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939687284438032,
+      "loss": 0.1156,
+      "step": 3178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939639247013823,
+      "loss": 0.0945,
+      "step": 3179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019395911910623002,
+      "loss": 0.1355,
+      "step": 3180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019395431165844116,
+      "loss": 0.1295,
+      "step": 3181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939495023581105,
+      "loss": 0.1556,
+      "step": 3182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939446912053329,
+      "loss": 0.1415,
+      "step": 3183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019393987820020319,
+      "loss": 0.1366,
+      "step": 3184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019393506334281626,
+      "loss": 0.1763,
+      "step": 3185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019393024663326706,
+      "loss": 0.0962,
+      "step": 3186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019392542807165059,
+      "loss": 0.1157,
+      "step": 3187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019392060765806183,
+      "loss": 0.1855,
+      "step": 3188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019391578539259583,
+      "loss": 0.1403,
+      "step": 3189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939109612753477,
+      "loss": 0.1346,
+      "step": 3190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001939061353064125,
+      "loss": 0.121,
+      "step": 3191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019390130748588544,
+      "loss": 0.123,
+      "step": 3192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001938964778138617,
+      "loss": 0.129,
+      "step": 3193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001938916462904365,
+      "loss": 0.1428,
+      "step": 3194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019388681291570512,
+      "loss": 0.1021,
+      "step": 3195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001938819776897628,
+      "loss": 0.1346,
+      "step": 3196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00193877140612705,
+      "loss": 0.1182,
+      "step": 3197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019387230168462697,
+      "loss": 0.0895,
+      "step": 3198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019386746090562421,
+      "loss": 0.0975,
+      "step": 3199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019386261827579211,
+      "loss": 0.1783,
+      "step": 3200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.001938577737952262,
+      "loss": 0.1182,
+      "step": 3201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019385292746402199,
+      "loss": 0.1433,
+      "step": 3202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00193848079282275,
+      "loss": 0.0943,
+      "step": 3203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019384322925008086,
+      "loss": 0.1407,
+      "step": 3204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019383837736753518,
+      "loss": 0.0796,
+      "step": 3205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0019383352363473365,
+      "loss": 0.0815,
+      "step": 3206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019382866805177197,
+      "loss": 0.1088,
+      "step": 3207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019382381061874585,
+      "loss": 0.0727,
+      "step": 3208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001938189513357511,
+      "loss": 0.1171,
+      "step": 3209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001938140902028835,
+      "loss": 0.1774,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019380922722023895,
+      "loss": 0.1083,
+      "step": 3211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019380436238791328,
+      "loss": 0.1906,
+      "step": 3212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019379949570600245,
+      "loss": 0.0952,
+      "step": 3213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019379462717460242,
+      "loss": 0.0753,
+      "step": 3214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019378975679380914,
+      "loss": 0.1559,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001937848845637187,
+      "loss": 0.1583,
+      "step": 3216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019378001048442712,
+      "loss": 0.1448,
+      "step": 3217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019377513455603053,
+      "loss": 0.1813,
+      "step": 3218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019377025677862507,
+      "loss": 0.1937,
+      "step": 3219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001937653771523069,
+      "loss": 0.1158,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001937604956771723,
+      "loss": 0.1265,
+      "step": 3221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019375561235331741,
+      "loss": 0.1734,
+      "step": 3222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019375072718083861,
+      "loss": 0.0961,
+      "step": 3223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019374584015983216,
+      "loss": 0.1047,
+      "step": 3224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019374095129039448,
+      "loss": 0.1367,
+      "step": 3225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001937360605726219,
+      "loss": 0.1033,
+      "step": 3226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001937311680066109,
+      "loss": 0.1406,
+      "step": 3227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019372627359245797,
+      "loss": 0.1334,
+      "step": 3228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019372137733025953,
+      "loss": 0.0993,
+      "step": 3229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019371647922011219,
+      "loss": 0.1434,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001937115792621125,
+      "loss": 0.1071,
+      "step": 3231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019370667745635714,
+      "loss": 0.1506,
+      "step": 3232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019370177380294265,
+      "loss": 0.1387,
+      "step": 3233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001936968683019658,
+      "loss": 0.1282,
+      "step": 3234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019369196095352327,
+      "loss": 0.1049,
+      "step": 3235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019368705175771186,
+      "loss": 0.1086,
+      "step": 3236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019368214071462835,
+      "loss": 0.1359,
+      "step": 3237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019367722782436957,
+      "loss": 0.1404,
+      "step": 3238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019367231308703237,
+      "loss": 0.1106,
+      "step": 3239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019366739650271368,
+      "loss": 0.1487,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019366247807151042,
+      "loss": 0.1796,
+      "step": 3241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019365755779351963,
+      "loss": 0.1602,
+      "step": 3242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019365263566883827,
+      "loss": 0.1553,
+      "step": 3243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001936477116975634,
+      "loss": 0.1121,
+      "step": 3244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019364278587979211,
+      "loss": 0.1077,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019363785821562155,
+      "loss": 0.1331,
+      "step": 3246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019363292870514881,
+      "loss": 0.111,
+      "step": 3247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019362799734847118,
+      "loss": 0.1295,
+      "step": 3248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019362306414568585,
+      "loss": 0.0784,
+      "step": 3249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.001936181290968901,
+      "loss": 0.151,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019361319220218118,
+      "loss": 0.1715,
+      "step": 3251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0019360825346165652,
+      "loss": 0.13,
+      "step": 3252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019360331287541345,
+      "loss": 0.12,
+      "step": 3253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019359837044354937,
+      "loss": 0.142,
+      "step": 3254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001935934261661618,
+      "loss": 0.1301,
+      "step": 3255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019358848004334818,
+      "loss": 0.1901,
+      "step": 3256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019358353207520602,
+      "loss": 0.1965,
+      "step": 3257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019357858226183291,
+      "loss": 0.136,
+      "step": 3258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019357363060332645,
+      "loss": 0.1379,
+      "step": 3259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019356867709978427,
+      "loss": 0.1218,
+      "step": 3260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019356372175130404,
+      "loss": 0.1073,
+      "step": 3261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019355876455798345,
+      "loss": 0.0816,
+      "step": 3262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019355380551992023,
+      "loss": 0.1401,
+      "step": 3263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019354884463721224,
+      "loss": 0.1755,
+      "step": 3264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019354388190995723,
+      "loss": 0.1116,
+      "step": 3265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019353891733825304,
+      "loss": 0.1115,
+      "step": 3266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001935339509221976,
+      "loss": 0.1394,
+      "step": 3267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019352898266188881,
+      "loss": 0.1126,
+      "step": 3268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019352401255742465,
+      "loss": 0.129,
+      "step": 3269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001935190406089031,
+      "loss": 0.1205,
+      "step": 3270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019351406681642218,
+      "loss": 0.1245,
+      "step": 3271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019350909118008002,
+      "loss": 0.1066,
+      "step": 3272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019350411369997468,
+      "loss": 0.1838,
+      "step": 3273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001934991343762043,
+      "loss": 0.1128,
+      "step": 3274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001934941532088671,
+      "loss": 0.0667,
+      "step": 3275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019348917019806125,
+      "loss": 0.1335,
+      "step": 3276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019348418534388503,
+      "loss": 0.1211,
+      "step": 3277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001934791986464367,
+      "loss": 0.1259,
+      "step": 3278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019347421010581464,
+      "loss": 0.1594,
+      "step": 3279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019346921972211715,
+      "loss": 0.1412,
+      "step": 3280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019346422749544265,
+      "loss": 0.1367,
+      "step": 3281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019345923342588961,
+      "loss": 0.1272,
+      "step": 3282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019345423751355644,
+      "loss": 0.0778,
+      "step": 3283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019344923975854167,
+      "loss": 0.0679,
+      "step": 3284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019344424016094388,
+      "loss": 0.0754,
+      "step": 3285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001934392387208616,
+      "loss": 0.1313,
+      "step": 3286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019343423543839347,
+      "loss": 0.1254,
+      "step": 3287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019342923031363812,
+      "loss": 0.0989,
+      "step": 3288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019342422334669427,
+      "loss": 0.0814,
+      "step": 3289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019341921453766063,
+      "loss": 0.1169,
+      "step": 3290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019341420388663596,
+      "loss": 0.0753,
+      "step": 3291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019340919139371904,
+      "loss": 0.1129,
+      "step": 3292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019340417705900875,
+      "loss": 0.0876,
+      "step": 3293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019339916088260392,
+      "loss": 0.1299,
+      "step": 3294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019339414286460347,
+      "loss": 0.1002,
+      "step": 3295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0019338912300510631,
+      "loss": 0.1229,
+      "step": 3296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.001933841013042115,
+      "loss": 0.0894,
+      "step": 3297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00193379077762018,
+      "loss": 0.1219,
+      "step": 3298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019337405237862485,
+      "loss": 0.1251,
+      "step": 3299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019336902515413113,
+      "loss": 0.1451,
+      "step": 3300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00193363996088636,
+      "loss": 0.1199,
+      "step": 3301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019335896518223863,
+      "loss": 0.1594,
+      "step": 3302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019335393243503818,
+      "loss": 0.1066,
+      "step": 3303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019334889784713392,
+      "loss": 0.0757,
+      "step": 3304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019334386141862505,
+      "loss": 0.0775,
+      "step": 3305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019333882314961097,
+      "loss": 0.1425,
+      "step": 3306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019333378304019098,
+      "loss": 0.1027,
+      "step": 3307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019332874109046442,
+      "loss": 0.1029,
+      "step": 3308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019332369730053075,
+      "loss": 0.1353,
+      "step": 3309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.001933186516704894,
+      "loss": 0.1206,
+      "step": 3310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019331360420043989,
+      "loss": 0.1233,
+      "step": 3311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019330855489048172,
+      "loss": 0.0784,
+      "step": 3312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019330350374071443,
+      "loss": 0.072,
+      "step": 3313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019329845075123765,
+      "loss": 0.0973,
+      "step": 3314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00193293395922151,
+      "loss": 0.1855,
+      "step": 3315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019328833925355412,
+      "loss": 0.0842,
+      "step": 3316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019328328074554675,
+      "loss": 0.0972,
+      "step": 3317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019327822039822862,
+      "loss": 0.1447,
+      "step": 3318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019327315821169953,
+      "loss": 0.0941,
+      "step": 3319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019326809418605925,
+      "loss": 0.1261,
+      "step": 3320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019326302832140768,
+      "loss": 0.1313,
+      "step": 3321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019325796061784467,
+      "loss": 0.1102,
+      "step": 3322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019325289107547012,
+      "loss": 0.1361,
+      "step": 3323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019324781969438404,
+      "loss": 0.1405,
+      "step": 3324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019324274647468643,
+      "loss": 0.0839,
+      "step": 3325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019323767141647724,
+      "loss": 0.1321,
+      "step": 3326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019323259451985663,
+      "loss": 0.1656,
+      "step": 3327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019322751578492465,
+      "loss": 0.1093,
+      "step": 3328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019322243521178147,
+      "loss": 0.1263,
+      "step": 3329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019321735280052727,
+      "loss": 0.1284,
+      "step": 3330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.001932122685512622,
+      "loss": 0.1361,
+      "step": 3331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019320718246408656,
+      "loss": 0.0905,
+      "step": 3332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019320209453910066,
+      "loss": 0.1079,
+      "step": 3333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019319700477640476,
+      "loss": 0.166,
+      "step": 3334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019319191317609926,
+      "loss": 0.1061,
+      "step": 3335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019318681973828453,
+      "loss": 0.1156,
+      "step": 3336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019318172446306102,
+      "loss": 0.1119,
+      "step": 3337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019317662735052917,
+      "loss": 0.15,
+      "step": 3338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019317152840078952,
+      "loss": 0.1561,
+      "step": 3339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019316642761394256,
+      "loss": 0.0936,
+      "step": 3340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.001931613249900889,
+      "loss": 0.22,
+      "step": 3341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019315622052932912,
+      "loss": 0.0818,
+      "step": 3342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019315111423176396,
+      "loss": 0.1169,
+      "step": 3343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0019314600609749396,
+      "loss": 0.1184,
+      "step": 3344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019314089612661994,
+      "loss": 0.1246,
+      "step": 3345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019313578431924261,
+      "loss": 0.1008,
+      "step": 3346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001931306706754628,
+      "loss": 0.1359,
+      "step": 3347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019312555519538132,
+      "loss": 0.2112,
+      "step": 3348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019312043787909898,
+      "loss": 0.0704,
+      "step": 3349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019311531872671676,
+      "loss": 0.1669,
+      "step": 3350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001931101977383356,
+      "loss": 0.1183,
+      "step": 3351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019310507491405642,
+      "loss": 0.1224,
+      "step": 3352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019309995025398026,
+      "loss": 0.0749,
+      "step": 3353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019309482375820814,
+      "loss": 0.1288,
+      "step": 3354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019308969542684116,
+      "loss": 0.1548,
+      "step": 3355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019308456525998043,
+      "loss": 0.1587,
+      "step": 3356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019307943325772713,
+      "loss": 0.0912,
+      "step": 3357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019307429942018242,
+      "loss": 0.1219,
+      "step": 3358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019306916374744753,
+      "loss": 0.1323,
+      "step": 3359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019306402623962373,
+      "loss": 0.1282,
+      "step": 3360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019305888689681232,
+      "loss": 0.1027,
+      "step": 3361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019305374571911462,
+      "loss": 0.1453,
+      "step": 3362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019304860270663205,
+      "loss": 0.1566,
+      "step": 3363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019304345785946593,
+      "loss": 0.0763,
+      "step": 3364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019303831117771777,
+      "loss": 0.1705,
+      "step": 3365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019303316266148904,
+      "loss": 0.1698,
+      "step": 3366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019302801231088128,
+      "loss": 0.1436,
+      "step": 3367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019302286012599598,
+      "loss": 0.1345,
+      "step": 3368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019301770610693474,
+      "loss": 0.1658,
+      "step": 3369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001930125502537992,
+      "loss": 0.1411,
+      "step": 3370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019300739256669106,
+      "loss": 0.2321,
+      "step": 3371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019300223304571196,
+      "loss": 0.1691,
+      "step": 3372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019299707169096365,
+      "loss": 0.1357,
+      "step": 3373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001929919085025479,
+      "loss": 0.1809,
+      "step": 3374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019298674348056655,
+      "loss": 0.0941,
+      "step": 3375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019298157662512136,
+      "loss": 0.123,
+      "step": 3376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019297640793631428,
+      "loss": 0.2239,
+      "step": 3377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001929712374142472,
+      "loss": 0.1098,
+      "step": 3378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019296606505902205,
+      "loss": 0.1078,
+      "step": 3379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019296089087074088,
+      "loss": 0.1816,
+      "step": 3380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019295571484950562,
+      "loss": 0.1569,
+      "step": 3381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001929505369954184,
+      "loss": 0.1117,
+      "step": 3382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019294535730858128,
+      "loss": 0.1162,
+      "step": 3383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019294017578909638,
+      "loss": 0.103,
+      "step": 3384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001929349924370659,
+      "loss": 0.1976,
+      "step": 3385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019292980725259202,
+      "loss": 0.1285,
+      "step": 3386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019292462023577702,
+      "loss": 0.1841,
+      "step": 3387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.001929194313867231,
+      "loss": 0.116,
+      "step": 3388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019291424070553263,
+      "loss": 0.2593,
+      "step": 3389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0019290904819230792,
+      "loss": 0.1443,
+      "step": 3390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019290385384715138,
+      "loss": 0.1191,
+      "step": 3391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019289865767016545,
+      "loss": 0.1627,
+      "step": 3392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019289345966145253,
+      "loss": 0.1489,
+      "step": 3393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019288825982111514,
+      "loss": 0.1348,
+      "step": 3394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019288305814925582,
+      "loss": 0.0761,
+      "step": 3395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019287785464597712,
+      "loss": 0.126,
+      "step": 3396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019287264931138165,
+      "loss": 0.1259,
+      "step": 3397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019286744214557202,
+      "loss": 0.1472,
+      "step": 3398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019286223314865093,
+      "loss": 0.0954,
+      "step": 3399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019285702232072106,
+      "loss": 0.1105,
+      "step": 3400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019285180966188518,
+      "loss": 0.109,
+      "step": 3401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019284659517224605,
+      "loss": 0.2114,
+      "step": 3402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001928413788519065,
+      "loss": 0.1213,
+      "step": 3403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019283616070096944,
+      "loss": 0.1112,
+      "step": 3404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019283094071953762,
+      "loss": 0.075,
+      "step": 3405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001928257189077141,
+      "loss": 0.104,
+      "step": 3406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019282049526560174,
+      "loss": 0.092,
+      "step": 3407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001928152697933036,
+      "loss": 0.1075,
+      "step": 3408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001928100424909227,
+      "loss": 0.0933,
+      "step": 3409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019280481335856212,
+      "loss": 0.1572,
+      "step": 3410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019279958239632492,
+      "loss": 0.1801,
+      "step": 3411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001927943496043143,
+      "loss": 0.0931,
+      "step": 3412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019278911498263338,
+      "loss": 0.1318,
+      "step": 3413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019278387853138544,
+      "loss": 0.149,
+      "step": 3414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019277864025067366,
+      "loss": 0.1731,
+      "step": 3415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001927734001406014,
+      "loss": 0.1547,
+      "step": 3416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001927681582012719,
+      "loss": 0.1548,
+      "step": 3417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019276291443278857,
+      "loss": 0.0804,
+      "step": 3418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019275766883525478,
+      "loss": 0.1182,
+      "step": 3419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019275242140877398,
+      "loss": 0.1006,
+      "step": 3420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019274717215344963,
+      "loss": 0.1012,
+      "step": 3421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019274192106938522,
+      "loss": 0.1036,
+      "step": 3422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019273666815668431,
+      "loss": 0.086,
+      "step": 3423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019273141341545046,
+      "loss": 0.1359,
+      "step": 3424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001927261568457873,
+      "loss": 0.1124,
+      "step": 3425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001927208984477984,
+      "loss": 0.1326,
+      "step": 3426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019271563822158755,
+      "loss": 0.1251,
+      "step": 3427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001927103761672584,
+      "loss": 0.1234,
+      "step": 3428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019270511228491475,
+      "loss": 0.1479,
+      "step": 3429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019269984657466033,
+      "loss": 0.1077,
+      "step": 3430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00192694579036599,
+      "loss": 0.1729,
+      "step": 3431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019268930967083465,
+      "loss": 0.1382,
+      "step": 3432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019268403847747114,
+      "loss": 0.1677,
+      "step": 3433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.001926787654566124,
+      "loss": 0.1593,
+      "step": 3434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019267349060836244,
+      "loss": 0.2001,
+      "step": 3435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0019266821393282526,
+      "loss": 0.1661,
+      "step": 3436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019266293543010482,
+      "loss": 0.1956,
+      "step": 3437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019265765510030532,
+      "loss": 0.0844,
+      "step": 3438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019265237294353082,
+      "loss": 0.1915,
+      "step": 3439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019264708895988542,
+      "loss": 0.1417,
+      "step": 3440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019264180314947342,
+      "loss": 0.1334,
+      "step": 3441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019263651551239894,
+      "loss": 0.0751,
+      "step": 3442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.001926312260487663,
+      "loss": 0.1438,
+      "step": 3443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019262593475867975,
+      "loss": 0.1158,
+      "step": 3444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019262064164224366,
+      "loss": 0.114,
+      "step": 3445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.001926153466995624,
+      "loss": 0.1118,
+      "step": 3446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019261004993074034,
+      "loss": 0.0994,
+      "step": 3447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019260475133588194,
+      "loss": 0.0879,
+      "step": 3448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019259945091509164,
+      "loss": 0.1326,
+      "step": 3449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00192594148668474,
+      "loss": 0.1105,
+      "step": 3450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019258884459613355,
+      "loss": 0.0585,
+      "step": 3451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019258353869817486,
+      "loss": 0.1602,
+      "step": 3452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019257823097470258,
+      "loss": 0.1232,
+      "step": 3453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.001925729214258213,
+      "loss": 0.0978,
+      "step": 3454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019256761005163577,
+      "loss": 0.1002,
+      "step": 3455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019256229685225073,
+      "loss": 0.0849,
+      "step": 3456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019255698182777089,
+      "loss": 0.1355,
+      "step": 3457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019255166497830105,
+      "loss": 0.0828,
+      "step": 3458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019254634630394612,
+      "loss": 0.1243,
+      "step": 3459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019254102580481087,
+      "loss": 0.1405,
+      "step": 3460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.001925357034810003,
+      "loss": 0.1372,
+      "step": 3461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019253037933261927,
+      "loss": 0.1176,
+      "step": 3462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.001925250533597728,
+      "loss": 0.0975,
+      "step": 3463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019251972556256591,
+      "loss": 0.1462,
+      "step": 3464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019251439594110365,
+      "loss": 0.1365,
+      "step": 3465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019250906449549108,
+      "loss": 0.079,
+      "step": 3466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019250373122583337,
+      "loss": 0.0701,
+      "step": 3467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.001924983961322356,
+      "loss": 0.1016,
+      "step": 3468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019249305921480305,
+      "loss": 0.1064,
+      "step": 3469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019248772047364092,
+      "loss": 0.1427,
+      "step": 3470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019248237990885447,
+      "loss": 0.0989,
+      "step": 3471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019247703752054899,
+      "loss": 0.076,
+      "step": 3472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019247169330882984,
+      "loss": 0.0885,
+      "step": 3473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019246634727380237,
+      "loss": 0.0821,
+      "step": 3474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019246099941557203,
+      "loss": 0.129,
+      "step": 3475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019245564973424424,
+      "loss": 0.1304,
+      "step": 3476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019245029822992448,
+      "loss": 0.2061,
+      "step": 3477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019244494490271826,
+      "loss": 0.0879,
+      "step": 3478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019243958975273118,
+      "loss": 0.1283,
+      "step": 3479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019243423278006877,
+      "loss": 0.1313,
+      "step": 3480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019242887398483667,
+      "loss": 0.0902,
+      "step": 3481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0019242351336714058,
+      "loss": 0.093,
+      "step": 3482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019241815092708619,
+      "loss": 0.0833,
+      "step": 3483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019241278666477922,
+      "loss": 0.0928,
+      "step": 3484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001924074205803254,
+      "loss": 0.093,
+      "step": 3485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001924020526738306,
+      "loss": 0.1128,
+      "step": 3486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923966829454006,
+      "loss": 0.1029,
+      "step": 3487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019239131139514134,
+      "loss": 0.1755,
+      "step": 3488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923859380231587,
+      "loss": 0.179,
+      "step": 3489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019238056282955866,
+      "loss": 0.1281,
+      "step": 3490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019237518581444713,
+      "loss": 0.1002,
+      "step": 3491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923698069779302,
+      "loss": 0.1226,
+      "step": 3492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923644263201139,
+      "loss": 0.1677,
+      "step": 3493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019235904384110436,
+      "loss": 0.0905,
+      "step": 3494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019235365954100767,
+      "loss": 0.0963,
+      "step": 3495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019234827341993,
+      "loss": 0.1211,
+      "step": 3496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019234288547797754,
+      "loss": 0.1377,
+      "step": 3497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019233749571525656,
+      "loss": 0.1355,
+      "step": 3498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923321041318733,
+      "loss": 0.1122,
+      "step": 3499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923267107279341,
+      "loss": 0.1167,
+      "step": 3500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019232131550354528,
+      "loss": 0.0615,
+      "step": 3501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019231591845881321,
+      "loss": 0.1103,
+      "step": 3502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019231051959384432,
+      "loss": 0.1268,
+      "step": 3503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001923051189087451,
+      "loss": 0.135,
+      "step": 3504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00192299716403622,
+      "loss": 0.0973,
+      "step": 3505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019229431207858153,
+      "loss": 0.1169,
+      "step": 3506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019228890593373028,
+      "loss": 0.1567,
+      "step": 3507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001922834979691748,
+      "loss": 0.0923,
+      "step": 3508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001922780881850218,
+      "loss": 0.116,
+      "step": 3509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019227267658137785,
+      "loss": 0.1071,
+      "step": 3510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019226726315834976,
+      "loss": 0.1158,
+      "step": 3511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019226184791604419,
+      "loss": 0.1521,
+      "step": 3512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019225643085456791,
+      "loss": 0.0693,
+      "step": 3513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019225101197402781,
+      "loss": 0.1136,
+      "step": 3514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019224559127453062,
+      "loss": 0.1226,
+      "step": 3515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019224016875618334,
+      "loss": 0.0783,
+      "step": 3516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019223474441909284,
+      "loss": 0.0881,
+      "step": 3517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019222931826336604,
+      "loss": 0.0812,
+      "step": 3518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019222389028910997,
+      "loss": 0.141,
+      "step": 3519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019221846049643167,
+      "loss": 0.0933,
+      "step": 3520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019221302888543814,
+      "loss": 0.104,
+      "step": 3521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019220759545623652,
+      "loss": 0.0905,
+      "step": 3522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019220216020893396,
+      "loss": 0.1315,
+      "step": 3523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019219672314363758,
+      "loss": 0.0577,
+      "step": 3524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019219128426045465,
+      "loss": 0.2059,
+      "step": 3525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0019218584355949237,
+      "loss": 0.1116,
+      "step": 3526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00192180401040858,
+      "loss": 0.0867,
+      "step": 3527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.001921749567046589,
+      "loss": 0.1313,
+      "step": 3528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019216951055100236,
+      "loss": 0.0738,
+      "step": 3529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019216406257999584,
+      "loss": 0.0751,
+      "step": 3530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019215861279174669,
+      "loss": 0.1101,
+      "step": 3531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.001921531611863624,
+      "loss": 0.1198,
+      "step": 3532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019214770776395046,
+      "loss": 0.1108,
+      "step": 3533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019214225252461837,
+      "loss": 0.1565,
+      "step": 3534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019213679546847375,
+      "loss": 0.2145,
+      "step": 3535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019213133659562415,
+      "loss": 0.1244,
+      "step": 3536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019212587590617723,
+      "loss": 0.108,
+      "step": 3537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019212041340024065,
+      "loss": 0.087,
+      "step": 3538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019211494907792211,
+      "loss": 0.1683,
+      "step": 3539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019210948293932938,
+      "loss": 0.0665,
+      "step": 3540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019210401498457018,
+      "loss": 0.0503,
+      "step": 3541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.001920985452137524,
+      "loss": 0.1287,
+      "step": 3542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019209307362698384,
+      "loss": 0.1685,
+      "step": 3543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019208760022437243,
+      "loss": 0.1245,
+      "step": 3544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019208212500602601,
+      "loss": 0.1235,
+      "step": 3545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019207664797205265,
+      "loss": 0.1323,
+      "step": 3546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019207116912256024,
+      "loss": 0.2104,
+      "step": 3547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019206568845765686,
+      "loss": 0.0904,
+      "step": 3548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.001920602059774506,
+      "loss": 0.1674,
+      "step": 3549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.001920547216820495,
+      "loss": 0.0822,
+      "step": 3550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019204923557156173,
+      "loss": 0.1538,
+      "step": 3551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019204374764609546,
+      "loss": 0.1085,
+      "step": 3552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019203825790575885,
+      "loss": 0.1549,
+      "step": 3553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019203276635066024,
+      "loss": 0.0831,
+      "step": 3554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019202727298090784,
+      "loss": 0.1326,
+      "step": 3555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019202177779661,
+      "loss": 0.0987,
+      "step": 3556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019201628079787503,
+      "loss": 0.1226,
+      "step": 3557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019201078198481138,
+      "loss": 0.0818,
+      "step": 3558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.001920052813575274,
+      "loss": 0.12,
+      "step": 3559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019199977891613158,
+      "loss": 0.1095,
+      "step": 3560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019199427466073244,
+      "loss": 0.1251,
+      "step": 3561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019198876859143846,
+      "loss": 0.1704,
+      "step": 3562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019198326070835826,
+      "loss": 0.1466,
+      "step": 3563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.001919777510116004,
+      "loss": 0.1244,
+      "step": 3564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019197223950127352,
+      "loss": 0.0922,
+      "step": 3565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019196672617748632,
+      "loss": 0.0858,
+      "step": 3566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019196121104034748,
+      "loss": 0.0883,
+      "step": 3567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019195569408996579,
+      "loss": 0.1296,
+      "step": 3568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019195017532644997,
+      "loss": 0.1064,
+      "step": 3569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019194465474990886,
+      "loss": 0.1479,
+      "step": 3570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019193913236045133,
+      "loss": 0.1592,
+      "step": 3571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019193360815818626,
+      "loss": 0.146,
+      "step": 3572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019192808214322254,
+      "loss": 0.0941,
+      "step": 3573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019192255431566918,
+      "loss": 0.1187,
+      "step": 3574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0019191702467563515,
+      "loss": 0.1266,
+      "step": 3575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019191149322322948,
+      "loss": 0.1375,
+      "step": 3576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019190595995856121,
+      "loss": 0.1188,
+      "step": 3577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.001919004248817395,
+      "loss": 0.1138,
+      "step": 3578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019189488799287347,
+      "loss": 0.1003,
+      "step": 3579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019188934929207228,
+      "loss": 0.0874,
+      "step": 3580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.001918838087794451,
+      "loss": 0.1536,
+      "step": 3581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019187826645510123,
+      "loss": 0.1382,
+      "step": 3582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019187272231914996,
+      "loss": 0.0802,
+      "step": 3583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019186717637170055,
+      "loss": 0.0954,
+      "step": 3584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.001918616286128624,
+      "loss": 0.129,
+      "step": 3585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.001918560790427449,
+      "loss": 0.0842,
+      "step": 3586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019185052766145744,
+      "loss": 0.1104,
+      "step": 3587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019184497446910948,
+      "loss": 0.1379,
+      "step": 3588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019183941946581053,
+      "loss": 0.0815,
+      "step": 3589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019183386265167015,
+      "loss": 0.126,
+      "step": 3590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019182830402679784,
+      "loss": 0.1367,
+      "step": 3591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019182274359130327,
+      "loss": 0.1418,
+      "step": 3592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019181718134529606,
+      "loss": 0.104,
+      "step": 3593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019181161728888583,
+      "loss": 0.098,
+      "step": 3594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019180605142218234,
+      "loss": 0.1075,
+      "step": 3595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019180048374529536,
+      "loss": 0.1147,
+      "step": 3596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019179491425833457,
+      "loss": 0.1355,
+      "step": 3597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019178934296140988,
+      "loss": 0.1105,
+      "step": 3598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019178376985463112,
+      "loss": 0.1176,
+      "step": 3599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019177819493810817,
+      "loss": 0.1267,
+      "step": 3600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019177261821195096,
+      "loss": 0.1045,
+      "step": 3601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019176703967626943,
+      "loss": 0.1464,
+      "step": 3602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019176145933117359,
+      "loss": 0.1395,
+      "step": 3603
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019175587717677348,
+      "loss": 0.1342,
+      "step": 3604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019175029321317912,
+      "loss": 0.167,
+      "step": 3605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019174470744050068,
+      "loss": 0.1327,
+      "step": 3606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019173911985884825,
+      "loss": 0.1366,
+      "step": 3607
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019173353046833199,
+      "loss": 0.1233,
+      "step": 3608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019172793926906216,
+      "loss": 0.0883,
+      "step": 3609
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00191722346261149,
+      "loss": 0.2186,
+      "step": 3610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019171675144470273,
+      "loss": 0.1487,
+      "step": 3611
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019171115481983372,
+      "loss": 0.0986,
+      "step": 3612
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.001917055563866523,
+      "loss": 0.1107,
+      "step": 3613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019169995614526886,
+      "loss": 0.0971,
+      "step": 3614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019169435409579383,
+      "loss": 0.1661,
+      "step": 3615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019168875023833768,
+      "loss": 0.1514,
+      "step": 3616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019168314457301086,
+      "loss": 0.165,
+      "step": 3617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019167753709992393,
+      "loss": 0.1117,
+      "step": 3618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019167192781918746,
+      "loss": 0.1656,
+      "step": 3619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019166631673091204,
+      "loss": 0.0991,
+      "step": 3620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0019166070383520833,
+      "loss": 0.1415,
+      "step": 3621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019165508913218693,
+      "loss": 0.1283,
+      "step": 3622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019164947262195863,
+      "loss": 0.1469,
+      "step": 3623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019164385430463415,
+      "loss": 0.1997,
+      "step": 3624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019163823418032427,
+      "loss": 0.2092,
+      "step": 3625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019163261224913978,
+      "loss": 0.0863,
+      "step": 3626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019162698851119156,
+      "loss": 0.1478,
+      "step": 3627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019162136296659044,
+      "loss": 0.0955,
+      "step": 3628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019161573561544743,
+      "loss": 0.1964,
+      "step": 3629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019161010645787342,
+      "loss": 0.1254,
+      "step": 3630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019160447549397943,
+      "loss": 0.11,
+      "step": 3631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001915988427238765,
+      "loss": 0.1744,
+      "step": 3632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019159320814767565,
+      "loss": 0.1649,
+      "step": 3633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019158757176548802,
+      "loss": 0.1245,
+      "step": 3634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001915819335774247,
+      "loss": 0.1141,
+      "step": 3635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019157629358359692,
+      "loss": 0.1429,
+      "step": 3636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019157065178411585,
+      "loss": 0.1187,
+      "step": 3637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019156500817909276,
+      "loss": 0.13,
+      "step": 3638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001915593627686389,
+      "loss": 0.1104,
+      "step": 3639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019155371555286559,
+      "loss": 0.1357,
+      "step": 3640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019154806653188415,
+      "loss": 0.1282,
+      "step": 3641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019154241570580601,
+      "loss": 0.1041,
+      "step": 3642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019153676307474261,
+      "loss": 0.1158,
+      "step": 3643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019153110863880534,
+      "loss": 0.1119,
+      "step": 3644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019152545239810572,
+      "loss": 0.1349,
+      "step": 3645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001915197943527553,
+      "loss": 0.1693,
+      "step": 3646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001915141345028656,
+      "loss": 0.1029,
+      "step": 3647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019150847284854825,
+      "loss": 0.1353,
+      "step": 3648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019150280938991485,
+      "loss": 0.1545,
+      "step": 3649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019149714412707711,
+      "loss": 0.1914,
+      "step": 3650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019149147706014672,
+      "loss": 0.0899,
+      "step": 3651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001914858081892354,
+      "loss": 0.1147,
+      "step": 3652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019148013751445494,
+      "loss": 0.1265,
+      "step": 3653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019147446503591716,
+      "loss": 0.1112,
+      "step": 3654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001914687907537339,
+      "loss": 0.0906,
+      "step": 3655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019146311466801706,
+      "loss": 0.1264,
+      "step": 3656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019145743677887852,
+      "loss": 0.1047,
+      "step": 3657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019145175708643026,
+      "loss": 0.1368,
+      "step": 3658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019144607559078426,
+      "loss": 0.1366,
+      "step": 3659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019144039229205257,
+      "loss": 0.0966,
+      "step": 3660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001914347071903472,
+      "loss": 0.1509,
+      "step": 3661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001914290202857803,
+      "loss": 0.1411,
+      "step": 3662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019142333157846398,
+      "loss": 0.1459,
+      "step": 3663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019141764106851036,
+      "loss": 0.1079,
+      "step": 3664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.001914119487560317,
+      "loss": 0.1875,
+      "step": 3665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019140625464114022,
+      "loss": 0.1714,
+      "step": 3666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0019140055872394824,
+      "loss": 0.131,
+      "step": 3667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019139486100456797,
+      "loss": 0.1197,
+      "step": 3668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019138916148311186,
+      "loss": 0.1337,
+      "step": 3669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019138346015969221,
+      "loss": 0.1028,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019137775703442148,
+      "loss": 0.0735,
+      "step": 3671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019137205210741212,
+      "loss": 0.1207,
+      "step": 3672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019136634537877657,
+      "loss": 0.1877,
+      "step": 3673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001913606368486274,
+      "loss": 0.1528,
+      "step": 3674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019135492651707717,
+      "loss": 0.1104,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019134921438423845,
+      "loss": 0.0773,
+      "step": 3676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019134350045022389,
+      "loss": 0.1082,
+      "step": 3677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019133778471514611,
+      "loss": 0.1006,
+      "step": 3678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019133206717911785,
+      "loss": 0.1064,
+      "step": 3679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019132634784225185,
+      "loss": 0.11,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019132062670466085,
+      "loss": 0.1191,
+      "step": 3681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019131490376645768,
+      "loss": 0.1096,
+      "step": 3682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001913091790277552,
+      "loss": 0.1882,
+      "step": 3683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019130345248866622,
+      "loss": 0.1216,
+      "step": 3684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019129772414930373,
+      "loss": 0.1628,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019129199400978064,
+      "loss": 0.1586,
+      "step": 3686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001912862620702099,
+      "loss": 0.1006,
+      "step": 3687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001912805283307046,
+      "loss": 0.1124,
+      "step": 3688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019127479279137777,
+      "loss": 0.1498,
+      "step": 3689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001912690554523425,
+      "loss": 0.1263,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019126331631371188,
+      "loss": 0.1086,
+      "step": 3691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019125757537559911,
+      "loss": 0.095,
+      "step": 3692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019125183263811736,
+      "loss": 0.1437,
+      "step": 3693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019124608810137992,
+      "loss": 0.0762,
+      "step": 3694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019124034176549998,
+      "loss": 0.097,
+      "step": 3695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001912345936305909,
+      "loss": 0.0983,
+      "step": 3696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00191228843696766,
+      "loss": 0.1295,
+      "step": 3697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019122309196413863,
+      "loss": 0.0847,
+      "step": 3698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019121733843282223,
+      "loss": 0.1415,
+      "step": 3699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019121158310293027,
+      "loss": 0.1106,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019120582597457617,
+      "loss": 0.1982,
+      "step": 3701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019120006704787347,
+      "loss": 0.0847,
+      "step": 3702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019119430632293572,
+      "loss": 0.0898,
+      "step": 3703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019118854379987656,
+      "loss": 0.1138,
+      "step": 3704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019118277947880952,
+      "loss": 0.1235,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001911770133598483,
+      "loss": 0.1001,
+      "step": 3706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019117124544310662,
+      "loss": 0.1156,
+      "step": 3707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019116547572869816,
+      "loss": 0.0901,
+      "step": 3708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019115970421673674,
+      "loss": 0.1382,
+      "step": 3709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001911539309073361,
+      "loss": 0.1063,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019114815580061012,
+      "loss": 0.113,
+      "step": 3711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0019114237889667266,
+      "loss": 0.1182,
+      "step": 3712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.001911366001956376,
+      "loss": 0.135,
+      "step": 3713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019113081969761893,
+      "loss": 0.0897,
+      "step": 3714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019112503740273056,
+      "loss": 0.1257,
+      "step": 3715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019111925331108657,
+      "loss": 0.1566,
+      "step": 3716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019111346742280099,
+      "loss": 0.1211,
+      "step": 3717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019110767973798785,
+      "loss": 0.095,
+      "step": 3718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019110189025676135,
+      "loss": 0.1039,
+      "step": 3719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019109609897923559,
+      "loss": 0.1167,
+      "step": 3720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019109030590552477,
+      "loss": 0.0897,
+      "step": 3721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019108451103574309,
+      "loss": 0.1102,
+      "step": 3722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019107871437000488,
+      "loss": 0.2075,
+      "step": 3723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001910729159084244,
+      "loss": 0.1824,
+      "step": 3724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019106711565111594,
+      "loss": 0.1021,
+      "step": 3725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019106131359819393,
+      "loss": 0.156,
+      "step": 3726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019105550974977273,
+      "loss": 0.1052,
+      "step": 3727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019104970410596678,
+      "loss": 0.101,
+      "step": 3728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001910438966668906,
+      "loss": 0.0621,
+      "step": 3729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019103808743265863,
+      "loss": 0.1516,
+      "step": 3730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019103227640338543,
+      "loss": 0.1335,
+      "step": 3731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019102646357918563,
+      "loss": 0.0989,
+      "step": 3732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019102064896017379,
+      "loss": 0.0763,
+      "step": 3733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019101483254646459,
+      "loss": 0.0978,
+      "step": 3734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019100901433817268,
+      "loss": 0.1027,
+      "step": 3735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019100319433541283,
+      "loss": 0.1831,
+      "step": 3736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001909973725382997,
+      "loss": 0.1255,
+      "step": 3737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019099154894694824,
+      "loss": 0.1125,
+      "step": 3738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019098572356147315,
+      "loss": 0.111,
+      "step": 3739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019097989638198931,
+      "loss": 0.1127,
+      "step": 3740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019097406740861166,
+      "loss": 0.1337,
+      "step": 3741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001909682366414551,
+      "loss": 0.1293,
+      "step": 3742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001909624040806346,
+      "loss": 0.1299,
+      "step": 3743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001909565697262652,
+      "loss": 0.1432,
+      "step": 3744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019095073357846188,
+      "loss": 0.1831,
+      "step": 3745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019094489563733971,
+      "loss": 0.233,
+      "step": 3746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019093905590301388,
+      "loss": 0.1787,
+      "step": 3747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019093321437559947,
+      "loss": 0.0943,
+      "step": 3748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019092737105521169,
+      "loss": 0.1477,
+      "step": 3749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019092152594196575,
+      "loss": 0.1009,
+      "step": 3750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019091567903597687,
+      "loss": 0.2411,
+      "step": 3751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019090983033736037,
+      "loss": 0.092,
+      "step": 3752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019090397984623153,
+      "loss": 0.0923,
+      "step": 3753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.001908981275627058,
+      "loss": 0.1493,
+      "step": 3754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019089227348689846,
+      "loss": 0.0892,
+      "step": 3755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019088641761892497,
+      "loss": 0.0812,
+      "step": 3756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019088055995890088,
+      "loss": 0.1151,
+      "step": 3757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019087470050694155,
+      "loss": 0.1288,
+      "step": 3758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0019086883926316264,
+      "loss": 0.063,
+      "step": 3759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019086297622767964,
+      "loss": 0.1362,
+      "step": 3760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019085711140060815,
+      "loss": 0.1497,
+      "step": 3761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019085124478206387,
+      "loss": 0.1351,
+      "step": 3762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001908453763721624,
+      "loss": 0.0566,
+      "step": 3763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019083950617101953,
+      "loss": 0.0811,
+      "step": 3764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019083363417875094,
+      "loss": 0.2152,
+      "step": 3765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019082776039547244,
+      "loss": 0.106,
+      "step": 3766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019082188482129987,
+      "loss": 0.0866,
+      "step": 3767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019081600745634903,
+      "loss": 0.0667,
+      "step": 3768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019081012830073584,
+      "loss": 0.1017,
+      "step": 3769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001908042473545762,
+      "loss": 0.0941,
+      "step": 3770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001907983646179861,
+      "loss": 0.1007,
+      "step": 3771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001907924800910815,
+      "loss": 0.1445,
+      "step": 3772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019078659377397842,
+      "loss": 0.1351,
+      "step": 3773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00190780705666793,
+      "loss": 0.1525,
+      "step": 3774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019077481576964123,
+      "loss": 0.1404,
+      "step": 3775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001907689240826393,
+      "loss": 0.1011,
+      "step": 3776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019076303060590342,
+      "loss": 0.1572,
+      "step": 3777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019075713533954972,
+      "loss": 0.1049,
+      "step": 3778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001907512382836945,
+      "loss": 0.1586,
+      "step": 3779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019074533943845395,
+      "loss": 0.1597,
+      "step": 3780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019073943880394447,
+      "loss": 0.0943,
+      "step": 3781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019073353638028236,
+      "loss": 0.0918,
+      "step": 3782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019072763216758403,
+      "loss": 0.1429,
+      "step": 3783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001907217261659659,
+      "loss": 0.1183,
+      "step": 3784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019071581837554437,
+      "loss": 0.1467,
+      "step": 3785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019070990879643595,
+      "loss": 0.1261,
+      "step": 3786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019070399742875718,
+      "loss": 0.0883,
+      "step": 3787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001906980842726246,
+      "loss": 0.1897,
+      "step": 3788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.001906921693281548,
+      "loss": 0.1812,
+      "step": 3789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019068625259546443,
+      "loss": 0.1282,
+      "step": 3790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019068033407467014,
+      "loss": 0.1107,
+      "step": 3791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019067441376588864,
+      "loss": 0.1316,
+      "step": 3792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019066849166923662,
+      "loss": 0.1063,
+      "step": 3793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019066256778483088,
+      "loss": 0.1125,
+      "step": 3794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019065664211278825,
+      "loss": 0.1107,
+      "step": 3795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019065071465322553,
+      "loss": 0.1542,
+      "step": 3796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019064478540625962,
+      "loss": 0.099,
+      "step": 3797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019063885437200741,
+      "loss": 0.1362,
+      "step": 3798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019063292155058584,
+      "loss": 0.1353,
+      "step": 3799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019062698694211192,
+      "loss": 0.0892,
+      "step": 3800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019062105054670263,
+      "loss": 0.1183,
+      "step": 3801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019061511236447505,
+      "loss": 0.1276,
+      "step": 3802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019060917239554628,
+      "loss": 0.1318,
+      "step": 3803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019060323064003337,
+      "loss": 0.1199,
+      "step": 3804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0019059728709805354,
+      "loss": 0.0959,
+      "step": 3805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019059134176972396,
+      "loss": 0.1464,
+      "step": 3806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019058539465516189,
+      "loss": 0.1055,
+      "step": 3807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019057944575448455,
+      "loss": 0.1006,
+      "step": 3808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019057349506780923,
+      "loss": 0.1188,
+      "step": 3809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001905675425952533,
+      "loss": 0.119,
+      "step": 3810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001905615883369341,
+      "loss": 0.1165,
+      "step": 3811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019055563229296907,
+      "loss": 0.0739,
+      "step": 3812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001905496744634756,
+      "loss": 0.116,
+      "step": 3813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019054371484857118,
+      "loss": 0.1593,
+      "step": 3814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019053775344837335,
+      "loss": 0.131,
+      "step": 3815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019053179026299958,
+      "loss": 0.137,
+      "step": 3816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019052582529256754,
+      "loss": 0.1367,
+      "step": 3817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019051985853719478,
+      "loss": 0.1219,
+      "step": 3818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019051388999699898,
+      "loss": 0.0874,
+      "step": 3819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019050791967209781,
+      "loss": 0.1377,
+      "step": 3820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00190501947562609,
+      "loss": 0.1233,
+      "step": 3821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001904959736686503,
+      "loss": 0.1146,
+      "step": 3822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019048999799033948,
+      "loss": 0.1357,
+      "step": 3823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001904840205277944,
+      "loss": 0.0968,
+      "step": 3824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019047804128113294,
+      "loss": 0.0883,
+      "step": 3825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001904720602504729,
+      "loss": 0.1467,
+      "step": 3826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001904660774359323,
+      "loss": 0.1716,
+      "step": 3827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019046009283762908,
+      "loss": 0.1201,
+      "step": 3828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019045410645568122,
+      "loss": 0.1138,
+      "step": 3829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019044811829020678,
+      "loss": 0.099,
+      "step": 3830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019044212834132383,
+      "loss": 0.2002,
+      "step": 3831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019043613660915048,
+      "loss": 0.1295,
+      "step": 3832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019043014309380487,
+      "loss": 0.1101,
+      "step": 3833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019042414779540519,
+      "loss": 0.0891,
+      "step": 3834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001904181507140696,
+      "loss": 0.1333,
+      "step": 3835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001904121518499164,
+      "loss": 0.1599,
+      "step": 3836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019040615120306385,
+      "loss": 0.0979,
+      "step": 3837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019040014877363023,
+      "loss": 0.0985,
+      "step": 3838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019039414456173398,
+      "loss": 0.1211,
+      "step": 3839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019038813856749343,
+      "loss": 0.1231,
+      "step": 3840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019038213079102702,
+      "loss": 0.1379,
+      "step": 3841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001903761212324532,
+      "loss": 0.1462,
+      "step": 3842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019037010989189046,
+      "loss": 0.1635,
+      "step": 3843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019036409676945736,
+      "loss": 0.127,
+      "step": 3844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019035808186527242,
+      "loss": 0.1558,
+      "step": 3845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019035206517945425,
+      "loss": 0.1586,
+      "step": 3846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001903460467121215,
+      "loss": 0.1926,
+      "step": 3847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.001903400264633928,
+      "loss": 0.1305,
+      "step": 3848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019033400443338696,
+      "loss": 0.1357,
+      "step": 3849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019032798062222258,
+      "loss": 0.1188,
+      "step": 3850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0019032195503001852,
+      "loss": 0.1351,
+      "step": 3851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019031592765689357,
+      "loss": 0.0955,
+      "step": 3852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019030989850296657,
+      "loss": 0.1583,
+      "step": 3853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001903038675683564,
+      "loss": 0.125,
+      "step": 3854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019029783485318196,
+      "loss": 0.0927,
+      "step": 3855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019029180035756222,
+      "loss": 0.1083,
+      "step": 3856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001902857640816162,
+      "loss": 0.1298,
+      "step": 3857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019027972602546283,
+      "loss": 0.1129,
+      "step": 3858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019027368618922125,
+      "loss": 0.0997,
+      "step": 3859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001902676445730105,
+      "loss": 0.1569,
+      "step": 3860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019026160117694975,
+      "loss": 0.1415,
+      "step": 3861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019025555600115812,
+      "loss": 0.1198,
+      "step": 3862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001902495090457548,
+      "loss": 0.0828,
+      "step": 3863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019024346031085906,
+      "loss": 0.1368,
+      "step": 3864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019023740979659013,
+      "loss": 0.0998,
+      "step": 3865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019023135750306733,
+      "loss": 0.1133,
+      "step": 3866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019022530343041,
+      "loss": 0.1752,
+      "step": 3867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019021924757873748,
+      "loss": 0.1136,
+      "step": 3868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019021318994816923,
+      "loss": 0.1191,
+      "step": 3869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019020713053882466,
+      "loss": 0.1094,
+      "step": 3870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019020106935082322,
+      "loss": 0.1034,
+      "step": 3871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019019500638428445,
+      "loss": 0.0765,
+      "step": 3872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019018894163932792,
+      "loss": 0.078,
+      "step": 3873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019018287511607314,
+      "loss": 0.1173,
+      "step": 3874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019017680681463983,
+      "loss": 0.1326,
+      "step": 3875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019017073673514755,
+      "loss": 0.1346,
+      "step": 3876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019016466487771604,
+      "loss": 0.1429,
+      "step": 3877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019015859124246498,
+      "loss": 0.1343,
+      "step": 3878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019015251582951416,
+      "loss": 0.0911,
+      "step": 3879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019014643863898338,
+      "loss": 0.1101,
+      "step": 3880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019014035967099242,
+      "loss": 0.1031,
+      "step": 3881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019013427892566118,
+      "loss": 0.1292,
+      "step": 3882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019012819640310953,
+      "loss": 0.1042,
+      "step": 3883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019012211210345743,
+      "loss": 0.1782,
+      "step": 3884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019011602602682482,
+      "loss": 0.0939,
+      "step": 3885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019010993817333173,
+      "loss": 0.1382,
+      "step": 3886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019010384854309822,
+      "loss": 0.123,
+      "step": 3887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001900977571362443,
+      "loss": 0.1459,
+      "step": 3888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019009166395289009,
+      "loss": 0.1494,
+      "step": 3889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019008556899315575,
+      "loss": 0.1569,
+      "step": 3890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019007947225716144,
+      "loss": 0.1418,
+      "step": 3891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019007337374502742,
+      "loss": 0.0978,
+      "step": 3892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001900672734568739,
+      "loss": 0.1194,
+      "step": 3893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019006117139282112,
+      "loss": 0.1022,
+      "step": 3894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019005506755298948,
+      "loss": 0.1327,
+      "step": 3895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.001900489619374993,
+      "loss": 0.104,
+      "step": 3896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0019004285454647094,
+      "loss": 0.094,
+      "step": 3897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0019003674538002487,
+      "loss": 0.1653,
+      "step": 3898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001900306344382815,
+      "loss": 0.109,
+      "step": 3899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0019002452172136135,
+      "loss": 0.1636,
+      "step": 3900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0019001840722938496,
+      "loss": 0.127,
+      "step": 3901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0019001229096247284,
+      "loss": 0.1145,
+      "step": 3902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0019000617292074563,
+      "loss": 0.1146,
+      "step": 3903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0019000005310432395,
+      "loss": 0.1125,
+      "step": 3904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001899939315133285,
+      "loss": 0.0793,
+      "step": 3905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018998780814787992,
+      "loss": 0.1307,
+      "step": 3906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018998168300809902,
+      "loss": 0.1088,
+      "step": 3907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018997555609410648,
+      "loss": 0.1204,
+      "step": 3908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001899694274060232,
+      "loss": 0.1143,
+      "step": 3909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018996329694396996,
+      "loss": 0.1163,
+      "step": 3910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018995716470806764,
+      "loss": 0.1815,
+      "step": 3911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018995103069843723,
+      "loss": 0.1248,
+      "step": 3912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018994489491519957,
+      "loss": 0.0904,
+      "step": 3913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018993875735847572,
+      "loss": 0.14,
+      "step": 3914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018993261802838666,
+      "loss": 0.1603,
+      "step": 3915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018992647692505345,
+      "loss": 0.1941,
+      "step": 3916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018992033404859716,
+      "loss": 0.1013,
+      "step": 3917
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018991418939913896,
+      "loss": 0.1461,
+      "step": 3918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018990804297679997,
+      "loss": 0.0997,
+      "step": 3919
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018990189478170136,
+      "loss": 0.1576,
+      "step": 3920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018989574481396438,
+      "loss": 0.1191,
+      "step": 3921
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018988959307371033,
+      "loss": 0.1111,
+      "step": 3922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018988343956106047,
+      "loss": 0.0859,
+      "step": 3923
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018987728427613613,
+      "loss": 0.1887,
+      "step": 3924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018987112721905867,
+      "loss": 0.1144,
+      "step": 3925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001898649683899495,
+      "loss": 0.1086,
+      "step": 3926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001898588077889301,
+      "loss": 0.1312,
+      "step": 3927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018985264541612186,
+      "loss": 0.1249,
+      "step": 3928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018984648127164637,
+      "loss": 0.1193,
+      "step": 3929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001898403153556251,
+      "loss": 0.1785,
+      "step": 3930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018983414766817963,
+      "loss": 0.0864,
+      "step": 3931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018982797820943165,
+      "loss": 0.1593,
+      "step": 3932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001898218069795027,
+      "loss": 0.1293,
+      "step": 3933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018981563397851456,
+      "loss": 0.1132,
+      "step": 3934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018980945920658888,
+      "loss": 0.1152,
+      "step": 3935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018980328266384743,
+      "loss": 0.0975,
+      "step": 3936
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00189797104350412,
+      "loss": 0.177,
+      "step": 3937
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018979092426640442,
+      "loss": 0.1396,
+      "step": 3938
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001897847424119465,
+      "loss": 0.114,
+      "step": 3939
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.001897785587871602,
+      "loss": 0.1115,
+      "step": 3940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018977237339216737,
+      "loss": 0.0671,
+      "step": 3941
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018976618622709004,
+      "loss": 0.113,
+      "step": 3942
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018975999729205018,
+      "loss": 0.1047,
+      "step": 3943
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0018975380658716981,
+      "loss": 0.1404,
+      "step": 3944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00189747614112571,
+      "loss": 0.1292,
+      "step": 3945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018974141986837583,
+      "loss": 0.1069,
+      "step": 3946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018973522385470646,
+      "loss": 0.1088,
+      "step": 3947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018972902607168507,
+      "loss": 0.0869,
+      "step": 3948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018972282651943386,
+      "loss": 0.1058,
+      "step": 3949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018971662519807503,
+      "loss": 0.1416,
+      "step": 3950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018971042210773092,
+      "loss": 0.1263,
+      "step": 3951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018970421724852375,
+      "loss": 0.1205,
+      "step": 3952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018969801062057594,
+      "loss": 0.1038,
+      "step": 3953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018969180222400985,
+      "loss": 0.1079,
+      "step": 3954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001896855920589479,
+      "loss": 0.1082,
+      "step": 3955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001896793801255125,
+      "loss": 0.0613,
+      "step": 3956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001896731664238262,
+      "loss": 0.1495,
+      "step": 3957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018966695095401147,
+      "loss": 0.1478,
+      "step": 3958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018966073371619087,
+      "loss": 0.1278,
+      "step": 3959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018965451471048699,
+      "loss": 0.1187,
+      "step": 3960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018964829393702245,
+      "loss": 0.1271,
+      "step": 3961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001896420713959199,
+      "loss": 0.0833,
+      "step": 3962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018963584708730208,
+      "loss": 0.1008,
+      "step": 3963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018962962101129167,
+      "loss": 0.0869,
+      "step": 3964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018962339316801146,
+      "loss": 0.1198,
+      "step": 3965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001896171635575842,
+      "loss": 0.0903,
+      "step": 3966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018961093218013278,
+      "loss": 0.1118,
+      "step": 3967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018960469903578,
+      "loss": 0.1058,
+      "step": 3968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018959846412464885,
+      "loss": 0.0901,
+      "step": 3969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018959222744686223,
+      "loss": 0.1063,
+      "step": 3970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018958598900254307,
+      "loss": 0.0926,
+      "step": 3971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001895797487918144,
+      "loss": 0.098,
+      "step": 3972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018957350681479933,
+      "loss": 0.0939,
+      "step": 3973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018956726307162081,
+      "loss": 0.1542,
+      "step": 3974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018956101756240208,
+      "loss": 0.0926,
+      "step": 3975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018955477028726618,
+      "loss": 0.1296,
+      "step": 3976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018954852124633632,
+      "loss": 0.0663,
+      "step": 3977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018954227043973578,
+      "loss": 0.1897,
+      "step": 3978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018953601786758773,
+      "loss": 0.1036,
+      "step": 3979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018952976353001547,
+      "loss": 0.1497,
+      "step": 3980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018952350742714235,
+      "loss": 0.0775,
+      "step": 3981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018951724955909171,
+      "loss": 0.1746,
+      "step": 3982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018951098992598693,
+      "loss": 0.1345,
+      "step": 3983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018950472852795146,
+      "loss": 0.1162,
+      "step": 3984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001894984653651087,
+      "loss": 0.0933,
+      "step": 3985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018949220043758222,
+      "loss": 0.0593,
+      "step": 3986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018948593374549553,
+      "loss": 0.0998,
+      "step": 3987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018947966528897213,
+      "loss": 0.1331,
+      "step": 3988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0018947339506813568,
+      "loss": 0.0759,
+      "step": 3989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.001894671230831098,
+      "loss": 0.0981,
+      "step": 3990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018946084933401815,
+      "loss": 0.0981,
+      "step": 3991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018945457382098447,
+      "loss": 0.0999,
+      "step": 3992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018944829654413242,
+      "loss": 0.0818,
+      "step": 3993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018944201750358586,
+      "loss": 0.1185,
+      "step": 3994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018943573669946852,
+      "loss": 0.0734,
+      "step": 3995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001894294541319043,
+      "loss": 0.0695,
+      "step": 3996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018942316980101702,
+      "loss": 0.0901,
+      "step": 3997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018941688370693063,
+      "loss": 0.1405,
+      "step": 3998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018941059584976908,
+      "loss": 0.0969,
+      "step": 3999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018940430622965635,
+      "loss": 0.1188,
+      "step": 4000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001893980148467164,
+      "loss": 0.119,
+      "step": 4001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001893917217010734,
+      "loss": 0.1162,
+      "step": 4002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018938542679285129,
+      "loss": 0.096,
+      "step": 4003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001893791301221743,
+      "loss": 0.0762,
+      "step": 4004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018937283168916652,
+      "loss": 0.0895,
+      "step": 4005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018936653149395218,
+      "loss": 0.1484,
+      "step": 4006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018936022953665549,
+      "loss": 0.1497,
+      "step": 4007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018935392581740068,
+      "loss": 0.1201,
+      "step": 4008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018934762033631207,
+      "loss": 0.0828,
+      "step": 4009
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018934131309351404,
+      "loss": 0.1318,
+      "step": 4010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018933500408913084,
+      "loss": 0.0919,
+      "step": 4011
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018932869332328696,
+      "loss": 0.0975,
+      "step": 4012
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001893223807961068,
+      "loss": 0.1088,
+      "step": 4013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001893160665077148,
+      "loss": 0.1875,
+      "step": 4014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018930975045823553,
+      "loss": 0.1023,
+      "step": 4015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018930343264779346,
+      "loss": 0.0912,
+      "step": 4016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018929711307651318,
+      "loss": 0.0936,
+      "step": 4017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018929079174451933,
+      "loss": 0.1606,
+      "step": 4018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001892844686519365,
+      "loss": 0.1237,
+      "step": 4019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001892781437988894,
+      "loss": 0.0955,
+      "step": 4020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018927181718550272,
+      "loss": 0.1153,
+      "step": 4021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018926548881190125,
+      "loss": 0.082,
+      "step": 4022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018925915867820968,
+      "loss": 0.1127,
+      "step": 4023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.001892528267845529,
+      "loss": 0.0931,
+      "step": 4024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018924649313105575,
+      "loss": 0.1454,
+      "step": 4025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018924015771784309,
+      "loss": 0.1561,
+      "step": 4026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018923382054503986,
+      "loss": 0.1057,
+      "step": 4027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00189227481612771,
+      "loss": 0.1027,
+      "step": 4028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018922114092116147,
+      "loss": 0.1082,
+      "step": 4029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018921479847033634,
+      "loss": 0.088,
+      "step": 4030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018920845426042066,
+      "loss": 0.1196,
+      "step": 4031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018920210829153953,
+      "loss": 0.0923,
+      "step": 4032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018919576056381801,
+      "loss": 0.1409,
+      "step": 4033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018918941107738132,
+      "loss": 0.1466,
+      "step": 4034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018918305983235467,
+      "loss": 0.1112,
+      "step": 4035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0018917670682886323,
+      "loss": 0.1178,
+      "step": 4036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018917035206703235,
+      "loss": 0.0896,
+      "step": 4037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018916399554698721,
+      "loss": 0.0965,
+      "step": 4038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018915763726885327,
+      "loss": 0.0767,
+      "step": 4039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018915127723275584,
+      "loss": 0.1274,
+      "step": 4040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018914491543882033,
+      "loss": 0.0852,
+      "step": 4041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018913855188717214,
+      "loss": 0.115,
+      "step": 4042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018913218657793682,
+      "loss": 0.1289,
+      "step": 4043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001891258195112398,
+      "loss": 0.1023,
+      "step": 4044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001891194506872067,
+      "loss": 0.1228,
+      "step": 4045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018911308010596303,
+      "loss": 0.0887,
+      "step": 4046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018910670776763443,
+      "loss": 0.1042,
+      "step": 4047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018910033367234656,
+      "loss": 0.0884,
+      "step": 4048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018909395782022507,
+      "loss": 0.0863,
+      "step": 4049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018908758021139568,
+      "loss": 0.113,
+      "step": 4050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001890812008459842,
+      "loss": 0.089,
+      "step": 4051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018907481972411633,
+      "loss": 0.0892,
+      "step": 4052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018906843684591793,
+      "loss": 0.106,
+      "step": 4053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018906205221151484,
+      "loss": 0.1252,
+      "step": 4054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00189055665821033,
+      "loss": 0.1451,
+      "step": 4055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018904927767459825,
+      "loss": 0.1467,
+      "step": 4056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001890428877723366,
+      "loss": 0.0719,
+      "step": 4057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018903649611437406,
+      "loss": 0.0945,
+      "step": 4058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018903010270083663,
+      "loss": 0.1052,
+      "step": 4059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018902370753185035,
+      "loss": 0.0805,
+      "step": 4060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018901731060754132,
+      "loss": 0.1437,
+      "step": 4061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018901091192803574,
+      "loss": 0.0897,
+      "step": 4062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018900451149345971,
+      "loss": 0.1338,
+      "step": 4063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018899810930393943,
+      "loss": 0.1411,
+      "step": 4064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018899170535960118,
+      "loss": 0.0806,
+      "step": 4065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018898529966057118,
+      "loss": 0.0873,
+      "step": 4066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018897889220697577,
+      "loss": 0.1345,
+      "step": 4067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018897248299894127,
+      "loss": 0.132,
+      "step": 4068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018896607203659405,
+      "loss": 0.0925,
+      "step": 4069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018895965932006057,
+      "loss": 0.0902,
+      "step": 4070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018895324484946717,
+      "loss": 0.1339,
+      "step": 4071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018894682862494042,
+      "loss": 0.124,
+      "step": 4072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018894041064660679,
+      "loss": 0.1342,
+      "step": 4073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018893399091459283,
+      "loss": 0.1071,
+      "step": 4074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001889275694290251,
+      "loss": 0.0978,
+      "step": 4075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001889211461900303,
+      "loss": 0.1683,
+      "step": 4076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018891472119773498,
+      "loss": 0.1309,
+      "step": 4077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001889082944522659,
+      "loss": 0.106,
+      "step": 4078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001889018659537497,
+      "loss": 0.1084,
+      "step": 4079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.001888954357023132,
+      "loss": 0.0957,
+      "step": 4080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018888900369808316,
+      "loss": 0.1133,
+      "step": 4081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0018888256994118642,
+      "loss": 0.1056,
+      "step": 4082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018887613443174982,
+      "loss": 0.1481,
+      "step": 4083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018886969716990025,
+      "loss": 0.1753,
+      "step": 4084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018886325815576468,
+      "loss": 0.0851,
+      "step": 4085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018885681738947,
+      "loss": 0.1151,
+      "step": 4086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018885037487114324,
+      "loss": 0.1387,
+      "step": 4087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018884393060091145,
+      "loss": 0.103,
+      "step": 4088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.001888374845789017,
+      "loss": 0.0834,
+      "step": 4089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018883103680524102,
+      "loss": 0.1025,
+      "step": 4090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018882458728005662,
+      "loss": 0.1132,
+      "step": 4091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018881813600347564,
+      "loss": 0.1694,
+      "step": 4092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018881168297562529,
+      "loss": 0.1248,
+      "step": 4093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018880522819663275,
+      "loss": 0.1145,
+      "step": 4094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018879877166662537,
+      "loss": 0.1134,
+      "step": 4095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018879231338573045,
+      "loss": 0.1787,
+      "step": 4096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018878585335407527,
+      "loss": 0.1041,
+      "step": 4097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018877939157178724,
+      "loss": 0.0947,
+      "step": 4098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018877292803899381,
+      "loss": 0.1411,
+      "step": 4099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.001887664627558224,
+      "loss": 0.1682,
+      "step": 4100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018875999572240043,
+      "loss": 0.1102,
+      "step": 4101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018875352693885547,
+      "loss": 0.1197,
+      "step": 4102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018874705640531508,
+      "loss": 0.1,
+      "step": 4103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018874058412190681,
+      "loss": 0.1044,
+      "step": 4104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.001887341100887583,
+      "loss": 0.0952,
+      "step": 4105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.001887276343059972,
+      "loss": 0.1238,
+      "step": 4106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018872115677375117,
+      "loss": 0.1039,
+      "step": 4107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018871467749214796,
+      "loss": 0.1116,
+      "step": 4108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018870819646131527,
+      "loss": 0.131,
+      "step": 4109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018870171368138098,
+      "loss": 0.103,
+      "step": 4110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018869522915247283,
+      "loss": 0.0599,
+      "step": 4111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018868874287471873,
+      "loss": 0.0878,
+      "step": 4112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018868225484824656,
+      "loss": 0.1296,
+      "step": 4113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018867576507318425,
+      "loss": 0.1293,
+      "step": 4114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018866927354965976,
+      "loss": 0.0643,
+      "step": 4115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018866278027780106,
+      "loss": 0.1727,
+      "step": 4116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.001886562852577362,
+      "loss": 0.0814,
+      "step": 4117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018864978848959328,
+      "loss": 0.1034,
+      "step": 4118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018864328997350035,
+      "loss": 0.1116,
+      "step": 4119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018863678970958558,
+      "loss": 0.1028,
+      "step": 4120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018863028769797712,
+      "loss": 0.0779,
+      "step": 4121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018862378393880315,
+      "loss": 0.1279,
+      "step": 4122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018861727843219197,
+      "loss": 0.0986,
+      "step": 4123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.001886107711782718,
+      "loss": 0.109,
+      "step": 4124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018860426217717095,
+      "loss": 0.1163,
+      "step": 4125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018859775142901778,
+      "loss": 0.14,
+      "step": 4126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0018859123893394069,
+      "loss": 0.1547,
+      "step": 4127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00188584724692068,
+      "loss": 0.1564,
+      "step": 4128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018857820870352823,
+      "loss": 0.1164,
+      "step": 4129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001885716909684499,
+      "loss": 0.118,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001885651714869614,
+      "loss": 0.1381,
+      "step": 4131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001885586502591914,
+      "loss": 0.1366,
+      "step": 4132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018855212728526836,
+      "loss": 0.1305,
+      "step": 4133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018854560256532098,
+      "loss": 0.1426,
+      "step": 4134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018853907609947794,
+      "loss": 0.1501,
+      "step": 4135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018853254788786782,
+      "loss": 0.1033,
+      "step": 4136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018852601793061944,
+      "loss": 0.1376,
+      "step": 4137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018851948622786148,
+      "loss": 0.1135,
+      "step": 4138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018851295277972277,
+      "loss": 0.1417,
+      "step": 4139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018850641758633214,
+      "loss": 0.1226,
+      "step": 4140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018849988064781843,
+      "loss": 0.1399,
+      "step": 4141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001884933419643105,
+      "loss": 0.1021,
+      "step": 4142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018848680153593733,
+      "loss": 0.0892,
+      "step": 4143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018848025936282786,
+      "loss": 0.0785,
+      "step": 4144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018847371544511112,
+      "loss": 0.124,
+      "step": 4145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018846716978291606,
+      "loss": 0.1196,
+      "step": 4146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001884606223763718,
+      "loss": 0.1329,
+      "step": 4147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018845407322560744,
+      "loss": 0.0856,
+      "step": 4148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018844752233075212,
+      "loss": 0.093,
+      "step": 4149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018844096969193494,
+      "loss": 0.0759,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018843441530928518,
+      "loss": 0.1417,
+      "step": 4151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018842785918293202,
+      "loss": 0.1003,
+      "step": 4152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018842130131300476,
+      "loss": 0.0916,
+      "step": 4153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018841474169963273,
+      "loss": 0.1121,
+      "step": 4154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001884081803429452,
+      "loss": 0.1182,
+      "step": 4155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001884016172430716,
+      "loss": 0.1598,
+      "step": 4156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018839505240014133,
+      "loss": 0.1281,
+      "step": 4157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001883884858142838,
+      "loss": 0.0933,
+      "step": 4158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018838191748562854,
+      "loss": 0.09,
+      "step": 4159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00188375347414305,
+      "loss": 0.1503,
+      "step": 4160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018836877560044277,
+      "loss": 0.1555,
+      "step": 4161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001883622020441714,
+      "loss": 0.0991,
+      "step": 4162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018835562674562052,
+      "loss": 0.1108,
+      "step": 4163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018834904970491978,
+      "loss": 0.0898,
+      "step": 4164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018834247092219887,
+      "loss": 0.0997,
+      "step": 4165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001883358903975875,
+      "loss": 0.1053,
+      "step": 4166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018832930813121541,
+      "loss": 0.1014,
+      "step": 4167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018832272412321238,
+      "loss": 0.1396,
+      "step": 4168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018831613837370824,
+      "loss": 0.093,
+      "step": 4169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018830955088283286,
+      "loss": 0.0996,
+      "step": 4170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001883029616507161,
+      "loss": 0.124,
+      "step": 4171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018829637067748794,
+      "loss": 0.1865,
+      "step": 4172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0018828977796327827,
+      "loss": 0.1105,
+      "step": 4173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.001882831835082171,
+      "loss": 0.0618,
+      "step": 4174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018827658731243447,
+      "loss": 0.1074,
+      "step": 4175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018826998937606044,
+      "loss": 0.1119,
+      "step": 4176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001882633896992251,
+      "loss": 0.0989,
+      "step": 4177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018825678828205854,
+      "loss": 0.0955,
+      "step": 4178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018825018512469102,
+      "loss": 0.1257,
+      "step": 4179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018824358022725264,
+      "loss": 0.0947,
+      "step": 4180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018823697358987368,
+      "loss": 0.1431,
+      "step": 4181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001882303652126844,
+      "loss": 0.1417,
+      "step": 4182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001882237550958151,
+      "loss": 0.0926,
+      "step": 4183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001882171432393961,
+      "loss": 0.156,
+      "step": 4184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018821052964355781,
+      "loss": 0.1372,
+      "step": 4185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018820391430843058,
+      "loss": 0.1296,
+      "step": 4186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018819729723414487,
+      "loss": 0.1448,
+      "step": 4187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018819067842083118,
+      "loss": 0.0644,
+      "step": 4188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018818405786861997,
+      "loss": 0.1149,
+      "step": 4189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001881774355776418,
+      "loss": 0.0786,
+      "step": 4190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018817081154802726,
+      "loss": 0.1437,
+      "step": 4191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018816418577990693,
+      "loss": 0.1108,
+      "step": 4192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018815755827341148,
+      "loss": 0.1467,
+      "step": 4193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018815092902867156,
+      "loss": 0.0891,
+      "step": 4194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018814429804581792,
+      "loss": 0.0951,
+      "step": 4195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018813766532498127,
+      "loss": 0.067,
+      "step": 4196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018813103086629242,
+      "loss": 0.0793,
+      "step": 4197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018812439466988217,
+      "loss": 0.085,
+      "step": 4198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018811775673588133,
+      "loss": 0.1104,
+      "step": 4199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018811111706442088,
+      "loss": 0.1017,
+      "step": 4200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018810447565563164,
+      "loss": 0.1053,
+      "step": 4201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001880978325096446,
+      "loss": 0.1251,
+      "step": 4202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018809118762659076,
+      "loss": 0.1299,
+      "step": 4203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018808454100660112,
+      "loss": 0.0886,
+      "step": 4204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018807789264980678,
+      "loss": 0.1081,
+      "step": 4205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018807124255633876,
+      "loss": 0.1068,
+      "step": 4206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018806459072632823,
+      "loss": 0.1195,
+      "step": 4207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018805793715990632,
+      "loss": 0.0719,
+      "step": 4208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018805128185720422,
+      "loss": 0.1326,
+      "step": 4209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018804462481835317,
+      "loss": 0.1444,
+      "step": 4210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018803796604348443,
+      "loss": 0.1544,
+      "step": 4211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001880313055327293,
+      "loss": 0.2042,
+      "step": 4212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.001880246432862191,
+      "loss": 0.1448,
+      "step": 4213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018801797930408522,
+      "loss": 0.0911,
+      "step": 4214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00188011313586459,
+      "loss": 0.146,
+      "step": 4215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018800464613347191,
+      "loss": 0.0977,
+      "step": 4216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018799797694525541,
+      "loss": 0.1694,
+      "step": 4217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018799130602194102,
+      "loss": 0.0665,
+      "step": 4218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018798463336366022,
+      "loss": 0.0723,
+      "step": 4219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0018797795897054464,
+      "loss": 0.1539,
+      "step": 4220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001879712828427258,
+      "loss": 0.1178,
+      "step": 4221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018796460498033544,
+      "loss": 0.1078,
+      "step": 4222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018795792538350514,
+      "loss": 0.1265,
+      "step": 4223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018795124405236664,
+      "loss": 0.1095,
+      "step": 4224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018794456098705174,
+      "loss": 0.0886,
+      "step": 4225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001879378761876921,
+      "loss": 0.08,
+      "step": 4226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001879311896544196,
+      "loss": 0.0801,
+      "step": 4227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018792450138736606,
+      "loss": 0.1207,
+      "step": 4228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018791781138666336,
+      "loss": 0.0953,
+      "step": 4229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001879111196524434,
+      "loss": 0.0925,
+      "step": 4230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018790442618483816,
+      "loss": 0.1093,
+      "step": 4231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018789773098397957,
+      "loss": 0.0761,
+      "step": 4232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001878910340499997,
+      "loss": 0.0912,
+      "step": 4233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018788433538303052,
+      "loss": 0.1178,
+      "step": 4234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018787763498320418,
+      "loss": 0.1345,
+      "step": 4235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018787093285065272,
+      "loss": 0.1815,
+      "step": 4236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001878642289855084,
+      "loss": 0.1637,
+      "step": 4237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001878575233879033,
+      "loss": 0.1293,
+      "step": 4238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018785081605796968,
+      "loss": 0.1368,
+      "step": 4239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018784410699583978,
+      "loss": 0.0902,
+      "step": 4240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018783739620164594,
+      "loss": 0.113,
+      "step": 4241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018783068367552037,
+      "loss": 0.1722,
+      "step": 4242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018782396941759553,
+      "loss": 0.0985,
+      "step": 4243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018781725342800375,
+      "loss": 0.1376,
+      "step": 4244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018781053570687746,
+      "loss": 0.1033,
+      "step": 4245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018780381625434917,
+      "loss": 0.1406,
+      "step": 4246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018779709507055129,
+      "loss": 0.172,
+      "step": 4247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018779037215561638,
+      "loss": 0.0964,
+      "step": 4248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018778364750967702,
+      "loss": 0.0935,
+      "step": 4249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018777692113286575,
+      "loss": 0.0787,
+      "step": 4250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018777019302531526,
+      "loss": 0.1108,
+      "step": 4251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018776346318715817,
+      "loss": 0.1439,
+      "step": 4252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018775673161852717,
+      "loss": 0.0555,
+      "step": 4253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018774999831955503,
+      "loss": 0.1204,
+      "step": 4254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018774326329037449,
+      "loss": 0.1392,
+      "step": 4255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018773652653111834,
+      "loss": 0.0846,
+      "step": 4256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001877297880419194,
+      "loss": 0.0688,
+      "step": 4257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001877230478229106,
+      "loss": 0.1465,
+      "step": 4258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018771630587422474,
+      "loss": 0.0974,
+      "step": 4259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018770956219599487,
+      "loss": 0.119,
+      "step": 4260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018770281678835387,
+      "loss": 0.131,
+      "step": 4261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018769606965143475,
+      "loss": 0.0888,
+      "step": 4262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018768932078537057,
+      "loss": 0.2004,
+      "step": 4263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0018768257019029443,
+      "loss": 0.0726,
+      "step": 4264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001876758178663394,
+      "loss": 0.1721,
+      "step": 4265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001876690638136386,
+      "loss": 0.1017,
+      "step": 4266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.001876623080323252,
+      "loss": 0.1132,
+      "step": 4267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001876555505225325,
+      "loss": 0.1296,
+      "step": 4268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001876487912843936,
+      "loss": 0.1945,
+      "step": 4269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001876420303180419,
+      "loss": 0.078,
+      "step": 4270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018763526762361061,
+      "loss": 0.1265,
+      "step": 4271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018762850320123315,
+      "loss": 0.0993,
+      "step": 4272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018762173705104287,
+      "loss": 0.1093,
+      "step": 4273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018761496917317315,
+      "loss": 0.103,
+      "step": 4274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001876081995677575,
+      "loss": 0.1453,
+      "step": 4275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018760142823492933,
+      "loss": 0.1656,
+      "step": 4276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001875946551748222,
+      "loss": 0.0758,
+      "step": 4277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018758788038756964,
+      "loss": 0.1342,
+      "step": 4278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018758110387330524,
+      "loss": 0.1095,
+      "step": 4279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018757432563216263,
+      "loss": 0.1,
+      "step": 4280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001875675456642754,
+      "loss": 0.0668,
+      "step": 4281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001875607639697773,
+      "loss": 0.1169,
+      "step": 4282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018755398054880205,
+      "loss": 0.1296,
+      "step": 4283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018754719540148336,
+      "loss": 0.1273,
+      "step": 4284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018754040852795501,
+      "loss": 0.0666,
+      "step": 4285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018753361992835083,
+      "loss": 0.0828,
+      "step": 4286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018752682960280472,
+      "loss": 0.1223,
+      "step": 4287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018752003755145054,
+      "loss": 0.135,
+      "step": 4288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018751324377442216,
+      "loss": 0.1143,
+      "step": 4289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018750644827185362,
+      "loss": 0.0773,
+      "step": 4290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018749965104387885,
+      "loss": 0.172,
+      "step": 4291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018749285209063188,
+      "loss": 0.1049,
+      "step": 4292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018748605141224682,
+      "loss": 0.1101,
+      "step": 4293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001874792490088577,
+      "loss": 0.1024,
+      "step": 4294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001874724448805987,
+      "loss": 0.1458,
+      "step": 4295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018746563902760395,
+      "loss": 0.108,
+      "step": 4296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018745883145000763,
+      "loss": 0.1238,
+      "step": 4297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00187452022147944,
+      "loss": 0.1101,
+      "step": 4298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018744521112154728,
+      "loss": 0.1437,
+      "step": 4299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018743839837095182,
+      "loss": 0.085,
+      "step": 4300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018743158389629192,
+      "loss": 0.0907,
+      "step": 4301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018742476769770195,
+      "loss": 0.1146,
+      "step": 4302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001874179497753163,
+      "loss": 0.0892,
+      "step": 4303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001874111301292694,
+      "loss": 0.085,
+      "step": 4304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018740430875969574,
+      "loss": 0.0748,
+      "step": 4305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018739748566672981,
+      "loss": 0.1189,
+      "step": 4306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.001873906608505061,
+      "loss": 0.1231,
+      "step": 4307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018738383431115926,
+      "loss": 0.0648,
+      "step": 4308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018737700604882384,
+      "loss": 0.1106,
+      "step": 4309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018737017606363447,
+      "loss": 0.0942,
+      "step": 4310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018736334435572584,
+      "loss": 0.1327,
+      "step": 4311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018735651092523266,
+      "loss": 0.0823,
+      "step": 4312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0018734967577228965,
+      "loss": 0.1219,
+      "step": 4313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001873428388970316,
+      "loss": 0.1029,
+      "step": 4314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018733600029959328,
+      "loss": 0.1281,
+      "step": 4315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018732915998010957,
+      "loss": 0.0801,
+      "step": 4316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018732231793871534,
+      "loss": 0.1345,
+      "step": 4317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001873154741755455,
+      "loss": 0.1068,
+      "step": 4318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018730862869073494,
+      "loss": 0.1571,
+      "step": 4319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018730178148441867,
+      "loss": 0.0736,
+      "step": 4320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018729493255673173,
+      "loss": 0.1156,
+      "step": 4321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018728808190780915,
+      "loss": 0.0857,
+      "step": 4322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018728122953778597,
+      "loss": 0.1002,
+      "step": 4323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018727437544679732,
+      "loss": 0.1178,
+      "step": 4324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001872675196349784,
+      "loss": 0.0848,
+      "step": 4325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018726066210246431,
+      "loss": 0.1138,
+      "step": 4326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001872538028493903,
+      "loss": 0.0994,
+      "step": 4327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018724694187589157,
+      "loss": 0.0853,
+      "step": 4328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001872400791821035,
+      "loss": 0.077,
+      "step": 4329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001872332147681613,
+      "loss": 0.0863,
+      "step": 4330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018722634863420039,
+      "loss": 0.0763,
+      "step": 4331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018721948078035612,
+      "loss": 0.1245,
+      "step": 4332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018721261120676395,
+      "loss": 0.0805,
+      "step": 4333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018720573991355925,
+      "loss": 0.1971,
+      "step": 4334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018719886690087759,
+      "loss": 0.0737,
+      "step": 4335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018719199216885444,
+      "loss": 0.1371,
+      "step": 4336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018718511571762532,
+      "loss": 0.0963,
+      "step": 4337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018717823754732593,
+      "loss": 0.1232,
+      "step": 4338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018717135765809177,
+      "loss": 0.0732,
+      "step": 4339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018716447605005855,
+      "loss": 0.1102,
+      "step": 4340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018715759272336194,
+      "loss": 0.1222,
+      "step": 4341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018715070767813772,
+      "loss": 0.1673,
+      "step": 4342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018714382091452153,
+      "loss": 0.1039,
+      "step": 4343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018713693243264926,
+      "loss": 0.1375,
+      "step": 4344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018713004223265672,
+      "loss": 0.0778,
+      "step": 4345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018712315031467973,
+      "loss": 0.1226,
+      "step": 4346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018711625667885419,
+      "loss": 0.1633,
+      "step": 4347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018710936132531604,
+      "loss": 0.0929,
+      "step": 4348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018710246425420124,
+      "loss": 0.1881,
+      "step": 4349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018709556546564578,
+      "loss": 0.1134,
+      "step": 4350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018708866495978568,
+      "loss": 0.1112,
+      "step": 4351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00187081762736757,
+      "loss": 0.1033,
+      "step": 4352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018707485879669585,
+      "loss": 0.1133,
+      "step": 4353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018706795313973832,
+      "loss": 0.1422,
+      "step": 4354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018706104576602064,
+      "loss": 0.0629,
+      "step": 4355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018705413667567893,
+      "loss": 0.0869,
+      "step": 4356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018704722586884946,
+      "loss": 0.1545,
+      "step": 4357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0018704031334566848,
+      "loss": 0.1027,
+      "step": 4358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.001870333991062723,
+      "loss": 0.1392,
+      "step": 4359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018702648315079729,
+      "loss": 0.1227,
+      "step": 4360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018701956547937974,
+      "loss": 0.0891,
+      "step": 4361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018701264609215604,
+      "loss": 0.1174,
+      "step": 4362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001870057249892627,
+      "loss": 0.1351,
+      "step": 4363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018699880217083616,
+      "loss": 0.0974,
+      "step": 4364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018699187763701291,
+      "loss": 0.1082,
+      "step": 4365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018698495138792946,
+      "loss": 0.1389,
+      "step": 4366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018697802342372244,
+      "loss": 0.1575,
+      "step": 4367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018697109374452838,
+      "loss": 0.095,
+      "step": 4368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00186964162350484,
+      "loss": 0.1244,
+      "step": 4369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018695722924172587,
+      "loss": 0.119,
+      "step": 4370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018695029441839076,
+      "loss": 0.0912,
+      "step": 4371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018694335788061538,
+      "loss": 0.1764,
+      "step": 4372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018693641962853654,
+      "loss": 0.1104,
+      "step": 4373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018692947966229099,
+      "loss": 0.0951,
+      "step": 4374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018692253798201562,
+      "loss": 0.1129,
+      "step": 4375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018691559458784725,
+      "loss": 0.1112,
+      "step": 4376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018690864947992286,
+      "loss": 0.0909,
+      "step": 4377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018690170265837928,
+      "loss": 0.1099,
+      "step": 4378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018689475412335362,
+      "loss": 0.098,
+      "step": 4379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018688780387498276,
+      "loss": 0.1626,
+      "step": 4380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018688085191340383,
+      "loss": 0.0853,
+      "step": 4381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018687389823875385,
+      "loss": 0.0987,
+      "step": 4382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018686694285116995,
+      "loss": 0.0955,
+      "step": 4383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001868599857507893,
+      "loss": 0.077,
+      "step": 4384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018685302693774903,
+      "loss": 0.0929,
+      "step": 4385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018684606641218638,
+      "loss": 0.1294,
+      "step": 4386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018683910417423857,
+      "loss": 0.0884,
+      "step": 4387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001868321402240429,
+      "loss": 0.113,
+      "step": 4388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001868251745617367,
+      "loss": 0.0757,
+      "step": 4389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018681820718745726,
+      "loss": 0.0967,
+      "step": 4390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018681123810134197,
+      "loss": 0.0715,
+      "step": 4391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018680426730352828,
+      "loss": 0.1669,
+      "step": 4392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018679729479415363,
+      "loss": 0.1232,
+      "step": 4393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018679032057335547,
+      "loss": 0.1111,
+      "step": 4394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018678334464127136,
+      "loss": 0.1077,
+      "step": 4395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018677636699803877,
+      "loss": 0.1495,
+      "step": 4396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018676938764379538,
+      "loss": 0.1127,
+      "step": 4397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018676240657867873,
+      "loss": 0.1038,
+      "step": 4398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001867554238028265,
+      "loss": 0.1227,
+      "step": 4399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001867484393163764,
+      "loss": 0.1041,
+      "step": 4400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018674145311946605,
+      "loss": 0.1013,
+      "step": 4401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001867344652122333,
+      "loss": 0.0753,
+      "step": 4402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.001867274755948159,
+      "loss": 0.1076,
+      "step": 4403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018672048426735166,
+      "loss": 0.1722,
+      "step": 4404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0018671349122997844,
+      "loss": 0.2014,
+      "step": 4405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018670649648283412,
+      "loss": 0.1078,
+      "step": 4406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018669950002605665,
+      "loss": 0.1616,
+      "step": 4407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001866925018597839,
+      "loss": 0.1486,
+      "step": 4408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018668550198415396,
+      "loss": 0.109,
+      "step": 4409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018667850039930476,
+      "loss": 0.1414,
+      "step": 4410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018667149710537441,
+      "loss": 0.1561,
+      "step": 4411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018666449210250097,
+      "loss": 0.1182,
+      "step": 4412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001866574853908226,
+      "loss": 0.1904,
+      "step": 4413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001866504769704774,
+      "loss": 0.0764,
+      "step": 4414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018664346684160358,
+      "loss": 0.1041,
+      "step": 4415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001866364550043394,
+      "loss": 0.0873,
+      "step": 4416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018662944145882306,
+      "loss": 0.0876,
+      "step": 4417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018662242620519284,
+      "loss": 0.074,
+      "step": 4418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018661540924358715,
+      "loss": 0.1437,
+      "step": 4419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018660839057414427,
+      "loss": 0.1058,
+      "step": 4420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001866013701970026,
+      "loss": 0.0741,
+      "step": 4421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018659434811230061,
+      "loss": 0.1005,
+      "step": 4422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001865873243201767,
+      "loss": 0.1526,
+      "step": 4423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018658029882076936,
+      "loss": 0.1106,
+      "step": 4424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018657327161421718,
+      "loss": 0.1039,
+      "step": 4425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018656624270065865,
+      "loss": 0.1144,
+      "step": 4426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001865592120802324,
+      "loss": 0.219,
+      "step": 4427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018655217975307706,
+      "loss": 0.1058,
+      "step": 4428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018654514571933127,
+      "loss": 0.0851,
+      "step": 4429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018653810997913375,
+      "loss": 0.0911,
+      "step": 4430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018653107253262318,
+      "loss": 0.1069,
+      "step": 4431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018652403337993835,
+      "loss": 0.0877,
+      "step": 4432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018651699252121806,
+      "loss": 0.0867,
+      "step": 4433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018650994995660112,
+      "loss": 0.111,
+      "step": 4434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018650290568622643,
+      "loss": 0.0881,
+      "step": 4435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018649585971023284,
+      "loss": 0.0769,
+      "step": 4436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001864888120287593,
+      "loss": 0.1074,
+      "step": 4437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018648176264194477,
+      "loss": 0.0671,
+      "step": 4438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018647471154992824,
+      "loss": 0.1025,
+      "step": 4439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018646765875284875,
+      "loss": 0.119,
+      "step": 4440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018646060425084533,
+      "loss": 0.1221,
+      "step": 4441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018645354804405715,
+      "loss": 0.1401,
+      "step": 4442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018644649013262328,
+      "loss": 0.0959,
+      "step": 4443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018643943051668288,
+      "loss": 0.1472,
+      "step": 4444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001864323691963752,
+      "loss": 0.0977,
+      "step": 4445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018642530617183942,
+      "loss": 0.0762,
+      "step": 4446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018641824144321481,
+      "loss": 0.1107,
+      "step": 4447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001864111750106407,
+      "loss": 0.1018,
+      "step": 4448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.001864041068742564,
+      "loss": 0.0648,
+      "step": 4449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018639703703420129,
+      "loss": 0.095,
+      "step": 4450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0018638996549061476,
+      "loss": 0.1294,
+      "step": 4451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018638289224363622,
+      "loss": 0.0696,
+      "step": 4452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018637581729340517,
+      "loss": 0.0576,
+      "step": 4453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001863687406400611,
+      "loss": 0.1061,
+      "step": 4454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018636166228374355,
+      "loss": 0.1013,
+      "step": 4455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001863545822245921,
+      "loss": 0.0935,
+      "step": 4456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018634750046274632,
+      "loss": 0.0588,
+      "step": 4457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018634041699834584,
+      "loss": 0.1041,
+      "step": 4458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018633333183153034,
+      "loss": 0.0983,
+      "step": 4459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018632624496243953,
+      "loss": 0.0978,
+      "step": 4460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018631915639121318,
+      "loss": 0.0734,
+      "step": 4461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018631206611799099,
+      "loss": 0.1274,
+      "step": 4462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018630497414291276,
+      "loss": 0.0961,
+      "step": 4463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001862978804661184,
+      "loss": 0.0573,
+      "step": 4464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018629078508774772,
+      "loss": 0.1089,
+      "step": 4465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018628368800794064,
+      "loss": 0.1416,
+      "step": 4466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018627658922683713,
+      "loss": 0.1344,
+      "step": 4467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018626948874457708,
+      "loss": 0.0698,
+      "step": 4468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018626238656130054,
+      "loss": 0.0947,
+      "step": 4469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018625528267714757,
+      "loss": 0.141,
+      "step": 4470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018624817709225823,
+      "loss": 0.1215,
+      "step": 4471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001862410698067726,
+      "loss": 0.1101,
+      "step": 4472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018623396082083081,
+      "loss": 0.126,
+      "step": 4473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018622685013457308,
+      "loss": 0.1299,
+      "step": 4474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018621973774813956,
+      "loss": 0.1384,
+      "step": 4475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018621262366167054,
+      "loss": 0.151,
+      "step": 4476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018620550787530625,
+      "loss": 0.129,
+      "step": 4477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018619839038918702,
+      "loss": 0.1172,
+      "step": 4478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018619127120345317,
+      "loss": 0.1245,
+      "step": 4479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001861841503182451,
+      "loss": 0.0974,
+      "step": 4480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018617702773370318,
+      "loss": 0.1138,
+      "step": 4481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018616990344996788,
+      "loss": 0.1099,
+      "step": 4482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018616277746717964,
+      "loss": 0.1058,
+      "step": 4483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00186155649785479,
+      "loss": 0.1376,
+      "step": 4484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001861485204050065,
+      "loss": 0.0957,
+      "step": 4485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001861413893259027,
+      "loss": 0.121,
+      "step": 4486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018613425654830818,
+      "loss": 0.1401,
+      "step": 4487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018612712207236365,
+      "loss": 0.2102,
+      "step": 4488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001861199858982097,
+      "loss": 0.1154,
+      "step": 4489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.001861128480259871,
+      "loss": 0.1191,
+      "step": 4490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018610570845583657,
+      "loss": 0.1379,
+      "step": 4491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018609856718789888,
+      "loss": 0.1212,
+      "step": 4492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018609142422231484,
+      "loss": 0.0991,
+      "step": 4493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018608427955922528,
+      "loss": 0.1429,
+      "step": 4494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018607713319877112,
+      "loss": 0.1583,
+      "step": 4495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018606998514109322,
+      "loss": 0.0941,
+      "step": 4496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0018606283538633252,
+      "loss": 0.1844,
+      "step": 4497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018605568393463003,
+      "loss": 0.1353,
+      "step": 4498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018604853078612673,
+      "loss": 0.1846,
+      "step": 4499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001860413759409637,
+      "loss": 0.1561,
+      "step": 4500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018603421939928197,
+      "loss": 0.1454,
+      "step": 4501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018602706116122266,
+      "loss": 0.1037,
+      "step": 4502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018601990122692692,
+      "loss": 0.1224,
+      "step": 4503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018601273959653595,
+      "loss": 0.2445,
+      "step": 4504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018600557627019092,
+      "loss": 0.0934,
+      "step": 4505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018599841124803307,
+      "loss": 0.0981,
+      "step": 4506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018599124453020372,
+      "loss": 0.0846,
+      "step": 4507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018598407611684416,
+      "loss": 0.1147,
+      "step": 4508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018597690600809572,
+      "loss": 0.1411,
+      "step": 4509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018596973420409975,
+      "loss": 0.0702,
+      "step": 4510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018596256070499773,
+      "loss": 0.0744,
+      "step": 4511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018595538551093105,
+      "loss": 0.1386,
+      "step": 4512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001859482086220412,
+      "loss": 0.1317,
+      "step": 4513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001859410300384697,
+      "loss": 0.1013,
+      "step": 4514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001859338497603581,
+      "loss": 0.0923,
+      "step": 4515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018592666778784795,
+      "loss": 0.1387,
+      "step": 4516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018591948412108087,
+      "loss": 0.1177,
+      "step": 4517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018591229876019854,
+      "loss": 0.0837,
+      "step": 4518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018590511170534256,
+      "loss": 0.1059,
+      "step": 4519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001858979229566547,
+      "loss": 0.1262,
+      "step": 4520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018589073251427668,
+      "loss": 0.1146,
+      "step": 4521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001858835403783503,
+      "loss": 0.1162,
+      "step": 4522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018587634654901738,
+      "loss": 0.075,
+      "step": 4523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001858691510264197,
+      "loss": 0.1151,
+      "step": 4524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018586195381069918,
+      "loss": 0.1414,
+      "step": 4525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018585475490199776,
+      "loss": 0.1052,
+      "step": 4526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018584755430045734,
+      "loss": 0.0928,
+      "step": 4527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001858403520062199,
+      "loss": 0.1118,
+      "step": 4528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018583314801942747,
+      "loss": 0.1237,
+      "step": 4529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018582594234022212,
+      "loss": 0.0696,
+      "step": 4530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018581873496874585,
+      "loss": 0.0458,
+      "step": 4531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018581152590514086,
+      "loss": 0.126,
+      "step": 4532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018580431514954921,
+      "loss": 0.1522,
+      "step": 4533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018579710270211314,
+      "loss": 0.0812,
+      "step": 4534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018578988856297483,
+      "loss": 0.0607,
+      "step": 4535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018578267273227655,
+      "loss": 0.1442,
+      "step": 4536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018577545521016056,
+      "loss": 0.1595,
+      "step": 4537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018576823599676918,
+      "loss": 0.0948,
+      "step": 4538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018576101509224473,
+      "loss": 0.0829,
+      "step": 4539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001857537924967296,
+      "loss": 0.1108,
+      "step": 4540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018574656821036624,
+      "loss": 0.1447,
+      "step": 4541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0018573934223329702,
+      "loss": 0.1233,
+      "step": 4542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.001857321145656645,
+      "loss": 0.167,
+      "step": 4543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018572488520761112,
+      "loss": 0.1039,
+      "step": 4544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001857176541592795,
+      "loss": 0.0869,
+      "step": 4545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018571042142081216,
+      "loss": 0.1663,
+      "step": 4546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001857031869923517,
+      "loss": 0.0818,
+      "step": 4547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018569595087404081,
+      "loss": 0.0938,
+      "step": 4548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018568871306602213,
+      "loss": 0.118,
+      "step": 4549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001856814735684384,
+      "loss": 0.1167,
+      "step": 4550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018567423238143236,
+      "loss": 0.1011,
+      "step": 4551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001856669895051468,
+      "loss": 0.127,
+      "step": 4552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001856597449397245,
+      "loss": 0.1277,
+      "step": 4553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018565249868530832,
+      "loss": 0.1325,
+      "step": 4554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018564525074204112,
+      "loss": 0.168,
+      "step": 4555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018563800111006583,
+      "loss": 0.1411,
+      "step": 4556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001856307497895254,
+      "loss": 0.0507,
+      "step": 4557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018562349678056282,
+      "loss": 0.0735,
+      "step": 4558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018561624208332107,
+      "loss": 0.1486,
+      "step": 4559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018560898569794319,
+      "loss": 0.0767,
+      "step": 4560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018560172762457229,
+      "loss": 0.0743,
+      "step": 4561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018559446786335144,
+      "loss": 0.0741,
+      "step": 4562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001855872064144238,
+      "loss": 0.1285,
+      "step": 4563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001855799432779326,
+      "loss": 0.0943,
+      "step": 4564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018557267845402097,
+      "loss": 0.167,
+      "step": 4565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001855654119428322,
+      "loss": 0.1038,
+      "step": 4566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018555814374450956,
+      "loss": 0.1301,
+      "step": 4567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018555087385919635,
+      "loss": 0.076,
+      "step": 4568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018554360228703595,
+      "loss": 0.1184,
+      "step": 4569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018553632902817165,
+      "loss": 0.1426,
+      "step": 4570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018552905408274697,
+      "loss": 0.127,
+      "step": 4571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001855217774509053,
+      "loss": 0.0903,
+      "step": 4572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001855144991327901,
+      "loss": 0.145,
+      "step": 4573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018550721912854492,
+      "loss": 0.1208,
+      "step": 4574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018549993743831323,
+      "loss": 0.1135,
+      "step": 4575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018549265406223871,
+      "loss": 0.1086,
+      "step": 4576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018548536900046492,
+      "loss": 0.1057,
+      "step": 4577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018547808225313548,
+      "loss": 0.0626,
+      "step": 4578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018547079382039407,
+      "loss": 0.0798,
+      "step": 4579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018546350370238446,
+      "loss": 0.0931,
+      "step": 4580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018545621189925033,
+      "loss": 0.0919,
+      "step": 4581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018544891841113546,
+      "loss": 0.0933,
+      "step": 4582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001854416232381837,
+      "loss": 0.0913,
+      "step": 4583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018543432638053888,
+      "loss": 0.077,
+      "step": 4584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018542702783834483,
+      "loss": 0.0336,
+      "step": 4585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018541972761174552,
+      "loss": 0.1451,
+      "step": 4586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018541242570088484,
+      "loss": 0.0934,
+      "step": 4587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0018540512210590682,
+      "loss": 0.0933,
+      "step": 4588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.001853978168269554,
+      "loss": 0.0733,
+      "step": 4589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018539050986417468,
+      "loss": 0.116,
+      "step": 4590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018538320121770872,
+      "loss": 0.0954,
+      "step": 4591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001853758908877016,
+      "loss": 0.0883,
+      "step": 4592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001853685788742975,
+      "loss": 0.106,
+      "step": 4593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018536126517764058,
+      "loss": 0.0934,
+      "step": 4594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018535394979787502,
+      "loss": 0.0725,
+      "step": 4595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001853466327351451,
+      "loss": 0.0933,
+      "step": 4596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001853393139895951,
+      "loss": 0.0823,
+      "step": 4597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018533199356136933,
+      "loss": 0.1107,
+      "step": 4598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018532467145061205,
+      "loss": 0.1221,
+      "step": 4599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018531734765746771,
+      "loss": 0.113,
+      "step": 4600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001853100221820807,
+      "loss": 0.0575,
+      "step": 4601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018530269502459546,
+      "loss": 0.132,
+      "step": 4602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018529536618515647,
+      "loss": 0.1172,
+      "step": 4603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001852880356639082,
+      "loss": 0.0955,
+      "step": 4604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018528070346099526,
+      "loss": 0.1179,
+      "step": 4605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018527336957656216,
+      "loss": 0.0844,
+      "step": 4606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018526603401075353,
+      "loss": 0.1552,
+      "step": 4607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00185258696763714,
+      "loss": 0.2546,
+      "step": 4608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018525135783558824,
+      "loss": 0.1464,
+      "step": 4609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018524401722652096,
+      "loss": 0.2011,
+      "step": 4610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018523667493665688,
+      "loss": 0.1393,
+      "step": 4611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018522933096614084,
+      "loss": 0.1237,
+      "step": 4612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018522198531511756,
+      "loss": 0.0814,
+      "step": 4613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018521463798373192,
+      "loss": 0.1418,
+      "step": 4614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001852072889721288,
+      "loss": 0.1171,
+      "step": 4615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018519993828045305,
+      "loss": 0.0982,
+      "step": 4616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018519258590884969,
+      "loss": 0.0732,
+      "step": 4617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001851852318574636,
+      "loss": 0.1135,
+      "step": 4618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018517787612643986,
+      "loss": 0.0854,
+      "step": 4619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018517051871592347,
+      "loss": 0.1713,
+      "step": 4620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001851631596260595,
+      "loss": 0.0758,
+      "step": 4621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018515579885699305,
+      "loss": 0.1438,
+      "step": 4622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018514843640886926,
+      "loss": 0.1086,
+      "step": 4623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018514107228183331,
+      "loss": 0.186,
+      "step": 4624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001851337064760304,
+      "loss": 0.0944,
+      "step": 4625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018512633899160577,
+      "loss": 0.0929,
+      "step": 4626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018511896982870467,
+      "loss": 0.1322,
+      "step": 4627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018511159898747238,
+      "loss": 0.15,
+      "step": 4628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001851042264680543,
+      "loss": 0.0419,
+      "step": 4629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018509685227059572,
+      "loss": 0.1138,
+      "step": 4630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018508947639524211,
+      "loss": 0.0983,
+      "step": 4631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018508209884213888,
+      "loss": 0.1273,
+      "step": 4632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018507471961143147,
+      "loss": 0.093,
+      "step": 4633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001850673387032654,
+      "loss": 0.1473,
+      "step": 4634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.001850599561177862,
+      "loss": 0.1553,
+      "step": 4635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0018505257185513946,
+      "loss": 0.1055,
+      "step": 4636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018504518591547076,
+      "loss": 0.1682,
+      "step": 4637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001850377982989257,
+      "loss": 0.1675,
+      "step": 4638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018503040900565003,
+      "loss": 0.1447,
+      "step": 4639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018502301803578934,
+      "loss": 0.0763,
+      "step": 4640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018501562538948946,
+      "loss": 0.0977,
+      "step": 4641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001850082310668961,
+      "loss": 0.071,
+      "step": 4642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018500083506815504,
+      "loss": 0.2024,
+      "step": 4643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018499343739341212,
+      "loss": 0.1306,
+      "step": 4644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018498603804281326,
+      "loss": 0.2092,
+      "step": 4645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018497863701650431,
+      "loss": 0.177,
+      "step": 4646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001849712343146312,
+      "loss": 0.0765,
+      "step": 4647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018496382993733991,
+      "loss": 0.1136,
+      "step": 4648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018495642388477642,
+      "loss": 0.1106,
+      "step": 4649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018494901615708677,
+      "loss": 0.1379,
+      "step": 4650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018494160675441701,
+      "loss": 0.1058,
+      "step": 4651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018493419567691322,
+      "loss": 0.1406,
+      "step": 4652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001849267829247216,
+      "loss": 0.1172,
+      "step": 4653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018491936849798823,
+      "loss": 0.0862,
+      "step": 4654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001849119523968593,
+      "loss": 0.1417,
+      "step": 4655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001849045346214811,
+      "loss": 0.1516,
+      "step": 4656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018489711517199988,
+      "loss": 0.1034,
+      "step": 4657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001848896940485619,
+      "loss": 0.0785,
+      "step": 4658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018488227125131346,
+      "loss": 0.1274,
+      "step": 4659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018487484678040098,
+      "loss": 0.151,
+      "step": 4660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018486742063597084,
+      "loss": 0.1283,
+      "step": 4661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018485999281816943,
+      "loss": 0.1213,
+      "step": 4662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018485256332714326,
+      "loss": 0.1455,
+      "step": 4663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018484513216303878,
+      "loss": 0.1543,
+      "step": 4664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018483769932600253,
+      "loss": 0.0968,
+      "step": 4665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018483026481618107,
+      "loss": 0.0927,
+      "step": 4666
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018482282863372095,
+      "loss": 0.0831,
+      "step": 4667
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018481539077876885,
+      "loss": 0.0901,
+      "step": 4668
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018480795125147141,
+      "loss": 0.0909,
+      "step": 4669
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018480051005197527,
+      "loss": 0.0816,
+      "step": 4670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018479306718042722,
+      "loss": 0.14,
+      "step": 4671
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018478562263697395,
+      "loss": 0.1776,
+      "step": 4672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018477817642176234,
+      "loss": 0.165,
+      "step": 4673
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018477072853493913,
+      "loss": 0.0767,
+      "step": 4674
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018476327897665118,
+      "loss": 0.1982,
+      "step": 4675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.001847558277470454,
+      "loss": 0.1101,
+      "step": 4676
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018474837484626873,
+      "loss": 0.0831,
+      "step": 4677
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018474092027446808,
+      "loss": 0.0738,
+      "step": 4678
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018473346403179043,
+      "loss": 0.0834,
+      "step": 4679
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018472600611838285,
+      "loss": 0.1312,
+      "step": 4680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0018471854653439234,
+      "loss": 0.1492,
+      "step": 4681
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00184711085279966,
+      "loss": 0.1368,
+      "step": 4682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018470362235525097,
+      "loss": 0.1407,
+      "step": 4683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018469615776039437,
+      "loss": 0.0972,
+      "step": 4684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018468869149554342,
+      "loss": 0.0699,
+      "step": 4685
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018468122356084526,
+      "loss": 0.1426,
+      "step": 4686
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018467375395644723,
+      "loss": 0.105,
+      "step": 4687
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018466628268249657,
+      "loss": 0.1077,
+      "step": 4688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018465880973914058,
+      "loss": 0.144,
+      "step": 4689
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018465133512652662,
+      "loss": 0.1276,
+      "step": 4690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018464385884480207,
+      "loss": 0.0824,
+      "step": 4691
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018463638089411436,
+      "loss": 0.125,
+      "step": 4692
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018462890127461094,
+      "loss": 0.0718,
+      "step": 4693
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018462141998643925,
+      "loss": 0.1403,
+      "step": 4694
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018461393702974681,
+      "loss": 0.13,
+      "step": 4695
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018460645240468124,
+      "loss": 0.1416,
+      "step": 4696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018459896611139,
+      "loss": 0.062,
+      "step": 4697
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018459147815002078,
+      "loss": 0.1479,
+      "step": 4698
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001845839885207212,
+      "loss": 0.1929,
+      "step": 4699
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018457649722363894,
+      "loss": 0.1012,
+      "step": 4700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001845690042589217,
+      "loss": 0.1177,
+      "step": 4701
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018456150962671725,
+      "loss": 0.1195,
+      "step": 4702
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018455401332717332,
+      "loss": 0.1562,
+      "step": 4703
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018454651536043776,
+      "loss": 0.158,
+      "step": 4704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018453901572665838,
+      "loss": 0.0818,
+      "step": 4705
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001845315144259831,
+      "loss": 0.059,
+      "step": 4706
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018452401145855975,
+      "loss": 0.1204,
+      "step": 4707
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018451650682453635,
+      "loss": 0.0817,
+      "step": 4708
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018450900052406083,
+      "loss": 0.1763,
+      "step": 4709
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001845014925572812,
+      "loss": 0.0861,
+      "step": 4710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018449398292434552,
+      "loss": 0.1239,
+      "step": 4711
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018448647162540182,
+      "loss": 0.0851,
+      "step": 4712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001844789586605982,
+      "loss": 0.0771,
+      "step": 4713
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001844714440300829,
+      "loss": 0.1304,
+      "step": 4714
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018446392773400393,
+      "loss": 0.1229,
+      "step": 4715
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018445640977250962,
+      "loss": 0.1009,
+      "step": 4716
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018444889014574814,
+      "loss": 0.0706,
+      "step": 4717
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.001844413688538678,
+      "loss": 0.1503,
+      "step": 4718
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018443384589701686,
+      "loss": 0.0681,
+      "step": 4719
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018442632127534367,
+      "loss": 0.152,
+      "step": 4720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018441879498899661,
+      "loss": 0.126,
+      "step": 4721
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018441126703812407,
+      "loss": 0.1262,
+      "step": 4722
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018440373742287445,
+      "loss": 0.0508,
+      "step": 4723
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018439620614339627,
+      "loss": 0.1079,
+      "step": 4724
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018438867319983802,
+      "loss": 0.0892,
+      "step": 4725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018438113859234816,
+      "loss": 0.1428,
+      "step": 4726
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018437360232107535,
+      "loss": 0.114,
+      "step": 4727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0018436606438616816,
+      "loss": 0.0762,
+      "step": 4728
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018435852478777518,
+      "loss": 0.1726,
+      "step": 4729
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001843509835260451,
+      "loss": 0.1575,
+      "step": 4730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018434344060112658,
+      "loss": 0.0842,
+      "step": 4731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018433589601316842,
+      "loss": 0.0961,
+      "step": 4732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001843283497623193,
+      "loss": 0.1519,
+      "step": 4733
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018432080184872808,
+      "loss": 0.1364,
+      "step": 4734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018431325227254351,
+      "loss": 0.0923,
+      "step": 4735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018430570103391455,
+      "loss": 0.1028,
+      "step": 4736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018429814813299,
+      "loss": 0.0892,
+      "step": 4737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018429059356991884,
+      "loss": 0.1006,
+      "step": 4738
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018428303734485,
+      "loss": 0.1621,
+      "step": 4739
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018427547945793247,
+      "loss": 0.1273,
+      "step": 4740
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018426791990931526,
+      "loss": 0.1045,
+      "step": 4741
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018426035869914747,
+      "loss": 0.1281,
+      "step": 4742
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018425279582757817,
+      "loss": 0.1991,
+      "step": 4743
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018424523129475646,
+      "loss": 0.0885,
+      "step": 4744
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018423766510083151,
+      "loss": 0.1609,
+      "step": 4745
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018423009724595249,
+      "loss": 0.0648,
+      "step": 4746
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001842225277302686,
+      "loss": 0.0853,
+      "step": 4747
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001842149565539292,
+      "loss": 0.1204,
+      "step": 4748
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018420738371708345,
+      "loss": 0.06,
+      "step": 4749
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001841998092198807,
+      "loss": 0.1233,
+      "step": 4750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018419223306247038,
+      "loss": 0.1317,
+      "step": 4751
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018418465524500174,
+      "loss": 0.1169,
+      "step": 4752
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001841770757676243,
+      "loss": 0.1367,
+      "step": 4753
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018416949463048744,
+      "loss": 0.1398,
+      "step": 4754
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018416191183374072,
+      "loss": 0.1423,
+      "step": 4755
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018415432737753355,
+      "loss": 0.1112,
+      "step": 4756
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018414674126201556,
+      "loss": 0.1033,
+      "step": 4757
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001841391534873363,
+      "loss": 0.1179,
+      "step": 4758
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001841315640536454,
+      "loss": 0.1732,
+      "step": 4759
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018412397296109244,
+      "loss": 0.115,
+      "step": 4760
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018411638020982717,
+      "loss": 0.1208,
+      "step": 4761
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018410878579999928,
+      "loss": 0.0754,
+      "step": 4762
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.001841011897317585,
+      "loss": 0.2537,
+      "step": 4763
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018409359200525462,
+      "loss": 0.1342,
+      "step": 4764
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018408599262063744,
+      "loss": 0.0549,
+      "step": 4765
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018407839157805678,
+      "loss": 0.0981,
+      "step": 4766
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018407078887766257,
+      "loss": 0.1718,
+      "step": 4767
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018406318451960465,
+      "loss": 0.1703,
+      "step": 4768
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018405557850403298,
+      "loss": 0.0808,
+      "step": 4769
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018404797083109757,
+      "loss": 0.1732,
+      "step": 4770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018404036150094839,
+      "loss": 0.1295,
+      "step": 4771
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018403275051373546,
+      "loss": 0.0697,
+      "step": 4772
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018402513786960888,
+      "loss": 0.2295,
+      "step": 4773
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0018401752356871875,
+      "loss": 0.1102,
+      "step": 4774
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001840099076112152,
+      "loss": 0.1017,
+      "step": 4775
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018400228999724839,
+      "loss": 0.0823,
+      "step": 4776
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018399467072696852,
+      "loss": 0.0767,
+      "step": 4777
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018398704980052583,
+      "loss": 0.1387,
+      "step": 4778
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018397942721807058,
+      "loss": 0.1426,
+      "step": 4779
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018397180297975307,
+      "loss": 0.1426,
+      "step": 4780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018396417708572365,
+      "loss": 0.1156,
+      "step": 4781
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018395654953613264,
+      "loss": 0.15,
+      "step": 4782
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018394892033113046,
+      "loss": 0.1075,
+      "step": 4783
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018394128947086755,
+      "loss": 0.167,
+      "step": 4784
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018393365695549438,
+      "loss": 0.1655,
+      "step": 4785
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001839260227851614,
+      "loss": 0.1074,
+      "step": 4786
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018391838696001916,
+      "loss": 0.0622,
+      "step": 4787
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018391074948021825,
+      "loss": 0.1644,
+      "step": 4788
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001839031103459092,
+      "loss": 0.1917,
+      "step": 4789
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001838954695572427,
+      "loss": 0.1313,
+      "step": 4790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018388782711436936,
+      "loss": 0.0992,
+      "step": 4791
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018388018301743988,
+      "loss": 0.1396,
+      "step": 4792
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018387253726660498,
+      "loss": 0.1155,
+      "step": 4793
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018386488986201544,
+      "loss": 0.1447,
+      "step": 4794
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00183857240803822,
+      "loss": 0.0845,
+      "step": 4795
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018384959009217555,
+      "loss": 0.196,
+      "step": 4796
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018384193772722685,
+      "loss": 0.1296,
+      "step": 4797
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018383428370912687,
+      "loss": 0.1432,
+      "step": 4798
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018382662803802649,
+      "loss": 0.0924,
+      "step": 4799
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018381897071407666,
+      "loss": 0.1783,
+      "step": 4800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018381131173742838,
+      "loss": 0.0814,
+      "step": 4801
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018380365110823263,
+      "loss": 0.0806,
+      "step": 4802
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001837959888266405,
+      "loss": 0.0906,
+      "step": 4803
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018378832489280305,
+      "loss": 0.0947,
+      "step": 4804
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001837806593068714,
+      "loss": 0.1333,
+      "step": 4805
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001837729920689967,
+      "loss": 0.0682,
+      "step": 4806
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001837653231793301,
+      "loss": 0.1586,
+      "step": 4807
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018375765263802285,
+      "loss": 0.1403,
+      "step": 4808
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018374998044522616,
+      "loss": 0.1951,
+      "step": 4809
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018374230660109135,
+      "loss": 0.1301,
+      "step": 4810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018373463110576966,
+      "loss": 0.1022,
+      "step": 4811
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018372695395941251,
+      "loss": 0.1165,
+      "step": 4812
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.001837192751621712,
+      "loss": 0.089,
+      "step": 4813
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018371159471419721,
+      "loss": 0.0742,
+      "step": 4814
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018370391261564192,
+      "loss": 0.1901,
+      "step": 4815
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018369622886665682,
+      "loss": 0.0938,
+      "step": 4816
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018368854346739343,
+      "loss": 0.1083,
+      "step": 4817
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018368085641800329,
+      "loss": 0.0862,
+      "step": 4818
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018367316771863793,
+      "loss": 0.0995,
+      "step": 4819
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0018366547736944896,
+      "loss": 0.104,
+      "step": 4820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018365778537058808,
+      "loss": 0.1189,
+      "step": 4821
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018365009172220685,
+      "loss": 0.1171,
+      "step": 4822
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001836423964244571,
+      "loss": 0.1169,
+      "step": 4823
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018363469947749041,
+      "loss": 0.1019,
+      "step": 4824
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018362700088145865,
+      "loss": 0.1027,
+      "step": 4825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018361930063651357,
+      "loss": 0.0991,
+      "step": 4826
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018361159874280704,
+      "loss": 0.088,
+      "step": 4827
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001836038952004909,
+      "loss": 0.1062,
+      "step": 4828
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018359619000971705,
+      "loss": 0.1428,
+      "step": 4829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018358848317063742,
+      "loss": 0.1431,
+      "step": 4830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018358077468340393,
+      "loss": 0.1334,
+      "step": 4831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018357306454816862,
+      "loss": 0.1584,
+      "step": 4832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018356535276508347,
+      "loss": 0.1208,
+      "step": 4833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018355763933430062,
+      "loss": 0.1786,
+      "step": 4834
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018354992425597206,
+      "loss": 0.0959,
+      "step": 4835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018354220753024997,
+      "loss": 0.1262,
+      "step": 4836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018353448915728648,
+      "loss": 0.0589,
+      "step": 4837
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018352676913723379,
+      "loss": 0.0851,
+      "step": 4838
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018351904747024412,
+      "loss": 0.0583,
+      "step": 4839
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001835113241564697,
+      "loss": 0.1144,
+      "step": 4840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018350359919606287,
+      "loss": 0.1082,
+      "step": 4841
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018349587258917586,
+      "loss": 0.1028,
+      "step": 4842
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001834881443359611,
+      "loss": 0.1042,
+      "step": 4843
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018348041443657093,
+      "loss": 0.082,
+      "step": 4844
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001834726828911578,
+      "loss": 0.0883,
+      "step": 4845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018346494969987408,
+      "loss": 0.075,
+      "step": 4846
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018345721486287233,
+      "loss": 0.1515,
+      "step": 4847
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018344947838030503,
+      "loss": 0.1227,
+      "step": 4848
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001834417402523247,
+      "loss": 0.1129,
+      "step": 4849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018343400047908397,
+      "loss": 0.1602,
+      "step": 4850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001834262590607354,
+      "loss": 0.15,
+      "step": 4851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018341851599743167,
+      "loss": 0.2004,
+      "step": 4852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018341077128932542,
+      "loss": 0.1401,
+      "step": 4853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018340302493656938,
+      "loss": 0.2052,
+      "step": 4854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018339527693931625,
+      "loss": 0.095,
+      "step": 4855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018338752729771886,
+      "loss": 0.1494,
+      "step": 4856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018337977601192996,
+      "loss": 0.1156,
+      "step": 4857
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018337202308210242,
+      "loss": 0.1534,
+      "step": 4858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018336426850838912,
+      "loss": 0.0775,
+      "step": 4859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001833565122909429,
+      "loss": 0.1116,
+      "step": 4860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018334875442991676,
+      "loss": 0.0819,
+      "step": 4861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018334099492546363,
+      "loss": 0.1282,
+      "step": 4862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018333323377773651,
+      "loss": 0.0854,
+      "step": 4863
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018332547098688843,
+      "loss": 0.1433,
+      "step": 4864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0018331770655307246,
+      "loss": 0.1382,
+      "step": 4865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.001833099404764417,
+      "loss": 0.098,
+      "step": 4866
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018330217275714924,
+      "loss": 0.0847,
+      "step": 4867
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001832944033953483,
+      "loss": 0.0792,
+      "step": 4868
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018328663239119203,
+      "loss": 0.1042,
+      "step": 4869
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018327885974483368,
+      "loss": 0.0779,
+      "step": 4870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018327108545642648,
+      "loss": 0.1152,
+      "step": 4871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001832633095261237,
+      "loss": 0.1129,
+      "step": 4872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018325553195407873,
+      "loss": 0.1428,
+      "step": 4873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001832477527404449,
+      "loss": 0.0823,
+      "step": 4874
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018323997188537558,
+      "loss": 0.0751,
+      "step": 4875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018323218938902418,
+      "loss": 0.116,
+      "step": 4876
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018322440525154417,
+      "loss": 0.0986,
+      "step": 4877
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018321661947308901,
+      "loss": 0.1195,
+      "step": 4878
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018320883205381225,
+      "loss": 0.1339,
+      "step": 4879
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001832010429938674,
+      "loss": 0.1011,
+      "step": 4880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018319325229340811,
+      "loss": 0.1167,
+      "step": 4881
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001831854599525879,
+      "loss": 0.125,
+      "step": 4882
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018317766597156049,
+      "loss": 0.1019,
+      "step": 4883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001831698703504795,
+      "loss": 0.132,
+      "step": 4884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001831620730894987,
+      "loss": 0.1318,
+      "step": 4885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001831542741887718,
+      "loss": 0.1136,
+      "step": 4886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018314647364845254,
+      "loss": 0.1614,
+      "step": 4887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001831386714686948,
+      "loss": 0.0934,
+      "step": 4888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018313086764965238,
+      "loss": 0.1581,
+      "step": 4889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018312306219147915,
+      "loss": 0.0968,
+      "step": 4890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00183115255094329,
+      "loss": 0.1437,
+      "step": 4891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018310744635835587,
+      "loss": 0.1215,
+      "step": 4892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018309963598371378,
+      "loss": 0.1799,
+      "step": 4893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018309182397055665,
+      "loss": 0.0782,
+      "step": 4894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.001830840103190386,
+      "loss": 0.0961,
+      "step": 4895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018307619502931363,
+      "loss": 0.0995,
+      "step": 4896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018306837810153588,
+      "loss": 0.1077,
+      "step": 4897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018306055953585944,
+      "loss": 0.1487,
+      "step": 4898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018305273933243848,
+      "loss": 0.1316,
+      "step": 4899
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018304491749142721,
+      "loss": 0.1459,
+      "step": 4900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018303709401297987,
+      "loss": 0.1654,
+      "step": 4901
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018302926889725072,
+      "loss": 0.0753,
+      "step": 4902
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00183021442144394,
+      "loss": 0.0705,
+      "step": 4903
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018301361375456407,
+      "loss": 0.1024,
+      "step": 4904
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018300578372791529,
+      "loss": 0.1271,
+      "step": 4905
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018299795206460205,
+      "loss": 0.1193,
+      "step": 4906
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018299011876477876,
+      "loss": 0.1323,
+      "step": 4907
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018298228382859988,
+      "loss": 0.1262,
+      "step": 4908
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018297444725621985,
+      "loss": 0.1041,
+      "step": 4909
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018296660904779328,
+      "loss": 0.1292,
+      "step": 4910
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018295876920347465,
+      "loss": 0.1544,
+      "step": 4911
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0018295092772341857,
+      "loss": 0.1315,
+      "step": 4912
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001829430846077796,
+      "loss": 0.0665,
+      "step": 4913
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001829352398567125,
+      "loss": 0.1553,
+      "step": 4914
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018292739347037187,
+      "loss": 0.1656,
+      "step": 4915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001829195454489124,
+      "loss": 0.1107,
+      "step": 4916
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018291169579248885,
+      "loss": 0.118,
+      "step": 4917
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018290384450125603,
+      "loss": 0.1088,
+      "step": 4918
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018289599157536873,
+      "loss": 0.1704,
+      "step": 4919
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001828881370149818,
+      "loss": 0.0979,
+      "step": 4920
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018288028082025006,
+      "loss": 0.1013,
+      "step": 4921
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018287242299132847,
+      "loss": 0.0971,
+      "step": 4922
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018286456352837196,
+      "loss": 0.1071,
+      "step": 4923
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018285670243153548,
+      "loss": 0.1335,
+      "step": 4924
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018284883970097403,
+      "loss": 0.1041,
+      "step": 4925
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018284097533684267,
+      "loss": 0.093,
+      "step": 4926
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018283310933929645,
+      "loss": 0.1035,
+      "step": 4927
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018282524170849044,
+      "loss": 0.1774,
+      "step": 4928
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001828173724445798,
+      "loss": 0.1143,
+      "step": 4929
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001828095015477197,
+      "loss": 0.1069,
+      "step": 4930
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018280162901806534,
+      "loss": 0.1414,
+      "step": 4931
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001827937548557719,
+      "loss": 0.1162,
+      "step": 4932
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018278587906099468,
+      "loss": 0.1725,
+      "step": 4933
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018277800163388896,
+      "loss": 0.1295,
+      "step": 4934
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018277012257461004,
+      "loss": 0.074,
+      "step": 4935
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001827622418833133,
+      "loss": 0.177,
+      "step": 4936
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018275435956015414,
+      "loss": 0.1038,
+      "step": 4937
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018274647560528794,
+      "loss": 0.1549,
+      "step": 4938
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001827385900188702,
+      "loss": 0.0928,
+      "step": 4939
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018273070280105635,
+      "loss": 0.1335,
+      "step": 4940
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018272281395200194,
+      "loss": 0.0652,
+      "step": 4941
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018271492347186253,
+      "loss": 0.1244,
+      "step": 4942
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018270703136079366,
+      "loss": 0.179,
+      "step": 4943
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00182699137618951,
+      "loss": 0.1255,
+      "step": 4944
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001826912422464901,
+      "loss": 0.0895,
+      "step": 4945
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018268334524356674,
+      "loss": 0.1732,
+      "step": 4946
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018267544661033659,
+      "loss": 0.0825,
+      "step": 4947
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018266754634695538,
+      "loss": 0.1479,
+      "step": 4948
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018265964445357886,
+      "loss": 0.0916,
+      "step": 4949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018265174093036287,
+      "loss": 0.1484,
+      "step": 4950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018264383577746327,
+      "loss": 0.1204,
+      "step": 4951
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001826359289950359,
+      "loss": 0.1554,
+      "step": 4952
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018262802058323663,
+      "loss": 0.1642,
+      "step": 4953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018262011054222147,
+      "loss": 0.0701,
+      "step": 4954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018261219887214632,
+      "loss": 0.1569,
+      "step": 4955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.001826042855731672,
+      "loss": 0.0873,
+      "step": 4956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018259637064544016,
+      "loss": 0.152,
+      "step": 4957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0018258845408912125,
+      "loss": 0.0578,
+      "step": 4958
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018258053590436655,
+      "loss": 0.1034,
+      "step": 4959
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018257261609133217,
+      "loss": 0.1091,
+      "step": 4960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001825646946501743,
+      "loss": 0.1129,
+      "step": 4961
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018255677158104918,
+      "loss": 0.0902,
+      "step": 4962
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001825488468841129,
+      "loss": 0.1387,
+      "step": 4963
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018254092055952184,
+      "loss": 0.1176,
+      "step": 4964
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018253299260743224,
+      "loss": 0.1194,
+      "step": 4965
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018252506302800042,
+      "loss": 0.0819,
+      "step": 4966
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001825171318213827,
+      "loss": 0.1422,
+      "step": 4967
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018250919898773553,
+      "loss": 0.1284,
+      "step": 4968
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018250126452721528,
+      "loss": 0.1221,
+      "step": 4969
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018249332843997838,
+      "loss": 0.0684,
+      "step": 4970
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018248539072618138,
+      "loss": 0.0737,
+      "step": 4971
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018247745138598073,
+      "loss": 0.1423,
+      "step": 4972
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00182469510419533,
+      "loss": 0.0952,
+      "step": 4973
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018246156782699475,
+      "loss": 0.0911,
+      "step": 4974
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018245362360852259,
+      "loss": 0.0955,
+      "step": 4975
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018244567776427316,
+      "loss": 0.1398,
+      "step": 4976
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018243773029440313,
+      "loss": 0.1017,
+      "step": 4977
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018242978119906922,
+      "loss": 0.1188,
+      "step": 4978
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018242183047842819,
+      "loss": 0.1392,
+      "step": 4979
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018241387813263674,
+      "loss": 0.1066,
+      "step": 4980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018240592416185173,
+      "loss": 0.1052,
+      "step": 4981
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018239796856622996,
+      "loss": 0.15,
+      "step": 4982
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001823900113459283,
+      "loss": 0.0871,
+      "step": 4983
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018238205250110365,
+      "loss": 0.0894,
+      "step": 4984
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018237409203191294,
+      "loss": 0.1605,
+      "step": 4985
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018236612993851311,
+      "loss": 0.1088,
+      "step": 4986
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001823581662210612,
+      "loss": 0.0745,
+      "step": 4987
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018235020087971417,
+      "loss": 0.0704,
+      "step": 4988
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018234223391462915,
+      "loss": 0.1088,
+      "step": 4989
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018233426532596317,
+      "loss": 0.1006,
+      "step": 4990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018232629511387335,
+      "loss": 0.0911,
+      "step": 4991
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018231832327851687,
+      "loss": 0.0902,
+      "step": 4992
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018231034982005095,
+      "loss": 0.1375,
+      "step": 4993
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001823023747386327,
+      "loss": 0.0995,
+      "step": 4994
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018229439803441948,
+      "loss": 0.0989,
+      "step": 4995
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018228641970756852,
+      "loss": 0.09,
+      "step": 4996
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001822784397582371,
+      "loss": 0.093,
+      "step": 4997
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001822704581865826,
+      "loss": 0.0933,
+      "step": 4998
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018226247499276242,
+      "loss": 0.079,
+      "step": 4999
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018225449017693395,
+      "loss": 0.1006,
+      "step": 5000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018224650373925462,
+      "loss": 0.0957,
+      "step": 5001
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001822385156798819,
+      "loss": 0.1139,
+      "step": 5002
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.001822305259989733,
+      "loss": 0.0917,
+      "step": 5003
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018222253469668632,
+      "loss": 0.103,
+      "step": 5004
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0018221454177317863,
+      "loss": 0.166,
+      "step": 5005
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018220654722860773,
+      "loss": 0.1542,
+      "step": 5006
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001821985510631313,
+      "loss": 0.0851,
+      "step": 5007
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00182190553276907,
+      "loss": 0.1317,
+      "step": 5008
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018218255387009248,
+      "loss": 0.1199,
+      "step": 5009
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018217455284284553,
+      "loss": 0.1335,
+      "step": 5010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018216655019532388,
+      "loss": 0.102,
+      "step": 5011
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018215854592768537,
+      "loss": 0.116,
+      "step": 5012
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018215054004008773,
+      "loss": 0.1306,
+      "step": 5013
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018214253253268891,
+      "loss": 0.1298,
+      "step": 5014
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018213452340564672,
+      "loss": 0.1079,
+      "step": 5015
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018212651265911912,
+      "loss": 0.2363,
+      "step": 5016
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018211850029326406,
+      "loss": 0.0741,
+      "step": 5017
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018211048630823955,
+      "loss": 0.0901,
+      "step": 5018
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018210247070420357,
+      "loss": 0.1324,
+      "step": 5019
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018209445348131414,
+      "loss": 0.1025,
+      "step": 5020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001820864346397294,
+      "loss": 0.1211,
+      "step": 5021
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018207841417960744,
+      "loss": 0.099,
+      "step": 5022
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018207039210110638,
+      "loss": 0.1171,
+      "step": 5023
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001820623684043844,
+      "loss": 0.1024,
+      "step": 5024
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001820543430895998,
+      "loss": 0.1532,
+      "step": 5025
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018204631615691064,
+      "loss": 0.11,
+      "step": 5026
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018203828760647537,
+      "loss": 0.0875,
+      "step": 5027
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018203025743845217,
+      "loss": 0.1448,
+      "step": 5028
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018202222565299939,
+      "loss": 0.0878,
+      "step": 5029
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018201419225027547,
+      "loss": 0.1152,
+      "step": 5030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018200615723043875,
+      "loss": 0.0683,
+      "step": 5031
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018199812059364767,
+      "loss": 0.0884,
+      "step": 5032
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001819900823400607,
+      "loss": 0.1443,
+      "step": 5033
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018198204246983633,
+      "loss": 0.1147,
+      "step": 5034
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018197400098313309,
+      "loss": 0.1149,
+      "step": 5035
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018196595788010953,
+      "loss": 0.1616,
+      "step": 5036
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018195791316092424,
+      "loss": 0.0684,
+      "step": 5037
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018194986682573582,
+      "loss": 0.0696,
+      "step": 5038
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018194181887470299,
+      "loss": 0.1072,
+      "step": 5039
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018193376930798436,
+      "loss": 0.1084,
+      "step": 5040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018192571812573868,
+      "loss": 0.1099,
+      "step": 5041
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001819176653281247,
+      "loss": 0.0832,
+      "step": 5042
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018190961091530122,
+      "loss": 0.1129,
+      "step": 5043
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018190155488742703,
+      "loss": 0.1047,
+      "step": 5044
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018189349724466097,
+      "loss": 0.0959,
+      "step": 5045
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018188543798716194,
+      "loss": 0.0634,
+      "step": 5046
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001818773771150888,
+      "loss": 0.0845,
+      "step": 5047
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018186931462860051,
+      "loss": 0.1025,
+      "step": 5048
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0018186125052785612,
+      "loss": 0.0881,
+      "step": 5049
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001818531848130145,
+      "loss": 0.0766,
+      "step": 5050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.001818451174842348,
+      "loss": 0.1355,
+      "step": 5051
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00181837048541676,
+      "loss": 0.1198,
+      "step": 5052
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018182897798549728,
+      "loss": 0.0728,
+      "step": 5053
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001818209058158577,
+      "loss": 0.1183,
+      "step": 5054
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018181283203291647,
+      "loss": 0.1329,
+      "step": 5055
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018180475663683276,
+      "loss": 0.1221,
+      "step": 5056
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001817966796277658,
+      "loss": 0.0687,
+      "step": 5057
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018178860100587487,
+      "loss": 0.1284,
+      "step": 5058
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018178052077131924,
+      "loss": 0.0909,
+      "step": 5059
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001817724389242582,
+      "loss": 0.1261,
+      "step": 5060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018176435546485114,
+      "loss": 0.0765,
+      "step": 5061
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001817562703932575,
+      "loss": 0.0731,
+      "step": 5062
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018174818370963659,
+      "loss": 0.0742,
+      "step": 5063
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018174009541414794,
+      "loss": 0.1019,
+      "step": 5064
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018173200550695098,
+      "loss": 0.1188,
+      "step": 5065
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018172391398820526,
+      "loss": 0.1036,
+      "step": 5066
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018171582085807028,
+      "loss": 0.0698,
+      "step": 5067
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018170772611670564,
+      "loss": 0.0839,
+      "step": 5068
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018169962976427097,
+      "loss": 0.0714,
+      "step": 5069
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001816915318009259,
+      "loss": 0.2056,
+      "step": 5070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018168343222683006,
+      "loss": 0.0771,
+      "step": 5071
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018167533104214322,
+      "loss": 0.0724,
+      "step": 5072
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018166722824702506,
+      "loss": 0.125,
+      "step": 5073
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018165912384163537,
+      "loss": 0.0888,
+      "step": 5074
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018165101782613396,
+      "loss": 0.1307,
+      "step": 5075
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018164291020068063,
+      "loss": 0.0927,
+      "step": 5076
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018163480096543526,
+      "loss": 0.1045,
+      "step": 5077
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018162669012055774,
+      "loss": 0.0771,
+      "step": 5078
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00181618577666208,
+      "loss": 0.1129,
+      "step": 5079
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018161046360254598,
+      "loss": 0.119,
+      "step": 5080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018160234792973168,
+      "loss": 0.0964,
+      "step": 5081
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018159423064792516,
+      "loss": 0.1161,
+      "step": 5082
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001815861117572864,
+      "loss": 0.1021,
+      "step": 5083
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018157799125797555,
+      "loss": 0.0839,
+      "step": 5084
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018156986915015265,
+      "loss": 0.1411,
+      "step": 5085
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018156174543397792,
+      "loss": 0.0759,
+      "step": 5086
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018155362010961152,
+      "loss": 0.1242,
+      "step": 5087
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018154549317721362,
+      "loss": 0.1537,
+      "step": 5088
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018153736463694453,
+      "loss": 0.0842,
+      "step": 5089
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018152923448896447,
+      "loss": 0.1816,
+      "step": 5090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018152110273343377,
+      "loss": 0.1235,
+      "step": 5091
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018151296937051274,
+      "loss": 0.0934,
+      "step": 5092
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001815048344003618,
+      "loss": 0.0986,
+      "step": 5093
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018149669782314133,
+      "loss": 0.1682,
+      "step": 5094
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018148855963901172,
+      "loss": 0.1094,
+      "step": 5095
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.001814804198481335,
+      "loss": 0.0621,
+      "step": 5096
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0018147227845066714,
+      "loss": 0.0991,
+      "step": 5097
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018146413544677315,
+      "loss": 0.1609,
+      "step": 5098
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018145599083661211,
+      "loss": 0.1665,
+      "step": 5099
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.001814478446203446,
+      "loss": 0.0776,
+      "step": 5100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018143969679813123,
+      "loss": 0.084,
+      "step": 5101
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018143154737013272,
+      "loss": 0.0876,
+      "step": 5102
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018142339633650968,
+      "loss": 0.1311,
+      "step": 5103
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018141524369742286,
+      "loss": 0.1121,
+      "step": 5104
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00181407089453033,
+      "loss": 0.0854,
+      "step": 5105
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018139893360350089,
+      "loss": 0.1053,
+      "step": 5106
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018139077614898733,
+      "loss": 0.1301,
+      "step": 5107
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018138261708965319,
+      "loss": 0.162,
+      "step": 5108
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018137445642565932,
+      "loss": 0.091,
+      "step": 5109
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018136629415716663,
+      "loss": 0.1207,
+      "step": 5110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018135813028433607,
+      "loss": 0.0722,
+      "step": 5111
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018134996480732863,
+      "loss": 0.1547,
+      "step": 5112
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.001813417977263053,
+      "loss": 0.1781,
+      "step": 5113
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018133362904142708,
+      "loss": 0.0746,
+      "step": 5114
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018132545875285504,
+      "loss": 0.0666,
+      "step": 5115
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018131728686075032,
+      "loss": 0.0823,
+      "step": 5116
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018130911336527405,
+      "loss": 0.1356,
+      "step": 5117
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018130093826658734,
+      "loss": 0.0952,
+      "step": 5118
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018129276156485141,
+      "loss": 0.1225,
+      "step": 5119
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018128458326022749,
+      "loss": 0.1104,
+      "step": 5120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018127640335287682,
+      "loss": 0.2035,
+      "step": 5121
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018126822184296072,
+      "loss": 0.0893,
+      "step": 5122
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018126003873064048,
+      "loss": 0.0868,
+      "step": 5123
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018125185401607743,
+      "loss": 0.1085,
+      "step": 5124
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018124366769943301,
+      "loss": 0.1034,
+      "step": 5125
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018123547978086858,
+      "loss": 0.1393,
+      "step": 5126
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018122729026054563,
+      "loss": 0.0964,
+      "step": 5127
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.001812190991386256,
+      "loss": 0.1637,
+      "step": 5128
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018121090641527,
+      "loss": 0.0707,
+      "step": 5129
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.001812027120906404,
+      "loss": 0.1368,
+      "step": 5130
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018119451616489832,
+      "loss": 0.1006,
+      "step": 5131
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018118631863820542,
+      "loss": 0.1736,
+      "step": 5132
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.001811781195107233,
+      "loss": 0.1104,
+      "step": 5133
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018116991878261363,
+      "loss": 0.1094,
+      "step": 5134
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018116171645403815,
+      "loss": 0.1143,
+      "step": 5135
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018115351252515853,
+      "loss": 0.1717,
+      "step": 5136
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018114530699613655,
+      "loss": 0.0865,
+      "step": 5137
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00181137099867134,
+      "loss": 0.1917,
+      "step": 5138
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018112889113831275,
+      "loss": 0.1117,
+      "step": 5139
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018112068080983455,
+      "loss": 0.1028,
+      "step": 5140
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.001811124688818614,
+      "loss": 0.1057,
+      "step": 5141
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018110425535455515,
+      "loss": 0.0947,
+      "step": 5142
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0018109604022807778,
+      "loss": 0.124,
+      "step": 5143
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018108782350259125,
+      "loss": 0.1129,
+      "step": 5144
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001810796051782576,
+      "loss": 0.0638,
+      "step": 5145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018107138525523885,
+      "loss": 0.0967,
+      "step": 5146
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018106316373369708,
+      "loss": 0.064,
+      "step": 5147
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001810549406137944,
+      "loss": 0.0798,
+      "step": 5148
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00181046715895693,
+      "loss": 0.0864,
+      "step": 5149
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018103848957955495,
+      "loss": 0.0808,
+      "step": 5150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018103026166554253,
+      "loss": 0.0945,
+      "step": 5151
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018102203215381793,
+      "loss": 0.0829,
+      "step": 5152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018101380104454346,
+      "loss": 0.1604,
+      "step": 5153
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018100556833788137,
+      "loss": 0.1465,
+      "step": 5154
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018099733403399406,
+      "loss": 0.1454,
+      "step": 5155
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018098909813304377,
+      "loss": 0.073,
+      "step": 5156
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018098086063519298,
+      "loss": 0.1179,
+      "step": 5157
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001809726215406041,
+      "loss": 0.0802,
+      "step": 5158
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018096438084943959,
+      "loss": 0.0956,
+      "step": 5159
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018095613856186192,
+      "loss": 0.1481,
+      "step": 5160
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018094789467803358,
+      "loss": 0.0942,
+      "step": 5161
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018093964919811717,
+      "loss": 0.1378,
+      "step": 5162
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018093140212227525,
+      "loss": 0.1655,
+      "step": 5163
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018092315345067042,
+      "loss": 0.1185,
+      "step": 5164
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018091490318346538,
+      "loss": 0.1335,
+      "step": 5165
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018090665132082272,
+      "loss": 0.1407,
+      "step": 5166
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018089839786290516,
+      "loss": 0.0875,
+      "step": 5167
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018089014280987547,
+      "loss": 0.1323,
+      "step": 5168
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018088188616189643,
+      "loss": 0.1273,
+      "step": 5169
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018087362791913081,
+      "loss": 0.1458,
+      "step": 5170
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018086536808174145,
+      "loss": 0.0991,
+      "step": 5171
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018085710664989124,
+      "loss": 0.0984,
+      "step": 5172
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00180848843623743,
+      "loss": 0.106,
+      "step": 5173
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018084057900345973,
+      "loss": 0.0839,
+      "step": 5174
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018083231278920439,
+      "loss": 0.0845,
+      "step": 5175
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018082404498113989,
+      "loss": 0.1039,
+      "step": 5176
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018081577557942934,
+      "loss": 0.1172,
+      "step": 5177
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018080750458423574,
+      "loss": 0.0999,
+      "step": 5178
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018079923199572218,
+      "loss": 0.1035,
+      "step": 5179
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018079095781405178,
+      "loss": 0.1403,
+      "step": 5180
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001807826820393877,
+      "loss": 0.152,
+      "step": 5181
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001807744046718931,
+      "loss": 0.0968,
+      "step": 5182
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001807661257117312,
+      "loss": 0.1127,
+      "step": 5183
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018075784515906522,
+      "loss": 0.0989,
+      "step": 5184
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018074956301405846,
+      "loss": 0.1143,
+      "step": 5185
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.001807412792768742,
+      "loss": 0.1532,
+      "step": 5186
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018073299394767576,
+      "loss": 0.1028,
+      "step": 5187
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018072470702662656,
+      "loss": 0.1219,
+      "step": 5188
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0018071641851388996,
+      "loss": 0.1238,
+      "step": 5189
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001807081284096294,
+      "loss": 0.1365,
+      "step": 5190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018069983671400829,
+      "loss": 0.07,
+      "step": 5191
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018069154342719022,
+      "loss": 0.109,
+      "step": 5192
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018068324854933861,
+      "loss": 0.1575,
+      "step": 5193
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001806749520806171,
+      "loss": 0.0822,
+      "step": 5194
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001806666540211892,
+      "loss": 0.1201,
+      "step": 5195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001806583543712186,
+      "loss": 0.0692,
+      "step": 5196
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001806500531308689,
+      "loss": 0.0716,
+      "step": 5197
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001806417503003038,
+      "loss": 0.0828,
+      "step": 5198
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00180633445879687,
+      "loss": 0.0918,
+      "step": 5199
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018062513986918222,
+      "loss": 0.1002,
+      "step": 5200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001806168322689533,
+      "loss": 0.0974,
+      "step": 5201
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018060852307916395,
+      "loss": 0.0793,
+      "step": 5202
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018060021229997808,
+      "loss": 0.1245,
+      "step": 5203
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018059189993155959,
+      "loss": 0.0858,
+      "step": 5204
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018058358597407228,
+      "loss": 0.1193,
+      "step": 5205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001805752704276801,
+      "loss": 0.1179,
+      "step": 5206
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018056695329254709,
+      "loss": 0.0966,
+      "step": 5207
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018055863456883715,
+      "loss": 0.0911,
+      "step": 5208
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018055031425671436,
+      "loss": 0.0499,
+      "step": 5209
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018054199235634277,
+      "loss": 0.0936,
+      "step": 5210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018053366886788642,
+      "loss": 0.1171,
+      "step": 5211
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018052534379150952,
+      "loss": 0.1722,
+      "step": 5212
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018051701712737613,
+      "loss": 0.0437,
+      "step": 5213
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018050868887565046,
+      "loss": 0.1422,
+      "step": 5214
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018050035903649675,
+      "loss": 0.1382,
+      "step": 5215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018049202761007922,
+      "loss": 0.0489,
+      "step": 5216
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018048369459656214,
+      "loss": 0.1636,
+      "step": 5217
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018047535999610982,
+      "loss": 0.0673,
+      "step": 5218
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001804670238088866,
+      "loss": 0.1514,
+      "step": 5219
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018045868603505687,
+      "loss": 0.1158,
+      "step": 5220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018045034667478496,
+      "loss": 0.1437,
+      "step": 5221
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001804420057282354,
+      "loss": 0.0668,
+      "step": 5222
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018043366319557255,
+      "loss": 0.0918,
+      "step": 5223
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018042531907696098,
+      "loss": 0.0947,
+      "step": 5224
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.001804169733725652,
+      "loss": 0.1272,
+      "step": 5225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018040862608254974,
+      "loss": 0.1207,
+      "step": 5226
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018040027720707921,
+      "loss": 0.08,
+      "step": 5227
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018039192674631822,
+      "loss": 0.0703,
+      "step": 5228
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018038357470043143,
+      "loss": 0.1273,
+      "step": 5229
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018037522106958347,
+      "loss": 0.0819,
+      "step": 5230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018036686585393914,
+      "loss": 0.0789,
+      "step": 5231
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018035850905366315,
+      "loss": 0.0913,
+      "step": 5232
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018035015066892024,
+      "loss": 0.1005,
+      "step": 5233
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018034179069987527,
+      "loss": 0.1609,
+      "step": 5234
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0018033342914669302,
+      "loss": 0.0845,
+      "step": 5235
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001803250660095384,
+      "loss": 0.1044,
+      "step": 5236
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018031670128857628,
+      "loss": 0.0775,
+      "step": 5237
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001803083349839716,
+      "loss": 0.1511,
+      "step": 5238
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018029996709588937,
+      "loss": 0.115,
+      "step": 5239
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018029159762449453,
+      "loss": 0.131,
+      "step": 5240
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001802832265699521,
+      "loss": 0.1268,
+      "step": 5241
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018027485393242717,
+      "loss": 0.093,
+      "step": 5242
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001802664797120848,
+      "loss": 0.1116,
+      "step": 5243
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018025810390909015,
+      "loss": 0.0975,
+      "step": 5244
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018024972652360832,
+      "loss": 0.068,
+      "step": 5245
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001802413475558045,
+      "loss": 0.1437,
+      "step": 5246
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018023296700584394,
+      "loss": 0.1239,
+      "step": 5247
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018022458487389184,
+      "loss": 0.0931,
+      "step": 5248
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001802162011601135,
+      "loss": 0.1205,
+      "step": 5249
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018020781586467422,
+      "loss": 0.129,
+      "step": 5250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001801994289877393,
+      "loss": 0.0936,
+      "step": 5251
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018019104052947417,
+      "loss": 0.1094,
+      "step": 5252
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001801826504900442,
+      "loss": 0.1542,
+      "step": 5253
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018017425886961483,
+      "loss": 0.1133,
+      "step": 5254
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001801658656683515,
+      "loss": 0.1115,
+      "step": 5255
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018015747088641972,
+      "loss": 0.0848,
+      "step": 5256
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00180149074523985,
+      "loss": 0.1357,
+      "step": 5257
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018014067658121294,
+      "loss": 0.1692,
+      "step": 5258
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018013227705826905,
+      "loss": 0.1072,
+      "step": 5259
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00180123875955319,
+      "loss": 0.1421,
+      "step": 5260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018011547327252844,
+      "loss": 0.1128,
+      "step": 5261
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018010706901006302,
+      "loss": 0.0968,
+      "step": 5262
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001800986631680885,
+      "loss": 0.1152,
+      "step": 5263
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018009025574677055,
+      "loss": 0.1505,
+      "step": 5264
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00180081846746275,
+      "loss": 0.0936,
+      "step": 5265
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001800734361667676,
+      "loss": 0.0939,
+      "step": 5266
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018006502400841429,
+      "loss": 0.097,
+      "step": 5267
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018005661027138084,
+      "loss": 0.1206,
+      "step": 5268
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018004819495583315,
+      "loss": 0.0953,
+      "step": 5269
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018003977806193721,
+      "loss": 0.0788,
+      "step": 5270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018003135958985892,
+      "loss": 0.1501,
+      "step": 5271
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001800229395397643,
+      "loss": 0.1113,
+      "step": 5272
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0018001451791181938,
+      "loss": 0.0962,
+      "step": 5273
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.001800060947061902,
+      "loss": 0.1047,
+      "step": 5274
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017999766992304283,
+      "loss": 0.14,
+      "step": 5275
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017998924356254341,
+      "loss": 0.0694,
+      "step": 5276
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017998081562485809,
+      "loss": 0.0623,
+      "step": 5277
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017997238611015304,
+      "loss": 0.0997,
+      "step": 5278
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017996395501859443,
+      "loss": 0.0963,
+      "step": 5279
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017995552235034857,
+      "loss": 0.0762,
+      "step": 5280
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0017994708810558167,
+      "loss": 0.0654,
+      "step": 5281
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017993865228446008,
+      "loss": 0.1152,
+      "step": 5282
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017993021488715009,
+      "loss": 0.113,
+      "step": 5283
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017992177591381808,
+      "loss": 0.1179,
+      "step": 5284
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017991333536463046,
+      "loss": 0.134,
+      "step": 5285
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017990489323975365,
+      "loss": 0.0917,
+      "step": 5286
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017989644953935412,
+      "loss": 0.1019,
+      "step": 5287
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017988800426359832,
+      "loss": 0.1848,
+      "step": 5288
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017987955741265277,
+      "loss": 0.0895,
+      "step": 5289
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017987110898668407,
+      "loss": 0.1357,
+      "step": 5290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017986265898585875,
+      "loss": 0.1163,
+      "step": 5291
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017985420741034347,
+      "loss": 0.0974,
+      "step": 5292
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017984575426030487,
+      "loss": 0.0675,
+      "step": 5293
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017983729953590956,
+      "loss": 0.0996,
+      "step": 5294
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017982884323732432,
+      "loss": 0.1189,
+      "step": 5295
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017982038536471583,
+      "loss": 0.081,
+      "step": 5296
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017981192591825092,
+      "loss": 0.0883,
+      "step": 5297
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017980346489809632,
+      "loss": 0.0739,
+      "step": 5298
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.001797950023044189,
+      "loss": 0.099,
+      "step": 5299
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017978653813738555,
+      "loss": 0.0879,
+      "step": 5300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017977807239716308,
+      "loss": 0.1162,
+      "step": 5301
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.001797696050839185,
+      "loss": 0.0803,
+      "step": 5302
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.001797611361978187,
+      "loss": 0.1191,
+      "step": 5303
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017975266573903067,
+      "loss": 0.0911,
+      "step": 5304
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017974419370772145,
+      "loss": 0.0629,
+      "step": 5305
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017973572010405812,
+      "loss": 0.0995,
+      "step": 5306
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.001797272449282077,
+      "loss": 0.1428,
+      "step": 5307
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017971876818033733,
+      "loss": 0.0778,
+      "step": 5308
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017971028986061414,
+      "loss": 0.0806,
+      "step": 5309
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017970180996920528,
+      "loss": 0.0736,
+      "step": 5310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017969332850627797,
+      "loss": 0.1133,
+      "step": 5311
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017968484547199945,
+      "loss": 0.0922,
+      "step": 5312
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00179676360866537,
+      "loss": 0.1409,
+      "step": 5313
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017966787469005788,
+      "loss": 0.1385,
+      "step": 5314
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017965938694272942,
+      "loss": 0.1062,
+      "step": 5315
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00179650897624719,
+      "loss": 0.1055,
+      "step": 5316
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00179642406736194,
+      "loss": 0.0982,
+      "step": 5317
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017963391427732183,
+      "loss": 0.1276,
+      "step": 5318
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017962542024826994,
+      "loss": 0.1202,
+      "step": 5319
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017961692464920583,
+      "loss": 0.0994,
+      "step": 5320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017960842748029698,
+      "loss": 0.1205,
+      "step": 5321
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017959992874171096,
+      "loss": 0.1578,
+      "step": 5322
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017959142843361532,
+      "loss": 0.1117,
+      "step": 5323
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017958292655617769,
+      "loss": 0.1095,
+      "step": 5324
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017957442310956569,
+      "loss": 0.0656,
+      "step": 5325
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017956591809394698,
+      "loss": 0.0958,
+      "step": 5326
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0017955741150948927,
+      "loss": 0.1418,
+      "step": 5327
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.001795489033563603,
+      "loss": 0.1171,
+      "step": 5328
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017954039363472778,
+      "loss": 0.1335,
+      "step": 5329
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017953188234475959,
+      "loss": 0.1317,
+      "step": 5330
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017952336948662345,
+      "loss": 0.1083,
+      "step": 5331
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017951485506048728,
+      "loss": 0.1348,
+      "step": 5332
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017950633906651894,
+      "loss": 0.1448,
+      "step": 5333
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.001794978215048863,
+      "loss": 0.0956,
+      "step": 5334
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017948930237575743,
+      "loss": 0.0864,
+      "step": 5335
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017948078167930018,
+      "loss": 0.146,
+      "step": 5336
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.001794722594156826,
+      "loss": 0.1161,
+      "step": 5337
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017946373558507273,
+      "loss": 0.1094,
+      "step": 5338
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017945521018763867,
+      "loss": 0.0732,
+      "step": 5339
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017944668322354845,
+      "loss": 0.1121,
+      "step": 5340
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017943815469297025,
+      "loss": 0.1481,
+      "step": 5341
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017942962459607222,
+      "loss": 0.0767,
+      "step": 5342
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017942109293302257,
+      "loss": 0.0809,
+      "step": 5343
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017941255970398947,
+      "loss": 0.1168,
+      "step": 5344
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017940402490914122,
+      "loss": 0.1268,
+      "step": 5345
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.001793954885486461,
+      "loss": 0.063,
+      "step": 5346
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.001793869506226724,
+      "loss": 0.12,
+      "step": 5347
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.001793784111313885,
+      "loss": 0.1708,
+      "step": 5348
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017936987007496276,
+      "loss": 0.1392,
+      "step": 5349
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017936132745356359,
+      "loss": 0.1036,
+      "step": 5350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017935278326735942,
+      "loss": 0.144,
+      "step": 5351
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017934423751651875,
+      "loss": 0.0914,
+      "step": 5352
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017933569020121003,
+      "loss": 0.2056,
+      "step": 5353
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017932714132160182,
+      "loss": 0.0968,
+      "step": 5354
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017931859087786269,
+      "loss": 0.1236,
+      "step": 5355
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017931003887016123,
+      "loss": 0.1185,
+      "step": 5356
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017930148529866606,
+      "loss": 0.0354,
+      "step": 5357
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017929293016354583,
+      "loss": 0.1203,
+      "step": 5358
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017928437346496921,
+      "loss": 0.0907,
+      "step": 5359
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017927581520310496,
+      "loss": 0.0626,
+      "step": 5360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017926725537812183,
+      "loss": 0.0884,
+      "step": 5361
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017925869399018854,
+      "loss": 0.103,
+      "step": 5362
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017925013103947392,
+      "loss": 0.1329,
+      "step": 5363
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017924156652614682,
+      "loss": 0.0611,
+      "step": 5364
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017923300045037615,
+      "loss": 0.1055,
+      "step": 5365
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017922443281233073,
+      "loss": 0.1083,
+      "step": 5366
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017921586361217957,
+      "loss": 0.0996,
+      "step": 5367
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017920729285009157,
+      "loss": 0.0524,
+      "step": 5368
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017919872052623576,
+      "loss": 0.1121,
+      "step": 5369
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017919014664078115,
+      "loss": 0.0942,
+      "step": 5370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.001791815711938968,
+      "loss": 0.1151,
+      "step": 5371
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017917299418575181,
+      "loss": 0.0942,
+      "step": 5372
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017916441561651526,
+      "loss": 0.1177,
+      "step": 5373
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0017915583548635636,
+      "loss": 0.087,
+      "step": 5374
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017914725379544426,
+      "loss": 0.0834,
+      "step": 5375
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001791386705439481,
+      "loss": 0.0672,
+      "step": 5376
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017913008573203723,
+      "loss": 0.0679,
+      "step": 5377
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017912149935988084,
+      "loss": 0.0898,
+      "step": 5378
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017911291142764828,
+      "loss": 0.0637,
+      "step": 5379
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017910432193550885,
+      "loss": 0.0829,
+      "step": 5380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017909573088363193,
+      "loss": 0.0903,
+      "step": 5381
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017908713827218695,
+      "loss": 0.0774,
+      "step": 5382
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017907854410134326,
+      "loss": 0.0581,
+      "step": 5383
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017906994837127036,
+      "loss": 0.2158,
+      "step": 5384
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017906135108213773,
+      "loss": 0.0898,
+      "step": 5385
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017905275223411485,
+      "loss": 0.0805,
+      "step": 5386
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017904415182737135,
+      "loss": 0.0822,
+      "step": 5387
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017903554986207673,
+      "loss": 0.0895,
+      "step": 5388
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017902694633840067,
+      "loss": 0.1095,
+      "step": 5389
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017901834125651273,
+      "loss": 0.1198,
+      "step": 5390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017900973461658263,
+      "loss": 0.1041,
+      "step": 5391
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017900112641878005,
+      "loss": 0.0704,
+      "step": 5392
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017899251666327477,
+      "loss": 0.1298,
+      "step": 5393
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017898390535023647,
+      "loss": 0.1154,
+      "step": 5394
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017897529247983502,
+      "loss": 0.1261,
+      "step": 5395
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001789666780522402,
+      "loss": 0.0707,
+      "step": 5396
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017895806206762188,
+      "loss": 0.097,
+      "step": 5397
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017894944452614993,
+      "loss": 0.1224,
+      "step": 5398
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001789408254279943,
+      "loss": 0.098,
+      "step": 5399
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017893220477332493,
+      "loss": 0.1277,
+      "step": 5400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017892358256231175,
+      "loss": 0.1503,
+      "step": 5401
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017891495879512483,
+      "loss": 0.1089,
+      "step": 5402
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017890633347193415,
+      "loss": 0.0863,
+      "step": 5403
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017889770659290983,
+      "loss": 0.1221,
+      "step": 5404
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017888907815822193,
+      "loss": 0.0936,
+      "step": 5405
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017888044816804062,
+      "loss": 0.064,
+      "step": 5406
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017887181662253604,
+      "loss": 0.1435,
+      "step": 5407
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001788631835218784,
+      "loss": 0.1057,
+      "step": 5408
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001788545488662379,
+      "loss": 0.0823,
+      "step": 5409
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017884591265578478,
+      "loss": 0.0765,
+      "step": 5410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017883727489068936,
+      "loss": 0.0786,
+      "step": 5411
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017882863557112196,
+      "loss": 0.1276,
+      "step": 5412
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001788199946972529,
+      "loss": 0.1288,
+      "step": 5413
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017881135226925254,
+      "loss": 0.0507,
+      "step": 5414
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017880270828729136,
+      "loss": 0.0991,
+      "step": 5415
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017879406275153969,
+      "loss": 0.2065,
+      "step": 5416
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017878541566216809,
+      "loss": 0.1292,
+      "step": 5417
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00178776767019347,
+      "loss": 0.0627,
+      "step": 5418
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0017876811682324698,
+      "loss": 0.1198,
+      "step": 5419
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.001787594650740386,
+      "loss": 0.1213,
+      "step": 5420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017875081177189244,
+      "loss": 0.1337,
+      "step": 5421
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001787421569169791,
+      "loss": 0.0519,
+      "step": 5422
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017873350050946924,
+      "loss": 0.0707,
+      "step": 5423
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017872484254953353,
+      "loss": 0.0974,
+      "step": 5424
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017871618303734276,
+      "loss": 0.1344,
+      "step": 5425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017870752197306757,
+      "loss": 0.0662,
+      "step": 5426
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017869885935687879,
+      "loss": 0.1044,
+      "step": 5427
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017869019518894724,
+      "loss": 0.1824,
+      "step": 5428
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017868152946944372,
+      "loss": 0.1556,
+      "step": 5429
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001786728621985391,
+      "loss": 0.0879,
+      "step": 5430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001786641933764043,
+      "loss": 0.0926,
+      "step": 5431
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001786555230032102,
+      "loss": 0.1115,
+      "step": 5432
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017864685107912782,
+      "loss": 0.1014,
+      "step": 5433
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001786381776043281,
+      "loss": 0.084,
+      "step": 5434
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001786295025789821,
+      "loss": 0.1082,
+      "step": 5435
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017862082600326081,
+      "loss": 0.0839,
+      "step": 5436
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001786121478773354,
+      "loss": 0.1733,
+      "step": 5437
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017860346820137687,
+      "loss": 0.0896,
+      "step": 5438
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017859478697555646,
+      "loss": 0.0747,
+      "step": 5439
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017858610420004527,
+      "loss": 0.1273,
+      "step": 5440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017857741987501456,
+      "loss": 0.1039,
+      "step": 5441
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017856873400063553,
+      "loss": 0.0763,
+      "step": 5442
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017856004657707946,
+      "loss": 0.0817,
+      "step": 5443
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001785513576045176,
+      "loss": 0.1421,
+      "step": 5444
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017854266708312132,
+      "loss": 0.0792,
+      "step": 5445
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017853397501306197,
+      "loss": 0.1138,
+      "step": 5446
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017852528139451094,
+      "loss": 0.1035,
+      "step": 5447
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017851658622763961,
+      "loss": 0.1044,
+      "step": 5448
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017850788951261948,
+      "loss": 0.1105,
+      "step": 5449
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017849919124962197,
+      "loss": 0.1066,
+      "step": 5450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017849049143881864,
+      "loss": 0.0736,
+      "step": 5451
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017848179008038102,
+      "loss": 0.108,
+      "step": 5452
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017847308717448063,
+      "loss": 0.083,
+      "step": 5453
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001784643827212891,
+      "loss": 0.0621,
+      "step": 5454
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017845567672097808,
+      "loss": 0.059,
+      "step": 5455
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017844696917371922,
+      "loss": 0.156,
+      "step": 5456
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001784382600796842,
+      "loss": 0.099,
+      "step": 5457
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017842954943904474,
+      "loss": 0.0828,
+      "step": 5458
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017842083725197258,
+      "loss": 0.1063,
+      "step": 5459
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017841212351863956,
+      "loss": 0.0862,
+      "step": 5460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017840340823921741,
+      "loss": 0.1023,
+      "step": 5461
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017839469141387805,
+      "loss": 0.0844,
+      "step": 5462
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017838597304279335,
+      "loss": 0.1804,
+      "step": 5463
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001783772531261351,
+      "loss": 0.0961,
+      "step": 5464
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.001783685316640754,
+      "loss": 0.1151,
+      "step": 5465
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0017835980865678607,
+      "loss": 0.0966,
+      "step": 5466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017835108410443922,
+      "loss": 0.1101,
+      "step": 5467
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017834235800720678,
+      "loss": 0.1088,
+      "step": 5468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017833363036526086,
+      "loss": 0.1216,
+      "step": 5469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017832490117877357,
+      "loss": 0.1189,
+      "step": 5470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00178316170447917,
+      "loss": 0.0804,
+      "step": 5471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017830743817286328,
+      "loss": 0.0916,
+      "step": 5472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017829870435378456,
+      "loss": 0.17,
+      "step": 5473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017828996899085317,
+      "loss": 0.1377,
+      "step": 5474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.001782812320842412,
+      "loss": 0.0966,
+      "step": 5475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017827249363412104,
+      "loss": 0.1597,
+      "step": 5476
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017826375364066495,
+      "loss": 0.1663,
+      "step": 5477
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017825501210404521,
+      "loss": 0.1219,
+      "step": 5478
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017824626902443424,
+      "loss": 0.0638,
+      "step": 5479
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017823752440200443,
+      "loss": 0.1099,
+      "step": 5480
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017822877823692816,
+      "loss": 0.1121,
+      "step": 5481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017822003052937795,
+      "loss": 0.1935,
+      "step": 5482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017821128127952618,
+      "loss": 0.0844,
+      "step": 5483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017820253048754548,
+      "loss": 0.0732,
+      "step": 5484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017819377815360832,
+      "loss": 0.1533,
+      "step": 5485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.001781850242778873,
+      "loss": 0.1047,
+      "step": 5486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00178176268860555,
+      "loss": 0.0804,
+      "step": 5487
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017816751190178409,
+      "loss": 0.1229,
+      "step": 5488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017815875340174718,
+      "loss": 0.1193,
+      "step": 5489
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017814999336061704,
+      "loss": 0.0966,
+      "step": 5490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017814123177856635,
+      "loss": 0.0728,
+      "step": 5491
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017813246865576787,
+      "loss": 0.1093,
+      "step": 5492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017812370399239437,
+      "loss": 0.0992,
+      "step": 5493
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017811493778861869,
+      "loss": 0.1108,
+      "step": 5494
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017810617004461369,
+      "loss": 0.1234,
+      "step": 5495
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017809740076055226,
+      "loss": 0.1045,
+      "step": 5496
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017808862993660727,
+      "loss": 0.0767,
+      "step": 5497
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017807985757295165,
+      "loss": 0.1189,
+      "step": 5498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017807108366975836,
+      "loss": 0.075,
+      "step": 5499
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017806230822720046,
+      "loss": 0.1591,
+      "step": 5500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017805353124545097,
+      "loss": 0.0355,
+      "step": 5501
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.001780447527246829,
+      "loss": 0.13,
+      "step": 5502
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017803597266506936,
+      "loss": 0.0723,
+      "step": 5503
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.001780271910667835,
+      "loss": 0.0905,
+      "step": 5504
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017801840792999843,
+      "loss": 0.0774,
+      "step": 5505
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017800962325488737,
+      "loss": 0.0876,
+      "step": 5506
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.001780008370416235,
+      "loss": 0.0799,
+      "step": 5507
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017799204929038006,
+      "loss": 0.0712,
+      "step": 5508
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017798326000133034,
+      "loss": 0.1534,
+      "step": 5509
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017797446917464765,
+      "loss": 0.132,
+      "step": 5510
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0017796567681050532,
+      "loss": 0.0989,
+      "step": 5511
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.001779568829090767,
+      "loss": 0.1267,
+      "step": 5512
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017794808747053517,
+      "loss": 0.1007,
+      "step": 5513
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017793929049505419,
+      "loss": 0.0956,
+      "step": 5514
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001779304919828072,
+      "loss": 0.0647,
+      "step": 5515
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017792169193396768,
+      "loss": 0.126,
+      "step": 5516
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017791289034870914,
+      "loss": 0.1061,
+      "step": 5517
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017790408722720512,
+      "loss": 0.0942,
+      "step": 5518
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001778952825696292,
+      "loss": 0.087,
+      "step": 5519
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00177886476376155,
+      "loss": 0.1237,
+      "step": 5520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017787766864695617,
+      "loss": 0.0916,
+      "step": 5521
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017786885938220634,
+      "loss": 0.0653,
+      "step": 5522
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001778600485820792,
+      "loss": 0.0557,
+      "step": 5523
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001778512362467485,
+      "loss": 0.1624,
+      "step": 5524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00177842422376388,
+      "loss": 0.1506,
+      "step": 5525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001778336069711715,
+      "loss": 0.1349,
+      "step": 5526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001778247900312728,
+      "loss": 0.125,
+      "step": 5527
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017781597155686574,
+      "loss": 0.095,
+      "step": 5528
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017780715154812419,
+      "loss": 0.1426,
+      "step": 5529
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017779833000522208,
+      "loss": 0.1411,
+      "step": 5530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017778950692833332,
+      "loss": 0.1121,
+      "step": 5531
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001777806823176319,
+      "loss": 0.0961,
+      "step": 5532
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017777185617329186,
+      "loss": 0.1229,
+      "step": 5533
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017776302849548716,
+      "loss": 0.1029,
+      "step": 5534
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001777541992843919,
+      "loss": 0.1245,
+      "step": 5535
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017774536854018015,
+      "loss": 0.1572,
+      "step": 5536
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017773653626302604,
+      "loss": 0.1665,
+      "step": 5537
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001777277024531037,
+      "loss": 0.1124,
+      "step": 5538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017771886711058736,
+      "loss": 0.0638,
+      "step": 5539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017771003023565118,
+      "loss": 0.1559,
+      "step": 5540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017770119182846942,
+      "loss": 0.0813,
+      "step": 5541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017769235188921635,
+      "loss": 0.119,
+      "step": 5542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001776835104180663,
+      "loss": 0.1678,
+      "step": 5543
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017767466741519353,
+      "loss": 0.1281,
+      "step": 5544
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017766582288077247,
+      "loss": 0.0818,
+      "step": 5545
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017765697681497749,
+      "loss": 0.1853,
+      "step": 5546
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00177648129217983,
+      "loss": 0.1449,
+      "step": 5547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017763928008996347,
+      "loss": 0.1083,
+      "step": 5548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017763042943109335,
+      "loss": 0.0874,
+      "step": 5549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001776215772415472,
+      "loss": 0.1306,
+      "step": 5550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017761272352149952,
+      "loss": 0.1011,
+      "step": 5551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.001776038682711249,
+      "loss": 0.0584,
+      "step": 5552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017759501149059798,
+      "loss": 0.1159,
+      "step": 5553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017758615318009333,
+      "loss": 0.1107,
+      "step": 5554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017757729333978565,
+      "loss": 0.1614,
+      "step": 5555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017756843196984962,
+      "loss": 0.2063,
+      "step": 5556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017755956907045999,
+      "loss": 0.1569,
+      "step": 5557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0017755070464179148,
+      "loss": 0.1428,
+      "step": 5558
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001775418386840189,
+      "loss": 0.1399,
+      "step": 5559
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017753297119731705,
+      "loss": 0.1227,
+      "step": 5560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017752410218186076,
+      "loss": 0.1042,
+      "step": 5561
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001775152316378249,
+      "loss": 0.1483,
+      "step": 5562
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017750635956538443,
+      "loss": 0.1077,
+      "step": 5563
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017749748596471427,
+      "loss": 0.094,
+      "step": 5564
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017748861083598934,
+      "loss": 0.0875,
+      "step": 5565
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017747973417938465,
+      "loss": 0.1053,
+      "step": 5566
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017747085599507525,
+      "loss": 0.0951,
+      "step": 5567
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017746197628323618,
+      "loss": 0.0797,
+      "step": 5568
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017745309504404252,
+      "loss": 0.1097,
+      "step": 5569
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001774442122776694,
+      "loss": 0.0973,
+      "step": 5570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017743532798429197,
+      "loss": 0.1536,
+      "step": 5571
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001774264421640854,
+      "loss": 0.1467,
+      "step": 5572
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017741755481722485,
+      "loss": 0.1646,
+      "step": 5573
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017740866594388562,
+      "loss": 0.0751,
+      "step": 5574
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017739977554424296,
+      "loss": 0.0768,
+      "step": 5575
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017739088361847215,
+      "loss": 0.2007,
+      "step": 5576
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017738199016674853,
+      "loss": 0.1301,
+      "step": 5577
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017737309518924744,
+      "loss": 0.0611,
+      "step": 5578
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017736419868614429,
+      "loss": 0.0631,
+      "step": 5579
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017735530065761448,
+      "loss": 0.1213,
+      "step": 5580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017734640110383347,
+      "loss": 0.12,
+      "step": 5581
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001773375000249767,
+      "loss": 0.0879,
+      "step": 5582
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017732859742121973,
+      "loss": 0.1119,
+      "step": 5583
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017731969329273806,
+      "loss": 0.0745,
+      "step": 5584
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001773107876397073,
+      "loss": 0.1124,
+      "step": 5585
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00177301880462303,
+      "loss": 0.0941,
+      "step": 5586
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017729297176070078,
+      "loss": 0.0896,
+      "step": 5587
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017728406153507633,
+      "loss": 0.1578,
+      "step": 5588
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017727514978560532,
+      "loss": 0.1753,
+      "step": 5589
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017726623651246349,
+      "loss": 0.0966,
+      "step": 5590
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017725732171582655,
+      "loss": 0.1602,
+      "step": 5591
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017724840539587033,
+      "loss": 0.0847,
+      "step": 5592
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017723948755277062,
+      "loss": 0.1403,
+      "step": 5593
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001772305681867032,
+      "loss": 0.0795,
+      "step": 5594
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00177221647297844,
+      "loss": 0.1305,
+      "step": 5595
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017721272488636888,
+      "loss": 0.1373,
+      "step": 5596
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001772038009524538,
+      "loss": 0.1189,
+      "step": 5597
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001771948754962747,
+      "loss": 0.0869,
+      "step": 5598
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017718594851800756,
+      "loss": 0.1036,
+      "step": 5599
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017717702001782843,
+      "loss": 0.0956,
+      "step": 5600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017716808999591331,
+      "loss": 0.0974,
+      "step": 5601
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.001771591584524383,
+      "loss": 0.1064,
+      "step": 5602
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017715022538757955,
+      "loss": 0.1063,
+      "step": 5603
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0017714129080151314,
+      "loss": 0.1278,
+      "step": 5604
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017713235469441525,
+      "loss": 0.1487,
+      "step": 5605
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001771234170664621,
+      "loss": 0.0917,
+      "step": 5606
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017711447791782984,
+      "loss": 0.0955,
+      "step": 5607
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017710553724869485,
+      "loss": 0.1134,
+      "step": 5608
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017709659505923331,
+      "loss": 0.1486,
+      "step": 5609
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001770876513496216,
+      "loss": 0.0973,
+      "step": 5610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017707870612003608,
+      "loss": 0.0676,
+      "step": 5611
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017706975937065309,
+      "loss": 0.1516,
+      "step": 5612
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00177060811101649,
+      "loss": 0.139,
+      "step": 5613
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017705186131320035,
+      "loss": 0.0891,
+      "step": 5614
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001770429100054835,
+      "loss": 0.0765,
+      "step": 5615
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017703395717867502,
+      "loss": 0.1646,
+      "step": 5616
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017702500283295142,
+      "loss": 0.137,
+      "step": 5617
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017701604696848923,
+      "loss": 0.1283,
+      "step": 5618
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017700708958546506,
+      "loss": 0.0607,
+      "step": 5619
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017699813068405552,
+      "loss": 0.1144,
+      "step": 5620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017698917026443728,
+      "loss": 0.1459,
+      "step": 5621
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017698020832678698,
+      "loss": 0.1486,
+      "step": 5622
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017697124487128138,
+      "loss": 0.1025,
+      "step": 5623
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017696227989809714,
+      "loss": 0.0962,
+      "step": 5624
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001769533134074111,
+      "loss": 0.1113,
+      "step": 5625
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001769443453994,
+      "loss": 0.1177,
+      "step": 5626
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001769353758742407,
+      "loss": 0.1038,
+      "step": 5627
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017692640483211004,
+      "loss": 0.1047,
+      "step": 5628
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017691743227318494,
+      "loss": 0.0969,
+      "step": 5629
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017690845819764225,
+      "loss": 0.0912,
+      "step": 5630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00176899482605659,
+      "loss": 0.0831,
+      "step": 5631
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001768905054974121,
+      "loss": 0.1273,
+      "step": 5632
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017688152687307856,
+      "loss": 0.0722,
+      "step": 5633
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017687254673283547,
+      "loss": 0.1066,
+      "step": 5634
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017686356507685983,
+      "loss": 0.1678,
+      "step": 5635
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001768545819053288,
+      "loss": 0.109,
+      "step": 5636
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017684559721841944,
+      "loss": 0.0779,
+      "step": 5637
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017683661101630893,
+      "loss": 0.1018,
+      "step": 5638
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017682762329917449,
+      "loss": 0.0924,
+      "step": 5639
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017681863406719329,
+      "loss": 0.086,
+      "step": 5640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001768096433205426,
+      "loss": 0.1047,
+      "step": 5641
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017680065105939967,
+      "loss": 0.1216,
+      "step": 5642
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017679165728394183,
+      "loss": 0.1296,
+      "step": 5643
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001767826619943464,
+      "loss": 0.1437,
+      "step": 5644
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017677366519079074,
+      "loss": 0.0666,
+      "step": 5645
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001767646668734523,
+      "loss": 0.1921,
+      "step": 5646
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017675566704250841,
+      "loss": 0.182,
+      "step": 5647
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.001767466656981366,
+      "loss": 0.074,
+      "step": 5648
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017673766284051432,
+      "loss": 0.0778,
+      "step": 5649
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0017672865846981908,
+      "loss": 0.135,
+      "step": 5650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017671965258622844,
+      "loss": 0.1189,
+      "step": 5651
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017671064518991995,
+      "loss": 0.0758,
+      "step": 5652
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017670163628107126,
+      "loss": 0.1061,
+      "step": 5653
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017669262585985994,
+      "loss": 0.104,
+      "step": 5654
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001766836139264637,
+      "loss": 0.171,
+      "step": 5655
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017667460048106022,
+      "loss": 0.0875,
+      "step": 5656
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017666558552382724,
+      "loss": 0.0738,
+      "step": 5657
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017665656905494246,
+      "loss": 0.1052,
+      "step": 5658
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001766475510745837,
+      "loss": 0.1271,
+      "step": 5659
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017663853158292877,
+      "loss": 0.1121,
+      "step": 5660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001766295105801555,
+      "loss": 0.0945,
+      "step": 5661
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001766204880664418,
+      "loss": 0.1304,
+      "step": 5662
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001766114640419655,
+      "loss": 0.0883,
+      "step": 5663
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017660243850690458,
+      "loss": 0.0843,
+      "step": 5664
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017659341146143701,
+      "loss": 0.1736,
+      "step": 5665
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017658438290574074,
+      "loss": 0.1207,
+      "step": 5666
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017657535283999385,
+      "loss": 0.0956,
+      "step": 5667
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001765663212643743,
+      "loss": 0.1473,
+      "step": 5668
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017655728817906028,
+      "loss": 0.1021,
+      "step": 5669
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017654825358422981,
+      "loss": 0.0505,
+      "step": 5670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017653921748006106,
+      "loss": 0.1235,
+      "step": 5671
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001765301798667322,
+      "loss": 0.1411,
+      "step": 5672
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017652114074442143,
+      "loss": 0.0805,
+      "step": 5673
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00176512100113307,
+      "loss": 0.0635,
+      "step": 5674
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017650305797356715,
+      "loss": 0.1844,
+      "step": 5675
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001764940143253801,
+      "loss": 0.1627,
+      "step": 5676
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001764849691689243,
+      "loss": 0.1021,
+      "step": 5677
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017647592250437801,
+      "loss": 0.0629,
+      "step": 5678
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017646687433191964,
+      "loss": 0.1118,
+      "step": 5679
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017645782465172757,
+      "loss": 0.1816,
+      "step": 5680
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017644877346398025,
+      "loss": 0.0798,
+      "step": 5681
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017643972076885612,
+      "loss": 0.0632,
+      "step": 5682
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017643066656653377,
+      "loss": 0.136,
+      "step": 5683
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001764216108571916,
+      "loss": 0.1365,
+      "step": 5684
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017641255364100829,
+      "loss": 0.09,
+      "step": 5685
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017640349491816233,
+      "loss": 0.0779,
+      "step": 5686
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017639443468883233,
+      "loss": 0.1838,
+      "step": 5687
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017638537295319703,
+      "loss": 0.0696,
+      "step": 5688
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017637630971143504,
+      "loss": 0.0898,
+      "step": 5689
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017636724496372502,
+      "loss": 0.0957,
+      "step": 5690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001763581787102458,
+      "loss": 0.0773,
+      "step": 5691
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017634911095117608,
+      "loss": 0.0782,
+      "step": 5692
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017634004168669468,
+      "loss": 0.1038,
+      "step": 5693
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017633097091698043,
+      "loss": 0.1201,
+      "step": 5694
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017632189864221215,
+      "loss": 0.1129,
+      "step": 5695
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0017631282486256872,
+      "loss": 0.0931,
+      "step": 5696
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.001763037495782291,
+      "loss": 0.0766,
+      "step": 5697
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001762946727893722,
+      "loss": 0.1083,
+      "step": 5698
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017628559449617699,
+      "loss": 0.094,
+      "step": 5699
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017627651469882247,
+      "loss": 0.1184,
+      "step": 5700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017626743339748767,
+      "loss": 0.0753,
+      "step": 5701
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017625835059235167,
+      "loss": 0.1326,
+      "step": 5702
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017624926628359355,
+      "loss": 0.069,
+      "step": 5703
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001762401804713924,
+      "loss": 0.1171,
+      "step": 5704
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017623109315592744,
+      "loss": 0.056,
+      "step": 5705
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017622200433737778,
+      "loss": 0.0958,
+      "step": 5706
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017621291401592265,
+      "loss": 0.0818,
+      "step": 5707
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017620382219174129,
+      "loss": 0.0983,
+      "step": 5708
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017619472886501297,
+      "loss": 0.0906,
+      "step": 5709
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017618563403591701,
+      "loss": 0.125,
+      "step": 5710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001761765377046327,
+      "loss": 0.1412,
+      "step": 5711
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001761674398713394,
+      "loss": 0.1616,
+      "step": 5712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001761583405362165,
+      "loss": 0.1594,
+      "step": 5713
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017614923969944344,
+      "loss": 0.1357,
+      "step": 5714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017614013736119967,
+      "loss": 0.0856,
+      "step": 5715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001761310335216646,
+      "loss": 0.1931,
+      "step": 5716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001761219281810178,
+      "loss": 0.1377,
+      "step": 5717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017611282133943877,
+      "loss": 0.085,
+      "step": 5718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017610371299710708,
+      "loss": 0.0739,
+      "step": 5719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017609460315420235,
+      "loss": 0.0885,
+      "step": 5720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017608549181090418,
+      "loss": 0.0809,
+      "step": 5721
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017607637896739222,
+      "loss": 0.0829,
+      "step": 5722
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017606726462384614,
+      "loss": 0.1318,
+      "step": 5723
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001760581487804457,
+      "loss": 0.0746,
+      "step": 5724
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001760490314373706,
+      "loss": 0.0603,
+      "step": 5725
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017603991259480063,
+      "loss": 0.0898,
+      "step": 5726
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017603079225291555,
+      "loss": 0.0822,
+      "step": 5727
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017602167041189528,
+      "loss": 0.0751,
+      "step": 5728
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017601254707191956,
+      "loss": 0.0858,
+      "step": 5729
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017600342223316838,
+      "loss": 0.1334,
+      "step": 5730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001759942958958216,
+      "loss": 0.1123,
+      "step": 5731
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001759851680600592,
+      "loss": 0.1211,
+      "step": 5732
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017597603872606116,
+      "loss": 0.1025,
+      "step": 5733
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017596690789400747,
+      "loss": 0.1346,
+      "step": 5734
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017595777556407813,
+      "loss": 0.1385,
+      "step": 5735
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001759486417364533,
+      "loss": 0.0824,
+      "step": 5736
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00175939506411313,
+      "loss": 0.0746,
+      "step": 5737
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017593036958883737,
+      "loss": 0.1869,
+      "step": 5738
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017592123126920658,
+      "loss": 0.1439,
+      "step": 5739
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017591209145260082,
+      "loss": 0.0358,
+      "step": 5740
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017590295013920028,
+      "loss": 0.0693,
+      "step": 5741
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0017589380732918522,
+      "loss": 0.1506,
+      "step": 5742
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.001758846630227359,
+      "loss": 0.1195,
+      "step": 5743
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017587551722003263,
+      "loss": 0.1251,
+      "step": 5744
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017586636992125578,
+      "loss": 0.0808,
+      "step": 5745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017585722112658565,
+      "loss": 0.1274,
+      "step": 5746
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017584807083620264,
+      "loss": 0.0715,
+      "step": 5747
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017583891905028723,
+      "loss": 0.1797,
+      "step": 5748
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017582976576901981,
+      "loss": 0.099,
+      "step": 5749
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017582061099258085,
+      "loss": 0.1129,
+      "step": 5750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017581145472115093,
+      "loss": 0.1167,
+      "step": 5751
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001758022969549105,
+      "loss": 0.0803,
+      "step": 5752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017579313769404022,
+      "loss": 0.1622,
+      "step": 5753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001757839769387206,
+      "loss": 0.0967,
+      "step": 5754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017577481468913235,
+      "loss": 0.0916,
+      "step": 5755
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017576565094545602,
+      "loss": 0.1234,
+      "step": 5756
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017575648570787238,
+      "loss": 0.1403,
+      "step": 5757
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017574731897656215,
+      "loss": 0.1032,
+      "step": 5758
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017573815075170604,
+      "loss": 0.0589,
+      "step": 5759
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017572898103348487,
+      "loss": 0.0761,
+      "step": 5760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017571980982207937,
+      "loss": 0.1448,
+      "step": 5761
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001757106371176704,
+      "loss": 0.095,
+      "step": 5762
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017570146292043886,
+      "loss": 0.0988,
+      "step": 5763
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001756922872305656,
+      "loss": 0.1411,
+      "step": 5764
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017568311004823155,
+      "loss": 0.0957,
+      "step": 5765
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017567393137361768,
+      "loss": 0.1272,
+      "step": 5766
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017566475120690496,
+      "loss": 0.1272,
+      "step": 5767
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001756555695482744,
+      "loss": 0.1334,
+      "step": 5768
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017564638639790701,
+      "loss": 0.0853,
+      "step": 5769
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017563720175598392,
+      "loss": 0.101,
+      "step": 5770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001756280156226862,
+      "loss": 0.0909,
+      "step": 5771
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017561882799819496,
+      "loss": 0.1339,
+      "step": 5772
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017560963888269136,
+      "loss": 0.092,
+      "step": 5773
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017560044827635657,
+      "loss": 0.1115,
+      "step": 5774
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017559125617937186,
+      "loss": 0.1108,
+      "step": 5775
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017558206259191845,
+      "loss": 0.1636,
+      "step": 5776
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017557286751417758,
+      "loss": 0.0771,
+      "step": 5777
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001755636709463306,
+      "loss": 0.1543,
+      "step": 5778
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017555447288855881,
+      "loss": 0.1324,
+      "step": 5779
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001755452733410436,
+      "loss": 0.0975,
+      "step": 5780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017553607230396636,
+      "loss": 0.1183,
+      "step": 5781
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017552686977750848,
+      "loss": 0.114,
+      "step": 5782
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017551766576185143,
+      "loss": 0.1157,
+      "step": 5783
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017550846025717673,
+      "loss": 0.0637,
+      "step": 5784
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001754992532636658,
+      "loss": 0.1649,
+      "step": 5785
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017549004478150025,
+      "loss": 0.1284,
+      "step": 5786
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017548083481086164,
+      "loss": 0.1374,
+      "step": 5787
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0017547162335193156,
+      "loss": 0.1016,
+      "step": 5788
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.001754624104048916,
+      "loss": 0.1147,
+      "step": 5789
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001754531959699235,
+      "loss": 0.1279,
+      "step": 5790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017544398004720884,
+      "loss": 0.1378,
+      "step": 5791
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001754347626369294,
+      "loss": 0.1992,
+      "step": 5792
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017542554373926689,
+      "loss": 0.1107,
+      "step": 5793
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017541632335440313,
+      "loss": 0.116,
+      "step": 5794
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001754071014825199,
+      "loss": 0.0944,
+      "step": 5795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017539787812379903,
+      "loss": 0.1388,
+      "step": 5796
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001753886532784224,
+      "loss": 0.0894,
+      "step": 5797
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017537942694657185,
+      "loss": 0.1027,
+      "step": 5798
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017537019912842935,
+      "loss": 0.1426,
+      "step": 5799
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017536096982417684,
+      "loss": 0.1005,
+      "step": 5800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017535173903399626,
+      "loss": 0.1338,
+      "step": 5801
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017534250675806965,
+      "loss": 0.1384,
+      "step": 5802
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017533327299657907,
+      "loss": 0.1008,
+      "step": 5803
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017532403774970656,
+      "loss": 0.1169,
+      "step": 5804
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001753148010176342,
+      "loss": 0.1499,
+      "step": 5805
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017530556280054414,
+      "loss": 0.064,
+      "step": 5806
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017529632309861852,
+      "loss": 0.1777,
+      "step": 5807
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017528708191203953,
+      "loss": 0.1672,
+      "step": 5808
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017527783924098936,
+      "loss": 0.1566,
+      "step": 5809
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001752685950856503,
+      "loss": 0.1271,
+      "step": 5810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017525934944620462,
+      "loss": 0.146,
+      "step": 5811
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017525010232283455,
+      "loss": 0.1838,
+      "step": 5812
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001752408537157225,
+      "loss": 0.0969,
+      "step": 5813
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017523160362505079,
+      "loss": 0.1038,
+      "step": 5814
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017522235205100181,
+      "loss": 0.0801,
+      "step": 5815
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017521309899375798,
+      "loss": 0.0791,
+      "step": 5816
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017520384445350176,
+      "loss": 0.106,
+      "step": 5817
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001751945884304156,
+      "loss": 0.1071,
+      "step": 5818
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017518533092468203,
+      "loss": 0.0712,
+      "step": 5819
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017517607193648357,
+      "loss": 0.0873,
+      "step": 5820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017516681146600276,
+      "loss": 0.0891,
+      "step": 5821
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001751575495134223,
+      "loss": 0.0895,
+      "step": 5822
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017514828607892467,
+      "loss": 0.0513,
+      "step": 5823
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001751390211626926,
+      "loss": 0.1418,
+      "step": 5824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017512975476490875,
+      "loss": 0.2118,
+      "step": 5825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017512048688575585,
+      "loss": 0.0947,
+      "step": 5826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017511121752541664,
+      "loss": 0.1212,
+      "step": 5827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017510194668407385,
+      "loss": 0.1207,
+      "step": 5828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001750926743619103,
+      "loss": 0.1552,
+      "step": 5829
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017508340055910883,
+      "loss": 0.1042,
+      "step": 5830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001750741252758523,
+      "loss": 0.1405,
+      "step": 5831
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017506484851232357,
+      "loss": 0.1904,
+      "step": 5832
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0017505557026870555,
+      "loss": 0.1028,
+      "step": 5833
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001750462905451812,
+      "loss": 0.0977,
+      "step": 5834
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.001750370093419335,
+      "loss": 0.0959,
+      "step": 5835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017502772665914544,
+      "loss": 0.1309,
+      "step": 5836
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017501844249700005,
+      "loss": 0.0853,
+      "step": 5837
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017500915685568037,
+      "loss": 0.1481,
+      "step": 5838
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017499986973536954,
+      "loss": 0.1155,
+      "step": 5839
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017499058113625066,
+      "loss": 0.1123,
+      "step": 5840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017498129105850686,
+      "loss": 0.0927,
+      "step": 5841
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017497199950232134,
+      "loss": 0.1268,
+      "step": 5842
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017496270646787727,
+      "loss": 0.1069,
+      "step": 5843
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017495341195535791,
+      "loss": 0.0744,
+      "step": 5844
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017494411596494653,
+      "loss": 0.106,
+      "step": 5845
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017493481849682642,
+      "loss": 0.1011,
+      "step": 5846
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017492551955118087,
+      "loss": 0.1085,
+      "step": 5847
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017491621912819329,
+      "loss": 0.0773,
+      "step": 5848
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00174906917228047,
+      "loss": 0.1045,
+      "step": 5849
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017489761385092546,
+      "loss": 0.1179,
+      "step": 5850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017488830899701209,
+      "loss": 0.127,
+      "step": 5851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017487900266649036,
+      "loss": 0.0657,
+      "step": 5852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017486969485954378,
+      "loss": 0.0776,
+      "step": 5853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017486038557635584,
+      "loss": 0.1254,
+      "step": 5854
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.001748510748171101,
+      "loss": 0.0717,
+      "step": 5855
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017484176258199021,
+      "loss": 0.1318,
+      "step": 5856
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.001748324488711797,
+      "loss": 0.1487,
+      "step": 5857
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017482313368486224,
+      "loss": 0.1202,
+      "step": 5858
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017481381702322155,
+      "loss": 0.0969,
+      "step": 5859
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017480449888644123,
+      "loss": 0.0883,
+      "step": 5860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017479517927470511,
+      "loss": 0.1032,
+      "step": 5861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017478585818819692,
+      "loss": 0.0861,
+      "step": 5862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017477653562710045,
+      "loss": 0.0681,
+      "step": 5863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017476721159159947,
+      "loss": 0.1033,
+      "step": 5864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017475788608187787,
+      "loss": 0.093,
+      "step": 5865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017474855909811952,
+      "loss": 0.0948,
+      "step": 5866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017473923064050836,
+      "loss": 0.1481,
+      "step": 5867
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017472990070922824,
+      "loss": 0.1514,
+      "step": 5868
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017472056930446318,
+      "loss": 0.0997,
+      "step": 5869
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017471123642639715,
+      "loss": 0.1151,
+      "step": 5870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017470190207521422,
+      "loss": 0.126,
+      "step": 5871
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017469256625109837,
+      "loss": 0.1,
+      "step": 5872
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017468322895423372,
+      "loss": 0.0757,
+      "step": 5873
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017467389018480436,
+      "loss": 0.1002,
+      "step": 5874
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017466454994299445,
+      "loss": 0.0591,
+      "step": 5875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017465520822898812,
+      "loss": 0.0557,
+      "step": 5876
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.001746458650429696,
+      "loss": 0.1151,
+      "step": 5877
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.001746365203851231,
+      "loss": 0.123,
+      "step": 5878
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.001746271742556329,
+      "loss": 0.1174,
+      "step": 5879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0017461782665468323,
+      "loss": 0.0763,
+      "step": 5880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.001746084775824584,
+      "loss": 0.1443,
+      "step": 5881
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017459912703914281,
+      "loss": 0.0973,
+      "step": 5882
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001745897750249208,
+      "loss": 0.1489,
+      "step": 5883
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017458042153997674,
+      "loss": 0.1146,
+      "step": 5884
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001745710665844951,
+      "loss": 0.1471,
+      "step": 5885
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001745617101586603,
+      "loss": 0.0833,
+      "step": 5886
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017455235226265685,
+      "loss": 0.1504,
+      "step": 5887
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017454299289666926,
+      "loss": 0.1033,
+      "step": 5888
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017453363206088206,
+      "loss": 0.1444,
+      "step": 5889
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017452426975547984,
+      "loss": 0.0728,
+      "step": 5890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001745149059806472,
+      "loss": 0.0653,
+      "step": 5891
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017450554073656873,
+      "loss": 0.0982,
+      "step": 5892
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017449617402342916,
+      "loss": 0.1227,
+      "step": 5893
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017448680584141313,
+      "loss": 0.0808,
+      "step": 5894
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017447743619070533,
+      "loss": 0.1402,
+      "step": 5895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017446806507149058,
+      "loss": 0.1262,
+      "step": 5896
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017445869248395361,
+      "loss": 0.1063,
+      "step": 5897
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017444931842827925,
+      "loss": 0.134,
+      "step": 5898
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001744399429046523,
+      "loss": 0.0661,
+      "step": 5899
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017443056591325758,
+      "loss": 0.0958,
+      "step": 5900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001744211874542801,
+      "loss": 0.0975,
+      "step": 5901
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001744118075279047,
+      "loss": 0.1014,
+      "step": 5902
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017440242613431635,
+      "loss": 0.0864,
+      "step": 5903
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001743930432737,
+      "loss": 0.0839,
+      "step": 5904
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001743836589462407,
+      "loss": 0.2007,
+      "step": 5905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017437427315212344,
+      "loss": 0.1067,
+      "step": 5906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017436488589153332,
+      "loss": 0.0789,
+      "step": 5907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017435549716465543,
+      "loss": 0.0775,
+      "step": 5908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017434610697167486,
+      "loss": 0.1521,
+      "step": 5909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017433671531277679,
+      "loss": 0.1069,
+      "step": 5910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001743273221881464,
+      "loss": 0.1367,
+      "step": 5911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001743179275979689,
+      "loss": 0.0829,
+      "step": 5912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017430853154242949,
+      "loss": 0.1398,
+      "step": 5913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017429913402171348,
+      "loss": 0.1328,
+      "step": 5914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017428973503600613,
+      "loss": 0.0901,
+      "step": 5915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017428033458549283,
+      "loss": 0.061,
+      "step": 5916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017427093267035885,
+      "loss": 0.1692,
+      "step": 5917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017426152929078962,
+      "loss": 0.1436,
+      "step": 5918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017425212444697055,
+      "loss": 0.1036,
+      "step": 5919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017424271813908706,
+      "loss": 0.0909,
+      "step": 5920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017423331036732464,
+      "loss": 0.1588,
+      "step": 5921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017422390113186876,
+      "loss": 0.1052,
+      "step": 5922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017421449043290495,
+      "loss": 0.1355,
+      "step": 5923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017420507827061883,
+      "loss": 0.1234,
+      "step": 5924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001741956646451959,
+      "loss": 0.1416,
+      "step": 5925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.001741862495568218,
+      "loss": 0.1063,
+      "step": 5926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0017417683300568218,
+      "loss": 0.0786,
+      "step": 5927
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017416741499196272,
+      "loss": 0.1097,
+      "step": 5928
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001741579955158491,
+      "loss": 0.0984,
+      "step": 5929
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017414857457752706,
+      "loss": 0.1228,
+      "step": 5930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017413915217718236,
+      "loss": 0.0975,
+      "step": 5931
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017412972831500075,
+      "loss": 0.0876,
+      "step": 5932
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017412030299116812,
+      "loss": 0.1272,
+      "step": 5933
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001741108762058702,
+      "loss": 0.1514,
+      "step": 5934
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017410144795929298,
+      "loss": 0.1196,
+      "step": 5935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001740920182516223,
+      "loss": 0.0638,
+      "step": 5936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001740825870830441,
+      "loss": 0.1194,
+      "step": 5937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017407315445374435,
+      "loss": 0.1381,
+      "step": 5938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017406372036390901,
+      "loss": 0.1803,
+      "step": 5939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017405428481372409,
+      "loss": 0.1196,
+      "step": 5940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017404484780337572,
+      "loss": 0.1437,
+      "step": 5941
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017403540933304987,
+      "loss": 0.1359,
+      "step": 5942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001740259694029327,
+      "loss": 0.0865,
+      "step": 5943
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017401652801321032,
+      "loss": 0.1273,
+      "step": 5944
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001740070851640689,
+      "loss": 0.1521,
+      "step": 5945
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017399764085569464,
+      "loss": 0.0935,
+      "step": 5946
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017398819508827371,
+      "loss": 0.2277,
+      "step": 5947
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017397874786199243,
+      "loss": 0.0838,
+      "step": 5948
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017396929917703702,
+      "loss": 0.1094,
+      "step": 5949
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017395984903359378,
+      "loss": 0.1213,
+      "step": 5950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017395039743184908,
+      "loss": 0.1205,
+      "step": 5951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017394094437198928,
+      "loss": 0.0715,
+      "step": 5952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017393148985420073,
+      "loss": 0.093,
+      "step": 5953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001739220338786699,
+      "loss": 0.1203,
+      "step": 5954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001739125764455832,
+      "loss": 0.0799,
+      "step": 5955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017390311755512713,
+      "loss": 0.1121,
+      "step": 5956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017389365720748818,
+      "loss": 0.1028,
+      "step": 5957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001738841954028529,
+      "loss": 0.0828,
+      "step": 5958
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017387473214140782,
+      "loss": 0.068,
+      "step": 5959
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017386526742333957,
+      "loss": 0.1396,
+      "step": 5960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017385580124883475,
+      "loss": 0.0815,
+      "step": 5961
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017384633361808,
+      "loss": 0.0818,
+      "step": 5962
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00173836864531262,
+      "loss": 0.0624,
+      "step": 5963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001738273939885675,
+      "loss": 0.1144,
+      "step": 5964
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001738179219901832,
+      "loss": 0.1243,
+      "step": 5965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017380844853629588,
+      "loss": 0.0911,
+      "step": 5966
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001737989736270923,
+      "loss": 0.0872,
+      "step": 5967
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.001737894972627593,
+      "loss": 0.0828,
+      "step": 5968
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017378001944348372,
+      "loss": 0.1506,
+      "step": 5969
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017377054016945246,
+      "loss": 0.0935,
+      "step": 5970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017376105944085242,
+      "loss": 0.0935,
+      "step": 5971
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017375157725787052,
+      "loss": 0.0914,
+      "step": 5972
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0017374209362069374,
+      "loss": 0.0609,
+      "step": 5973
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001737326085295091,
+      "loss": 0.1158,
+      "step": 5974
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017372312198450355,
+      "loss": 0.1544,
+      "step": 5975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017371363398586421,
+      "loss": 0.118,
+      "step": 5976
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017370414453377813,
+      "loss": 0.0832,
+      "step": 5977
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017369465362843243,
+      "loss": 0.0735,
+      "step": 5978
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017368516127001423,
+      "loss": 0.087,
+      "step": 5979
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017367566745871068,
+      "loss": 0.165,
+      "step": 5980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017366617219470905,
+      "loss": 0.0918,
+      "step": 5981
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017365667547819643,
+      "loss": 0.1284,
+      "step": 5982
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017364717730936024,
+      "loss": 0.059,
+      "step": 5983
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001736376776883876,
+      "loss": 0.0797,
+      "step": 5984
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017362817661546594,
+      "loss": 0.1331,
+      "step": 5985
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001736186740907825,
+      "loss": 0.1228,
+      "step": 5986
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017360917011452473,
+      "loss": 0.0984,
+      "step": 5987
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017359966468687995,
+      "loss": 0.0792,
+      "step": 5988
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017359015780803566,
+      "loss": 0.1191,
+      "step": 5989
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017358064947817925,
+      "loss": 0.126,
+      "step": 5990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001735711396974982,
+      "loss": 0.1277,
+      "step": 5991
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017356162846618006,
+      "loss": 0.0667,
+      "step": 5992
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001735521157844124,
+      "loss": 0.1306,
+      "step": 5993
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017354260165238268,
+      "loss": 0.1338,
+      "step": 5994
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017353308607027853,
+      "loss": 0.0899,
+      "step": 5995
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017352356903828762,
+      "loss": 0.1932,
+      "step": 5996
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017351405055659756,
+      "loss": 0.0751,
+      "step": 5997
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017350453062539607,
+      "loss": 0.1307,
+      "step": 5998
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017349500924487083,
+      "loss": 0.145,
+      "step": 5999
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017348548641520957,
+      "loss": 0.1304,
+      "step": 6000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017347596213660003,
+      "loss": 0.1053,
+      "step": 6001
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017346643640923007,
+      "loss": 0.097,
+      "step": 6002
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017345690923328751,
+      "loss": 0.0981,
+      "step": 6003
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017344738060896018,
+      "loss": 0.1364,
+      "step": 6004
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017343785053643593,
+      "loss": 0.0892,
+      "step": 6005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017342831901590269,
+      "loss": 0.1003,
+      "step": 6006
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017341878604754841,
+      "loss": 0.1261,
+      "step": 6007
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017340925163156107,
+      "loss": 0.1149,
+      "step": 6008
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017339971576812864,
+      "loss": 0.1595,
+      "step": 6009
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017339017845743913,
+      "loss": 0.0856,
+      "step": 6010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017338063969968062,
+      "loss": 0.0654,
+      "step": 6011
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017337109949504118,
+      "loss": 0.0758,
+      "step": 6012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017336155784370888,
+      "loss": 0.0947,
+      "step": 6013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017335201474587194,
+      "loss": 0.0939,
+      "step": 6014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017334247020171847,
+      "loss": 0.0855,
+      "step": 6015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017333292421143668,
+      "loss": 0.1219,
+      "step": 6016
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001733233767752148,
+      "loss": 0.075,
+      "step": 6017
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0017331382789324103,
+      "loss": 0.1509,
+      "step": 6018
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.001733042775657037,
+      "loss": 0.1226,
+      "step": 6019
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017329472579279113,
+      "loss": 0.1246,
+      "step": 6020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017328517257469162,
+      "loss": 0.0734,
+      "step": 6021
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017327561791159356,
+      "loss": 0.1078,
+      "step": 6022
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017326606180368535,
+      "loss": 0.1799,
+      "step": 6023
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017325650425115535,
+      "loss": 0.161,
+      "step": 6024
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001732469452541921,
+      "loss": 0.1173,
+      "step": 6025
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017323738481298402,
+      "loss": 0.0809,
+      "step": 6026
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017322782292771963,
+      "loss": 0.0699,
+      "step": 6027
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017321825959858749,
+      "loss": 0.1309,
+      "step": 6028
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001732086948257761,
+      "loss": 0.1344,
+      "step": 6029
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017319912860947416,
+      "loss": 0.1205,
+      "step": 6030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001731895609498702,
+      "loss": 0.0684,
+      "step": 6031
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017317999184715287,
+      "loss": 0.1367,
+      "step": 6032
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017317042130151094,
+      "loss": 0.1108,
+      "step": 6033
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00173160849313133,
+      "loss": 0.1057,
+      "step": 6034
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017315127588220785,
+      "loss": 0.0571,
+      "step": 6035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017314170100892427,
+      "loss": 0.1431,
+      "step": 6036
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00173132124693471,
+      "loss": 0.1091,
+      "step": 6037
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001731225469360369,
+      "loss": 0.0913,
+      "step": 6038
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001731129677368108,
+      "loss": 0.1147,
+      "step": 6039
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017310338709598159,
+      "loss": 0.1317,
+      "step": 6040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017309380501373819,
+      "loss": 0.0845,
+      "step": 6041
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001730842214902695,
+      "loss": 0.1274,
+      "step": 6042
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001730746365257645,
+      "loss": 0.079,
+      "step": 6043
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017306505012041216,
+      "loss": 0.0995,
+      "step": 6044
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017305546227440155,
+      "loss": 0.1099,
+      "step": 6045
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017304587298792168,
+      "loss": 0.0935,
+      "step": 6046
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017303628226116162,
+      "loss": 0.0909,
+      "step": 6047
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001730266900943105,
+      "loss": 0.1007,
+      "step": 6048
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001730170964875574,
+      "loss": 0.1202,
+      "step": 6049
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017300750144109158,
+      "loss": 0.0764,
+      "step": 6050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017299790495510215,
+      "loss": 0.0785,
+      "step": 6051
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017298830702977836,
+      "loss": 0.0963,
+      "step": 6052
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017297870766530945,
+      "loss": 0.1189,
+      "step": 6053
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017296910686188467,
+      "loss": 0.1063,
+      "step": 6054
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017295950461969337,
+      "loss": 0.075,
+      "step": 6055
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017294990093892487,
+      "loss": 0.1095,
+      "step": 6056
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017294029581976852,
+      "loss": 0.0581,
+      "step": 6057
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017293068926241366,
+      "loss": 0.0796,
+      "step": 6058
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.001729210812670498,
+      "loss": 0.0981,
+      "step": 6059
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017291147183386632,
+      "loss": 0.0916,
+      "step": 6060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017290186096305274,
+      "loss": 0.0914,
+      "step": 6061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017289224865479854,
+      "loss": 0.1087,
+      "step": 6062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017288263490929322,
+      "loss": 0.074,
+      "step": 6063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017287301972672639,
+      "loss": 0.0972,
+      "step": 6064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017286340310728758,
+      "loss": 0.0818,
+      "step": 6065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0017285378505116646,
+      "loss": 0.0755,
+      "step": 6066
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017284416555855267,
+      "loss": 0.1205,
+      "step": 6067
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017283454462963586,
+      "loss": 0.0803,
+      "step": 6068
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001728249222646057,
+      "loss": 0.1058,
+      "step": 6069
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017281529846365198,
+      "loss": 0.085,
+      "step": 6070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017280567322696446,
+      "loss": 0.1077,
+      "step": 6071
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017279604655473285,
+      "loss": 0.0741,
+      "step": 6072
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017278641844714702,
+      "loss": 0.1237,
+      "step": 6073
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001727767889043968,
+      "loss": 0.1184,
+      "step": 6074
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017276715792667208,
+      "loss": 0.1143,
+      "step": 6075
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017275752551416273,
+      "loss": 0.1121,
+      "step": 6076
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017274789166705868,
+      "loss": 0.0797,
+      "step": 6077
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017273825638554991,
+      "loss": 0.0904,
+      "step": 6078
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017272861966982638,
+      "loss": 0.1807,
+      "step": 6079
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001727189815200781,
+      "loss": 0.1547,
+      "step": 6080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017270934193649513,
+      "loss": 0.0533,
+      "step": 6081
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017269970091926751,
+      "loss": 0.1051,
+      "step": 6082
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017269005846858537,
+      "loss": 0.2758,
+      "step": 6083
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001726804145846388,
+      "loss": 0.1996,
+      "step": 6084
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017267076926761797,
+      "loss": 0.1237,
+      "step": 6085
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001726611225177131,
+      "loss": 0.1053,
+      "step": 6086
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017265147433511432,
+      "loss": 0.1145,
+      "step": 6087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017264182472001193,
+      "loss": 0.1216,
+      "step": 6088
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017263217367259618,
+      "loss": 0.1088,
+      "step": 6089
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017262252119305734,
+      "loss": 0.0972,
+      "step": 6090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017261286728158579,
+      "loss": 0.1027,
+      "step": 6091
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017260321193837181,
+      "loss": 0.1448,
+      "step": 6092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001725935551636058,
+      "loss": 0.1306,
+      "step": 6093
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017258389695747817,
+      "loss": 0.0931,
+      "step": 6094
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017257423732017943,
+      "loss": 0.1298,
+      "step": 6095
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017256457625189992,
+      "loss": 0.1388,
+      "step": 6096
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017255491375283022,
+      "loss": 0.1184,
+      "step": 6097
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001725452498231608,
+      "loss": 0.1255,
+      "step": 6098
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017253558446308221,
+      "loss": 0.1235,
+      "step": 6099
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001725259176727851,
+      "loss": 0.1053,
+      "step": 6100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017251624945245998,
+      "loss": 0.1177,
+      "step": 6101
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017250657980229754,
+      "loss": 0.1203,
+      "step": 6102
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017249690872248842,
+      "loss": 0.0949,
+      "step": 6103
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017248723621322332,
+      "loss": 0.0583,
+      "step": 6104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017247756227469293,
+      "loss": 0.0979,
+      "step": 6105
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00172467886907088,
+      "loss": 0.115,
+      "step": 6106
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017245821011059937,
+      "loss": 0.1252,
+      "step": 6107
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017244853188541779,
+      "loss": 0.1066,
+      "step": 6108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017243885223173403,
+      "loss": 0.0848,
+      "step": 6109
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0017242917114973907,
+      "loss": 0.0955,
+      "step": 6110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001724194886396237,
+      "loss": 0.0738,
+      "step": 6111
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.001724098047015789,
+      "loss": 0.0752,
+      "step": 6112
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017240011933579557,
+      "loss": 0.1218,
+      "step": 6113
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001723904325424647,
+      "loss": 0.1364,
+      "step": 6114
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017238074432177728,
+      "loss": 0.1241,
+      "step": 6115
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017237105467392435,
+      "loss": 0.1077,
+      "step": 6116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017236136359909694,
+      "loss": 0.0743,
+      "step": 6117
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017235167109748617,
+      "loss": 0.1423,
+      "step": 6118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017234197716928313,
+      "loss": 0.1498,
+      "step": 6119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017233228181467895,
+      "loss": 0.0818,
+      "step": 6120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017232258503386481,
+      "loss": 0.0958,
+      "step": 6121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001723128868270319,
+      "loss": 0.1515,
+      "step": 6122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017230318719437148,
+      "loss": 0.1039,
+      "step": 6123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017229348613607475,
+      "loss": 0.1165,
+      "step": 6124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017228378365233298,
+      "loss": 0.0734,
+      "step": 6125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017227407974333758,
+      "loss": 0.158,
+      "step": 6126
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017226437440927978,
+      "loss": 0.125,
+      "step": 6127
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017225466765035098,
+      "loss": 0.0784,
+      "step": 6128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001722449594667426,
+      "loss": 0.144,
+      "step": 6129
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00172235249858646,
+      "loss": 0.1384,
+      "step": 6130
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017222553882625267,
+      "loss": 0.1569,
+      "step": 6131
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017221582636975411,
+      "loss": 0.1266,
+      "step": 6132
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017220611248934176,
+      "loss": 0.1144,
+      "step": 6133
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017219639718520723,
+      "loss": 0.1335,
+      "step": 6134
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017218668045754205,
+      "loss": 0.1377,
+      "step": 6135
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017217696230653777,
+      "loss": 0.1032,
+      "step": 6136
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017216724273238602,
+      "loss": 0.1118,
+      "step": 6137
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017215752173527848,
+      "loss": 0.0894,
+      "step": 6138
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017214779931540682,
+      "loss": 0.2277,
+      "step": 6139
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001721380754729627,
+      "loss": 0.1283,
+      "step": 6140
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017212835020813794,
+      "loss": 0.0962,
+      "step": 6141
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001721186235211242,
+      "loss": 0.1003,
+      "step": 6142
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017210889541211328,
+      "loss": 0.0605,
+      "step": 6143
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017209916588129704,
+      "loss": 0.0959,
+      "step": 6144
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001720894349288673,
+      "loss": 0.0964,
+      "step": 6145
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001720797025550159,
+      "loss": 0.0884,
+      "step": 6146
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001720699687599348,
+      "loss": 0.084,
+      "step": 6147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017206023354381585,
+      "loss": 0.084,
+      "step": 6148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017205049690685109,
+      "loss": 0.0681,
+      "step": 6149
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017204075884923245,
+      "loss": 0.097,
+      "step": 6150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017203101937115193,
+      "loss": 0.0662,
+      "step": 6151
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017202127847280157,
+      "loss": 0.0826,
+      "step": 6152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017201153615437348,
+      "loss": 0.1131,
+      "step": 6153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017200179241605971,
+      "loss": 0.0831,
+      "step": 6154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001719920472580524,
+      "loss": 0.0903,
+      "step": 6155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.001719823006805437,
+      "loss": 0.084,
+      "step": 6156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017197255268372577,
+      "loss": 0.1255,
+      "step": 6157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0017196280326779085,
+      "loss": 0.0782,
+      "step": 6158
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017195305243293112,
+      "loss": 0.0714,
+      "step": 6159
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001719433001793389,
+      "loss": 0.0804,
+      "step": 6160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017193354650720646,
+      "loss": 0.1306,
+      "step": 6161
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017192379141672612,
+      "loss": 0.1149,
+      "step": 6162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017191403490809018,
+      "loss": 0.1234,
+      "step": 6163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001719042769814911,
+      "loss": 0.1021,
+      "step": 6164
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001718945176371212,
+      "loss": 0.0981,
+      "step": 6165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00171884756875173,
+      "loss": 0.1389,
+      "step": 6166
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017187499469583885,
+      "loss": 0.1011,
+      "step": 6167
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017186523109931128,
+      "loss": 0.0909,
+      "step": 6168
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017185546608578283,
+      "loss": 0.1191,
+      "step": 6169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00171845699655446,
+      "loss": 0.136,
+      "step": 6170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017183593180849343,
+      "loss": 0.1248,
+      "step": 6171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017182616254511763,
+      "loss": 0.1234,
+      "step": 6172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017181639186551125,
+      "loss": 0.1241,
+      "step": 6173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017180661976986702,
+      "loss": 0.1553,
+      "step": 6174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001717968462583775,
+      "loss": 0.0665,
+      "step": 6175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017178707133123547,
+      "loss": 0.1877,
+      "step": 6176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001717772949886337,
+      "loss": 0.0877,
+      "step": 6177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017176751723076483,
+      "loss": 0.0977,
+      "step": 6178
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017175773805782177,
+      "loss": 0.1224,
+      "step": 6179
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001717479574699973,
+      "loss": 0.1282,
+      "step": 6180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017173817546748426,
+      "loss": 0.0934,
+      "step": 6181
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017172839205047558,
+      "loss": 0.1426,
+      "step": 6182
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017171860721916408,
+      "loss": 0.1224,
+      "step": 6183
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017170882097374274,
+      "loss": 0.1155,
+      "step": 6184
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017169903331440454,
+      "loss": 0.0423,
+      "step": 6185
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017168924424134238,
+      "loss": 0.0886,
+      "step": 6186
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017167945375474937,
+      "loss": 0.0995,
+      "step": 6187
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017166966185481853,
+      "loss": 0.1279,
+      "step": 6188
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017165986854174294,
+      "loss": 0.071,
+      "step": 6189
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017165007381571564,
+      "loss": 0.1127,
+      "step": 6190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017164027767692983,
+      "loss": 0.1949,
+      "step": 6191
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017163048012557862,
+      "loss": 0.1056,
+      "step": 6192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017162068116185522,
+      "loss": 0.0921,
+      "step": 6193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.001716108807859528,
+      "loss": 0.1243,
+      "step": 6194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017160107899806465,
+      "loss": 0.1526,
+      "step": 6195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00171591275798384,
+      "loss": 0.1591,
+      "step": 6196
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017158147118710417,
+      "loss": 0.1898,
+      "step": 6197
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017157166516441847,
+      "loss": 0.0927,
+      "step": 6198
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017156185773052025,
+      "loss": 0.08,
+      "step": 6199
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017155204888560287,
+      "loss": 0.0934,
+      "step": 6200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017154223862985978,
+      "loss": 0.1082,
+      "step": 6201
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017153242696348436,
+      "loss": 0.1027,
+      "step": 6202
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017152261388667012,
+      "loss": 0.1582,
+      "step": 6203
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0017151279939961051,
+      "loss": 0.0939,
+      "step": 6204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017150298350249906,
+      "loss": 0.0957,
+      "step": 6205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017149316619552934,
+      "loss": 0.1009,
+      "step": 6206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017148334747889486,
+      "loss": 0.1232,
+      "step": 6207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017147352735278933,
+      "loss": 0.1383,
+      "step": 6208
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017146370581740628,
+      "loss": 0.0715,
+      "step": 6209
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017145388287293937,
+      "loss": 0.0675,
+      "step": 6210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017144405851958232,
+      "loss": 0.0823,
+      "step": 6211
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017143423275752883,
+      "loss": 0.0871,
+      "step": 6212
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017142440558697266,
+      "loss": 0.0983,
+      "step": 6213
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017141457700810751,
+      "loss": 0.0691,
+      "step": 6214
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017140474702112725,
+      "loss": 0.1118,
+      "step": 6215
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017139491562622568,
+      "loss": 0.0863,
+      "step": 6216
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017138508282359666,
+      "loss": 0.129,
+      "step": 6217
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017137524861343403,
+      "loss": 0.0983,
+      "step": 6218
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017136541299593172,
+      "loss": 0.0858,
+      "step": 6219
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017135557597128365,
+      "loss": 0.1476,
+      "step": 6220
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.001713457375396838,
+      "loss": 0.0771,
+      "step": 6221
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017133589770132619,
+      "loss": 0.1274,
+      "step": 6222
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017132605645640475,
+      "loss": 0.1389,
+      "step": 6223
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.001713162138051136,
+      "loss": 0.1263,
+      "step": 6224
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017130636974764678,
+      "loss": 0.0842,
+      "step": 6225
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017129652428419838,
+      "loss": 0.1183,
+      "step": 6226
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017128667741496257,
+      "loss": 0.0808,
+      "step": 6227
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017127682914013342,
+      "loss": 0.1133,
+      "step": 6228
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017126697945990524,
+      "loss": 0.0737,
+      "step": 6229
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017125712837447215,
+      "loss": 0.1323,
+      "step": 6230
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.001712472758840284,
+      "loss": 0.0998,
+      "step": 6231
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017123742198876827,
+      "loss": 0.0797,
+      "step": 6232
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017122756668888606,
+      "loss": 0.1016,
+      "step": 6233
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017121770998457606,
+      "loss": 0.0996,
+      "step": 6234
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017120785187603267,
+      "loss": 0.1653,
+      "step": 6235
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017119799236345022,
+      "loss": 0.0844,
+      "step": 6236
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017118813144702317,
+      "loss": 0.0565,
+      "step": 6237
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017117826912694588,
+      "loss": 0.0977,
+      "step": 6238
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017116840540341285,
+      "loss": 0.1217,
+      "step": 6239
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017115854027661857,
+      "loss": 0.0576,
+      "step": 6240
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017114867374675756,
+      "loss": 0.0974,
+      "step": 6241
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017113880581402432,
+      "loss": 0.076,
+      "step": 6242
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017112893647861346,
+      "loss": 0.0599,
+      "step": 6243
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.001711190657407196,
+      "loss": 0.1152,
+      "step": 6244
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017110919360053731,
+      "loss": 0.0792,
+      "step": 6245
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.001710993200582613,
+      "loss": 0.1163,
+      "step": 6246
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.001710894451140862,
+      "loss": 0.047,
+      "step": 6247
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017107956876820673,
+      "loss": 0.1018,
+      "step": 6248
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017106969102081769,
+      "loss": 0.1343,
+      "step": 6249
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0017105981187211375,
+      "loss": 0.126,
+      "step": 6250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017104993132228974,
+      "loss": 0.124,
+      "step": 6251
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017104004937154049,
+      "loss": 0.1229,
+      "step": 6252
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017103016602006086,
+      "loss": 0.09,
+      "step": 6253
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017102028126804569,
+      "loss": 0.1019,
+      "step": 6254
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017101039511568988,
+      "loss": 0.071,
+      "step": 6255
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017100050756318841,
+      "loss": 0.0898,
+      "step": 6256
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001709906186107362,
+      "loss": 0.0798,
+      "step": 6257
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017098072825852823,
+      "loss": 0.1004,
+      "step": 6258
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017097083650675954,
+      "loss": 0.0829,
+      "step": 6259
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017096094335562514,
+      "loss": 0.0848,
+      "step": 6260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001709510488053201,
+      "loss": 0.0763,
+      "step": 6261
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017094115285603952,
+      "loss": 0.0772,
+      "step": 6262
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017093125550797855,
+      "loss": 0.0936,
+      "step": 6263
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001709213567613323,
+      "loss": 0.1288,
+      "step": 6264
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017091145661629601,
+      "loss": 0.0952,
+      "step": 6265
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001709015550730648,
+      "loss": 0.0632,
+      "step": 6266
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00170891652131834,
+      "loss": 0.1938,
+      "step": 6267
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001708817477927988,
+      "loss": 0.1013,
+      "step": 6268
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001708718420561545,
+      "loss": 0.058,
+      "step": 6269
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017086193492209642,
+      "loss": 0.0932,
+      "step": 6270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017085202639081992,
+      "loss": 0.1335,
+      "step": 6271
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017084211646252036,
+      "loss": 0.1101,
+      "step": 6272
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017083220513739313,
+      "loss": 0.1104,
+      "step": 6273
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017082229241563364,
+      "loss": 0.0603,
+      "step": 6274
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001708123782974374,
+      "loss": 0.133,
+      "step": 6275
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017080246278299988,
+      "loss": 0.1346,
+      "step": 6276
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017079254587251652,
+      "loss": 0.1135,
+      "step": 6277
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017078262756618293,
+      "loss": 0.1078,
+      "step": 6278
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017077270786419463,
+      "loss": 0.0636,
+      "step": 6279
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017076278676674726,
+      "loss": 0.1125,
+      "step": 6280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001707528642740364,
+      "loss": 0.1351,
+      "step": 6281
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017074294038625765,
+      "loss": 0.0791,
+      "step": 6282
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017073301510360682,
+      "loss": 0.0672,
+      "step": 6283
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001707230884262795,
+      "loss": 0.0828,
+      "step": 6284
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017071316035447144,
+      "loss": 0.1367,
+      "step": 6285
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017070323088837843,
+      "loss": 0.1375,
+      "step": 6286
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001706933000281962,
+      "loss": 0.0895,
+      "step": 6287
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001706833677741206,
+      "loss": 0.1229,
+      "step": 6288
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017067343412634748,
+      "loss": 0.0995,
+      "step": 6289
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001706634990850727,
+      "loss": 0.136,
+      "step": 6290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.001706535626504921,
+      "loss": 0.0685,
+      "step": 6291
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017064362482280167,
+      "loss": 0.1204,
+      "step": 6292
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017063368560219733,
+      "loss": 0.1433,
+      "step": 6293
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017062374498887506,
+      "loss": 0.1639,
+      "step": 6294
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017061380298303084,
+      "loss": 0.1218,
+      "step": 6295
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0017060385958486072,
+      "loss": 0.0869,
+      "step": 6296
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001705939147945608,
+      "loss": 0.1438,
+      "step": 6297
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017058396861232713,
+      "loss": 0.1108,
+      "step": 6298
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017057402103835577,
+      "loss": 0.1389,
+      "step": 6299
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017056407207284296,
+      "loss": 0.1545,
+      "step": 6300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001705541217159848,
+      "loss": 0.07,
+      "step": 6301
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001705441699679775,
+      "loss": 0.1764,
+      "step": 6302
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017053421682901733,
+      "loss": 0.1475,
+      "step": 6303
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017052426229930044,
+      "loss": 0.0729,
+      "step": 6304
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017051430637902323,
+      "loss": 0.0861,
+      "step": 6305
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001705043490683819,
+      "loss": 0.1656,
+      "step": 6306
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017049439036757286,
+      "loss": 0.098,
+      "step": 6307
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001704844302767924,
+      "loss": 0.1066,
+      "step": 6308
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00170474468796237,
+      "loss": 0.1617,
+      "step": 6309
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017046450592610295,
+      "loss": 0.1602,
+      "step": 6310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017045454166658683,
+      "loss": 0.0991,
+      "step": 6311
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017044457601788504,
+      "loss": 0.0659,
+      "step": 6312
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017043460898019404,
+      "loss": 0.1189,
+      "step": 6313
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017042464055371043,
+      "loss": 0.1371,
+      "step": 6314
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001704146707386307,
+      "loss": 0.1106,
+      "step": 6315
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001704046995351515,
+      "loss": 0.1566,
+      "step": 6316
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017039472694346937,
+      "loss": 0.1005,
+      "step": 6317
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017038475296378098,
+      "loss": 0.1089,
+      "step": 6318
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017037477759628294,
+      "loss": 0.1295,
+      "step": 6319
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017036480084117202,
+      "loss": 0.1074,
+      "step": 6320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017035482269864488,
+      "loss": 0.1476,
+      "step": 6321
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017034484316889832,
+      "loss": 0.1106,
+      "step": 6322
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00170334862252129,
+      "loss": 0.0726,
+      "step": 6323
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017032487994853385,
+      "loss": 0.0945,
+      "step": 6324
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017031489625830962,
+      "loss": 0.1185,
+      "step": 6325
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017030491118165318,
+      "loss": 0.1261,
+      "step": 6326
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001702949247187614,
+      "loss": 0.1266,
+      "step": 6327
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017028493686983119,
+      "loss": 0.0778,
+      "step": 6328
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017027494763505954,
+      "loss": 0.1002,
+      "step": 6329
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017026495701464332,
+      "loss": 0.1348,
+      "step": 6330
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017025496500877957,
+      "loss": 0.0917,
+      "step": 6331
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017024497161766533,
+      "loss": 0.0907,
+      "step": 6332
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017023497684149759,
+      "loss": 0.1163,
+      "step": 6333
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017022498068047342,
+      "loss": 0.127,
+      "step": 6334
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017021498313479,
+      "loss": 0.1285,
+      "step": 6335
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017020498420464438,
+      "loss": 0.1105,
+      "step": 6336
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017019498389023373,
+      "loss": 0.1307,
+      "step": 6337
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017018498219175528,
+      "loss": 0.0951,
+      "step": 6338
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017017497910940616,
+      "loss": 0.2546,
+      "step": 6339
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017016497464338364,
+      "loss": 0.0854,
+      "step": 6340
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0017015496879388499,
+      "loss": 0.0999,
+      "step": 6341
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.001701449615611075,
+      "loss": 0.1024,
+      "step": 6342
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001701349529452485,
+      "loss": 0.1923,
+      "step": 6343
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017012494294650527,
+      "loss": 0.1163,
+      "step": 6344
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017011493156507526,
+      "loss": 0.1531,
+      "step": 6345
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017010491880115584,
+      "loss": 0.125,
+      "step": 6346
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017009490465494442,
+      "loss": 0.1298,
+      "step": 6347
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001700848891266385,
+      "loss": 0.126,
+      "step": 6348
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001700748722164355,
+      "loss": 0.0939,
+      "step": 6349
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017006485392453295,
+      "loss": 0.1013,
+      "step": 6350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001700548342511284,
+      "loss": 0.0786,
+      "step": 6351
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001700448131964194,
+      "loss": 0.0554,
+      "step": 6352
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017003479076060357,
+      "loss": 0.0831,
+      "step": 6353
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001700247669438785,
+      "loss": 0.1143,
+      "step": 6354
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001700147417464418,
+      "loss": 0.0986,
+      "step": 6355
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0017000471516849123,
+      "loss": 0.1165,
+      "step": 6356
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001699946872102244,
+      "loss": 0.0747,
+      "step": 6357
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016998465787183908,
+      "loss": 0.1338,
+      "step": 6358
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016997462715353303,
+      "loss": 0.1575,
+      "step": 6359
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00169964595055504,
+      "loss": 0.0754,
+      "step": 6360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016995456157794981,
+      "loss": 0.0837,
+      "step": 6361
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016994452672106832,
+      "loss": 0.124,
+      "step": 6362
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016993449048505736,
+      "loss": 0.1342,
+      "step": 6363
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016992445287011481,
+      "loss": 0.0934,
+      "step": 6364
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016991441387643864,
+      "loss": 0.0781,
+      "step": 6365
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016990437350422676,
+      "loss": 0.0872,
+      "step": 6366
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001698943317536771,
+      "loss": 0.1125,
+      "step": 6367
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016988428862498774,
+      "loss": 0.1392,
+      "step": 6368
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016987424411835665,
+      "loss": 0.0452,
+      "step": 6369
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016986419823398192,
+      "loss": 0.1028,
+      "step": 6370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016985415097206156,
+      "loss": 0.0847,
+      "step": 6371
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016984410233279376,
+      "loss": 0.1599,
+      "step": 6372
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001698340523163766,
+      "loss": 0.0712,
+      "step": 6373
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016982400092300828,
+      "loss": 0.1373,
+      "step": 6374
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016981394815288695,
+      "loss": 0.1923,
+      "step": 6375
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016980389400621083,
+      "loss": 0.1527,
+      "step": 6376
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016979383848317819,
+      "loss": 0.1231,
+      "step": 6377
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016978378158398729,
+      "loss": 0.0999,
+      "step": 6378
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016977372330883639,
+      "loss": 0.1177,
+      "step": 6379
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016976366365792385,
+      "loss": 0.1533,
+      "step": 6380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016975360263144802,
+      "loss": 0.0808,
+      "step": 6381
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001697435402296073,
+      "loss": 0.0694,
+      "step": 6382
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016973347645260004,
+      "loss": 0.1399,
+      "step": 6383
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016972341130062469,
+      "loss": 0.1393,
+      "step": 6384
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016971334477387972,
+      "loss": 0.1263,
+      "step": 6385
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016970327687256363,
+      "loss": 0.1466,
+      "step": 6386
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.001696932075968749,
+      "loss": 0.1318,
+      "step": 6387
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016968313694701212,
+      "loss": 0.0966,
+      "step": 6388
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0016967306492317383,
+      "loss": 0.0984,
+      "step": 6389
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001696629915255586,
+      "loss": 0.0815,
+      "step": 6390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016965291675436504,
+      "loss": 0.0752,
+      "step": 6391
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001696428406097919,
+      "loss": 0.0847,
+      "step": 6392
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016963276309203777,
+      "loss": 0.0949,
+      "step": 6393
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016962268420130137,
+      "loss": 0.1014,
+      "step": 6394
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016961260393778142,
+      "loss": 0.0667,
+      "step": 6395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001696025223016767,
+      "loss": 0.0959,
+      "step": 6396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016959243929318597,
+      "loss": 0.081,
+      "step": 6397
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016958235491250809,
+      "loss": 0.1558,
+      "step": 6398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016957226915984185,
+      "loss": 0.063,
+      "step": 6399
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016956218203538615,
+      "loss": 0.1064,
+      "step": 6400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001695520935393398,
+      "loss": 0.1183,
+      "step": 6401
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016954200367190182,
+      "loss": 0.1152,
+      "step": 6402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016953191243327113,
+      "loss": 0.1112,
+      "step": 6403
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016952181982364668,
+      "loss": 0.1252,
+      "step": 6404
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001695117258432275,
+      "loss": 0.1146,
+      "step": 6405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016950163049221257,
+      "loss": 0.1167,
+      "step": 6406
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00169491533770801,
+      "loss": 0.1443,
+      "step": 6407
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016948143567919183,
+      "loss": 0.1868,
+      "step": 6408
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001694713362175842,
+      "loss": 0.1195,
+      "step": 6409
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016946123538617726,
+      "loss": 0.1334,
+      "step": 6410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016945113318517012,
+      "loss": 0.1715,
+      "step": 6411
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016944102961476198,
+      "loss": 0.0611,
+      "step": 6412
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001694309246751521,
+      "loss": 0.1094,
+      "step": 6413
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001694208183665397,
+      "loss": 0.1132,
+      "step": 6414
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016941071068912403,
+      "loss": 0.1757,
+      "step": 6415
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001694006016431044,
+      "loss": 0.1149,
+      "step": 6416
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016939049122868016,
+      "loss": 0.0978,
+      "step": 6417
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016938037944605063,
+      "loss": 0.1085,
+      "step": 6418
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001693702662954152,
+      "loss": 0.1017,
+      "step": 6419
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016936015177697326,
+      "loss": 0.0996,
+      "step": 6420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016935003589092427,
+      "loss": 0.093,
+      "step": 6421
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001693399186374677,
+      "loss": 0.1581,
+      "step": 6422
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00169329800016803,
+      "loss": 0.0872,
+      "step": 6423
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001693196800291297,
+      "loss": 0.1838,
+      "step": 6424
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016930955867464736,
+      "loss": 0.0839,
+      "step": 6425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016929943595355552,
+      "loss": 0.1555,
+      "step": 6426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016928931186605376,
+      "loss": 0.0911,
+      "step": 6427
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016927918641234175,
+      "loss": 0.0612,
+      "step": 6428
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001692690595926191,
+      "loss": 0.142,
+      "step": 6429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.001692589314070855,
+      "loss": 0.0969,
+      "step": 6430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016924880185594064,
+      "loss": 0.1012,
+      "step": 6431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016923867093938429,
+      "loss": 0.0884,
+      "step": 6432
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016922853865761615,
+      "loss": 0.207,
+      "step": 6433
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016921840501083602,
+      "loss": 0.1007,
+      "step": 6434
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0016920826999924372,
+      "loss": 0.0919,
+      "step": 6435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001691981336230391,
+      "loss": 0.1333,
+      "step": 6436
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016918799588242198,
+      "loss": 0.097,
+      "step": 6437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001691778567775923,
+      "loss": 0.2056,
+      "step": 6438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016916771630874996,
+      "loss": 0.1095,
+      "step": 6439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016915757447609492,
+      "loss": 0.1141,
+      "step": 6440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001691474312798271,
+      "loss": 0.0764,
+      "step": 6441
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016913728672014652,
+      "loss": 0.1372,
+      "step": 6442
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016912714079725325,
+      "loss": 0.1796,
+      "step": 6443
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016911699351134727,
+      "loss": 0.0627,
+      "step": 6444
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016910684486262876,
+      "loss": 0.0816,
+      "step": 6445
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001690966948512977,
+      "loss": 0.1242,
+      "step": 6446
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016908654347755432,
+      "loss": 0.1718,
+      "step": 6447
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016907639074159873,
+      "loss": 0.1583,
+      "step": 6448
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016906623664363113,
+      "loss": 0.0929,
+      "step": 6449
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016905608118385174,
+      "loss": 0.1427,
+      "step": 6450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001690459243624608,
+      "loss": 0.1536,
+      "step": 6451
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016903576617965856,
+      "loss": 0.0887,
+      "step": 6452
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016902560663564535,
+      "loss": 0.108,
+      "step": 6453
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016901544573062143,
+      "loss": 0.1278,
+      "step": 6454
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001690052834647872,
+      "loss": 0.0858,
+      "step": 6455
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00168995119838343,
+      "loss": 0.0623,
+      "step": 6456
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016898495485148924,
+      "loss": 0.1111,
+      "step": 6457
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016897478850442637,
+      "loss": 0.1592,
+      "step": 6458
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016896462079735485,
+      "loss": 0.1416,
+      "step": 6459
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016895445173047512,
+      "loss": 0.1501,
+      "step": 6460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016894428130398774,
+      "loss": 0.1084,
+      "step": 6461
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016893410951809317,
+      "loss": 0.1431,
+      "step": 6462
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016892393637299207,
+      "loss": 0.0758,
+      "step": 6463
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016891376186888491,
+      "loss": 0.1082,
+      "step": 6464
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001689035860059724,
+      "loss": 0.1309,
+      "step": 6465
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016889340878445517,
+      "loss": 0.1042,
+      "step": 6466
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016888323020453387,
+      "loss": 0.0721,
+      "step": 6467
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016887305026640918,
+      "loss": 0.1116,
+      "step": 6468
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016886286897028187,
+      "loss": 0.078,
+      "step": 6469
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016885268631635264,
+      "loss": 0.1133,
+      "step": 6470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.001688425023048223,
+      "loss": 0.1084,
+      "step": 6471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016883231693589163,
+      "loss": 0.1008,
+      "step": 6472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016882213020976147,
+      "loss": 0.1071,
+      "step": 6473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016881194212663266,
+      "loss": 0.1041,
+      "step": 6474
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016880175268670615,
+      "loss": 0.1729,
+      "step": 6475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016879156189018275,
+      "loss": 0.1641,
+      "step": 6476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016878136973726345,
+      "loss": 0.1054,
+      "step": 6477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016877117622814924,
+      "loss": 0.1373,
+      "step": 6478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016876098136304105,
+      "loss": 0.1035,
+      "step": 6479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016875078514213995,
+      "loss": 0.0935,
+      "step": 6480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0016874058756564693,
+      "loss": 0.1333,
+      "step": 6481
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016873038863376314,
+      "loss": 0.0688,
+      "step": 6482
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016872018834668961,
+      "loss": 0.0828,
+      "step": 6483
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016870998670462745,
+      "loss": 0.127,
+      "step": 6484
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016869978370777788,
+      "loss": 0.1197,
+      "step": 6485
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016868957935634202,
+      "loss": 0.1035,
+      "step": 6486
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001686793736505211,
+      "loss": 0.0914,
+      "step": 6487
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016866916659051636,
+      "loss": 0.1067,
+      "step": 6488
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016865895817652903,
+      "loss": 0.1498,
+      "step": 6489
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016864874840876044,
+      "loss": 0.0927,
+      "step": 6490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016863853728741182,
+      "loss": 0.0927,
+      "step": 6491
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016862832481268455,
+      "loss": 0.1205,
+      "step": 6492
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016861811098478005,
+      "loss": 0.0812,
+      "step": 6493
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001686078958038996,
+      "loss": 0.1505,
+      "step": 6494
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016859767927024471,
+      "loss": 0.0905,
+      "step": 6495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016858746138401681,
+      "loss": 0.16,
+      "step": 6496
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016857724214541729,
+      "loss": 0.0844,
+      "step": 6497
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016856702155464778,
+      "loss": 0.0807,
+      "step": 6498
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016855679961190974,
+      "loss": 0.1058,
+      "step": 6499
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016854657631740465,
+      "loss": 0.114,
+      "step": 6500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001685363516713342,
+      "loss": 0.1233,
+      "step": 6501
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016852612567389994,
+      "loss": 0.0867,
+      "step": 6502
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001685158983253035,
+      "loss": 0.0914,
+      "step": 6503
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016850566962574654,
+      "loss": 0.0712,
+      "step": 6504
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016849543957543078,
+      "loss": 0.1388,
+      "step": 6505
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016848520817455788,
+      "loss": 0.1539,
+      "step": 6506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001684749754233296,
+      "loss": 0.1697,
+      "step": 6507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001684647413219477,
+      "loss": 0.0972,
+      "step": 6508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00168454505870614,
+      "loss": 0.1066,
+      "step": 6509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016844426906953026,
+      "loss": 0.1283,
+      "step": 6510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016843403091889835,
+      "loss": 0.1469,
+      "step": 6511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016842379141892014,
+      "loss": 0.0699,
+      "step": 6512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016841355056979755,
+      "loss": 0.1039,
+      "step": 6513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016840330837173249,
+      "loss": 0.0951,
+      "step": 6514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016839306482492689,
+      "loss": 0.2401,
+      "step": 6515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016838281992958275,
+      "loss": 0.0956,
+      "step": 6516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016837257368590205,
+      "loss": 0.1671,
+      "step": 6517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016836232609408685,
+      "loss": 0.1323,
+      "step": 6518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001683520771543392,
+      "loss": 0.1064,
+      "step": 6519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016834182686686118,
+      "loss": 0.0684,
+      "step": 6520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016833157523185487,
+      "loss": 0.1195,
+      "step": 6521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016832132224952245,
+      "loss": 0.1446,
+      "step": 6522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016831106792006608,
+      "loss": 0.0821,
+      "step": 6523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.001683008122436879,
+      "loss": 0.1885,
+      "step": 6524
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016829055522059018,
+      "loss": 0.0925,
+      "step": 6525
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016828029685097515,
+      "loss": 0.1423,
+      "step": 6526
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0016827003713504505,
+      "loss": 0.0811,
+      "step": 6527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.001682597760730022,
+      "loss": 0.0684,
+      "step": 6528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016824951366504895,
+      "loss": 0.0783,
+      "step": 6529
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016823924991138757,
+      "loss": 0.1121,
+      "step": 6530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016822898481222053,
+      "loss": 0.1166,
+      "step": 6531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016821871836775015,
+      "loss": 0.065,
+      "step": 6532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016820845057817894,
+      "loss": 0.099,
+      "step": 6533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016819818144370926,
+      "loss": 0.1053,
+      "step": 6534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016818791096454367,
+      "loss": 0.0829,
+      "step": 6535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016817763914088464,
+      "loss": 0.0884,
+      "step": 6536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016816736597293472,
+      "loss": 0.1141,
+      "step": 6537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016815709146089644,
+      "loss": 0.0911,
+      "step": 6538
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016814681560497246,
+      "loss": 0.1931,
+      "step": 6539
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.001681365384053653,
+      "loss": 0.075,
+      "step": 6540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016812625986227766,
+      "loss": 0.0744,
+      "step": 6541
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.001681159799759122,
+      "loss": 0.0901,
+      "step": 6542
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.001681056987464716,
+      "loss": 0.1401,
+      "step": 6543
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016809541617415857,
+      "loss": 0.1193,
+      "step": 6544
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016808513225917588,
+      "loss": 0.1125,
+      "step": 6545
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016807484700172633,
+      "loss": 0.0676,
+      "step": 6546
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016806456040201264,
+      "loss": 0.0985,
+      "step": 6547
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016805427246023767,
+      "loss": 0.0732,
+      "step": 6548
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016804398317660433,
+      "loss": 0.1283,
+      "step": 6549
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016803369255131541,
+      "loss": 0.0819,
+      "step": 6550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.001680234005845739,
+      "loss": 0.0956,
+      "step": 6551
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016801310727658266,
+      "loss": 0.1148,
+      "step": 6552
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016800281262754468,
+      "loss": 0.0887,
+      "step": 6553
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016799251663766294,
+      "loss": 0.1033,
+      "step": 6554
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016798221930714044,
+      "loss": 0.068,
+      "step": 6555
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016797192063618027,
+      "loss": 0.0947,
+      "step": 6556
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016796162062498539,
+      "loss": 0.1129,
+      "step": 6557
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016795131927375899,
+      "loss": 0.129,
+      "step": 6558
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016794101658270417,
+      "loss": 0.0806,
+      "step": 6559
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016793071255202401,
+      "loss": 0.0823,
+      "step": 6560
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016792040718192174,
+      "loss": 0.0958,
+      "step": 6561
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016791010047260055,
+      "loss": 0.1111,
+      "step": 6562
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016789979242426364,
+      "loss": 0.1481,
+      "step": 6563
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016788948303711425,
+      "loss": 0.1003,
+      "step": 6564
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016787917231135569,
+      "loss": 0.0856,
+      "step": 6565
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016786886024719126,
+      "loss": 0.1471,
+      "step": 6566
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016785854684482425,
+      "loss": 0.1275,
+      "step": 6567
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016784823210445805,
+      "loss": 0.0956,
+      "step": 6568
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00167837916026296,
+      "loss": 0.0711,
+      "step": 6569
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016782759861054158,
+      "loss": 0.0914,
+      "step": 6570
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016781727985739814,
+      "loss": 0.0797,
+      "step": 6571
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.001678069597670692,
+      "loss": 0.0858,
+      "step": 6572
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0016779663833975823,
+      "loss": 0.1075,
+      "step": 6573
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001677863155756687,
+      "loss": 0.0968,
+      "step": 6574
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016777599147500422,
+      "loss": 0.0656,
+      "step": 6575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016776566603796833,
+      "loss": 0.1079,
+      "step": 6576
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016775533926476455,
+      "loss": 0.1143,
+      "step": 6577
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001677450111555966,
+      "loss": 0.1414,
+      "step": 6578
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001677346817106681,
+      "loss": 0.0785,
+      "step": 6579
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001677243509301827,
+      "loss": 0.163,
+      "step": 6580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016771401881434409,
+      "loss": 0.1234,
+      "step": 6581
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016770368536335603,
+      "loss": 0.0831,
+      "step": 6582
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016769335057742222,
+      "loss": 0.1249,
+      "step": 6583
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016768301445674646,
+      "loss": 0.0933,
+      "step": 6584
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016767267700153257,
+      "loss": 0.073,
+      "step": 6585
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016766233821198435,
+      "loss": 0.0822,
+      "step": 6586
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016765199808830566,
+      "loss": 0.1082,
+      "step": 6587
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016764165663070041,
+      "loss": 0.175,
+      "step": 6588
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016763131383937247,
+      "loss": 0.0875,
+      "step": 6589
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016762096971452577,
+      "loss": 0.1201,
+      "step": 6590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016761062425636431,
+      "loss": 0.0875,
+      "step": 6591
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016760027746509202,
+      "loss": 0.1295,
+      "step": 6592
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016758992934091301,
+      "loss": 0.0766,
+      "step": 6593
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016757957988403121,
+      "loss": 0.1169,
+      "step": 6594
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016756922909465074,
+      "loss": 0.0809,
+      "step": 6595
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016755887697297571,
+      "loss": 0.0922,
+      "step": 6596
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016754852351921017,
+      "loss": 0.0901,
+      "step": 6597
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016753816873355833,
+      "loss": 0.1675,
+      "step": 6598
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016752781261622433,
+      "loss": 0.1108,
+      "step": 6599
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016751745516741237,
+      "loss": 0.1829,
+      "step": 6600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016750709638732669,
+      "loss": 0.0927,
+      "step": 6601
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001674967362761715,
+      "loss": 0.1303,
+      "step": 6602
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016748637483415111,
+      "loss": 0.1093,
+      "step": 6603
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016747601206146979,
+      "loss": 0.0712,
+      "step": 6604
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001674656479583319,
+      "loss": 0.0895,
+      "step": 6605
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001674552825249418,
+      "loss": 0.084,
+      "step": 6606
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001674449157615038,
+      "loss": 0.1514,
+      "step": 6607
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016743454766822241,
+      "loss": 0.1112,
+      "step": 6608
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00167424178245302,
+      "loss": 0.082,
+      "step": 6609
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016741380749294701,
+      "loss": 0.0916,
+      "step": 6610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016740343541136197,
+      "loss": 0.1086,
+      "step": 6611
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016739306200075142,
+      "loss": 0.0709,
+      "step": 6612
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001673826872613198,
+      "loss": 0.1602,
+      "step": 6613
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016737231119327172,
+      "loss": 0.0958,
+      "step": 6614
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016736193379681182,
+      "loss": 0.1093,
+      "step": 6615
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016735155507214463,
+      "loss": 0.1351,
+      "step": 6616
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016734117501947489,
+      "loss": 0.0929,
+      "step": 6617
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0016733079363900716,
+      "loss": 0.0578,
+      "step": 6618
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.001673204109309462,
+      "loss": 0.1003,
+      "step": 6619
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016731002689549673,
+      "loss": 0.1107,
+      "step": 6620
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001672996415328635,
+      "loss": 0.1169,
+      "step": 6621
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016728925484325123,
+      "loss": 0.0693,
+      "step": 6622
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001672788668268648,
+      "loss": 0.0914,
+      "step": 6623
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016726847748390896,
+      "loss": 0.0992,
+      "step": 6624
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001672580868145886,
+      "loss": 0.0875,
+      "step": 6625
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016724769481910862,
+      "loss": 0.088,
+      "step": 6626
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016723730149767384,
+      "loss": 0.1968,
+      "step": 6627
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001672269068504893,
+      "loss": 0.0555,
+      "step": 6628
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001672165108777599,
+      "loss": 0.0818,
+      "step": 6629
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001672061135796906,
+      "loss": 0.0959,
+      "step": 6630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016719571495648644,
+      "loss": 0.1846,
+      "step": 6631
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016718531500835249,
+      "loss": 0.079,
+      "step": 6632
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016717491373549375,
+      "loss": 0.1121,
+      "step": 6633
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016716451113811528,
+      "loss": 0.1096,
+      "step": 6634
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016715410721642226,
+      "loss": 0.153,
+      "step": 6635
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016714370197061983,
+      "loss": 0.1102,
+      "step": 6636
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016713329540091312,
+      "loss": 0.1555,
+      "step": 6637
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016712288750750735,
+      "loss": 0.0835,
+      "step": 6638
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016711247829060772,
+      "loss": 0.1044,
+      "step": 6639
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016710206775041945,
+      "loss": 0.1105,
+      "step": 6640
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016709165588714783,
+      "loss": 0.1483,
+      "step": 6641
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016708124270099819,
+      "loss": 0.1079,
+      "step": 6642
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016707082819217582,
+      "loss": 0.1755,
+      "step": 6643
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016706041236088606,
+      "loss": 0.1185,
+      "step": 6644
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001670499952073343,
+      "loss": 0.097,
+      "step": 6645
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001670395767317259,
+      "loss": 0.1431,
+      "step": 6646
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016702915693426637,
+      "loss": 0.1578,
+      "step": 6647
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016701873581516109,
+      "loss": 0.1508,
+      "step": 6648
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016700831337461558,
+      "loss": 0.1228,
+      "step": 6649
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016699788961283529,
+      "loss": 0.1445,
+      "step": 6650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016698746453002581,
+      "loss": 0.1399,
+      "step": 6651
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016697703812639264,
+      "loss": 0.1238,
+      "step": 6652
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016696661040214145,
+      "loss": 0.1005,
+      "step": 6653
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016695618135747772,
+      "loss": 0.1025,
+      "step": 6654
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016694575099260721,
+      "loss": 0.0675,
+      "step": 6655
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001669353193077355,
+      "loss": 0.1411,
+      "step": 6656
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016692488630306833,
+      "loss": 0.0709,
+      "step": 6657
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016691445197881136,
+      "loss": 0.0879,
+      "step": 6658
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016690401633517038,
+      "loss": 0.0939,
+      "step": 6659
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001668935793723511,
+      "loss": 0.0725,
+      "step": 6660
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016688314109055936,
+      "loss": 0.0698,
+      "step": 6661
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016687270149000095,
+      "loss": 0.1219,
+      "step": 6662
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016686226057088173,
+      "loss": 0.2103,
+      "step": 6663
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0016685181833340753,
+      "loss": 0.0793,
+      "step": 6664
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.001668413747777843,
+      "loss": 0.1069,
+      "step": 6665
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001668309299042179,
+      "loss": 0.1083,
+      "step": 6666
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001668204837129143,
+      "loss": 0.0967,
+      "step": 6667
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016681003620407956,
+      "loss": 0.0643,
+      "step": 6668
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016679958737791953,
+      "loss": 0.1105,
+      "step": 6669
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016678913723464032,
+      "loss": 0.0837,
+      "step": 6670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016677868577444797,
+      "loss": 0.102,
+      "step": 6671
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016676823299754855,
+      "loss": 0.0837,
+      "step": 6672
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016675777890414816,
+      "loss": 0.0983,
+      "step": 6673
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016674732349445294,
+      "loss": 0.0698,
+      "step": 6674
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016673686676866902,
+      "loss": 0.08,
+      "step": 6675
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016672640872700262,
+      "loss": 0.0908,
+      "step": 6676
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001667159493696599,
+      "loss": 0.0911,
+      "step": 6677
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016670548869684712,
+      "loss": 0.109,
+      "step": 6678
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016669502670877054,
+      "loss": 0.1081,
+      "step": 6679
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016668456340563642,
+      "loss": 0.0638,
+      "step": 6680
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016667409878765112,
+      "loss": 0.0727,
+      "step": 6681
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016666363285502092,
+      "loss": 0.0985,
+      "step": 6682
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016665316560795218,
+      "loss": 0.1244,
+      "step": 6683
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016664269704665133,
+      "loss": 0.0587,
+      "step": 6684
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016663222717132479,
+      "loss": 0.1361,
+      "step": 6685
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016662175598217894,
+      "loss": 0.1381,
+      "step": 6686
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016661128347942031,
+      "loss": 0.1285,
+      "step": 6687
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016660080966325534,
+      "loss": 0.0721,
+      "step": 6688
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001665903345338906,
+      "loss": 0.0936,
+      "step": 6689
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016657985809153258,
+      "loss": 0.0724,
+      "step": 6690
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016656938033638787,
+      "loss": 0.1659,
+      "step": 6691
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016655890126866307,
+      "loss": 0.0942,
+      "step": 6692
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016654842088856478,
+      "loss": 0.0852,
+      "step": 6693
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016653793919629964,
+      "loss": 0.0781,
+      "step": 6694
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001665274561920744,
+      "loss": 0.1124,
+      "step": 6695
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016651697187609572,
+      "loss": 0.1024,
+      "step": 6696
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016650648624857026,
+      "loss": 0.0994,
+      "step": 6697
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016649599930970483,
+      "loss": 0.0908,
+      "step": 6698
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001664855110597062,
+      "loss": 0.0892,
+      "step": 6699
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016647502149878114,
+      "loss": 0.0809,
+      "step": 6700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016646453062713654,
+      "loss": 0.1041,
+      "step": 6701
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.001664540384449792,
+      "loss": 0.14,
+      "step": 6702
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016644354495251603,
+      "loss": 0.1109,
+      "step": 6703
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016643305014995388,
+      "loss": 0.054,
+      "step": 6704
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016642255403749976,
+      "loss": 0.1058,
+      "step": 6705
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016641205661536058,
+      "loss": 0.1283,
+      "step": 6706
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016640155788374335,
+      "loss": 0.0801,
+      "step": 6707
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016639105784285505,
+      "loss": 0.0963,
+      "step": 6708
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016638055649290271,
+      "loss": 0.1223,
+      "step": 6709
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016637005383409343,
+      "loss": 0.1504,
+      "step": 6710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0016635954986663427,
+      "loss": 0.0826,
+      "step": 6711
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016634904459073233,
+      "loss": 0.0817,
+      "step": 6712
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016633853800659475,
+      "loss": 0.1118,
+      "step": 6713
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016632803011442873,
+      "loss": 0.1265,
+      "step": 6714
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001663175209144414,
+      "loss": 0.0593,
+      "step": 6715
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016630701040684004,
+      "loss": 0.0789,
+      "step": 6716
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016629649859183182,
+      "loss": 0.1361,
+      "step": 6717
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016628598546962411,
+      "loss": 0.098,
+      "step": 6718
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016627547104042409,
+      "loss": 0.1443,
+      "step": 6719
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001662649553044391,
+      "loss": 0.0801,
+      "step": 6720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016625443826187656,
+      "loss": 0.1472,
+      "step": 6721
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016624391991294377,
+      "loss": 0.1132,
+      "step": 6722
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016623340025784814,
+      "loss": 0.0729,
+      "step": 6723
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001662228792967971,
+      "loss": 0.0906,
+      "step": 6724
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016621235702999808,
+      "loss": 0.0836,
+      "step": 6725
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016620183345765858,
+      "loss": 0.1094,
+      "step": 6726
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016619130857998606,
+      "loss": 0.1384,
+      "step": 6727
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016618078239718807,
+      "loss": 0.1191,
+      "step": 6728
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016617025490947217,
+      "loss": 0.0931,
+      "step": 6729
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001661597261170459,
+      "loss": 0.0691,
+      "step": 6730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016614919602011692,
+      "loss": 0.0615,
+      "step": 6731
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001661386646188928,
+      "loss": 0.1112,
+      "step": 6732
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016612813191358123,
+      "loss": 0.0991,
+      "step": 6733
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016611759790438982,
+      "loss": 0.1263,
+      "step": 6734
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016610706259152638,
+      "loss": 0.097,
+      "step": 6735
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016609652597519855,
+      "loss": 0.1143,
+      "step": 6736
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016608598805561412,
+      "loss": 0.1447,
+      "step": 6737
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016607544883298088,
+      "loss": 0.104,
+      "step": 6738
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016606490830750665,
+      "loss": 0.1528,
+      "step": 6739
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001660543664793992,
+      "loss": 0.1218,
+      "step": 6740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016604382334886646,
+      "loss": 0.1073,
+      "step": 6741
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016603327891611629,
+      "loss": 0.1586,
+      "step": 6742
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016602273318135656,
+      "loss": 0.1765,
+      "step": 6743
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016601218614479523,
+      "loss": 0.145,
+      "step": 6744
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016600163780664033,
+      "loss": 0.1104,
+      "step": 6745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016599108816709972,
+      "loss": 0.1208,
+      "step": 6746
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001659805372263815,
+      "loss": 0.0907,
+      "step": 6747
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001659699849846937,
+      "loss": 0.0743,
+      "step": 6748
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016595943144224435,
+      "loss": 0.1064,
+      "step": 6749
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016594887659924153,
+      "loss": 0.1171,
+      "step": 6750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016593832045589344,
+      "loss": 0.1389,
+      "step": 6751
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016592776301240812,
+      "loss": 0.0986,
+      "step": 6752
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016591720426899382,
+      "loss": 0.0645,
+      "step": 6753
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016590664422585868,
+      "loss": 0.1295,
+      "step": 6754
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016589608288321092,
+      "loss": 0.0682,
+      "step": 6755
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.001658855202412588,
+      "loss": 0.0664,
+      "step": 6756
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016587495630021054,
+      "loss": 0.0784,
+      "step": 6757
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0016586439106027451,
+      "loss": 0.0986,
+      "step": 6758
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016585382452165899,
+      "loss": 0.1323,
+      "step": 6759
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016584325668457232,
+      "loss": 0.0806,
+      "step": 6760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001658326875492229,
+      "loss": 0.0775,
+      "step": 6761
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001658221171158191,
+      "loss": 0.0768,
+      "step": 6762
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016581154538456936,
+      "loss": 0.0645,
+      "step": 6763
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016580097235568208,
+      "loss": 0.0847,
+      "step": 6764
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001657903980293658,
+      "loss": 0.0963,
+      "step": 6765
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016577982240582898,
+      "loss": 0.113,
+      "step": 6766
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016576924548528016,
+      "loss": 0.0928,
+      "step": 6767
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016575866726792788,
+      "loss": 0.1387,
+      "step": 6768
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016574808775398071,
+      "loss": 0.1044,
+      "step": 6769
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016573750694364728,
+      "loss": 0.1182,
+      "step": 6770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016572692483713622,
+      "loss": 0.1008,
+      "step": 6771
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016571634143465613,
+      "loss": 0.1323,
+      "step": 6772
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001657057567364157,
+      "loss": 0.0976,
+      "step": 6773
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016569517074262368,
+      "loss": 0.0933,
+      "step": 6774
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016568458345348877,
+      "loss": 0.0879,
+      "step": 6775
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016567399486921972,
+      "loss": 0.0829,
+      "step": 6776
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016566340499002533,
+      "loss": 0.0847,
+      "step": 6777
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016565281381611438,
+      "loss": 0.1338,
+      "step": 6778
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001656422213476957,
+      "loss": 0.1268,
+      "step": 6779
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016563162758497822,
+      "loss": 0.1038,
+      "step": 6780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001656210325281707,
+      "loss": 0.1093,
+      "step": 6781
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016561043617748213,
+      "loss": 0.0916,
+      "step": 6782
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016559983853312139,
+      "loss": 0.1555,
+      "step": 6783
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016558923959529752,
+      "loss": 0.0598,
+      "step": 6784
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016557863936421944,
+      "loss": 0.1055,
+      "step": 6785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001655680378400962,
+      "loss": 0.1262,
+      "step": 6786
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016555743502313675,
+      "loss": 0.0746,
+      "step": 6787
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016554683091355028,
+      "loss": 0.0882,
+      "step": 6788
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016553622551154577,
+      "loss": 0.1741,
+      "step": 6789
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016552561881733236,
+      "loss": 0.0865,
+      "step": 6790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016551501083111925,
+      "loss": 0.0872,
+      "step": 6791
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001655044015531155,
+      "loss": 0.1012,
+      "step": 6792
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016549379098353034,
+      "loss": 0.1375,
+      "step": 6793
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016548317912257304,
+      "loss": 0.1403,
+      "step": 6794
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016547256597045277,
+      "loss": 0.1407,
+      "step": 6795
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016546195152737882,
+      "loss": 0.0723,
+      "step": 6796
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016545133579356044,
+      "loss": 0.1051,
+      "step": 6797
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016544071876920703,
+      "loss": 0.0807,
+      "step": 6798
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016543010045452786,
+      "loss": 0.0792,
+      "step": 6799
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001654194808497323,
+      "loss": 0.1105,
+      "step": 6800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.001654088599550298,
+      "loss": 0.0784,
+      "step": 6801
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016539823777062972,
+      "loss": 0.1057,
+      "step": 6802
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016538761429674151,
+      "loss": 0.0665,
+      "step": 6803
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0016537698953357465,
+      "loss": 0.08,
+      "step": 6804
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016536636348133866,
+      "loss": 0.0748,
+      "step": 6805
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00165355736140243,
+      "loss": 0.0917,
+      "step": 6806
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016534510751049726,
+      "loss": 0.1443,
+      "step": 6807
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00165334477592311,
+      "loss": 0.1108,
+      "step": 6808
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001653238463858938,
+      "loss": 0.064,
+      "step": 6809
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016531321389145527,
+      "loss": 0.1391,
+      "step": 6810
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001653025801092051,
+      "loss": 0.1399,
+      "step": 6811
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016529194503935294,
+      "loss": 0.0841,
+      "step": 6812
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016528130868210845,
+      "loss": 0.0576,
+      "step": 6813
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016527067103768141,
+      "loss": 0.1345,
+      "step": 6814
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016526003210628156,
+      "loss": 0.0896,
+      "step": 6815
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016524939188811865,
+      "loss": 0.1025,
+      "step": 6816
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016523875038340248,
+      "loss": 0.1111,
+      "step": 6817
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016522810759234288,
+      "loss": 0.079,
+      "step": 6818
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001652174635151497,
+      "loss": 0.1251,
+      "step": 6819
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001652068181520328,
+      "loss": 0.0923,
+      "step": 6820
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016519617150320208,
+      "loss": 0.0881,
+      "step": 6821
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016518552356886752,
+      "loss": 0.0875,
+      "step": 6822
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016517487434923898,
+      "loss": 0.0977,
+      "step": 6823
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001651642238445265,
+      "loss": 0.1293,
+      "step": 6824
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016515357205494007,
+      "loss": 0.1006,
+      "step": 6825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016514291898068972,
+      "loss": 0.0727,
+      "step": 6826
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016513226462198547,
+      "loss": 0.123,
+      "step": 6827
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016512160897903746,
+      "loss": 0.0818,
+      "step": 6828
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016511095205205568,
+      "loss": 0.1254,
+      "step": 6829
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001651002938412504,
+      "loss": 0.0826,
+      "step": 6830
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016508963434683167,
+      "loss": 0.08,
+      "step": 6831
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001650789735690097,
+      "loss": 0.1072,
+      "step": 6832
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016506831150799469,
+      "loss": 0.1021,
+      "step": 6833
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001650576481639969,
+      "loss": 0.1022,
+      "step": 6834
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016504698353722652,
+      "loss": 0.1144,
+      "step": 6835
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016503631762789389,
+      "loss": 0.1072,
+      "step": 6836
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016502565043620928,
+      "loss": 0.1093,
+      "step": 6837
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016501498196238302,
+      "loss": 0.0931,
+      "step": 6838
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001650043122066255,
+      "loss": 0.0721,
+      "step": 6839
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016499364116914705,
+      "loss": 0.1124,
+      "step": 6840
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016498296885015815,
+      "loss": 0.105,
+      "step": 6841
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016497229524986916,
+      "loss": 0.1097,
+      "step": 6842
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016496162036849053,
+      "loss": 0.1036,
+      "step": 6843
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016495094420623283,
+      "loss": 0.0674,
+      "step": 6844
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016494026676330646,
+      "loss": 0.177,
+      "step": 6845
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016492958803992206,
+      "loss": 0.1071,
+      "step": 6846
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.001649189080362901,
+      "loss": 0.0701,
+      "step": 6847
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016490822675262115,
+      "loss": 0.0442,
+      "step": 6848
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016489754418912593,
+      "loss": 0.1067,
+      "step": 6849
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0016488686034601496,
+      "loss": 0.0812,
+      "step": 6850
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016487617522349895,
+      "loss": 0.0972,
+      "step": 6851
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016486548882178858,
+      "loss": 0.0858,
+      "step": 6852
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016485480114109452,
+      "loss": 0.0775,
+      "step": 6853
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016484411218162759,
+      "loss": 0.082,
+      "step": 6854
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016483342194359848,
+      "loss": 0.1235,
+      "step": 6855
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016482273042721799,
+      "loss": 0.0961,
+      "step": 6856
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016481203763269693,
+      "loss": 0.1135,
+      "step": 6857
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016480134356024613,
+      "loss": 0.0947,
+      "step": 6858
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016479064821007645,
+      "loss": 0.1823,
+      "step": 6859
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001647799515823988,
+      "loss": 0.1167,
+      "step": 6860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016476925367742408,
+      "loss": 0.1122,
+      "step": 6861
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016475855449536322,
+      "loss": 0.1217,
+      "step": 6862
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016474785403642715,
+      "loss": 0.1428,
+      "step": 6863
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001647371523008269,
+      "loss": 0.1075,
+      "step": 6864
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016472644928877348,
+      "loss": 0.0873,
+      "step": 6865
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016471574500047789,
+      "loss": 0.1106,
+      "step": 6866
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016470503943615121,
+      "loss": 0.113,
+      "step": 6867
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016469433259600455,
+      "loss": 0.1577,
+      "step": 6868
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00164683624480249,
+      "loss": 0.0851,
+      "step": 6869
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016467291508909568,
+      "loss": 0.0726,
+      "step": 6870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001646622044227558,
+      "loss": 0.0895,
+      "step": 6871
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016465149248144051,
+      "loss": 0.143,
+      "step": 6872
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016464077926536102,
+      "loss": 0.0889,
+      "step": 6873
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001646300647747286,
+      "loss": 0.0735,
+      "step": 6874
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016461934900975448,
+      "loss": 0.0868,
+      "step": 6875
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016460863197064995,
+      "loss": 0.1233,
+      "step": 6876
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016459791365762634,
+      "loss": 0.092,
+      "step": 6877
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00164587194070895,
+      "loss": 0.1096,
+      "step": 6878
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016457647321066725,
+      "loss": 0.0917,
+      "step": 6879
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001645657510771545,
+      "loss": 0.1049,
+      "step": 6880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001645550276705682,
+      "loss": 0.0968,
+      "step": 6881
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016454430299111973,
+      "loss": 0.1161,
+      "step": 6882
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016453357703902056,
+      "loss": 0.0901,
+      "step": 6883
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016452284981448221,
+      "loss": 0.1107,
+      "step": 6884
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001645121213177162,
+      "loss": 0.1179,
+      "step": 6885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016450139154893404,
+      "loss": 0.0994,
+      "step": 6886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001644906605083473,
+      "loss": 0.106,
+      "step": 6887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016447992819616758,
+      "loss": 0.0747,
+      "step": 6888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016446919461260646,
+      "loss": 0.0682,
+      "step": 6889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016445845975787562,
+      "loss": 0.111,
+      "step": 6890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016444772363218671,
+      "loss": 0.1388,
+      "step": 6891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001644369862357514,
+      "loss": 0.0981,
+      "step": 6892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016442624756878143,
+      "loss": 0.0953,
+      "step": 6893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016441550763148856,
+      "loss": 0.0703,
+      "step": 6894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.001644047664240845,
+      "loss": 0.1536,
+      "step": 6895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0016439402394678112,
+      "loss": 0.09,
+      "step": 6896
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016438328019979013,
+      "loss": 0.085,
+      "step": 6897
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001643725351833234,
+      "loss": 0.0863,
+      "step": 6898
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016436178889759289,
+      "loss": 0.1001,
+      "step": 6899
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016435104134281033,
+      "loss": 0.1376,
+      "step": 6900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001643402925191878,
+      "loss": 0.1437,
+      "step": 6901
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016432954242693708,
+      "loss": 0.0939,
+      "step": 6902
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016431879106627025,
+      "loss": 0.0879,
+      "step": 6903
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016430803843739928,
+      "loss": 0.1465,
+      "step": 6904
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016429728454053615,
+      "loss": 0.1572,
+      "step": 6905
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016428652937589292,
+      "loss": 0.118,
+      "step": 6906
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016427577294368166,
+      "loss": 0.1226,
+      "step": 6907
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016426501524411444,
+      "loss": 0.1328,
+      "step": 6908
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001642542562774034,
+      "loss": 0.089,
+      "step": 6909
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016424349604376065,
+      "loss": 0.1259,
+      "step": 6910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001642327345433984,
+      "loss": 0.0776,
+      "step": 6911
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001642219717765288,
+      "loss": 0.1324,
+      "step": 6912
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016421120774336405,
+      "loss": 0.1001,
+      "step": 6913
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016420044244411644,
+      "loss": 0.1298,
+      "step": 6914
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001641896758789982,
+      "loss": 0.1126,
+      "step": 6915
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016417890804822162,
+      "loss": 0.0725,
+      "step": 6916
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016416813895199904,
+      "loss": 0.1465,
+      "step": 6917
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016415736859054278,
+      "loss": 0.0864,
+      "step": 6918
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001641465969640652,
+      "loss": 0.1104,
+      "step": 6919
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001641358240727787,
+      "loss": 0.1031,
+      "step": 6920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001641250499168957,
+      "loss": 0.1405,
+      "step": 6921
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016411427449662863,
+      "loss": 0.0992,
+      "step": 6922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016410349781218994,
+      "loss": 0.1312,
+      "step": 6923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016409271986379214,
+      "loss": 0.106,
+      "step": 6924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016408194065164773,
+      "loss": 0.1008,
+      "step": 6925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016407116017596928,
+      "loss": 0.1113,
+      "step": 6926
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001640603784369693,
+      "loss": 0.1055,
+      "step": 6927
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016404959543486041,
+      "loss": 0.1691,
+      "step": 6928
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016403881116985523,
+      "loss": 0.1099,
+      "step": 6929
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016402802564216637,
+      "loss": 0.0814,
+      "step": 6930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016401723885200654,
+      "loss": 0.084,
+      "step": 6931
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001640064507995884,
+      "loss": 0.182,
+      "step": 6932
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016399566148512467,
+      "loss": 0.1057,
+      "step": 6933
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016398487090882807,
+      "loss": 0.0965,
+      "step": 6934
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016397407907091138,
+      "loss": 0.0997,
+      "step": 6935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016396328597158738,
+      "loss": 0.1154,
+      "step": 6936
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016395249161106889,
+      "loss": 0.1373,
+      "step": 6937
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016394169598956873,
+      "loss": 0.0855,
+      "step": 6938
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.001639308991072998,
+      "loss": 0.104,
+      "step": 6939
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016392010096447495,
+      "loss": 0.1066,
+      "step": 6940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016390930156130713,
+      "loss": 0.1738,
+      "step": 6941
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0016389850089800922,
+      "loss": 0.0681,
+      "step": 6942
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001638876989747942,
+      "loss": 0.1324,
+      "step": 6943
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016387689579187513,
+      "loss": 0.0732,
+      "step": 6944
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016386609134946492,
+      "loss": 0.1318,
+      "step": 6945
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016385528564777665,
+      "loss": 0.0768,
+      "step": 6946
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016384447868702337,
+      "loss": 0.1393,
+      "step": 6947
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016383367046741819,
+      "loss": 0.0956,
+      "step": 6948
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016382286098917422,
+      "loss": 0.1396,
+      "step": 6949
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001638120502525046,
+      "loss": 0.091,
+      "step": 6950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016380123825762243,
+      "loss": 0.0994,
+      "step": 6951
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016379042500474096,
+      "loss": 0.0748,
+      "step": 6952
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016377961049407338,
+      "loss": 0.0905,
+      "step": 6953
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016376879472583287,
+      "loss": 0.0847,
+      "step": 6954
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016375797770023283,
+      "loss": 0.0776,
+      "step": 6955
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016374715941748642,
+      "loss": 0.0764,
+      "step": 6956
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016373633987780693,
+      "loss": 0.0789,
+      "step": 6957
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016372551908140785,
+      "loss": 0.1281,
+      "step": 6958
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016371469702850237,
+      "loss": 0.1312,
+      "step": 6959
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016370387371930396,
+      "loss": 0.0945,
+      "step": 6960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00163693049154026,
+      "loss": 0.0863,
+      "step": 6961
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016368222333288196,
+      "loss": 0.1182,
+      "step": 6962
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016367139625608528,
+      "loss": 0.1367,
+      "step": 6963
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016366056792384938,
+      "loss": 0.1421,
+      "step": 6964
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016364973833638788,
+      "loss": 0.0681,
+      "step": 6965
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016363890749391424,
+      "loss": 0.1006,
+      "step": 6966
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00163628075396642,
+      "loss": 0.1306,
+      "step": 6967
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001636172420447848,
+      "loss": 0.1119,
+      "step": 6968
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001636064074385562,
+      "loss": 0.0946,
+      "step": 6969
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016359557157816986,
+      "loss": 0.1691,
+      "step": 6970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016358473446383944,
+      "loss": 0.1091,
+      "step": 6971
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016357389609577856,
+      "loss": 0.0811,
+      "step": 6972
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016356305647420102,
+      "loss": 0.1247,
+      "step": 6973
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001635522155993205,
+      "loss": 0.1003,
+      "step": 6974
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016354137347135073,
+      "loss": 0.1132,
+      "step": 6975
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016353053009050553,
+      "loss": 0.077,
+      "step": 6976
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016351968545699868,
+      "loss": 0.0714,
+      "step": 6977
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016350883957104406,
+      "loss": 0.1057,
+      "step": 6978
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016349799243285546,
+      "loss": 0.0997,
+      "step": 6979
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016348714404264672,
+      "loss": 0.1016,
+      "step": 6980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016347629440063188,
+      "loss": 0.0912,
+      "step": 6981
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016346544350702478,
+      "loss": 0.0677,
+      "step": 6982
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016345459136203934,
+      "loss": 0.1585,
+      "step": 6983
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001634437379658896,
+      "loss": 0.1177,
+      "step": 6984
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016343288331878956,
+      "loss": 0.0812,
+      "step": 6985
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.001634220274209532,
+      "loss": 0.1404,
+      "step": 6986
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016341117027259462,
+      "loss": 0.0791,
+      "step": 6987
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0016340031187392786,
+      "loss": 0.1584,
+      "step": 6988
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016338945222516706,
+      "loss": 0.1161,
+      "step": 6989
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016337859132652627,
+      "loss": 0.1188,
+      "step": 6990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016336772917821972,
+      "loss": 0.1008,
+      "step": 6991
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016335686578046156,
+      "loss": 0.1552,
+      "step": 6992
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016334600113346593,
+      "loss": 0.0918,
+      "step": 6993
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016333513523744717,
+      "loss": 0.1661,
+      "step": 6994
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016332426809261944,
+      "loss": 0.0784,
+      "step": 6995
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016331339969919703,
+      "loss": 0.1165,
+      "step": 6996
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016330253005739424,
+      "loss": 0.1041,
+      "step": 6997
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016329165916742542,
+      "loss": 0.1781,
+      "step": 6998
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016328078702950487,
+      "loss": 0.1342,
+      "step": 6999
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016326991364384695,
+      "loss": 0.0757,
+      "step": 7000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016325903901066613,
+      "loss": 0.1129,
+      "step": 7001
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001632481631301768,
+      "loss": 0.137,
+      "step": 7002
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016323728600259335,
+      "loss": 0.1573,
+      "step": 7003
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016322640762813033,
+      "loss": 0.0673,
+      "step": 7004
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016321552800700218,
+      "loss": 0.0844,
+      "step": 7005
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016320464713942343,
+      "loss": 0.1096,
+      "step": 7006
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016319376502560864,
+      "loss": 0.0883,
+      "step": 7007
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016318288166577237,
+      "loss": 0.0744,
+      "step": 7008
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001631719970601292,
+      "loss": 0.1451,
+      "step": 7009
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016316111120889373,
+      "loss": 0.0826,
+      "step": 7010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016315022411228068,
+      "loss": 0.1224,
+      "step": 7011
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016313933577050462,
+      "loss": 0.0872,
+      "step": 7012
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001631284461837803,
+      "loss": 0.1038,
+      "step": 7013
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016311755535232241,
+      "loss": 0.112,
+      "step": 7014
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016310666327634573,
+      "loss": 0.0842,
+      "step": 7015
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016309576995606498,
+      "loss": 0.09,
+      "step": 7016
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016308487539169493,
+      "loss": 0.0787,
+      "step": 7017
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016307397958345046,
+      "loss": 0.123,
+      "step": 7018
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016306308253154636,
+      "loss": 0.098,
+      "step": 7019
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016305218423619747,
+      "loss": 0.0933,
+      "step": 7020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016304128469761877,
+      "loss": 0.101,
+      "step": 7021
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001630303839160251,
+      "loss": 0.1005,
+      "step": 7022
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016301948189163138,
+      "loss": 0.124,
+      "step": 7023
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016300857862465263,
+      "loss": 0.0953,
+      "step": 7024
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016299767411530382,
+      "loss": 0.1094,
+      "step": 7025
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016298676836379992,
+      "loss": 0.1622,
+      "step": 7026
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016297586137035598,
+      "loss": 0.0853,
+      "step": 7027
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016296495313518706,
+      "loss": 0.1122,
+      "step": 7028
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016295404365850828,
+      "loss": 0.0895,
+      "step": 7029
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001629431329405347,
+      "loss": 0.085,
+      "step": 7030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016293222098148145,
+      "loss": 0.1433,
+      "step": 7031
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0016292130778156374,
+      "loss": 0.0862,
+      "step": 7032
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001629103933409967,
+      "loss": 0.1379,
+      "step": 7033
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.001628994776599955,
+      "loss": 0.1287,
+      "step": 7034
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001628885607387755,
+      "loss": 0.1129,
+      "step": 7035
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001628776425775518,
+      "loss": 0.1,
+      "step": 7036
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016286672317653982,
+      "loss": 0.1138,
+      "step": 7037
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016285580253595478,
+      "loss": 0.1198,
+      "step": 7038
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016284488065601201,
+      "loss": 0.1123,
+      "step": 7039
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016283395753692686,
+      "loss": 0.0914,
+      "step": 7040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016282303317891476,
+      "loss": 0.1292,
+      "step": 7041
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016281210758219106,
+      "loss": 0.1105,
+      "step": 7042
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001628011807469712,
+      "loss": 0.1165,
+      "step": 7043
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016279025267347058,
+      "loss": 0.1238,
+      "step": 7044
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016277932336190477,
+      "loss": 0.1537,
+      "step": 7045
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001627683928124892,
+      "loss": 0.1003,
+      "step": 7046
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016275746102543942,
+      "loss": 0.0946,
+      "step": 7047
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00162746528000971,
+      "loss": 0.2115,
+      "step": 7048
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016273559373929944,
+      "loss": 0.142,
+      "step": 7049
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016272465824064039,
+      "loss": 0.0956,
+      "step": 7050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016271372150520945,
+      "loss": 0.094,
+      "step": 7051
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016270278353322227,
+      "loss": 0.1636,
+      "step": 7052
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016269184432489451,
+      "loss": 0.113,
+      "step": 7053
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016268090388044192,
+      "loss": 0.0842,
+      "step": 7054
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016266996220008015,
+      "loss": 0.1256,
+      "step": 7055
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016265901928402497,
+      "loss": 0.1069,
+      "step": 7056
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016264807513249214,
+      "loss": 0.0862,
+      "step": 7057
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016263712974569744,
+      "loss": 0.1185,
+      "step": 7058
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016262618312385669,
+      "loss": 0.0853,
+      "step": 7059
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016261523526718575,
+      "loss": 0.1583,
+      "step": 7060
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016260428617590052,
+      "loss": 0.0836,
+      "step": 7061
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016259333585021678,
+      "loss": 0.0824,
+      "step": 7062
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001625823842903505,
+      "loss": 0.1221,
+      "step": 7063
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016257143149651764,
+      "loss": 0.1292,
+      "step": 7064
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016256047746893417,
+      "loss": 0.1692,
+      "step": 7065
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016254952220781604,
+      "loss": 0.0722,
+      "step": 7066
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016253856571337923,
+      "loss": 0.1169,
+      "step": 7067
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016252760798583983,
+      "loss": 0.1101,
+      "step": 7068
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001625166490254139,
+      "loss": 0.1191,
+      "step": 7069
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001625056888323175,
+      "loss": 0.0933,
+      "step": 7070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001624947274067667,
+      "loss": 0.0623,
+      "step": 7071
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016248376474897771,
+      "loss": 0.1676,
+      "step": 7072
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016247280085916664,
+      "loss": 0.1669,
+      "step": 7073
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016246183573754968,
+      "loss": 0.1041,
+      "step": 7074
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016245086938434306,
+      "loss": 0.0799,
+      "step": 7075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016243990179976293,
+      "loss": 0.1235,
+      "step": 7076
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016242893298402564,
+      "loss": 0.123,
+      "step": 7077
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001624179629373474,
+      "loss": 0.1011,
+      "step": 7078
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0016240699165994458,
+      "loss": 0.0869,
+      "step": 7079
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.001623960191520334,
+      "loss": 0.1389,
+      "step": 7080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016238504541383027,
+      "loss": 0.1448,
+      "step": 7081
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.001623740704455516,
+      "loss": 0.0944,
+      "step": 7082
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016236309424741374,
+      "loss": 0.1042,
+      "step": 7083
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016235211681963314,
+      "loss": 0.0908,
+      "step": 7084
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016234113816242623,
+      "loss": 0.1232,
+      "step": 7085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016233015827600945,
+      "loss": 0.0651,
+      "step": 7086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016231917716059937,
+      "loss": 0.0771,
+      "step": 7087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016230819481641246,
+      "loss": 0.104,
+      "step": 7088
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016229721124366527,
+      "loss": 0.0845,
+      "step": 7089
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016228622644257438,
+      "loss": 0.0917,
+      "step": 7090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016227524041335637,
+      "loss": 0.1303,
+      "step": 7091
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016226425315622789,
+      "loss": 0.101,
+      "step": 7092
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.001622532646714055,
+      "loss": 0.1835,
+      "step": 7093
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016224227495910597,
+      "loss": 0.0842,
+      "step": 7094
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016223128401954593,
+      "loss": 0.0727,
+      "step": 7095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016222029185294206,
+      "loss": 0.106,
+      "step": 7096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016220929845951115,
+      "loss": 0.1068,
+      "step": 7097
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016219830383947,
+      "loss": 0.1305,
+      "step": 7098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016218730799303532,
+      "loss": 0.0571,
+      "step": 7099
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016217631092042392,
+      "loss": 0.1251,
+      "step": 7100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.001621653126218527,
+      "loss": 0.1082,
+      "step": 7101
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016215431309753846,
+      "loss": 0.1169,
+      "step": 7102
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016214331234769813,
+      "loss": 0.0988,
+      "step": 7103
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016213231037254857,
+      "loss": 0.1433,
+      "step": 7104
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016212130717230672,
+      "loss": 0.1064,
+      "step": 7105
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016211030274718957,
+      "loss": 0.0861,
+      "step": 7106
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016209929709741406,
+      "loss": 0.0873,
+      "step": 7107
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016208829022319722,
+      "loss": 0.095,
+      "step": 7108
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016207728212475608,
+      "loss": 0.1053,
+      "step": 7109
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016206627280230764,
+      "loss": 0.1012,
+      "step": 7110
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016205526225606906,
+      "loss": 0.1296,
+      "step": 7111
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016204425048625738,
+      "loss": 0.0522,
+      "step": 7112
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016203323749308975,
+      "loss": 0.1053,
+      "step": 7113
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016202222327678327,
+      "loss": 0.0762,
+      "step": 7114
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016201120783755519,
+      "loss": 0.1096,
+      "step": 7115
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016200019117562263,
+      "loss": 0.1211,
+      "step": 7116
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016198917329120288,
+      "loss": 0.1327,
+      "step": 7117
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016197815418451315,
+      "loss": 0.0936,
+      "step": 7118
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016196713385577075,
+      "loss": 0.1216,
+      "step": 7119
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016195611230519287,
+      "loss": 0.1039,
+      "step": 7120
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016194508953299694,
+      "loss": 0.1056,
+      "step": 7121
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016193406553940025,
+      "loss": 0.1066,
+      "step": 7122
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016192304032462013,
+      "loss": 0.1066,
+      "step": 7123
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016191201388887405,
+      "loss": 0.0734,
+      "step": 7124
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016190098623237936,
+      "loss": 0.1062,
+      "step": 7125
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0016188995735535352,
+      "loss": 0.1163,
+      "step": 7126
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00161878927258014,
+      "loss": 0.0976,
+      "step": 7127
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016186789594057828,
+      "loss": 0.1168,
+      "step": 7128
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016185686340326386,
+      "loss": 0.1216,
+      "step": 7129
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016184582964628826,
+      "loss": 0.171,
+      "step": 7130
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001618347946698691,
+      "loss": 0.1067,
+      "step": 7131
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016182375847422392,
+      "loss": 0.1648,
+      "step": 7132
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001618127210595703,
+      "loss": 0.114,
+      "step": 7133
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001618016824261259,
+      "loss": 0.1227,
+      "step": 7134
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016179064257410838,
+      "loss": 0.092,
+      "step": 7135
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016177960150373539,
+      "loss": 0.1204,
+      "step": 7136
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016176855921522467,
+      "loss": 0.0936,
+      "step": 7137
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016175751570879394,
+      "loss": 0.1508,
+      "step": 7138
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001617464709846609,
+      "loss": 0.1377,
+      "step": 7139
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016173542504304337,
+      "loss": 0.1442,
+      "step": 7140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016172437788415915,
+      "loss": 0.0858,
+      "step": 7141
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016171332950822607,
+      "loss": 0.0818,
+      "step": 7142
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016170227991546189,
+      "loss": 0.177,
+      "step": 7143
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016169122910608463,
+      "loss": 0.0862,
+      "step": 7144
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016168017708031205,
+      "loss": 0.0836,
+      "step": 7145
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016166912383836212,
+      "loss": 0.1199,
+      "step": 7146
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001616580693804528,
+      "loss": 0.1268,
+      "step": 7147
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016164701370680202,
+      "loss": 0.1173,
+      "step": 7148
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001616359568176278,
+      "loss": 0.0955,
+      "step": 7149
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016162489871314812,
+      "loss": 0.0856,
+      "step": 7150
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016161383939358104,
+      "loss": 0.1152,
+      "step": 7151
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016160277885914466,
+      "loss": 0.1044,
+      "step": 7152
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016159171711005698,
+      "loss": 0.0876,
+      "step": 7153
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001615806541465362,
+      "loss": 0.062,
+      "step": 7154
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001615695899688004,
+      "loss": 0.1082,
+      "step": 7155
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001615585245770677,
+      "loss": 0.0731,
+      "step": 7156
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001615474579715564,
+      "loss": 0.1569,
+      "step": 7157
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016153639015248457,
+      "loss": 0.064,
+      "step": 7158
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016152532112007055,
+      "loss": 0.0942,
+      "step": 7159
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016151425087453251,
+      "loss": 0.097,
+      "step": 7160
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001615031794160888,
+      "loss": 0.1213,
+      "step": 7161
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016149210674495766,
+      "loss": 0.0801,
+      "step": 7162
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016148103286135744,
+      "loss": 0.1077,
+      "step": 7163
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001614699577655065,
+      "loss": 0.1097,
+      "step": 7164
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016145888145762321,
+      "loss": 0.0919,
+      "step": 7165
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016144780393792595,
+      "loss": 0.0674,
+      "step": 7166
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016143672520663316,
+      "loss": 0.1034,
+      "step": 7167
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016142564526396327,
+      "loss": 0.1417,
+      "step": 7168
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016141456411013476,
+      "loss": 0.1134,
+      "step": 7169
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001614034817453661,
+      "loss": 0.1396,
+      "step": 7170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001613923981698758,
+      "loss": 0.0762,
+      "step": 7171
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.001613813133838825,
+      "loss": 0.1389,
+      "step": 7172
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0016137022738760463,
+      "loss": 0.0881,
+      "step": 7173
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016135914018126083,
+      "loss": 0.0995,
+      "step": 7174
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016134805176506975,
+      "loss": 0.1456,
+      "step": 7175
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016133696213924992,
+      "loss": 0.1318,
+      "step": 7176
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016132587130402018,
+      "loss": 0.118,
+      "step": 7177
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016131477925959902,
+      "loss": 0.1021,
+      "step": 7178
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016130368600620526,
+      "loss": 0.1323,
+      "step": 7179
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001612925915440576,
+      "loss": 0.1033,
+      "step": 7180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001612814958733748,
+      "loss": 0.1213,
+      "step": 7181
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016127039899437564,
+      "loss": 0.1083,
+      "step": 7182
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001612593009072789,
+      "loss": 0.1183,
+      "step": 7183
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016124820161230345,
+      "loss": 0.0983,
+      "step": 7184
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001612371011096681,
+      "loss": 0.1251,
+      "step": 7185
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001612259993995917,
+      "loss": 0.0895,
+      "step": 7186
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016121489648229324,
+      "loss": 0.1108,
+      "step": 7187
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016120379235799156,
+      "loss": 0.1124,
+      "step": 7188
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016119268702690565,
+      "loss": 0.1633,
+      "step": 7189
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016118158048925447,
+      "loss": 0.1526,
+      "step": 7190
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016117047274525698,
+      "loss": 0.1199,
+      "step": 7191
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016115936379513223,
+      "loss": 0.0814,
+      "step": 7192
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016114825363909924,
+      "loss": 0.1235,
+      "step": 7193
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016113714227737712,
+      "loss": 0.1234,
+      "step": 7194
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001611260297101849,
+      "loss": 0.0842,
+      "step": 7195
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001611149159377417,
+      "loss": 0.0893,
+      "step": 7196
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.001611038009602667,
+      "loss": 0.1095,
+      "step": 7197
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016109268477797903,
+      "loss": 0.1824,
+      "step": 7198
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016108156739109787,
+      "loss": 0.0638,
+      "step": 7199
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016107044879984241,
+      "loss": 0.0706,
+      "step": 7200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016105932900443193,
+      "loss": 0.1113,
+      "step": 7201
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016104820800508565,
+      "loss": 0.1113,
+      "step": 7202
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016103708580202283,
+      "loss": 0.0849,
+      "step": 7203
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016102596239546282,
+      "loss": 0.0869,
+      "step": 7204
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016101483778562492,
+      "loss": 0.1294,
+      "step": 7205
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016100371197272842,
+      "loss": 0.1368,
+      "step": 7206
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016099258495699282,
+      "loss": 0.0743,
+      "step": 7207
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016098145673863744,
+      "loss": 0.101,
+      "step": 7208
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016097032731788171,
+      "loss": 0.124,
+      "step": 7209
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016095919669494505,
+      "loss": 0.106,
+      "step": 7210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016094806487004698,
+      "loss": 0.066,
+      "step": 7211
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016093693184340695,
+      "loss": 0.1098,
+      "step": 7212
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016092579761524447,
+      "loss": 0.0994,
+      "step": 7213
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016091466218577913,
+      "loss": 0.1478,
+      "step": 7214
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016090352555523046,
+      "loss": 0.1338,
+      "step": 7215
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016089238772381803,
+      "loss": 0.155,
+      "step": 7216
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016088124869176146,
+      "loss": 0.1121,
+      "step": 7217
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016087010845928038,
+      "loss": 0.1367,
+      "step": 7218
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0016085896702659449,
+      "loss": 0.1111,
+      "step": 7219
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001608478243939234,
+      "loss": 0.1283,
+      "step": 7220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016083668056148688,
+      "loss": 0.1282,
+      "step": 7221
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016082553552950463,
+      "loss": 0.1205,
+      "step": 7222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016081438929819638,
+      "loss": 0.1102,
+      "step": 7223
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016080324186778198,
+      "loss": 0.1,
+      "step": 7224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016079209323848112,
+      "loss": 0.1099,
+      "step": 7225
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016078094341051368,
+      "loss": 0.0986,
+      "step": 7226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016076979238409954,
+      "loss": 0.0472,
+      "step": 7227
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016075864015945853,
+      "loss": 0.1044,
+      "step": 7228
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016074748673681055,
+      "loss": 0.1255,
+      "step": 7229
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001607363321163755,
+      "loss": 0.1229,
+      "step": 7230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016072517629837336,
+      "loss": 0.1031,
+      "step": 7231
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016071401928302404,
+      "loss": 0.0828,
+      "step": 7232
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001607028610705476,
+      "loss": 0.0746,
+      "step": 7233
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016069170166116398,
+      "loss": 0.0914,
+      "step": 7234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016068054105509329,
+      "loss": 0.11,
+      "step": 7235
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001606693792525555,
+      "loss": 0.1171,
+      "step": 7236
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016065821625377077,
+      "loss": 0.0928,
+      "step": 7237
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016064705205895917,
+      "loss": 0.0999,
+      "step": 7238
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016063588666834084,
+      "loss": 0.1237,
+      "step": 7239
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016062472008213592,
+      "loss": 0.0781,
+      "step": 7240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001606135523005646,
+      "loss": 0.1096,
+      "step": 7241
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016060238332384707,
+      "loss": 0.0705,
+      "step": 7242
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016059121315220356,
+      "loss": 0.1108,
+      "step": 7243
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016058004178585432,
+      "loss": 0.0717,
+      "step": 7244
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016056886922501963,
+      "loss": 0.0807,
+      "step": 7245
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016055769546991976,
+      "loss": 0.1003,
+      "step": 7246
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016054652052077505,
+      "loss": 0.1028,
+      "step": 7247
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016053534437780583,
+      "loss": 0.1292,
+      "step": 7248
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016052416704123247,
+      "loss": 0.1022,
+      "step": 7249
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016051298851127536,
+      "loss": 0.0903,
+      "step": 7250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001605018087881549,
+      "loss": 0.1122,
+      "step": 7251
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016049062787209154,
+      "loss": 0.0981,
+      "step": 7252
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016047944576330574,
+      "loss": 0.0791,
+      "step": 7253
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00160468262462018,
+      "loss": 0.0702,
+      "step": 7254
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016045707796844878,
+      "loss": 0.2101,
+      "step": 7255
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016044589228281863,
+      "loss": 0.1106,
+      "step": 7256
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001604347054053481,
+      "loss": 0.1484,
+      "step": 7257
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016042351733625783,
+      "loss": 0.0884,
+      "step": 7258
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001604123280757683,
+      "loss": 0.1044,
+      "step": 7259
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016040113762410021,
+      "loss": 0.0841,
+      "step": 7260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016038994598147425,
+      "loss": 0.141,
+      "step": 7261
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00160378753148111,
+      "loss": 0.0907,
+      "step": 7262
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.001603675591242312,
+      "loss": 0.1117,
+      "step": 7263
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016035636391005554,
+      "loss": 0.101,
+      "step": 7264
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0016034516750580484,
+      "loss": 0.0834,
+      "step": 7265
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016033396991169972,
+      "loss": 0.1447,
+      "step": 7266
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016032277112796112,
+      "loss": 0.1205,
+      "step": 7267
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016031157115480976,
+      "loss": 0.0625,
+      "step": 7268
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016030036999246653,
+      "loss": 0.2107,
+      "step": 7269
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016028916764115224,
+      "loss": 0.0734,
+      "step": 7270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001602779641010878,
+      "loss": 0.123,
+      "step": 7271
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001602667593724941,
+      "loss": 0.1348,
+      "step": 7272
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016025555345559207,
+      "loss": 0.0928,
+      "step": 7273
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001602443463506027,
+      "loss": 0.0916,
+      "step": 7274
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001602331380577469,
+      "loss": 0.0703,
+      "step": 7275
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016022192857724571,
+      "loss": 0.1239,
+      "step": 7276
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016021071790932018,
+      "loss": 0.1207,
+      "step": 7277
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001601995060541913,
+      "loss": 0.0817,
+      "step": 7278
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016018829301208016,
+      "loss": 0.1069,
+      "step": 7279
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016017707878320784,
+      "loss": 0.0862,
+      "step": 7280
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001601658633677955,
+      "loss": 0.1144,
+      "step": 7281
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016015464676606425,
+      "loss": 0.1302,
+      "step": 7282
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001601434289782352,
+      "loss": 0.068,
+      "step": 7283
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016013221000452966,
+      "loss": 0.0709,
+      "step": 7284
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016012098984516875,
+      "loss": 0.0946,
+      "step": 7285
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016010976850037372,
+      "loss": 0.0849,
+      "step": 7286
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016009854597036585,
+      "loss": 0.1187,
+      "step": 7287
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001600873222553664,
+      "loss": 0.0996,
+      "step": 7288
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016007609735559664,
+      "loss": 0.1035,
+      "step": 7289
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016006487127127795,
+      "loss": 0.1596,
+      "step": 7290
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016005364400263167,
+      "loss": 0.1049,
+      "step": 7291
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016004241554987916,
+      "loss": 0.0872,
+      "step": 7292
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001600311859132418,
+      "loss": 0.1124,
+      "step": 7293
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016001995509294104,
+      "loss": 0.1057,
+      "step": 7294
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0016000872308919835,
+      "loss": 0.0995,
+      "step": 7295
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015999748990223512,
+      "loss": 0.1245,
+      "step": 7296
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001599862555322729,
+      "loss": 0.1384,
+      "step": 7297
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015997501997953316,
+      "loss": 0.1079,
+      "step": 7298
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015996378324423751,
+      "loss": 0.0949,
+      "step": 7299
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001599525453266074,
+      "loss": 0.1305,
+      "step": 7300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015994130622686452,
+      "loss": 0.0967,
+      "step": 7301
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015993006594523043,
+      "loss": 0.1364,
+      "step": 7302
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015991882448192674,
+      "loss": 0.0891,
+      "step": 7303
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015990758183717514,
+      "loss": 0.1477,
+      "step": 7304
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.001598963380111973,
+      "loss": 0.1304,
+      "step": 7305
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015988509300421487,
+      "loss": 0.141,
+      "step": 7306
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015987384681644965,
+      "loss": 0.1159,
+      "step": 7307
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015986259944812335,
+      "loss": 0.1066,
+      "step": 7308
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015985135089945774,
+      "loss": 0.1128,
+      "step": 7309
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015984010117067458,
+      "loss": 0.1075,
+      "step": 7310
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0015982885026199574,
+      "loss": 0.0981,
+      "step": 7311
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015981759817364307,
+      "loss": 0.0573,
+      "step": 7312
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015980634490583836,
+      "loss": 0.1196,
+      "step": 7313
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015979509045880357,
+      "loss": 0.0845,
+      "step": 7314
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015978383483276056,
+      "loss": 0.0712,
+      "step": 7315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015977257802793127,
+      "loss": 0.0857,
+      "step": 7316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015976132004453768,
+      "loss": 0.0725,
+      "step": 7317
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015975006088280175,
+      "loss": 0.1058,
+      "step": 7318
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001597388005429455,
+      "loss": 0.1394,
+      "step": 7319
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015972753902519093,
+      "loss": 0.088,
+      "step": 7320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001597162763297601,
+      "loss": 0.1189,
+      "step": 7321
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015970501245687507,
+      "loss": 0.1229,
+      "step": 7322
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015969374740675796,
+      "loss": 0.0831,
+      "step": 7323
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001596824811796309,
+      "loss": 0.0604,
+      "step": 7324
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015967121377571595,
+      "loss": 0.0776,
+      "step": 7325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015965994519523538,
+      "loss": 0.1201,
+      "step": 7326
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015964867543841134,
+      "loss": 0.1049,
+      "step": 7327
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00159637404505466,
+      "loss": 0.0853,
+      "step": 7328
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015962613239662163,
+      "loss": 0.0747,
+      "step": 7329
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001596148591121005,
+      "loss": 0.0782,
+      "step": 7330
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015960358465212487,
+      "loss": 0.0752,
+      "step": 7331
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015959230901691703,
+      "loss": 0.0864,
+      "step": 7332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015958103220669935,
+      "loss": 0.15,
+      "step": 7333
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015956975422169416,
+      "loss": 0.129,
+      "step": 7334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015955847506212383,
+      "loss": 0.0775,
+      "step": 7335
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015954719472821071,
+      "loss": 0.1071,
+      "step": 7336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001595359132201773,
+      "loss": 0.0992,
+      "step": 7337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00159524630538246,
+      "loss": 0.1003,
+      "step": 7338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015951334668263927,
+      "loss": 0.1044,
+      "step": 7339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015950206165357963,
+      "loss": 0.1099,
+      "step": 7340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015949077545128957,
+      "loss": 0.0952,
+      "step": 7341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015947948807599165,
+      "loss": 0.092,
+      "step": 7342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001594681995279084,
+      "loss": 0.1017,
+      "step": 7343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001594569098072624,
+      "loss": 0.1279,
+      "step": 7344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015944561891427628,
+      "loss": 0.1332,
+      "step": 7345
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015943432684917263,
+      "loss": 0.1394,
+      "step": 7346
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015942303361217414,
+      "loss": 0.0774,
+      "step": 7347
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001594117392035035,
+      "loss": 0.0812,
+      "step": 7348
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015940044362338331,
+      "loss": 0.0715,
+      "step": 7349
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001593891468720364,
+      "loss": 0.0662,
+      "step": 7350
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015937784894968546,
+      "loss": 0.0703,
+      "step": 7351
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015936654985655328,
+      "loss": 0.0825,
+      "step": 7352
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001593552495928626,
+      "loss": 0.16,
+      "step": 7353
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001593439481588363,
+      "loss": 0.1049,
+      "step": 7354
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.001593326455546972,
+      "loss": 0.0667,
+      "step": 7355
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015932134178066808,
+      "loss": 0.0851,
+      "step": 7356
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0015931003683697193,
+      "loss": 0.0942,
+      "step": 7357
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001592987307238316,
+      "loss": 0.0504,
+      "step": 7358
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015928742344147005,
+      "loss": 0.1387,
+      "step": 7359
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015927611499011018,
+      "loss": 0.0661,
+      "step": 7360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00159264805369975,
+      "loss": 0.104,
+      "step": 7361
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015925349458128753,
+      "loss": 0.0637,
+      "step": 7362
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001592421826242707,
+      "loss": 0.1519,
+      "step": 7363
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015923086949914769,
+      "loss": 0.2049,
+      "step": 7364
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015921955520614142,
+      "loss": 0.131,
+      "step": 7365
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015920823974547508,
+      "loss": 0.1235,
+      "step": 7366
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015919692311737177,
+      "loss": 0.1166,
+      "step": 7367
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015918560532205458,
+      "loss": 0.1056,
+      "step": 7368
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001591742863597467,
+      "loss": 0.1011,
+      "step": 7369
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001591629662306713,
+      "loss": 0.1556,
+      "step": 7370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015915164493505158,
+      "loss": 0.0906,
+      "step": 7371
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015914032247311082,
+      "loss": 0.0856,
+      "step": 7372
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001591289988450722,
+      "loss": 0.1553,
+      "step": 7373
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015911767405115903,
+      "loss": 0.1414,
+      "step": 7374
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015910634809159462,
+      "loss": 0.0762,
+      "step": 7375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015909502096660222,
+      "loss": 0.187,
+      "step": 7376
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015908369267640526,
+      "loss": 0.1135,
+      "step": 7377
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015907236322122702,
+      "loss": 0.1377,
+      "step": 7378
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015906103260129097,
+      "loss": 0.1013,
+      "step": 7379
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015904970081682048,
+      "loss": 0.094,
+      "step": 7380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015903836786803898,
+      "loss": 0.0743,
+      "step": 7381
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015902703375516997,
+      "loss": 0.0853,
+      "step": 7382
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015901569847843685,
+      "loss": 0.0709,
+      "step": 7383
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015900436203806318,
+      "loss": 0.0985,
+      "step": 7384
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015899302443427251,
+      "loss": 0.0723,
+      "step": 7385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001589816856672883,
+      "loss": 0.0736,
+      "step": 7386
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001589703457373342,
+      "loss": 0.122,
+      "step": 7387
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001589590046446338,
+      "loss": 0.0885,
+      "step": 7388
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015894766238941066,
+      "loss": 0.0837,
+      "step": 7389
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001589363189718885,
+      "loss": 0.0958,
+      "step": 7390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015892497439229092,
+      "loss": 0.1125,
+      "step": 7391
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015891362865084165,
+      "loss": 0.0931,
+      "step": 7392
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015890228174776434,
+      "loss": 0.0953,
+      "step": 7393
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001588909336832828,
+      "loss": 0.0933,
+      "step": 7394
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001588795844576207,
+      "loss": 0.1356,
+      "step": 7395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001588682340710019,
+      "loss": 0.0896,
+      "step": 7396
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015885688252365014,
+      "loss": 0.0588,
+      "step": 7397
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001588455298157893,
+      "loss": 0.124,
+      "step": 7398
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015883417594764317,
+      "loss": 0.0721,
+      "step": 7399
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015882282091943564,
+      "loss": 0.0779,
+      "step": 7400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015881146473139065,
+      "loss": 0.0967,
+      "step": 7401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0015880010738373205,
+      "loss": 0.0931,
+      "step": 7402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.001587887488766838,
+      "loss": 0.1077,
+      "step": 7403
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015877738921046985,
+      "loss": 0.0797,
+      "step": 7404
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001587660283853142,
+      "loss": 0.1115,
+      "step": 7405
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015875466640144086,
+      "loss": 0.0681,
+      "step": 7406
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015874330325907386,
+      "loss": 0.137,
+      "step": 7407
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015873193895843724,
+      "loss": 0.0553,
+      "step": 7408
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015872057349975506,
+      "loss": 0.0832,
+      "step": 7409
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015870920688325146,
+      "loss": 0.0702,
+      "step": 7410
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015869783910915055,
+      "loss": 0.1289,
+      "step": 7411
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015868647017767646,
+      "loss": 0.0782,
+      "step": 7412
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015867510008905333,
+      "loss": 0.1261,
+      "step": 7413
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001586637288435054,
+      "loss": 0.0726,
+      "step": 7414
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015865235644125686,
+      "loss": 0.1013,
+      "step": 7415
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015864098288253194,
+      "loss": 0.0871,
+      "step": 7416
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015862960816755492,
+      "loss": 0.0912,
+      "step": 7417
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015861823229655005,
+      "loss": 0.0905,
+      "step": 7418
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015860685526974164,
+      "loss": 0.0646,
+      "step": 7419
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015859547708735407,
+      "loss": 0.1003,
+      "step": 7420
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001585840977496116,
+      "loss": 0.0894,
+      "step": 7421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015857271725673865,
+      "loss": 0.1061,
+      "step": 7422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015856133560895962,
+      "loss": 0.0543,
+      "step": 7423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001585499528064989,
+      "loss": 0.1516,
+      "step": 7424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00158538568849581,
+      "loss": 0.0818,
+      "step": 7425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015852718373843028,
+      "loss": 0.0832,
+      "step": 7426
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001585157974732713,
+      "loss": 0.0984,
+      "step": 7427
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015850441005432854,
+      "loss": 0.0881,
+      "step": 7428
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015849302148182654,
+      "loss": 0.0582,
+      "step": 7429
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015848163175598983,
+      "loss": 0.1276,
+      "step": 7430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015847024087704302,
+      "loss": 0.2887,
+      "step": 7431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001584588488452107,
+      "loss": 0.0789,
+      "step": 7432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015844745566071748,
+      "loss": 0.0897,
+      "step": 7433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015843606132378801,
+      "loss": 0.0909,
+      "step": 7434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015842466583464697,
+      "loss": 0.1748,
+      "step": 7435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015841326919351904,
+      "loss": 0.1066,
+      "step": 7436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015840187140062892,
+      "loss": 0.1102,
+      "step": 7437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015839047245620138,
+      "loss": 0.1124,
+      "step": 7438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015837907236046114,
+      "loss": 0.082,
+      "step": 7439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00158367671113633,
+      "loss": 0.1016,
+      "step": 7440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015835626871594179,
+      "loss": 0.105,
+      "step": 7441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015834486516761227,
+      "loss": 0.1263,
+      "step": 7442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015833346046886935,
+      "loss": 0.1143,
+      "step": 7443
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015832205461993787,
+      "loss": 0.0773,
+      "step": 7444
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015831064762104276,
+      "loss": 0.0942,
+      "step": 7445
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015829923947240886,
+      "loss": 0.1262,
+      "step": 7446
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.001582878301742612,
+      "loss": 0.1541,
+      "step": 7447
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015827641972682473,
+      "loss": 0.0739,
+      "step": 7448
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0015826500813032433,
+      "loss": 0.0775,
+      "step": 7449
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015825359538498517,
+      "loss": 0.0837,
+      "step": 7450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015824218149103216,
+      "loss": 0.0999,
+      "step": 7451
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001582307664486904,
+      "loss": 0.1417,
+      "step": 7452
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015821935025818492,
+      "loss": 0.0975,
+      "step": 7453
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001582079329197409,
+      "loss": 0.1365,
+      "step": 7454
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001581965144335834,
+      "loss": 0.0876,
+      "step": 7455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015818509479993757,
+      "loss": 0.1128,
+      "step": 7456
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001581736740190286,
+      "loss": 0.0854,
+      "step": 7457
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015816225209108166,
+      "loss": 0.1245,
+      "step": 7458
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015815082901632195,
+      "loss": 0.0898,
+      "step": 7459
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015813940479497478,
+      "loss": 0.0937,
+      "step": 7460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001581279794272653,
+      "loss": 0.1382,
+      "step": 7461
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015811655291341883,
+      "loss": 0.0793,
+      "step": 7462
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015810512525366073,
+      "loss": 0.0919,
+      "step": 7463
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015809369644821621,
+      "loss": 0.0715,
+      "step": 7464
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015808226649731071,
+      "loss": 0.0941,
+      "step": 7465
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001580708354011696,
+      "loss": 0.1083,
+      "step": 7466
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015805940316001822,
+      "loss": 0.144,
+      "step": 7467
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00158047969774082,
+      "loss": 0.0822,
+      "step": 7468
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015803653524358637,
+      "loss": 0.1135,
+      "step": 7469
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015802509956875683,
+      "loss": 0.1028,
+      "step": 7470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015801366274981884,
+      "loss": 0.1431,
+      "step": 7471
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001580022247869979,
+      "loss": 0.0869,
+      "step": 7472
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015799078568051951,
+      "loss": 0.1227,
+      "step": 7473
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001579793454306093,
+      "loss": 0.0719,
+      "step": 7474
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015796790403749279,
+      "loss": 0.1384,
+      "step": 7475
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015795646150139554,
+      "loss": 0.0881,
+      "step": 7476
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015794501782254325,
+      "loss": 0.1274,
+      "step": 7477
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001579335730011615,
+      "loss": 0.0911,
+      "step": 7478
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015792212703747597,
+      "loss": 0.0964,
+      "step": 7479
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015791067993171236,
+      "loss": 0.0804,
+      "step": 7480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015789923168409636,
+      "loss": 0.1381,
+      "step": 7481
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015788778229485368,
+      "loss": 0.0988,
+      "step": 7482
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015787633176421012,
+      "loss": 0.1267,
+      "step": 7483
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015786488009239144,
+      "loss": 0.0828,
+      "step": 7484
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001578534272796234,
+      "loss": 0.0779,
+      "step": 7485
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.001578419733261319,
+      "loss": 0.0884,
+      "step": 7486
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015783051823214272,
+      "loss": 0.1207,
+      "step": 7487
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015781906199788171,
+      "loss": 0.0933,
+      "step": 7488
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015780760462357482,
+      "loss": 0.0812,
+      "step": 7489
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015779614610944793,
+      "loss": 0.0818,
+      "step": 7490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015778468645572695,
+      "loss": 0.1224,
+      "step": 7491
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015777322566263787,
+      "loss": 0.1089,
+      "step": 7492
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015776176373040664,
+      "loss": 0.0937,
+      "step": 7493
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015775030065925933,
+      "loss": 0.0773,
+      "step": 7494
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015773883644942185,
+      "loss": 0.1276,
+      "step": 7495
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0015772737110112032,
+      "loss": 0.0864,
+      "step": 7496
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015771590461458082,
+      "loss": 0.0934,
+      "step": 7497
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015770443699002939,
+      "loss": 0.0955,
+      "step": 7498
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015769296822769215,
+      "loss": 0.0956,
+      "step": 7499
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015768149832779527,
+      "loss": 0.1155,
+      "step": 7500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001576700272905649,
+      "loss": 0.1179,
+      "step": 7501
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001576585551162272,
+      "loss": 0.118,
+      "step": 7502
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015764708180500836,
+      "loss": 0.0983,
+      "step": 7503
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015763560735713462,
+      "loss": 0.1112,
+      "step": 7504
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015762413177283227,
+      "loss": 0.1031,
+      "step": 7505
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001576126550523275,
+      "loss": 0.0942,
+      "step": 7506
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015760117719584665,
+      "loss": 0.0879,
+      "step": 7507
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015758969820361605,
+      "loss": 0.1382,
+      "step": 7508
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015757821807586198,
+      "loss": 0.1213,
+      "step": 7509
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015756673681281087,
+      "loss": 0.1003,
+      "step": 7510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015755525441468906,
+      "loss": 0.1162,
+      "step": 7511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015754377088172294,
+      "loss": 0.0967,
+      "step": 7512
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015753228621413896,
+      "loss": 0.082,
+      "step": 7513
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015752080041216356,
+      "loss": 0.085,
+      "step": 7514
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015750931347602321,
+      "loss": 0.0862,
+      "step": 7515
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015749782540594444,
+      "loss": 0.1426,
+      "step": 7516
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001574863362021537,
+      "loss": 0.0832,
+      "step": 7517
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001574748458648776,
+      "loss": 0.1365,
+      "step": 7518
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001574633543943426,
+      "loss": 0.0613,
+      "step": 7519
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015745186179077538,
+      "loss": 0.1282,
+      "step": 7520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015744036805440255,
+      "loss": 0.1083,
+      "step": 7521
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015742887318545063,
+      "loss": 0.0803,
+      "step": 7522
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015741737718414637,
+      "loss": 0.0947,
+      "step": 7523
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015740588005071642,
+      "loss": 0.1227,
+      "step": 7524
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015739438178538742,
+      "loss": 0.1306,
+      "step": 7525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015738288238838615,
+      "loss": 0.0856,
+      "step": 7526
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015737138185993934,
+      "loss": 0.1417,
+      "step": 7527
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015735988020027377,
+      "loss": 0.0867,
+      "step": 7528
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015734837740961614,
+      "loss": 0.1083,
+      "step": 7529
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015733687348819334,
+      "loss": 0.0778,
+      "step": 7530
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015732536843623217,
+      "loss": 0.0863,
+      "step": 7531
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015731386225395947,
+      "loss": 0.0946,
+      "step": 7532
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015730235494160211,
+      "loss": 0.111,
+      "step": 7533
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00157290846499387,
+      "loss": 0.0814,
+      "step": 7534
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015727933692754109,
+      "loss": 0.0916,
+      "step": 7535
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015726782622629127,
+      "loss": 0.064,
+      "step": 7536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001572563143958645,
+      "loss": 0.0951,
+      "step": 7537
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001572448014364878,
+      "loss": 0.0642,
+      "step": 7538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.001572332873483881,
+      "loss": 0.0912,
+      "step": 7539
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015722177213179256,
+      "loss": 0.0687,
+      "step": 7540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015721025578692813,
+      "loss": 0.0813,
+      "step": 7541
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0015719873831402192,
+      "loss": 0.0604,
+      "step": 7542
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00157187219713301,
+      "loss": 0.1297,
+      "step": 7543
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015717569998499252,
+      "loss": 0.0707,
+      "step": 7544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001571641791293236,
+      "loss": 0.1016,
+      "step": 7545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015715265714652139,
+      "loss": 0.1178,
+      "step": 7546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001571411340368131,
+      "loss": 0.1388,
+      "step": 7547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015712960980042592,
+      "loss": 0.0819,
+      "step": 7548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015711808443758708,
+      "loss": 0.1193,
+      "step": 7549
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015710655794852386,
+      "loss": 0.0717,
+      "step": 7550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001570950303334635,
+      "loss": 0.1027,
+      "step": 7551
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015708350159263332,
+      "loss": 0.1011,
+      "step": 7552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015707197172626058,
+      "loss": 0.0914,
+      "step": 7553
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015706044073457268,
+      "loss": 0.0874,
+      "step": 7554
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00157048908617797,
+      "loss": 0.1155,
+      "step": 7555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015703737537616086,
+      "loss": 0.1611,
+      "step": 7556
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001570258410098917,
+      "loss": 0.0851,
+      "step": 7557
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015701430551921694,
+      "loss": 0.1121,
+      "step": 7558
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015700276890436404,
+      "loss": 0.0715,
+      "step": 7559
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015699123116556044,
+      "loss": 0.1035,
+      "step": 7560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015697969230303373,
+      "loss": 0.1414,
+      "step": 7561
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015696815231701132,
+      "loss": 0.1273,
+      "step": 7562
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015695661120772076,
+      "loss": 0.0769,
+      "step": 7563
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015694506897538968,
+      "loss": 0.1146,
+      "step": 7564
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001569335256202456,
+      "loss": 0.0906,
+      "step": 7565
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015692198114251616,
+      "loss": 0.1284,
+      "step": 7566
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015691043554242898,
+      "loss": 0.1194,
+      "step": 7567
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015689888882021172,
+      "loss": 0.1182,
+      "step": 7568
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015688734097609202,
+      "loss": 0.1111,
+      "step": 7569
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015687579201029763,
+      "loss": 0.0815,
+      "step": 7570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001568642419230562,
+      "loss": 0.1228,
+      "step": 7571
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015685269071459551,
+      "loss": 0.1924,
+      "step": 7572
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001568411383851433,
+      "loss": 0.1238,
+      "step": 7573
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015682958493492738,
+      "loss": 0.1134,
+      "step": 7574
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015681803036417552,
+      "loss": 0.1425,
+      "step": 7575
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015680647467311557,
+      "loss": 0.1065,
+      "step": 7576
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.001567949178619754,
+      "loss": 0.0831,
+      "step": 7577
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015678335993098284,
+      "loss": 0.0873,
+      "step": 7578
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015677180088036577,
+      "loss": 0.1694,
+      "step": 7579
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015676024071035217,
+      "loss": 0.1274,
+      "step": 7580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015674867942116993,
+      "loss": 0.0878,
+      "step": 7581
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015673711701304702,
+      "loss": 0.1228,
+      "step": 7582
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015672555348621143,
+      "loss": 0.0887,
+      "step": 7583
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015671398884089115,
+      "loss": 0.1251,
+      "step": 7584
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015670242307731421,
+      "loss": 0.1379,
+      "step": 7585
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015669085619570864,
+      "loss": 0.1479,
+      "step": 7586
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015667928819630256,
+      "loss": 0.0746,
+      "step": 7587
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0015666771907932402,
+      "loss": 0.0852,
+      "step": 7588
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015665614884500115,
+      "loss": 0.1199,
+      "step": 7589
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001566445774935621,
+      "loss": 0.1267,
+      "step": 7590
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015663300502523497,
+      "loss": 0.1113,
+      "step": 7591
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015662143144024798,
+      "loss": 0.1409,
+      "step": 7592
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015660985673882934,
+      "loss": 0.0974,
+      "step": 7593
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015659828092120724,
+      "loss": 0.083,
+      "step": 7594
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015658670398761001,
+      "loss": 0.0975,
+      "step": 7595
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001565751259382658,
+      "loss": 0.1208,
+      "step": 7596
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015656354677340296,
+      "loss": 0.1176,
+      "step": 7597
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015655196649324986,
+      "loss": 0.0678,
+      "step": 7598
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015654038509803469,
+      "loss": 0.0935,
+      "step": 7599
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015652880258798594,
+      "loss": 0.165,
+      "step": 7600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015651721896333193,
+      "loss": 0.0967,
+      "step": 7601
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015650563422430104,
+      "loss": 0.1031,
+      "step": 7602
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015649404837112173,
+      "loss": 0.0775,
+      "step": 7603
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015648246140402247,
+      "loss": 0.106,
+      "step": 7604
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015647087332323166,
+      "loss": 0.0954,
+      "step": 7605
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015645928412897781,
+      "loss": 0.1207,
+      "step": 7606
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015644769382148942,
+      "loss": 0.1,
+      "step": 7607
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001564361024009951,
+      "loss": 0.0992,
+      "step": 7608
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015642450986772327,
+      "loss": 0.0657,
+      "step": 7609
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015641291622190262,
+      "loss": 0.1042,
+      "step": 7610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015640132146376167,
+      "loss": 0.0655,
+      "step": 7611
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015638972559352908,
+      "loss": 0.1157,
+      "step": 7612
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015637812861143351,
+      "loss": 0.1005,
+      "step": 7613
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015636653051770357,
+      "loss": 0.0798,
+      "step": 7614
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00156354931312568,
+      "loss": 0.0945,
+      "step": 7615
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015634333099625547,
+      "loss": 0.1525,
+      "step": 7616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015633172956899468,
+      "loss": 0.0979,
+      "step": 7617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015632012703101449,
+      "loss": 0.109,
+      "step": 7618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001563085233825435,
+      "loss": 0.0977,
+      "step": 7619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001562969186238107,
+      "loss": 0.1038,
+      "step": 7620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015628531275504477,
+      "loss": 0.1055,
+      "step": 7621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015627370577647461,
+      "loss": 0.119,
+      "step": 7622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015626209768832906,
+      "loss": 0.1382,
+      "step": 7623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015625048849083698,
+      "loss": 0.137,
+      "step": 7624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015623887818422733,
+      "loss": 0.1117,
+      "step": 7625
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015622726676872898,
+      "loss": 0.099,
+      "step": 7626
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001562156542445709,
+      "loss": 0.0825,
+      "step": 7627
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.001562040406119821,
+      "loss": 0.1035,
+      "step": 7628
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015619242587119149,
+      "loss": 0.0778,
+      "step": 7629
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015618081002242814,
+      "loss": 0.0856,
+      "step": 7630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015616919306592108,
+      "loss": 0.1041,
+      "step": 7631
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015615757500189935,
+      "loss": 0.107,
+      "step": 7632
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015614595583059207,
+      "loss": 0.1044,
+      "step": 7633
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0015613433555222826,
+      "loss": 0.1007,
+      "step": 7634
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001561227141670371,
+      "loss": 0.1071,
+      "step": 7635
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015611109167524772,
+      "loss": 0.1032,
+      "step": 7636
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015609946807708932,
+      "loss": 0.0878,
+      "step": 7637
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015608784337279107,
+      "loss": 0.1422,
+      "step": 7638
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001560762175625821,
+      "loss": 0.1094,
+      "step": 7639
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015606459064669176,
+      "loss": 0.1068,
+      "step": 7640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015605296262534928,
+      "loss": 0.0541,
+      "step": 7641
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015604133349878385,
+      "loss": 0.092,
+      "step": 7642
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015602970326722488,
+      "loss": 0.1965,
+      "step": 7643
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015601807193090158,
+      "loss": 0.0892,
+      "step": 7644
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015600643949004339,
+      "loss": 0.0724,
+      "step": 7645
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001559948059448796,
+      "loss": 0.0784,
+      "step": 7646
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015598317129563966,
+      "loss": 0.1039,
+      "step": 7647
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015597153554255292,
+      "loss": 0.1055,
+      "step": 7648
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001559598986858488,
+      "loss": 0.0708,
+      "step": 7649
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015594826072575679,
+      "loss": 0.1237,
+      "step": 7650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015593662166250636,
+      "loss": 0.1575,
+      "step": 7651
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015592498149632696,
+      "loss": 0.1202,
+      "step": 7652
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015591334022744816,
+      "loss": 0.1154,
+      "step": 7653
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015590169785609947,
+      "loss": 0.0822,
+      "step": 7654
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015589005438251045,
+      "loss": 0.0936,
+      "step": 7655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015587840980691068,
+      "loss": 0.0844,
+      "step": 7656
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015586676412952974,
+      "loss": 0.0854,
+      "step": 7657
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001558551173505973,
+      "loss": 0.1119,
+      "step": 7658
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015584346947034296,
+      "loss": 0.131,
+      "step": 7659
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001558318204889964,
+      "loss": 0.1033,
+      "step": 7660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015582017040678734,
+      "loss": 0.1161,
+      "step": 7661
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015580851922394542,
+      "loss": 0.0649,
+      "step": 7662
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015579686694070045,
+      "loss": 0.0865,
+      "step": 7663
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015578521355728214,
+      "loss": 0.097,
+      "step": 7664
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001557735590739203,
+      "loss": 0.1034,
+      "step": 7665
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015576190349084467,
+      "loss": 0.0768,
+      "step": 7666
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001557502468082851,
+      "loss": 0.0855,
+      "step": 7667
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015573858902647145,
+      "loss": 0.1087,
+      "step": 7668
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015572693014563356,
+      "loss": 0.1105,
+      "step": 7669
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001557152701660013,
+      "loss": 0.0764,
+      "step": 7670
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015570360908780461,
+      "loss": 0.0905,
+      "step": 7671
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015569194691127338,
+      "loss": 0.059,
+      "step": 7672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001556802836366376,
+      "loss": 0.1217,
+      "step": 7673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015566861926412721,
+      "loss": 0.093,
+      "step": 7674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015565695379397223,
+      "loss": 0.0708,
+      "step": 7675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.001556452872264026,
+      "loss": 0.0595,
+      "step": 7676
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015563361956164846,
+      "loss": 0.0924,
+      "step": 7677
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015562195079993977,
+      "loss": 0.1396,
+      "step": 7678
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015561028094150668,
+      "loss": 0.101,
+      "step": 7679
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0015559860998657928,
+      "loss": 0.0905,
+      "step": 7680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015558693793538767,
+      "loss": 0.0992,
+      "step": 7681
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015557526478816198,
+      "loss": 0.1523,
+      "step": 7682
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001555635905451324,
+      "loss": 0.0765,
+      "step": 7683
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015555191520652913,
+      "loss": 0.0626,
+      "step": 7684
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015554023877258235,
+      "loss": 0.1371,
+      "step": 7685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015552856124352228,
+      "loss": 0.0704,
+      "step": 7686
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001555168826195792,
+      "loss": 0.0866,
+      "step": 7687
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001555052029009834,
+      "loss": 0.1155,
+      "step": 7688
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001554935220879651,
+      "loss": 0.0865,
+      "step": 7689
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015548184018075468,
+      "loss": 0.0757,
+      "step": 7690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001554701571795825,
+      "loss": 0.2178,
+      "step": 7691
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015545847308467884,
+      "loss": 0.0548,
+      "step": 7692
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015544678789627415,
+      "loss": 0.0759,
+      "step": 7693
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001554351016145988,
+      "loss": 0.1532,
+      "step": 7694
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001554234142398832,
+      "loss": 0.0953,
+      "step": 7695
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015541172577235782,
+      "loss": 0.0845,
+      "step": 7696
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015540003621225312,
+      "loss": 0.1151,
+      "step": 7697
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001553883455597996,
+      "loss": 0.1118,
+      "step": 7698
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015537665381522773,
+      "loss": 0.1339,
+      "step": 7699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015536496097876811,
+      "loss": 0.1019,
+      "step": 7700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015535326705065126,
+      "loss": 0.0984,
+      "step": 7701
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001553415720311077,
+      "loss": 0.1433,
+      "step": 7702
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001553298759203681,
+      "loss": 0.1465,
+      "step": 7703
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015531817871866308,
+      "loss": 0.1224,
+      "step": 7704
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015530648042622322,
+      "loss": 0.1009,
+      "step": 7705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001552947810432792,
+      "loss": 0.108,
+      "step": 7706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015528308057006176,
+      "loss": 0.1892,
+      "step": 7707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015527137900680152,
+      "loss": 0.102,
+      "step": 7708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001552596763537293,
+      "loss": 0.0728,
+      "step": 7709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015524797261107573,
+      "loss": 0.0939,
+      "step": 7710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015523626777907166,
+      "loss": 0.1498,
+      "step": 7711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015522456185794787,
+      "loss": 0.099,
+      "step": 7712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015521285484793514,
+      "loss": 0.0886,
+      "step": 7713
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015520114674926436,
+      "loss": 0.1269,
+      "step": 7714
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015518943756216632,
+      "loss": 0.1589,
+      "step": 7715
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015517772728687192,
+      "loss": 0.1052,
+      "step": 7716
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015516601592361209,
+      "loss": 0.1593,
+      "step": 7717
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001551543034726177,
+      "loss": 0.1238,
+      "step": 7718
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001551425899341197,
+      "loss": 0.0831,
+      "step": 7719
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001551308753083491,
+      "loss": 0.144,
+      "step": 7720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.001551191595955368,
+      "loss": 0.0568,
+      "step": 7721
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015510744279591381,
+      "loss": 0.12,
+      "step": 7722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015509572490971127,
+      "loss": 0.1643,
+      "step": 7723
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015508400593716011,
+      "loss": 0.1429,
+      "step": 7724
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015507228587849147,
+      "loss": 0.1508,
+      "step": 7725
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0015506056473393638,
+      "loss": 0.1141,
+      "step": 7726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015504884250372598,
+      "loss": 0.0939,
+      "step": 7727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015503711918809142,
+      "loss": 0.1172,
+      "step": 7728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015502539478726384,
+      "loss": 0.0691,
+      "step": 7729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015501366930147438,
+      "loss": 0.1033,
+      "step": 7730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015500194273095431,
+      "loss": 0.0932,
+      "step": 7731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015499021507593476,
+      "loss": 0.1675,
+      "step": 7732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015497848633664706,
+      "loss": 0.1641,
+      "step": 7733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.001549667565133224,
+      "loss": 0.1276,
+      "step": 7734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015495502560619212,
+      "loss": 0.0972,
+      "step": 7735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015494329361548746,
+      "loss": 0.0894,
+      "step": 7736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015493156054143982,
+      "loss": 0.0929,
+      "step": 7737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015491982638428051,
+      "loss": 0.101,
+      "step": 7738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015490809114424085,
+      "loss": 0.1125,
+      "step": 7739
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015489635482155229,
+      "loss": 0.069,
+      "step": 7740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015488461741644624,
+      "loss": 0.0859,
+      "step": 7741
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.001548728789291541,
+      "loss": 0.0829,
+      "step": 7742
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015486113935990736,
+      "loss": 0.0704,
+      "step": 7743
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015484939870893746,
+      "loss": 0.1384,
+      "step": 7744
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015483765697647592,
+      "loss": 0.1135,
+      "step": 7745
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015482591416275421,
+      "loss": 0.1454,
+      "step": 7746
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015481417026800396,
+      "loss": 0.07,
+      "step": 7747
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015480242529245665,
+      "loss": 0.1176,
+      "step": 7748
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.001547906792363439,
+      "loss": 0.0705,
+      "step": 7749
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.001547789320998973,
+      "loss": 0.1501,
+      "step": 7750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015476718388334847,
+      "loss": 0.1041,
+      "step": 7751
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015475543458692905,
+      "loss": 0.0714,
+      "step": 7752
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015474368421087071,
+      "loss": 0.1111,
+      "step": 7753
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015473193275540517,
+      "loss": 0.1029,
+      "step": 7754
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.001547201802207641,
+      "loss": 0.1306,
+      "step": 7755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015470842660717926,
+      "loss": 0.0957,
+      "step": 7756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015469667191488235,
+      "loss": 0.1017,
+      "step": 7757
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.001546849161441052,
+      "loss": 0.0889,
+      "step": 7758
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015467315929507959,
+      "loss": 0.1174,
+      "step": 7759
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015466140136803732,
+      "loss": 0.0851,
+      "step": 7760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015464964236321021,
+      "loss": 0.1178,
+      "step": 7761
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015463788228083015,
+      "loss": 0.1381,
+      "step": 7762
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015462612112112904,
+      "loss": 0.0838,
+      "step": 7763
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015461435888433873,
+      "loss": 0.1302,
+      "step": 7764
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015460259557069115,
+      "loss": 0.1296,
+      "step": 7765
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015459083118041829,
+      "loss": 0.109,
+      "step": 7766
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015457906571375206,
+      "loss": 0.1005,
+      "step": 7767
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015456729917092448,
+      "loss": 0.0963,
+      "step": 7768
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015455553155216749,
+      "loss": 0.1472,
+      "step": 7769
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015454376285771324,
+      "loss": 0.1729,
+      "step": 7770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015453199308779367,
+      "loss": 0.0953,
+      "step": 7771
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0015452022224264085,
+      "loss": 0.0822,
+      "step": 7772
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015450845032248697,
+      "loss": 0.1001,
+      "step": 7773
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015449667732756408,
+      "loss": 0.1099,
+      "step": 7774
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001544849032581043,
+      "loss": 0.101,
+      "step": 7775
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001544731281143398,
+      "loss": 0.0776,
+      "step": 7776
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015446135189650274,
+      "loss": 0.119,
+      "step": 7777
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015444957460482537,
+      "loss": 0.0868,
+      "step": 7778
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015443779623953984,
+      "loss": 0.0642,
+      "step": 7779
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015442601680087842,
+      "loss": 0.0997,
+      "step": 7780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001544142362890734,
+      "loss": 0.1163,
+      "step": 7781
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015440245470435703,
+      "loss": 0.1172,
+      "step": 7782
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015439067204696163,
+      "loss": 0.0941,
+      "step": 7783
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015437888831711947,
+      "loss": 0.0775,
+      "step": 7784
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015436710351506297,
+      "loss": 0.0969,
+      "step": 7785
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015435531764102443,
+      "loss": 0.092,
+      "step": 7786
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015434353069523632,
+      "loss": 0.063,
+      "step": 7787
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015433174267793098,
+      "loss": 0.1056,
+      "step": 7788
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015431995358934085,
+      "loss": 0.0743,
+      "step": 7789
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001543081634296984,
+      "loss": 0.0696,
+      "step": 7790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015429637219923611,
+      "loss": 0.085,
+      "step": 7791
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015428457989818643,
+      "loss": 0.0977,
+      "step": 7792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015427278652678192,
+      "loss": 0.0984,
+      "step": 7793
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015426099208525508,
+      "loss": 0.0867,
+      "step": 7794
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001542491965738385,
+      "loss": 0.0968,
+      "step": 7795
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015423739999276473,
+      "loss": 0.096,
+      "step": 7796
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015422560234226639,
+      "loss": 0.0654,
+      "step": 7797
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001542138036225761,
+      "loss": 0.0697,
+      "step": 7798
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015420200383392646,
+      "loss": 0.118,
+      "step": 7799
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001541902029765502,
+      "loss": 0.0773,
+      "step": 7800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015417840105067992,
+      "loss": 0.0938,
+      "step": 7801
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001541665980565484,
+      "loss": 0.1213,
+      "step": 7802
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015415479399438837,
+      "loss": 0.0784,
+      "step": 7803
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015414298886443248,
+      "loss": 0.0947,
+      "step": 7804
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001541311826669136,
+      "loss": 0.0994,
+      "step": 7805
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015411937540206445,
+      "loss": 0.0723,
+      "step": 7806
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015410756707011789,
+      "loss": 0.062,
+      "step": 7807
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001540957576713067,
+      "loss": 0.0988,
+      "step": 7808
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015408394720586376,
+      "loss": 0.0975,
+      "step": 7809
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015407213567402197,
+      "loss": 0.0675,
+      "step": 7810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015406032307601419,
+      "loss": 0.1803,
+      "step": 7811
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015404850941207334,
+      "loss": 0.1364,
+      "step": 7812
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015403669468243236,
+      "loss": 0.0785,
+      "step": 7813
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015402487888732416,
+      "loss": 0.0682,
+      "step": 7814
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015401306202698182,
+      "loss": 0.1124,
+      "step": 7815
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015400124410163824,
+      "loss": 0.0875,
+      "step": 7816
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015398942511152649,
+      "loss": 0.1366,
+      "step": 7817
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0015397760505687958,
+      "loss": 0.0885,
+      "step": 7818
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.001539657839379306,
+      "loss": 0.0946,
+      "step": 7819
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015395396175491263,
+      "loss": 0.124,
+      "step": 7820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015394213850805879,
+      "loss": 0.0911,
+      "step": 7821
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015393031419760213,
+      "loss": 0.0724,
+      "step": 7822
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015391848882377588,
+      "loss": 0.0651,
+      "step": 7823
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015390666238681314,
+      "loss": 0.1427,
+      "step": 7824
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001538948348869472,
+      "loss": 0.1113,
+      "step": 7825
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015388300632441113,
+      "loss": 0.1122,
+      "step": 7826
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015387117669943826,
+      "loss": 0.0892,
+      "step": 7827
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001538593460122618,
+      "loss": 0.1476,
+      "step": 7828
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015384751426311502,
+      "loss": 0.0743,
+      "step": 7829
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015383568145223125,
+      "loss": 0.082,
+      "step": 7830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015382384757984377,
+      "loss": 0.1222,
+      "step": 7831
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001538120126461859,
+      "loss": 0.1239,
+      "step": 7832
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015380017665149103,
+      "loss": 0.0974,
+      "step": 7833
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015378833959599253,
+      "loss": 0.0831,
+      "step": 7834
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015377650147992378,
+      "loss": 0.0651,
+      "step": 7835
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015376466230351818,
+      "loss": 0.0941,
+      "step": 7836
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001537528220670092,
+      "loss": 0.0656,
+      "step": 7837
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015374098077063035,
+      "loss": 0.0948,
+      "step": 7838
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015372913841461499,
+      "loss": 0.124,
+      "step": 7839
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015371729499919669,
+      "loss": 0.1409,
+      "step": 7840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00153705450524609,
+      "loss": 0.1026,
+      "step": 7841
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001536936049910854,
+      "loss": 0.0698,
+      "step": 7842
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001536817583988595,
+      "loss": 0.1622,
+      "step": 7843
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015366991074816487,
+      "loss": 0.1079,
+      "step": 7844
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001536580620392351,
+      "loss": 0.1097,
+      "step": 7845
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015364621227230385,
+      "loss": 0.1003,
+      "step": 7846
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015363436144760474,
+      "loss": 0.1511,
+      "step": 7847
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015362250956537148,
+      "loss": 0.1004,
+      "step": 7848
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015361065662583768,
+      "loss": 0.1019,
+      "step": 7849
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015359880262923711,
+      "loss": 0.0762,
+      "step": 7850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001535869475758035,
+      "loss": 0.1843,
+      "step": 7851
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015357509146577055,
+      "loss": 0.1345,
+      "step": 7852
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015356323429937212,
+      "loss": 0.0999,
+      "step": 7853
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015355137607684195,
+      "loss": 0.0911,
+      "step": 7854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015353951679841385,
+      "loss": 0.0829,
+      "step": 7855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015352765646432162,
+      "loss": 0.1213,
+      "step": 7856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015351579507479922,
+      "loss": 0.0759,
+      "step": 7857
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001535039326300805,
+      "loss": 0.0914,
+      "step": 7858
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015349206913039923,
+      "loss": 0.1353,
+      "step": 7859
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001534802045759895,
+      "loss": 0.1009,
+      "step": 7860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015346833896708514,
+      "loss": 0.131,
+      "step": 7861
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.001534564723039201,
+      "loss": 0.0667,
+      "step": 7862
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015344460458672847,
+      "loss": 0.0968,
+      "step": 7863
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015343273581574417,
+      "loss": 0.1503,
+      "step": 7864
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0015342086599120122,
+      "loss": 0.0911,
+      "step": 7865
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015340899511333368,
+      "loss": 0.1086,
+      "step": 7866
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015339712318237562,
+      "loss": 0.0917,
+      "step": 7867
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015338525019856113,
+      "loss": 0.098,
+      "step": 7868
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015337337616212429,
+      "loss": 0.0842,
+      "step": 7869
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015336150107329923,
+      "loss": 0.0833,
+      "step": 7870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015334962493232008,
+      "loss": 0.1609,
+      "step": 7871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015333774773942106,
+      "loss": 0.0767,
+      "step": 7872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015332586949483634,
+      "loss": 0.0719,
+      "step": 7873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015331399019880014,
+      "loss": 0.0662,
+      "step": 7874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015330210985154663,
+      "loss": 0.0992,
+      "step": 7875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015329022845331011,
+      "loss": 0.1868,
+      "step": 7876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015327834600432485,
+      "loss": 0.104,
+      "step": 7877
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015326646250482514,
+      "loss": 0.0974,
+      "step": 7878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015325457795504526,
+      "loss": 0.1109,
+      "step": 7879
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001532426923552196,
+      "loss": 0.1429,
+      "step": 7880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015323080570558245,
+      "loss": 0.0732,
+      "step": 7881
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015321891800636823,
+      "loss": 0.1187,
+      "step": 7882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015320702925781134,
+      "loss": 0.1344,
+      "step": 7883
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015319513946014617,
+      "loss": 0.1154,
+      "step": 7884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015318324861360716,
+      "loss": 0.0971,
+      "step": 7885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001531713567184288,
+      "loss": 0.2708,
+      "step": 7886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015315946377484548,
+      "loss": 0.0936,
+      "step": 7887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001531475697830918,
+      "loss": 0.1758,
+      "step": 7888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015313567474340224,
+      "loss": 0.107,
+      "step": 7889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015312377865601134,
+      "loss": 0.1212,
+      "step": 7890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015311188152115367,
+      "loss": 0.0984,
+      "step": 7891
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015309998333906379,
+      "loss": 0.1254,
+      "step": 7892
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015308808410997633,
+      "loss": 0.1199,
+      "step": 7893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015307618383412589,
+      "loss": 0.1174,
+      "step": 7894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015306428251174711,
+      "loss": 0.1313,
+      "step": 7895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015305238014307468,
+      "loss": 0.1558,
+      "step": 7896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015304047672834327,
+      "loss": 0.1624,
+      "step": 7897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015302857226778756,
+      "loss": 0.1128,
+      "step": 7898
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001530166667616423,
+      "loss": 0.0978,
+      "step": 7899
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015300476021014226,
+      "loss": 0.1047,
+      "step": 7900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001529928526135222,
+      "loss": 0.1003,
+      "step": 7901
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015298094397201685,
+      "loss": 0.1654,
+      "step": 7902
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015296903428586109,
+      "loss": 0.0872,
+      "step": 7903
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001529571235552897,
+      "loss": 0.1558,
+      "step": 7904
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015294521178053756,
+      "loss": 0.1154,
+      "step": 7905
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015293329896183955,
+      "loss": 0.0938,
+      "step": 7906
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015292138509943052,
+      "loss": 0.0616,
+      "step": 7907
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.001529094701935454,
+      "loss": 0.0861,
+      "step": 7908
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015289755424441908,
+      "loss": 0.0765,
+      "step": 7909
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015288563725228662,
+      "loss": 0.2135,
+      "step": 7910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0015287371921738294,
+      "loss": 0.1194,
+      "step": 7911
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015286180013994297,
+      "loss": 0.1255,
+      "step": 7912
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001528498800202018,
+      "loss": 0.1177,
+      "step": 7913
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015283795885839442,
+      "loss": 0.0913,
+      "step": 7914
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001528260366547559,
+      "loss": 0.0548,
+      "step": 7915
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015281411340952137,
+      "loss": 0.1058,
+      "step": 7916
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015280218912292584,
+      "loss": 0.1196,
+      "step": 7917
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001527902637952045,
+      "loss": 0.0969,
+      "step": 7918
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001527783374265924,
+      "loss": 0.1099,
+      "step": 7919
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015276641001732477,
+      "loss": 0.0805,
+      "step": 7920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015275448156763678,
+      "loss": 0.1389,
+      "step": 7921
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001527425520777636,
+      "loss": 0.0967,
+      "step": 7922
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001527306215479405,
+      "loss": 0.0574,
+      "step": 7923
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015271868997840264,
+      "loss": 0.0839,
+      "step": 7924
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015270675736938535,
+      "loss": 0.1653,
+      "step": 7925
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015269482372112391,
+      "loss": 0.0662,
+      "step": 7926
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015268288903385358,
+      "loss": 0.0834,
+      "step": 7927
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001526709533078097,
+      "loss": 0.1187,
+      "step": 7928
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001526590165432276,
+      "loss": 0.1125,
+      "step": 7929
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015264707874034268,
+      "loss": 0.111,
+      "step": 7930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015263513989939028,
+      "loss": 0.0805,
+      "step": 7931
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015262320002060582,
+      "loss": 0.0773,
+      "step": 7932
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015261125910422475,
+      "loss": 0.0911,
+      "step": 7933
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015259931715048251,
+      "loss": 0.0902,
+      "step": 7934
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015258737415961452,
+      "loss": 0.1004,
+      "step": 7935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015257543013185626,
+      "loss": 0.0773,
+      "step": 7936
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015256348506744333,
+      "loss": 0.1058,
+      "step": 7937
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015255153896661117,
+      "loss": 0.0896,
+      "step": 7938
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015253959182959535,
+      "loss": 0.1151,
+      "step": 7939
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015252764365663142,
+      "loss": 0.1038,
+      "step": 7940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015251569444795503,
+      "loss": 0.085,
+      "step": 7941
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015250374420380172,
+      "loss": 0.1003,
+      "step": 7942
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015249179292440719,
+      "loss": 0.0836,
+      "step": 7943
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00152479840610007,
+      "loss": 0.1288,
+      "step": 7944
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015246788726083684,
+      "loss": 0.0748,
+      "step": 7945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015245593287713246,
+      "loss": 0.1112,
+      "step": 7946
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015244397745912952,
+      "loss": 0.0956,
+      "step": 7947
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001524320210070638,
+      "loss": 0.077,
+      "step": 7948
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015242006352117097,
+      "loss": 0.1,
+      "step": 7949
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015240810500168687,
+      "loss": 0.0815,
+      "step": 7950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015239614544884726,
+      "loss": 0.055,
+      "step": 7951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015238418486288796,
+      "loss": 0.1018,
+      "step": 7952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.001523722232440448,
+      "loss": 0.1376,
+      "step": 7953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015236026059255365,
+      "loss": 0.1144,
+      "step": 7954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015234829690865036,
+      "loss": 0.1141,
+      "step": 7955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0015233633219257081,
+      "loss": 0.1193,
+      "step": 7956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00152324366444551,
+      "loss": 0.0992,
+      "step": 7957
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015231239966482675,
+      "loss": 0.101,
+      "step": 7958
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015230043185363406,
+      "loss": 0.0964,
+      "step": 7959
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015228846301120896,
+      "loss": 0.1113,
+      "step": 7960
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015227649313778735,
+      "loss": 0.0874,
+      "step": 7961
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001522645222336053,
+      "loss": 0.089,
+      "step": 7962
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015225255029889886,
+      "loss": 0.1257,
+      "step": 7963
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015224057733390407,
+      "loss": 0.0637,
+      "step": 7964
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015222860333885696,
+      "loss": 0.0812,
+      "step": 7965
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001522166283139937,
+      "loss": 0.0624,
+      "step": 7966
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015220465225955035,
+      "loss": 0.1023,
+      "step": 7967
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001521926751757631,
+      "loss": 0.0757,
+      "step": 7968
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015218069706286808,
+      "loss": 0.1324,
+      "step": 7969
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015216871792110146,
+      "loss": 0.0777,
+      "step": 7970
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015215673775069946,
+      "loss": 0.1387,
+      "step": 7971
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015214475655189825,
+      "loss": 0.121,
+      "step": 7972
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001521327743249341,
+      "loss": 0.0917,
+      "step": 7973
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015212079107004334,
+      "loss": 0.1154,
+      "step": 7974
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015210880678746214,
+      "loss": 0.0699,
+      "step": 7975
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015209682147742684,
+      "loss": 0.101,
+      "step": 7976
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015208483514017378,
+      "loss": 0.1038,
+      "step": 7977
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015207284777593926,
+      "loss": 0.0934,
+      "step": 7978
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015206085938495968,
+      "loss": 0.1696,
+      "step": 7979
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001520488699674714,
+      "loss": 0.1237,
+      "step": 7980
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015203687952371086,
+      "loss": 0.0989,
+      "step": 7981
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001520248880539144,
+      "loss": 0.0689,
+      "step": 7982
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015201289555831852,
+      "loss": 0.1096,
+      "step": 7983
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015200090203715969,
+      "loss": 0.0851,
+      "step": 7984
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015198890749067435,
+      "loss": 0.077,
+      "step": 7985
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015197691191909907,
+      "loss": 0.0908,
+      "step": 7986
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001519649153226703,
+      "loss": 0.1128,
+      "step": 7987
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015195291770162458,
+      "loss": 0.0879,
+      "step": 7988
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015194091905619852,
+      "loss": 0.098,
+      "step": 7989
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015192891938662872,
+      "loss": 0.0876,
+      "step": 7990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015191691869315173,
+      "loss": 0.0845,
+      "step": 7991
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015190491697600415,
+      "loss": 0.1887,
+      "step": 7992
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015189291423542272,
+      "loss": 0.1082,
+      "step": 7993
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00151880910471644,
+      "loss": 0.0859,
+      "step": 7994
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015186890568490478,
+      "loss": 0.1312,
+      "step": 7995
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015185689987544167,
+      "loss": 0.0991,
+      "step": 7996
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015184489304349148,
+      "loss": 0.0612,
+      "step": 7997
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015183288518929084,
+      "loss": 0.0611,
+      "step": 7998
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015182087631307659,
+      "loss": 0.0676,
+      "step": 7999
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015180886641508553,
+      "loss": 0.1003,
+      "step": 8000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015179685549555444,
+      "loss": 0.1455,
+      "step": 8001
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0015178484355472016,
+      "loss": 0.0739,
+      "step": 8002
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.001517728305928195,
+      "loss": 0.0839,
+      "step": 8003
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015176081661008934,
+      "loss": 0.0917,
+      "step": 8004
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015174880160676656,
+      "loss": 0.1204,
+      "step": 8005
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.001517367855830881,
+      "loss": 0.0927,
+      "step": 8006
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015172476853929089,
+      "loss": 0.1155,
+      "step": 8007
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015171275047561183,
+      "loss": 0.1372,
+      "step": 8008
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015170073139228787,
+      "loss": 0.1384,
+      "step": 8009
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015168871128955603,
+      "loss": 0.1206,
+      "step": 8010
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015167669016765335,
+      "loss": 0.1096,
+      "step": 8011
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.001516646680268168,
+      "loss": 0.1528,
+      "step": 8012
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015165264486728348,
+      "loss": 0.1643,
+      "step": 8013
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015164062068929037,
+      "loss": 0.0747,
+      "step": 8014
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015162859549307465,
+      "loss": 0.1628,
+      "step": 8015
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015161656927887338,
+      "loss": 0.1642,
+      "step": 8016
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015160454204692368,
+      "loss": 0.1255,
+      "step": 8017
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015159251379746276,
+      "loss": 0.0959,
+      "step": 8018
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015158048453072768,
+      "loss": 0.1112,
+      "step": 8019
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015156845424695571,
+      "loss": 0.1034,
+      "step": 8020
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015155642294638404,
+      "loss": 0.0984,
+      "step": 8021
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015154439062924987,
+      "loss": 0.1469,
+      "step": 8022
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015153235729579047,
+      "loss": 0.1846,
+      "step": 8023
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015152032294624312,
+      "loss": 0.1592,
+      "step": 8024
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015150828758084508,
+      "loss": 0.0696,
+      "step": 8025
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015149625119983364,
+      "loss": 0.0938,
+      "step": 8026
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015148421380344619,
+      "loss": 0.1176,
+      "step": 8027
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015147217539192002,
+      "loss": 0.0795,
+      "step": 8028
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015146013596549252,
+      "loss": 0.1057,
+      "step": 8029
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.001514480955244011,
+      "loss": 0.0885,
+      "step": 8030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015143605406888307,
+      "loss": 0.1279,
+      "step": 8031
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015142401159917598,
+      "loss": 0.1093,
+      "step": 8032
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015141196811551724,
+      "loss": 0.0617,
+      "step": 8033
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015139992361814427,
+      "loss": 0.1222,
+      "step": 8034
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015138787810729459,
+      "loss": 0.0828,
+      "step": 8035
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015137583158320572,
+      "loss": 0.0864,
+      "step": 8036
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.001513637840461152,
+      "loss": 0.0847,
+      "step": 8037
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015135173549626051,
+      "loss": 0.0838,
+      "step": 8038
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015133968593387928,
+      "loss": 0.0696,
+      "step": 8039
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015132763535920905,
+      "loss": 0.0972,
+      "step": 8040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015131558377248747,
+      "loss": 0.0896,
+      "step": 8041
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015130353117395212,
+      "loss": 0.0682,
+      "step": 8042
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015129147756384072,
+      "loss": 0.0871,
+      "step": 8043
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.001512794229423909,
+      "loss": 0.0896,
+      "step": 8044
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.001512673673098403,
+      "loss": 0.0927,
+      "step": 8045
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015125531066642669,
+      "loss": 0.0545,
+      "step": 8046
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015124325301238775,
+      "loss": 0.0866,
+      "step": 8047
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015123119434796127,
+      "loss": 0.103,
+      "step": 8048
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0015121913467338497,
+      "loss": 0.0889,
+      "step": 8049
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015120707398889668,
+      "loss": 0.0984,
+      "step": 8050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015119501229473417,
+      "loss": 0.1069,
+      "step": 8051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001511829495911353,
+      "loss": 0.0727,
+      "step": 8052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001511708858783379,
+      "loss": 0.097,
+      "step": 8053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015115882115657984,
+      "loss": 0.095,
+      "step": 8054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015114675542609896,
+      "loss": 0.0596,
+      "step": 8055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015113468868713329,
+      "loss": 0.1182,
+      "step": 8056
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001511226209399206,
+      "loss": 0.1458,
+      "step": 8057
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015111055218469894,
+      "loss": 0.1115,
+      "step": 8058
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015109848242170624,
+      "loss": 0.1313,
+      "step": 8059
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001510864116511805,
+      "loss": 0.1038,
+      "step": 8060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001510743398733597,
+      "loss": 0.1077,
+      "step": 8061
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015106226708848185,
+      "loss": 0.0851,
+      "step": 8062
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015105019329678506,
+      "loss": 0.1171,
+      "step": 8063
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015103811849850736,
+      "loss": 0.0952,
+      "step": 8064
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001510260426938868,
+      "loss": 0.1276,
+      "step": 8065
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015101396588316155,
+      "loss": 0.1227,
+      "step": 8066
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015100188806656972,
+      "loss": 0.0714,
+      "step": 8067
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001509898092443494,
+      "loss": 0.1053,
+      "step": 8068
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015097772941673878,
+      "loss": 0.0859,
+      "step": 8069
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015096564858397608,
+      "loss": 0.1252,
+      "step": 8070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015095356674629948,
+      "loss": 0.0804,
+      "step": 8071
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015094148390394717,
+      "loss": 0.0945,
+      "step": 8072
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015092940005715745,
+      "loss": 0.1041,
+      "step": 8073
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015091731520616857,
+      "loss": 0.1226,
+      "step": 8074
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015090522935121875,
+      "loss": 0.072,
+      "step": 8075
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015089314249254638,
+      "loss": 0.1094,
+      "step": 8076
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015088105463038973,
+      "loss": 0.1195,
+      "step": 8077
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015086896576498718,
+      "loss": 0.0768,
+      "step": 8078
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00150856875896577,
+      "loss": 0.08,
+      "step": 8079
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.001508447850253977,
+      "loss": 0.0968,
+      "step": 8080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015083269315168764,
+      "loss": 0.1532,
+      "step": 8081
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015082060027568516,
+      "loss": 0.0909,
+      "step": 8082
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015080850639762882,
+      "loss": 0.0893,
+      "step": 8083
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015079641151775698,
+      "loss": 0.0948,
+      "step": 8084
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015078431563630818,
+      "loss": 0.1002,
+      "step": 8085
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015077221875352091,
+      "loss": 0.0794,
+      "step": 8086
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015076012086963368,
+      "loss": 0.0719,
+      "step": 8087
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015074802198488506,
+      "loss": 0.1063,
+      "step": 8088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015073592209951352,
+      "loss": 0.1317,
+      "step": 8089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015072382121375778,
+      "loss": 0.0836,
+      "step": 8090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015071171932785632,
+      "loss": 0.0565,
+      "step": 8091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015069961644204781,
+      "loss": 0.0864,
+      "step": 8092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015068751255657086,
+      "loss": 0.0908,
+      "step": 8093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015067540767166416,
+      "loss": 0.0936,
+      "step": 8094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0015066330178756637,
+      "loss": 0.0909,
+      "step": 8095
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015065119490451622,
+      "loss": 0.0854,
+      "step": 8096
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001506390870227524,
+      "loss": 0.1344,
+      "step": 8097
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015062697814251364,
+      "loss": 0.1011,
+      "step": 8098
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015061486826403866,
+      "loss": 0.1133,
+      "step": 8099
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001506027573875663,
+      "loss": 0.1287,
+      "step": 8100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015059064551333535,
+      "loss": 0.1165,
+      "step": 8101
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001505785326415846,
+      "loss": 0.0718,
+      "step": 8102
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015056641877255288,
+      "loss": 0.0568,
+      "step": 8103
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015055430390647911,
+      "loss": 0.0887,
+      "step": 8104
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015054218804360204,
+      "loss": 0.1287,
+      "step": 8105
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015053007118416071,
+      "loss": 0.064,
+      "step": 8106
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015051795332839393,
+      "loss": 0.0812,
+      "step": 8107
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015050583447654069,
+      "loss": 0.1019,
+      "step": 8108
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015049371462883992,
+      "loss": 0.1187,
+      "step": 8109
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015048159378553054,
+      "loss": 0.0861,
+      "step": 8110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015046947194685164,
+      "loss": 0.0776,
+      "step": 8111
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015045734911304219,
+      "loss": 0.065,
+      "step": 8112
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001504452252843412,
+      "loss": 0.0731,
+      "step": 8113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015043310046098776,
+      "loss": 0.0755,
+      "step": 8114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001504209746432209,
+      "loss": 0.0841,
+      "step": 8115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015040884783127976,
+      "loss": 0.0914,
+      "step": 8116
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015039672002540342,
+      "loss": 0.104,
+      "step": 8117
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015038459122583099,
+      "loss": 0.0818,
+      "step": 8118
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015037246143280163,
+      "loss": 0.072,
+      "step": 8119
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015036033064655455,
+      "loss": 0.1317,
+      "step": 8120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001503481988673289,
+      "loss": 0.0975,
+      "step": 8121
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015033606609536389,
+      "loss": 0.0671,
+      "step": 8122
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015032393233089874,
+      "loss": 0.0784,
+      "step": 8123
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015031179757417274,
+      "loss": 0.0864,
+      "step": 8124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001502996618254251,
+      "loss": 0.1515,
+      "step": 8125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015028752508489513,
+      "loss": 0.0819,
+      "step": 8126
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015027538735282219,
+      "loss": 0.1371,
+      "step": 8127
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001502632486294455,
+      "loss": 0.1447,
+      "step": 8128
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015025110891500448,
+      "loss": 0.1093,
+      "step": 8129
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015023896820973851,
+      "loss": 0.09,
+      "step": 8130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015022682651388692,
+      "loss": 0.1625,
+      "step": 8131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015021468382768908,
+      "loss": 0.088,
+      "step": 8132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001502025401513845,
+      "loss": 0.2114,
+      "step": 8133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015019039548521264,
+      "loss": 0.1278,
+      "step": 8134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015017824982941283,
+      "loss": 0.1112,
+      "step": 8135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.001501661031842247,
+      "loss": 0.1074,
+      "step": 8136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015015395554988762,
+      "loss": 0.105,
+      "step": 8137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015014180692664123,
+      "loss": 0.0765,
+      "step": 8138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015012965731472495,
+      "loss": 0.111,
+      "step": 8139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015011750671437847,
+      "loss": 0.0522,
+      "step": 8140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0015010535512584129,
+      "loss": 0.0974,
+      "step": 8141
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00150093202549353,
+      "loss": 0.112,
+      "step": 8142
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0015008104898515322,
+      "loss": 0.1214,
+      "step": 8143
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0015006889443348165,
+      "loss": 0.0864,
+      "step": 8144
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001500567388945779,
+      "loss": 0.1233,
+      "step": 8145
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0015004458236868164,
+      "loss": 0.1569,
+      "step": 8146
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0015003242485603257,
+      "loss": 0.0908,
+      "step": 8147
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0015002026635687043,
+      "loss": 0.0623,
+      "step": 8148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0015000810687143492,
+      "loss": 0.1218,
+      "step": 8149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014999594639996583,
+      "loss": 0.0962,
+      "step": 8150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001499837849427029,
+      "loss": 0.1213,
+      "step": 8151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014997162249988595,
+      "loss": 0.1328,
+      "step": 8152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014995945907175475,
+      "loss": 0.1023,
+      "step": 8153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001499472946585492,
+      "loss": 0.1299,
+      "step": 8154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014993512926050909,
+      "loss": 0.1148,
+      "step": 8155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014992296287787434,
+      "loss": 0.1097,
+      "step": 8156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014991079551088483,
+      "loss": 0.1086,
+      "step": 8157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014989862715978043,
+      "loss": 0.1207,
+      "step": 8158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014988645782480106,
+      "loss": 0.0963,
+      "step": 8159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014987428750618675,
+      "loss": 0.1088,
+      "step": 8160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014986211620417742,
+      "loss": 0.0801,
+      "step": 8161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014984994391901306,
+      "loss": 0.1261,
+      "step": 8162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014983777065093364,
+      "loss": 0.0812,
+      "step": 8163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014982559640017924,
+      "loss": 0.1405,
+      "step": 8164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001498134211669899,
+      "loss": 0.0793,
+      "step": 8165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001498012449516057,
+      "loss": 0.0727,
+      "step": 8166
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014978906775426667,
+      "loss": 0.0539,
+      "step": 8167
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014977688957521292,
+      "loss": 0.0909,
+      "step": 8168
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001497647104146846,
+      "loss": 0.1235,
+      "step": 8169
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014975253027292186,
+      "loss": 0.151,
+      "step": 8170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014974034915016486,
+      "loss": 0.1125,
+      "step": 8171
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014972816704665374,
+      "loss": 0.0727,
+      "step": 8172
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014971598396262874,
+      "loss": 0.132,
+      "step": 8173
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014970379989833005,
+      "loss": 0.0701,
+      "step": 8174
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014969161485399792,
+      "loss": 0.1331,
+      "step": 8175
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014967942882987263,
+      "loss": 0.0776,
+      "step": 8176
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001496672418261945,
+      "loss": 0.1403,
+      "step": 8177
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014965505384320367,
+      "loss": 0.1138,
+      "step": 8178
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014964286488114062,
+      "loss": 0.1305,
+      "step": 8179
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.001496306749402456,
+      "loss": 0.0753,
+      "step": 8180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014961848402075901,
+      "loss": 0.1062,
+      "step": 8181
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014960629212292119,
+      "loss": 0.1053,
+      "step": 8182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014959409924697252,
+      "loss": 0.1254,
+      "step": 8183
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014958190539315344,
+      "loss": 0.0814,
+      "step": 8184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014956971056170438,
+      "loss": 0.0898,
+      "step": 8185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014955751475286584,
+      "loss": 0.0942,
+      "step": 8186
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014954531796687821,
+      "loss": 0.0989,
+      "step": 8187
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0014953312020398197,
+      "loss": 0.1089,
+      "step": 8188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014952092146441772,
+      "loss": 0.1229,
+      "step": 8189
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001495087217484259,
+      "loss": 0.0762,
+      "step": 8190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014949652105624713,
+      "loss": 0.0859,
+      "step": 8191
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014948431938812193,
+      "loss": 0.0892,
+      "step": 8192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001494721167442909,
+      "loss": 0.152,
+      "step": 8193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014945991312499466,
+      "loss": 0.0956,
+      "step": 8194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014944770853047378,
+      "loss": 0.0714,
+      "step": 8195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014943550296096894,
+      "loss": 0.1494,
+      "step": 8196
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014942329641672086,
+      "loss": 0.0956,
+      "step": 8197
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014941108889797011,
+      "loss": 0.0969,
+      "step": 8198
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014939888040495749,
+      "loss": 0.0955,
+      "step": 8199
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014938667093792365,
+      "loss": 0.1383,
+      "step": 8200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014937446049710936,
+      "loss": 0.1595,
+      "step": 8201
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001493622490827554,
+      "loss": 0.114,
+      "step": 8202
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014935003669510252,
+      "loss": 0.0897,
+      "step": 8203
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014933782333439152,
+      "loss": 0.1794,
+      "step": 8204
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014932560900086322,
+      "loss": 0.126,
+      "step": 8205
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014931339369475842,
+      "loss": 0.1114,
+      "step": 8206
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014930117741631809,
+      "loss": 0.0621,
+      "step": 8207
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014928896016578295,
+      "loss": 0.1495,
+      "step": 8208
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014927674194339401,
+      "loss": 0.083,
+      "step": 8209
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014926452274939216,
+      "loss": 0.1062,
+      "step": 8210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014925230258401828,
+      "loss": 0.2062,
+      "step": 8211
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014924008144751332,
+      "loss": 0.0906,
+      "step": 8212
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014922785934011835,
+      "loss": 0.1005,
+      "step": 8213
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014921563626207428,
+      "loss": 0.1185,
+      "step": 8214
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001492034122136221,
+      "loss": 0.0698,
+      "step": 8215
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014919118719500292,
+      "loss": 0.0858,
+      "step": 8216
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014917896120645766,
+      "loss": 0.093,
+      "step": 8217
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001491667342482275,
+      "loss": 0.1119,
+      "step": 8218
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001491545063205535,
+      "loss": 0.0452,
+      "step": 8219
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014914227742367674,
+      "loss": 0.0976,
+      "step": 8220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014913004755783834,
+      "loss": 0.097,
+      "step": 8221
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014911781672327944,
+      "loss": 0.1298,
+      "step": 8222
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014910558492024126,
+      "loss": 0.1117,
+      "step": 8223
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014909335214896493,
+      "loss": 0.1022,
+      "step": 8224
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014908111840969164,
+      "loss": 0.1067,
+      "step": 8225
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014906888370266262,
+      "loss": 0.1311,
+      "step": 8226
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014905664802811911,
+      "loss": 0.1395,
+      "step": 8227
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014904441138630237,
+      "loss": 0.0783,
+      "step": 8228
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014903217377745365,
+      "loss": 0.1005,
+      "step": 8229
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014901993520181431,
+      "loss": 0.0846,
+      "step": 8230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.001490076956596256,
+      "loss": 0.0991,
+      "step": 8231
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014899545515112888,
+      "loss": 0.1121,
+      "step": 8232
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014898321367656552,
+      "loss": 0.1114,
+      "step": 8233
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0014897097123617682,
+      "loss": 0.1126,
+      "step": 8234
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014895872783020426,
+      "loss": 0.1041,
+      "step": 8235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001489464834588892,
+      "loss": 0.0757,
+      "step": 8236
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014893423812247307,
+      "loss": 0.1478,
+      "step": 8237
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014892199182119728,
+      "loss": 0.0955,
+      "step": 8238
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014890974455530337,
+      "loss": 0.1133,
+      "step": 8239
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014889749632503281,
+      "loss": 0.1023,
+      "step": 8240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001488852471306271,
+      "loss": 0.1173,
+      "step": 8241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001488729969723277,
+      "loss": 0.0986,
+      "step": 8242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014886074585037622,
+      "loss": 0.0988,
+      "step": 8243
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014884849376501422,
+      "loss": 0.1099,
+      "step": 8244
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014883624071648325,
+      "loss": 0.1451,
+      "step": 8245
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014882398670502494,
+      "loss": 0.0803,
+      "step": 8246
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001488117317308809,
+      "loss": 0.1049,
+      "step": 8247
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014879947579429272,
+      "loss": 0.088,
+      "step": 8248
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001487872188955021,
+      "loss": 0.1144,
+      "step": 8249
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014877496103475071,
+      "loss": 0.08,
+      "step": 8250
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001487627022122803,
+      "loss": 0.1456,
+      "step": 8251
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014875044242833246,
+      "loss": 0.0865,
+      "step": 8252
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014873818168314903,
+      "loss": 0.1437,
+      "step": 8253
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001487259199769717,
+      "loss": 0.1035,
+      "step": 8254
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014871365731004226,
+      "loss": 0.098,
+      "step": 8255
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014870139368260249,
+      "loss": 0.114,
+      "step": 8256
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014868912909489422,
+      "loss": 0.121,
+      "step": 8257
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014867686354715927,
+      "loss": 0.0704,
+      "step": 8258
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014866459703963945,
+      "loss": 0.0854,
+      "step": 8259
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014865232957257668,
+      "loss": 0.0911,
+      "step": 8260
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014864006114621282,
+      "loss": 0.1375,
+      "step": 8261
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014862779176078975,
+      "loss": 0.0712,
+      "step": 8262
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014861552141654944,
+      "loss": 0.087,
+      "step": 8263
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014860325011373378,
+      "loss": 0.1248,
+      "step": 8264
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014859097785258474,
+      "loss": 0.0811,
+      "step": 8265
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001485787046333443,
+      "loss": 0.108,
+      "step": 8266
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014856643045625451,
+      "loss": 0.1099,
+      "step": 8267
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001485541553215573,
+      "loss": 0.0695,
+      "step": 8268
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014854187922949476,
+      "loss": 0.1231,
+      "step": 8269
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014852960218030894,
+      "loss": 0.1379,
+      "step": 8270
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014851732417424188,
+      "loss": 0.086,
+      "step": 8271
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001485050452115357,
+      "loss": 0.0842,
+      "step": 8272
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001484927652924325,
+      "loss": 0.1546,
+      "step": 8273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014848048441717444,
+      "loss": 0.1713,
+      "step": 8274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014846820258600357,
+      "loss": 0.0851,
+      "step": 8275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014845591979916217,
+      "loss": 0.0571,
+      "step": 8276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001484436360568924,
+      "loss": 0.0954,
+      "step": 8277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014843135135943642,
+      "loss": 0.1152,
+      "step": 8278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.001484190657070365,
+      "loss": 0.1094,
+      "step": 8279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0014840677909993482,
+      "loss": 0.0626,
+      "step": 8280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014839449153837374,
+      "loss": 0.0972,
+      "step": 8281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014838220302259542,
+      "loss": 0.1118,
+      "step": 8282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.001483699135528423,
+      "loss": 0.1288,
+      "step": 8283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014835762312935654,
+      "loss": 0.1132,
+      "step": 8284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014834533175238059,
+      "loss": 0.135,
+      "step": 8285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014833303942215674,
+      "loss": 0.095,
+      "step": 8286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014832074613892743,
+      "loss": 0.166,
+      "step": 8287
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014830845190293498,
+      "loss": 0.0837,
+      "step": 8288
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014829615671442186,
+      "loss": 0.1022,
+      "step": 8289
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014828386057363049,
+      "loss": 0.084,
+      "step": 8290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014827156348080326,
+      "loss": 0.1543,
+      "step": 8291
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014825926543618271,
+      "loss": 0.1019,
+      "step": 8292
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.001482469664400113,
+      "loss": 0.0867,
+      "step": 8293
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014823466649253154,
+      "loss": 0.1116,
+      "step": 8294
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014822236559398595,
+      "loss": 0.1493,
+      "step": 8295
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014821006374461704,
+      "loss": 0.1198,
+      "step": 8296
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.001481977609446674,
+      "loss": 0.1124,
+      "step": 8297
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014818545719437965,
+      "loss": 0.1051,
+      "step": 8298
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014817315249399635,
+      "loss": 0.12,
+      "step": 8299
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014816084684376011,
+      "loss": 0.0895,
+      "step": 8300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014814854024391355,
+      "loss": 0.1021,
+      "step": 8301
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014813623269469937,
+      "loss": 0.1542,
+      "step": 8302
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014812392419636022,
+      "loss": 0.1118,
+      "step": 8303
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014811161474913882,
+      "loss": 0.0657,
+      "step": 8304
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014809930435327784,
+      "loss": 0.0896,
+      "step": 8305
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014808699300902003,
+      "loss": 0.2001,
+      "step": 8306
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014807468071660812,
+      "loss": 0.1489,
+      "step": 8307
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.001480623674762849,
+      "loss": 0.0945,
+      "step": 8308
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014805005328829318,
+      "loss": 0.1422,
+      "step": 8309
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014803773815287569,
+      "loss": 0.0726,
+      "step": 8310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014802542207027535,
+      "loss": 0.1061,
+      "step": 8311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014801310504073486,
+      "loss": 0.0719,
+      "step": 8312
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014800078706449725,
+      "loss": 0.1235,
+      "step": 8313
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014798846814180532,
+      "loss": 0.082,
+      "step": 8314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014797614827290193,
+      "loss": 0.1069,
+      "step": 8315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014796382745803005,
+      "loss": 0.0901,
+      "step": 8316
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.001479515056974326,
+      "loss": 0.1337,
+      "step": 8317
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.001479391829913525,
+      "loss": 0.0795,
+      "step": 8318
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014792685934003282,
+      "loss": 0.1135,
+      "step": 8319
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014791453474371648,
+      "loss": 0.085,
+      "step": 8320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014790220920264648,
+      "loss": 0.0794,
+      "step": 8321
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014788988271706585,
+      "loss": 0.0963,
+      "step": 8322
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014787755528721765,
+      "loss": 0.1221,
+      "step": 8323
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00147865226913345,
+      "loss": 0.0913,
+      "step": 8324
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014785289759569088,
+      "loss": 0.0918,
+      "step": 8325
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0014784056733449847,
+      "loss": 0.0757,
+      "step": 8326
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014782823613001086,
+      "loss": 0.1256,
+      "step": 8327
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001478159039824712,
+      "loss": 0.0887,
+      "step": 8328
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014780357089212263,
+      "loss": 0.078,
+      "step": 8329
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014779123685920838,
+      "loss": 0.0959,
+      "step": 8330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001477789018839716,
+      "loss": 0.0963,
+      "step": 8331
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014776656596665553,
+      "loss": 0.125,
+      "step": 8332
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014775422910750335,
+      "loss": 0.1055,
+      "step": 8333
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014774189130675837,
+      "loss": 0.1124,
+      "step": 8334
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014772955256466384,
+      "loss": 0.1002,
+      "step": 8335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014771721288146305,
+      "loss": 0.102,
+      "step": 8336
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014770487225739932,
+      "loss": 0.0904,
+      "step": 8337
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014769253069271593,
+      "loss": 0.1345,
+      "step": 8338
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014768018818765627,
+      "loss": 0.0869,
+      "step": 8339
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001476678447424637,
+      "loss": 0.1228,
+      "step": 8340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014765550035738162,
+      "loss": 0.0635,
+      "step": 8341
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014764315503265335,
+      "loss": 0.0942,
+      "step": 8342
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001476308087685224,
+      "loss": 0.0836,
+      "step": 8343
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014761846156523214,
+      "loss": 0.0985,
+      "step": 8344
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014760611342302605,
+      "loss": 0.0977,
+      "step": 8345
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014759376434214761,
+      "loss": 0.192,
+      "step": 8346
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014758141432284035,
+      "loss": 0.0688,
+      "step": 8347
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001475690633653477,
+      "loss": 0.0714,
+      "step": 8348
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014755671146991325,
+      "loss": 0.1074,
+      "step": 8349
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014754435863678051,
+      "loss": 0.0994,
+      "step": 8350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014753200486619308,
+      "loss": 0.0658,
+      "step": 8351
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014751965015839452,
+      "loss": 0.0988,
+      "step": 8352
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014750729451362848,
+      "loss": 0.1011,
+      "step": 8353
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001474949379321385,
+      "loss": 0.0891,
+      "step": 8354
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014748258041416829,
+      "loss": 0.0767,
+      "step": 8355
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001474702219599615,
+      "loss": 0.0581,
+      "step": 8356
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014745786256976177,
+      "loss": 0.0618,
+      "step": 8357
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014744550224381281,
+      "loss": 0.1077,
+      "step": 8358
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014743314098235839,
+      "loss": 0.0524,
+      "step": 8359
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014742077878564215,
+      "loss": 0.087,
+      "step": 8360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014740841565390793,
+      "loss": 0.1049,
+      "step": 8361
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014739605158739942,
+      "loss": 0.0709,
+      "step": 8362
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014738368658636048,
+      "loss": 0.0807,
+      "step": 8363
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001473713206510349,
+      "loss": 0.0966,
+      "step": 8364
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014735895378166642,
+      "loss": 0.0955,
+      "step": 8365
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00147346585978499,
+      "loss": 0.0887,
+      "step": 8366
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014733421724177642,
+      "loss": 0.1111,
+      "step": 8367
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001473218475717426,
+      "loss": 0.0737,
+      "step": 8368
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014730947696864147,
+      "loss": 0.0856,
+      "step": 8369
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014729710543271689,
+      "loss": 0.0881,
+      "step": 8370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.001472847329642128,
+      "loss": 0.0914,
+      "step": 8371
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0014727235956337318,
+      "loss": 0.1192,
+      "step": 8372
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00147259985230442,
+      "loss": 0.1226,
+      "step": 8373
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014724760996566325,
+      "loss": 0.1035,
+      "step": 8374
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001472352337692809,
+      "loss": 0.0793,
+      "step": 8375
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00147222856641539,
+      "loss": 0.1317,
+      "step": 8376
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014721047858268165,
+      "loss": 0.1018,
+      "step": 8377
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001471980995929528,
+      "loss": 0.1648,
+      "step": 8378
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014718571967259667,
+      "loss": 0.1406,
+      "step": 8379
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014717333882185728,
+      "loss": 0.0919,
+      "step": 8380
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014716095704097871,
+      "loss": 0.1195,
+      "step": 8381
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014714857433020519,
+      "loss": 0.0665,
+      "step": 8382
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014713619068978084,
+      "loss": 0.1382,
+      "step": 8383
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001471238061199498,
+      "loss": 0.1469,
+      "step": 8384
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014711142062095628,
+      "loss": 0.172,
+      "step": 8385
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014709903419304453,
+      "loss": 0.0863,
+      "step": 8386
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014708664683645875,
+      "loss": 0.1262,
+      "step": 8387
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014707425855144316,
+      "loss": 0.0966,
+      "step": 8388
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014706186933824208,
+      "loss": 0.1074,
+      "step": 8389
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014704947919709976,
+      "loss": 0.1096,
+      "step": 8390
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014703708812826052,
+      "loss": 0.0744,
+      "step": 8391
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014702469613196863,
+      "loss": 0.0911,
+      "step": 8392
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001470123032084685,
+      "loss": 0.0989,
+      "step": 8393
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014699990935800446,
+      "loss": 0.0839,
+      "step": 8394
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014698751458082087,
+      "loss": 0.0749,
+      "step": 8395
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014697511887716212,
+      "loss": 0.1143,
+      "step": 8396
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014696272224727265,
+      "loss": 0.1161,
+      "step": 8397
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001469503246913969,
+      "loss": 0.1616,
+      "step": 8398
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014693792620977925,
+      "loss": 0.1162,
+      "step": 8399
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014692552680266426,
+      "loss": 0.0834,
+      "step": 8400
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014691312647029636,
+      "loss": 0.131,
+      "step": 8401
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014690072521292002,
+      "loss": 0.1167,
+      "step": 8402
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001468883230307798,
+      "loss": 0.0849,
+      "step": 8403
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001468759199241203,
+      "loss": 0.0732,
+      "step": 8404
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014686351589318597,
+      "loss": 0.1027,
+      "step": 8405
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014685111093822147,
+      "loss": 0.1523,
+      "step": 8406
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014683870505947131,
+      "loss": 0.2102,
+      "step": 8407
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014682629825718018,
+      "loss": 0.0849,
+      "step": 8408
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014681389053159272,
+      "loss": 0.13,
+      "step": 8409
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014680148188295351,
+      "loss": 0.1344,
+      "step": 8410
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014678907231150727,
+      "loss": 0.0826,
+      "step": 8411
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014677666181749864,
+      "loss": 0.079,
+      "step": 8412
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014676425040117233,
+      "loss": 0.0958,
+      "step": 8413
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014675183806277313,
+      "loss": 0.1029,
+      "step": 8414
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001467394248025457,
+      "loss": 0.129,
+      "step": 8415
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014672701062073484,
+      "loss": 0.0581,
+      "step": 8416
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0014671459551758533,
+      "loss": 0.0942,
+      "step": 8417
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.001467021794933419,
+      "loss": 0.151,
+      "step": 8418
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014668976254824944,
+      "loss": 0.0936,
+      "step": 8419
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014667734468255275,
+      "loss": 0.1011,
+      "step": 8420
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001466649258964967,
+      "loss": 0.1052,
+      "step": 8421
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014665250619032614,
+      "loss": 0.0962,
+      "step": 8422
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014664008556428594,
+      "loss": 0.0811,
+      "step": 8423
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014662766401862101,
+      "loss": 0.1171,
+      "step": 8424
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014661524155357628,
+      "loss": 0.1256,
+      "step": 8425
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001466028181693967,
+      "loss": 0.1109,
+      "step": 8426
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014659039386632725,
+      "loss": 0.0996,
+      "step": 8427
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014657796864461282,
+      "loss": 0.1365,
+      "step": 8428
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014656554250449844,
+      "loss": 0.1115,
+      "step": 8429
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001465531154462292,
+      "loss": 0.1498,
+      "step": 8430
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014654068747005002,
+      "loss": 0.1232,
+      "step": 8431
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014652825857620604,
+      "loss": 0.1195,
+      "step": 8432
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014651582876494228,
+      "loss": 0.0919,
+      "step": 8433
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001465033980365038,
+      "loss": 0.0786,
+      "step": 8434
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001464909663911357,
+      "loss": 0.1002,
+      "step": 8435
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001464785338290832,
+      "loss": 0.1476,
+      "step": 8436
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014646610035059134,
+      "loss": 0.1034,
+      "step": 8437
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001464536659559053,
+      "loss": 0.1682,
+      "step": 8438
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014644123064527028,
+      "loss": 0.0589,
+      "step": 8439
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014642879441893145,
+      "loss": 0.0937,
+      "step": 8440
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00146416357277134,
+      "loss": 0.1555,
+      "step": 8441
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014640391922012322,
+      "loss": 0.0992,
+      "step": 8442
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001463914802481443,
+      "loss": 0.0751,
+      "step": 8443
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014637904036144255,
+      "loss": 0.109,
+      "step": 8444
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014636659956026316,
+      "loss": 0.1002,
+      "step": 8445
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001463541578448516,
+      "loss": 0.054,
+      "step": 8446
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014634171521545306,
+      "loss": 0.093,
+      "step": 8447
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014632927167231287,
+      "loss": 0.0928,
+      "step": 8448
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014631682721567645,
+      "loss": 0.0786,
+      "step": 8449
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014630438184578913,
+      "loss": 0.0938,
+      "step": 8450
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014629193556289632,
+      "loss": 0.0801,
+      "step": 8451
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014627948836724346,
+      "loss": 0.0885,
+      "step": 8452
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001462670402590759,
+      "loss": 0.1093,
+      "step": 8453
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014625459123863917,
+      "loss": 0.0872,
+      "step": 8454
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014624214130617866,
+      "loss": 0.0956,
+      "step": 8455
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014622969046193988,
+      "loss": 0.0683,
+      "step": 8456
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014621723870616836,
+      "loss": 0.085,
+      "step": 8457
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014620478603910958,
+      "loss": 0.0826,
+      "step": 8458
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.001461923324610091,
+      "loss": 0.0677,
+      "step": 8459
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014617987797211243,
+      "loss": 0.0901,
+      "step": 8460
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014616742257266516,
+      "loss": 0.0877,
+      "step": 8461
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014615496626291292,
+      "loss": 0.09,
+      "step": 8462
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014614250904310126,
+      "loss": 0.0733,
+      "step": 8463
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0014613005091347587,
+      "loss": 0.0684,
+      "step": 8464
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001461175918742823,
+      "loss": 0.1536,
+      "step": 8465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001461051319257663,
+      "loss": 0.1074,
+      "step": 8466
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014609267106817352,
+      "loss": 0.0592,
+      "step": 8467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001460802093017496,
+      "loss": 0.1079,
+      "step": 8468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014606774662674037,
+      "loss": 0.124,
+      "step": 8469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014605528304339147,
+      "loss": 0.0563,
+      "step": 8470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014604281855194867,
+      "loss": 0.0573,
+      "step": 8471
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014603035315265773,
+      "loss": 0.0793,
+      "step": 8472
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014601788684576447,
+      "loss": 0.1176,
+      "step": 8473
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014600541963151468,
+      "loss": 0.0622,
+      "step": 8474
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014599295151015417,
+      "loss": 0.141,
+      "step": 8475
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014598048248192878,
+      "loss": 0.0531,
+      "step": 8476
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014596801254708439,
+      "loss": 0.071,
+      "step": 8477
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014595554170586684,
+      "loss": 0.1053,
+      "step": 8478
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014594306995852208,
+      "loss": 0.0828,
+      "step": 8479
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014593059730529595,
+      "loss": 0.1154,
+      "step": 8480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014591812374643443,
+      "loss": 0.1846,
+      "step": 8481
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014590564928218344,
+      "loss": 0.0961,
+      "step": 8482
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014589317391278899,
+      "loss": 0.0726,
+      "step": 8483
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00145880697638497,
+      "loss": 0.083,
+      "step": 8484
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001458682204595535,
+      "loss": 0.0967,
+      "step": 8485
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001458557423762045,
+      "loss": 0.1122,
+      "step": 8486
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014584326338869607,
+      "loss": 0.1069,
+      "step": 8487
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001458307834972742,
+      "loss": 0.1304,
+      "step": 8488
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014581830270218505,
+      "loss": 0.1173,
+      "step": 8489
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014580582100367466,
+      "loss": 0.1033,
+      "step": 8490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014579333840198912,
+      "loss": 0.0919,
+      "step": 8491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014578085489737457,
+      "loss": 0.1051,
+      "step": 8492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014576837049007714,
+      "loss": 0.1188,
+      "step": 8493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014575588518034304,
+      "loss": 0.0875,
+      "step": 8494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001457433989684184,
+      "loss": 0.0873,
+      "step": 8495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014573091185454945,
+      "loss": 0.1508,
+      "step": 8496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014571842383898238,
+      "loss": 0.0961,
+      "step": 8497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014570593492196342,
+      "loss": 0.1115,
+      "step": 8498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014569344510373882,
+      "loss": 0.1293,
+      "step": 8499
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014568095438455488,
+      "loss": 0.1149,
+      "step": 8500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014566846276465783,
+      "loss": 0.0909,
+      "step": 8501
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014565597024429407,
+      "loss": 0.0927,
+      "step": 8502
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001456434768237098,
+      "loss": 0.1088,
+      "step": 8503
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014563098250315143,
+      "loss": 0.089,
+      "step": 8504
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001456184872828653,
+      "loss": 0.1044,
+      "step": 8505
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.001456059911630978,
+      "loss": 0.0679,
+      "step": 8506
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014559349414409533,
+      "loss": 0.0745,
+      "step": 8507
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014558099622610423,
+      "loss": 0.0797,
+      "step": 8508
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014556849740937096,
+      "loss": 0.0969,
+      "step": 8509
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0014555599769414203,
+      "loss": 0.1368,
+      "step": 8510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014554349708066382,
+      "loss": 0.1085,
+      "step": 8511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001455309955691829,
+      "loss": 0.0853,
+      "step": 8512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014551849315994569,
+      "loss": 0.1399,
+      "step": 8513
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014550598985319867,
+      "loss": 0.1196,
+      "step": 8514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014549348564918848,
+      "loss": 0.1514,
+      "step": 8515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014548098054816161,
+      "loss": 0.1355,
+      "step": 8516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014546847455036468,
+      "loss": 0.1043,
+      "step": 8517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014545596765604418,
+      "loss": 0.1165,
+      "step": 8518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001454434598654468,
+      "loss": 0.1226,
+      "step": 8519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014543095117881912,
+      "loss": 0.101,
+      "step": 8520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001454184415964078,
+      "loss": 0.129,
+      "step": 8521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014540593111845952,
+      "loss": 0.1301,
+      "step": 8522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014539341974522091,
+      "loss": 0.1543,
+      "step": 8523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014538090747693868,
+      "loss": 0.1067,
+      "step": 8524
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001453683943138595,
+      "loss": 0.1118,
+      "step": 8525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014535588025623022,
+      "loss": 0.1267,
+      "step": 8526
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014534336530429746,
+      "loss": 0.1506,
+      "step": 8527
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014533084945830802,
+      "loss": 0.1068,
+      "step": 8528
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001453183327185087,
+      "loss": 0.2101,
+      "step": 8529
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014530581508514627,
+      "loss": 0.1202,
+      "step": 8530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014529329655846755,
+      "loss": 0.1302,
+      "step": 8531
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014528077713871943,
+      "loss": 0.0722,
+      "step": 8532
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001452682568261487,
+      "loss": 0.0707,
+      "step": 8533
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014525573562100226,
+      "loss": 0.0855,
+      "step": 8534
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014524321352352696,
+      "loss": 0.1387,
+      "step": 8535
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001452306905339697,
+      "loss": 0.0756,
+      "step": 8536
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001452181666525775,
+      "loss": 0.0864,
+      "step": 8537
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014520564187959718,
+      "loss": 0.0981,
+      "step": 8538
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014519311621527576,
+      "loss": 0.105,
+      "step": 8539
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014518058965986018,
+      "loss": 0.0667,
+      "step": 8540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014516806221359748,
+      "loss": 0.1228,
+      "step": 8541
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014515553387673464,
+      "loss": 0.0591,
+      "step": 8542
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001451430046495187,
+      "loss": 0.0892,
+      "step": 8543
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014513047453219665,
+      "loss": 0.0824,
+      "step": 8544
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014511794352501562,
+      "loss": 0.0891,
+      "step": 8545
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014510541162822267,
+      "loss": 0.1201,
+      "step": 8546
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014509287884206488,
+      "loss": 0.056,
+      "step": 8547
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014508034516678937,
+      "loss": 0.123,
+      "step": 8548
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014506781060264328,
+      "loss": 0.0763,
+      "step": 8549
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001450552751498738,
+      "loss": 0.0945,
+      "step": 8550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014504273880872802,
+      "loss": 0.0847,
+      "step": 8551
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014503020157945318,
+      "loss": 0.1046,
+      "step": 8552
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014501766346229645,
+      "loss": 0.106,
+      "step": 8553
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.001450051244575051,
+      "loss": 0.0781,
+      "step": 8554
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014499258456532628,
+      "loss": 0.1185,
+      "step": 8555
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014498004378600733,
+      "loss": 0.0883,
+      "step": 8556
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0014496750211979547,
+      "loss": 0.1231,
+      "step": 8557
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014495495956693802,
+      "loss": 0.1001,
+      "step": 8558
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001449424161276823,
+      "loss": 0.0786,
+      "step": 8559
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001449298718022756,
+      "loss": 0.0715,
+      "step": 8560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014491732659096526,
+      "loss": 0.0739,
+      "step": 8561
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014490478049399864,
+      "loss": 0.0784,
+      "step": 8562
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014489223351162317,
+      "loss": 0.0984,
+      "step": 8563
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001448796856440862,
+      "loss": 0.0968,
+      "step": 8564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014486713689163514,
+      "loss": 0.0724,
+      "step": 8565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014485458725451744,
+      "loss": 0.1246,
+      "step": 8566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001448420367329805,
+      "loss": 0.0597,
+      "step": 8567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014482948532727185,
+      "loss": 0.0911,
+      "step": 8568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014481693303763895,
+      "loss": 0.0778,
+      "step": 8569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014480437986432928,
+      "loss": 0.0948,
+      "step": 8570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014479182580759036,
+      "loss": 0.1001,
+      "step": 8571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014477927086766974,
+      "loss": 0.0999,
+      "step": 8572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014476671504481497,
+      "loss": 0.097,
+      "step": 8573
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014475415833927364,
+      "loss": 0.1086,
+      "step": 8574
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014474160075129325,
+      "loss": 0.0807,
+      "step": 8575
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014472904228112152,
+      "loss": 0.0963,
+      "step": 8576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00144716482929006,
+      "loss": 0.1195,
+      "step": 8577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014470392269519431,
+      "loss": 0.133,
+      "step": 8578
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014469136157993418,
+      "loss": 0.1094,
+      "step": 8579
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014467879958347325,
+      "loss": 0.1143,
+      "step": 8580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014466623670605916,
+      "loss": 0.1589,
+      "step": 8581
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014465367294793972,
+      "loss": 0.1165,
+      "step": 8582
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014464110830936256,
+      "loss": 0.0834,
+      "step": 8583
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001446285427905755,
+      "loss": 0.0841,
+      "step": 8584
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001446159763918262,
+      "loss": 0.1315,
+      "step": 8585
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014460340911336255,
+      "loss": 0.072,
+      "step": 8586
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014459084095543228,
+      "loss": 0.0887,
+      "step": 8587
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001445782719182832,
+      "loss": 0.0956,
+      "step": 8588
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014456570200216316,
+      "loss": 0.0685,
+      "step": 8589
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014455313120732005,
+      "loss": 0.151,
+      "step": 8590
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014454055953400165,
+      "loss": 0.0905,
+      "step": 8591
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014452798698245588,
+      "loss": 0.0861,
+      "step": 8592
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014451541355293065,
+      "loss": 0.0895,
+      "step": 8593
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014450283924567384,
+      "loss": 0.0917,
+      "step": 8594
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014449026406093343,
+      "loss": 0.1006,
+      "step": 8595
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014447768799895734,
+      "loss": 0.071,
+      "step": 8596
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014446511105999358,
+      "loss": 0.0702,
+      "step": 8597
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014445253324429006,
+      "loss": 0.0999,
+      "step": 8598
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014443995455209484,
+      "loss": 0.088,
+      "step": 8599
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014442737498365594,
+      "loss": 0.1187,
+      "step": 8600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014441479453922135,
+      "loss": 0.0712,
+      "step": 8601
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.001444022132190392,
+      "loss": 0.0999,
+      "step": 8602
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0014438963102335752,
+      "loss": 0.0761,
+      "step": 8603
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014437704795242437,
+      "loss": 0.1044,
+      "step": 8604
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001443644640064879,
+      "loss": 0.058,
+      "step": 8605
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001443518791857962,
+      "loss": 0.0906,
+      "step": 8606
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014433929349059746,
+      "loss": 0.1173,
+      "step": 8607
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001443267069211398,
+      "loss": 0.0934,
+      "step": 8608
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014431411947767136,
+      "loss": 0.1256,
+      "step": 8609
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014430153116044041,
+      "loss": 0.0773,
+      "step": 8610
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014428894196969514,
+      "loss": 0.0944,
+      "step": 8611
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014427635190568376,
+      "loss": 0.1379,
+      "step": 8612
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001442637609686545,
+      "loss": 0.0901,
+      "step": 8613
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014425116915885562,
+      "loss": 0.0712,
+      "step": 8614
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014423857647653544,
+      "loss": 0.0597,
+      "step": 8615
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014422598292194225,
+      "loss": 0.1132,
+      "step": 8616
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014421338849532435,
+      "loss": 0.118,
+      "step": 8617
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014420079319693003,
+      "loss": 0.0703,
+      "step": 8618
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001441881970270077,
+      "loss": 0.0951,
+      "step": 8619
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014417559998580569,
+      "loss": 0.0983,
+      "step": 8620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014416300207357239,
+      "loss": 0.0797,
+      "step": 8621
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014415040329055619,
+      "loss": 0.0945,
+      "step": 8622
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014413780363700555,
+      "loss": 0.0911,
+      "step": 8623
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014412520311316887,
+      "loss": 0.0818,
+      "step": 8624
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014411260171929455,
+      "loss": 0.1422,
+      "step": 8625
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014409999945563113,
+      "loss": 0.1132,
+      "step": 8626
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014408739632242708,
+      "loss": 0.0883,
+      "step": 8627
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014407479231993087,
+      "loss": 0.0737,
+      "step": 8628
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014406218744839109,
+      "loss": 0.0933,
+      "step": 8629
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001440495817080562,
+      "loss": 0.0625,
+      "step": 8630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014403697509917475,
+      "loss": 0.1201,
+      "step": 8631
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001440243676219954,
+      "loss": 0.1299,
+      "step": 8632
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014401175927676666,
+      "loss": 0.0797,
+      "step": 8633
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014399915006373717,
+      "loss": 0.1216,
+      "step": 8634
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014398653998315553,
+      "loss": 0.1167,
+      "step": 8635
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014397392903527035,
+      "loss": 0.0796,
+      "step": 8636
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014396131722033037,
+      "loss": 0.0729,
+      "step": 8637
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014394870453858417,
+      "loss": 0.0983,
+      "step": 8638
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014393609099028054,
+      "loss": 0.1158,
+      "step": 8639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014392347657566811,
+      "loss": 0.1829,
+      "step": 8640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001439108612949956,
+      "loss": 0.123,
+      "step": 8641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001438982451485118,
+      "loss": 0.1176,
+      "step": 8642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014388562813646545,
+      "loss": 0.1128,
+      "step": 8643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014387301025910535,
+      "loss": 0.1061,
+      "step": 8644
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014386039151668023,
+      "loss": 0.0833,
+      "step": 8645
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014384777190943894,
+      "loss": 0.1128,
+      "step": 8646
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.001438351514376303,
+      "loss": 0.0929,
+      "step": 8647
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014382253010150317,
+      "loss": 0.1481,
+      "step": 8648
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0014380990790130639,
+      "loss": 0.1263,
+      "step": 8649
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014379728483728885,
+      "loss": 0.0569,
+      "step": 8650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014378466090969942,
+      "loss": 0.1746,
+      "step": 8651
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014377203611878704,
+      "loss": 0.0984,
+      "step": 8652
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014375941046480066,
+      "loss": 0.1464,
+      "step": 8653
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014374678394798916,
+      "loss": 0.072,
+      "step": 8654
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014373415656860156,
+      "loss": 0.0795,
+      "step": 8655
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014372152832688684,
+      "loss": 0.0956,
+      "step": 8656
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014370889922309393,
+      "loss": 0.1313,
+      "step": 8657
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014369626925747193,
+      "loss": 0.0874,
+      "step": 8658
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001436836384302698,
+      "loss": 0.1027,
+      "step": 8659
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014367100674173665,
+      "loss": 0.054,
+      "step": 8660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001436583741921215,
+      "loss": 0.073,
+      "step": 8661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014364574078167345,
+      "loss": 0.0806,
+      "step": 8662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014363310651064158,
+      "loss": 0.1016,
+      "step": 8663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014362047137927506,
+      "loss": 0.0791,
+      "step": 8664
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014360783538782296,
+      "loss": 0.0806,
+      "step": 8665
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014359519853653446,
+      "loss": 0.0555,
+      "step": 8666
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014358256082565872,
+      "loss": 0.1014,
+      "step": 8667
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001435699222554449,
+      "loss": 0.0742,
+      "step": 8668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014355728282614227,
+      "loss": 0.1176,
+      "step": 8669
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014354464253799998,
+      "loss": 0.0492,
+      "step": 8670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001435320013912673,
+      "loss": 0.0687,
+      "step": 8671
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014351935938619346,
+      "loss": 0.1146,
+      "step": 8672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014350671652302773,
+      "loss": 0.1089,
+      "step": 8673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014349407280201943,
+      "loss": 0.0729,
+      "step": 8674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001434814282234178,
+      "loss": 0.0645,
+      "step": 8675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014346878278747224,
+      "loss": 0.0527,
+      "step": 8676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014345613649443202,
+      "loss": 0.1091,
+      "step": 8677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001434434893445465,
+      "loss": 0.0529,
+      "step": 8678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014343084133806505,
+      "loss": 0.1504,
+      "step": 8679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014341819247523713,
+      "loss": 0.0676,
+      "step": 8680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014340554275631204,
+      "loss": 0.113,
+      "step": 8681
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014339289218153924,
+      "loss": 0.0576,
+      "step": 8682
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001433802407511682,
+      "loss": 0.1551,
+      "step": 8683
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014336758846544833,
+      "loss": 0.0829,
+      "step": 8684
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001433549353246291,
+      "loss": 0.1503,
+      "step": 8685
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014334228132896005,
+      "loss": 0.0988,
+      "step": 8686
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014332962647869065,
+      "loss": 0.1492,
+      "step": 8687
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001433169707740704,
+      "loss": 0.1028,
+      "step": 8688
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014330431421534886,
+      "loss": 0.1577,
+      "step": 8689
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014329165680277561,
+      "loss": 0.067,
+      "step": 8690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014327899853660015,
+      "loss": 0.082,
+      "step": 8691
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014326633941707217,
+      "loss": 0.1054,
+      "step": 8692
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.001432536794444412,
+      "loss": 0.0897,
+      "step": 8693
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014324101861895688,
+      "loss": 0.074,
+      "step": 8694
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0014322835694086887,
+      "loss": 0.0665,
+      "step": 8695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.001432156944104268,
+      "loss": 0.0966,
+      "step": 8696
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014320303102788037,
+      "loss": 0.0795,
+      "step": 8697
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014319036679347923,
+      "loss": 0.0765,
+      "step": 8698
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014317770170747316,
+      "loss": 0.1163,
+      "step": 8699
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014316503577011177,
+      "loss": 0.0933,
+      "step": 8700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.001431523689816449,
+      "loss": 0.1185,
+      "step": 8701
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014313970134232227,
+      "loss": 0.0797,
+      "step": 8702
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014312703285239365,
+      "loss": 0.0714,
+      "step": 8703
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014311436351210885,
+      "loss": 0.1048,
+      "step": 8704
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014310169332171765,
+      "loss": 0.1064,
+      "step": 8705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.001430890222814699,
+      "loss": 0.0983,
+      "step": 8706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014307635039161543,
+      "loss": 0.1262,
+      "step": 8707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.001430636776524041,
+      "loss": 0.1174,
+      "step": 8708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014305100406408577,
+      "loss": 0.2144,
+      "step": 8709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014303832962691031,
+      "loss": 0.0585,
+      "step": 8710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.001430256543411277,
+      "loss": 0.0658,
+      "step": 8711
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014301297820698783,
+      "loss": 0.1165,
+      "step": 8712
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014300030122474064,
+      "loss": 0.0811,
+      "step": 8713
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014298762339463606,
+      "loss": 0.0863,
+      "step": 8714
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014297494471692412,
+      "loss": 0.0834,
+      "step": 8715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014296226519185473,
+      "loss": 0.0919,
+      "step": 8716
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014294958481967802,
+      "loss": 0.1456,
+      "step": 8717
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014293690360064387,
+      "loss": 0.0746,
+      "step": 8718
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014292422153500243,
+      "loss": 0.1238,
+      "step": 8719
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014291153862300374,
+      "loss": 0.1455,
+      "step": 8720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014289885486489784,
+      "loss": 0.0865,
+      "step": 8721
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014288617026093485,
+      "loss": 0.0886,
+      "step": 8722
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014287348481136486,
+      "loss": 0.0978,
+      "step": 8723
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00142860798516438,
+      "loss": 0.1294,
+      "step": 8724
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014284811137640443,
+      "loss": 0.1173,
+      "step": 8725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014283542339151428,
+      "loss": 0.1156,
+      "step": 8726
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014282273456201775,
+      "loss": 0.0721,
+      "step": 8727
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00142810044888165,
+      "loss": 0.0599,
+      "step": 8728
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014279735437020629,
+      "loss": 0.0688,
+      "step": 8729
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014278466300839182,
+      "loss": 0.1019,
+      "step": 8730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014277197080297178,
+      "loss": 0.081,
+      "step": 8731
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014275927775419647,
+      "loss": 0.1014,
+      "step": 8732
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014274658386231622,
+      "loss": 0.1104,
+      "step": 8733
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014273388912758125,
+      "loss": 0.0874,
+      "step": 8734
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014272119355024189,
+      "loss": 0.0798,
+      "step": 8735
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014270849713054847,
+      "loss": 0.0856,
+      "step": 8736
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.001426957998687513,
+      "loss": 0.0768,
+      "step": 8737
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014268310176510078,
+      "loss": 0.1351,
+      "step": 8738
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014267040281984726,
+      "loss": 0.1007,
+      "step": 8739
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014265770303324115,
+      "loss": 0.1548,
+      "step": 8740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0014264500240553284,
+      "loss": 0.0717,
+      "step": 8741
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014263230093697274,
+      "loss": 0.0826,
+      "step": 8742
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014261959862781133,
+      "loss": 0.1227,
+      "step": 8743
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014260689547829905,
+      "loss": 0.1426,
+      "step": 8744
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014259419148868638,
+      "loss": 0.0864,
+      "step": 8745
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001425814866592238,
+      "loss": 0.1174,
+      "step": 8746
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001425687809901618,
+      "loss": 0.067,
+      "step": 8747
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014255607448175093,
+      "loss": 0.0875,
+      "step": 8748
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014254336713424172,
+      "loss": 0.1277,
+      "step": 8749
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014253065894788476,
+      "loss": 0.0735,
+      "step": 8750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014251794992293056,
+      "loss": 0.0864,
+      "step": 8751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014250524005962977,
+      "loss": 0.2159,
+      "step": 8752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014249252935823295,
+      "loss": 0.0656,
+      "step": 8753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014247981781899076,
+      "loss": 0.0947,
+      "step": 8754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014246710544215382,
+      "loss": 0.0507,
+      "step": 8755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001424543922279728,
+      "loss": 0.1625,
+      "step": 8756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014244167817669836,
+      "loss": 0.0933,
+      "step": 8757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014242896328858117,
+      "loss": 0.0704,
+      "step": 8758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014241624756387195,
+      "loss": 0.1138,
+      "step": 8759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014240353100282146,
+      "loss": 0.0881,
+      "step": 8760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014239081360568036,
+      "loss": 0.0908,
+      "step": 8761
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001423780953726995,
+      "loss": 0.0692,
+      "step": 8762
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014236537630412957,
+      "loss": 0.0907,
+      "step": 8763
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001423526564002214,
+      "loss": 0.1281,
+      "step": 8764
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001423399356612258,
+      "loss": 0.079,
+      "step": 8765
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014232721408739357,
+      "loss": 0.1013,
+      "step": 8766
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014231449167897558,
+      "loss": 0.0927,
+      "step": 8767
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001423017684362226,
+      "loss": 0.0896,
+      "step": 8768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014228904435938557,
+      "loss": 0.1158,
+      "step": 8769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014227631944871539,
+      "loss": 0.0919,
+      "step": 8770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014226359370446292,
+      "loss": 0.0687,
+      "step": 8771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014225086712687908,
+      "loss": 0.0982,
+      "step": 8772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014223813971621486,
+      "loss": 0.0975,
+      "step": 8773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014222541147272116,
+      "loss": 0.1259,
+      "step": 8774
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014221268239664896,
+      "loss": 0.0746,
+      "step": 8775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014219995248824927,
+      "loss": 0.0754,
+      "step": 8776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014218722174777305,
+      "loss": 0.0858,
+      "step": 8777
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014217449017547132,
+      "loss": 0.0832,
+      "step": 8778
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001421617577715952,
+      "loss": 0.0616,
+      "step": 8779
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001421490245363956,
+      "loss": 0.1182,
+      "step": 8780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001421362904701237,
+      "loss": 0.0886,
+      "step": 8781
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014212355557303052,
+      "loss": 0.1306,
+      "step": 8782
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.001421108198453672,
+      "loss": 0.1198,
+      "step": 8783
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014209808328738485,
+      "loss": 0.1063,
+      "step": 8784
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014208534589933456,
+      "loss": 0.0973,
+      "step": 8785
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014207260768146757,
+      "loss": 0.0897,
+      "step": 8786
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0014205986863403495,
+      "loss": 0.0633,
+      "step": 8787
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014204712875728793,
+      "loss": 0.1068,
+      "step": 8788
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001420343880514777,
+      "loss": 0.1234,
+      "step": 8789
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014202164651685547,
+      "loss": 0.153,
+      "step": 8790
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014200890415367246,
+      "loss": 0.0916,
+      "step": 8791
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014199616096217996,
+      "loss": 0.0661,
+      "step": 8792
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001419834169426292,
+      "loss": 0.1891,
+      "step": 8793
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014197067209527144,
+      "loss": 0.0999,
+      "step": 8794
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014195792642035803,
+      "loss": 0.0645,
+      "step": 8795
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014194517991814024,
+      "loss": 0.1049,
+      "step": 8796
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001419324325888694,
+      "loss": 0.1107,
+      "step": 8797
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014191968443279688,
+      "loss": 0.1465,
+      "step": 8798
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014190693545017403,
+      "loss": 0.1064,
+      "step": 8799
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014189418564125222,
+      "loss": 0.0941,
+      "step": 8800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014188143500628286,
+      "loss": 0.0743,
+      "step": 8801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014186868354551735,
+      "loss": 0.0929,
+      "step": 8802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014185593125920716,
+      "loss": 0.0906,
+      "step": 8803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014184317814760362,
+      "loss": 0.1571,
+      "step": 8804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001418304242109583,
+      "loss": 0.1165,
+      "step": 8805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001418176694495226,
+      "loss": 0.1152,
+      "step": 8806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014180491386354807,
+      "loss": 0.0999,
+      "step": 8807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001417921574532862,
+      "loss": 0.1245,
+      "step": 8808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014177940021898852,
+      "loss": 0.0881,
+      "step": 8809
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014176664216090655,
+      "loss": 0.1085,
+      "step": 8810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014175388327929185,
+      "loss": 0.0782,
+      "step": 8811
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014174112357439599,
+      "loss": 0.0854,
+      "step": 8812
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001417283630464706,
+      "loss": 0.0865,
+      "step": 8813
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001417156016957672,
+      "loss": 0.082,
+      "step": 8814
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001417028395225375,
+      "loss": 0.0811,
+      "step": 8815
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014169007652703313,
+      "loss": 0.0765,
+      "step": 8816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014167731270950568,
+      "loss": 0.067,
+      "step": 8817
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014166454807020687,
+      "loss": 0.0804,
+      "step": 8818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014165178260938836,
+      "loss": 0.0999,
+      "step": 8819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001416390163273019,
+      "loss": 0.0888,
+      "step": 8820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014162624922419913,
+      "loss": 0.1006,
+      "step": 8821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014161348130033184,
+      "loss": 0.0737,
+      "step": 8822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.001416007125559518,
+      "loss": 0.1536,
+      "step": 8823
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014158794299131072,
+      "loss": 0.0841,
+      "step": 8824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014157517260666044,
+      "loss": 0.0741,
+      "step": 8825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014156240140225272,
+      "loss": 0.1214,
+      "step": 8826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014154962937833937,
+      "loss": 0.0969,
+      "step": 8827
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014153685653517224,
+      "loss": 0.0923,
+      "step": 8828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014152408287300322,
+      "loss": 0.085,
+      "step": 8829
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014151130839208412,
+      "loss": 0.0885,
+      "step": 8830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014149853309266681,
+      "loss": 0.113,
+      "step": 8831
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014148575697500324,
+      "loss": 0.1124,
+      "step": 8832
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0014147298003934525,
+      "loss": 0.0766,
+      "step": 8833
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014146020228594483,
+      "loss": 0.1473,
+      "step": 8834
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014144742371505392,
+      "loss": 0.1343,
+      "step": 8835
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014143464432692448,
+      "loss": 0.1425,
+      "step": 8836
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014142186412180846,
+      "loss": 0.1129,
+      "step": 8837
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014140908309995784,
+      "loss": 0.084,
+      "step": 8838
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001413963012616247,
+      "loss": 0.0701,
+      "step": 8839
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014138351860706103,
+      "loss": 0.1866,
+      "step": 8840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014137073513651882,
+      "loss": 0.08,
+      "step": 8841
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014135795085025021,
+      "loss": 0.0801,
+      "step": 8842
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001413451657485072,
+      "loss": 0.1279,
+      "step": 8843
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014133237983154194,
+      "loss": 0.092,
+      "step": 8844
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001413195930996065,
+      "loss": 0.0845,
+      "step": 8845
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014130680555295306,
+      "loss": 0.0938,
+      "step": 8846
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001412940171918337,
+      "loss": 0.088,
+      "step": 8847
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014128122801650056,
+      "loss": 0.1105,
+      "step": 8848
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014126843802720584,
+      "loss": 0.1068,
+      "step": 8849
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014125564722420171,
+      "loss": 0.0582,
+      "step": 8850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014124285560774043,
+      "loss": 0.1083,
+      "step": 8851
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014123006317807415,
+      "loss": 0.141,
+      "step": 8852
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014121726993545514,
+      "loss": 0.0889,
+      "step": 8853
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014120447588013558,
+      "loss": 0.1686,
+      "step": 8854
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014119168101236787,
+      "loss": 0.0764,
+      "step": 8855
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014117888533240423,
+      "loss": 0.1346,
+      "step": 8856
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001411660888404969,
+      "loss": 0.0989,
+      "step": 8857
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014115329153689825,
+      "loss": 0.1105,
+      "step": 8858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014114049342186057,
+      "loss": 0.0681,
+      "step": 8859
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014112769449563629,
+      "loss": 0.1204,
+      "step": 8860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001411148947584777,
+      "loss": 0.1042,
+      "step": 8861
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014110209421063723,
+      "loss": 0.1316,
+      "step": 8862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014108929285236722,
+      "loss": 0.0652,
+      "step": 8863
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014107649068392009,
+      "loss": 0.0999,
+      "step": 8864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014106368770554828,
+      "loss": 0.1419,
+      "step": 8865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014105088391750424,
+      "loss": 0.0972,
+      "step": 8866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014103807932004044,
+      "loss": 0.0844,
+      "step": 8867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014102527391340932,
+      "loss": 0.1001,
+      "step": 8868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001410124676978634,
+      "loss": 0.093,
+      "step": 8869
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014099966067365516,
+      "loss": 0.1136,
+      "step": 8870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014098685284103711,
+      "loss": 0.1576,
+      "step": 8871
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014097404420026186,
+      "loss": 0.1519,
+      "step": 8872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014096123475158188,
+      "loss": 0.0867,
+      "step": 8873
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001409484244952498,
+      "loss": 0.0812,
+      "step": 8874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014093561343151817,
+      "loss": 0.106,
+      "step": 8875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014092280156063962,
+      "loss": 0.1511,
+      "step": 8876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014090998888286674,
+      "loss": 0.0975,
+      "step": 8877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014089717539845217,
+      "loss": 0.0575,
+      "step": 8878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.001408843611076486,
+      "loss": 0.1142,
+      "step": 8879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0014087154601070862,
+      "loss": 0.1281,
+      "step": 8880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014085873010788496,
+      "loss": 0.0589,
+      "step": 8881
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014084591339943032,
+      "loss": 0.0961,
+      "step": 8882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001408330958855974,
+      "loss": 0.1144,
+      "step": 8883
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014082027756663893,
+      "loss": 0.1127,
+      "step": 8884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014080745844280767,
+      "loss": 0.1002,
+      "step": 8885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014079463851435635,
+      "loss": 0.0884,
+      "step": 8886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014078181778153778,
+      "loss": 0.0939,
+      "step": 8887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001407689962446047,
+      "loss": 0.1136,
+      "step": 8888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014075617390381,
+      "loss": 0.0864,
+      "step": 8889
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014074335075940643,
+      "loss": 0.0737,
+      "step": 8890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014073052681164687,
+      "loss": 0.1006,
+      "step": 8891
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014071770206078415,
+      "loss": 0.1523,
+      "step": 8892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014070487650707118,
+      "loss": 0.1048,
+      "step": 8893
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001406920501507608,
+      "loss": 0.0721,
+      "step": 8894
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014067922299210596,
+      "loss": 0.1027,
+      "step": 8895
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014066639503135954,
+      "loss": 0.1055,
+      "step": 8896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001406535662687745,
+      "loss": 0.1029,
+      "step": 8897
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014064073670460378,
+      "loss": 0.1198,
+      "step": 8898
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014062790633910037,
+      "loss": 0.1426,
+      "step": 8899
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014061507517251722,
+      "loss": 0.1082,
+      "step": 8900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014060224320510731,
+      "loss": 0.0831,
+      "step": 8901
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001405894104371237,
+      "loss": 0.0963,
+      "step": 8902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014057657686881943,
+      "loss": 0.104,
+      "step": 8903
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014056374250044747,
+      "loss": 0.0824,
+      "step": 8904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014055090733226096,
+      "loss": 0.1013,
+      "step": 8905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014053807136451294,
+      "loss": 0.0901,
+      "step": 8906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014052523459745652,
+      "loss": 0.0906,
+      "step": 8907
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001405123970313448,
+      "loss": 0.0856,
+      "step": 8908
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001404995586664309,
+      "loss": 0.1109,
+      "step": 8909
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014048671950296797,
+      "loss": 0.1332,
+      "step": 8910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014047387954120914,
+      "loss": 0.0759,
+      "step": 8911
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001404610387814076,
+      "loss": 0.0669,
+      "step": 8912
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014044819722381656,
+      "loss": 0.1029,
+      "step": 8913
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014043535486868918,
+      "loss": 0.1162,
+      "step": 8914
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014042251171627873,
+      "loss": 0.1108,
+      "step": 8915
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001404096677668384,
+      "loss": 0.1249,
+      "step": 8916
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014039682302062143,
+      "loss": 0.0857,
+      "step": 8917
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014038397747788115,
+      "loss": 0.0794,
+      "step": 8918
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014037113113887078,
+      "loss": 0.0887,
+      "step": 8919
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014035828400384366,
+      "loss": 0.0981,
+      "step": 8920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014034543607305302,
+      "loss": 0.1324,
+      "step": 8921
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014033258734675233,
+      "loss": 0.0875,
+      "step": 8922
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014031973782519483,
+      "loss": 0.1035,
+      "step": 8923
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014030688750863388,
+      "loss": 0.124,
+      "step": 8924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0014029403639732292,
+      "loss": 0.1309,
+      "step": 8925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.001402811844915153,
+      "loss": 0.1265,
+      "step": 8926
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014026833179146443,
+      "loss": 0.1208,
+      "step": 8927
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014025547829742372,
+      "loss": 0.1101,
+      "step": 8928
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001402426240096466,
+      "loss": 0.1073,
+      "step": 8929
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001402297689283866,
+      "loss": 0.0804,
+      "step": 8930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014021691305389707,
+      "loss": 0.1404,
+      "step": 8931
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014020405638643161,
+      "loss": 0.1224,
+      "step": 8932
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014019119892624366,
+      "loss": 0.1201,
+      "step": 8933
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014017834067358672,
+      "loss": 0.0792,
+      "step": 8934
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001401654816287144,
+      "loss": 0.1379,
+      "step": 8935
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014015262179188019,
+      "loss": 0.0842,
+      "step": 8936
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014013976116333767,
+      "loss": 0.0887,
+      "step": 8937
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014012689974334039,
+      "loss": 0.1064,
+      "step": 8938
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014011403753214195,
+      "loss": 0.1322,
+      "step": 8939
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00140101174529996,
+      "loss": 0.0732,
+      "step": 8940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014008831073715614,
+      "loss": 0.0838,
+      "step": 8941
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014007544615387603,
+      "loss": 0.0856,
+      "step": 8942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001400625807804093,
+      "loss": 0.0956,
+      "step": 8943
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001400497146170096,
+      "loss": 0.1334,
+      "step": 8944
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001400368476639307,
+      "loss": 0.0895,
+      "step": 8945
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014002397992142624,
+      "loss": 0.1398,
+      "step": 8946
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0014001111138974996,
+      "loss": 0.0957,
+      "step": 8947
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013999824206915557,
+      "loss": 0.0737,
+      "step": 8948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013998537195989685,
+      "loss": 0.0624,
+      "step": 8949
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013997250106222757,
+      "loss": 0.0828,
+      "step": 8950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013995962937640148,
+      "loss": 0.0892,
+      "step": 8951
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013994675690267243,
+      "loss": 0.0709,
+      "step": 8952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013993388364129417,
+      "loss": 0.0801,
+      "step": 8953
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013992100959252056,
+      "loss": 0.1392,
+      "step": 8954
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013990813475660542,
+      "loss": 0.1011,
+      "step": 8955
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013989525913380265,
+      "loss": 0.1276,
+      "step": 8956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013988238272436614,
+      "loss": 0.066,
+      "step": 8957
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013986950552854968,
+      "loss": 0.099,
+      "step": 8958
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013985662754660729,
+      "loss": 0.1326,
+      "step": 8959
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013984374877879283,
+      "loss": 0.0862,
+      "step": 8960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013983086922536021,
+      "loss": 0.0931,
+      "step": 8961
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013981798888656348,
+      "loss": 0.0796,
+      "step": 8962
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001398051077626565,
+      "loss": 0.0795,
+      "step": 8963
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013979222585389333,
+      "loss": 0.1351,
+      "step": 8964
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013977934316052793,
+      "loss": 0.0957,
+      "step": 8965
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013976645968281432,
+      "loss": 0.1351,
+      "step": 8966
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013975357542100656,
+      "loss": 0.0541,
+      "step": 8967
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013974069037535864,
+      "loss": 0.0782,
+      "step": 8968
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013972780454612468,
+      "loss": 0.1296,
+      "step": 8969
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001397149179335587,
+      "loss": 0.0824,
+      "step": 8970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.001397020305379148,
+      "loss": 0.0848,
+      "step": 8971
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0013968914235944715,
+      "loss": 0.1455,
+      "step": 8972
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013967625339840982,
+      "loss": 0.1142,
+      "step": 8973
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013966336365505692,
+      "loss": 0.0908,
+      "step": 8974
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013965047312964266,
+      "loss": 0.1017,
+      "step": 8975
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001396375818224212,
+      "loss": 0.1497,
+      "step": 8976
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013962468973364671,
+      "loss": 0.0876,
+      "step": 8977
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013961179686357336,
+      "loss": 0.095,
+      "step": 8978
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013959890321245541,
+      "loss": 0.0754,
+      "step": 8979
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013958600878054709,
+      "loss": 0.124,
+      "step": 8980
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001395731135681026,
+      "loss": 0.1295,
+      "step": 8981
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013956021757537621,
+      "loss": 0.1069,
+      "step": 8982
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001395473208026223,
+      "loss": 0.1537,
+      "step": 8983
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013953442325009503,
+      "loss": 0.1215,
+      "step": 8984
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013952152491804874,
+      "loss": 0.0742,
+      "step": 8985
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013950862580673783,
+      "loss": 0.0798,
+      "step": 8986
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013949572591641648,
+      "loss": 0.1043,
+      "step": 8987
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001394828252473392,
+      "loss": 0.1035,
+      "step": 8988
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001394699237997603,
+      "loss": 0.1209,
+      "step": 8989
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001394570215739342,
+      "loss": 0.0913,
+      "step": 8990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001394441185701152,
+      "loss": 0.0952,
+      "step": 8991
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013943121478855781,
+      "loss": 0.0964,
+      "step": 8992
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013941831022951643,
+      "loss": 0.0878,
+      "step": 8993
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013940540489324547,
+      "loss": 0.1168,
+      "step": 8994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013939249877999945,
+      "loss": 0.1011,
+      "step": 8995
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013937959189003282,
+      "loss": 0.0678,
+      "step": 8996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013936668422360006,
+      "loss": 0.0629,
+      "step": 8997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013935377578095567,
+      "loss": 0.0641,
+      "step": 8998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013934086656235422,
+      "loss": 0.0844,
+      "step": 8999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001393279565680502,
+      "loss": 0.1034,
+      "step": 9000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001393150457982982,
+      "loss": 0.119,
+      "step": 9001
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013930213425335277,
+      "loss": 0.0778,
+      "step": 9002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013928922193346845,
+      "loss": 0.0585,
+      "step": 9003
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013927630883889988,
+      "loss": 0.1211,
+      "step": 9004
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001392633949699017,
+      "loss": 0.0925,
+      "step": 9005
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001392504803267285,
+      "loss": 0.1033,
+      "step": 9006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013923756490963493,
+      "loss": 0.0975,
+      "step": 9007
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013922464871887565,
+      "loss": 0.1055,
+      "step": 9008
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013921173175470535,
+      "loss": 0.0763,
+      "step": 9009
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013919881401737873,
+      "loss": 0.0969,
+      "step": 9010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013918589550715042,
+      "loss": 0.1498,
+      "step": 9011
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013917297622427524,
+      "loss": 0.1235,
+      "step": 9012
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013916005616900783,
+      "loss": 0.0962,
+      "step": 9013
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013914713534160304,
+      "loss": 0.1018,
+      "step": 9014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013913421374231558,
+      "loss": 0.214,
+      "step": 9015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013912129137140024,
+      "loss": 0.0968,
+      "step": 9016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.001391083682291118,
+      "loss": 0.1122,
+      "step": 9017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0013909544431570508,
+      "loss": 0.1364,
+      "step": 9018
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013908251963143492,
+      "loss": 0.1119,
+      "step": 9019
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013906959417655615,
+      "loss": 0.1091,
+      "step": 9020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013905666795132362,
+      "loss": 0.1251,
+      "step": 9021
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013904374095599227,
+      "loss": 0.1056,
+      "step": 9022
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013903081319081689,
+      "loss": 0.078,
+      "step": 9023
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001390178846560524,
+      "loss": 0.1461,
+      "step": 9024
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001390049553519538,
+      "loss": 0.0852,
+      "step": 9025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013899202527877595,
+      "loss": 0.0576,
+      "step": 9026
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001389790944367738,
+      "loss": 0.0983,
+      "step": 9027
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013896616282620233,
+      "loss": 0.1006,
+      "step": 9028
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013895323044731652,
+      "loss": 0.1253,
+      "step": 9029
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013894029730037137,
+      "loss": 0.137,
+      "step": 9030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001389273633856219,
+      "loss": 0.1384,
+      "step": 9031
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013891442870332307,
+      "loss": 0.0994,
+      "step": 9032
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013890149325373,
+      "loss": 0.0745,
+      "step": 9033
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013888855703709769,
+      "loss": 0.0797,
+      "step": 9034
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001388756200536812,
+      "loss": 0.1016,
+      "step": 9035
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013886268230373568,
+      "loss": 0.1053,
+      "step": 9036
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001388497437875162,
+      "loss": 0.1235,
+      "step": 9037
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013883680450527786,
+      "loss": 0.103,
+      "step": 9038
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001388238644572758,
+      "loss": 0.0889,
+      "step": 9039
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013881092364376514,
+      "loss": 0.0703,
+      "step": 9040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013879798206500107,
+      "loss": 0.0895,
+      "step": 9041
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013878503972123877,
+      "loss": 0.1002,
+      "step": 9042
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013877209661273342,
+      "loss": 0.0826,
+      "step": 9043
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001387591527397402,
+      "loss": 0.0707,
+      "step": 9044
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013874620810251435,
+      "loss": 0.0878,
+      "step": 9045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013873326270131114,
+      "loss": 0.0726,
+      "step": 9046
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013872031653638576,
+      "loss": 0.1167,
+      "step": 9047
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013870736960799354,
+      "loss": 0.1334,
+      "step": 9048
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013869442191638973,
+      "loss": 0.1285,
+      "step": 9049
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013868147346182958,
+      "loss": 0.1124,
+      "step": 9050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013866852424456845,
+      "loss": 0.1117,
+      "step": 9051
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001386555742648617,
+      "loss": 0.1064,
+      "step": 9052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013864262352296462,
+      "loss": 0.0554,
+      "step": 9053
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013862967201913254,
+      "loss": 0.1069,
+      "step": 9054
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.001386167197536209,
+      "loss": 0.0899,
+      "step": 9055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013860376672668502,
+      "loss": 0.1033,
+      "step": 9056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013859081293858034,
+      "loss": 0.0928,
+      "step": 9057
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013857785838956229,
+      "loss": 0.094,
+      "step": 9058
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013856490307988629,
+      "loss": 0.1344,
+      "step": 9059
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013855194700980779,
+      "loss": 0.1133,
+      "step": 9060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013853899017958215,
+      "loss": 0.1146,
+      "step": 9061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013852603258946502,
+      "loss": 0.0759,
+      "step": 9062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013851307423971178,
+      "loss": 0.1123,
+      "step": 9063
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0013850011513057794,
+      "loss": 0.1278,
+      "step": 9064
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013848715526231905,
+      "loss": 0.093,
+      "step": 9065
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013847419463519064,
+      "loss": 0.1198,
+      "step": 9066
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013846123324944826,
+      "loss": 0.115,
+      "step": 9067
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013844827110534748,
+      "loss": 0.1136,
+      "step": 9068
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001384353082031439,
+      "loss": 0.0972,
+      "step": 9069
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013842234454309306,
+      "loss": 0.127,
+      "step": 9070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001384093801254506,
+      "loss": 0.0857,
+      "step": 9071
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013839641495047212,
+      "loss": 0.0735,
+      "step": 9072
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013838344901841333,
+      "loss": 0.1052,
+      "step": 9073
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013837048232952982,
+      "loss": 0.0848,
+      "step": 9074
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001383575148840773,
+      "loss": 0.1224,
+      "step": 9075
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013834454668231145,
+      "loss": 0.0786,
+      "step": 9076
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013833157772448793,
+      "loss": 0.0798,
+      "step": 9077
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013831860801086246,
+      "loss": 0.1366,
+      "step": 9078
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013830563754169086,
+      "loss": 0.1149,
+      "step": 9079
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013829266631722877,
+      "loss": 0.0775,
+      "step": 9080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013827969433773198,
+      "loss": 0.1046,
+      "step": 9081
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013826672160345628,
+      "loss": 0.1545,
+      "step": 9082
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013825374811465747,
+      "loss": 0.0593,
+      "step": 9083
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001382407738715913,
+      "loss": 0.0784,
+      "step": 9084
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013822779887451367,
+      "loss": 0.1191,
+      "step": 9085
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013821482312368037,
+      "loss": 0.0948,
+      "step": 9086
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013820184661934723,
+      "loss": 0.0638,
+      "step": 9087
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001381888693617701,
+      "loss": 0.1898,
+      "step": 9088
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013817589135120497,
+      "loss": 0.0875,
+      "step": 9089
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013816291258790761,
+      "loss": 0.069,
+      "step": 9090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013814993307213394,
+      "loss": 0.0763,
+      "step": 9091
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013813695280413998,
+      "loss": 0.1248,
+      "step": 9092
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013812397178418158,
+      "loss": 0.14,
+      "step": 9093
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001381109900125147,
+      "loss": 0.1165,
+      "step": 9094
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013809800748939536,
+      "loss": 0.0673,
+      "step": 9095
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001380850242150795,
+      "loss": 0.0662,
+      "step": 9096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001380720401898231,
+      "loss": 0.0975,
+      "step": 9097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013805905541388222,
+      "loss": 0.0895,
+      "step": 9098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013804606988751288,
+      "loss": 0.1188,
+      "step": 9099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001380330836109711,
+      "loss": 0.0753,
+      "step": 9100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013802009658451292,
+      "loss": 0.0759,
+      "step": 9101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013800710880839445,
+      "loss": 0.1157,
+      "step": 9102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013799412028287176,
+      "loss": 0.0765,
+      "step": 9103
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013798113100820094,
+      "loss": 0.1204,
+      "step": 9104
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013796814098463814,
+      "loss": 0.0782,
+      "step": 9105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013795515021243947,
+      "loss": 0.0744,
+      "step": 9106
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013794215869186103,
+      "loss": 0.0709,
+      "step": 9107
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013792916642315909,
+      "loss": 0.0884,
+      "step": 9108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.001379161734065897,
+      "loss": 0.071,
+      "step": 9109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0013790317964240916,
+      "loss": 0.1383,
+      "step": 9110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013789018513087356,
+      "loss": 0.0676,
+      "step": 9111
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013787718987223925,
+      "loss": 0.0751,
+      "step": 9112
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013786419386676236,
+      "loss": 0.0833,
+      "step": 9113
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013785119711469916,
+      "loss": 0.0955,
+      "step": 9114
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013783819961630593,
+      "loss": 0.0625,
+      "step": 9115
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013782520137183897,
+      "loss": 0.0708,
+      "step": 9116
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001378122023815545,
+      "loss": 0.0913,
+      "step": 9117
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013779920264570893,
+      "loss": 0.1383,
+      "step": 9118
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013778620216455847,
+      "loss": 0.1373,
+      "step": 9119
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013777320093835957,
+      "loss": 0.082,
+      "step": 9120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013776019896736848,
+      "loss": 0.0874,
+      "step": 9121
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013774719625184164,
+      "loss": 0.1096,
+      "step": 9122
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001377341927920354,
+      "loss": 0.0886,
+      "step": 9123
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013772118858820612,
+      "loss": 0.0862,
+      "step": 9124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013770818364061026,
+      "loss": 0.1051,
+      "step": 9125
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013769517794950425,
+      "loss": 0.1083,
+      "step": 9126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001376821715151445,
+      "loss": 0.1118,
+      "step": 9127
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013766916433778748,
+      "loss": 0.0897,
+      "step": 9128
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013765615641768965,
+      "loss": 0.0891,
+      "step": 9129
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013764314775510749,
+      "loss": 0.1208,
+      "step": 9130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001376301383502975,
+      "loss": 0.2007,
+      "step": 9131
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013761712820351622,
+      "loss": 0.1014,
+      "step": 9132
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013760411731502014,
+      "loss": 0.0983,
+      "step": 9133
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013759110568506581,
+      "loss": 0.1254,
+      "step": 9134
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001375780933139098,
+      "loss": 0.1578,
+      "step": 9135
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013756508020180867,
+      "loss": 0.0862,
+      "step": 9136
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00137552066349019,
+      "loss": 0.1046,
+      "step": 9137
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013753905175579741,
+      "loss": 0.0847,
+      "step": 9138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013752603642240052,
+      "loss": 0.1324,
+      "step": 9139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001375130203490849,
+      "loss": 0.1252,
+      "step": 9140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013750000353610727,
+      "loss": 0.1141,
+      "step": 9141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013748698598372426,
+      "loss": 0.1548,
+      "step": 9142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013747396769219257,
+      "loss": 0.0999,
+      "step": 9143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001374609486617688,
+      "loss": 0.0847,
+      "step": 9144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013744792889270974,
+      "loss": 0.1281,
+      "step": 9145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013743490838527203,
+      "loss": 0.1667,
+      "step": 9146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.001374218871397125,
+      "loss": 0.0986,
+      "step": 9147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013740886515628784,
+      "loss": 0.1385,
+      "step": 9148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013739584243525483,
+      "loss": 0.0756,
+      "step": 9149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013738281897687022,
+      "loss": 0.0785,
+      "step": 9150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013736979478139078,
+      "loss": 0.1157,
+      "step": 9151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013735676984907336,
+      "loss": 0.092,
+      "step": 9152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013734374418017477,
+      "loss": 0.1134,
+      "step": 9153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013733071777495183,
+      "loss": 0.0851,
+      "step": 9154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013731769063366139,
+      "loss": 0.1075,
+      "step": 9155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0013730466275656034,
+      "loss": 0.114,
+      "step": 9156
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013729163414390548,
+      "loss": 0.0958,
+      "step": 9157
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013727860479595382,
+      "loss": 0.0947,
+      "step": 9158
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001372655747129622,
+      "loss": 0.1912,
+      "step": 9159
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013725254389518752,
+      "loss": 0.0946,
+      "step": 9160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013723951234288672,
+      "loss": 0.1056,
+      "step": 9161
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013722648005631676,
+      "loss": 0.0893,
+      "step": 9162
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013721344703573465,
+      "loss": 0.0959,
+      "step": 9163
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001372004132813973,
+      "loss": 0.1445,
+      "step": 9164
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013718737879356174,
+      "loss": 0.1078,
+      "step": 9165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013717434357248498,
+      "loss": 0.0872,
+      "step": 9166
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013716130761842397,
+      "loss": 0.115,
+      "step": 9167
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013714827093163585,
+      "loss": 0.0885,
+      "step": 9168
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013713523351237765,
+      "loss": 0.1665,
+      "step": 9169
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013712219536090638,
+      "loss": 0.0892,
+      "step": 9170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013710915647747915,
+      "loss": 0.0831,
+      "step": 9171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013709611686235307,
+      "loss": 0.063,
+      "step": 9172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001370830765157852,
+      "loss": 0.0866,
+      "step": 9173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001370700354380327,
+      "loss": 0.1,
+      "step": 9174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013705699362935274,
+      "loss": 0.1,
+      "step": 9175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013704395109000243,
+      "loss": 0.0983,
+      "step": 9176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001370309078202389,
+      "loss": 0.1281,
+      "step": 9177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013701786382031937,
+      "loss": 0.0811,
+      "step": 9178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013700481909050106,
+      "loss": 0.0939,
+      "step": 9179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013699177363104115,
+      "loss": 0.0892,
+      "step": 9180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013697872744219688,
+      "loss": 0.1276,
+      "step": 9181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013696568052422547,
+      "loss": 0.0882,
+      "step": 9182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013695263287738415,
+      "loss": 0.0931,
+      "step": 9183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013693958450193022,
+      "loss": 0.0792,
+      "step": 9184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013692653539812097,
+      "loss": 0.1039,
+      "step": 9185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001369134855662137,
+      "loss": 0.0716,
+      "step": 9186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013690043500646568,
+      "loss": 0.1291,
+      "step": 9187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013688738371913426,
+      "loss": 0.1874,
+      "step": 9188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013687433170447673,
+      "loss": 0.1039,
+      "step": 9189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013686127896275053,
+      "loss": 0.0806,
+      "step": 9190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013684822549421298,
+      "loss": 0.1119,
+      "step": 9191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013683517129912145,
+      "loss": 0.1523,
+      "step": 9192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013682211637773337,
+      "loss": 0.1385,
+      "step": 9193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013680906073030611,
+      "loss": 0.0694,
+      "step": 9194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013679600435709711,
+      "loss": 0.1142,
+      "step": 9195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013678294725836383,
+      "loss": 0.0758,
+      "step": 9196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001367698894343637,
+      "loss": 0.1315,
+      "step": 9197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013675683088535421,
+      "loss": 0.0836,
+      "step": 9198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013674377161159278,
+      "loss": 0.1359,
+      "step": 9199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013673071161333702,
+      "loss": 0.11,
+      "step": 9200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.001367176508908443,
+      "loss": 0.0595,
+      "step": 9201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0013670458944437228,
+      "loss": 0.1344,
+      "step": 9202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001366915272741784,
+      "loss": 0.1219,
+      "step": 9203
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013667846438052025,
+      "loss": 0.066,
+      "step": 9204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013666540076365536,
+      "loss": 0.1132,
+      "step": 9205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001366523364238414,
+      "loss": 0.111,
+      "step": 9206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013663927136133589,
+      "loss": 0.0856,
+      "step": 9207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013662620557639647,
+      "loss": 0.0897,
+      "step": 9208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013661313906928078,
+      "loss": 0.1163,
+      "step": 9209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013660007184024636,
+      "loss": 0.1061,
+      "step": 9210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013658700388955098,
+      "loss": 0.101,
+      "step": 9211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013657393521745228,
+      "loss": 0.0848,
+      "step": 9212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013656086582420793,
+      "loss": 0.1497,
+      "step": 9213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013654779571007558,
+      "loss": 0.1147,
+      "step": 9214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00136534724875313,
+      "loss": 0.113,
+      "step": 9215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001365216533201779,
+      "loss": 0.0937,
+      "step": 9216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00136508581044928,
+      "loss": 0.1086,
+      "step": 9217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001364955080498211,
+      "loss": 0.09,
+      "step": 9218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001364824343351149,
+      "loss": 0.1589,
+      "step": 9219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013646935990106722,
+      "loss": 0.1078,
+      "step": 9220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013645628474793583,
+      "loss": 0.1144,
+      "step": 9221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013644320887597858,
+      "loss": 0.0817,
+      "step": 9222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013643013228545329,
+      "loss": 0.0791,
+      "step": 9223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013641705497661773,
+      "loss": 0.0597,
+      "step": 9224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013640397694972984,
+      "loss": 0.0919,
+      "step": 9225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001363908982050474,
+      "loss": 0.1017,
+      "step": 9226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013637781874282836,
+      "loss": 0.1195,
+      "step": 9227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013636473856333058,
+      "loss": 0.1448,
+      "step": 9228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013635165766681199,
+      "loss": 0.1115,
+      "step": 9229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013633857605353048,
+      "loss": 0.1559,
+      "step": 9230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00136325493723744,
+      "loss": 0.1559,
+      "step": 9231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013631241067771052,
+      "loss": 0.2095,
+      "step": 9232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013629932691568798,
+      "loss": 0.0389,
+      "step": 9233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013628624243793438,
+      "loss": 0.1028,
+      "step": 9234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001362731572447077,
+      "loss": 0.0777,
+      "step": 9235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013626007133626591,
+      "loss": 0.0806,
+      "step": 9236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013624698471286709,
+      "loss": 0.1416,
+      "step": 9237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013623389737476928,
+      "loss": 0.0605,
+      "step": 9238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001362208093222305,
+      "loss": 0.1094,
+      "step": 9239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013620772055550883,
+      "loss": 0.0598,
+      "step": 9240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001361946310748623,
+      "loss": 0.0521,
+      "step": 9241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00136181540880549,
+      "loss": 0.0989,
+      "step": 9242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013616844997282713,
+      "loss": 0.1335,
+      "step": 9243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013615535835195473,
+      "loss": 0.089,
+      "step": 9244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013614226601818998,
+      "loss": 0.1067,
+      "step": 9245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00136129172971791,
+      "loss": 0.0508,
+      "step": 9246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001361160792130159,
+      "loss": 0.1101,
+      "step": 9247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0013610298474212295,
+      "loss": 0.064,
+      "step": 9248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.001360898895593703,
+      "loss": 0.0714,
+      "step": 9249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013607679366501617,
+      "loss": 0.1011,
+      "step": 9250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013606369705931873,
+      "loss": 0.0648,
+      "step": 9251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013605059974253627,
+      "loss": 0.1263,
+      "step": 9252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013603750171492699,
+      "loss": 0.1584,
+      "step": 9253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001360244029767492,
+      "loss": 0.07,
+      "step": 9254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001360113035282611,
+      "loss": 0.1514,
+      "step": 9255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013599820336972106,
+      "loss": 0.1335,
+      "step": 9256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013598510250138732,
+      "loss": 0.0814,
+      "step": 9257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013597200092351823,
+      "loss": 0.0826,
+      "step": 9258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001359588986363721,
+      "loss": 0.0933,
+      "step": 9259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013594579564020729,
+      "loss": 0.0609,
+      "step": 9260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013593269193528214,
+      "loss": 0.1338,
+      "step": 9261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013591958752185507,
+      "loss": 0.1187,
+      "step": 9262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001359064824001844,
+      "loss": 0.1051,
+      "step": 9263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013589337657052855,
+      "loss": 0.1667,
+      "step": 9264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013588027003314598,
+      "loss": 0.1725,
+      "step": 9265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013586716278829508,
+      "loss": 0.1006,
+      "step": 9266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013585405483623428,
+      "loss": 0.0861,
+      "step": 9267
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013584094617722202,
+      "loss": 0.0636,
+      "step": 9268
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013582783681151684,
+      "loss": 0.0494,
+      "step": 9269
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013581472673937715,
+      "loss": 0.0919,
+      "step": 9270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013580161596106152,
+      "loss": 0.075,
+      "step": 9271
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001357885044768284,
+      "loss": 0.1241,
+      "step": 9272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013577539228693633,
+      "loss": 0.1417,
+      "step": 9273
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013576227939164384,
+      "loss": 0.0631,
+      "step": 9274
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013574916579120954,
+      "loss": 0.0709,
+      "step": 9275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013573605148589193,
+      "loss": 0.1047,
+      "step": 9276
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013572293647594964,
+      "loss": 0.3179,
+      "step": 9277
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013570982076164125,
+      "loss": 0.1191,
+      "step": 9278
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013569670434322532,
+      "loss": 0.1134,
+      "step": 9279
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013568358722096054,
+      "loss": 0.1304,
+      "step": 9280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001356704693951055,
+      "loss": 0.119,
+      "step": 9281
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001356573508659189,
+      "loss": 0.1263,
+      "step": 9282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013564423163365934,
+      "loss": 0.1069,
+      "step": 9283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013563111169858555,
+      "loss": 0.0853,
+      "step": 9284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001356179910609562,
+      "loss": 0.0988,
+      "step": 9285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013560486972103,
+      "loss": 0.0869,
+      "step": 9286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001355917476790657,
+      "loss": 0.0934,
+      "step": 9287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013557862493532198,
+      "loss": 0.1649,
+      "step": 9288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013556550149005759,
+      "loss": 0.0707,
+      "step": 9289
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013555237734353134,
+      "loss": 0.0785,
+      "step": 9290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013553925249600199,
+      "loss": 0.0784,
+      "step": 9291
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.001355261269477283,
+      "loss": 0.0886,
+      "step": 9292
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013551300069896911,
+      "loss": 0.131,
+      "step": 9293
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013549987374998319,
+      "loss": 0.1129,
+      "step": 9294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0013548674610102938,
+      "loss": 0.0869,
+      "step": 9295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013547361775236655,
+      "loss": 0.0948,
+      "step": 9296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013546048870425357,
+      "loss": 0.0858,
+      "step": 9297
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013544735895694928,
+      "loss": 0.131,
+      "step": 9298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.001354342285107126,
+      "loss": 0.0852,
+      "step": 9299
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013542109736580236,
+      "loss": 0.0739,
+      "step": 9300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013540796552247753,
+      "loss": 0.1298,
+      "step": 9301
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013539483298099701,
+      "loss": 0.1069,
+      "step": 9302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013538169974161981,
+      "loss": 0.0601,
+      "step": 9303
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013536856580460476,
+      "loss": 0.0709,
+      "step": 9304
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.001353554311702109,
+      "loss": 0.0979,
+      "step": 9305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013534229583869726,
+      "loss": 0.1058,
+      "step": 9306
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013532915981032277,
+      "loss": 0.1626,
+      "step": 9307
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013531602308534646,
+      "loss": 0.1416,
+      "step": 9308
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013530288566402733,
+      "loss": 0.1041,
+      "step": 9309
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.001352897475466244,
+      "loss": 0.0824,
+      "step": 9310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.001352766087333968,
+      "loss": 0.0864,
+      "step": 9311
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013526346922460358,
+      "loss": 0.118,
+      "step": 9312
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013525032902050374,
+      "loss": 0.1174,
+      "step": 9313
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013523718812135641,
+      "loss": 0.1165,
+      "step": 9314
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013522404652742074,
+      "loss": 0.1295,
+      "step": 9315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013521090423895576,
+      "loss": 0.0994,
+      "step": 9316
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013519776125622068,
+      "loss": 0.0933,
+      "step": 9317
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013518461757947465,
+      "loss": 0.0546,
+      "step": 9318
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013517147320897679,
+      "loss": 0.082,
+      "step": 9319
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013515832814498623,
+      "loss": 0.1289,
+      "step": 9320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013514518238776228,
+      "loss": 0.0743,
+      "step": 9321
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013513203593756404,
+      "loss": 0.1316,
+      "step": 9322
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013511888879465077,
+      "loss": 0.1273,
+      "step": 9323
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013510574095928168,
+      "loss": 0.1479,
+      "step": 9324
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013509259243171604,
+      "loss": 0.1102,
+      "step": 9325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013507944321221306,
+      "loss": 0.1266,
+      "step": 9326
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013506629330103203,
+      "loss": 0.0765,
+      "step": 9327
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013505314269843228,
+      "loss": 0.0583,
+      "step": 9328
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013503999140467302,
+      "loss": 0.1294,
+      "step": 9329
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013502683942001362,
+      "loss": 0.1165,
+      "step": 9330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013501368674471341,
+      "loss": 0.1091,
+      "step": 9331
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013500053337903166,
+      "loss": 0.0656,
+      "step": 9332
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013498737932322782,
+      "loss": 0.1663,
+      "step": 9333
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013497422457756114,
+      "loss": 0.077,
+      "step": 9334
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013496106914229111,
+      "loss": 0.1227,
+      "step": 9335
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013494791301767708,
+      "loss": 0.1216,
+      "step": 9336
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013493475620397842,
+      "loss": 0.1011,
+      "step": 9337
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013492159870145458,
+      "loss": 0.0774,
+      "step": 9338
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00134908440510365,
+      "loss": 0.1843,
+      "step": 9339
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.001348952816309691,
+      "loss": 0.1565,
+      "step": 9340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0013488212206352634,
+      "loss": 0.0916,
+      "step": 9341
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013486896180829623,
+      "loss": 0.0446,
+      "step": 9342
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013485580086553825,
+      "loss": 0.109,
+      "step": 9343
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013484263923551187,
+      "loss": 0.106,
+      "step": 9344
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013482947691847665,
+      "loss": 0.0905,
+      "step": 9345
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013481631391469206,
+      "loss": 0.1071,
+      "step": 9346
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013480315022441768,
+      "loss": 0.1771,
+      "step": 9347
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013478998584791305,
+      "loss": 0.113,
+      "step": 9348
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013477682078543776,
+      "loss": 0.0617,
+      "step": 9349
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013476365503725133,
+      "loss": 0.0897,
+      "step": 9350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013475048860361345,
+      "loss": 0.1144,
+      "step": 9351
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013473732148478366,
+      "loss": 0.2217,
+      "step": 9352
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.001347241536810216,
+      "loss": 0.1315,
+      "step": 9353
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013471098519258688,
+      "loss": 0.137,
+      "step": 9354
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013469781601973922,
+      "loss": 0.1263,
+      "step": 9355
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013468464616273824,
+      "loss": 0.0782,
+      "step": 9356
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.001346714756218436,
+      "loss": 0.1265,
+      "step": 9357
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00134658304397315,
+      "loss": 0.0848,
+      "step": 9358
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013464513248941212,
+      "loss": 0.0747,
+      "step": 9359
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013463195989839474,
+      "loss": 0.0881,
+      "step": 9360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013461878662452255,
+      "loss": 0.1075,
+      "step": 9361
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013460561266805532,
+      "loss": 0.0655,
+      "step": 9362
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013459243802925276,
+      "loss": 0.0857,
+      "step": 9363
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013457926270837466,
+      "loss": 0.067,
+      "step": 9364
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013456608670568086,
+      "loss": 0.1503,
+      "step": 9365
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013455291002143106,
+      "loss": 0.0635,
+      "step": 9366
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013453973265588513,
+      "loss": 0.0944,
+      "step": 9367
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013452655460930289,
+      "loss": 0.0993,
+      "step": 9368
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013451337588194416,
+      "loss": 0.0602,
+      "step": 9369
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013450019647406881,
+      "loss": 0.0972,
+      "step": 9370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013448701638593672,
+      "loss": 0.0493,
+      "step": 9371
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013447383561780773,
+      "loss": 0.0921,
+      "step": 9372
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013446065416994175,
+      "loss": 0.0933,
+      "step": 9373
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.001344474720425987,
+      "loss": 0.14,
+      "step": 9374
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013443428923603843,
+      "loss": 0.0493,
+      "step": 9375
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013442110575052095,
+      "loss": 0.0709,
+      "step": 9376
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013440792158630616,
+      "loss": 0.1216,
+      "step": 9377
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013439473674365405,
+      "loss": 0.102,
+      "step": 9378
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.001343815512228246,
+      "loss": 0.109,
+      "step": 9379
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013436836502407773,
+      "loss": 0.126,
+      "step": 9380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.001343551781476735,
+      "loss": 0.1375,
+      "step": 9381
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.001343419905938719,
+      "loss": 0.0966,
+      "step": 9382
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013432880236293298,
+      "loss": 0.0504,
+      "step": 9383
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013431561345511674,
+      "loss": 0.1147,
+      "step": 9384
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013430242387068322,
+      "loss": 0.1165,
+      "step": 9385
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013428923360989256,
+      "loss": 0.1671,
+      "step": 9386
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0013427604267300477,
+      "loss": 0.104,
+      "step": 9387
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013426285106027998,
+      "loss": 0.0873,
+      "step": 9388
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013424965877197829,
+      "loss": 0.1333,
+      "step": 9389
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013423646580835978,
+      "loss": 0.1495,
+      "step": 9390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001342232721696846,
+      "loss": 0.0763,
+      "step": 9391
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013421007785621294,
+      "loss": 0.0602,
+      "step": 9392
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001341968828682049,
+      "loss": 0.0919,
+      "step": 9393
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013418368720592069,
+      "loss": 0.136,
+      "step": 9394
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013417049086962049,
+      "loss": 0.0834,
+      "step": 9395
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013415729385956446,
+      "loss": 0.1067,
+      "step": 9396
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013414409617601285,
+      "loss": 0.0822,
+      "step": 9397
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001341308978192259,
+      "loss": 0.1046,
+      "step": 9398
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001341176987894638,
+      "loss": 0.1061,
+      "step": 9399
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013410449908698684,
+      "loss": 0.1344,
+      "step": 9400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013409129871205523,
+      "loss": 0.1271,
+      "step": 9401
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013407809766492935,
+      "loss": 0.0851,
+      "step": 9402
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013406489594586942,
+      "loss": 0.0918,
+      "step": 9403
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001340516935551357,
+      "loss": 0.0864,
+      "step": 9404
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013403849049298865,
+      "loss": 0.1145,
+      "step": 9405
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013402528675968844,
+      "loss": 0.0775,
+      "step": 9406
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001340120823554955,
+      "loss": 0.0433,
+      "step": 9407
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001339988772806702,
+      "loss": 0.0967,
+      "step": 9408
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001339856715354729,
+      "loss": 0.0723,
+      "step": 9409
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013397246512016393,
+      "loss": 0.0733,
+      "step": 9410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013395925803500375,
+      "loss": 0.0697,
+      "step": 9411
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013394605028025273,
+      "loss": 0.127,
+      "step": 9412
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013393284185617134,
+      "loss": 0.0706,
+      "step": 9413
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013391963276301994,
+      "loss": 0.1215,
+      "step": 9414
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001339064230010591,
+      "loss": 0.0956,
+      "step": 9415
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013389321257054915,
+      "loss": 0.1281,
+      "step": 9416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013388000147175063,
+      "loss": 0.1149,
+      "step": 9417
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013386678970492404,
+      "loss": 0.1185,
+      "step": 9418
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001338535772703299,
+      "loss": 0.0848,
+      "step": 9419
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013384036416822865,
+      "loss": 0.1174,
+      "step": 9420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001338271503988809,
+      "loss": 0.134,
+      "step": 9421
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013381393596254715,
+      "loss": 0.1707,
+      "step": 9422
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013380072085948794,
+      "loss": 0.0679,
+      "step": 9423
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013378750508996388,
+      "loss": 0.0733,
+      "step": 9424
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013377428865423554,
+      "loss": 0.0923,
+      "step": 9425
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001337610715525635,
+      "loss": 0.099,
+      "step": 9426
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013374785378520834,
+      "loss": 0.0876,
+      "step": 9427
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013373463535243074,
+      "loss": 0.1207,
+      "step": 9428
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013372141625449135,
+      "loss": 0.0467,
+      "step": 9429
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.001337081964916507,
+      "loss": 0.0598,
+      "step": 9430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013369497606416962,
+      "loss": 0.1106,
+      "step": 9431
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013368175497230862,
+      "loss": 0.1191,
+      "step": 9432
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0013366853321632845,
+      "loss": 0.0421,
+      "step": 9433
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013365531079648984,
+      "loss": 0.0886,
+      "step": 9434
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013364208771305351,
+      "loss": 0.0968,
+      "step": 9435
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013362886396628016,
+      "loss": 0.126,
+      "step": 9436
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013361563955643049,
+      "loss": 0.1375,
+      "step": 9437
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013360241448376526,
+      "loss": 0.0729,
+      "step": 9438
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001335891887485453,
+      "loss": 0.0835,
+      "step": 9439
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013357596235103134,
+      "loss": 0.1586,
+      "step": 9440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001335627352914842,
+      "loss": 0.1188,
+      "step": 9441
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013354950757016467,
+      "loss": 0.0988,
+      "step": 9442
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001335362791873335,
+      "loss": 0.0922,
+      "step": 9443
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013352305014325164,
+      "loss": 0.0777,
+      "step": 9444
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013350982043817984,
+      "loss": 0.1055,
+      "step": 9445
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013349659007237905,
+      "loss": 0.0952,
+      "step": 9446
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013348335904611003,
+      "loss": 0.0869,
+      "step": 9447
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013347012735963374,
+      "loss": 0.0715,
+      "step": 9448
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013345689501321103,
+      "loss": 0.0675,
+      "step": 9449
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001334436620071028,
+      "loss": 0.0832,
+      "step": 9450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013343042834157005,
+      "loss": 0.1017,
+      "step": 9451
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013341719401687366,
+      "loss": 0.0734,
+      "step": 9452
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001334039590332746,
+      "loss": 0.0659,
+      "step": 9453
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013339072339103376,
+      "loss": 0.1204,
+      "step": 9454
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001333774870904122,
+      "loss": 0.1199,
+      "step": 9455
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013336425013167085,
+      "loss": 0.1064,
+      "step": 9456
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013335101251507075,
+      "loss": 0.0743,
+      "step": 9457
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001333377742408729,
+      "loss": 0.1245,
+      "step": 9458
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013332453530933827,
+      "loss": 0.0639,
+      "step": 9459
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00133311295720728,
+      "loss": 0.0652,
+      "step": 9460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013329805547530307,
+      "loss": 0.1433,
+      "step": 9461
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013328481457332458,
+      "loss": 0.0684,
+      "step": 9462
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001332715730150536,
+      "loss": 0.033,
+      "step": 9463
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013325833080075117,
+      "loss": 0.078,
+      "step": 9464
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013324508793067843,
+      "loss": 0.1914,
+      "step": 9465
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013323184440509654,
+      "loss": 0.0664,
+      "step": 9466
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013321860022426652,
+      "loss": 0.1323,
+      "step": 9467
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013320535538844964,
+      "loss": 0.1035,
+      "step": 9468
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00133192109897907,
+      "loss": 0.1188,
+      "step": 9469
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013317886375289972,
+      "loss": 0.0928,
+      "step": 9470
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013316561695368902,
+      "loss": 0.0748,
+      "step": 9471
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013315236950053615,
+      "loss": 0.1021,
+      "step": 9472
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013313912139370223,
+      "loss": 0.0786,
+      "step": 9473
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001331258726334485,
+      "loss": 0.0721,
+      "step": 9474
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013311262322003622,
+      "loss": 0.1166,
+      "step": 9475
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013309937315372662,
+      "loss": 0.0948,
+      "step": 9476
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013308612243478092,
+      "loss": 0.1877,
+      "step": 9477
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.001330728710634605,
+      "loss": 0.1374,
+      "step": 9478
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0013305961904002653,
+      "loss": 0.1145,
+      "step": 9479
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013304636636474031,
+      "loss": 0.0782,
+      "step": 9480
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013303311303786325,
+      "loss": 0.0739,
+      "step": 9481
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001330198590596566,
+      "loss": 0.0623,
+      "step": 9482
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013300660443038169,
+      "loss": 0.073,
+      "step": 9483
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013299334915029986,
+      "loss": 0.07,
+      "step": 9484
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013298009321967253,
+      "loss": 0.074,
+      "step": 9485
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00132966836638761,
+      "loss": 0.0752,
+      "step": 9486
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013295357940782669,
+      "loss": 0.0674,
+      "step": 9487
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013294032152713102,
+      "loss": 0.0764,
+      "step": 9488
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001329270629969354,
+      "loss": 0.0834,
+      "step": 9489
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001329138038175012,
+      "loss": 0.077,
+      "step": 9490
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013290054398908994,
+      "loss": 0.0614,
+      "step": 9491
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013288728351196297,
+      "loss": 0.078,
+      "step": 9492
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013287402238638185,
+      "loss": 0.1462,
+      "step": 9493
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00132860760612608,
+      "loss": 0.115,
+      "step": 9494
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013284749819090294,
+      "loss": 0.1086,
+      "step": 9495
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013283423512152812,
+      "loss": 0.0797,
+      "step": 9496
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013282097140474508,
+      "loss": 0.1357,
+      "step": 9497
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013280770704081541,
+      "loss": 0.071,
+      "step": 9498
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001327944420300006,
+      "loss": 0.0595,
+      "step": 9499
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013278117637256215,
+      "loss": 0.1089,
+      "step": 9500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001327679100687617,
+      "loss": 0.182,
+      "step": 9501
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013275464311886078,
+      "loss": 0.0973,
+      "step": 9502
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013274137552312104,
+      "loss": 0.0569,
+      "step": 9503
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00132728107281804,
+      "loss": 0.0973,
+      "step": 9504
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013271483839517137,
+      "loss": 0.0691,
+      "step": 9505
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013270156886348472,
+      "loss": 0.0659,
+      "step": 9506
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013268829868700567,
+      "loss": 0.0796,
+      "step": 9507
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013267502786599593,
+      "loss": 0.0838,
+      "step": 9508
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013266175640071715,
+      "loss": 0.05,
+      "step": 9509
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013264848429143097,
+      "loss": 0.0749,
+      "step": 9510
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013263521153839915,
+      "loss": 0.0818,
+      "step": 9511
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013262193814188334,
+      "loss": 0.1,
+      "step": 9512
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001326086641021453,
+      "loss": 0.1019,
+      "step": 9513
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013259538941944671,
+      "loss": 0.0874,
+      "step": 9514
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001325821140940494,
+      "loss": 0.0596,
+      "step": 9515
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013256883812621503,
+      "loss": 0.1131,
+      "step": 9516
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001325555615162054,
+      "loss": 0.1387,
+      "step": 9517
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001325422842642823,
+      "loss": 0.1123,
+      "step": 9518
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013252900637070754,
+      "loss": 0.0681,
+      "step": 9519
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.001325157278357429,
+      "loss": 0.1058,
+      "step": 9520
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013250244865965022,
+      "loss": 0.061,
+      "step": 9521
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013248916884269133,
+      "loss": 0.1329,
+      "step": 9522
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013247588838512802,
+      "loss": 0.0795,
+      "step": 9523
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013246260728722221,
+      "loss": 0.0886,
+      "step": 9524
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0013244932554923577,
+      "loss": 0.0853,
+      "step": 9525
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013243604317143057,
+      "loss": 0.0721,
+      "step": 9526
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013242276015406848,
+      "loss": 0.0963,
+      "step": 9527
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013240947649741143,
+      "loss": 0.136,
+      "step": 9528
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013239619220172132,
+      "loss": 0.0906,
+      "step": 9529
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001323829072672601,
+      "loss": 0.0995,
+      "step": 9530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013236962169428975,
+      "loss": 0.0731,
+      "step": 9531
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013235633548307217,
+      "loss": 0.1134,
+      "step": 9532
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013234304863386935,
+      "loss": 0.1302,
+      "step": 9533
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013232976114694326,
+      "loss": 0.1339,
+      "step": 9534
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013231647302255594,
+      "loss": 0.1837,
+      "step": 9535
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001323031842609694,
+      "loss": 0.0969,
+      "step": 9536
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013228989486244558,
+      "loss": 0.0682,
+      "step": 9537
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013227660482724656,
+      "loss": 0.1138,
+      "step": 9538
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013226331415563443,
+      "loss": 0.0895,
+      "step": 9539
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013225002284787118,
+      "loss": 0.0706,
+      "step": 9540
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001322367309042189,
+      "loss": 0.0655,
+      "step": 9541
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013222343832493973,
+      "loss": 0.1201,
+      "step": 9542
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013221014511029569,
+      "loss": 0.0812,
+      "step": 9543
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013219685126054893,
+      "loss": 0.1064,
+      "step": 9544
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013218355677596153,
+      "loss": 0.0563,
+      "step": 9545
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001321702616567957,
+      "loss": 0.0748,
+      "step": 9546
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013215696590331348,
+      "loss": 0.075,
+      "step": 9547
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013214366951577712,
+      "loss": 0.1075,
+      "step": 9548
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013213037249444875,
+      "loss": 0.1228,
+      "step": 9549
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013211707483959058,
+      "loss": 0.0907,
+      "step": 9550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013210377655146477,
+      "loss": 0.1182,
+      "step": 9551
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013209047763033358,
+      "loss": 0.0978,
+      "step": 9552
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013207717807645915,
+      "loss": 0.0626,
+      "step": 9553
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013206387789010381,
+      "loss": 0.1069,
+      "step": 9554
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013205057707152972,
+      "loss": 0.0768,
+      "step": 9555
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013203727562099922,
+      "loss": 0.0667,
+      "step": 9556
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013202397353877448,
+      "loss": 0.1058,
+      "step": 9557
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001320106708251179,
+      "loss": 0.0793,
+      "step": 9558
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001319973674802917,
+      "loss": 0.0815,
+      "step": 9559
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013198406350455817,
+      "loss": 0.0822,
+      "step": 9560
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001319707588981797,
+      "loss": 0.0957,
+      "step": 9561
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013195745366141858,
+      "loss": 0.0688,
+      "step": 9562
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013194414779453716,
+      "loss": 0.1102,
+      "step": 9563
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013193084129779784,
+      "loss": 0.062,
+      "step": 9564
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013191753417146294,
+      "loss": 0.084,
+      "step": 9565
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013190422641579482,
+      "loss": 0.1008,
+      "step": 9566
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013189091803105593,
+      "loss": 0.1089,
+      "step": 9567
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013187760901750866,
+      "loss": 0.0632,
+      "step": 9568
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013186429937541546,
+      "loss": 0.1379,
+      "step": 9569
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.001318509891050387,
+      "loss": 0.0693,
+      "step": 9570
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0013183767820664088,
+      "loss": 0.1084,
+      "step": 9571
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013182436668048442,
+      "loss": 0.0948,
+      "step": 9572
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001318110545268318,
+      "loss": 0.0662,
+      "step": 9573
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013179774174594551,
+      "loss": 0.0891,
+      "step": 9574
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013178442833808805,
+      "loss": 0.0912,
+      "step": 9575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013177111430352194,
+      "loss": 0.1012,
+      "step": 9576
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013175779964250962,
+      "loss": 0.0871,
+      "step": 9577
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013174448435531371,
+      "loss": 0.1461,
+      "step": 9578
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013173116844219675,
+      "loss": 0.0787,
+      "step": 9579
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013171785190342123,
+      "loss": 0.0809,
+      "step": 9580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013170453473924978,
+      "loss": 0.0753,
+      "step": 9581
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013169121694994494,
+      "loss": 0.1383,
+      "step": 9582
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001316778985357693,
+      "loss": 0.0948,
+      "step": 9583
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013166457949698553,
+      "loss": 0.1006,
+      "step": 9584
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001316512598338562,
+      "loss": 0.0956,
+      "step": 9585
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013163793954664397,
+      "loss": 0.0787,
+      "step": 9586
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013162461863561138,
+      "loss": 0.1368,
+      "step": 9587
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001316112971010212,
+      "loss": 0.091,
+      "step": 9588
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013159797494313608,
+      "loss": 0.0361,
+      "step": 9589
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013158465216221865,
+      "loss": 0.0969,
+      "step": 9590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013157132875853165,
+      "loss": 0.0917,
+      "step": 9591
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013155800473233774,
+      "loss": 0.0739,
+      "step": 9592
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013154468008389965,
+      "loss": 0.09,
+      "step": 9593
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013153135481348015,
+      "loss": 0.0792,
+      "step": 9594
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013151802892134197,
+      "loss": 0.0914,
+      "step": 9595
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001315047024077478,
+      "loss": 0.0782,
+      "step": 9596
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013149137527296046,
+      "loss": 0.0838,
+      "step": 9597
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001314780475172427,
+      "loss": 0.0591,
+      "step": 9598
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013146471914085736,
+      "loss": 0.1071,
+      "step": 9599
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013145139014406717,
+      "loss": 0.065,
+      "step": 9600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013143806052713502,
+      "loss": 0.0883,
+      "step": 9601
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013142473029032367,
+      "loss": 0.0659,
+      "step": 9602
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013141139943389595,
+      "loss": 0.0831,
+      "step": 9603
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001313980679581148,
+      "loss": 0.0807,
+      "step": 9604
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00131384735863243,
+      "loss": 0.106,
+      "step": 9605
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001313714031495435,
+      "loss": 0.1128,
+      "step": 9606
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013135806981727908,
+      "loss": 0.0593,
+      "step": 9607
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013134473586671274,
+      "loss": 0.1179,
+      "step": 9608
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013133140129810734,
+      "loss": 0.0891,
+      "step": 9609
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.001313180661117258,
+      "loss": 0.092,
+      "step": 9610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013130473030783113,
+      "loss": 0.1116,
+      "step": 9611
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013129139388668618,
+      "loss": 0.0653,
+      "step": 9612
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013127805684855396,
+      "loss": 0.1143,
+      "step": 9613
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013126471919369742,
+      "loss": 0.0455,
+      "step": 9614
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013125138092237958,
+      "loss": 0.1934,
+      "step": 9615
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013123804203486343,
+      "loss": 0.0852,
+      "step": 9616
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013122470253141194,
+      "loss": 0.1686,
+      "step": 9617
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0013121136241228818,
+      "loss": 0.0804,
+      "step": 9618
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013119802167775511,
+      "loss": 0.1494,
+      "step": 9619
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013118468032807587,
+      "loss": 0.1117,
+      "step": 9620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013117133836351349,
+      "loss": 0.09,
+      "step": 9621
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00131157995784331,
+      "loss": 0.1239,
+      "step": 9622
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001311446525907915,
+      "loss": 0.0543,
+      "step": 9623
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001311313087831581,
+      "loss": 0.1044,
+      "step": 9624
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001311179643616939,
+      "loss": 0.082,
+      "step": 9625
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013110461932666203,
+      "loss": 0.0751,
+      "step": 9626
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013109127367832555,
+      "loss": 0.0894,
+      "step": 9627
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001310779274169477,
+      "loss": 0.1085,
+      "step": 9628
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001310645805427916,
+      "loss": 0.0927,
+      "step": 9629
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013105123305612036,
+      "loss": 0.1561,
+      "step": 9630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013103788495719725,
+      "loss": 0.1019,
+      "step": 9631
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013102453624628542,
+      "loss": 0.0579,
+      "step": 9632
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013101118692364806,
+      "loss": 0.0744,
+      "step": 9633
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001309978369895484,
+      "loss": 0.11,
+      "step": 9634
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013098448644424966,
+      "loss": 0.0734,
+      "step": 9635
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001309711352880151,
+      "loss": 0.1163,
+      "step": 9636
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013095778352110795,
+      "loss": 0.104,
+      "step": 9637
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013094443114379146,
+      "loss": 0.0494,
+      "step": 9638
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013093107815632893,
+      "loss": 0.0937,
+      "step": 9639
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013091772455898364,
+      "loss": 0.1215,
+      "step": 9640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013090437035201885,
+      "loss": 0.0975,
+      "step": 9641
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013089101553569796,
+      "loss": 0.0677,
+      "step": 9642
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013087766011028422,
+      "loss": 0.0867,
+      "step": 9643
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00130864304076041,
+      "loss": 0.1564,
+      "step": 9644
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013085094743323163,
+      "loss": 0.0878,
+      "step": 9645
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013083759018211946,
+      "loss": 0.0552,
+      "step": 9646
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001308242323229679,
+      "loss": 0.0649,
+      "step": 9647
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013081087385604028,
+      "loss": 0.0842,
+      "step": 9648
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013079751478160006,
+      "loss": 0.1427,
+      "step": 9649
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001307841550999106,
+      "loss": 0.1124,
+      "step": 9650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013077079481123531,
+      "loss": 0.127,
+      "step": 9651
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013075743391583767,
+      "loss": 0.0733,
+      "step": 9652
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013074407241398106,
+      "loss": 0.0696,
+      "step": 9653
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00130730710305929,
+      "loss": 0.0768,
+      "step": 9654
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013071734759194493,
+      "loss": 0.0879,
+      "step": 9655
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001307039842722923,
+      "loss": 0.089,
+      "step": 9656
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013069062034723462,
+      "loss": 0.0886,
+      "step": 9657
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013067725581703544,
+      "loss": 0.1461,
+      "step": 9658
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013066389068195823,
+      "loss": 0.0696,
+      "step": 9659
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001306505249422665,
+      "loss": 0.1105,
+      "step": 9660
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.001306371585982238,
+      "loss": 0.1038,
+      "step": 9661
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013062379165009372,
+      "loss": 0.0784,
+      "step": 9662
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013061042409813976,
+      "loss": 0.0845,
+      "step": 9663
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0013059705594262556,
+      "loss": 0.1063,
+      "step": 9664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013058368718381467,
+      "loss": 0.1482,
+      "step": 9665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013057031782197069,
+      "loss": 0.108,
+      "step": 9666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013055694785735722,
+      "loss": 0.09,
+      "step": 9667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001305435772902379,
+      "loss": 0.0962,
+      "step": 9668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013053020612087637,
+      "loss": 0.0931,
+      "step": 9669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013051683434953624,
+      "loss": 0.092,
+      "step": 9670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001305034619764812,
+      "loss": 0.0753,
+      "step": 9671
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001304900890019749,
+      "loss": 0.0804,
+      "step": 9672
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013047671542628103,
+      "loss": 0.0821,
+      "step": 9673
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013046334124966331,
+      "loss": 0.0986,
+      "step": 9674
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001304499664723854,
+      "loss": 0.1487,
+      "step": 9675
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013043659109471108,
+      "loss": 0.0844,
+      "step": 9676
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013042321511690397,
+      "loss": 0.0917,
+      "step": 9677
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001304098385392279,
+      "loss": 0.1792,
+      "step": 9678
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013039646136194661,
+      "loss": 0.0845,
+      "step": 9679
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013038308358532381,
+      "loss": 0.1274,
+      "step": 9680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001303697052096234,
+      "loss": 0.0714,
+      "step": 9681
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013035632623510903,
+      "loss": 0.1392,
+      "step": 9682
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013034294666204455,
+      "loss": 0.0854,
+      "step": 9683
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013032956649069378,
+      "loss": 0.0808,
+      "step": 9684
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013031618572132057,
+      "loss": 0.0726,
+      "step": 9685
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013030280435418872,
+      "loss": 0.1116,
+      "step": 9686
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013028942238956205,
+      "loss": 0.1135,
+      "step": 9687
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013027603982770446,
+      "loss": 0.0498,
+      "step": 9688
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013026265666887984,
+      "loss": 0.0746,
+      "step": 9689
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00130249272913352,
+      "loss": 0.045,
+      "step": 9690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013023588856138494,
+      "loss": 0.118,
+      "step": 9691
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013022250361324247,
+      "loss": 0.1409,
+      "step": 9692
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013020911806918849,
+      "loss": 0.1041,
+      "step": 9693
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013019573192948701,
+      "loss": 0.1243,
+      "step": 9694
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013018234519440197,
+      "loss": 0.0959,
+      "step": 9695
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013016895786419728,
+      "loss": 0.0875,
+      "step": 9696
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001301555699391369,
+      "loss": 0.1805,
+      "step": 9697
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013014218141948481,
+      "loss": 0.0997,
+      "step": 9698
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013012879230550505,
+      "loss": 0.0961,
+      "step": 9699
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013011540259746158,
+      "loss": 0.1333,
+      "step": 9700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013010201229561838,
+      "loss": 0.0834,
+      "step": 9701
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013008862140023955,
+      "loss": 0.1104,
+      "step": 9702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013007522991158901,
+      "loss": 0.1071,
+      "step": 9703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013006183782993093,
+      "loss": 0.062,
+      "step": 9704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001300484451555293,
+      "loss": 0.0755,
+      "step": 9705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013003505188864822,
+      "loss": 0.0566,
+      "step": 9706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.001300216580295517,
+      "loss": 0.0985,
+      "step": 9707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0013000826357850396,
+      "loss": 0.0659,
+      "step": 9708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0012999486853576896,
+      "loss": 0.1313,
+      "step": 9709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0012998147290161092,
+      "loss": 0.0717,
+      "step": 9710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012996807667629395,
+      "loss": 0.0975,
+      "step": 9711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012995467986008217,
+      "loss": 0.0439,
+      "step": 9712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012994128245323972,
+      "loss": 0.0927,
+      "step": 9713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001299278844560308,
+      "loss": 0.1501,
+      "step": 9714
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012991448586871957,
+      "loss": 0.1167,
+      "step": 9715
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012990108669157022,
+      "loss": 0.0941,
+      "step": 9716
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012988768692484692,
+      "loss": 0.0803,
+      "step": 9717
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012987428656881389,
+      "loss": 0.1173,
+      "step": 9718
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012986088562373537,
+      "loss": 0.1057,
+      "step": 9719
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012984748408987558,
+      "loss": 0.1449,
+      "step": 9720
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012983408196749876,
+      "loss": 0.0995,
+      "step": 9721
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012982067925686922,
+      "loss": 0.0885,
+      "step": 9722
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012980727595825114,
+      "loss": 0.0494,
+      "step": 9723
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012979387207190888,
+      "loss": 0.1453,
+      "step": 9724
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012978046759810663,
+      "loss": 0.0423,
+      "step": 9725
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001297670625371088,
+      "loss": 0.1444,
+      "step": 9726
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012975365688917965,
+      "loss": 0.0677,
+      "step": 9727
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001297402506545835,
+      "loss": 0.1052,
+      "step": 9728
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001297268438335847,
+      "loss": 0.1495,
+      "step": 9729
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012971343642644761,
+      "loss": 0.0942,
+      "step": 9730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012970002843343655,
+      "loss": 0.0963,
+      "step": 9731
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012968661985481597,
+      "loss": 0.0981,
+      "step": 9732
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001296732106908502,
+      "loss": 0.0664,
+      "step": 9733
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012965980094180362,
+      "loss": 0.077,
+      "step": 9734
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012964639060794065,
+      "loss": 0.1104,
+      "step": 9735
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012963297968952572,
+      "loss": 0.1122,
+      "step": 9736
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012961956818682324,
+      "loss": 0.0812,
+      "step": 9737
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012960615610009768,
+      "loss": 0.1743,
+      "step": 9738
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001295927434296135,
+      "loss": 0.103,
+      "step": 9739
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001295793301756351,
+      "loss": 0.0821,
+      "step": 9740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00129565916338427,
+      "loss": 0.1268,
+      "step": 9741
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.001295525019182537,
+      "loss": 0.0693,
+      "step": 9742
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012953908691537967,
+      "loss": 0.1261,
+      "step": 9743
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012952567133006946,
+      "loss": 0.085,
+      "step": 9744
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012951225516258753,
+      "loss": 0.0656,
+      "step": 9745
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012949883841319844,
+      "loss": 0.1587,
+      "step": 9746
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012948542108216676,
+      "loss": 0.0972,
+      "step": 9747
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012947200316975702,
+      "loss": 0.1127,
+      "step": 9748
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012945858467623382,
+      "loss": 0.1384,
+      "step": 9749
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012944516560186169,
+      "loss": 0.0518,
+      "step": 9750
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012943174594690523,
+      "loss": 0.0814,
+      "step": 9751
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012941832571162907,
+      "loss": 0.0885,
+      "step": 9752
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012940490489629778,
+      "loss": 0.1049,
+      "step": 9753
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012939148350117604,
+      "loss": 0.1272,
+      "step": 9754
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012937806152652847,
+      "loss": 0.0508,
+      "step": 9755
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0012936463897261967,
+      "loss": 0.0717,
+      "step": 9756
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012935121583971434,
+      "loss": 0.0864,
+      "step": 9757
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012933779212807718,
+      "loss": 0.1115,
+      "step": 9758
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012932436783797282,
+      "loss": 0.0963,
+      "step": 9759
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012931094296966595,
+      "loss": 0.0791,
+      "step": 9760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012929751752342132,
+      "loss": 0.0559,
+      "step": 9761
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001292840914995036,
+      "loss": 0.0637,
+      "step": 9762
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012927066489817755,
+      "loss": 0.0695,
+      "step": 9763
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012925723771970788,
+      "loss": 0.094,
+      "step": 9764
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012924380996435939,
+      "loss": 0.0832,
+      "step": 9765
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012923038163239677,
+      "loss": 0.1053,
+      "step": 9766
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012921695272408485,
+      "loss": 0.0848,
+      "step": 9767
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012920352323968838,
+      "loss": 0.0701,
+      "step": 9768
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012919009317947218,
+      "loss": 0.0789,
+      "step": 9769
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012917666254370103,
+      "loss": 0.0526,
+      "step": 9770
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012916323133263978,
+      "loss": 0.147,
+      "step": 9771
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012914979954655327,
+      "loss": 0.0991,
+      "step": 9772
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012913636718570624,
+      "loss": 0.0735,
+      "step": 9773
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001291229342503637,
+      "loss": 0.0803,
+      "step": 9774
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001291095007407904,
+      "loss": 0.1121,
+      "step": 9775
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012909606665725127,
+      "loss": 0.0811,
+      "step": 9776
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012908263200001114,
+      "loss": 0.0907,
+      "step": 9777
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012906919676933493,
+      "loss": 0.1564,
+      "step": 9778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012905576096548757,
+      "loss": 0.1152,
+      "step": 9779
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012904232458873397,
+      "loss": 0.0826,
+      "step": 9780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012902888763933907,
+      "loss": 0.0621,
+      "step": 9781
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012901545011756782,
+      "loss": 0.0639,
+      "step": 9782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012900201202368513,
+      "loss": 0.1035,
+      "step": 9783
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012898857335795597,
+      "loss": 0.0903,
+      "step": 9784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001289751341206454,
+      "loss": 0.0925,
+      "step": 9785
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001289616943120183,
+      "loss": 0.0873,
+      "step": 9786
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012894825393233971,
+      "loss": 0.146,
+      "step": 9787
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012893481298187467,
+      "loss": 0.0779,
+      "step": 9788
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012892137146088818,
+      "loss": 0.0484,
+      "step": 9789
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012890792936964526,
+      "loss": 0.0643,
+      "step": 9790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012889448670841098,
+      "loss": 0.1285,
+      "step": 9791
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012888104347745038,
+      "loss": 0.083,
+      "step": 9792
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001288675996770285,
+      "loss": 0.0797,
+      "step": 9793
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012885415530741043,
+      "loss": 0.0609,
+      "step": 9794
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012884071036886131,
+      "loss": 0.0939,
+      "step": 9795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012882726486164622,
+      "loss": 0.131,
+      "step": 9796
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.001288138187860302,
+      "loss": 0.083,
+      "step": 9797
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012880037214227848,
+      "loss": 0.0527,
+      "step": 9798
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012878692493065612,
+      "loss": 0.0864,
+      "step": 9799
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012877347715142826,
+      "loss": 0.1593,
+      "step": 9800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012876002880486013,
+      "loss": 0.1108,
+      "step": 9801
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0012874657989121682,
+      "loss": 0.0957,
+      "step": 9802
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012873313041076356,
+      "loss": 0.0961,
+      "step": 9803
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012871968036376551,
+      "loss": 0.1411,
+      "step": 9804
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012870622975048786,
+      "loss": 0.104,
+      "step": 9805
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012869277857119588,
+      "loss": 0.1202,
+      "step": 9806
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012867932682615471,
+      "loss": 0.1268,
+      "step": 9807
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012866587451562967,
+      "loss": 0.0655,
+      "step": 9808
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012865242163988596,
+      "loss": 0.0933,
+      "step": 9809
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001286389681991888,
+      "loss": 0.1494,
+      "step": 9810
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012862551419380354,
+      "loss": 0.1068,
+      "step": 9811
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012861205962399543,
+      "loss": 0.1051,
+      "step": 9812
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001285986044900297,
+      "loss": 0.0591,
+      "step": 9813
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012858514879217173,
+      "loss": 0.1984,
+      "step": 9814
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001285716925306868,
+      "loss": 0.0887,
+      "step": 9815
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012855823570584021,
+      "loss": 0.0999,
+      "step": 9816
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012854477831789736,
+      "loss": 0.0833,
+      "step": 9817
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012853132036712356,
+      "loss": 0.0814,
+      "step": 9818
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012851786185378415,
+      "loss": 0.0652,
+      "step": 9819
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012850440277814453,
+      "loss": 0.1019,
+      "step": 9820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012849094314047,
+      "loss": 0.0654,
+      "step": 9821
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012847748294102606,
+      "loss": 0.0792,
+      "step": 9822
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012846402218007805,
+      "loss": 0.0848,
+      "step": 9823
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001284505608578914,
+      "loss": 0.0888,
+      "step": 9824
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012843709897473153,
+      "loss": 0.0674,
+      "step": 9825
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012842363653086384,
+      "loss": 0.0698,
+      "step": 9826
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012841017352655385,
+      "loss": 0.1241,
+      "step": 9827
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012839670996206694,
+      "loss": 0.0627,
+      "step": 9828
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012838324583766864,
+      "loss": 0.095,
+      "step": 9829
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012836978115362437,
+      "loss": 0.1271,
+      "step": 9830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012835631591019965,
+      "loss": 0.1449,
+      "step": 9831
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012834285010766,
+      "loss": 0.1125,
+      "step": 9832
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012832938374627088,
+      "loss": 0.0577,
+      "step": 9833
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012831591682629787,
+      "loss": 0.1912,
+      "step": 9834
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012830244934800648,
+      "loss": 0.084,
+      "step": 9835
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012828898131166224,
+      "loss": 0.0683,
+      "step": 9836
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001282755127175307,
+      "loss": 0.1013,
+      "step": 9837
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012826204356587748,
+      "loss": 0.1158,
+      "step": 9838
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012824857385696812,
+      "loss": 0.1036,
+      "step": 9839
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001282351035910682,
+      "loss": 0.0897,
+      "step": 9840
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012822163276844333,
+      "loss": 0.0677,
+      "step": 9841
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001282081613893591,
+      "loss": 0.1499,
+      "step": 9842
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012819468945408114,
+      "loss": 0.073,
+      "step": 9843
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012818121696287514,
+      "loss": 0.0786,
+      "step": 9844
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.001281677439160067,
+      "loss": 0.036,
+      "step": 9845
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012815427031374143,
+      "loss": 0.0781,
+      "step": 9846
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012814079615634506,
+      "loss": 0.0716,
+      "step": 9847
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0012812732144408326,
+      "loss": 0.0765,
+      "step": 9848
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012811384617722168,
+      "loss": 0.0798,
+      "step": 9849
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012810037035602604,
+      "loss": 0.0777,
+      "step": 9850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012808689398076207,
+      "loss": 0.0593,
+      "step": 9851
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012807341705169547,
+      "loss": 0.121,
+      "step": 9852
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012805993956909194,
+      "loss": 0.0984,
+      "step": 9853
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012804646153321728,
+      "loss": 0.0922,
+      "step": 9854
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012803298294433722,
+      "loss": 0.0839,
+      "step": 9855
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.001280195038027175,
+      "loss": 0.1394,
+      "step": 9856
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012800602410862393,
+      "loss": 0.0817,
+      "step": 9857
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012799254386232225,
+      "loss": 0.1065,
+      "step": 9858
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012797906306407831,
+      "loss": 0.1124,
+      "step": 9859
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012796558171415788,
+      "loss": 0.098,
+      "step": 9860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.001279520998128268,
+      "loss": 0.1649,
+      "step": 9861
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012793861736035092,
+      "loss": 0.1808,
+      "step": 9862
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00127925134356996,
+      "loss": 0.1218,
+      "step": 9863
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012791165080302794,
+      "loss": 0.0828,
+      "step": 9864
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012789816669871263,
+      "loss": 0.1084,
+      "step": 9865
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012788468204431589,
+      "loss": 0.1126,
+      "step": 9866
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012787119684010367,
+      "loss": 0.1005,
+      "step": 9867
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012785771108634178,
+      "loss": 0.1907,
+      "step": 9868
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.001278442247832962,
+      "loss": 0.1011,
+      "step": 9869
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012783073793123277,
+      "loss": 0.1447,
+      "step": 9870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.001278172505304175,
+      "loss": 0.0988,
+      "step": 9871
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012780376258111628,
+      "loss": 0.0833,
+      "step": 9872
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012779027408359506,
+      "loss": 0.0958,
+      "step": 9873
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012777678503811977,
+      "loss": 0.0925,
+      "step": 9874
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012776329544495649,
+      "loss": 0.1041,
+      "step": 9875
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012774980530437107,
+      "loss": 0.092,
+      "step": 9876
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012773631461662958,
+      "loss": 0.1041,
+      "step": 9877
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00127722823381998,
+      "loss": 0.1152,
+      "step": 9878
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012770933160074231,
+      "loss": 0.0906,
+      "step": 9879
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012769583927312858,
+      "loss": 0.1003,
+      "step": 9880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012768234639942285,
+      "loss": 0.0807,
+      "step": 9881
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012766885297989116,
+      "loss": 0.15,
+      "step": 9882
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.001276553590147995,
+      "loss": 0.0735,
+      "step": 9883
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012764186450441404,
+      "loss": 0.1096,
+      "step": 9884
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012762836944900076,
+      "loss": 0.0784,
+      "step": 9885
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012761487384882581,
+      "loss": 0.0823,
+      "step": 9886
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.001276013777041553,
+      "loss": 0.1021,
+      "step": 9887
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012758788101525532,
+      "loss": 0.0911,
+      "step": 9888
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012757438378239194,
+      "loss": 0.1002,
+      "step": 9889
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012756088600583135,
+      "loss": 0.1311,
+      "step": 9890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012754738768583971,
+      "loss": 0.0861,
+      "step": 9891
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012753388882268313,
+      "loss": 0.094,
+      "step": 9892
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012752038941662776,
+      "loss": 0.1285,
+      "step": 9893
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0012750688946793981,
+      "loss": 0.0711,
+      "step": 9894
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012749338897688545,
+      "loss": 0.097,
+      "step": 9895
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.001274798879437309,
+      "loss": 0.1141,
+      "step": 9896
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012746638636874233,
+      "loss": 0.1008,
+      "step": 9897
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00127452884252186,
+      "loss": 0.0919,
+      "step": 9898
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012743938159432808,
+      "loss": 0.1049,
+      "step": 9899
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012742587839543484,
+      "loss": 0.1208,
+      "step": 9900
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012741237465577254,
+      "loss": 0.082,
+      "step": 9901
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.001273988703756074,
+      "loss": 0.1018,
+      "step": 9902
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012738536555520574,
+      "loss": 0.1196,
+      "step": 9903
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012737186019483383,
+      "loss": 0.1617,
+      "step": 9904
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012735835429475793,
+      "loss": 0.0891,
+      "step": 9905
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012734484785524437,
+      "loss": 0.104,
+      "step": 9906
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012733134087655944,
+      "loss": 0.0895,
+      "step": 9907
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012731783335896952,
+      "loss": 0.1429,
+      "step": 9908
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012730432530274087,
+      "loss": 0.0837,
+      "step": 9909
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012729081670813987,
+      "loss": 0.1198,
+      "step": 9910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012727730757543289,
+      "loss": 0.1379,
+      "step": 9911
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012726379790488625,
+      "loss": 0.1162,
+      "step": 9912
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012725028769676636,
+      "loss": 0.0858,
+      "step": 9913
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012723677695133963,
+      "loss": 0.0543,
+      "step": 9914
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012722326566887243,
+      "loss": 0.0784,
+      "step": 9915
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012720975384963114,
+      "loss": 0.0749,
+      "step": 9916
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012719624149388222,
+      "loss": 0.0812,
+      "step": 9917
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012718272860189211,
+      "loss": 0.1417,
+      "step": 9918
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.001271692151739272,
+      "loss": 0.0762,
+      "step": 9919
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012715570121025397,
+      "loss": 0.0823,
+      "step": 9920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.001271421867111389,
+      "loss": 0.1018,
+      "step": 9921
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012712867167684841,
+      "loss": 0.0621,
+      "step": 9922
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012711515610764903,
+      "loss": 0.0646,
+      "step": 9923
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012710164000380724,
+      "loss": 0.1166,
+      "step": 9924
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012708812336558954,
+      "loss": 0.1167,
+      "step": 9925
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012707460619326245,
+      "loss": 0.1221,
+      "step": 9926
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012706108848709245,
+      "loss": 0.066,
+      "step": 9927
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012704757024734616,
+      "loss": 0.0645,
+      "step": 9928
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012703405147429009,
+      "loss": 0.0857,
+      "step": 9929
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012702053216819074,
+      "loss": 0.1885,
+      "step": 9930
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012700701232931475,
+      "loss": 0.0817,
+      "step": 9931
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012699349195792865,
+      "loss": 0.0589,
+      "step": 9932
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012697997105429906,
+      "loss": 0.0714,
+      "step": 9933
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012696644961869255,
+      "loss": 0.1108,
+      "step": 9934
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012695292765137576,
+      "loss": 0.0811,
+      "step": 9935
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012693940515261528,
+      "loss": 0.0829,
+      "step": 9936
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012692588212267777,
+      "loss": 0.0878,
+      "step": 9937
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012691235856182983,
+      "loss": 0.1426,
+      "step": 9938
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012689883447033817,
+      "loss": 0.1558,
+      "step": 9939
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012688530984846936,
+      "loss": 0.0684,
+      "step": 9940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0012687178469649016,
+      "loss": 0.0658,
+      "step": 9941
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012685825901466719,
+      "loss": 0.0861,
+      "step": 9942
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012684473280326718,
+      "loss": 0.0971,
+      "step": 9943
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012683120606255683,
+      "loss": 0.1014,
+      "step": 9944
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012681767879280286,
+      "loss": 0.0945,
+      "step": 9945
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012680415099427195,
+      "loss": 0.0544,
+      "step": 9946
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012679062266723088,
+      "loss": 0.1229,
+      "step": 9947
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012677709381194635,
+      "loss": 0.0892,
+      "step": 9948
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012676356442868518,
+      "loss": 0.1255,
+      "step": 9949
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012675003451771405,
+      "loss": 0.1047,
+      "step": 9950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012673650407929982,
+      "loss": 0.12,
+      "step": 9951
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012672297311370924,
+      "loss": 0.0552,
+      "step": 9952
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012670944162120907,
+      "loss": 0.0853,
+      "step": 9953
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012669590960206617,
+      "loss": 0.0895,
+      "step": 9954
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012668237705654734,
+      "loss": 0.1051,
+      "step": 9955
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012666884398491941,
+      "loss": 0.1118,
+      "step": 9956
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012665531038744922,
+      "loss": 0.071,
+      "step": 9957
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012664177626440364,
+      "loss": 0.084,
+      "step": 9958
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012662824161604944,
+      "loss": 0.2236,
+      "step": 9959
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012661470644265357,
+      "loss": 0.0837,
+      "step": 9960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012660117074448291,
+      "loss": 0.0901,
+      "step": 9961
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012658763452180434,
+      "loss": 0.0714,
+      "step": 9962
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012657409777488472,
+      "loss": 0.1963,
+      "step": 9963
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012656056050399098,
+      "loss": 0.1082,
+      "step": 9964
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012654702270939009,
+      "loss": 0.0727,
+      "step": 9965
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012653348439134891,
+      "loss": 0.144,
+      "step": 9966
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012651994555013444,
+      "loss": 0.1403,
+      "step": 9967
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012650640618601361,
+      "loss": 0.1033,
+      "step": 9968
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012649286629925337,
+      "loss": 0.0837,
+      "step": 9969
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012647932589012068,
+      "loss": 0.087,
+      "step": 9970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012646578495888257,
+      "loss": 0.1244,
+      "step": 9971
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00126452243505806,
+      "loss": 0.1152,
+      "step": 9972
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012643870153115795,
+      "loss": 0.1167,
+      "step": 9973
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.001264251590352055,
+      "loss": 0.106,
+      "step": 9974
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.001264116160182156,
+      "loss": 0.1306,
+      "step": 9975
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012639807248045534,
+      "loss": 0.0605,
+      "step": 9976
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012638452842219177,
+      "loss": 0.1036,
+      "step": 9977
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.001263709838436919,
+      "loss": 0.1028,
+      "step": 9978
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012635743874522282,
+      "loss": 0.119,
+      "step": 9979
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012634389312705158,
+      "loss": 0.1272,
+      "step": 9980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.001263303469894453,
+      "loss": 0.0788,
+      "step": 9981
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012631680033267107,
+      "loss": 0.1034,
+      "step": 9982
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012630325315699596,
+      "loss": 0.1083,
+      "step": 9983
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012628970546268714,
+      "loss": 0.0715,
+      "step": 9984
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.001262761572500117,
+      "loss": 0.0969,
+      "step": 9985
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012626260851923677,
+      "loss": 0.0985,
+      "step": 9986
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0012624905927062953,
+      "loss": 0.1008,
+      "step": 9987
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012623550950445713,
+      "loss": 0.1248,
+      "step": 9988
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001262219592209867,
+      "loss": 0.0986,
+      "step": 9989
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012620840842048544,
+      "loss": 0.1173,
+      "step": 9990
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012619485710322056,
+      "loss": 0.0749,
+      "step": 9991
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012618130526945925,
+      "loss": 0.1028,
+      "step": 9992
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012616775291946869,
+      "loss": 0.1116,
+      "step": 9993
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012615420005351612,
+      "loss": 0.0836,
+      "step": 9994
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012614064667186879,
+      "loss": 0.0743,
+      "step": 9995
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012612709277479385,
+      "loss": 0.1558,
+      "step": 9996
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012611353836255864,
+      "loss": 0.0707,
+      "step": 9997
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001260999834354304,
+      "loss": 0.1417,
+      "step": 9998
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001260864279936764,
+      "loss": 0.0887,
+      "step": 9999
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012607287203756387,
+      "loss": 0.108,
+      "step": 10000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012605931556736016,
+      "loss": 0.053,
+      "step": 10001
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012604575858333256,
+      "loss": 0.0917,
+      "step": 10002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012603220108574832,
+      "loss": 0.1026,
+      "step": 10003
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012601864307487485,
+      "loss": 0.1146,
+      "step": 10004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012600508455097943,
+      "loss": 0.0848,
+      "step": 10005
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001259915255143294,
+      "loss": 0.0551,
+      "step": 10006
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001259779659651921,
+      "loss": 0.1089,
+      "step": 10007
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012596440590383492,
+      "loss": 0.0917,
+      "step": 10008
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012595084533052522,
+      "loss": 0.0856,
+      "step": 10009
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012593728424553036,
+      "loss": 0.099,
+      "step": 10010
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012592372264911774,
+      "loss": 0.1782,
+      "step": 10011
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012591016054155477,
+      "loss": 0.1105,
+      "step": 10012
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012589659792310885,
+      "loss": 0.0649,
+      "step": 10013
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012588303479404745,
+      "loss": 0.0648,
+      "step": 10014
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012586947115463794,
+      "loss": 0.0739,
+      "step": 10015
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012585590700514774,
+      "loss": 0.073,
+      "step": 10016
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012584234234584435,
+      "loss": 0.1339,
+      "step": 10017
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012582877717699525,
+      "loss": 0.0999,
+      "step": 10018
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012581521149886787,
+      "loss": 0.0899,
+      "step": 10019
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001258016453117297,
+      "loss": 0.0851,
+      "step": 10020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012578807861584823,
+      "loss": 0.0936,
+      "step": 10021
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012577451141149096,
+      "loss": 0.1541,
+      "step": 10022
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001257609436989254,
+      "loss": 0.094,
+      "step": 10023
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001257473754784191,
+      "loss": 0.0721,
+      "step": 10024
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012573380675023958,
+      "loss": 0.0872,
+      "step": 10025
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012572023751465432,
+      "loss": 0.0912,
+      "step": 10026
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012570666777193095,
+      "loss": 0.0681,
+      "step": 10027
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012569309752233698,
+      "loss": 0.0808,
+      "step": 10028
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012567952676614002,
+      "loss": 0.0963,
+      "step": 10029
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012566595550360762,
+      "loss": 0.1193,
+      "step": 10030
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.001256523837350074,
+      "loss": 0.1059,
+      "step": 10031
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012563881146060694,
+      "loss": 0.0801,
+      "step": 10032
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0012562523868067384,
+      "loss": 0.1694,
+      "step": 10033
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012561166539547576,
+      "loss": 0.1316,
+      "step": 10034
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001255980916052803,
+      "loss": 0.0594,
+      "step": 10035
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012558451731035508,
+      "loss": 0.0733,
+      "step": 10036
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001255709425109678,
+      "loss": 0.1024,
+      "step": 10037
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012555736720738608,
+      "loss": 0.0764,
+      "step": 10038
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012554379139987763,
+      "loss": 0.1111,
+      "step": 10039
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012553021508871012,
+      "loss": 0.0988,
+      "step": 10040
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012551663827415124,
+      "loss": 0.0657,
+      "step": 10041
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012550306095646866,
+      "loss": 0.1479,
+      "step": 10042
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012548948313593012,
+      "loss": 0.0917,
+      "step": 10043
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001254759048128033,
+      "loss": 0.0668,
+      "step": 10044
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00125462325987356,
+      "loss": 0.0968,
+      "step": 10045
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012544874665985591,
+      "loss": 0.1243,
+      "step": 10046
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012543516683057082,
+      "loss": 0.0729,
+      "step": 10047
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012542158649976845,
+      "loss": 0.1013,
+      "step": 10048
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012540800566771654,
+      "loss": 0.0745,
+      "step": 10049
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012539442433468296,
+      "loss": 0.1046,
+      "step": 10050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012538084250093545,
+      "loss": 0.0903,
+      "step": 10051
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001253672601667418,
+      "loss": 0.0879,
+      "step": 10052
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012535367733236985,
+      "loss": 0.0876,
+      "step": 10053
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012534009399808735,
+      "loss": 0.1663,
+      "step": 10054
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012532651016416224,
+      "loss": 0.1091,
+      "step": 10055
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012531292583086225,
+      "loss": 0.0912,
+      "step": 10056
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001252993409984553,
+      "loss": 0.0807,
+      "step": 10057
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012528575566720923,
+      "loss": 0.1198,
+      "step": 10058
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012527216983739189,
+      "loss": 0.1094,
+      "step": 10059
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012525858350927116,
+      "loss": 0.0549,
+      "step": 10060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00125244996683115,
+      "loss": 0.0966,
+      "step": 10061
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012523140935919122,
+      "loss": 0.1379,
+      "step": 10062
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012521782153776773,
+      "loss": 0.1851,
+      "step": 10063
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001252042332191125,
+      "loss": 0.0768,
+      "step": 10064
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012519064440349342,
+      "loss": 0.0736,
+      "step": 10065
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001251770550911784,
+      "loss": 0.0485,
+      "step": 10066
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012516346528243546,
+      "loss": 0.1473,
+      "step": 10067
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012514987497753256,
+      "loss": 0.1193,
+      "step": 10068
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001251362841767376,
+      "loss": 0.1281,
+      "step": 10069
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012512269288031857,
+      "loss": 0.1281,
+      "step": 10070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012510910108854347,
+      "loss": 0.0883,
+      "step": 10071
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012509550880168033,
+      "loss": 0.0979,
+      "step": 10072
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012508191601999707,
+      "loss": 0.1034,
+      "step": 10073
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001250683227437618,
+      "loss": 0.1041,
+      "step": 10074
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012505472897324248,
+      "loss": 0.0617,
+      "step": 10075
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012504113470870716,
+      "loss": 0.1104,
+      "step": 10076
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001250275399504239,
+      "loss": 0.1107,
+      "step": 10077
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0012501394469866076,
+      "loss": 0.1385,
+      "step": 10078
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.001250003489536858,
+      "loss": 0.1064,
+      "step": 10079
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012498675271576708,
+      "loss": 0.1182,
+      "step": 10080
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012497315598517264,
+      "loss": 0.078,
+      "step": 10081
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012495955876217068,
+      "loss": 0.1084,
+      "step": 10082
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001249459610470292,
+      "loss": 0.1206,
+      "step": 10083
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012493236284001638,
+      "loss": 0.0929,
+      "step": 10084
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012491876414140034,
+      "loss": 0.1764,
+      "step": 10085
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012490516495144912,
+      "loss": 0.093,
+      "step": 10086
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012489156527043098,
+      "loss": 0.0811,
+      "step": 10087
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012487796509861402,
+      "loss": 0.0832,
+      "step": 10088
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012486436443626642,
+      "loss": 0.0858,
+      "step": 10089
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001248507632836563,
+      "loss": 0.1057,
+      "step": 10090
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012483716164105188,
+      "loss": 0.0723,
+      "step": 10091
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012482355950872138,
+      "loss": 0.0977,
+      "step": 10092
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012480995688693294,
+      "loss": 0.1251,
+      "step": 10093
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001247963537759548,
+      "loss": 0.1564,
+      "step": 10094
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001247827501760552,
+      "loss": 0.087,
+      "step": 10095
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001247691460875023,
+      "loss": 0.0978,
+      "step": 10096
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012475554151056437,
+      "loss": 0.0834,
+      "step": 10097
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001247419364455097,
+      "loss": 0.1548,
+      "step": 10098
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012472833089260651,
+      "loss": 0.0732,
+      "step": 10099
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012471472485212308,
+      "loss": 0.1284,
+      "step": 10100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012470111832432768,
+      "loss": 0.1095,
+      "step": 10101
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012468751130948858,
+      "loss": 0.1146,
+      "step": 10102
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012467390380787409,
+      "loss": 0.0798,
+      "step": 10103
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012466029581975253,
+      "loss": 0.1001,
+      "step": 10104
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012464668734539221,
+      "loss": 0.0817,
+      "step": 10105
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012463307838506143,
+      "loss": 0.0604,
+      "step": 10106
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012461946893902857,
+      "loss": 0.0756,
+      "step": 10107
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012460585900756192,
+      "loss": 0.0901,
+      "step": 10108
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012459224859092985,
+      "loss": 0.0956,
+      "step": 10109
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012457863768940076,
+      "loss": 0.0881,
+      "step": 10110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00124565026303243,
+      "loss": 0.1072,
+      "step": 10111
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012455141443272492,
+      "loss": 0.0936,
+      "step": 10112
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012453780207811497,
+      "loss": 0.1038,
+      "step": 10113
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012452418923968149,
+      "loss": 0.0677,
+      "step": 10114
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012451057591769295,
+      "loss": 0.0759,
+      "step": 10115
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012449696211241771,
+      "loss": 0.0486,
+      "step": 10116
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012448334782412427,
+      "loss": 0.0798,
+      "step": 10117
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012446973305308099,
+      "loss": 0.1252,
+      "step": 10118
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012445611779955637,
+      "loss": 0.0848,
+      "step": 10119
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012444250206381886,
+      "loss": 0.069,
+      "step": 10120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012442888584613696,
+      "loss": 0.0968,
+      "step": 10121
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.001244152691467791,
+      "loss": 0.135,
+      "step": 10122
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012440165196601373,
+      "loss": 0.0986,
+      "step": 10123
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012438803430410944,
+      "loss": 0.1106,
+      "step": 10124
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0012437441616133468,
+      "loss": 0.1066,
+      "step": 10125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012436079753795797,
+      "loss": 0.1118,
+      "step": 10126
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012434717843424787,
+      "loss": 0.1478,
+      "step": 10127
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012433355885047288,
+      "loss": 0.1249,
+      "step": 10128
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012431993878690153,
+      "loss": 0.0545,
+      "step": 10129
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001243063182438024,
+      "loss": 0.0936,
+      "step": 10130
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012429269722144408,
+      "loss": 0.1084,
+      "step": 10131
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001242790757200951,
+      "loss": 0.0899,
+      "step": 10132
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012426545374002403,
+      "loss": 0.1163,
+      "step": 10133
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012425183128149946,
+      "loss": 0.1062,
+      "step": 10134
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012423820834479005,
+      "loss": 0.0797,
+      "step": 10135
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012422458493016434,
+      "loss": 0.0929,
+      "step": 10136
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012421096103789103,
+      "loss": 0.0884,
+      "step": 10137
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012419733666823869,
+      "loss": 0.085,
+      "step": 10138
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012418371182147594,
+      "loss": 0.097,
+      "step": 10139
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012417008649787144,
+      "loss": 0.0993,
+      "step": 10140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012415646069769393,
+      "loss": 0.0725,
+      "step": 10141
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012414283442121197,
+      "loss": 0.0705,
+      "step": 10142
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012412920766869426,
+      "loss": 0.1152,
+      "step": 10143
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012411558044040953,
+      "loss": 0.0809,
+      "step": 10144
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012410195273662643,
+      "loss": 0.0881,
+      "step": 10145
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012408832455761368,
+      "loss": 0.0605,
+      "step": 10146
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012407469590363998,
+      "loss": 0.1018,
+      "step": 10147
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001240610667749741,
+      "loss": 0.0654,
+      "step": 10148
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012404743717188472,
+      "loss": 0.0552,
+      "step": 10149
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012403380709464056,
+      "loss": 0.0919,
+      "step": 10150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012402017654351046,
+      "loss": 0.0735,
+      "step": 10151
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012400654551876311,
+      "loss": 0.0678,
+      "step": 10152
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001239929140206673,
+      "loss": 0.0978,
+      "step": 10153
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001239792820494918,
+      "loss": 0.078,
+      "step": 10154
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001239656496055054,
+      "loss": 0.1239,
+      "step": 10155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012395201668897692,
+      "loss": 0.0536,
+      "step": 10156
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012393838330017515,
+      "loss": 0.1121,
+      "step": 10157
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012392474943936891,
+      "loss": 0.0835,
+      "step": 10158
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00123911115106827,
+      "loss": 0.1086,
+      "step": 10159
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001238974803028183,
+      "loss": 0.0732,
+      "step": 10160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012388384502761158,
+      "loss": 0.1415,
+      "step": 10161
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012387020928147578,
+      "loss": 0.1959,
+      "step": 10162
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012385657306467973,
+      "loss": 0.089,
+      "step": 10163
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012384293637749228,
+      "loss": 0.0858,
+      "step": 10164
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012382929922018235,
+      "loss": 0.1127,
+      "step": 10165
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012381566159301875,
+      "loss": 0.0763,
+      "step": 10166
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.001238020234962705,
+      "loss": 0.1107,
+      "step": 10167
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012378838493020644,
+      "loss": 0.1121,
+      "step": 10168
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012377474589509545,
+      "loss": 0.1193,
+      "step": 10169
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012376110639120655,
+      "loss": 0.0848,
+      "step": 10170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0012374746641880862,
+      "loss": 0.0825,
+      "step": 10171
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001237338259781706,
+      "loss": 0.1053,
+      "step": 10172
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012372018506956145,
+      "loss": 0.1182,
+      "step": 10173
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012370654369325015,
+      "loss": 0.1089,
+      "step": 10174
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001236929018495057,
+      "loss": 0.1209,
+      "step": 10175
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00123679259538597,
+      "loss": 0.0824,
+      "step": 10176
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001236656167607931,
+      "loss": 0.0619,
+      "step": 10177
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012365197351636302,
+      "loss": 0.1083,
+      "step": 10178
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001236383298055757,
+      "loss": 0.1095,
+      "step": 10179
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012362468562870025,
+      "loss": 0.0927,
+      "step": 10180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012361104098600564,
+      "loss": 0.0977,
+      "step": 10181
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012359739587776092,
+      "loss": 0.0916,
+      "step": 10182
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001235837503042351,
+      "loss": 0.0984,
+      "step": 10183
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001235701042656973,
+      "loss": 0.138,
+      "step": 10184
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012355645776241657,
+      "loss": 0.0876,
+      "step": 10185
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012354281079466193,
+      "loss": 0.0856,
+      "step": 10186
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012352916336270255,
+      "loss": 0.0787,
+      "step": 10187
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012351551546680747,
+      "loss": 0.1459,
+      "step": 10188
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012350186710724579,
+      "loss": 0.1417,
+      "step": 10189
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012348821828428664,
+      "loss": 0.1173,
+      "step": 10190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012347456899819914,
+      "loss": 0.0963,
+      "step": 10191
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001234609192492524,
+      "loss": 0.1238,
+      "step": 10192
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012344726903771557,
+      "loss": 0.0718,
+      "step": 10193
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001234336183638578,
+      "loss": 0.0683,
+      "step": 10194
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012341996722794826,
+      "loss": 0.0969,
+      "step": 10195
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012340631563025608,
+      "loss": 0.0888,
+      "step": 10196
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012339266357105046,
+      "loss": 0.0944,
+      "step": 10197
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012337901105060058,
+      "loss": 0.0988,
+      "step": 10198
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012336535806917563,
+      "loss": 0.157,
+      "step": 10199
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012335170462704484,
+      "loss": 0.1792,
+      "step": 10200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012333805072447738,
+      "loss": 0.0912,
+      "step": 10201
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001233243963617425,
+      "loss": 0.1564,
+      "step": 10202
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012331074153910937,
+      "loss": 0.0895,
+      "step": 10203
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012329708625684731,
+      "loss": 0.0602,
+      "step": 10204
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012328343051522554,
+      "loss": 0.127,
+      "step": 10205
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012326977431451332,
+      "loss": 0.1287,
+      "step": 10206
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.001232561176549799,
+      "loss": 0.0854,
+      "step": 10207
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012324246053689455,
+      "loss": 0.1006,
+      "step": 10208
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012322880296052655,
+      "loss": 0.0712,
+      "step": 10209
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012321514492614524,
+      "loss": 0.1293,
+      "step": 10210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012320148643401986,
+      "loss": 0.1709,
+      "step": 10211
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012318782748441978,
+      "loss": 0.1165,
+      "step": 10212
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012317416807761429,
+      "loss": 0.0894,
+      "step": 10213
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012316050821387268,
+      "loss": 0.1278,
+      "step": 10214
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012314684789346436,
+      "loss": 0.1151,
+      "step": 10215
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012313318711665864,
+      "loss": 0.055,
+      "step": 10216
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0012311952588372488,
+      "loss": 0.0856,
+      "step": 10217
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012310586419493246,
+      "loss": 0.0391,
+      "step": 10218
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012309220205055072,
+      "loss": 0.1221,
+      "step": 10219
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012307853945084904,
+      "loss": 0.0762,
+      "step": 10220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012306487639609687,
+      "loss": 0.1158,
+      "step": 10221
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012305121288656358,
+      "loss": 0.0742,
+      "step": 10222
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012303754892251855,
+      "loss": 0.0875,
+      "step": 10223
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012302388450423124,
+      "loss": 0.0668,
+      "step": 10224
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012301021963197105,
+      "loss": 0.1718,
+      "step": 10225
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012299655430600745,
+      "loss": 0.1168,
+      "step": 10226
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012298288852660986,
+      "loss": 0.1323,
+      "step": 10227
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012296922229404774,
+      "loss": 0.0931,
+      "step": 10228
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012295555560859052,
+      "loss": 0.1185,
+      "step": 10229
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012294188847050773,
+      "loss": 0.1381,
+      "step": 10230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012292822088006885,
+      "loss": 0.0563,
+      "step": 10231
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012291455283754333,
+      "loss": 0.1436,
+      "step": 10232
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012290088434320067,
+      "loss": 0.0884,
+      "step": 10233
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012288721539731042,
+      "loss": 0.089,
+      "step": 10234
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012287354600014205,
+      "loss": 0.1025,
+      "step": 10235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001228598761519651,
+      "loss": 0.0801,
+      "step": 10236
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012284620585304915,
+      "loss": 0.1412,
+      "step": 10237
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001228325351036637,
+      "loss": 0.0705,
+      "step": 10238
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012281886390407828,
+      "loss": 0.1028,
+      "step": 10239
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001228051922545625,
+      "loss": 0.0664,
+      "step": 10240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001227915201553859,
+      "loss": 0.1631,
+      "step": 10241
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012277784760681809,
+      "loss": 0.0911,
+      "step": 10242
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012276417460912862,
+      "loss": 0.102,
+      "step": 10243
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012275050116258713,
+      "loss": 0.0445,
+      "step": 10244
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012273682726746317,
+      "loss": 0.0808,
+      "step": 10245
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001227231529240264,
+      "loss": 0.1234,
+      "step": 10246
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012270947813254646,
+      "loss": 0.0972,
+      "step": 10247
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012269580289329292,
+      "loss": 0.096,
+      "step": 10248
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012268212720653545,
+      "loss": 0.1068,
+      "step": 10249
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012266845107254372,
+      "loss": 0.0536,
+      "step": 10250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012265477449158735,
+      "loss": 0.0721,
+      "step": 10251
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012264109746393604,
+      "loss": 0.0574,
+      "step": 10252
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012262741998985945,
+      "loss": 0.1191,
+      "step": 10253
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001226137420696273,
+      "loss": 0.0573,
+      "step": 10254
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012260006370350925,
+      "loss": 0.0815,
+      "step": 10255
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00122586384891775,
+      "loss": 0.0869,
+      "step": 10256
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012257270563469423,
+      "loss": 0.1423,
+      "step": 10257
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012255902593253673,
+      "loss": 0.0833,
+      "step": 10258
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001225453457855722,
+      "loss": 0.1292,
+      "step": 10259
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001225316651940704,
+      "loss": 0.0877,
+      "step": 10260
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012251798415830102,
+      "loss": 0.1268,
+      "step": 10261
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0012250430267853388,
+      "loss": 0.1191,
+      "step": 10262
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.001224906207550387,
+      "loss": 0.1011,
+      "step": 10263
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012247693838808528,
+      "loss": 0.1255,
+      "step": 10264
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012246325557794336,
+      "loss": 0.0848,
+      "step": 10265
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012244957232488277,
+      "loss": 0.0895,
+      "step": 10266
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.001224358886291733,
+      "loss": 0.0923,
+      "step": 10267
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012242220449108477,
+      "loss": 0.1077,
+      "step": 10268
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012240851991088696,
+      "loss": 0.1175,
+      "step": 10269
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012239483488884972,
+      "loss": 0.0757,
+      "step": 10270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012238114942524291,
+      "loss": 0.1155,
+      "step": 10271
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012236746352033631,
+      "loss": 0.0742,
+      "step": 10272
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012235377717439982,
+      "loss": 0.1848,
+      "step": 10273
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.001223400903877033,
+      "loss": 0.0661,
+      "step": 10274
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012232640316051661,
+      "loss": 0.0672,
+      "step": 10275
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.001223127154931096,
+      "loss": 0.0657,
+      "step": 10276
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012229902738575221,
+      "loss": 0.1093,
+      "step": 10277
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012228533883871426,
+      "loss": 0.1056,
+      "step": 10278
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012227164985226572,
+      "loss": 0.1207,
+      "step": 10279
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012225796042667648,
+      "loss": 0.0615,
+      "step": 10280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012224427056221647,
+      "loss": 0.1113,
+      "step": 10281
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.001222305802591556,
+      "loss": 0.1442,
+      "step": 10282
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012221688951776382,
+      "loss": 0.0878,
+      "step": 10283
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012220319833831108,
+      "loss": 0.101,
+      "step": 10284
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012218950672106734,
+      "loss": 0.1193,
+      "step": 10285
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012217581466630253,
+      "loss": 0.0887,
+      "step": 10286
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.001221621221742867,
+      "loss": 0.0581,
+      "step": 10287
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012214842924528973,
+      "loss": 0.0559,
+      "step": 10288
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012213473587958164,
+      "loss": 0.0874,
+      "step": 10289
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012212104207743248,
+      "loss": 0.0715,
+      "step": 10290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012210734783911225,
+      "loss": 0.0861,
+      "step": 10291
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012209365316489093,
+      "loss": 0.1096,
+      "step": 10292
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012207995805503854,
+      "loss": 0.1134,
+      "step": 10293
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012206626250982511,
+      "loss": 0.0988,
+      "step": 10294
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012205256652952072,
+      "loss": 0.109,
+      "step": 10295
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.001220388701143954,
+      "loss": 0.0862,
+      "step": 10296
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012202517326471922,
+      "loss": 0.0818,
+      "step": 10297
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012201147598076224,
+      "loss": 0.1083,
+      "step": 10298
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012199777826279448,
+      "loss": 0.1005,
+      "step": 10299
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012198408011108613,
+      "loss": 0.1052,
+      "step": 10300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012197038152590722,
+      "loss": 0.0854,
+      "step": 10301
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012195668250752788,
+      "loss": 0.1534,
+      "step": 10302
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012194298305621817,
+      "loss": 0.1083,
+      "step": 10303
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012192928317224822,
+      "loss": 0.1174,
+      "step": 10304
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012191558285588823,
+      "loss": 0.0925,
+      "step": 10305
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012190188210740825,
+      "loss": 0.1,
+      "step": 10306
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012188818092707845,
+      "loss": 0.0953,
+      "step": 10307
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012187447931516903,
+      "loss": 0.1034,
+      "step": 10308
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012186077727195006,
+      "loss": 0.0626,
+      "step": 10309
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0012184707479769177,
+      "loss": 0.0948,
+      "step": 10310
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012183337189266435,
+      "loss": 0.0881,
+      "step": 10311
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012181966855713798,
+      "loss": 0.0969,
+      "step": 10312
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001218059647913828,
+      "loss": 0.0895,
+      "step": 10313
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012179226059566908,
+      "loss": 0.1262,
+      "step": 10314
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012177855597026697,
+      "loss": 0.1204,
+      "step": 10315
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012176485091544675,
+      "loss": 0.0593,
+      "step": 10316
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012175114543147862,
+      "loss": 0.1096,
+      "step": 10317
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012173743951863286,
+      "loss": 0.1479,
+      "step": 10318
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012172373317717964,
+      "loss": 0.1509,
+      "step": 10319
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012171002640738923,
+      "loss": 0.0664,
+      "step": 10320
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012169631920953196,
+      "loss": 0.1426,
+      "step": 10321
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012168261158387807,
+      "loss": 0.0809,
+      "step": 10322
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001216689035306978,
+      "loss": 0.0764,
+      "step": 10323
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012165519505026147,
+      "loss": 0.0652,
+      "step": 10324
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012164148614283935,
+      "loss": 0.1138,
+      "step": 10325
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001216277768087018,
+      "loss": 0.0867,
+      "step": 10326
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001216140670481191,
+      "loss": 0.0674,
+      "step": 10327
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012160035686136156,
+      "loss": 0.0828,
+      "step": 10328
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012158664624869952,
+      "loss": 0.1145,
+      "step": 10329
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012157293521040337,
+      "loss": 0.0601,
+      "step": 10330
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012155922374674335,
+      "loss": 0.0674,
+      "step": 10331
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001215455118579899,
+      "loss": 0.0891,
+      "step": 10332
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012153179954441334,
+      "loss": 0.1183,
+      "step": 10333
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001215180868062841,
+      "loss": 0.1406,
+      "step": 10334
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012150437364387245,
+      "loss": 0.1177,
+      "step": 10335
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012149066005744888,
+      "loss": 0.1306,
+      "step": 10336
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001214769460472838,
+      "loss": 0.1135,
+      "step": 10337
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012146323161364755,
+      "loss": 0.0649,
+      "step": 10338
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012144951675681053,
+      "loss": 0.0936,
+      "step": 10339
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012143580147704325,
+      "loss": 0.1306,
+      "step": 10340
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012142208577461604,
+      "loss": 0.0549,
+      "step": 10341
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012140836964979944,
+      "loss": 0.1066,
+      "step": 10342
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001213946531028638,
+      "loss": 0.1146,
+      "step": 10343
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012138093613407966,
+      "loss": 0.0941,
+      "step": 10344
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012136721874371743,
+      "loss": 0.1025,
+      "step": 10345
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001213535009320476,
+      "loss": 0.1194,
+      "step": 10346
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012133978269934062,
+      "loss": 0.0718,
+      "step": 10347
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012132606404586705,
+      "loss": 0.0762,
+      "step": 10348
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012131234497189732,
+      "loss": 0.0925,
+      "step": 10349
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00121298625477702,
+      "loss": 0.0905,
+      "step": 10350
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001212849055635515,
+      "loss": 0.1054,
+      "step": 10351
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012127118522971645,
+      "loss": 0.0845,
+      "step": 10352
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.001212574644764673,
+      "loss": 0.0909,
+      "step": 10353
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012124374330407465,
+      "loss": 0.1089,
+      "step": 10354
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0012123002171280906,
+      "loss": 0.1107,
+      "step": 10355
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00121216299702941,
+      "loss": 0.0844,
+      "step": 10356
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012120257727474106,
+      "loss": 0.0599,
+      "step": 10357
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012118885442847988,
+      "loss": 0.118,
+      "step": 10358
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012117513116442795,
+      "loss": 0.1056,
+      "step": 10359
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012116140748285595,
+      "loss": 0.103,
+      "step": 10360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012114768338403442,
+      "loss": 0.0525,
+      "step": 10361
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012113395886823393,
+      "loss": 0.1081,
+      "step": 10362
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012112023393572513,
+      "loss": 0.0943,
+      "step": 10363
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001211065085867787,
+      "loss": 0.1526,
+      "step": 10364
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012109278282166518,
+      "loss": 0.0556,
+      "step": 10365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012107905664065523,
+      "loss": 0.1012,
+      "step": 10366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012106533004401954,
+      "loss": 0.1147,
+      "step": 10367
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001210516030320287,
+      "loss": 0.092,
+      "step": 10368
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001210378756049534,
+      "loss": 0.0834,
+      "step": 10369
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012102414776306435,
+      "loss": 0.1784,
+      "step": 10370
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012101041950663219,
+      "loss": 0.0746,
+      "step": 10371
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012099669083592756,
+      "loss": 0.0948,
+      "step": 10372
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012098296175122123,
+      "loss": 0.0934,
+      "step": 10373
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001209692322527839,
+      "loss": 0.1299,
+      "step": 10374
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012095550234088623,
+      "loss": 0.0718,
+      "step": 10375
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012094177201579896,
+      "loss": 0.1093,
+      "step": 10376
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012092804127779285,
+      "loss": 0.1232,
+      "step": 10377
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012091431012713858,
+      "loss": 0.1409,
+      "step": 10378
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012090057856410693,
+      "loss": 0.1017,
+      "step": 10379
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012088684658896867,
+      "loss": 0.0847,
+      "step": 10380
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001208731142019945,
+      "loss": 0.0991,
+      "step": 10381
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012085938140345522,
+      "loss": 0.1105,
+      "step": 10382
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012084564819362165,
+      "loss": 0.1197,
+      "step": 10383
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001208319145727645,
+      "loss": 0.0903,
+      "step": 10384
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.001208181805411546,
+      "loss": 0.0855,
+      "step": 10385
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012080444609906273,
+      "loss": 0.0994,
+      "step": 10386
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012079071124675976,
+      "loss": 0.058,
+      "step": 10387
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012077697598451642,
+      "loss": 0.1027,
+      "step": 10388
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012076324031260357,
+      "loss": 0.1041,
+      "step": 10389
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012074950423129206,
+      "loss": 0.085,
+      "step": 10390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012073576774085273,
+      "loss": 0.0714,
+      "step": 10391
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012072203084155639,
+      "loss": 0.1071,
+      "step": 10392
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012070829353367396,
+      "loss": 0.1114,
+      "step": 10393
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012069455581747623,
+      "loss": 0.0767,
+      "step": 10394
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012068081769323416,
+      "loss": 0.0814,
+      "step": 10395
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012066707916121855,
+      "loss": 0.1458,
+      "step": 10396
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012065334022170033,
+      "loss": 0.1425,
+      "step": 10397
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012063960087495041,
+      "loss": 0.0526,
+      "step": 10398
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012062586112123964,
+      "loss": 0.1274,
+      "step": 10399
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012061212096083898,
+      "loss": 0.1483,
+      "step": 10400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012059838039401936,
+      "loss": 0.0674,
+      "step": 10401
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0012058463942105164,
+      "loss": 0.0635,
+      "step": 10402
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012057089804220686,
+      "loss": 0.0996,
+      "step": 10403
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001205571562577559,
+      "loss": 0.1061,
+      "step": 10404
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001205434140679697,
+      "loss": 0.0688,
+      "step": 10405
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012052967147311926,
+      "loss": 0.0794,
+      "step": 10406
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012051592847347554,
+      "loss": 0.1058,
+      "step": 10407
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012050218506930955,
+      "loss": 0.0851,
+      "step": 10408
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012048844126089219,
+      "loss": 0.0627,
+      "step": 10409
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012047469704849448,
+      "loss": 0.1075,
+      "step": 10410
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012046095243238748,
+      "loss": 0.1301,
+      "step": 10411
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012044720741284213,
+      "loss": 0.0516,
+      "step": 10412
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012043346199012953,
+      "loss": 0.1345,
+      "step": 10413
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012041971616452063,
+      "loss": 0.0726,
+      "step": 10414
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012040596993628647,
+      "loss": 0.0958,
+      "step": 10415
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012039222330569811,
+      "loss": 0.0869,
+      "step": 10416
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012037847627302664,
+      "loss": 0.1234,
+      "step": 10417
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012036472883854303,
+      "loss": 0.09,
+      "step": 10418
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012035098100251838,
+      "loss": 0.1141,
+      "step": 10419
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012033723276522379,
+      "loss": 0.1245,
+      "step": 10420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001203234841269303,
+      "loss": 0.1205,
+      "step": 10421
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012030973508790903,
+      "loss": 0.1339,
+      "step": 10422
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001202959856484311,
+      "loss": 0.0964,
+      "step": 10423
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012028223580876755,
+      "loss": 0.126,
+      "step": 10424
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012026848556918955,
+      "loss": 0.0909,
+      "step": 10425
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012025473492996814,
+      "loss": 0.0715,
+      "step": 10426
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012024098389137453,
+      "loss": 0.0822,
+      "step": 10427
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012022723245367981,
+      "loss": 0.0916,
+      "step": 10428
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012021348061715515,
+      "loss": 0.095,
+      "step": 10429
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012019972838207172,
+      "loss": 0.0515,
+      "step": 10430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001201859757487006,
+      "loss": 0.1028,
+      "step": 10431
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00120172222717313,
+      "loss": 0.0847,
+      "step": 10432
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012015846928818015,
+      "loss": 0.1547,
+      "step": 10433
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012014471546157317,
+      "loss": 0.0881,
+      "step": 10434
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012013096123776328,
+      "loss": 0.1851,
+      "step": 10435
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012011720661702164,
+      "loss": 0.0695,
+      "step": 10436
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001201034515996195,
+      "loss": 0.0759,
+      "step": 10437
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012008969618582804,
+      "loss": 0.1105,
+      "step": 10438
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001200759403759185,
+      "loss": 0.0721,
+      "step": 10439
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001200621841701621,
+      "loss": 0.0955,
+      "step": 10440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012004842756883009,
+      "loss": 0.1012,
+      "step": 10441
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001200346705721937,
+      "loss": 0.1192,
+      "step": 10442
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012002091318052421,
+      "loss": 0.0955,
+      "step": 10443
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0012000715539409288,
+      "loss": 0.0793,
+      "step": 10444
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0011999339721317096,
+      "loss": 0.0808,
+      "step": 10445
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001199796386380297,
+      "loss": 0.0767,
+      "step": 10446
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.001199658796689404,
+      "loss": 0.1628,
+      "step": 10447
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0011995212030617441,
+      "loss": 0.0961,
+      "step": 10448
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011993836055000295,
+      "loss": 0.1053,
+      "step": 10449
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001199246004006974,
+      "loss": 0.1265,
+      "step": 10450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011991083985852904,
+      "loss": 0.0767,
+      "step": 10451
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011989707892376916,
+      "loss": 0.0849,
+      "step": 10452
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011988331759668913,
+      "loss": 0.0894,
+      "step": 10453
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011986955587756029,
+      "loss": 0.1082,
+      "step": 10454
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011985579376665402,
+      "loss": 0.0893,
+      "step": 10455
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011984203126424156,
+      "loss": 0.1145,
+      "step": 10456
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011982826837059438,
+      "loss": 0.1093,
+      "step": 10457
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001198145050859838,
+      "loss": 0.089,
+      "step": 10458
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001198007414106812,
+      "loss": 0.0527,
+      "step": 10459
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011978697734495803,
+      "loss": 0.0871,
+      "step": 10460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011977321288908559,
+      "loss": 0.0796,
+      "step": 10461
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011975944804333532,
+      "loss": 0.0975,
+      "step": 10462
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011974568280797863,
+      "loss": 0.069,
+      "step": 10463
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011973191718328692,
+      "loss": 0.0562,
+      "step": 10464
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011971815116953164,
+      "loss": 0.1136,
+      "step": 10465
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001197043847669842,
+      "loss": 0.1565,
+      "step": 10466
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011969061797591606,
+      "loss": 0.1257,
+      "step": 10467
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011967685079659863,
+      "loss": 0.0776,
+      "step": 10468
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001196630832293034,
+      "loss": 0.1499,
+      "step": 10469
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001196493152743018,
+      "loss": 0.1621,
+      "step": 10470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011963554693186537,
+      "loss": 0.1086,
+      "step": 10471
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011962177820226548,
+      "loss": 0.0709,
+      "step": 10472
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011960800908577367,
+      "loss": 0.088,
+      "step": 10473
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011959423958266145,
+      "loss": 0.1222,
+      "step": 10474
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001195804696932003,
+      "loss": 0.1353,
+      "step": 10475
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001195666994176617,
+      "loss": 0.1111,
+      "step": 10476
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011955292875631724,
+      "loss": 0.1008,
+      "step": 10477
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011953915770943836,
+      "loss": 0.1255,
+      "step": 10478
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011952538627729664,
+      "loss": 0.1412,
+      "step": 10479
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011951161446016358,
+      "loss": 0.1035,
+      "step": 10480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001194978422583108,
+      "loss": 0.1101,
+      "step": 10481
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011948406967200976,
+      "loss": 0.0975,
+      "step": 10482
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011947029670153206,
+      "loss": 0.1606,
+      "step": 10483
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001194565233471493,
+      "loss": 0.136,
+      "step": 10484
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011944274960913302,
+      "loss": 0.0775,
+      "step": 10485
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011942897548775477,
+      "loss": 0.1255,
+      "step": 10486
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011941520098328624,
+      "loss": 0.1459,
+      "step": 10487
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011940142609599895,
+      "loss": 0.1144,
+      "step": 10488
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011938765082616452,
+      "loss": 0.1243,
+      "step": 10489
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011937387517405456,
+      "loss": 0.1035,
+      "step": 10490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001193600991399407,
+      "loss": 0.0806,
+      "step": 10491
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011934632272409455,
+      "loss": 0.1105,
+      "step": 10492
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0011933254592678781,
+      "loss": 0.1022,
+      "step": 10493
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.001193187687482921,
+      "loss": 0.1205,
+      "step": 10494
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011930499118887898,
+      "loss": 0.1013,
+      "step": 10495
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001192912132488202,
+      "loss": 0.0648,
+      "step": 10496
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011927743492838744,
+      "loss": 0.1244,
+      "step": 10497
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011926365622785232,
+      "loss": 0.0942,
+      "step": 10498
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001192498771474865,
+      "loss": 0.1324,
+      "step": 10499
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011923609768756177,
+      "loss": 0.1057,
+      "step": 10500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001192223178483497,
+      "loss": 0.1462,
+      "step": 10501
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001192085376301221,
+      "loss": 0.0829,
+      "step": 10502
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001191947570331506,
+      "loss": 0.0752,
+      "step": 10503
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011918097605770701,
+      "loss": 0.0643,
+      "step": 10504
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011916719470406298,
+      "loss": 0.0621,
+      "step": 10505
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011915341297249024,
+      "loss": 0.0715,
+      "step": 10506
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001191396308632606,
+      "loss": 0.1058,
+      "step": 10507
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011912584837664578,
+      "loss": 0.1172,
+      "step": 10508
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011911206551291746,
+      "loss": 0.0864,
+      "step": 10509
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011909828227234754,
+      "loss": 0.0515,
+      "step": 10510
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011908449865520765,
+      "loss": 0.0823,
+      "step": 10511
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011907071466176963,
+      "loss": 0.0546,
+      "step": 10512
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011905693029230532,
+      "loss": 0.083,
+      "step": 10513
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011904314554708646,
+      "loss": 0.1026,
+      "step": 10514
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011902936042638486,
+      "loss": 0.0784,
+      "step": 10515
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001190155749304723,
+      "loss": 0.0825,
+      "step": 10516
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011900178905962058,
+      "loss": 0.0792,
+      "step": 10517
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011898800281410161,
+      "loss": 0.0753,
+      "step": 10518
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011897421619418713,
+      "loss": 0.0941,
+      "step": 10519
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011896042920014906,
+      "loss": 0.0806,
+      "step": 10520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011894664183225918,
+      "loss": 0.0616,
+      "step": 10521
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011893285409078933,
+      "loss": 0.1249,
+      "step": 10522
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011891906597601145,
+      "loss": 0.0722,
+      "step": 10523
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011890527748819736,
+      "loss": 0.1163,
+      "step": 10524
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001188914886276189,
+      "loss": 0.0733,
+      "step": 10525
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00118877699394548,
+      "loss": 0.1256,
+      "step": 10526
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011886390978925654,
+      "loss": 0.152,
+      "step": 10527
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011885011981201642,
+      "loss": 0.1371,
+      "step": 10528
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001188363294630995,
+      "loss": 0.0813,
+      "step": 10529
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011882253874277776,
+      "loss": 0.1263,
+      "step": 10530
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011880874765132306,
+      "loss": 0.1307,
+      "step": 10531
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011879495618900735,
+      "loss": 0.1529,
+      "step": 10532
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011878116435610256,
+      "loss": 0.049,
+      "step": 10533
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011876737215288063,
+      "loss": 0.105,
+      "step": 10534
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011875357957961353,
+      "loss": 0.1488,
+      "step": 10535
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011873978663657317,
+      "loss": 0.0822,
+      "step": 10536
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011872599332403153,
+      "loss": 0.0691,
+      "step": 10537
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011871219964226059,
+      "loss": 0.0849,
+      "step": 10538
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.001186984055915323,
+      "loss": 0.0933,
+      "step": 10539
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0011868461117211873,
+      "loss": 0.1377,
+      "step": 10540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011867081638429176,
+      "loss": 0.0967,
+      "step": 10541
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011865702122832342,
+      "loss": 0.1386,
+      "step": 10542
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011864322570448575,
+      "loss": 0.1141,
+      "step": 10543
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011862942981305075,
+      "loss": 0.1565,
+      "step": 10544
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011861563355429045,
+      "loss": 0.1185,
+      "step": 10545
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011860183692847682,
+      "loss": 0.1201,
+      "step": 10546
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011858803993588197,
+      "loss": 0.1641,
+      "step": 10547
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011857424257677787,
+      "loss": 0.0665,
+      "step": 10548
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011856044485143663,
+      "loss": 0.1332,
+      "step": 10549
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011854664676013028,
+      "loss": 0.1,
+      "step": 10550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001185328483031309,
+      "loss": 0.1251,
+      "step": 10551
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001185190494807105,
+      "loss": 0.0707,
+      "step": 10552
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011850525029314123,
+      "loss": 0.15,
+      "step": 10553
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011849145074069515,
+      "loss": 0.1046,
+      "step": 10554
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011847765082364436,
+      "loss": 0.1198,
+      "step": 10555
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011846385054226094,
+      "loss": 0.0854,
+      "step": 10556
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011845004989681702,
+      "loss": 0.1543,
+      "step": 10557
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011843624888758471,
+      "loss": 0.0698,
+      "step": 10558
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011842244751483608,
+      "loss": 0.0883,
+      "step": 10559
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011840864577884335,
+      "loss": 0.0999,
+      "step": 10560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001183948436798786,
+      "loss": 0.0999,
+      "step": 10561
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011838104121821393,
+      "loss": 0.1021,
+      "step": 10562
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001183672383941216,
+      "loss": 0.093,
+      "step": 10563
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011835343520787368,
+      "loss": 0.0838,
+      "step": 10564
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011833963165974236,
+      "loss": 0.0811,
+      "step": 10565
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001183258277499998,
+      "loss": 0.1003,
+      "step": 10566
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001183120234789182,
+      "loss": 0.1222,
+      "step": 10567
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011829821884676974,
+      "loss": 0.0687,
+      "step": 10568
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001182844138538266,
+      "loss": 0.1047,
+      "step": 10569
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011827060850036098,
+      "loss": 0.1064,
+      "step": 10570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001182568027866451,
+      "loss": 0.0622,
+      "step": 10571
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011824299671295115,
+      "loss": 0.0747,
+      "step": 10572
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011822919027955142,
+      "loss": 0.0796,
+      "step": 10573
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011821538348671806,
+      "loss": 0.0815,
+      "step": 10574
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001182015763347233,
+      "loss": 0.0857,
+      "step": 10575
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011818776882383942,
+      "loss": 0.0656,
+      "step": 10576
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001181739609543387,
+      "loss": 0.0843,
+      "step": 10577
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011816015272649333,
+      "loss": 0.0934,
+      "step": 10578
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011814634414057561,
+      "loss": 0.157,
+      "step": 10579
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011813253519685778,
+      "loss": 0.0646,
+      "step": 10580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.001181187258956122,
+      "loss": 0.0875,
+      "step": 10581
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011810491623711103,
+      "loss": 0.1244,
+      "step": 10582
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011809110622162666,
+      "loss": 0.1007,
+      "step": 10583
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011807729584943135,
+      "loss": 0.087,
+      "step": 10584
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011806348512079742,
+      "loss": 0.0662,
+      "step": 10585
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0011804967403599713,
+      "loss": 0.1929,
+      "step": 10586
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011803586259530288,
+      "loss": 0.1067,
+      "step": 10587
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011802205079898698,
+      "loss": 0.1013,
+      "step": 10588
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001180082386473217,
+      "loss": 0.1088,
+      "step": 10589
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011799442614057944,
+      "loss": 0.1338,
+      "step": 10590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011798061327903254,
+      "loss": 0.1013,
+      "step": 10591
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011796680006295334,
+      "loss": 0.0782,
+      "step": 10592
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001179529864926142,
+      "loss": 0.1097,
+      "step": 10593
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011793917256828752,
+      "loss": 0.1587,
+      "step": 10594
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011792535829024565,
+      "loss": 0.0812,
+      "step": 10595
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011791154365876093,
+      "loss": 0.0796,
+      "step": 10596
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011789772867410585,
+      "loss": 0.1267,
+      "step": 10597
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011788391333655273,
+      "loss": 0.1416,
+      "step": 10598
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011787009764637397,
+      "loss": 0.055,
+      "step": 10599
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011785628160384206,
+      "loss": 0.0723,
+      "step": 10600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011784246520922933,
+      "loss": 0.1256,
+      "step": 10601
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011782864846280822,
+      "loss": 0.1696,
+      "step": 10602
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011781483136485123,
+      "loss": 0.0732,
+      "step": 10603
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001178010139156307,
+      "loss": 0.0718,
+      "step": 10604
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011778719611541916,
+      "loss": 0.0822,
+      "step": 10605
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011777337796448901,
+      "loss": 0.1344,
+      "step": 10606
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011775955946311271,
+      "loss": 0.0694,
+      "step": 10607
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011774574061156277,
+      "loss": 0.0905,
+      "step": 10608
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001177319214101116,
+      "loss": 0.1205,
+      "step": 10609
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011771810185903172,
+      "loss": 0.1366,
+      "step": 10610
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011770428195859565,
+      "loss": 0.0608,
+      "step": 10611
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001176904617090758,
+      "loss": 0.1113,
+      "step": 10612
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001176766411107447,
+      "loss": 0.0969,
+      "step": 10613
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001176628201638749,
+      "loss": 0.1061,
+      "step": 10614
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001176489988687389,
+      "loss": 0.0803,
+      "step": 10615
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011763517722560917,
+      "loss": 0.1018,
+      "step": 10616
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011762135523475828,
+      "loss": 0.0975,
+      "step": 10617
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011760753289645877,
+      "loss": 0.0576,
+      "step": 10618
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011759371021098316,
+      "loss": 0.0927,
+      "step": 10619
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00117579887178604,
+      "loss": 0.0604,
+      "step": 10620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011756606379959389,
+      "loss": 0.1217,
+      "step": 10621
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011755224007422532,
+      "loss": 0.1101,
+      "step": 10622
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.001175384160027709,
+      "loss": 0.0694,
+      "step": 10623
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011752459158550321,
+      "loss": 0.0893,
+      "step": 10624
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011751076682269483,
+      "loss": 0.0753,
+      "step": 10625
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011749694171461834,
+      "loss": 0.0869,
+      "step": 10626
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011748311626154633,
+      "loss": 0.1038,
+      "step": 10627
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011746929046375142,
+      "loss": 0.136,
+      "step": 10628
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011745546432150617,
+      "loss": 0.0738,
+      "step": 10629
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011744163783508328,
+      "loss": 0.1373,
+      "step": 10630
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011742781100475535,
+      "loss": 0.0815,
+      "step": 10631
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0011741398383079496,
+      "loss": 0.0729,
+      "step": 10632
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011740015631347479,
+      "loss": 0.1121,
+      "step": 10633
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011738632845306744,
+      "loss": 0.1573,
+      "step": 10634
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011737250024984565,
+      "loss": 0.0867,
+      "step": 10635
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011735867170408196,
+      "loss": 0.1042,
+      "step": 10636
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011734484281604915,
+      "loss": 0.1016,
+      "step": 10637
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001173310135860198,
+      "loss": 0.0977,
+      "step": 10638
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011731718401426661,
+      "loss": 0.1107,
+      "step": 10639
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011730335410106232,
+      "loss": 0.0909,
+      "step": 10640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011728952384667958,
+      "loss": 0.1196,
+      "step": 10641
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011727569325139103,
+      "loss": 0.0896,
+      "step": 10642
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001172618623154695,
+      "loss": 0.088,
+      "step": 10643
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011724803103918754,
+      "loss": 0.0866,
+      "step": 10644
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011723419942281802,
+      "loss": 0.1178,
+      "step": 10645
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001172203674666336,
+      "loss": 0.0713,
+      "step": 10646
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011720653517090702,
+      "loss": 0.0972,
+      "step": 10647
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011719270253591101,
+      "loss": 0.0775,
+      "step": 10648
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001171788695619183,
+      "loss": 0.1305,
+      "step": 10649
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011716503624920164,
+      "loss": 0.0511,
+      "step": 10650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011715120259803387,
+      "loss": 0.0791,
+      "step": 10651
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011713736860868764,
+      "loss": 0.0695,
+      "step": 10652
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011712353428143581,
+      "loss": 0.1257,
+      "step": 10653
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011710969961655107,
+      "loss": 0.1172,
+      "step": 10654
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011709586461430628,
+      "loss": 0.0757,
+      "step": 10655
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011708202927497421,
+      "loss": 0.0659,
+      "step": 10656
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011706819359882767,
+      "loss": 0.0778,
+      "step": 10657
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011705435758613944,
+      "loss": 0.084,
+      "step": 10658
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011704052123718233,
+      "loss": 0.1001,
+      "step": 10659
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011702668455222916,
+      "loss": 0.0811,
+      "step": 10660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001170128475315528,
+      "loss": 0.0881,
+      "step": 10661
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011699901017542598,
+      "loss": 0.1126,
+      "step": 10662
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011698517248412165,
+      "loss": 0.0968,
+      "step": 10663
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001169713344579126,
+      "loss": 0.0721,
+      "step": 10664
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011695749609707167,
+      "loss": 0.0782,
+      "step": 10665
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011694365740187174,
+      "loss": 0.145,
+      "step": 10666
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001169298183725857,
+      "loss": 0.0989,
+      "step": 10667
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011691597900948636,
+      "loss": 0.1051,
+      "step": 10668
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.001169021393128466,
+      "loss": 0.1143,
+      "step": 10669
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011688829928293938,
+      "loss": 0.0661,
+      "step": 10670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011687445892003751,
+      "loss": 0.098,
+      "step": 10671
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011686061822441395,
+      "loss": 0.095,
+      "step": 10672
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011684677719634154,
+      "loss": 0.0902,
+      "step": 10673
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011683293583609324,
+      "loss": 0.0898,
+      "step": 10674
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011681909414394192,
+      "loss": 0.0967,
+      "step": 10675
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011680525212016052,
+      "loss": 0.1006,
+      "step": 10676
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011679140976502205,
+      "loss": 0.1279,
+      "step": 10677
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011677756707879933,
+      "loss": 0.1589,
+      "step": 10678
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0011676372406176536,
+      "loss": 0.0658,
+      "step": 10679
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011674988071419309,
+      "loss": 0.0981,
+      "step": 10680
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011673603703635544,
+      "loss": 0.1753,
+      "step": 10681
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001167221930285254,
+      "loss": 0.0905,
+      "step": 10682
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011670834869097594,
+      "loss": 0.0847,
+      "step": 10683
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011669450402398008,
+      "loss": 0.115,
+      "step": 10684
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011668065902781068,
+      "loss": 0.0768,
+      "step": 10685
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011666681370274086,
+      "loss": 0.1279,
+      "step": 10686
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011665296804904352,
+      "loss": 0.1212,
+      "step": 10687
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011663912206699171,
+      "loss": 0.1219,
+      "step": 10688
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011662527575685843,
+      "loss": 0.1139,
+      "step": 10689
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001166114291189167,
+      "loss": 0.1071,
+      "step": 10690
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001165975821534395,
+      "loss": 0.075,
+      "step": 10691
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001165837348606999,
+      "loss": 0.0649,
+      "step": 10692
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011656988724097093,
+      "loss": 0.1223,
+      "step": 10693
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011655603929452566,
+      "loss": 0.1008,
+      "step": 10694
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011654219102163704,
+      "loss": 0.0847,
+      "step": 10695
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001165283424225782,
+      "loss": 0.0762,
+      "step": 10696
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011651449349762218,
+      "loss": 0.1031,
+      "step": 10697
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011650064424704204,
+      "loss": 0.1033,
+      "step": 10698
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011648679467111088,
+      "loss": 0.1591,
+      "step": 10699
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011647294477010175,
+      "loss": 0.1232,
+      "step": 10700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011645909454428773,
+      "loss": 0.0946,
+      "step": 10701
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011644524399394194,
+      "loss": 0.1033,
+      "step": 10702
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011643139311933743,
+      "loss": 0.1321,
+      "step": 10703
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011641754192074738,
+      "loss": 0.0699,
+      "step": 10704
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011640369039844483,
+      "loss": 0.1104,
+      "step": 10705
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011638983855270296,
+      "loss": 0.1752,
+      "step": 10706
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011637598638379483,
+      "loss": 0.1184,
+      "step": 10707
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011636213389199358,
+      "loss": 0.1434,
+      "step": 10708
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011634828107757239,
+      "loss": 0.1149,
+      "step": 10709
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011633442794080438,
+      "loss": 0.0801,
+      "step": 10710
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001163205744819627,
+      "loss": 0.0885,
+      "step": 10711
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011630672070132046,
+      "loss": 0.0869,
+      "step": 10712
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011629286659915088,
+      "loss": 0.0693,
+      "step": 10713
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001162790121757271,
+      "loss": 0.1191,
+      "step": 10714
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011626515743132233,
+      "loss": 0.0784,
+      "step": 10715
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011625130236620972,
+      "loss": 0.135,
+      "step": 10716
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011623744698066247,
+      "loss": 0.0714,
+      "step": 10717
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011622359127495374,
+      "loss": 0.0707,
+      "step": 10718
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011620973524935677,
+      "loss": 0.1057,
+      "step": 10719
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011619587890414475,
+      "loss": 0.092,
+      "step": 10720
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011618202223959088,
+      "loss": 0.0615,
+      "step": 10721
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011616816525596842,
+      "loss": 0.0886,
+      "step": 10722
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011615430795355053,
+      "loss": 0.1316,
+      "step": 10723
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.001161404503326105,
+      "loss": 0.1077,
+      "step": 10724
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0011612659239342152,
+      "loss": 0.0922,
+      "step": 10725
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011611273413625687,
+      "loss": 0.1061,
+      "step": 10726
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011609887556138978,
+      "loss": 0.0688,
+      "step": 10727
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011608501666909355,
+      "loss": 0.1591,
+      "step": 10728
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011607115745964135,
+      "loss": 0.0796,
+      "step": 10729
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011605729793330653,
+      "loss": 0.0644,
+      "step": 10730
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011604343809036234,
+      "loss": 0.1045,
+      "step": 10731
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011602957793108207,
+      "loss": 0.075,
+      "step": 10732
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011601571745573898,
+      "loss": 0.0621,
+      "step": 10733
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.001160018566646064,
+      "loss": 0.0912,
+      "step": 10734
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011598799555795757,
+      "loss": 0.0938,
+      "step": 10735
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011597413413606589,
+      "loss": 0.0814,
+      "step": 10736
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011596027239920458,
+      "loss": 0.0767,
+      "step": 10737
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011594641034764703,
+      "loss": 0.1016,
+      "step": 10738
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.001159325479816665,
+      "loss": 0.0896,
+      "step": 10739
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011591868530153638,
+      "loss": 0.0811,
+      "step": 10740
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011590482230752997,
+      "loss": 0.0729,
+      "step": 10741
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.001158909589999206,
+      "loss": 0.1032,
+      "step": 10742
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011587709537898168,
+      "loss": 0.104,
+      "step": 10743
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011586323144498652,
+      "loss": 0.0666,
+      "step": 10744
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011584936719820844,
+      "loss": 0.0778,
+      "step": 10745
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011583550263892094,
+      "loss": 0.1263,
+      "step": 10746
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011582163776739727,
+      "loss": 0.1572,
+      "step": 10747
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011580777258391086,
+      "loss": 0.097,
+      "step": 10748
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011579390708873507,
+      "loss": 0.0858,
+      "step": 10749
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011578004128214331,
+      "loss": 0.1012,
+      "step": 10750
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011576617516440902,
+      "loss": 0.1669,
+      "step": 10751
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011575230873580554,
+      "loss": 0.1257,
+      "step": 10752
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.001157384419966063,
+      "loss": 0.072,
+      "step": 10753
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011572457494708475,
+      "loss": 0.0563,
+      "step": 10754
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011571070758751424,
+      "loss": 0.1086,
+      "step": 10755
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011569683991816827,
+      "loss": 0.103,
+      "step": 10756
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011568297193932026,
+      "loss": 0.1258,
+      "step": 10757
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011566910365124366,
+      "loss": 0.0935,
+      "step": 10758
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011565523505421185,
+      "loss": 0.1147,
+      "step": 10759
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011564136614849838,
+      "loss": 0.1196,
+      "step": 10760
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011562749693437662,
+      "loss": 0.1483,
+      "step": 10761
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011561362741212009,
+      "loss": 0.0687,
+      "step": 10762
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011559975758200227,
+      "loss": 0.1221,
+      "step": 10763
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011558588744429662,
+      "loss": 0.1276,
+      "step": 10764
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011557201699927662,
+      "loss": 0.0995,
+      "step": 10765
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011555814624721575,
+      "loss": 0.0951,
+      "step": 10766
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011554427518838755,
+      "loss": 0.0805,
+      "step": 10767
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011553040382306548,
+      "loss": 0.1182,
+      "step": 10768
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011551653215152303,
+      "loss": 0.0699,
+      "step": 10769
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0011550266017403379,
+      "loss": 0.0794,
+      "step": 10770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.001154887878908712,
+      "loss": 0.1228,
+      "step": 10771
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011547491530230882,
+      "loss": 0.0802,
+      "step": 10772
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001154610424086202,
+      "loss": 0.0755,
+      "step": 10773
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011544716921007888,
+      "loss": 0.0984,
+      "step": 10774
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011543329570695836,
+      "loss": 0.1396,
+      "step": 10775
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011541942189953222,
+      "loss": 0.1124,
+      "step": 10776
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011540554778807402,
+      "loss": 0.08,
+      "step": 10777
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011539167337285732,
+      "loss": 0.1353,
+      "step": 10778
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011537779865415567,
+      "loss": 0.1088,
+      "step": 10779
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011536392363224268,
+      "loss": 0.1041,
+      "step": 10780
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001153500483073919,
+      "loss": 0.1007,
+      "step": 10781
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011533617267987693,
+      "loss": 0.1162,
+      "step": 10782
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011532229674997131,
+      "loss": 0.0786,
+      "step": 10783
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011530842051794877,
+      "loss": 0.0992,
+      "step": 10784
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011529454398408277,
+      "loss": 0.1075,
+      "step": 10785
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00115280667148647,
+      "loss": 0.1037,
+      "step": 10786
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011526679001191507,
+      "loss": 0.0873,
+      "step": 10787
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011525291257416057,
+      "loss": 0.12,
+      "step": 10788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011523903483565713,
+      "loss": 0.0981,
+      "step": 10789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011522515679667847,
+      "loss": 0.112,
+      "step": 10790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011521127845749812,
+      "loss": 0.0614,
+      "step": 10791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011519739981838974,
+      "loss": 0.1357,
+      "step": 10792
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011518352087962703,
+      "loss": 0.0786,
+      "step": 10793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011516964164148362,
+      "loss": 0.0778,
+      "step": 10794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001151557621042332,
+      "loss": 0.0901,
+      "step": 10795
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001151418822681494,
+      "loss": 0.1689,
+      "step": 10796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011512800213350593,
+      "loss": 0.0812,
+      "step": 10797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011511412170057645,
+      "loss": 0.0917,
+      "step": 10798
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011510024096963462,
+      "loss": 0.0892,
+      "step": 10799
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011508635994095422,
+      "loss": 0.0743,
+      "step": 10800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001150724786148089,
+      "loss": 0.0624,
+      "step": 10801
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011505859699147232,
+      "loss": 0.0923,
+      "step": 10802
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001150447150712182,
+      "loss": 0.1129,
+      "step": 10803
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011503083285432038,
+      "loss": 0.1345,
+      "step": 10804
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001150169503410524,
+      "loss": 0.1221,
+      "step": 10805
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011500306753168812,
+      "loss": 0.0952,
+      "step": 10806
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011498918442650124,
+      "loss": 0.0731,
+      "step": 10807
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011497530102576546,
+      "loss": 0.0785,
+      "step": 10808
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011496141732975456,
+      "loss": 0.0784,
+      "step": 10809
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001149475333387423,
+      "loss": 0.1324,
+      "step": 10810
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011493364905300243,
+      "loss": 0.1042,
+      "step": 10811
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001149197644728087,
+      "loss": 0.1051,
+      "step": 10812
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011490587959843489,
+      "loss": 0.0745,
+      "step": 10813
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011489199443015475,
+      "loss": 0.1239,
+      "step": 10814
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001148781089682421,
+      "loss": 0.0867,
+      "step": 10815
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.001148642232129707,
+      "loss": 0.1145,
+      "step": 10816
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0011485033716461437,
+      "loss": 0.0868,
+      "step": 10817
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011483645082344689,
+      "loss": 0.1056,
+      "step": 10818
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011482256418974203,
+      "loss": 0.1022,
+      "step": 10819
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011480867726377363,
+      "loss": 0.1077,
+      "step": 10820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011479479004581554,
+      "loss": 0.0796,
+      "step": 10821
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011478090253614153,
+      "loss": 0.1288,
+      "step": 10822
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011476701473502543,
+      "loss": 0.1056,
+      "step": 10823
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011475312664274112,
+      "loss": 0.0663,
+      "step": 10824
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011473923825956237,
+      "loss": 0.0711,
+      "step": 10825
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011472534958576306,
+      "loss": 0.1556,
+      "step": 10826
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011471146062161707,
+      "loss": 0.0877,
+      "step": 10827
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001146975713673982,
+      "loss": 0.1346,
+      "step": 10828
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011468368182338032,
+      "loss": 0.1143,
+      "step": 10829
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001146697919898373,
+      "loss": 0.1318,
+      "step": 10830
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011465590186704308,
+      "loss": 0.144,
+      "step": 10831
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011464201145527143,
+      "loss": 0.0862,
+      "step": 10832
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011462812075479633,
+      "loss": 0.1497,
+      "step": 10833
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001146142297658916,
+      "loss": 0.156,
+      "step": 10834
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011460033848883114,
+      "loss": 0.085,
+      "step": 10835
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011458644692388888,
+      "loss": 0.1053,
+      "step": 10836
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011457255507133872,
+      "loss": 0.0938,
+      "step": 10837
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011455866293145459,
+      "loss": 0.1056,
+      "step": 10838
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011454477050451036,
+      "loss": 0.0884,
+      "step": 10839
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011453087779077999,
+      "loss": 0.1115,
+      "step": 10840
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001145169847905374,
+      "loss": 0.0663,
+      "step": 10841
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001145030915040565,
+      "loss": 0.1401,
+      "step": 10842
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011448919793161129,
+      "loss": 0.1018,
+      "step": 10843
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001144753040734757,
+      "loss": 0.1174,
+      "step": 10844
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011446140992992361,
+      "loss": 0.1027,
+      "step": 10845
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011444751550122908,
+      "loss": 0.1504,
+      "step": 10846
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00114433620787666,
+      "loss": 0.0682,
+      "step": 10847
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011441972578950838,
+      "loss": 0.1,
+      "step": 10848
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011440583050703016,
+      "loss": 0.1041,
+      "step": 10849
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011439193494050533,
+      "loss": 0.0813,
+      "step": 10850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011437803909020788,
+      "loss": 0.0742,
+      "step": 10851
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001143641429564118,
+      "loss": 0.0994,
+      "step": 10852
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011435024653939111,
+      "loss": 0.1267,
+      "step": 10853
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001143363498394198,
+      "loss": 0.1332,
+      "step": 10854
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011432245285677184,
+      "loss": 0.0543,
+      "step": 10855
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011430855559172132,
+      "loss": 0.1007,
+      "step": 10856
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011429465804454217,
+      "loss": 0.1149,
+      "step": 10857
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001142807602155085,
+      "loss": 0.0568,
+      "step": 10858
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011426686210489424,
+      "loss": 0.062,
+      "step": 10859
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001142529637129735,
+      "loss": 0.1198,
+      "step": 10860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011423906504002032,
+      "loss": 0.1052,
+      "step": 10861
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.001142251660863087,
+      "loss": 0.0738,
+      "step": 10862
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0011421126685211276,
+      "loss": 0.0656,
+      "step": 10863
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011419736733770652,
+      "loss": 0.0876,
+      "step": 10864
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011418346754336402,
+      "loss": 0.1215,
+      "step": 10865
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011416956746935938,
+      "loss": 0.0905,
+      "step": 10866
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011415566711596662,
+      "loss": 0.0946,
+      "step": 10867
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011414176648345987,
+      "loss": 0.1064,
+      "step": 10868
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011412786557211316,
+      "loss": 0.1151,
+      "step": 10869
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011411396438220066,
+      "loss": 0.0561,
+      "step": 10870
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001141000629139964,
+      "loss": 0.1041,
+      "step": 10871
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011408616116777449,
+      "loss": 0.0861,
+      "step": 10872
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011407225914380904,
+      "loss": 0.1239,
+      "step": 10873
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001140583568423742,
+      "loss": 0.1094,
+      "step": 10874
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011404445426374404,
+      "loss": 0.0909,
+      "step": 10875
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001140305514081927,
+      "loss": 0.0783,
+      "step": 10876
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011401664827599433,
+      "loss": 0.1072,
+      "step": 10877
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011400274486742304,
+      "loss": 0.0987,
+      "step": 10878
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011398884118275297,
+      "loss": 0.0845,
+      "step": 10879
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011397493722225828,
+      "loss": 0.0924,
+      "step": 10880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011396103298621314,
+      "loss": 0.1334,
+      "step": 10881
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011394712847489165,
+      "loss": 0.1385,
+      "step": 10882
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00113933223688568,
+      "loss": 0.1001,
+      "step": 10883
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011391931862751638,
+      "loss": 0.163,
+      "step": 10884
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011390541329201092,
+      "loss": 0.1089,
+      "step": 10885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011389150768232584,
+      "loss": 0.1415,
+      "step": 10886
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011387760179873531,
+      "loss": 0.0925,
+      "step": 10887
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001138636956415135,
+      "loss": 0.1118,
+      "step": 10888
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001138497892109346,
+      "loss": 0.0683,
+      "step": 10889
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011383588250727286,
+      "loss": 0.1315,
+      "step": 10890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011382197553080245,
+      "loss": 0.1169,
+      "step": 10891
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011380806828179754,
+      "loss": 0.069,
+      "step": 10892
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011379416076053243,
+      "loss": 0.0925,
+      "step": 10893
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001137802529672813,
+      "loss": 0.0811,
+      "step": 10894
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011376634490231833,
+      "loss": 0.1488,
+      "step": 10895
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011375243656591785,
+      "loss": 0.1387,
+      "step": 10896
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011373852795835404,
+      "loss": 0.1229,
+      "step": 10897
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011372461907990113,
+      "loss": 0.1318,
+      "step": 10898
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001137107099308334,
+      "loss": 0.0895,
+      "step": 10899
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011369680051142507,
+      "loss": 0.0678,
+      "step": 10900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011368289082195045,
+      "loss": 0.1008,
+      "step": 10901
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011366898086268373,
+      "loss": 0.0691,
+      "step": 10902
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011365507063389926,
+      "loss": 0.09,
+      "step": 10903
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011364116013587125,
+      "loss": 0.0655,
+      "step": 10904
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00113627249368874,
+      "loss": 0.0767,
+      "step": 10905
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011361333833318184,
+      "loss": 0.099,
+      "step": 10906
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00113599427029069,
+      "loss": 0.077,
+      "step": 10907
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.001135855154568098,
+      "loss": 0.0898,
+      "step": 10908
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0011357160361667852,
+      "loss": 0.1189,
+      "step": 10909
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001135576915089495,
+      "loss": 0.085,
+      "step": 10910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011354377913389705,
+      "loss": 0.0795,
+      "step": 10911
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011352986649179541,
+      "loss": 0.0883,
+      "step": 10912
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011351595358291904,
+      "loss": 0.1104,
+      "step": 10913
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011350204040754218,
+      "loss": 0.1227,
+      "step": 10914
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011348812696593913,
+      "loss": 0.0853,
+      "step": 10915
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011347421325838432,
+      "loss": 0.0721,
+      "step": 10916
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011346029928515201,
+      "loss": 0.0897,
+      "step": 10917
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001134463850465166,
+      "loss": 0.1309,
+      "step": 10918
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011343247054275245,
+      "loss": 0.0673,
+      "step": 10919
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001134185557741339,
+      "loss": 0.0825,
+      "step": 10920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011340464074093529,
+      "loss": 0.1176,
+      "step": 10921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00113390725443431,
+      "loss": 0.0745,
+      "step": 10922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011337680988189542,
+      "loss": 0.1204,
+      "step": 10923
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011336289405660296,
+      "loss": 0.1055,
+      "step": 10924
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011334897796782793,
+      "loss": 0.0682,
+      "step": 10925
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011333506161584476,
+      "loss": 0.0851,
+      "step": 10926
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001133211450009279,
+      "loss": 0.1273,
+      "step": 10927
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011330722812335165,
+      "loss": 0.0451,
+      "step": 10928
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011329331098339046,
+      "loss": 0.0934,
+      "step": 10929
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011327939358131876,
+      "loss": 0.1302,
+      "step": 10930
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011326547591741093,
+      "loss": 0.0947,
+      "step": 10931
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001132515579919414,
+      "loss": 0.1558,
+      "step": 10932
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011323763980518465,
+      "loss": 0.0906,
+      "step": 10933
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011322372135741505,
+      "loss": 0.1201,
+      "step": 10934
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011320980264890703,
+      "loss": 0.1216,
+      "step": 10935
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011319588367993507,
+      "loss": 0.075,
+      "step": 10936
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011318196445077363,
+      "loss": 0.1274,
+      "step": 10937
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001131680449616971,
+      "loss": 0.0876,
+      "step": 10938
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011315412521298,
+      "loss": 0.073,
+      "step": 10939
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011314020520489675,
+      "loss": 0.0626,
+      "step": 10940
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011312628493772181,
+      "loss": 0.0901,
+      "step": 10941
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001131123644117297,
+      "loss": 0.1259,
+      "step": 10942
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011309844362719487,
+      "loss": 0.0996,
+      "step": 10943
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011308452258439181,
+      "loss": 0.09,
+      "step": 10944
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00113070601283595,
+      "loss": 0.061,
+      "step": 10945
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011305667972507894,
+      "loss": 0.0844,
+      "step": 10946
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001130427579091181,
+      "loss": 0.098,
+      "step": 10947
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011302883583598701,
+      "loss": 0.149,
+      "step": 10948
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001130149135059602,
+      "loss": 0.0883,
+      "step": 10949
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011300099091931216,
+      "loss": 0.0995,
+      "step": 10950
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001129870680763174,
+      "loss": 0.0759,
+      "step": 10951
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001129731449772504,
+      "loss": 0.0914,
+      "step": 10952
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011295922162238578,
+      "loss": 0.0727,
+      "step": 10953
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0011294529801199806,
+      "loss": 0.1356,
+      "step": 10954
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.001129313741463617,
+      "loss": 0.0659,
+      "step": 10955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011291745002575133,
+      "loss": 0.0884,
+      "step": 10956
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011290352565044145,
+      "loss": 0.0738,
+      "step": 10957
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001128896010207066,
+      "loss": 0.05,
+      "step": 10958
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001128756761368214,
+      "loss": 0.1227,
+      "step": 10959
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011286175099906035,
+      "loss": 0.1071,
+      "step": 10960
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011284782560769808,
+      "loss": 0.0779,
+      "step": 10961
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001128338999630091,
+      "loss": 0.113,
+      "step": 10962
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011281997406526801,
+      "loss": 0.0675,
+      "step": 10963
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011280604791474942,
+      "loss": 0.0774,
+      "step": 10964
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011279212151172789,
+      "loss": 0.0969,
+      "step": 10965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011277819485647806,
+      "loss": 0.0728,
+      "step": 10966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011276426794927445,
+      "loss": 0.1024,
+      "step": 10967
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001127503407903917,
+      "loss": 0.1877,
+      "step": 10968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011273641338010443,
+      "loss": 0.098,
+      "step": 10969
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001127224857186873,
+      "loss": 0.0975,
+      "step": 10970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011270855780641483,
+      "loss": 0.0806,
+      "step": 10971
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011269462964356167,
+      "loss": 0.1001,
+      "step": 10972
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011268070123040248,
+      "loss": 0.0788,
+      "step": 10973
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001126667725672119,
+      "loss": 0.069,
+      "step": 10974
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011265284365426455,
+      "loss": 0.0878,
+      "step": 10975
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011263891449183505,
+      "loss": 0.0778,
+      "step": 10976
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001126249850801981,
+      "loss": 0.0941,
+      "step": 10977
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011261105541962829,
+      "loss": 0.1146,
+      "step": 10978
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011259712551040032,
+      "loss": 0.0947,
+      "step": 10979
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011258319535278883,
+      "loss": 0.113,
+      "step": 10980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001125692649470685,
+      "loss": 0.1458,
+      "step": 10981
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011255533429351404,
+      "loss": 0.0983,
+      "step": 10982
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011254140339240007,
+      "loss": 0.0619,
+      "step": 10983
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011252747224400129,
+      "loss": 0.0836,
+      "step": 10984
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011251354084859235,
+      "loss": 0.1179,
+      "step": 10985
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011249960920644805,
+      "loss": 0.1418,
+      "step": 10986
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011248567731784297,
+      "loss": 0.0732,
+      "step": 10987
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011247174518305189,
+      "loss": 0.0939,
+      "step": 10988
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011245781280234944,
+      "loss": 0.0881,
+      "step": 10989
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011244388017601042,
+      "loss": 0.1002,
+      "step": 10990
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001124299473043095,
+      "loss": 0.0533,
+      "step": 10991
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001124160141875214,
+      "loss": 0.0629,
+      "step": 10992
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011240208082592084,
+      "loss": 0.0758,
+      "step": 10993
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011238814721978255,
+      "loss": 0.0749,
+      "step": 10994
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011237421336938128,
+      "loss": 0.1537,
+      "step": 10995
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001123602792749918,
+      "loss": 0.1141,
+      "step": 10996
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001123463449368888,
+      "loss": 0.1208,
+      "step": 10997
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011233241035534705,
+      "loss": 0.1449,
+      "step": 10998
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011231847553064133,
+      "loss": 0.0873,
+      "step": 10999
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011230454046304633,
+      "loss": 0.0781,
+      "step": 11000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.001122906051528369,
+      "loss": 0.1037,
+      "step": 11001
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0011227666960028776,
+      "loss": 0.1071,
+      "step": 11002
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011226273380567369,
+      "loss": 0.1028,
+      "step": 11003
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011224879776926947,
+      "loss": 0.0761,
+      "step": 11004
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001122348614913499,
+      "loss": 0.1172,
+      "step": 11005
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011222092497218971,
+      "loss": 0.0627,
+      "step": 11006
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011220698821206379,
+      "loss": 0.12,
+      "step": 11007
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011219305121124685,
+      "loss": 0.0888,
+      "step": 11008
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011217911397001378,
+      "loss": 0.1125,
+      "step": 11009
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011216517648863927,
+      "loss": 0.1055,
+      "step": 11010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011215123876739824,
+      "loss": 0.223,
+      "step": 11011
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011213730080656542,
+      "loss": 0.0964,
+      "step": 11012
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001121233626064157,
+      "loss": 0.0785,
+      "step": 11013
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001121094241672239,
+      "loss": 0.078,
+      "step": 11014
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001120954854892648,
+      "loss": 0.1242,
+      "step": 11015
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011208154657281328,
+      "loss": 0.1198,
+      "step": 11016
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011206760741814417,
+      "loss": 0.0638,
+      "step": 11017
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011205366802553229,
+      "loss": 0.0793,
+      "step": 11018
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011203972839525255,
+      "loss": 0.1062,
+      "step": 11019
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011202578852757977,
+      "loss": 0.0777,
+      "step": 11020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011201184842278876,
+      "loss": 0.0681,
+      "step": 11021
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011199790808115444,
+      "loss": 0.1021,
+      "step": 11022
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001119839675029517,
+      "loss": 0.098,
+      "step": 11023
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011197002668845535,
+      "loss": 0.0792,
+      "step": 11024
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001119560856379403,
+      "loss": 0.0601,
+      "step": 11025
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011194214435168144,
+      "loss": 0.1545,
+      "step": 11026
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011192820282995364,
+      "loss": 0.0919,
+      "step": 11027
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001119142610730318,
+      "loss": 0.1172,
+      "step": 11028
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011190031908119083,
+      "loss": 0.0928,
+      "step": 11029
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001118863768547056,
+      "loss": 0.1156,
+      "step": 11030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011187243439385102,
+      "loss": 0.0668,
+      "step": 11031
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00111858491698902,
+      "loss": 0.0726,
+      "step": 11032
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011184454877013353,
+      "loss": 0.0515,
+      "step": 11033
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011183060560782043,
+      "loss": 0.1069,
+      "step": 11034
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011181666221223763,
+      "loss": 0.0791,
+      "step": 11035
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011180271858366013,
+      "loss": 0.1436,
+      "step": 11036
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001117887747223628,
+      "loss": 0.0906,
+      "step": 11037
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.001117748306286206,
+      "loss": 0.1378,
+      "step": 11038
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011176088630270847,
+      "loss": 0.0579,
+      "step": 11039
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011174694174490136,
+      "loss": 0.0641,
+      "step": 11040
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011173299695547425,
+      "loss": 0.0599,
+      "step": 11041
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011171905193470201,
+      "loss": 0.1275,
+      "step": 11042
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011170510668285969,
+      "loss": 0.0969,
+      "step": 11043
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011169116120022223,
+      "loss": 0.0664,
+      "step": 11044
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011167721548706456,
+      "loss": 0.0608,
+      "step": 11045
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011166326954366171,
+      "loss": 0.103,
+      "step": 11046
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011164932337028861,
+      "loss": 0.0887,
+      "step": 11047
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0011163537696722028,
+      "loss": 0.0867,
+      "step": 11048
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001116214303347317,
+      "loss": 0.12,
+      "step": 11049
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011160748347309787,
+      "loss": 0.162,
+      "step": 11050
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011159353638259377,
+      "loss": 0.1035,
+      "step": 11051
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001115795890634944,
+      "loss": 0.1164,
+      "step": 11052
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011156564151607474,
+      "loss": 0.1493,
+      "step": 11053
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011155169374060987,
+      "loss": 0.1079,
+      "step": 11054
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011153774573737475,
+      "loss": 0.1232,
+      "step": 11055
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011152379750664443,
+      "loss": 0.0805,
+      "step": 11056
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011150984904869392,
+      "loss": 0.1128,
+      "step": 11057
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011149590036379825,
+      "loss": 0.1307,
+      "step": 11058
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001114819514522324,
+      "loss": 0.063,
+      "step": 11059
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011146800231427154,
+      "loss": 0.0918,
+      "step": 11060
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011145405295019059,
+      "loss": 0.0649,
+      "step": 11061
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011144010336026464,
+      "loss": 0.0957,
+      "step": 11062
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011142615354476876,
+      "loss": 0.0865,
+      "step": 11063
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011141220350397794,
+      "loss": 0.0928,
+      "step": 11064
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001113982532381673,
+      "loss": 0.1154,
+      "step": 11065
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011138430274761188,
+      "loss": 0.0797,
+      "step": 11066
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011137035203258679,
+      "loss": 0.0701,
+      "step": 11067
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011135640109336705,
+      "loss": 0.0636,
+      "step": 11068
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011134244993022773,
+      "loss": 0.0876,
+      "step": 11069
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011132849854344395,
+      "loss": 0.0795,
+      "step": 11070
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001113145469332908,
+      "loss": 0.0717,
+      "step": 11071
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011130059510004334,
+      "loss": 0.0828,
+      "step": 11072
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011128664304397668,
+      "loss": 0.0784,
+      "step": 11073
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011127269076536593,
+      "loss": 0.1093,
+      "step": 11074
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011125873826448613,
+      "loss": 0.1038,
+      "step": 11075
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001112447855416125,
+      "loss": 0.0689,
+      "step": 11076
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001112308325970201,
+      "loss": 0.0968,
+      "step": 11077
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011121687943098397,
+      "loss": 0.1146,
+      "step": 11078
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011120292604377938,
+      "loss": 0.064,
+      "step": 11079
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001111889724356813,
+      "loss": 0.0873,
+      "step": 11080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00111175018606965,
+      "loss": 0.1098,
+      "step": 11081
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001111610645579055,
+      "loss": 0.088,
+      "step": 11082
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011114711028877803,
+      "loss": 0.0838,
+      "step": 11083
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011113315579985765,
+      "loss": 0.0911,
+      "step": 11084
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011111920109141955,
+      "loss": 0.1609,
+      "step": 11085
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.001111052461637389,
+      "loss": 0.1679,
+      "step": 11086
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011109129101709086,
+      "loss": 0.1013,
+      "step": 11087
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011107733565175054,
+      "loss": 0.1608,
+      "step": 11088
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011106338006799312,
+      "loss": 0.0759,
+      "step": 11089
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011104942426609378,
+      "loss": 0.0773,
+      "step": 11090
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011103546824632773,
+      "loss": 0.1045,
+      "step": 11091
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011102151200897007,
+      "loss": 0.155,
+      "step": 11092
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011100755555429605,
+      "loss": 0.1099,
+      "step": 11093
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0011099359888258082,
+      "loss": 0.1037,
+      "step": 11094
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011097964199409955,
+      "loss": 0.1004,
+      "step": 11095
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011096568488912749,
+      "loss": 0.1212,
+      "step": 11096
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011095172756793983,
+      "loss": 0.0938,
+      "step": 11097
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011093777003081172,
+      "loss": 0.1285,
+      "step": 11098
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011092381227801843,
+      "loss": 0.1027,
+      "step": 11099
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011090985430983514,
+      "loss": 0.0953,
+      "step": 11100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011089589612653706,
+      "loss": 0.1487,
+      "step": 11101
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011088193772839938,
+      "loss": 0.0465,
+      "step": 11102
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011086797911569744,
+      "loss": 0.0972,
+      "step": 11103
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011085402028870638,
+      "loss": 0.0751,
+      "step": 11104
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001108400612477014,
+      "loss": 0.0931,
+      "step": 11105
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011082610199295779,
+      "loss": 0.1078,
+      "step": 11106
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011081214252475083,
+      "loss": 0.1132,
+      "step": 11107
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011079818284335569,
+      "loss": 0.0725,
+      "step": 11108
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011078422294904765,
+      "loss": 0.0968,
+      "step": 11109
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011077026284210196,
+      "loss": 0.1052,
+      "step": 11110
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011075630252279389,
+      "loss": 0.0852,
+      "step": 11111
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011074234199139868,
+      "loss": 0.0936,
+      "step": 11112
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011072838124819163,
+      "loss": 0.0794,
+      "step": 11113
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011071442029344798,
+      "loss": 0.0994,
+      "step": 11114
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011070045912744302,
+      "loss": 0.1178,
+      "step": 11115
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011068649775045204,
+      "loss": 0.071,
+      "step": 11116
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011067253616275025,
+      "loss": 0.0659,
+      "step": 11117
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011065857436461303,
+      "loss": 0.0776,
+      "step": 11118
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011064461235631567,
+      "loss": 0.0616,
+      "step": 11119
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001106306501381334,
+      "loss": 0.0942,
+      "step": 11120
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011061668771034157,
+      "loss": 0.1272,
+      "step": 11121
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001106027250732154,
+      "loss": 0.1097,
+      "step": 11122
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011058876222703033,
+      "loss": 0.1036,
+      "step": 11123
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011057479917206158,
+      "loss": 0.1133,
+      "step": 11124
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001105608359085845,
+      "loss": 0.1093,
+      "step": 11125
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011054687243687443,
+      "loss": 0.0732,
+      "step": 11126
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001105329087572066,
+      "loss": 0.1071,
+      "step": 11127
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011051894486985643,
+      "loss": 0.1087,
+      "step": 11128
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011050498077509925,
+      "loss": 0.0981,
+      "step": 11129
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011049101647321038,
+      "loss": 0.0593,
+      "step": 11130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011047705196446515,
+      "loss": 0.1556,
+      "step": 11131
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011046308724913887,
+      "loss": 0.1427,
+      "step": 11132
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011044912232750694,
+      "loss": 0.0851,
+      "step": 11133
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011043515719984475,
+      "loss": 0.0852,
+      "step": 11134
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011042119186642756,
+      "loss": 0.0797,
+      "step": 11135
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001104072263275308,
+      "loss": 0.1306,
+      "step": 11136
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011039326058342983,
+      "loss": 0.0534,
+      "step": 11137
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011037929463439994,
+      "loss": 0.0445,
+      "step": 11138
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0011036532848071661,
+      "loss": 0.0955,
+      "step": 11139
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.001103513621226552,
+      "loss": 0.1162,
+      "step": 11140
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011033739556049105,
+      "loss": 0.0956,
+      "step": 11141
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011032342879449954,
+      "loss": 0.0952,
+      "step": 11142
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011030946182495607,
+      "loss": 0.0748,
+      "step": 11143
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011029549465213608,
+      "loss": 0.0814,
+      "step": 11144
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011028152727631491,
+      "loss": 0.1152,
+      "step": 11145
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00110267559697768,
+      "loss": 0.0645,
+      "step": 11146
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011025359191677074,
+      "loss": 0.1199,
+      "step": 11147
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011023962393359851,
+      "loss": 0.0761,
+      "step": 11148
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011022565574852676,
+      "loss": 0.0717,
+      "step": 11149
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011021168736183092,
+      "loss": 0.0999,
+      "step": 11150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011019771877378637,
+      "loss": 0.0972,
+      "step": 11151
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011018374998466856,
+      "loss": 0.1265,
+      "step": 11152
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011016978099475292,
+      "loss": 0.09,
+      "step": 11153
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011015581180431487,
+      "loss": 0.1274,
+      "step": 11154
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011014184241362984,
+      "loss": 0.0674,
+      "step": 11155
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001101278728229733,
+      "loss": 0.0785,
+      "step": 11156
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001101139030326207,
+      "loss": 0.1262,
+      "step": 11157
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011009993304284742,
+      "loss": 0.099,
+      "step": 11158
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00110085962853929,
+      "loss": 0.086,
+      "step": 11159
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011007199246614086,
+      "loss": 0.0748,
+      "step": 11160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011005802187975845,
+      "loss": 0.0983,
+      "step": 11161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011004405109505725,
+      "loss": 0.089,
+      "step": 11162
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011003008011231273,
+      "loss": 0.0922,
+      "step": 11163
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011001610893180032,
+      "loss": 0.0728,
+      "step": 11164
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0011000213755379556,
+      "loss": 0.1261,
+      "step": 11165
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001099881659785739,
+      "loss": 0.0826,
+      "step": 11166
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010997419420641084,
+      "loss": 0.0721,
+      "step": 11167
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010996022223758184,
+      "loss": 0.0818,
+      "step": 11168
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010994625007236243,
+      "loss": 0.0862,
+      "step": 11169
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010993227771102804,
+      "loss": 0.1234,
+      "step": 11170
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010991830515385424,
+      "loss": 0.1906,
+      "step": 11171
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001099043324011165,
+      "loss": 0.1091,
+      "step": 11172
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010989035945309035,
+      "loss": 0.0929,
+      "step": 11173
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010987638631005125,
+      "loss": 0.1144,
+      "step": 11174
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010986241297227478,
+      "loss": 0.0963,
+      "step": 11175
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001098484394400364,
+      "loss": 0.1088,
+      "step": 11176
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010983446571361168,
+      "loss": 0.1178,
+      "step": 11177
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001098204917932761,
+      "loss": 0.1046,
+      "step": 11178
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010980651767930527,
+      "loss": 0.1332,
+      "step": 11179
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010979254337197463,
+      "loss": 0.0859,
+      "step": 11180
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001097785688715598,
+      "loss": 0.0912,
+      "step": 11181
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010976459417833624,
+      "loss": 0.078,
+      "step": 11182
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0010975061929257956,
+      "loss": 0.1226,
+      "step": 11183
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001097366442145653,
+      "loss": 0.086,
+      "step": 11184
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00109722668944569,
+      "loss": 0.1107,
+      "step": 11185
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.001097086934828662,
+      "loss": 0.0983,
+      "step": 11186
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010969471782973246,
+      "loss": 0.0618,
+      "step": 11187
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010968074198544338,
+      "loss": 0.0756,
+      "step": 11188
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010966676595027453,
+      "loss": 0.0855,
+      "step": 11189
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010965278972450145,
+      "loss": 0.0925,
+      "step": 11190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010963881330839974,
+      "loss": 0.0729,
+      "step": 11191
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010962483670224495,
+      "loss": 0.0726,
+      "step": 11192
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001096108599063127,
+      "loss": 0.0802,
+      "step": 11193
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010959688292087855,
+      "loss": 0.1359,
+      "step": 11194
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001095829057462181,
+      "loss": 0.0699,
+      "step": 11195
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010956892838260694,
+      "loss": 0.0779,
+      "step": 11196
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001095549508303207,
+      "loss": 0.0729,
+      "step": 11197
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010954097308963492,
+      "loss": 0.0856,
+      "step": 11198
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010952699516082528,
+      "loss": 0.0859,
+      "step": 11199
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010951301704416733,
+      "loss": 0.1265,
+      "step": 11200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001094990387399367,
+      "loss": 0.0964,
+      "step": 11201
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010948506024840898,
+      "loss": 0.0994,
+      "step": 11202
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010947108156985988,
+      "loss": 0.0986,
+      "step": 11203
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010945710270456495,
+      "loss": 0.0949,
+      "step": 11204
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010944312365279982,
+      "loss": 0.0692,
+      "step": 11205
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010942914441484015,
+      "loss": 0.1377,
+      "step": 11206
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010941516499096153,
+      "loss": 0.122,
+      "step": 11207
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010940118538143966,
+      "loss": 0.12,
+      "step": 11208
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010938720558655015,
+      "loss": 0.089,
+      "step": 11209
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010937322560656867,
+      "loss": 0.098,
+      "step": 11210
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010935924544177079,
+      "loss": 0.0795,
+      "step": 11211
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010934526509243227,
+      "loss": 0.0558,
+      "step": 11212
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010933128455882868,
+      "loss": 0.0823,
+      "step": 11213
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010931730384123575,
+      "loss": 0.1011,
+      "step": 11214
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010930332293992908,
+      "loss": 0.0731,
+      "step": 11215
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010928934185518442,
+      "loss": 0.0942,
+      "step": 11216
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010927536058727735,
+      "loss": 0.0922,
+      "step": 11217
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001092613791364836,
+      "loss": 0.0981,
+      "step": 11218
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010924739750307886,
+      "loss": 0.0996,
+      "step": 11219
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001092334156873388,
+      "loss": 0.0839,
+      "step": 11220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010921943368953906,
+      "loss": 0.0897,
+      "step": 11221
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.001092054515099554,
+      "loss": 0.1005,
+      "step": 11222
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010919146914886344,
+      "loss": 0.072,
+      "step": 11223
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010917748660653895,
+      "loss": 0.0897,
+      "step": 11224
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010916350388325757,
+      "loss": 0.0852,
+      "step": 11225
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010914952097929506,
+      "loss": 0.094,
+      "step": 11226
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010913553789492709,
+      "loss": 0.0905,
+      "step": 11227
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010912155463042937,
+      "loss": 0.1182,
+      "step": 11228
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010910757118607762,
+      "loss": 0.0774,
+      "step": 11229
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010909358756214758,
+      "loss": 0.0869,
+      "step": 11230
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010907960375891495,
+      "loss": 0.0989,
+      "step": 11231
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0010906561977665546,
+      "loss": 0.0834,
+      "step": 11232
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010905163561564482,
+      "loss": 0.0938,
+      "step": 11233
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001090376512761588,
+      "loss": 0.1304,
+      "step": 11234
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001090236667584731,
+      "loss": 0.0897,
+      "step": 11235
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001090096820628635,
+      "loss": 0.0567,
+      "step": 11236
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010899569718960571,
+      "loss": 0.0766,
+      "step": 11237
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010898171213897547,
+      "loss": 0.1178,
+      "step": 11238
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010896772691124856,
+      "loss": 0.0567,
+      "step": 11239
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010895374150670073,
+      "loss": 0.1106,
+      "step": 11240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001089397559256077,
+      "loss": 0.0824,
+      "step": 11241
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010892577016824528,
+      "loss": 0.1163,
+      "step": 11242
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001089117842348892,
+      "loss": 0.0923,
+      "step": 11243
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001088977981258152,
+      "loss": 0.0918,
+      "step": 11244
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010888381184129912,
+      "loss": 0.0569,
+      "step": 11245
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010886982538161672,
+      "loss": 0.096,
+      "step": 11246
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010885583874704373,
+      "loss": 0.0694,
+      "step": 11247
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010884185193785598,
+      "loss": 0.0995,
+      "step": 11248
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010882786495432924,
+      "loss": 0.0808,
+      "step": 11249
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010881387779673928,
+      "loss": 0.0735,
+      "step": 11250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001087998904653619,
+      "loss": 0.1139,
+      "step": 11251
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001087859029604729,
+      "loss": 0.1526,
+      "step": 11252
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001087719152823481,
+      "loss": 0.1274,
+      "step": 11253
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010875792743126325,
+      "loss": 0.0787,
+      "step": 11254
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010874393940749419,
+      "loss": 0.1024,
+      "step": 11255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010872995121131675,
+      "loss": 0.0836,
+      "step": 11256
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010871596284300672,
+      "loss": 0.0704,
+      "step": 11257
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010870197430283986,
+      "loss": 0.1149,
+      "step": 11258
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010868798559109207,
+      "loss": 0.1239,
+      "step": 11259
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010867399670803912,
+      "loss": 0.1105,
+      "step": 11260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010866000765395688,
+      "loss": 0.0795,
+      "step": 11261
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010864601842912114,
+      "loss": 0.1091,
+      "step": 11262
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010863202903380778,
+      "loss": 0.1003,
+      "step": 11263
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010861803946829258,
+      "loss": 0.0769,
+      "step": 11264
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001086040497328514,
+      "loss": 0.129,
+      "step": 11265
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010859005982776007,
+      "loss": 0.0995,
+      "step": 11266
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010857606975329449,
+      "loss": 0.1061,
+      "step": 11267
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010856207950973042,
+      "loss": 0.0712,
+      "step": 11268
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001085480890973438,
+      "loss": 0.1221,
+      "step": 11269
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010853409851641043,
+      "loss": 0.1005,
+      "step": 11270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010852010776720615,
+      "loss": 0.1166,
+      "step": 11271
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010850611685000692,
+      "loss": 0.1946,
+      "step": 11272
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010849212576508854,
+      "loss": 0.1394,
+      "step": 11273
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010847813451272685,
+      "loss": 0.1699,
+      "step": 11274
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010846414309319777,
+      "loss": 0.123,
+      "step": 11275
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010845015150677713,
+      "loss": 0.0774,
+      "step": 11276
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.001084361597537409,
+      "loss": 0.0919,
+      "step": 11277
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0010842216783436482,
+      "loss": 0.0915,
+      "step": 11278
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010840817574892493,
+      "loss": 0.0847,
+      "step": 11279
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010839418349769703,
+      "loss": 0.1221,
+      "step": 11280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00108380191080957,
+      "loss": 0.0868,
+      "step": 11281
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010836619849898075,
+      "loss": 0.0933,
+      "step": 11282
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010835220575204424,
+      "loss": 0.0793,
+      "step": 11283
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001083382128404233,
+      "loss": 0.0973,
+      "step": 11284
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010832421976439384,
+      "loss": 0.1206,
+      "step": 11285
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010831022652423182,
+      "loss": 0.0889,
+      "step": 11286
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010829623312021306,
+      "loss": 0.1152,
+      "step": 11287
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010828223955261355,
+      "loss": 0.0891,
+      "step": 11288
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010826824582170923,
+      "loss": 0.1246,
+      "step": 11289
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010825425192777596,
+      "loss": 0.1234,
+      "step": 11290
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010824025787108966,
+      "loss": 0.1198,
+      "step": 11291
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001082262636519263,
+      "loss": 0.0917,
+      "step": 11292
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010821226927056177,
+      "loss": 0.0825,
+      "step": 11293
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010819827472727207,
+      "loss": 0.0714,
+      "step": 11294
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010818428002233305,
+      "loss": 0.1136,
+      "step": 11295
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010817028515602073,
+      "loss": 0.1014,
+      "step": 11296
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010815629012861099,
+      "loss": 0.0702,
+      "step": 11297
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010814229494037978,
+      "loss": 0.0598,
+      "step": 11298
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010812829959160313,
+      "loss": 0.0884,
+      "step": 11299
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010811430408255692,
+      "loss": 0.0996,
+      "step": 11300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001081003084135171,
+      "loss": 0.1297,
+      "step": 11301
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010808631258475967,
+      "loss": 0.1482,
+      "step": 11302
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010807231659656054,
+      "loss": 0.07,
+      "step": 11303
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010805832044919574,
+      "loss": 0.1019,
+      "step": 11304
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001080443241429412,
+      "loss": 0.0943,
+      "step": 11305
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010803032767807293,
+      "loss": 0.085,
+      "step": 11306
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010801633105486682,
+      "loss": 0.1439,
+      "step": 11307
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010800233427359892,
+      "loss": 0.1027,
+      "step": 11308
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001079883373345452,
+      "loss": 0.1522,
+      "step": 11309
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010797434023798167,
+      "loss": 0.0955,
+      "step": 11310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010796034298418424,
+      "loss": 0.0896,
+      "step": 11311
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010794634557342896,
+      "loss": 0.0806,
+      "step": 11312
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010793234800599181,
+      "loss": 0.1353,
+      "step": 11313
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001079183502821488,
+      "loss": 0.0957,
+      "step": 11314
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001079043524021759,
+      "loss": 0.0947,
+      "step": 11315
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010789035436634914,
+      "loss": 0.1141,
+      "step": 11316
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001078763561749445,
+      "loss": 0.0908,
+      "step": 11317
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00107862357828238,
+      "loss": 0.0787,
+      "step": 11318
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010784835932650567,
+      "loss": 0.0688,
+      "step": 11319
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010783436067002352,
+      "loss": 0.0883,
+      "step": 11320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010782036185906754,
+      "loss": 0.1113,
+      "step": 11321
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.001078063628939138,
+      "loss": 0.1254,
+      "step": 11322
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0010779236377483827,
+      "loss": 0.0983,
+      "step": 11323
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00107778364502117,
+      "loss": 0.1689,
+      "step": 11324
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00107764365076026,
+      "loss": 0.1079,
+      "step": 11325
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010775036549684138,
+      "loss": 0.1019,
+      "step": 11326
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010773636576483908,
+      "loss": 0.1172,
+      "step": 11327
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001077223658802952,
+      "loss": 0.0505,
+      "step": 11328
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010770836584348576,
+      "loss": 0.1442,
+      "step": 11329
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001076943656546868,
+      "loss": 0.1246,
+      "step": 11330
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010768036531417437,
+      "loss": 0.0792,
+      "step": 11331
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010766636482222457,
+      "loss": 0.1533,
+      "step": 11332
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001076523641791134,
+      "loss": 0.12,
+      "step": 11333
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001076383633851169,
+      "loss": 0.0873,
+      "step": 11334
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010762436244051118,
+      "loss": 0.0863,
+      "step": 11335
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001076103613455723,
+      "loss": 0.1046,
+      "step": 11336
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001075963601005763,
+      "loss": 0.1127,
+      "step": 11337
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010758235870579921,
+      "loss": 0.0848,
+      "step": 11338
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010756835716151723,
+      "loss": 0.1398,
+      "step": 11339
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001075543554680063,
+      "loss": 0.082,
+      "step": 11340
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010754035362554256,
+      "loss": 0.0652,
+      "step": 11341
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010752635163440213,
+      "loss": 0.0789,
+      "step": 11342
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010751234949486102,
+      "loss": 0.136,
+      "step": 11343
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010749834720719536,
+      "loss": 0.1074,
+      "step": 11344
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010748434477168116,
+      "loss": 0.0875,
+      "step": 11345
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010747034218859465,
+      "loss": 0.0952,
+      "step": 11346
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010745633945821184,
+      "loss": 0.1256,
+      "step": 11347
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010744233658080884,
+      "loss": 0.1003,
+      "step": 11348
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010742833355666178,
+      "loss": 0.0801,
+      "step": 11349
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010741433038604668,
+      "loss": 0.0941,
+      "step": 11350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010740032706923976,
+      "loss": 0.1516,
+      "step": 11351
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010738632360651704,
+      "loss": 0.1252,
+      "step": 11352
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001073723199981547,
+      "loss": 0.0938,
+      "step": 11353
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010735831624442882,
+      "loss": 0.0942,
+      "step": 11354
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001073443123456155,
+      "loss": 0.0864,
+      "step": 11355
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001073303083019909,
+      "loss": 0.0931,
+      "step": 11356
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001073163041138311,
+      "loss": 0.0813,
+      "step": 11357
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010730229978141229,
+      "loss": 0.1534,
+      "step": 11358
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010728829530501057,
+      "loss": 0.1273,
+      "step": 11359
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010727429068490206,
+      "loss": 0.1232,
+      "step": 11360
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010726028592136288,
+      "loss": 0.106,
+      "step": 11361
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001072462810146692,
+      "loss": 0.0922,
+      "step": 11362
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001072322759650972,
+      "loss": 0.167,
+      "step": 11363
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010721827077292295,
+      "loss": 0.1895,
+      "step": 11364
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001072042654384226,
+      "loss": 0.0931,
+      "step": 11365
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010719025996187234,
+      "loss": 0.1096,
+      "step": 11366
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010717625434354834,
+      "loss": 0.0563,
+      "step": 11367
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.001071622485837267,
+      "loss": 0.1057,
+      "step": 11368
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010714824268268361,
+      "loss": 0.1002,
+      "step": 11369
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010713423664069524,
+      "loss": 0.1104,
+      "step": 11370
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0010712023045803772,
+      "loss": 0.0858,
+      "step": 11371
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010710622413498722,
+      "loss": 0.0952,
+      "step": 11372
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010709221767181993,
+      "loss": 0.1136,
+      "step": 11373
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010707821106881203,
+      "loss": 0.1411,
+      "step": 11374
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010706420432623966,
+      "loss": 0.0996,
+      "step": 11375
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010705019744437906,
+      "loss": 0.1442,
+      "step": 11376
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.001070361904235063,
+      "loss": 0.0939,
+      "step": 11377
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010702218326389765,
+      "loss": 0.0731,
+      "step": 11378
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010700817596582932,
+      "loss": 0.0685,
+      "step": 11379
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010699416852957744,
+      "loss": 0.0956,
+      "step": 11380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010698016095541819,
+      "loss": 0.075,
+      "step": 11381
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010696615324362782,
+      "loss": 0.1998,
+      "step": 11382
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010695214539448248,
+      "loss": 0.1014,
+      "step": 11383
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.001069381374082584,
+      "loss": 0.121,
+      "step": 11384
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010692412928523173,
+      "loss": 0.0912,
+      "step": 11385
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010691012102567877,
+      "loss": 0.0748,
+      "step": 11386
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010689611262987562,
+      "loss": 0.0782,
+      "step": 11387
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010688210409809855,
+      "loss": 0.0846,
+      "step": 11388
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010686809543062382,
+      "loss": 0.0841,
+      "step": 11389
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010685408662772756,
+      "loss": 0.0775,
+      "step": 11390
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010684007768968597,
+      "loss": 0.0851,
+      "step": 11391
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010682606861677538,
+      "loss": 0.0686,
+      "step": 11392
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010681205940927191,
+      "loss": 0.0806,
+      "step": 11393
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010679805006745186,
+      "loss": 0.1398,
+      "step": 11394
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010678404059159142,
+      "loss": 0.0677,
+      "step": 11395
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010677003098196682,
+      "loss": 0.0729,
+      "step": 11396
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010675602123885433,
+      "loss": 0.1207,
+      "step": 11397
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010674201136253011,
+      "loss": 0.0674,
+      "step": 11398
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010672800135327048,
+      "loss": 0.1198,
+      "step": 11399
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010671399121135166,
+      "loss": 0.088,
+      "step": 11400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010669998093704987,
+      "loss": 0.0864,
+      "step": 11401
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.001066859705306414,
+      "loss": 0.1062,
+      "step": 11402
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010667195999240241,
+      "loss": 0.1194,
+      "step": 11403
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010665794932260928,
+      "loss": 0.0812,
+      "step": 11404
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010664393852153817,
+      "loss": 0.0801,
+      "step": 11405
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010662992758946538,
+      "loss": 0.0807,
+      "step": 11406
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010661591652666718,
+      "loss": 0.1506,
+      "step": 11407
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010660190533341976,
+      "loss": 0.0732,
+      "step": 11408
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010658789400999947,
+      "loss": 0.0939,
+      "step": 11409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010657388255668253,
+      "loss": 0.1206,
+      "step": 11410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010655987097374523,
+      "loss": 0.0814,
+      "step": 11411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010654585926146386,
+      "loss": 0.0768,
+      "step": 11412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010653184742011464,
+      "loss": 0.1697,
+      "step": 11413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.001065178354499739,
+      "loss": 0.0795,
+      "step": 11414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010650382335131788,
+      "loss": 0.13,
+      "step": 11415
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.001064898111244229,
+      "loss": 0.0769,
+      "step": 11416
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0010647579876956526,
+      "loss": 0.0919,
+      "step": 11417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010646178628702116,
+      "loss": 0.1217,
+      "step": 11418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00106447773677067,
+      "loss": 0.0884,
+      "step": 11419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00106433760939979,
+      "loss": 0.092,
+      "step": 11420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010641974807603348,
+      "loss": 0.0869,
+      "step": 11421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010640573508550674,
+      "loss": 0.1045,
+      "step": 11422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001063917219686751,
+      "loss": 0.1377,
+      "step": 11423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001063777087258148,
+      "loss": 0.063,
+      "step": 11424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001063636953572022,
+      "loss": 0.068,
+      "step": 11425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010634968186311363,
+      "loss": 0.1097,
+      "step": 11426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010633566824382534,
+      "loss": 0.1136,
+      "step": 11427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010632165449961366,
+      "loss": 0.075,
+      "step": 11428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010630764063075495,
+      "loss": 0.0694,
+      "step": 11429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010629362663752545,
+      "loss": 0.0684,
+      "step": 11430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010627961252020156,
+      "loss": 0.1073,
+      "step": 11431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010626559827905955,
+      "loss": 0.0901,
+      "step": 11432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010625158391437577,
+      "loss": 0.0495,
+      "step": 11433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010623756942642656,
+      "loss": 0.067,
+      "step": 11434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010622355481548819,
+      "loss": 0.0816,
+      "step": 11435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010620954008183703,
+      "loss": 0.0905,
+      "step": 11436
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010619552522574945,
+      "loss": 0.0618,
+      "step": 11437
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010618151024750174,
+      "loss": 0.0792,
+      "step": 11438
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010616749514737024,
+      "loss": 0.0746,
+      "step": 11439
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010615347992563133,
+      "loss": 0.1044,
+      "step": 11440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010613946458256133,
+      "loss": 0.044,
+      "step": 11441
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001061254491184366,
+      "loss": 0.0963,
+      "step": 11442
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010611143353353347,
+      "loss": 0.0861,
+      "step": 11443
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001060974178281283,
+      "loss": 0.0649,
+      "step": 11444
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010608340200249746,
+      "loss": 0.0869,
+      "step": 11445
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010606938605691727,
+      "loss": 0.101,
+      "step": 11446
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010605536999166415,
+      "loss": 0.0704,
+      "step": 11447
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010604135380701438,
+      "loss": 0.0767,
+      "step": 11448
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001060273375032444,
+      "loss": 0.0681,
+      "step": 11449
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010601332108063054,
+      "loss": 0.0801,
+      "step": 11450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010599930453944915,
+      "loss": 0.0475,
+      "step": 11451
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001059852878799766,
+      "loss": 0.0798,
+      "step": 11452
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010597127110248935,
+      "loss": 0.0653,
+      "step": 11453
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010595725420726367,
+      "loss": 0.1538,
+      "step": 11454
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00105943237194576,
+      "loss": 0.0812,
+      "step": 11455
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010592922006470269,
+      "loss": 0.0876,
+      "step": 11456
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001059152028179201,
+      "loss": 0.0879,
+      "step": 11457
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010590118545450468,
+      "loss": 0.1223,
+      "step": 11458
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001058871679747328,
+      "loss": 0.0724,
+      "step": 11459
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001058731503788808,
+      "loss": 0.1147,
+      "step": 11460
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001058591326672251,
+      "loss": 0.1176,
+      "step": 11461
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.001058451148400421,
+      "loss": 0.1163,
+      "step": 11462
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0010583109689760823,
+      "loss": 0.1002,
+      "step": 11463
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010581707884019984,
+      "loss": 0.1071,
+      "step": 11464
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010580306066809333,
+      "loss": 0.1052,
+      "step": 11465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010578904238156513,
+      "loss": 0.1001,
+      "step": 11466
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.001057750239808916,
+      "loss": 0.1036,
+      "step": 11467
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.001057610054663492,
+      "loss": 0.0775,
+      "step": 11468
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010574698683821432,
+      "loss": 0.1266,
+      "step": 11469
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.001057329680967634,
+      "loss": 0.1041,
+      "step": 11470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010571894924227276,
+      "loss": 0.113,
+      "step": 11471
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010570493027501893,
+      "loss": 0.083,
+      "step": 11472
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010569091119527825,
+      "loss": 0.0698,
+      "step": 11473
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010567689200332718,
+      "loss": 0.1191,
+      "step": 11474
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010566287269944215,
+      "loss": 0.1237,
+      "step": 11475
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010564885328389958,
+      "loss": 0.0997,
+      "step": 11476
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010563483375697587,
+      "loss": 0.1093,
+      "step": 11477
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010562081411894743,
+      "loss": 0.114,
+      "step": 11478
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010560679437009075,
+      "loss": 0.082,
+      "step": 11479
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010559277451068227,
+      "loss": 0.1379,
+      "step": 11480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010557875454099838,
+      "loss": 0.1148,
+      "step": 11481
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010556473446131553,
+      "loss": 0.1854,
+      "step": 11482
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010555071427191017,
+      "loss": 0.1133,
+      "step": 11483
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010553669397305875,
+      "loss": 0.0831,
+      "step": 11484
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010552267356503768,
+      "loss": 0.098,
+      "step": 11485
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010550865304812347,
+      "loss": 0.0702,
+      "step": 11486
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010549463242259252,
+      "loss": 0.1421,
+      "step": 11487
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010548061168872127,
+      "loss": 0.1385,
+      "step": 11488
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010546659084678621,
+      "loss": 0.108,
+      "step": 11489
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010545256989706378,
+      "loss": 0.0515,
+      "step": 11490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010543854883983041,
+      "loss": 0.1249,
+      "step": 11491
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010542452767536264,
+      "loss": 0.1008,
+      "step": 11492
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010541050640393687,
+      "loss": 0.0624,
+      "step": 11493
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010539648502582955,
+      "loss": 0.0745,
+      "step": 11494
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010538246354131718,
+      "loss": 0.1267,
+      "step": 11495
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010536844195067623,
+      "loss": 0.1674,
+      "step": 11496
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010535442025418318,
+      "loss": 0.102,
+      "step": 11497
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010534039845211446,
+      "loss": 0.0596,
+      "step": 11498
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010532637654474652,
+      "loss": 0.1217,
+      "step": 11499
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010531235453235595,
+      "loss": 0.1183,
+      "step": 11500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010529833241521914,
+      "loss": 0.0789,
+      "step": 11501
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.001052843101936126,
+      "loss": 0.0719,
+      "step": 11502
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.001052702878678128,
+      "loss": 0.0841,
+      "step": 11503
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010525626543809623,
+      "loss": 0.1244,
+      "step": 11504
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010524224290473937,
+      "loss": 0.0883,
+      "step": 11505
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010522822026801872,
+      "loss": 0.0959,
+      "step": 11506
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.001052141975282108,
+      "loss": 0.0692,
+      "step": 11507
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00105200174685592,
+      "loss": 0.2319,
+      "step": 11508
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0010518615174043896,
+      "loss": 0.1074,
+      "step": 11509
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010517212869302806,
+      "loss": 0.0773,
+      "step": 11510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010515810554363582,
+      "loss": 0.0729,
+      "step": 11511
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010514408229253883,
+      "loss": 0.0923,
+      "step": 11512
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010513005894001348,
+      "loss": 0.1196,
+      "step": 11513
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010511603548633633,
+      "loss": 0.0797,
+      "step": 11514
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010510201193178385,
+      "loss": 0.0789,
+      "step": 11515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010508798827663264,
+      "loss": 0.0994,
+      "step": 11516
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010507396452115912,
+      "loss": 0.0609,
+      "step": 11517
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010505994066563982,
+      "loss": 0.0624,
+      "step": 11518
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010504591671035128,
+      "loss": 0.0775,
+      "step": 11519
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010503189265556997,
+      "loss": 0.1585,
+      "step": 11520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010501786850157246,
+      "loss": 0.1218,
+      "step": 11521
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010500384424863528,
+      "loss": 0.1224,
+      "step": 11522
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010498981989703492,
+      "loss": 0.0986,
+      "step": 11523
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010497579544704787,
+      "loss": 0.0915,
+      "step": 11524
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010496177089895076,
+      "loss": 0.1189,
+      "step": 11525
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010494774625302,
+      "loss": 0.0973,
+      "step": 11526
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.001049337215095322,
+      "loss": 0.1049,
+      "step": 11527
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010491969666876386,
+      "loss": 0.1028,
+      "step": 11528
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010490567173099155,
+      "loss": 0.0941,
+      "step": 11529
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010489164669649177,
+      "loss": 0.0862,
+      "step": 11530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010487762156554104,
+      "loss": 0.0826,
+      "step": 11531
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010486359633841594,
+      "loss": 0.0996,
+      "step": 11532
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010484957101539302,
+      "loss": 0.124,
+      "step": 11533
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010483554559674879,
+      "loss": 0.1217,
+      "step": 11534
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010482152008275982,
+      "loss": 0.0538,
+      "step": 11535
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010480749447370264,
+      "loss": 0.0825,
+      "step": 11536
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.001047934687698538,
+      "loss": 0.1143,
+      "step": 11537
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010477944297148988,
+      "loss": 0.1368,
+      "step": 11538
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.001047654170788874,
+      "loss": 0.1149,
+      "step": 11539
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010475139109232296,
+      "loss": 0.108,
+      "step": 11540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010473736501207304,
+      "loss": 0.1044,
+      "step": 11541
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010472333883841422,
+      "loss": 0.1121,
+      "step": 11542
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010470931257162315,
+      "loss": 0.0628,
+      "step": 11543
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010469528621197628,
+      "loss": 0.1082,
+      "step": 11544
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010468125975975026,
+      "loss": 0.1345,
+      "step": 11545
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010466723321522161,
+      "loss": 0.1355,
+      "step": 11546
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010465320657866688,
+      "loss": 0.0609,
+      "step": 11547
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010463917985036267,
+      "loss": 0.0953,
+      "step": 11548
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010462515303058556,
+      "loss": 0.0963,
+      "step": 11549
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010461112611961211,
+      "loss": 0.066,
+      "step": 11550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.001045970991177189,
+      "loss": 0.1011,
+      "step": 11551
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010458307202518247,
+      "loss": 0.1367,
+      "step": 11552
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010456904484227947,
+      "loss": 0.1641,
+      "step": 11553
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010455501756928642,
+      "loss": 0.1125,
+      "step": 11554
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0010454099020647996,
+      "loss": 0.1467,
+      "step": 11555
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010452696275413662,
+      "loss": 0.1105,
+      "step": 11556
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010451293521253299,
+      "loss": 0.0996,
+      "step": 11557
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010449890758194567,
+      "loss": 0.0708,
+      "step": 11558
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010448487986265127,
+      "loss": 0.1199,
+      "step": 11559
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010447085205492637,
+      "loss": 0.1075,
+      "step": 11560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010445682415904753,
+      "loss": 0.1078,
+      "step": 11561
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001044427961752914,
+      "loss": 0.1155,
+      "step": 11562
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010442876810393452,
+      "loss": 0.1207,
+      "step": 11563
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010441473994525353,
+      "loss": 0.1144,
+      "step": 11564
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00104400711699525,
+      "loss": 0.1545,
+      "step": 11565
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010438668336702557,
+      "loss": 0.0816,
+      "step": 11566
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010437265494803183,
+      "loss": 0.1443,
+      "step": 11567
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010435862644282035,
+      "loss": 0.0971,
+      "step": 11568
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010434459785166775,
+      "loss": 0.0897,
+      "step": 11569
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010433056917485068,
+      "loss": 0.1005,
+      "step": 11570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010431654041264569,
+      "loss": 0.0795,
+      "step": 11571
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010430251156532946,
+      "loss": 0.1324,
+      "step": 11572
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010428848263317853,
+      "loss": 0.1207,
+      "step": 11573
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010427445361646957,
+      "loss": 0.1004,
+      "step": 11574
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010426042451547918,
+      "loss": 0.1465,
+      "step": 11575
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010424639533048398,
+      "loss": 0.1263,
+      "step": 11576
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001042323660617606,
+      "loss": 0.076,
+      "step": 11577
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001042183367095856,
+      "loss": 0.1112,
+      "step": 11578
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010420430727423567,
+      "loss": 0.0873,
+      "step": 11579
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010419027775598743,
+      "loss": 0.0816,
+      "step": 11580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010417624815511745,
+      "loss": 0.0626,
+      "step": 11581
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010416221847190247,
+      "loss": 0.0853,
+      "step": 11582
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010414818870661902,
+      "loss": 0.0576,
+      "step": 11583
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010413415885954377,
+      "loss": 0.1267,
+      "step": 11584
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001041201289309533,
+      "loss": 0.0846,
+      "step": 11585
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010410609892112434,
+      "loss": 0.0433,
+      "step": 11586
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010409206883033348,
+      "loss": 0.1088,
+      "step": 11587
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010407803865885733,
+      "loss": 0.0966,
+      "step": 11588
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010406400840697254,
+      "loss": 0.0804,
+      "step": 11589
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001040499780749558,
+      "loss": 0.0911,
+      "step": 11590
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001040359476630837,
+      "loss": 0.0703,
+      "step": 11591
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001040219171716329,
+      "loss": 0.0666,
+      "step": 11592
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001040078866008801,
+      "loss": 0.0763,
+      "step": 11593
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010399385595110185,
+      "loss": 0.0834,
+      "step": 11594
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010397982522257483,
+      "loss": 0.0604,
+      "step": 11595
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010396579441557575,
+      "loss": 0.1052,
+      "step": 11596
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010395176353038117,
+      "loss": 0.1204,
+      "step": 11597
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010393773256726782,
+      "loss": 0.138,
+      "step": 11598
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010392370152651233,
+      "loss": 0.1245,
+      "step": 11599
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0010390967040839133,
+      "loss": 0.1323,
+      "step": 11600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.001038956392131815,
+      "loss": 0.0811,
+      "step": 11601
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010388160794115955,
+      "loss": 0.0806,
+      "step": 11602
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010386757659260207,
+      "loss": 0.0819,
+      "step": 11603
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010385354516778572,
+      "loss": 0.0797,
+      "step": 11604
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010383951366698722,
+      "loss": 0.1042,
+      "step": 11605
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010382548209048317,
+      "loss": 0.0919,
+      "step": 11606
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.001038114504385503,
+      "loss": 0.0982,
+      "step": 11607
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010379741871146522,
+      "loss": 0.0796,
+      "step": 11608
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010378338690950467,
+      "loss": 0.1116,
+      "step": 11609
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010376935503294527,
+      "loss": 0.0717,
+      "step": 11610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010375532308206368,
+      "loss": 0.0939,
+      "step": 11611
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010374129105713663,
+      "loss": 0.0594,
+      "step": 11612
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010372725895844077,
+      "loss": 0.095,
+      "step": 11613
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010371322678625276,
+      "loss": 0.0978,
+      "step": 11614
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010369919454084932,
+      "loss": 0.0731,
+      "step": 11615
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010368516222250706,
+      "loss": 0.0936,
+      "step": 11616
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010367112983150275,
+      "loss": 0.1053,
+      "step": 11617
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00103657097368113,
+      "loss": 0.1292,
+      "step": 11618
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010364306483261455,
+      "loss": 0.0798,
+      "step": 11619
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010362903222528403,
+      "loss": 0.0818,
+      "step": 11620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010361499954639817,
+      "loss": 0.1394,
+      "step": 11621
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010360096679623363,
+      "loss": 0.1294,
+      "step": 11622
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010358693397506716,
+      "loss": 0.0671,
+      "step": 11623
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010357290108317537,
+      "loss": 0.0911,
+      "step": 11624
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010355886812083502,
+      "loss": 0.1113,
+      "step": 11625
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010354483508832274,
+      "loss": 0.1299,
+      "step": 11626
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.001035308019859153,
+      "loss": 0.1345,
+      "step": 11627
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010351676881388933,
+      "loss": 0.1071,
+      "step": 11628
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.001035027355725216,
+      "loss": 0.0715,
+      "step": 11629
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010348870226208873,
+      "loss": 0.0868,
+      "step": 11630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010347466888286747,
+      "loss": 0.1003,
+      "step": 11631
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010346063543513451,
+      "loss": 0.0802,
+      "step": 11632
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010344660191916658,
+      "loss": 0.0905,
+      "step": 11633
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010343256833524033,
+      "loss": 0.0738,
+      "step": 11634
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010341853468363253,
+      "loss": 0.0929,
+      "step": 11635
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010340450096461985,
+      "loss": 0.14,
+      "step": 11636
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010339046717847897,
+      "loss": 0.08,
+      "step": 11637
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010337643332548668,
+      "loss": 0.1155,
+      "step": 11638
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010336239940591963,
+      "loss": 0.0737,
+      "step": 11639
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010334836542005457,
+      "loss": 0.0943,
+      "step": 11640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.001033343313681682,
+      "loss": 0.1106,
+      "step": 11641
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010332029725053722,
+      "loss": 0.0916,
+      "step": 11642
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010330626306743835,
+      "loss": 0.0595,
+      "step": 11643
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010329222881914832,
+      "loss": 0.0789,
+      "step": 11644
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010327819450594389,
+      "loss": 0.1223,
+      "step": 11645
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010326416012810173,
+      "loss": 0.1442,
+      "step": 11646
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0010325012568589856,
+      "loss": 0.058,
+      "step": 11647
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001032360911796111,
+      "loss": 0.0831,
+      "step": 11648
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010322205660951613,
+      "loss": 0.1107,
+      "step": 11649
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010320802197589035,
+      "loss": 0.1208,
+      "step": 11650
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010319398727901045,
+      "loss": 0.0994,
+      "step": 11651
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010317995251915317,
+      "loss": 0.0816,
+      "step": 11652
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010316591769659526,
+      "loss": 0.0705,
+      "step": 11653
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010315188281161344,
+      "loss": 0.0772,
+      "step": 11654
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001031378478644845,
+      "loss": 0.1764,
+      "step": 11655
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010312381285548509,
+      "loss": 0.0981,
+      "step": 11656
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00103109777784892,
+      "loss": 0.0972,
+      "step": 11657
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010309574265298192,
+      "loss": 0.1145,
+      "step": 11658
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001030817074600316,
+      "loss": 0.0742,
+      "step": 11659
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001030676722063178,
+      "loss": 0.0974,
+      "step": 11660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010305363689211726,
+      "loss": 0.1042,
+      "step": 11661
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010303960151770671,
+      "loss": 0.0872,
+      "step": 11662
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001030255660833629,
+      "loss": 0.1154,
+      "step": 11663
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010301153058936252,
+      "loss": 0.0653,
+      "step": 11664
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010299749503598242,
+      "loss": 0.1056,
+      "step": 11665
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010298345942349927,
+      "loss": 0.0884,
+      "step": 11666
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010296942375218983,
+      "loss": 0.0861,
+      "step": 11667
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010295538802233084,
+      "loss": 0.0929,
+      "step": 11668
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010294135223419905,
+      "loss": 0.0935,
+      "step": 11669
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010292731638807124,
+      "loss": 0.1011,
+      "step": 11670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010291328048422416,
+      "loss": 0.0788,
+      "step": 11671
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010289924452293454,
+      "loss": 0.0765,
+      "step": 11672
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010288520850447911,
+      "loss": 0.059,
+      "step": 11673
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010287117242913466,
+      "loss": 0.0856,
+      "step": 11674
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010285713629717794,
+      "loss": 0.1021,
+      "step": 11675
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010284310010888573,
+      "loss": 0.0458,
+      "step": 11676
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010282906386453474,
+      "loss": 0.0951,
+      "step": 11677
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010281502756440176,
+      "loss": 0.0994,
+      "step": 11678
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010280099120876355,
+      "loss": 0.0818,
+      "step": 11679
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010278695479789687,
+      "loss": 0.1185,
+      "step": 11680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010277291833207846,
+      "loss": 0.1288,
+      "step": 11681
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001027588818115851,
+      "loss": 0.0796,
+      "step": 11682
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010274484523669358,
+      "loss": 0.0912,
+      "step": 11683
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010273080860768063,
+      "loss": 0.0799,
+      "step": 11684
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010271677192482303,
+      "loss": 0.0901,
+      "step": 11685
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010270273518839755,
+      "loss": 0.0882,
+      "step": 11686
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010268869839868099,
+      "loss": 0.1071,
+      "step": 11687
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010267466155595001,
+      "loss": 0.0742,
+      "step": 11688
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010266062466048151,
+      "loss": 0.0771,
+      "step": 11689
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001026465877125522,
+      "loss": 0.0993,
+      "step": 11690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010263255071243885,
+      "loss": 0.1227,
+      "step": 11691
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0010261851366041828,
+      "loss": 0.1027,
+      "step": 11692
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.001026044765567672,
+      "loss": 0.077,
+      "step": 11693
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010259043940176243,
+      "loss": 0.1049,
+      "step": 11694
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010257640219568074,
+      "loss": 0.0873,
+      "step": 11695
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001025623649387989,
+      "loss": 0.0945,
+      "step": 11696
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001025483276313937,
+      "loss": 0.0835,
+      "step": 11697
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001025342902737419,
+      "loss": 0.0539,
+      "step": 11698
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001025202528661203,
+      "loss": 0.0831,
+      "step": 11699
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001025062154088057,
+      "loss": 0.0795,
+      "step": 11700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001024921779020748,
+      "loss": 0.0901,
+      "step": 11701
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010247814034620448,
+      "loss": 0.0909,
+      "step": 11702
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001024641027414715,
+      "loss": 0.0735,
+      "step": 11703
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010245006508815264,
+      "loss": 0.1125,
+      "step": 11704
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010243602738652467,
+      "loss": 0.1204,
+      "step": 11705
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010242198963686438,
+      "loss": 0.0659,
+      "step": 11706
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001024079518394486,
+      "loss": 0.104,
+      "step": 11707
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010239391399455406,
+      "loss": 0.0604,
+      "step": 11708
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001023798761024576,
+      "loss": 0.1649,
+      "step": 11709
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010236583816343602,
+      "loss": 0.0877,
+      "step": 11710
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010235180017776602,
+      "loss": 0.0876,
+      "step": 11711
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010233776214572448,
+      "loss": 0.0682,
+      "step": 11712
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010232372406758818,
+      "loss": 0.1039,
+      "step": 11713
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001023096859436339,
+      "loss": 0.0586,
+      "step": 11714
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010229564777413846,
+      "loss": 0.0928,
+      "step": 11715
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010228160955937865,
+      "loss": 0.0998,
+      "step": 11716
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010226757129963124,
+      "loss": 0.1481,
+      "step": 11717
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010225353299517306,
+      "loss": 0.0542,
+      "step": 11718
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001022394946462809,
+      "loss": 0.1157,
+      "step": 11719
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010222545625323157,
+      "loss": 0.0869,
+      "step": 11720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010221141781630184,
+      "loss": 0.0865,
+      "step": 11721
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010219737933576853,
+      "loss": 0.1538,
+      "step": 11722
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010218334081190847,
+      "loss": 0.0994,
+      "step": 11723
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010216930224499843,
+      "loss": 0.1619,
+      "step": 11724
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010215526363531525,
+      "loss": 0.1073,
+      "step": 11725
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001021412249831357,
+      "loss": 0.0852,
+      "step": 11726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010212718628873658,
+      "loss": 0.0999,
+      "step": 11727
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001021131475523947,
+      "loss": 0.1301,
+      "step": 11728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010209910877438694,
+      "loss": 0.0807,
+      "step": 11729
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010208506995499004,
+      "loss": 0.1202,
+      "step": 11730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001020710310944808,
+      "loss": 0.1127,
+      "step": 11731
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010205699219313611,
+      "loss": 0.0994,
+      "step": 11732
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010204295325123268,
+      "loss": 0.0964,
+      "step": 11733
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010202891426904738,
+      "loss": 0.1056,
+      "step": 11734
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010201487524685701,
+      "loss": 0.0574,
+      "step": 11735
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.001020008361849384,
+      "loss": 0.1254,
+      "step": 11736
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010198679708356836,
+      "loss": 0.074,
+      "step": 11737
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010197275794302364,
+      "loss": 0.0831,
+      "step": 11738
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010195871876358118,
+      "loss": 0.0698,
+      "step": 11739
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0010194467954551772,
+      "loss": 0.0902,
+      "step": 11740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010193064028911007,
+      "loss": 0.0673,
+      "step": 11741
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.001019166009946351,
+      "loss": 0.1216,
+      "step": 11742
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010190256166236955,
+      "loss": 0.0791,
+      "step": 11743
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010188852229259032,
+      "loss": 0.0756,
+      "step": 11744
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010187448288557419,
+      "loss": 0.0968,
+      "step": 11745
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00101860443441598,
+      "loss": 0.077,
+      "step": 11746
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010184640396093854,
+      "loss": 0.0941,
+      "step": 11747
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.001018323644438727,
+      "loss": 0.0575,
+      "step": 11748
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010181832489067721,
+      "loss": 0.093,
+      "step": 11749
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010180428530162898,
+      "loss": 0.0969,
+      "step": 11750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010179024567700476,
+      "loss": 0.1293,
+      "step": 11751
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010177620601708147,
+      "loss": 0.0638,
+      "step": 11752
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010176216632213587,
+      "loss": 0.0892,
+      "step": 11753
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010174812659244476,
+      "loss": 0.0775,
+      "step": 11754
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010173408682828502,
+      "loss": 0.0712,
+      "step": 11755
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010172004702993349,
+      "loss": 0.0998,
+      "step": 11756
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010170600719766696,
+      "loss": 0.0898,
+      "step": 11757
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010169196733176226,
+      "loss": 0.0725,
+      "step": 11758
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010167792743249627,
+      "loss": 0.1085,
+      "step": 11759
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010166388750014576,
+      "loss": 0.1196,
+      "step": 11760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.001016498475349876,
+      "loss": 0.071,
+      "step": 11761
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010163580753729865,
+      "loss": 0.0911,
+      "step": 11762
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010162176750735565,
+      "loss": 0.1324,
+      "step": 11763
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.001016077274454355,
+      "loss": 0.1382,
+      "step": 11764
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010159368735181504,
+      "loss": 0.1196,
+      "step": 11765
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010157964722677108,
+      "loss": 0.1387,
+      "step": 11766
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010156560707058047,
+      "loss": 0.1238,
+      "step": 11767
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010155156688352006,
+      "loss": 0.0703,
+      "step": 11768
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010153752666586665,
+      "loss": 0.0933,
+      "step": 11769
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010152348641789708,
+      "loss": 0.1011,
+      "step": 11770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010150944613988822,
+      "loss": 0.0972,
+      "step": 11771
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010149540583211688,
+      "loss": 0.1028,
+      "step": 11772
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010148136549485995,
+      "loss": 0.0917,
+      "step": 11773
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.001014673251283942,
+      "loss": 0.0744,
+      "step": 11774
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010145328473299654,
+      "loss": 0.1464,
+      "step": 11775
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010143924430894373,
+      "loss": 0.1165,
+      "step": 11776
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010142520385651266,
+      "loss": 0.1289,
+      "step": 11777
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.001014111633759802,
+      "loss": 0.1013,
+      "step": 11778
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010139712286762314,
+      "loss": 0.077,
+      "step": 11779
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010138308233171834,
+      "loss": 0.0984,
+      "step": 11780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010136904176854264,
+      "loss": 0.14,
+      "step": 11781
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010135500117837294,
+      "loss": 0.0951,
+      "step": 11782
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00101340960561486,
+      "loss": 0.1703,
+      "step": 11783
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010132691991815873,
+      "loss": 0.0742,
+      "step": 11784
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010131287924866792,
+      "loss": 0.1131,
+      "step": 11785
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0010129883855329045,
+      "loss": 0.1674,
+      "step": 11786
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010128479783230316,
+      "loss": 0.1854,
+      "step": 11787
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010127075708598293,
+      "loss": 0.0963,
+      "step": 11788
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010125671631460655,
+      "loss": 0.0918,
+      "step": 11789
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010124267551845094,
+      "loss": 0.1455,
+      "step": 11790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010122863469779286,
+      "loss": 0.1041,
+      "step": 11791
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001012145938529092,
+      "loss": 0.1165,
+      "step": 11792
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010120055298407688,
+      "loss": 0.1044,
+      "step": 11793
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001011865120915726,
+      "loss": 0.0956,
+      "step": 11794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010117247117567337,
+      "loss": 0.0944,
+      "step": 11795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010115843023665591,
+      "loss": 0.0894,
+      "step": 11796
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010114438927479716,
+      "loss": 0.0687,
+      "step": 11797
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010113034829037394,
+      "loss": 0.1176,
+      "step": 11798
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010111630728366313,
+      "loss": 0.097,
+      "step": 11799
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010110226625494154,
+      "loss": 0.0911,
+      "step": 11800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010108822520448603,
+      "loss": 0.0864,
+      "step": 11801
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010107418413257347,
+      "loss": 0.1154,
+      "step": 11802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010106014303948075,
+      "loss": 0.1054,
+      "step": 11803
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010104610192548462,
+      "loss": 0.1547,
+      "step": 11804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010103206079086207,
+      "loss": 0.0609,
+      "step": 11805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010101801963588986,
+      "loss": 0.0878,
+      "step": 11806
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010100397846084486,
+      "loss": 0.116,
+      "step": 11807
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010098993726600392,
+      "loss": 0.0984,
+      "step": 11808
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010097589605164397,
+      "loss": 0.1013,
+      "step": 11809
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010096185481804178,
+      "loss": 0.1467,
+      "step": 11810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010094781356547425,
+      "loss": 0.0463,
+      "step": 11811
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010093377229421824,
+      "loss": 0.0906,
+      "step": 11812
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010091973100455056,
+      "loss": 0.0873,
+      "step": 11813
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010090568969674813,
+      "loss": 0.1716,
+      "step": 11814
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010089164837108779,
+      "loss": 0.0932,
+      "step": 11815
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010087760702784641,
+      "loss": 0.0952,
+      "step": 11816
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010086356566730082,
+      "loss": 0.106,
+      "step": 11817
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001008495242897279,
+      "loss": 0.1282,
+      "step": 11818
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010083548289540449,
+      "loss": 0.097,
+      "step": 11819
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001008214414846075,
+      "loss": 0.0994,
+      "step": 11820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010080740005761373,
+      "loss": 0.1398,
+      "step": 11821
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010079335861470006,
+      "loss": 0.0717,
+      "step": 11822
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001007793171561434,
+      "loss": 0.0798,
+      "step": 11823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001007652756822205,
+      "loss": 0.1477,
+      "step": 11824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010075123419320837,
+      "loss": 0.069,
+      "step": 11825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001007371926893838,
+      "loss": 0.1264,
+      "step": 11826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010072315117102362,
+      "loss": 0.0911,
+      "step": 11827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010070910963840473,
+      "loss": 0.0762,
+      "step": 11828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00100695068091804,
+      "loss": 0.0737,
+      "step": 11829
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010068102653149826,
+      "loss": 0.0548,
+      "step": 11830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.001006669849577644,
+      "loss": 0.1696,
+      "step": 11831
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0010065294337087932,
+      "loss": 0.1134,
+      "step": 11832
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010063890177111979,
+      "loss": 0.077,
+      "step": 11833
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010062486015876275,
+      "loss": 0.0953,
+      "step": 11834
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010061081853408507,
+      "loss": 0.0458,
+      "step": 11835
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010059677689736358,
+      "loss": 0.1076,
+      "step": 11836
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010058273524887514,
+      "loss": 0.1183,
+      "step": 11837
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010056869358889664,
+      "loss": 0.0795,
+      "step": 11838
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010055465191770492,
+      "loss": 0.0729,
+      "step": 11839
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.001005406102355769,
+      "loss": 0.0842,
+      "step": 11840
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010052656854278939,
+      "loss": 0.1069,
+      "step": 11841
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010051252683961928,
+      "loss": 0.0693,
+      "step": 11842
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010049848512634345,
+      "loss": 0.1066,
+      "step": 11843
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010048444340323875,
+      "loss": 0.1088,
+      "step": 11844
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00100470401670582,
+      "loss": 0.1139,
+      "step": 11845
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.001004563599286502,
+      "loss": 0.1158,
+      "step": 11846
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010044231817772007,
+      "loss": 0.103,
+      "step": 11847
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010042827641806855,
+      "loss": 0.0669,
+      "step": 11848
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010041423464997254,
+      "loss": 0.1196,
+      "step": 11849
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010040019287370883,
+      "loss": 0.0961,
+      "step": 11850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010038615108955433,
+      "loss": 0.0792,
+      "step": 11851
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010037210929778593,
+      "loss": 0.0638,
+      "step": 11852
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010035806749868045,
+      "loss": 0.1084,
+      "step": 11853
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.001003440256925148,
+      "loss": 0.1062,
+      "step": 11854
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.001003299838795658,
+      "loss": 0.1063,
+      "step": 11855
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010031594206011039,
+      "loss": 0.0839,
+      "step": 11856
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010030190023442539,
+      "loss": 0.0884,
+      "step": 11857
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010028785840278766,
+      "loss": 0.1112,
+      "step": 11858
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010027381656547412,
+      "loss": 0.0714,
+      "step": 11859
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010025977472276158,
+      "loss": 0.1017,
+      "step": 11860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010024573287492693,
+      "loss": 0.0906,
+      "step": 11861
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.001002316910222471,
+      "loss": 0.0897,
+      "step": 11862
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010021764916499888,
+      "loss": 0.1198,
+      "step": 11863
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010020360730345914,
+      "loss": 0.0793,
+      "step": 11864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010018956543790484,
+      "loss": 0.097,
+      "step": 11865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010017552356861273,
+      "loss": 0.0769,
+      "step": 11866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010016148169585974,
+      "loss": 0.1013,
+      "step": 11867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010014743981992277,
+      "loss": 0.1124,
+      "step": 11868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010013339794107865,
+      "loss": 0.1417,
+      "step": 11869
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010011935605960427,
+      "loss": 0.0963,
+      "step": 11870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010010531417577644,
+      "loss": 0.0963,
+      "step": 11871
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010009127228987214,
+      "loss": 0.0641,
+      "step": 11872
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010007723040216816,
+      "loss": 0.0818,
+      "step": 11873
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010006318851294139,
+      "loss": 0.0728,
+      "step": 11874
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010004914662246872,
+      "loss": 0.1096,
+      "step": 11875
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010003510473102697,
+      "loss": 0.0934,
+      "step": 11876
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010002106283889304,
+      "loss": 0.1221,
+      "step": 11877
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0010000702094634384,
+      "loss": 0.1513,
+      "step": 11878
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009999297905365619,
+      "loss": 0.067,
+      "step": 11879
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009997893716110697,
+      "loss": 0.1326,
+      "step": 11880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009996489526897306,
+      "loss": 0.1118,
+      "step": 11881
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000999508533775313,
+      "loss": 0.064,
+      "step": 11882
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009993681148705862,
+      "loss": 0.0786,
+      "step": 11883
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009992276959783185,
+      "loss": 0.0908,
+      "step": 11884
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009990872771012788,
+      "loss": 0.0944,
+      "step": 11885
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009989468582422356,
+      "loss": 0.1399,
+      "step": 11886
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009988064394039577,
+      "loss": 0.0529,
+      "step": 11887
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009986660205892138,
+      "loss": 0.0593,
+      "step": 11888
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009985256018007723,
+      "loss": 0.1121,
+      "step": 11889
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009983851830414024,
+      "loss": 0.0909,
+      "step": 11890
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000998244764313873,
+      "loss": 0.0743,
+      "step": 11891
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009981043456209519,
+      "loss": 0.0997,
+      "step": 11892
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009979639269654087,
+      "loss": 0.0873,
+      "step": 11893
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009978235083500115,
+      "loss": 0.1779,
+      "step": 11894
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009976830897775295,
+      "loss": 0.1113,
+      "step": 11895
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009975426712507305,
+      "loss": 0.0929,
+      "step": 11896
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009974022527723843,
+      "loss": 0.0856,
+      "step": 11897
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000997261834345259,
+      "loss": 0.1049,
+      "step": 11898
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009971214159721234,
+      "loss": 0.0569,
+      "step": 11899
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009969809976557464,
+      "loss": 0.0849,
+      "step": 11900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009968405793988964,
+      "loss": 0.0658,
+      "step": 11901
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009967001612043419,
+      "loss": 0.1537,
+      "step": 11902
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009965597430748522,
+      "loss": 0.0818,
+      "step": 11903
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009964193250131955,
+      "loss": 0.077,
+      "step": 11904
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000996278907022141,
+      "loss": 0.1219,
+      "step": 11905
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000996138489104457,
+      "loss": 0.087,
+      "step": 11906
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009959980712629121,
+      "loss": 0.077,
+      "step": 11907
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009958576535002751,
+      "loss": 0.0964,
+      "step": 11908
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009957172358193143,
+      "loss": 0.075,
+      "step": 11909
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009955768182227994,
+      "loss": 0.0697,
+      "step": 11910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009954364007134983,
+      "loss": 0.0544,
+      "step": 11911
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00099529598329418,
+      "loss": 0.093,
+      "step": 11912
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009951555659676128,
+      "loss": 0.1249,
+      "step": 11913
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000995015148736566,
+      "loss": 0.0966,
+      "step": 11914
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009948747316038075,
+      "loss": 0.1163,
+      "step": 11915
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009947343145721062,
+      "loss": 0.0743,
+      "step": 11916
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000994593897644231,
+      "loss": 0.1502,
+      "step": 11917
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009944534808229509,
+      "loss": 0.0746,
+      "step": 11918
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009943130641110337,
+      "loss": 0.1168,
+      "step": 11919
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009941726475112488,
+      "loss": 0.0546,
+      "step": 11920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009940322310263647,
+      "loss": 0.134,
+      "step": 11921
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009938918146591494,
+      "loss": 0.1204,
+      "step": 11922
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009937513984123725,
+      "loss": 0.0489,
+      "step": 11923
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0009936109822888022,
+      "loss": 0.0978,
+      "step": 11924
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000993470566291207,
+      "loss": 0.1158,
+      "step": 11925
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009933301504223562,
+      "loss": 0.0917,
+      "step": 11926
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009931897346850177,
+      "loss": 0.1018,
+      "step": 11927
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009930493190819603,
+      "loss": 0.1261,
+      "step": 11928
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009929089036159527,
+      "loss": 0.1296,
+      "step": 11929
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009927684882897636,
+      "loss": 0.1343,
+      "step": 11930
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009926280731061621,
+      "loss": 0.0962,
+      "step": 11931
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009924876580679164,
+      "loss": 0.0952,
+      "step": 11932
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000992347243177795,
+      "loss": 0.0809,
+      "step": 11933
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009922068284385663,
+      "loss": 0.1019,
+      "step": 11934
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009920664138529996,
+      "loss": 0.1052,
+      "step": 11935
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009919259994238627,
+      "loss": 0.1455,
+      "step": 11936
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009917855851539253,
+      "loss": 0.0549,
+      "step": 11937
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009916451710459552,
+      "loss": 0.089,
+      "step": 11938
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009915047571027213,
+      "loss": 0.1292,
+      "step": 11939
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000991364343326992,
+      "loss": 0.0951,
+      "step": 11940
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009912239297215361,
+      "loss": 0.0786,
+      "step": 11941
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000991083516289122,
+      "loss": 0.0748,
+      "step": 11942
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009909431030325187,
+      "loss": 0.1298,
+      "step": 11943
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009908026899544946,
+      "loss": 0.0947,
+      "step": 11944
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009906622770578179,
+      "loss": 0.1033,
+      "step": 11945
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009905218643452578,
+      "loss": 0.0881,
+      "step": 11946
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009903814518195824,
+      "loss": 0.0582,
+      "step": 11947
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009902410394835607,
+      "loss": 0.1155,
+      "step": 11948
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009901006273399606,
+      "loss": 0.0876,
+      "step": 11949
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009899602153915516,
+      "loss": 0.0659,
+      "step": 11950
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009898198036411017,
+      "loss": 0.095,
+      "step": 11951
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009896793920913796,
+      "loss": 0.0801,
+      "step": 11952
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009895389807451538,
+      "loss": 0.1327,
+      "step": 11953
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000989398569605193,
+      "loss": 0.1047,
+      "step": 11954
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009892581586742654,
+      "loss": 0.0929,
+      "step": 11955
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009891177479551397,
+      "loss": 0.0762,
+      "step": 11956
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009889773374505847,
+      "loss": 0.1141,
+      "step": 11957
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009888369271633687,
+      "loss": 0.0759,
+      "step": 11958
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009886965170962606,
+      "loss": 0.0994,
+      "step": 11959
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009885561072520287,
+      "loss": 0.075,
+      "step": 11960
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000988415697633441,
+      "loss": 0.1025,
+      "step": 11961
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009882752882432663,
+      "loss": 0.1031,
+      "step": 11962
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000988134879084274,
+      "loss": 0.088,
+      "step": 11963
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009879944701592317,
+      "loss": 0.0785,
+      "step": 11964
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000987854061470908,
+      "loss": 0.0872,
+      "step": 11965
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009877136530220716,
+      "loss": 0.0599,
+      "step": 11966
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000987573244815491,
+      "loss": 0.0864,
+      "step": 11967
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009874328368539348,
+      "loss": 0.0917,
+      "step": 11968
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009872924291401707,
+      "loss": 0.0743,
+      "step": 11969
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0009871520216769684,
+      "loss": 0.0879,
+      "step": 11970
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009870116144670956,
+      "loss": 0.1009,
+      "step": 11971
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000986871207513321,
+      "loss": 0.075,
+      "step": 11972
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000986730800818413,
+      "loss": 0.1108,
+      "step": 11973
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009865903943851403,
+      "loss": 0.1364,
+      "step": 11974
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000986449988216271,
+      "loss": 0.0723,
+      "step": 11975
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009863095823145734,
+      "loss": 0.1099,
+      "step": 11976
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009861691766828167,
+      "loss": 0.0563,
+      "step": 11977
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009860287713237686,
+      "loss": 0.1013,
+      "step": 11978
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009858883662401982,
+      "loss": 0.0638,
+      "step": 11979
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009857479614348735,
+      "loss": 0.0738,
+      "step": 11980
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009856075569105631,
+      "loss": 0.0944,
+      "step": 11981
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009854671526700349,
+      "loss": 0.1298,
+      "step": 11982
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009853267487160578,
+      "loss": 0.0875,
+      "step": 11983
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009851863450514005,
+      "loss": 0.0804,
+      "step": 11984
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009850459416788312,
+      "loss": 0.0931,
+      "step": 11985
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000984905538601118,
+      "loss": 0.0986,
+      "step": 11986
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009847651358210294,
+      "loss": 0.1197,
+      "step": 11987
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009846247333413338,
+      "loss": 0.1268,
+      "step": 11988
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009844843311647995,
+      "loss": 0.054,
+      "step": 11989
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009843439292941954,
+      "loss": 0.1484,
+      "step": 11990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009842035277322892,
+      "loss": 0.0738,
+      "step": 11991
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009840631264818498,
+      "loss": 0.0707,
+      "step": 11992
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000983922725545645,
+      "loss": 0.0997,
+      "step": 11993
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009837823249264437,
+      "loss": 0.0742,
+      "step": 11994
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000983641924627014,
+      "loss": 0.094,
+      "step": 11995
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009835015246501238,
+      "loss": 0.0947,
+      "step": 11996
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009833611249985424,
+      "loss": 0.1233,
+      "step": 11997
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009832207256750376,
+      "loss": 0.0992,
+      "step": 11998
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009830803266823774,
+      "loss": 0.1121,
+      "step": 11999
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009829399280233307,
+      "loss": 0.14,
+      "step": 12000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009827995297006656,
+      "loss": 0.088,
+      "step": 12001
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009826591317171497,
+      "loss": 0.1647,
+      "step": 12002
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009825187340755527,
+      "loss": 0.0729,
+      "step": 12003
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009823783367786415,
+      "loss": 0.0801,
+      "step": 12004
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009822379398291856,
+      "loss": 0.0916,
+      "step": 12005
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009820975432299525,
+      "loss": 0.0948,
+      "step": 12006
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009819571469837107,
+      "loss": 0.0732,
+      "step": 12007
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009818167510932281,
+      "loss": 0.0883,
+      "step": 12008
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009816763555612733,
+      "loss": 0.1564,
+      "step": 12009
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009815359603906144,
+      "loss": 0.1035,
+      "step": 12010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00098139556558402,
+      "loss": 0.1005,
+      "step": 12011
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009812551711442584,
+      "loss": 0.0901,
+      "step": 12012
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000981114777074097,
+      "loss": 0.0916,
+      "step": 12013
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009809743833763046,
+      "loss": 0.0766,
+      "step": 12014
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009808339900536495,
+      "loss": 0.1174,
+      "step": 12015
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0009806935971088994,
+      "loss": 0.0803,
+      "step": 12016
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000980553204544823,
+      "loss": 0.1465,
+      "step": 12017
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009804128123641882,
+      "loss": 0.0891,
+      "step": 12018
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009802724205697636,
+      "loss": 0.0882,
+      "step": 12019
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009801320291643167,
+      "loss": 0.0728,
+      "step": 12020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009799916381506163,
+      "loss": 0.0837,
+      "step": 12021
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00097985124753143,
+      "loss": 0.102,
+      "step": 12022
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009797108573095263,
+      "loss": 0.0839,
+      "step": 12023
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009795704674876735,
+      "loss": 0.0914,
+      "step": 12024
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009794300780686391,
+      "loss": 0.0777,
+      "step": 12025
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000979289689055192,
+      "loss": 0.0646,
+      "step": 12026
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009791493004500998,
+      "loss": 0.0809,
+      "step": 12027
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009790089122561308,
+      "loss": 0.0669,
+      "step": 12028
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009788685244760528,
+      "loss": 0.1301,
+      "step": 12029
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009787281371126345,
+      "loss": 0.1007,
+      "step": 12030
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009785877501686433,
+      "loss": 0.0833,
+      "step": 12031
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009784473636468478,
+      "loss": 0.0665,
+      "step": 12032
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000978306977550016,
+      "loss": 0.0691,
+      "step": 12033
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009781665918809155,
+      "loss": 0.1146,
+      "step": 12034
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009780262066423145,
+      "loss": 0.088,
+      "step": 12035
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009778858218369814,
+      "loss": 0.0699,
+      "step": 12036
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009777454374676844,
+      "loss": 0.1139,
+      "step": 12037
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000977605053537191,
+      "loss": 0.0974,
+      "step": 12038
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009774646700482696,
+      "loss": 0.0858,
+      "step": 12039
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009773242870036876,
+      "loss": 0.1123,
+      "step": 12040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009771839044062137,
+      "loss": 0.053,
+      "step": 12041
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009770435222586152,
+      "loss": 0.065,
+      "step": 12042
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009769031405636609,
+      "loss": 0.1015,
+      "step": 12043
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009767627593241183,
+      "loss": 0.1042,
+      "step": 12044
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009766223785427554,
+      "loss": 0.0709,
+      "step": 12045
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009764819982223399,
+      "loss": 0.0692,
+      "step": 12046
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009763416183656404,
+      "loss": 0.079,
+      "step": 12047
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009762012389754243,
+      "loss": 0.069,
+      "step": 12048
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009760608600544593,
+      "loss": 0.1039,
+      "step": 12049
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009759204816055142,
+      "loss": 0.08,
+      "step": 12050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009757801036313563,
+      "loss": 0.1019,
+      "step": 12051
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009756397261347534,
+      "loss": 0.1144,
+      "step": 12052
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009754993491184738,
+      "loss": 0.063,
+      "step": 12053
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009753589725852852,
+      "loss": 0.1307,
+      "step": 12054
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000975218596537955,
+      "loss": 0.0869,
+      "step": 12055
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009750782209792519,
+      "loss": 0.0948,
+      "step": 12056
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009749378459119432,
+      "loss": 0.1157,
+      "step": 12057
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009747974713387971,
+      "loss": 0.1655,
+      "step": 12058
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009746570972625812,
+      "loss": 0.1352,
+      "step": 12059
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009745167236860634,
+      "loss": 0.1373,
+      "step": 12060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009743763506120112,
+      "loss": 0.1154,
+      "step": 12061
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0009742359780431927,
+      "loss": 0.129,
+      "step": 12062
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009740956059823756,
+      "loss": 0.1667,
+      "step": 12063
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000973955234432328,
+      "loss": 0.0956,
+      "step": 12064
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009738148633958174,
+      "loss": 0.0856,
+      "step": 12065
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009736744928756117,
+      "loss": 0.1445,
+      "step": 12066
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009735341228744781,
+      "loss": 0.1193,
+      "step": 12067
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009733937533951852,
+      "loss": 0.0886,
+      "step": 12068
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009732533844404997,
+      "loss": 0.0703,
+      "step": 12069
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009731130160131905,
+      "loss": 0.1035,
+      "step": 12070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009729726481160246,
+      "loss": 0.1138,
+      "step": 12071
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009728322807517699,
+      "loss": 0.0844,
+      "step": 12072
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009726919139231938,
+      "loss": 0.1127,
+      "step": 12073
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009725515476330644,
+      "loss": 0.0897,
+      "step": 12074
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009724111818841488,
+      "loss": 0.1055,
+      "step": 12075
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009722708166792155,
+      "loss": 0.0941,
+      "step": 12076
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009721304520210316,
+      "loss": 0.2036,
+      "step": 12077
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009719900879123645,
+      "loss": 0.1069,
+      "step": 12078
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009718497243559825,
+      "loss": 0.08,
+      "step": 12079
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009717093613546529,
+      "loss": 0.0774,
+      "step": 12080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009715689989111432,
+      "loss": 0.0739,
+      "step": 12081
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009714286370282206,
+      "loss": 0.0632,
+      "step": 12082
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009712882757086536,
+      "loss": 0.103,
+      "step": 12083
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000971147914955209,
+      "loss": 0.0786,
+      "step": 12084
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000971007554770655,
+      "loss": 0.1127,
+      "step": 12085
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009708671951577587,
+      "loss": 0.0696,
+      "step": 12086
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009707268361192878,
+      "loss": 0.1194,
+      "step": 12087
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009705864776580096,
+      "loss": 0.115,
+      "step": 12088
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009704461197766917,
+      "loss": 0.1516,
+      "step": 12089
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009703057624781019,
+      "loss": 0.0698,
+      "step": 12090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009701654057650076,
+      "loss": 0.0695,
+      "step": 12091
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000970025049640176,
+      "loss": 0.0919,
+      "step": 12092
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009698846941063749,
+      "loss": 0.0776,
+      "step": 12093
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009697443391663713,
+      "loss": 0.0618,
+      "step": 12094
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009696039848229328,
+      "loss": 0.0883,
+      "step": 12095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009694636310788274,
+      "loss": 0.1147,
+      "step": 12096
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009693232779368221,
+      "loss": 0.084,
+      "step": 12097
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009691829253996843,
+      "loss": 0.0707,
+      "step": 12098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009690425734701811,
+      "loss": 0.0819,
+      "step": 12099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009689022221510804,
+      "loss": 0.0852,
+      "step": 12100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009687618714451494,
+      "loss": 0.0839,
+      "step": 12101
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000968621521355155,
+      "loss": 0.1713,
+      "step": 12102
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009684811718838654,
+      "loss": 0.162,
+      "step": 12103
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009683408230340474,
+      "loss": 0.1365,
+      "step": 12104
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009682004748084684,
+      "loss": 0.086,
+      "step": 12105
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009680601272098959,
+      "loss": 0.0659,
+      "step": 12106
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009679197802410969,
+      "loss": 0.1105,
+      "step": 12107
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000967779433904839,
+      "loss": 0.1135,
+      "step": 12108
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0009676390882038888,
+      "loss": 0.1051,
+      "step": 12109
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009674987431410146,
+      "loss": 0.0618,
+      "step": 12110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009673583987189828,
+      "loss": 0.0792,
+      "step": 12111
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009672180549405613,
+      "loss": 0.0889,
+      "step": 12112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009670777118085167,
+      "loss": 0.0918,
+      "step": 12113
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009669373693256168,
+      "loss": 0.0675,
+      "step": 12114
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009667970274946279,
+      "loss": 0.0699,
+      "step": 12115
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000966656686318318,
+      "loss": 0.0784,
+      "step": 12116
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009665163457994543,
+      "loss": 0.099,
+      "step": 12117
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009663760059408037,
+      "loss": 0.0856,
+      "step": 12118
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009662356667451336,
+      "loss": 0.0779,
+      "step": 12119
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009660953282152104,
+      "loss": 0.1503,
+      "step": 12120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000965954990353802,
+      "loss": 0.1174,
+      "step": 12121
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009658146531636748,
+      "loss": 0.1044,
+      "step": 12122
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009656743166475967,
+      "loss": 0.0846,
+      "step": 12123
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009655339808083346,
+      "loss": 0.1906,
+      "step": 12124
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009653936456486551,
+      "loss": 0.0793,
+      "step": 12125
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009652533111713255,
+      "loss": 0.131,
+      "step": 12126
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009651129773791129,
+      "loss": 0.1294,
+      "step": 12127
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009649726442747846,
+      "loss": 0.1595,
+      "step": 12128
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009648323118611066,
+      "loss": 0.1008,
+      "step": 12129
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009646919801408472,
+      "loss": 0.0834,
+      "step": 12130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009645516491167725,
+      "loss": 0.0889,
+      "step": 12131
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009644113187916501,
+      "loss": 0.1021,
+      "step": 12132
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009642709891682465,
+      "loss": 0.0627,
+      "step": 12133
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009641306602493288,
+      "loss": 0.1205,
+      "step": 12134
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009639903320376636,
+      "loss": 0.0934,
+      "step": 12135
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009638500045360184,
+      "loss": 0.0898,
+      "step": 12136
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009637096777471598,
+      "loss": 0.1195,
+      "step": 12137
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009635693516738548,
+      "loss": 0.0964,
+      "step": 12138
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009634290263188702,
+      "loss": 0.1041,
+      "step": 12139
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000963288701684973,
+      "loss": 0.095,
+      "step": 12140
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009631483777749294,
+      "loss": 0.0881,
+      "step": 12141
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000963008054591507,
+      "loss": 0.0737,
+      "step": 12142
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009628677321374723,
+      "loss": 0.0966,
+      "step": 12143
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009627274104155924,
+      "loss": 0.1141,
+      "step": 12144
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009625870894286339,
+      "loss": 0.074,
+      "step": 12145
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009624467691793634,
+      "loss": 0.0654,
+      "step": 12146
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009623064496705476,
+      "loss": 0.1388,
+      "step": 12147
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009621661309049536,
+      "loss": 0.0823,
+      "step": 12148
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009620258128853476,
+      "loss": 0.098,
+      "step": 12149
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009618854956144972,
+      "loss": 0.0779,
+      "step": 12150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009617451790951685,
+      "loss": 0.1195,
+      "step": 12151
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009616048633301281,
+      "loss": 0.1228,
+      "step": 12152
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000961464548322143,
+      "loss": 0.1002,
+      "step": 12153
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009613242340739798,
+      "loss": 0.0867,
+      "step": 12154
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0009611839205884046,
+      "loss": 0.0927,
+      "step": 12155
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009610436078681848,
+      "loss": 0.077,
+      "step": 12156
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009609032959160868,
+      "loss": 0.0933,
+      "step": 12157
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009607629847348767,
+      "loss": 0.0809,
+      "step": 12158
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000960622674327322,
+      "loss": 0.1121,
+      "step": 12159
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009604823646961884,
+      "loss": 0.1066,
+      "step": 12160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009603420558442431,
+      "loss": 0.1215,
+      "step": 12161
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009602017477742516,
+      "loss": 0.1204,
+      "step": 12162
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009600614404889818,
+      "loss": 0.0616,
+      "step": 12163
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009599211339911992,
+      "loss": 0.0956,
+      "step": 12164
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009597808282836708,
+      "loss": 0.1055,
+      "step": 12165
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009596405233691631,
+      "loss": 0.1071,
+      "step": 12166
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009595002192504423,
+      "loss": 0.08,
+      "step": 12167
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009593599159302747,
+      "loss": 0.0869,
+      "step": 12168
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009592196134114267,
+      "loss": 0.0651,
+      "step": 12169
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009590793116966654,
+      "loss": 0.1233,
+      "step": 12170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009589390107887567,
+      "loss": 0.0929,
+      "step": 12171
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000958798710690467,
+      "loss": 0.1157,
+      "step": 12172
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009586584114045627,
+      "loss": 0.094,
+      "step": 12173
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009585181129338102,
+      "loss": 0.1106,
+      "step": 12174
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009583778152809754,
+      "loss": 0.0979,
+      "step": 12175
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009582375184488253,
+      "loss": 0.0934,
+      "step": 12176
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000958097222440126,
+      "loss": 0.1191,
+      "step": 12177
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009579569272576436,
+      "loss": 0.0803,
+      "step": 12178
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009578166329041442,
+      "loss": 0.1038,
+      "step": 12179
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009576763393823947,
+      "loss": 0.114,
+      "step": 12180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009575360466951607,
+      "loss": 0.0743,
+      "step": 12181
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009573957548452082,
+      "loss": 0.066,
+      "step": 12182
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009572554638353043,
+      "loss": 0.0514,
+      "step": 12183
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009571151736682149,
+      "loss": 0.1035,
+      "step": 12184
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009569748843467056,
+      "loss": 0.0848,
+      "step": 12185
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009568345958735432,
+      "loss": 0.1194,
+      "step": 12186
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009566943082514936,
+      "loss": 0.1042,
+      "step": 12187
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009565540214833229,
+      "loss": 0.2062,
+      "step": 12188
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009564137355717968,
+      "loss": 0.0727,
+      "step": 12189
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009562734505196819,
+      "loss": 0.0624,
+      "step": 12190
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009561331663297443,
+      "loss": 0.0907,
+      "step": 12191
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009559928830047501,
+      "loss": 0.1445,
+      "step": 12192
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000955852600547465,
+      "loss": 0.0908,
+      "step": 12193
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.000955712318960655,
+      "loss": 0.103,
+      "step": 12194
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009555720382470862,
+      "loss": 0.1068,
+      "step": 12195
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009554317584095247,
+      "loss": 0.0988,
+      "step": 12196
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009552914794507364,
+      "loss": 0.1157,
+      "step": 12197
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009551512013734874,
+      "loss": 0.0724,
+      "step": 12198
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009550109241805435,
+      "loss": 0.0895,
+      "step": 12199
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009548706478746703,
+      "loss": 0.1116,
+      "step": 12200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0009547303724586342,
+      "loss": 0.0919,
+      "step": 12201
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009545900979352005,
+      "loss": 0.0516,
+      "step": 12202
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009544498243071357,
+      "loss": 0.1064,
+      "step": 12203
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009543095515772053,
+      "loss": 0.1086,
+      "step": 12204
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009541692797481754,
+      "loss": 0.0929,
+      "step": 12205
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009540290088228112,
+      "loss": 0.0948,
+      "step": 12206
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009538887388038791,
+      "loss": 0.0616,
+      "step": 12207
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009537484696941447,
+      "loss": 0.1458,
+      "step": 12208
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009536082014963733,
+      "loss": 0.1058,
+      "step": 12209
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009534679342133314,
+      "loss": 0.0786,
+      "step": 12210
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009533276678477841,
+      "loss": 0.0808,
+      "step": 12211
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009531874024024976,
+      "loss": 0.0729,
+      "step": 12212
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009530471378802373,
+      "loss": 0.0875,
+      "step": 12213
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000952906874283769,
+      "loss": 0.0988,
+      "step": 12214
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009527666116158576,
+      "loss": 0.0652,
+      "step": 12215
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009526263498792699,
+      "loss": 0.0801,
+      "step": 12216
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009524860890767707,
+      "loss": 0.218,
+      "step": 12217
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009523458292111261,
+      "loss": 0.0938,
+      "step": 12218
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009522055702851015,
+      "loss": 0.0766,
+      "step": 12219
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009520653123014623,
+      "loss": 0.0791,
+      "step": 12220
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009519250552629739,
+      "loss": 0.0668,
+      "step": 12221
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000951784799172402,
+      "loss": 0.1308,
+      "step": 12222
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009516445440325122,
+      "loss": 0.0941,
+      "step": 12223
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009515042898460699,
+      "loss": 0.1321,
+      "step": 12224
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009513640366158407,
+      "loss": 0.0644,
+      "step": 12225
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009512237843445897,
+      "loss": 0.0598,
+      "step": 12226
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009510835330350827,
+      "loss": 0.1002,
+      "step": 12227
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009509432826900849,
+      "loss": 0.057,
+      "step": 12228
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009508030333123613,
+      "loss": 0.0608,
+      "step": 12229
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000950662784904678,
+      "loss": 0.1414,
+      "step": 12230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009505225374698002,
+      "loss": 0.0828,
+      "step": 12231
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009503822910104927,
+      "loss": 0.0903,
+      "step": 12232
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009502420455295214,
+      "loss": 0.1066,
+      "step": 12233
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009501018010296513,
+      "loss": 0.0876,
+      "step": 12234
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009499615575136472,
+      "loss": 0.101,
+      "step": 12235
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009498213149842753,
+      "loss": 0.084,
+      "step": 12236
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009496810734443003,
+      "loss": 0.0879,
+      "step": 12237
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009495408328964874,
+      "loss": 0.0643,
+      "step": 12238
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009494005933436021,
+      "loss": 0.1697,
+      "step": 12239
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009492603547884092,
+      "loss": 0.063,
+      "step": 12240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009491201172336741,
+      "loss": 0.088,
+      "step": 12241
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009489798806821613,
+      "loss": 0.1028,
+      "step": 12242
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009488396451366366,
+      "loss": 0.1066,
+      "step": 12243
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009486994105998652,
+      "loss": 0.1178,
+      "step": 12244
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009485591770746119,
+      "loss": 0.1185,
+      "step": 12245
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009484189445636417,
+      "loss": 0.1084,
+      "step": 12246
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0009482787130697196,
+      "loss": 0.0517,
+      "step": 12247
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009481384825956108,
+      "loss": 0.1096,
+      "step": 12248
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009479982531440797,
+      "loss": 0.0916,
+      "step": 12249
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009478580247178923,
+      "loss": 0.0851,
+      "step": 12250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009477177973198129,
+      "loss": 0.0582,
+      "step": 12251
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009475775709526066,
+      "loss": 0.112,
+      "step": 12252
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009474373456190379,
+      "loss": 0.1024,
+      "step": 12253
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009472971213218723,
+      "loss": 0.0807,
+      "step": 12254
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009471568980638739,
+      "loss": 0.134,
+      "step": 12255
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009470166758478086,
+      "loss": 0.0988,
+      "step": 12256
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009468764546764405,
+      "loss": 0.0692,
+      "step": 12257
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009467362345525347,
+      "loss": 0.0939,
+      "step": 12258
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009465960154788558,
+      "loss": 0.058,
+      "step": 12259
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009464557974581685,
+      "loss": 0.108,
+      "step": 12260
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000946315580493238,
+      "loss": 0.0642,
+      "step": 12261
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009461753645868281,
+      "loss": 0.0952,
+      "step": 12262
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009460351497417046,
+      "loss": 0.0901,
+      "step": 12263
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009458949359606313,
+      "loss": 0.1464,
+      "step": 12264
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009457547232463736,
+      "loss": 0.0984,
+      "step": 12265
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009456145116016959,
+      "loss": 0.077,
+      "step": 12266
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009454743010293626,
+      "loss": 0.0944,
+      "step": 12267
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009453340915321382,
+      "loss": 0.0912,
+      "step": 12268
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009451938831127874,
+      "loss": 0.0917,
+      "step": 12269
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000945053675774075,
+      "loss": 0.0676,
+      "step": 12270
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009449134695187655,
+      "loss": 0.1272,
+      "step": 12271
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009447732643496233,
+      "loss": 0.0542,
+      "step": 12272
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009446330602694129,
+      "loss": 0.1078,
+      "step": 12273
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009444928572808985,
+      "loss": 0.0575,
+      "step": 12274
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009443526553868448,
+      "loss": 0.0787,
+      "step": 12275
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009442124545900163,
+      "loss": 0.1237,
+      "step": 12276
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009440722548931775,
+      "loss": 0.1089,
+      "step": 12277
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009439320562990925,
+      "loss": 0.0745,
+      "step": 12278
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000943791858810526,
+      "loss": 0.1467,
+      "step": 12279
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009436516624302417,
+      "loss": 0.0748,
+      "step": 12280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009435114671610047,
+      "loss": 0.0859,
+      "step": 12281
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009433712730055785,
+      "loss": 0.0977,
+      "step": 12282
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009432310799667281,
+      "loss": 0.0882,
+      "step": 12283
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009430908880472175,
+      "loss": 0.0472,
+      "step": 12284
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009429506972498109,
+      "loss": 0.073,
+      "step": 12285
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009428105075772725,
+      "loss": 0.108,
+      "step": 12286
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009426703190323666,
+      "loss": 0.1294,
+      "step": 12287
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009425301316178571,
+      "loss": 0.093,
+      "step": 12288
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009423899453365079,
+      "loss": 0.0833,
+      "step": 12289
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009422497601910841,
+      "loss": 0.1157,
+      "step": 12290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009421095761843489,
+      "loss": 0.0827,
+      "step": 12291
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0009419693933190668,
+      "loss": 0.0958,
+      "step": 12292
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000941829211598002,
+      "loss": 0.111,
+      "step": 12293
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000941689031023918,
+      "loss": 0.1017,
+      "step": 12294
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009415488515995787,
+      "loss": 0.082,
+      "step": 12295
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009414086733277489,
+      "loss": 0.0771,
+      "step": 12296
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000941268496211192,
+      "loss": 0.1016,
+      "step": 12297
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009411283202526722,
+      "loss": 0.1055,
+      "step": 12298
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009409881454549533,
+      "loss": 0.1046,
+      "step": 12299
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009408479718207992,
+      "loss": 0.0875,
+      "step": 12300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009407077993529735,
+      "loss": 0.0889,
+      "step": 12301
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00094056762805424,
+      "loss": 0.1376,
+      "step": 12302
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009404274579273633,
+      "loss": 0.0666,
+      "step": 12303
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009402872889751067,
+      "loss": 0.0741,
+      "step": 12304
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000940147121200234,
+      "loss": 0.0734,
+      "step": 12305
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009400069546055087,
+      "loss": 0.0955,
+      "step": 12306
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000939866789193695,
+      "loss": 0.0754,
+      "step": 12307
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000939726624967556,
+      "loss": 0.0714,
+      "step": 12308
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009395864619298562,
+      "loss": 0.0563,
+      "step": 12309
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009394463000833587,
+      "loss": 0.1675,
+      "step": 12310
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009393061394308274,
+      "loss": 0.0811,
+      "step": 12311
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009391659799750256,
+      "loss": 0.1119,
+      "step": 12312
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009390258217187171,
+      "loss": 0.098,
+      "step": 12313
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009388856646646656,
+      "loss": 0.1136,
+      "step": 12314
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009387455088156339,
+      "loss": 0.0751,
+      "step": 12315
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009386053541743867,
+      "loss": 0.1031,
+      "step": 12316
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009384652007436868,
+      "loss": 0.0751,
+      "step": 12317
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009383250485262975,
+      "loss": 0.1776,
+      "step": 12318
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009381848975249829,
+      "loss": 0.0688,
+      "step": 12319
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009380447477425059,
+      "loss": 0.0707,
+      "step": 12320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009379045991816298,
+      "loss": 0.1062,
+      "step": 12321
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009377644518451184,
+      "loss": 0.1569,
+      "step": 12322
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009376243057357347,
+      "loss": 0.0704,
+      "step": 12323
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009374841608562424,
+      "loss": 0.0714,
+      "step": 12324
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009373440172094048,
+      "loss": 0.0731,
+      "step": 12325
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009372038747979848,
+      "loss": 0.1444,
+      "step": 12326
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009370637336247456,
+      "loss": 0.0866,
+      "step": 12327
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009369235936924508,
+      "loss": 0.1232,
+      "step": 12328
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009367834550038632,
+      "loss": 0.083,
+      "step": 12329
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009366433175617467,
+      "loss": 0.0801,
+      "step": 12330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009365031813688639,
+      "loss": 0.0762,
+      "step": 12331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009363630464279781,
+      "loss": 0.0854,
+      "step": 12332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009362229127418522,
+      "loss": 0.1054,
+      "step": 12333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009360827803132494,
+      "loss": 0.0785,
+      "step": 12334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009359426491449325,
+      "loss": 0.1143,
+      "step": 12335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009358025192396652,
+      "loss": 0.1166,
+      "step": 12336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009356623906002101,
+      "loss": 0.0908,
+      "step": 12337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009355222632293301,
+      "loss": 0.0972,
+      "step": 12338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0009353821371297884,
+      "loss": 0.1083,
+      "step": 12339
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009352420123043479,
+      "loss": 0.0681,
+      "step": 12340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009351018887557714,
+      "loss": 0.1041,
+      "step": 12341
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009349617664868212,
+      "loss": 0.0959,
+      "step": 12342
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009348216455002613,
+      "loss": 0.0607,
+      "step": 12343
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009346815257988537,
+      "loss": 0.1031,
+      "step": 12344
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009345414073853618,
+      "loss": 0.1261,
+      "step": 12345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009344012902625479,
+      "loss": 0.1112,
+      "step": 12346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009342611744331751,
+      "loss": 0.0872,
+      "step": 12347
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009341210599000054,
+      "loss": 0.0828,
+      "step": 12348
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009339809466658024,
+      "loss": 0.1285,
+      "step": 12349
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009338408347333285,
+      "loss": 0.1185,
+      "step": 12350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009337007241053462,
+      "loss": 0.0793,
+      "step": 12351
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009335606147846184,
+      "loss": 0.0416,
+      "step": 12352
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009334205067739076,
+      "loss": 0.0608,
+      "step": 12353
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009332804000759759,
+      "loss": 0.1155,
+      "step": 12354
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009331402946935861,
+      "loss": 0.1019,
+      "step": 12355
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009330001906295012,
+      "loss": 0.058,
+      "step": 12356
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009328600878864834,
+      "loss": 0.058,
+      "step": 12357
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009327199864672953,
+      "loss": 0.1013,
+      "step": 12358
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009325798863746988,
+      "loss": 0.062,
+      "step": 12359
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009324397876114571,
+      "loss": 0.0683,
+      "step": 12360
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000932299690180332,
+      "loss": 0.0958,
+      "step": 12361
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009321595940840857,
+      "loss": 0.0486,
+      "step": 12362
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009320194993254814,
+      "loss": 0.0629,
+      "step": 12363
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009318794059072809,
+      "loss": 0.075,
+      "step": 12364
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009317393138322463,
+      "loss": 0.0974,
+      "step": 12365
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009315992231031403,
+      "loss": 0.0736,
+      "step": 12366
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000931459133722725,
+      "loss": 0.0869,
+      "step": 12367
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009313190456937619,
+      "loss": 0.0674,
+      "step": 12368
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009311789590190143,
+      "loss": 0.0802,
+      "step": 12369
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009310388737012439,
+      "loss": 0.094,
+      "step": 12370
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009308987897432125,
+      "loss": 0.0977,
+      "step": 12371
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009307587071476827,
+      "loss": 0.0851,
+      "step": 12372
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009306186259174164,
+      "loss": 0.0948,
+      "step": 12373
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009304785460551757,
+      "loss": 0.1057,
+      "step": 12374
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000930338467563722,
+      "loss": 0.0544,
+      "step": 12375
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000930198390445818,
+      "loss": 0.0887,
+      "step": 12376
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009300583147042258,
+      "loss": 0.078,
+      "step": 12377
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000929918240341707,
+      "loss": 0.0625,
+      "step": 12378
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009297781673610235,
+      "loss": 0.1693,
+      "step": 12379
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009296380957649371,
+      "loss": 0.0635,
+      "step": 12380
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00092949802555621,
+      "loss": 0.074,
+      "step": 12381
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009293579567376032,
+      "loss": 0.0802,
+      "step": 12382
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009292178893118798,
+      "loss": 0.1876,
+      "step": 12383
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009290778232818007,
+      "loss": 0.1104,
+      "step": 12384
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0009289377586501281,
+      "loss": 0.0938,
+      "step": 12385
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009287976954196231,
+      "loss": 0.067,
+      "step": 12386
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000928657633593048,
+      "loss": 0.0971,
+      "step": 12387
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009285175731731638,
+      "loss": 0.0784,
+      "step": 12388
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000928377514162733,
+      "loss": 0.075,
+      "step": 12389
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009282374565645166,
+      "loss": 0.0641,
+      "step": 12390
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009280974003812765,
+      "loss": 0.1,
+      "step": 12391
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000927957345615774,
+      "loss": 0.0778,
+      "step": 12392
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009278172922707708,
+      "loss": 0.0927,
+      "step": 12393
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009276772403490285,
+      "loss": 0.0974,
+      "step": 12394
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009275371898533079,
+      "loss": 0.0917,
+      "step": 12395
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009273971407863712,
+      "loss": 0.0941,
+      "step": 12396
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009272570931509796,
+      "loss": 0.0515,
+      "step": 12397
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009271170469498946,
+      "loss": 0.1155,
+      "step": 12398
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009269770021858773,
+      "loss": 0.1158,
+      "step": 12399
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009268369588616893,
+      "loss": 0.088,
+      "step": 12400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009266969169800914,
+      "loss": 0.093,
+      "step": 12401
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009265568765438452,
+      "loss": 0.0927,
+      "step": 12402
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009264168375557121,
+      "loss": 0.0785,
+      "step": 12403
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009262768000184533,
+      "loss": 0.0566,
+      "step": 12404
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009261367639348297,
+      "loss": 0.0891,
+      "step": 12405
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009259967293076028,
+      "loss": 0.1069,
+      "step": 12406
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009258566961395332,
+      "loss": 0.1292,
+      "step": 12407
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009257166644333825,
+      "loss": 0.0783,
+      "step": 12408
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009255766341919116,
+      "loss": 0.0671,
+      "step": 12409
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009254366054178817,
+      "loss": 0.1306,
+      "step": 12410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009252965781140535,
+      "loss": 0.1249,
+      "step": 12411
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009251565522831884,
+      "loss": 0.0943,
+      "step": 12412
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009250165279280468,
+      "loss": 0.1014,
+      "step": 12413
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009248765050513902,
+      "loss": 0.1018,
+      "step": 12414
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009247364836559788,
+      "loss": 0.0813,
+      "step": 12415
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009245964637445743,
+      "loss": 0.0568,
+      "step": 12416
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009244564453199371,
+      "loss": 0.069,
+      "step": 12417
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009243164283848279,
+      "loss": 0.111,
+      "step": 12418
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009241764129420078,
+      "loss": 0.0923,
+      "step": 12419
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009240363989942375,
+      "loss": 0.0549,
+      "step": 12420
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009238963865442775,
+      "loss": 0.1082,
+      "step": 12421
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009237563755948883,
+      "loss": 0.0867,
+      "step": 12422
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009236163661488311,
+      "loss": 0.0773,
+      "step": 12423
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009234763582088662,
+      "loss": 0.0529,
+      "step": 12424
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009233363517777545,
+      "loss": 0.0679,
+      "step": 12425
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009231963468582564,
+      "loss": 0.12,
+      "step": 12426
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009230563434531323,
+      "loss": 0.1265,
+      "step": 12427
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009229163415651425,
+      "loss": 0.053,
+      "step": 12428
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009227763411970481,
+      "loss": 0.0894,
+      "step": 12429
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009226363423516092,
+      "loss": 0.0822,
+      "step": 12430
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009224963450315864,
+      "loss": 0.0995,
+      "step": 12431
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0009223563492397401,
+      "loss": 0.0743,
+      "step": 12432
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009222163549788303,
+      "loss": 0.0938,
+      "step": 12433
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009220763622516178,
+      "loss": 0.0874,
+      "step": 12434
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000921936371060862,
+      "loss": 0.0804,
+      "step": 12435
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009217963814093245,
+      "loss": 0.056,
+      "step": 12436
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009216563932997649,
+      "loss": 0.0611,
+      "step": 12437
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009215164067349434,
+      "loss": 0.1241,
+      "step": 12438
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00092137642171762,
+      "loss": 0.1215,
+      "step": 12439
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009212364382505553,
+      "loss": 0.1125,
+      "step": 12440
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000921096456336509,
+      "loss": 0.0746,
+      "step": 12441
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000920956475978241,
+      "loss": 0.0856,
+      "step": 12442
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009208164971785121,
+      "loss": 0.101,
+      "step": 12443
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000920676519940082,
+      "loss": 0.0673,
+      "step": 12444
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009205365442657104,
+      "loss": 0.1018,
+      "step": 12445
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009203965701581579,
+      "loss": 0.0889,
+      "step": 12446
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009202565976201837,
+      "loss": 0.1133,
+      "step": 12447
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009201166266545479,
+      "loss": 0.0828,
+      "step": 12448
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009199766572640107,
+      "loss": 0.0709,
+      "step": 12449
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009198366894513317,
+      "loss": 0.1199,
+      "step": 12450
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000919696723219271,
+      "loss": 0.0788,
+      "step": 12451
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009195567585705882,
+      "loss": 0.0837,
+      "step": 12452
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009194167955080428,
+      "loss": 0.07,
+      "step": 12453
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009192768340343946,
+      "loss": 0.1042,
+      "step": 12454
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009191368741524035,
+      "loss": 0.0946,
+      "step": 12455
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000918996915864829,
+      "loss": 0.0876,
+      "step": 12456
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000918856959174431,
+      "loss": 0.0901,
+      "step": 12457
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009187170040839689,
+      "loss": 0.1428,
+      "step": 12458
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009185770505962022,
+      "loss": 0.1481,
+      "step": 12459
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009184370987138904,
+      "loss": 0.0982,
+      "step": 12460
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009182971484397931,
+      "loss": 0.0917,
+      "step": 12461
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009181571997766693,
+      "loss": 0.0583,
+      "step": 12462
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009180172527272795,
+      "loss": 0.0989,
+      "step": 12463
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009178773072943823,
+      "loss": 0.1155,
+      "step": 12464
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009177373634807373,
+      "loss": 0.0935,
+      "step": 12465
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009175974212891035,
+      "loss": 0.1386,
+      "step": 12466
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009174574807222408,
+      "loss": 0.0897,
+      "step": 12467
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009173175417829077,
+      "loss": 0.1186,
+      "step": 12468
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009171776044738643,
+      "loss": 0.0621,
+      "step": 12469
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009170376687978694,
+      "loss": 0.0886,
+      "step": 12470
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009168977347576822,
+      "loss": 0.1033,
+      "step": 12471
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009167578023560617,
+      "loss": 0.1049,
+      "step": 12472
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009166178715957673,
+      "loss": 0.0924,
+      "step": 12473
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009164779424795581,
+      "loss": 0.1086,
+      "step": 12474
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009163380150101924,
+      "loss": 0.1439,
+      "step": 12475
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009161980891904302,
+      "loss": 0.0942,
+      "step": 12476
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009160581650230299,
+      "loss": 0.0981,
+      "step": 12477
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0009159182425107509,
+      "loss": 0.1292,
+      "step": 12478
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009157783216563517,
+      "loss": 0.1704,
+      "step": 12479
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009156384024625916,
+      "loss": 0.0755,
+      "step": 12480
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009154984849322288,
+      "loss": 0.0677,
+      "step": 12481
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009153585690680225,
+      "loss": 0.0847,
+      "step": 12482
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009152186548727315,
+      "loss": 0.1448,
+      "step": 12483
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000915078742349115,
+      "loss": 0.1313,
+      "step": 12484
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000914938831499931,
+      "loss": 0.1173,
+      "step": 12485
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009147989223279386,
+      "loss": 0.081,
+      "step": 12486
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009146590148358959,
+      "loss": 0.0719,
+      "step": 12487
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000914519109026562,
+      "loss": 0.1116,
+      "step": 12488
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009143792049026958,
+      "loss": 0.0707,
+      "step": 12489
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009142393024670555,
+      "loss": 0.0955,
+      "step": 12490
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009140994017223995,
+      "loss": 0.1333,
+      "step": 12491
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009139595026714862,
+      "loss": 0.1149,
+      "step": 12492
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009138196053170745,
+      "loss": 0.0812,
+      "step": 12493
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009136797096619226,
+      "loss": 0.103,
+      "step": 12494
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009135398157087884,
+      "loss": 0.0855,
+      "step": 12495
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009133999234604312,
+      "loss": 0.1169,
+      "step": 12496
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009132600329196089,
+      "loss": 0.0714,
+      "step": 12497
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009131201440890794,
+      "loss": 0.0925,
+      "step": 12498
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009129802569716016,
+      "loss": 0.0658,
+      "step": 12499
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009128403715699334,
+      "loss": 0.074,
+      "step": 12500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009127004878868329,
+      "loss": 0.0747,
+      "step": 12501
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009125606059250579,
+      "loss": 0.0878,
+      "step": 12502
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009124207256873676,
+      "loss": 0.0431,
+      "step": 12503
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009122808471765192,
+      "loss": 0.0964,
+      "step": 12504
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009121409703952711,
+      "loss": 0.1278,
+      "step": 12505
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009120010953463812,
+      "loss": 0.0648,
+      "step": 12506
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009118612220326076,
+      "loss": 0.1141,
+      "step": 12507
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009117213504567077,
+      "loss": 0.0735,
+      "step": 12508
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009115814806214401,
+      "loss": 0.0746,
+      "step": 12509
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009114416125295626,
+      "loss": 0.0886,
+      "step": 12510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000911301746183833,
+      "loss": 0.094,
+      "step": 12511
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009111618815870089,
+      "loss": 0.0578,
+      "step": 12512
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000911022018741848,
+      "loss": 0.0989,
+      "step": 12513
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009108821576511083,
+      "loss": 0.0857,
+      "step": 12514
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009107422983175473,
+      "loss": 0.0599,
+      "step": 12515
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009106024407439231,
+      "loss": 0.0804,
+      "step": 12516
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009104625849329929,
+      "loss": 0.0851,
+      "step": 12517
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009103227308875147,
+      "loss": 0.1117,
+      "step": 12518
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009101828786102454,
+      "loss": 0.0789,
+      "step": 12519
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009100430281039432,
+      "loss": 0.1104,
+      "step": 12520
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009099031793713653,
+      "loss": 0.1036,
+      "step": 12521
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009097633324152689,
+      "loss": 0.1084,
+      "step": 12522
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009096234872384122,
+      "loss": 0.0903,
+      "step": 12523
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0009094836438435519,
+      "loss": 0.084,
+      "step": 12524
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009093438022334456,
+      "loss": 0.0767,
+      "step": 12525
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009092039624108508,
+      "loss": 0.0786,
+      "step": 12526
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009090641243785246,
+      "loss": 0.0803,
+      "step": 12527
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009089242881392238,
+      "loss": 0.0836,
+      "step": 12528
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009087844536957064,
+      "loss": 0.117,
+      "step": 12529
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009086446210507292,
+      "loss": 0.0786,
+      "step": 12530
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009085047902070496,
+      "loss": 0.0573,
+      "step": 12531
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009083649611674243,
+      "loss": 0.0568,
+      "step": 12532
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009082251339346108,
+      "loss": 0.1201,
+      "step": 12533
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009080853085113657,
+      "loss": 0.0818,
+      "step": 12534
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009079454849004462,
+      "loss": 0.0983,
+      "step": 12535
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009078056631046093,
+      "loss": 0.0742,
+      "step": 12536
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009076658431266122,
+      "loss": 0.0945,
+      "step": 12537
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009075260249692115,
+      "loss": 0.0769,
+      "step": 12538
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009073862086351641,
+      "loss": 0.1239,
+      "step": 12539
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009072463941272265,
+      "loss": 0.0816,
+      "step": 12540
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009071065814481562,
+      "loss": 0.0968,
+      "step": 12541
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009069667706007089,
+      "loss": 0.0978,
+      "step": 12542
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009068269615876426,
+      "loss": 0.0931,
+      "step": 12543
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009066871544117133,
+      "loss": 0.0625,
+      "step": 12544
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009065473490756776,
+      "loss": 0.1471,
+      "step": 12545
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009064075455822924,
+      "loss": 0.0648,
+      "step": 12546
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009062677439343139,
+      "loss": 0.0834,
+      "step": 12547
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009061279441344985,
+      "loss": 0.0786,
+      "step": 12548
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009059881461856035,
+      "loss": 0.1346,
+      "step": 12549
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009058483500903846,
+      "loss": 0.0947,
+      "step": 12550
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009057085558515987,
+      "loss": 0.0696,
+      "step": 12551
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009055687634720019,
+      "loss": 0.0864,
+      "step": 12552
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009054289729543508,
+      "loss": 0.0825,
+      "step": 12553
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009052891843014015,
+      "loss": 0.0811,
+      "step": 12554
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00090514939751591,
+      "loss": 0.0991,
+      "step": 12555
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009050096126006332,
+      "loss": 0.0844,
+      "step": 12556
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009048698295583269,
+      "loss": 0.0903,
+      "step": 12557
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009047300483917474,
+      "loss": 0.1312,
+      "step": 12558
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000904590269103651,
+      "loss": 0.0856,
+      "step": 12559
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009044504916967935,
+      "loss": 0.119,
+      "step": 12560
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009043107161739308,
+      "loss": 0.0749,
+      "step": 12561
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000904170942537819,
+      "loss": 0.1136,
+      "step": 12562
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009040311707912145,
+      "loss": 0.1249,
+      "step": 12563
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009038914009368732,
+      "loss": 0.0951,
+      "step": 12564
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009037516329775509,
+      "loss": 0.0964,
+      "step": 12565
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009036118669160029,
+      "loss": 0.0706,
+      "step": 12566
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009034721027549857,
+      "loss": 0.0828,
+      "step": 12567
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009033323404972547,
+      "loss": 0.0992,
+      "step": 12568
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009031925801455662,
+      "loss": 0.1002,
+      "step": 12569
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0009030528217026754,
+      "loss": 0.0901,
+      "step": 12570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009029130651713385,
+      "loss": 0.1211,
+      "step": 12571
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009027733105543103,
+      "loss": 0.0806,
+      "step": 12572
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009026335578543473,
+      "loss": 0.0825,
+      "step": 12573
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009024938070742046,
+      "loss": 0.275,
+      "step": 12574
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009023540582166375,
+      "loss": 0.0666,
+      "step": 12575
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009022143112844023,
+      "loss": 0.1088,
+      "step": 12576
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009020745662802537,
+      "loss": 0.0985,
+      "step": 12577
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009019348232069474,
+      "loss": 0.0798,
+      "step": 12578
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009017950820672389,
+      "loss": 0.0781,
+      "step": 12579
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009016553428638835,
+      "loss": 0.0892,
+      "step": 12580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000901515605599636,
+      "loss": 0.0735,
+      "step": 12581
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009013758702772524,
+      "loss": 0.0995,
+      "step": 12582
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009012361368994874,
+      "loss": 0.1163,
+      "step": 12583
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009010964054690968,
+      "loss": 0.0967,
+      "step": 12584
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009009566759888353,
+      "loss": 0.0905,
+      "step": 12585
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000900816948461458,
+      "loss": 0.0634,
+      "step": 12586
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009006772228897197,
+      "loss": 0.1001,
+      "step": 12587
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000900537499276376,
+      "loss": 0.054,
+      "step": 12588
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009003977776241816,
+      "loss": 0.1271,
+      "step": 12589
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0009002580579358917,
+      "loss": 0.0831,
+      "step": 12590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000900118340214261,
+      "loss": 0.097,
+      "step": 12591
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008999786244620446,
+      "loss": 0.1055,
+      "step": 12592
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008998389106819969,
+      "loss": 0.0796,
+      "step": 12593
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008996991988768731,
+      "loss": 0.1117,
+      "step": 12594
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008995594890494276,
+      "loss": 0.0856,
+      "step": 12595
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008994197812024156,
+      "loss": 0.0858,
+      "step": 12596
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008992800753385916,
+      "loss": 0.0919,
+      "step": 12597
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008991403714607103,
+      "loss": 0.1234,
+      "step": 12598
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008990006695715258,
+      "loss": 0.075,
+      "step": 12599
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008988609696737935,
+      "loss": 0.0723,
+      "step": 12600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000898721271770267,
+      "loss": 0.1468,
+      "step": 12601
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008985815758637016,
+      "loss": 0.0942,
+      "step": 12602
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008984418819568517,
+      "loss": 0.1017,
+      "step": 12603
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008983021900524711,
+      "loss": 0.0928,
+      "step": 12604
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008981625001533146,
+      "loss": 0.1061,
+      "step": 12605
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008980228122621366,
+      "loss": 0.1134,
+      "step": 12606
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008978831263816913,
+      "loss": 0.0798,
+      "step": 12607
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008977434425147324,
+      "loss": 0.089,
+      "step": 12608
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000897603760664015,
+      "loss": 0.0911,
+      "step": 12609
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008974640808322928,
+      "loss": 0.0674,
+      "step": 12610
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008973244030223202,
+      "loss": 0.0946,
+      "step": 12611
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008971847272368511,
+      "loss": 0.066,
+      "step": 12612
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008970450534786397,
+      "loss": 0.0695,
+      "step": 12613
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008969053817504395,
+      "loss": 0.0906,
+      "step": 12614
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008967657120550048,
+      "loss": 0.1184,
+      "step": 12615
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0008966260443950898,
+      "loss": 0.0822,
+      "step": 12616
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008964863787734483,
+      "loss": 0.0675,
+      "step": 12617
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008963467151928339,
+      "loss": 0.078,
+      "step": 12618
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008962070536560009,
+      "loss": 0.1381,
+      "step": 12619
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008960673941657023,
+      "loss": 0.0634,
+      "step": 12620
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008959277367246919,
+      "loss": 0.0902,
+      "step": 12621
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008957880813357244,
+      "loss": 0.0641,
+      "step": 12622
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008956484280015528,
+      "loss": 0.0703,
+      "step": 12623
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008955087767249307,
+      "loss": 0.1139,
+      "step": 12624
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008953691275086113,
+      "loss": 0.0551,
+      "step": 12625
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008952294803553489,
+      "loss": 0.1133,
+      "step": 12626
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008950898352678967,
+      "loss": 0.0616,
+      "step": 12627
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008949501922490074,
+      "loss": 0.0839,
+      "step": 12628
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008948105513014356,
+      "loss": 0.0726,
+      "step": 12629
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008946709124279341,
+      "loss": 0.1049,
+      "step": 12630
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000894531275631256,
+      "loss": 0.1694,
+      "step": 12631
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008943916409141551,
+      "loss": 0.1155,
+      "step": 12632
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008942520082793845,
+      "loss": 0.1014,
+      "step": 12633
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008941123777296971,
+      "loss": 0.1213,
+      "step": 12634
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008939727492678458,
+      "loss": 0.1068,
+      "step": 12635
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008938331228965845,
+      "loss": 0.0696,
+      "step": 12636
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008936934986186662,
+      "loss": 0.1054,
+      "step": 12637
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008935538764368436,
+      "loss": 0.116,
+      "step": 12638
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008934142563538698,
+      "loss": 0.0744,
+      "step": 12639
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008932746383724975,
+      "loss": 0.0674,
+      "step": 12640
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00089313502249548,
+      "loss": 0.0812,
+      "step": 12641
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008929954087255698,
+      "loss": 0.0661,
+      "step": 12642
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008928557970655203,
+      "loss": 0.1077,
+      "step": 12643
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000892716187518084,
+      "loss": 0.0908,
+      "step": 12644
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008925765800860134,
+      "loss": 0.1005,
+      "step": 12645
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008924369747720613,
+      "loss": 0.0949,
+      "step": 12646
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008922973715789807,
+      "loss": 0.1049,
+      "step": 12647
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008921577705095236,
+      "loss": 0.0889,
+      "step": 12648
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008920181715664433,
+      "loss": 0.0963,
+      "step": 12649
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000891878574752492,
+      "loss": 0.08,
+      "step": 12650
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008917389800704222,
+      "loss": 0.0696,
+      "step": 12651
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008915993875229862,
+      "loss": 0.0663,
+      "step": 12652
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008914597971129368,
+      "loss": 0.1077,
+      "step": 12653
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008913202088430261,
+      "loss": 0.0683,
+      "step": 12654
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008911806227160059,
+      "loss": 0.0754,
+      "step": 12655
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008910410387346297,
+      "loss": 0.0983,
+      "step": 12656
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008909014569016488,
+      "loss": 0.0721,
+      "step": 12657
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008907618772198159,
+      "loss": 0.1272,
+      "step": 12658
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008906222996918831,
+      "loss": 0.1149,
+      "step": 12659
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008904827243206021,
+      "loss": 0.0919,
+      "step": 12660
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000890343151108725,
+      "loss": 0.0957,
+      "step": 12661
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0008902035800590045,
+      "loss": 0.0967,
+      "step": 12662
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000890064011174192,
+      "loss": 0.1378,
+      "step": 12663
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008899244444570397,
+      "loss": 0.0917,
+      "step": 12664
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008897848799102995,
+      "loss": 0.09,
+      "step": 12665
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008896453175367232,
+      "loss": 0.1371,
+      "step": 12666
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008895057573390623,
+      "loss": 0.0824,
+      "step": 12667
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008893661993200689,
+      "loss": 0.0881,
+      "step": 12668
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008892266434824947,
+      "loss": 0.1194,
+      "step": 12669
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008890870898290916,
+      "loss": 0.1125,
+      "step": 12670
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008889475383626111,
+      "loss": 0.0791,
+      "step": 12671
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008888079890858046,
+      "loss": 0.0774,
+      "step": 12672
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008886684420014238,
+      "loss": 0.0789,
+      "step": 12673
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008885288971122202,
+      "loss": 0.0762,
+      "step": 12674
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008883893544209449,
+      "loss": 0.1024,
+      "step": 12675
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008882498139303503,
+      "loss": 0.0866,
+      "step": 12676
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008881102756431871,
+      "loss": 0.0981,
+      "step": 12677
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008879707395622066,
+      "loss": 0.1245,
+      "step": 12678
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008878312056901603,
+      "loss": 0.1513,
+      "step": 12679
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008876916740297996,
+      "loss": 0.0876,
+      "step": 12680
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008875521445838749,
+      "loss": 0.1517,
+      "step": 12681
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008874126173551384,
+      "loss": 0.054,
+      "step": 12682
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000887273092346341,
+      "loss": 0.0754,
+      "step": 12683
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008871335695602333,
+      "loss": 0.0874,
+      "step": 12684
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008869940489995667,
+      "loss": 0.0741,
+      "step": 12685
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008868545306670924,
+      "loss": 0.085,
+      "step": 12686
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008867150145655608,
+      "loss": 0.1405,
+      "step": 12687
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008865755006977228,
+      "loss": 0.0661,
+      "step": 12688
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008864359890663298,
+      "loss": 0.1365,
+      "step": 12689
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008862964796741322,
+      "loss": 0.1407,
+      "step": 12690
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008861569725238811,
+      "loss": 0.0847,
+      "step": 12691
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008860174676183272,
+      "loss": 0.1276,
+      "step": 12692
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000885877964960221,
+      "loss": 0.1166,
+      "step": 12693
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008857384645523129,
+      "loss": 0.0609,
+      "step": 12694
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008855989663973534,
+      "loss": 0.078,
+      "step": 12695
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008854594704980942,
+      "loss": 0.1058,
+      "step": 12696
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008853199768572848,
+      "loss": 0.1046,
+      "step": 12697
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008851804854776759,
+      "loss": 0.0745,
+      "step": 12698
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008850409963620178,
+      "loss": 0.1069,
+      "step": 12699
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008849015095130612,
+      "loss": 0.0852,
+      "step": 12700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008847620249335556,
+      "loss": 0.1183,
+      "step": 12701
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008846225426262524,
+      "loss": 0.0789,
+      "step": 12702
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008844830625939015,
+      "loss": 0.0979,
+      "step": 12703
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008843435848392527,
+      "loss": 0.0737,
+      "step": 12704
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008842041093650564,
+      "loss": 0.0901,
+      "step": 12705
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008840646361740627,
+      "loss": 0.1442,
+      "step": 12706
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008839251652690217,
+      "loss": 0.0721,
+      "step": 12707
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0008837856966526829,
+      "loss": 0.1257,
+      "step": 12708
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008836462303277972,
+      "loss": 0.0993,
+      "step": 12709
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000883506766297114,
+      "loss": 0.0837,
+      "step": 12710
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008833673045633831,
+      "loss": 0.0721,
+      "step": 12711
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008832278451293546,
+      "loss": 0.0769,
+      "step": 12712
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008830883879977781,
+      "loss": 0.0745,
+      "step": 12713
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008829489331714036,
+      "loss": 0.075,
+      "step": 12714
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008828094806529799,
+      "loss": 0.1129,
+      "step": 12715
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008826700304452578,
+      "loss": 0.1107,
+      "step": 12716
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008825305825509864,
+      "loss": 0.0743,
+      "step": 12717
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008823911369729154,
+      "loss": 0.1216,
+      "step": 12718
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008822516937137943,
+      "loss": 0.1121,
+      "step": 12719
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008821122527763722,
+      "loss": 0.1143,
+      "step": 12720
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008819728141633988,
+      "loss": 0.106,
+      "step": 12721
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008818333778776236,
+      "loss": 0.0992,
+      "step": 12722
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008816939439217959,
+      "loss": 0.0877,
+      "step": 12723
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000881554512298665,
+      "loss": 0.1161,
+      "step": 12724
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008814150830109799,
+      "loss": 0.0746,
+      "step": 12725
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00088127565606149,
+      "loss": 0.0897,
+      "step": 12726
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008811362314529444,
+      "loss": 0.0781,
+      "step": 12727
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008809968091880917,
+      "loss": 0.0998,
+      "step": 12728
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008808573892696821,
+      "loss": 0.0857,
+      "step": 12729
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008807179717004637,
+      "loss": 0.131,
+      "step": 12730
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008805785564831856,
+      "loss": 0.0861,
+      "step": 12731
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008804391436205972,
+      "loss": 0.1057,
+      "step": 12732
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008802997331154468,
+      "loss": 0.0922,
+      "step": 12733
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008801603249704835,
+      "loss": 0.0864,
+      "step": 12734
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008800209191884555,
+      "loss": 0.08,
+      "step": 12735
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008798815157721125,
+      "loss": 0.0688,
+      "step": 12736
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008797421147242026,
+      "loss": 0.1125,
+      "step": 12737
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008796027160474746,
+      "loss": 0.1115,
+      "step": 12738
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008794633197446771,
+      "loss": 0.0771,
+      "step": 12739
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008793239258185585,
+      "loss": 0.0867,
+      "step": 12740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008791845342718672,
+      "loss": 0.0737,
+      "step": 12741
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008790451451073521,
+      "loss": 0.0764,
+      "step": 12742
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008789057583277611,
+      "loss": 0.1028,
+      "step": 12743
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000878766373935843,
+      "loss": 0.0732,
+      "step": 12744
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008786269919343459,
+      "loss": 0.1019,
+      "step": 12745
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008784876123260182,
+      "loss": 0.0872,
+      "step": 12746
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008783482351136076,
+      "loss": 0.1123,
+      "step": 12747
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008782088602998624,
+      "loss": 0.1113,
+      "step": 12748
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008780694878875314,
+      "loss": 0.0706,
+      "step": 12749
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008779301178793623,
+      "loss": 0.1216,
+      "step": 12750
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008777907502781029,
+      "loss": 0.056,
+      "step": 12751
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008776513850865013,
+      "loss": 0.1376,
+      "step": 12752
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008775120223073055,
+      "loss": 0.0724,
+      "step": 12753
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0008773726619432635,
+      "loss": 0.1133,
+      "step": 12754
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008772333039971226,
+      "loss": 0.0695,
+      "step": 12755
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008770939484716311,
+      "loss": 0.0746,
+      "step": 12756
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008769545953695368,
+      "loss": 0.1018,
+      "step": 12757
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000876815244693587,
+      "loss": 0.0953,
+      "step": 12758
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008766758964465297,
+      "loss": 0.0812,
+      "step": 12759
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008765365506311123,
+      "loss": 0.0925,
+      "step": 12760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008763972072500821,
+      "loss": 0.0821,
+      "step": 12761
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008762578663061871,
+      "loss": 0.0627,
+      "step": 12762
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008761185278021747,
+      "loss": 0.0579,
+      "step": 12763
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008759791917407918,
+      "loss": 0.1487,
+      "step": 12764
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008758398581247864,
+      "loss": 0.0878,
+      "step": 12765
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008757005269569054,
+      "loss": 0.1417,
+      "step": 12766
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008755611982398962,
+      "loss": 0.0685,
+      "step": 12767
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008754218719765056,
+      "loss": 0.0884,
+      "step": 12768
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008752825481694813,
+      "loss": 0.0765,
+      "step": 12769
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008751432268215703,
+      "loss": 0.0721,
+      "step": 12770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008750039079355199,
+      "loss": 0.095,
+      "step": 12771
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008748645915140766,
+      "loss": 0.1108,
+      "step": 12772
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008747252775599874,
+      "loss": 0.0934,
+      "step": 12773
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008745859660759997,
+      "loss": 0.0797,
+      "step": 12774
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008744466570648596,
+      "loss": 0.0585,
+      "step": 12775
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008743073505293148,
+      "loss": 0.0894,
+      "step": 12776
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008741680464721118,
+      "loss": 0.1135,
+      "step": 12777
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008740287448959971,
+      "loss": 0.0812,
+      "step": 12778
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008738894458037173,
+      "loss": 0.1202,
+      "step": 12779
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008737501491980195,
+      "loss": 0.1085,
+      "step": 12780
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008736108550816494,
+      "loss": 0.0886,
+      "step": 12781
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008734715634573547,
+      "loss": 0.0944,
+      "step": 12782
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000873332274327881,
+      "loss": 0.1212,
+      "step": 12783
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008731929876959753,
+      "loss": 0.104,
+      "step": 12784
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008730537035643834,
+      "loss": 0.0968,
+      "step": 12785
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008729144219358522,
+      "loss": 0.0787,
+      "step": 12786
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008727751428131276,
+      "loss": 0.1541,
+      "step": 12787
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008726358661989555,
+      "loss": 0.084,
+      "step": 12788
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008724965920960832,
+      "loss": 0.0764,
+      "step": 12789
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008723573205072556,
+      "loss": 0.0728,
+      "step": 12790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008722180514352198,
+      "loss": 0.111,
+      "step": 12791
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008720787848827213,
+      "loss": 0.0792,
+      "step": 12792
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008719395208525062,
+      "loss": 0.1117,
+      "step": 12793
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008718002593473201,
+      "loss": 0.076,
+      "step": 12794
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008716610003699092,
+      "loss": 0.0934,
+      "step": 12795
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008715217439230194,
+      "loss": 0.11,
+      "step": 12796
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008713824900093965,
+      "loss": 0.1094,
+      "step": 12797
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008712432386317864,
+      "loss": 0.1077,
+      "step": 12798
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008711039897929343,
+      "loss": 0.0917,
+      "step": 12799
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008709647434955858,
+      "loss": 0.101,
+      "step": 12800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0008708254997424869,
+      "loss": 0.1311,
+      "step": 12801
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008706862585363828,
+      "loss": 0.0908,
+      "step": 12802
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008705470198800196,
+      "loss": 0.1511,
+      "step": 12803
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008704077837761422,
+      "loss": 0.1338,
+      "step": 12804
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000870268550227496,
+      "loss": 0.082,
+      "step": 12805
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008701293192368264,
+      "loss": 0.0847,
+      "step": 12806
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008699900908068788,
+      "loss": 0.053,
+      "step": 12807
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008698508649403979,
+      "loss": 0.0781,
+      "step": 12808
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008697116416401298,
+      "loss": 0.1477,
+      "step": 12809
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008695724209088192,
+      "loss": 0.0749,
+      "step": 12810
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008694332027492108,
+      "loss": 0.0858,
+      "step": 12811
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008692939871640502,
+      "loss": 0.0844,
+      "step": 12812
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008691547741560821,
+      "loss": 0.1013,
+      "step": 12813
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008690155637280517,
+      "loss": 0.0511,
+      "step": 12814
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000868876355882703,
+      "loss": 0.1115,
+      "step": 12815
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008687371506227819,
+      "loss": 0.0854,
+      "step": 12816
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008685979479510326,
+      "loss": 0.1031,
+      "step": 12817
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008684587478702002,
+      "loss": 0.1321,
+      "step": 12818
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008683195503830292,
+      "loss": 0.0656,
+      "step": 12819
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008681803554922642,
+      "loss": 0.0693,
+      "step": 12820
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008680411632006492,
+      "loss": 0.13,
+      "step": 12821
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008679019735109298,
+      "loss": 0.1303,
+      "step": 12822
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008677627864258497,
+      "loss": 0.1269,
+      "step": 12823
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008676236019481537,
+      "loss": 0.0849,
+      "step": 12824
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008674844200805861,
+      "loss": 0.0941,
+      "step": 12825
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008673452408258909,
+      "loss": 0.0717,
+      "step": 12826
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008672060641868128,
+      "loss": 0.0807,
+      "step": 12827
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008670668901660955,
+      "loss": 0.113,
+      "step": 12828
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008669277187664837,
+      "loss": 0.0952,
+      "step": 12829
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008667885499907213,
+      "loss": 0.0771,
+      "step": 12830
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008666493838415524,
+      "loss": 0.1003,
+      "step": 12831
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008665102203217207,
+      "loss": 0.0641,
+      "step": 12832
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008663710594339708,
+      "loss": 0.1373,
+      "step": 12833
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.000866231901181046,
+      "loss": 0.0791,
+      "step": 12834
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00086609274556569,
+      "loss": 0.1421,
+      "step": 12835
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008659535925906473,
+      "loss": 0.1229,
+      "step": 12836
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008658144422586613,
+      "loss": 0.095,
+      "step": 12837
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008656752945724756,
+      "loss": 0.1282,
+      "step": 12838
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008655361495348339,
+      "loss": 0.0815,
+      "step": 12839
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00086539700714848,
+      "loss": 0.0977,
+      "step": 12840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008652578674161568,
+      "loss": 0.15,
+      "step": 12841
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008651187303406087,
+      "loss": 0.0944,
+      "step": 12842
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008649795959245783,
+      "loss": 0.0968,
+      "step": 12843
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008648404641708097,
+      "loss": 0.0776,
+      "step": 12844
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008647013350820458,
+      "loss": 0.1059,
+      "step": 12845
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008645622086610301,
+      "loss": 0.1055,
+      "step": 12846
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0008644230849105053,
+      "loss": 0.1201,
+      "step": 12847
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008642839638332149,
+      "loss": 0.0988,
+      "step": 12848
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008641448454319021,
+      "loss": 0.1213,
+      "step": 12849
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008640057297093103,
+      "loss": 0.0958,
+      "step": 12850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008638666166681819,
+      "loss": 0.0473,
+      "step": 12851
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008637275063112602,
+      "loss": 0.0928,
+      "step": 12852
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008635883986412877,
+      "loss": 0.1268,
+      "step": 12853
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008634492936610076,
+      "loss": 0.0718,
+      "step": 12854
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008633101913731626,
+      "loss": 0.121,
+      "step": 12855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008631710917804957,
+      "loss": 0.0839,
+      "step": 12856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008630319948857493,
+      "loss": 0.1058,
+      "step": 12857
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008628929006916664,
+      "loss": 0.0824,
+      "step": 12858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008627538092009888,
+      "loss": 0.1124,
+      "step": 12859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008626147204164598,
+      "loss": 0.082,
+      "step": 12860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008624756343408213,
+      "loss": 0.0989,
+      "step": 12861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008623365509768165,
+      "loss": 0.0699,
+      "step": 12862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008621974703271873,
+      "loss": 0.1076,
+      "step": 12863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008620583923946757,
+      "loss": 0.0905,
+      "step": 12864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008619193171820246,
+      "loss": 0.0895,
+      "step": 12865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008617802446919759,
+      "loss": 0.1127,
+      "step": 12866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008616411749272718,
+      "loss": 0.1118,
+      "step": 12867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000861502107890654,
+      "loss": 0.0963,
+      "step": 12868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008613630435848652,
+      "loss": 0.0749,
+      "step": 12869
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000861223982012647,
+      "loss": 0.1294,
+      "step": 12870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008610849231767417,
+      "loss": 0.0989,
+      "step": 12871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000860945867079891,
+      "loss": 0.0812,
+      "step": 12872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008608068137248366,
+      "loss": 0.0817,
+      "step": 12873
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00086066776311432,
+      "loss": 0.0983,
+      "step": 12874
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008605287152510837,
+      "loss": 0.0923,
+      "step": 12875
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008603896701378687,
+      "loss": 0.0718,
+      "step": 12876
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008602506277774173,
+      "loss": 0.0933,
+      "step": 12877
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008601115881724705,
+      "loss": 0.1135,
+      "step": 12878
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00085997255132577,
+      "loss": 0.0718,
+      "step": 12879
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008598335172400569,
+      "loss": 0.1411,
+      "step": 12880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008596944859180729,
+      "loss": 0.1073,
+      "step": 12881
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008595554573625597,
+      "loss": 0.0925,
+      "step": 12882
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008594164315762582,
+      "loss": 0.0803,
+      "step": 12883
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008592774085619099,
+      "loss": 0.0717,
+      "step": 12884
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008591383883222554,
+      "loss": 0.1217,
+      "step": 12885
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008589993708600364,
+      "loss": 0.083,
+      "step": 12886
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008588603561779939,
+      "loss": 0.1372,
+      "step": 12887
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008587213442788682,
+      "loss": 0.0674,
+      "step": 12888
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008585823351654015,
+      "loss": 0.1365,
+      "step": 12889
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000858443328840334,
+      "loss": 0.0927,
+      "step": 12890
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008583043253064064,
+      "loss": 0.1366,
+      "step": 12891
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008581653245663599,
+      "loss": 0.0999,
+      "step": 12892
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0008580263266229352,
+      "loss": 0.0725,
+      "step": 12893
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008578873314788723,
+      "loss": 0.0941,
+      "step": 12894
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000857748339136913,
+      "loss": 0.0977,
+      "step": 12895
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008576093495997971,
+      "loss": 0.116,
+      "step": 12896
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008574703628702651,
+      "loss": 0.1145,
+      "step": 12897
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008573313789510578,
+      "loss": 0.0748,
+      "step": 12898
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008571923978449157,
+      "loss": 0.0848,
+      "step": 12899
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008570534195545787,
+      "loss": 0.1074,
+      "step": 12900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000856914444082787,
+      "loss": 0.0724,
+      "step": 12901
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008567754714322814,
+      "loss": 0.072,
+      "step": 12902
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008566365016058021,
+      "loss": 0.1177,
+      "step": 12903
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008564975346060889,
+      "loss": 0.0698,
+      "step": 12904
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000856358570435882,
+      "loss": 0.1013,
+      "step": 12905
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008562196090979213,
+      "loss": 0.0786,
+      "step": 12906
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000856080650594947,
+      "loss": 0.1345,
+      "step": 12907
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008559416949296986,
+      "loss": 0.0976,
+      "step": 12908
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008558027421049164,
+      "loss": 0.0696,
+      "step": 12909
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008556637921233401,
+      "loss": 0.1123,
+      "step": 12910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008555248449877094,
+      "loss": 0.0896,
+      "step": 12911
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008553859007007638,
+      "loss": 0.1095,
+      "step": 12912
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008552469592652433,
+      "loss": 0.0907,
+      "step": 12913
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008551080206838868,
+      "loss": 0.0692,
+      "step": 12914
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008549690849594347,
+      "loss": 0.0831,
+      "step": 12915
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008548301520946261,
+      "loss": 0.0656,
+      "step": 12916
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008546912220922003,
+      "loss": 0.0981,
+      "step": 12917
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008545522949548965,
+      "loss": 0.0956,
+      "step": 12918
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008544133706854545,
+      "loss": 0.0712,
+      "step": 12919
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008542744492866131,
+      "loss": 0.0885,
+      "step": 12920
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008541355307611112,
+      "loss": 0.1293,
+      "step": 12921
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008539966151116888,
+      "loss": 0.1131,
+      "step": 12922
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008538577023410842,
+      "loss": 0.0682,
+      "step": 12923
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000853718792452037,
+      "loss": 0.0969,
+      "step": 12924
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008535798854472858,
+      "loss": 0.0948,
+      "step": 12925
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008534409813295696,
+      "loss": 0.0762,
+      "step": 12926
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008533020801016271,
+      "loss": 0.0742,
+      "step": 12927
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008531631817661969,
+      "loss": 0.0953,
+      "step": 12928
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008530242863260181,
+      "loss": 0.0825,
+      "step": 12929
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008528853937838296,
+      "loss": 0.1117,
+      "step": 12930
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008527465041423695,
+      "loss": 0.0981,
+      "step": 12931
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008526076174043767,
+      "loss": 0.0885,
+      "step": 12932
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008524687335725892,
+      "loss": 0.1232,
+      "step": 12933
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008523298526497455,
+      "loss": 0.0961,
+      "step": 12934
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008521909746385847,
+      "loss": 0.0656,
+      "step": 12935
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008520520995418447,
+      "loss": 0.1136,
+      "step": 12936
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008519132273622637,
+      "loss": 0.0641,
+      "step": 12937
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008517743581025798,
+      "loss": 0.0764,
+      "step": 12938
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0008516354917655315,
+      "loss": 0.066,
+      "step": 12939
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008514966283538567,
+      "loss": 0.0833,
+      "step": 12940
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008513577678702928,
+      "loss": 0.0679,
+      "step": 12941
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008512189103175791,
+      "loss": 0.1091,
+      "step": 12942
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008510800556984526,
+      "loss": 0.1071,
+      "step": 12943
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008509412040156513,
+      "loss": 0.0891,
+      "step": 12944
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008508023552719133,
+      "loss": 0.1243,
+      "step": 12945
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008506635094699761,
+      "loss": 0.0748,
+      "step": 12946
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008505246666125774,
+      "loss": 0.0923,
+      "step": 12947
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008503858267024545,
+      "loss": 0.1125,
+      "step": 12948
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008502469897423456,
+      "loss": 0.0728,
+      "step": 12949
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008501081557349878,
+      "loss": 0.1636,
+      "step": 12950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008499693246831189,
+      "loss": 0.0967,
+      "step": 12951
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008498304965894762,
+      "loss": 0.0975,
+      "step": 12952
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008496916714567968,
+      "loss": 0.0664,
+      "step": 12953
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008495528492878178,
+      "loss": 0.0684,
+      "step": 12954
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008494140300852769,
+      "loss": 0.1409,
+      "step": 12955
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008492752138519114,
+      "loss": 0.0561,
+      "step": 12956
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000849136400590458,
+      "loss": 0.0818,
+      "step": 12957
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008489975903036537,
+      "loss": 0.1083,
+      "step": 12958
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008488587829942357,
+      "loss": 0.0755,
+      "step": 12959
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008487199786649409,
+      "loss": 0.113,
+      "step": 12960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008485811773185059,
+      "loss": 0.0939,
+      "step": 12961
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008484423789576681,
+      "loss": 0.1144,
+      "step": 12962
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008483035835851638,
+      "loss": 0.1273,
+      "step": 12963
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008481647912037299,
+      "loss": 0.0779,
+      "step": 12964
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008480260018161027,
+      "loss": 0.0899,
+      "step": 12965
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008478872154250193,
+      "loss": 0.1091,
+      "step": 12966
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008477484320332158,
+      "loss": 0.0621,
+      "step": 12967
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008476096516434284,
+      "loss": 0.1432,
+      "step": 12968
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008474708742583944,
+      "loss": 0.0856,
+      "step": 12969
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008473320998808496,
+      "loss": 0.0801,
+      "step": 12970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008471933285135301,
+      "loss": 0.0912,
+      "step": 12971
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008470545601591725,
+      "loss": 0.1426,
+      "step": 12972
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008469157948205127,
+      "loss": 0.0606,
+      "step": 12973
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008467770325002866,
+      "loss": 0.0945,
+      "step": 12974
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000846638273201231,
+      "loss": 0.0901,
+      "step": 12975
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008464995169260811,
+      "loss": 0.1088,
+      "step": 12976
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008463607636775734,
+      "loss": 0.0815,
+      "step": 12977
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008462220134584436,
+      "loss": 0.0809,
+      "step": 12978
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008460832662714272,
+      "loss": 0.0718,
+      "step": 12979
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00084594452211926,
+      "loss": 0.1001,
+      "step": 12980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008458057810046779,
+      "loss": 0.0754,
+      "step": 12981
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008456670429304164,
+      "loss": 0.0848,
+      "step": 12982
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008455283078992115,
+      "loss": 0.1046,
+      "step": 12983
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0008453895759137982,
+      "loss": 0.0701,
+      "step": 12984
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000845250846976912,
+      "loss": 0.1469,
+      "step": 12985
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008451121210912882,
+      "loss": 0.1322,
+      "step": 12986
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008449733982596626,
+      "loss": 0.0675,
+      "step": 12987
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008448346784847696,
+      "loss": 0.0637,
+      "step": 12988
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008446959617693455,
+      "loss": 0.0971,
+      "step": 12989
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008445572481161248,
+      "loss": 0.1456,
+      "step": 12990
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008444185375278426,
+      "loss": 0.0973,
+      "step": 12991
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008442798300072339,
+      "loss": 0.0652,
+      "step": 12992
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000844141125557034,
+      "loss": 0.0927,
+      "step": 12993
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008440024241799772,
+      "loss": 0.1318,
+      "step": 12994
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000843863725878799,
+      "loss": 0.1066,
+      "step": 12995
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000843725030656234,
+      "loss": 0.0978,
+      "step": 12996
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008435863385150164,
+      "loss": 0.1128,
+      "step": 12997
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008434476494578817,
+      "loss": 0.1313,
+      "step": 12998
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008433089634875639,
+      "loss": 0.0823,
+      "step": 12999
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008431702806067977,
+      "loss": 0.0978,
+      "step": 13000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008430316008183172,
+      "loss": 0.0721,
+      "step": 13001
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008428929241248576,
+      "loss": 0.1042,
+      "step": 13002
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008427542505291527,
+      "loss": 0.0825,
+      "step": 13003
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008426155800339371,
+      "loss": 0.0887,
+      "step": 13004
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000842476912641945,
+      "loss": 0.0887,
+      "step": 13005
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008423382483559103,
+      "loss": 0.0828,
+      "step": 13006
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008421995871785669,
+      "loss": 0.0643,
+      "step": 13007
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008420609291126494,
+      "loss": 0.0989,
+      "step": 13008
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008419222741608915,
+      "loss": 0.1089,
+      "step": 13009
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008417836223260274,
+      "loss": 0.0651,
+      "step": 13010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000841644973610791,
+      "loss": 0.0746,
+      "step": 13011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008415063280179156,
+      "loss": 0.0911,
+      "step": 13012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008413676855501351,
+      "loss": 0.1194,
+      "step": 13013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008412290462101831,
+      "loss": 0.0792,
+      "step": 13014
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008410904100007939,
+      "loss": 0.1182,
+      "step": 13015
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008409517769247005,
+      "loss": 0.0954,
+      "step": 13016
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008408131469846365,
+      "loss": 0.123,
+      "step": 13017
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000840674520183335,
+      "loss": 0.0815,
+      "step": 13018
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00084053589652353,
+      "loss": 0.1053,
+      "step": 13019
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008403972760079544,
+      "loss": 0.1073,
+      "step": 13020
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008402586586393412,
+      "loss": 0.0836,
+      "step": 13021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008401200444204242,
+      "loss": 0.0915,
+      "step": 13022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008399814333539363,
+      "loss": 0.1176,
+      "step": 13023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008398428254426102,
+      "loss": 0.127,
+      "step": 13024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008397042206891796,
+      "loss": 0.0854,
+      "step": 13025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008395656190963769,
+      "loss": 0.0799,
+      "step": 13026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008394270206669351,
+      "loss": 0.1445,
+      "step": 13027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008392884254035866,
+      "loss": 0.0901,
+      "step": 13028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008391498333090647,
+      "loss": 0.1631,
+      "step": 13029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008390112443861023,
+      "loss": 0.0922,
+      "step": 13030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0008388726586374315,
+      "loss": 0.1333,
+      "step": 13031
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008387340760657851,
+      "loss": 0.0793,
+      "step": 13032
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008385954966738953,
+      "loss": 0.1072,
+      "step": 13033
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008384569204644949,
+      "loss": 0.0593,
+      "step": 13034
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000838318347440316,
+      "loss": 0.0751,
+      "step": 13035
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008381797776040912,
+      "loss": 0.1145,
+      "step": 13036
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008380412109585527,
+      "loss": 0.088,
+      "step": 13037
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008379026475064327,
+      "loss": 0.0751,
+      "step": 13038
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008377640872504627,
+      "loss": 0.1392,
+      "step": 13039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008376255301933756,
+      "loss": 0.1158,
+      "step": 13040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008374869763379027,
+      "loss": 0.0942,
+      "step": 13041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008373484256867766,
+      "loss": 0.0628,
+      "step": 13042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008372098782427289,
+      "loss": 0.0891,
+      "step": 13043
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008370713340084914,
+      "loss": 0.1005,
+      "step": 13044
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008369327929867956,
+      "loss": 0.1292,
+      "step": 13045
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008367942551803735,
+      "loss": 0.0785,
+      "step": 13046
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008366557205919566,
+      "loss": 0.0863,
+      "step": 13047
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008365171892242761,
+      "loss": 0.0922,
+      "step": 13048
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008363786610800642,
+      "loss": 0.125,
+      "step": 13049
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008362401361620517,
+      "loss": 0.0638,
+      "step": 13050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008361016144729707,
+      "loss": 0.1074,
+      "step": 13051
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008359630960155517,
+      "loss": 0.069,
+      "step": 13052
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008358245807925265,
+      "loss": 0.0692,
+      "step": 13053
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008356860688066254,
+      "loss": 0.0875,
+      "step": 13054
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008355475600605808,
+      "loss": 0.0815,
+      "step": 13055
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008354090545571227,
+      "loss": 0.0836,
+      "step": 13056
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008352705522989827,
+      "loss": 0.0948,
+      "step": 13057
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008351320532888916,
+      "loss": 0.0916,
+      "step": 13058
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008349935575295798,
+      "loss": 0.0544,
+      "step": 13059
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008348550650237785,
+      "loss": 0.0916,
+      "step": 13060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008347165757742181,
+      "loss": 0.1223,
+      "step": 13061
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008345780897836297,
+      "loss": 0.0636,
+      "step": 13062
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008344396070547438,
+      "loss": 0.124,
+      "step": 13063
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008343011275902907,
+      "loss": 0.0633,
+      "step": 13064
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008341626513930012,
+      "loss": 0.1362,
+      "step": 13065
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008340241784656052,
+      "loss": 0.0576,
+      "step": 13066
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008338857088108335,
+      "loss": 0.0719,
+      "step": 13067
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008337472424314157,
+      "loss": 0.0752,
+      "step": 13068
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008336087793300829,
+      "loss": 0.1666,
+      "step": 13069
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000833470319509565,
+      "loss": 0.0873,
+      "step": 13070
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008333318629725916,
+      "loss": 0.0716,
+      "step": 13071
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008331934097218933,
+      "loss": 0.0751,
+      "step": 13072
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008330549597601998,
+      "loss": 0.0903,
+      "step": 13073
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008329165130902403,
+      "loss": 0.0705,
+      "step": 13074
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008327780697147459,
+      "loss": 0.0895,
+      "step": 13075
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008326396296364458,
+      "loss": 0.1069,
+      "step": 13076
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0008325011928580693,
+      "loss": 0.0716,
+      "step": 13077
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008323627593823465,
+      "loss": 0.1379,
+      "step": 13078
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008322243292120069,
+      "loss": 0.1522,
+      "step": 13079
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00083208590234978,
+      "loss": 0.1095,
+      "step": 13080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008319474787983945,
+      "loss": 0.0926,
+      "step": 13081
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008318090585605808,
+      "loss": 0.173,
+      "step": 13082
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008316706416390678,
+      "loss": 0.0846,
+      "step": 13083
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008315322280365848,
+      "loss": 0.1104,
+      "step": 13084
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008313938177558609,
+      "loss": 0.0605,
+      "step": 13085
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008312554107996251,
+      "loss": 0.0865,
+      "step": 13086
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008311170071706065,
+      "loss": 0.1382,
+      "step": 13087
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008309786068715337,
+      "loss": 0.0834,
+      "step": 13088
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008308402099051366,
+      "loss": 0.0826,
+      "step": 13089
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008307018162741433,
+      "loss": 0.1031,
+      "step": 13090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008305634259812827,
+      "loss": 0.0667,
+      "step": 13091
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008304250390292834,
+      "loss": 0.0892,
+      "step": 13092
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008302866554208742,
+      "loss": 0.0948,
+      "step": 13093
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008301482751587834,
+      "loss": 0.0644,
+      "step": 13094
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00083000989824574,
+      "loss": 0.1041,
+      "step": 13095
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008298715246844723,
+      "loss": 0.0827,
+      "step": 13096
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008297331544777085,
+      "loss": 0.0623,
+      "step": 13097
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008295947876281769,
+      "loss": 0.1012,
+      "step": 13098
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008294564241386059,
+      "loss": 0.0851,
+      "step": 13099
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008293180640117238,
+      "loss": 0.0936,
+      "step": 13100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008291797072502578,
+      "loss": 0.0949,
+      "step": 13101
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008290413538569371,
+      "loss": 0.1001,
+      "step": 13102
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008289030038344893,
+      "loss": 0.0915,
+      "step": 13103
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008287646571856422,
+      "loss": 0.0685,
+      "step": 13104
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008286263139131239,
+      "loss": 0.1196,
+      "step": 13105
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008284879740196617,
+      "loss": 0.0739,
+      "step": 13106
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008283496375079837,
+      "loss": 0.0818,
+      "step": 13107
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000828211304380817,
+      "loss": 0.1057,
+      "step": 13108
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008280729746408899,
+      "loss": 0.0807,
+      "step": 13109
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00082793464829093,
+      "loss": 0.1237,
+      "step": 13110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008277963253336641,
+      "loss": 0.0879,
+      "step": 13111
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00082765800577182,
+      "loss": 0.1188,
+      "step": 13112
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008275196896081245,
+      "loss": 0.1328,
+      "step": 13113
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008273813768453054,
+      "loss": 0.0894,
+      "step": 13114
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008272430674860896,
+      "loss": 0.097,
+      "step": 13115
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008271047615332045,
+      "loss": 0.1688,
+      "step": 13116
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008269664589893769,
+      "loss": 0.0837,
+      "step": 13117
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008268281598573339,
+      "loss": 0.0964,
+      "step": 13118
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008266898641398022,
+      "loss": 0.054,
+      "step": 13119
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008265515718395089,
+      "loss": 0.1587,
+      "step": 13120
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008264132829591802,
+      "loss": 0.064,
+      "step": 13121
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008262749975015438,
+      "loss": 0.1201,
+      "step": 13122
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0008261367154693256,
+      "loss": 0.0862,
+      "step": 13123
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008259984368652522,
+      "loss": 0.1258,
+      "step": 13124
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008258601616920507,
+      "loss": 0.064,
+      "step": 13125
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008257218899524469,
+      "loss": 0.1135,
+      "step": 13126
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000825583621649167,
+      "loss": 0.0718,
+      "step": 13127
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008254453567849381,
+      "loss": 0.1464,
+      "step": 13128
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008253070953624862,
+      "loss": 0.1123,
+      "step": 13129
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008251688373845369,
+      "loss": 0.0909,
+      "step": 13130
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008250305828538168,
+      "loss": 0.0641,
+      "step": 13131
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000824892331773052,
+      "loss": 0.0907,
+      "step": 13132
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008247540841449683,
+      "loss": 0.0894,
+      "step": 13133
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000824615839972291,
+      "loss": 0.097,
+      "step": 13134
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008244775992577469,
+      "loss": 0.0645,
+      "step": 13135
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008243393620040613,
+      "loss": 0.0673,
+      "step": 13136
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00082420112821396,
+      "loss": 0.103,
+      "step": 13137
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008240628978901687,
+      "loss": 0.0671,
+      "step": 13138
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008239246710354126,
+      "loss": 0.0685,
+      "step": 13139
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008237864476524174,
+      "loss": 0.0571,
+      "step": 13140
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008236482277439082,
+      "loss": 0.0634,
+      "step": 13141
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008235100113126112,
+      "loss": 0.1229,
+      "step": 13142
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000823371798361251,
+      "loss": 0.1066,
+      "step": 13143
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000823233588892553,
+      "loss": 0.2196,
+      "step": 13144
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008230953829092422,
+      "loss": 0.0964,
+      "step": 13145
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008229571804140439,
+      "loss": 0.1076,
+      "step": 13146
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008228189814096825,
+      "loss": 0.0785,
+      "step": 13147
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008226807858988838,
+      "loss": 0.0823,
+      "step": 13148
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008225425938843725,
+      "loss": 0.1904,
+      "step": 13149
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008224044053688729,
+      "loss": 0.0975,
+      "step": 13150
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00082226622035511,
+      "loss": 0.1411,
+      "step": 13151
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008221280388458086,
+      "loss": 0.097,
+      "step": 13152
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008219898608436931,
+      "loss": 0.0645,
+      "step": 13153
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008218516863514878,
+      "loss": 0.0859,
+      "step": 13154
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008217135153719176,
+      "loss": 0.1085,
+      "step": 13155
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008215753479077069,
+      "loss": 0.0915,
+      "step": 13156
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008214371839615795,
+      "loss": 0.1553,
+      "step": 13157
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008212990235362603,
+      "loss": 0.0934,
+      "step": 13158
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008211608666344731,
+      "loss": 0.1322,
+      "step": 13159
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000821022713258942,
+      "loss": 0.0475,
+      "step": 13160
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008208845634123907,
+      "loss": 0.0714,
+      "step": 13161
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008207464170975437,
+      "loss": 0.1038,
+      "step": 13162
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000820608274317125,
+      "loss": 0.0997,
+      "step": 13163
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008204701350738581,
+      "loss": 0.1041,
+      "step": 13164
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000820331999370467,
+      "loss": 0.1027,
+      "step": 13165
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008201938672096748,
+      "loss": 0.1222,
+      "step": 13166
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008200557385942056,
+      "loss": 0.0895,
+      "step": 13167
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000819917613526783,
+      "loss": 0.1412,
+      "step": 13168
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008197794920101304,
+      "loss": 0.054,
+      "step": 13169
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0008196413740469712,
+      "loss": 0.1345,
+      "step": 13170
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008195032596400288,
+      "loss": 0.0578,
+      "step": 13171
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008193651487920262,
+      "loss": 0.1169,
+      "step": 13172
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008192270415056869,
+      "loss": 0.1228,
+      "step": 13173
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008190889377837334,
+      "loss": 0.1146,
+      "step": 13174
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008189508376288897,
+      "loss": 0.0651,
+      "step": 13175
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008188127410438784,
+      "loss": 0.1593,
+      "step": 13176
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008186746480314223,
+      "loss": 0.0568,
+      "step": 13177
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008185365585942441,
+      "loss": 0.0861,
+      "step": 13178
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008183984727350669,
+      "loss": 0.0738,
+      "step": 13179
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008182603904566135,
+      "loss": 0.1068,
+      "step": 13180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008181223117616057,
+      "loss": 0.0595,
+      "step": 13181
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008179842366527673,
+      "loss": 0.1704,
+      "step": 13182
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008178461651328197,
+      "loss": 0.0941,
+      "step": 13183
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008177080972044861,
+      "loss": 0.1289,
+      "step": 13184
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008175700328704886,
+      "loss": 0.0811,
+      "step": 13185
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008174319721335494,
+      "loss": 0.1125,
+      "step": 13186
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008172939149963903,
+      "loss": 0.0804,
+      "step": 13187
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008171558614617343,
+      "loss": 0.1237,
+      "step": 13188
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008170178115323027,
+      "loss": 0.1069,
+      "step": 13189
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008168797652108182,
+      "loss": 0.075,
+      "step": 13190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008167417225000023,
+      "loss": 0.073,
+      "step": 13191
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008166036834025769,
+      "loss": 0.096,
+      "step": 13192
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008164656479212636,
+      "loss": 0.0677,
+      "step": 13193
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008163276160587842,
+      "loss": 0.1295,
+      "step": 13194
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008161895878178606,
+      "loss": 0.0979,
+      "step": 13195
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008160515632012143,
+      "loss": 0.1194,
+      "step": 13196
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008159135422115668,
+      "loss": 0.1078,
+      "step": 13197
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008157755248516394,
+      "loss": 0.1006,
+      "step": 13198
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008156375111241532,
+      "loss": 0.1078,
+      "step": 13199
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008154995010318301,
+      "loss": 0.0813,
+      "step": 13200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008153614945773906,
+      "loss": 0.0684,
+      "step": 13201
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008152234917635564,
+      "loss": 0.1179,
+      "step": 13202
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008150854925930486,
+      "loss": 0.1018,
+      "step": 13203
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008149474970685877,
+      "loss": 0.0857,
+      "step": 13204
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008148095051928952,
+      "loss": 0.1536,
+      "step": 13205
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008146715169686916,
+      "loss": 0.0778,
+      "step": 13206
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008145335323986972,
+      "loss": 0.0554,
+      "step": 13207
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008143955514856338,
+      "loss": 0.1201,
+      "step": 13208
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008142575742322214,
+      "loss": 0.1273,
+      "step": 13209
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008141196006411805,
+      "loss": 0.0853,
+      "step": 13210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008139816307152319,
+      "loss": 0.0849,
+      "step": 13211
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008138436644570959,
+      "loss": 0.1118,
+      "step": 13212
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008137057018694928,
+      "loss": 0.1561,
+      "step": 13213
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008135677429551424,
+      "loss": 0.0974,
+      "step": 13214
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008134297877167658,
+      "loss": 0.0892,
+      "step": 13215
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0008132918361570824,
+      "loss": 0.1282,
+      "step": 13216
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000813153888278813,
+      "loss": 0.0707,
+      "step": 13217
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008130159440846769,
+      "loss": 0.1443,
+      "step": 13218
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008128780035773942,
+      "loss": 0.0817,
+      "step": 13219
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000812740066759685,
+      "loss": 0.0867,
+      "step": 13220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008126021336342684,
+      "loss": 0.1573,
+      "step": 13221
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008124642042038649,
+      "loss": 0.0754,
+      "step": 13222
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008123262784711937,
+      "loss": 0.0809,
+      "step": 13223
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008121883564389747,
+      "loss": 0.1038,
+      "step": 13224
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008120504381099267,
+      "loss": 0.1141,
+      "step": 13225
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008119125234867697,
+      "loss": 0.0735,
+      "step": 13226
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008117746125722224,
+      "loss": 0.114,
+      "step": 13227
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000811636705369005,
+      "loss": 0.1025,
+      "step": 13228
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008114988018798361,
+      "loss": 0.0739,
+      "step": 13229
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008113609021074349,
+      "loss": 0.098,
+      "step": 13230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008112230060545201,
+      "loss": 0.0888,
+      "step": 13231
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008110851137238111,
+      "loss": 0.0724,
+      "step": 13232
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008109472251180268,
+      "loss": 0.0898,
+      "step": 13233
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008108093402398854,
+      "loss": 0.0851,
+      "step": 13234
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008106714590921066,
+      "loss": 0.0958,
+      "step": 13235
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008105335816774082,
+      "loss": 0.0658,
+      "step": 13236
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008103957079985096,
+      "loss": 0.0911,
+      "step": 13237
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008102578380581287,
+      "loss": 0.0908,
+      "step": 13238
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008101199718589842,
+      "loss": 0.057,
+      "step": 13239
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008099821094037943,
+      "loss": 0.1507,
+      "step": 13240
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008098442506952774,
+      "loss": 0.108,
+      "step": 13241
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008097063957361515,
+      "loss": 0.1074,
+      "step": 13242
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008095685445291354,
+      "loss": 0.081,
+      "step": 13243
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008094306970769468,
+      "loss": 0.0828,
+      "step": 13244
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008092928533823036,
+      "loss": 0.066,
+      "step": 13245
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008091550134479236,
+      "loss": 0.1313,
+      "step": 13246
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008090171772765249,
+      "loss": 0.0665,
+      "step": 13247
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008088793448708251,
+      "loss": 0.0873,
+      "step": 13248
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008087415162335424,
+      "loss": 0.1951,
+      "step": 13249
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000808603691367394,
+      "loss": 0.1162,
+      "step": 13250
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008084658702750975,
+      "loss": 0.1246,
+      "step": 13251
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008083280529593704,
+      "loss": 0.075,
+      "step": 13252
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008081902394229301,
+      "loss": 0.0876,
+      "step": 13253
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008080524296684937,
+      "loss": 0.0811,
+      "step": 13254
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008079146236987791,
+      "loss": 0.077,
+      "step": 13255
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008077768215165031,
+      "loss": 0.0923,
+      "step": 13256
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008076390231243827,
+      "loss": 0.1041,
+      "step": 13257
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008075012285251351,
+      "loss": 0.0797,
+      "step": 13258
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008073634377214774,
+      "loss": 0.0939,
+      "step": 13259
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008072256507161262,
+      "loss": 0.0568,
+      "step": 13260
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008070878675117979,
+      "loss": 0.1165,
+      "step": 13261
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0008069500881112103,
+      "loss": 0.0839,
+      "step": 13262
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008068123125170793,
+      "loss": 0.0897,
+      "step": 13263
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008066745407321219,
+      "loss": 0.0757,
+      "step": 13264
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008065367727590545,
+      "loss": 0.1244,
+      "step": 13265
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008063990086005933,
+      "loss": 0.0658,
+      "step": 13266
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008062612482594545,
+      "loss": 0.0868,
+      "step": 13267
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008061234917383551,
+      "loss": 0.077,
+      "step": 13268
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008059857390400106,
+      "loss": 0.0827,
+      "step": 13269
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008058479901671378,
+      "loss": 0.084,
+      "step": 13270
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008057102451224522,
+      "loss": 0.124,
+      "step": 13271
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008055725039086702,
+      "loss": 0.0865,
+      "step": 13272
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008054347665285072,
+      "loss": 0.0908,
+      "step": 13273
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008052970329846792,
+      "loss": 0.0918,
+      "step": 13274
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008051593032799024,
+      "loss": 0.0744,
+      "step": 13275
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008050215774168924,
+      "loss": 0.0803,
+      "step": 13276
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008048838553983643,
+      "loss": 0.0728,
+      "step": 13277
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008047461372270337,
+      "loss": 0.1031,
+      "step": 13278
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008046084229056166,
+      "loss": 0.0754,
+      "step": 13279
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000804470712436828,
+      "loss": 0.0811,
+      "step": 13280
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008043330058233829,
+      "loss": 0.1105,
+      "step": 13281
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008041953030679972,
+      "loss": 0.0786,
+      "step": 13282
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008040576041733856,
+      "loss": 0.1036,
+      "step": 13283
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008039199091422634,
+      "loss": 0.0572,
+      "step": 13284
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008037822179773455,
+      "loss": 0.1066,
+      "step": 13285
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008036445306813469,
+      "loss": 0.0784,
+      "step": 13286
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008035068472569819,
+      "loss": 0.0927,
+      "step": 13287
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008033691677069661,
+      "loss": 0.1013,
+      "step": 13288
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008032314920340137,
+      "loss": 0.0798,
+      "step": 13289
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008030938202408395,
+      "loss": 0.0638,
+      "step": 13290
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008029561523301582,
+      "loss": 0.1267,
+      "step": 13291
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008028184883046839,
+      "loss": 0.0958,
+      "step": 13292
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008026808281671312,
+      "loss": 0.0995,
+      "step": 13293
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008025431719202138,
+      "loss": 0.0901,
+      "step": 13294
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008024055195666468,
+      "loss": 0.072,
+      "step": 13295
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008022678711091442,
+      "loss": 0.0709,
+      "step": 13296
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00080213022655042,
+      "loss": 0.0792,
+      "step": 13297
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000801992585893188,
+      "loss": 0.0848,
+      "step": 13298
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008018549491401621,
+      "loss": 0.0882,
+      "step": 13299
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008017173162940566,
+      "loss": 0.1057,
+      "step": 13300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008015796873575843,
+      "loss": 0.0936,
+      "step": 13301
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008014420623334602,
+      "loss": 0.1003,
+      "step": 13302
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008013044412243972,
+      "loss": 0.135,
+      "step": 13303
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000801166824033109,
+      "loss": 0.0925,
+      "step": 13304
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008010292107623085,
+      "loss": 0.0515,
+      "step": 13305
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008008916014147101,
+      "loss": 0.1057,
+      "step": 13306
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008007539959930259,
+      "loss": 0.1163,
+      "step": 13307
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0008006163944999703,
+      "loss": 0.1171,
+      "step": 13308
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0008004787969382561,
+      "loss": 0.1407,
+      "step": 13309
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0008003412033105961,
+      "loss": 0.0917,
+      "step": 13310
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0008002036136197033,
+      "loss": 0.0837,
+      "step": 13311
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000800066027868291,
+      "loss": 0.1168,
+      "step": 13312
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007999284460590717,
+      "loss": 0.1062,
+      "step": 13313
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007997908681947579,
+      "loss": 0.0845,
+      "step": 13314
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000799653294278063,
+      "loss": 0.109,
+      "step": 13315
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000799515724311699,
+      "loss": 0.1078,
+      "step": 13316
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007993781582983791,
+      "loss": 0.101,
+      "step": 13317
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007992405962408154,
+      "loss": 0.0837,
+      "step": 13318
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00079910303814172,
+      "loss": 0.0638,
+      "step": 13319
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007989654840038055,
+      "loss": 0.0884,
+      "step": 13320
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007988279338297837,
+      "loss": 0.1111,
+      "step": 13321
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007986903876223673,
+      "loss": 0.1223,
+      "step": 13322
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007985528453842683,
+      "loss": 0.0787,
+      "step": 13323
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007984153071181985,
+      "loss": 0.124,
+      "step": 13324
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00079827777282687,
+      "loss": 0.1027,
+      "step": 13325
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007981402425129942,
+      "loss": 0.1201,
+      "step": 13326
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007980027161792831,
+      "loss": 0.0869,
+      "step": 13327
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007978651938284484,
+      "loss": 0.0903,
+      "step": 13328
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007977276754632019,
+      "loss": 0.0889,
+      "step": 13329
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007975901610862549,
+      "loss": 0.0601,
+      "step": 13330
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007974526507003186,
+      "loss": 0.1188,
+      "step": 13331
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000797315144308105,
+      "loss": 0.0792,
+      "step": 13332
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007971776419123248,
+      "loss": 0.0945,
+      "step": 13333
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000797040143515689,
+      "loss": 0.0437,
+      "step": 13334
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007969026491209095,
+      "loss": 0.0907,
+      "step": 13335
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007967651587306969,
+      "loss": 0.0887,
+      "step": 13336
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007966276723477622,
+      "loss": 0.0754,
+      "step": 13337
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007964901899748163,
+      "loss": 0.0859,
+      "step": 13338
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007963527116145702,
+      "loss": 0.1097,
+      "step": 13339
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007962152372697342,
+      "loss": 0.0807,
+      "step": 13340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007960777669430188,
+      "loss": 0.1234,
+      "step": 13341
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007959403006371354,
+      "loss": 0.1218,
+      "step": 13342
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007958028383547939,
+      "loss": 0.1072,
+      "step": 13343
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007956653800987049,
+      "loss": 0.091,
+      "step": 13344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007955279258715787,
+      "loss": 0.1405,
+      "step": 13345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007953904756761255,
+      "loss": 0.1044,
+      "step": 13346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007952530295150552,
+      "loss": 0.0645,
+      "step": 13347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007951155873910783,
+      "loss": 0.1298,
+      "step": 13348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007949781493069048,
+      "loss": 0.0796,
+      "step": 13349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007948407152652446,
+      "loss": 0.1365,
+      "step": 13350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007947032852688075,
+      "loss": 0.0844,
+      "step": 13351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000794565859320303,
+      "loss": 0.0892,
+      "step": 13352
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007944284374224413,
+      "loss": 0.1017,
+      "step": 13353
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0007942910195779312,
+      "loss": 0.0872,
+      "step": 13354
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007941536057894834,
+      "loss": 0.0852,
+      "step": 13355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007940161960598067,
+      "loss": 0.1194,
+      "step": 13356
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007938787903916104,
+      "loss": 0.0956,
+      "step": 13357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007937413887876037,
+      "loss": 0.1948,
+      "step": 13358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007936039912504963,
+      "loss": 0.0874,
+      "step": 13359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.000793466597782997,
+      "loss": 0.1165,
+      "step": 13360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007933292083878146,
+      "loss": 0.0813,
+      "step": 13361
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007931918230676586,
+      "loss": 0.114,
+      "step": 13362
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007930544418252377,
+      "loss": 0.17,
+      "step": 13363
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007929170646632605,
+      "loss": 0.099,
+      "step": 13364
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007927796915844361,
+      "loss": 0.0991,
+      "step": 13365
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.000792642322591473,
+      "loss": 0.1133,
+      "step": 13366
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007925049576870794,
+      "loss": 0.0727,
+      "step": 13367
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007923675968739643,
+      "loss": 0.1223,
+      "step": 13368
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007922302401548358,
+      "loss": 0.1071,
+      "step": 13369
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007920928875324027,
+      "loss": 0.1194,
+      "step": 13370
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007919555390093728,
+      "loss": 0.0897,
+      "step": 13371
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007918181945884544,
+      "loss": 0.059,
+      "step": 13372
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007916808542723552,
+      "loss": 0.0895,
+      "step": 13373
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007915435180637836,
+      "loss": 0.1201,
+      "step": 13374
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007914061859654476,
+      "loss": 0.1355,
+      "step": 13375
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007912688579800551,
+      "loss": 0.0566,
+      "step": 13376
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007911315341103135,
+      "loss": 0.0695,
+      "step": 13377
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007909942143589309,
+      "loss": 0.1428,
+      "step": 13378
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007908568987286144,
+      "loss": 0.1149,
+      "step": 13379
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007907195872220718,
+      "loss": 0.0773,
+      "step": 13380
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007905822798420102,
+      "loss": 0.0655,
+      "step": 13381
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007904449765911378,
+      "loss": 0.087,
+      "step": 13382
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007903076774721612,
+      "loss": 0.114,
+      "step": 13383
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007901703824877878,
+      "loss": 0.1113,
+      "step": 13384
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007900330916407244,
+      "loss": 0.0883,
+      "step": 13385
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007898958049336785,
+      "loss": 0.0736,
+      "step": 13386
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007897585223693564,
+      "loss": 0.0876,
+      "step": 13387
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007896212439504658,
+      "loss": 0.0572,
+      "step": 13388
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007894839696797132,
+      "loss": 0.0891,
+      "step": 13389
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007893466995598047,
+      "loss": 0.0816,
+      "step": 13390
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007892094335934477,
+      "loss": 0.0641,
+      "step": 13391
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007890721717833484,
+      "loss": 0.0994,
+      "step": 13392
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007889349141322135,
+      "loss": 0.0468,
+      "step": 13393
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007887976606427485,
+      "loss": 0.1136,
+      "step": 13394
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007886604113176609,
+      "loss": 0.056,
+      "step": 13395
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007885231661596561,
+      "loss": 0.0682,
+      "step": 13396
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007883859251714407,
+      "loss": 0.1353,
+      "step": 13397
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007882486883557205,
+      "loss": 0.0784,
+      "step": 13398
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007881114557152016,
+      "loss": 0.0555,
+      "step": 13399
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0007879742272525892,
+      "loss": 0.1053,
+      "step": 13400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007878370029705902,
+      "loss": 0.1631,
+      "step": 13401
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007876997828719097,
+      "loss": 0.066,
+      "step": 13402
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007875625669592534,
+      "loss": 0.075,
+      "step": 13403
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007874253552353271,
+      "loss": 0.0891,
+      "step": 13404
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000787288147702836,
+      "loss": 0.0856,
+      "step": 13405
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007871509443644852,
+      "loss": 0.1575,
+      "step": 13406
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007870137452229801,
+      "loss": 0.1361,
+      "step": 13407
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007868765502810267,
+      "loss": 0.0767,
+      "step": 13408
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007867393595413297,
+      "loss": 0.092,
+      "step": 13409
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007866021730065938,
+      "loss": 0.1054,
+      "step": 13410
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007864649906795243,
+      "loss": 0.0763,
+      "step": 13411
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000786327812562826,
+      "loss": 0.1364,
+      "step": 13412
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007861906386592038,
+      "loss": 0.0844,
+      "step": 13413
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007860534689713618,
+      "loss": 0.0763,
+      "step": 13414
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007859163035020059,
+      "loss": 0.0631,
+      "step": 13415
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007857791422538397,
+      "loss": 0.0809,
+      "step": 13416
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007856419852295678,
+      "loss": 0.067,
+      "step": 13417
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007855048324318948,
+      "loss": 0.1031,
+      "step": 13418
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000785367683863525,
+      "loss": 0.0961,
+      "step": 13419
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000785230539527162,
+      "loss": 0.1045,
+      "step": 13420
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007850933994255111,
+      "loss": 0.1167,
+      "step": 13421
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007849562635612756,
+      "loss": 0.1229,
+      "step": 13422
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007848191319371594,
+      "loss": 0.0841,
+      "step": 13423
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007846820045558668,
+      "loss": 0.0688,
+      "step": 13424
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007845448814201014,
+      "loss": 0.0836,
+      "step": 13425
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007844077625325667,
+      "loss": 0.12,
+      "step": 13426
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007842706478959666,
+      "loss": 0.0917,
+      "step": 13427
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007841335375130045,
+      "loss": 0.1072,
+      "step": 13428
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007839964313863843,
+      "loss": 0.0914,
+      "step": 13429
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007838593295188091,
+      "loss": 0.0839,
+      "step": 13430
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007837222319129822,
+      "loss": 0.0917,
+      "step": 13431
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007835851385716064,
+      "loss": 0.1041,
+      "step": 13432
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007834480494973855,
+      "loss": 0.1228,
+      "step": 13433
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.000783310964693022,
+      "loss": 0.163,
+      "step": 13434
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007831738841612194,
+      "loss": 0.0665,
+      "step": 13435
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007830368079046803,
+      "loss": 0.0649,
+      "step": 13436
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007828997359261076,
+      "loss": 0.0704,
+      "step": 13437
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007827626682282039,
+      "loss": 0.0706,
+      "step": 13438
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007826256048136718,
+      "loss": 0.0698,
+      "step": 13439
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007824885456852135,
+      "loss": 0.0807,
+      "step": 13440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007823514908455324,
+      "loss": 0.0815,
+      "step": 13441
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007822144402973303,
+      "loss": 0.0961,
+      "step": 13442
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007820773940433094,
+      "loss": 0.0784,
+      "step": 13443
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007819403520861722,
+      "loss": 0.1005,
+      "step": 13444
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007818033144286207,
+      "loss": 0.0838,
+      "step": 13445
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0007816662810733568,
+      "loss": 0.076,
+      "step": 13446
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007815292520230822,
+      "loss": 0.1241,
+      "step": 13447
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007813922272804995,
+      "loss": 0.1523,
+      "step": 13448
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00078125520684831,
+      "loss": 0.1536,
+      "step": 13449
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007811181907292155,
+      "loss": 0.0684,
+      "step": 13450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007809811789259178,
+      "loss": 0.0925,
+      "step": 13451
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007808441714411181,
+      "loss": 0.0858,
+      "step": 13452
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007807071682775178,
+      "loss": 0.1132,
+      "step": 13453
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007805701694378186,
+      "loss": 0.0633,
+      "step": 13454
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007804331749247213,
+      "loss": 0.1078,
+      "step": 13455
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007802961847409278,
+      "loss": 0.0818,
+      "step": 13456
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007801591988891388,
+      "loss": 0.0729,
+      "step": 13457
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007800222173720553,
+      "loss": 0.1025,
+      "step": 13458
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007798852401923779,
+      "loss": 0.079,
+      "step": 13459
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007797482673528077,
+      "loss": 0.1124,
+      "step": 13460
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007796112988560459,
+      "loss": 0.1285,
+      "step": 13461
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007794743347047928,
+      "loss": 0.0639,
+      "step": 13462
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000779337374901749,
+      "loss": 0.0544,
+      "step": 13463
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007792004194496149,
+      "loss": 0.1211,
+      "step": 13464
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007790634683510911,
+      "loss": 0.095,
+      "step": 13465
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007789265216088779,
+      "loss": 0.1146,
+      "step": 13466
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007787895792256749,
+      "loss": 0.1666,
+      "step": 13467
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007786526412041834,
+      "loss": 0.0814,
+      "step": 13468
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000778515707547103,
+      "loss": 0.1032,
+      "step": 13469
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007783787782571334,
+      "loss": 0.1682,
+      "step": 13470
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007782418533369748,
+      "loss": 0.1061,
+      "step": 13471
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000778104932789327,
+      "loss": 0.0974,
+      "step": 13472
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007779680166168896,
+      "loss": 0.0937,
+      "step": 13473
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007778311048223618,
+      "loss": 0.0501,
+      "step": 13474
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007776941974084442,
+      "loss": 0.0896,
+      "step": 13475
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007775572943778354,
+      "loss": 0.0771,
+      "step": 13476
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007774203957332353,
+      "loss": 0.1193,
+      "step": 13477
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007772835014773431,
+      "loss": 0.0594,
+      "step": 13478
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007771466116128578,
+      "loss": 0.1129,
+      "step": 13479
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007770097261424782,
+      "loss": 0.1178,
+      "step": 13480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000776872845068904,
+      "loss": 0.0555,
+      "step": 13481
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000776735968394834,
+      "loss": 0.1034,
+      "step": 13482
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007765990961229672,
+      "loss": 0.0961,
+      "step": 13483
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007764622282560019,
+      "loss": 0.1082,
+      "step": 13484
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007763253647966369,
+      "loss": 0.0872,
+      "step": 13485
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007761885057475712,
+      "loss": 0.0673,
+      "step": 13486
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007760516511115026,
+      "loss": 0.1327,
+      "step": 13487
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007759148008911304,
+      "loss": 0.0779,
+      "step": 13488
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007757779550891526,
+      "loss": 0.0816,
+      "step": 13489
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007756411137082673,
+      "loss": 0.1167,
+      "step": 13490
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007755042767511725,
+      "loss": 0.1358,
+      "step": 13491
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007753674442205667,
+      "loss": 0.0984,
+      "step": 13492
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0007752306161191478,
+      "loss": 0.1108,
+      "step": 13493
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007750937924496131,
+      "loss": 0.0908,
+      "step": 13494
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007749569732146615,
+      "loss": 0.1318,
+      "step": 13495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00077482015841699,
+      "loss": 0.1013,
+      "step": 13496
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007746833480592962,
+      "loss": 0.0883,
+      "step": 13497
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007745465421442781,
+      "loss": 0.0972,
+      "step": 13498
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000774409740674633,
+      "loss": 0.0933,
+      "step": 13499
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007742729436530577,
+      "loss": 0.1177,
+      "step": 13500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007741361510822505,
+      "loss": 0.0897,
+      "step": 13501
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007739993629649078,
+      "loss": 0.1161,
+      "step": 13502
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007738625793037273,
+      "loss": 0.1006,
+      "step": 13503
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007737258001014055,
+      "loss": 0.0696,
+      "step": 13504
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007735890253606399,
+      "loss": 0.0992,
+      "step": 13505
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007734522550841268,
+      "loss": 0.0815,
+      "step": 13506
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000773315489274563,
+      "loss": 0.0779,
+      "step": 13507
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007731787279346455,
+      "loss": 0.1164,
+      "step": 13508
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000773041971067071,
+      "loss": 0.1273,
+      "step": 13509
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007729052186745358,
+      "loss": 0.1189,
+      "step": 13510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007727684707597363,
+      "loss": 0.0682,
+      "step": 13511
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007726317273253683,
+      "loss": 0.0724,
+      "step": 13512
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000772494988374129,
+      "loss": 0.0779,
+      "step": 13513
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007723582539087136,
+      "loss": 0.0839,
+      "step": 13514
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007722215239318191,
+      "loss": 0.0616,
+      "step": 13515
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007720847984461409,
+      "loss": 0.0959,
+      "step": 13516
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007719480774543752,
+      "loss": 0.072,
+      "step": 13517
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007718113609592173,
+      "loss": 0.1292,
+      "step": 13518
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007716746489633634,
+      "loss": 0.0619,
+      "step": 13519
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007715379414695083,
+      "loss": 0.0663,
+      "step": 13520
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007714012384803489,
+      "loss": 0.111,
+      "step": 13521
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007712645399985796,
+      "loss": 0.0499,
+      "step": 13522
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007711278460268959,
+      "loss": 0.1031,
+      "step": 13523
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007709911565679934,
+      "loss": 0.0977,
+      "step": 13524
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007708544716245672,
+      "loss": 0.0992,
+      "step": 13525
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000770717791199312,
+      "loss": 0.1443,
+      "step": 13526
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007705811152949226,
+      "loss": 0.0778,
+      "step": 13527
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007704444439140948,
+      "loss": 0.0884,
+      "step": 13528
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007703077770595229,
+      "loss": 0.0652,
+      "step": 13529
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007701711147339017,
+      "loss": 0.0519,
+      "step": 13530
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007700344569399258,
+      "loss": 0.1637,
+      "step": 13531
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007698978036802897,
+      "loss": 0.0923,
+      "step": 13532
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007697611549576878,
+      "loss": 0.1129,
+      "step": 13533
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007696245107748144,
+      "loss": 0.1222,
+      "step": 13534
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007694878711343643,
+      "loss": 0.0895,
+      "step": 13535
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007693512360390314,
+      "loss": 0.1227,
+      "step": 13536
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007692146054915096,
+      "loss": 0.0851,
+      "step": 13537
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007690779794944931,
+      "loss": 0.0738,
+      "step": 13538
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0007689413580506759,
+      "loss": 0.071,
+      "step": 13539
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007688047411627513,
+      "loss": 0.0976,
+      "step": 13540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007686681288334137,
+      "loss": 0.0653,
+      "step": 13541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007685315210653565,
+      "loss": 0.1327,
+      "step": 13542
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007683949178612734,
+      "loss": 0.0807,
+      "step": 13543
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007682583192238575,
+      "loss": 0.0912,
+      "step": 13544
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007681217251558025,
+      "loss": 0.1071,
+      "step": 13545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007679851356598016,
+      "loss": 0.1006,
+      "step": 13546
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007678485507385477,
+      "loss": 0.093,
+      "step": 13547
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007677119703947345,
+      "loss": 0.1181,
+      "step": 13548
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007675753946310549,
+      "loss": 0.0635,
+      "step": 13549
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007674388234502012,
+      "loss": 0.0842,
+      "step": 13550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000767302256854867,
+      "loss": 0.0858,
+      "step": 13551
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007671656948477448,
+      "loss": 0.1307,
+      "step": 13552
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007670291374315271,
+      "loss": 0.1072,
+      "step": 13553
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007668925846089062,
+      "loss": 0.0604,
+      "step": 13554
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007667560363825752,
+      "loss": 0.0743,
+      "step": 13555
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007666194927552263,
+      "loss": 0.1411,
+      "step": 13556
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007664829537295519,
+      "loss": 0.0825,
+      "step": 13557
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007663464193082439,
+      "loss": 0.0894,
+      "step": 13558
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007662098894939944,
+      "loss": 0.0598,
+      "step": 13559
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007660733642894955,
+      "loss": 0.0991,
+      "step": 13560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007659368436974392,
+      "loss": 0.0822,
+      "step": 13561
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007658003277205176,
+      "loss": 0.1141,
+      "step": 13562
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007656638163614221,
+      "loss": 0.07,
+      "step": 13563
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007655273096228445,
+      "loss": 0.1008,
+      "step": 13564
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007653908075074763,
+      "loss": 0.1416,
+      "step": 13565
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007652543100180089,
+      "loss": 0.077,
+      "step": 13566
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007651178171571335,
+      "loss": 0.0718,
+      "step": 13567
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000764981328927542,
+      "loss": 0.1487,
+      "step": 13568
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007648448453319254,
+      "loss": 0.0951,
+      "step": 13569
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007647083663729747,
+      "loss": 0.0953,
+      "step": 13570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007645718920533806,
+      "loss": 0.0673,
+      "step": 13571
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007644354223758347,
+      "loss": 0.0968,
+      "step": 13572
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007642989573430273,
+      "loss": 0.1339,
+      "step": 13573
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000764162496957649,
+      "loss": 0.0884,
+      "step": 13574
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007640260412223912,
+      "loss": 0.0899,
+      "step": 13575
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007638895901399438,
+      "loss": 0.0901,
+      "step": 13576
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007637531437129976,
+      "loss": 0.0957,
+      "step": 13577
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000763616701944243,
+      "loss": 0.0573,
+      "step": 13578
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007634802648363703,
+      "loss": 0.0959,
+      "step": 13579
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000763343832392069,
+      "loss": 0.1017,
+      "step": 13580
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007632074046140301,
+      "loss": 0.116,
+      "step": 13581
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007630709815049432,
+      "loss": 0.0986,
+      "step": 13582
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007629345630674986,
+      "loss": 0.0828,
+      "step": 13583
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007627981493043858,
+      "loss": 0.1537,
+      "step": 13584
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0007626617402182945,
+      "loss": 0.0803,
+      "step": 13585
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007625253358119142,
+      "loss": 0.0596,
+      "step": 13586
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007623889360879346,
+      "loss": 0.0873,
+      "step": 13587
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007622525410490452,
+      "loss": 0.082,
+      "step": 13588
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007621161506979359,
+      "loss": 0.1135,
+      "step": 13589
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007619797650372952,
+      "loss": 0.0732,
+      "step": 13590
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007618433840698125,
+      "loss": 0.131,
+      "step": 13591
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000761707007798177,
+      "loss": 0.0914,
+      "step": 13592
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007615706362250775,
+      "loss": 0.1298,
+      "step": 13593
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007614342693532027,
+      "loss": 0.0751,
+      "step": 13594
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007612979071852422,
+      "loss": 0.1005,
+      "step": 13595
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007611615497238843,
+      "loss": 0.1053,
+      "step": 13596
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007610251969718172,
+      "loss": 0.055,
+      "step": 13597
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007608888489317303,
+      "loss": 0.0941,
+      "step": 13598
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007607525056063113,
+      "loss": 0.0754,
+      "step": 13599
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007606161669982484,
+      "loss": 0.0862,
+      "step": 13600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007604798331102308,
+      "loss": 0.1226,
+      "step": 13601
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000760343503944946,
+      "loss": 0.0605,
+      "step": 13602
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000760207179505082,
+      "loss": 0.0792,
+      "step": 13603
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007600708597933272,
+      "loss": 0.0803,
+      "step": 13604
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007599345448123692,
+      "loss": 0.1565,
+      "step": 13605
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007597982345648958,
+      "loss": 0.0648,
+      "step": 13606
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007596619290535943,
+      "loss": 0.1129,
+      "step": 13607
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007595256282811532,
+      "loss": 0.0746,
+      "step": 13608
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007593893322502592,
+      "loss": 0.0655,
+      "step": 13609
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007592530409636002,
+      "loss": 0.1079,
+      "step": 13610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007591167544238636,
+      "loss": 0.0783,
+      "step": 13611
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007589804726337361,
+      "loss": 0.1199,
+      "step": 13612
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.000758844195595905,
+      "loss": 0.0914,
+      "step": 13613
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007587079233130573,
+      "loss": 0.0594,
+      "step": 13614
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007585716557878805,
+      "loss": 0.103,
+      "step": 13615
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007584353930230609,
+      "loss": 0.0813,
+      "step": 13616
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007582991350212855,
+      "loss": 0.1676,
+      "step": 13617
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007581628817852408,
+      "loss": 0.0599,
+      "step": 13618
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007580266333176136,
+      "loss": 0.0978,
+      "step": 13619
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007578903896210896,
+      "loss": 0.0535,
+      "step": 13620
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007577541506983564,
+      "loss": 0.0801,
+      "step": 13621
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007576179165520996,
+      "loss": 0.0793,
+      "step": 13622
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007574816871850056,
+      "loss": 0.1665,
+      "step": 13623
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00075734546259976,
+      "loss": 0.0919,
+      "step": 13624
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007572092427990494,
+      "loss": 0.1468,
+      "step": 13625
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007570730277855595,
+      "loss": 0.0607,
+      "step": 13626
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007569368175619758,
+      "loss": 0.1069,
+      "step": 13627
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007568006121309848,
+      "loss": 0.0659,
+      "step": 13628
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007566644114952712,
+      "loss": 0.1108,
+      "step": 13629
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007565282156575214,
+      "loss": 0.0536,
+      "step": 13630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0007563920246204203,
+      "loss": 0.1056,
+      "step": 13631
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007562558383866535,
+      "loss": 0.0873,
+      "step": 13632
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007561196569589059,
+      "loss": 0.0842,
+      "step": 13633
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007559834803398627,
+      "loss": 0.1464,
+      "step": 13634
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007558473085322093,
+      "loss": 0.1237,
+      "step": 13635
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007557111415386306,
+      "loss": 0.0887,
+      "step": 13636
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007555749793618114,
+      "loss": 0.0606,
+      "step": 13637
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007554388220044365,
+      "loss": 0.1454,
+      "step": 13638
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007553026694691903,
+      "loss": 0.085,
+      "step": 13639
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007551665217587576,
+      "loss": 0.0577,
+      "step": 13640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007550303788758228,
+      "loss": 0.0605,
+      "step": 13641
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007548942408230707,
+      "loss": 0.0801,
+      "step": 13642
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007547581076031852,
+      "loss": 0.168,
+      "step": 13643
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007546219792188507,
+      "loss": 0.0804,
+      "step": 13644
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007544858556727508,
+      "loss": 0.1221,
+      "step": 13645
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007543497369675703,
+      "loss": 0.0767,
+      "step": 13646
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007542136231059921,
+      "loss": 0.148,
+      "step": 13647
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007540775140907013,
+      "loss": 0.0732,
+      "step": 13648
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007539414099243808,
+      "loss": 0.0856,
+      "step": 13649
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007538053106097145,
+      "loss": 0.1354,
+      "step": 13650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007536692161493858,
+      "loss": 0.0959,
+      "step": 13651
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007535331265460781,
+      "loss": 0.0638,
+      "step": 13652
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007533970418024749,
+      "loss": 0.0654,
+      "step": 13653
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007532609619212591,
+      "loss": 0.0596,
+      "step": 13654
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007531248869051144,
+      "loss": 0.091,
+      "step": 13655
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007529888167567234,
+      "loss": 0.0621,
+      "step": 13656
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007528527514787694,
+      "loss": 0.1199,
+      "step": 13657
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007527166910739351,
+      "loss": 0.0717,
+      "step": 13658
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007525806355449033,
+      "loss": 0.1505,
+      "step": 13659
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007524445848943562,
+      "loss": 0.1516,
+      "step": 13660
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007523085391249773,
+      "loss": 0.062,
+      "step": 13661
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007521724982394484,
+      "loss": 0.127,
+      "step": 13662
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007520364622404522,
+      "loss": 0.1067,
+      "step": 13663
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007519004311306709,
+      "loss": 0.0876,
+      "step": 13664
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007517644049127866,
+      "loss": 0.0688,
+      "step": 13665
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007516283835894813,
+      "loss": 0.0672,
+      "step": 13666
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007514923671634368,
+      "loss": 0.1133,
+      "step": 13667
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000751356355637336,
+      "loss": 0.0927,
+      "step": 13668
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00075122034901386,
+      "loss": 0.0668,
+      "step": 13669
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007510843472956904,
+      "loss": 0.1052,
+      "step": 13670
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007509483504855088,
+      "loss": 0.1461,
+      "step": 13671
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007508123585859972,
+      "loss": 0.108,
+      "step": 13672
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007506763715998365,
+      "loss": 0.1111,
+      "step": 13673
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007505403895297079,
+      "loss": 0.1437,
+      "step": 13674
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007504044123782935,
+      "loss": 0.1555,
+      "step": 13675
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007502684401482736,
+      "loss": 0.0783,
+      "step": 13676
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0007501324728423296,
+      "loss": 0.0898,
+      "step": 13677
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007499965104631423,
+      "loss": 0.0803,
+      "step": 13678
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007498605530133927,
+      "loss": 0.1923,
+      "step": 13679
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007497246004957608,
+      "loss": 0.2213,
+      "step": 13680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007495886529129283,
+      "loss": 0.0798,
+      "step": 13681
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007494527102675754,
+      "loss": 0.1001,
+      "step": 13682
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007493167725623822,
+      "loss": 0.0955,
+      "step": 13683
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007491808398000294,
+      "loss": 0.1095,
+      "step": 13684
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007490449119831972,
+      "loss": 0.1081,
+      "step": 13685
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007489089891145657,
+      "loss": 0.1272,
+      "step": 13686
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007487730711968144,
+      "loss": 0.1252,
+      "step": 13687
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007486371582326241,
+      "loss": 0.0638,
+      "step": 13688
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007485012502246745,
+      "loss": 0.0641,
+      "step": 13689
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007483653471756453,
+      "loss": 0.1534,
+      "step": 13690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000748229449088216,
+      "loss": 0.0919,
+      "step": 13691
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007480935559650661,
+      "loss": 0.0848,
+      "step": 13692
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007479576678088752,
+      "loss": 0.0996,
+      "step": 13693
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007478217846223227,
+      "loss": 0.092,
+      "step": 13694
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007476859064080881,
+      "loss": 0.1237,
+      "step": 13695
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007475500331688503,
+      "loss": 0.1171,
+      "step": 13696
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007474141649072883,
+      "loss": 0.104,
+      "step": 13697
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007472783016260811,
+      "loss": 0.0954,
+      "step": 13698
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007471424433279079,
+      "loss": 0.1097,
+      "step": 13699
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007470065900154469,
+      "loss": 0.0787,
+      "step": 13700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007468707416913774,
+      "loss": 0.1042,
+      "step": 13701
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007467348983583779,
+      "loss": 0.0917,
+      "step": 13702
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007465990600191266,
+      "loss": 0.0753,
+      "step": 13703
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007464632266763018,
+      "loss": 0.0691,
+      "step": 13704
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007463273983325823,
+      "loss": 0.0853,
+      "step": 13705
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007461915749906459,
+      "loss": 0.1232,
+      "step": 13706
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007460557566531705,
+      "loss": 0.0737,
+      "step": 13707
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007459199433228345,
+      "loss": 0.0523,
+      "step": 13708
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007457841350023157,
+      "loss": 0.0842,
+      "step": 13709
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007456483316942922,
+      "loss": 0.0721,
+      "step": 13710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000745512533401441,
+      "loss": 0.0858,
+      "step": 13711
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007453767401264403,
+      "loss": 0.0898,
+      "step": 13712
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000745240951871967,
+      "loss": 0.0754,
+      "step": 13713
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007451051686406991,
+      "loss": 0.0868,
+      "step": 13714
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007449693904353137,
+      "loss": 0.0692,
+      "step": 13715
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000744833617258488,
+      "loss": 0.0803,
+      "step": 13716
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000744697849112899,
+      "loss": 0.1233,
+      "step": 13717
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007445620860012239,
+      "loss": 0.0833,
+      "step": 13718
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007444263279261393,
+      "loss": 0.0992,
+      "step": 13719
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007442905748903222,
+      "loss": 0.0845,
+      "step": 13720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007441548268964493,
+      "loss": 0.1609,
+      "step": 13721
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007440190839471975,
+      "loss": 0.08,
+      "step": 13722
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0007438833460452428,
+      "loss": 0.0654,
+      "step": 13723
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007437476131932617,
+      "loss": 0.1054,
+      "step": 13724
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007436118853939308,
+      "loss": 0.1342,
+      "step": 13725
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007434761626499264,
+      "loss": 0.0682,
+      "step": 13726
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007433404449639238,
+      "loss": 0.0796,
+      "step": 13727
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007432047323385998,
+      "loss": 0.0837,
+      "step": 13728
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007430690247766302,
+      "loss": 0.0891,
+      "step": 13729
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007429333222806906,
+      "loss": 0.1311,
+      "step": 13730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000742797624853457,
+      "loss": 0.1231,
+      "step": 13731
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007426619324976047,
+      "loss": 0.1215,
+      "step": 13732
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007425262452158089,
+      "loss": 0.071,
+      "step": 13733
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007423905630107459,
+      "loss": 0.0711,
+      "step": 13734
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007422548858850905,
+      "loss": 0.1245,
+      "step": 13735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007421192138415178,
+      "loss": 0.0751,
+      "step": 13736
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007419835468827032,
+      "loss": 0.0837,
+      "step": 13737
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007418478850113215,
+      "loss": 0.0819,
+      "step": 13738
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007417122282300479,
+      "loss": 0.1127,
+      "step": 13739
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007415765765415564,
+      "loss": 0.0911,
+      "step": 13740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007414409299485225,
+      "loss": 0.0559,
+      "step": 13741
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007413052884536208,
+      "loss": 0.0874,
+      "step": 13742
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007411696520595258,
+      "loss": 0.0635,
+      "step": 13743
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007410340207689116,
+      "loss": 0.119,
+      "step": 13744
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007408983945844524,
+      "loss": 0.0912,
+      "step": 13745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007407627735088229,
+      "loss": 0.1014,
+      "step": 13746
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007406271575446965,
+      "loss": 0.0918,
+      "step": 13747
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000740491546694748,
+      "loss": 0.1057,
+      "step": 13748
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007403559409616511,
+      "loss": 0.1071,
+      "step": 13749
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007402203403480792,
+      "loss": 0.0696,
+      "step": 13750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007400847448567063,
+      "loss": 0.0952,
+      "step": 13751
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000739949154490206,
+      "loss": 0.1036,
+      "step": 13752
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007398135692512514,
+      "loss": 0.0671,
+      "step": 13753
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007396779891425166,
+      "loss": 0.0743,
+      "step": 13754
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007395424141666746,
+      "loss": 0.0828,
+      "step": 13755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007394068443263986,
+      "loss": 0.0695,
+      "step": 13756
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007392712796243613,
+      "loss": 0.0709,
+      "step": 13757
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007391357200632364,
+      "loss": 0.0918,
+      "step": 13758
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007390001656456962,
+      "loss": 0.0691,
+      "step": 13759
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007388646163744135,
+      "loss": 0.0887,
+      "step": 13760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007387290722520617,
+      "loss": 0.0518,
+      "step": 13761
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007385935332813124,
+      "loss": 0.093,
+      "step": 13762
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007384579994648389,
+      "loss": 0.1136,
+      "step": 13763
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007383224708053134,
+      "loss": 0.0642,
+      "step": 13764
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007381869473054079,
+      "loss": 0.0586,
+      "step": 13765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007380514289677945,
+      "loss": 0.0923,
+      "step": 13766
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007379159157951456,
+      "loss": 0.1316,
+      "step": 13767
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0007377804077901331,
+      "loss": 0.1238,
+      "step": 13768
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000737644904955429,
+      "loss": 0.0924,
+      "step": 13769
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000737509407293705,
+      "loss": 0.1154,
+      "step": 13770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007373739148076326,
+      "loss": 0.0771,
+      "step": 13771
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007372384274998833,
+      "loss": 0.0682,
+      "step": 13772
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007371029453731289,
+      "loss": 0.0665,
+      "step": 13773
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007369674684300404,
+      "loss": 0.1509,
+      "step": 13774
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007368319966732894,
+      "loss": 0.1075,
+      "step": 13775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007366965301055472,
+      "loss": 0.0775,
+      "step": 13776
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007365610687294845,
+      "loss": 0.0568,
+      "step": 13777
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007364256125477721,
+      "loss": 0.0957,
+      "step": 13778
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007362901615630814,
+      "loss": 0.108,
+      "step": 13779
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007361547157780824,
+      "loss": 0.0716,
+      "step": 13780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007360192751954464,
+      "loss": 0.0767,
+      "step": 13781
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007358838398178441,
+      "loss": 0.098,
+      "step": 13782
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007357484096479452,
+      "loss": 0.0715,
+      "step": 13783
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007356129846884206,
+      "loss": 0.1141,
+      "step": 13784
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007354775649419406,
+      "loss": 0.0947,
+      "step": 13785
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007353421504111748,
+      "loss": 0.073,
+      "step": 13786
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007352067410987933,
+      "loss": 0.0641,
+      "step": 13787
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007350713370074666,
+      "loss": 0.0846,
+      "step": 13788
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000734935938139864,
+      "loss": 0.0692,
+      "step": 13789
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007348005444986558,
+      "loss": 0.073,
+      "step": 13790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007346651560865109,
+      "loss": 0.0638,
+      "step": 13791
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007345297729060996,
+      "loss": 0.1249,
+      "step": 13792
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007343943949600901,
+      "loss": 0.0715,
+      "step": 13793
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007342590222511531,
+      "loss": 0.0798,
+      "step": 13794
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007341236547819568,
+      "loss": 0.0723,
+      "step": 13795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000733988292555171,
+      "loss": 0.1053,
+      "step": 13796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007338529355734644,
+      "loss": 0.1249,
+      "step": 13797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000733717583839506,
+      "loss": 0.1487,
+      "step": 13798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007335822373559642,
+      "loss": 0.0946,
+      "step": 13799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007334468961255077,
+      "loss": 0.093,
+      "step": 13800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007333115601508058,
+      "loss": 0.115,
+      "step": 13801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007331762294345266,
+      "loss": 0.1161,
+      "step": 13802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007330409039793384,
+      "loss": 0.0789,
+      "step": 13803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007329055837879093,
+      "loss": 0.1102,
+      "step": 13804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000732770268862908,
+      "loss": 0.0809,
+      "step": 13805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007326349592070021,
+      "loss": 0.0691,
+      "step": 13806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007324996548228594,
+      "loss": 0.0715,
+      "step": 13807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007323643557131485,
+      "loss": 0.0857,
+      "step": 13808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007322290618805368,
+      "loss": 0.0803,
+      "step": 13809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007320937733276914,
+      "loss": 0.0641,
+      "step": 13810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007319584900572808,
+      "loss": 0.0951,
+      "step": 13811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007318232120719719,
+      "loss": 0.1272,
+      "step": 13812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007316879393744316,
+      "loss": 0.1044,
+      "step": 13813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007315526719673282,
+      "loss": 0.1019,
+      "step": 13814
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0007314174098533282,
+      "loss": 0.0593,
+      "step": 13815
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007312821530350986,
+      "loss": 0.0889,
+      "step": 13816
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007311469015153065,
+      "loss": 0.0723,
+      "step": 13817
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007310116552966189,
+      "loss": 0.0717,
+      "step": 13818
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007308764143817018,
+      "loss": 0.07,
+      "step": 13819
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007307411787732223,
+      "loss": 0.0671,
+      "step": 13820
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000730605948473847,
+      "loss": 0.0925,
+      "step": 13821
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007304707234862425,
+      "loss": 0.063,
+      "step": 13822
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007303355038130745,
+      "loss": 0.0952,
+      "step": 13823
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007302002894570097,
+      "loss": 0.0701,
+      "step": 13824
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007300650804207136,
+      "loss": 0.0978,
+      "step": 13825
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007299298767068529,
+      "loss": 0.1277,
+      "step": 13826
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007297946783180925,
+      "loss": 0.0586,
+      "step": 13827
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007296594852570993,
+      "loss": 0.0677,
+      "step": 13828
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007295242975265384,
+      "loss": 0.0704,
+      "step": 13829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007293891151290754,
+      "loss": 0.0801,
+      "step": 13830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007292539380673757,
+      "loss": 0.1024,
+      "step": 13831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007291187663441048,
+      "loss": 0.077,
+      "step": 13832
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007289835999619275,
+      "loss": 0.1184,
+      "step": 13833
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007288484389235097,
+      "loss": 0.1333,
+      "step": 13834
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000728713283231516,
+      "loss": 0.1237,
+      "step": 13835
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007285781328886112,
+      "loss": 0.0637,
+      "step": 13836
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007284429878974605,
+      "loss": 0.1543,
+      "step": 13837
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007283078482607282,
+      "loss": 0.092,
+      "step": 13838
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007281727139810795,
+      "loss": 0.0869,
+      "step": 13839
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007280375850611777,
+      "loss": 0.1053,
+      "step": 13840
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007279024615036887,
+      "loss": 0.0903,
+      "step": 13841
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007277673433112759,
+      "loss": 0.1045,
+      "step": 13842
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007276322304866038,
+      "loss": 0.12,
+      "step": 13843
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007274971230323364,
+      "loss": 0.0612,
+      "step": 13844
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007273620209511378,
+      "loss": 0.0765,
+      "step": 13845
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007272269242456715,
+      "loss": 0.0688,
+      "step": 13846
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007270918329186014,
+      "loss": 0.0666,
+      "step": 13847
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007269567469725913,
+      "loss": 0.0657,
+      "step": 13848
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000726821666410305,
+      "loss": 0.0605,
+      "step": 13849
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007266865912344057,
+      "loss": 0.0654,
+      "step": 13850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007265515214475567,
+      "loss": 0.0928,
+      "step": 13851
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000726416457052421,
+      "loss": 0.1582,
+      "step": 13852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007262813980516616,
+      "loss": 0.1478,
+      "step": 13853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007261463444479426,
+      "loss": 0.0962,
+      "step": 13854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000726011296243926,
+      "loss": 0.1022,
+      "step": 13855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000725876253442275,
+      "loss": 0.1572,
+      "step": 13856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007257412160456518,
+      "loss": 0.1018,
+      "step": 13857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007256061840567195,
+      "loss": 0.0876,
+      "step": 13858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007254711574781405,
+      "loss": 0.105,
+      "step": 13859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007253361363125766,
+      "loss": 0.0571,
+      "step": 13860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007252011205626911,
+      "loss": 0.1189,
+      "step": 13861
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0007250661102311455,
+      "loss": 0.094,
+      "step": 13862
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007249311053206019,
+      "loss": 0.1178,
+      "step": 13863
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007247961058337225,
+      "loss": 0.0721,
+      "step": 13864
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007246611117731692,
+      "loss": 0.0802,
+      "step": 13865
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007245261231416033,
+      "loss": 0.0914,
+      "step": 13866
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007243911399416865,
+      "loss": 0.0779,
+      "step": 13867
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007242561621760807,
+      "loss": 0.0891,
+      "step": 13868
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007241211898474471,
+      "loss": 0.0777,
+      "step": 13869
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007239862229584471,
+      "loss": 0.0839,
+      "step": 13870
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000723851261511742,
+      "loss": 0.0679,
+      "step": 13871
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007237163055099926,
+      "loss": 0.1418,
+      "step": 13872
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007235813549558599,
+      "loss": 0.1134,
+      "step": 13873
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000723446409852005,
+      "loss": 0.0583,
+      "step": 13874
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007233114702010888,
+      "loss": 0.0634,
+      "step": 13875
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007231765360057716,
+      "loss": 0.0798,
+      "step": 13876
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007230416072687144,
+      "loss": 0.0696,
+      "step": 13877
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000722906683992577,
+      "loss": 0.0627,
+      "step": 13878
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007227717661800205,
+      "loss": 0.0895,
+      "step": 13879
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007226368538337043,
+      "loss": 0.0801,
+      "step": 13880
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007225019469562895,
+      "loss": 0.0598,
+      "step": 13881
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007223670455504354,
+      "loss": 0.1281,
+      "step": 13882
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007222321496188023,
+      "loss": 0.1073,
+      "step": 13883
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007220972591640497,
+      "loss": 0.1069,
+      "step": 13884
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007219623741888376,
+      "loss": 0.1293,
+      "step": 13885
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007218274946958254,
+      "loss": 0.0738,
+      "step": 13886
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007216926206876723,
+      "loss": 0.1184,
+      "step": 13887
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007215577521670383,
+      "loss": 0.0874,
+      "step": 13888
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007214228891365824,
+      "loss": 0.0431,
+      "step": 13889
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007212880315989636,
+      "loss": 0.0983,
+      "step": 13890
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007211531795568413,
+      "loss": 0.0751,
+      "step": 13891
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007210183330128741,
+      "loss": 0.0716,
+      "step": 13892
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007208834919697206,
+      "loss": 0.1051,
+      "step": 13893
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007207486564300403,
+      "loss": 0.109,
+      "step": 13894
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007206138263964912,
+      "loss": 0.0658,
+      "step": 13895
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007204790018717322,
+      "loss": 0.1154,
+      "step": 13896
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007203441828584214,
+      "loss": 0.0923,
+      "step": 13897
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007202093693592173,
+      "loss": 0.0985,
+      "step": 13898
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007200745613767777,
+      "loss": 0.1221,
+      "step": 13899
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000719939758913761,
+      "loss": 0.1117,
+      "step": 13900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007198049619728251,
+      "loss": 0.0715,
+      "step": 13901
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000719670170556628,
+      "loss": 0.0448,
+      "step": 13902
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007195353846678274,
+      "loss": 0.0762,
+      "step": 13903
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007194006043090808,
+      "loss": 0.0659,
+      "step": 13904
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007192658294830456,
+      "loss": 0.0874,
+      "step": 13905
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007191310601923795,
+      "loss": 0.0587,
+      "step": 13906
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007189962964397394,
+      "loss": 0.0703,
+      "step": 13907
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0007188615382277832,
+      "loss": 0.1251,
+      "step": 13908
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007187267855591676,
+      "loss": 0.1014,
+      "step": 13909
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007185920384365496,
+      "loss": 0.1166,
+      "step": 13910
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007184572968625858,
+      "loss": 0.0796,
+      "step": 13911
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007183225608399335,
+      "loss": 0.1066,
+      "step": 13912
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007181878303712486,
+      "loss": 0.0851,
+      "step": 13913
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007180531054591885,
+      "loss": 0.1312,
+      "step": 13914
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007179183861064093,
+      "loss": 0.0684,
+      "step": 13915
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007177836723155671,
+      "loss": 0.0728,
+      "step": 13916
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007176489640893185,
+      "loss": 0.1251,
+      "step": 13917
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007175142614303193,
+      "loss": 0.1023,
+      "step": 13918
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007173795643412256,
+      "loss": 0.0834,
+      "step": 13919
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000717244872824693,
+      "loss": 0.1451,
+      "step": 13920
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007171101868833778,
+      "loss": 0.1292,
+      "step": 13921
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007169755065199353,
+      "loss": 0.0592,
+      "step": 13922
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007168408317370214,
+      "loss": 0.1189,
+      "step": 13923
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007167061625372914,
+      "loss": 0.0656,
+      "step": 13924
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007165714989234005,
+      "loss": 0.1027,
+      "step": 13925
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007164368408980038,
+      "loss": 0.105,
+      "step": 13926
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007163021884637564,
+      "loss": 0.0894,
+      "step": 13927
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007161675416233138,
+      "loss": 0.056,
+      "step": 13928
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007160329003793307,
+      "loss": 0.1167,
+      "step": 13929
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007158982647344619,
+      "loss": 0.0695,
+      "step": 13930
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007157636346913617,
+      "loss": 0.1044,
+      "step": 13931
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007156290102526851,
+      "loss": 0.0504,
+      "step": 13932
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007154943914210859,
+      "loss": 0.0753,
+      "step": 13933
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007153597781992194,
+      "loss": 0.0782,
+      "step": 13934
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007152251705897395,
+      "loss": 0.0782,
+      "step": 13935
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007150905685953,
+      "loss": 0.0719,
+      "step": 13936
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000714955972218555,
+      "loss": 0.0683,
+      "step": 13937
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007148213814621587,
+      "loss": 0.0473,
+      "step": 13938
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007146867963287646,
+      "loss": 0.089,
+      "step": 13939
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007145522168210262,
+      "loss": 0.0723,
+      "step": 13940
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007144176429415976,
+      "loss": 0.0985,
+      "step": 13941
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007142830746931321,
+      "loss": 0.1483,
+      "step": 13942
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007141485120782827,
+      "loss": 0.0946,
+      "step": 13943
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007140139550997031,
+      "loss": 0.0665,
+      "step": 13944
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007138794037600463,
+      "loss": 0.1138,
+      "step": 13945
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000713744858061965,
+      "loss": 0.0469,
+      "step": 13946
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000713610318008112,
+      "loss": 0.0975,
+      "step": 13947
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007134757836011406,
+      "loss": 0.1464,
+      "step": 13948
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007133412548437035,
+      "loss": 0.083,
+      "step": 13949
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000713206731738453,
+      "loss": 0.0892,
+      "step": 13950
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007130722142880417,
+      "loss": 0.1357,
+      "step": 13951
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007129377024951216,
+      "loss": 0.0573,
+      "step": 13952
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007128031963623451,
+      "loss": 0.1108,
+      "step": 13953
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0007126686958923645,
+      "loss": 0.0826,
+      "step": 13954
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007125342010878317,
+      "loss": 0.0697,
+      "step": 13955
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007123997119513989,
+      "loss": 0.0718,
+      "step": 13956
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007122652284857176,
+      "loss": 0.1084,
+      "step": 13957
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007121307506934391,
+      "loss": 0.0793,
+      "step": 13958
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007119962785772156,
+      "loss": 0.0904,
+      "step": 13959
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007118618121396978,
+      "loss": 0.0851,
+      "step": 13960
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000711727351383538,
+      "loss": 0.1265,
+      "step": 13961
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007115928963113869,
+      "loss": 0.1144,
+      "step": 13962
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007114584469258958,
+      "loss": 0.0831,
+      "step": 13963
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007113240032297153,
+      "loss": 0.076,
+      "step": 13964
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007111895652254967,
+      "loss": 0.0832,
+      "step": 13965
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007110551329158903,
+      "loss": 0.0741,
+      "step": 13966
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007109207063035474,
+      "loss": 0.0685,
+      "step": 13967
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007107862853911183,
+      "loss": 0.0914,
+      "step": 13968
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007106518701812532,
+      "loss": 0.072,
+      "step": 13969
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007105174606766029,
+      "loss": 0.1194,
+      "step": 13970
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007103830568798173,
+      "loss": 0.1113,
+      "step": 13971
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007102486587935466,
+      "loss": 0.103,
+      "step": 13972
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007101142664204402,
+      "loss": 0.0792,
+      "step": 13973
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000709979879763149,
+      "loss": 0.0473,
+      "step": 13974
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007098454988243219,
+      "loss": 0.0644,
+      "step": 13975
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007097111236066093,
+      "loss": 0.0836,
+      "step": 13976
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007095767541126604,
+      "loss": 0.0741,
+      "step": 13977
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007094423903451246,
+      "loss": 0.0808,
+      "step": 13978
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000709308032306651,
+      "loss": 0.0809,
+      "step": 13979
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007091736799998889,
+      "loss": 0.0491,
+      "step": 13980
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007090393334274876,
+      "loss": 0.066,
+      "step": 13981
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007089049925920961,
+      "loss": 0.1207,
+      "step": 13982
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007087706574963633,
+      "loss": 0.0889,
+      "step": 13983
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007086363281429375,
+      "loss": 0.1005,
+      "step": 13984
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007085020045344677,
+      "loss": 0.0538,
+      "step": 13985
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000708367686673602,
+      "loss": 0.0642,
+      "step": 13986
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007082333745629896,
+      "loss": 0.0853,
+      "step": 13987
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007080990682052784,
+      "loss": 0.0356,
+      "step": 13988
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007079647676031164,
+      "loss": 0.0933,
+      "step": 13989
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007078304727591518,
+      "loss": 0.0772,
+      "step": 13990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007076961836760326,
+      "loss": 0.0991,
+      "step": 13991
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007075619003564066,
+      "loss": 0.0696,
+      "step": 13992
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007074276228029212,
+      "loss": 0.0769,
+      "step": 13993
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007072933510182247,
+      "loss": 0.0434,
+      "step": 13994
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007071590850049643,
+      "loss": 0.106,
+      "step": 13995
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000707024824765787,
+      "loss": 0.121,
+      "step": 13996
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007068905703033407,
+      "loss": 0.0938,
+      "step": 13997
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007067563216202722,
+      "loss": 0.067,
+      "step": 13998
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007066220787192287,
+      "loss": 0.1218,
+      "step": 13999
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0007064878416028565,
+      "loss": 0.0621,
+      "step": 14000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007063536102738036,
+      "loss": 0.1309,
+      "step": 14001
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007062193847347156,
+      "loss": 0.0914,
+      "step": 14002
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007060851649882398,
+      "loss": 0.0783,
+      "step": 14003
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007059509510370224,
+      "loss": 0.0704,
+      "step": 14004
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007058167428837099,
+      "loss": 0.1074,
+      "step": 14005
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007056825405309479,
+      "loss": 0.1414,
+      "step": 14006
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007055483439813833,
+      "loss": 0.0864,
+      "step": 14007
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007054141532376622,
+      "loss": 0.055,
+      "step": 14008
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007052799683024299,
+      "loss": 0.1036,
+      "step": 14009
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007051457891783326,
+      "loss": 0.0717,
+      "step": 14010
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007050116158680156,
+      "loss": 0.0638,
+      "step": 14011
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007048774483741251,
+      "loss": 0.0645,
+      "step": 14012
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007047432866993055,
+      "loss": 0.0928,
+      "step": 14013
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007046091308462033,
+      "loss": 0.0591,
+      "step": 14014
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007044749808174631,
+      "loss": 0.0733,
+      "step": 14015
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007043408366157303,
+      "loss": 0.145,
+      "step": 14016
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007042066982436493,
+      "loss": 0.1219,
+      "step": 14017
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007040725657038654,
+      "loss": 0.0533,
+      "step": 14018
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007039384389990235,
+      "loss": 0.0736,
+      "step": 14019
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007038043181317675,
+      "loss": 0.0969,
+      "step": 14020
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007036702031047431,
+      "loss": 0.1489,
+      "step": 14021
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007035360939205936,
+      "loss": 0.108,
+      "step": 14022
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007034019905819641,
+      "loss": 0.0793,
+      "step": 14023
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007032678930914984,
+      "loss": 0.0549,
+      "step": 14024
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007031338014518405,
+      "loss": 0.0789,
+      "step": 14025
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007029997156656342,
+      "loss": 0.105,
+      "step": 14026
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007028656357355241,
+      "loss": 0.1355,
+      "step": 14027
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000702731561664153,
+      "loss": 0.1571,
+      "step": 14028
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007025974934541652,
+      "loss": 0.0502,
+      "step": 14029
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007024634311082039,
+      "loss": 0.0881,
+      "step": 14030
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007023293746289123,
+      "loss": 0.1267,
+      "step": 14031
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007021953240189338,
+      "loss": 0.0933,
+      "step": 14032
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007020612792809116,
+      "loss": 0.0555,
+      "step": 14033
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007019272404174885,
+      "loss": 0.0763,
+      "step": 14034
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007017932074313081,
+      "loss": 0.0874,
+      "step": 14035
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007016591803250123,
+      "loss": 0.098,
+      "step": 14036
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007015251591012444,
+      "loss": 0.0941,
+      "step": 14037
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007013911437626465,
+      "loss": 0.066,
+      "step": 14038
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007012571343118615,
+      "loss": 0.1276,
+      "step": 14039
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007011231307515309,
+      "loss": 0.0614,
+      "step": 14040
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007009891330842981,
+      "loss": 0.0558,
+      "step": 14041
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007008551413128046,
+      "loss": 0.0767,
+      "step": 14042
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007007211554396919,
+      "loss": 0.0978,
+      "step": 14043
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007005871754676028,
+      "loss": 0.1368,
+      "step": 14044
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007004532013991787,
+      "loss": 0.1509,
+      "step": 14045
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0007003192332370604,
+      "loss": 0.1041,
+      "step": 14046
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0007001852709838907,
+      "loss": 0.1097,
+      "step": 14047
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0007000513146423105,
+      "loss": 0.0757,
+      "step": 14048
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006999173642149608,
+      "loss": 0.1136,
+      "step": 14049
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000699783419704483,
+      "loss": 0.0792,
+      "step": 14050
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006996494811135184,
+      "loss": 0.0465,
+      "step": 14051
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006995155484447075,
+      "loss": 0.092,
+      "step": 14052
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006993816217006907,
+      "loss": 0.096,
+      "step": 14053
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006992477008841099,
+      "loss": 0.086,
+      "step": 14054
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006991137859976048,
+      "loss": 0.0657,
+      "step": 14055
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006989798770438162,
+      "loss": 0.092,
+      "step": 14056
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006988459740253846,
+      "loss": 0.1067,
+      "step": 14057
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006987120769449497,
+      "loss": 0.0758,
+      "step": 14058
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006985781858051519,
+      "loss": 0.0683,
+      "step": 14059
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000698444300608631,
+      "loss": 0.0792,
+      "step": 14060
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006983104213580274,
+      "loss": 0.097,
+      "step": 14061
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006981765480559805,
+      "loss": 0.0764,
+      "step": 14062
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006980426807051299,
+      "loss": 0.0927,
+      "step": 14063
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006979088193081151,
+      "loss": 0.1342,
+      "step": 14064
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006977749638675759,
+      "loss": 0.0791,
+      "step": 14065
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006976411143861508,
+      "loss": 0.0906,
+      "step": 14066
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006975072708664798,
+      "loss": 0.0909,
+      "step": 14067
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006973734333112019,
+      "loss": 0.1055,
+      "step": 14068
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006972396017229555,
+      "loss": 0.1088,
+      "step": 14069
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006971057761043795,
+      "loss": 0.0384,
+      "step": 14070
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006969719564581131,
+      "loss": 0.0942,
+      "step": 14071
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006968381427867947,
+      "loss": 0.064,
+      "step": 14072
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006967043350930619,
+      "loss": 0.0663,
+      "step": 14073
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006965705333795546,
+      "loss": 0.0894,
+      "step": 14074
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00069643673764891,
+      "loss": 0.0841,
+      "step": 14075
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006963029479037663,
+      "loss": 0.0463,
+      "step": 14076
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006961691641467618,
+      "loss": 0.0751,
+      "step": 14077
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006960353863805342,
+      "loss": 0.092,
+      "step": 14078
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006959016146077213,
+      "loss": 0.0868,
+      "step": 14079
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006957678488309604,
+      "loss": 0.1245,
+      "step": 14080
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006956340890528895,
+      "loss": 0.1058,
+      "step": 14081
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000695500335276146,
+      "loss": 0.0572,
+      "step": 14082
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.000695366587503367,
+      "loss": 0.089,
+      "step": 14083
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00069523284573719,
+      "loss": 0.141,
+      "step": 14084
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006950991099802512,
+      "loss": 0.1294,
+      "step": 14085
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006949653802351881,
+      "loss": 0.0741,
+      "step": 14086
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006948316565046377,
+      "loss": 0.1145,
+      "step": 14087
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006946979387912366,
+      "loss": 0.1105,
+      "step": 14088
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006945642270976213,
+      "loss": 0.0775,
+      "step": 14089
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006944305214264282,
+      "loss": 0.088,
+      "step": 14090
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006942968217802934,
+      "loss": 0.08,
+      "step": 14091
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0006941631281618536,
+      "loss": 0.1089,
+      "step": 14092
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006940294405737441,
+      "loss": 0.0851,
+      "step": 14093
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006938957590186022,
+      "loss": 0.0529,
+      "step": 14094
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000693762083499063,
+      "loss": 0.119,
+      "step": 14095
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006936284140177621,
+      "loss": 0.0915,
+      "step": 14096
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006934947505773352,
+      "loss": 0.0852,
+      "step": 14097
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000693361093180418,
+      "loss": 0.0594,
+      "step": 14098
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006932274418296459,
+      "loss": 0.1292,
+      "step": 14099
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006930937965276536,
+      "loss": 0.0836,
+      "step": 14100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006929601572770772,
+      "loss": 0.1309,
+      "step": 14101
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006928265240805508,
+      "loss": 0.1428,
+      "step": 14102
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006926928969407101,
+      "loss": 0.136,
+      "step": 14103
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006925592758601895,
+      "loss": 0.065,
+      "step": 14104
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006924256608416237,
+      "loss": 0.0835,
+      "step": 14105
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006922920518876469,
+      "loss": 0.0992,
+      "step": 14106
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006921584490008942,
+      "loss": 0.0947,
+      "step": 14107
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006920248521839994,
+      "loss": 0.074,
+      "step": 14108
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006918912614395971,
+      "loss": 0.1143,
+      "step": 14109
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006917576767703213,
+      "loss": 0.1502,
+      "step": 14110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006916240981788056,
+      "loss": 0.0835,
+      "step": 14111
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000691490525667684,
+      "loss": 0.0897,
+      "step": 14112
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006913569592395901,
+      "loss": 0.0983,
+      "step": 14113
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006912233988971577,
+      "loss": 0.1702,
+      "step": 14114
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006910898446430205,
+      "loss": 0.0963,
+      "step": 14115
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006909562964798115,
+      "loss": 0.1049,
+      "step": 14116
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006908227544101639,
+      "loss": 0.0726,
+      "step": 14117
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006906892184367111,
+      "loss": 0.1052,
+      "step": 14118
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006905556885620859,
+      "loss": 0.1303,
+      "step": 14119
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006904221647889206,
+      "loss": 0.0715,
+      "step": 14120
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006902886471198491,
+      "loss": 0.0468,
+      "step": 14121
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006901551355575035,
+      "loss": 0.0728,
+      "step": 14122
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000690021630104516,
+      "loss": 0.1359,
+      "step": 14123
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006898881307635194,
+      "loss": 0.0833,
+      "step": 14124
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000689754637537146,
+      "loss": 0.1082,
+      "step": 14125
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006896211504280273,
+      "loss": 0.0798,
+      "step": 14126
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006894876694387962,
+      "loss": 0.1786,
+      "step": 14127
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006893541945720842,
+      "loss": 0.0715,
+      "step": 14128
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000689220725830523,
+      "loss": 0.1498,
+      "step": 14129
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006890872632167446,
+      "loss": 0.0531,
+      "step": 14130
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006889538067333803,
+      "loss": 0.0679,
+      "step": 14131
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006888203563830616,
+      "loss": 0.1279,
+      "step": 14132
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.000688686912168419,
+      "loss": 0.0612,
+      "step": 14133
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006885534740920849,
+      "loss": 0.0717,
+      "step": 14134
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006884200421566902,
+      "loss": 0.0834,
+      "step": 14135
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006882866163648654,
+      "loss": 0.1031,
+      "step": 14136
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006881531967192415,
+      "loss": 0.0787,
+      "step": 14137
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0006880197832224488,
+      "loss": 0.0681,
+      "step": 14138
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006878863758771187,
+      "loss": 0.1047,
+      "step": 14139
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006877529746858806,
+      "loss": 0.0671,
+      "step": 14140
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006876195796513659,
+      "loss": 0.0914,
+      "step": 14141
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006874861907762043,
+      "loss": 0.0651,
+      "step": 14142
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000687352808063026,
+      "loss": 0.0778,
+      "step": 14143
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006872194315144606,
+      "loss": 0.073,
+      "step": 14144
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006870860611331384,
+      "loss": 0.0862,
+      "step": 14145
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006869526969216888,
+      "loss": 0.0667,
+      "step": 14146
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006868193388827417,
+      "loss": 0.1339,
+      "step": 14147
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006866859870189267,
+      "loss": 0.0749,
+      "step": 14148
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006865526413328729,
+      "loss": 0.0798,
+      "step": 14149
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006864193018272092,
+      "loss": 0.16,
+      "step": 14150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006862859685045655,
+      "loss": 0.0812,
+      "step": 14151
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006861526413675702,
+      "loss": 0.077,
+      "step": 14152
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006860193204188521,
+      "loss": 0.1092,
+      "step": 14153
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006858860056610405,
+      "loss": 0.1125,
+      "step": 14154
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006857526970967634,
+      "loss": 0.1063,
+      "step": 14155
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006856193947286501,
+      "loss": 0.1329,
+      "step": 14156
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006854860985593286,
+      "loss": 0.1422,
+      "step": 14157
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006853528085914267,
+      "loss": 0.0701,
+      "step": 14158
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006852195248275732,
+      "loss": 0.0724,
+      "step": 14159
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006850862472703956,
+      "loss": 0.084,
+      "step": 14160
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000684952975922522,
+      "loss": 0.0906,
+      "step": 14161
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006848197107865806,
+      "loss": 0.1316,
+      "step": 14162
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006846864518651985,
+      "loss": 0.0747,
+      "step": 14163
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006845531991610035,
+      "loss": 0.1335,
+      "step": 14164
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006844199526766228,
+      "loss": 0.0836,
+      "step": 14165
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006842867124146837,
+      "loss": 0.0888,
+      "step": 14166
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006841534783778136,
+      "loss": 0.1187,
+      "step": 14167
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006840202505686394,
+      "loss": 0.0577,
+      "step": 14168
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006838870289897882,
+      "loss": 0.1078,
+      "step": 14169
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006837538136438866,
+      "loss": 0.1002,
+      "step": 14170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006836206045335609,
+      "loss": 0.0716,
+      "step": 14171
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006834874016614383,
+      "loss": 0.1048,
+      "step": 14172
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006833542050301447,
+      "loss": 0.0778,
+      "step": 14173
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006832210146423068,
+      "loss": 0.0947,
+      "step": 14174
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006830878305005508,
+      "loss": 0.0703,
+      "step": 14175
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006829546526075024,
+      "loss": 0.0961,
+      "step": 14176
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006828214809657879,
+      "loss": 0.1016,
+      "step": 14177
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006826883155780329,
+      "loss": 0.0974,
+      "step": 14178
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.000682555156446863,
+      "loss": 0.0923,
+      "step": 14179
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006824220035749037,
+      "loss": 0.0787,
+      "step": 14180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006822888569647809,
+      "loss": 0.0651,
+      "step": 14181
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006821557166191194,
+      "loss": 0.0856,
+      "step": 14182
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006820225825405449,
+      "loss": 0.0918,
+      "step": 14183
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0006818894547316822,
+      "loss": 0.1119,
+      "step": 14184
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006817563331951562,
+      "loss": 0.1191,
+      "step": 14185
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006816232179335913,
+      "loss": 0.0747,
+      "step": 14186
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006814901089496131,
+      "loss": 0.1317,
+      "step": 14187
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006813570062458454,
+      "loss": 0.0515,
+      "step": 14188
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006812239098249133,
+      "loss": 0.1339,
+      "step": 14189
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006810908196894408,
+      "loss": 0.1453,
+      "step": 14190
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000680957735842052,
+      "loss": 0.0995,
+      "step": 14191
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000680824658285371,
+      "loss": 0.0593,
+      "step": 14192
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006806915870220216,
+      "loss": 0.1605,
+      "step": 14193
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006805585220546282,
+      "loss": 0.1243,
+      "step": 14194
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006804254633858142,
+      "loss": 0.1171,
+      "step": 14195
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006802924110182032,
+      "loss": 0.0964,
+      "step": 14196
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006801593649544182,
+      "loss": 0.0785,
+      "step": 14197
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006800263251970833,
+      "loss": 0.1289,
+      "step": 14198
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006798932917488214,
+      "loss": 0.1012,
+      "step": 14199
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006797602646122549,
+      "loss": 0.1544,
+      "step": 14200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006796272437900081,
+      "loss": 0.0698,
+      "step": 14201
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006794942292847029,
+      "loss": 0.0671,
+      "step": 14202
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000679361221098962,
+      "loss": 0.1078,
+      "step": 14203
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006792282192354085,
+      "loss": 0.0798,
+      "step": 14204
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006790952236966645,
+      "loss": 0.0672,
+      "step": 14205
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006789622344853521,
+      "loss": 0.0704,
+      "step": 14206
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006788292516040943,
+      "loss": 0.113,
+      "step": 14207
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006786962750555126,
+      "loss": 0.0764,
+      "step": 14208
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006785633048422288,
+      "loss": 0.0904,
+      "step": 14209
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006784303409668653,
+      "loss": 0.0779,
+      "step": 14210
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006782973834320436,
+      "loss": 0.105,
+      "step": 14211
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006781644322403848,
+      "loss": 0.0988,
+      "step": 14212
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006780314873945111,
+      "loss": 0.0696,
+      "step": 14213
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006778985488970432,
+      "loss": 0.0819,
+      "step": 14214
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006777656167506029,
+      "loss": 0.0828,
+      "step": 14215
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000677632690957811,
+      "loss": 0.0742,
+      "step": 14216
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006774997715212885,
+      "loss": 0.1101,
+      "step": 14217
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000677366858443656,
+      "loss": 0.0983,
+      "step": 14218
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006772339517275345,
+      "loss": 0.0686,
+      "step": 14219
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006771010513755443,
+      "loss": 0.0983,
+      "step": 14220
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006769681573903064,
+      "loss": 0.0615,
+      "step": 14221
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006768352697744406,
+      "loss": 0.0836,
+      "step": 14222
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006767023885305675,
+      "loss": 0.0908,
+      "step": 14223
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006765695136613067,
+      "loss": 0.1411,
+      "step": 14224
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006764366451692786,
+      "loss": 0.0764,
+      "step": 14225
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006763037830571026,
+      "loss": 0.0956,
+      "step": 14226
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006761709273273989,
+      "loss": 0.0656,
+      "step": 14227
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.000676038077982787,
+      "loss": 0.0787,
+      "step": 14228
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006759052350258859,
+      "loss": 0.0979,
+      "step": 14229
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006757723984593155,
+      "loss": 0.1119,
+      "step": 14230
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0006756395682856948,
+      "loss": 0.0552,
+      "step": 14231
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006755067445076426,
+      "loss": 0.061,
+      "step": 14232
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006753739271277779,
+      "loss": 0.0966,
+      "step": 14233
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00067524111614872,
+      "loss": 0.0817,
+      "step": 14234
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000675108311573087,
+      "loss": 0.0452,
+      "step": 14235
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006749755134034981,
+      "loss": 0.1686,
+      "step": 14236
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006748427216425713,
+      "loss": 0.1426,
+      "step": 14237
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000674709936292925,
+      "loss": 0.1294,
+      "step": 14238
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000674577157357177,
+      "loss": 0.1237,
+      "step": 14239
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006744443848379461,
+      "loss": 0.1946,
+      "step": 14240
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006743116187378498,
+      "loss": 0.1007,
+      "step": 14241
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006741788590595063,
+      "loss": 0.0894,
+      "step": 14242
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006740461058055329,
+      "loss": 0.1261,
+      "step": 14243
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006739133589785474,
+      "loss": 0.0695,
+      "step": 14244
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006737806185811668,
+      "loss": 0.1157,
+      "step": 14245
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006736478846160085,
+      "loss": 0.1653,
+      "step": 14246
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006735151570856903,
+      "loss": 0.0812,
+      "step": 14247
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006733824359928288,
+      "loss": 0.1324,
+      "step": 14248
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000673249721340041,
+      "loss": 0.1299,
+      "step": 14249
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006731170131299434,
+      "loss": 0.1337,
+      "step": 14250
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006729843113651532,
+      "loss": 0.0652,
+      "step": 14251
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006728516160482867,
+      "loss": 0.0657,
+      "step": 14252
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006727189271819598,
+      "loss": 0.1139,
+      "step": 14253
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006725862447687898,
+      "loss": 0.1644,
+      "step": 14254
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006724535688113923,
+      "loss": 0.0977,
+      "step": 14255
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000672320899312383,
+      "loss": 0.114,
+      "step": 14256
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006721882362743786,
+      "loss": 0.1166,
+      "step": 14257
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006720555796999946,
+      "loss": 0.1124,
+      "step": 14258
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006719229295918457,
+      "loss": 0.0732,
+      "step": 14259
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006717902859525489,
+      "loss": 0.0927,
+      "step": 14260
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006716576487847189,
+      "loss": 0.0516,
+      "step": 14261
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006715250180909709,
+      "loss": 0.1156,
+      "step": 14262
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006713923938739202,
+      "loss": 0.0754,
+      "step": 14263
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006712597761361818,
+      "loss": 0.0734,
+      "step": 14264
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006711271648803706,
+      "loss": 0.0828,
+      "step": 14265
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006709945601091009,
+      "loss": 0.0898,
+      "step": 14266
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000670861961824988,
+      "loss": 0.0769,
+      "step": 14267
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006707293700306462,
+      "loss": 0.0723,
+      "step": 14268
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006705967847286899,
+      "loss": 0.1152,
+      "step": 14269
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006704642059217333,
+      "loss": 0.0767,
+      "step": 14270
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006703316336123903,
+      "loss": 0.0713,
+      "step": 14271
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006701990678032753,
+      "loss": 0.104,
+      "step": 14272
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006700665084970014,
+      "loss": 0.0814,
+      "step": 14273
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006699339556961834,
+      "loss": 0.1269,
+      "step": 14274
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006698014094034344,
+      "loss": 0.0682,
+      "step": 14275
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006696688696213678,
+      "loss": 0.0945,
+      "step": 14276
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0006695363363525968,
+      "loss": 0.1343,
+      "step": 14277
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000669403809599735,
+      "loss": 0.118,
+      "step": 14278
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006692712893653951,
+      "loss": 0.1066,
+      "step": 14279
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006691387756521905,
+      "loss": 0.0834,
+      "step": 14280
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006690062684627339,
+      "loss": 0.0868,
+      "step": 14281
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000668873767799638,
+      "loss": 0.072,
+      "step": 14282
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006687412736655151,
+      "loss": 0.0645,
+      "step": 14283
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000668608786062978,
+      "loss": 0.075,
+      "step": 14284
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000668476304994639,
+      "loss": 0.0764,
+      "step": 14285
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006683438304631096,
+      "loss": 0.1053,
+      "step": 14286
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006682113624710029,
+      "loss": 0.0635,
+      "step": 14287
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006680789010209303,
+      "loss": 0.0792,
+      "step": 14288
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006679464461155037,
+      "loss": 0.098,
+      "step": 14289
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006678139977573348,
+      "loss": 0.1012,
+      "step": 14290
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006676815559490352,
+      "loss": 0.0539,
+      "step": 14291
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000667549120693216,
+      "loss": 0.06,
+      "step": 14292
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006674166919924887,
+      "loss": 0.0612,
+      "step": 14293
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006672842698494644,
+      "loss": 0.133,
+      "step": 14294
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006671518542667544,
+      "loss": 0.0706,
+      "step": 14295
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006670194452469694,
+      "loss": 0.0734,
+      "step": 14296
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006668870427927202,
+      "loss": 0.0673,
+      "step": 14297
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006667546469066172,
+      "loss": 0.1294,
+      "step": 14298
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006666222575912713,
+      "loss": 0.1053,
+      "step": 14299
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006664898748492924,
+      "loss": 0.098,
+      "step": 14300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006663574986832915,
+      "loss": 0.0996,
+      "step": 14301
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006662251290958783,
+      "loss": 0.1398,
+      "step": 14302
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006660927660896626,
+      "loss": 0.1063,
+      "step": 14303
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006659604096672545,
+      "loss": 0.1202,
+      "step": 14304
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006658280598312636,
+      "loss": 0.0931,
+      "step": 14305
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006656957165842993,
+      "loss": 0.0706,
+      "step": 14306
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006655633799289717,
+      "loss": 0.0745,
+      "step": 14307
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00066543104986789,
+      "loss": 0.0912,
+      "step": 14308
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006652987264036627,
+      "loss": 0.0956,
+      "step": 14309
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006651664095388999,
+      "loss": 0.0578,
+      "step": 14310
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00066503409927621,
+      "loss": 0.0858,
+      "step": 14311
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006649017956182017,
+      "loss": 0.1267,
+      "step": 14312
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006647694985674837,
+      "loss": 0.1069,
+      "step": 14313
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000664637208126665,
+      "loss": 0.069,
+      "step": 14314
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006645049242983537,
+      "loss": 0.0626,
+      "step": 14315
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006643726470851583,
+      "loss": 0.1338,
+      "step": 14316
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006642403764896867,
+      "loss": 0.0981,
+      "step": 14317
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006641081125145474,
+      "loss": 0.1028,
+      "step": 14318
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006639758551623474,
+      "loss": 0.0911,
+      "step": 14319
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006638436044356953,
+      "loss": 0.0842,
+      "step": 14320
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006637113603371987,
+      "loss": 0.0699,
+      "step": 14321
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000663579122869465,
+      "loss": 0.0967,
+      "step": 14322
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0006634468920351015,
+      "loss": 0.1534,
+      "step": 14323
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006633146678367154,
+      "loss": 0.0876,
+      "step": 14324
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006631824502769142,
+      "loss": 0.0889,
+      "step": 14325
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000663050239358304,
+      "loss": 0.0821,
+      "step": 14326
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006629180350834928,
+      "loss": 0.1072,
+      "step": 14327
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006627858374550868,
+      "loss": 0.0672,
+      "step": 14328
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006626536464756927,
+      "loss": 0.0624,
+      "step": 14329
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006625214621479166,
+      "loss": 0.1212,
+      "step": 14330
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006623892844743654,
+      "loss": 0.0681,
+      "step": 14331
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000662257113457645,
+      "loss": 0.0768,
+      "step": 14332
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006621249491003613,
+      "loss": 0.0881,
+      "step": 14333
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006619927914051207,
+      "loss": 0.1552,
+      "step": 14334
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006618606403745289,
+      "loss": 0.1183,
+      "step": 14335
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006617284960111911,
+      "loss": 0.0679,
+      "step": 14336
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006615963583177136,
+      "loss": 0.1019,
+      "step": 14337
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006614642272967014,
+      "loss": 0.0917,
+      "step": 14338
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006613321029507595,
+      "loss": 0.0869,
+      "step": 14339
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006611999852824937,
+      "loss": 0.1038,
+      "step": 14340
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006610678742945086,
+      "loss": 0.0535,
+      "step": 14341
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006609357699894094,
+      "loss": 0.1406,
+      "step": 14342
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006608036723698006,
+      "loss": 0.1097,
+      "step": 14343
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006606715814382871,
+      "loss": 0.0826,
+      "step": 14344
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000660539497197473,
+      "loss": 0.0443,
+      "step": 14345
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006604074196499628,
+      "loss": 0.0985,
+      "step": 14346
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006602753487983608,
+      "loss": 0.0858,
+      "step": 14347
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006601432846452713,
+      "loss": 0.0523,
+      "step": 14348
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006600112271932982,
+      "loss": 0.1052,
+      "step": 14349
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006598791764450451,
+      "loss": 0.1096,
+      "step": 14350
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006597471324031158,
+      "loss": 0.0848,
+      "step": 14351
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006596150950701141,
+      "loss": 0.0933,
+      "step": 14352
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006594830644486428,
+      "loss": 0.1591,
+      "step": 14353
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006593510405413062,
+      "loss": 0.0734,
+      "step": 14354
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006592190233507067,
+      "loss": 0.1121,
+      "step": 14355
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006590870128794478,
+      "loss": 0.1102,
+      "step": 14356
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000658955009130132,
+      "loss": 0.155,
+      "step": 14357
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006588230121053624,
+      "loss": 0.101,
+      "step": 14358
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006586910218077411,
+      "loss": 0.0712,
+      "step": 14359
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006585590382398716,
+      "loss": 0.0889,
+      "step": 14360
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006584270614043557,
+      "loss": 0.0585,
+      "step": 14361
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006582950913037954,
+      "loss": 0.1958,
+      "step": 14362
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006581631279407934,
+      "loss": 0.0629,
+      "step": 14363
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006580311713179513,
+      "loss": 0.1226,
+      "step": 14364
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006578992214378711,
+      "loss": 0.0832,
+      "step": 14365
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006577672783031539,
+      "loss": 0.1408,
+      "step": 14366
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006576353419164025,
+      "loss": 0.1194,
+      "step": 14367
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006575034122802174,
+      "loss": 0.1082,
+      "step": 14368
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0006573714893972003,
+      "loss": 0.0772,
+      "step": 14369
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006572395732699525,
+      "loss": 0.1281,
+      "step": 14370
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006571076639010747,
+      "loss": 0.1179,
+      "step": 14371
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006569757612931678,
+      "loss": 0.124,
+      "step": 14372
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006568438654488328,
+      "loss": 0.0544,
+      "step": 14373
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006567119763706702,
+      "loss": 0.148,
+      "step": 14374
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006565800940612809,
+      "loss": 0.1641,
+      "step": 14375
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006564482185232651,
+      "loss": 0.1394,
+      "step": 14376
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000656316349759223,
+      "loss": 0.0663,
+      "step": 14377
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006561844877717544,
+      "loss": 0.0597,
+      "step": 14378
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006560526325634593,
+      "loss": 0.0604,
+      "step": 14379
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006559207841369384,
+      "loss": 0.1278,
+      "step": 14380
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006557889424947907,
+      "loss": 0.0555,
+      "step": 14381
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006556571076396161,
+      "loss": 0.0451,
+      "step": 14382
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006555252795740135,
+      "loss": 0.1282,
+      "step": 14383
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006553934583005827,
+      "loss": 0.1343,
+      "step": 14384
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006552616438219229,
+      "loss": 0.0398,
+      "step": 14385
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006551298361406328,
+      "loss": 0.0465,
+      "step": 14386
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006549980352593118,
+      "loss": 0.0908,
+      "step": 14387
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006548662411805584,
+      "loss": 0.0828,
+      "step": 14388
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006547344539069711,
+      "loss": 0.098,
+      "step": 14389
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006546026734411488,
+      "loss": 0.0617,
+      "step": 14390
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006544708997856896,
+      "loss": 0.1466,
+      "step": 14391
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006543391329431919,
+      "loss": 0.0594,
+      "step": 14392
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006542073729162532,
+      "loss": 0.1337,
+      "step": 14393
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006540756197074726,
+      "loss": 0.0847,
+      "step": 14394
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000653943873319447,
+      "loss": 0.1796,
+      "step": 14395
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006538121337547745,
+      "loss": 0.0805,
+      "step": 14396
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006536804010160527,
+      "loss": 0.1571,
+      "step": 14397
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000653548675105879,
+      "loss": 0.0684,
+      "step": 14398
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006534169560268502,
+      "loss": 0.1406,
+      "step": 14399
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006532852437815641,
+      "loss": 0.1155,
+      "step": 14400
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006531535383726178,
+      "loss": 0.0808,
+      "step": 14401
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.000653021839802608,
+      "loss": 0.0683,
+      "step": 14402
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006528901480741312,
+      "loss": 0.1415,
+      "step": 14403
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006527584631897843,
+      "loss": 0.1173,
+      "step": 14404
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006526267851521637,
+      "loss": 0.1311,
+      "step": 14405
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006524951139638655,
+      "loss": 0.0831,
+      "step": 14406
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006523634496274866,
+      "loss": 0.0777,
+      "step": 14407
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006522317921456227,
+      "loss": 0.0917,
+      "step": 14408
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006521001415208698,
+      "loss": 0.0924,
+      "step": 14409
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006519684977558234,
+      "loss": 0.0601,
+      "step": 14410
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006518368608530796,
+      "loss": 0.0581,
+      "step": 14411
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006517052308152338,
+      "loss": 0.0588,
+      "step": 14412
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006515736076448812,
+      "loss": 0.1034,
+      "step": 14413
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006514419913446175,
+      "loss": 0.1317,
+      "step": 14414
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0006513103819170377,
+      "loss": 0.1049,
+      "step": 14415
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006511787793647366,
+      "loss": 0.1121,
+      "step": 14416
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006510471836903094,
+      "loss": 0.1119,
+      "step": 14417
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006509155948963506,
+      "loss": 0.0694,
+      "step": 14418
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006507840129854544,
+      "loss": 0.1061,
+      "step": 14419
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.000650652437960216,
+      "loss": 0.069,
+      "step": 14420
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006505208698232294,
+      "loss": 0.1206,
+      "step": 14421
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006503893085770889,
+      "loss": 0.1182,
+      "step": 14422
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006502577542243886,
+      "loss": 0.0646,
+      "step": 14423
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006501262067677223,
+      "loss": 0.054,
+      "step": 14424
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006499946662096835,
+      "loss": 0.1189,
+      "step": 14425
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006498631325528661,
+      "loss": 0.0765,
+      "step": 14426
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006497316057998637,
+      "loss": 0.106,
+      "step": 14427
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006496000859532699,
+      "loss": 0.1108,
+      "step": 14428
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006494685730156775,
+      "loss": 0.1992,
+      "step": 14429
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006493370669896799,
+      "loss": 0.0991,
+      "step": 14430
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006492055678778695,
+      "loss": 0.1238,
+      "step": 14431
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006490740756828399,
+      "loss": 0.0809,
+      "step": 14432
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006489425904071831,
+      "loss": 0.0779,
+      "step": 14433
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006488111120534923,
+      "loss": 0.0551,
+      "step": 14434
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006486796406243598,
+      "loss": 0.1083,
+      "step": 14435
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006485481761223774,
+      "loss": 0.0789,
+      "step": 14436
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006484167185501378,
+      "loss": 0.0784,
+      "step": 14437
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006482852679102327,
+      "loss": 0.0653,
+      "step": 14438
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006481538242052536,
+      "loss": 0.101,
+      "step": 14439
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.000648022387437793,
+      "loss": 0.0784,
+      "step": 14440
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006478909576104425,
+      "loss": 0.0756,
+      "step": 14441
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006477595347257929,
+      "loss": 0.0953,
+      "step": 14442
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006476281187864361,
+      "loss": 0.074,
+      "step": 14443
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.000647496709794963,
+      "loss": 0.0986,
+      "step": 14444
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006473653077539648,
+      "loss": 0.0908,
+      "step": 14445
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006472339126660318,
+      "loss": 0.0873,
+      "step": 14446
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006471025245337558,
+      "loss": 0.0623,
+      "step": 14447
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006469711433597267,
+      "loss": 0.0684,
+      "step": 14448
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006468397691465356,
+      "loss": 0.0859,
+      "step": 14449
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006467084018967724,
+      "loss": 0.0654,
+      "step": 14450
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006465770416130277,
+      "loss": 0.0759,
+      "step": 14451
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006464456882978908,
+      "loss": 0.074,
+      "step": 14452
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006463143419539523,
+      "loss": 0.0844,
+      "step": 14453
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006461830025838022,
+      "loss": 0.0873,
+      "step": 14454
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006460516701900299,
+      "loss": 0.1129,
+      "step": 14455
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006459203447752251,
+      "loss": 0.0681,
+      "step": 14456
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006457890263419767,
+      "loss": 0.0663,
+      "step": 14457
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006456577148928746,
+      "loss": 0.0723,
+      "step": 14458
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006455264104305072,
+      "loss": 0.0696,
+      "step": 14459
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006453951129574643,
+      "loss": 0.076,
+      "step": 14460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0006452638224763345,
+      "loss": 0.0612,
+      "step": 14461
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006451325389897065,
+      "loss": 0.0528,
+      "step": 14462
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006450012625001685,
+      "loss": 0.0778,
+      "step": 14463
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006448699930103094,
+      "loss": 0.0916,
+      "step": 14464
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006447387305227174,
+      "loss": 0.0862,
+      "step": 14465
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006446074750399802,
+      "loss": 0.0824,
+      "step": 14466
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006444762265646867,
+      "loss": 0.1064,
+      "step": 14467
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000644344985099424,
+      "loss": 0.0539,
+      "step": 14468
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006442137506467803,
+      "loss": 0.1499,
+      "step": 14469
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006440825232093433,
+      "loss": 0.0722,
+      "step": 14470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006439513027897001,
+      "loss": 0.0876,
+      "step": 14471
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006438200893904382,
+      "loss": 0.0651,
+      "step": 14472
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006436888830141446,
+      "loss": 0.1323,
+      "step": 14473
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006435576836634065,
+      "loss": 0.0709,
+      "step": 14474
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006434264913408112,
+      "loss": 0.0758,
+      "step": 14475
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006432953060489452,
+      "loss": 0.0862,
+      "step": 14476
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000643164127790395,
+      "loss": 0.0362,
+      "step": 14477
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006430329565677471,
+      "loss": 0.0595,
+      "step": 14478
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006429017923835879,
+      "loss": 0.1361,
+      "step": 14479
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006427706352405036,
+      "loss": 0.1499,
+      "step": 14480
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006426394851410806,
+      "loss": 0.0388,
+      "step": 14481
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006425083420879047,
+      "loss": 0.0777,
+      "step": 14482
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006423772060835616,
+      "loss": 0.0765,
+      "step": 14483
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006422460771306369,
+      "loss": 0.0799,
+      "step": 14484
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006421149552317163,
+      "loss": 0.0817,
+      "step": 14485
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006419838403893848,
+      "loss": 0.0864,
+      "step": 14486
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006418527326062284,
+      "loss": 0.1271,
+      "step": 14487
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006417216318848318,
+      "loss": 0.0745,
+      "step": 14488
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006415905382277799,
+      "loss": 0.0795,
+      "step": 14489
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006414594516376575,
+      "loss": 0.1277,
+      "step": 14490
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006413283721170496,
+      "loss": 0.1203,
+      "step": 14491
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006411972996685405,
+      "loss": 0.0803,
+      "step": 14492
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006410662342947143,
+      "loss": 0.0857,
+      "step": 14493
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006409351759981561,
+      "loss": 0.0897,
+      "step": 14494
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006408041247814493,
+      "loss": 0.0786,
+      "step": 14495
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006406730806471785,
+      "loss": 0.0798,
+      "step": 14496
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006405420435979274,
+      "loss": 0.0748,
+      "step": 14497
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006404110136362793,
+      "loss": 0.0625,
+      "step": 14498
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006402799907648178,
+      "loss": 0.0641,
+      "step": 14499
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.000640148974986127,
+      "loss": 0.0778,
+      "step": 14500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006400179663027896,
+      "loss": 0.0596,
+      "step": 14501
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006398869647173891,
+      "loss": 0.1021,
+      "step": 14502
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006397559702325085,
+      "loss": 0.0858,
+      "step": 14503
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006396249828507304,
+      "loss": 0.1248,
+      "step": 14504
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006394940025746377,
+      "loss": 0.0881,
+      "step": 14505
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006393630294068127,
+      "loss": 0.0909,
+      "step": 14506
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0006392320633498383,
+      "loss": 0.0859,
+      "step": 14507
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.000639101104406297,
+      "loss": 0.1421,
+      "step": 14508
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006389701525787706,
+      "loss": 0.0665,
+      "step": 14509
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006388392078698412,
+      "loss": 0.1655,
+      "step": 14510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006387082702820904,
+      "loss": 0.124,
+      "step": 14511
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006385773398181001,
+      "loss": 0.0895,
+      "step": 14512
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006384464164804525,
+      "loss": 0.1206,
+      "step": 14513
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006383155002717286,
+      "loss": 0.0834,
+      "step": 14514
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00063818459119451,
+      "loss": 0.1081,
+      "step": 14515
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006380536892513773,
+      "loss": 0.0824,
+      "step": 14516
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006379227944449122,
+      "loss": 0.1202,
+      "step": 14517
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006377919067776954,
+      "loss": 0.0731,
+      "step": 14518
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006376610262523071,
+      "loss": 0.0718,
+      "step": 14519
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006375301528713289,
+      "loss": 0.171,
+      "step": 14520
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006373992866373409,
+      "loss": 0.0774,
+      "step": 14521
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006372684275529233,
+      "loss": 0.1063,
+      "step": 14522
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006371375756206566,
+      "loss": 0.1049,
+      "step": 14523
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006370067308431205,
+      "loss": 0.1793,
+      "step": 14524
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006368758932228954,
+      "loss": 0.1195,
+      "step": 14525
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006367450627625601,
+      "loss": 0.0749,
+      "step": 14526
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006366142394646952,
+      "loss": 0.1036,
+      "step": 14527
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006364834233318804,
+      "loss": 0.1323,
+      "step": 14528
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006363526143666944,
+      "loss": 0.0695,
+      "step": 14529
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006362218125717167,
+      "loss": 0.111,
+      "step": 14530
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006360910179495261,
+      "loss": 0.0703,
+      "step": 14531
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006359602305027019,
+      "loss": 0.0519,
+      "step": 14532
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006358294502338226,
+      "loss": 0.07,
+      "step": 14533
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006356986771454674,
+      "loss": 0.0795,
+      "step": 14534
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006355679112402142,
+      "loss": 0.08,
+      "step": 14535
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006354371525206417,
+      "loss": 0.0613,
+      "step": 14536
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006353064009893279,
+      "loss": 0.1066,
+      "step": 14537
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006351756566488513,
+      "loss": 0.1226,
+      "step": 14538
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.000635044919501789,
+      "loss": 0.1179,
+      "step": 14539
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006349141895507199,
+      "loss": 0.1927,
+      "step": 14540
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.000634783466798221,
+      "loss": 0.0831,
+      "step": 14541
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006346527512468702,
+      "loss": 0.0856,
+      "step": 14542
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006345220428992443,
+      "loss": 0.082,
+      "step": 14543
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006343913417579212,
+      "loss": 0.0981,
+      "step": 14544
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006342606478254775,
+      "loss": 0.0995,
+      "step": 14545
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006341299611044901,
+      "loss": 0.0665,
+      "step": 14546
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006339992815975365,
+      "loss": 0.0802,
+      "step": 14547
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006338686093071926,
+      "loss": 0.0804,
+      "step": 14548
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006337379442360356,
+      "loss": 0.0678,
+      "step": 14549
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006336072863866413,
+      "loss": 0.0983,
+      "step": 14550
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006334766357615864,
+      "loss": 0.1892,
+      "step": 14551
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006333459923634463,
+      "loss": 0.1133,
+      "step": 14552
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006332153561947977,
+      "loss": 0.0904,
+      "step": 14553
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0006330847272582161,
+      "loss": 0.0892,
+      "step": 14554
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006329541055562775,
+      "loss": 0.1688,
+      "step": 14555
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006328234910915571,
+      "loss": 0.0771,
+      "step": 14556
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006326928838666303,
+      "loss": 0.0808,
+      "step": 14557
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006325622838840721,
+      "loss": 0.0475,
+      "step": 14558
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.000632431691146458,
+      "loss": 0.0807,
+      "step": 14559
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006323011056563629,
+      "loss": 0.1152,
+      "step": 14560
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006321705274163617,
+      "loss": 0.1753,
+      "step": 14561
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006320399564290289,
+      "loss": 0.1094,
+      "step": 14562
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006319093926969391,
+      "loss": 0.0975,
+      "step": 14563
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006317788362226666,
+      "loss": 0.0784,
+      "step": 14564
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006316482870087857,
+      "loss": 0.0854,
+      "step": 14565
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006315177450578702,
+      "loss": 0.1097,
+      "step": 14566
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006313872103724947,
+      "loss": 0.0521,
+      "step": 14567
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006312566829552328,
+      "loss": 0.0614,
+      "step": 14568
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006311261628086578,
+      "loss": 0.0855,
+      "step": 14569
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006309956499353436,
+      "loss": 0.0637,
+      "step": 14570
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006308651443378635,
+      "loss": 0.0873,
+      "step": 14571
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006307346460187902,
+      "loss": 0.0671,
+      "step": 14572
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006306041549806978,
+      "loss": 0.1206,
+      "step": 14573
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006304736712261587,
+      "loss": 0.0772,
+      "step": 14574
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006303431947577456,
+      "loss": 0.0936,
+      "step": 14575
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006302127255780315,
+      "loss": 0.078,
+      "step": 14576
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006300822636895886,
+      "loss": 0.1626,
+      "step": 14577
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006299518090949896,
+      "loss": 0.077,
+      "step": 14578
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006298213617968063,
+      "loss": 0.0864,
+      "step": 14579
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006296909217976112,
+      "loss": 0.0596,
+      "step": 14580
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006295604890999759,
+      "loss": 0.167,
+      "step": 14581
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006294300637064729,
+      "loss": 0.098,
+      "step": 14582
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006292996456196731,
+      "loss": 0.141,
+      "step": 14583
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006291692348421483,
+      "loss": 0.0697,
+      "step": 14584
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006290388313764697,
+      "loss": 0.0858,
+      "step": 14585
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006289084352252085,
+      "loss": 0.0905,
+      "step": 14586
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006287780463909363,
+      "loss": 0.0746,
+      "step": 14587
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006286476648762238,
+      "loss": 0.0471,
+      "step": 14588
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006285172906836416,
+      "loss": 0.0881,
+      "step": 14589
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006283869238157602,
+      "loss": 0.0757,
+      "step": 14590
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006282565642751507,
+      "loss": 0.0533,
+      "step": 14591
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006281262120643826,
+      "loss": 0.0648,
+      "step": 14592
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006279958671860271,
+      "loss": 0.0773,
+      "step": 14593
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006278655296426536,
+      "loss": 0.0675,
+      "step": 14594
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006277351994368325,
+      "loss": 0.054,
+      "step": 14595
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.000627604876571133,
+      "loss": 0.0781,
+      "step": 14596
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006274745610481252,
+      "loss": 0.0973,
+      "step": 14597
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006273442528703786,
+      "loss": 0.0681,
+      "step": 14598
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0006272139520404617,
+      "loss": 0.0811,
+      "step": 14599
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.000627083658560945,
+      "loss": 0.0739,
+      "step": 14600
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006269533724343969,
+      "loss": 0.1477,
+      "step": 14601
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006268230936633861,
+      "loss": 0.0745,
+      "step": 14602
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000626692822250482,
+      "loss": 0.119,
+      "step": 14603
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006265625581982527,
+      "loss": 0.178,
+      "step": 14604
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006264323015092668,
+      "loss": 0.1199,
+      "step": 14605
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006263020521860924,
+      "loss": 0.1033,
+      "step": 14606
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006261718102312981,
+      "loss": 0.1104,
+      "step": 14607
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000626041575647452,
+      "loss": 0.0872,
+      "step": 14608
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006259113484371218,
+      "loss": 0.0649,
+      "step": 14609
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006257811286028752,
+      "loss": 0.0915,
+      "step": 14610
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006256509161472796,
+      "loss": 0.0851,
+      "step": 14611
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006255207110729029,
+      "loss": 0.0734,
+      "step": 14612
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006253905133823121,
+      "loss": 0.0817,
+      "step": 14613
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006252603230780747,
+      "loss": 0.0953,
+      "step": 14614
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006251301401627575,
+      "loss": 0.182,
+      "step": 14615
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006249999646389275,
+      "loss": 0.1011,
+      "step": 14616
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000624869796509151,
+      "loss": 0.1492,
+      "step": 14617
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006247396357759951,
+      "loss": 0.0811,
+      "step": 14618
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006246094824420258,
+      "loss": 0.1063,
+      "step": 14619
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006244793365098101,
+      "loss": 0.097,
+      "step": 14620
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006243491979819135,
+      "loss": 0.2292,
+      "step": 14621
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006242190668609021,
+      "loss": 0.0717,
+      "step": 14622
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000624088943149342,
+      "loss": 0.0769,
+      "step": 14623
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006239588268497989,
+      "loss": 0.0891,
+      "step": 14624
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006238287179648382,
+      "loss": 0.0813,
+      "step": 14625
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000623698616497025,
+      "loss": 0.0992,
+      "step": 14626
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006235685224489252,
+      "loss": 0.1034,
+      "step": 14627
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006234384358231035,
+      "loss": 0.1453,
+      "step": 14628
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006233083566221253,
+      "loss": 0.1849,
+      "step": 14629
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006231782848485552,
+      "loss": 0.1016,
+      "step": 14630
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006230482205049578,
+      "loss": 0.0641,
+      "step": 14631
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006229181635938972,
+      "loss": 0.0948,
+      "step": 14632
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006227881141179388,
+      "loss": 0.0806,
+      "step": 14633
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006226580720796462,
+      "loss": 0.0624,
+      "step": 14634
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006225280374815838,
+      "loss": 0.1101,
+      "step": 14635
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006223980103263154,
+      "loss": 0.1193,
+      "step": 14636
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006222679906164046,
+      "loss": 0.1271,
+      "step": 14637
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006221379783544152,
+      "loss": 0.0984,
+      "step": 14638
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006220079735429107,
+      "loss": 0.0956,
+      "step": 14639
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006218779761844548,
+      "loss": 0.1102,
+      "step": 14640
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006217479862816105,
+      "loss": 0.1091,
+      "step": 14641
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006216180038369407,
+      "loss": 0.0868,
+      "step": 14642
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006214880288530085,
+      "loss": 0.0588,
+      "step": 14643
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006213580613323768,
+      "loss": 0.146,
+      "step": 14644
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000621228101277608,
+      "loss": 0.059,
+      "step": 14645
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0006210981486912642,
+      "loss": 0.0759,
+      "step": 14646
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006209682035759087,
+      "loss": 0.0748,
+      "step": 14647
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.000620838265934103,
+      "loss": 0.0624,
+      "step": 14648
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006207083357684094,
+      "loss": 0.0382,
+      "step": 14649
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006205784130813896,
+      "loss": 0.1555,
+      "step": 14650
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006204484978756058,
+      "loss": 0.1022,
+      "step": 14651
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006203185901536186,
+      "loss": 0.0693,
+      "step": 14652
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006201886899179906,
+      "loss": 0.0664,
+      "step": 14653
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006200587971712827,
+      "loss": 0.073,
+      "step": 14654
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006199289119160556,
+      "loss": 0.075,
+      "step": 14655
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006197990341548711,
+      "loss": 0.0687,
+      "step": 14656
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006196691638902895,
+      "loss": 0.046,
+      "step": 14657
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006195393011248717,
+      "loss": 0.087,
+      "step": 14658
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006194094458611779,
+      "loss": 0.0865,
+      "step": 14659
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006192795981017689,
+      "loss": 0.08,
+      "step": 14660
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006191497578492051,
+      "loss": 0.1843,
+      "step": 14661
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006190199251060466,
+      "loss": 0.0892,
+      "step": 14662
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006188900998748532,
+      "loss": 0.0892,
+      "step": 14663
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006187602821581844,
+      "loss": 0.0936,
+      "step": 14664
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006186304719586005,
+      "loss": 0.1508,
+      "step": 14665
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006185006692786603,
+      "loss": 0.1,
+      "step": 14666
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006183708741209242,
+      "loss": 0.0679,
+      "step": 14667
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006182410864879506,
+      "loss": 0.1395,
+      "step": 14668
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006181113063822989,
+      "loss": 0.1807,
+      "step": 14669
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.000617981533806528,
+      "loss": 0.0856,
+      "step": 14670
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006178517687631966,
+      "loss": 0.1281,
+      "step": 14671
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006177220112548632,
+      "loss": 0.075,
+      "step": 14672
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006175922612840867,
+      "loss": 0.0632,
+      "step": 14673
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006174625188534254,
+      "loss": 0.0815,
+      "step": 14674
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006173327839654372,
+      "loss": 0.0843,
+      "step": 14675
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006172030566226803,
+      "loss": 0.1373,
+      "step": 14676
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006170733368277126,
+      "loss": 0.077,
+      "step": 14677
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006169436245830918,
+      "loss": 0.0743,
+      "step": 14678
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006168139198913751,
+      "loss": 0.1051,
+      "step": 14679
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006166842227551209,
+      "loss": 0.0498,
+      "step": 14680
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006165545331768857,
+      "loss": 0.0531,
+      "step": 14681
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.000616424851159227,
+      "loss": 0.0891,
+      "step": 14682
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.000616295176704702,
+      "loss": 0.1218,
+      "step": 14683
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006161655098158669,
+      "loss": 0.0602,
+      "step": 14684
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006160358504952789,
+      "loss": 0.1124,
+      "step": 14685
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006159061987454943,
+      "loss": 0.073,
+      "step": 14686
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006157765545690697,
+      "loss": 0.1115,
+      "step": 14687
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006156469179685613,
+      "loss": 0.0718,
+      "step": 14688
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006155172889465252,
+      "loss": 0.1159,
+      "step": 14689
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006153876675055177,
+      "loss": 0.0682,
+      "step": 14690
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006152580536480937,
+      "loss": 0.1628,
+      "step": 14691
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0006151284473768095,
+      "loss": 0.0685,
+      "step": 14692
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006149988486942207,
+      "loss": 0.0967,
+      "step": 14693
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006148692576028825,
+      "loss": 0.0856,
+      "step": 14694
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006147396741053501,
+      "loss": 0.0798,
+      "step": 14695
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006146100982041786,
+      "loss": 0.0884,
+      "step": 14696
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006144805299019227,
+      "loss": 0.1272,
+      "step": 14697
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006143509692011376,
+      "loss": 0.1382,
+      "step": 14698
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006142214161043771,
+      "loss": 0.16,
+      "step": 14699
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006140918706141966,
+      "loss": 0.0919,
+      "step": 14700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006139623327331501,
+      "loss": 0.1077,
+      "step": 14701
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006138328024637913,
+      "loss": 0.099,
+      "step": 14702
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006137032798086748,
+      "loss": 0.0457,
+      "step": 14703
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006135737647703543,
+      "loss": 0.0914,
+      "step": 14704
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006134442573513835,
+      "loss": 0.104,
+      "step": 14705
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006133147575543154,
+      "loss": 0.0858,
+      "step": 14706
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006131852653817043,
+      "loss": 0.1053,
+      "step": 14707
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000613055780836103,
+      "loss": 0.0854,
+      "step": 14708
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006129263039200648,
+      "loss": 0.0626,
+      "step": 14709
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006127968346361424,
+      "loss": 0.0835,
+      "step": 14710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006126673729868889,
+      "loss": 0.1057,
+      "step": 14711
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006125379189748563,
+      "loss": 0.0517,
+      "step": 14712
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006124084726025983,
+      "loss": 0.0684,
+      "step": 14713
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000612279033872666,
+      "loss": 0.0652,
+      "step": 14714
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006121496027876126,
+      "loss": 0.0586,
+      "step": 14715
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006120201793499895,
+      "loss": 0.2722,
+      "step": 14716
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006118907635623488,
+      "loss": 0.0834,
+      "step": 14717
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006117613554272423,
+      "loss": 0.0735,
+      "step": 14718
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006116319549472213,
+      "loss": 0.1062,
+      "step": 14719
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000611502562124838,
+      "loss": 0.0812,
+      "step": 14720
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006113731769626431,
+      "loss": 0.0998,
+      "step": 14721
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006112437994631881,
+      "loss": 0.1089,
+      "step": 14722
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006111144296290234,
+      "loss": 0.1313,
+      "step": 14723
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006109850674627004,
+      "loss": 0.2184,
+      "step": 14724
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006108557129667696,
+      "loss": 0.0684,
+      "step": 14725
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006107263661437812,
+      "loss": 0.1027,
+      "step": 14726
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006105970269962864,
+      "loss": 0.0861,
+      "step": 14727
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006104676955268349,
+      "loss": 0.0887,
+      "step": 14728
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006103383717379768,
+      "loss": 0.1438,
+      "step": 14729
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006102090556322623,
+      "loss": 0.0741,
+      "step": 14730
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000610079747212241,
+      "loss": 0.093,
+      "step": 14731
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000609950446480462,
+      "loss": 0.0625,
+      "step": 14732
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006098211534394759,
+      "loss": 0.1821,
+      "step": 14733
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006096918680918313,
+      "loss": 0.0865,
+      "step": 14734
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006095625904400777,
+      "loss": 0.0662,
+      "step": 14735
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006094333204867638,
+      "loss": 0.0927,
+      "step": 14736
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0006093040582344388,
+      "loss": 0.0803,
+      "step": 14737
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000609174803685651,
+      "loss": 0.0998,
+      "step": 14738
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006090455568429494,
+      "loss": 0.0948,
+      "step": 14739
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006089163177088821,
+      "loss": 0.0611,
+      "step": 14740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006087870862859978,
+      "loss": 0.0981,
+      "step": 14741
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006086578625768444,
+      "loss": 0.0675,
+      "step": 14742
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006085286465839697,
+      "loss": 0.2368,
+      "step": 14743
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006083994383099216,
+      "loss": 0.0794,
+      "step": 14744
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000608270237757248,
+      "loss": 0.0608,
+      "step": 14745
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006081410449284957,
+      "loss": 0.0751,
+      "step": 14746
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000608011859826213,
+      "loss": 0.0961,
+      "step": 14747
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006078826824529466,
+      "loss": 0.0832,
+      "step": 14748
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006077535128112437,
+      "loss": 0.1229,
+      "step": 14749
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006076243509036507,
+      "loss": 0.0814,
+      "step": 14750
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006074951967327153,
+      "loss": 0.1236,
+      "step": 14751
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006073660503009829,
+      "loss": 0.0693,
+      "step": 14752
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000607236911611001,
+      "loss": 0.1047,
+      "step": 14753
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006071077806653157,
+      "loss": 0.0784,
+      "step": 14754
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006069786574664726,
+      "loss": 0.1144,
+      "step": 14755
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006068495420170181,
+      "loss": 0.1128,
+      "step": 14756
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006067204343194981,
+      "loss": 0.1018,
+      "step": 14757
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006065913343764581,
+      "loss": 0.0791,
+      "step": 14758
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006064622421904432,
+      "loss": 0.0842,
+      "step": 14759
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006063331577639996,
+      "loss": 0.1462,
+      "step": 14760
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006062040810996718,
+      "loss": 0.0524,
+      "step": 14761
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006060750122000056,
+      "loss": 0.0792,
+      "step": 14762
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006059459510675455,
+      "loss": 0.1227,
+      "step": 14763
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006058168977048362,
+      "loss": 0.0721,
+      "step": 14764
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006056878521144221,
+      "loss": 0.0905,
+      "step": 14765
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000605558814298848,
+      "loss": 0.087,
+      "step": 14766
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006054297842606583,
+      "loss": 0.1525,
+      "step": 14767
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000605300762002397,
+      "loss": 0.1127,
+      "step": 14768
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006051717475266079,
+      "loss": 0.0878,
+      "step": 14769
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006050427408358352,
+      "loss": 0.1077,
+      "step": 14770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006049137419326223,
+      "loss": 0.0927,
+      "step": 14771
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006047847508195125,
+      "loss": 0.0931,
+      "step": 14772
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006046557674990498,
+      "loss": 0.1157,
+      "step": 14773
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006045267919737773,
+      "loss": 0.1094,
+      "step": 14774
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006043978242462378,
+      "loss": 0.0756,
+      "step": 14775
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006042688643189742,
+      "loss": 0.1176,
+      "step": 14776
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006041399121945296,
+      "loss": 0.1011,
+      "step": 14777
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006040109678754463,
+      "loss": 0.121,
+      "step": 14778
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006038820313642666,
+      "loss": 0.078,
+      "step": 14779
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006037531026635332,
+      "loss": 0.1063,
+      "step": 14780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006036241817757884,
+      "loss": 0.1129,
+      "step": 14781
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006034952687035735,
+      "loss": 0.1013,
+      "step": 14782
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006033663634494309,
+      "loss": 0.0735,
+      "step": 14783
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0006032374660159023,
+      "loss": 0.0419,
+      "step": 14784
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006031085764055284,
+      "loss": 0.0731,
+      "step": 14785
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006029796946208518,
+      "loss": 0.1625,
+      "step": 14786
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006028508206644132,
+      "loss": 0.071,
+      "step": 14787
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006027219545387535,
+      "loss": 0.0751,
+      "step": 14788
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006025930962464136,
+      "loss": 0.097,
+      "step": 14789
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006024642457899348,
+      "loss": 0.1022,
+      "step": 14790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006023354031718571,
+      "loss": 0.08,
+      "step": 14791
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006022065683947207,
+      "loss": 0.0909,
+      "step": 14792
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006020777414610666,
+      "loss": 0.0813,
+      "step": 14793
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006019489223734349,
+      "loss": 0.085,
+      "step": 14794
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006018201111343655,
+      "loss": 0.0709,
+      "step": 14795
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000601691307746398,
+      "loss": 0.1735,
+      "step": 14796
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006015625122120721,
+      "loss": 0.0966,
+      "step": 14797
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006014337245339274,
+      "loss": 0.058,
+      "step": 14798
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000601304944714503,
+      "loss": 0.0853,
+      "step": 14799
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006011761727563389,
+      "loss": 0.0765,
+      "step": 14800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006010474086619735,
+      "loss": 0.0603,
+      "step": 14801
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006009186524339458,
+      "loss": 0.0818,
+      "step": 14802
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006007899040747946,
+      "loss": 0.1251,
+      "step": 14803
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006006611635870587,
+      "loss": 0.0805,
+      "step": 14804
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006005324309732758,
+      "loss": 0.0867,
+      "step": 14805
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006004037062359851,
+      "loss": 0.1003,
+      "step": 14806
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006002749893777244,
+      "loss": 0.0978,
+      "step": 14807
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006001462804010316,
+      "loss": 0.0682,
+      "step": 14808
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0006000175793084443,
+      "loss": 0.0798,
+      "step": 14809
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005998888861025006,
+      "loss": 0.0841,
+      "step": 14810
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000599760200785738,
+      "loss": 0.1118,
+      "step": 14811
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000599631523360693,
+      "loss": 0.0994,
+      "step": 14812
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000599502853829904,
+      "loss": 0.0538,
+      "step": 14813
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005993741921959071,
+      "loss": 0.0925,
+      "step": 14814
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005992455384612399,
+      "loss": 0.1108,
+      "step": 14815
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005991168926284387,
+      "loss": 0.127,
+      "step": 14816
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005989882547000403,
+      "loss": 0.0756,
+      "step": 14817
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005988596246785807,
+      "loss": 0.1547,
+      "step": 14818
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005987310025665964,
+      "loss": 0.0837,
+      "step": 14819
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005986023883666234,
+      "loss": 0.1302,
+      "step": 14820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005984737820811982,
+      "loss": 0.0668,
+      "step": 14821
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000598345183712856,
+      "loss": 0.084,
+      "step": 14822
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005982165932641329,
+      "loss": 0.1198,
+      "step": 14823
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005980880107375636,
+      "loss": 0.0729,
+      "step": 14824
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005979594361356837,
+      "loss": 0.1078,
+      "step": 14825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005978308694610291,
+      "loss": 0.0916,
+      "step": 14826
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005977023107161343,
+      "loss": 0.0634,
+      "step": 14827
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005975737599035341,
+      "loss": 0.1238,
+      "step": 14828
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005974452170257632,
+      "loss": 0.1022,
+      "step": 14829
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0005973166820853562,
+      "loss": 0.0613,
+      "step": 14830
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005971881550848474,
+      "loss": 0.1238,
+      "step": 14831
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005970596360267708,
+      "loss": 0.1167,
+      "step": 14832
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005969311249136611,
+      "loss": 0.1215,
+      "step": 14833
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.000596802621748052,
+      "loss": 0.092,
+      "step": 14834
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005966741265324769,
+      "loss": 0.1015,
+      "step": 14835
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005965456392694697,
+      "loss": 0.1282,
+      "step": 14836
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005964171599615639,
+      "loss": 0.1268,
+      "step": 14837
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005962886886112926,
+      "loss": 0.0459,
+      "step": 14838
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005961602252211886,
+      "loss": 0.1362,
+      "step": 14839
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005960317697937858,
+      "loss": 0.1113,
+      "step": 14840
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005959033223316161,
+      "loss": 0.1007,
+      "step": 14841
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005957748828372129,
+      "loss": 0.0609,
+      "step": 14842
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005956464513131083,
+      "loss": 0.1289,
+      "step": 14843
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005955180277618348,
+      "loss": 0.1091,
+      "step": 14844
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005953896121859239,
+      "loss": 0.1372,
+      "step": 14845
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005952612045879086,
+      "loss": 0.0753,
+      "step": 14846
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005951328049703204,
+      "loss": 0.1068,
+      "step": 14847
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005950044133356911,
+      "loss": 0.0975,
+      "step": 14848
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005948760296865523,
+      "loss": 0.0612,
+      "step": 14849
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005947476540254349,
+      "loss": 0.1133,
+      "step": 14850
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005946192863548707,
+      "loss": 0.0701,
+      "step": 14851
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005944909266773903,
+      "loss": 0.0597,
+      "step": 14852
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005943625749955252,
+      "loss": 0.061,
+      "step": 14853
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005942342313118061,
+      "loss": 0.1163,
+      "step": 14854
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005941058956287633,
+      "loss": 0.1587,
+      "step": 14855
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.000593977567948927,
+      "loss": 0.1588,
+      "step": 14856
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005938492482748283,
+      "loss": 0.0794,
+      "step": 14857
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005937209366089968,
+      "loss": 0.1233,
+      "step": 14858
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005935926329539621,
+      "loss": 0.0812,
+      "step": 14859
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.000593464337312255,
+      "loss": 0.0519,
+      "step": 14860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005933360496864047,
+      "loss": 0.0847,
+      "step": 14861
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005932077700789405,
+      "loss": 0.162,
+      "step": 14862
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005930794984923922,
+      "loss": 0.1252,
+      "step": 14863
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005929512349292886,
+      "loss": 0.0591,
+      "step": 14864
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005928229793921584,
+      "loss": 0.1043,
+      "step": 14865
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005926947318835314,
+      "loss": 0.1112,
+      "step": 14866
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005925664924059358,
+      "loss": 0.098,
+      "step": 14867
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005924382609619002,
+      "loss": 0.1097,
+      "step": 14868
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005923100375539532,
+      "loss": 0.1232,
+      "step": 14869
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005921818221846227,
+      "loss": 0.1182,
+      "step": 14870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005920536148564369,
+      "loss": 0.1442,
+      "step": 14871
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005919254155719235,
+      "loss": 0.0891,
+      "step": 14872
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005917972243336107,
+      "loss": 0.0682,
+      "step": 14873
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005916690411440261,
+      "loss": 0.1136,
+      "step": 14874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005915408660056969,
+      "loss": 0.0798,
+      "step": 14875
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0005914126989211506,
+      "loss": 0.0604,
+      "step": 14876
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000591284539892914,
+      "loss": 0.0676,
+      "step": 14877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005911563889235144,
+      "loss": 0.0918,
+      "step": 14878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005910282460154781,
+      "loss": 0.0784,
+      "step": 14879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005909001111713326,
+      "loss": 0.0885,
+      "step": 14880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005907719843936039,
+      "loss": 0.0817,
+      "step": 14881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005906438656848185,
+      "loss": 0.0666,
+      "step": 14882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005905157550475021,
+      "loss": 0.1145,
+      "step": 14883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005903876524841813,
+      "loss": 0.1004,
+      "step": 14884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005902595579973814,
+      "loss": 0.0929,
+      "step": 14885
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005901314715896288,
+      "loss": 0.1076,
+      "step": 14886
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005900033932634487,
+      "loss": 0.1183,
+      "step": 14887
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005898753230213662,
+      "loss": 0.0659,
+      "step": 14888
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000589747260865907,
+      "loss": 0.0732,
+      "step": 14889
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005896192067995959,
+      "loss": 0.0774,
+      "step": 14890
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005894911608249578,
+      "loss": 0.0707,
+      "step": 14891
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005893631229445172,
+      "loss": 0.0568,
+      "step": 14892
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005892350931607993,
+      "loss": 0.0589,
+      "step": 14893
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005891070714763281,
+      "loss": 0.1389,
+      "step": 14894
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000588979057893628,
+      "loss": 0.1195,
+      "step": 14895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000588851052415223,
+      "loss": 0.0711,
+      "step": 14896
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005887230550436373,
+      "loss": 0.1028,
+      "step": 14897
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005885950657813944,
+      "loss": 0.105,
+      "step": 14898
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005884670846310178,
+      "loss": 0.1111,
+      "step": 14899
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005883391115950312,
+      "loss": 0.0472,
+      "step": 14900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005882111466759581,
+      "loss": 0.1361,
+      "step": 14901
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005880831898763215,
+      "loss": 0.0784,
+      "step": 14902
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005879552411986442,
+      "loss": 0.1199,
+      "step": 14903
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005878273006454489,
+      "loss": 0.058,
+      "step": 14904
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005876993682192584,
+      "loss": 0.0773,
+      "step": 14905
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005875714439225958,
+      "loss": 0.0887,
+      "step": 14906
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005874435277579828,
+      "loss": 0.1263,
+      "step": 14907
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005873156197279419,
+      "loss": 0.1233,
+      "step": 14908
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005871877198349946,
+      "loss": 0.0605,
+      "step": 14909
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005870598280816635,
+      "loss": 0.0841,
+      "step": 14910
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005869319444704697,
+      "loss": 0.1273,
+      "step": 14911
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005868040690039346,
+      "loss": 0.0902,
+      "step": 14912
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005866762016845806,
+      "loss": 0.0912,
+      "step": 14913
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005865483425149281,
+      "loss": 0.1284,
+      "step": 14914
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005864204914974981,
+      "loss": 0.1078,
+      "step": 14915
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000586292648634812,
+      "loss": 0.1152,
+      "step": 14916
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005861648139293903,
+      "loss": 0.1272,
+      "step": 14917
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005860369873837535,
+      "loss": 0.1183,
+      "step": 14918
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005859091690004214,
+      "loss": 0.1106,
+      "step": 14919
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005857813587819154,
+      "loss": 0.1044,
+      "step": 14920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005856535567307554,
+      "loss": 0.0529,
+      "step": 14921
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005855257628494608,
+      "loss": 0.1171,
+      "step": 14922
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0005853979771405517,
+      "loss": 0.0934,
+      "step": 14923
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005852701996065476,
+      "loss": 0.0994,
+      "step": 14924
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005851424302499678,
+      "loss": 0.0821,
+      "step": 14925
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005850146690733318,
+      "loss": 0.1532,
+      "step": 14926
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000584886916079159,
+      "loss": 0.1205,
+      "step": 14927
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005847591712699681,
+      "loss": 0.1294,
+      "step": 14928
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005846314346482776,
+      "loss": 0.1243,
+      "step": 14929
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005845037062166065,
+      "loss": 0.0822,
+      "step": 14930
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005843759859774731,
+      "loss": 0.1328,
+      "step": 14931
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005842482739333956,
+      "loss": 0.1078,
+      "step": 14932
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005841205700868928,
+      "loss": 0.0817,
+      "step": 14933
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005839928744404822,
+      "loss": 0.1178,
+      "step": 14934
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005838651869966817,
+      "loss": 0.0797,
+      "step": 14935
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005837375077580089,
+      "loss": 0.1018,
+      "step": 14936
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005836098367269816,
+      "loss": 0.1008,
+      "step": 14937
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005834821739061168,
+      "loss": 0.0812,
+      "step": 14938
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005833545192979315,
+      "loss": 0.0761,
+      "step": 14939
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005832268729049435,
+      "loss": 0.0574,
+      "step": 14940
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005830992347296691,
+      "loss": 0.0747,
+      "step": 14941
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005829716047746252,
+      "loss": 0.1131,
+      "step": 14942
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005828439830423279,
+      "loss": 0.0991,
+      "step": 14943
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005827163695352943,
+      "loss": 0.1161,
+      "step": 14944
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00058258876425604,
+      "loss": 0.0496,
+      "step": 14945
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005824611672070815,
+      "loss": 0.1534,
+      "step": 14946
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005823335783909346,
+      "loss": 0.1106,
+      "step": 14947
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005822059978101149,
+      "loss": 0.0941,
+      "step": 14948
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005820784254671381,
+      "loss": 0.0516,
+      "step": 14949
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005819508613645194,
+      "loss": 0.0966,
+      "step": 14950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005818233055047742,
+      "loss": 0.0931,
+      "step": 14951
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005816957578904172,
+      "loss": 0.1044,
+      "step": 14952
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005815682185239639,
+      "loss": 0.0761,
+      "step": 14953
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000581440687407929,
+      "loss": 0.1129,
+      "step": 14954
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005813131645448263,
+      "loss": 0.0904,
+      "step": 14955
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005811856499371713,
+      "loss": 0.1012,
+      "step": 14956
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005810581435874778,
+      "loss": 0.1294,
+      "step": 14957
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005809306454982599,
+      "loss": 0.0897,
+      "step": 14958
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000580803155672031,
+      "loss": 0.0684,
+      "step": 14959
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000580675674111306,
+      "loss": 0.1039,
+      "step": 14960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005805482008185978,
+      "loss": 0.1064,
+      "step": 14961
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005804207357964201,
+      "loss": 0.1162,
+      "step": 14962
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005802932790472859,
+      "loss": 0.0508,
+      "step": 14963
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005801658305737082,
+      "loss": 0.1401,
+      "step": 14964
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005800383903782006,
+      "loss": 0.0499,
+      "step": 14965
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005799109584632751,
+      "loss": 0.0884,
+      "step": 14966
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005797835348314454,
+      "loss": 0.0793,
+      "step": 14967
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005796561194852232,
+      "loss": 0.0556,
+      "step": 14968
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0005795287124271209,
+      "loss": 0.0661,
+      "step": 14969
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005794013136596508,
+      "loss": 0.0488,
+      "step": 14970
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005792739231853248,
+      "loss": 0.0687,
+      "step": 14971
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005791465410066543,
+      "loss": 0.12,
+      "step": 14972
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005790191671261517,
+      "loss": 0.0907,
+      "step": 14973
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005788918015463282,
+      "loss": 0.0728,
+      "step": 14974
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005787644442696951,
+      "loss": 0.1023,
+      "step": 14975
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005786370952987631,
+      "loss": 0.1053,
+      "step": 14976
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005785097546360441,
+      "loss": 0.1051,
+      "step": 14977
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005783824222840486,
+      "loss": 0.1005,
+      "step": 14978
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005782550982452865,
+      "loss": 0.0623,
+      "step": 14979
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005781277825222696,
+      "loss": 0.1037,
+      "step": 14980
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005780004751175076,
+      "loss": 0.0936,
+      "step": 14981
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005778731760335105,
+      "loss": 0.0676,
+      "step": 14982
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005777458852727887,
+      "loss": 0.1112,
+      "step": 14983
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005776186028378518,
+      "loss": 0.0725,
+      "step": 14984
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005774913287312091,
+      "loss": 0.0961,
+      "step": 14985
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000577364062955371,
+      "loss": 0.0832,
+      "step": 14986
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005772368055128461,
+      "loss": 0.1191,
+      "step": 14987
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005771095564061442,
+      "loss": 0.0746,
+      "step": 14988
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005769823156377741,
+      "loss": 0.1001,
+      "step": 14989
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005768550832102447,
+      "loss": 0.062,
+      "step": 14990
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005767278591260646,
+      "loss": 0.1584,
+      "step": 14991
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000576600643387742,
+      "loss": 0.0793,
+      "step": 14992
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005764734359977859,
+      "loss": 0.1657,
+      "step": 14993
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005763462369587044,
+      "loss": 0.0779,
+      "step": 14994
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005762190462730054,
+      "loss": 0.1138,
+      "step": 14995
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005760918639431966,
+      "loss": 0.0547,
+      "step": 14996
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005759646899717856,
+      "loss": 0.0885,
+      "step": 14997
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005758375243612807,
+      "loss": 0.0871,
+      "step": 14998
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005757103671141883,
+      "loss": 0.0785,
+      "step": 14999
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005755832182330166,
+      "loss": 0.0735,
+      "step": 15000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005754560777202722,
+      "loss": 0.1192,
+      "step": 15001
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005753289455784619,
+      "loss": 0.1512,
+      "step": 15002
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005752018218100926,
+      "loss": 0.0999,
+      "step": 15003
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005750747064176707,
+      "loss": 0.092,
+      "step": 15004
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005749475994037024,
+      "loss": 0.1063,
+      "step": 15005
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005748205007706945,
+      "loss": 0.1274,
+      "step": 15006
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005746934105211528,
+      "loss": 0.0873,
+      "step": 15007
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005745663286575826,
+      "loss": 0.1018,
+      "step": 15008
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005744392551824907,
+      "loss": 0.0865,
+      "step": 15009
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005743121900983822,
+      "loss": 0.1561,
+      "step": 15010
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005741851334077624,
+      "loss": 0.0724,
+      "step": 15011
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005740580851131363,
+      "loss": 0.0904,
+      "step": 15012
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005739310452170096,
+      "loss": 0.0579,
+      "step": 15013
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005738040137218868,
+      "loss": 0.1223,
+      "step": 15014
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0005736769906302728,
+      "loss": 0.0424,
+      "step": 15015
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005735499759446721,
+      "loss": 0.0807,
+      "step": 15016
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005734229696675886,
+      "loss": 0.09,
+      "step": 15017
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005732959718015276,
+      "loss": 0.097,
+      "step": 15018
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005731689823489921,
+      "loss": 0.0687,
+      "step": 15019
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005730420013124869,
+      "loss": 0.0928,
+      "step": 15020
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005729150286945155,
+      "loss": 0.1117,
+      "step": 15021
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005727880644975813,
+      "loss": 0.1128,
+      "step": 15022
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005726611087241878,
+      "loss": 0.1473,
+      "step": 15023
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005725341613768381,
+      "loss": 0.0947,
+      "step": 15024
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.000572407222458035,
+      "loss": 0.0748,
+      "step": 15025
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005722802919702824,
+      "loss": 0.0973,
+      "step": 15026
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005721533699160824,
+      "loss": 0.0833,
+      "step": 15027
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005720264562979375,
+      "loss": 0.0844,
+      "step": 15028
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005718995511183499,
+      "loss": 0.0688,
+      "step": 15029
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005717726543798227,
+      "loss": 0.087,
+      "step": 15030
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005716457660848574,
+      "loss": 0.1573,
+      "step": 15031
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005715188862359557,
+      "loss": 0.0968,
+      "step": 15032
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00057139201483562,
+      "loss": 0.1289,
+      "step": 15033
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005712651518863516,
+      "loss": 0.1052,
+      "step": 15034
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005711382973906518,
+      "loss": 0.0757,
+      "step": 15035
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005710114513510219,
+      "loss": 0.0873,
+      "step": 15036
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.000570884613769963,
+      "loss": 0.1086,
+      "step": 15037
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005707577846499756,
+      "loss": 0.0692,
+      "step": 15038
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005706309639935614,
+      "loss": 0.1063,
+      "step": 15039
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00057050415180322,
+      "loss": 0.1158,
+      "step": 15040
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005703773480814525,
+      "loss": 0.1071,
+      "step": 15041
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005702505528307592,
+      "loss": 0.1017,
+      "step": 15042
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005701237660536395,
+      "loss": 0.0623,
+      "step": 15043
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005699969877525939,
+      "loss": 0.0868,
+      "step": 15044
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005698702179301216,
+      "loss": 0.1005,
+      "step": 15045
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005697434565887229,
+      "loss": 0.0748,
+      "step": 15046
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005696167037308968,
+      "loss": 0.0989,
+      "step": 15047
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005694899593591428,
+      "loss": 0.09,
+      "step": 15048
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005693632234759596,
+      "loss": 0.0897,
+      "step": 15049
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005692364960838459,
+      "loss": 0.0787,
+      "step": 15050
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005691097771853011,
+      "loss": 0.0916,
+      "step": 15051
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005689830667828233,
+      "loss": 0.1042,
+      "step": 15052
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005688563648789114,
+      "loss": 0.0522,
+      "step": 15053
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005687296714760635,
+      "loss": 0.0676,
+      "step": 15054
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005686029865767775,
+      "loss": 0.08,
+      "step": 15055
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005684763101835513,
+      "loss": 0.1198,
+      "step": 15056
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005683496422988825,
+      "loss": 0.0773,
+      "step": 15057
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005682229829252688,
+      "loss": 0.0619,
+      "step": 15058
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005680963320652077,
+      "loss": 0.0592,
+      "step": 15059
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005679696897211966,
+      "loss": 0.0667,
+      "step": 15060
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0005678430558957319,
+      "loss": 0.0903,
+      "step": 15061
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005677164305913114,
+      "loss": 0.0895,
+      "step": 15062
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005675898138104313,
+      "loss": 0.0593,
+      "step": 15063
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005674632055555883,
+      "loss": 0.0498,
+      "step": 15064
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005673366058292783,
+      "loss": 0.0663,
+      "step": 15065
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005672100146339985,
+      "loss": 0.1604,
+      "step": 15066
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005670834319722442,
+      "loss": 0.1144,
+      "step": 15067
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005669568578465117,
+      "loss": 0.1072,
+      "step": 15068
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005668302922592965,
+      "loss": 0.055,
+      "step": 15069
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005667037352130942,
+      "loss": 0.1016,
+      "step": 15070
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005665771867103997,
+      "loss": 0.0629,
+      "step": 15071
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005664506467537088,
+      "loss": 0.0959,
+      "step": 15072
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005663241153455168,
+      "loss": 0.139,
+      "step": 15073
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000566197592488318,
+      "loss": 0.0821,
+      "step": 15074
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005660710781846077,
+      "loss": 0.0786,
+      "step": 15075
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005659445724368799,
+      "loss": 0.0842,
+      "step": 15076
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005658180752476291,
+      "loss": 0.0834,
+      "step": 15077
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005656915866193492,
+      "loss": 0.0932,
+      "step": 15078
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005655651065545351,
+      "loss": 0.0574,
+      "step": 15079
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005654386350556802,
+      "loss": 0.0989,
+      "step": 15080
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005653121721252781,
+      "loss": 0.08,
+      "step": 15081
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000565185717765822,
+      "loss": 0.113,
+      "step": 15082
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005650592719798059,
+      "loss": 0.0829,
+      "step": 15083
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005649328347697228,
+      "loss": 0.1086,
+      "step": 15084
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005648064061380653,
+      "loss": 0.0985,
+      "step": 15085
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000564679986087327,
+      "loss": 0.0765,
+      "step": 15086
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005645535746200003,
+      "loss": 0.14,
+      "step": 15087
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005644271717385775,
+      "loss": 0.0737,
+      "step": 15088
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005643007774455511,
+      "loss": 0.08,
+      "step": 15089
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005641743917434132,
+      "loss": 0.0844,
+      "step": 15090
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005640480146346559,
+      "loss": 0.1102,
+      "step": 15091
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005639216461217707,
+      "loss": 0.0891,
+      "step": 15092
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005637952862072493,
+      "loss": 0.0605,
+      "step": 15093
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005636689348935839,
+      "loss": 0.137,
+      "step": 15094
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005635425921832655,
+      "loss": 0.1172,
+      "step": 15095
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005634162580787852,
+      "loss": 0.0572,
+      "step": 15096
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005632899325826338,
+      "loss": 0.0903,
+      "step": 15097
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005631636156973019,
+      "loss": 0.0981,
+      "step": 15098
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005630373074252808,
+      "loss": 0.0794,
+      "step": 15099
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005629110077690608,
+      "loss": 0.0818,
+      "step": 15100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000562784716731132,
+      "loss": 0.0963,
+      "step": 15101
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005626584343139847,
+      "loss": 0.1213,
+      "step": 15102
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005625321605201084,
+      "loss": 0.1233,
+      "step": 15103
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005624058953519936,
+      "loss": 0.0967,
+      "step": 15104
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005622796388121293,
+      "loss": 0.0474,
+      "step": 15105
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005621533909030056,
+      "loss": 0.0775,
+      "step": 15106
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0005620271516271116,
+      "loss": 0.0883,
+      "step": 15107
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005619009209869364,
+      "loss": 0.0645,
+      "step": 15108
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005617746989849686,
+      "loss": 0.0507,
+      "step": 15109
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005616484856236973,
+      "loss": 0.0955,
+      "step": 15110
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005615222809056111,
+      "loss": 0.086,
+      "step": 15111
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005613960848331979,
+      "loss": 0.0817,
+      "step": 15112
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005612698974089469,
+      "loss": 0.0471,
+      "step": 15113
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005611437186353454,
+      "loss": 0.1138,
+      "step": 15114
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.000561017548514882,
+      "loss": 0.0962,
+      "step": 15115
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.000560891387050044,
+      "loss": 0.0589,
+      "step": 15116
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005607652342433194,
+      "loss": 0.0983,
+      "step": 15117
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005606390900971947,
+      "loss": 0.077,
+      "step": 15118
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005605129546141582,
+      "loss": 0.1069,
+      "step": 15119
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005603868277966966,
+      "loss": 0.1215,
+      "step": 15120
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005602607096472967,
+      "loss": 0.1055,
+      "step": 15121
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005601346001684453,
+      "loss": 0.0535,
+      "step": 15122
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005600084993626289,
+      "loss": 0.1028,
+      "step": 15123
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005598824072323335,
+      "loss": 0.0978,
+      "step": 15124
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005597563237800459,
+      "loss": 0.0665,
+      "step": 15125
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005596302490082523,
+      "loss": 0.067,
+      "step": 15126
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.000559504182919438,
+      "loss": 0.0735,
+      "step": 15127
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005593781255160893,
+      "loss": 0.1823,
+      "step": 15128
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005592520768006914,
+      "loss": 0.0992,
+      "step": 15129
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005591260367757296,
+      "loss": 0.0871,
+      "step": 15130
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005590000054436892,
+      "loss": 0.0952,
+      "step": 15131
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005588739828070546,
+      "loss": 0.0974,
+      "step": 15132
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005587479688683118,
+      "loss": 0.0991,
+      "step": 15133
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005586219636299449,
+      "loss": 0.0743,
+      "step": 15134
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.000558495967094438,
+      "loss": 0.0828,
+      "step": 15135
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005583699792642761,
+      "loss": 0.094,
+      "step": 15136
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005582440001419434,
+      "loss": 0.0635,
+      "step": 15137
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005581180297299229,
+      "loss": 0.0963,
+      "step": 15138
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005579920680306996,
+      "loss": 0.0998,
+      "step": 15139
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005578661150467567,
+      "loss": 0.097,
+      "step": 15140
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005577401707805775,
+      "loss": 0.1309,
+      "step": 15141
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005576142352346456,
+      "loss": 0.1537,
+      "step": 15142
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005574883084114439,
+      "loss": 0.0629,
+      "step": 15143
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005573623903134554,
+      "loss": 0.0889,
+      "step": 15144
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005572364809431627,
+      "loss": 0.0916,
+      "step": 15145
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005571105803030485,
+      "loss": 0.0895,
+      "step": 15146
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005569846883955958,
+      "loss": 0.0892,
+      "step": 15147
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005568588052232864,
+      "loss": 0.0521,
+      "step": 15148
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005567329307886024,
+      "loss": 0.0999,
+      "step": 15149
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005566070650940259,
+      "loss": 0.0901,
+      "step": 15150
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005564812081420383,
+      "loss": 0.1205,
+      "step": 15151
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005563553599351212,
+      "loss": 0.0901,
+      "step": 15152
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0005562295204757566,
+      "loss": 0.1124,
+      "step": 15153
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005561036897664252,
+      "loss": 0.088,
+      "step": 15154
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005559778678096084,
+      "loss": 0.1288,
+      "step": 15155
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005558520546077863,
+      "loss": 0.1448,
+      "step": 15156
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005557262501634408,
+      "loss": 0.0927,
+      "step": 15157
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005556004544790514,
+      "loss": 0.0789,
+      "step": 15158
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005554746675570991,
+      "loss": 0.0989,
+      "step": 15159
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005553488894000643,
+      "loss": 0.0723,
+      "step": 15160
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005552231200104265,
+      "loss": 0.1209,
+      "step": 15161
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005550973593906658,
+      "loss": 0.1404,
+      "step": 15162
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005549716075432618,
+      "loss": 0.0908,
+      "step": 15163
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005548458644706939,
+      "loss": 0.1473,
+      "step": 15164
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005547201301754414,
+      "loss": 0.0646,
+      "step": 15165
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005545944046599837,
+      "loss": 0.1049,
+      "step": 15166
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005544686879267995,
+      "loss": 0.0726,
+      "step": 15167
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005543429799783681,
+      "loss": 0.0647,
+      "step": 15168
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005542172808171681,
+      "loss": 0.0537,
+      "step": 15169
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005540915904456774,
+      "loss": 0.174,
+      "step": 15170
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005539659088663749,
+      "loss": 0.0508,
+      "step": 15171
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005538402360817379,
+      "loss": 0.0909,
+      "step": 15172
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005537145720942453,
+      "loss": 0.0657,
+      "step": 15173
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005535889169063745,
+      "loss": 0.1185,
+      "step": 15174
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005534632705206032,
+      "loss": 0.069,
+      "step": 15175
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005533376329394087,
+      "loss": 0.074,
+      "step": 15176
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005532120041652677,
+      "loss": 0.1145,
+      "step": 15177
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005530863842006584,
+      "loss": 0.0792,
+      "step": 15178
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005529607730480566,
+      "loss": 0.162,
+      "step": 15179
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005528351707099401,
+      "loss": 0.1165,
+      "step": 15180
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005527095771887849,
+      "loss": 0.1119,
+      "step": 15181
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005525839924870675,
+      "loss": 0.0939,
+      "step": 15182
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000552458416607264,
+      "loss": 0.0684,
+      "step": 15183
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005523328495518506,
+      "loss": 0.1365,
+      "step": 15184
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005522072913233025,
+      "loss": 0.1332,
+      "step": 15185
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005520817419240964,
+      "loss": 0.0804,
+      "step": 15186
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005519562013567074,
+      "loss": 0.1066,
+      "step": 15187
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005518306696236105,
+      "loss": 0.0841,
+      "step": 15188
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005517051467272816,
+      "loss": 0.1163,
+      "step": 15189
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005515796326701951,
+      "loss": 0.0953,
+      "step": 15190
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000551454127454826,
+      "loss": 0.1055,
+      "step": 15191
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005513286310836486,
+      "loss": 0.1066,
+      "step": 15192
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000551203143559138,
+      "loss": 0.065,
+      "step": 15193
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005510776648837685,
+      "loss": 0.0654,
+      "step": 15194
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005509521950600136,
+      "loss": 0.0692,
+      "step": 15195
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005508267340903478,
+      "loss": 0.0885,
+      "step": 15196
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005507012819772445,
+      "loss": 0.058,
+      "step": 15197
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0005505758387231772,
+      "loss": 0.0821,
+      "step": 15198
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00055045040433062,
+      "loss": 0.0858,
+      "step": 15199
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005503249788020451,
+      "loss": 0.0711,
+      "step": 15200
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005501995621399268,
+      "loss": 0.127,
+      "step": 15201
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005500741543467373,
+      "loss": 0.123,
+      "step": 15202
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005499487554249493,
+      "loss": 0.0779,
+      "step": 15203
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005498233653770357,
+      "loss": 0.0958,
+      "step": 15204
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005496979842054682,
+      "loss": 0.1022,
+      "step": 15205
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00054957261191272,
+      "loss": 0.0956,
+      "step": 15206
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005494472485012622,
+      "loss": 0.0775,
+      "step": 15207
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005493218939735673,
+      "loss": 0.1154,
+      "step": 15208
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005491965483321062,
+      "loss": 0.0852,
+      "step": 15209
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005490712115793515,
+      "loss": 0.1461,
+      "step": 15210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005489458837177737,
+      "loss": 0.085,
+      "step": 15211
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005488205647498438,
+      "loss": 0.0557,
+      "step": 15212
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005486952546780335,
+      "loss": 0.083,
+      "step": 15213
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005485699535048135,
+      "loss": 0.0895,
+      "step": 15214
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000548444661232654,
+      "loss": 0.1267,
+      "step": 15215
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005483193778640255,
+      "loss": 0.1096,
+      "step": 15216
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005481941034013984,
+      "loss": 0.0793,
+      "step": 15217
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005480688378472425,
+      "loss": 0.1046,
+      "step": 15218
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005479435812040283,
+      "loss": 0.0885,
+      "step": 15219
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005478183334742251,
+      "loss": 0.0723,
+      "step": 15220
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005476930946603027,
+      "loss": 0.0578,
+      "step": 15221
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005475678647647306,
+      "loss": 0.0621,
+      "step": 15222
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005474426437899778,
+      "loss": 0.1006,
+      "step": 15223
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005473174317385132,
+      "loss": 0.0961,
+      "step": 15224
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005471922286128057,
+      "loss": 0.0797,
+      "step": 15225
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005470670344153244,
+      "loss": 0.0685,
+      "step": 15226
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005469418491485375,
+      "loss": 0.1006,
+      "step": 15227
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005468166728149134,
+      "loss": 0.0654,
+      "step": 15228
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005466915054169202,
+      "loss": 0.0862,
+      "step": 15229
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005465663469570257,
+      "loss": 0.0787,
+      "step": 15230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005464411974376982,
+      "loss": 0.0639,
+      "step": 15231
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005463160568614045,
+      "loss": 0.0769,
+      "step": 15232
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005461909252306132,
+      "loss": 0.1104,
+      "step": 15233
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005460658025477909,
+      "loss": 0.0884,
+      "step": 15234
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005459406888154051,
+      "loss": 0.0905,
+      "step": 15235
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000545815584035922,
+      "loss": 0.0667,
+      "step": 15236
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000545690488211809,
+      "loss": 0.0757,
+      "step": 15237
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000545565401345532,
+      "loss": 0.1347,
+      "step": 15238
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005454403234395585,
+      "loss": 0.093,
+      "step": 15239
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005453152544963538,
+      "loss": 0.0815,
+      "step": 15240
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005451901945183838,
+      "loss": 0.0817,
+      "step": 15241
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005450651435081153,
+      "loss": 0.0699,
+      "step": 15242
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005449401014680133,
+      "loss": 0.0732,
+      "step": 15243
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005448150684005437,
+      "loss": 0.0549,
+      "step": 15244
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0005446900443081711,
+      "loss": 0.0757,
+      "step": 15245
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005445650291933616,
+      "loss": 0.1082,
+      "step": 15246
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005444400230585798,
+      "loss": 0.0605,
+      "step": 15247
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005443150259062903,
+      "loss": 0.073,
+      "step": 15248
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005441900377389582,
+      "loss": 0.119,
+      "step": 15249
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005440650585590474,
+      "loss": 0.1044,
+      "step": 15250
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005439400883690221,
+      "loss": 0.0649,
+      "step": 15251
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005438151271713471,
+      "loss": 0.0897,
+      "step": 15252
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005436901749684855,
+      "loss": 0.0997,
+      "step": 15253
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000543565231762902,
+      "loss": 0.0668,
+      "step": 15254
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005434402975570596,
+      "loss": 0.06,
+      "step": 15255
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005433153723534217,
+      "loss": 0.0745,
+      "step": 15256
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005431904561544515,
+      "loss": 0.1257,
+      "step": 15257
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005430655489626117,
+      "loss": 0.0612,
+      "step": 15258
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000542940650780366,
+      "loss": 0.0639,
+      "step": 15259
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005428157616101765,
+      "loss": 0.0857,
+      "step": 15260
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005426908814545058,
+      "loss": 0.1259,
+      "step": 15261
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005425660103158159,
+      "loss": 0.0862,
+      "step": 15262
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005424411481965697,
+      "loss": 0.1652,
+      "step": 15263
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005423162950992286,
+      "loss": 0.0923,
+      "step": 15264
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005421914510262542,
+      "loss": 0.1021,
+      "step": 15265
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005420666159801088,
+      "loss": 0.1004,
+      "step": 15266
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005419417899632534,
+      "loss": 0.1421,
+      "step": 15267
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005418169729781496,
+      "loss": 0.044,
+      "step": 15268
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005416921650272579,
+      "loss": 0.1535,
+      "step": 15269
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005415673661130396,
+      "loss": 0.0715,
+      "step": 15270
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005414425762379553,
+      "loss": 0.1085,
+      "step": 15271
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005413177954044652,
+      "loss": 0.1224,
+      "step": 15272
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005411930236150304,
+      "loss": 0.0852,
+      "step": 15273
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005410682608721103,
+      "loss": 0.097,
+      "step": 15274
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005409435071781656,
+      "loss": 0.1349,
+      "step": 15275
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005408187625356559,
+      "loss": 0.0767,
+      "step": 15276
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005406940269470407,
+      "loss": 0.0685,
+      "step": 15277
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005405693004147793,
+      "loss": 0.1316,
+      "step": 15278
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005404445829413315,
+      "loss": 0.074,
+      "step": 15279
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005403198745291562,
+      "loss": 0.0654,
+      "step": 15280
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005401951751807124,
+      "loss": 0.0642,
+      "step": 15281
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005400704848984587,
+      "loss": 0.0837,
+      "step": 15282
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005399458036848533,
+      "loss": 0.0963,
+      "step": 15283
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005398211315423554,
+      "loss": 0.0818,
+      "step": 15284
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005396964684734225,
+      "loss": 0.0497,
+      "step": 15285
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005395718144805133,
+      "loss": 0.0786,
+      "step": 15286
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005394471695660854,
+      "loss": 0.0723,
+      "step": 15287
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005393225337325964,
+      "loss": 0.0662,
+      "step": 15288
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005391979069825039,
+      "loss": 0.0432,
+      "step": 15289
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005390732893182652,
+      "loss": 0.0755,
+      "step": 15290
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0005389486807423374,
+      "loss": 0.066,
+      "step": 15291
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000538824081257177,
+      "loss": 0.0593,
+      "step": 15292
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005386994908652416,
+      "loss": 0.0418,
+      "step": 15293
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005385749095689876,
+      "loss": 0.1015,
+      "step": 15294
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005384503373708709,
+      "loss": 0.1075,
+      "step": 15295
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005383257742733484,
+      "loss": 0.0754,
+      "step": 15296
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000538201220278876,
+      "loss": 0.0581,
+      "step": 15297
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005380766753899091,
+      "loss": 0.0878,
+      "step": 15298
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005379521396089043,
+      "loss": 0.048,
+      "step": 15299
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005378276129383165,
+      "loss": 0.1069,
+      "step": 15300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005377030953806013,
+      "loss": 0.1164,
+      "step": 15301
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005375785869382137,
+      "loss": 0.1097,
+      "step": 15302
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005374540876136089,
+      "loss": 0.0523,
+      "step": 15303
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000537329597409241,
+      "loss": 0.1104,
+      "step": 15304
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005372051163275658,
+      "loss": 0.0675,
+      "step": 15305
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005370806443710366,
+      "loss": 0.0895,
+      "step": 15306
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005369561815421087,
+      "loss": 0.0574,
+      "step": 15307
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005368317278432358,
+      "loss": 0.0953,
+      "step": 15308
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005367072832768716,
+      "loss": 0.0762,
+      "step": 15309
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005365828478454699,
+      "loss": 0.1422,
+      "step": 15310
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005364584215514845,
+      "loss": 0.0758,
+      "step": 15311
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005363340043973682,
+      "loss": 0.1097,
+      "step": 15312
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005362095963855749,
+      "loss": 0.0698,
+      "step": 15313
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005360851975185572,
+      "loss": 0.105,
+      "step": 15314
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005359608077987679,
+      "loss": 0.0659,
+      "step": 15315
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00053583642722866,
+      "loss": 0.1041,
+      "step": 15316
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005357120558106859,
+      "loss": 0.0696,
+      "step": 15317
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005355876935472972,
+      "loss": 0.1012,
+      "step": 15318
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005354633404409469,
+      "loss": 0.0633,
+      "step": 15319
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005353389964940867,
+      "loss": 0.1196,
+      "step": 15320
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005352146617091683,
+      "loss": 0.1309,
+      "step": 15321
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000535090336088643,
+      "loss": 0.0892,
+      "step": 15322
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005349660196349625,
+      "loss": 0.1151,
+      "step": 15323
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005348417123505779,
+      "loss": 0.1368,
+      "step": 15324
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005347174142379398,
+      "loss": 0.1437,
+      "step": 15325
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005345931252994999,
+      "loss": 0.0945,
+      "step": 15326
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005344688455377081,
+      "loss": 0.0793,
+      "step": 15327
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005343445749550154,
+      "loss": 0.1112,
+      "step": 15328
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005342203135538721,
+      "loss": 0.0829,
+      "step": 15329
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005340960613367281,
+      "loss": 0.0913,
+      "step": 15330
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005339718183060332,
+      "loss": 0.079,
+      "step": 15331
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000533847584464237,
+      "loss": 0.0756,
+      "step": 15332
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00053372335981379,
+      "loss": 0.1039,
+      "step": 15333
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005335991443571409,
+      "loss": 0.0533,
+      "step": 15334
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000533474938096739,
+      "loss": 0.1001,
+      "step": 15335
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005333507410350329,
+      "loss": 0.1138,
+      "step": 15336
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005332265531744724,
+      "loss": 0.067,
+      "step": 15337
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0005331023745175054,
+      "loss": 0.1182,
+      "step": 15338
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005329782050665808,
+      "loss": 0.1129,
+      "step": 15339
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000532854044824147,
+      "loss": 0.1012,
+      "step": 15340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005327298937926517,
+      "loss": 0.13,
+      "step": 15341
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005326057519745432,
+      "loss": 0.115,
+      "step": 15342
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000532481619372269,
+      "loss": 0.0727,
+      "step": 15343
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005323574959882768,
+      "loss": 0.0743,
+      "step": 15344
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005322333818250138,
+      "loss": 0.1216,
+      "step": 15345
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005321092768849276,
+      "loss": 0.0867,
+      "step": 15346
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005319851811704653,
+      "loss": 0.0831,
+      "step": 15347
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005318610946840728,
+      "loss": 0.0732,
+      "step": 15348
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000531737017428198,
+      "loss": 0.1129,
+      "step": 15349
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005316129494052868,
+      "loss": 0.098,
+      "step": 15350
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005314888906177852,
+      "loss": 0.1188,
+      "step": 15351
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005313648410681403,
+      "loss": 0.0372,
+      "step": 15352
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005312408007587973,
+      "loss": 0.0963,
+      "step": 15353
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005311167696922019,
+      "loss": 0.1432,
+      "step": 15354
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005309927478708001,
+      "loss": 0.0959,
+      "step": 15355
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000530868735297037,
+      "loss": 0.0553,
+      "step": 15356
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005307447319733576,
+      "loss": 0.0833,
+      "step": 15357
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005306207379022075,
+      "loss": 0.1746,
+      "step": 15358
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005304967530860309,
+      "loss": 0.0679,
+      "step": 15359
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005303727775272733,
+      "loss": 0.0697,
+      "step": 15360
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005302488112283787,
+      "loss": 0.0713,
+      "step": 15361
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005301248541917916,
+      "loss": 0.1327,
+      "step": 15362
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005300009064199558,
+      "loss": 0.0775,
+      "step": 15363
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005298769679153153,
+      "loss": 0.0613,
+      "step": 15364
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005297530386803136,
+      "loss": 0.0897,
+      "step": 15365
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005296291187173952,
+      "loss": 0.1014,
+      "step": 15366
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005295052080290025,
+      "loss": 0.0864,
+      "step": 15367
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005293813066175794,
+      "loss": 0.0891,
+      "step": 15368
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005292574144855683,
+      "loss": 0.1273,
+      "step": 15369
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005291335316354127,
+      "loss": 0.0977,
+      "step": 15370
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005290096580695545,
+      "loss": 0.0632,
+      "step": 15371
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000528885793790437,
+      "loss": 0.1178,
+      "step": 15372
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005287619388005021,
+      "loss": 0.1006,
+      "step": 15373
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005286380931021919,
+      "loss": 0.0858,
+      "step": 15374
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005285142566979484,
+      "loss": 0.098,
+      "step": 15375
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005283904295902131,
+      "loss": 0.0797,
+      "step": 15376
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005282666117814279,
+      "loss": 0.0968,
+      "step": 15377
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005281428032740334,
+      "loss": 0.1064,
+      "step": 15378
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.000528019004070472,
+      "loss": 0.1448,
+      "step": 15379
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005278952141731837,
+      "loss": 0.0889,
+      "step": 15380
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00052777143358461,
+      "loss": 0.1373,
+      "step": 15381
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005276476623071913,
+      "loss": 0.0605,
+      "step": 15382
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005275239003433679,
+      "loss": 0.0848,
+      "step": 15383
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0005274001476955803,
+      "loss": 0.0516,
+      "step": 15384
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005272764043662681,
+      "loss": 0.1256,
+      "step": 15385
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005271526703578721,
+      "loss": 0.0731,
+      "step": 15386
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005270289456728314,
+      "loss": 0.0725,
+      "step": 15387
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005269052303135856,
+      "loss": 0.0776,
+      "step": 15388
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005267815242825742,
+      "loss": 0.0919,
+      "step": 15389
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005266578275822357,
+      "loss": 0.0929,
+      "step": 15390
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005265341402150099,
+      "loss": 0.1302,
+      "step": 15391
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005264104621833357,
+      "loss": 0.1264,
+      "step": 15392
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005262867934896513,
+      "loss": 0.0669,
+      "step": 15393
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005261631341363953,
+      "loss": 0.1141,
+      "step": 15394
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005260394841260058,
+      "loss": 0.0935,
+      "step": 15395
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005259158434609211,
+      "loss": 0.0763,
+      "step": 15396
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005257922121435787,
+      "loss": 0.0836,
+      "step": 15397
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005256685901764162,
+      "loss": 0.1115,
+      "step": 15398
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005255449775618719,
+      "loss": 0.119,
+      "step": 15399
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005254213743023826,
+      "loss": 0.0733,
+      "step": 15400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005252977804003851,
+      "loss": 0.0913,
+      "step": 15401
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005251741958583172,
+      "loss": 0.0791,
+      "step": 15402
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000525050620678615,
+      "loss": 0.0606,
+      "step": 15403
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005249270548637156,
+      "loss": 0.1213,
+      "step": 15404
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005248034984160546,
+      "loss": 0.0715,
+      "step": 15405
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005246799513380692,
+      "loss": 0.0508,
+      "step": 15406
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005245564136321949,
+      "loss": 0.1241,
+      "step": 15407
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005244328853008679,
+      "loss": 0.0897,
+      "step": 15408
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005243093663465234,
+      "loss": 0.0861,
+      "step": 15409
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005241858567715967,
+      "loss": 0.0931,
+      "step": 15410
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005240623565785238,
+      "loss": 0.099,
+      "step": 15411
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005239388657697393,
+      "loss": 0.0781,
+      "step": 15412
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005238153843476786,
+      "loss": 0.0962,
+      "step": 15413
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005236919123147762,
+      "loss": 0.1307,
+      "step": 15414
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005235684496734666,
+      "loss": 0.0806,
+      "step": 15415
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005234449964261843,
+      "loss": 0.1138,
+      "step": 15416
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005233215525753632,
+      "loss": 0.0901,
+      "step": 15417
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005231981181234371,
+      "loss": 0.1095,
+      "step": 15418
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005230746930728408,
+      "loss": 0.0864,
+      "step": 15419
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005229512774260071,
+      "loss": 0.0809,
+      "step": 15420
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005228278711853697,
+      "loss": 0.0676,
+      "step": 15421
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005227044743533615,
+      "loss": 0.0731,
+      "step": 15422
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005225810869324164,
+      "loss": 0.1463,
+      "step": 15423
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005224577089249667,
+      "loss": 0.1127,
+      "step": 15424
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005223343403334449,
+      "loss": 0.0927,
+      "step": 15425
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005222109811602839,
+      "loss": 0.0587,
+      "step": 15426
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005220876314079163,
+      "loss": 0.0935,
+      "step": 15427
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005219642910787737,
+      "loss": 0.126,
+      "step": 15428
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005218409601752882,
+      "loss": 0.1066,
+      "step": 15429
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0005217176386998919,
+      "loss": 0.0786,
+      "step": 15430
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005215943266550155,
+      "loss": 0.1262,
+      "step": 15431
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005214710240430915,
+      "loss": 0.0976,
+      "step": 15432
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005213477308665501,
+      "loss": 0.1328,
+      "step": 15433
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005212244471278233,
+      "loss": 0.1087,
+      "step": 15434
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005211011728293418,
+      "loss": 0.1445,
+      "step": 15435
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005209779079735356,
+      "loss": 0.0829,
+      "step": 15436
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005208546525628358,
+      "loss": 0.0824,
+      "step": 15437
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005207314065996718,
+      "loss": 0.0911,
+      "step": 15438
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005206081700864747,
+      "loss": 0.0901,
+      "step": 15439
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005204849430256741,
+      "loss": 0.1121,
+      "step": 15440
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005203617254196997,
+      "loss": 0.095,
+      "step": 15441
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005202385172709812,
+      "loss": 0.1303,
+      "step": 15442
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005201153185819471,
+      "loss": 0.112,
+      "step": 15443
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005199921293550276,
+      "loss": 0.1592,
+      "step": 15444
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.000519868949592651,
+      "loss": 0.0778,
+      "step": 15445
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005197457792972468,
+      "loss": 0.0851,
+      "step": 15446
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005196226184712432,
+      "loss": 0.0803,
+      "step": 15447
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005194994671170685,
+      "loss": 0.1159,
+      "step": 15448
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005193763252371511,
+      "loss": 0.1003,
+      "step": 15449
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005192531928339191,
+      "loss": 0.0752,
+      "step": 15450
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005191300699097999,
+      "loss": 0.0812,
+      "step": 15451
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005190069564672219,
+      "loss": 0.116,
+      "step": 15452
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005188838525086121,
+      "loss": 0.076,
+      "step": 15453
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005187607580363976,
+      "loss": 0.0547,
+      "step": 15454
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005186376730530062,
+      "loss": 0.1438,
+      "step": 15455
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005185145975608645,
+      "loss": 0.0842,
+      "step": 15456
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005183915315623992,
+      "loss": 0.0895,
+      "step": 15457
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005182684750600365,
+      "loss": 0.06,
+      "step": 15458
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005181454280562035,
+      "loss": 0.0873,
+      "step": 15459
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005180223905533259,
+      "loss": 0.0517,
+      "step": 15460
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005178993625538298,
+      "loss": 0.0715,
+      "step": 15461
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.000517776344060141,
+      "loss": 0.1336,
+      "step": 15462
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005176533350746851,
+      "loss": 0.0944,
+      "step": 15463
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005175303355998871,
+      "loss": 0.0705,
+      "step": 15464
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005174073456381727,
+      "loss": 0.0852,
+      "step": 15465
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005172843651919672,
+      "loss": 0.1085,
+      "step": 15466
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005171613942636953,
+      "loss": 0.0919,
+      "step": 15467
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005170384328557814,
+      "loss": 0.0658,
+      "step": 15468
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005169154809706503,
+      "loss": 0.1176,
+      "step": 15469
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005167925386107261,
+      "loss": 0.1156,
+      "step": 15470
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005166696057784326,
+      "loss": 0.0831,
+      "step": 15471
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005165466824761944,
+      "loss": 0.0533,
+      "step": 15472
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005164237687064348,
+      "loss": 0.1228,
+      "step": 15473
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005163008644715777,
+      "loss": 0.0948,
+      "step": 15474
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005161779697740456,
+      "loss": 0.1211,
+      "step": 15475
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0005160550846162629,
+      "loss": 0.0884,
+      "step": 15476
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005159322090006518,
+      "loss": 0.0792,
+      "step": 15477
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.000515809342929635,
+      "loss": 0.1329,
+      "step": 15478
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005156864864056357,
+      "loss": 0.0848,
+      "step": 15479
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005155636394310761,
+      "loss": 0.0716,
+      "step": 15480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005154408020083783,
+      "loss": 0.0685,
+      "step": 15481
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005153179741399644,
+      "loss": 0.0801,
+      "step": 15482
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005151951558282562,
+      "loss": 0.0572,
+      "step": 15483
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005150723470756755,
+      "loss": 0.1183,
+      "step": 15484
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005149495478846433,
+      "loss": 0.093,
+      "step": 15485
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005148267582575812,
+      "loss": 0.1293,
+      "step": 15486
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005147039781969108,
+      "loss": 0.1053,
+      "step": 15487
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005145812077050524,
+      "loss": 0.111,
+      "step": 15488
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005144584467844271,
+      "loss": 0.125,
+      "step": 15489
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005143356954374552,
+      "loss": 0.0855,
+      "step": 15490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005142129536665568,
+      "loss": 0.093,
+      "step": 15491
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005140902214741526,
+      "loss": 0.1317,
+      "step": 15492
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005139674988626624,
+      "loss": 0.1161,
+      "step": 15493
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005138447858345059,
+      "loss": 0.0927,
+      "step": 15494
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005137220823921027,
+      "loss": 0.1116,
+      "step": 15495
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005135993885378718,
+      "loss": 0.1036,
+      "step": 15496
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005134767042742332,
+      "loss": 0.1432,
+      "step": 15497
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005133540296036052,
+      "loss": 0.0891,
+      "step": 15498
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005132313645284072,
+      "loss": 0.0795,
+      "step": 15499
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005131087090510577,
+      "loss": 0.1964,
+      "step": 15500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005129860631739752,
+      "loss": 0.0843,
+      "step": 15501
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005128634268995777,
+      "loss": 0.0942,
+      "step": 15502
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005127408002302833,
+      "loss": 0.095,
+      "step": 15503
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005126181831685101,
+      "loss": 0.1323,
+      "step": 15504
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005124955757166755,
+      "loss": 0.0959,
+      "step": 15505
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005123729778771975,
+      "loss": 0.0773,
+      "step": 15506
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005122503896524926,
+      "loss": 0.0973,
+      "step": 15507
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005121278110449789,
+      "loss": 0.1693,
+      "step": 15508
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.000512005242057073,
+      "loss": 0.1179,
+      "step": 15509
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005118826826911915,
+      "loss": 0.0698,
+      "step": 15510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005117601329497506,
+      "loss": 0.0705,
+      "step": 15511
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005116375928351676,
+      "loss": 0.1541,
+      "step": 15512
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005115150623498578,
+      "loss": 0.0698,
+      "step": 15513
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005113925414962379,
+      "loss": 0.1366,
+      "step": 15514
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005112700302767233,
+      "loss": 0.0859,
+      "step": 15515
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005111475286937297,
+      "loss": 0.0966,
+      "step": 15516
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.000511025036749672,
+      "loss": 0.0815,
+      "step": 15517
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.000510902554446966,
+      "loss": 0.0806,
+      "step": 15518
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.000510780081788027,
+      "loss": 0.0962,
+      "step": 15519
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005106576187752696,
+      "loss": 0.106,
+      "step": 15520
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005105351654111083,
+      "loss": 0.0952,
+      "step": 15521
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0005104127216979577,
+      "loss": 0.0627,
+      "step": 15522
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005102902876382321,
+      "loss": 0.0951,
+      "step": 15523
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005101678632343454,
+      "loss": 0.0863,
+      "step": 15524
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005100454484887111,
+      "loss": 0.0651,
+      "step": 15525
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000509923043403744,
+      "loss": 0.0898,
+      "step": 15526
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005098006479818571,
+      "loss": 0.0731,
+      "step": 15527
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005096782622254632,
+      "loss": 0.1638,
+      "step": 15528
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005095558861369765,
+      "loss": 0.0884,
+      "step": 15529
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005094335197188091,
+      "loss": 0.0628,
+      "step": 15530
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005093111629733738,
+      "loss": 0.0685,
+      "step": 15531
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005091888159030837,
+      "loss": 0.0956,
+      "step": 15532
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005090664785103509,
+      "loss": 0.0717,
+      "step": 15533
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005089441507975876,
+      "loss": 0.0655,
+      "step": 15534
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005088218327672056,
+      "loss": 0.1309,
+      "step": 15535
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000508699524421617,
+      "loss": 0.1083,
+      "step": 15536
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000508577225763233,
+      "loss": 0.1174,
+      "step": 15537
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005084549367944652,
+      "loss": 0.1221,
+      "step": 15538
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005083326575177247,
+      "loss": 0.0946,
+      "step": 15539
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005082103879354233,
+      "loss": 0.1042,
+      "step": 15540
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005080881280499712,
+      "loss": 0.079,
+      "step": 15541
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005079658778637792,
+      "loss": 0.0807,
+      "step": 15542
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005078436373792577,
+      "loss": 0.0927,
+      "step": 15543
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005077214065988169,
+      "loss": 0.0806,
+      "step": 15544
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005075991855248666,
+      "loss": 0.0742,
+      "step": 15545
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005074769741598176,
+      "loss": 0.0984,
+      "step": 15546
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005073547725060789,
+      "loss": 0.0969,
+      "step": 15547
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005072325805660602,
+      "loss": 0.0767,
+      "step": 15548
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005071103983421704,
+      "loss": 0.0798,
+      "step": 15549
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005069882258368195,
+      "loss": 0.1014,
+      "step": 15550
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005068660630524153,
+      "loss": 0.1517,
+      "step": 15551
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005067439099913679,
+      "loss": 0.0699,
+      "step": 15552
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005066217666560849,
+      "loss": 0.0908,
+      "step": 15553
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005064996330489751,
+      "loss": 0.111,
+      "step": 15554
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005063775091724463,
+      "loss": 0.0847,
+      "step": 15555
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005062553950289066,
+      "loss": 0.0844,
+      "step": 15556
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005061332906207639,
+      "loss": 0.1167,
+      "step": 15557
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005060111959504254,
+      "loss": 0.0957,
+      "step": 15558
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000505889111020299,
+      "loss": 0.0833,
+      "step": 15559
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005057670358327914,
+      "loss": 0.0682,
+      "step": 15560
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005056449703903104,
+      "loss": 0.0806,
+      "step": 15561
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005055229146952624,
+      "loss": 0.114,
+      "step": 15562
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005054008687500538,
+      "loss": 0.0925,
+      "step": 15563
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005052788325570913,
+      "loss": 0.0746,
+      "step": 15564
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005051568061187807,
+      "loss": 0.0857,
+      "step": 15565
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005050347894375287,
+      "loss": 0.1141,
+      "step": 15566
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.000504912782515741,
+      "loss": 0.0972,
+      "step": 15567
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0005047907853558231,
+      "loss": 0.0624,
+      "step": 15568
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005046687979601805,
+      "loss": 0.091,
+      "step": 15569
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005045468203312182,
+      "loss": 0.1196,
+      "step": 15570
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000504424852471342,
+      "loss": 0.0957,
+      "step": 15571
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005043028943829557,
+      "loss": 0.1159,
+      "step": 15572
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005041809460684654,
+      "loss": 0.0717,
+      "step": 15573
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005040590075302749,
+      "loss": 0.1166,
+      "step": 15574
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005039370787707884,
+      "loss": 0.1196,
+      "step": 15575
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005038151597924103,
+      "loss": 0.0955,
+      "step": 15576
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005036932505975443,
+      "loss": 0.0688,
+      "step": 15577
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005035713511885938,
+      "loss": 0.0917,
+      "step": 15578
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005034494615679632,
+      "loss": 0.0969,
+      "step": 15579
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005033275817380555,
+      "loss": 0.0738,
+      "step": 15580
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005032057117012734,
+      "loss": 0.0718,
+      "step": 15581
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005030838514600206,
+      "loss": 0.0685,
+      "step": 15582
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005029620010166998,
+      "loss": 0.1199,
+      "step": 15583
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005028401603737131,
+      "loss": 0.0731,
+      "step": 15584
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005027183295334628,
+      "loss": 0.0717,
+      "step": 15585
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005025965084983517,
+      "loss": 0.0814,
+      "step": 15586
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005024746972707815,
+      "loss": 0.1005,
+      "step": 15587
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005023528958531542,
+      "loss": 0.0634,
+      "step": 15588
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005022311042478711,
+      "loss": 0.0705,
+      "step": 15589
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005021093224573339,
+      "loss": 0.0967,
+      "step": 15590
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005019875504839433,
+      "loss": 0.0845,
+      "step": 15591
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000501865788330101,
+      "loss": 0.1071,
+      "step": 15592
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005017440359982073,
+      "loss": 0.0703,
+      "step": 15593
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005016222934906636,
+      "loss": 0.0763,
+      "step": 15594
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005015005608098697,
+      "loss": 0.0768,
+      "step": 15595
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000501378837958226,
+      "loss": 0.1113,
+      "step": 15596
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005012571249381328,
+      "loss": 0.0778,
+      "step": 15597
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005011354217519892,
+      "loss": 0.1134,
+      "step": 15598
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000501013728402196,
+      "loss": 0.0928,
+      "step": 15599
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005008920448911522,
+      "loss": 0.1078,
+      "step": 15600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005007703712212569,
+      "loss": 0.0891,
+      "step": 15601
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005006487073949091,
+      "loss": 0.0853,
+      "step": 15602
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005005270534145081,
+      "loss": 0.0608,
+      "step": 15603
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005004054092824526,
+      "loss": 0.1077,
+      "step": 15604
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005002837750011406,
+      "loss": 0.0701,
+      "step": 15605
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0005001621505729712,
+      "loss": 0.1016,
+      "step": 15606
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000500040536000342,
+      "loss": 0.0622,
+      "step": 15607
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000499918931285651,
+      "loss": 0.0791,
+      "step": 15608
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004997973364312961,
+      "loss": 0.0891,
+      "step": 15609
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004996757514396747,
+      "loss": 0.1289,
+      "step": 15610
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004995541763131838,
+      "loss": 0.1118,
+      "step": 15611
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004994326110542212,
+      "loss": 0.1105,
+      "step": 15612
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004993110556651834,
+      "loss": 0.0876,
+      "step": 15613
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004991895101484676,
+      "loss": 0.0729,
+      "step": 15614
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0004990679745064703,
+      "loss": 0.085,
+      "step": 15615
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004989464487415874,
+      "loss": 0.1342,
+      "step": 15616
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004988249328562157,
+      "loss": 0.0933,
+      "step": 15617
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004987034268527503,
+      "loss": 0.0692,
+      "step": 15618
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000498581930733588,
+      "loss": 0.061,
+      "step": 15619
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004984604445011239,
+      "loss": 0.0965,
+      "step": 15620
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004983389681577535,
+      "loss": 0.1,
+      "step": 15621
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000498217501705872,
+      "loss": 0.0869,
+      "step": 15622
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004980960451478739,
+      "loss": 0.0721,
+      "step": 15623
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000497974598486155,
+      "loss": 0.074,
+      "step": 15624
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004978531617231089,
+      "loss": 0.0817,
+      "step": 15625
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004977317348611311,
+      "loss": 0.0849,
+      "step": 15626
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004976103179026151,
+      "loss": 0.0647,
+      "step": 15627
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004974889108499552,
+      "loss": 0.0875,
+      "step": 15628
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004973675137055451,
+      "loss": 0.0519,
+      "step": 15629
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004972461264717786,
+      "loss": 0.0739,
+      "step": 15630
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004971247491510485,
+      "loss": 0.1487,
+      "step": 15631
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004970033817457491,
+      "loss": 0.0712,
+      "step": 15632
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000496882024258273,
+      "loss": 0.0715,
+      "step": 15633
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004967606766910125,
+      "loss": 0.0434,
+      "step": 15634
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004966393390463613,
+      "loss": 0.0437,
+      "step": 15635
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004965180113267112,
+      "loss": 0.1431,
+      "step": 15636
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004963966935344549,
+      "loss": 0.082,
+      "step": 15637
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004962753856719836,
+      "loss": 0.1115,
+      "step": 15638
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004961540877416904,
+      "loss": 0.0704,
+      "step": 15639
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004960327997459662,
+      "loss": 0.087,
+      "step": 15640
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004959115216872027,
+      "loss": 0.0713,
+      "step": 15641
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004957902535677912,
+      "loss": 0.0515,
+      "step": 15642
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004956689953901228,
+      "loss": 0.075,
+      "step": 15643
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004955477471565881,
+      "loss": 0.0876,
+      "step": 15644
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004954265088695783,
+      "loss": 0.1273,
+      "step": 15645
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004953052805314834,
+      "loss": 0.097,
+      "step": 15646
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004951840621446945,
+      "loss": 0.0858,
+      "step": 15647
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004950628537116011,
+      "loss": 0.0973,
+      "step": 15648
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004949416552345933,
+      "loss": 0.0787,
+      "step": 15649
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000494820466716061,
+      "loss": 0.1052,
+      "step": 15650
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004946992881583929,
+      "loss": 0.0919,
+      "step": 15651
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004945781195639795,
+      "loss": 0.1272,
+      "step": 15652
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004944569609352093,
+      "loss": 0.1,
+      "step": 15653
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004943358122744714,
+      "loss": 0.0829,
+      "step": 15654
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000494214673584154,
+      "loss": 0.0989,
+      "step": 15655
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004940935448666467,
+      "loss": 0.1079,
+      "step": 15656
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004939724261243373,
+      "loss": 0.0807,
+      "step": 15657
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004938513173596134,
+      "loss": 0.1016,
+      "step": 15658
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000493730218574864,
+      "loss": 0.0955,
+      "step": 15659
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0004936091297724763,
+      "loss": 0.0995,
+      "step": 15660
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000493488050954838,
+      "loss": 0.075,
+      "step": 15661
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004933669821243364,
+      "loss": 0.1138,
+      "step": 15662
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004932459232833587,
+      "loss": 0.0963,
+      "step": 15663
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004931248744342914,
+      "loss": 0.069,
+      "step": 15664
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004930038355795221,
+      "loss": 0.0753,
+      "step": 15665
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004928828067214372,
+      "loss": 0.1201,
+      "step": 15666
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004927617878624223,
+      "loss": 0.1086,
+      "step": 15667
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004926407790048645,
+      "loss": 0.0654,
+      "step": 15668
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004925197801511497,
+      "loss": 0.0659,
+      "step": 15669
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004923987913036633,
+      "loss": 0.0971,
+      "step": 15670
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004922778124647908,
+      "loss": 0.0962,
+      "step": 15671
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004921568436369181,
+      "loss": 0.114,
+      "step": 15672
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004920358848224303,
+      "loss": 0.0757,
+      "step": 15673
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004919149360237122,
+      "loss": 0.0935,
+      "step": 15674
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004917939972431487,
+      "loss": 0.0598,
+      "step": 15675
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004916730684831239,
+      "loss": 0.0925,
+      "step": 15676
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000491552149746023,
+      "loss": 0.0692,
+      "step": 15677
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004914312410342295,
+      "loss": 0.1235,
+      "step": 15678
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004913103423501284,
+      "loss": 0.0779,
+      "step": 15679
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004911894536961028,
+      "loss": 0.0862,
+      "step": 15680
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004910685750745364,
+      "loss": 0.1016,
+      "step": 15681
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004909477064878127,
+      "loss": 0.0838,
+      "step": 15682
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004908268479383149,
+      "loss": 0.0823,
+      "step": 15683
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004907059994284256,
+      "loss": 0.1022,
+      "step": 15684
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004905851609605284,
+      "loss": 0.1316,
+      "step": 15685
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004904643325370056,
+      "loss": 0.0846,
+      "step": 15686
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004903435141602396,
+      "loss": 0.1061,
+      "step": 15687
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004902227058326122,
+      "loss": 0.1188,
+      "step": 15688
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004901019075565063,
+      "loss": 0.073,
+      "step": 15689
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004899811193343032,
+      "loss": 0.0955,
+      "step": 15690
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004898603411683844,
+      "loss": 0.1373,
+      "step": 15691
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000489739573061132,
+      "loss": 0.1095,
+      "step": 15692
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004896188150149267,
+      "loss": 0.0679,
+      "step": 15693
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004894980670321496,
+      "loss": 0.1046,
+      "step": 15694
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004893773291151818,
+      "loss": 0.1729,
+      "step": 15695
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004892566012664036,
+      "loss": 0.1084,
+      "step": 15696
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004891358834881954,
+      "loss": 0.0888,
+      "step": 15697
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004890151757829379,
+      "loss": 0.1094,
+      "step": 15698
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004888944781530106,
+      "loss": 0.1002,
+      "step": 15699
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000488773790600794,
+      "loss": 0.1273,
+      "step": 15700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004886531131286674,
+      "loss": 0.0673,
+      "step": 15701
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004885324457390102,
+      "loss": 0.0705,
+      "step": 15702
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0004884117884342018,
+      "loss": 0.0965,
+      "step": 15703
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.000488291141216621,
+      "loss": 0.1023,
+      "step": 15704
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00048817050408864703,
+      "loss": 0.0996,
+      "step": 15705
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00048804987705265837,
+      "loss": 0.1353,
+      "step": 15706
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00048792926011103347,
+      "loss": 0.0842,
+      "step": 15707
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004878086532661502,
+      "loss": 0.0779,
+      "step": 15708
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004876880565203875,
+      "loss": 0.0816,
+      "step": 15709
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048756746987612266,
+      "loss": 0.0848,
+      "step": 15710
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048744689333573313,
+      "loss": 0.0584,
+      "step": 15711
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048732632690159694,
+      "loss": 0.1266,
+      "step": 15712
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048720577057609116,
+      "loss": 0.103,
+      "step": 15713
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048708522436159284,
+      "loss": 0.094,
+      "step": 15714
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048696468826047877,
+      "loss": 0.1107,
+      "step": 15715
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048684416227512563,
+      "loss": 0.1121,
+      "step": 15716
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004867236464079099,
+      "loss": 0.0712,
+      "step": 15717
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048660314066120746,
+      "loss": 0.1358,
+      "step": 15718
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048648264503739513,
+      "loss": 0.0869,
+      "step": 15719
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004863621595388481,
+      "loss": 0.1093,
+      "step": 15720
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004862416841679428,
+      "loss": 0.0818,
+      "step": 15721
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048612121892705417,
+      "loss": 0.0884,
+      "step": 15722
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004860007638185576,
+      "loss": 0.1128,
+      "step": 15723
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048588031884482777,
+      "loss": 0.1163,
+      "step": 15724
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048575988400824025,
+      "loss": 0.098,
+      "step": 15725
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004856394593111694,
+      "loss": 0.0995,
+      "step": 15726
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004855190447559896,
+      "loss": 0.0778,
+      "step": 15727
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004853986403450752,
+      "loss": 0.1359,
+      "step": 15728
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048527824608079995,
+      "loss": 0.0852,
+      "step": 15729
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048515786196553834,
+      "loss": 0.1075,
+      "step": 15730
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048503748800166334,
+      "loss": 0.0643,
+      "step": 15731
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004849171241915493,
+      "loss": 0.152,
+      "step": 15732
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004847967705375689,
+      "loss": 0.1241,
+      "step": 15733
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048467642704209534,
+      "loss": 0.0978,
+      "step": 15734
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048455609370750153,
+      "loss": 0.0756,
+      "step": 15735
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048443577053615996,
+      "loss": 0.1173,
+      "step": 15736
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048431545753044325,
+      "loss": 0.1152,
+      "step": 15737
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048419515469272324,
+      "loss": 0.1115,
+      "step": 15738
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048407486202537277,
+      "loss": 0.0783,
+      "step": 15739
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048395457953076326,
+      "loss": 0.1271,
+      "step": 15740
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004838343072112662,
+      "loss": 0.1165,
+      "step": 15741
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048371404506925356,
+      "loss": 0.1206,
+      "step": 15742
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048359379310709637,
+      "loss": 0.065,
+      "step": 15743
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004834735513271653,
+      "loss": 0.0749,
+      "step": 15744
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048335331973183194,
+      "loss": 0.0815,
+      "step": 15745
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004832330983234666,
+      "loss": 0.0677,
+      "step": 15746
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048311288710443977,
+      "loss": 0.0781,
+      "step": 15747
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048299268607712167,
+      "loss": 0.0655,
+      "step": 15748
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048287249524388244,
+      "loss": 0.0906,
+      "step": 15749
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048275231460709133,
+      "loss": 0.0786,
+      "step": 15750
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048263214416911906,
+      "loss": 0.0898,
+      "step": 15751
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0004825119839323341,
+      "loss": 0.0963,
+      "step": 15752
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00048239183389910654,
+      "loss": 0.0988,
+      "step": 15753
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004822716940718052,
+      "loss": 0.0994,
+      "step": 15754
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048215156445279863,
+      "loss": 0.0945,
+      "step": 15755
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004820314450444557,
+      "loss": 0.074,
+      "step": 15756
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048191133584914495,
+      "loss": 0.0969,
+      "step": 15757
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000481791236869234,
+      "loss": 0.083,
+      "step": 15758
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004816711481070918,
+      "loss": 0.0883,
+      "step": 15759
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048155106956508577,
+      "loss": 0.0687,
+      "step": 15760
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004814310012455836,
+      "loss": 0.0664,
+      "step": 15761
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048131094315095223,
+      "loss": 0.0655,
+      "step": 15762
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048119089528355995,
+      "loss": 0.1113,
+      "step": 15763
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048107085764577276,
+      "loss": 0.1055,
+      "step": 15764
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048095083023995833,
+      "loss": 0.0401,
+      "step": 15765
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000480830813068483,
+      "loss": 0.0962,
+      "step": 15766
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004807108061337131,
+      "loss": 0.0887,
+      "step": 15767
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004805908094380149,
+      "loss": 0.1332,
+      "step": 15768
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004804708229837544,
+      "loss": 0.0834,
+      "step": 15769
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048035084677329763,
+      "loss": 0.0843,
+      "step": 15770
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00048023088080900955,
+      "loss": 0.1346,
+      "step": 15771
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004801109250932566,
+      "loss": 0.0925,
+      "step": 15772
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000479990979628403,
+      "loss": 0.0703,
+      "step": 15773
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004798710444168147,
+      "loss": 0.0808,
+      "step": 15774
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000479751119460856,
+      "loss": 0.1221,
+      "step": 15775
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004796312047628918,
+      "loss": 0.0972,
+      "step": 15776
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047951130032528613,
+      "loss": 0.0983,
+      "step": 15777
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004793914061504031,
+      "loss": 0.0978,
+      "step": 15778
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047927152224060745,
+      "loss": 0.0975,
+      "step": 15779
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004791516485982624,
+      "loss": 0.0927,
+      "step": 15780
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047903178522573186,
+      "loss": 0.0804,
+      "step": 15781
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004789119321253791,
+      "loss": 0.1239,
+      "step": 15782
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047879208929956684,
+      "loss": 0.0919,
+      "step": 15783
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004786722567506585,
+      "loss": 0.0712,
+      "step": 15784
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004785524344810174,
+      "loss": 0.1035,
+      "step": 15785
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004784326224930057,
+      "loss": 0.0787,
+      "step": 15786
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047831282078898566,
+      "loss": 0.0977,
+      "step": 15787
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047819302937131947,
+      "loss": 0.0709,
+      "step": 15788
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004780732482423693,
+      "loss": 0.1071,
+      "step": 15789
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047795347740449667,
+      "loss": 0.0846,
+      "step": 15790
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047783371686006316,
+      "loss": 0.1346,
+      "step": 15791
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047771396661143054,
+      "loss": 0.0826,
+      "step": 15792
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047759422666095977,
+      "loss": 0.0917,
+      "step": 15793
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004774744970110113,
+      "loss": 0.0449,
+      "step": 15794
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047735477766394695,
+      "loss": 0.1139,
+      "step": 15795
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004772350686221266,
+      "loss": 0.0976,
+      "step": 15796
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047711536988791084,
+      "loss": 0.1313,
+      "step": 15797
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00047699568146365923,
+      "loss": 0.0956,
+      "step": 15798
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0004768760033517325,
+      "loss": 0.0731,
+      "step": 15799
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004767563355544903,
+      "loss": 0.0682,
+      "step": 15800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004766366780742919,
+      "loss": 0.0725,
+      "step": 15801
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047651703091349673,
+      "loss": 0.0549,
+      "step": 15802
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004763973940744636,
+      "loss": 0.0728,
+      "step": 15803
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004762777675595521,
+      "loss": 0.0829,
+      "step": 15804
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004761581513711203,
+      "loss": 0.0793,
+      "step": 15805
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047603854551152736,
+      "loss": 0.0872,
+      "step": 15806
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047591894998313136,
+      "loss": 0.0858,
+      "step": 15807
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047579936478829045,
+      "loss": 0.0522,
+      "step": 15808
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047567978992936233,
+      "loss": 0.105,
+      "step": 15809
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047556022540870503,
+      "loss": 0.0842,
+      "step": 15810
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004754406712286754,
+      "loss": 0.1046,
+      "step": 15811
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004753211273916316,
+      "loss": 0.0735,
+      "step": 15812
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047520159389993034,
+      "loss": 0.0933,
+      "step": 15813
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047508207075592867,
+      "loss": 0.076,
+      "step": 15814
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004749625579619827,
+      "loss": 0.0945,
+      "step": 15815
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004748430555204498,
+      "loss": 0.0521,
+      "step": 15816
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047472356343368586,
+      "loss": 0.0525,
+      "step": 15817
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047460408170404647,
+      "loss": 0.0779,
+      "step": 15818
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047448461033388836,
+      "loss": 0.1056,
+      "step": 15819
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047436514932556686,
+      "loss": 0.1077,
+      "step": 15820
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004742456986814374,
+      "loss": 0.0846,
+      "step": 15821
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004741262584038553,
+      "loss": 0.1356,
+      "step": 15822
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004740068284951754,
+      "loss": 0.0834,
+      "step": 15823
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004738874089577525,
+      "loss": 0.0688,
+      "step": 15824
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004737679997939418,
+      "loss": 0.0858,
+      "step": 15825
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047364860100609706,
+      "loss": 0.1019,
+      "step": 15826
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004735292125965732,
+      "loss": 0.1016,
+      "step": 15827
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.000473409834567724,
+      "loss": 0.0825,
+      "step": 15828
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047329046692190323,
+      "loss": 0.0726,
+      "step": 15829
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004731711096614646,
+      "loss": 0.0753,
+      "step": 15830
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047305176278876097,
+      "loss": 0.1293,
+      "step": 15831
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047293242630614643,
+      "loss": 0.0909,
+      "step": 15832
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004728131002159737,
+      "loss": 0.093,
+      "step": 15833
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004726937845205954,
+      "loss": 0.108,
+      "step": 15834
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004725744792223643,
+      "loss": 0.1029,
+      "step": 15835
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004724551843236323,
+      "loss": 0.1188,
+      "step": 15836
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004723358998267524,
+      "loss": 0.0739,
+      "step": 15837
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047221662573407585,
+      "loss": 0.1558,
+      "step": 15838
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004720973620479553,
+      "loss": 0.084,
+      "step": 15839
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047197810877074167,
+      "loss": 0.0734,
+      "step": 15840
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047185886590478653,
+      "loss": 0.0983,
+      "step": 15841
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004717396334524411,
+      "loss": 0.0994,
+      "step": 15842
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004716204114160562,
+      "loss": 0.0822,
+      "step": 15843
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00047150119979798223,
+      "loss": 0.1014,
+      "step": 15844
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0004713819986005705,
+      "loss": 0.0961,
+      "step": 15845
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00047126280782617115,
+      "loss": 0.1243,
+      "step": 15846
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00047114362747713366,
+      "loss": 0.1036,
+      "step": 15847
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004710244575558089,
+      "loss": 0.0491,
+      "step": 15848
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004709052980645463,
+      "loss": 0.1306,
+      "step": 15849
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00047078614900569515,
+      "loss": 0.1172,
+      "step": 15850
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004706670103816045,
+      "loss": 0.0856,
+      "step": 15851
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004705478821946243,
+      "loss": 0.1139,
+      "step": 15852
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000470428764447103,
+      "loss": 0.0939,
+      "step": 15853
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004703096571413893,
+      "loss": 0.0562,
+      "step": 15854
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00047019056027983164,
+      "loss": 0.0775,
+      "step": 15855
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00047007147386477854,
+      "loss": 0.0719,
+      "step": 15856
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046995239789857736,
+      "loss": 0.0739,
+      "step": 15857
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004698333323835766,
+      "loss": 0.0911,
+      "step": 15858
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046971427732212434,
+      "loss": 0.0699,
+      "step": 15859
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046959523271656743,
+      "loss": 0.1135,
+      "step": 15860
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004694761985692534,
+      "loss": 0.0914,
+      "step": 15861
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046935717488252903,
+      "loss": 0.0958,
+      "step": 15862
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046923816165874143,
+      "loss": 0.1044,
+      "step": 15863
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004691191589002367,
+      "loss": 0.1414,
+      "step": 15864
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046900016660936206,
+      "loss": 0.0751,
+      "step": 15865
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046888118478846343,
+      "loss": 0.0704,
+      "step": 15866
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004687622134398868,
+      "loss": 0.065,
+      "step": 15867
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004686432525659775,
+      "loss": 0.086,
+      "step": 15868
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004685243021690819,
+      "loss": 0.085,
+      "step": 15869
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004684053622515453,
+      "loss": 0.0685,
+      "step": 15870
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046828643281571214,
+      "loss": 0.0701,
+      "step": 15871
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004681675138639284,
+      "loss": 0.0915,
+      "step": 15872
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004680486053985384,
+      "loss": 0.0917,
+      "step": 15873
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046792970742188687,
+      "loss": 0.1171,
+      "step": 15874
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046781081993631793,
+      "loss": 0.0747,
+      "step": 15875
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004676919429441758,
+      "loss": 0.0511,
+      "step": 15876
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004675730764478047,
+      "loss": 0.0725,
+      "step": 15877
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004674542204495476,
+      "loss": 0.111,
+      "step": 15878
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004673353749517486,
+      "loss": 0.098,
+      "step": 15879
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004672165399567515,
+      "loss": 0.1321,
+      "step": 15880
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004670977154668989,
+      "loss": 0.1197,
+      "step": 15881
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046697890148453394,
+      "loss": 0.0967,
+      "step": 15882
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000466860098011999,
+      "loss": 0.0563,
+      "step": 15883
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046674130505163647,
+      "loss": 0.092,
+      "step": 15884
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004666225226057893,
+      "loss": 0.0724,
+      "step": 15885
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046650375067679926,
+      "loss": 0.1511,
+      "step": 15886
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004663849892670081,
+      "loss": 0.0765,
+      "step": 15887
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004662662383787577,
+      "loss": 0.0644,
+      "step": 15888
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00046614749801438893,
+      "loss": 0.0898,
+      "step": 15889
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.000466028768176244,
+      "loss": 0.0753,
+      "step": 15890
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0004659100488666631,
+      "loss": 0.0668,
+      "step": 15891
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046579134008798784,
+      "loss": 0.0765,
+      "step": 15892
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004656726418425584,
+      "loss": 0.0665,
+      "step": 15893
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004655539541327154,
+      "loss": 0.1157,
+      "step": 15894
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004654352769607989,
+      "loss": 0.1018,
+      "step": 15895
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046531661032914905,
+      "loss": 0.1044,
+      "step": 15896
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046519795424010527,
+      "loss": 0.0947,
+      "step": 15897
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004650793086960077,
+      "loss": 0.134,
+      "step": 15898
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004649606736991955,
+      "loss": 0.0924,
+      "step": 15899
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046484204925200756,
+      "loss": 0.0665,
+      "step": 15900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004647234353567835,
+      "loss": 0.0746,
+      "step": 15901
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046460483201586167,
+      "loss": 0.0771,
+      "step": 15902
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046448623923158085,
+      "loss": 0.1533,
+      "step": 15903
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004643676570062787,
+      "loss": 0.0914,
+      "step": 15904
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004642490853422944,
+      "loss": 0.0663,
+      "step": 15905
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004641305242419652,
+      "loss": 0.0759,
+      "step": 15906
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046401197370762913,
+      "loss": 0.0811,
+      "step": 15907
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004638934337416236,
+      "loss": 0.0848,
+      "step": 15908
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046377490434628587,
+      "loss": 0.0875,
+      "step": 15909
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004636563855239527,
+      "loss": 0.0901,
+      "step": 15910
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004635378772769617,
+      "loss": 0.1028,
+      "step": 15911
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004634193796076489,
+      "loss": 0.0961,
+      "step": 15912
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046330089251835137,
+      "loss": 0.0603,
+      "step": 15913
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004631824160114052,
+      "loss": 0.1184,
+      "step": 15914
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046306395008914624,
+      "loss": 0.058,
+      "step": 15915
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046294549475391045,
+      "loss": 0.0655,
+      "step": 15916
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046282705000803305,
+      "loss": 0.1089,
+      "step": 15917
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046270861585385027,
+      "loss": 0.0903,
+      "step": 15918
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046259019229369705,
+      "loss": 0.1248,
+      "step": 15919
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004624717793299081,
+      "loss": 0.0742,
+      "step": 15920
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004623533769648182,
+      "loss": 0.0669,
+      "step": 15921
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046223498520076245,
+      "loss": 0.0744,
+      "step": 15922
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046211660404007506,
+      "loss": 0.1044,
+      "step": 15923
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046199823348508964,
+      "loss": 0.073,
+      "step": 15924
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046187987353814097,
+      "loss": 0.0751,
+      "step": 15925
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046176152420156246,
+      "loss": 0.0804,
+      "step": 15926
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004616431854776877,
+      "loss": 0.0677,
+      "step": 15927
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046152485736884985,
+      "loss": 0.069,
+      "step": 15928
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046140653987738236,
+      "loss": 0.0564,
+      "step": 15929
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046128823300561786,
+      "loss": 0.0952,
+      "step": 15930
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046116993675588883,
+      "loss": 0.0774,
+      "step": 15931
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046105165113052805,
+      "loss": 0.0814,
+      "step": 15932
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046093337613186834,
+      "loss": 0.0776,
+      "step": 15933
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046081511176224133,
+      "loss": 0.061,
+      "step": 15934
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0004606968580239789,
+      "loss": 0.0811,
+      "step": 15935
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046057861491941254,
+      "loss": 0.0873,
+      "step": 15936
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00046046038245087364,
+      "loss": 0.0768,
+      "step": 15937
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.000460342160620694,
+      "loss": 0.0744,
+      "step": 15938
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004602239494312044,
+      "loss": 0.1078,
+      "step": 15939
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00046010574888473557,
+      "loss": 0.077,
+      "step": 15940
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004599875589836181,
+      "loss": 0.1027,
+      "step": 15941
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045986937973018204,
+      "loss": 0.1063,
+      "step": 15942
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045975121112675843,
+      "loss": 0.0909,
+      "step": 15943
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004596330531756765,
+      "loss": 0.0926,
+      "step": 15944
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045951490587926657,
+      "loss": 0.102,
+      "step": 15945
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045939676923985816,
+      "loss": 0.1177,
+      "step": 15946
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045927864325978043,
+      "loss": 0.0975,
+      "step": 15947
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004591605279413624,
+      "loss": 0.103,
+      "step": 15948
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045904242328693323,
+      "loss": 0.0809,
+      "step": 15949
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045892432929882165,
+      "loss": 0.071,
+      "step": 15950
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004588062459793556,
+      "loss": 0.1252,
+      "step": 15951
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004586881733308643,
+      "loss": 0.0609,
+      "step": 15952
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045857011135567506,
+      "loss": 0.0562,
+      "step": 15953
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045845206005611653,
+      "loss": 0.1057,
+      "step": 15954
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004583340194345159,
+      "loss": 0.0628,
+      "step": 15955
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045821598949320085,
+      "loss": 0.0878,
+      "step": 15956
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004580979702344981,
+      "loss": 0.0971,
+      "step": 15957
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004579799616607353,
+      "loss": 0.167,
+      "step": 15958
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004578619637742392,
+      "loss": 0.0739,
+      "step": 15959
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045774397657733624,
+      "loss": 0.0634,
+      "step": 15960
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045762600007235286,
+      "loss": 0.1016,
+      "step": 15961
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004575080342616154,
+      "loss": 0.08,
+      "step": 15962
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045739007914744924,
+      "loss": 0.0582,
+      "step": 15963
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.000457272134732181,
+      "loss": 0.0782,
+      "step": 15964
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045715420101813555,
+      "loss": 0.0633,
+      "step": 15965
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004570362780076389,
+      "loss": 0.0957,
+      "step": 15966
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045691836570301605,
+      "loss": 0.0917,
+      "step": 15967
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004568004641065916,
+      "loss": 0.0771,
+      "step": 15968
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004566825732206905,
+      "loss": 0.0591,
+      "step": 15969
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045656469304763714,
+      "loss": 0.0981,
+      "step": 15970
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045644682358975553,
+      "loss": 0.0664,
+      "step": 15971
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045632896484937047,
+      "loss": 0.0907,
+      "step": 15972
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045621111682880557,
+      "loss": 0.0623,
+      "step": 15973
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045609327953038395,
+      "loss": 0.0989,
+      "step": 15974
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045597545295642986,
+      "loss": 0.0779,
+      "step": 15975
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004558576371092661,
+      "loss": 0.1361,
+      "step": 15976
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004557398319912155,
+      "loss": 0.0609,
+      "step": 15977
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045562203760460165,
+      "loss": 0.1085,
+      "step": 15978
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045550425395174653,
+      "loss": 0.0632,
+      "step": 15979
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004553864810349727,
+      "loss": 0.0782,
+      "step": 15980
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004552687188566024,
+      "loss": 0.0889,
+      "step": 15981
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0004551509674189575,
+      "loss": 0.1143,
+      "step": 15982
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045503322672435975,
+      "loss": 0.0798,
+      "step": 15983
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00045491549677513044,
+      "loss": 0.1274,
+      "step": 15984
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004547977775735915,
+      "loss": 0.0706,
+      "step": 15985
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004546800691220634,
+      "loss": 0.0799,
+      "step": 15986
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004545623714228678,
+      "loss": 0.1069,
+      "step": 15987
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004544446844783251,
+      "loss": 0.1168,
+      "step": 15988
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004543270082907557,
+      "loss": 0.0606,
+      "step": 15989
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045420934286247985,
+      "loss": 0.0643,
+      "step": 15990
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045409168819581723,
+      "loss": 0.0604,
+      "step": 15991
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004539740442930885,
+      "loss": 0.0643,
+      "step": 15992
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004538564111566129,
+      "loss": 0.0726,
+      "step": 15993
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045373878878871,
+      "loss": 0.086,
+      "step": 15994
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045362117719169835,
+      "loss": 0.0633,
+      "step": 15995
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000453503576367898,
+      "loss": 0.1498,
+      "step": 15996
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045338598631962677,
+      "loss": 0.1232,
+      "step": 15997
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045326840704920413,
+      "loss": 0.0651,
+      "step": 15998
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000453150838558948,
+      "loss": 0.1149,
+      "step": 15999
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045303328085117655,
+      "loss": 0.142,
+      "step": 16000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045291573392820773,
+      "loss": 0.0964,
+      "step": 16001
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004527981977923592,
+      "loss": 0.0781,
+      "step": 16002
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004526806724459487,
+      "loss": 0.0805,
+      "step": 16003
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004525631578912929,
+      "loss": 0.0862,
+      "step": 16004
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004524456541307097,
+      "loss": 0.0779,
+      "step": 16005
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004523281611665157,
+      "loss": 0.0612,
+      "step": 16006
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004522106790010271,
+      "loss": 0.0827,
+      "step": 16007
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045209320763656114,
+      "loss": 0.0943,
+      "step": 16008
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045197574707543366,
+      "loss": 0.0801,
+      "step": 16009
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045185829731996067,
+      "loss": 0.0579,
+      "step": 16010
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004517408583724577,
+      "loss": 0.0811,
+      "step": 16011
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000451623430235241,
+      "loss": 0.0989,
+      "step": 16012
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045150601291062565,
+      "loss": 0.0465,
+      "step": 16013
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004513886064009267,
+      "loss": 0.1111,
+      "step": 16014
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045127121070845936,
+      "loss": 0.0931,
+      "step": 16015
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004511538258355378,
+      "loss": 0.0719,
+      "step": 16016
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045103645178447727,
+      "loss": 0.069,
+      "step": 16017
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004509190885575915,
+      "loss": 0.0796,
+      "step": 16018
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004508017361571951,
+      "loss": 0.0686,
+      "step": 16019
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045068439458560196,
+      "loss": 0.1031,
+      "step": 16020
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045056706384512534,
+      "loss": 0.0762,
+      "step": 16021
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004504497439380791,
+      "loss": 0.0551,
+      "step": 16022
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004503324348667762,
+      "loss": 0.0868,
+      "step": 16023
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045021513663352953,
+      "loss": 0.0755,
+      "step": 16024
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00045009784924065246,
+      "loss": 0.0627,
+      "step": 16025
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00044998057269045734,
+      "loss": 0.072,
+      "step": 16026
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0004498633069852561,
+      "loss": 0.0661,
+      "step": 16027
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00044974605212736183,
+      "loss": 0.1491,
+      "step": 16028
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000449628808119086,
+      "loss": 0.0978,
+      "step": 16029
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00044951157496274044,
+      "loss": 0.0629,
+      "step": 16030
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004493943526606362,
+      "loss": 0.0703,
+      "step": 16031
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044927714121508546,
+      "loss": 0.0781,
+      "step": 16032
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044915994062839893,
+      "loss": 0.0788,
+      "step": 16033
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004490427509028876,
+      "loss": 0.1191,
+      "step": 16034
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004489255720408619,
+      "loss": 0.0883,
+      "step": 16035
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044880840404463263,
+      "loss": 0.0836,
+      "step": 16036
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004486912469165094,
+      "loss": 0.0898,
+      "step": 16037
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004485741006588032,
+      "loss": 0.0501,
+      "step": 16038
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044845696527382297,
+      "loss": 0.0795,
+      "step": 16039
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004483398407638792,
+      "loss": 0.1081,
+      "step": 16040
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044822272713128077,
+      "loss": 0.1099,
+      "step": 16041
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004481056243783369,
+      "loss": 0.0557,
+      "step": 16042
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004479885325073567,
+      "loss": 0.0618,
+      "step": 16043
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004478714515206485,
+      "loss": 0.0739,
+      "step": 16044
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004477543814205214,
+      "loss": 0.0813,
+      "step": 16045
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044763732220928353,
+      "loss": 0.1041,
+      "step": 16046
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000447520273889243,
+      "loss": 0.1315,
+      "step": 16047
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004474032364627072,
+      "loss": 0.0624,
+      "step": 16048
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044728620993198475,
+      "loss": 0.0917,
+      "step": 16049
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044716919429938265,
+      "loss": 0.1509,
+      "step": 16050
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004470521895672077,
+      "loss": 0.0988,
+      "step": 16051
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004469351957377679,
+      "loss": 0.0633,
+      "step": 16052
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044681821281336944,
+      "loss": 0.1111,
+      "step": 16053
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004467012407963191,
+      "loss": 0.0802,
+      "step": 16054
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044658427968892326,
+      "loss": 0.1075,
+      "step": 16055
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044646732949348803,
+      "loss": 0.0884,
+      "step": 16056
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044635039021231903,
+      "loss": 0.1158,
+      "step": 16057
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044623346184772275,
+      "loss": 0.0681,
+      "step": 16058
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004461165444020043,
+      "loss": 0.061,
+      "step": 16059
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044599963787746876,
+      "loss": 0.0557,
+      "step": 16060
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004458827422764219,
+      "loss": 0.0859,
+      "step": 16061
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004457658576011683,
+      "loss": 0.1182,
+      "step": 16062
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004456489838540124,
+      "loss": 0.1219,
+      "step": 16063
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004455321210372585,
+      "loss": 0.1051,
+      "step": 16064
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004454152691532116,
+      "loss": 0.0818,
+      "step": 16065
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004452984282041752,
+      "loss": 0.0624,
+      "step": 16066
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004451815981924532,
+      "loss": 0.1074,
+      "step": 16067
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044506477912034924,
+      "loss": 0.0629,
+      "step": 16068
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044494797099016615,
+      "loss": 0.101,
+      "step": 16069
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044483117380420814,
+      "loss": 0.1044,
+      "step": 16070
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000444714387564777,
+      "loss": 0.1575,
+      "step": 16071
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004445976122741766,
+      "loss": 0.1209,
+      "step": 16072
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004444808479347088,
+      "loss": 0.0693,
+      "step": 16073
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004443640945486761,
+      "loss": 0.0624,
+      "step": 16074
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0004442473521183804,
+      "loss": 0.0795,
+      "step": 16075
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00044413062064612373,
+      "loss": 0.0688,
+      "step": 16076
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004440139001342073,
+      "loss": 0.0902,
+      "step": 16077
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044389719058493315,
+      "loss": 0.0898,
+      "step": 16078
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004437804920006023,
+      "loss": 0.1147,
+      "step": 16079
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044366380438351586,
+      "loss": 0.0984,
+      "step": 16080
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044354712773597395,
+      "loss": 0.0837,
+      "step": 16081
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044343046206027803,
+      "loss": 0.0845,
+      "step": 16082
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004433138073587281,
+      "loss": 0.1014,
+      "step": 16083
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044319716363362395,
+      "loss": 0.0876,
+      "step": 16084
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044308053088726605,
+      "loss": 0.1239,
+      "step": 16085
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044296390912195404,
+      "loss": 0.066,
+      "step": 16086
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044284729833998717,
+      "loss": 0.1171,
+      "step": 16087
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044273069854366476,
+      "loss": 0.0914,
+      "step": 16088
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044261410973528583,
+      "loss": 0.0967,
+      "step": 16089
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044249753191714903,
+      "loss": 0.0612,
+      "step": 16090
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044238096509155347,
+      "loss": 0.1134,
+      "step": 16091
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000442264409260797,
+      "loss": 0.0757,
+      "step": 16092
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004421478644271785,
+      "loss": 0.0792,
+      "step": 16093
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004420313305929955,
+      "loss": 0.0564,
+      "step": 16094
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004419148077605458,
+      "loss": 0.0789,
+      "step": 16095
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004417982959321271,
+      "loss": 0.0623,
+      "step": 16096
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044168179511003595,
+      "loss": 0.0895,
+      "step": 16097
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044156530529657057,
+      "loss": 0.0665,
+      "step": 16098
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004414488264940273,
+      "loss": 0.0636,
+      "step": 16099
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004413323587047028,
+      "loss": 0.0622,
+      "step": 16100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004412159019308937,
+      "loss": 0.0981,
+      "step": 16101
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004410994561748956,
+      "loss": 0.0872,
+      "step": 16102
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044098302143900535,
+      "loss": 0.1129,
+      "step": 16103
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044086659772551816,
+      "loss": 0.1019,
+      "step": 16104
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044075018503673025,
+      "loss": 0.074,
+      "step": 16105
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044063378337493654,
+      "loss": 0.0779,
+      "step": 16106
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004405173927424322,
+      "loss": 0.0825,
+      "step": 16107
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004404010131415123,
+      "loss": 0.0859,
+      "step": 16108
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044028464457447137,
+      "loss": 0.1057,
+      "step": 16109
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004401682870436039,
+      "loss": 0.0804,
+      "step": 16110
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00044005194055120404,
+      "loss": 0.0946,
+      "step": 16111
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00043993560509956634,
+      "loss": 0.0792,
+      "step": 16112
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000439819280690984,
+      "loss": 0.1078,
+      "step": 16113
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004397029673277514,
+      "loss": 0.0739,
+      "step": 16114
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004395866650121615,
+      "loss": 0.0869,
+      "step": 16115
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00043947037374650756,
+      "loss": 0.0914,
+      "step": 16116
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004393540935330822,
+      "loss": 0.062,
+      "step": 16117
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0004392378243741788,
+      "loss": 0.0719,
+      "step": 16118
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00043912156627208966,
+      "loss": 0.1042,
+      "step": 16119
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00043900531922910703,
+      "loss": 0.0923,
+      "step": 16120
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00043888908324752284,
+      "loss": 0.061,
+      "step": 16121
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000438772858329629,
+      "loss": 0.0704,
+      "step": 16122
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004386566444777176,
+      "loss": 0.1401,
+      "step": 16123
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004385404416940794,
+      "loss": 0.0701,
+      "step": 16124
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004384242499810064,
+      "loss": 0.1132,
+      "step": 16125
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004383080693407891,
+      "loss": 0.084,
+      "step": 16126
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004381918997757186,
+      "loss": 0.0997,
+      "step": 16127
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043807574128808523,
+      "loss": 0.0912,
+      "step": 16128
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004379595938801794,
+      "loss": 0.0758,
+      "step": 16129
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004378434575542911,
+      "loss": 0.0529,
+      "step": 16130
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043772733231271023,
+      "loss": 0.0831,
+      "step": 16131
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004376112181577269,
+      "loss": 0.1276,
+      "step": 16132
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043749511509163033,
+      "loss": 0.0596,
+      "step": 16133
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004373790231167094,
+      "loss": 0.1212,
+      "step": 16134
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043726294223525396,
+      "loss": 0.0812,
+      "step": 16135
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004371468724495524,
+      "loss": 0.1111,
+      "step": 16136
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004370308137618929,
+      "loss": 0.1085,
+      "step": 16137
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004369147661745647,
+      "loss": 0.0632,
+      "step": 16138
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004367987296898555,
+      "loss": 0.0862,
+      "step": 16139
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004366827043100533,
+      "loss": 0.1017,
+      "step": 16140
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043656669003744574,
+      "loss": 0.0771,
+      "step": 16141
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043645068687432045,
+      "loss": 0.0863,
+      "step": 16142
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004363346948229643,
+      "loss": 0.0847,
+      "step": 16143
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043621871388566514,
+      "loss": 0.1503,
+      "step": 16144
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043610274406470894,
+      "loss": 0.0835,
+      "step": 16145
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004359867853623832,
+      "loss": 0.087,
+      "step": 16146
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043587083778097403,
+      "loss": 0.1134,
+      "step": 16147
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004357549013227674,
+      "loss": 0.065,
+      "step": 16148
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043563897599004954,
+      "loss": 0.0569,
+      "step": 16149
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043552306178510605,
+      "loss": 0.0945,
+      "step": 16150
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043540715871022195,
+      "loss": 0.0674,
+      "step": 16151
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004352912667676836,
+      "loss": 0.0927,
+      "step": 16152
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004351753859597757,
+      "loss": 0.0873,
+      "step": 16153
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043505951628878284,
+      "loss": 0.0756,
+      "step": 16154
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004349436577569895,
+      "loss": 0.064,
+      "step": 16155
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000434827810366681,
+      "loss": 0.0826,
+      "step": 16156
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004347119741201405,
+      "loss": 0.1115,
+      "step": 16157
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043459614901965294,
+      "loss": 0.1052,
+      "step": 16158
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043448033506750174,
+      "loss": 0.1021,
+      "step": 16159
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004343645322659704,
+      "loss": 0.0654,
+      "step": 16160
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043424874061734234,
+      "loss": 0.1219,
+      "step": 16161
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043413296012390047,
+      "loss": 0.0867,
+      "step": 16162
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004340171907879279,
+      "loss": 0.1188,
+      "step": 16163
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043390143261170676,
+      "loss": 0.0653,
+      "step": 16164
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004337856855975205,
+      "loss": 0.0819,
+      "step": 16165
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004336699497476504,
+      "loss": 0.0613,
+      "step": 16166
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00043355422506437925,
+      "loss": 0.1006,
+      "step": 16167
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0004334385115499886,
+      "loss": 0.0549,
+      "step": 16168
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004333228092067599,
+      "loss": 0.0654,
+      "step": 16169
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004332071180369742,
+      "loss": 0.0782,
+      "step": 16170
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004330914380429134,
+      "loss": 0.0971,
+      "step": 16171
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000432975769226858,
+      "loss": 0.1096,
+      "step": 16172
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004328601115910887,
+      "loss": 0.1309,
+      "step": 16173
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000432744465137886,
+      "loss": 0.0881,
+      "step": 16174
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043262882986953014,
+      "loss": 0.0737,
+      "step": 16175
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043251320578830077,
+      "loss": 0.0781,
+      "step": 16176
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043239759289647806,
+      "loss": 0.0779,
+      "step": 16177
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043228199119634204,
+      "loss": 0.1642,
+      "step": 16178
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043216640069017163,
+      "loss": 0.0831,
+      "step": 16179
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004320508213802461,
+      "loss": 0.0985,
+      "step": 16180
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004319352532688443,
+      "loss": 0.0674,
+      "step": 16181
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043181969635824505,
+      "loss": 0.1121,
+      "step": 16182
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004317041506507265,
+      "loss": 0.0623,
+      "step": 16183
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000431588616148567,
+      "loss": 0.0921,
+      "step": 16184
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004314730928540451,
+      "loss": 0.0697,
+      "step": 16185
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043135758076943834,
+      "loss": 0.0865,
+      "step": 16186
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004312420798970238,
+      "loss": 0.101,
+      "step": 16187
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004311265902390797,
+      "loss": 0.0902,
+      "step": 16188
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000431011111797883,
+      "loss": 0.0905,
+      "step": 16189
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043089564457570996,
+      "loss": 0.103,
+      "step": 16190
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043078018857483827,
+      "loss": 0.0524,
+      "step": 16191
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043066474379754405,
+      "loss": 0.0859,
+      "step": 16192
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00043054931024610345,
+      "loss": 0.1011,
+      "step": 16193
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004304338879227926,
+      "loss": 0.127,
+      "step": 16194
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004303184768298873,
+      "loss": 0.1526,
+      "step": 16195
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004302030769696633,
+      "loss": 0.1577,
+      "step": 16196
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004300876883443956,
+      "loss": 0.0762,
+      "step": 16197
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004299723109563596,
+      "loss": 0.1115,
+      "step": 16198
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042985694480783066,
+      "loss": 0.1022,
+      "step": 16199
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042974158990108314,
+      "loss": 0.1041,
+      "step": 16200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004296262462383916,
+      "loss": 0.0499,
+      "step": 16201
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042951091382203033,
+      "loss": 0.046,
+      "step": 16202
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004293955926542734,
+      "loss": 0.101,
+      "step": 16203
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004292802827373942,
+      "loss": 0.0886,
+      "step": 16204
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004291649840736671,
+      "loss": 0.0848,
+      "step": 16205
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004290496966653653,
+      "loss": 0.0781,
+      "step": 16206
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042893442051476175,
+      "loss": 0.0846,
+      "step": 16207
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004288191556241291,
+      "loss": 0.0948,
+      "step": 16208
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042870390199574096,
+      "loss": 0.1047,
+      "step": 16209
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042858865963186886,
+      "loss": 0.0892,
+      "step": 16210
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004284734285347861,
+      "loss": 0.0666,
+      "step": 16211
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0004283582087067642,
+      "loss": 0.1012,
+      "step": 16212
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000428243000150075,
+      "loss": 0.072,
+      "step": 16213
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00042812780286699016,
+      "loss": 0.0547,
+      "step": 16214
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004280126168597811,
+      "loss": 0.0615,
+      "step": 16215
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.000427897442130719,
+      "loss": 0.0667,
+      "step": 16216
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004277822786820744,
+      "loss": 0.1055,
+      "step": 16217
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004276671265161188,
+      "loss": 0.0768,
+      "step": 16218
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004275519856351221,
+      "loss": 0.0892,
+      "step": 16219
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042743685604135505,
+      "loss": 0.0731,
+      "step": 16220
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004273217377370875,
+      "loss": 0.0495,
+      "step": 16221
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004272066307245893,
+      "loss": 0.08,
+      "step": 16222
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042709153500613005,
+      "loss": 0.125,
+      "step": 16223
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042697645058397885,
+      "loss": 0.1207,
+      "step": 16224
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042686137746040555,
+      "loss": 0.097,
+      "step": 16225
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042674631563767863,
+      "loss": 0.0764,
+      "step": 16226
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042663126511806695,
+      "loss": 0.0604,
+      "step": 16227
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.000426516225903839,
+      "loss": 0.1605,
+      "step": 16228
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004264011979972626,
+      "loss": 0.1088,
+      "step": 16229
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004262861814006064,
+      "loss": 0.1116,
+      "step": 16230
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004261711761161383,
+      "loss": 0.0807,
+      "step": 16231
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004260561821461257,
+      "loss": 0.1252,
+      "step": 16232
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042594119949283614,
+      "loss": 0.1039,
+      "step": 16233
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004258262281585366,
+      "loss": 0.0499,
+      "step": 16234
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.000425711268145494,
+      "loss": 0.1111,
+      "step": 16235
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004255963194559751,
+      "loss": 0.1134,
+      "step": 16236
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004254813820922462,
+      "loss": 0.0983,
+      "step": 16237
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004253664560565741,
+      "loss": 0.0988,
+      "step": 16238
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042525154135122445,
+      "loss": 0.104,
+      "step": 16239
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004251366379784629,
+      "loss": 0.0953,
+      "step": 16240
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004250217459405558,
+      "loss": 0.1044,
+      "step": 16241
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.000424906865239768,
+      "loss": 0.0861,
+      "step": 16242
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004247919958783646,
+      "loss": 0.1122,
+      "step": 16243
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004246771378586104,
+      "loss": 0.0918,
+      "step": 16244
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004245622911827707,
+      "loss": 0.1427,
+      "step": 16245
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004244474558531096,
+      "loss": 0.0613,
+      "step": 16246
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042433263187189154,
+      "loss": 0.0807,
+      "step": 16247
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042421781924138036,
+      "loss": 0.087,
+      "step": 16248
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004241030179638399,
+      "loss": 0.0878,
+      "step": 16249
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004239882280415335,
+      "loss": 0.0771,
+      "step": 16250
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004238734494767248,
+      "loss": 0.0586,
+      "step": 16251
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042375868227167735,
+      "loss": 0.1165,
+      "step": 16252
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004236439264286538,
+      "loss": 0.0811,
+      "step": 16253
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042352918194991663,
+      "loss": 0.0745,
+      "step": 16254
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00042341444883772837,
+      "loss": 0.1254,
+      "step": 16255
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004232997270943514,
+      "loss": 0.0853,
+      "step": 16256
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004231850167220472,
+      "loss": 0.1581,
+      "step": 16257
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004230703177230785,
+      "loss": 0.0894,
+      "step": 16258
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004229556300997063,
+      "loss": 0.0743,
+      "step": 16259
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0004228409538541921,
+      "loss": 0.1059,
+      "step": 16260
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004227262889887966,
+      "loss": 0.1,
+      "step": 16261
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004226116355057815,
+      "loss": 0.0828,
+      "step": 16262
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042249699340740697,
+      "loss": 0.1409,
+      "step": 16263
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042238236269593335,
+      "loss": 0.0695,
+      "step": 16264
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042226774337362127,
+      "loss": 0.0923,
+      "step": 16265
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004221531354427306,
+      "loss": 0.0844,
+      "step": 16266
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000422038538905521,
+      "loss": 0.1232,
+      "step": 16267
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004219239537642521,
+      "loss": 0.0873,
+      "step": 16268
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004218093800211832,
+      "loss": 0.0753,
+      "step": 16269
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004216948176785731,
+      "loss": 0.0944,
+      "step": 16270
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042158026673868123,
+      "loss": 0.1115,
+      "step": 16271
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042146572720376573,
+      "loss": 0.0569,
+      "step": 16272
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004213511990760857,
+      "loss": 0.0771,
+      "step": 16273
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004212366823578988,
+      "loss": 0.0836,
+      "step": 16274
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004211221770514633,
+      "loss": 0.0786,
+      "step": 16275
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004210076831590368,
+      "loss": 0.0767,
+      "step": 16276
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004208932006828765,
+      "loss": 0.0905,
+      "step": 16277
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042077872962524034,
+      "loss": 0.0708,
+      "step": 16278
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004206642699883851,
+      "loss": 0.0635,
+      "step": 16279
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004205498217745678,
+      "loss": 0.0588,
+      "step": 16280
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004204353849860448,
+      "loss": 0.0789,
+      "step": 16281
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042032095962507224,
+      "loss": 0.1068,
+      "step": 16282
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00042020654569390717,
+      "loss": 0.0988,
+      "step": 16283
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004200921431948045,
+      "loss": 0.101,
+      "step": 16284
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000419977752130021,
+      "loss": 0.1053,
+      "step": 16285
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004198633725018117,
+      "loss": 0.0711,
+      "step": 16286
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004197490043124318,
+      "loss": 0.0828,
+      "step": 16287
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004196346475641365,
+      "loss": 0.085,
+      "step": 16288
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041952030225918046,
+      "loss": 0.084,
+      "step": 16289
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041940596839981805,
+      "loss": 0.1317,
+      "step": 16290
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041929164598830427,
+      "loss": 0.1171,
+      "step": 16291
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000419177335026893,
+      "loss": 0.0784,
+      "step": 16292
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041906303551783776,
+      "loss": 0.0698,
+      "step": 16293
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000418948747463393,
+      "loss": 0.1182,
+      "step": 16294
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004188344708658117,
+      "loss": 0.0834,
+      "step": 16295
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041872020572734724,
+      "loss": 0.1166,
+      "step": 16296
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041860595205025234,
+      "loss": 0.12,
+      "step": 16297
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004184917098367803,
+      "loss": 0.1259,
+      "step": 16298
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041837747908918347,
+      "loss": 0.0964,
+      "step": 16299
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041826325980971426,
+      "loss": 0.0944,
+      "step": 16300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004181490520006246,
+      "loss": 0.0995,
+      "step": 16301
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041803485566416644,
+      "loss": 0.0873,
+      "step": 16302
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004179206708025912,
+      "loss": 0.083,
+      "step": 16303
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000417806497418151,
+      "loss": 0.0948,
+      "step": 16304
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00041769233551309615,
+      "loss": 0.077,
+      "step": 16305
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0004175781850896786,
+      "loss": 0.1095,
+      "step": 16306
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004174640461501487,
+      "loss": 0.0663,
+      "step": 16307
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041734991869675676,
+      "loss": 0.0497,
+      "step": 16308
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004172358027317533,
+      "loss": 0.0903,
+      "step": 16309
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.000417121698257388,
+      "loss": 0.088,
+      "step": 16310
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004170076052759114,
+      "loss": 0.0916,
+      "step": 16311
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004168935237895729,
+      "loss": 0.0604,
+      "step": 16312
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041677945380062163,
+      "loss": 0.0771,
+      "step": 16313
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004166653953113067,
+      "loss": 0.1239,
+      "step": 16314
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004165513483238774,
+      "loss": 0.0783,
+      "step": 16315
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004164373128405825,
+      "loss": 0.092,
+      "step": 16316
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004163232888636699,
+      "loss": 0.1082,
+      "step": 16317
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041620927639538863,
+      "loss": 0.0878,
+      "step": 16318
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041609527543798644,
+      "loss": 0.076,
+      "step": 16319
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.000415981285993711,
+      "loss": 0.0519,
+      "step": 16320
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041586730806481,
+      "loss": 0.1464,
+      "step": 16321
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041575334165353076,
+      "loss": 0.0839,
+      "step": 16322
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004156393867621203,
+      "loss": 0.0867,
+      "step": 16323
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004155254433928255,
+      "loss": 0.0757,
+      "step": 16324
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041541151154789305,
+      "loss": 0.1366,
+      "step": 16325
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041529759122956977,
+      "loss": 0.1296,
+      "step": 16326
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041518368244010184,
+      "loss": 0.0778,
+      "step": 16327
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004150697851817349,
+      "loss": 0.0571,
+      "step": 16328
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041495589945671495,
+      "loss": 0.0854,
+      "step": 16329
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041484202526728697,
+      "loss": 0.077,
+      "step": 16330
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004147281626156972,
+      "loss": 0.0699,
+      "step": 16331
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004146143115041903,
+      "loss": 0.0778,
+      "step": 16332
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004145004719350111,
+      "loss": 0.1598,
+      "step": 16333
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041438664391040404,
+      "loss": 0.1422,
+      "step": 16334
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004142728274326135,
+      "loss": 0.1018,
+      "step": 16335
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041415902250388416,
+      "loss": 0.0796,
+      "step": 16336
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004140452291264594,
+      "loss": 0.1104,
+      "step": 16337
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041393144730258346,
+      "loss": 0.13,
+      "step": 16338
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041381767703449956,
+      "loss": 0.272,
+      "step": 16339
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041370391832445095,
+      "loss": 0.048,
+      "step": 16340
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004135901711746808,
+      "loss": 0.0789,
+      "step": 16341
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041347643558743176,
+      "loss": 0.0704,
+      "step": 16342
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004133627115649464,
+      "loss": 0.1045,
+      "step": 16343
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004132489991094668,
+      "loss": 0.0826,
+      "step": 16344
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004131352982232358,
+      "loss": 0.1252,
+      "step": 16345
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004130216089084945,
+      "loss": 0.1182,
+      "step": 16346
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00041290793116748536,
+      "loss": 0.126,
+      "step": 16347
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004127942650024494,
+      "loss": 0.0587,
+      "step": 16348
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004126806104156279,
+      "loss": 0.0981,
+      "step": 16349
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004125669674092614,
+      "loss": 0.0864,
+      "step": 16350
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004124533359855913,
+      "loss": 0.0714,
+      "step": 16351
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004123397161468581,
+      "loss": 0.0471,
+      "step": 16352
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0004122261078953017,
+      "loss": 0.0945,
+      "step": 16353
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004121125112331624,
+      "loss": 0.1261,
+      "step": 16354
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041199892616268,
+      "loss": 0.0441,
+      "step": 16355
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004118853526860936,
+      "loss": 0.0765,
+      "step": 16356
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041177179080564353,
+      "loss": 0.0888,
+      "step": 16357
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041165824052356815,
+      "loss": 0.0795,
+      "step": 16358
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.000411544701842107,
+      "loss": 0.1448,
+      "step": 16359
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041143117476349857,
+      "loss": 0.1055,
+      "step": 16360
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041131765928998124,
+      "loss": 0.0929,
+      "step": 16361
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041120415542379317,
+      "loss": 0.1315,
+      "step": 16362
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041109066316717257,
+      "loss": 0.0911,
+      "step": 16363
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041097718252235675,
+      "loss": 0.0861,
+      "step": 16364
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041086371349158393,
+      "loss": 0.0744,
+      "step": 16365
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041075025607709115,
+      "loss": 0.0839,
+      "step": 16366
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041063681028111513,
+      "loss": 0.0726,
+      "step": 16367
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041052337610589343,
+      "loss": 0.0757,
+      "step": 16368
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004104099535536623,
+      "loss": 0.0852,
+      "step": 16369
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041029654262665784,
+      "loss": 0.1125,
+      "step": 16370
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041018314332711694,
+      "loss": 0.0778,
+      "step": 16371
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00041006975565727524,
+      "loss": 0.0756,
+      "step": 16372
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004099563796193683,
+      "loss": 0.0757,
+      "step": 16373
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004098430152156317,
+      "loss": 0.1071,
+      "step": 16374
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004097296624483008,
+      "loss": 0.0779,
+      "step": 16375
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004096163213196106,
+      "loss": 0.0811,
+      "step": 16376
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004095029918317953,
+      "loss": 0.0488,
+      "step": 16377
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004093896739870905,
+      "loss": 0.0701,
+      "step": 16378
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004092763677877297,
+      "loss": 0.1464,
+      "step": 16379
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040916307323594757,
+      "loss": 0.0554,
+      "step": 16380
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040904979033397793,
+      "loss": 0.1045,
+      "step": 16381
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040893651908405427,
+      "loss": 0.1105,
+      "step": 16382
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040882325948841003,
+      "loss": 0.0864,
+      "step": 16383
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.000408710011549278,
+      "loss": 0.058,
+      "step": 16384
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040859677526889195,
+      "loss": 0.0607,
+      "step": 16385
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004084835506494842,
+      "loss": 0.114,
+      "step": 16386
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040837033769328726,
+      "loss": 0.0757,
+      "step": 16387
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004082571364025331,
+      "loss": 0.1069,
+      "step": 16388
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040814394677945445,
+      "loss": 0.0673,
+      "step": 16389
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004080307688262823,
+      "loss": 0.1001,
+      "step": 16390
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040791760254524915,
+      "loss": 0.0911,
+      "step": 16391
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040780444793858575,
+      "loss": 0.074,
+      "step": 16392
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004076913050085234,
+      "loss": 0.0861,
+      "step": 16393
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.000407578173757293,
+      "loss": 0.1099,
+      "step": 16394
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004074650541871251,
+      "loss": 0.1307,
+      "step": 16395
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004073519463002502,
+      "loss": 0.0962,
+      "step": 16396
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040723885009889816,
+      "loss": 0.0694,
+      "step": 16397
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0004071257655852997,
+      "loss": 0.099,
+      "step": 16398
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00040701269276168406,
+      "loss": 0.1398,
+      "step": 16399
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040689963163028053,
+      "loss": 0.0869,
+      "step": 16400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004067865821933191,
+      "loss": 0.1114,
+      "step": 16401
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004066735444530284,
+      "loss": 0.0811,
+      "step": 16402
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004065605184116372,
+      "loss": 0.0935,
+      "step": 16403
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004064475040713739,
+      "loss": 0.0785,
+      "step": 16404
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040633450143446736,
+      "loss": 0.0847,
+      "step": 16405
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004062215105031456,
+      "loss": 0.0865,
+      "step": 16406
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004061085312796363,
+      "loss": 0.0834,
+      "step": 16407
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004059955637661672,
+      "loss": 0.0546,
+      "step": 16408
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040588260796496536,
+      "loss": 0.0937,
+      "step": 16409
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040576966387825873,
+      "loss": 0.1843,
+      "step": 16410
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040565673150827356,
+      "loss": 0.0846,
+      "step": 16411
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004055438108572372,
+      "loss": 0.0804,
+      "step": 16412
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040543090192737607,
+      "loss": 0.0972,
+      "step": 16413
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004053180047209163,
+      "loss": 0.0707,
+      "step": 16414
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040520511924008374,
+      "loss": 0.0865,
+      "step": 16415
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040509224548710457,
+      "loss": 0.0571,
+      "step": 16416
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004049793834642037,
+      "loss": 0.0798,
+      "step": 16417
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040486653317360735,
+      "loss": 0.094,
+      "step": 16418
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004047536946175403,
+      "loss": 0.1186,
+      "step": 16419
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040464086779822705,
+      "loss": 0.0562,
+      "step": 16420
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040452805271789296,
+      "loss": 0.0876,
+      "step": 16421
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004044152493787622,
+      "loss": 0.0798,
+      "step": 16422
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040430245778305874,
+      "loss": 0.0622,
+      "step": 16423
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040418967793300645,
+      "loss": 0.0828,
+      "step": 16424
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040407690983082967,
+      "loss": 0.0861,
+      "step": 16425
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004039641534787515,
+      "loss": 0.0449,
+      "step": 16426
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040385140887899516,
+      "loss": 0.0832,
+      "step": 16427
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004037386760337839,
+      "loss": 0.0539,
+      "step": 16428
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040362595494534037,
+      "loss": 0.0671,
+      "step": 16429
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004035132456158869,
+      "loss": 0.0609,
+      "step": 16430
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004034005480476464,
+      "loss": 0.1002,
+      "step": 16431
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040328786224284017,
+      "loss": 0.0547,
+      "step": 16432
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040317518820369116,
+      "loss": 0.1013,
+      "step": 16433
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040306252593242043,
+      "loss": 0.0637,
+      "step": 16434
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004029498754312495,
+      "loss": 0.0782,
+      "step": 16435
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040283723670239935,
+      "loss": 0.1268,
+      "step": 16436
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040272460974809086,
+      "loss": 0.1195,
+      "step": 16437
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004026119945705452,
+      "loss": 0.0846,
+      "step": 16438
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040249939117198273,
+      "loss": 0.074,
+      "step": 16439
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040238679955462354,
+      "loss": 0.0748,
+      "step": 16440
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00040227421972068733,
+      "loss": 0.098,
+      "step": 16441
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004021616516723946,
+      "loss": 0.2683,
+      "step": 16442
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004020490954119642,
+      "loss": 0.0984,
+      "step": 16443
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004019365509416163,
+      "loss": 0.0601,
+      "step": 16444
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0004018240182635694,
+      "loss": 0.0824,
+      "step": 16445
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00040171149738004255,
+      "loss": 0.0732,
+      "step": 16446
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004015989882932544,
+      "loss": 0.0624,
+      "step": 16447
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004014864910054231,
+      "loss": 0.0792,
+      "step": 16448
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000401374005518767,
+      "loss": 0.0826,
+      "step": 16449
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004012615318355036,
+      "loss": 0.0669,
+      "step": 16450
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00040114906995785137,
+      "loss": 0.0936,
+      "step": 16451
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004010366198880274,
+      "loss": 0.0933,
+      "step": 16452
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004009241816282486,
+      "loss": 0.0753,
+      "step": 16453
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00040081175518073267,
+      "loss": 0.0875,
+      "step": 16454
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00040069934054769597,
+      "loss": 0.0745,
+      "step": 16455
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000400586937731355,
+      "loss": 0.0908,
+      "step": 16456
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004004745467339258,
+      "loss": 0.0868,
+      "step": 16457
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00040036216755762514,
+      "loss": 0.0842,
+      "step": 16458
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004002498002046684,
+      "loss": 0.1346,
+      "step": 16459
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0004001374446772713,
+      "loss": 0.077,
+      "step": 16460
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00040002510097764913,
+      "loss": 0.0692,
+      "step": 16461
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003999127691080167,
+      "loss": 0.0784,
+      "step": 16462
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039980044907058954,
+      "loss": 0.0729,
+      "step": 16463
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039968814086758175,
+      "loss": 0.1289,
+      "step": 16464
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039957584450120843,
+      "loss": 0.0471,
+      "step": 16465
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039946355997368333,
+      "loss": 0.0714,
+      "step": 16466
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003993512872872206,
+      "loss": 0.1205,
+      "step": 16467
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003992390264440338,
+      "loss": 0.0822,
+      "step": 16468
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003991267774463364,
+      "loss": 0.0847,
+      "step": 16469
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003990145402963415,
+      "loss": 0.0545,
+      "step": 16470
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003989023149962627,
+      "loss": 0.0608,
+      "step": 16471
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003987901015483126,
+      "loss": 0.0651,
+      "step": 16472
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003986778999547036,
+      "loss": 0.1278,
+      "step": 16473
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039856571021764764,
+      "loss": 0.0786,
+      "step": 16474
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039845353233935776,
+      "loss": 0.086,
+      "step": 16475
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003983413663220452,
+      "loss": 0.0914,
+      "step": 16476
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003982292121679215,
+      "loss": 0.0908,
+      "step": 16477
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003981170698791986,
+      "loss": 0.0628,
+      "step": 16478
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039800493945808726,
+      "loss": 0.0869,
+      "step": 16479
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039789282090679854,
+      "loss": 0.0915,
+      "step": 16480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039778071422754313,
+      "loss": 0.0832,
+      "step": 16481
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003976686194225314,
+      "loss": 0.0608,
+      "step": 16482
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003975565364939733,
+      "loss": 0.0776,
+      "step": 16483
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039744446544407954,
+      "loss": 0.121,
+      "step": 16484
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.000397332406275059,
+      "loss": 0.0533,
+      "step": 16485
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003972203589891221,
+      "loss": 0.0571,
+      "step": 16486
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003971083235884777,
+      "loss": 0.1194,
+      "step": 16487
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039699630007533496,
+      "loss": 0.086,
+      "step": 16488
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003968842884519025,
+      "loss": 0.106,
+      "step": 16489
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039677228872038873,
+      "loss": 0.0601,
+      "step": 16490
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00039666030088300266,
+      "loss": 0.1631,
+      "step": 16491
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000396548324941952,
+      "loss": 0.0847,
+      "step": 16492
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039643636089944467,
+      "loss": 0.093,
+      "step": 16493
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003963244087576884,
+      "loss": 0.0969,
+      "step": 16494
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039621246851889014,
+      "loss": 0.0951,
+      "step": 16495
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003961005401852578,
+      "loss": 0.0707,
+      "step": 16496
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003959886237589976,
+      "loss": 0.1214,
+      "step": 16497
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039587671924231694,
+      "loss": 0.0867,
+      "step": 16498
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000395764826637422,
+      "loss": 0.0981,
+      "step": 16499
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000395652945946519,
+      "loss": 0.1063,
+      "step": 16500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000395541077171814,
+      "loss": 0.0824,
+      "step": 16501
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039542922031551267,
+      "loss": 0.0873,
+      "step": 16502
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003953173753798203,
+      "loss": 0.0789,
+      "step": 16503
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039520554236694273,
+      "loss": 0.0806,
+      "step": 16504
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003950937212790848,
+      "loss": 0.0464,
+      "step": 16505
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003949819121184509,
+      "loss": 0.1024,
+      "step": 16506
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039487011488724655,
+      "loss": 0.1067,
+      "step": 16507
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039475832958767556,
+      "loss": 0.0975,
+      "step": 16508
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000394646556221942,
+      "loss": 0.0686,
+      "step": 16509
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003945347947922495,
+      "loss": 0.0919,
+      "step": 16510
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003944230453008025,
+      "loss": 0.1058,
+      "step": 16511
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039431130774980394,
+      "loss": 0.1062,
+      "step": 16512
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000394199582141457,
+      "loss": 0.1266,
+      "step": 16513
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003940878684779647,
+      "loss": 0.0894,
+      "step": 16514
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003939761667615294,
+      "loss": 0.074,
+      "step": 16515
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039386447699435423,
+      "loss": 0.0794,
+      "step": 16516
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003937527991786407,
+      "loss": 0.1289,
+      "step": 16517
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003936411333165917,
+      "loss": 0.0762,
+      "step": 16518
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003935294794104084,
+      "loss": 0.0828,
+      "step": 16519
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003934178374622924,
+      "loss": 0.134,
+      "step": 16520
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003933062074744451,
+      "loss": 0.0745,
+      "step": 16521
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003931945894490675,
+      "loss": 0.0807,
+      "step": 16522
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003930829833883601,
+      "loss": 0.104,
+      "step": 16523
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003929713892945241,
+      "loss": 0.0687,
+      "step": 16524
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039285980716975965,
+      "loss": 0.071,
+      "step": 16525
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003927482370162668,
+      "loss": 0.0906,
+      "step": 16526
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000392636678836245,
+      "loss": 0.099,
+      "step": 16527
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003925251326318947,
+      "loss": 0.0778,
+      "step": 16528
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003924135984054149,
+      "loss": 0.1128,
+      "step": 16529
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003923020761590045,
+      "loss": 0.0798,
+      "step": 16530
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039219056589486303,
+      "loss": 0.0758,
+      "step": 16531
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.000392079067615189,
+      "loss": 0.0728,
+      "step": 16532
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00039196758132218056,
+      "loss": 0.1012,
+      "step": 16533
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003918561070180363,
+      "loss": 0.0742,
+      "step": 16534
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003917446447049541,
+      "loss": 0.0594,
+      "step": 16535
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003916331943851312,
+      "loss": 0.073,
+      "step": 16536
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003915217560607661,
+      "loss": 0.0802,
+      "step": 16537
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003914103297340551,
+      "loss": 0.0784,
+      "step": 16538
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003912989154071961,
+      "loss": 0.0728,
+      "step": 16539
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00039118751308238553,
+      "loss": 0.0728,
+      "step": 16540
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00039107612276182004,
+      "loss": 0.0779,
+      "step": 16541
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003909647444476958,
+      "loss": 0.0894,
+      "step": 16542
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00039085337814220877,
+      "loss": 0.0641,
+      "step": 16543
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003907420238475553,
+      "loss": 0.0841,
+      "step": 16544
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00039063068156593075,
+      "loss": 0.1071,
+      "step": 16545
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003905193512995305,
+      "loss": 0.076,
+      "step": 16546
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00039040803305054984,
+      "loss": 0.0927,
+      "step": 16547
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003902967268211831,
+      "loss": 0.0749,
+      "step": 16548
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00039018543261362584,
+      "loss": 0.074,
+      "step": 16549
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003900741504300717,
+      "loss": 0.0621,
+      "step": 16550
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038996288027271555,
+      "loss": 0.0713,
+      "step": 16551
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038985162214375115,
+      "loss": 0.0522,
+      "step": 16552
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038974037604537216,
+      "loss": 0.1163,
+      "step": 16553
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003896291419797721,
+      "loss": 0.1125,
+      "step": 16554
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003895179199491441,
+      "loss": 0.0578,
+      "step": 16555
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003894067099556813,
+      "loss": 0.0822,
+      "step": 16556
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038929551200157596,
+      "loss": 0.1113,
+      "step": 16557
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038918432608902166,
+      "loss": 0.0723,
+      "step": 16558
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003890731522202098,
+      "loss": 0.0532,
+      "step": 16559
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038896199039733305,
+      "loss": 0.1007,
+      "step": 16560
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038885084062258304,
+      "loss": 0.1107,
+      "step": 16561
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003887397028981513,
+      "loss": 0.064,
+      "step": 16562
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003886285772262289,
+      "loss": 0.0814,
+      "step": 16563
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038851746360900753,
+      "loss": 0.0884,
+      "step": 16564
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038840636204867787,
+      "loss": 0.0742,
+      "step": 16565
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038829527254743046,
+      "loss": 0.0728,
+      "step": 16566
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003881841951074557,
+      "loss": 0.0979,
+      "step": 16567
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003880731297309439,
+      "loss": 0.0588,
+      "step": 16568
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003879620764200844,
+      "loss": 0.0919,
+      "step": 16569
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038785103517706746,
+      "loss": 0.0778,
+      "step": 16570
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003877400060040828,
+      "loss": 0.0666,
+      "step": 16571
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003876289889033192,
+      "loss": 0.0668,
+      "step": 16572
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003875179838769658,
+      "loss": 0.0889,
+      "step": 16573
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003874069909272112,
+      "loss": 0.0561,
+      "step": 16574
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038729601005624406,
+      "loss": 0.1085,
+      "step": 16575
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003871850412662524,
+      "loss": 0.0857,
+      "step": 16576
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003870740845594242,
+      "loss": 0.1193,
+      "step": 16577
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003869631399379476,
+      "loss": 0.0695,
+      "step": 16578
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00038685220740401007,
+      "loss": 0.0981,
+      "step": 16579
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003867412869597985,
+      "loss": 0.079,
+      "step": 16580
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003866303786075006,
+      "loss": 0.146,
+      "step": 16581
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003865194823493029,
+      "loss": 0.0819,
+      "step": 16582
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0003864085981873916,
+      "loss": 0.0952,
+      "step": 16583
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038629772612395384,
+      "loss": 0.1435,
+      "step": 16584
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003861868661611753,
+      "loss": 0.1209,
+      "step": 16585
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038607601830124193,
+      "loss": 0.0712,
+      "step": 16586
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038596518254633926,
+      "loss": 0.0679,
+      "step": 16587
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038585435889865286,
+      "loss": 0.0729,
+      "step": 16588
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003857435473603678,
+      "loss": 0.109,
+      "step": 16589
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003856327479336686,
+      "loss": 0.0952,
+      "step": 16590
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003855219606207404,
+      "loss": 0.0849,
+      "step": 16591
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038541118542376787,
+      "loss": 0.1807,
+      "step": 16592
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000385300422344935,
+      "loss": 0.0648,
+      "step": 16593
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003851896713864257,
+      "loss": 0.093,
+      "step": 16594
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003850789325504237,
+      "loss": 0.0646,
+      "step": 16595
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038496820583911254,
+      "loss": 0.0909,
+      "step": 16596
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038485749125467497,
+      "loss": 0.0641,
+      "step": 16597
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003847467887992948,
+      "loss": 0.0938,
+      "step": 16598
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038463609847515446,
+      "loss": 0.0685,
+      "step": 16599
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003845254202844366,
+      "loss": 0.0847,
+      "step": 16600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003844147542293229,
+      "loss": 0.0669,
+      "step": 16601
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038430410031199637,
+      "loss": 0.095,
+      "step": 16602
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038419345853463795,
+      "loss": 0.0948,
+      "step": 16603
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038408282889943005,
+      "loss": 0.0719,
+      "step": 16604
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038397221140855353,
+      "loss": 0.081,
+      "step": 16605
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003838616060641895,
+      "loss": 0.0983,
+      "step": 16606
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038375101286851885,
+      "loss": 0.092,
+      "step": 16607
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003836404318237223,
+      "loss": 0.1241,
+      "step": 16608
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038352986293198023,
+      "loss": 0.0745,
+      "step": 16609
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038341930619547216,
+      "loss": 0.1101,
+      "step": 16610
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038330876161637897,
+      "loss": 0.108,
+      "step": 16611
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038319822919687953,
+      "loss": 0.1045,
+      "step": 16612
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003830877089391539,
+      "loss": 0.1288,
+      "step": 16613
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000382977200845381,
+      "loss": 0.0599,
+      "step": 16614
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003828667049177397,
+      "loss": 0.0881,
+      "step": 16615
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003827562211584087,
+      "loss": 0.0994,
+      "step": 16616
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003826457495695662,
+      "loss": 0.0698,
+      "step": 16617
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038253529015339104,
+      "loss": 0.1176,
+      "step": 16618
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003824248429120609,
+      "loss": 0.1156,
+      "step": 16619
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038231440784775353,
+      "loss": 0.0817,
+      "step": 16620
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003822039849626463,
+      "loss": 0.0892,
+      "step": 16621
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003820935742589163,
+      "loss": 0.1144,
+      "step": 16622
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038198317573874074,
+      "loss": 0.0845,
+      "step": 16623
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038187278940429693,
+      "loss": 0.1089,
+      "step": 16624
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003817624152577609,
+      "loss": 0.1075,
+      "step": 16625
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000381652053301309,
+      "loss": 0.1134,
+      "step": 16626
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038154170353711723,
+      "loss": 0.0698,
+      "step": 16627
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003814313659673616,
+      "loss": 0.084,
+      "step": 16628
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00038132104059421755,
+      "loss": 0.1144,
+      "step": 16629
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00038121072741986007,
+      "loss": 0.0683,
+      "step": 16630
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00038110042644646494,
+      "loss": 0.1767,
+      "step": 16631
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003809901376762066,
+      "loss": 0.1078,
+      "step": 16632
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003808798611112596,
+      "loss": 0.0915,
+      "step": 16633
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00038076959675379876,
+      "loss": 0.1001,
+      "step": 16634
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003806593446059978,
+      "loss": 0.0999,
+      "step": 16635
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003805491046700309,
+      "loss": 0.1133,
+      "step": 16636
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003804388769480711,
+      "loss": 0.0623,
+      "step": 16637
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00038032866144229274,
+      "loss": 0.0847,
+      "step": 16638
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00038021845815486847,
+      "loss": 0.0948,
+      "step": 16639
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003801082670879712,
+      "loss": 0.1055,
+      "step": 16640
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003799980882437738,
+      "loss": 0.0642,
+      "step": 16641
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003798879216244485,
+      "loss": 0.0582,
+      "step": 16642
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037977776723216737,
+      "loss": 0.0842,
+      "step": 16643
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003796676250691026,
+      "loss": 0.0872,
+      "step": 16644
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003795574951374262,
+      "loss": 0.0837,
+      "step": 16645
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037944737743930956,
+      "loss": 0.0969,
+      "step": 16646
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037933727197692356,
+      "loss": 0.1271,
+      "step": 16647
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003792271787524396,
+      "loss": 0.1388,
+      "step": 16648
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037911709776802815,
+      "loss": 0.0716,
+      "step": 16649
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037900702902585946,
+      "loss": 0.093,
+      "step": 16650
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037889697252810453,
+      "loss": 0.0891,
+      "step": 16651
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037878692827693296,
+      "loss": 0.1003,
+      "step": 16652
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003786768962745147,
+      "loss": 0.0943,
+      "step": 16653
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003785668765230188,
+      "loss": 0.166,
+      "step": 16654
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003784568690246154,
+      "loss": 0.1062,
+      "step": 16655
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003783468737814731,
+      "loss": 0.1157,
+      "step": 16656
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003782368907957606,
+      "loss": 0.0676,
+      "step": 16657
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003781269200696469,
+      "loss": 0.0807,
+      "step": 16658
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037801696160530017,
+      "loss": 0.1123,
+      "step": 16659
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003779070154048884,
+      "loss": 0.0658,
+      "step": 16660
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037779708147057955,
+      "loss": 0.0839,
+      "step": 16661
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037768715980454127,
+      "loss": 0.1166,
+      "step": 16662
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003775772504089404,
+      "loss": 0.0905,
+      "step": 16663
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037746735328594497,
+      "loss": 0.0552,
+      "step": 16664
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003773574684377211,
+      "loss": 0.0891,
+      "step": 16665
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003772475958664362,
+      "loss": 0.1096,
+      "step": 16666
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037713773557425623,
+      "loss": 0.1068,
+      "step": 16667
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037702788756334736,
+      "loss": 0.0654,
+      "step": 16668
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037691805183587556,
+      "loss": 0.1123,
+      "step": 16669
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003768082283940062,
+      "loss": 0.0795,
+      "step": 16670
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003766984172399054,
+      "loss": 0.1022,
+      "step": 16671
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000376588618375738,
+      "loss": 0.0888,
+      "step": 16672
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00037647883180366894,
+      "loss": 0.0738,
+      "step": 16673
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0003763690575258628,
+      "loss": 0.0698,
+      "step": 16674
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000376259295544484,
+      "loss": 0.09,
+      "step": 16675
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037614954586169725,
+      "loss": 0.1394,
+      "step": 16676
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000376039808479666,
+      "loss": 0.0951,
+      "step": 16677
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037593008340055443,
+      "loss": 0.0509,
+      "step": 16678
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037582037062652596,
+      "loss": 0.0754,
+      "step": 16679
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003757106701597438,
+      "loss": 0.0573,
+      "step": 16680
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003756009820023708,
+      "loss": 0.1082,
+      "step": 16681
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003754913061565699,
+      "loss": 0.0807,
+      "step": 16682
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037538164262450313,
+      "loss": 0.1608,
+      "step": 16683
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003752719914083337,
+      "loss": 0.0584,
+      "step": 16684
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003751623525102231,
+      "loss": 0.0809,
+      "step": 16685
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037505272593233287,
+      "loss": 0.1107,
+      "step": 16686
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037494311167682526,
+      "loss": 0.1497,
+      "step": 16687
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003748335097458613,
+      "loss": 0.1805,
+      "step": 16688
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037472392014160184,
+      "loss": 0.1199,
+      "step": 16689
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037461434286620765,
+      "loss": 0.1368,
+      "step": 16690
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003745047779218399,
+      "loss": 0.0839,
+      "step": 16691
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003743952253106585,
+      "loss": 0.1099,
+      "step": 16692
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037428568503482365,
+      "loss": 0.0645,
+      "step": 16693
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003741761570964951,
+      "loss": 0.1248,
+      "step": 16694
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037406664149783266,
+      "loss": 0.1026,
+      "step": 16695
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003739571382409952,
+      "loss": 0.1149,
+      "step": 16696
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003738476473281426,
+      "loss": 0.1074,
+      "step": 16697
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000373738168761433,
+      "loss": 0.0865,
+      "step": 16698
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037362870254302574,
+      "loss": 0.071,
+      "step": 16699
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003735192486750789,
+      "loss": 0.087,
+      "step": 16700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003734098071597506,
+      "loss": 0.0798,
+      "step": 16701
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003733003779991988,
+      "loss": 0.0948,
+      "step": 16702
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037319096119558083,
+      "loss": 0.1696,
+      "step": 16703
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003730815567510548,
+      "loss": 0.089,
+      "step": 16704
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003729721646677775,
+      "loss": 0.0745,
+      "step": 16705
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037286278494790584,
+      "loss": 0.0631,
+      "step": 16706
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003727534175935963,
+      "loss": 0.1388,
+      "step": 16707
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003726440626070058,
+      "loss": 0.1151,
+      "step": 16708
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003725347199902904,
+      "loss": 0.0809,
+      "step": 16709
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003724253897456056,
+      "loss": 0.0813,
+      "step": 16710
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037231607187510784,
+      "loss": 0.0568,
+      "step": 16711
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003722067663809524,
+      "loss": 0.0723,
+      "step": 16712
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003720974732652942,
+      "loss": 0.1008,
+      "step": 16713
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003719881925302885,
+      "loss": 0.0752,
+      "step": 16714
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037187892417808987,
+      "loss": 0.0864,
+      "step": 16715
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003717696682108526,
+      "loss": 0.0702,
+      "step": 16716
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037166042463073135,
+      "loss": 0.0769,
+      "step": 16717
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037155119343987986,
+      "loss": 0.0874,
+      "step": 16718
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037144197464045224,
+      "loss": 0.0591,
+      "step": 16719
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00037133276823460184,
+      "loss": 0.0839,
+      "step": 16720
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003712235742244818,
+      "loss": 0.0752,
+      "step": 16721
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003711143926122453,
+      "loss": 0.1176,
+      "step": 16722
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003710052234000447,
+      "loss": 0.1307,
+      "step": 16723
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003708960665900333,
+      "loss": 0.1103,
+      "step": 16724
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00037078692218436295,
+      "loss": 0.0684,
+      "step": 16725
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00037067779018518576,
+      "loss": 0.0626,
+      "step": 16726
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003705686705946535,
+      "loss": 0.0939,
+      "step": 16727
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003704595634149174,
+      "loss": 0.1065,
+      "step": 16728
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003703504686481295,
+      "loss": 0.0917,
+      "step": 16729
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003702413862964402,
+      "loss": 0.0867,
+      "step": 16730
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00037013231636200095,
+      "loss": 0.1157,
+      "step": 16731
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00037002325884696196,
+      "loss": 0.1599,
+      "step": 16732
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003699142137534738,
+      "loss": 0.1287,
+      "step": 16733
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003698051810836862,
+      "loss": 0.1124,
+      "step": 16734
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003696961608397493,
+      "loss": 0.1049,
+      "step": 16735
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003695871530238123,
+      "loss": 0.085,
+      "step": 16736
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036947815763802515,
+      "loss": 0.0856,
+      "step": 16737
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036936917468453667,
+      "loss": 0.0911,
+      "step": 16738
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036926020416549543,
+      "loss": 0.0882,
+      "step": 16739
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003691512460830506,
+      "loss": 0.0953,
+      "step": 16740
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003690423004393505,
+      "loss": 0.1425,
+      "step": 16741
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036893336723654304,
+      "loss": 0.0726,
+      "step": 16742
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003688244464767757,
+      "loss": 0.1285,
+      "step": 16743
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003687155381621969,
+      "loss": 0.0864,
+      "step": 16744
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003686066422949539,
+      "loss": 0.1337,
+      "step": 16745
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003684977588771935,
+      "loss": 0.061,
+      "step": 16746
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036838888791106285,
+      "loss": 0.0733,
+      "step": 16747
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003682800293987084,
+      "loss": 0.0737,
+      "step": 16748
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003681711833422765,
+      "loss": 0.0809,
+      "step": 16749
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036806234974391384,
+      "loss": 0.0624,
+      "step": 16750
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036795352860576557,
+      "loss": 0.1989,
+      "step": 16751
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036784471992997825,
+      "loss": 0.1132,
+      "step": 16752
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003677359237186969,
+      "loss": 0.0842,
+      "step": 16753
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003676271399740666,
+      "loss": 0.0929,
+      "step": 16754
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003675183686982324,
+      "loss": 0.1061,
+      "step": 16755
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003674096098933386,
+      "loss": 0.1222,
+      "step": 16756
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003673008635615304,
+      "loss": 0.0612,
+      "step": 16757
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036719212970495166,
+      "loss": 0.0627,
+      "step": 16758
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003670834083257463,
+      "loss": 0.0961,
+      "step": 16759
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036697469942605754,
+      "loss": 0.0939,
+      "step": 16760
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036686600300802973,
+      "loss": 0.0784,
+      "step": 16761
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036675731907380584,
+      "loss": 0.0922,
+      "step": 16762
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003666486476255282,
+      "loss": 0.0875,
+      "step": 16763
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036653998866534044,
+      "loss": 0.0753,
+      "step": 16764
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0003664313421953847,
+      "loss": 0.1023,
+      "step": 16765
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000366322708217803,
+      "loss": 0.1035,
+      "step": 16766
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036621408673473757,
+      "loss": 0.1039,
+      "step": 16767
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036610547774833,
+      "loss": 0.085,
+      "step": 16768
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036599688126072184,
+      "loss": 0.0687,
+      "step": 16769
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036588829727405404,
+      "loss": 0.0933,
+      "step": 16770
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036577972579046825,
+      "loss": 0.0871,
+      "step": 16771
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003656711668121045,
+      "loss": 0.0919,
+      "step": 16772
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.000365562620341104,
+      "loss": 0.0646,
+      "step": 16773
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036545408637960673,
+      "loss": 0.0996,
+      "step": 16774
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003653455649297527,
+      "loss": 0.07,
+      "step": 16775
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003652370559936813,
+      "loss": 0.0633,
+      "step": 16776
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036512855957353263,
+      "loss": 0.1004,
+      "step": 16777
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003650200756714458,
+      "loss": 0.0833,
+      "step": 16778
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036491160428955973,
+      "loss": 0.0881,
+      "step": 16779
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003648031454300134,
+      "loss": 0.131,
+      "step": 16780
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003646946990949447,
+      "loss": 0.0839,
+      "step": 16781
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003645862652864927,
+      "loss": 0.08,
+      "step": 16782
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036447784400679483,
+      "loss": 0.0751,
+      "step": 16783
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003643694352579896,
+      "loss": 0.0622,
+      "step": 16784
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003642610390422142,
+      "loss": 0.0808,
+      "step": 16785
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003641526553616058,
+      "loss": 0.0682,
+      "step": 16786
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003640442842183015,
+      "loss": 0.1132,
+      "step": 16787
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036393592561443825,
+      "loss": 0.0789,
+      "step": 16788
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003638275795521524,
+      "loss": 0.0966,
+      "step": 16789
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003637192460335801,
+      "loss": 0.0916,
+      "step": 16790
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036361092506085804,
+      "loss": 0.0756,
+      "step": 16791
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003635026166361216,
+      "loss": 0.0945,
+      "step": 16792
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.000363394320761506,
+      "loss": 0.0778,
+      "step": 16793
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003632860374391475,
+      "loss": 0.1171,
+      "step": 16794
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036317776667118054,
+      "loss": 0.0733,
+      "step": 16795
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036306950845973974,
+      "loss": 0.0962,
+      "step": 16796
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003629612628069604,
+      "loss": 0.0731,
+      "step": 16797
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036285302971497647,
+      "loss": 0.074,
+      "step": 16798
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003627448091859219,
+      "loss": 0.0918,
+      "step": 16799
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036263660122193064,
+      "loss": 0.1365,
+      "step": 16800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036252840582513634,
+      "loss": 0.0884,
+      "step": 16801
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036242022299767194,
+      "loss": 0.1062,
+      "step": 16802
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003623120527416712,
+      "loss": 0.0903,
+      "step": 16803
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003622038950592663,
+      "loss": 0.0673,
+      "step": 16804
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036209574995259053,
+      "loss": 0.1014,
+      "step": 16805
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003619876174237758,
+      "loss": 0.0859,
+      "step": 16806
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003618794974749544,
+      "loss": 0.0831,
+      "step": 16807
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036177139010825797,
+      "loss": 0.0784,
+      "step": 16808
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003616632953258182,
+      "loss": 0.0781,
+      "step": 16809
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003615552131297662,
+      "loss": 0.0712,
+      "step": 16810
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036144714352223374,
+      "loss": 0.0959,
+      "step": 16811
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003613390865053512,
+      "loss": 0.0957,
+      "step": 16812
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003612310420812489,
+      "loss": 0.0754,
+      "step": 16813
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036112301025205797,
+      "loss": 0.0814,
+      "step": 16814
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003610149910199081,
+      "loss": 0.0988,
+      "step": 16815
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003609069843869288,
+      "loss": 0.1558,
+      "step": 16816
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036079899035525045,
+      "loss": 0.0931,
+      "step": 16817
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003606910089270021,
+      "loss": 0.0848,
+      "step": 16818
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036058304010431276,
+      "loss": 0.074,
+      "step": 16819
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036047508388931137,
+      "loss": 0.0928,
+      "step": 16820
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036036714028412655,
+      "loss": 0.0947,
+      "step": 16821
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036025920929088663,
+      "loss": 0.0922,
+      "step": 16822
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003601512909117195,
+      "loss": 0.0744,
+      "step": 16823
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036004338514875354,
+      "loss": 0.0718,
+      "step": 16824
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003599354920041159,
+      "loss": 0.0756,
+      "step": 16825
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035982761147993457,
+      "loss": 0.1075,
+      "step": 16826
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035971974357833626,
+      "loss": 0.1292,
+      "step": 16827
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.000359611888301448,
+      "loss": 0.0892,
+      "step": 16828
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035950404565139626,
+      "loss": 0.0964,
+      "step": 16829
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003593962156303071,
+      "loss": 0.0593,
+      "step": 16830
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035928839824030757,
+      "loss": 0.118,
+      "step": 16831
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003591805934835229,
+      "loss": 0.0856,
+      "step": 16832
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035907280136207896,
+      "loss": 0.123,
+      "step": 16833
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035896502187810067,
+      "loss": 0.092,
+      "step": 16834
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003588572550337139,
+      "loss": 0.1293,
+      "step": 16835
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003587495008310429,
+      "loss": 0.1283,
+      "step": 16836
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035864175927221285,
+      "loss": 0.0653,
+      "step": 16837
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035853403035934794,
+      "loss": 0.0828,
+      "step": 16838
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035842631409457225,
+      "loss": 0.0783,
+      "step": 16839
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003583186104800098,
+      "loss": 0.1272,
+      "step": 16840
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035821091951778394,
+      "loss": 0.1215,
+      "step": 16841
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035810324121001835,
+      "loss": 0.0814,
+      "step": 16842
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003579955755588358,
+      "loss": 0.0596,
+      "step": 16843
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003578879225663596,
+      "loss": 0.0736,
+      "step": 16844
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035778028223471236,
+      "loss": 0.0819,
+      "step": 16845
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035767265456601604,
+      "loss": 0.0856,
+      "step": 16846
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003575650395623935,
+      "loss": 0.0761,
+      "step": 16847
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035745743722596614,
+      "loss": 0.1491,
+      "step": 16848
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003573498475588558,
+      "loss": 0.0897,
+      "step": 16849
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035724227056318337,
+      "loss": 0.0765,
+      "step": 16850
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035713470624107083,
+      "loss": 0.1266,
+      "step": 16851
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003570271545946386,
+      "loss": 0.1188,
+      "step": 16852
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003569196156260075,
+      "loss": 0.0822,
+      "step": 16853
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003568120893372977,
+      "loss": 0.0853,
+      "step": 16854
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003567045757306292,
+      "loss": 0.1024,
+      "step": 16855
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003565970748081224,
+      "loss": 0.1482,
+      "step": 16856
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035648958657189645,
+      "loss": 0.1155,
+      "step": 16857
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035638211102407137,
+      "loss": 0.074,
+      "step": 16858
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035627464816676593,
+      "loss": 0.0981,
+      "step": 16859
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00035616719800209894,
+      "loss": 0.0837,
+      "step": 16860
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035605976053218924,
+      "loss": 0.0725,
+      "step": 16861
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003559523357591551,
+      "loss": 0.0538,
+      "step": 16862
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035584492368511436,
+      "loss": 0.0767,
+      "step": 16863
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035573752431218553,
+      "loss": 0.077,
+      "step": 16864
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035563013764248607,
+      "loss": 0.0759,
+      "step": 16865
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003555227636781333,
+      "loss": 0.0929,
+      "step": 16866
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035541540242124394,
+      "loss": 0.064,
+      "step": 16867
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003553080538739356,
+      "loss": 0.0895,
+      "step": 16868
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035520071803832467,
+      "loss": 0.0751,
+      "step": 16869
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035509339491652706,
+      "loss": 0.089,
+      "step": 16870
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003549860845106597,
+      "loss": 0.0604,
+      "step": 16871
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035487878682283815,
+      "loss": 0.0945,
+      "step": 16872
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035477150185517795,
+      "loss": 0.1516,
+      "step": 16873
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003546642296097946,
+      "loss": 0.0728,
+      "step": 16874
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003545569700888032,
+      "loss": 0.0809,
+      "step": 16875
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003544497232943182,
+      "loss": 0.0673,
+      "step": 16876
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.000354342489228455,
+      "loss": 0.1058,
+      "step": 16877
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035423526789332737,
+      "loss": 0.1208,
+      "step": 16878
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003541280592910501,
+      "loss": 0.101,
+      "step": 16879
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003540208634237365,
+      "loss": 0.0753,
+      "step": 16880
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003539136802935006,
+      "loss": 0.1155,
+      "step": 16881
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035380650990245546,
+      "loss": 0.0825,
+      "step": 16882
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.000353699352252714,
+      "loss": 0.0935,
+      "step": 16883
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035359220734638974,
+      "loss": 0.0689,
+      "step": 16884
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.000353485075185595,
+      "loss": 0.0797,
+      "step": 16885
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035337795577244224,
+      "loss": 0.0823,
+      "step": 16886
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035327084910904337,
+      "loss": 0.0772,
+      "step": 16887
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003531637551975101,
+      "loss": 0.0789,
+      "step": 16888
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003530566740399547,
+      "loss": 0.0845,
+      "step": 16889
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003529496056384878,
+      "loss": 0.0783,
+      "step": 16890
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035284254999522115,
+      "loss": 0.048,
+      "step": 16891
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035273550711226545,
+      "loss": 0.0686,
+      "step": 16892
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003526284769917312,
+      "loss": 0.1291,
+      "step": 16893
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003525214596357288,
+      "loss": 0.0812,
+      "step": 16894
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003524144550463684,
+      "loss": 0.0895,
+      "step": 16895
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035230746322575935,
+      "loss": 0.0803,
+      "step": 16896
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035220048417601216,
+      "loss": 0.0806,
+      "step": 16897
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003520935178992356,
+      "loss": 0.0822,
+      "step": 16898
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035198656439753873,
+      "loss": 0.0923,
+      "step": 16899
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035187962367303085,
+      "loss": 0.0718,
+      "step": 16900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035177269572782033,
+      "loss": 0.0956,
+      "step": 16901
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003516657805640155,
+      "loss": 0.1519,
+      "step": 16902
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003515588781837241,
+      "loss": 0.0703,
+      "step": 16903
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003514519885890547,
+      "loss": 0.0554,
+      "step": 16904
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00035134511178211447,
+      "loss": 0.1049,
+      "step": 16905
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003512382477650108,
+      "loss": 0.0559,
+      "step": 16906
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035113139653985084,
+      "loss": 0.0845,
+      "step": 16907
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003510245581087409,
+      "loss": 0.1154,
+      "step": 16908
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035091773247378845,
+      "loss": 0.0652,
+      "step": 16909
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003508109196370991,
+      "loss": 0.0834,
+      "step": 16910
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035070411960077955,
+      "loss": 0.1027,
+      "step": 16911
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035059733236693526,
+      "loss": 0.0995,
+      "step": 16912
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003504905579376718,
+      "loss": 0.0945,
+      "step": 16913
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035038379631509466,
+      "loss": 0.1379,
+      "step": 16914
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035027704750130873,
+      "loss": 0.0812,
+      "step": 16915
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003501703114984186,
+      "loss": 0.0872,
+      "step": 16916
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00035006358830852935,
+      "loss": 0.098,
+      "step": 16917
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003499568779337452,
+      "loss": 0.0679,
+      "step": 16918
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034985018037617,
+      "loss": 0.0603,
+      "step": 16919
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003497434956379073,
+      "loss": 0.113,
+      "step": 16920
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003496368237210613,
+      "loss": 0.0684,
+      "step": 16921
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.000349530164627735,
+      "loss": 0.0929,
+      "step": 16922
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003494235183600312,
+      "loss": 0.059,
+      "step": 16923
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003493168849200531,
+      "loss": 0.0748,
+      "step": 16924
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003492102643099032,
+      "loss": 0.0897,
+      "step": 16925
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034910365653168364,
+      "loss": 0.0768,
+      "step": 16926
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003489970615874963,
+      "loss": 0.0756,
+      "step": 16927
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034889047947944343,
+      "loss": 0.0946,
+      "step": 16928
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034878391020962574,
+      "loss": 0.0854,
+      "step": 16929
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003486773537801454,
+      "loss": 0.1041,
+      "step": 16930
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003485708101931028,
+      "loss": 0.0645,
+      "step": 16931
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003484642794505992,
+      "loss": 0.1619,
+      "step": 16932
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034835776155473495,
+      "loss": 0.0671,
+      "step": 16933
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034825125650761034,
+      "loss": 0.1445,
+      "step": 16934
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034814476431132527,
+      "loss": 0.0831,
+      "step": 16935
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003480382849679792,
+      "loss": 0.0527,
+      "step": 16936
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034793181847967217,
+      "loss": 0.0735,
+      "step": 16937
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034782536484850336,
+      "loss": 0.0933,
+      "step": 16938
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034771892407657156,
+      "loss": 0.0841,
+      "step": 16939
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003476124961659757,
+      "loss": 0.1,
+      "step": 16940
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003475060811188137,
+      "loss": 0.0652,
+      "step": 16941
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003473996789371846,
+      "loss": 0.0967,
+      "step": 16942
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003472932896231857,
+      "loss": 0.0856,
+      "step": 16943
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003471869131789154,
+      "loss": 0.1312,
+      "step": 16944
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034708054960647084,
+      "loss": 0.0549,
+      "step": 16945
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034697419890794925,
+      "loss": 0.0852,
+      "step": 16946
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003468678610854475,
+      "loss": 0.0785,
+      "step": 16947
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034676153614106254,
+      "loss": 0.0528,
+      "step": 16948
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003466552240768905,
+      "loss": 0.0961,
+      "step": 16949
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034654892489502766,
+      "loss": 0.1548,
+      "step": 16950
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003464426385975702,
+      "loss": 0.0844,
+      "step": 16951
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00034633636518661346,
+      "loss": 0.0763,
+      "step": 16952
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034623010466425344,
+      "loss": 0.0825,
+      "step": 16953
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000346123857032585,
+      "loss": 0.0812,
+      "step": 16954
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034601762229370306,
+      "loss": 0.0954,
+      "step": 16955
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034591140044970195,
+      "loss": 0.1447,
+      "step": 16956
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034580519150267686,
+      "loss": 0.0839,
+      "step": 16957
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003456989954547215,
+      "loss": 0.0932,
+      "step": 16958
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034559281230792994,
+      "loss": 0.0665,
+      "step": 16959
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034548664206439573,
+      "loss": 0.0699,
+      "step": 16960
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003453804847262123,
+      "loss": 0.1425,
+      "step": 16961
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003452743402954724,
+      "loss": 0.1154,
+      "step": 16962
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003451682087742695,
+      "loss": 0.0725,
+      "step": 16963
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003450620901646964,
+      "loss": 0.0909,
+      "step": 16964
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000344955984468845,
+      "loss": 0.0842,
+      "step": 16965
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034484989168880774,
+      "loss": 0.0783,
+      "step": 16966
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034474381182667647,
+      "loss": 0.08,
+      "step": 16967
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034463774488454257,
+      "loss": 0.0801,
+      "step": 16968
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034453169086449763,
+      "loss": 0.085,
+      "step": 16969
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034442564976863244,
+      "loss": 0.1034,
+      "step": 16970
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003443196215990384,
+      "loss": 0.0788,
+      "step": 16971
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003442136063578059,
+      "loss": 0.1201,
+      "step": 16972
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034410760404702476,
+      "loss": 0.0842,
+      "step": 16973
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000344001614668786,
+      "loss": 0.0906,
+      "step": 16974
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000343895638225179,
+      "loss": 0.0978,
+      "step": 16975
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000343789674718293,
+      "loss": 0.0962,
+      "step": 16976
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003436837241502181,
+      "loss": 0.0903,
+      "step": 16977
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034357778652304295,
+      "loss": 0.0789,
+      "step": 16978
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003434718618388564,
+      "loss": 0.0678,
+      "step": 16979
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003433659500997469,
+      "loss": 0.1128,
+      "step": 16980
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000343260051307803,
+      "loss": 0.0664,
+      "step": 16981
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034315416546511267,
+      "loss": 0.0851,
+      "step": 16982
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003430482925737634,
+      "loss": 0.0817,
+      "step": 16983
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003429424326358428,
+      "loss": 0.1193,
+      "step": 16984
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003428365856534389,
+      "loss": 0.1241,
+      "step": 16985
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003427307516286381,
+      "loss": 0.0933,
+      "step": 16986
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003426249305635273,
+      "loss": 0.0829,
+      "step": 16987
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034251912246019303,
+      "loss": 0.1187,
+      "step": 16988
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034241332732072116,
+      "loss": 0.0792,
+      "step": 16989
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003423075451471985,
+      "loss": 0.071,
+      "step": 16990
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034220177594171045,
+      "loss": 0.0956,
+      "step": 16991
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003420960197063423,
+      "loss": 0.0981,
+      "step": 16992
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034199027644317946,
+      "loss": 0.0978,
+      "step": 16993
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034188454615430677,
+      "loss": 0.0895,
+      "step": 16994
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0003417788288418092,
+      "loss": 0.1082,
+      "step": 16995
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000341673124507771,
+      "loss": 0.0977,
+      "step": 16996
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034156743315427675,
+      "loss": 0.0977,
+      "step": 16997
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00034146175478341013,
+      "loss": 0.0789,
+      "step": 16998
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000341356089397255,
+      "loss": 0.1001,
+      "step": 16999
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003412504369978947,
+      "loss": 0.0581,
+      "step": 17000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00034114479758741255,
+      "loss": 0.1017,
+      "step": 17001
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003410391711678913,
+      "loss": 0.1226,
+      "step": 17002
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00034093355774141346,
+      "loss": 0.0995,
+      "step": 17003
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000340827957310062,
+      "loss": 0.1161,
+      "step": 17004
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00034072236987591853,
+      "loss": 0.1125,
+      "step": 17005
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00034061679544106563,
+      "loss": 0.0753,
+      "step": 17006
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00034051123400758457,
+      "loss": 0.098,
+      "step": 17007
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003404056855775567,
+      "loss": 0.0878,
+      "step": 17008
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00034030015015306303,
+      "loss": 0.1163,
+      "step": 17009
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000340194627736185,
+      "loss": 0.0966,
+      "step": 17010
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003400891183290029,
+      "loss": 0.1178,
+      "step": 17011
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003399836219335971,
+      "loss": 0.1155,
+      "step": 17012
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033987813855204777,
+      "loss": 0.1086,
+      "step": 17013
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003397726681864347,
+      "loss": 0.1588,
+      "step": 17014
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003396672108388373,
+      "loss": 0.0878,
+      "step": 17015
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003395617665113352,
+      "loss": 0.0937,
+      "step": 17016
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003394563352060077,
+      "loss": 0.1276,
+      "step": 17017
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003393509169249336,
+      "loss": 0.1333,
+      "step": 17018
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033924551167019114,
+      "loss": 0.1011,
+      "step": 17019
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003391401194438588,
+      "loss": 0.0914,
+      "step": 17020
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033903474024801473,
+      "loss": 0.0997,
+      "step": 17021
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033892937408473657,
+      "loss": 0.0927,
+      "step": 17022
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033882402095610176,
+      "loss": 0.0992,
+      "step": 17023
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.000338718680864188,
+      "loss": 0.1575,
+      "step": 17024
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033861335381107216,
+      "loss": 0.0883,
+      "step": 17025
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033850803979883074,
+      "loss": 0.1254,
+      "step": 17026
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033840273882954086,
+      "loss": 0.0874,
+      "step": 17027
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033829745090527844,
+      "loss": 0.1216,
+      "step": 17028
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033819217602811913,
+      "loss": 0.1235,
+      "step": 17029
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003380869142001394,
+      "loss": 0.0718,
+      "step": 17030
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033798166542341434,
+      "loss": 0.1022,
+      "step": 17031
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003378764297000194,
+      "loss": 0.0746,
+      "step": 17032
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003377712070320293,
+      "loss": 0.0725,
+      "step": 17033
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033766599742151904,
+      "loss": 0.054,
+      "step": 17034
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003375608008705627,
+      "loss": 0.0724,
+      "step": 17035
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033745561738123464,
+      "loss": 0.0886,
+      "step": 17036
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003373504469556088,
+      "loss": 0.0754,
+      "step": 17037
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003372452895957593,
+      "loss": 0.0825,
+      "step": 17038
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033714014530375914,
+      "loss": 0.1448,
+      "step": 17039
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003370350140816818,
+      "loss": 0.0801,
+      "step": 17040
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003369298959315998,
+      "loss": 0.0779,
+      "step": 17041
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033682479085558626,
+      "loss": 0.0919,
+      "step": 17042
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033671969885571286,
+      "loss": 0.0624,
+      "step": 17043
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003366146199340526,
+      "loss": 0.0816,
+      "step": 17044
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00033650955409267704,
+      "loss": 0.0574,
+      "step": 17045
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033640450133365775,
+      "loss": 0.0663,
+      "step": 17046
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033629946165906577,
+      "loss": 0.0779,
+      "step": 17047
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003361944350709729,
+      "loss": 0.1069,
+      "step": 17048
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003360894215714494,
+      "loss": 0.0539,
+      "step": 17049
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033598442116256647,
+      "loss": 0.0659,
+      "step": 17050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033587943384639406,
+      "loss": 0.1174,
+      "step": 17051
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003357744596250024,
+      "loss": 0.1086,
+      "step": 17052
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033566949850046126,
+      "loss": 0.0867,
+      "step": 17053
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033556455047484014,
+      "loss": 0.0602,
+      "step": 17054
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033545961555020844,
+      "loss": 0.0796,
+      "step": 17055
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003353546937286348,
+      "loss": 0.1038,
+      "step": 17056
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033524978501218875,
+      "loss": 0.0896,
+      "step": 17057
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033514488940293807,
+      "loss": 0.0598,
+      "step": 17058
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003350400069029518,
+      "loss": 0.0784,
+      "step": 17059
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003349351375142976,
+      "loss": 0.1429,
+      "step": 17060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003348302812390432,
+      "loss": 0.0663,
+      "step": 17061
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033472543807925616,
+      "loss": 0.0974,
+      "step": 17062
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033462060803700333,
+      "loss": 0.0786,
+      "step": 17063
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003345157911143524,
+      "loss": 0.0753,
+      "step": 17064
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003344109873133697,
+      "loss": 0.0913,
+      "step": 17065
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003343061966361217,
+      "loss": 0.0824,
+      "step": 17066
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033420141908467474,
+      "loss": 0.0646,
+      "step": 17067
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003340966546610942,
+      "loss": 0.082,
+      "step": 17068
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033399190336744666,
+      "loss": 0.0851,
+      "step": 17069
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033388716520579674,
+      "loss": 0.0787,
+      "step": 17070
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003337824401782105,
+      "loss": 0.1049,
+      "step": 17071
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033367772828675225,
+      "loss": 0.1467,
+      "step": 17072
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033357302953348664,
+      "loss": 0.0818,
+      "step": 17073
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003334683439204783,
+      "loss": 0.0704,
+      "step": 17074
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003333636714497913,
+      "loss": 0.0638,
+      "step": 17075
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000333259012123489,
+      "loss": 0.0677,
+      "step": 17076
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003331543659436359,
+      "loss": 0.0759,
+      "step": 17077
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003330497329122949,
+      "loss": 0.0757,
+      "step": 17078
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033294511303152876,
+      "loss": 0.0883,
+      "step": 17079
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003328405063034011,
+      "loss": 0.0792,
+      "step": 17080
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033273591272997403,
+      "loss": 0.1026,
+      "step": 17081
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033263133231331,
+      "loss": 0.0999,
+      "step": 17082
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003325267650554706,
+      "loss": 0.123,
+      "step": 17083
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003324222109585184,
+      "loss": 0.1044,
+      "step": 17084
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033231767002451465,
+      "loss": 0.0958,
+      "step": 17085
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033221314225552045,
+      "loss": 0.0609,
+      "step": 17086
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003321086276535971,
+      "loss": 0.0904,
+      "step": 17087
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000332004126220805,
+      "loss": 0.1169,
+      "step": 17088
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033189963795920464,
+      "loss": 0.1552,
+      "step": 17089
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033179516287085685,
+      "loss": 0.0917,
+      "step": 17090
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00033169070095782086,
+      "loss": 0.0965,
+      "step": 17091
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003315862522221571,
+      "loss": 0.0793,
+      "step": 17092
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00033148181666592483,
+      "loss": 0.0587,
+      "step": 17093
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.000331377394291183,
+      "loss": 0.0811,
+      "step": 17094
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003312729850999907,
+      "loss": 0.1257,
+      "step": 17095
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003311685890944064,
+      "loss": 0.0774,
+      "step": 17096
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00033106420627648905,
+      "loss": 0.0984,
+      "step": 17097
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003309598366482964,
+      "loss": 0.0855,
+      "step": 17098
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003308554802118865,
+      "loss": 0.0681,
+      "step": 17099
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003307511369693167,
+      "loss": 0.0776,
+      "step": 17100
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00033064680692264495,
+      "loss": 0.1257,
+      "step": 17101
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.000330542490073928,
+      "loss": 0.0657,
+      "step": 17102
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003304381864252225,
+      "loss": 0.0956,
+      "step": 17103
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00033033389597858563,
+      "loss": 0.0756,
+      "step": 17104
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00033022961873607347,
+      "loss": 0.1123,
+      "step": 17105
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00033012535469974214,
+      "loss": 0.0824,
+      "step": 17106
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003300211038716474,
+      "loss": 0.1204,
+      "step": 17107
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003299168662538448,
+      "loss": 0.0611,
+      "step": 17108
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032981264184838924,
+      "loss": 0.1674,
+      "step": 17109
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032970843065733646,
+      "loss": 0.0724,
+      "step": 17110
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032960423268274075,
+      "loss": 0.0853,
+      "step": 17111
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003295000479266571,
+      "loss": 0.1545,
+      "step": 17112
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003293958763911395,
+      "loss": 0.083,
+      "step": 17113
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003292917180782421,
+      "loss": 0.1008,
+      "step": 17114
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003291875729900183,
+      "loss": 0.1299,
+      "step": 17115
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003290834411285216,
+      "loss": 0.0776,
+      "step": 17116
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032897932249580564,
+      "loss": 0.1255,
+      "step": 17117
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003288752170939232,
+      "loss": 0.0864,
+      "step": 17118
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003287711249249269,
+      "loss": 0.0953,
+      "step": 17119
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032866704599086903,
+      "loss": 0.0757,
+      "step": 17120
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003285629802938018,
+      "loss": 0.0998,
+      "step": 17121
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032845892783577745,
+      "loss": 0.0817,
+      "step": 17122
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003283548886188471,
+      "loss": 0.0801,
+      "step": 17123
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003282508626450628,
+      "loss": 0.0984,
+      "step": 17124
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003281468499164754,
+      "loss": 0.071,
+      "step": 17125
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003280428504351356,
+      "loss": 0.1251,
+      "step": 17126
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003279388642030942,
+      "loss": 0.0898,
+      "step": 17127
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032783489122240126,
+      "loss": 0.092,
+      "step": 17128
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003277309314951069,
+      "loss": 0.0973,
+      "step": 17129
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003276269850232615,
+      "loss": 0.0543,
+      "step": 17130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003275230518089142,
+      "loss": 0.0836,
+      "step": 17131
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032741913185411385,
+      "loss": 0.112,
+      "step": 17132
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003273152251609105,
+      "loss": 0.0908,
+      "step": 17133
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003272113317313523,
+      "loss": 0.0611,
+      "step": 17134
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032710745156748787,
+      "loss": 0.0612,
+      "step": 17135
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00032700358467136517,
+      "loss": 0.0651,
+      "step": 17136
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003268997310450327,
+      "loss": 0.0669,
+      "step": 17137
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032679589069053804,
+      "loss": 0.0913,
+      "step": 17138
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032669206360992864,
+      "loss": 0.1127,
+      "step": 17139
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032658824980525157,
+      "loss": 0.0757,
+      "step": 17140
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003264844492785539,
+      "loss": 0.1227,
+      "step": 17141
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.000326380662031882,
+      "loss": 0.0752,
+      "step": 17142
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003262768880672828,
+      "loss": 0.0637,
+      "step": 17143
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032617312738680203,
+      "loss": 0.0732,
+      "step": 17144
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003260693799924861,
+      "loss": 0.0986,
+      "step": 17145
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032596564588638024,
+      "loss": 0.0882,
+      "step": 17146
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032586192507053004,
+      "loss": 0.076,
+      "step": 17147
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003257582175469804,
+      "loss": 0.0668,
+      "step": 17148
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032565452331777603,
+      "loss": 0.1443,
+      "step": 17149
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003255508423849619,
+      "loss": 0.0811,
+      "step": 17150
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032544717475058236,
+      "loss": 0.0961,
+      "step": 17151
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032534352041668123,
+      "loss": 0.1121,
+      "step": 17152
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003252398793853022,
+      "loss": 0.0939,
+      "step": 17153
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032513625165848914,
+      "loss": 0.0786,
+      "step": 17154
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032503263723828525,
+      "loss": 0.0878,
+      "step": 17155
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003249290361267332,
+      "loss": 0.0771,
+      "step": 17156
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032482544832587636,
+      "loss": 0.0978,
+      "step": 17157
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032472187383775675,
+      "loss": 0.0692,
+      "step": 17158
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032461831266441686,
+      "loss": 0.0601,
+      "step": 17159
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003245147648078984,
+      "loss": 0.0995,
+      "step": 17160
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032441123027024334,
+      "loss": 0.1086,
+      "step": 17161
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003243077090534928,
+      "loss": 0.1211,
+      "step": 17162
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032420420115968795,
+      "loss": 0.1146,
+      "step": 17163
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003241007065908702,
+      "loss": 0.1036,
+      "step": 17164
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003239972253490795,
+      "loss": 0.0854,
+      "step": 17165
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032389375743635697,
+      "loss": 0.0677,
+      "step": 17166
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032379030285474244,
+      "loss": 0.1146,
+      "step": 17167
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003236868616062757,
+      "loss": 0.0908,
+      "step": 17168
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.000323583433692996,
+      "loss": 0.1693,
+      "step": 17169
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032348001911694345,
+      "loss": 0.1148,
+      "step": 17170
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003233766178801567,
+      "loss": 0.2172,
+      "step": 17171
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032327322998467454,
+      "loss": 0.124,
+      "step": 17172
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003231698554325356,
+      "loss": 0.1057,
+      "step": 17173
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003230664942257778,
+      "loss": 0.116,
+      "step": 17174
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032296314636643986,
+      "loss": 0.0613,
+      "step": 17175
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032285981185655875,
+      "loss": 0.0949,
+      "step": 17176
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003227564906981728,
+      "loss": 0.1833,
+      "step": 17177
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032265318289331893,
+      "loss": 0.1648,
+      "step": 17178
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003225498884440339,
+      "loss": 0.0894,
+      "step": 17179
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032244660735235444,
+      "loss": 0.0996,
+      "step": 17180
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00032234333962031717,
+      "loss": 0.0822,
+      "step": 17181
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0003222400852499582,
+      "loss": 0.0997,
+      "step": 17182
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.000322136844243313,
+      "loss": 0.0897,
+      "step": 17183
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032203361660241805,
+      "loss": 0.072,
+      "step": 17184
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032193040232930827,
+      "loss": 0.1061,
+      "step": 17185
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003218272014260186,
+      "loss": 0.0561,
+      "step": 17186
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003217240138945844,
+      "loss": 0.0842,
+      "step": 17187
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032162083973704016,
+      "loss": 0.079,
+      "step": 17188
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003215176789554196,
+      "loss": 0.0877,
+      "step": 17189
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003214145315517576,
+      "loss": 0.0775,
+      "step": 17190
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032131139752808766,
+      "loss": 0.0811,
+      "step": 17191
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032120827688644326,
+      "loss": 0.0872,
+      "step": 17192
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003211051696288577,
+      "loss": 0.0953,
+      "step": 17193
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032100207575736406,
+      "loss": 0.065,
+      "step": 17194
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003208989952739947,
+      "loss": 0.131,
+      "step": 17195
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032079592818078274,
+      "loss": 0.0847,
+      "step": 17196
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003206928744797598,
+      "loss": 0.1224,
+      "step": 17197
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003205898341729584,
+      "loss": 0.0594,
+      "step": 17198
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003204868072624101,
+      "loss": 0.0908,
+      "step": 17199
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032038379375014605,
+      "loss": 0.1307,
+      "step": 17200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032028079363819774,
+      "loss": 0.1091,
+      "step": 17201
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032017780692859577,
+      "loss": 0.0676,
+      "step": 17202
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00032007483362337064,
+      "loss": 0.1293,
+      "step": 17203
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031997187372455327,
+      "loss": 0.1274,
+      "step": 17204
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003198689272341736,
+      "loss": 0.0997,
+      "step": 17205
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000319765994154261,
+      "loss": 0.0474,
+      "step": 17206
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003196630744868457,
+      "loss": 0.1067,
+      "step": 17207
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031956016823395686,
+      "loss": 0.0922,
+      "step": 17208
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031945727539762293,
+      "loss": 0.0991,
+      "step": 17209
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003193543959798736,
+      "loss": 0.0547,
+      "step": 17210
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000319251529982737,
+      "loss": 0.0681,
+      "step": 17211
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031914867740824126,
+      "loss": 0.0662,
+      "step": 17212
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003190458382584145,
+      "loss": 0.1053,
+      "step": 17213
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031894301253528447,
+      "loss": 0.108,
+      "step": 17214
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031884020024087843,
+      "loss": 0.0647,
+      "step": 17215
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031873740137722354,
+      "loss": 0.0964,
+      "step": 17216
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031863461594634724,
+      "loss": 0.1018,
+      "step": 17217
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003185318439502756,
+      "loss": 0.0724,
+      "step": 17218
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031842908539103556,
+      "loss": 0.044,
+      "step": 17219
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000318326340270653,
+      "loss": 0.0376,
+      "step": 17220
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00031822360859115373,
+      "loss": 0.0618,
+      "step": 17221
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003181208903545636,
+      "loss": 0.1094,
+      "step": 17222
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003180181855629073,
+      "loss": 0.0908,
+      "step": 17223
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003179154942182109,
+      "loss": 0.0723,
+      "step": 17224
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003178128163224986,
+      "loss": 0.0925,
+      "step": 17225
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000317710151877795,
+      "loss": 0.069,
+      "step": 17226
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003176075008861242,
+      "loss": 0.0947,
+      "step": 17227
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003175048633495108,
+      "loss": 0.0934,
+      "step": 17228
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003174022392699778,
+      "loss": 0.0792,
+      "step": 17229
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003172996286495494,
+      "loss": 0.0803,
+      "step": 17230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003171970314902486,
+      "loss": 0.1021,
+      "step": 17231
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003170944477940984,
+      "loss": 0.0586,
+      "step": 17232
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003169918775631212,
+      "loss": 0.0887,
+      "step": 17233
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003168893207993396,
+      "loss": 0.0398,
+      "step": 17234
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003167867775047758,
+      "loss": 0.0942,
+      "step": 17235
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003166842476814513,
+      "loss": 0.0833,
+      "step": 17236
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031658173133138843,
+      "loss": 0.0684,
+      "step": 17237
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031647922845660817,
+      "loss": 0.0628,
+      "step": 17238
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003163767390591313,
+      "loss": 0.0529,
+      "step": 17239
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003162742631409795,
+      "loss": 0.0641,
+      "step": 17240
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031617180070417274,
+      "loss": 0.0978,
+      "step": 17241
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003160693517507314,
+      "loss": 0.0693,
+      "step": 17242
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031596691628267516,
+      "loss": 0.0647,
+      "step": 17243
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031586449430202456,
+      "loss": 0.0453,
+      "step": 17244
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031576208581079866,
+      "loss": 0.0862,
+      "step": 17245
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003156596908110169,
+      "loss": 0.0464,
+      "step": 17246
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031555730930469796,
+      "loss": 0.0671,
+      "step": 17247
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003154549412938603,
+      "loss": 0.0662,
+      "step": 17248
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003153525867805231,
+      "loss": 0.0796,
+      "step": 17249
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003152502457667039,
+      "loss": 0.094,
+      "step": 17250
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031514791825442123,
+      "loss": 0.0549,
+      "step": 17251
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003150456042456923,
+      "loss": 0.1078,
+      "step": 17252
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031494330374253463,
+      "loss": 0.084,
+      "step": 17253
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003148410167469652,
+      "loss": 0.1177,
+      "step": 17254
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031473874326100093,
+      "loss": 0.0756,
+      "step": 17255
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031463648328665806,
+      "loss": 0.06,
+      "step": 17256
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003145342368259535,
+      "loss": 0.1415,
+      "step": 17257
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003144320038809031,
+      "loss": 0.075,
+      "step": 17258
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031432978445352245,
+      "loss": 0.0916,
+      "step": 17259
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031422757854582685,
+      "loss": 0.1425,
+      "step": 17260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003141253861598322,
+      "loss": 0.0869,
+      "step": 17261
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031402320729755306,
+      "loss": 0.1228,
+      "step": 17262
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003139210419610039,
+      "loss": 0.058,
+      "step": 17263
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003138188901521998,
+      "loss": 0.0858,
+      "step": 17264
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031371675187315463,
+      "loss": 0.0979,
+      "step": 17265
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003136146271258822,
+      "loss": 0.0648,
+      "step": 17266
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031351251591239614,
+      "loss": 0.0784,
+      "step": 17267
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031341041823471007,
+      "loss": 0.0623,
+      "step": 17268
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003133083340948365,
+      "loss": 0.0758,
+      "step": 17269
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003132062634947891,
+      "loss": 0.0724,
+      "step": 17270
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031310420643657967,
+      "loss": 0.1341,
+      "step": 17271
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003130021629222213,
+      "loss": 0.0706,
+      "step": 17272
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0003129001329537256,
+      "loss": 0.0717,
+      "step": 17273
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031279811653310417,
+      "loss": 0.0979,
+      "step": 17274
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00031269611366236896,
+      "loss": 0.0863,
+      "step": 17275
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003125941243435305,
+      "loss": 0.0807,
+      "step": 17276
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031249214857860064,
+      "loss": 0.071,
+      "step": 17277
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003123901863695896,
+      "loss": 0.0711,
+      "step": 17278
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003122882377185079,
+      "loss": 0.0869,
+      "step": 17279
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003121863026273658,
+      "loss": 0.1005,
+      "step": 17280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003120843810981726,
+      "loss": 0.0886,
+      "step": 17281
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031198247313293857,
+      "loss": 0.0769,
+      "step": 17282
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031188057873367324,
+      "loss": 0.1073,
+      "step": 17283
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003117786979023853,
+      "loss": 0.0815,
+      "step": 17284
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003116768306410839,
+      "loss": 0.158,
+      "step": 17285
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003115749769517773,
+      "loss": 0.1125,
+      "step": 17286
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031147313683647384,
+      "loss": 0.1063,
+      "step": 17287
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003113713102971817,
+      "loss": 0.095,
+      "step": 17288
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031126949733590816,
+      "loss": 0.1046,
+      "step": 17289
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003111676979546615,
+      "loss": 0.0858,
+      "step": 17290
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003110659121554484,
+      "loss": 0.0745,
+      "step": 17291
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031096413994027584,
+      "loss": 0.1088,
+      "step": 17292
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031086238131115086,
+      "loss": 0.0875,
+      "step": 17293
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003107606362700797,
+      "loss": 0.0975,
+      "step": 17294
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003106589048190684,
+      "loss": 0.08,
+      "step": 17295
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003105571869601228,
+      "loss": 0.0812,
+      "step": 17296
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003104554826952488,
+      "loss": 0.0822,
+      "step": 17297
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031035379202645167,
+      "loss": 0.0925,
+      "step": 17298
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00031025211495573635,
+      "loss": 0.132,
+      "step": 17299
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003101504514851078,
+      "loss": 0.1406,
+      "step": 17300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003100488016165701,
+      "loss": 0.0664,
+      "step": 17301
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003099471653521283,
+      "loss": 0.0764,
+      "step": 17302
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030984554269378574,
+      "loss": 0.1416,
+      "step": 17303
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003097439336435467,
+      "loss": 0.1118,
+      "step": 17304
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003096423382034145,
+      "loss": 0.0903,
+      "step": 17305
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030954075637539214,
+      "loss": 0.0817,
+      "step": 17306
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030943918816148276,
+      "loss": 0.0889,
+      "step": 17307
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030933763356368895,
+      "loss": 0.0643,
+      "step": 17308
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030923609258401273,
+      "loss": 0.0895,
+      "step": 17309
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030913456522445695,
+      "loss": 0.0493,
+      "step": 17310
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003090330514870231,
+      "loss": 0.1098,
+      "step": 17311
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003089315513737128,
+      "loss": 0.0967,
+      "step": 17312
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003088300648865271,
+      "loss": 0.0959,
+      "step": 17313
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003087285920274676,
+      "loss": 0.0413,
+      "step": 17314
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003086271327985348,
+      "loss": 0.0984,
+      "step": 17315
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030852568720172904,
+      "loss": 0.0836,
+      "step": 17316
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00030842425523905094,
+      "loss": 0.1094,
+      "step": 17317
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003083228369125004,
+      "loss": 0.1072,
+      "step": 17318
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003082214322240771,
+      "loss": 0.0751,
+      "step": 17319
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003081200411757804,
+      "loss": 0.0834,
+      "step": 17320
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003080186637696094,
+      "loss": 0.0563,
+      "step": 17321
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003079173000075629,
+      "loss": 0.0807,
+      "step": 17322
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003078159498916401,
+      "loss": 0.1206,
+      "step": 17323
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030771461342383864,
+      "loss": 0.1396,
+      "step": 17324
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003076132906061574,
+      "loss": 0.1042,
+      "step": 17325
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003075119814405937,
+      "loss": 0.1024,
+      "step": 17326
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030741068592914523,
+      "loss": 0.1735,
+      "step": 17327
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030730940407380927,
+      "loss": 0.0803,
+      "step": 17328
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030720813587658256,
+      "loss": 0.0741,
+      "step": 17329
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003071068813394624,
+      "loss": 0.0844,
+      "step": 17330
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000307005640464445,
+      "loss": 0.0746,
+      "step": 17331
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003069044132535267,
+      "loss": 0.0773,
+      "step": 17332
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003068031997087032,
+      "loss": 0.072,
+      "step": 17333
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030670199983197003,
+      "loss": 0.0759,
+      "step": 17334
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003066008136253231,
+      "loss": 0.0708,
+      "step": 17335
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000306499641090757,
+      "loss": 0.0681,
+      "step": 17336
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003063984822302673,
+      "loss": 0.0783,
+      "step": 17337
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003062973370458482,
+      "loss": 0.067,
+      "step": 17338
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000306196205539494,
+      "loss": 0.0497,
+      "step": 17339
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030609508771319884,
+      "loss": 0.0595,
+      "step": 17340
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003059939835689564,
+      "loss": 0.0927,
+      "step": 17341
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003058928931087599,
+      "loss": 0.0575,
+      "step": 17342
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003057918163346034,
+      "loss": 0.0751,
+      "step": 17343
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003056907532484793,
+      "loss": 0.0641,
+      "step": 17344
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003055897038523802,
+      "loss": 0.1069,
+      "step": 17345
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030548866814829903,
+      "loss": 0.1012,
+      "step": 17346
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003053876461382277,
+      "loss": 0.1075,
+      "step": 17347
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003052866378241581,
+      "loss": 0.0808,
+      "step": 17348
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003051856432080815,
+      "loss": 0.1244,
+      "step": 17349
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030508466229199003,
+      "loss": 0.0831,
+      "step": 17350
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030498369507787437,
+      "loss": 0.0759,
+      "step": 17351
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030488274156772533,
+      "loss": 0.0583,
+      "step": 17352
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030478180176353356,
+      "loss": 0.0585,
+      "step": 17353
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003046808756672892,
+      "loss": 0.1016,
+      "step": 17354
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003045799632809819,
+      "loss": 0.0789,
+      "step": 17355
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003044790646066019,
+      "loss": 0.0765,
+      "step": 17356
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003043781796461388,
+      "loss": 0.0977,
+      "step": 17357
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030427730840158166,
+      "loss": 0.0792,
+      "step": 17358
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003041764508749193,
+      "loss": 0.0933,
+      "step": 17359
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003040756070681404,
+      "loss": 0.1156,
+      "step": 17360
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003039747769832333,
+      "loss": 0.0661,
+      "step": 17361
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003038739606221859,
+      "loss": 0.1028,
+      "step": 17362
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00030377315798698647,
+      "loss": 0.0602,
+      "step": 17363
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003036723690796225,
+      "loss": 0.0862,
+      "step": 17364
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003035715939020812,
+      "loss": 0.0834,
+      "step": 17365
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003034708324563493,
+      "loss": 0.1051,
+      "step": 17366
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003033700847444142,
+      "loss": 0.0887,
+      "step": 17367
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000303269350768262,
+      "loss": 0.0765,
+      "step": 17368
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003031686305298788,
+      "loss": 0.1082,
+      "step": 17369
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030306792403125084,
+      "loss": 0.077,
+      "step": 17370
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030296723127436377,
+      "loss": 0.0715,
+      "step": 17371
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030286655226120286,
+      "loss": 0.0919,
+      "step": 17372
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003027658869937534,
+      "loss": 0.0689,
+      "step": 17373
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030266523547400003,
+      "loss": 0.1107,
+      "step": 17374
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030256459770392765,
+      "loss": 0.0822,
+      "step": 17375
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003024639736855199,
+      "loss": 0.0907,
+      "step": 17376
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003023633634207614,
+      "loss": 0.1404,
+      "step": 17377
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003022627669116361,
+      "loss": 0.0714,
+      "step": 17378
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030216218416012744,
+      "loss": 0.0735,
+      "step": 17379
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003020616151682183,
+      "loss": 0.0725,
+      "step": 17380
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003019610599378919,
+      "loss": 0.1182,
+      "step": 17381
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003018605184711306,
+      "loss": 0.0746,
+      "step": 17382
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030175999076991736,
+      "loss": 0.0983,
+      "step": 17383
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003016594768362341,
+      "loss": 0.0589,
+      "step": 17384
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030155897667206256,
+      "loss": 0.0903,
+      "step": 17385
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030145849027938453,
+      "loss": 0.0609,
+      "step": 17386
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000301358017660181,
+      "loss": 0.0834,
+      "step": 17387
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003012575588164336,
+      "loss": 0.1053,
+      "step": 17388
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030115711375012247,
+      "loss": 0.1014,
+      "step": 17389
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003010566824632288,
+      "loss": 0.0854,
+      "step": 17390
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003009562649577325,
+      "loss": 0.0485,
+      "step": 17391
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003008558612356137,
+      "loss": 0.1135,
+      "step": 17392
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030075547129885193,
+      "loss": 0.0764,
+      "step": 17393
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003006550951494268,
+      "loss": 0.1003,
+      "step": 17394
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003005547327893172,
+      "loss": 0.0588,
+      "step": 17395
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030045438422050195,
+      "loss": 0.0833,
+      "step": 17396
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030035404944496024,
+      "loss": 0.0775,
+      "step": 17397
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003002537284646698,
+      "loss": 0.1082,
+      "step": 17398
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00030015342128160926,
+      "loss": 0.0772,
+      "step": 17399
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003000531278977562,
+      "loss": 0.067,
+      "step": 17400
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002999528483150881,
+      "loss": 0.1204,
+      "step": 17401
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002998525825355818,
+      "loss": 0.1004,
+      "step": 17402
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002997523305612151,
+      "loss": 0.1049,
+      "step": 17403
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002996520923939644,
+      "loss": 0.0665,
+      "step": 17404
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00029955186803580605,
+      "loss": 0.1146,
+      "step": 17405
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00029945165748871614,
+      "loss": 0.0791,
+      "step": 17406
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002993514607546708,
+      "loss": 0.0732,
+      "step": 17407
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002992512778356452,
+      "loss": 0.0969,
+      "step": 17408
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00029915110873361504,
+      "loss": 0.1024,
+      "step": 17409
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002990509534505557,
+      "loss": 0.1399,
+      "step": 17410
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00029895081198844163,
+      "loss": 0.0873,
+      "step": 17411
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00029885068434924744,
+      "loss": 0.0833,
+      "step": 17412
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002987505705349474,
+      "loss": 0.1136,
+      "step": 17413
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00029865047054751547,
+      "loss": 0.0865,
+      "step": 17414
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029855038438892533,
+      "loss": 0.1224,
+      "step": 17415
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029845031206115016,
+      "loss": 0.1072,
+      "step": 17416
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029835025356616374,
+      "loss": 0.1011,
+      "step": 17417
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002982502089059387,
+      "loss": 0.0745,
+      "step": 17418
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002981501780824474,
+      "loss": 0.1373,
+      "step": 17419
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029805016109766256,
+      "loss": 0.0748,
+      "step": 17420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029795015795355627,
+      "loss": 0.1418,
+      "step": 17421
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002978501686520999,
+      "loss": 0.0699,
+      "step": 17422
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002977501931952656,
+      "loss": 0.1409,
+      "step": 17423
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029765023158502427,
+      "loss": 0.0651,
+      "step": 17424
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029755028382334704,
+      "loss": 0.0969,
+      "step": 17425
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029745034991220446,
+      "loss": 0.0991,
+      "step": 17426
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029735042985356707,
+      "loss": 0.0757,
+      "step": 17427
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002972505236494051,
+      "loss": 0.0727,
+      "step": 17428
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029715063130168805,
+      "loss": 0.0894,
+      "step": 17429
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002970507528123858,
+      "loss": 0.1401,
+      "step": 17430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029695088818346816,
+      "loss": 0.0653,
+      "step": 17431
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029685103741690377,
+      "loss": 0.0748,
+      "step": 17432
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029675120051466156,
+      "loss": 0.1316,
+      "step": 17433
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029665137747870995,
+      "loss": 0.1124,
+      "step": 17434
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029655156831101727,
+      "loss": 0.1061,
+      "step": 17435
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002964517730135511,
+      "loss": 0.0698,
+      "step": 17436
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029635199158827984,
+      "loss": 0.0629,
+      "step": 17437
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029625222403717065,
+      "loss": 0.1029,
+      "step": 17438
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002961524703621907,
+      "loss": 0.0805,
+      "step": 17439
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002960527305653065,
+      "loss": 0.0586,
+      "step": 17440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029595300464848527,
+      "loss": 0.1648,
+      "step": 17441
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029585329261369277,
+      "loss": 0.0554,
+      "step": 17442
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029575359446289574,
+      "loss": 0.0995,
+      "step": 17443
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029565391019805955,
+      "loss": 0.0955,
+      "step": 17444
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029555423982114993,
+      "loss": 0.1207,
+      "step": 17445
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002954545833341319,
+      "loss": 0.0566,
+      "step": 17446
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002953549407389705,
+      "loss": 0.0713,
+      "step": 17447
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029525531203763045,
+      "loss": 0.0715,
+      "step": 17448
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002951556972320759,
+      "loss": 0.0898,
+      "step": 17449
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029505609632427167,
+      "loss": 0.0859,
+      "step": 17450
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002949565093161809,
+      "loss": 0.079,
+      "step": 17451
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029485693620976783,
+      "loss": 0.0895,
+      "step": 17452
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029475737700699554,
+      "loss": 0.0862,
+      "step": 17453
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002946578317098271,
+      "loss": 0.0822,
+      "step": 17454
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002945583003202251,
+      "loss": 0.0858,
+      "step": 17455
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.000294458782840152,
+      "loss": 0.1432,
+      "step": 17456
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029435927927157056,
+      "loss": 0.1042,
+      "step": 17457
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002942597896164424,
+      "loss": 0.0825,
+      "step": 17458
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00029416031387672915,
+      "loss": 0.0749,
+      "step": 17459
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002940608520543923,
+      "loss": 0.0701,
+      "step": 17460
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002939614041513926,
+      "loss": 0.1277,
+      "step": 17461
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029386197016969173,
+      "loss": 0.1583,
+      "step": 17462
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002937625501112494,
+      "loss": 0.0966,
+      "step": 17463
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029366314397802676,
+      "loss": 0.1143,
+      "step": 17464
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029356375177198336,
+      "loss": 0.0957,
+      "step": 17465
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029346437349507906,
+      "loss": 0.1167,
+      "step": 17466
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002933650091492733,
+      "loss": 0.0726,
+      "step": 17467
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002932656587365254,
+      "loss": 0.0526,
+      "step": 17468
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002931663222587939,
+      "loss": 0.105,
+      "step": 17469
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002930669997180381,
+      "loss": 0.0944,
+      "step": 17470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000292967691116216,
+      "loss": 0.0575,
+      "step": 17471
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002928683964552855,
+      "loss": 0.0822,
+      "step": 17472
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002927691157372051,
+      "loss": 0.0953,
+      "step": 17473
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000292669848963932,
+      "loss": 0.1063,
+      "step": 17474
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002925705961374234,
+      "loss": 0.0558,
+      "step": 17475
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029247135725963625,
+      "loss": 0.0728,
+      "step": 17476
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002923721323325276,
+      "loss": 0.1328,
+      "step": 17477
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002922729213580538,
+      "loss": 0.1108,
+      "step": 17478
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000292173724338171,
+      "loss": 0.0692,
+      "step": 17479
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029207454127483514,
+      "loss": 0.1202,
+      "step": 17480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029197537217000183,
+      "loss": 0.0821,
+      "step": 17481
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029187621702562604,
+      "loss": 0.0677,
+      "step": 17482
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029177707584366377,
+      "loss": 0.0773,
+      "step": 17483
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029167794862606876,
+      "loss": 0.1061,
+      "step": 17484
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002915788353747966,
+      "loss": 0.13,
+      "step": 17485
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000291479736091801,
+      "loss": 0.1199,
+      "step": 17486
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029138065077903596,
+      "loss": 0.1064,
+      "step": 17487
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029128157943845534,
+      "loss": 0.1126,
+      "step": 17488
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002911825220720121,
+      "loss": 0.1594,
+      "step": 17489
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002910834786816601,
+      "loss": 0.0566,
+      "step": 17490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029098444926935186,
+      "loss": 0.123,
+      "step": 17491
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029088543383704014,
+      "loss": 0.1055,
+      "step": 17492
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002907864323866768,
+      "loss": 0.1105,
+      "step": 17493
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029068744492021447,
+      "loss": 0.063,
+      "step": 17494
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029058847143960485,
+      "loss": 0.1211,
+      "step": 17495
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002904895119467991,
+      "loss": 0.1144,
+      "step": 17496
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029039056644374886,
+      "loss": 0.1899,
+      "step": 17497
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029029163493240486,
+      "loss": 0.0811,
+      "step": 17498
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029019271741471795,
+      "loss": 0.084,
+      "step": 17499
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00029009381389263835,
+      "loss": 0.1289,
+      "step": 17500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00028999492436811625,
+      "loss": 0.0678,
+      "step": 17501
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00028989604884310116,
+      "loss": 0.0717,
+      "step": 17502
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002897971873195433,
+      "loss": 0.0894,
+      "step": 17503
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00028969833979939144,
+      "loss": 0.0931,
+      "step": 17504
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00028959950628459506,
+      "loss": 0.1199,
+      "step": 17505
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002895006867771026,
+      "loss": 0.1055,
+      "step": 17506
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002894018812788627,
+      "loss": 0.0845,
+      "step": 17507
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028930308979182353,
+      "loss": 0.0942,
+      "step": 17508
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002892043123179325,
+      "loss": 0.1078,
+      "step": 17509
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002891055488591381,
+      "loss": 0.0972,
+      "step": 17510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002890067994173872,
+      "loss": 0.1121,
+      "step": 17511
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000288908063994627,
+      "loss": 0.1056,
+      "step": 17512
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028880934259280434,
+      "loss": 0.0645,
+      "step": 17513
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002887106352138653,
+      "loss": 0.0939,
+      "step": 17514
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028861194185975694,
+      "loss": 0.078,
+      "step": 17515
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002885132625324245,
+      "loss": 0.078,
+      "step": 17516
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028841459723381434,
+      "loss": 0.057,
+      "step": 17517
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028831594596587164,
+      "loss": 0.0895,
+      "step": 17518
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002882173087305414,
+      "loss": 0.1062,
+      "step": 17519
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002881186855297686,
+      "loss": 0.0936,
+      "step": 17520
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028802007636549796,
+      "loss": 0.0807,
+      "step": 17521
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028792148123967334,
+      "loss": 0.0782,
+      "step": 17522
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002878229001542394,
+      "loss": 0.1301,
+      "step": 17523
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028772433311113964,
+      "loss": 0.1245,
+      "step": 17524
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002876257801123172,
+      "loss": 0.1085,
+      "step": 17525
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028752724115971604,
+      "loss": 0.1321,
+      "step": 17526
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028742871625527867,
+      "loss": 0.1065,
+      "step": 17527
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002873302054009478,
+      "loss": 0.0945,
+      "step": 17528
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002872317085986655,
+      "loss": 0.1196,
+      "step": 17529
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002871332258503745,
+      "loss": 0.1125,
+      "step": 17530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028703475715801633,
+      "loss": 0.101,
+      "step": 17531
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028693630252353254,
+      "loss": 0.0897,
+      "step": 17532
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002868378619488643,
+      "loss": 0.0777,
+      "step": 17533
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002867394354359528,
+      "loss": 0.1219,
+      "step": 17534
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028664102298673835,
+      "loss": 0.1378,
+      "step": 17535
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002865426246031619,
+      "loss": 0.1006,
+      "step": 17536
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028644424028716333,
+      "loss": 0.0838,
+      "step": 17537
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002863458700406828,
+      "loss": 0.1453,
+      "step": 17538
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002862475138656598,
+      "loss": 0.089,
+      "step": 17539
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028614917176403363,
+      "loss": 0.0941,
+      "step": 17540
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028605084373774323,
+      "loss": 0.0853,
+      "step": 17541
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028595252978872734,
+      "loss": 0.1343,
+      "step": 17542
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028585422991892487,
+      "loss": 0.1056,
+      "step": 17543
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028575594413027375,
+      "loss": 0.097,
+      "step": 17544
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000285657672424712,
+      "loss": 0.0809,
+      "step": 17545
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002855594148041769,
+      "loss": 0.1104,
+      "step": 17546
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028546117127060646,
+      "loss": 0.0817,
+      "step": 17547
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002853629418259376,
+      "loss": 0.1077,
+      "step": 17548
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002852647264721068,
+      "loss": 0.042,
+      "step": 17549
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00028516652521105114,
+      "loss": 0.0807,
+      "step": 17550
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002850683380447068,
+      "loss": 0.1094,
+      "step": 17551
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002849701649750095,
+      "loss": 0.063,
+      "step": 17552
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028487200600389517,
+      "loss": 0.112,
+      "step": 17553
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028477386113329915,
+      "loss": 0.1214,
+      "step": 17554
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002846757303651565,
+      "loss": 0.1038,
+      "step": 17555
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028457761370140246,
+      "loss": 0.0751,
+      "step": 17556
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002844795111439714,
+      "loss": 0.1239,
+      "step": 17557
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002843814226947975,
+      "loss": 0.0574,
+      "step": 17558
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028428334835581536,
+      "loss": 0.0635,
+      "step": 17559
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028418528812895836,
+      "loss": 0.1361,
+      "step": 17560
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002840872420161601,
+      "loss": 0.0641,
+      "step": 17561
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002839892100193534,
+      "loss": 0.0813,
+      "step": 17562
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028389119214047197,
+      "loss": 0.0886,
+      "step": 17563
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000283793188381448,
+      "loss": 0.0714,
+      "step": 17564
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000283695198744214,
+      "loss": 0.1028,
+      "step": 17565
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000283597223230702,
+      "loss": 0.1169,
+      "step": 17566
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028349926184284357,
+      "loss": 0.0582,
+      "step": 17567
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002834013145825708,
+      "loss": 0.1329,
+      "step": 17568
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028330338145181447,
+      "loss": 0.0958,
+      "step": 17569
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002832054624525061,
+      "loss": 0.0846,
+      "step": 17570
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028310755758657614,
+      "loss": 0.1016,
+      "step": 17571
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000283009666855955,
+      "loss": 0.0592,
+      "step": 17572
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028291179026257283,
+      "loss": 0.0992,
+      "step": 17573
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028281392780835947,
+      "loss": 0.076,
+      "step": 17574
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028271607949524434,
+      "loss": 0.095,
+      "step": 17575
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002826182453251573,
+      "loss": 0.0677,
+      "step": 17576
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002825204253000271,
+      "loss": 0.0936,
+      "step": 17577
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002824226194217825,
+      "loss": 0.0884,
+      "step": 17578
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002823248276923517,
+      "loss": 0.1402,
+      "step": 17579
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028222705011366346,
+      "loss": 0.101,
+      "step": 17580
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028212928668764535,
+      "loss": 0.103,
+      "step": 17581
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002820315374162249,
+      "loss": 0.111,
+      "step": 17582
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028193380230133,
+      "loss": 0.0638,
+      "step": 17583
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028183608134488736,
+      "loss": 0.0931,
+      "step": 17584
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002817383745488239,
+      "loss": 0.0978,
+      "step": 17585
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002816406819150661,
+      "loss": 0.0839,
+      "step": 17586
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028154300344554006,
+      "loss": 0.0895,
+      "step": 17587
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002814453391421717,
+      "loss": 0.067,
+      "step": 17588
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028134768900688734,
+      "loss": 0.1229,
+      "step": 17589
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028125005304161155,
+      "loss": 0.0668,
+      "step": 17590
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002811524312482703,
+      "loss": 0.0804,
+      "step": 17591
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002810548236287879,
+      "loss": 0.0753,
+      "step": 17592
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002809572301850891,
+      "loss": 0.0835,
+      "step": 17593
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028085965091909814,
+      "loss": 0.0687,
+      "step": 17594
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00028076208583273875,
+      "loss": 0.0809,
+      "step": 17595
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002806645349279353,
+      "loss": 0.1456,
+      "step": 17596
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000280566998206611,
+      "loss": 0.0902,
+      "step": 17597
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002804694756706888,
+      "loss": 0.0646,
+      "step": 17598
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002803719673220916,
+      "loss": 0.0935,
+      "step": 17599
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00028027447316274234,
+      "loss": 0.0652,
+      "step": 17600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00028017699319456314,
+      "loss": 0.0651,
+      "step": 17601
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00028007952741947585,
+      "loss": 0.176,
+      "step": 17602
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002799820758394028,
+      "loss": 0.0741,
+      "step": 17603
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002798846384562652,
+      "loss": 0.0683,
+      "step": 17604
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027978721527198436,
+      "loss": 0.0873,
+      "step": 17605
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002796898062884809,
+      "loss": 0.0604,
+      "step": 17606
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002795924115076759,
+      "loss": 0.0984,
+      "step": 17607
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002794950309314894,
+      "loss": 0.0535,
+      "step": 17608
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027939766456184145,
+      "loss": 0.063,
+      "step": 17609
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027930031240065223,
+      "loss": 0.0742,
+      "step": 17610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002792029744498409,
+      "loss": 0.0978,
+      "step": 17611
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002791056507113271,
+      "loss": 0.0612,
+      "step": 17612
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027900834118702977,
+      "loss": 0.1116,
+      "step": 17613
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002789110458788674,
+      "loss": 0.1444,
+      "step": 17614
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002788137647887581,
+      "loss": 0.087,
+      "step": 17615
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027871649791862076,
+      "loss": 0.0544,
+      "step": 17616
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002786192452703729,
+      "loss": 0.0621,
+      "step": 17617
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000278522006845932,
+      "loss": 0.1195,
+      "step": 17618
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027842478264721536,
+      "loss": 0.0849,
+      "step": 17619
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027832757267613983,
+      "loss": 0.0756,
+      "step": 17620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002782303769346226,
+      "loss": 0.1053,
+      "step": 17621
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002781331954245796,
+      "loss": 0.0922,
+      "step": 17622
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002780360281479276,
+      "loss": 0.0643,
+      "step": 17623
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027793887510658223,
+      "loss": 0.1055,
+      "step": 17624
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000277841736302459,
+      "loss": 0.0883,
+      "step": 17625
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027774461173747336,
+      "loss": 0.0803,
+      "step": 17626
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027764750141354025,
+      "loss": 0.0647,
+      "step": 17627
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027755040533257447,
+      "loss": 0.0875,
+      "step": 17628
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002774533234964902,
+      "loss": 0.082,
+      "step": 17629
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027735625590720236,
+      "loss": 0.088,
+      "step": 17630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002772592025666245,
+      "loss": 0.0596,
+      "step": 17631
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027716216347667,
+      "loss": 0.0581,
+      "step": 17632
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027706513863925264,
+      "loss": 0.0895,
+      "step": 17633
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027696812805628546,
+      "loss": 0.0909,
+      "step": 17634
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002768711317296808,
+      "loss": 0.0734,
+      "step": 17635
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027677414966135197,
+      "loss": 0.077,
+      "step": 17636
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002766771818532107,
+      "loss": 0.0746,
+      "step": 17637
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000276580228307169,
+      "loss": 0.1094,
+      "step": 17638
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002764832890251385,
+      "loss": 0.0615,
+      "step": 17639
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002763863640090308,
+      "loss": 0.1302,
+      "step": 17640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002762894532607566,
+      "loss": 0.0827,
+      "step": 17641
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027619255678222734,
+      "loss": 0.101,
+      "step": 17642
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002760956745753529,
+      "loss": 0.0848,
+      "step": 17643
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00027599880664204424,
+      "loss": 0.0573,
+      "step": 17644
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002759019529842111,
+      "loss": 0.0773,
+      "step": 17645
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000275805113603763,
+      "loss": 0.0808,
+      "step": 17646
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027570828850260955,
+      "loss": 0.144,
+      "step": 17647
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002756114776826598,
+      "loss": 0.0602,
+      "step": 17648
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027551468114582245,
+      "loss": 0.0678,
+      "step": 17649
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002754178988940065,
+      "loss": 0.1272,
+      "step": 17650
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027532113092912006,
+      "loss": 0.1268,
+      "step": 17651
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027522437725307103,
+      "loss": 0.0651,
+      "step": 17652
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027512763786776697,
+      "loss": 0.0698,
+      "step": 17653
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027503091277511596,
+      "loss": 0.0845,
+      "step": 17654
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027493420197702447,
+      "loss": 0.1357,
+      "step": 17655
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002748375054754001,
+      "loss": 0.098,
+      "step": 17656
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002747408232721491,
+      "loss": 0.0934,
+      "step": 17657
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027464415536917775,
+      "loss": 0.0854,
+      "step": 17658
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002745475017683923,
+      "loss": 0.0761,
+      "step": 17659
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027445086247169815,
+      "loss": 0.0785,
+      "step": 17660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002743542374810011,
+      "loss": 0.0882,
+      "step": 17661
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000274257626798206,
+      "loss": 0.0826,
+      "step": 17662
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002741610304252182,
+      "loss": 0.0533,
+      "step": 17663
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027406444836394197,
+      "loss": 0.0795,
+      "step": 17664
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027396788061628206,
+      "loss": 0.0738,
+      "step": 17665
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027387132718414243,
+      "loss": 0.1211,
+      "step": 17666
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002737747880694267,
+      "loss": 0.1207,
+      "step": 17667
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027367826327403844,
+      "loss": 0.1026,
+      "step": 17668
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002735817527998806,
+      "loss": 0.0729,
+      "step": 17669
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002734852566488567,
+      "loss": 0.0842,
+      "step": 17670
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002733887748228692,
+      "loss": 0.1552,
+      "step": 17671
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027329230732382027,
+      "loss": 0.1436,
+      "step": 17672
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027319585415361215,
+      "loss": 0.0938,
+      "step": 17673
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002730994153141464,
+      "loss": 0.0621,
+      "step": 17674
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002730029908073247,
+      "loss": 0.0718,
+      "step": 17675
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002729065806350487,
+      "loss": 0.094,
+      "step": 17676
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002728101847992189,
+      "loss": 0.0762,
+      "step": 17677
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027271380330173633,
+      "loss": 0.0681,
+      "step": 17678
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027261743614450106,
+      "loss": 0.0622,
+      "step": 17679
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027252108332941337,
+      "loss": 0.0613,
+      "step": 17680
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.000272424744858373,
+      "loss": 0.0477,
+      "step": 17681
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027232842073327934,
+      "loss": 0.0909,
+      "step": 17682
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002722321109560321,
+      "loss": 0.0723,
+      "step": 17683
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002721358155285301,
+      "loss": 0.0878,
+      "step": 17684
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002720395344526716,
+      "loss": 0.0901,
+      "step": 17685
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002719432677303557,
+      "loss": 0.0824,
+      "step": 17686
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002718470153634802,
+      "loss": 0.0716,
+      "step": 17687
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027175077735394317,
+      "loss": 0.1436,
+      "step": 17688
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027165455370364156,
+      "loss": 0.0891,
+      "step": 17689
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00027155834441447337,
+      "loss": 0.1055,
+      "step": 17690
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002714621494883354,
+      "loss": 0.0756,
+      "step": 17691
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00027136596892712427,
+      "loss": 0.0803,
+      "step": 17692
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002712698027327365,
+      "loss": 0.0929,
+      "step": 17693
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002711736509070678,
+      "loss": 0.1024,
+      "step": 17694
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002710775134520148,
+      "loss": 0.1299,
+      "step": 17695
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002709813903694724,
+      "loss": 0.0885,
+      "step": 17696
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002708852816613366,
+      "loss": 0.1121,
+      "step": 17697
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00027078918732950197,
+      "loss": 0.0834,
+      "step": 17698
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002706931073758634,
+      "loss": 0.13,
+      "step": 17699
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00027059704180231527,
+      "loss": 0.1738,
+      "step": 17700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002705009906107516,
+      "loss": 0.1023,
+      "step": 17701
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002704049538030663,
+      "loss": 0.0966,
+      "step": 17702
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00027030893138115333,
+      "loss": 0.1169,
+      "step": 17703
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00027021292334690586,
+      "loss": 0.1019,
+      "step": 17704
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002701169297022167,
+      "loss": 0.0658,
+      "step": 17705
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002700209504489786,
+      "loss": 0.1033,
+      "step": 17706
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002699249855890844,
+      "loss": 0.0742,
+      "step": 17707
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000269829035124426,
+      "loss": 0.091,
+      "step": 17708
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026973309905689515,
+      "loss": 0.0941,
+      "step": 17709
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000269637177388384,
+      "loss": 0.0907,
+      "step": 17710
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002695412701207834,
+      "loss": 0.1125,
+      "step": 17711
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026944537725598485,
+      "loss": 0.066,
+      "step": 17712
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026934949879587857,
+      "loss": 0.2113,
+      "step": 17713
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002692536347423554,
+      "loss": 0.0593,
+      "step": 17714
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026915778509730516,
+      "loss": 0.1033,
+      "step": 17715
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002690619498626183,
+      "loss": 0.0997,
+      "step": 17716
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002689661290401839,
+      "loss": 0.1218,
+      "step": 17717
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026887032263189195,
+      "loss": 0.0593,
+      "step": 17718
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002687745306396311,
+      "loss": 0.1021,
+      "step": 17719
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002686787530652901,
+      "loss": 0.0623,
+      "step": 17720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002685829899107577,
+      "loss": 0.0908,
+      "step": 17721
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026848724117792147,
+      "loss": 0.1249,
+      "step": 17722
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002683915068686702,
+      "loss": 0.1264,
+      "step": 17723
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002682957869848911,
+      "loss": 0.0955,
+      "step": 17724
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026820008152847154,
+      "loss": 0.0946,
+      "step": 17725
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002681043905012985,
+      "loss": 0.1176,
+      "step": 17726
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002680087139052586,
+      "loss": 0.1389,
+      "step": 17727
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.000267913051742239,
+      "loss": 0.0683,
+      "step": 17728
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002678174040141251,
+      "loss": 0.1324,
+      "step": 17729
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002677217707228037,
+      "loss": 0.0805,
+      "step": 17730
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002676261518701598,
+      "loss": 0.0685,
+      "step": 17731
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002675305474580791,
+      "loss": 0.0822,
+      "step": 17732
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002674349574884466,
+      "loss": 0.1136,
+      "step": 17733
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002673393819631469,
+      "loss": 0.1443,
+      "step": 17734
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002672438208840644,
+      "loss": 0.0909,
+      "step": 17735
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026714827425308377,
+      "loss": 0.0899,
+      "step": 17736
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002670527420720888,
+      "loss": 0.086,
+      "step": 17737
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002669572243429628,
+      "loss": 0.0795,
+      "step": 17738
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026686172106758966,
+      "loss": 0.0869,
+      "step": 17739
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026676623224785234,
+      "loss": 0.1268,
+      "step": 17740
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002666707578856334,
+      "loss": 0.1079,
+      "step": 17741
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002665752979828153,
+      "loss": 0.0768,
+      "step": 17742
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002664798525412806,
+      "loss": 0.076,
+      "step": 17743
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002663844215629112,
+      "loss": 0.1151,
+      "step": 17744
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026628900504958863,
+      "loss": 0.0836,
+      "step": 17745
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026619360300319427,
+      "loss": 0.073,
+      "step": 17746
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002660982154256092,
+      "loss": 0.0582,
+      "step": 17747
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026600284231871396,
+      "loss": 0.1373,
+      "step": 17748
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002659074836843893,
+      "loss": 0.0898,
+      "step": 17749
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002658121395245158,
+      "loss": 0.0986,
+      "step": 17750
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026571680984097313,
+      "loss": 0.0862,
+      "step": 17751
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002656214946356409,
+      "loss": 0.084,
+      "step": 17752
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026552619391039857,
+      "loss": 0.1212,
+      "step": 17753
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002654309076671252,
+      "loss": 0.0656,
+      "step": 17754
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026533563590769917,
+      "loss": 0.0951,
+      "step": 17755
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002652403786339996,
+      "loss": 0.1217,
+      "step": 17756
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002651451358479047,
+      "loss": 0.1393,
+      "step": 17757
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002650499075512921,
+      "loss": 0.0596,
+      "step": 17758
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002649546937460394,
+      "loss": 0.0817,
+      "step": 17759
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002648594944340245,
+      "loss": 0.0947,
+      "step": 17760
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026476430961712395,
+      "loss": 0.0758,
+      "step": 17761
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002646691392972146,
+      "loss": 0.062,
+      "step": 17762
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002645739834761734,
+      "loss": 0.1095,
+      "step": 17763
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026447884215587635,
+      "loss": 0.0758,
+      "step": 17764
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002643837153381994,
+      "loss": 0.0875,
+      "step": 17765
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002642886030250181,
+      "loss": 0.0949,
+      "step": 17766
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026419350521820795,
+      "loss": 0.1285,
+      "step": 17767
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026409842191964396,
+      "loss": 0.1042,
+      "step": 17768
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002640033531312006,
+      "loss": 0.0936,
+      "step": 17769
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002639082988547528,
+      "loss": 0.0804,
+      "step": 17770
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.000263813259092175,
+      "loss": 0.0935,
+      "step": 17771
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002637182338453409,
+      "loss": 0.0948,
+      "step": 17772
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002636232231161241,
+      "loss": 0.0568,
+      "step": 17773
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.000263528226906398,
+      "loss": 0.0693,
+      "step": 17774
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002634332452180355,
+      "loss": 0.0843,
+      "step": 17775
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026333827805290975,
+      "loss": 0.0657,
+      "step": 17776
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002632433254128932,
+      "loss": 0.1022,
+      "step": 17777
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026314838729985803,
+      "loss": 0.0652,
+      "step": 17778
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026305346371567606,
+      "loss": 0.0955,
+      "step": 17779
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026295855466221865,
+      "loss": 0.0619,
+      "step": 17780
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00026286366014135787,
+      "loss": 0.1085,
+      "step": 17781
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002627687801549642,
+      "loss": 0.1295,
+      "step": 17782
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.000262673914704909,
+      "loss": 0.0792,
+      "step": 17783
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002625790637930625,
+      "loss": 0.0772,
+      "step": 17784
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026248422742129485,
+      "loss": 0.0742,
+      "step": 17785
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026238940559147596,
+      "loss": 0.0957,
+      "step": 17786
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026229459830547563,
+      "loss": 0.1256,
+      "step": 17787
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002621998055651631,
+      "loss": 0.0662,
+      "step": 17788
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002621050273724072,
+      "loss": 0.0847,
+      "step": 17789
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002620102637290772,
+      "loss": 0.0776,
+      "step": 17790
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002619155146370412,
+      "loss": 0.0981,
+      "step": 17791
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002618207800981679,
+      "loss": 0.1194,
+      "step": 17792
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026172606011432497,
+      "loss": 0.1031,
+      "step": 17793
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026163135468737985,
+      "loss": 0.0753,
+      "step": 17794
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026153666381919995,
+      "loss": 0.0919,
+      "step": 17795
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002614419875116526,
+      "loss": 0.0869,
+      "step": 17796
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002613473257666045,
+      "loss": 0.1285,
+      "step": 17797
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026125267858592206,
+      "loss": 0.0819,
+      "step": 17798
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002611580459714714,
+      "loss": 0.118,
+      "step": 17799
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002610634279251185,
+      "loss": 0.1437,
+      "step": 17800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026096882444872884,
+      "loss": 0.1111,
+      "step": 17801
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002608742355441678,
+      "loss": 0.1168,
+      "step": 17802
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002607796612133009,
+      "loss": 0.0604,
+      "step": 17803
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00026068510145799264,
+      "loss": 0.1244,
+      "step": 17804
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002605905562801073,
+      "loss": 0.0885,
+      "step": 17805
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002604960256815093,
+      "loss": 0.1141,
+      "step": 17806
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002604015096640624,
+      "loss": 0.0633,
+      "step": 17807
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002603070082296303,
+      "loss": 0.0822,
+      "step": 17808
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002602125213800759,
+      "loss": 0.1177,
+      "step": 17809
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.000260118049117263,
+      "loss": 0.0873,
+      "step": 17810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.000260023591443054,
+      "loss": 0.0903,
+      "step": 17811
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025992914835931105,
+      "loss": 0.0697,
+      "step": 17812
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025983471986789687,
+      "loss": 0.1689,
+      "step": 17813
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002597403059706732,
+      "loss": 0.1111,
+      "step": 17814
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002596459066695013,
+      "loss": 0.1058,
+      "step": 17815
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025955152196624286,
+      "loss": 0.0714,
+      "step": 17816
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025945715186275896,
+      "loss": 0.1185,
+      "step": 17817
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025936279636091006,
+      "loss": 0.1139,
+      "step": 17818
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025926845546255683,
+      "loss": 0.0994,
+      "step": 17819
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002591741291695593,
+      "loss": 0.1024,
+      "step": 17820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002590798174837773,
+      "loss": 0.0712,
+      "step": 17821
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025898552040707024,
+      "loss": 0.1085,
+      "step": 17822
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025889123794129777,
+      "loss": 0.0764,
+      "step": 17823
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.000258796970088319,
+      "loss": 0.0717,
+      "step": 17824
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002587027168499925,
+      "loss": 0.0612,
+      "step": 17825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002586084782281766,
+      "loss": 0.0771,
+      "step": 17826
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025851425422472965,
+      "loss": 0.0957,
+      "step": 17827
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00025842004484150896,
+      "loss": 0.1141,
+      "step": 17828
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002583258500803728,
+      "loss": 0.0981,
+      "step": 17829
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002582316699431783,
+      "loss": 0.1279,
+      "step": 17830
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002581375044317822,
+      "loss": 0.1074,
+      "step": 17831
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025804335354804133,
+      "loss": 0.1155,
+      "step": 17832
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025794921729381184,
+      "loss": 0.0989,
+      "step": 17833
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025785509567095047,
+      "loss": 0.1017,
+      "step": 17834
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002577609886813124,
+      "loss": 0.0564,
+      "step": 17835
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002576668963267537,
+      "loss": 0.0789,
+      "step": 17836
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002575728186091295,
+      "loss": 0.0776,
+      "step": 17837
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025747875553029474,
+      "loss": 0.0928,
+      "step": 17838
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025738470709210404,
+      "loss": 0.1007,
+      "step": 17839
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025729067329641176,
+      "loss": 0.0864,
+      "step": 17840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002571966541450722,
+      "loss": 0.0852,
+      "step": 17841
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002571026496399387,
+      "loss": 0.1113,
+      "step": 17842
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025700865978286547,
+      "loss": 0.0609,
+      "step": 17843
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002569146845757051,
+      "loss": 0.1621,
+      "step": 17844
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002568207240203113,
+      "loss": 0.1546,
+      "step": 17845
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025672677811853617,
+      "loss": 0.0775,
+      "step": 17846
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002566328468722323,
+      "loss": 0.0694,
+      "step": 17847
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025653893028325137,
+      "loss": 0.0649,
+      "step": 17848
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025644502835344584,
+      "loss": 0.0725,
+      "step": 17849
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025635114108466683,
+      "loss": 0.1136,
+      "step": 17850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002562572684787657,
+      "loss": 0.0561,
+      "step": 17851
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002561634105375933,
+      "loss": 0.0867,
+      "step": 17852
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025606956726300024,
+      "loss": 0.0822,
+      "step": 17853
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002559757386568367,
+      "loss": 0.0513,
+      "step": 17854
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002558819247209532,
+      "loss": 0.0873,
+      "step": 17855
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002557881254571989,
+      "loss": 0.0732,
+      "step": 17856
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002556943408674239,
+      "loss": 0.1353,
+      "step": 17857
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025560057095347735,
+      "loss": 0.0858,
+      "step": 17858
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025550681571720777,
+      "loss": 0.1058,
+      "step": 17859
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002554130751604641,
+      "loss": 0.0811,
+      "step": 17860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002553193492850944,
+      "loss": 0.1224,
+      "step": 17861
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025522563809294663,
+      "loss": 0.0975,
+      "step": 17862
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002551319415858689,
+      "loss": 0.0781,
+      "step": 17863
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002550382597657087,
+      "loss": 0.109,
+      "step": 17864
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025494459263431256,
+      "loss": 0.0988,
+      "step": 17865
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025485094019352816,
+      "loss": 0.1052,
+      "step": 17866
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002547573024452018,
+      "loss": 0.0743,
+      "step": 17867
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002546636793911793,
+      "loss": 0.1014,
+      "step": 17868
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002545700710333074,
+      "loss": 0.1001,
+      "step": 17869
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002544764773734316,
+      "loss": 0.0909,
+      "step": 17870
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002543828984133971,
+      "loss": 0.0628,
+      "step": 17871
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025428933415504927,
+      "loss": 0.1232,
+      "step": 17872
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025419578460023296,
+      "loss": 0.1084,
+      "step": 17873
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025410224975079256,
+      "loss": 0.0797,
+      "step": 17874
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00025400872960857203,
+      "loss": 0.0682,
+      "step": 17875
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025391522417541614,
+      "loss": 0.0645,
+      "step": 17876
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002538217334531678,
+      "loss": 0.1669,
+      "step": 17877
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025372825744367125,
+      "loss": 0.0564,
+      "step": 17878
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025363479614876907,
+      "loss": 0.0932,
+      "step": 17879
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025354134957030405,
+      "loss": 0.0735,
+      "step": 17880
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.000253447917710119,
+      "loss": 0.1184,
+      "step": 17881
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002533545005700556,
+      "loss": 0.0794,
+      "step": 17882
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002532610981519564,
+      "loss": 0.1161,
+      "step": 17883
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025316771045766297,
+      "loss": 0.084,
+      "step": 17884
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025307433748901655,
+      "loss": 0.1301,
+      "step": 17885
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002529809792478579,
+      "loss": 0.0882,
+      "step": 17886
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025288763573602847,
+      "loss": 0.062,
+      "step": 17887
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002527943069553681,
+      "loss": 0.0581,
+      "step": 17888
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025270099290771754,
+      "loss": 0.1199,
+      "step": 17889
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025260769359491665,
+      "loss": 0.0573,
+      "step": 17890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025251440901880474,
+      "loss": 0.0557,
+      "step": 17891
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025242113918122136,
+      "loss": 0.0691,
+      "step": 17892
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025232788408400555,
+      "loss": 0.118,
+      "step": 17893
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025223464372899594,
+      "loss": 0.0818,
+      "step": 17894
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002521414181180308,
+      "loss": 0.0712,
+      "step": 17895
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025204820725294885,
+      "loss": 0.0925,
+      "step": 17896
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002519550111355874,
+      "loss": 0.1478,
+      "step": 17897
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025186182976778474,
+      "loss": 0.0555,
+      "step": 17898
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002517686631513776,
+      "loss": 0.0647,
+      "step": 17899
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002516755112882032,
+      "loss": 0.0902,
+      "step": 17900
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025158237418009823,
+      "loss": 0.0866,
+      "step": 17901
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002514892518288988,
+      "loss": 0.0743,
+      "step": 17902
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002513961442364417,
+      "loss": 0.1055,
+      "step": 17903
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025130305140456244,
+      "loss": 0.1763,
+      "step": 17904
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025120997333509656,
+      "loss": 0.1709,
+      "step": 17905
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025111691002987927,
+      "loss": 0.0787,
+      "step": 17906
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002510238614907453,
+      "loss": 0.1335,
+      "step": 17907
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025093082771953,
+      "loss": 0.1255,
+      "step": 17908
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.000250837808718067,
+      "loss": 0.0839,
+      "step": 17909
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025074480448819124,
+      "loss": 0.1403,
+      "step": 17910
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025065181503173596,
+      "loss": 0.0768,
+      "step": 17911
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002505588403505349,
+      "loss": 0.0712,
+      "step": 17912
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002504658804464212,
+      "loss": 0.0455,
+      "step": 17913
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025037293532122765,
+      "loss": 0.0752,
+      "step": 17914
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002502800049767868,
+      "loss": 0.0578,
+      "step": 17915
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025018708941493154,
+      "loss": 0.0655,
+      "step": 17916
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025009418863749356,
+      "loss": 0.1133,
+      "step": 17917
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00025000130264630436,
+      "loss": 0.088,
+      "step": 17918
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002499084314431962,
+      "loss": 0.0664,
+      "step": 17919
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00024981557502999976,
+      "loss": 0.0761,
+      "step": 17920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0002497227334085459,
+      "loss": 0.1351,
+      "step": 17921
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000249629906580665,
+      "loss": 0.0675,
+      "step": 17922
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024953709454818805,
+      "loss": 0.0916,
+      "step": 17923
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002494442973129447,
+      "loss": 0.0727,
+      "step": 17924
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002493515148767647,
+      "loss": 0.069,
+      "step": 17925
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024925874724147733,
+      "loss": 0.0777,
+      "step": 17926
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024916599440891196,
+      "loss": 0.1132,
+      "step": 17927
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000249073256380897,
+      "loss": 0.0676,
+      "step": 17928
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024898053315926163,
+      "loss": 0.1211,
+      "step": 17929
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002488878247458336,
+      "loss": 0.1506,
+      "step": 17930
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002487951311424415,
+      "loss": 0.0644,
+      "step": 17931
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002487024523509125,
+      "loss": 0.0444,
+      "step": 17932
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024860978837307413,
+      "loss": 0.0689,
+      "step": 17933
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024851713921075357,
+      "loss": 0.0924,
+      "step": 17934
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002484245048657773,
+      "loss": 0.0784,
+      "step": 17935
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002483318853399722,
+      "loss": 0.1042,
+      "step": 17936
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024823928063516455,
+      "loss": 0.1321,
+      "step": 17937
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024814669075317995,
+      "loss": 0.1079,
+      "step": 17938
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000248054115695844,
+      "loss": 0.0632,
+      "step": 17939
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024796155546498254,
+      "loss": 0.0828,
+      "step": 17940
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024786901006242036,
+      "loss": 0.1031,
+      "step": 17941
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024777647948998185,
+      "loss": 0.1095,
+      "step": 17942
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002476839637494922,
+      "loss": 0.0863,
+      "step": 17943
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024759146284277514,
+      "loss": 0.0727,
+      "step": 17944
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002474989767716546,
+      "loss": 0.0724,
+      "step": 17945
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002474065055379542,
+      "loss": 0.0711,
+      "step": 17946
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002473140491434972,
+      "loss": 0.0752,
+      "step": 17947
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024722160759010637,
+      "loss": 0.0929,
+      "step": 17948
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024712918087960503,
+      "loss": 0.0815,
+      "step": 17949
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002470367690138151,
+      "loss": 0.094,
+      "step": 17950
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024694437199455877,
+      "loss": 0.0576,
+      "step": 17951
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002468519898236582,
+      "loss": 0.1503,
+      "step": 17952
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002467596225029347,
+      "loss": 0.1111,
+      "step": 17953
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002466672700342095,
+      "loss": 0.1427,
+      "step": 17954
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002465749324193034,
+      "loss": 0.0917,
+      "step": 17955
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002464826096600374,
+      "loss": 0.0781,
+      "step": 17956
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002463903017582318,
+      "loss": 0.2356,
+      "step": 17957
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024629800871570675,
+      "loss": 0.0589,
+      "step": 17958
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024620573053428177,
+      "loss": 0.0703,
+      "step": 17959
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002461134672157761,
+      "loss": 0.0495,
+      "step": 17960
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024602121876200967,
+      "loss": 0.0616,
+      "step": 17961
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024592898517480075,
+      "loss": 0.1053,
+      "step": 17962
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024583676645596855,
+      "loss": 0.0765,
+      "step": 17963
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000245744562607331,
+      "loss": 0.0979,
+      "step": 17964
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024565237363070625,
+      "loss": 0.1002,
+      "step": 17965
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024556019952791196,
+      "loss": 0.12,
+      "step": 17966
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00024546804030076564,
+      "loss": 0.0742,
+      "step": 17967
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000245375895951084,
+      "loss": 0.1001,
+      "step": 17968
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024528376648068464,
+      "loss": 0.0626,
+      "step": 17969
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002451916518913837,
+      "loss": 0.0914,
+      "step": 17970
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002450995521849977,
+      "loss": 0.1977,
+      "step": 17971
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002450074673633419,
+      "loss": 0.0817,
+      "step": 17972
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024491539742823296,
+      "loss": 0.1205,
+      "step": 17973
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024482334238148576,
+      "loss": 0.1002,
+      "step": 17974
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002447313022249151,
+      "loss": 0.1204,
+      "step": 17975
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002446392769603365,
+      "loss": 0.0861,
+      "step": 17976
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002445472665895641,
+      "loss": 0.1099,
+      "step": 17977
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024445527111441204,
+      "loss": 0.1299,
+      "step": 17978
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024436329053669436,
+      "loss": 0.0723,
+      "step": 17979
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002442713248582246,
+      "loss": 0.0995,
+      "step": 17980
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002441793740808158,
+      "loss": 0.0958,
+      "step": 17981
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024408743820628165,
+      "loss": 0.1123,
+      "step": 17982
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002439955172364342,
+      "loss": 0.0792,
+      "step": 17983
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024390361117308657,
+      "loss": 0.1235,
+      "step": 17984
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024381172001805063,
+      "loss": 0.0892,
+      "step": 17985
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024371984377313826,
+      "loss": 0.0576,
+      "step": 17986
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024362798244016094,
+      "loss": 0.0889,
+      "step": 17987
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002435361360209296,
+      "loss": 0.0813,
+      "step": 17988
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024344430451725608,
+      "loss": 0.119,
+      "step": 17989
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024335248793095043,
+      "loss": 0.0825,
+      "step": 17990
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024326068626382335,
+      "loss": 0.074,
+      "step": 17991
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002431688995176844,
+      "loss": 0.1162,
+      "step": 17992
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002430771276943441,
+      "loss": 0.072,
+      "step": 17993
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024298537079561156,
+      "loss": 0.08,
+      "step": 17994
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002428936288232958,
+      "loss": 0.0875,
+      "step": 17995
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024280190177920636,
+      "loss": 0.0464,
+      "step": 17996
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024271018966515146,
+      "loss": 0.0482,
+      "step": 17997
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024261849248293955,
+      "loss": 0.094,
+      "step": 17998
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002425268102343785,
+      "loss": 0.0695,
+      "step": 17999
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024243514292127623,
+      "loss": 0.071,
+      "step": 18000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024234349054544002,
+      "loss": 0.0815,
+      "step": 18001
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024225185310867693,
+      "loss": 0.0694,
+      "step": 18002
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024216023061279424,
+      "loss": 0.0591,
+      "step": 18003
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.000242068623059598,
+      "loss": 0.0908,
+      "step": 18004
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024197703045089492,
+      "loss": 0.0603,
+      "step": 18005
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024188545278849095,
+      "loss": 0.0951,
+      "step": 18006
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024179389007419162,
+      "loss": 0.077,
+      "step": 18007
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024170234230980203,
+      "loss": 0.1527,
+      "step": 18008
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024161080949712787,
+      "loss": 0.0518,
+      "step": 18009
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024151929163797358,
+      "loss": 0.0994,
+      "step": 18010
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024142778873414373,
+      "loss": 0.1144,
+      "step": 18011
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00024133630078744251,
+      "loss": 0.0773,
+      "step": 18012
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002412448277996736,
+      "loss": 0.099,
+      "step": 18013
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024115336977264103,
+      "loss": 0.1262,
+      "step": 18014
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024106192670814776,
+      "loss": 0.0602,
+      "step": 18015
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024097049860799725,
+      "loss": 0.1351,
+      "step": 18016
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024087908547399195,
+      "loss": 0.1061,
+      "step": 18017
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024078768730793432,
+      "loss": 0.0566,
+      "step": 18018
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024069630411162647,
+      "loss": 0.1095,
+      "step": 18019
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002406049358868704,
+      "loss": 0.0699,
+      "step": 18020
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002405135826354674,
+      "loss": 0.0633,
+      "step": 18021
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024042224435921866,
+      "loss": 0.0745,
+      "step": 18022
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002403309210599257,
+      "loss": 0.0963,
+      "step": 18023
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024023961273938877,
+      "loss": 0.0728,
+      "step": 18024
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024014831939940794,
+      "loss": 0.0728,
+      "step": 18025
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00024005704104178405,
+      "loss": 0.0966,
+      "step": 18026
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002399657776683164,
+      "loss": 0.0942,
+      "step": 18027
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023987452928080422,
+      "loss": 0.0836,
+      "step": 18028
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023978329588104742,
+      "loss": 0.0605,
+      "step": 18029
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023969207747084442,
+      "loss": 0.057,
+      "step": 18030
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023960087405199395,
+      "loss": 0.0985,
+      "step": 18031
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023950968562629416,
+      "loss": 0.079,
+      "step": 18032
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023941851219554323,
+      "loss": 0.043,
+      "step": 18033
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023932735376153846,
+      "loss": 0.1296,
+      "step": 18034
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023923621032607802,
+      "loss": 0.0981,
+      "step": 18035
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002391450818909582,
+      "loss": 0.108,
+      "step": 18036
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023905396845797644,
+      "loss": 0.0679,
+      "step": 18037
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023896287002892914,
+      "loss": 0.0775,
+      "step": 18038
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023887178660561247,
+      "loss": 0.0913,
+      "step": 18039
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023878071818982227,
+      "loss": 0.1333,
+      "step": 18040
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002386896647833543,
+      "loss": 0.0905,
+      "step": 18041
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023859862638800355,
+      "loss": 0.0784,
+      "step": 18042
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023850760300556562,
+      "loss": 0.0757,
+      "step": 18043
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023841659463783504,
+      "loss": 0.0876,
+      "step": 18044
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023832560128660629,
+      "loss": 0.0776,
+      "step": 18045
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023823462295367303,
+      "loss": 0.071,
+      "step": 18046
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023814365964082996,
+      "loss": 0.0737,
+      "step": 18047
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023805271134987006,
+      "loss": 0.0751,
+      "step": 18048
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002379617780825869,
+      "loss": 0.0786,
+      "step": 18049
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002378708598407735,
+      "loss": 0.1036,
+      "step": 18050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002377799566262223,
+      "loss": 0.0981,
+      "step": 18051
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002376890684407258,
+      "loss": 0.0637,
+      "step": 18052
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023759819528607607,
+      "loss": 0.1099,
+      "step": 18053
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023750733716406482,
+      "loss": 0.0532,
+      "step": 18054
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023741649407648337,
+      "loss": 0.0994,
+      "step": 18055
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0002373256660251234,
+      "loss": 0.1307,
+      "step": 18056
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023723485301177527,
+      "loss": 0.0737,
+      "step": 18057
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023714405503823022,
+      "loss": 0.1411,
+      "step": 18058
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00023705327210627814,
+      "loss": 0.1083,
+      "step": 18059
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023696250421770916,
+      "loss": 0.1316,
+      "step": 18060
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023687175137431294,
+      "loss": 0.0922,
+      "step": 18061
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023678101357787863,
+      "loss": 0.1543,
+      "step": 18062
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023669029083019588,
+      "loss": 0.0852,
+      "step": 18063
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002365995831330533,
+      "loss": 0.0682,
+      "step": 18064
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023650889048823943,
+      "loss": 0.1185,
+      "step": 18065
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023641821289754227,
+      "loss": 0.0652,
+      "step": 18066
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023632755036274977,
+      "loss": 0.092,
+      "step": 18067
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002362369028856497,
+      "loss": 0.1108,
+      "step": 18068
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002361462704680297,
+      "loss": 0.1587,
+      "step": 18069
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023605565311167655,
+      "loss": 0.067,
+      "step": 18070
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023596505081837694,
+      "loss": 0.0934,
+      "step": 18071
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002358744635899174,
+      "loss": 0.0917,
+      "step": 18072
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023578389142808398,
+      "loss": 0.067,
+      "step": 18073
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023569333433466266,
+      "loss": 0.0674,
+      "step": 18074
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023560279231143866,
+      "loss": 0.094,
+      "step": 18075
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023551226536019777,
+      "loss": 0.0748,
+      "step": 18076
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023542175348272465,
+      "loss": 0.0801,
+      "step": 18077
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023533125668080378,
+      "loss": 0.0661,
+      "step": 18078
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023524077495622,
+      "loss": 0.0703,
+      "step": 18079
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023515030831075724,
+      "loss": 0.0771,
+      "step": 18080
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.000235059856746199,
+      "loss": 0.0948,
+      "step": 18081
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023496942026432878,
+      "loss": 0.0881,
+      "step": 18082
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002348789988669302,
+      "loss": 0.153,
+      "step": 18083
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023478859255578579,
+      "loss": 0.111,
+      "step": 18084
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023469820133267816,
+      "loss": 0.071,
+      "step": 18085
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023460782519938974,
+      "loss": 0.1019,
+      "step": 18086
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023451746415770238,
+      "loss": 0.0949,
+      "step": 18087
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023442711820939743,
+      "loss": 0.0719,
+      "step": 18088
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023433678735625675,
+      "loss": 0.101,
+      "step": 18089
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023424647160006162,
+      "loss": 0.0763,
+      "step": 18090
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023415617094259256,
+      "loss": 0.0596,
+      "step": 18091
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023406588538563,
+      "loss": 0.0986,
+      "step": 18092
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002339756149309543,
+      "loss": 0.0657,
+      "step": 18093
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002338853595803453,
+      "loss": 0.0867,
+      "step": 18094
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023379511933558218,
+      "loss": 0.071,
+      "step": 18095
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.000233704894198445,
+      "loss": 0.0679,
+      "step": 18096
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023361468417071253,
+      "loss": 0.0582,
+      "step": 18097
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023352448925416326,
+      "loss": 0.0948,
+      "step": 18098
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023343430945057554,
+      "loss": 0.0645,
+      "step": 18099
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023334414476172783,
+      "loss": 0.1158,
+      "step": 18100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002332539951893976,
+      "loss": 0.0494,
+      "step": 18101
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023316386073536288,
+      "loss": 0.0845,
+      "step": 18102
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002330737414014006,
+      "loss": 0.1229,
+      "step": 18103
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0002329836371892876,
+      "loss": 0.0786,
+      "step": 18104
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023289354810080055,
+      "loss": 0.0787,
+      "step": 18105
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00023280347413771587,
+      "loss": 0.0967,
+      "step": 18106
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023271341530180955,
+      "loss": 0.1079,
+      "step": 18107
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023262337159485702,
+      "loss": 0.0663,
+      "step": 18108
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023253334301863416,
+      "loss": 0.0657,
+      "step": 18109
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023244332957491576,
+      "loss": 0.1013,
+      "step": 18110
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002323533312654772,
+      "loss": 0.08,
+      "step": 18111
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002322633480920925,
+      "loss": 0.0858,
+      "step": 18112
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002321733800565361,
+      "loss": 0.1191,
+      "step": 18113
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023208342716058195,
+      "loss": 0.1389,
+      "step": 18114
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002319934894060034,
+      "loss": 0.0984,
+      "step": 18115
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023190356679457424,
+      "loss": 0.1219,
+      "step": 18116
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002318136593280673,
+      "loss": 0.076,
+      "step": 18117
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023172376700825536,
+      "loss": 0.0915,
+      "step": 18118
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002316338898369109,
+      "loss": 0.0775,
+      "step": 18119
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023154402781580573,
+      "loss": 0.1047,
+      "step": 18120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000231454180946712,
+      "loss": 0.0709,
+      "step": 18121
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002313643492314016,
+      "loss": 0.103,
+      "step": 18122
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023127453267164543,
+      "loss": 0.0626,
+      "step": 18123
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023118473126921435,
+      "loss": 0.059,
+      "step": 18124
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023109494502587924,
+      "loss": 0.097,
+      "step": 18125
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023100517394341035,
+      "loss": 0.0676,
+      "step": 18126
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023091541802357763,
+      "loss": 0.0925,
+      "step": 18127
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023082567726815073,
+      "loss": 0.0742,
+      "step": 18128
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023073595167889973,
+      "loss": 0.0944,
+      "step": 18129
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023064624125759326,
+      "loss": 0.0738,
+      "step": 18130
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023055654600600008,
+      "loss": 0.0952,
+      "step": 18131
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023046686592588918,
+      "loss": 0.1156,
+      "step": 18132
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023037720101902871,
+      "loss": 0.1111,
+      "step": 18133
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023028755128718647,
+      "loss": 0.0813,
+      "step": 18134
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00023019791673213009,
+      "loss": 0.0899,
+      "step": 18135
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002301082973556272,
+      "loss": 0.0659,
+      "step": 18136
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002300186931594448,
+      "loss": 0.0654,
+      "step": 18137
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022992910414534963,
+      "loss": 0.0681,
+      "step": 18138
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022983953031510795,
+      "loss": 0.092,
+      "step": 18139
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022974997167048627,
+      "loss": 0.0823,
+      "step": 18140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022966042821324995,
+      "loss": 0.0929,
+      "step": 18141
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002295708999451649,
+      "loss": 0.142,
+      "step": 18142
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002294813868679966,
+      "loss": 0.0551,
+      "step": 18143
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022939188898350992,
+      "loss": 0.0872,
+      "step": 18144
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022930240629346932,
+      "loss": 0.0775,
+      "step": 18145
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022921293879963933,
+      "loss": 0.0969,
+      "step": 18146
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000229123486503784,
+      "loss": 0.0884,
+      "step": 18147
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0002290340494076667,
+      "loss": 0.0801,
+      "step": 18148
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022894462751305166,
+      "loss": 0.0958,
+      "step": 18149
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022885522082170164,
+      "loss": 0.0665,
+      "step": 18150
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022876582933537947,
+      "loss": 0.0898,
+      "step": 18151
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00022867645305584762,
+      "loss": 0.1099,
+      "step": 18152
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022858709198486882,
+      "loss": 0.0874,
+      "step": 18153
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022849774612420472,
+      "loss": 0.1104,
+      "step": 18154
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022840841547561676,
+      "loss": 0.1046,
+      "step": 18155
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022831910004086686,
+      "loss": 0.0825,
+      "step": 18156
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022822979982171587,
+      "loss": 0.0764,
+      "step": 18157
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022814051481992448,
+      "loss": 0.0917,
+      "step": 18158
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022805124503725327,
+      "loss": 0.0807,
+      "step": 18159
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002279619904754624,
+      "loss": 0.1306,
+      "step": 18160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022787275113631133,
+      "loss": 0.0739,
+      "step": 18161
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022778352702156036,
+      "loss": 0.0922,
+      "step": 18162
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022769431813296804,
+      "loss": 0.0847,
+      "step": 18163
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002276051244722941,
+      "loss": 0.0667,
+      "step": 18164
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022751594604129667,
+      "loss": 0.0699,
+      "step": 18165
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022742678284173447,
+      "loss": 0.0609,
+      "step": 18166
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022733763487536518,
+      "loss": 0.0969,
+      "step": 18167
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022724850214394667,
+      "loss": 0.1415,
+      "step": 18168
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002271593846492367,
+      "loss": 0.0735,
+      "step": 18169
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022707028239299232,
+      "loss": 0.0921,
+      "step": 18170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022698119537697037,
+      "loss": 0.0615,
+      "step": 18171
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022689212360292731,
+      "loss": 0.0853,
+      "step": 18172
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022680306707261922,
+      "loss": 0.0734,
+      "step": 18173
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022671402578780265,
+      "loss": 0.0703,
+      "step": 18174
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022662499975023277,
+      "loss": 0.0848,
+      "step": 18175
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022653598896166529,
+      "loss": 0.0734,
+      "step": 18176
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022644699342385522,
+      "loss": 0.0929,
+      "step": 18177
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002263580131385572,
+      "loss": 0.0895,
+      "step": 18178
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002262690481075257,
+      "loss": 0.1013,
+      "step": 18179
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022618009833251496,
+      "loss": 0.0852,
+      "step": 18180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002260911638152785,
+      "loss": 0.0506,
+      "step": 18181
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022600224455757056,
+      "loss": 0.0711,
+      "step": 18182
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022591334056114388,
+      "loss": 0.0855,
+      "step": 18183
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022582445182775147,
+      "loss": 0.0764,
+      "step": 18184
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022573557835914627,
+      "loss": 0.1089,
+      "step": 18185
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022564672015708043,
+      "loss": 0.1267,
+      "step": 18186
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022555787722330613,
+      "loss": 0.1263,
+      "step": 18187
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002254690495595747,
+      "loss": 0.1248,
+      "step": 18188
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022538023716763822,
+      "loss": 0.0852,
+      "step": 18189
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022529144004924763,
+      "loss": 0.1109,
+      "step": 18190
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002252026582061537,
+      "loss": 0.1146,
+      "step": 18191
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022511389164010697,
+      "loss": 0.1057,
+      "step": 18192
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022502514035285782,
+      "loss": 0.1198,
+      "step": 18193
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002249364043461557,
+      "loss": 0.0878,
+      "step": 18194
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002248476836217511,
+      "loss": 0.0668,
+      "step": 18195
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022475897818139258,
+      "loss": 0.0847,
+      "step": 18196
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00022467028802682977,
+      "loss": 0.0953,
+      "step": 18197
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002245816131598113,
+      "loss": 0.0859,
+      "step": 18198
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022449295358208544,
+      "loss": 0.0603,
+      "step": 18199
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022440430929540034,
+      "loss": 0.0613,
+      "step": 18200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022431568030150374,
+      "loss": 0.1055,
+      "step": 18201
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022422706660214353,
+      "loss": 0.0745,
+      "step": 18202
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002241384681990668,
+      "loss": 0.1653,
+      "step": 18203
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022404988509402046,
+      "loss": 0.0613,
+      "step": 18204
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022396131728875092,
+      "loss": 0.0706,
+      "step": 18205
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022387276478500484,
+      "loss": 0.0912,
+      "step": 18206
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022378422758452822,
+      "loss": 0.1407,
+      "step": 18207
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002236957056890665,
+      "loss": 0.1042,
+      "step": 18208
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022360719910036552,
+      "loss": 0.0916,
+      "step": 18209
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022351870782017015,
+      "loss": 0.0827,
+      "step": 18210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022343023185022537,
+      "loss": 0.0937,
+      "step": 18211
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002233417711922755,
+      "loss": 0.0726,
+      "step": 18212
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002232533258480649,
+      "loss": 0.0852,
+      "step": 18213
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022316489581933753,
+      "loss": 0.0659,
+      "step": 18214
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022307648110783652,
+      "loss": 0.0713,
+      "step": 18215
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022298808171530572,
+      "loss": 0.0634,
+      "step": 18216
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022289969764348815,
+      "loss": 0.0542,
+      "step": 18217
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022281132889412648,
+      "loss": 0.1151,
+      "step": 18218
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022272297546896302,
+      "loss": 0.076,
+      "step": 18219
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022263463736973976,
+      "loss": 0.0555,
+      "step": 18220
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022254631459819852,
+      "loss": 0.069,
+      "step": 18221
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002224580071560811,
+      "loss": 0.1147,
+      "step": 18222
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022236971504512848,
+      "loss": 0.0911,
+      "step": 18223
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022228143826708157,
+      "loss": 0.0639,
+      "step": 18224
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.000222193176823681,
+      "loss": 0.0771,
+      "step": 18225
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002221049307166668,
+      "loss": 0.0783,
+      "step": 18226
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022201669994777952,
+      "loss": 0.0632,
+      "step": 18227
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022192848451875813,
+      "loss": 0.06,
+      "step": 18228
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022184028443134274,
+      "loss": 0.0889,
+      "step": 18229
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022175209968727216,
+      "loss": 0.1007,
+      "step": 18230
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022166393028828503,
+      "loss": 0.0646,
+      "step": 18231
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022157577623612,
+      "loss": 0.0542,
+      "step": 18232
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022148763753251512,
+      "loss": 0.0977,
+      "step": 18233
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022139951417920835,
+      "loss": 0.1144,
+      "step": 18234
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022131140617793677,
+      "loss": 0.0869,
+      "step": 18235
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022122331353043856,
+      "loss": 0.0646,
+      "step": 18236
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022113523623844988,
+      "loss": 0.1364,
+      "step": 18237
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0002210471743037079,
+      "loss": 0.0754,
+      "step": 18238
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022095912772794902,
+      "loss": 0.0844,
+      "step": 18239
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022087109651290892,
+      "loss": 0.0526,
+      "step": 18240
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022078308066032326,
+      "loss": 0.1295,
+      "step": 18241
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022069508017192807,
+      "loss": 0.0742,
+      "step": 18242
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022060709504945807,
+      "loss": 0.0784,
+      "step": 18243
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00022051912529464835,
+      "loss": 0.0832,
+      "step": 18244
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00022043117090923326,
+      "loss": 0.0548,
+      "step": 18245
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002203432318949471,
+      "loss": 0.0823,
+      "step": 18246
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00022025530825352347,
+      "loss": 0.0636,
+      "step": 18247
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00022016739998669655,
+      "loss": 0.0771,
+      "step": 18248
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002200795070961993,
+      "loss": 0.1059,
+      "step": 18249
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021999162958376506,
+      "loss": 0.0673,
+      "step": 18250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002199037674511265,
+      "loss": 0.08,
+      "step": 18251
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021981592070001576,
+      "loss": 0.0697,
+      "step": 18252
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021972808933216526,
+      "loss": 0.114,
+      "step": 18253
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021964027334930657,
+      "loss": 0.0952,
+      "step": 18254
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021955247275317114,
+      "loss": 0.0789,
+      "step": 18255
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021946468754549054,
+      "loss": 0.0861,
+      "step": 18256
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002193769177279955,
+      "loss": 0.1202,
+      "step": 18257
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021928916330241622,
+      "loss": 0.0672,
+      "step": 18258
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021920142427048374,
+      "loss": 0.0856,
+      "step": 18259
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021911370063392767,
+      "loss": 0.0903,
+      "step": 18260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002190259923944774,
+      "loss": 0.1042,
+      "step": 18261
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002189382995538629,
+      "loss": 0.0803,
+      "step": 18262
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021885062211381303,
+      "loss": 0.0656,
+      "step": 18263
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021876296007605645,
+      "loss": 0.0623,
+      "step": 18264
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021867531344232162,
+      "loss": 0.1056,
+      "step": 18265
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002185876822143369,
+      "loss": 0.0932,
+      "step": 18266
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021850006639383002,
+      "loss": 0.0629,
+      "step": 18267
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021841246598252828,
+      "loss": 0.0877,
+      "step": 18268
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021832488098215942,
+      "loss": 0.0663,
+      "step": 18269
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021823731139445002,
+      "loss": 0.1431,
+      "step": 18270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021814975722112717,
+      "loss": 0.0896,
+      "step": 18271
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021806221846391695,
+      "loss": 0.0945,
+      "step": 18272
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021797469512454536,
+      "loss": 0.0961,
+      "step": 18273
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021788718720473823,
+      "loss": 0.1222,
+      "step": 18274
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021779969470622074,
+      "loss": 0.0773,
+      "step": 18275
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021771221763071836,
+      "loss": 0.0845,
+      "step": 18276
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002176247559799559,
+      "loss": 0.0747,
+      "step": 18277
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021753730975565777,
+      "loss": 0.0906,
+      "step": 18278
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021744987895954792,
+      "loss": 0.0821,
+      "step": 18279
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021736246359335077,
+      "loss": 0.0611,
+      "step": 18280
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021727506365878946,
+      "loss": 0.1052,
+      "step": 18281
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021718767915758776,
+      "loss": 0.1246,
+      "step": 18282
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021710031009146847,
+      "loss": 0.0684,
+      "step": 18283
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021701295646215423,
+      "loss": 0.0648,
+      "step": 18284
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002169256182713676,
+      "loss": 0.1111,
+      "step": 18285
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021683829552083034,
+      "loss": 0.08,
+      "step": 18286
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021675098821226458,
+      "loss": 0.0895,
+      "step": 18287
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021666369634739137,
+      "loss": 0.1407,
+      "step": 18288
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0002165764199279323,
+      "loss": 0.1053,
+      "step": 18289
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00021648915895560827,
+      "loss": 0.0798,
+      "step": 18290
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021640191343213932,
+      "loss": 0.0944,
+      "step": 18291
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002163146833592463,
+      "loss": 0.0911,
+      "step": 18292
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021622746873864907,
+      "loss": 0.1345,
+      "step": 18293
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.000216140269572067,
+      "loss": 0.1115,
+      "step": 18294
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002160530858612194,
+      "loss": 0.0836,
+      "step": 18295
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002159659176078258,
+      "loss": 0.0748,
+      "step": 18296
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002158787648136046,
+      "loss": 0.0827,
+      "step": 18297
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021579162748027426,
+      "loss": 0.1324,
+      "step": 18298
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021570450560955292,
+      "loss": 0.1118,
+      "step": 18299
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021561739920315814,
+      "loss": 0.1018,
+      "step": 18300
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021553030826280796,
+      "loss": 0.1022,
+      "step": 18301
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021544323279021904,
+      "loss": 0.0978,
+      "step": 18302
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021535617278710894,
+      "loss": 0.0948,
+      "step": 18303
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021526912825519385,
+      "loss": 0.1752,
+      "step": 18304
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021518209919619015,
+      "loss": 0.094,
+      "step": 18305
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021509508561181378,
+      "loss": 0.0999,
+      "step": 18306
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021500808750378055,
+      "loss": 0.0885,
+      "step": 18307
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021492110487380533,
+      "loss": 0.1249,
+      "step": 18308
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021483413772360393,
+      "loss": 0.085,
+      "step": 18309
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021474718605489086,
+      "loss": 0.157,
+      "step": 18310
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021466024986938027,
+      "loss": 0.0801,
+      "step": 18311
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002145733291687868,
+      "loss": 0.0939,
+      "step": 18312
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021448642395482422,
+      "loss": 0.0917,
+      "step": 18313
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021439953422920576,
+      "loss": 0.1375,
+      "step": 18314
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002143126599936448,
+      "loss": 0.1347,
+      "step": 18315
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021422580124985446,
+      "loss": 0.1026,
+      "step": 18316
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021413895799954731,
+      "loss": 0.0977,
+      "step": 18317
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002140521302444357,
+      "loss": 0.0875,
+      "step": 18318
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021396531798623155,
+      "loss": 0.0664,
+      "step": 18319
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021387852122664653,
+      "loss": 0.0873,
+      "step": 18320
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021379173996739187,
+      "loss": 0.1236,
+      "step": 18321
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021370497421017932,
+      "loss": 0.0954,
+      "step": 18322
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021361822395671892,
+      "loss": 0.0861,
+      "step": 18323
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021353148920872188,
+      "loss": 0.0575,
+      "step": 18324
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021344476996789809,
+      "loss": 0.1027,
+      "step": 18325
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021335806623595734,
+      "loss": 0.0602,
+      "step": 18326
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021327137801460926,
+      "loss": 0.1006,
+      "step": 18327
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002131847053055629,
+      "loss": 0.0913,
+      "step": 18328
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002130980481105278,
+      "loss": 0.0906,
+      "step": 18329
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021301140643121218,
+      "loss": 0.0657,
+      "step": 18330
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002129247802693245,
+      "loss": 0.0507,
+      "step": 18331
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021283816962657255,
+      "loss": 0.0497,
+      "step": 18332
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021275157450466464,
+      "loss": 0.1395,
+      "step": 18333
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021266499490530787,
+      "loss": 0.0835,
+      "step": 18334
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021257843083020923,
+      "loss": 0.064,
+      "step": 18335
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00021249188228107584,
+      "loss": 0.1054,
+      "step": 18336
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002124053492596142,
+      "loss": 0.0812,
+      "step": 18337
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002123188317675303,
+      "loss": 0.0566,
+      "step": 18338
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021223232980653028,
+      "loss": 0.1061,
+      "step": 18339
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021214584337831956,
+      "loss": 0.0521,
+      "step": 18340
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021205937248460328,
+      "loss": 0.0623,
+      "step": 18341
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021197291712708678,
+      "loss": 0.1046,
+      "step": 18342
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021188647730747468,
+      "loss": 0.0944,
+      "step": 18343
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021180005302747108,
+      "loss": 0.0635,
+      "step": 18344
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002117136442887805,
+      "loss": 0.0877,
+      "step": 18345
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021162725109310644,
+      "loss": 0.0652,
+      "step": 18346
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021154087344215234,
+      "loss": 0.1169,
+      "step": 18347
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021145451133762106,
+      "loss": 0.0964,
+      "step": 18348
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021136816478121612,
+      "loss": 0.0955,
+      "step": 18349
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021128183377463962,
+      "loss": 0.1019,
+      "step": 18350
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021119551831959394,
+      "loss": 0.0663,
+      "step": 18351
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021110921841778087,
+      "loss": 0.0865,
+      "step": 18352
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002110229340709018,
+      "loss": 0.1141,
+      "step": 18353
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002109366652806587,
+      "loss": 0.1508,
+      "step": 18354
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021085041204875178,
+      "loss": 0.0754,
+      "step": 18355
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021076417437688246,
+      "loss": 0.0594,
+      "step": 18356
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002106779522667509,
+      "loss": 0.1017,
+      "step": 18357
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000210591745720057,
+      "loss": 0.1227,
+      "step": 18358
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021050555473850076,
+      "loss": 0.0931,
+      "step": 18359
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021041937932378142,
+      "loss": 0.0666,
+      "step": 18360
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002103332194775981,
+      "loss": 0.0764,
+      "step": 18361
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021024707520165,
+      "loss": 0.1006,
+      "step": 18362
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002101609464976354,
+      "loss": 0.0643,
+      "step": 18363
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00021007483336725263,
+      "loss": 0.0978,
+      "step": 18364
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020998873581219946,
+      "loss": 0.1626,
+      "step": 18365
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020990265383417385,
+      "loss": 0.0691,
+      "step": 18366
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020981658743487287,
+      "loss": 0.0547,
+      "step": 18367
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002097305366159934,
+      "loss": 0.0706,
+      "step": 18368
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020964450137923263,
+      "loss": 0.0935,
+      "step": 18369
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020955848172628656,
+      "loss": 0.0907,
+      "step": 18370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020947247765885148,
+      "loss": 0.0716,
+      "step": 18371
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020938648917862303,
+      "loss": 0.0576,
+      "step": 18372
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020930051628729673,
+      "loss": 0.1266,
+      "step": 18373
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002092145589865675,
+      "loss": 0.0727,
+      "step": 18374
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0002091286172781308,
+      "loss": 0.0865,
+      "step": 18375
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020904269116368046,
+      "loss": 0.0769,
+      "step": 18376
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020895678064491142,
+      "loss": 0.0944,
+      "step": 18377
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020887088572351731,
+      "loss": 0.0696,
+      "step": 18378
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020878500640119168,
+      "loss": 0.0973,
+      "step": 18379
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020869914267962798,
+      "loss": 0.0829,
+      "step": 18380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020861329456051892,
+      "loss": 0.0988,
+      "step": 18381
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00020852746204555773,
+      "loss": 0.0738,
+      "step": 18382
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020844164513643648,
+      "loss": 0.1173,
+      "step": 18383
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002083558438348474,
+      "loss": 0.1115,
+      "step": 18384
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020827005814248213,
+      "loss": 0.1163,
+      "step": 18385
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020818428806103196,
+      "loss": 0.0782,
+      "step": 18386
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020809853359218868,
+      "loss": 0.1093,
+      "step": 18387
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020801279473764246,
+      "loss": 0.0872,
+      "step": 18388
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002079270714990844,
+      "loss": 0.0731,
+      "step": 18389
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002078413638782045,
+      "loss": 0.0625,
+      "step": 18390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020775567187669288,
+      "loss": 0.1139,
+      "step": 18391
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002076699954962389,
+      "loss": 0.0889,
+      "step": 18392
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.000207584334738532,
+      "loss": 0.1205,
+      "step": 18393
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020749868960526086,
+      "loss": 0.1135,
+      "step": 18394
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002074130600981149,
+      "loss": 0.1169,
+      "step": 18395
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020732744621878207,
+      "loss": 0.0799,
+      "step": 18396
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020724184796895028,
+      "loss": 0.1196,
+      "step": 18397
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020715626535030773,
+      "loss": 0.1168,
+      "step": 18398
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020707069836454185,
+      "loss": 0.0668,
+      "step": 18399
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002069851470133396,
+      "loss": 0.1377,
+      "step": 18400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020689961129838764,
+      "loss": 0.0815,
+      "step": 18401
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020681409122137308,
+      "loss": 0.111,
+      "step": 18402
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020672858678398188,
+      "loss": 0.066,
+      "step": 18403
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020664309798789992,
+      "loss": 0.0731,
+      "step": 18404
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020655762483481289,
+      "loss": 0.0651,
+      "step": 18405
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020647216732640585,
+      "loss": 0.0655,
+      "step": 18406
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002063867254643642,
+      "loss": 0.084,
+      "step": 18407
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020630129925037222,
+      "loss": 0.07,
+      "step": 18408
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020621588868611484,
+      "loss": 0.1089,
+      "step": 18409
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020613049377327585,
+      "loss": 0.089,
+      "step": 18410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020604511451353914,
+      "loss": 0.064,
+      "step": 18411
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020595975090858788,
+      "loss": 0.1851,
+      "step": 18412
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002058744029601055,
+      "loss": 0.0699,
+      "step": 18413
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020578907066977437,
+      "loss": 0.0672,
+      "step": 18414
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020570375403927777,
+      "loss": 0.0663,
+      "step": 18415
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020561845307029758,
+      "loss": 0.0992,
+      "step": 18416
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002055331677645157,
+      "loss": 0.0897,
+      "step": 18417
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020544789812361342,
+      "loss": 0.067,
+      "step": 18418
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020536264414927263,
+      "loss": 0.0708,
+      "step": 18419
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.000205277405843174,
+      "loss": 0.0568,
+      "step": 18420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020519218320699817,
+      "loss": 0.0679,
+      "step": 18421
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002051069762424258,
+      "loss": 0.0715,
+      "step": 18422
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020502178495113678,
+      "loss": 0.0682,
+      "step": 18423
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020493660933481084,
+      "loss": 0.0991,
+      "step": 18424
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020485144939512746,
+      "loss": 0.1095,
+      "step": 18425
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020476630513376572,
+      "loss": 0.0854,
+      "step": 18426
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00020468117655240425,
+      "loss": 0.0677,
+      "step": 18427
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0002045960636527221,
+      "loss": 0.1566,
+      "step": 18428
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.000204510966436397,
+      "loss": 0.0842,
+      "step": 18429
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020442588490510727,
+      "loss": 0.0724,
+      "step": 18430
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020434081906053027,
+      "loss": 0.0955,
+      "step": 18431
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020425576890434327,
+      "loss": 0.1471,
+      "step": 18432
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002041707344382233,
+      "loss": 0.1086,
+      "step": 18433
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020408571566384682,
+      "loss": 0.096,
+      "step": 18434
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020400071258289053,
+      "loss": 0.0974,
+      "step": 18435
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002039157251970304,
+      "loss": 0.0904,
+      "step": 18436
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002038307535079419,
+      "loss": 0.1091,
+      "step": 18437
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002037457975173008,
+      "loss": 0.0999,
+      "step": 18438
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020366085722678173,
+      "loss": 0.068,
+      "step": 18439
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020357593263806007,
+      "loss": 0.0661,
+      "step": 18440
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002034910237528098,
+      "loss": 0.0935,
+      "step": 18441
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020340613057270564,
+      "loss": 0.0778,
+      "step": 18442
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002033212530994213,
+      "loss": 0.0602,
+      "step": 18443
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020323639133463012,
+      "loss": 0.1154,
+      "step": 18444
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020315154528000558,
+      "loss": 0.0742,
+      "step": 18445
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020306671493722052,
+      "loss": 0.0662,
+      "step": 18446
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002029819003079476,
+      "loss": 0.0688,
+      "step": 18447
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020289710139385886,
+      "loss": 0.0747,
+      "step": 18448
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020281231819662692,
+      "loss": 0.0793,
+      "step": 18449
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020272755071792294,
+      "loss": 0.0489,
+      "step": 18450
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020264279895941884,
+      "loss": 0.0968,
+      "step": 18451
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020255806292278534,
+      "loss": 0.1115,
+      "step": 18452
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002024733426096934,
+      "loss": 0.112,
+      "step": 18453
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020238863802181318,
+      "loss": 0.0704,
+      "step": 18454
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002023039491608152,
+      "loss": 0.0543,
+      "step": 18455
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002022192760283693,
+      "loss": 0.0799,
+      "step": 18456
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002021346186261449,
+      "loss": 0.0939,
+      "step": 18457
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002020499769558112,
+      "loss": 0.119,
+      "step": 18458
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020196535101903713,
+      "loss": 0.0946,
+      "step": 18459
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020188074081749097,
+      "loss": 0.1052,
+      "step": 18460
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020179614635284148,
+      "loss": 0.093,
+      "step": 18461
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002017115676267568,
+      "loss": 0.0663,
+      "step": 18462
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020162700464090435,
+      "loss": 0.0823,
+      "step": 18463
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020154245739695153,
+      "loss": 0.069,
+      "step": 18464
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002014579258965653,
+      "loss": 0.0778,
+      "step": 18465
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020137341014141253,
+      "loss": 0.0634,
+      "step": 18466
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020128891013315953,
+      "loss": 0.0679,
+      "step": 18467
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020120442587347233,
+      "loss": 0.085,
+      "step": 18468
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020111995736401712,
+      "loss": 0.078,
+      "step": 18469
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020103550460645915,
+      "loss": 0.0815,
+      "step": 18470
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002009510676024634,
+      "loss": 0.1082,
+      "step": 18471
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020086664635369535,
+      "loss": 0.0665,
+      "step": 18472
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020078224086181917,
+      "loss": 0.0631,
+      "step": 18473
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020069785112849904,
+      "loss": 0.0621,
+      "step": 18474
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00020061347715539924,
+      "loss": 0.0933,
+      "step": 18475
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00020052911894418335,
+      "loss": 0.1041,
+      "step": 18476
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002004447764965145,
+      "loss": 0.1498,
+      "step": 18477
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00020036044981405578,
+      "loss": 0.0922,
+      "step": 18478
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00020027613889846997,
+      "loss": 0.0923,
+      "step": 18479
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002001918437514194,
+      "loss": 0.1022,
+      "step": 18480
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002001075643745659,
+      "loss": 0.0577,
+      "step": 18481
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00020002330076957154,
+      "loss": 0.0931,
+      "step": 18482
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000199939052938098,
+      "loss": 0.1327,
+      "step": 18483
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019985482088180628,
+      "loss": 0.064,
+      "step": 18484
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019977060460235708,
+      "loss": 0.0737,
+      "step": 18485
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019968640410141102,
+      "loss": 0.0897,
+      "step": 18486
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019960221938062828,
+      "loss": 0.1115,
+      "step": 18487
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019951805044166848,
+      "loss": 0.0379,
+      "step": 18488
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019943389728619188,
+      "loss": 0.0961,
+      "step": 18489
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019934975991585735,
+      "loss": 0.1235,
+      "step": 18490
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019926563833232402,
+      "loss": 0.0742,
+      "step": 18491
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019918153253725013,
+      "loss": 0.0562,
+      "step": 18492
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019909744253229466,
+      "loss": 0.0524,
+      "step": 18493
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001990133683191151,
+      "loss": 0.1018,
+      "step": 18494
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019892930989936974,
+      "loss": 0.1284,
+      "step": 18495
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019884526727471564,
+      "loss": 0.087,
+      "step": 18496
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019876124044681,
+      "loss": 0.1119,
+      "step": 18497
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019867722941730958,
+      "loss": 0.0903,
+      "step": 18498
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001985932341878709,
+      "loss": 0.0585,
+      "step": 18499
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019850925476015013,
+      "loss": 0.0805,
+      "step": 18500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019842529113580287,
+      "loss": 0.0869,
+      "step": 18501
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019834134331648512,
+      "loss": 0.0836,
+      "step": 18502
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019825741130385168,
+      "loss": 0.1489,
+      "step": 18503
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019817349509955795,
+      "loss": 0.0974,
+      "step": 18504
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019808959470525822,
+      "loss": 0.0881,
+      "step": 18505
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019800571012260705,
+      "loss": 0.0723,
+      "step": 18506
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019792184135325819,
+      "loss": 0.066,
+      "step": 18507
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019783798839886503,
+      "loss": 0.1034,
+      "step": 18508
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019775415126108164,
+      "loss": 0.0642,
+      "step": 18509
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019767032994156075,
+      "loss": 0.1046,
+      "step": 18510
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019758652444195512,
+      "loss": 0.1,
+      "step": 18511
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019750273476391712,
+      "loss": 0.0699,
+      "step": 18512
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019741896090909861,
+      "loss": 0.0921,
+      "step": 18513
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019733520287915175,
+      "loss": 0.0676,
+      "step": 18514
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019725146067572818,
+      "loss": 0.0674,
+      "step": 18515
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019716773430047897,
+      "loss": 0.0757,
+      "step": 18516
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019708402375505485,
+      "loss": 0.0673,
+      "step": 18517
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019700032904110642,
+      "loss": 0.0887,
+      "step": 18518
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019691665016028403,
+      "loss": 0.069,
+      "step": 18519
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019683298711423747,
+      "loss": 0.0697,
+      "step": 18520
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001967493399046162,
+      "loss": 0.0777,
+      "step": 18521
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019666570853306997,
+      "loss": 0.1676,
+      "step": 18522
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019658209300124762,
+      "loss": 0.0856,
+      "step": 18523
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019649849331079749,
+      "loss": 0.1268,
+      "step": 18524
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019641490946336847,
+      "loss": 0.108,
+      "step": 18525
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001963313414606085,
+      "loss": 0.1207,
+      "step": 18526
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001962477893041652,
+      "loss": 0.0741,
+      "step": 18527
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019616425299568574,
+      "loss": 0.1046,
+      "step": 18528
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001960807325368179,
+      "loss": 0.1194,
+      "step": 18529
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019599722792920803,
+      "loss": 0.1191,
+      "step": 18530
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019591373917450284,
+      "loss": 0.1282,
+      "step": 18531
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001958302662743483,
+      "loss": 0.1279,
+      "step": 18532
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019574680923039047,
+      "loss": 0.1189,
+      "step": 18533
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019566336804427464,
+      "loss": 0.0681,
+      "step": 18534
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001955799427176461,
+      "loss": 0.0662,
+      "step": 18535
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019549653325215034,
+      "loss": 0.0536,
+      "step": 18536
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019541313964943153,
+      "loss": 0.0727,
+      "step": 18537
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019532976191113417,
+      "loss": 0.172,
+      "step": 18538
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019524640003890204,
+      "loss": 0.0812,
+      "step": 18539
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019516305403437885,
+      "loss": 0.0745,
+      "step": 18540
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001950797238992078,
+      "loss": 0.0618,
+      "step": 18541
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001949964096350325,
+      "loss": 0.0829,
+      "step": 18542
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019491311124349544,
+      "loss": 0.0836,
+      "step": 18543
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.000194829828726239,
+      "loss": 0.1008,
+      "step": 18544
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019474656208490494,
+      "loss": 0.0955,
+      "step": 18545
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019466331132113578,
+      "loss": 0.0667,
+      "step": 18546
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019458007643657262,
+      "loss": 0.1116,
+      "step": 18547
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019449685743285638,
+      "loss": 0.0858,
+      "step": 18548
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001944136543116286,
+      "loss": 0.0796,
+      "step": 18549
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019433046707452938,
+      "loss": 0.0638,
+      "step": 18550
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019424729572319911,
+      "loss": 0.1166,
+      "step": 18551
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019416414025927764,
+      "loss": 0.0695,
+      "step": 18552
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001940810006844046,
+      "loss": 0.1196,
+      "step": 18553
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001939978770002191,
+      "loss": 0.0741,
+      "step": 18554
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001939147692083606,
+      "loss": 0.0776,
+      "step": 18555
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019383167731046715,
+      "loss": 0.0571,
+      "step": 18556
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019374860130817772,
+      "loss": 0.1075,
+      "step": 18557
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019366554120313017,
+      "loss": 0.0781,
+      "step": 18558
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019358249699696217,
+      "loss": 0.1342,
+      "step": 18559
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019349946869131118,
+      "loss": 0.1854,
+      "step": 18560
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.000193416456287814,
+      "loss": 0.1051,
+      "step": 18561
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001933334597881079,
+      "loss": 0.0494,
+      "step": 18562
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001932504791938292,
+      "loss": 0.0643,
+      "step": 18563
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019316751450661407,
+      "loss": 0.0886,
+      "step": 18564
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001930845657280983,
+      "loss": 0.1016,
+      "step": 18565
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019300163285991713,
+      "loss": 0.0401,
+      "step": 18566
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019291871590370636,
+      "loss": 0.0661,
+      "step": 18567
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019283581486110048,
+      "loss": 0.1089,
+      "step": 18568
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019275292973373447,
+      "loss": 0.0911,
+      "step": 18569
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019267006052324243,
+      "loss": 0.1079,
+      "step": 18570
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019258720723125834,
+      "loss": 0.1193,
+      "step": 18571
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019250436985941577,
+      "loss": 0.0652,
+      "step": 18572
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019242154840934812,
+      "loss": 0.0949,
+      "step": 18573
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019233874288268816,
+      "loss": 0.1036,
+      "step": 18574
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019225595328106916,
+      "loss": 0.0676,
+      "step": 18575
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019217317960612323,
+      "loss": 0.0605,
+      "step": 18576
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019209042185948212,
+      "loss": 0.1198,
+      "step": 18577
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019200768004277825,
+      "loss": 0.0862,
+      "step": 18578
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019192495415764276,
+      "loss": 0.065,
+      "step": 18579
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019184224420570683,
+      "loss": 0.0526,
+      "step": 18580
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019175955018860104,
+      "loss": 0.1191,
+      "step": 18581
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019167687210795638,
+      "loss": 0.0613,
+      "step": 18582
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019159420996540266,
+      "loss": 0.0756,
+      "step": 18583
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019151156376257005,
+      "loss": 0.0515,
+      "step": 18584
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019142893350108803,
+      "loss": 0.0783,
+      "step": 18585
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019134631918258572,
+      "loss": 0.1116,
+      "step": 18586
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019126372080869192,
+      "loss": 0.1404,
+      "step": 18587
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019118113838103578,
+      "loss": 0.0786,
+      "step": 18588
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019109857190124514,
+      "loss": 0.0728,
+      "step": 18589
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019101602137094843,
+      "loss": 0.053,
+      "step": 18590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019093348679177313,
+      "loss": 0.0956,
+      "step": 18591
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019085096816534654,
+      "loss": 0.092,
+      "step": 18592
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019076846549329586,
+      "loss": 0.0809,
+      "step": 18593
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019068597877724736,
+      "loss": 0.0981,
+      "step": 18594
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001906035080188282,
+      "loss": 0.0673,
+      "step": 18595
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019052105321966416,
+      "loss": 0.064,
+      "step": 18596
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001904386143813811,
+      "loss": 0.0684,
+      "step": 18597
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019035619150560413,
+      "loss": 0.0905,
+      "step": 18598
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019027378459395895,
+      "loss": 0.1048,
+      "step": 18599
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019019139364807026,
+      "loss": 0.094,
+      "step": 18600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001901090186695622,
+      "loss": 0.0735,
+      "step": 18601
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001900266596600596,
+      "loss": 0.0975,
+      "step": 18602
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018994431662118616,
+      "loss": 0.0621,
+      "step": 18603
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018986198955456545,
+      "loss": 0.0863,
+      "step": 18604
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018977967846182077,
+      "loss": 0.0602,
+      "step": 18605
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018969738334457498,
+      "loss": 0.109,
+      "step": 18606
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018961510420445073,
+      "loss": 0.0934,
+      "step": 18607
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018953284104307022,
+      "loss": 0.155,
+      "step": 18608
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018945059386205577,
+      "loss": 0.0842,
+      "step": 18609
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018936836266302914,
+      "loss": 0.1042,
+      "step": 18610
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001892861474476115,
+      "loss": 0.0736,
+      "step": 18611
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018920394821742416,
+      "loss": 0.1064,
+      "step": 18612
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00018912176497408762,
+      "loss": 0.1011,
+      "step": 18613
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001890395977192223,
+      "loss": 0.1141,
+      "step": 18614
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018895744645444857,
+      "loss": 0.09,
+      "step": 18615
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018887531118138624,
+      "loss": 0.1206,
+      "step": 18616
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018879319190165457,
+      "loss": 0.0613,
+      "step": 18617
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018871108861687303,
+      "loss": 0.0896,
+      "step": 18618
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018862900132866,
+      "loss": 0.0929,
+      "step": 18619
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018854693003863467,
+      "loss": 0.1157,
+      "step": 18620
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018846487474841468,
+      "loss": 0.1018,
+      "step": 18621
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018838283545961855,
+      "loss": 0.0723,
+      "step": 18622
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018830081217386353,
+      "loss": 0.1107,
+      "step": 18623
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000188218804892767,
+      "loss": 0.1129,
+      "step": 18624
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018813681361794587,
+      "loss": 0.0802,
+      "step": 18625
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000188054838351017,
+      "loss": 0.0865,
+      "step": 18626
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001879728790935964,
+      "loss": 0.0504,
+      "step": 18627
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001878909358473001,
+      "loss": 0.1023,
+      "step": 18628
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018780900861374428,
+      "loss": 0.0821,
+      "step": 18629
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018772709739454374,
+      "loss": 0.1221,
+      "step": 18630
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001876452021913141,
+      "loss": 0.0723,
+      "step": 18631
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018756332300566992,
+      "loss": 0.0997,
+      "step": 18632
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001874814598392257,
+      "loss": 0.0767,
+      "step": 18633
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018739961269359517,
+      "loss": 0.0687,
+      "step": 18634
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018731778157039282,
+      "loss": 0.0974,
+      "step": 18635
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018723596647123164,
+      "loss": 0.1075,
+      "step": 18636
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018715416739772516,
+      "loss": 0.0683,
+      "step": 18637
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018707238435148598,
+      "loss": 0.062,
+      "step": 18638
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018699061733412682,
+      "loss": 0.1073,
+      "step": 18639
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018690886634725967,
+      "loss": 0.0938,
+      "step": 18640
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018682713139249674,
+      "loss": 0.0427,
+      "step": 18641
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018674541247144937,
+      "loss": 0.1047,
+      "step": 18642
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001866637095857293,
+      "loss": 0.1133,
+      "step": 18643
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018658202273694714,
+      "loss": 0.097,
+      "step": 18644
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018650035192671377,
+      "loss": 0.0741,
+      "step": 18645
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001864186971566393,
+      "loss": 0.0642,
+      "step": 18646
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018633705842833358,
+      "loss": 0.0819,
+      "step": 18647
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018625543574340688,
+      "loss": 0.09,
+      "step": 18648
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018617382910346826,
+      "loss": 0.1021,
+      "step": 18649
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001860922385101268,
+      "loss": 0.0817,
+      "step": 18650
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001860106639649911,
+      "loss": 0.1198,
+      "step": 18651
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018592910546967012,
+      "loss": 0.0593,
+      "step": 18652
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018584756302577167,
+      "loss": 0.1099,
+      "step": 18653
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018576603663490322,
+      "loss": 0.0887,
+      "step": 18654
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018568452629867283,
+      "loss": 0.0414,
+      "step": 18655
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018560303201868757,
+      "loss": 0.074,
+      "step": 18656
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018552155379655412,
+      "loss": 0.0693,
+      "step": 18657
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001854400916338791,
+      "loss": 0.0857,
+      "step": 18658
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00018535864553226878,
+      "loss": 0.0836,
+      "step": 18659
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.000185277215493329,
+      "loss": 0.0676,
+      "step": 18660
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001851958015186651,
+      "loss": 0.1301,
+      "step": 18661
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018511440360988285,
+      "loss": 0.1569,
+      "step": 18662
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018503302176858683,
+      "loss": 0.0648,
+      "step": 18663
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001849516559963821,
+      "loss": 0.0916,
+      "step": 18664
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018487030629487267,
+      "loss": 0.0816,
+      "step": 18665
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018478897266566263,
+      "loss": 0.0893,
+      "step": 18666
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018470765511035536,
+      "loss": 0.1278,
+      "step": 18667
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018462635363055492,
+      "loss": 0.0527,
+      "step": 18668
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018454506822786389,
+      "loss": 0.0802,
+      "step": 18669
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001844637989038852,
+      "loss": 0.0819,
+      "step": 18670
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018438254566022105,
+      "loss": 0.0903,
+      "step": 18671
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001843013084984736,
+      "loss": 0.0667,
+      "step": 18672
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018422008742024487,
+      "loss": 0.069,
+      "step": 18673
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018413888242713595,
+      "loss": 0.0366,
+      "step": 18674
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018405769352074854,
+      "loss": 0.0717,
+      "step": 18675
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018397652070268312,
+      "loss": 0.0562,
+      "step": 18676
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001838953639745403,
+      "loss": 0.1178,
+      "step": 18677
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001838142233379202,
+      "loss": 0.0791,
+      "step": 18678
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018373309879442292,
+      "loss": 0.0832,
+      "step": 18679
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018365199034564773,
+      "loss": 0.0767,
+      "step": 18680
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001835708979931938,
+      "loss": 0.1012,
+      "step": 18681
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018348982173866057,
+      "loss": 0.0764,
+      "step": 18682
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001834087615836465,
+      "loss": 0.1022,
+      "step": 18683
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001833277175297494,
+      "loss": 0.1018,
+      "step": 18684
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018324668957856794,
+      "loss": 0.0681,
+      "step": 18685
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018316567773169945,
+      "loss": 0.1152,
+      "step": 18686
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018308468199074103,
+      "loss": 0.0855,
+      "step": 18687
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018300370235729024,
+      "loss": 0.0969,
+      "step": 18688
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018292273883294363,
+      "loss": 0.0629,
+      "step": 18689
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018284179141929747,
+      "loss": 0.1115,
+      "step": 18690
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001827608601179478,
+      "loss": 0.0826,
+      "step": 18691
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018267994493049056,
+      "loss": 0.1235,
+      "step": 18692
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018259904585852084,
+      "loss": 0.0755,
+      "step": 18693
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018251816290363422,
+      "loss": 0.0685,
+      "step": 18694
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001824372960674251,
+      "loss": 0.1455,
+      "step": 18695
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018235644535148833,
+      "loss": 0.1348,
+      "step": 18696
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018227561075741805,
+      "loss": 0.1078,
+      "step": 18697
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001821947922868079,
+      "loss": 0.0673,
+      "step": 18698
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001821139899412516,
+      "loss": 0.0482,
+      "step": 18699
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018203320372234223,
+      "loss": 0.0848,
+      "step": 18700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018195243363167246,
+      "loss": 0.0706,
+      "step": 18701
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001818716796708354,
+      "loss": 0.1217,
+      "step": 18702
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018179094184142308,
+      "loss": 0.0789,
+      "step": 18703
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018171022014502726,
+      "loss": 0.053,
+      "step": 18704
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00018162951458323983,
+      "loss": 0.0724,
+      "step": 18705
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018154882515765213,
+      "loss": 0.0645,
+      "step": 18706
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018146815186985478,
+      "loss": 0.1245,
+      "step": 18707
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001813874947214389,
+      "loss": 0.0739,
+      "step": 18708
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018130685371399468,
+      "loss": 0.1031,
+      "step": 18709
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001812262288491121,
+      "loss": 0.0299,
+      "step": 18710
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018114562012838087,
+      "loss": 0.0293,
+      "step": 18711
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001810650275533905,
+      "loss": 0.1057,
+      "step": 18712
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018098445112572993,
+      "loss": 0.1133,
+      "step": 18713
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018090389084698778,
+      "loss": 0.0483,
+      "step": 18714
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000180823346718753,
+      "loss": 0.0941,
+      "step": 18715
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001807428187426131,
+      "loss": 0.1055,
+      "step": 18716
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001806623069201565,
+      "loss": 0.1561,
+      "step": 18717
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018058181125297034,
+      "loss": 0.0806,
+      "step": 18718
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018050133174264194,
+      "loss": 0.0979,
+      "step": 18719
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000180420868390758,
+      "loss": 0.095,
+      "step": 18720
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018034042119890493,
+      "loss": 0.0846,
+      "step": 18721
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018025999016866922,
+      "loss": 0.0821,
+      "step": 18722
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018017957530163688,
+      "loss": 0.0806,
+      "step": 18723
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00018009917659939323,
+      "loss": 0.0833,
+      "step": 18724
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001800187940635233,
+      "loss": 0.069,
+      "step": 18725
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017993842769561253,
+      "loss": 0.0579,
+      "step": 18726
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017985807749724514,
+      "loss": 0.1067,
+      "step": 18727
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017977774347000596,
+      "loss": 0.0724,
+      "step": 18728
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017969742561547853,
+      "loss": 0.0704,
+      "step": 18729
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017961712393524664,
+      "loss": 0.0516,
+      "step": 18730
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017953683843089364,
+      "loss": 0.0616,
+      "step": 18731
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017945656910400254,
+      "loss": 0.091,
+      "step": 18732
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.000179376315956156,
+      "loss": 0.0969,
+      "step": 18733
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017929607898893617,
+      "loss": 0.0851,
+      "step": 18734
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017921585820392571,
+      "loss": 0.062,
+      "step": 18735
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001791356536027061,
+      "loss": 0.0746,
+      "step": 18736
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017905546518685856,
+      "loss": 0.1055,
+      "step": 18737
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017897529295796445,
+      "loss": 0.0632,
+      "step": 18738
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017889513691760463,
+      "loss": 0.0789,
+      "step": 18739
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017881499706735938,
+      "loss": 0.0845,
+      "step": 18740
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017873487340880868,
+      "loss": 0.1109,
+      "step": 18741
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017865476594353282,
+      "loss": 0.0882,
+      "step": 18742
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017857467467311116,
+      "loss": 0.0509,
+      "step": 18743
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017849459959912284,
+      "loss": 0.0862,
+      "step": 18744
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017841454072314668,
+      "loss": 0.0655,
+      "step": 18745
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017833449804676105,
+      "loss": 0.0749,
+      "step": 18746
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017825447157154473,
+      "loss": 0.0662,
+      "step": 18747
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001781744612990751,
+      "loss": 0.1311,
+      "step": 18748
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017809446723093014,
+      "loss": 0.1123,
+      "step": 18749
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017801448936868714,
+      "loss": 0.0804,
+      "step": 18750
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00017793452771392283,
+      "loss": 0.0815,
+      "step": 18751
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017785458226821395,
+      "loss": 0.1148,
+      "step": 18752
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017777465303313688,
+      "loss": 0.1269,
+      "step": 18753
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001776947400102672,
+      "loss": 0.0736,
+      "step": 18754
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017761484320118125,
+      "loss": 0.1024,
+      "step": 18755
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017753496260745406,
+      "loss": 0.1086,
+      "step": 18756
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017745509823066075,
+      "loss": 0.0732,
+      "step": 18757
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017737525007237577,
+      "loss": 0.0719,
+      "step": 18758
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017729541813417393,
+      "loss": 0.0448,
+      "step": 18759
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017721560241762912,
+      "loss": 0.1621,
+      "step": 18760
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017713580292431496,
+      "loss": 0.1119,
+      "step": 18761
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001770560196558053,
+      "loss": 0.1057,
+      "step": 18762
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001769762526136729,
+      "loss": 0.1218,
+      "step": 18763
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001768965017994908,
+      "loss": 0.1237,
+      "step": 18764
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017681676721483132,
+      "loss": 0.1013,
+      "step": 18765
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017673704886126673,
+      "loss": 0.0696,
+      "step": 18766
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017665734674036847,
+      "loss": 0.0528,
+      "step": 18767
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017657766085370875,
+      "loss": 0.0656,
+      "step": 18768
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017649799120285814,
+      "loss": 0.0613,
+      "step": 18769
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017641833778938809,
+      "loss": 0.0736,
+      "step": 18770
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001763387006148689,
+      "loss": 0.0493,
+      "step": 18771
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017625907968087085,
+      "loss": 0.0762,
+      "step": 18772
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017617947498896368,
+      "loss": 0.113,
+      "step": 18773
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017609988654071707,
+      "loss": 0.0962,
+      "step": 18774
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001760203143377005,
+      "loss": 0.118,
+      "step": 18775
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017594075838148283,
+      "loss": 0.0714,
+      "step": 18776
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001758612186736327,
+      "loss": 0.1295,
+      "step": 18777
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001757816952157184,
+      "loss": 0.1148,
+      "step": 18778
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001757021880093077,
+      "loss": 0.0848,
+      "step": 18779
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017562269705596867,
+      "loss": 0.0973,
+      "step": 18780
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017554322235726838,
+      "loss": 0.1287,
+      "step": 18781
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017546376391477426,
+      "loss": 0.1139,
+      "step": 18782
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001753843217300527,
+      "loss": 0.1375,
+      "step": 18783
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017530489580467023,
+      "loss": 0.0816,
+      "step": 18784
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017522548614019295,
+      "loss": 0.0676,
+      "step": 18785
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017514609273818648,
+      "loss": 0.0494,
+      "step": 18786
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017506671560021614,
+      "loss": 0.1107,
+      "step": 18787
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001749873547278473,
+      "loss": 0.1044,
+      "step": 18788
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001749080101226449,
+      "loss": 0.0756,
+      "step": 18789
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017482868178617285,
+      "loss": 0.0818,
+      "step": 18790
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.000174749369719996,
+      "loss": 0.0829,
+      "step": 18791
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017467007392567768,
+      "loss": 0.0807,
+      "step": 18792
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017459079440478164,
+      "loss": 0.0576,
+      "step": 18793
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017451153115887087,
+      "loss": 0.069,
+      "step": 18794
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001744322841895084,
+      "loss": 0.0594,
+      "step": 18795
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001743530534982569,
+      "loss": 0.071,
+      "step": 18796
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00017427383908667838,
+      "loss": 0.1212,
+      "step": 18797
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017419464095633485,
+      "loss": 0.0828,
+      "step": 18798
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017411545910878769,
+      "loss": 0.0786,
+      "step": 18799
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017403629354559847,
+      "loss": 0.0904,
+      "step": 18800
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017395714426832775,
+      "loss": 0.1013,
+      "step": 18801
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017387801127853676,
+      "loss": 0.129,
+      "step": 18802
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001737988945777853,
+      "loss": 0.1078,
+      "step": 18803
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017371979416763361,
+      "loss": 0.0743,
+      "step": 18804
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001736407100496412,
+      "loss": 0.1129,
+      "step": 18805
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017356164222536753,
+      "loss": 0.1073,
+      "step": 18806
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001734825906963712,
+      "loss": 0.1215,
+      "step": 18807
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001734035554642115,
+      "loss": 0.0489,
+      "step": 18808
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017332453653044655,
+      "loss": 0.0968,
+      "step": 18809
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017324553389663443,
+      "loss": 0.0872,
+      "step": 18810
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001731665475643326,
+      "loss": 0.079,
+      "step": 18811
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017308757753509897,
+      "loss": 0.0901,
+      "step": 18812
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017300862381049043,
+      "loss": 0.1014,
+      "step": 18813
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001729296863920634,
+      "loss": 0.0728,
+      "step": 18814
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017285076528137478,
+      "loss": 0.0742,
+      "step": 18815
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017277186047998062,
+      "loss": 0.0569,
+      "step": 18816
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017269297198943667,
+      "loss": 0.1404,
+      "step": 18817
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017261409981129838,
+      "loss": 0.0923,
+      "step": 18818
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017253524394712083,
+      "loss": 0.0581,
+      "step": 18819
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001724564043984588,
+      "loss": 0.1206,
+      "step": 18820
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001723775811668672,
+      "loss": 0.0887,
+      "step": 18821
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001722987742538996,
+      "loss": 0.0934,
+      "step": 18822
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017221998366111059,
+      "loss": 0.0836,
+      "step": 18823
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001721412093900533,
+      "loss": 0.09,
+      "step": 18824
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017206245144228116,
+      "loss": 0.1251,
+      "step": 18825
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017198370981934686,
+      "loss": 0.1467,
+      "step": 18826
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001719049845228029,
+      "loss": 0.0728,
+      "step": 18827
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017182627555420194,
+      "loss": 0.0926,
+      "step": 18828
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017174758291509562,
+      "loss": 0.1188,
+      "step": 18829
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017166890660703582,
+      "loss": 0.0897,
+      "step": 18830
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001715902466315735,
+      "loss": 0.1202,
+      "step": 18831
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001715116029902597,
+      "loss": 0.1171,
+      "step": 18832
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017143297568464534,
+      "loss": 0.1202,
+      "step": 18833
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001713543647162804,
+      "loss": 0.066,
+      "step": 18834
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017127577008671524,
+      "loss": 0.0817,
+      "step": 18835
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017119719179749948,
+      "loss": 0.0618,
+      "step": 18836
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017111862985018234,
+      "loss": 0.1141,
+      "step": 18837
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001710400842463129,
+      "loss": 0.1032,
+      "step": 18838
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017096155498743992,
+      "loss": 0.095,
+      "step": 18839
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017088304207511174,
+      "loss": 0.1346,
+      "step": 18840
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001708045455108763,
+      "loss": 0.0869,
+      "step": 18841
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017072606529628165,
+      "loss": 0.0807,
+      "step": 18842
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017064760143287495,
+      "loss": 0.1191,
+      "step": 18843
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00017056915392220373,
+      "loss": 0.0623,
+      "step": 18844
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00017049072276581446,
+      "loss": 0.0642,
+      "step": 18845
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00017041230796525363,
+      "loss": 0.0901,
+      "step": 18846
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00017033390952206717,
+      "loss": 0.1071,
+      "step": 18847
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00017025552743780138,
+      "loss": 0.1025,
+      "step": 18848
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00017017716171400144,
+      "loss": 0.0684,
+      "step": 18849
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00017009881235221258,
+      "loss": 0.0667,
+      "step": 18850
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001700204793539798,
+      "loss": 0.0908,
+      "step": 18851
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001699421627208474,
+      "loss": 0.0508,
+      "step": 18852
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016986386245435947,
+      "loss": 0.0684,
+      "step": 18853
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016978557855606004,
+      "loss": 0.0666,
+      "step": 18854
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016970731102749305,
+      "loss": 0.0895,
+      "step": 18855
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016962905987020127,
+      "loss": 0.0457,
+      "step": 18856
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016955082508572793,
+      "loss": 0.0677,
+      "step": 18857
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016947260667561536,
+      "loss": 0.0656,
+      "step": 18858
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016939440464140598,
+      "loss": 0.0918,
+      "step": 18859
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016931621898464168,
+      "loss": 0.1018,
+      "step": 18860
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016923804970686373,
+      "loss": 0.0928,
+      "step": 18861
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001691598968096142,
+      "loss": 0.086,
+      "step": 18862
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016908176029443357,
+      "loss": 0.1267,
+      "step": 18863
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001690036401628623,
+      "loss": 0.0939,
+      "step": 18864
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016892553641644138,
+      "loss": 0.1041,
+      "step": 18865
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016884744905671035,
+      "loss": 0.1204,
+      "step": 18866
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016876937808520866,
+      "loss": 0.1281,
+      "step": 18867
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016869132350347637,
+      "loss": 0.0815,
+      "step": 18868
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001686132853130521,
+      "loss": 0.061,
+      "step": 18869
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016853526351547454,
+      "loss": 0.1667,
+      "step": 18870
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016845725811228228,
+      "loss": 0.2322,
+      "step": 18871
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016837926910501322,
+      "loss": 0.151,
+      "step": 18872
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016830129649520508,
+      "loss": 0.1333,
+      "step": 18873
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016822334028439523,
+      "loss": 0.1193,
+      "step": 18874
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016814540047412085,
+      "loss": 0.0936,
+      "step": 18875
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016806747706591896,
+      "loss": 0.1307,
+      "step": 18876
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016798957006132587,
+      "loss": 0.1044,
+      "step": 18877
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016791167946187758,
+      "loss": 0.1188,
+      "step": 18878
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016783380526911006,
+      "loss": 0.0638,
+      "step": 18879
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001677559474845587,
+      "loss": 0.0988,
+      "step": 18880
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001676781061097583,
+      "loss": 0.0883,
+      "step": 18881
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016760028114624437,
+      "loss": 0.0818,
+      "step": 18882
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001675224725955512,
+      "loss": 0.1016,
+      "step": 18883
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016744468045921278,
+      "loss": 0.1236,
+      "step": 18884
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001673669047387628,
+      "loss": 0.0872,
+      "step": 18885
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016728914543573547,
+      "loss": 0.0714,
+      "step": 18886
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016721140255166324,
+      "loss": 0.0803,
+      "step": 18887
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016713367608807973,
+      "loss": 0.0776,
+      "step": 18888
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00016705596604651708,
+      "loss": 0.0673,
+      "step": 18889
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001669782724285076,
+      "loss": 0.0764,
+      "step": 18890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001669005952355833,
+      "loss": 0.0708,
+      "step": 18891
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016682293446927565,
+      "loss": 0.0784,
+      "step": 18892
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016674529013111606,
+      "loss": 0.0825,
+      "step": 18893
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016666766222263508,
+      "loss": 0.0593,
+      "step": 18894
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016659005074536393,
+      "loss": 0.1158,
+      "step": 18895
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001665124557008324,
+      "loss": 0.112,
+      "step": 18896
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016643487709057093,
+      "loss": 0.0755,
+      "step": 18897
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016635731491610896,
+      "loss": 0.0975,
+      "step": 18898
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001662797691789758,
+      "loss": 0.0903,
+      "step": 18899
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001662022398807005,
+      "loss": 0.0931,
+      "step": 18900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016612472702281144,
+      "loss": 0.1093,
+      "step": 18901
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016604723060683745,
+      "loss": 0.0594,
+      "step": 18902
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016596975063430653,
+      "loss": 0.1017,
+      "step": 18903
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001658922871067461,
+      "loss": 0.1505,
+      "step": 18904
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016581484002568358,
+      "loss": 0.0803,
+      "step": 18905
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016573740939264603,
+      "loss": 0.0736,
+      "step": 18906
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001656599952091602,
+      "loss": 0.0978,
+      "step": 18907
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016558259747675286,
+      "loss": 0.0665,
+      "step": 18908
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001655052161969498,
+      "loss": 0.0637,
+      "step": 18909
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001654278513712768,
+      "loss": 0.0787,
+      "step": 18910
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001653505030012593,
+      "loss": 0.0831,
+      "step": 18911
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016527317108842244,
+      "loss": 0.0753,
+      "step": 18912
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016519585563429086,
+      "loss": 0.0733,
+      "step": 18913
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000165118556640389,
+      "loss": 0.053,
+      "step": 18914
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016504127410824144,
+      "loss": 0.0775,
+      "step": 18915
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001649640080393716,
+      "loss": 0.0766,
+      "step": 18916
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016488675843530286,
+      "loss": 0.0913,
+      "step": 18917
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016480952529755887,
+      "loss": 0.1063,
+      "step": 18918
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001647323086276622,
+      "loss": 0.0583,
+      "step": 18919
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016465510842713537,
+      "loss": 0.0663,
+      "step": 18920
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016457792469750033,
+      "loss": 0.0945,
+      "step": 18921
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016450075744027948,
+      "loss": 0.1281,
+      "step": 18922
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016442360665699408,
+      "loss": 0.1157,
+      "step": 18923
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016434647234916534,
+      "loss": 0.1005,
+      "step": 18924
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016426935451831405,
+      "loss": 0.072,
+      "step": 18925
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016419225316596098,
+      "loss": 0.0511,
+      "step": 18926
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016411516829362605,
+      "loss": 0.072,
+      "step": 18927
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016403809990282937,
+      "loss": 0.1198,
+      "step": 18928
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016396104799509093,
+      "loss": 0.067,
+      "step": 18929
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016388401257192952,
+      "loss": 0.097,
+      "step": 18930
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016380699363486429,
+      "loss": 0.1138,
+      "step": 18931
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016372999118541377,
+      "loss": 0.1123,
+      "step": 18932
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001636530052250962,
+      "loss": 0.1082,
+      "step": 18933
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016357603575542935,
+      "loss": 0.0883,
+      "step": 18934
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016349908277793146,
+      "loss": 0.0403,
+      "step": 18935
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00016342214629411943,
+      "loss": 0.0878,
+      "step": 18936
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001633452263055104,
+      "loss": 0.0648,
+      "step": 18937
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001632683228136208,
+      "loss": 0.0818,
+      "step": 18938
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001631914358199673,
+      "loss": 0.0779,
+      "step": 18939
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001631145653260655,
+      "loss": 0.1166,
+      "step": 18940
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001630377113334316,
+      "loss": 0.1024,
+      "step": 18941
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016296087384358083,
+      "loss": 0.089,
+      "step": 18942
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016288405285802798,
+      "loss": 0.066,
+      "step": 18943
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016280724837828796,
+      "loss": 0.1639,
+      "step": 18944
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016273046040587514,
+      "loss": 0.1149,
+      "step": 18945
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016265368894230359,
+      "loss": 0.0928,
+      "step": 18946
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001625769339890867,
+      "loss": 0.0659,
+      "step": 18947
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016250019554773853,
+      "loss": 0.0668,
+      "step": 18948
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001624234736197715,
+      "loss": 0.099,
+      "step": 18949
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016234676820669903,
+      "loss": 0.0969,
+      "step": 18950
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016227007931003313,
+      "loss": 0.0833,
+      "step": 18951
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016219340693128615,
+      "loss": 0.1464,
+      "step": 18952
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016211675107196966,
+      "loss": 0.0787,
+      "step": 18953
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016204011173359502,
+      "loss": 0.0488,
+      "step": 18954
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016196348891767376,
+      "loss": 0.0792,
+      "step": 18955
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001618868826257165,
+      "loss": 0.1161,
+      "step": 18956
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016181029285923366,
+      "loss": 0.0642,
+      "step": 18957
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001617337196197354,
+      "loss": 0.0822,
+      "step": 18958
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016165716290873145,
+      "loss": 0.0732,
+      "step": 18959
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016158062272773156,
+      "loss": 0.0692,
+      "step": 18960
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016150409907824472,
+      "loss": 0.0593,
+      "step": 18961
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016142759196178003,
+      "loss": 0.1133,
+      "step": 18962
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016135110137984577,
+      "loss": 0.0689,
+      "step": 18963
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016127462733395037,
+      "loss": 0.1328,
+      "step": 18964
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016119816982560143,
+      "loss": 0.0652,
+      "step": 18965
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001611217288563067,
+      "loss": 0.0757,
+      "step": 18966
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016104530442757314,
+      "loss": 0.1046,
+      "step": 18967
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016096889654090797,
+      "loss": 0.1133,
+      "step": 18968
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016089250519781773,
+      "loss": 0.061,
+      "step": 18969
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016081613039980823,
+      "loss": 0.085,
+      "step": 18970
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016073977214838597,
+      "loss": 0.1099,
+      "step": 18971
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016066343044505638,
+      "loss": 0.152,
+      "step": 18972
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016058710529132458,
+      "loss": 0.1599,
+      "step": 18973
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001605107966886953,
+      "loss": 0.0875,
+      "step": 18974
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016043450463867372,
+      "loss": 0.0522,
+      "step": 18975
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016035822914276378,
+      "loss": 0.0724,
+      "step": 18976
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001602819702024695,
+      "loss": 0.121,
+      "step": 18977
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016020572781929443,
+      "loss": 0.052,
+      "step": 18978
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016012950199474206,
+      "loss": 0.05,
+      "step": 18979
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00016005329273031498,
+      "loss": 0.0809,
+      "step": 18980
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001599771000275163,
+      "loss": 0.0936,
+      "step": 18981
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00015990092388784805,
+      "loss": 0.0648,
+      "step": 18982
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015982476431281256,
+      "loss": 0.0432,
+      "step": 18983
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001597486213039112,
+      "loss": 0.076,
+      "step": 18984
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015967249486264557,
+      "loss": 0.0901,
+      "step": 18985
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015959638499051643,
+      "loss": 0.0879,
+      "step": 18986
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015952029168902438,
+      "loss": 0.0471,
+      "step": 18987
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015944421495967022,
+      "loss": 0.0541,
+      "step": 18988
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015936815480395374,
+      "loss": 0.0706,
+      "step": 18989
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015929211122337462,
+      "loss": 0.1326,
+      "step": 18990
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015921608421943213,
+      "loss": 0.0912,
+      "step": 18991
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015914007379362572,
+      "loss": 0.0499,
+      "step": 18992
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015906407994745398,
+      "loss": 0.048,
+      "step": 18993
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015898810268241492,
+      "loss": 0.076,
+      "step": 18994
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015891214200000714,
+      "loss": 0.0656,
+      "step": 18995
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015883619790172833,
+      "loss": 0.0649,
+      "step": 18996
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015876027038907558,
+      "loss": 0.064,
+      "step": 18997
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001586843594635464,
+      "loss": 0.0954,
+      "step": 18998
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015860846512663718,
+      "loss": 0.0504,
+      "step": 18999
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015853258737984444,
+      "loss": 0.0653,
+      "step": 19000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015845672622466457,
+      "loss": 0.063,
+      "step": 19001
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001583808816625929,
+      "loss": 0.0717,
+      "step": 19002
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001583050536951255,
+      "loss": 0.0397,
+      "step": 19003
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015822924232375714,
+      "loss": 0.1143,
+      "step": 19004
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015815344754998263,
+      "loss": 0.0565,
+      "step": 19005
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015807766937529655,
+      "loss": 0.1109,
+      "step": 19006
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015800190780119273,
+      "loss": 0.0659,
+      "step": 19007
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001579261628291655,
+      "loss": 0.072,
+      "step": 19008
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015785043446070823,
+      "loss": 0.066,
+      "step": 19009
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015777472269731384,
+      "loss": 0.1326,
+      "step": 19010
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015769902754047538,
+      "loss": 0.048,
+      "step": 19011
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015762334899168507,
+      "loss": 0.1073,
+      "step": 19012
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001575476870524356,
+      "loss": 0.0934,
+      "step": 19013
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015747204172421826,
+      "loss": 0.1395,
+      "step": 19014
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015739641300852524,
+      "loss": 0.0978,
+      "step": 19015
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015732080090684732,
+      "loss": 0.1029,
+      "step": 19016
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015724520542067545,
+      "loss": 0.1136,
+      "step": 19017
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015716962655150024,
+      "loss": 0.0623,
+      "step": 19018
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015709406430081186,
+      "loss": 0.0983,
+      "step": 19019
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015701851867009999,
+      "loss": 0.1204,
+      "step": 19020
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015694298966085463,
+      "loss": 0.1001,
+      "step": 19021
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015686747727456485,
+      "loss": 0.0871,
+      "step": 19022
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015679198151271923,
+      "loss": 0.0784,
+      "step": 19023
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000156716502376807,
+      "loss": 0.0924,
+      "step": 19024
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000156641039868316,
+      "loss": 0.079,
+      "step": 19025
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015656559398873426,
+      "loss": 0.1143,
+      "step": 19026
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00015649016473954903,
+      "loss": 0.0549,
+      "step": 19027
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001564147521222483,
+      "loss": 0.1653,
+      "step": 19028
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001563393561383185,
+      "loss": 0.0933,
+      "step": 19029
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015626397678924643,
+      "loss": 0.0817,
+      "step": 19030
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001561886140765183,
+      "loss": 0.0798,
+      "step": 19031
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015611326800162018,
+      "loss": 0.06,
+      "step": 19032
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015603793856603732,
+      "loss": 0.1005,
+      "step": 19033
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015596262577125553,
+      "loss": 0.0644,
+      "step": 19034
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001558873296187594,
+      "loss": 0.0642,
+      "step": 19035
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015581205011003395,
+      "loss": 0.0924,
+      "step": 19036
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015573678724656336,
+      "loss": 0.1357,
+      "step": 19037
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015566154102983167,
+      "loss": 0.1018,
+      "step": 19038
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001555863114613223,
+      "loss": 0.0726,
+      "step": 19039
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015551109854251854,
+      "loss": 0.0635,
+      "step": 19040
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001554359022749039,
+      "loss": 0.1301,
+      "step": 19041
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015536072265996082,
+      "loss": 0.0601,
+      "step": 19042
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015528555969917146,
+      "loss": 0.0836,
+      "step": 19043
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015521041339401775,
+      "loss": 0.0833,
+      "step": 19044
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015513528374598196,
+      "loss": 0.0699,
+      "step": 19045
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001550601707565451,
+      "loss": 0.0519,
+      "step": 19046
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001549850744271879,
+      "loss": 0.0579,
+      "step": 19047
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015490999475939172,
+      "loss": 0.0984,
+      "step": 19048
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001548349317546365,
+      "loss": 0.1154,
+      "step": 19049
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015475988541440244,
+      "loss": 0.0507,
+      "step": 19050
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015468485574016933,
+      "loss": 0.0872,
+      "step": 19051
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015460984273341639,
+      "loss": 0.075,
+      "step": 19052
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015453484639562276,
+      "loss": 0.0773,
+      "step": 19053
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015445986672826695,
+      "loss": 0.1805,
+      "step": 19054
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015438490373282775,
+      "loss": 0.0906,
+      "step": 19055
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015430995741078302,
+      "loss": 0.0688,
+      "step": 19056
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015423502776361065,
+      "loss": 0.0847,
+      "step": 19057
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015416011479278812,
+      "loss": 0.1149,
+      "step": 19058
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001540852184997924,
+      "loss": 0.1208,
+      "step": 19059
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015401033888609995,
+      "loss": 0.1007,
+      "step": 19060
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001539354759531878,
+      "loss": 0.0944,
+      "step": 19061
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015386062970253178,
+      "loss": 0.116,
+      "step": 19062
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015378580013560762,
+      "loss": 0.1047,
+      "step": 19063
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015371098725389088,
+      "loss": 0.1037,
+      "step": 19064
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015363619105885628,
+      "loss": 0.0793,
+      "step": 19065
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001535614115519792,
+      "loss": 0.1314,
+      "step": 19066
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015348664873473373,
+      "loss": 0.0802,
+      "step": 19067
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015341190260859428,
+      "loss": 0.0839,
+      "step": 19068
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015333717317503448,
+      "loss": 0.0792,
+      "step": 19069
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015326246043552782,
+      "loss": 0.0923,
+      "step": 19070
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015318776439154748,
+      "loss": 0.0842,
+      "step": 19071
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00015311308504456622,
+      "loss": 0.0886,
+      "step": 19072
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001530384223960566,
+      "loss": 0.075,
+      "step": 19073
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001529637764474904,
+      "loss": 0.0607,
+      "step": 19074
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015288914720034007,
+      "loss": 0.0878,
+      "step": 19075
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015281453465607688,
+      "loss": 0.0828,
+      "step": 19076
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015273993881617166,
+      "loss": 0.0685,
+      "step": 19077
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015266535968209573,
+      "loss": 0.1312,
+      "step": 19078
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015259079725531943,
+      "loss": 0.1201,
+      "step": 19079
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015251625153731274,
+      "loss": 0.0645,
+      "step": 19080
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015244172252954592,
+      "loss": 0.1781,
+      "step": 19081
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001523672102334882,
+      "loss": 0.0834,
+      "step": 19082
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001522927146506089,
+      "loss": 0.0745,
+      "step": 19083
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001522182357823768,
+      "loss": 0.0894,
+      "step": 19084
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015214377363026056,
+      "loss": 0.1035,
+      "step": 19085
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015206932819572793,
+      "loss": 0.1163,
+      "step": 19086
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015199489948024747,
+      "loss": 0.1038,
+      "step": 19087
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015192048748528607,
+      "loss": 0.0796,
+      "step": 19088
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001518460922123116,
+      "loss": 0.0833,
+      "step": 19089
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015177171366279052,
+      "loss": 0.0905,
+      "step": 19090
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015169735183818956,
+      "loss": 0.072,
+      "step": 19091
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015162300673997488,
+      "loss": 0.0748,
+      "step": 19092
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001515486783696124,
+      "loss": 0.0588,
+      "step": 19093
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001514743667285674,
+      "loss": 0.0894,
+      "step": 19094
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001514000718183056,
+      "loss": 0.0887,
+      "step": 19095
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015132579364029176,
+      "loss": 0.0859,
+      "step": 19096
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015125153219599008,
+      "loss": 0.0967,
+      "step": 19097
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015117728748686532,
+      "loss": 0.0924,
+      "step": 19098
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001511030595143813,
+      "loss": 0.1138,
+      "step": 19099
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001510288482800012,
+      "loss": 0.0651,
+      "step": 19100
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001509546537851889,
+      "loss": 0.0917,
+      "step": 19101
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015088047603140686,
+      "loss": 0.1306,
+      "step": 19102
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.000150806315020118,
+      "loss": 0.0427,
+      "step": 19103
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015073217075278422,
+      "loss": 0.083,
+      "step": 19104
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015065804323086785,
+      "loss": 0.0635,
+      "step": 19105
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015058393245583014,
+      "loss": 0.0767,
+      "step": 19106
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015050983842913236,
+      "loss": 0.1373,
+      "step": 19107
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001504357611522359,
+      "loss": 0.0592,
+      "step": 19108
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015036170062660082,
+      "loss": 0.1035,
+      "step": 19109
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00015028765685368796,
+      "loss": 0.0927,
+      "step": 19110
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001502136298349569,
+      "loss": 0.0903,
+      "step": 19111
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001501396195718675,
+      "loss": 0.0771,
+      "step": 19112
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001500656260658788,
+      "loss": 0.0591,
+      "step": 19113
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014999164931844976,
+      "loss": 0.1401,
+      "step": 19114
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014991768933103933,
+      "loss": 0.0865,
+      "step": 19115
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014984374610510566,
+      "loss": 0.0945,
+      "step": 19116
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014976981964210666,
+      "loss": 0.0778,
+      "step": 19117
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001496959099434999,
+      "loss": 0.0977,
+      "step": 19118
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014962201701074286,
+      "loss": 0.0475,
+      "step": 19119
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014954814084529232,
+      "loss": 0.0688,
+      "step": 19120
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001494742814486053,
+      "loss": 0.0582,
+      "step": 19121
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014940043882213783,
+      "loss": 0.097,
+      "step": 19122
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014932661296734607,
+      "loss": 0.0527,
+      "step": 19123
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014925280388568553,
+      "loss": 0.0808,
+      "step": 19124
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014917901157861146,
+      "loss": 0.1056,
+      "step": 19125
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001491052360475792,
+      "loss": 0.0674,
+      "step": 19126
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014903147729404277,
+      "loss": 0.0541,
+      "step": 19127
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014895773531945722,
+      "loss": 0.0709,
+      "step": 19128
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001488840101252763,
+      "loss": 0.0739,
+      "step": 19129
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001488103017129535,
+      "loss": 0.0778,
+      "step": 19130
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014873661008394246,
+      "loss": 0.0717,
+      "step": 19131
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.000148662935239696,
+      "loss": 0.1115,
+      "step": 19132
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014858927718166694,
+      "loss": 0.1008,
+      "step": 19133
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014851563591130722,
+      "loss": 0.0826,
+      "step": 19134
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014844201143006952,
+      "loss": 0.0736,
+      "step": 19135
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014836840373940518,
+      "loss": 0.121,
+      "step": 19136
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014829481284076552,
+      "loss": 0.071,
+      "step": 19137
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014822123873560168,
+      "loss": 0.1658,
+      "step": 19138
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014814768142536395,
+      "loss": 0.1383,
+      "step": 19139
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001480741409115034,
+      "loss": 0.1046,
+      "step": 19140
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001480006171954693,
+      "loss": 0.0768,
+      "step": 19141
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001479271102787122,
+      "loss": 0.0706,
+      "step": 19142
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001478536201626809,
+      "loss": 0.0516,
+      "step": 19143
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014778014684882458,
+      "loss": 0.0892,
+      "step": 19144
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014770669033859184,
+      "loss": 0.0894,
+      "step": 19145
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014763325063343126,
+      "loss": 0.1279,
+      "step": 19146
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014755982773479047,
+      "loss": 0.0571,
+      "step": 19147
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014748642164411784,
+      "loss": 0.1021,
+      "step": 19148
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001474130323628603,
+      "loss": 0.1181,
+      "step": 19149
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014733965989246502,
+      "loss": 0.1356,
+      "step": 19150
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014726630423437848,
+      "loss": 0.0634,
+      "step": 19151
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014719296539004745,
+      "loss": 0.0957,
+      "step": 19152
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014711964336091787,
+      "loss": 0.1042,
+      "step": 19153
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014704633814843527,
+      "loss": 0.1241,
+      "step": 19154
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014697304975404534,
+      "loss": 0.0811,
+      "step": 19155
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.000146899778179193,
+      "loss": 0.1478,
+      "step": 19156
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.000146826523425323,
+      "loss": 0.067,
+      "step": 19157
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001467532854938798,
+      "loss": 0.1373,
+      "step": 19158
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001466800643863072,
+      "loss": 0.0635,
+      "step": 19159
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.000146606860104049,
+      "loss": 0.0521,
+      "step": 19160
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.000146533672648549,
+      "loss": 0.0464,
+      "step": 19161
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001464605020212496,
+      "loss": 0.0913,
+      "step": 19162
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014638734822359435,
+      "loss": 0.075,
+      "step": 19163
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014631421125702515,
+      "loss": 0.1226,
+      "step": 19164
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014624109112298413,
+      "loss": 0.0838,
+      "step": 19165
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014616798782291317,
+      "loss": 0.0826,
+      "step": 19166
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00014609490135825332,
+      "loss": 0.0532,
+      "step": 19167
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014602183173044613,
+      "loss": 0.0641,
+      "step": 19168
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014594877894093217,
+      "loss": 0.0767,
+      "step": 19169
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001458757429911518,
+      "loss": 0.0682,
+      "step": 19170
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014580272388254513,
+      "loss": 0.0865,
+      "step": 19171
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014572972161655174,
+      "loss": 0.0425,
+      "step": 19172
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001456567361946114,
+      "loss": 0.1063,
+      "step": 19173
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014558376761816285,
+      "loss": 0.0928,
+      "step": 19174
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014551081588864523,
+      "loss": 0.1345,
+      "step": 19175
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014543788100749677,
+      "loss": 0.0723,
+      "step": 19176
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014536496297615554,
+      "loss": 0.0743,
+      "step": 19177
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014529206179605936,
+      "loss": 0.0767,
+      "step": 19178
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014521917746864555,
+      "loss": 0.0683,
+      "step": 19179
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014514630999535094,
+      "loss": 0.0679,
+      "step": 19180
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014507345937761295,
+      "loss": 0.149,
+      "step": 19181
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014500062561686767,
+      "loss": 0.0545,
+      "step": 19182
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014492780871455102,
+      "loss": 0.0875,
+      "step": 19183
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001448550086720991,
+      "loss": 0.1392,
+      "step": 19184
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014478222549094721,
+      "loss": 0.1136,
+      "step": 19185
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014470945917253043,
+      "loss": 0.0815,
+      "step": 19186
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014463670971828324,
+      "loss": 0.0743,
+      "step": 19187
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014456397712964063,
+      "loss": 0.0693,
+      "step": 19188
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014449126140803648,
+      "loss": 0.0757,
+      "step": 19189
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001444185625549045,
+      "loss": 0.0745,
+      "step": 19190
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001443458805716782,
+      "loss": 0.0577,
+      "step": 19191
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014427321545979033,
+      "loss": 0.0787,
+      "step": 19192
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001442005672206742,
+      "loss": 0.1292,
+      "step": 19193
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014412793585576178,
+      "loss": 0.0925,
+      "step": 19194
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001440553213664856,
+      "loss": 0.1548,
+      "step": 19195
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014398272375427723,
+      "loss": 0.0514,
+      "step": 19196
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014391014302056826,
+      "loss": 0.0781,
+      "step": 19197
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014383757916678952,
+      "loss": 0.1263,
+      "step": 19198
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014376503219437197,
+      "loss": 0.1201,
+      "step": 19199
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014369250210474583,
+      "loss": 0.0522,
+      "step": 19200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001436199888993417,
+      "loss": 0.0756,
+      "step": 19201
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014354749257958888,
+      "loss": 0.0875,
+      "step": 19202
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014347501314691702,
+      "loss": 0.1572,
+      "step": 19203
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001434025506027551,
+      "loss": 0.0742,
+      "step": 19204
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014333010494853204,
+      "loss": 0.1147,
+      "step": 19205
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014325767618567642,
+      "loss": 0.1127,
+      "step": 19206
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014318526431561585,
+      "loss": 0.0856,
+      "step": 19207
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014311286933977864,
+      "loss": 0.0973,
+      "step": 19208
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014304049125959207,
+      "loss": 0.0868,
+      "step": 19209
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014296813007648313,
+      "loss": 0.0939,
+      "step": 19210
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014289578579187878,
+      "loss": 0.1681,
+      "step": 19211
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014282345840720535,
+      "loss": 0.1374,
+      "step": 19212
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00014275114792388865,
+      "loss": 0.1265,
+      "step": 19213
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014267885434335515,
+      "loss": 0.0875,
+      "step": 19214
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014260657766702967,
+      "loss": 0.0849,
+      "step": 19215
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014253431789633774,
+      "loss": 0.0697,
+      "step": 19216
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014246207503270403,
+      "loss": 0.0873,
+      "step": 19217
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014238984907755293,
+      "loss": 0.1108,
+      "step": 19218
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014231764003230852,
+      "loss": 0.1093,
+      "step": 19219
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014224544789839444,
+      "loss": 0.0694,
+      "step": 19220
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014217327267723457,
+      "loss": 0.0714,
+      "step": 19221
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014210111437025174,
+      "loss": 0.0878,
+      "step": 19222
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014202897297886874,
+      "loss": 0.1158,
+      "step": 19223
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014195684850450808,
+      "loss": 0.0928,
+      "step": 19224
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001418847409485915,
+      "loss": 0.0985,
+      "step": 19225
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014181265031254142,
+      "loss": 0.0777,
+      "step": 19226
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014174057659777873,
+      "loss": 0.12,
+      "step": 19227
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014166851980572503,
+      "loss": 0.082,
+      "step": 19228
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001415964799378009,
+      "loss": 0.0627,
+      "step": 19229
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001415244569954267,
+      "loss": 0.1074,
+      "step": 19230
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014145245098002259,
+      "loss": 0.0808,
+      "step": 19231
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014138046189300825,
+      "loss": 0.0773,
+      "step": 19232
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014130848973580314,
+      "loss": 0.0737,
+      "step": 19233
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014123653450982655,
+      "loss": 0.0975,
+      "step": 19234
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014116459621649714,
+      "loss": 0.0691,
+      "step": 19235
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014109267485723308,
+      "loss": 0.0711,
+      "step": 19236
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014102077043345307,
+      "loss": 0.083,
+      "step": 19237
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014094888294657458,
+      "loss": 0.1147,
+      "step": 19238
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.000140877012398015,
+      "loss": 0.1649,
+      "step": 19239
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001408051587891913,
+      "loss": 0.069,
+      "step": 19240
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014073332212152057,
+      "loss": 0.0723,
+      "step": 19241
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014066150239641919,
+      "loss": 0.1099,
+      "step": 19242
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.000140589699615303,
+      "loss": 0.0812,
+      "step": 19243
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014051791377958813,
+      "loss": 0.0415,
+      "step": 19244
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014044614489068974,
+      "loss": 0.0782,
+      "step": 19245
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014037439295002286,
+      "loss": 0.1332,
+      "step": 19246
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014030265795900233,
+      "loss": 0.0674,
+      "step": 19247
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001402309399190429,
+      "loss": 0.0782,
+      "step": 19248
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014015923883155845,
+      "loss": 0.1023,
+      "step": 19249
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014008755469796286,
+      "loss": 0.0991,
+      "step": 19250
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001400158875196693,
+      "loss": 0.0743,
+      "step": 19251
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00013994423729809104,
+      "loss": 0.0612,
+      "step": 19252
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00013987260403464053,
+      "loss": 0.1125,
+      "step": 19253
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001398009877307308,
+      "loss": 0.1799,
+      "step": 19254
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001397293883877735,
+      "loss": 0.1234,
+      "step": 19255
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00013965780600718058,
+      "loss": 0.0864,
+      "step": 19256
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00013958624059036306,
+      "loss": 0.1102,
+      "step": 19257
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001395146921387327,
+      "loss": 0.0943,
+      "step": 19258
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00013944316065369976,
+      "loss": 0.1464,
+      "step": 19259
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013937164613667474,
+      "loss": 0.0663,
+      "step": 19260
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013930014858906791,
+      "loss": 0.1062,
+      "step": 19261
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013922866801228896,
+      "loss": 0.1334,
+      "step": 19262
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013915720440774726,
+      "loss": 0.1147,
+      "step": 19263
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013908575777685185,
+      "loss": 0.0845,
+      "step": 19264
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013901432812101156,
+      "loss": 0.0754,
+      "step": 19265
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013894291544163463,
+      "loss": 0.0665,
+      "step": 19266
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013887151974012913,
+      "loss": 0.1301,
+      "step": 19267
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013880014101790296,
+      "loss": 0.0918,
+      "step": 19268
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013872877927636362,
+      "loss": 0.1113,
+      "step": 19269
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013865743451691814,
+      "loss": 0.0782,
+      "step": 19270
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013858610674097316,
+      "loss": 0.085,
+      "step": 19271
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001385147959499351,
+      "loss": 0.1361,
+      "step": 19272
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013844350214520986,
+      "loss": 0.0979,
+      "step": 19273
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001383722253282036,
+      "loss": 0.0859,
+      "step": 19274
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013830096550032135,
+      "loss": 0.0803,
+      "step": 19275
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013822972266296841,
+      "loss": 0.1561,
+      "step": 19276
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013815849681754933,
+      "loss": 0.08,
+      "step": 19277
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013808728796546832,
+      "loss": 0.0646,
+      "step": 19278
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013801609610812992,
+      "loss": 0.1188,
+      "step": 19279
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013794492124693736,
+      "loss": 0.0834,
+      "step": 19280
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001378737633832946,
+      "loss": 0.1121,
+      "step": 19281
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001378026225186043,
+      "loss": 0.0685,
+      "step": 19282
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013773149865426926,
+      "loss": 0.0752,
+      "step": 19283
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013766039179169198,
+      "loss": 0.0925,
+      "step": 19284
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013758930193227427,
+      "loss": 0.099,
+      "step": 19285
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013751822907741794,
+      "loss": 0.0754,
+      "step": 19286
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013744717322852429,
+      "loss": 0.1077,
+      "step": 19287
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001373761343869946,
+      "loss": 0.0757,
+      "step": 19288
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013730511255422918,
+      "loss": 0.0964,
+      "step": 19289
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013723410773162892,
+      "loss": 0.0656,
+      "step": 19290
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001371631199205936,
+      "loss": 0.0809,
+      "step": 19291
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013709214912252288,
+      "loss": 0.0606,
+      "step": 19292
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013702119533881595,
+      "loss": 0.1146,
+      "step": 19293
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013695025857087228,
+      "loss": 0.076,
+      "step": 19294
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001368793388200904,
+      "loss": 0.0707,
+      "step": 19295
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013680843608786852,
+      "loss": 0.0827,
+      "step": 19296
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013673755037560477,
+      "loss": 0.0878,
+      "step": 19297
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013666668168469675,
+      "loss": 0.0927,
+      "step": 19298
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013659583001654173,
+      "loss": 0.0861,
+      "step": 19299
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013652499537253682,
+      "loss": 0.0648,
+      "step": 19300
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013645417775407897,
+      "loss": 0.0848,
+      "step": 19301
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013638337716256432,
+      "loss": 0.0743,
+      "step": 19302
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013631259359938886,
+      "loss": 0.121,
+      "step": 19303
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00013624182706594833,
+      "loss": 0.069,
+      "step": 19304
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001361710775636379,
+      "loss": 0.0501,
+      "step": 19305
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013610034509385272,
+      "loss": 0.0992,
+      "step": 19306
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013602962965798715,
+      "loss": 0.0794,
+      "step": 19307
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001359589312574361,
+      "loss": 0.0767,
+      "step": 19308
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013588824989359317,
+      "loss": 0.0806,
+      "step": 19309
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013581758556785185,
+      "loss": 0.0891,
+      "step": 19310
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013574693828160601,
+      "loss": 0.1128,
+      "step": 19311
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001356763080362482,
+      "loss": 0.0691,
+      "step": 19312
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001356056948331711,
+      "loss": 0.0714,
+      "step": 19313
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013553509867376734,
+      "loss": 0.0814,
+      "step": 19314
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013546451955942862,
+      "loss": 0.1385,
+      "step": 19315
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001353939574915466,
+      "loss": 0.0748,
+      "step": 19316
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001353234124715127,
+      "loss": 0.0696,
+      "step": 19317
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013525288450071782,
+      "loss": 0.1111,
+      "step": 19318
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013518237358055264,
+      "loss": 0.1154,
+      "step": 19319
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013511187971240713,
+      "loss": 0.1351,
+      "step": 19320
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013504140289767152,
+      "loss": 0.1008,
+      "step": 19321
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013497094313773573,
+      "loss": 0.0777,
+      "step": 19322
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001349005004339887,
+      "loss": 0.1046,
+      "step": 19323
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001348300747878194,
+      "loss": 0.0698,
+      "step": 19324
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013475966620061665,
+      "loss": 0.0805,
+      "step": 19325
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013468927467376847,
+      "loss": 0.1047,
+      "step": 19326
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013461890020866264,
+      "loss": 0.095,
+      "step": 19327
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001345485428066874,
+      "loss": 0.0654,
+      "step": 19328
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013447820246922947,
+      "loss": 0.1123,
+      "step": 19329
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013440787919767605,
+      "loss": 0.1748,
+      "step": 19330
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013433757299341343,
+      "loss": 0.0937,
+      "step": 19331
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001342672838578284,
+      "loss": 0.0613,
+      "step": 19332
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013419701179230624,
+      "loss": 0.0892,
+      "step": 19333
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013412675679823317,
+      "loss": 0.1053,
+      "step": 19334
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001340565188769941,
+      "loss": 0.0687,
+      "step": 19335
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013398629802997396,
+      "loss": 0.093,
+      "step": 19336
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013391609425855743,
+      "loss": 0.1319,
+      "step": 19337
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013384590756412872,
+      "loss": 0.1054,
+      "step": 19338
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013377573794807164,
+      "loss": 0.0925,
+      "step": 19339
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001337055854117696,
+      "loss": 0.1145,
+      "step": 19340
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013363544995660627,
+      "loss": 0.0648,
+      "step": 19341
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000133565331583964,
+      "loss": 0.0773,
+      "step": 19342
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013349523029522603,
+      "loss": 0.0747,
+      "step": 19343
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013342514609177415,
+      "loss": 0.0981,
+      "step": 19344
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013335507897499032,
+      "loss": 0.0809,
+      "step": 19345
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013328502894625606,
+      "loss": 0.0615,
+      "step": 19346
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001332149960069524,
+      "loss": 0.0662,
+      "step": 19347
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013314498015846066,
+      "loss": 0.0918,
+      "step": 19348
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013307498140216101,
+      "loss": 0.0464,
+      "step": 19349
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001330049997394338,
+      "loss": 0.0748,
+      "step": 19350
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00013293503517165895,
+      "loss": 0.063,
+      "step": 19351
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013286508770021555,
+      "loss": 0.0942,
+      "step": 19352
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001327951573264834,
+      "loss": 0.0944,
+      "step": 19353
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013272524405184095,
+      "loss": 0.0925,
+      "step": 19354
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013265534787766696,
+      "loss": 0.062,
+      "step": 19355
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001325854688053395,
+      "loss": 0.0808,
+      "step": 19356
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001325156068362363,
+      "loss": 0.074,
+      "step": 19357
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013244576197173508,
+      "loss": 0.0878,
+      "step": 19358
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013237593421321292,
+      "loss": 0.1023,
+      "step": 19359
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013230612356204629,
+      "loss": 0.0649,
+      "step": 19360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001322363300196122,
+      "loss": 0.0751,
+      "step": 19361
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013216655358728668,
+      "loss": 0.0649,
+      "step": 19362
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013209679426644515,
+      "loss": 0.0933,
+      "step": 19363
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013202705205846377,
+      "loss": 0.1425,
+      "step": 19364
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001319573269647172,
+      "loss": 0.0707,
+      "step": 19365
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001318876189865804,
+      "loss": 0.0522,
+      "step": 19366
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001318179281254276,
+      "loss": 0.1262,
+      "step": 19367
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013174825438263326,
+      "loss": 0.0875,
+      "step": 19368
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013167859775957102,
+      "loss": 0.1293,
+      "step": 19369
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013160895825761444,
+      "loss": 0.0593,
+      "step": 19370
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001315393358781365,
+      "loss": 0.0681,
+      "step": 19371
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013146973062251,
+      "loss": 0.0696,
+      "step": 19372
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001314001424921072,
+      "loss": 0.0981,
+      "step": 19373
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001313305714883005,
+      "loss": 0.1028,
+      "step": 19374
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013126101761246146,
+      "loss": 0.1028,
+      "step": 19375
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013119148086596178,
+      "loss": 0.0907,
+      "step": 19376
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001311219612501725,
+      "loss": 0.0615,
+      "step": 19377
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001310524587664641,
+      "loss": 0.075,
+      "step": 19378
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013098297341620723,
+      "loss": 0.0891,
+      "step": 19379
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013091350520077162,
+      "loss": 0.0681,
+      "step": 19380
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013084405412152743,
+      "loss": 0.0544,
+      "step": 19381
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013077462017984398,
+      "loss": 0.079,
+      "step": 19382
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013070520337709024,
+      "loss": 0.0625,
+      "step": 19383
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013063580371463468,
+      "loss": 0.1155,
+      "step": 19384
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001305664211938462,
+      "loss": 0.0631,
+      "step": 19385
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013049705581609261,
+      "loss": 0.1022,
+      "step": 19386
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013042770758274136,
+      "loss": 0.0751,
+      "step": 19387
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013035837649516026,
+      "loss": 0.0557,
+      "step": 19388
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013028906255471617,
+      "loss": 0.0684,
+      "step": 19389
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013021976576277583,
+      "loss": 0.0699,
+      "step": 19390
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013015048612070545,
+      "loss": 0.0587,
+      "step": 19391
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013008122362987118,
+      "loss": 0.0911,
+      "step": 19392
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00013001197829163848,
+      "loss": 0.0854,
+      "step": 19393
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.000129942750107373,
+      "loss": 0.0432,
+      "step": 19394
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00012987353907843946,
+      "loss": 0.1362,
+      "step": 19395
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001298043452062029,
+      "loss": 0.0811,
+      "step": 19396
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00012973516849202737,
+      "loss": 0.0754,
+      "step": 19397
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012966600893727699,
+      "loss": 0.0569,
+      "step": 19398
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012959686654331526,
+      "loss": 0.1144,
+      "step": 19399
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012952774131150548,
+      "loss": 0.1222,
+      "step": 19400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012945863324321084,
+      "loss": 0.0999,
+      "step": 19401
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012938954233979383,
+      "loss": 0.0952,
+      "step": 19402
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012932046860261693,
+      "loss": 0.0835,
+      "step": 19403
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001292514120330418,
+      "loss": 0.114,
+      "step": 19404
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012918237263243005,
+      "loss": 0.0721,
+      "step": 19405
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012911335040214334,
+      "loss": 0.0962,
+      "step": 19406
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012904434534354215,
+      "loss": 0.078,
+      "step": 19407
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001289753574579875,
+      "loss": 0.0526,
+      "step": 19408
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012890638674683953,
+      "loss": 0.0875,
+      "step": 19409
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012883743321145813,
+      "loss": 0.1021,
+      "step": 19410
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012876849685320292,
+      "loss": 0.0852,
+      "step": 19411
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001286995776734331,
+      "loss": 0.0891,
+      "step": 19412
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001286306756735074,
+      "loss": 0.1067,
+      "step": 19413
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001285617908547847,
+      "loss": 0.066,
+      "step": 19414
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012849292321862315,
+      "loss": 0.0837,
+      "step": 19415
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012842407276638046,
+      "loss": 0.1647,
+      "step": 19416
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012835523949941463,
+      "loss": 0.0834,
+      "step": 19417
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012828642341908248,
+      "loss": 0.099,
+      "step": 19418
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012821762452674103,
+      "loss": 0.0908,
+      "step": 19419
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001281488428237466,
+      "loss": 0.0804,
+      "step": 19420
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012808007831145585,
+      "loss": 0.0872,
+      "step": 19421
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012801133099122432,
+      "loss": 0.0714,
+      "step": 19422
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012794260086440758,
+      "loss": 0.0739,
+      "step": 19423
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012787388793236078,
+      "loss": 0.0922,
+      "step": 19424
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012780519219643893,
+      "loss": 0.0979,
+      "step": 19425
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001277365136579961,
+      "loss": 0.1462,
+      "step": 19426
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012766785231838707,
+      "loss": 0.0633,
+      "step": 19427
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001275992081789652,
+      "loss": 0.1023,
+      "step": 19428
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001275305812410843,
+      "loss": 0.1256,
+      "step": 19429
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001274619715060974,
+      "loss": 0.09,
+      "step": 19430
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012739337897535718,
+      "loss": 0.0658,
+      "step": 19431
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012732480365021636,
+      "loss": 0.1342,
+      "step": 19432
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012725624553202664,
+      "loss": 0.0665,
+      "step": 19433
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012718770462214035,
+      "loss": 0.0698,
+      "step": 19434
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012711918092190867,
+      "loss": 0.1221,
+      "step": 19435
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001270506744326827,
+      "loss": 0.072,
+      "step": 19436
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012698218515581317,
+      "loss": 0.1114,
+      "step": 19437
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012691371309265064,
+      "loss": 0.0623,
+      "step": 19438
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001268452582445453,
+      "loss": 0.0626,
+      "step": 19439
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012677682061284658,
+      "loss": 0.0626,
+      "step": 19440
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012670840019890416,
+      "loss": 0.0698,
+      "step": 19441
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012663999700406715,
+      "loss": 0.0801,
+      "step": 19442
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00012657161102968417,
+      "loss": 0.2061,
+      "step": 19443
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012650324227710363,
+      "loss": 0.0804,
+      "step": 19444
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012643489074767368,
+      "loss": 0.1278,
+      "step": 19445
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012636655644274186,
+      "loss": 0.0656,
+      "step": 19446
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012629823936365547,
+      "loss": 0.0816,
+      "step": 19447
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012622993951176187,
+      "loss": 0.0906,
+      "step": 19448
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012616165688840753,
+      "loss": 0.0906,
+      "step": 19449
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012609339149493893,
+      "loss": 0.099,
+      "step": 19450
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012602514333270222,
+      "loss": 0.0972,
+      "step": 19451
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012595691240304286,
+      "loss": 0.1351,
+      "step": 19452
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012588869870730601,
+      "loss": 0.0631,
+      "step": 19453
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012582050224683718,
+      "loss": 0.0497,
+      "step": 19454
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012575232302298068,
+      "loss": 0.0551,
+      "step": 19455
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012568416103708103,
+      "loss": 0.1102,
+      "step": 19456
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012561601629048204,
+      "loss": 0.0924,
+      "step": 19457
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012554788878452717,
+      "loss": 0.0918,
+      "step": 19458
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012547977852056026,
+      "loss": 0.0498,
+      "step": 19459
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012541168549992367,
+      "loss": 0.0956,
+      "step": 19460
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012534360972396054,
+      "loss": 0.0825,
+      "step": 19461
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000125275551194013,
+      "loss": 0.1144,
+      "step": 19462
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012520750991142294,
+      "loss": 0.0718,
+      "step": 19463
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001251394858775319,
+      "loss": 0.056,
+      "step": 19464
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001250714790936812,
+      "loss": 0.056,
+      "step": 19465
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012500348956121177,
+      "loss": 0.0635,
+      "step": 19466
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000124935517281464,
+      "loss": 0.0602,
+      "step": 19467
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012486756225577855,
+      "loss": 0.1201,
+      "step": 19468
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012479962448549498,
+      "loss": 0.0702,
+      "step": 19469
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012473170397195288,
+      "loss": 0.1251,
+      "step": 19470
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012466380071649163,
+      "loss": 0.0443,
+      "step": 19471
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012459591472045007,
+      "loss": 0.076,
+      "step": 19472
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012452804598516655,
+      "loss": 0.079,
+      "step": 19473
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012446019451197953,
+      "loss": 0.1208,
+      "step": 19474
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012439236030222688,
+      "loss": 0.075,
+      "step": 19475
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012432454335724595,
+      "loss": 0.0746,
+      "step": 19476
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012425674367837393,
+      "loss": 0.0916,
+      "step": 19477
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012418896126694767,
+      "loss": 0.076,
+      "step": 19478
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001241211961243035,
+      "loss": 0.0827,
+      "step": 19479
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000124053448251778,
+      "loss": 0.0601,
+      "step": 19480
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012398571765070664,
+      "loss": 0.1121,
+      "step": 19481
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012391800432242506,
+      "loss": 0.084,
+      "step": 19482
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001238503082682685,
+      "loss": 0.0941,
+      "step": 19483
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012378262948957154,
+      "loss": 0.0594,
+      "step": 19484
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012371496798766867,
+      "loss": 0.1979,
+      "step": 19485
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012364732376389376,
+      "loss": 0.1101,
+      "step": 19486
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012357969681958125,
+      "loss": 0.0654,
+      "step": 19487
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.000123512087156064,
+      "loss": 0.0919,
+      "step": 19488
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00012344449477467545,
+      "loss": 0.1218,
+      "step": 19489
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001233769196767478,
+      "loss": 0.0829,
+      "step": 19490
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012330936186361418,
+      "loss": 0.0918,
+      "step": 19491
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001232418213366062,
+      "loss": 0.0914,
+      "step": 19492
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001231742980970556,
+      "loss": 0.0999,
+      "step": 19493
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012310679214629417,
+      "loss": 0.0815,
+      "step": 19494
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012303930348565251,
+      "loss": 0.0587,
+      "step": 19495
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012297183211646158,
+      "loss": 0.0793,
+      "step": 19496
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012290437804005162,
+      "loss": 0.0867,
+      "step": 19497
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012283694125775257,
+      "loss": 0.1515,
+      "step": 19498
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001227695217708944,
+      "loss": 0.1082,
+      "step": 19499
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001227021195808059,
+      "loss": 0.0724,
+      "step": 19500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001226347346888168,
+      "loss": 0.0871,
+      "step": 19501
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012256736709625516,
+      "loss": 0.0563,
+      "step": 19502
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012250001680444967,
+      "loss": 0.0878,
+      "step": 19503
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012243268381472827,
+      "loss": 0.1024,
+      "step": 19504
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012236536812841847,
+      "loss": 0.095,
+      "step": 19505
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012229806974684742,
+      "loss": 0.0733,
+      "step": 19506
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001222307886713424,
+      "loss": 0.0583,
+      "step": 19507
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012216352490323001,
+      "loss": 0.1099,
+      "step": 19508
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001220962784438363,
+      "loss": 0.0612,
+      "step": 19509
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012202904929448732,
+      "loss": 0.114,
+      "step": 19510
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012196183745650846,
+      "loss": 0.0905,
+      "step": 19511
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012189464293122532,
+      "loss": 0.1305,
+      "step": 19512
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012182746571996239,
+      "loss": 0.0602,
+      "step": 19513
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012176030582404462,
+      "loss": 0.1351,
+      "step": 19514
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012169316324479618,
+      "loss": 0.0863,
+      "step": 19515
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012162603798354088,
+      "loss": 0.0716,
+      "step": 19516
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012155893004160223,
+      "loss": 0.1048,
+      "step": 19517
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001214918394203034,
+      "loss": 0.0817,
+      "step": 19518
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001214247661209673,
+      "loss": 0.0891,
+      "step": 19519
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012135771014491626,
+      "loss": 0.0841,
+      "step": 19520
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012129067149347272,
+      "loss": 0.1118,
+      "step": 19521
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012122365016795855,
+      "loss": 0.0548,
+      "step": 19522
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001211566461696948,
+      "loss": 0.0638,
+      "step": 19523
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012108965950000328,
+      "loss": 0.1349,
+      "step": 19524
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012102269016020429,
+      "loss": 0.1127,
+      "step": 19525
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012095573815161832,
+      "loss": 0.0973,
+      "step": 19526
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012088880347556585,
+      "loss": 0.0864,
+      "step": 19527
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012082188613336642,
+      "loss": 0.0893,
+      "step": 19528
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012075498612633951,
+      "loss": 0.0841,
+      "step": 19529
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012068810345580417,
+      "loss": 0.1326,
+      "step": 19530
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012062123812307923,
+      "loss": 0.0924,
+      "step": 19531
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012055439012948277,
+      "loss": 0.0683,
+      "step": 19532
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001204875594763335,
+      "loss": 0.0881,
+      "step": 19533
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012042074616494848,
+      "loss": 0.0798,
+      "step": 19534
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012035395019664575,
+      "loss": 0.0681,
+      "step": 19535
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00012028717157274194,
+      "loss": 0.1227,
+      "step": 19536
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00012022041029455388,
+      "loss": 0.0518,
+      "step": 19537
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00012015366636339786,
+      "loss": 0.0901,
+      "step": 19538
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00012008693978059005,
+      "loss": 0.1525,
+      "step": 19539
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00012002023054744582,
+      "loss": 0.11,
+      "step": 19540
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001199535386652808,
+      "loss": 0.1089,
+      "step": 19541
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011988686413541006,
+      "loss": 0.1284,
+      "step": 19542
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011982020695914797,
+      "loss": 0.0744,
+      "step": 19543
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011975356713780883,
+      "loss": 0.0829,
+      "step": 19544
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011968694467270691,
+      "loss": 0.0925,
+      "step": 19545
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001196203395651555,
+      "loss": 0.1411,
+      "step": 19546
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011955375181646833,
+      "loss": 0.1478,
+      "step": 19547
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011948718142795789,
+      "loss": 0.1221,
+      "step": 19548
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011942062840093715,
+      "loss": 0.0705,
+      "step": 19549
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011935409273671804,
+      "loss": 0.1088,
+      "step": 19550
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011928757443661264,
+      "loss": 0.0905,
+      "step": 19551
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011922107350193257,
+      "loss": 0.0623,
+      "step": 19552
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011915458993398875,
+      "loss": 0.0616,
+      "step": 19553
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011908812373409251,
+      "loss": 0.0887,
+      "step": 19554
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011902167490355387,
+      "loss": 0.1157,
+      "step": 19555
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001189552434436837,
+      "loss": 0.0676,
+      "step": 19556
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011888882935579149,
+      "loss": 0.1245,
+      "step": 19557
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011882243264118675,
+      "loss": 0.111,
+      "step": 19558
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011875605330117867,
+      "loss": 0.0862,
+      "step": 19559
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011868969133707597,
+      "loss": 0.0712,
+      "step": 19560
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011862334675018738,
+      "loss": 0.0842,
+      "step": 19561
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011855701954182097,
+      "loss": 0.0917,
+      "step": 19562
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011849070971328457,
+      "loss": 0.1377,
+      "step": 19563
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011842441726588549,
+      "loss": 0.0701,
+      "step": 19564
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011835814220093078,
+      "loss": 0.1564,
+      "step": 19565
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011829188451972738,
+      "loss": 0.0777,
+      "step": 19566
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011822564422358195,
+      "loss": 0.0989,
+      "step": 19567
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011815942131380042,
+      "loss": 0.1204,
+      "step": 19568
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011809321579168841,
+      "loss": 0.0746,
+      "step": 19569
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011802702765855134,
+      "loss": 0.0639,
+      "step": 19570
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011796085691569436,
+      "loss": 0.1312,
+      "step": 19571
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011789470356442222,
+      "loss": 0.0973,
+      "step": 19572
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011782856760603899,
+      "loss": 0.1222,
+      "step": 19573
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011776244904184908,
+      "loss": 0.0471,
+      "step": 19574
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011769634787315608,
+      "loss": 0.0769,
+      "step": 19575
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011763026410126321,
+      "loss": 0.0798,
+      "step": 19576
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011756419772747362,
+      "loss": 0.0665,
+      "step": 19577
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011749814875309007,
+      "loss": 0.0784,
+      "step": 19578
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001174321171794146,
+      "loss": 0.1051,
+      "step": 19579
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001173661030077492,
+      "loss": 0.0937,
+      "step": 19580
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001173001062393958,
+      "loss": 0.1049,
+      "step": 19581
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011723412687565549,
+      "loss": 0.069,
+      "step": 19582
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011716816491782922,
+      "loss": 0.1122,
+      "step": 19583
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011710222036721763,
+      "loss": 0.1486,
+      "step": 19584
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011703629322512077,
+      "loss": 0.0677,
+      "step": 19585
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011697038349283895,
+      "loss": 0.1425,
+      "step": 19586
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011690449117167135,
+      "loss": 0.1385,
+      "step": 19587
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011683861626291759,
+      "loss": 0.1328,
+      "step": 19588
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001167727587678763,
+      "loss": 0.0738,
+      "step": 19589
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011670691868784622,
+      "loss": 0.0656,
+      "step": 19590
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011664109602412532,
+      "loss": 0.0886,
+      "step": 19591
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011657529077801155,
+      "loss": 0.0862,
+      "step": 19592
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011650950295080209,
+      "loss": 0.0632,
+      "step": 19593
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011644373254379481,
+      "loss": 0.0729,
+      "step": 19594
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001163779795582861,
+      "loss": 0.1165,
+      "step": 19595
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011631224399557261,
+      "loss": 0.0756,
+      "step": 19596
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011624652585695006,
+      "loss": 0.1155,
+      "step": 19597
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011618082514371475,
+      "loss": 0.0887,
+      "step": 19598
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001161151418571621,
+      "loss": 0.0332,
+      "step": 19599
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011604947599858672,
+      "loss": 0.068,
+      "step": 19600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011598382756928394,
+      "loss": 0.0621,
+      "step": 19601
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011591819657054803,
+      "loss": 0.0641,
+      "step": 19602
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011585258300367297,
+      "loss": 0.0613,
+      "step": 19603
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011578698686995248,
+      "loss": 0.0865,
+      "step": 19604
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011572140817068,
+      "loss": 0.0566,
+      "step": 19605
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011565584690714836,
+      "loss": 0.0641,
+      "step": 19606
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011559030308065077,
+      "loss": 0.1104,
+      "step": 19607
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011552477669247896,
+      "loss": 0.1183,
+      "step": 19608
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011545926774392556,
+      "loss": 0.0707,
+      "step": 19609
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.000115393776236282,
+      "loss": 0.1288,
+      "step": 19610
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011532830217083945,
+      "loss": 0.0704,
+      "step": 19611
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001152628455488891,
+      "loss": 0.1005,
+      "step": 19612
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011519740637172127,
+      "loss": 0.0627,
+      "step": 19613
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011513198464062669,
+      "loss": 0.1077,
+      "step": 19614
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011506658035689499,
+      "loss": 0.1104,
+      "step": 19615
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011500119352181604,
+      "loss": 0.0771,
+      "step": 19616
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011493582413667891,
+      "loss": 0.1089,
+      "step": 19617
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011487047220277236,
+      "loss": 0.1134,
+      "step": 19618
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011480513772138534,
+      "loss": 0.1119,
+      "step": 19619
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011473982069380573,
+      "loss": 0.1365,
+      "step": 19620
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011467452112132182,
+      "loss": 0.0829,
+      "step": 19621
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011460923900522091,
+      "loss": 0.1016,
+      "step": 19622
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001145439743467902,
+      "loss": 0.137,
+      "step": 19623
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011447872714731656,
+      "loss": 0.1257,
+      "step": 19624
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011441349740808648,
+      "loss": 0.0858,
+      "step": 19625
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011434828513038597,
+      "loss": 0.0717,
+      "step": 19626
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011428309031550132,
+      "loss": 0.0911,
+      "step": 19627
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011421791296471761,
+      "loss": 0.0587,
+      "step": 19628
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011415275307931993,
+      "loss": 0.0727,
+      "step": 19629
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011408761066059336,
+      "loss": 0.0786,
+      "step": 19630
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011402248570982222,
+      "loss": 0.1167,
+      "step": 19631
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011395737822829067,
+      "loss": 0.0745,
+      "step": 19632
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011389228821728203,
+      "loss": 0.0992,
+      "step": 19633
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011382721567808041,
+      "loss": 0.0603,
+      "step": 19634
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011376216061196843,
+      "loss": 0.1145,
+      "step": 19635
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011369712302022906,
+      "loss": 0.0975,
+      "step": 19636
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011363210290414438,
+      "loss": 0.13,
+      "step": 19637
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011356710026499661,
+      "loss": 0.0539,
+      "step": 19638
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011350211510406727,
+      "loss": 0.0739,
+      "step": 19639
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011343714742263778,
+      "loss": 0.098,
+      "step": 19640
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011337219722198932,
+      "loss": 0.1354,
+      "step": 19641
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011330726450340256,
+      "loss": 0.0561,
+      "step": 19642
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011324234926815758,
+      "loss": 0.0777,
+      "step": 19643
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011317745151753445,
+      "loss": 0.1019,
+      "step": 19644
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011311257125281282,
+      "loss": 0.0749,
+      "step": 19645
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011304770847527168,
+      "loss": 0.067,
+      "step": 19646
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011298286318619045,
+      "loss": 0.0936,
+      "step": 19647
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011291803538684731,
+      "loss": 0.0748,
+      "step": 19648
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001128532250785208,
+      "loss": 0.0811,
+      "step": 19649
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011278843226248835,
+      "loss": 0.0588,
+      "step": 19650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011272365694002817,
+      "loss": 0.0493,
+      "step": 19651
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.000112658899112417,
+      "loss": 0.0705,
+      "step": 19652
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011259415878093171,
+      "loss": 0.1519,
+      "step": 19653
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011252943594684906,
+      "loss": 0.0748,
+      "step": 19654
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011246473061144514,
+      "loss": 0.0455,
+      "step": 19655
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001124000427759957,
+      "loss": 0.0869,
+      "step": 19656
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011233537244177627,
+      "loss": 0.0465,
+      "step": 19657
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011227071961006196,
+      "loss": 0.062,
+      "step": 19658
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011220608428212764,
+      "loss": 0.0716,
+      "step": 19659
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011214146645924739,
+      "loss": 0.1166,
+      "step": 19660
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011207686614269563,
+      "loss": 0.0963,
+      "step": 19661
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011201228333374625,
+      "loss": 0.0682,
+      "step": 19662
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011194771803367254,
+      "loss": 0.1205,
+      "step": 19663
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001118831702437475,
+      "loss": 0.1057,
+      "step": 19664
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011181863996524389,
+      "loss": 0.1249,
+      "step": 19665
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011175412719943378,
+      "loss": 0.0862,
+      "step": 19666
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011168963194758974,
+      "loss": 0.1044,
+      "step": 19667
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011162515421098318,
+      "loss": 0.1088,
+      "step": 19668
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011156069399088553,
+      "loss": 0.0797,
+      "step": 19669
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011149625128856766,
+      "loss": 0.0646,
+      "step": 19670
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011143182610530001,
+      "loss": 0.0753,
+      "step": 19671
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011136741844235343,
+      "loss": 0.1077,
+      "step": 19672
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011130302830099737,
+      "loss": 0.0665,
+      "step": 19673
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001112386556825018,
+      "loss": 0.0627,
+      "step": 19674
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011117430058813582,
+      "loss": 0.1028,
+      "step": 19675
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011110996301916842,
+      "loss": 0.077,
+      "step": 19676
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011104564297686815,
+      "loss": 0.1566,
+      "step": 19677
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011098134046250319,
+      "loss": 0.0748,
+      "step": 19678
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011091705547734144,
+      "loss": 0.1033,
+      "step": 19679
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011085278802265031,
+      "loss": 0.0931,
+      "step": 19680
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011078853809969724,
+      "loss": 0.119,
+      "step": 19681
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011072430570974878,
+      "loss": 0.069,
+      "step": 19682
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001106600908540718,
+      "loss": 0.1012,
+      "step": 19683
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011059589353393229,
+      "loss": 0.0774,
+      "step": 19684
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011053171375059601,
+      "loss": 0.0602,
+      "step": 19685
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011046755150532827,
+      "loss": 0.0533,
+      "step": 19686
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011040340679939454,
+      "loss": 0.1031,
+      "step": 19687
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011033927963405943,
+      "loss": 0.0598,
+      "step": 19688
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001102751700105874,
+      "loss": 0.0997,
+      "step": 19689
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011021107793024244,
+      "loss": 0.0564,
+      "step": 19690
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001101470033942884,
+      "loss": 0.1807,
+      "step": 19691
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011008294640398831,
+      "loss": 0.088,
+      "step": 19692
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011001890696060557,
+      "loss": 0.0737,
+      "step": 19693
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010995488506540296,
+      "loss": 0.0521,
+      "step": 19694
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001098908807196426,
+      "loss": 0.0963,
+      "step": 19695
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010982689392458667,
+      "loss": 0.0631,
+      "step": 19696
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010976292468149673,
+      "loss": 0.0574,
+      "step": 19697
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.000109698972991634,
+      "loss": 0.1116,
+      "step": 19698
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001096350388562597,
+      "loss": 0.0898,
+      "step": 19699
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010957112227663391,
+      "loss": 0.0975,
+      "step": 19700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010950722325401763,
+      "loss": 0.1045,
+      "step": 19701
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001094433417896703,
+      "loss": 0.085,
+      "step": 19702
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010937947788485148,
+      "loss": 0.1746,
+      "step": 19703
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001093156315408208,
+      "loss": 0.0923,
+      "step": 19704
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010925180275883684,
+      "loss": 0.1,
+      "step": 19705
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010918799154015812,
+      "loss": 0.0831,
+      "step": 19706
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.000109124197886043,
+      "loss": 0.0689,
+      "step": 19707
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010906042179774933,
+      "loss": 0.1018,
+      "step": 19708
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010899666327653456,
+      "loss": 0.1556,
+      "step": 19709
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010893292232365581,
+      "loss": 0.0759,
+      "step": 19710
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010886919894036995,
+      "loss": 0.0993,
+      "step": 19711
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001088054931279333,
+      "loss": 0.087,
+      "step": 19712
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.000108741804887602,
+      "loss": 0.0857,
+      "step": 19713
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010867813422063178,
+      "loss": 0.0726,
+      "step": 19714
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010861448112827854,
+      "loss": 0.063,
+      "step": 19715
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010855084561179685,
+      "loss": 0.0814,
+      "step": 19716
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010848722767244168,
+      "loss": 0.0734,
+      "step": 19717
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010842362731146738,
+      "loss": 0.0866,
+      "step": 19718
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010836004453012782,
+      "loss": 0.0574,
+      "step": 19719
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00010829647932967668,
+      "loss": 0.093,
+      "step": 19720
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010823293171136761,
+      "loss": 0.139,
+      "step": 19721
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001081694016764535,
+      "loss": 0.0981,
+      "step": 19722
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010810588922618692,
+      "loss": 0.0561,
+      "step": 19723
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010804239436181995,
+      "loss": 0.0715,
+      "step": 19724
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010797891708460505,
+      "loss": 0.0839,
+      "step": 19725
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010791545739579333,
+      "loss": 0.0613,
+      "step": 19726
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010785201529663646,
+      "loss": 0.0812,
+      "step": 19727
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010778859078838533,
+      "loss": 0.0995,
+      "step": 19728
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010772518387229024,
+      "loss": 0.0529,
+      "step": 19729
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010766179454960157,
+      "loss": 0.1425,
+      "step": 19730
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010759842282156928,
+      "loss": 0.0692,
+      "step": 19731
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010753506868944274,
+      "loss": 0.0805,
+      "step": 19732
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010747173215447092,
+      "loss": 0.0589,
+      "step": 19733
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001074084132179033,
+      "loss": 0.1499,
+      "step": 19734
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010734511188098761,
+      "loss": 0.0723,
+      "step": 19735
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010728182814497278,
+      "loss": 0.0741,
+      "step": 19736
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010721856201110602,
+      "loss": 0.0409,
+      "step": 19737
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010715531348063511,
+      "loss": 0.0863,
+      "step": 19738
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010709208255480695,
+      "loss": 0.1072,
+      "step": 19739
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001070288692348681,
+      "loss": 0.0808,
+      "step": 19740
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010696567352206554,
+      "loss": 0.058,
+      "step": 19741
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010690249541764486,
+      "loss": 0.0929,
+      "step": 19742
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010683933492285202,
+      "loss": 0.1025,
+      "step": 19743
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010677619203893229,
+      "loss": 0.065,
+      "step": 19744
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010671306676713045,
+      "loss": 0.0699,
+      "step": 19745
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010664995910869157,
+      "loss": 0.1333,
+      "step": 19746
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010658686906485971,
+      "loss": 0.0735,
+      "step": 19747
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010652379663687906,
+      "loss": 0.0799,
+      "step": 19748
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010646074182599319,
+      "loss": 0.0595,
+      "step": 19749
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010639770463344533,
+      "loss": 0.0877,
+      "step": 19750
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010633468506047839,
+      "loss": 0.0702,
+      "step": 19751
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010627168310833502,
+      "loss": 0.056,
+      "step": 19752
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010620869877825712,
+      "loss": 0.0692,
+      "step": 19753
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010614573207148714,
+      "loss": 0.0907,
+      "step": 19754
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010608278298926632,
+      "loss": 0.0639,
+      "step": 19755
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010601985153283577,
+      "loss": 0.0682,
+      "step": 19756
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010595693770343661,
+      "loss": 0.1053,
+      "step": 19757
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001058940415023093,
+      "loss": 0.1262,
+      "step": 19758
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010583116293069362,
+      "loss": 0.0432,
+      "step": 19759
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010576830198982989,
+      "loss": 0.0712,
+      "step": 19760
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010570545868095726,
+      "loss": 0.0889,
+      "step": 19761
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010564263300531496,
+      "loss": 0.0627,
+      "step": 19762
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010557982496414176,
+      "loss": 0.0784,
+      "step": 19763
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001055170345586759,
+      "loss": 0.0687,
+      "step": 19764
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001054542617901556,
+      "loss": 0.1464,
+      "step": 19765
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00010539150665981845,
+      "loss": 0.1182,
+      "step": 19766
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001053287691689021,
+      "loss": 0.0928,
+      "step": 19767
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010526604931864314,
+      "loss": 0.0607,
+      "step": 19768
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010520334711027867,
+      "loss": 0.1675,
+      "step": 19769
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010514066254504495,
+      "loss": 0.0985,
+      "step": 19770
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010507799562417785,
+      "loss": 0.1049,
+      "step": 19771
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010501534634891297,
+      "loss": 0.0602,
+      "step": 19772
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010495271472048551,
+      "loss": 0.0897,
+      "step": 19773
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010489010074013072,
+      "loss": 0.1142,
+      "step": 19774
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010482750440908307,
+      "loss": 0.1115,
+      "step": 19775
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010476492572857666,
+      "loss": 0.0621,
+      "step": 19776
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010470236469984529,
+      "loss": 0.0588,
+      "step": 19777
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010463982132412286,
+      "loss": 0.1091,
+      "step": 19778
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010457729560264228,
+      "loss": 0.0923,
+      "step": 19779
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010451478753663657,
+      "loss": 0.0845,
+      "step": 19780
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010445229712733829,
+      "loss": 0.066,
+      "step": 19781
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010438982437597943,
+      "loss": 0.1031,
+      "step": 19782
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010432736928379182,
+      "loss": 0.0756,
+      "step": 19783
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010426493185200703,
+      "loss": 0.0933,
+      "step": 19784
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010420251208185594,
+      "loss": 0.0469,
+      "step": 19785
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010414010997456936,
+      "loss": 0.1198,
+      "step": 19786
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010407772553137795,
+      "loss": 0.1085,
+      "step": 19787
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010401535875351142,
+      "loss": 0.0604,
+      "step": 19788
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010395300964219989,
+      "loss": 0.0472,
+      "step": 19789
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010389067819867248,
+      "loss": 0.0832,
+      "step": 19790
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010382836442415822,
+      "loss": 0.1166,
+      "step": 19791
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010376606831988578,
+      "loss": 0.0735,
+      "step": 19792
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001037037898870834,
+      "loss": 0.0895,
+      "step": 19793
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010364152912697933,
+      "loss": 0.062,
+      "step": 19794
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010357928604080102,
+      "loss": 0.1102,
+      "step": 19795
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010351706062977573,
+      "loss": 0.0675,
+      "step": 19796
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010345485289513035,
+      "loss": 0.079,
+      "step": 19797
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010339266283809145,
+      "loss": 0.0876,
+      "step": 19798
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010333049045988541,
+      "loss": 0.0912,
+      "step": 19799
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010326833576173789,
+      "loss": 0.1066,
+      "step": 19800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010320619874487468,
+      "loss": 0.0616,
+      "step": 19801
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010314407941052095,
+      "loss": 0.0898,
+      "step": 19802
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010308197775990147,
+      "loss": 0.07,
+      "step": 19803
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010301989379424059,
+      "loss": 0.0836,
+      "step": 19804
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010295782751476258,
+      "loss": 0.0989,
+      "step": 19805
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010289577892269109,
+      "loss": 0.1047,
+      "step": 19806
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010283374801924971,
+      "loss": 0.0873,
+      "step": 19807
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001027717348056616,
+      "loss": 0.1028,
+      "step": 19808
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010270973928314919,
+      "loss": 0.1063,
+      "step": 19809
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001026477614529353,
+      "loss": 0.1286,
+      "step": 19810
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010258580131624174,
+      "loss": 0.111,
+      "step": 19811
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00010252385887429027,
+      "loss": 0.0906,
+      "step": 19812
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010246193412830195,
+      "loss": 0.1097,
+      "step": 19813
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010240002707949825,
+      "loss": 0.0405,
+      "step": 19814
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010233813772909962,
+      "loss": 0.0893,
+      "step": 19815
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010227626607832629,
+      "loss": 0.1041,
+      "step": 19816
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010221441212839822,
+      "loss": 0.1372,
+      "step": 19817
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010215257588053517,
+      "loss": 0.0709,
+      "step": 19818
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010209075733595596,
+      "loss": 0.0739,
+      "step": 19819
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010202895649588017,
+      "loss": 0.1368,
+      "step": 19820
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010196717336152572,
+      "loss": 0.109,
+      "step": 19821
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010190540793411129,
+      "loss": 0.093,
+      "step": 19822
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010184366021485458,
+      "loss": 0.0627,
+      "step": 19823
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010178193020497306,
+      "loss": 0.088,
+      "step": 19824
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010172021790568386,
+      "loss": 0.0974,
+      "step": 19825
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001016585233182037,
+      "loss": 0.0905,
+      "step": 19826
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010159684644374922,
+      "loss": 0.0752,
+      "step": 19827
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010153518728353662,
+      "loss": 0.0786,
+      "step": 19828
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010147354583878155,
+      "loss": 0.1105,
+      "step": 19829
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010141192211069917,
+      "loss": 0.0814,
+      "step": 19830
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010135031610050494,
+      "loss": 0.0722,
+      "step": 19831
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010128872780941333,
+      "loss": 0.0718,
+      "step": 19832
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010122715723863873,
+      "loss": 0.1396,
+      "step": 19833
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010116560438939537,
+      "loss": 0.0627,
+      "step": 19834
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010110406926289673,
+      "loss": 0.0739,
+      "step": 19835
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010104255186035616,
+      "loss": 0.1746,
+      "step": 19836
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010098105218298658,
+      "loss": 0.1241,
+      "step": 19837
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010091957023200071,
+      "loss": 0.0914,
+      "step": 19838
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010085810600861056,
+      "loss": 0.0652,
+      "step": 19839
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010079665951402839,
+      "loss": 0.0933,
+      "step": 19840
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010073523074946567,
+      "loss": 0.1085,
+      "step": 19841
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010067381971613343,
+      "loss": 0.0777,
+      "step": 19842
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010061242641524282,
+      "loss": 0.0963,
+      "step": 19843
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010055105084800431,
+      "loss": 0.0828,
+      "step": 19844
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010048969301562794,
+      "loss": 0.0894,
+      "step": 19845
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001004283529193234,
+      "loss": 0.1354,
+      "step": 19846
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001003670305603005,
+      "loss": 0.1288,
+      "step": 19847
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010030572593976816,
+      "loss": 0.0718,
+      "step": 19848
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010024443905893532,
+      "loss": 0.1138,
+      "step": 19849
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010018316991901022,
+      "loss": 0.132,
+      "step": 19850
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010012191852120079,
+      "loss": 0.06,
+      "step": 19851
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00010006068486671515,
+      "loss": 0.0629,
+      "step": 19852
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.999946895676038e-05,
+      "loss": 0.0806,
+      "step": 19853
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.99382707925437e-05,
+      "loss": 0.0778,
+      "step": 19854
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.987709037527171e-05,
+      "loss": 0.0781,
+      "step": 19855
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.981592770615066e-05,
+      "loss": 0.064,
+      "step": 19856
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.975478278638661e-05,
+      "loss": 0.0754,
+      "step": 19857
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.969365561718524e-05,
+      "loss": 0.1067,
+      "step": 19858
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.963254619975137e-05,
+      "loss": 0.0859,
+      "step": 19859
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.957145453529059e-05,
+      "loss": 0.0926,
+      "step": 19860
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.951038062500717e-05,
+      "loss": 0.1312,
+      "step": 19861
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.944932447010525e-05,
+      "loss": 0.0997,
+      "step": 19862
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.938828607178874e-05,
+      "loss": 0.0713,
+      "step": 19863
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.932726543126125e-05,
+      "loss": 0.1213,
+      "step": 19864
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.926626254972593e-05,
+      "loss": 0.1047,
+      "step": 19865
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.920527742838548e-05,
+      "loss": 0.0836,
+      "step": 19866
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.91443100684426e-05,
+      "loss": 0.0864,
+      "step": 19867
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.908336047109923e-05,
+      "loss": 0.109,
+      "step": 19868
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.902242863755728e-05,
+      "loss": 0.0399,
+      "step": 19869
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.896151456901814e-05,
+      "loss": 0.0932,
+      "step": 19870
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.890061826668273e-05,
+      "loss": 0.1265,
+      "step": 19871
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.883973973175176e-05,
+      "loss": 0.0616,
+      "step": 19872
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.877887896542582e-05,
+      "loss": 0.0631,
+      "step": 19873
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.871803596890472e-05,
+      "loss": 0.092,
+      "step": 19874
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.86572107433884e-05,
+      "loss": 0.1134,
+      "step": 19875
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.859640329007602e-05,
+      "loss": 0.0893,
+      "step": 19876
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.85356136101665e-05,
+      "loss": 0.1047,
+      "step": 19877
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.847484170485854e-05,
+      "loss": 0.0785,
+      "step": 19878
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.84140875753502e-05,
+      "loss": 0.1213,
+      "step": 19879
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.835335122283972e-05,
+      "loss": 0.0917,
+      "step": 19880
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.829263264852451e-05,
+      "loss": 0.0685,
+      "step": 19881
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.82319318536019e-05,
+      "loss": 0.0778,
+      "step": 19882
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.817124883926842e-05,
+      "loss": 0.1208,
+      "step": 19883
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.811058360672098e-05,
+      "loss": 0.0984,
+      "step": 19884
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.804993615715552e-05,
+      "loss": 0.0894,
+      "step": 19885
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.798930649176785e-05,
+      "loss": 0.0795,
+      "step": 19886
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.792869461175357e-05,
+      "loss": 0.0511,
+      "step": 19887
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.786810051830774e-05,
+      "loss": 0.0638,
+      "step": 19888
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.780752421262518e-05,
+      "loss": 0.1627,
+      "step": 19889
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.774696569590025e-05,
+      "loss": 0.066,
+      "step": 19890
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.768642496932689e-05,
+      "loss": 0.1115,
+      "step": 19891
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.762590203409904e-05,
+      "loss": 0.0997,
+      "step": 19892
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.756539689140964e-05,
+      "loss": 0.1223,
+      "step": 19893
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.750490954245217e-05,
+      "loss": 0.1016,
+      "step": 19894
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.744443998841901e-05,
+      "loss": 0.0789,
+      "step": 19895
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.738398823050265e-05,
+      "loss": 0.0978,
+      "step": 19896
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.732355426989503e-05,
+      "loss": 0.1296,
+      "step": 19897
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.726313810778764e-05,
+      "loss": 0.084,
+      "step": 19898
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.720273974537164e-05,
+      "loss": 0.0796,
+      "step": 19899
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.714235918383818e-05,
+      "loss": 0.087,
+      "step": 19900
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.708199642437776e-05,
+      "loss": 0.0685,
+      "step": 19901
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.702165146818054e-05,
+      "loss": 0.0887,
+      "step": 19902
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.696132431643635e-05,
+      "loss": 0.0859,
+      "step": 19903
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.690101497033443e-05,
+      "loss": 0.062,
+      "step": 19904
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.684072343106443e-05,
+      "loss": 0.1069,
+      "step": 19905
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.67804496998147e-05,
+      "loss": 0.0671,
+      "step": 19906
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.672019377777419e-05,
+      "loss": 0.1054,
+      "step": 19907
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.665995566613062e-05,
+      "loss": 0.0752,
+      "step": 19908
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.65997353660718e-05,
+      "loss": 0.0828,
+      "step": 19909
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.653953287878514e-05,
+      "loss": 0.1266,
+      "step": 19910
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.647934820545767e-05,
+      "loss": 0.0955,
+      "step": 19911
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.641918134727612e-05,
+      "loss": 0.0895,
+      "step": 19912
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.635903230542665e-05,
+      "loss": 0.1062,
+      "step": 19913
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.629890108109551e-05,
+      "loss": 0.1061,
+      "step": 19914
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.623878767546824e-05,
+      "loss": 0.1375,
+      "step": 19915
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.617869208972985e-05,
+      "loss": 0.1027,
+      "step": 19916
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.611861432506574e-05,
+      "loss": 0.113,
+      "step": 19917
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.605855438266031e-05,
+      "loss": 0.0687,
+      "step": 19918
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.59985122636976e-05,
+      "loss": 0.0686,
+      "step": 19919
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.593848796936178e-05,
+      "loss": 0.1298,
+      "step": 19920
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.587848150083633e-05,
+      "loss": 0.0729,
+      "step": 19921
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.581849285930421e-05,
+      "loss": 0.0848,
+      "step": 19922
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.575852204594848e-05,
+      "loss": 0.1085,
+      "step": 19923
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.569856906195151e-05,
+      "loss": 0.1214,
+      "step": 19924
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.563863390849515e-05,
+      "loss": 0.1591,
+      "step": 19925
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.557871658676165e-05,
+      "loss": 0.0743,
+      "step": 19926
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.551881709793208e-05,
+      "loss": 0.1166,
+      "step": 19927
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.545893544318773e-05,
+      "loss": 0.0964,
+      "step": 19928
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.539907162370931e-05,
+      "loss": 0.0786,
+      "step": 19929
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.53392256406771e-05,
+      "loss": 0.0544,
+      "step": 19930
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.527939749527115e-05,
+      "loss": 0.1019,
+      "step": 19931
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.521958718867107e-05,
+      "loss": 0.0719,
+      "step": 19932
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.515979472205593e-05,
+      "loss": 0.0722,
+      "step": 19933
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.510002009660523e-05,
+      "loss": 0.0789,
+      "step": 19934
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.504026331349713e-05,
+      "loss": 0.1478,
+      "step": 19935
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.498052437391014e-05,
+      "loss": 0.0484,
+      "step": 19936
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.492080327902186e-05,
+      "loss": 0.0446,
+      "step": 19937
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.486110003001025e-05,
+      "loss": 0.0914,
+      "step": 19938
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.480141462805203e-05,
+      "loss": 0.0854,
+      "step": 19939
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.47417470743246e-05,
+      "loss": 0.0774,
+      "step": 19940
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.468209737000411e-05,
+      "loss": 0.1003,
+      "step": 19941
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.462246551626675e-05,
+      "loss": 0.0955,
+      "step": 19942
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.456285151428834e-05,
+      "loss": 0.0662,
+      "step": 19943
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.450325536524429e-05,
+      "loss": 0.0743,
+      "step": 19944
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.444367707030965e-05,
+      "loss": 0.0749,
+      "step": 19945
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.438411663065904e-05,
+      "loss": 0.0894,
+      "step": 19946
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.432457404746709e-05,
+      "loss": 0.0911,
+      "step": 19947
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.426504932190772e-05,
+      "loss": 0.0848,
+      "step": 19948
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.420554245515467e-05,
+      "loss": 0.0813,
+      "step": 19949
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.414605344838124e-05,
+      "loss": 0.0819,
+      "step": 19950
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.408658230276035e-05,
+      "loss": 0.0654,
+      "step": 19951
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.402712901946475e-05,
+      "loss": 0.0679,
+      "step": 19952
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.396769359966628e-05,
+      "loss": 0.1207,
+      "step": 19953
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.390827604453744e-05,
+      "loss": 0.0671,
+      "step": 19954
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.384887635524964e-05,
+      "loss": 0.0778,
+      "step": 19955
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.378949453297381e-05,
+      "loss": 0.1165,
+      "step": 19956
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.373013057888113e-05,
+      "loss": 0.0777,
+      "step": 19957
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.367078449414167e-05,
+      "loss": 0.0681,
+      "step": 19958
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.361145627992596e-05,
+      "loss": 0.0786,
+      "step": 19959
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.355214593740391e-05,
+      "loss": 0.0594,
+      "step": 19960
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.349285346774472e-05,
+      "loss": 0.097,
+      "step": 19961
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.343357887211757e-05,
+      "loss": 0.0889,
+      "step": 19962
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.337432215169129e-05,
+      "loss": 0.0742,
+      "step": 19963
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.331508330763405e-05,
+      "loss": 0.1135,
+      "step": 19964
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.325586234111404e-05,
+      "loss": 0.0724,
+      "step": 19965
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.319665925329878e-05,
+      "loss": 0.09,
+      "step": 19966
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.313747404535588e-05,
+      "loss": 0.0979,
+      "step": 19967
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.307830671845218e-05,
+      "loss": 0.0909,
+      "step": 19968
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.301915727375409e-05,
+      "loss": 0.0834,
+      "step": 19969
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.296002571242845e-05,
+      "loss": 0.0926,
+      "step": 19970
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.290091203564066e-05,
+      "loss": 0.0646,
+      "step": 19971
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.284181624455667e-05,
+      "loss": 0.0921,
+      "step": 19972
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.278273834034124e-05,
+      "loss": 0.1041,
+      "step": 19973
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.272367832415962e-05,
+      "loss": 0.0594,
+      "step": 19974
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.266463619717635e-05,
+      "loss": 0.119,
+      "step": 19975
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.260561196055539e-05,
+      "loss": 0.0934,
+      "step": 19976
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.254660561546058e-05,
+      "loss": 0.1497,
+      "step": 19977
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.248761716305521e-05,
+      "loss": 0.0739,
+      "step": 19978
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.24286466045029e-05,
+      "loss": 0.0595,
+      "step": 19979
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.236969394096584e-05,
+      "loss": 0.0964,
+      "step": 19980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.231075917360677e-05,
+      "loss": 0.0723,
+      "step": 19981
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.225184230358774e-05,
+      "loss": 0.0726,
+      "step": 19982
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.21929433320703e-05,
+      "loss": 0.082,
+      "step": 19983
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.213406226021582e-05,
+      "loss": 0.0994,
+      "step": 19984
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.207519908918527e-05,
+      "loss": 0.0724,
+      "step": 19985
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.201635382013917e-05,
+      "loss": 0.0745,
+      "step": 19986
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.195752645423805e-05,
+      "loss": 0.0941,
+      "step": 19987
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.189871699264185e-05,
+      "loss": 0.0562,
+      "step": 19988
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.183992543650988e-05,
+      "loss": 0.089,
+      "step": 19989
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.178115178700142e-05,
+      "loss": 0.0628,
+      "step": 19990
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.172239604527554e-05,
+      "loss": 0.1169,
+      "step": 19991
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.166365821249068e-05,
+      "loss": 0.0742,
+      "step": 19992
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.160493828980477e-05,
+      "loss": 0.0765,
+      "step": 19993
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.154623627837589e-05,
+      "loss": 0.1312,
+      "step": 19994
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.148755217936156e-05,
+      "loss": 0.0671,
+      "step": 19995
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.142888599391863e-05,
+      "loss": 0.0767,
+      "step": 19996
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.137023772320397e-05,
+      "loss": 0.1053,
+      "step": 19997
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.131160736837396e-05,
+      "loss": 0.08,
+      "step": 19998
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.125299493058437e-05,
+      "loss": 0.0829,
+      "step": 19999
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.119440041099147e-05,
+      "loss": 0.1262,
+      "step": 20000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.113582381075003e-05,
+      "loss": 0.0968,
+      "step": 20001
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.107726513101545e-05,
+      "loss": 0.1016,
+      "step": 20002
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.101872437294223e-05,
+      "loss": 0.0739,
+      "step": 20003
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.096020153768458e-05,
+      "loss": 0.0889,
+      "step": 20004
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.090169662639659e-05,
+      "loss": 0.1324,
+      "step": 20005
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.084320964023129e-05,
+      "loss": 0.0756,
+      "step": 20006
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.07847405803427e-05,
+      "loss": 0.0685,
+      "step": 20007
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.072628944788309e-05,
+      "loss": 0.0975,
+      "step": 20008
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.066785624400531e-05,
+      "loss": 0.0897,
+      "step": 20009
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.060944096986124e-05,
+      "loss": 0.0562,
+      "step": 20010
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.055104362660272e-05,
+      "loss": 0.055,
+      "step": 20011
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.049266421538139e-05,
+      "loss": 0.0815,
+      "step": 20012
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.04343027373481e-05,
+      "loss": 0.0635,
+      "step": 20013
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.037595919365394e-05,
+      "loss": 0.0801,
+      "step": 20014
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.031763358544898e-05,
+      "loss": 0.0839,
+      "step": 20015
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.025932591388341e-05,
+      "loss": 0.0745,
+      "step": 20016
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.020103618010688e-05,
+      "loss": 0.0624,
+      "step": 20017
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.01427643852687e-05,
+      "loss": 0.0747,
+      "step": 20018
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.00845105305177e-05,
+      "loss": 0.1029,
+      "step": 20019
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.002627461700274e-05,
+      "loss": 0.0566,
+      "step": 20020
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.996805664587204e-05,
+      "loss": 0.0758,
+      "step": 20021
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.990985661827322e-05,
+      "loss": 0.0649,
+      "step": 20022
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.985167453535436e-05,
+      "loss": 0.0722,
+      "step": 20023
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.979351039826223e-05,
+      "loss": 0.0884,
+      "step": 20024
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.973536420814398e-05,
+      "loss": 0.0967,
+      "step": 20025
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.967723596614563e-05,
+      "loss": 0.0629,
+      "step": 20026
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.96191256734139e-05,
+      "loss": 0.1299,
+      "step": 20027
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.956103333109433e-05,
+      "loss": 0.0676,
+      "step": 20028
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.950295894033233e-05,
+      "loss": 0.1345,
+      "step": 20029
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.9444902502273e-05,
+      "loss": 0.0781,
+      "step": 20030
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.938686401806106e-05,
+      "loss": 0.1232,
+      "step": 20031
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.932884348884063e-05,
+      "loss": 0.0819,
+      "step": 20032
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.92708409157561e-05,
+      "loss": 0.1152,
+      "step": 20033
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.921285629995112e-05,
+      "loss": 0.0846,
+      "step": 20034
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.915488964256901e-05,
+      "loss": 0.1081,
+      "step": 20035
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.909694094475251e-05,
+      "loss": 0.0788,
+      "step": 20036
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.903901020764438e-05,
+      "loss": 0.0771,
+      "step": 20037
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.89810974323868e-05,
+      "loss": 0.0638,
+      "step": 20038
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.892320262012143e-05,
+      "loss": 0.1147,
+      "step": 20039
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.886532577199035e-05,
+      "loss": 0.0885,
+      "step": 20040
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.880746688913444e-05,
+      "loss": 0.0994,
+      "step": 20041
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.874962597269443e-05,
+      "loss": 0.1282,
+      "step": 20042
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.869180302381085e-05,
+      "loss": 0.1835,
+      "step": 20043
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.863399804362404e-05,
+      "loss": 0.0823,
+      "step": 20044
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.857621103327362e-05,
+      "loss": 0.1218,
+      "step": 20045
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.85184419938988e-05,
+      "loss": 0.0996,
+      "step": 20046
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.8460690926639e-05,
+      "loss": 0.07,
+      "step": 20047
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.840295783263275e-05,
+      "loss": 0.0979,
+      "step": 20048
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.834524271301847e-05,
+      "loss": 0.1257,
+      "step": 20049
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.828754556893404e-05,
+      "loss": 0.0916,
+      "step": 20050
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.82298664015172e-05,
+      "loss": 0.0746,
+      "step": 20051
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.817220521190495e-05,
+      "loss": 0.1371,
+      "step": 20052
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.811456200123468e-05,
+      "loss": 0.1103,
+      "step": 20053
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.805693677064264e-05,
+      "loss": 0.0753,
+      "step": 20054
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.799932952126533e-05,
+      "loss": 0.0995,
+      "step": 20055
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.79417402542384e-05,
+      "loss": 0.1036,
+      "step": 20056
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.788416897069752e-05,
+      "loss": 0.0785,
+      "step": 20057
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.782661567177774e-05,
+      "loss": 0.0656,
+      "step": 20058
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.776908035861364e-05,
+      "loss": 0.0868,
+      "step": 20059
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.771156303234018e-05,
+      "loss": 0.1038,
+      "step": 20060
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.765406369409112e-05,
+      "loss": 0.0958,
+      "step": 20061
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.759658234500023e-05,
+      "loss": 0.1068,
+      "step": 20062
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.753911898620104e-05,
+      "loss": 0.1105,
+      "step": 20063
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.748167361882631e-05,
+      "loss": 0.0797,
+      "step": 20064
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.742424624400902e-05,
+      "loss": 0.165,
+      "step": 20065
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.736683686288116e-05,
+      "loss": 0.0426,
+      "step": 20066
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.730944547657516e-05,
+      "loss": 0.0977,
+      "step": 20067
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.725207208622232e-05,
+      "loss": 0.1027,
+      "step": 20068
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.719471669295399e-05,
+      "loss": 0.0817,
+      "step": 20069
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.713737929790099e-05,
+      "loss": 0.0546,
+      "step": 20070
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.708005990219392e-05,
+      "loss": 0.0848,
+      "step": 20071
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.702275850696284e-05,
+      "loss": 0.0824,
+      "step": 20072
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.696547511333785e-05,
+      "loss": 0.0534,
+      "step": 20073
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.690820972244829e-05,
+      "loss": 0.0972,
+      "step": 20074
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.685096233542311e-05,
+      "loss": 0.0829,
+      "step": 20075
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.679373295339155e-05,
+      "loss": 0.0952,
+      "step": 20076
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.673652157748169e-05,
+      "loss": 0.0619,
+      "step": 20077
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.667932820882163e-05,
+      "loss": 0.0568,
+      "step": 20078
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.662215284853892e-05,
+      "loss": 0.1073,
+      "step": 20079
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.656499549776131e-05,
+      "loss": 0.0674,
+      "step": 20080
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.650785615761558e-05,
+      "loss": 0.0652,
+      "step": 20081
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.64507348292285e-05,
+      "loss": 0.1428,
+      "step": 20082
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.639363151372604e-05,
+      "loss": 0.101,
+      "step": 20083
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.633654621223452e-05,
+      "loss": 0.1011,
+      "step": 20084
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.627947892587906e-05,
+      "loss": 0.0634,
+      "step": 20085
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.62224296557853e-05,
+      "loss": 0.0638,
+      "step": 20086
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.61653984030778e-05,
+      "loss": 0.0724,
+      "step": 20087
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.610838516888142e-05,
+      "loss": 0.1092,
+      "step": 20088
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.605138995432016e-05,
+      "loss": 0.0978,
+      "step": 20089
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.599441276051779e-05,
+      "loss": 0.0707,
+      "step": 20090
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.593745358859773e-05,
+      "loss": 0.0819,
+      "step": 20091
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.588051243968286e-05,
+      "loss": 0.1052,
+      "step": 20092
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.582358931489642e-05,
+      "loss": 0.0821,
+      "step": 20093
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.57666842153606e-05,
+      "loss": 0.0393,
+      "step": 20094
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.570979714219729e-05,
+      "loss": 0.0617,
+      "step": 20095
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.565292809652792e-05,
+      "loss": 0.0809,
+      "step": 20096
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.559607707947447e-05,
+      "loss": 0.0642,
+      "step": 20097
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.553924409215741e-05,
+      "loss": 0.0444,
+      "step": 20098
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.548242913569727e-05,
+      "loss": 0.0767,
+      "step": 20099
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.542563221121479e-05,
+      "loss": 0.1056,
+      "step": 20100
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.536885331982947e-05,
+      "loss": 0.0634,
+      "step": 20101
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.531209246266103e-05,
+      "loss": 0.058,
+      "step": 20102
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.525534964082848e-05,
+      "loss": 0.0566,
+      "step": 20103
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.519862485545071e-05,
+      "loss": 0.1022,
+      "step": 20104
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.514191810764627e-05,
+      "loss": 0.082,
+      "step": 20105
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.508522939853302e-05,
+      "loss": 0.0972,
+      "step": 20106
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.502855872922887e-05,
+      "loss": 0.0671,
+      "step": 20107
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.497190610085148e-05,
+      "loss": 0.1198,
+      "step": 20108
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.491527151451772e-05,
+      "loss": 0.0797,
+      "step": 20109
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.485865497134415e-05,
+      "loss": 0.072,
+      "step": 20110
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.480205647244721e-05,
+      "loss": 0.0594,
+      "step": 20111
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.474547601894279e-05,
+      "loss": 0.1382,
+      "step": 20112
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.468891361194663e-05,
+      "loss": 0.1191,
+      "step": 20113
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.4632369252574e-05,
+      "loss": 0.0825,
+      "step": 20114
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.457584294193976e-05,
+      "loss": 0.1134,
+      "step": 20115
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.451933468115858e-05,
+      "loss": 0.1198,
+      "step": 20116
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.446284447134423e-05,
+      "loss": 0.061,
+      "step": 20117
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.440637231361115e-05,
+      "loss": 0.0373,
+      "step": 20118
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.434991820907234e-05,
+      "loss": 0.152,
+      "step": 20119
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.429348215884136e-05,
+      "loss": 0.1112,
+      "step": 20120
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.423706416403076e-05,
+      "loss": 0.1147,
+      "step": 20121
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.418066422575298e-05,
+      "loss": 0.0526,
+      "step": 20122
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.412428234512004e-05,
+      "loss": 0.1658,
+      "step": 20123
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.406791852324369e-05,
+      "loss": 0.1184,
+      "step": 20124
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.401157276123539e-05,
+      "loss": 0.0662,
+      "step": 20125
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.395524506020568e-05,
+      "loss": 0.0883,
+      "step": 20126
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.389893542126592e-05,
+      "loss": 0.1595,
+      "step": 20127
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.384264384552564e-05,
+      "loss": 0.1044,
+      "step": 20128
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.378637033409542e-05,
+      "loss": 0.0599,
+      "step": 20129
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.373011488808457e-05,
+      "loss": 0.0637,
+      "step": 20130
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.367387750860233e-05,
+      "loss": 0.0771,
+      "step": 20131
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.361765819675737e-05,
+      "loss": 0.0903,
+      "step": 20132
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.356145695365846e-05,
+      "loss": 0.0877,
+      "step": 20133
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.350527378041362e-05,
+      "loss": 0.0786,
+      "step": 20134
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.34491086781306e-05,
+      "loss": 0.077,
+      "step": 20135
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.339296164791698e-05,
+      "loss": 0.073,
+      "step": 20136
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.333683269087977e-05,
+      "loss": 0.0868,
+      "step": 20137
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.32807218081254e-05,
+      "loss": 0.0814,
+      "step": 20138
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.322462900076078e-05,
+      "loss": 0.0508,
+      "step": 20139
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.316855426989145e-05,
+      "loss": 0.0718,
+      "step": 20140
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.311249761662332e-05,
+      "loss": 0.0864,
+      "step": 20141
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.305645904206171e-05,
+      "loss": 0.0941,
+      "step": 20142
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.300043854731143e-05,
+      "loss": 0.064,
+      "step": 20143
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.294443613347714e-05,
+      "loss": 0.0735,
+      "step": 20144
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.288845180166304e-05,
+      "loss": 0.1166,
+      "step": 20145
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.283248555297273e-05,
+      "loss": 0.0648,
+      "step": 20146
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.277653738851031e-05,
+      "loss": 0.0834,
+      "step": 20147
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.272060730937847e-05,
+      "loss": 0.0593,
+      "step": 20148
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.266469531667997e-05,
+      "loss": 0.1194,
+      "step": 20149
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.260880141151772e-05,
+      "loss": 0.0773,
+      "step": 20150
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.255292559499338e-05,
+      "loss": 0.0995,
+      "step": 20151
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.249706786820866e-05,
+      "loss": 0.0682,
+      "step": 20152
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.24412282322653e-05,
+      "loss": 0.079,
+      "step": 20153
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.238540668826411e-05,
+      "loss": 0.098,
+      "step": 20154
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.232960323730576e-05,
+      "loss": 0.0938,
+      "step": 20155
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.22738178804906e-05,
+      "loss": 0.0843,
+      "step": 20156
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.22180506189184e-05,
+      "loss": 0.0643,
+      "step": 20157
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.216230145368887e-05,
+      "loss": 0.0842,
+      "step": 20158
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.210657038590119e-05,
+      "loss": 0.09,
+      "step": 20159
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.205085741665431e-05,
+      "loss": 0.0512,
+      "step": 20160
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.199516254704664e-05,
+      "loss": 0.1432,
+      "step": 20161
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.193948577817656e-05,
+      "loss": 0.1621,
+      "step": 20162
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.188382711114173e-05,
+      "loss": 0.0856,
+      "step": 20163
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.18281865470396e-05,
+      "loss": 0.0484,
+      "step": 20164
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.177256408696731e-05,
+      "loss": 0.087,
+      "step": 20165
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.171695973202142e-05,
+      "loss": 0.1116,
+      "step": 20166
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.166137348329849e-05,
+      "loss": 0.0933,
+      "step": 20167
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.160580534189466e-05,
+      "loss": 0.046,
+      "step": 20168
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.155025530890525e-05,
+      "loss": 0.1169,
+      "step": 20169
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.149472338542574e-05,
+      "loss": 0.1113,
+      "step": 20170
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.143920957255113e-05,
+      "loss": 0.1099,
+      "step": 20171
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.138371387137589e-05,
+      "loss": 0.0898,
+      "step": 20172
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.132823628299435e-05,
+      "loss": 0.0813,
+      "step": 20173
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.127277680850053e-05,
+      "loss": 0.0814,
+      "step": 20174
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.121733544898768e-05,
+      "loss": 0.1399,
+      "step": 20175
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.116191220554914e-05,
+      "loss": 0.0585,
+      "step": 20176
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.110650707927758e-05,
+      "loss": 0.0588,
+      "step": 20177
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.105112007126559e-05,
+      "loss": 0.1853,
+      "step": 20178
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.099575118260494e-05,
+      "loss": 0.0617,
+      "step": 20179
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.094040041438788e-05,
+      "loss": 0.0815,
+      "step": 20180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.088506776770544e-05,
+      "loss": 0.0856,
+      "step": 20181
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.08297532436486e-05,
+      "loss": 0.0898,
+      "step": 20182
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.077445684330831e-05,
+      "loss": 0.0906,
+      "step": 20183
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.071917856777466e-05,
+      "loss": 0.0628,
+      "step": 20184
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.06639184181377e-05,
+      "loss": 0.0703,
+      "step": 20185
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.060867639548675e-05,
+      "loss": 0.0837,
+      "step": 20186
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.055345250091151e-05,
+      "loss": 0.0852,
+      "step": 20187
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.049824673550055e-05,
+      "loss": 0.1022,
+      "step": 20188
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.044305910034245e-05,
+      "loss": 0.0879,
+      "step": 20189
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.038788959652543e-05,
+      "loss": 0.0503,
+      "step": 20190
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.033273822513697e-05,
+      "loss": 0.109,
+      "step": 20191
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.0277604987265e-05,
+      "loss": 0.1031,
+      "step": 20192
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.022248988399617e-05,
+      "loss": 0.0983,
+      "step": 20193
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.016739291641762e-05,
+      "loss": 0.0681,
+      "step": 20194
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.011231408561548e-05,
+      "loss": 0.1295,
+      "step": 20195
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.005725339267588e-05,
+      "loss": 0.0812,
+      "step": 20196
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.000221083868441e-05,
+      "loss": 0.0836,
+      "step": 20197
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.994718642472631e-05,
+      "loss": 0.0475,
+      "step": 20198
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.989218015188648e-05,
+      "loss": 0.0787,
+      "step": 20199
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.983719202124973e-05,
+      "loss": 0.0792,
+      "step": 20200
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.978222203390018e-05,
+      "loss": 0.0972,
+      "step": 20201
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.972727019092152e-05,
+      "loss": 0.0753,
+      "step": 20202
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.967233649339767e-05,
+      "loss": 0.0751,
+      "step": 20203
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.961742094241143e-05,
+      "loss": 0.098,
+      "step": 20204
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.956252353904569e-05,
+      "loss": 0.071,
+      "step": 20205
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.950764428438284e-05,
+      "loss": 0.1062,
+      "step": 20206
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.94527831795051e-05,
+      "loss": 0.0696,
+      "step": 20207
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.939794022549418e-05,
+      "loss": 0.0801,
+      "step": 20208
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.934311542343142e-05,
+      "loss": 0.1027,
+      "step": 20209
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.928830877439774e-05,
+      "loss": 0.1085,
+      "step": 20210
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.923352027947383e-05,
+      "loss": 0.1202,
+      "step": 20211
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.917874993973972e-05,
+      "loss": 0.0724,
+      "step": 20212
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.912399775627588e-05,
+      "loss": 0.0945,
+      "step": 20213
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.906926373016143e-05,
+      "loss": 0.0779,
+      "step": 20214
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.901454786247597e-05,
+      "loss": 0.0779,
+      "step": 20215
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.895985015429818e-05,
+      "loss": 0.0637,
+      "step": 20216
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.890517060670654e-05,
+      "loss": 0.0797,
+      "step": 20217
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.885050922077907e-05,
+      "loss": 0.0795,
+      "step": 20218
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.87958659975937e-05,
+      "loss": 0.1069,
+      "step": 20219
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.874124093822788e-05,
+      "loss": 0.0849,
+      "step": 20220
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.868663404375864e-05,
+      "loss": 0.1351,
+      "step": 20221
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.86320453152627e-05,
+      "loss": 0.0793,
+      "step": 20222
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.857747475381627e-05,
+      "loss": 0.0906,
+      "step": 20223
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.852292236049552e-05,
+      "loss": 0.0512,
+      "step": 20224
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.846838813637613e-05,
+      "loss": 0.0831,
+      "step": 20225
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.841387208253315e-05,
+      "loss": 0.0859,
+      "step": 20226
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.83593742000417e-05,
+      "loss": 0.0722,
+      "step": 20227
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.830489448997635e-05,
+      "loss": 0.0898,
+      "step": 20228
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.825043295341116e-05,
+      "loss": 0.0763,
+      "step": 20229
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.819598959142005e-05,
+      "loss": 0.06,
+      "step": 20230
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.814156440507659e-05,
+      "loss": 0.1531,
+      "step": 20231
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.808715739545347e-05,
+      "loss": 0.0883,
+      "step": 20232
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.803276856362407e-05,
+      "loss": 0.1233,
+      "step": 20233
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.797839791066052e-05,
+      "loss": 0.1001,
+      "step": 20234
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.792404543763476e-05,
+      "loss": 0.1756,
+      "step": 20235
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.78697111456187e-05,
+      "loss": 0.0534,
+      "step": 20236
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.781539503568358e-05,
+      "loss": 0.0826,
+      "step": 20237
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.776109710890045e-05,
+      "loss": 0.1237,
+      "step": 20238
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.770681736633967e-05,
+      "loss": 0.1221,
+      "step": 20239
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.765255580907182e-05,
+      "loss": 0.0587,
+      "step": 20240
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.759831243816662e-05,
+      "loss": 0.1437,
+      "step": 20241
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.754408725469375e-05,
+      "loss": 0.082,
+      "step": 20242
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.748988025972225e-05,
+      "loss": 0.0778,
+      "step": 20243
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.743569145432084e-05,
+      "loss": 0.0825,
+      "step": 20244
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.73815208395583e-05,
+      "loss": 0.0831,
+      "step": 20245
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.732736841650234e-05,
+      "loss": 0.0527,
+      "step": 20246
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.727323418622123e-05,
+      "loss": 0.063,
+      "step": 20247
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.72191181497821e-05,
+      "loss": 0.0896,
+      "step": 20248
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.71650203082519e-05,
+      "loss": 0.0963,
+      "step": 20249
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.711094066269731e-05,
+      "loss": 0.0893,
+      "step": 20250
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.70568792141848e-05,
+      "loss": 0.0629,
+      "step": 20251
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.700283596377999e-05,
+      "loss": 0.0973,
+      "step": 20252
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.694881091254891e-05,
+      "loss": 0.0917,
+      "step": 20253
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.689480406155658e-05,
+      "loss": 0.0886,
+      "step": 20254
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.684081541186794e-05,
+      "loss": 0.0946,
+      "step": 20255
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.678684496454724e-05,
+      "loss": 0.063,
+      "step": 20256
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.673289272065908e-05,
+      "loss": 0.0927,
+      "step": 20257
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.667895868126706e-05,
+      "loss": 0.0749,
+      "step": 20258
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.662504284743443e-05,
+      "loss": 0.0639,
+      "step": 20259
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.657114522022457e-05,
+      "loss": 0.1171,
+      "step": 20260
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.651726580070017e-05,
+      "loss": 0.1107,
+      "step": 20261
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.646340458992351e-05,
+      "loss": 0.0903,
+      "step": 20262
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.640956158895651e-05,
+      "loss": 0.0729,
+      "step": 20263
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.635573679886099e-05,
+      "loss": 0.0405,
+      "step": 20264
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.630193022069798e-05,
+      "loss": 0.1913,
+      "step": 20265
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.624814185552875e-05,
+      "loss": 0.0853,
+      "step": 20266
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.619437170441357e-05,
+      "loss": 0.0979,
+      "step": 20267
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.614061976841291e-05,
+      "loss": 0.0804,
+      "step": 20268
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.60868860485866e-05,
+      "loss": 0.12,
+      "step": 20269
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.603317054599402e-05,
+      "loss": 0.0687,
+      "step": 20270
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.59794732616943e-05,
+      "loss": 0.0778,
+      "step": 20271
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.592579419674606e-05,
+      "loss": 0.1136,
+      "step": 20272
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.5872133352208e-05,
+      "loss": 0.0677,
+      "step": 20273
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.581849072913815e-05,
+      "loss": 0.0791,
+      "step": 20274
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.576486632859414e-05,
+      "loss": 0.101,
+      "step": 20275
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.571126015163321e-05,
+      "loss": 0.1151,
+      "step": 20276
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.56576721993123e-05,
+      "loss": 0.0881,
+      "step": 20277
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.560410247268834e-05,
+      "loss": 0.0556,
+      "step": 20278
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.555055097281726e-05,
+      "loss": 0.0847,
+      "step": 20279
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.549701770075524e-05,
+      "loss": 0.0772,
+      "step": 20280
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.544350265755772e-05,
+      "loss": 0.1196,
+      "step": 20281
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.53900058442798e-05,
+      "loss": 0.0789,
+      "step": 20282
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.533652726197637e-05,
+      "loss": 0.1703,
+      "step": 20283
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.528306691170184e-05,
+      "loss": 0.0721,
+      "step": 20284
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.522962479451034e-05,
+      "loss": 0.106,
+      "step": 20285
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.51762009114555e-05,
+      "loss": 0.0853,
+      "step": 20286
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.512279526359089e-05,
+      "loss": 0.1501,
+      "step": 20287
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.506940785196936e-05,
+      "loss": 0.0913,
+      "step": 20288
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.501603867764395e-05,
+      "loss": 0.073,
+      "step": 20289
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.496268774166659e-05,
+      "loss": 0.0841,
+      "step": 20290
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.490935504508933e-05,
+      "loss": 0.0714,
+      "step": 20291
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.485604058896356e-05,
+      "loss": 0.0727,
+      "step": 20292
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.480274437434098e-05,
+      "loss": 0.0802,
+      "step": 20293
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.47494664022721e-05,
+      "loss": 0.0947,
+      "step": 20294
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.46962066738075e-05,
+      "loss": 0.08,
+      "step": 20295
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.464296518999735e-05,
+      "loss": 0.069,
+      "step": 20296
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.458974195189128e-05,
+      "loss": 0.0717,
+      "step": 20297
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.453653696053897e-05,
+      "loss": 0.0653,
+      "step": 20298
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.448335021698926e-05,
+      "loss": 0.1049,
+      "step": 20299
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.443018172229122e-05,
+      "loss": 0.0688,
+      "step": 20300
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.437703147749286e-05,
+      "loss": 0.0946,
+      "step": 20301
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.432389948364226e-05,
+      "loss": 0.085,
+      "step": 20302
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.427078574178703e-05,
+      "loss": 0.0708,
+      "step": 20303
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.421769025297454e-05,
+      "loss": 0.0766,
+      "step": 20304
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.41646130182514e-05,
+      "loss": 0.0548,
+      "step": 20305
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.411155403866454e-05,
+      "loss": 0.055,
+      "step": 20306
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.405851331526004e-05,
+      "loss": 0.0682,
+      "step": 20307
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.40054908490837e-05,
+      "loss": 0.0606,
+      "step": 20308
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.395248664118071e-05,
+      "loss": 0.106,
+      "step": 20309
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.389950069259665e-05,
+      "loss": 0.117,
+      "step": 20310
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.384653300437605e-05,
+      "loss": 0.0518,
+      "step": 20311
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.379358357756316e-05,
+      "loss": 0.0831,
+      "step": 20312
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.374065241320238e-05,
+      "loss": 0.0845,
+      "step": 20313
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.368773951233699e-05,
+      "loss": 0.0871,
+      "step": 20314
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.363484487601058e-05,
+      "loss": 0.0908,
+      "step": 20315
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.358196850526599e-05,
+      "loss": 0.0595,
+      "step": 20316
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.352911040114574e-05,
+      "loss": 0.0696,
+      "step": 20317
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.347627056469197e-05,
+      "loss": 0.0767,
+      "step": 20318
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.342344899694686e-05,
+      "loss": 0.0676,
+      "step": 20319
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.337064569895158e-05,
+      "loss": 0.0895,
+      "step": 20320
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.331786067174762e-05,
+      "loss": 0.062,
+      "step": 20321
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.32650939163756e-05,
+      "loss": 0.0708,
+      "step": 20322
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.3212345433876e-05,
+      "loss": 0.0761,
+      "step": 20323
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.315961522528869e-05,
+      "loss": 0.0668,
+      "step": 20324
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.310690329165349e-05,
+      "loss": 0.0992,
+      "step": 20325
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.305420963400988e-05,
+      "loss": 0.1008,
+      "step": 20326
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.300153425339684e-05,
+      "loss": 0.075,
+      "step": 20327
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.294887715085275e-05,
+      "loss": 0.075,
+      "step": 20328
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.28962383274161e-05,
+      "loss": 0.0803,
+      "step": 20329
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.284361778412451e-05,
+      "loss": 0.0709,
+      "step": 20330
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.279101552201595e-05,
+      "loss": 0.0554,
+      "step": 20331
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.273843154212722e-05,
+      "loss": 0.102,
+      "step": 20332
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.268586584549541e-05,
+      "loss": 0.0574,
+      "step": 20333
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.26333184331569e-05,
+      "loss": 0.078,
+      "step": 20334
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.258078930614787e-05,
+      "loss": 0.0643,
+      "step": 20335
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.252827846550381e-05,
+      "loss": 0.0935,
+      "step": 20336
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.247578591226033e-05,
+      "loss": 0.0783,
+      "step": 20337
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.242331164745241e-05,
+      "loss": 0.0937,
+      "step": 20338
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.237085567211444e-05,
+      "loss": 0.0838,
+      "step": 20339
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.231841798728123e-05,
+      "loss": 0.0973,
+      "step": 20340
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.226599859398619e-05,
+      "loss": 0.0637,
+      "step": 20341
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.221359749326329e-05,
+      "loss": 0.0838,
+      "step": 20342
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.216121468614567e-05,
+      "loss": 0.1747,
+      "step": 20343
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.210885017366608e-05,
+      "loss": 0.0996,
+      "step": 20344
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.205650395685692e-05,
+      "loss": 0.0928,
+      "step": 20345
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.200417603675068e-05,
+      "loss": 0.1003,
+      "step": 20346
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.19518664143789e-05,
+      "loss": 0.1333,
+      "step": 20347
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.189957509077306e-05,
+      "loss": 0.0804,
+      "step": 20348
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.18473020669641e-05,
+      "loss": 0.0825,
+      "step": 20349
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.179504734398279e-05,
+      "loss": 0.0806,
+      "step": 20350
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.174281092285928e-05,
+      "loss": 0.1492,
+      "step": 20351
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.169059280462364e-05,
+      "loss": 0.1145,
+      "step": 20352
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.163839299030583e-05,
+      "loss": 0.1178,
+      "step": 20353
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.15862114809348e-05,
+      "loss": 0.1083,
+      "step": 20354
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.153404827753951e-05,
+      "loss": 0.0794,
+      "step": 20355
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.148190338114835e-05,
+      "loss": 0.0749,
+      "step": 20356
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.142977679278961e-05,
+      "loss": 0.0856,
+      "step": 20357
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.137766851349103e-05,
+      "loss": 0.1201,
+      "step": 20358
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.13255785442799e-05,
+      "loss": 0.0995,
+      "step": 20359
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.127350688618361e-05,
+      "loss": 0.0998,
+      "step": 20360
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.12214535402289e-05,
+      "loss": 0.0593,
+      "step": 20361
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.116941850744174e-05,
+      "loss": 0.1152,
+      "step": 20362
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.11174017888485e-05,
+      "loss": 0.1067,
+      "step": 20363
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.106540338547473e-05,
+      "loss": 0.105,
+      "step": 20364
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.101342329834548e-05,
+      "loss": 0.0952,
+      "step": 20365
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.096146152848604e-05,
+      "loss": 0.1082,
+      "step": 20366
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.090951807692081e-05,
+      "loss": 0.1007,
+      "step": 20367
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.085759294467386e-05,
+      "loss": 0.097,
+      "step": 20368
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.080568613276917e-05,
+      "loss": 0.1133,
+      "step": 20369
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.075379764223023e-05,
+      "loss": 0.0909,
+      "step": 20370
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.070192747408e-05,
+      "loss": 0.0909,
+      "step": 20371
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.065007562934111e-05,
+      "loss": 0.0945,
+      "step": 20372
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.05982421090362e-05,
+      "loss": 0.1171,
+      "step": 20373
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.054642691418734e-05,
+      "loss": 0.0857,
+      "step": 20374
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.049463004581613e-05,
+      "loss": 0.1298,
+      "step": 20375
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.04428515049439e-05,
+      "loss": 0.0568,
+      "step": 20376
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.039109129259147e-05,
+      "loss": 0.0658,
+      "step": 20377
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.033934940977949e-05,
+      "loss": 0.0931,
+      "step": 20378
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.028762585752802e-05,
+      "loss": 0.0988,
+      "step": 20379
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.023592063685713e-05,
+      "loss": 0.1285,
+      "step": 20380
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.018423374878635e-05,
+      "loss": 0.1372,
+      "step": 20381
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.013256519433475e-05,
+      "loss": 0.0663,
+      "step": 20382
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.008091497452085e-05,
+      "loss": 0.0863,
+      "step": 20383
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.002928309036349e-05,
+      "loss": 0.0678,
+      "step": 20384
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.997766954288032e-05,
+      "loss": 0.1198,
+      "step": 20385
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.99260743330894e-05,
+      "loss": 0.0805,
+      "step": 20386
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.987449746200791e-05,
+      "loss": 0.0582,
+      "step": 20387
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.982293893065272e-05,
+      "loss": 0.0761,
+      "step": 20388
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.977139874004057e-05,
+      "loss": 0.0731,
+      "step": 20389
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.971987689118753e-05,
+      "loss": 0.0981,
+      "step": 20390
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.966837338510967e-05,
+      "loss": 0.0957,
+      "step": 20391
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.961688822282231e-05,
+      "loss": 0.0806,
+      "step": 20392
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.956542140534072e-05,
+      "loss": 0.1224,
+      "step": 20393
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.951397293367968e-05,
+      "loss": 0.0737,
+      "step": 20394
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.946254280885378e-05,
+      "loss": 0.0898,
+      "step": 20395
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.941113103187691e-05,
+      "loss": 0.0546,
+      "step": 20396
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.93597376037628e-05,
+      "loss": 0.0625,
+      "step": 20397
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.930836252552485e-05,
+      "loss": 0.1173,
+      "step": 20398
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.925700579817584e-05,
+      "loss": 0.111,
+      "step": 20399
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.920566742272882e-05,
+      "loss": 0.0769,
+      "step": 20400
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.915434740019566e-05,
+      "loss": 0.0782,
+      "step": 20401
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.910304573158854e-05,
+      "loss": 0.0682,
+      "step": 20402
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.905176241791877e-05,
+      "loss": 0.1267,
+      "step": 20403
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.900049746019754e-05,
+      "loss": 0.1017,
+      "step": 20404
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.89492508594357e-05,
+      "loss": 0.1067,
+      "step": 20405
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.889802261664391e-05,
+      "loss": 0.0786,
+      "step": 20406
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.884681273283222e-05,
+      "loss": 0.097,
+      "step": 20407
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.879562120901006e-05,
+      "loss": 0.0681,
+      "step": 20408
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.874444804618718e-05,
+      "loss": 0.0792,
+      "step": 20409
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.869329324537221e-05,
+      "loss": 0.0669,
+      "step": 20410
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.864215680757402e-05,
+      "loss": 0.1333,
+      "step": 20411
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.859103873380079e-05,
+      "loss": 0.0752,
+      "step": 20412
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.853993902506051e-05,
+      "loss": 0.1165,
+      "step": 20413
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.84888576823608e-05,
+      "loss": 0.0681,
+      "step": 20414
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.843779470670864e-05,
+      "loss": 0.0681,
+      "step": 20415
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.838675009911111e-05,
+      "loss": 0.1132,
+      "step": 20416
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.833572386057451e-05,
+      "loss": 0.0756,
+      "step": 20417
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.828471599210507e-05,
+      "loss": 0.0891,
+      "step": 20418
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.823372649470827e-05,
+      "loss": 0.1047,
+      "step": 20419
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.818275536938989e-05,
+      "loss": 0.0625,
+      "step": 20420
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.813180261715479e-05,
+      "loss": 0.09,
+      "step": 20421
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.808086823900761e-05,
+      "loss": 0.1332,
+      "step": 20422
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.802995223595255e-05,
+      "loss": 0.0901,
+      "step": 20423
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.797905460899367e-05,
+      "loss": 0.0653,
+      "step": 20424
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.792817535913443e-05,
+      "loss": 0.1334,
+      "step": 20425
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.7877314487378e-05,
+      "loss": 0.1433,
+      "step": 20426
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.782647199472747e-05,
+      "loss": 0.0767,
+      "step": 20427
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.777564788218526e-05,
+      "loss": 0.0432,
+      "step": 20428
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.772484215075347e-05,
+      "loss": 0.0898,
+      "step": 20429
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.767405480143385e-05,
+      "loss": 0.1335,
+      "step": 20430
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.76232858352277e-05,
+      "loss": 0.0985,
+      "step": 20431
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.757253525313589e-05,
+      "loss": 0.0905,
+      "step": 20432
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.752180305615962e-05,
+      "loss": 0.0784,
+      "step": 20433
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.747108924529887e-05,
+      "loss": 0.0752,
+      "step": 20434
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.742039382155363e-05,
+      "loss": 0.0866,
+      "step": 20435
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.73697167859233e-05,
+      "loss": 0.0927,
+      "step": 20436
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.731905813940742e-05,
+      "loss": 0.0969,
+      "step": 20437
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.726841788300475e-05,
+      "loss": 0.0995,
+      "step": 20438
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.721779601771361e-05,
+      "loss": 0.0789,
+      "step": 20439
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.716719254453251e-05,
+      "loss": 0.0924,
+      "step": 20440
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.711660746445892e-05,
+      "loss": 0.0779,
+      "step": 20441
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.706604077849032e-05,
+      "loss": 0.131,
+      "step": 20442
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.701549248762373e-05,
+      "loss": 0.0772,
+      "step": 20443
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.696496259285589e-05,
+      "loss": 0.1015,
+      "step": 20444
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.69144510951829e-05,
+      "loss": 0.0641,
+      "step": 20445
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.686395799560118e-05,
+      "loss": 0.1046,
+      "step": 20446
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.681348329510584e-05,
+      "loss": 0.1002,
+      "step": 20447
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.67630269946925e-05,
+      "loss": 0.0578,
+      "step": 20448
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.671258909535582e-05,
+      "loss": 0.0879,
+      "step": 20449
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.666216959809046e-05,
+      "loss": 0.1214,
+      "step": 20450
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.66117685038904e-05,
+      "loss": 0.1118,
+      "step": 20451
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.656138581374937e-05,
+      "loss": 0.0438,
+      "step": 20452
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.651102152866095e-05,
+      "loss": 0.0541,
+      "step": 20453
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.646067564961821e-05,
+      "loss": 0.085,
+      "step": 20454
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.641034817761383e-05,
+      "loss": 0.0645,
+      "step": 20455
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.636003911364008e-05,
+      "loss": 0.0662,
+      "step": 20456
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.630974845868876e-05,
+      "loss": 0.0903,
+      "step": 20457
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.625947621375184e-05,
+      "loss": 0.095,
+      "step": 20458
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.62092223798202e-05,
+      "loss": 0.0771,
+      "step": 20459
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.615898695788502e-05,
+      "loss": 0.1202,
+      "step": 20460
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.610876994893677e-05,
+      "loss": 0.1379,
+      "step": 20461
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.605857135396542e-05,
+      "loss": 0.0825,
+      "step": 20462
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.600839117396096e-05,
+      "loss": 0.1211,
+      "step": 20463
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.595822940991269e-05,
+      "loss": 0.0755,
+      "step": 20464
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.590808606280951e-05,
+      "loss": 0.0972,
+      "step": 20465
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.58579611336405e-05,
+      "loss": 0.0864,
+      "step": 20466
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.580785462339378e-05,
+      "loss": 0.093,
+      "step": 20467
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.575776653305731e-05,
+      "loss": 0.0937,
+      "step": 20468
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.570769686361877e-05,
+      "loss": 0.136,
+      "step": 20469
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.565764561606546e-05,
+      "loss": 0.1191,
+      "step": 20470
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.560761279138416e-05,
+      "loss": 0.1201,
+      "step": 20471
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.55575983905613e-05,
+      "loss": 0.0789,
+      "step": 20472
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.55076024145832e-05,
+      "loss": 0.0881,
+      "step": 20473
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.545762486443574e-05,
+      "loss": 0.1519,
+      "step": 20474
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.540766574110412e-05,
+      "loss": 0.1025,
+      "step": 20475
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.535772504557358e-05,
+      "loss": 0.0693,
+      "step": 20476
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.530780277882875e-05,
+      "loss": 0.0922,
+      "step": 20477
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.525789894185386e-05,
+      "loss": 0.0927,
+      "step": 20478
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.520801353563311e-05,
+      "loss": 0.0955,
+      "step": 20479
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.515814656114983e-05,
+      "loss": 0.0747,
+      "step": 20480
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.510829801938756e-05,
+      "loss": 0.0795,
+      "step": 20481
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.505846791132919e-05,
+      "loss": 0.0865,
+      "step": 20482
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.500865623795704e-05,
+      "loss": 0.0736,
+      "step": 20483
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.495886300025333e-05,
+      "loss": 0.0868,
+      "step": 20484
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.490908819919983e-05,
+      "loss": 0.071,
+      "step": 20485
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.485933183577819e-05,
+      "loss": 0.0635,
+      "step": 20486
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.480959391096919e-05,
+      "loss": 0.0785,
+      "step": 20487
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.475987442575382e-05,
+      "loss": 0.0671,
+      "step": 20488
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.471017338111195e-05,
+      "loss": 0.0786,
+      "step": 20489
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.466049077802416e-05,
+      "loss": 0.0524,
+      "step": 20490
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.461082661746975e-05,
+      "loss": 0.0686,
+      "step": 20491
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.456118090042784e-05,
+      "loss": 0.0804,
+      "step": 20492
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.451155362787764e-05,
+      "loss": 0.0646,
+      "step": 20493
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.44619448007976e-05,
+      "loss": 0.1082,
+      "step": 20494
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.441235442016568e-05,
+      "loss": 0.0593,
+      "step": 20495
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.436278248695981e-05,
+      "loss": 0.0844,
+      "step": 20496
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.431322900215752e-05,
+      "loss": 0.1324,
+      "step": 20497
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.42636939667357e-05,
+      "loss": 0.0842,
+      "step": 20498
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.421417738167091e-05,
+      "loss": 0.0959,
+      "step": 20499
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.416467924793979e-05,
+      "loss": 0.0779,
+      "step": 20500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.411519956651835e-05,
+      "loss": 0.1154,
+      "step": 20501
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.406573833838203e-05,
+      "loss": 0.0892,
+      "step": 20502
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.401629556450628e-05,
+      "loss": 0.0532,
+      "step": 20503
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.396687124586575e-05,
+      "loss": 0.0937,
+      "step": 20504
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.39174653834349e-05,
+      "loss": 0.0652,
+      "step": 20505
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.386807797818816e-05,
+      "loss": 0.0865,
+      "step": 20506
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.381870903109931e-05,
+      "loss": 0.0674,
+      "step": 20507
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.37693585431417e-05,
+      "loss": 0.0877,
+      "step": 20508
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.37200265152883e-05,
+      "loss": 0.0793,
+      "step": 20509
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.367071294851178e-05,
+      "loss": 0.0805,
+      "step": 20510
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.362141784378473e-05,
+      "loss": 0.0771,
+      "step": 20511
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.357214120207889e-05,
+      "loss": 0.0799,
+      "step": 20512
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.352288302436604e-05,
+      "loss": 0.0815,
+      "step": 20513
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.347364331161731e-05,
+      "loss": 0.0665,
+      "step": 20514
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.342442206480381e-05,
+      "loss": 0.0828,
+      "step": 20515
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.337521928489575e-05,
+      "loss": 0.1019,
+      "step": 20516
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.332603497286337e-05,
+      "loss": 0.1243,
+      "step": 20517
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.327686912967656e-05,
+      "loss": 0.0713,
+      "step": 20518
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.322772175630465e-05,
+      "loss": 0.0991,
+      "step": 20519
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.317859285371675e-05,
+      "loss": 0.0978,
+      "step": 20520
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.312948242288141e-05,
+      "loss": 0.0601,
+      "step": 20521
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.308039046476733e-05,
+      "loss": 0.1019,
+      "step": 20522
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.303131698034215e-05,
+      "loss": 0.0919,
+      "step": 20523
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.298226197057367e-05,
+      "loss": 0.0859,
+      "step": 20524
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.293322543642876e-05,
+      "loss": 0.0656,
+      "step": 20525
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.288420737887479e-05,
+      "loss": 0.1036,
+      "step": 20526
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.283520779887808e-05,
+      "loss": 0.1051,
+      "step": 20527
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.278622669740475e-05,
+      "loss": 0.0603,
+      "step": 20528
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.273726407542069e-05,
+      "loss": 0.0773,
+      "step": 20529
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.268831993389102e-05,
+      "loss": 0.0767,
+      "step": 20530
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.263939427378096e-05,
+      "loss": 0.0797,
+      "step": 20531
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.259048709605542e-05,
+      "loss": 0.0744,
+      "step": 20532
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.254159840167828e-05,
+      "loss": 0.0911,
+      "step": 20533
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.2492728191614e-05,
+      "loss": 0.1171,
+      "step": 20534
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.24438764668258e-05,
+      "loss": 0.0519,
+      "step": 20535
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.239504322827716e-05,
+      "loss": 0.0674,
+      "step": 20536
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.234622847693083e-05,
+      "loss": 0.0836,
+      "step": 20537
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.22974322137494e-05,
+      "loss": 0.071,
+      "step": 20538
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.224865443969463e-05,
+      "loss": 0.0656,
+      "step": 20539
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.219989515572888e-05,
+      "loss": 0.1136,
+      "step": 20540
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.215115436281316e-05,
+      "loss": 0.0757,
+      "step": 20541
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.210243206190857e-05,
+      "loss": 0.0793,
+      "step": 20542
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.205372825397593e-05,
+      "loss": 0.0615,
+      "step": 20543
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.200504293997555e-05,
+      "loss": 0.1105,
+      "step": 20544
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.195637612086713e-05,
+      "loss": 0.0597,
+      "step": 20545
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.190772779761056e-05,
+      "loss": 0.1182,
+      "step": 20546
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.185909797116496e-05,
+      "loss": 0.0684,
+      "step": 20547
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.181048664248912e-05,
+      "loss": 0.0801,
+      "step": 20548
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.176189381254171e-05,
+      "loss": 0.0591,
+      "step": 20549
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.171331948228065e-05,
+      "loss": 0.106,
+      "step": 20550
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.166476365266371e-05,
+      "loss": 0.0617,
+      "step": 20551
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.161622632464825e-05,
+      "loss": 0.0836,
+      "step": 20552
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.15677074991916e-05,
+      "loss": 0.0564,
+      "step": 20553
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.151920717725001e-05,
+      "loss": 0.0785,
+      "step": 20554
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.147072535978026e-05,
+      "loss": 0.0761,
+      "step": 20555
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.142226204773804e-05,
+      "loss": 0.056,
+      "step": 20556
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.137381724207891e-05,
+      "loss": 0.0459,
+      "step": 20557
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.1325390943758e-05,
+      "loss": 0.0863,
+      "step": 20558
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.127698315373021e-05,
+      "loss": 0.0759,
+      "step": 20559
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.122859387295021e-05,
+      "loss": 0.0784,
+      "step": 20560
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.118022310237192e-05,
+      "loss": 0.0819,
+      "step": 20561
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.113187084294914e-05,
+      "loss": 0.1563,
+      "step": 20562
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.108353709563508e-05,
+      "loss": 0.1229,
+      "step": 20563
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.103522186138311e-05,
+      "loss": 0.0812,
+      "step": 20564
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0986925141145564e-05,
+      "loss": 0.0819,
+      "step": 20565
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0938646935875033e-05,
+      "loss": 0.0679,
+      "step": 20566
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.089038724652318e-05,
+      "loss": 0.0678,
+      "step": 20567
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.084214607404182e-05,
+      "loss": 0.113,
+      "step": 20568
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.079392341938184e-05,
+      "loss": 0.0918,
+      "step": 20569
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.074571928349426e-05,
+      "loss": 0.0725,
+      "step": 20570
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.069753366732955e-05,
+      "loss": 0.0765,
+      "step": 20571
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.064936657183762e-05,
+      "loss": 0.0593,
+      "step": 20572
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.060121799796847e-05,
+      "loss": 0.0601,
+      "step": 20573
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.055308794667125e-05,
+      "loss": 0.0688,
+      "step": 20574
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0504976418894966e-05,
+      "loss": 0.0761,
+      "step": 20575
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0456883415588524e-05,
+      "loss": 0.0992,
+      "step": 20576
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0408808937700064e-05,
+      "loss": 0.094,
+      "step": 20577
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.036075298617738e-05,
+      "loss": 0.0947,
+      "step": 20578
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.0312715561967936e-05,
+      "loss": 0.0645,
+      "step": 20579
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.026469666601919e-05,
+      "loss": 0.0856,
+      "step": 20580
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.021669629927784e-05,
+      "loss": 0.0588,
+      "step": 20581
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.016871446269034e-05,
+      "loss": 0.0546,
+      "step": 20582
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.012075115720272e-05,
+      "loss": 0.0879,
+      "step": 20583
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.007280638376056e-05,
+      "loss": 0.086,
+      "step": 20584
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.002488014330953e-05,
+      "loss": 0.0796,
+      "step": 20585
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.997697243679423e-05,
+      "loss": 0.0931,
+      "step": 20586
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.992908326515978e-05,
+      "loss": 0.0776,
+      "step": 20587
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.988121262934998e-05,
+      "loss": 0.0948,
+      "step": 20588
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.983336053030908e-05,
+      "loss": 0.1013,
+      "step": 20589
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.978552696898021e-05,
+      "loss": 0.0641,
+      "step": 20590
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.973771194630684e-05,
+      "loss": 0.1304,
+      "step": 20591
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.9689915463231545e-05,
+      "loss": 0.0562,
+      "step": 20592
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.964213752069691e-05,
+      "loss": 0.069,
+      "step": 20593
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.959437811964508e-05,
+      "loss": 0.0717,
+      "step": 20594
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.954663726101739e-05,
+      "loss": 0.075,
+      "step": 20595
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.9498914945755544e-05,
+      "loss": 0.0573,
+      "step": 20596
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.945121117480023e-05,
+      "loss": 0.0746,
+      "step": 20597
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.9403525949092153e-05,
+      "loss": 0.0698,
+      "step": 20598
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.9355859269571766e-05,
+      "loss": 0.1013,
+      "step": 20599
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.930821113717866e-05,
+      "loss": 0.0657,
+      "step": 20600
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.92605815528523e-05,
+      "loss": 0.1044,
+      "step": 20601
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.921297051753205e-05,
+      "loss": 0.1113,
+      "step": 20602
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.9165378032156604e-05,
+      "loss": 0.0829,
+      "step": 20603
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.911780409766421e-05,
+      "loss": 0.0626,
+      "step": 20604
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.907024871499289e-05,
+      "loss": 0.1266,
+      "step": 20605
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.902271188508068e-05,
+      "loss": 0.1016,
+      "step": 20606
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8975193608864384e-05,
+      "loss": 0.0607,
+      "step": 20607
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.892769388728137e-05,
+      "loss": 0.093,
+      "step": 20608
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.888021272126809e-05,
+      "loss": 0.1021,
+      "step": 20609
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8832750111760705e-05,
+      "loss": 0.081,
+      "step": 20610
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.878530605969512e-05,
+      "loss": 0.1044,
+      "step": 20611
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.873788056600649e-05,
+      "loss": 0.0732,
+      "step": 20612
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8690473631630384e-05,
+      "loss": 0.0695,
+      "step": 20613
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.86430852575014e-05,
+      "loss": 0.1063,
+      "step": 20614
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.859571544455378e-05,
+      "loss": 0.1004,
+      "step": 20615
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8548364193721556e-05,
+      "loss": 0.1014,
+      "step": 20616
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.850103150593866e-05,
+      "loss": 0.0833,
+      "step": 20617
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8453717382137897e-05,
+      "loss": 0.0699,
+      "step": 20618
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.840642182325262e-05,
+      "loss": 0.108,
+      "step": 20619
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8359144830215205e-05,
+      "loss": 0.0931,
+      "step": 20620
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.83118864039579e-05,
+      "loss": 0.1117,
+      "step": 20621
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.826464654541241e-05,
+      "loss": 0.0973,
+      "step": 20622
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.821742525551022e-05,
+      "loss": 0.1061,
+      "step": 20623
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.817022253518234e-05,
+      "loss": 0.0662,
+      "step": 20624
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8123038385359594e-05,
+      "loss": 0.1335,
+      "step": 20625
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.807587280697235e-05,
+      "loss": 0.0593,
+      "step": 20626
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.8028725800950646e-05,
+      "loss": 0.1129,
+      "step": 20627
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7981597368223836e-05,
+      "loss": 0.1127,
+      "step": 20628
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.793448750972141e-05,
+      "loss": 0.1128,
+      "step": 20629
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.788739622637229e-05,
+      "loss": 0.1185,
+      "step": 20630
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7840323519104845e-05,
+      "loss": 0.0951,
+      "step": 20631
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7793269388847215e-05,
+      "loss": 0.0995,
+      "step": 20632
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.774623383652744e-05,
+      "loss": 0.0941,
+      "step": 20633
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.769921686307267e-05,
+      "loss": 0.0706,
+      "step": 20634
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7652218469410155e-05,
+      "loss": 0.1251,
+      "step": 20635
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.760523865646638e-05,
+      "loss": 0.0735,
+      "step": 20636
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7558277425167725e-05,
+      "loss": 0.1092,
+      "step": 20637
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.751133477644022e-05,
+      "loss": 0.0823,
+      "step": 20638
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7464410711209444e-05,
+      "loss": 0.0933,
+      "step": 20639
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.741750523040068e-05,
+      "loss": 0.0647,
+      "step": 20640
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.737061833493862e-05,
+      "loss": 0.0991,
+      "step": 20641
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.732375002574797e-05,
+      "loss": 0.0651,
+      "step": 20642
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.7276900303752764e-05,
+      "loss": 0.0876,
+      "step": 20643
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.7230069169876606e-05,
+      "loss": 0.084,
+      "step": 20644
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.718325662504298e-05,
+      "loss": 0.0737,
+      "step": 20645
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.713646267017503e-05,
+      "loss": 0.0895,
+      "step": 20646
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.708968730619535e-05,
+      "loss": 0.1133,
+      "step": 20647
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.7042930534026204e-05,
+      "loss": 0.1165,
+      "step": 20648
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.699619235458942e-05,
+      "loss": 0.0746,
+      "step": 20649
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6949472768806796e-05,
+      "loss": 0.0945,
+      "step": 20650
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6902771777599286e-05,
+      "loss": 0.1073,
+      "step": 20651
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.68560893818878e-05,
+      "loss": 0.0764,
+      "step": 20652
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.680942558259294e-05,
+      "loss": 0.0555,
+      "step": 20653
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.676278038063465e-05,
+      "loss": 0.0577,
+      "step": 20654
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6716153776932625e-05,
+      "loss": 0.0824,
+      "step": 20655
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.666954577240635e-05,
+      "loss": 0.0614,
+      "step": 20656
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6622956367974764e-05,
+      "loss": 0.1128,
+      "step": 20657
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.657638556455635e-05,
+      "loss": 0.0807,
+      "step": 20658
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.652983336306949e-05,
+      "loss": 0.0973,
+      "step": 20659
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.6483299764432116e-05,
+      "loss": 0.0731,
+      "step": 20660
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.643678476956171e-05,
+      "loss": 0.131,
+      "step": 20661
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.639028837937554e-05,
+      "loss": 0.0767,
+      "step": 20662
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.634381059479021e-05,
+      "loss": 0.1262,
+      "step": 20663
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.629735141672232e-05,
+      "loss": 0.0742,
+      "step": 20664
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.625091084608758e-05,
+      "loss": 0.0854,
+      "step": 20665
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.620448888380214e-05,
+      "loss": 0.0772,
+      "step": 20666
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.615808553078117e-05,
+      "loss": 0.0604,
+      "step": 20667
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.611170078793948e-05,
+      "loss": 0.0648,
+      "step": 20668
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.606533465619179e-05,
+      "loss": 0.0665,
+      "step": 20669
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.601898713645215e-05,
+      "loss": 0.0758,
+      "step": 20670
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.597265822963471e-05,
+      "loss": 0.1176,
+      "step": 20671
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5926347936652633e-05,
+      "loss": 0.1155,
+      "step": 20672
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.588005625841941e-05,
+      "loss": 0.0878,
+      "step": 20673
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5833783195847645e-05,
+      "loss": 0.0577,
+      "step": 20674
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.57875287498496e-05,
+      "loss": 0.0591,
+      "step": 20675
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.574129292133734e-05,
+      "loss": 0.1448,
+      "step": 20676
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5695075711222564e-05,
+      "loss": 0.0818,
+      "step": 20677
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.564887712041644e-05,
+      "loss": 0.108,
+      "step": 20678
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.560269714983013e-05,
+      "loss": 0.0568,
+      "step": 20679
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.555653580037401e-05,
+      "loss": 0.1663,
+      "step": 20680
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5510393072958245e-05,
+      "loss": 0.0688,
+      "step": 20681
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.546426896849288e-05,
+      "loss": 0.0759,
+      "step": 20682
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.5418163487887084e-05,
+      "loss": 0.0717,
+      "step": 20683
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.537207663205013e-05,
+      "loss": 0.1011,
+      "step": 20684
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.53260084018905e-05,
+      "loss": 0.058,
+      "step": 20685
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.527995879831693e-05,
+      "loss": 0.0624,
+      "step": 20686
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.523392782223702e-05,
+      "loss": 0.1095,
+      "step": 20687
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.51879154745587e-05,
+      "loss": 0.0903,
+      "step": 20688
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.514192175618893e-05,
+      "loss": 0.0943,
+      "step": 20689
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.509594666803463e-05,
+      "loss": 0.1219,
+      "step": 20690
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.5049990211002544e-05,
+      "loss": 0.0482,
+      "step": 20691
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.500405238599859e-05,
+      "loss": 0.0915,
+      "step": 20692
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.495813319392873e-05,
+      "loss": 0.0847,
+      "step": 20693
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.491223263569822e-05,
+      "loss": 0.1046,
+      "step": 20694
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.486635071221224e-05,
+      "loss": 0.0561,
+      "step": 20695
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.48204874243754e-05,
+      "loss": 0.0569,
+      "step": 20696
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.477464277309196e-05,
+      "loss": 0.0781,
+      "step": 20697
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.472881675926578e-05,
+      "loss": 0.0796,
+      "step": 20698
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.468300938380055e-05,
+      "loss": 0.077,
+      "step": 20699
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4637220647599576e-05,
+      "loss": 0.0432,
+      "step": 20700
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.459145055156567e-05,
+      "loss": 0.0544,
+      "step": 20701
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.454569909660101e-05,
+      "loss": 0.0852,
+      "step": 20702
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.449996628360798e-05,
+      "loss": 0.0698,
+      "step": 20703
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.445425211348842e-05,
+      "loss": 0.0533,
+      "step": 20704
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4408556587143275e-05,
+      "loss": 0.0565,
+      "step": 20705
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.436287970547393e-05,
+      "loss": 0.0904,
+      "step": 20706
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4317221469380895e-05,
+      "loss": 0.0717,
+      "step": 20707
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.427158187976444e-05,
+      "loss": 0.0967,
+      "step": 20708
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.422596093752452e-05,
+      "loss": 0.0547,
+      "step": 20709
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.418035864356052e-05,
+      "loss": 0.0638,
+      "step": 20710
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.4134774998771505e-05,
+      "loss": 0.0551,
+      "step": 20711
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.408921000405664e-05,
+      "loss": 0.1127,
+      "step": 20712
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.404366366031399e-05,
+      "loss": 0.1217,
+      "step": 20713
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3998135968441943e-05,
+      "loss": 0.0591,
+      "step": 20714
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3952626929338e-05,
+      "loss": 0.1129,
+      "step": 20715
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.390713654389956e-05,
+      "loss": 0.0894,
+      "step": 20716
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.386166481302346e-05,
+      "loss": 0.0856,
+      "step": 20717
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.38162117376062e-05,
+      "loss": 0.0641,
+      "step": 20718
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3770777318544406e-05,
+      "loss": 0.0807,
+      "step": 20719
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.372536155673358e-05,
+      "loss": 0.1017,
+      "step": 20720
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.367996445306933e-05,
+      "loss": 0.1002,
+      "step": 20721
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3634586008446726e-05,
+      "loss": 0.071,
+      "step": 20722
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.358922622376039e-05,
+      "loss": 0.0577,
+      "step": 20723
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.354388509990504e-05,
+      "loss": 0.0652,
+      "step": 20724
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3498562637774304e-05,
+      "loss": 0.1023,
+      "step": 20725
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.345325883826224e-05,
+      "loss": 0.0728,
+      "step": 20726
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.340797370226191e-05,
+      "loss": 0.0814,
+      "step": 20727
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.336270723066616e-05,
+      "loss": 0.0753,
+      "step": 20728
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.33174594243675e-05,
+      "loss": 0.0618,
+      "step": 20729
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.327223028425832e-05,
+      "loss": 0.0455,
+      "step": 20730
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3227019811230236e-05,
+      "loss": 0.0738,
+      "step": 20731
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3181828006174656e-05,
+      "loss": 0.1187,
+      "step": 20732
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.313665486998287e-05,
+      "loss": 0.0765,
+      "step": 20733
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.3091500403545264e-05,
+      "loss": 0.1185,
+      "step": 20734
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.304636460775247e-05,
+      "loss": 0.1013,
+      "step": 20735
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.300124748349433e-05,
+      "loss": 0.0625,
+      "step": 20736
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2956149031660574e-05,
+      "loss": 0.1086,
+      "step": 20737
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.291106925314004e-05,
+      "loss": 0.0564,
+      "step": 20738
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.286600814882192e-05,
+      "loss": 0.08,
+      "step": 20739
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.282096571959472e-05,
+      "loss": 0.0819,
+      "step": 20740
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.27759419663465e-05,
+      "loss": 0.0487,
+      "step": 20741
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2730936889965e-05,
+      "loss": 0.0853,
+      "step": 20742
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.268595049133751e-05,
+      "loss": 0.0935,
+      "step": 20743
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2640982771350984e-05,
+      "loss": 0.1176,
+      "step": 20744
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.259603373089228e-05,
+      "loss": 0.0651,
+      "step": 20745
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.255110337084778e-05,
+      "loss": 0.0799,
+      "step": 20746
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2506191692103246e-05,
+      "loss": 0.0604,
+      "step": 20747
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.246129869554417e-05,
+      "loss": 0.0671,
+      "step": 20748
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2416424382055736e-05,
+      "loss": 0.1346,
+      "step": 20749
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2371568752522804e-05,
+      "loss": 0.0963,
+      "step": 20750
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.232673180782988e-05,
+      "loss": 0.0813,
+      "step": 20751
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2281913548860694e-05,
+      "loss": 0.0519,
+      "step": 20752
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.223711397649944e-05,
+      "loss": 0.1715,
+      "step": 20753
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.219233309162919e-05,
+      "loss": 0.0967,
+      "step": 20754
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.214757089513278e-05,
+      "loss": 0.0717,
+      "step": 20755
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.210282738789318e-05,
+      "loss": 0.0768,
+      "step": 20756
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.2058102570792356e-05,
+      "loss": 0.0922,
+      "step": 20757
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.201339644471204e-05,
+      "loss": 0.0836,
+      "step": 20758
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1968709010534075e-05,
+      "loss": 0.0881,
+      "step": 20759
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1924040269139436e-05,
+      "loss": 0.0764,
+      "step": 20760
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1879390221408865e-05,
+      "loss": 0.1085,
+      "step": 20761
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.183475886822264e-05,
+      "loss": 0.1281,
+      "step": 20762
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1790146210460965e-05,
+      "loss": 0.1042,
+      "step": 20763
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.174555224900346e-05,
+      "loss": 0.0736,
+      "step": 20764
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.170097698472909e-05,
+      "loss": 0.0673,
+      "step": 20765
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.165642041851704e-05,
+      "loss": 0.0714,
+      "step": 20766
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.161188255124605e-05,
+      "loss": 0.0711,
+      "step": 20767
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.156736338379397e-05,
+      "loss": 0.0692,
+      "step": 20768
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.152286291703878e-05,
+      "loss": 0.0844,
+      "step": 20769
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1478381151857876e-05,
+      "loss": 0.0683,
+      "step": 20770
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.143391808912823e-05,
+      "loss": 0.0565,
+      "step": 20771
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1389473729726574e-05,
+      "loss": 0.0784,
+      "step": 20772
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.134504807452933e-05,
+      "loss": 0.116,
+      "step": 20773
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1300641124412464e-05,
+      "loss": 0.0678,
+      "step": 20774
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.125625288025149e-05,
+      "loss": 0.0995,
+      "step": 20775
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.121188334292148e-05,
+      "loss": 0.1068,
+      "step": 20776
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.1167532513297645e-05,
+      "loss": 0.1172,
+      "step": 20777
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.112320039225415e-05,
+      "loss": 0.0618,
+      "step": 20778
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.107888698066543e-05,
+      "loss": 0.0609,
+      "step": 20779
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.103459227940499e-05,
+      "loss": 0.0916,
+      "step": 20780
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.099031628934636e-05,
+      "loss": 0.1031,
+      "step": 20781
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.09460590113624e-05,
+      "loss": 0.076,
+      "step": 20782
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.090182044632574e-05,
+      "loss": 0.0624,
+      "step": 20783
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.085760059510891e-05,
+      "loss": 0.0854,
+      "step": 20784
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.081339945858332e-05,
+      "loss": 0.075,
+      "step": 20785
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.076921703762105e-05,
+      "loss": 0.0853,
+      "step": 20786
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.072505333309296e-05,
+      "loss": 0.1022,
+      "step": 20787
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.068090834586991e-05,
+      "loss": 0.0674,
+      "step": 20788
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.063678207682243e-05,
+      "loss": 0.0604,
+      "step": 20789
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0592674526820595e-05,
+      "loss": 0.0789,
+      "step": 20790
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.054858569673382e-05,
+      "loss": 0.0679,
+      "step": 20791
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.050451558743152e-05,
+      "loss": 0.0514,
+      "step": 20792
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0460464199782896e-05,
+      "loss": 0.0685,
+      "step": 20793
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.041643153465625e-05,
+      "loss": 0.0674,
+      "step": 20794
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.037241759291999e-05,
+      "loss": 0.0637,
+      "step": 20795
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0328422375441775e-05,
+      "loss": 0.0612,
+      "step": 20796
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.028444588308911e-05,
+      "loss": 0.0602,
+      "step": 20797
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.024048811672921e-05,
+      "loss": 0.0922,
+      "step": 20798
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.019654907722881e-05,
+      "loss": 0.068,
+      "step": 20799
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0152628765454345e-05,
+      "loss": 0.121,
+      "step": 20800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.0108727182271554e-05,
+      "loss": 0.085,
+      "step": 20801
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.00648443285463e-05,
+      "loss": 0.0638,
+      "step": 20802
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.002098020514378e-05,
+      "loss": 0.0526,
+      "step": 20803
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.997713481292876e-05,
+      "loss": 0.0931,
+      "step": 20804
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9933308152765753e-05,
+      "loss": 0.0968,
+      "step": 20805
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.988950022551919e-05,
+      "loss": 0.0958,
+      "step": 20806
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.984571103205271e-05,
+      "loss": 0.0455,
+      "step": 20807
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.980194057322951e-05,
+      "loss": 0.1194,
+      "step": 20808
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.97581888499129e-05,
+      "loss": 0.0907,
+      "step": 20809
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9714455862965525e-05,
+      "loss": 0.0963,
+      "step": 20810
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.967074161324958e-05,
+      "loss": 0.1111,
+      "step": 20811
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.962704610162705e-05,
+      "loss": 0.0988,
+      "step": 20812
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.958336932895957e-05,
+      "loss": 0.0939,
+      "step": 20813
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.953971129610824e-05,
+      "loss": 0.1177,
+      "step": 20814
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.949607200393391e-05,
+      "loss": 0.1022,
+      "step": 20815
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.945245145329702e-05,
+      "loss": 0.077,
+      "step": 20816
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9408849645057766e-05,
+      "loss": 0.0942,
+      "step": 20817
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9365266580075566e-05,
+      "loss": 0.104,
+      "step": 20818
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9321702259209956e-05,
+      "loss": 0.0719,
+      "step": 20819
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.927815668332014e-05,
+      "loss": 0.127,
+      "step": 20820
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9234629853264545e-05,
+      "loss": 0.0769,
+      "step": 20821
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.919112176990137e-05,
+      "loss": 0.089,
+      "step": 20822
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9147632434088486e-05,
+      "loss": 0.1083,
+      "step": 20823
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9104161846683424e-05,
+      "loss": 0.0817,
+      "step": 20824
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.906071000854318e-05,
+      "loss": 0.0453,
+      "step": 20825
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.9017276920524714e-05,
+      "loss": 0.0717,
+      "step": 20826
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.897386258348435e-05,
+      "loss": 0.0857,
+      "step": 20827
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.893046699827819e-05,
+      "loss": 0.066,
+      "step": 20828
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8887090165761537e-05,
+      "loss": 0.0677,
+      "step": 20829
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.884373208679016e-05,
+      "loss": 0.0895,
+      "step": 20830
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8800392762218704e-05,
+      "loss": 0.0886,
+      "step": 20831
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8757072192901594e-05,
+      "loss": 0.072,
+      "step": 20832
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8713770379693265e-05,
+      "loss": 0.0547,
+      "step": 20833
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.867048732344748e-05,
+      "loss": 0.0975,
+      "step": 20834
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.862722302501765e-05,
+      "loss": 0.1149,
+      "step": 20835
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.858397748525667e-05,
+      "loss": 0.0505,
+      "step": 20836
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.85407507050174e-05,
+      "loss": 0.0424,
+      "step": 20837
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.849754268515194e-05,
+      "loss": 0.0861,
+      "step": 20838
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.845435342651261e-05,
+      "loss": 0.0886,
+      "step": 20839
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.841118292995072e-05,
+      "loss": 0.0621,
+      "step": 20840
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.8368031196317605e-05,
+      "loss": 0.059,
+      "step": 20841
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.832489822646402e-05,
+      "loss": 0.0778,
+      "step": 20842
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.828178402124062e-05,
+      "loss": 0.0825,
+      "step": 20843
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.823868858149727e-05,
+      "loss": 0.0798,
+      "step": 20844
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.819561190808375e-05,
+      "loss": 0.0944,
+      "step": 20845
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.815255400184959e-05,
+      "loss": 0.0738,
+      "step": 20846
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.810951486364368e-05,
+      "loss": 0.0704,
+      "step": 20847
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.806649449431455e-05,
+      "loss": 0.0441,
+      "step": 20848
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.802349289471064e-05,
+      "loss": 0.082,
+      "step": 20849
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7980510065679605e-05,
+      "loss": 0.1207,
+      "step": 20850
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.793754600806921e-05,
+      "loss": 0.0922,
+      "step": 20851
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7894600722726336e-05,
+      "loss": 0.1038,
+      "step": 20852
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7851674210497964e-05,
+      "loss": 0.0984,
+      "step": 20853
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7808766472230425e-05,
+      "loss": 0.0833,
+      "step": 20854
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.776587750876982e-05,
+      "loss": 0.0878,
+      "step": 20855
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.77230073209618e-05,
+      "loss": 0.0444,
+      "step": 20856
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7680155909651465e-05,
+      "loss": 0.0901,
+      "step": 20857
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.763732327568382e-05,
+      "loss": 0.083,
+      "step": 20858
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.759450941990362e-05,
+      "loss": 0.1319,
+      "step": 20859
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.755171434315475e-05,
+      "loss": 0.0644,
+      "step": 20860
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.75089380462812e-05,
+      "loss": 0.0757,
+      "step": 20861
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7466180530126414e-05,
+      "loss": 0.0801,
+      "step": 20862
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.742344179553348e-05,
+      "loss": 0.0769,
+      "step": 20863
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.738072184334496e-05,
+      "loss": 0.0624,
+      "step": 20864
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.733802067440318e-05,
+      "loss": 0.085,
+      "step": 20865
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.729533828955035e-05,
+      "loss": 0.0945,
+      "step": 20866
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7252674689627905e-05,
+      "loss": 0.088,
+      "step": 20867
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7210029875476954e-05,
+      "loss": 0.0529,
+      "step": 20868
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.716740384793861e-05,
+      "loss": 0.0444,
+      "step": 20869
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.712479660785307e-05,
+      "loss": 0.079,
+      "step": 20870
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.708220815606035e-05,
+      "loss": 0.0669,
+      "step": 20871
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.703963849340065e-05,
+      "loss": 0.0943,
+      "step": 20872
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.699708762071286e-05,
+      "loss": 0.1021,
+      "step": 20873
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.695455553883643e-05,
+      "loss": 0.0913,
+      "step": 20874
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6912042248609675e-05,
+      "loss": 0.0608,
+      "step": 20875
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6869547750870934e-05,
+      "loss": 0.048,
+      "step": 20876
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6827072046458084e-05,
+      "loss": 0.0672,
+      "step": 20877
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.678461513620846e-05,
+      "loss": 0.0848,
+      "step": 20878
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.674217702095962e-05,
+      "loss": 0.0533,
+      "step": 20879
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6699757701548e-05,
+      "loss": 0.0648,
+      "step": 20880
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6657357178810147e-05,
+      "loss": 0.1057,
+      "step": 20881
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6614975453581845e-05,
+      "loss": 0.081,
+      "step": 20882
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6572612526699196e-05,
+      "loss": 0.0783,
+      "step": 20883
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.65302683989971e-05,
+      "loss": 0.1411,
+      "step": 20884
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.648794307131066e-05,
+      "loss": 0.0894,
+      "step": 20885
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6445636544474425e-05,
+      "loss": 0.1201,
+      "step": 20886
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.640334881932251e-05,
+      "loss": 0.0529,
+      "step": 20887
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6361079896688805e-05,
+      "loss": 0.0773,
+      "step": 20888
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.631882977740664e-05,
+      "loss": 0.0765,
+      "step": 20889
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.627659846230925e-05,
+      "loss": 0.0641,
+      "step": 20890
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6234385952228955e-05,
+      "loss": 0.1355,
+      "step": 20891
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6192192247998534e-05,
+      "loss": 0.1017,
+      "step": 20892
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6150017350449545e-05,
+      "loss": 0.1541,
+      "step": 20893
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6107861260414e-05,
+      "loss": 0.09,
+      "step": 20894
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.606572397872277e-05,
+      "loss": 0.0891,
+      "step": 20895
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.602360550620688e-05,
+      "loss": 0.0634,
+      "step": 20896
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.598150584369665e-05,
+      "loss": 0.0913,
+      "step": 20897
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5939424992022194e-05,
+      "loss": 0.0862,
+      "step": 20898
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.589736295201341e-05,
+      "loss": 0.0999,
+      "step": 20899
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.585531972449952e-05,
+      "loss": 0.12,
+      "step": 20900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5813295310309535e-05,
+      "loss": 0.092,
+      "step": 20901
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5771289710272115e-05,
+      "loss": 0.1026,
+      "step": 20902
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.572930292521527e-05,
+      "loss": 0.1135,
+      "step": 20903
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.568733495596722e-05,
+      "loss": 0.1002,
+      "step": 20904
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.564538580335509e-05,
+      "loss": 0.0793,
+      "step": 20905
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5603455468206435e-05,
+      "loss": 0.1306,
+      "step": 20906
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.556154395134782e-05,
+      "loss": 0.0923,
+      "step": 20907
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5519651253605686e-05,
+      "loss": 0.0533,
+      "step": 20908
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.547777737580594e-05,
+      "loss": 0.1425,
+      "step": 20909
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5435922318774246e-05,
+      "loss": 0.0939,
+      "step": 20910
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.539408608333584e-05,
+      "loss": 0.0947,
+      "step": 20911
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.535226867031572e-05,
+      "loss": 0.0938,
+      "step": 20912
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.531047008053846e-05,
+      "loss": 0.1005,
+      "step": 20913
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5268690314828054e-05,
+      "loss": 0.1368,
+      "step": 20914
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.522692937400852e-05,
+      "loss": 0.1069,
+      "step": 20915
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5185187258903196e-05,
+      "loss": 0.0583,
+      "step": 20916
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5143463970335084e-05,
+      "loss": 0.1017,
+      "step": 20917
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.510175950912676e-05,
+      "loss": 0.1022,
+      "step": 20918
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.506007387610067e-05,
+      "loss": 0.0636,
+      "step": 20919
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.5018407072078826e-05,
+      "loss": 0.063,
+      "step": 20920
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.497675909788268e-05,
+      "loss": 0.1317,
+      "step": 20921
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4935129954333464e-05,
+      "loss": 0.1138,
+      "step": 20922
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4893519642251855e-05,
+      "loss": 0.0467,
+      "step": 20923
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.485192816245842e-05,
+      "loss": 0.1006,
+      "step": 20924
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.481035551577328e-05,
+      "loss": 0.076,
+      "step": 20925
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4768801703016e-05,
+      "loss": 0.0934,
+      "step": 20926
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.472726672500604e-05,
+      "loss": 0.0624,
+      "step": 20927
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4685750582562524e-05,
+      "loss": 0.0663,
+      "step": 20928
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.464425327650379e-05,
+      "loss": 0.0591,
+      "step": 20929
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.460277480764807e-05,
+      "loss": 0.129,
+      "step": 20930
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4561315176813165e-05,
+      "loss": 0.0743,
+      "step": 20931
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.451987438481686e-05,
+      "loss": 0.0681,
+      "step": 20932
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4478452432476056e-05,
+      "loss": 0.0824,
+      "step": 20933
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4437049320607435e-05,
+      "loss": 0.1304,
+      "step": 20934
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.439566505002735e-05,
+      "loss": 0.0699,
+      "step": 20935
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.435429962155202e-05,
+      "loss": 0.0688,
+      "step": 20936
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4312953035996914e-05,
+      "loss": 0.0781,
+      "step": 20937
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4271625294177165e-05,
+      "loss": 0.0845,
+      "step": 20938
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.423031639690789e-05,
+      "loss": 0.0535,
+      "step": 20939
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4189026345003436e-05,
+      "loss": 0.0968,
+      "step": 20940
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.414775513927804e-05,
+      "loss": 0.0775,
+      "step": 20941
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.41065027805454e-05,
+      "loss": 0.1451,
+      "step": 20942
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.406526926961885e-05,
+      "loss": 0.0818,
+      "step": 20943
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4024054607311515e-05,
+      "loss": 0.1094,
+      "step": 20944
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3982858794435976e-05,
+      "loss": 0.1185,
+      "step": 20945
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3941681831804584e-05,
+      "loss": 0.104,
+      "step": 20946
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.390052372022913e-05,
+      "loss": 0.1721,
+      "step": 20947
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.385938446052129e-05,
+      "loss": 0.0977,
+      "step": 20948
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.38182640534921e-05,
+      "loss": 0.111,
+      "step": 20949
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.377716249995256e-05,
+      "loss": 0.0812,
+      "step": 20950
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3736079800712705e-05,
+      "loss": 0.1426,
+      "step": 20951
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3695015956582986e-05,
+      "loss": 0.1521,
+      "step": 20952
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.365397096837287e-05,
+      "loss": 0.1131,
+      "step": 20953
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.361294483689171e-05,
+      "loss": 0.1555,
+      "step": 20954
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3571937562948414e-05,
+      "loss": 0.0894,
+      "step": 20955
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.353094914735145e-05,
+      "loss": 0.0753,
+      "step": 20956
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.348997959090928e-05,
+      "loss": 0.0664,
+      "step": 20957
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.344902889442937e-05,
+      "loss": 0.1323,
+      "step": 20958
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.34080970587194e-05,
+      "loss": 0.0769,
+      "step": 20959
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.336718408458651e-05,
+      "loss": 0.0648,
+      "step": 20960
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.332628997283716e-05,
+      "loss": 0.1095,
+      "step": 20961
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3285414724277825e-05,
+      "loss": 0.1104,
+      "step": 20962
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3244558339714525e-05,
+      "loss": 0.1048,
+      "step": 20963
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.320372081995272e-05,
+      "loss": 0.0829,
+      "step": 20964
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3162902165797435e-05,
+      "loss": 0.0548,
+      "step": 20965
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3122102378053915e-05,
+      "loss": 0.0928,
+      "step": 20966
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.308132145752641e-05,
+      "loss": 0.0807,
+      "step": 20967
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.304055940501905e-05,
+      "loss": 0.0679,
+      "step": 20968
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.299981622133553e-05,
+      "loss": 0.1034,
+      "step": 20969
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.295909190727931e-05,
+      "loss": 0.1321,
+      "step": 20970
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2918386463653205e-05,
+      "loss": 0.0909,
+      "step": 20971
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.287769989126e-05,
+      "loss": 0.1096,
+      "step": 20972
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2837032190901735e-05,
+      "loss": 0.0861,
+      "step": 20973
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2796383363380545e-05,
+      "loss": 0.0852,
+      "step": 20974
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2755753409497665e-05,
+      "loss": 0.0607,
+      "step": 20975
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.271514233005436e-05,
+      "loss": 0.1332,
+      "step": 20976
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2674550125851196e-05,
+      "loss": 0.0784,
+      "step": 20977
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2633976797688765e-05,
+      "loss": 0.0813,
+      "step": 20978
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2593422346366874e-05,
+      "loss": 0.0753,
+      "step": 20979
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.255288677268543e-05,
+      "loss": 0.1412,
+      "step": 20980
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2512370077443464e-05,
+      "loss": 0.0774,
+      "step": 20981
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2471872261440005e-05,
+      "loss": 0.0906,
+      "step": 20982
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.243139332547341e-05,
+      "loss": 0.1,
+      "step": 20983
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2390933270341934e-05,
+      "loss": 0.0987,
+      "step": 20984
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2350492096843274e-05,
+      "loss": 0.0845,
+      "step": 20985
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2310069805774896e-05,
+      "loss": 0.0743,
+      "step": 20986
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2269666397933836e-05,
+      "loss": 0.0826,
+      "step": 20987
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.222928187411656e-05,
+      "loss": 0.1014,
+      "step": 20988
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2188916235119665e-05,
+      "loss": 0.0939,
+      "step": 20989
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.214856948173884e-05,
+      "loss": 0.0812,
+      "step": 20990
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.210824161476956e-05,
+      "loss": 0.0994,
+      "step": 20991
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.20679326350073e-05,
+      "loss": 0.0931,
+      "step": 20992
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.202764254324654e-05,
+      "loss": 0.1523,
+      "step": 20993
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.198737134028186e-05,
+      "loss": 0.0903,
+      "step": 20994
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1947119026907285e-05,
+      "loss": 0.0851,
+      "step": 20995
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.190688560391642e-05,
+      "loss": 0.109,
+      "step": 20996
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1866671072102735e-05,
+      "loss": 0.0789,
+      "step": 20997
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.18264754322587e-05,
+      "loss": 0.0936,
+      "step": 20998
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.178629868517747e-05,
+      "loss": 0.0753,
+      "step": 20999
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.174614083165085e-05,
+      "loss": 0.0903,
+      "step": 21000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1706001872470864e-05,
+      "loss": 0.1094,
+      "step": 21001
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.166588180842878e-05,
+      "loss": 0.0847,
+      "step": 21002
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.162578064031586e-05,
+      "loss": 0.0938,
+      "step": 21003
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.158569836892256e-05,
+      "loss": 0.1023,
+      "step": 21004
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1545634995039273e-05,
+      "loss": 0.0853,
+      "step": 21005
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1505590519456125e-05,
+      "loss": 0.0785,
+      "step": 21006
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.14655649429625e-05,
+      "loss": 0.1494,
+      "step": 21007
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1425558266347644e-05,
+      "loss": 0.1008,
+      "step": 21008
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.138557049040037e-05,
+      "loss": 0.0782,
+      "step": 21009
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.134560161590917e-05,
+      "loss": 0.0685,
+      "step": 21010
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.130565164366207e-05,
+      "loss": 0.1532,
+      "step": 21011
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1265720574447e-05,
+      "loss": 0.1218,
+      "step": 21012
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1225808409051105e-05,
+      "loss": 0.0913,
+      "step": 21013
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.118591514826131e-05,
+      "loss": 0.1051,
+      "step": 21014
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1146040792864436e-05,
+      "loss": 0.1429,
+      "step": 21015
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.1106185343646405e-05,
+      "loss": 0.061,
+      "step": 21016
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.106634880139337e-05,
+      "loss": 0.1489,
+      "step": 21017
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.102653116689048e-05,
+      "loss": 0.176,
+      "step": 21018
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.09867324409231e-05,
+      "loss": 0.1206,
+      "step": 21019
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.094695262427583e-05,
+      "loss": 0.078,
+      "step": 21020
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.090719171773305e-05,
+      "loss": 0.098,
+      "step": 21021
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0867449722078784e-05,
+      "loss": 0.1216,
+      "step": 21022
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.082772663809664e-05,
+      "loss": 0.0818,
+      "step": 21023
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0788022466569874e-05,
+      "loss": 0.0717,
+      "step": 21024
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.074833720828119e-05,
+      "loss": 0.0559,
+      "step": 21025
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.070867086401342e-05,
+      "loss": 0.113,
+      "step": 21026
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.066902343454837e-05,
+      "loss": 0.0815,
+      "step": 21027
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0629394920667864e-05,
+      "loss": 0.0664,
+      "step": 21028
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0589785323153384e-05,
+      "loss": 0.0482,
+      "step": 21029
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.055019464278575e-05,
+      "loss": 0.1251,
+      "step": 21030
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.051062288034568e-05,
+      "loss": 0.0958,
+      "step": 21031
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.047107003661343e-05,
+      "loss": 0.061,
+      "step": 21032
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.043153611236905e-05,
+      "loss": 0.0817,
+      "step": 21033
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.03920211083918e-05,
+      "loss": 0.1134,
+      "step": 21034
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.035252502546094e-05,
+      "loss": 0.1231,
+      "step": 21035
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.031304786435519e-05,
+      "loss": 0.1002,
+      "step": 21036
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.027358962585292e-05,
+      "loss": 0.0729,
+      "step": 21037
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.023415031073208e-05,
+      "loss": 0.0762,
+      "step": 21038
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0194729919770576e-05,
+      "loss": 0.158,
+      "step": 21039
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0155328453745475e-05,
+      "loss": 0.1052,
+      "step": 21040
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0115945913433706e-05,
+      "loss": 0.0923,
+      "step": 21041
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.007658229961164e-05,
+      "loss": 0.0817,
+      "step": 21042
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0037237613055774e-05,
+      "loss": 0.1014,
+      "step": 21043
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9997911854541714e-05,
+      "loss": 0.1731,
+      "step": 21044
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.995860502484472e-05,
+      "loss": 0.0574,
+      "step": 21045
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9919317124740064e-05,
+      "loss": 0.0602,
+      "step": 21046
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.988004815500235e-05,
+      "loss": 0.1259,
+      "step": 21047
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.984079811640573e-05,
+      "loss": 0.0904,
+      "step": 21048
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9801567009724147e-05,
+      "loss": 0.061,
+      "step": 21049
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.976235483573132e-05,
+      "loss": 0.0878,
+      "step": 21050
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.972316159520006e-05,
+      "loss": 0.1109,
+      "step": 21051
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9683987288903544e-05,
+      "loss": 0.0969,
+      "step": 21052
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.964483191761392e-05,
+      "loss": 0.0804,
+      "step": 21053
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9605695482103356e-05,
+      "loss": 0.0886,
+      "step": 21054
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.956657798314356e-05,
+      "loss": 0.1174,
+      "step": 21055
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.952747942150581e-05,
+      "loss": 0.0646,
+      "step": 21056
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.948839979796104e-05,
+      "loss": 0.1132,
+      "step": 21057
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9449339113279525e-05,
+      "loss": 0.0585,
+      "step": 21058
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9410297368231876e-05,
+      "loss": 0.0893,
+      "step": 21059
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.937127456358769e-05,
+      "loss": 0.1117,
+      "step": 21060
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.933227070011636e-05,
+      "loss": 0.0795,
+      "step": 21061
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.929328577858704e-05,
+      "loss": 0.0845,
+      "step": 21062
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.925431979976824e-05,
+      "loss": 0.0894,
+      "step": 21063
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.921537276442844e-05,
+      "loss": 0.0804,
+      "step": 21064
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.917644467333548e-05,
+      "loss": 0.0732,
+      "step": 21065
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.9137535527257076e-05,
+      "loss": 0.1266,
+      "step": 21066
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.909864532696039e-05,
+      "loss": 0.0917,
+      "step": 21067
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.905977407321204e-05,
+      "loss": 0.0809,
+      "step": 21068
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.902092176677874e-05,
+      "loss": 0.0672,
+      "step": 21069
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.898208840842632e-05,
+      "loss": 0.061,
+      "step": 21070
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.894327399892039e-05,
+      "loss": 0.1038,
+      "step": 21071
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8904478539026675e-05,
+      "loss": 0.0635,
+      "step": 21072
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.88657020295099e-05,
+      "loss": 0.0782,
+      "step": 21073
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.882694447113444e-05,
+      "loss": 0.1102,
+      "step": 21074
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.878820586466492e-05,
+      "loss": 0.0724,
+      "step": 21075
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.874948621086483e-05,
+      "loss": 0.0543,
+      "step": 21076
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8710785510497895e-05,
+      "loss": 0.0845,
+      "step": 21077
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.867210376432684e-05,
+      "loss": 0.0936,
+      "step": 21078
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8633440973114606e-05,
+      "loss": 0.0908,
+      "step": 21079
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.859479713762359e-05,
+      "loss": 0.0776,
+      "step": 21080
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.855617225861563e-05,
+      "loss": 0.0829,
+      "step": 21081
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.851756633685233e-05,
+      "loss": 0.0664,
+      "step": 21082
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.847897937309475e-05,
+      "loss": 0.0682,
+      "step": 21083
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8440411368104055e-05,
+      "loss": 0.0557,
+      "step": 21084
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.840186232264031e-05,
+      "loss": 0.0823,
+      "step": 21085
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8363332237464135e-05,
+      "loss": 0.0798,
+      "step": 21086
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.83248211133348e-05,
+      "loss": 0.0847,
+      "step": 21087
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.828632895101181e-05,
+      "loss": 0.0836,
+      "step": 21088
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8247855751254114e-05,
+      "loss": 0.0734,
+      "step": 21089
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.820940151482033e-05,
+      "loss": 0.0959,
+      "step": 21090
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.817096624246852e-05,
+      "loss": 0.0666,
+      "step": 21091
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.813254993495674e-05,
+      "loss": 0.0829,
+      "step": 21092
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.809415259304239e-05,
+      "loss": 0.078,
+      "step": 21093
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.805577421748263e-05,
+      "loss": 0.0828,
+      "step": 21094
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.801741480903387e-05,
+      "loss": 0.0985,
+      "step": 21095
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7979074368452934e-05,
+      "loss": 0.0722,
+      "step": 21096
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.794075289649557e-05,
+      "loss": 0.1111,
+      "step": 21097
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.790245039391727e-05,
+      "loss": 0.0586,
+      "step": 21098
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7864166861473425e-05,
+      "loss": 0.0989,
+      "step": 21099
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.782590229991889e-05,
+      "loss": 0.0654,
+      "step": 21100
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.778765671000806e-05,
+      "loss": 0.0709,
+      "step": 21101
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.774943009249521e-05,
+      "loss": 0.1032,
+      "step": 21102
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7711222448133744e-05,
+      "loss": 0.0875,
+      "step": 21103
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.767303377767728e-05,
+      "loss": 0.0549,
+      "step": 21104
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7634864081878775e-05,
+      "loss": 0.1172,
+      "step": 21105
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.759671336149062e-05,
+      "loss": 0.0581,
+      "step": 21106
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.755858161726544e-05,
+      "loss": 0.1121,
+      "step": 21107
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.752046884995486e-05,
+      "loss": 0.091,
+      "step": 21108
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.748237506031038e-05,
+      "loss": 0.0547,
+      "step": 21109
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.744430024908319e-05,
+      "loss": 0.1063,
+      "step": 21110
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7406244417023785e-05,
+      "loss": 0.077,
+      "step": 21111
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.73682075648829e-05,
+      "loss": 0.1021,
+      "step": 21112
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.733018969341029e-05,
+      "loss": 0.067,
+      "step": 21113
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.729219080335555e-05,
+      "loss": 0.1045,
+      "step": 21114
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.72542108954681e-05,
+      "loss": 0.0569,
+      "step": 21115
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.721624997049655e-05,
+      "loss": 0.1016,
+      "step": 21116
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7178308029189756e-05,
+      "loss": 0.0962,
+      "step": 21117
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.714038507229533e-05,
+      "loss": 0.0609,
+      "step": 21118
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.710248110056158e-05,
+      "loss": 0.0414,
+      "step": 21119
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.706459611473556e-05,
+      "loss": 0.0822,
+      "step": 21120
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.702673011556434e-05,
+      "loss": 0.0596,
+      "step": 21121
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.698888310379445e-05,
+      "loss": 0.0631,
+      "step": 21122
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.695105508017227e-05,
+      "loss": 0.101,
+      "step": 21123
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.691324604544366e-05,
+      "loss": 0.0541,
+      "step": 21124
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6875456000353805e-05,
+      "loss": 0.0811,
+      "step": 21125
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.683768494564832e-05,
+      "loss": 0.1078,
+      "step": 21126
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.679993288207162e-05,
+      "loss": 0.0894,
+      "step": 21127
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6762199810368214e-05,
+      "loss": 0.0576,
+      "step": 21128
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.672448573128218e-05,
+      "loss": 0.0919,
+      "step": 21129
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.668679064555691e-05,
+      "loss": 0.1314,
+      "step": 21130
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.664911455393582e-05,
+      "loss": 0.0792,
+      "step": 21131
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.661145745716177e-05,
+      "loss": 0.0617,
+      "step": 21132
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6573819355977255e-05,
+      "loss": 0.0368,
+      "step": 21133
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.653620025112436e-05,
+      "loss": 0.106,
+      "step": 21134
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.649860014334494e-05,
+      "loss": 0.0554,
+      "step": 21135
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.64610190333804e-05,
+      "loss": 0.0446,
+      "step": 21136
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6423456921971356e-05,
+      "loss": 0.0789,
+      "step": 21137
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.638591380985878e-05,
+      "loss": 0.1095,
+      "step": 21138
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.634838969778309e-05,
+      "loss": 0.0718,
+      "step": 21139
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6310884586483906e-05,
+      "loss": 0.0432,
+      "step": 21140
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.627339847670075e-05,
+      "loss": 0.0883,
+      "step": 21141
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6235931369172806e-05,
+      "loss": 0.0958,
+      "step": 21142
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6198483264638926e-05,
+      "loss": 0.1588,
+      "step": 21143
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.616105416383708e-05,
+      "loss": 0.0601,
+      "step": 21144
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.612364406750579e-05,
+      "loss": 0.0767,
+      "step": 21145
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6086252976382463e-05,
+      "loss": 0.0659,
+      "step": 21146
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.604888089120439e-05,
+      "loss": 0.0933,
+      "step": 21147
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.601152781270833e-05,
+      "loss": 0.0629,
+      "step": 21148
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.597419374163102e-05,
+      "loss": 0.0667,
+      "step": 21149
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.593687867870843e-05,
+      "loss": 0.0631,
+      "step": 21150
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.589958262467619e-05,
+      "loss": 0.1454,
+      "step": 21151
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5862305580270035e-05,
+      "loss": 0.1407,
+      "step": 21152
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.582504754622484e-05,
+      "loss": 0.0958,
+      "step": 21153
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.578780852327512e-05,
+      "loss": 0.0536,
+      "step": 21154
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.575058851215518e-05,
+      "loss": 0.0952,
+      "step": 21155
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5713387513598984e-05,
+      "loss": 0.1028,
+      "step": 21156
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.567620552833995e-05,
+      "loss": 0.1308,
+      "step": 21157
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5639042557111144e-05,
+      "loss": 0.0996,
+      "step": 21158
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5601898600645445e-05,
+      "loss": 0.0947,
+      "step": 21159
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.556477365967525e-05,
+      "loss": 0.0605,
+      "step": 21160
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5527667734932655e-05,
+      "loss": 0.0804,
+      "step": 21161
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5490580827149175e-05,
+      "loss": 0.1045,
+      "step": 21162
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.545351293705601e-05,
+      "loss": 0.1113,
+      "step": 21163
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5416464065384124e-05,
+      "loss": 0.0647,
+      "step": 21164
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.537943421286405e-05,
+      "loss": 0.0728,
+      "step": 21165
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5342423380225865e-05,
+      "loss": 0.0819,
+      "step": 21166
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.530543156819943e-05,
+      "loss": 0.0854,
+      "step": 21167
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5268458777513946e-05,
+      "loss": 0.0698,
+      "step": 21168
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.523150500889849e-05,
+      "loss": 0.0906,
+      "step": 21169
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5194570263081814e-05,
+      "loss": 0.1115,
+      "step": 21170
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5157654540792113e-05,
+      "loss": 0.0887,
+      "step": 21171
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5120757842757254e-05,
+      "loss": 0.0721,
+      "step": 21172
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.508388016970476e-05,
+      "loss": 0.065,
+      "step": 21173
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.504702152236183e-05,
+      "loss": 0.0773,
+      "step": 21174
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5010181901455106e-05,
+      "loss": 0.1199,
+      "step": 21175
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4973361307711e-05,
+      "loss": 0.0902,
+      "step": 21176
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.49365597418555e-05,
+      "loss": 0.166,
+      "step": 21177
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.489977720461424e-05,
+      "loss": 0.0804,
+      "step": 21178
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.486301369671263e-05,
+      "loss": 0.1414,
+      "step": 21179
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.482626921887533e-05,
+      "loss": 0.0855,
+      "step": 21180
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.478954377182708e-05,
+      "loss": 0.1526,
+      "step": 21181
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4752837356291865e-05,
+      "loss": 0.056,
+      "step": 21182
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.471614997299355e-05,
+      "loss": 0.0818,
+      "step": 21183
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.467948162265522e-05,
+      "loss": 0.1094,
+      "step": 21184
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4642832306000295e-05,
+      "loss": 0.0853,
+      "step": 21185
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4606202023751196e-05,
+      "loss": 0.0847,
+      "step": 21186
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.456959077663013e-05,
+      "loss": 0.1439,
+      "step": 21187
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.453299856535908e-05,
+      "loss": 0.0771,
+      "step": 21188
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4496425390659556e-05,
+      "loss": 0.1343,
+      "step": 21189
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4459871253252564e-05,
+      "loss": 0.0483,
+      "step": 21190
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4423336153858844e-05,
+      "loss": 0.0945,
+      "step": 21191
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4386820093199044e-05,
+      "loss": 0.0621,
+      "step": 21192
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.435032307199304e-05,
+      "loss": 0.0912,
+      "step": 21193
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.431384509096036e-05,
+      "loss": 0.0803,
+      "step": 21194
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.427738615082043e-05,
+      "loss": 0.0826,
+      "step": 21195
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4240946252291906e-05,
+      "loss": 0.0561,
+      "step": 21196
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4204525396093425e-05,
+      "loss": 0.0775,
+      "step": 21197
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.416812358294297e-05,
+      "loss": 0.1029,
+      "step": 21198
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.413174081355863e-05,
+      "loss": 0.073,
+      "step": 21199
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.409537708865751e-05,
+      "loss": 0.0968,
+      "step": 21200
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4059032408956466e-05,
+      "loss": 0.0726,
+      "step": 21201
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.402270677517261e-05,
+      "loss": 0.1097,
+      "step": 21202
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3986400188021796e-05,
+      "loss": 0.1075,
+      "step": 21203
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.39501126482199e-05,
+      "loss": 0.0544,
+      "step": 21204
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.391384415648269e-05,
+      "loss": 0.0898,
+      "step": 21205
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.387759471352503e-05,
+      "loss": 0.1176,
+      "step": 21206
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.384136432006179e-05,
+      "loss": 0.0944,
+      "step": 21207
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3805152976807416e-05,
+      "loss": 0.084,
+      "step": 21208
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.376896068447577e-05,
+      "loss": 0.1031,
+      "step": 21209
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.37327874437805e-05,
+      "loss": 0.0748,
+      "step": 21210
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.369663325543471e-05,
+      "loss": 0.0598,
+      "step": 21211
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.366049812015148e-05,
+      "loss": 0.1071,
+      "step": 21212
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.362438203864338e-05,
+      "loss": 0.0573,
+      "step": 21213
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.358828501162226e-05,
+      "loss": 0.1226,
+      "step": 21214
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3552207039800106e-05,
+      "loss": 0.0751,
+      "step": 21215
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3516148123888144e-05,
+      "loss": 0.1197,
+      "step": 21216
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.348010826459747e-05,
+      "loss": 0.0786,
+      "step": 21217
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.344408746263838e-05,
+      "loss": 0.1185,
+      "step": 21218
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.340808571872156e-05,
+      "loss": 0.0683,
+      "step": 21219
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.337210303355665e-05,
+      "loss": 0.0413,
+      "step": 21220
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3336139407853074e-05,
+      "loss": 0.0552,
+      "step": 21221
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3300194842320056e-05,
+      "loss": 0.0908,
+      "step": 21222
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.326426933766624e-05,
+      "loss": 0.0728,
+      "step": 21223
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.322836289460007e-05,
+      "loss": 0.053,
+      "step": 21224
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.319247551382964e-05,
+      "loss": 0.0837,
+      "step": 21225
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.315660719606228e-05,
+      "loss": 0.1361,
+      "step": 21226
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.312075794200553e-05,
+      "loss": 0.0724,
+      "step": 21227
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3084927752366044e-05,
+      "loss": 0.0867,
+      "step": 21228
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.304911662785026e-05,
+      "loss": 0.0756,
+      "step": 21229
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3013324569164504e-05,
+      "loss": 0.1063,
+      "step": 21230
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2977551577014206e-05,
+      "loss": 0.09,
+      "step": 21231
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.294179765210492e-05,
+      "loss": 0.0929,
+      "step": 21232
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.290606279514152e-05,
+      "loss": 0.0854,
+      "step": 21233
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.287034700682889e-05,
+      "loss": 0.1251,
+      "step": 21234
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.283465028787091e-05,
+      "loss": 0.0692,
+      "step": 21235
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.279897263897158e-05,
+      "loss": 0.1024,
+      "step": 21236
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.276331406083444e-05,
+      "loss": 0.1371,
+      "step": 21237
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.272767455416226e-05,
+      "loss": 0.0946,
+      "step": 21238
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.269205411965814e-05,
+      "loss": 0.1582,
+      "step": 21239
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2656452758024316e-05,
+      "loss": 0.1097,
+      "step": 21240
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.262087046996276e-05,
+      "loss": 0.1205,
+      "step": 21241
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.258530725617503e-05,
+      "loss": 0.0814,
+      "step": 21242
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.254976311736213e-05,
+      "loss": 0.1119,
+      "step": 21243
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2514238054225266e-05,
+      "loss": 0.0493,
+      "step": 21244
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2478732067464764e-05,
+      "loss": 0.0919,
+      "step": 21245
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.244324515778063e-05,
+      "loss": 0.1045,
+      "step": 21246
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.240777732587275e-05,
+      "loss": 0.0595,
+      "step": 21247
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.237232857244032e-05,
+      "loss": 0.1123,
+      "step": 21248
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2336898898182367e-05,
+      "loss": 0.075,
+      "step": 21249
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.230148830379742e-05,
+      "loss": 0.0783,
+      "step": 21250
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.22660967899836e-05,
+      "loss": 0.0851,
+      "step": 21251
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.22307243574389e-05,
+      "loss": 0.1346,
+      "step": 21252
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.219537100686087e-05,
+      "loss": 0.1001,
+      "step": 21253
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2160036738946185e-05,
+      "loss": 0.0765,
+      "step": 21254
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2124721554391946e-05,
+      "loss": 0.074,
+      "step": 21255
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2089425453894375e-05,
+      "loss": 0.0852,
+      "step": 21256
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2054148438149354e-05,
+      "loss": 0.1012,
+      "step": 21257
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.201889050785245e-05,
+      "loss": 0.0931,
+      "step": 21258
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1983651663698984e-05,
+      "loss": 0.056,
+      "step": 21259
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1948431906383637e-05,
+      "loss": 0.0849,
+      "step": 21260
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.191323123660095e-05,
+      "loss": 0.1171,
+      "step": 21261
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1878049655044815e-05,
+      "loss": 0.0915,
+      "step": 21262
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.184288716240924e-05,
+      "loss": 0.0991,
+      "step": 21263
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.180774375938711e-05,
+      "loss": 0.0984,
+      "step": 21264
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.177261944667187e-05,
+      "loss": 0.1262,
+      "step": 21265
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.173751422495563e-05,
+      "loss": 0.069,
+      "step": 21266
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1702428094930845e-05,
+      "loss": 0.0675,
+      "step": 21267
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.166736105728929e-05,
+      "loss": 0.1547,
+      "step": 21268
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1632313112722414e-05,
+      "loss": 0.1121,
+      "step": 21269
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.15972842619211e-05,
+      "loss": 0.0682,
+      "step": 21270
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.156227450557614e-05,
+      "loss": 0.0775,
+      "step": 21271
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1527283844377864e-05,
+      "loss": 0.1304,
+      "step": 21272
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1492312279016274e-05,
+      "loss": 0.0921,
+      "step": 21273
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.145735981018072e-05,
+      "loss": 0.1022,
+      "step": 21274
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.142242643856053e-05,
+      "loss": 0.0617,
+      "step": 21275
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.138751216484448e-05,
+      "loss": 0.0722,
+      "step": 21276
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.135261698972092e-05,
+      "loss": 0.0585,
+      "step": 21277
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.131774091387784e-05,
+      "loss": 0.1164,
+      "step": 21278
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.128288393800327e-05,
+      "loss": 0.0898,
+      "step": 21279
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.124804606278408e-05,
+      "loss": 0.0889,
+      "step": 21280
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.121322728890741e-05,
+      "loss": 0.1061,
+      "step": 21281
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.117842761705969e-05,
+      "loss": 0.0818,
+      "step": 21282
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.114364704792716e-05,
+      "loss": 0.0922,
+      "step": 21283
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.110888558219549e-05,
+      "loss": 0.0952,
+      "step": 21284
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.107414322055024e-05,
+      "loss": 0.0753,
+      "step": 21285
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.103941996367632e-05,
+      "loss": 0.1296,
+      "step": 21286
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.10047158122585e-05,
+      "loss": 0.087,
+      "step": 21287
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0970030766981014e-05,
+      "loss": 0.0959,
+      "step": 21288
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.093536482852777e-05,
+      "loss": 0.0784,
+      "step": 21289
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0900717997582205e-05,
+      "loss": 0.0778,
+      "step": 21290
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0866090274827455e-05,
+      "loss": 0.1053,
+      "step": 21291
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.08314816609464e-05,
+      "loss": 0.097,
+      "step": 21292
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0796892156621516e-05,
+      "loss": 0.0813,
+      "step": 21293
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.076232176253468e-05,
+      "loss": 0.1145,
+      "step": 21294
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.072777047936748e-05,
+      "loss": 0.1462,
+      "step": 21295
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.069323830780124e-05,
+      "loss": 0.1141,
+      "step": 21296
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.065872524851687e-05,
+      "loss": 0.0622,
+      "step": 21297
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.062423130219483e-05,
+      "loss": 0.0701,
+      "step": 21298
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.058975646951545e-05,
+      "loss": 0.1342,
+      "step": 21299
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.055530075115831e-05,
+      "loss": 0.0885,
+      "step": 21300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.052086414780275e-05,
+      "loss": 0.0823,
+      "step": 21301
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0486446660127896e-05,
+      "loss": 0.1035,
+      "step": 21302
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0452048288812316e-05,
+      "loss": 0.0902,
+      "step": 21303
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0417669034534245e-05,
+      "loss": 0.1115,
+      "step": 21304
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0383308897971584e-05,
+      "loss": 0.0597,
+      "step": 21305
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0348967879801792e-05,
+      "loss": 0.0767,
+      "step": 21306
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0314645980701884e-05,
+      "loss": 0.058,
+      "step": 21307
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.028034320134887e-05,
+      "loss": 0.0613,
+      "step": 21308
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.024605954241899e-05,
+      "loss": 0.0851,
+      "step": 21309
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.021179500458815e-05,
+      "loss": 0.0795,
+      "step": 21310
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.017754958853203e-05,
+      "loss": 0.1006,
+      "step": 21311
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.014332329492586e-05,
+      "loss": 0.0555,
+      "step": 21312
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0109116124444447e-05,
+      "loss": 0.0812,
+      "step": 21313
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0074928077762465e-05,
+      "loss": 0.0947,
+      "step": 21314
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.004075915555371e-05,
+      "loss": 0.0697,
+      "step": 21315
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.00066093584922e-05,
+      "loss": 0.0601,
+      "step": 21316
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9972478687250947e-05,
+      "loss": 0.097,
+      "step": 21317
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.993836714250331e-05,
+      "loss": 0.0743,
+      "step": 21318
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.990427472492152e-05,
+      "loss": 0.1471,
+      "step": 21319
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9870201435178047e-05,
+      "loss": 0.0536,
+      "step": 21320
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.983614727394468e-05,
+      "loss": 0.084,
+      "step": 21321
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9802112241892887e-05,
+      "loss": 0.0951,
+      "step": 21322
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9768096339693683e-05,
+      "loss": 0.0801,
+      "step": 21323
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9734099568017646e-05,
+      "loss": 0.0571,
+      "step": 21324
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.970012192753546e-05,
+      "loss": 0.066,
+      "step": 21325
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9666163418916813e-05,
+      "loss": 0.0999,
+      "step": 21326
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.963222404283139e-05,
+      "loss": 0.0613,
+      "step": 21327
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9598303799948323e-05,
+      "loss": 0.1107,
+      "step": 21328
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.956440269093652e-05,
+      "loss": 0.1013,
+      "step": 21329
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.953052071646434e-05,
+      "loss": 0.0768,
+      "step": 21330
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9496657877199796e-05,
+      "loss": 0.0829,
+      "step": 21331
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9462814173810693e-05,
+      "loss": 0.0776,
+      "step": 21332
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9428989606964385e-05,
+      "loss": 0.0866,
+      "step": 21333
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9395184177327783e-05,
+      "loss": 0.0822,
+      "step": 21334
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9361397885567354e-05,
+      "loss": 0.0591,
+      "step": 21335
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9327630732349232e-05,
+      "loss": 0.1091,
+      "step": 21336
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.929388271833944e-05,
+      "loss": 0.0845,
+      "step": 21337
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9260153844203108e-05,
+      "loss": 0.0745,
+      "step": 21338
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.92264441106056e-05,
+      "loss": 0.0656,
+      "step": 21339
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9192753518211268e-05,
+      "loss": 0.1329,
+      "step": 21340
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9159082067684696e-05,
+      "loss": 0.0643,
+      "step": 21341
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.912542975968968e-05,
+      "loss": 0.0547,
+      "step": 21342
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9091796594889698e-05,
+      "loss": 0.081,
+      "step": 21343
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.905818257394799e-05,
+      "loss": 0.0928,
+      "step": 21344
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.902458769752736e-05,
+      "loss": 0.092,
+      "step": 21345
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.899101196629017e-05,
+      "loss": 0.0902,
+      "step": 21346
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8957455380898447e-05,
+      "loss": 0.1326,
+      "step": 21347
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8923917942013766e-05,
+      "loss": 0.0877,
+      "step": 21348
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8890399650297495e-05,
+      "loss": 0.0682,
+      "step": 21349
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8856900506410545e-05,
+      "loss": 0.0667,
+      "step": 21350
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.882342051101339e-05,
+      "loss": 0.0814,
+      "step": 21351
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8789959664766163e-05,
+      "loss": 0.0518,
+      "step": 21352
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.875651796832879e-05,
+      "loss": 0.0643,
+      "step": 21353
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.87230954223604e-05,
+      "loss": 0.1354,
+      "step": 21354
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8689692027520143e-05,
+      "loss": 0.0922,
+      "step": 21355
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8656307784466707e-05,
+      "loss": 0.0751,
+      "step": 21356
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8622942693858235e-05,
+      "loss": 0.1162,
+      "step": 21357
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.858959675635242e-05,
+      "loss": 0.0989,
+      "step": 21358
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8556269972607186e-05,
+      "loss": 0.0699,
+      "step": 21359
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8522962343279445e-05,
+      "loss": 0.0607,
+      "step": 21360
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.848967386902579e-05,
+      "loss": 0.0697,
+      "step": 21361
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.84564045505028e-05,
+      "loss": 0.0649,
+      "step": 21362
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8423154388366402e-05,
+      "loss": 0.0565,
+      "step": 21363
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8389923383272177e-05,
+      "loss": 0.0729,
+      "step": 21364
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.835671153587538e-05,
+      "loss": 0.091,
+      "step": 21365
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.832351884683082e-05,
+      "loss": 0.0886,
+      "step": 21366
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8290345316793086e-05,
+      "loss": 0.0858,
+      "step": 21367
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.825719094641621e-05,
+      "loss": 0.085,
+      "step": 21368
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8224055736353893e-05,
+      "loss": 0.0881,
+      "step": 21369
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8190939687259275e-05,
+      "loss": 0.1465,
+      "step": 21370
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8157842799785725e-05,
+      "loss": 0.058,
+      "step": 21371
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.812476507458561e-05,
+      "loss": 0.0797,
+      "step": 21372
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.809170651231119e-05,
+      "loss": 0.0805,
+      "step": 21373
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8058667113614267e-05,
+      "loss": 0.097,
+      "step": 21374
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.802564687914633e-05,
+      "loss": 0.0594,
+      "step": 21375
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7992645809558404e-05,
+      "loss": 0.092,
+      "step": 21376
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7959663905501197e-05,
+      "loss": 0.068,
+      "step": 21377
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.792670116762497e-05,
+      "loss": 0.0815,
+      "step": 21378
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.789375759657986e-05,
+      "loss": 0.0722,
+      "step": 21379
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7860833193015244e-05,
+      "loss": 0.1371,
+      "step": 21380
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7827927957580266e-05,
+      "loss": 0.0537,
+      "step": 21381
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7795041890923968e-05,
+      "loss": 0.1058,
+      "step": 21382
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7762174993694712e-05,
+      "loss": 0.071,
+      "step": 21383
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7729327266540316e-05,
+      "loss": 0.034,
+      "step": 21384
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.769649871010871e-05,
+      "loss": 0.0984,
+      "step": 21385
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.766368932504715e-05,
+      "loss": 0.1188,
+      "step": 21386
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.763089911200245e-05,
+      "loss": 0.0442,
+      "step": 21387
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.759812807162132e-05,
+      "loss": 0.1296,
+      "step": 21388
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7565376204549685e-05,
+      "loss": 0.0699,
+      "step": 21389
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7532643511433586e-05,
+      "loss": 0.1228,
+      "step": 21390
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.749992999291817e-05,
+      "loss": 0.0651,
+      "step": 21391
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.746723564964859e-05,
+      "loss": 0.084,
+      "step": 21392
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7434560482269556e-05,
+      "loss": 0.093,
+      "step": 21393
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.740190449142532e-05,
+      "loss": 0.1366,
+      "step": 21394
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7369267677759713e-05,
+      "loss": 0.0494,
+      "step": 21395
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7336650041916322e-05,
+      "loss": 0.084,
+      "step": 21396
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7304051584538193e-05,
+      "loss": 0.088,
+      "step": 21397
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.727147230626803e-05,
+      "loss": 0.0956,
+      "step": 21398
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7238912207748434e-05,
+      "loss": 0.13,
+      "step": 21399
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7206371289621336e-05,
+      "loss": 0.0946,
+      "step": 21400
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.717384955252833e-05,
+      "loss": 0.0632,
+      "step": 21401
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.714134699711046e-05,
+      "loss": 0.0686,
+      "step": 21402
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7108863624008995e-05,
+      "loss": 0.0894,
+      "step": 21403
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7076399433863975e-05,
+      "loss": 0.0825,
+      "step": 21404
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7043954427315997e-05,
+      "loss": 0.0473,
+      "step": 21405
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7011528605004443e-05,
+      "loss": 0.09,
+      "step": 21406
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.69791219675688e-05,
+      "loss": 0.0793,
+      "step": 21407
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6946734515647997e-05,
+      "loss": 0.0956,
+      "step": 21408
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.691436624988075e-05,
+      "loss": 0.0602,
+      "step": 21409
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6882017170905104e-05,
+      "loss": 0.0999,
+      "step": 21410
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6849687279358882e-05,
+      "loss": 0.1008,
+      "step": 21411
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6817376575879682e-05,
+      "loss": 0.076,
+      "step": 21412
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6785085061104663e-05,
+      "loss": 0.0573,
+      "step": 21413
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.675281273567032e-05,
+      "loss": 0.1356,
+      "step": 21414
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6720559600213134e-05,
+      "loss": 0.0655,
+      "step": 21415
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.668832565536905e-05,
+      "loss": 0.1123,
+      "step": 21416
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6656110901773445e-05,
+      "loss": 0.055,
+      "step": 21417
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.662391534006181e-05,
+      "loss": 0.0545,
+      "step": 21418
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.659173897086886e-05,
+      "loss": 0.1233,
+      "step": 21419
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6559581794828867e-05,
+      "loss": 0.1595,
+      "step": 21420
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.65274438125761e-05,
+      "loss": 0.0931,
+      "step": 21421
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6495325024744054e-05,
+      "loss": 0.0695,
+      "step": 21422
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6463225431966108e-05,
+      "loss": 0.0673,
+      "step": 21423
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6431145034875316e-05,
+      "loss": 0.0757,
+      "step": 21424
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6399083834103944e-05,
+      "loss": 0.0709,
+      "step": 21425
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.636704183028449e-05,
+      "loss": 0.0868,
+      "step": 21426
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6335019024048444e-05,
+      "loss": 0.0482,
+      "step": 21427
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.630301541602742e-05,
+      "loss": 0.1091,
+      "step": 21428
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6271031006852464e-05,
+      "loss": 0.0593,
+      "step": 21429
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.623906579715396e-05,
+      "loss": 0.1375,
+      "step": 21430
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6207119787562405e-05,
+      "loss": 0.0471,
+      "step": 21431
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.617519297870763e-05,
+      "loss": 0.0501,
+      "step": 21432
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.614328537121924e-05,
+      "loss": 0.0875,
+      "step": 21433
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6111396965726286e-05,
+      "loss": 0.0582,
+      "step": 21434
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.607952776285749e-05,
+      "loss": 0.087,
+      "step": 21435
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.604767776324124e-05,
+      "loss": 0.0515,
+      "step": 21436
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6015846967505584e-05,
+      "loss": 0.0761,
+      "step": 21437
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.598403537627825e-05,
+      "loss": 0.0634,
+      "step": 21438
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5952242990186283e-05,
+      "loss": 0.1176,
+      "step": 21439
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5920469809856738e-05,
+      "loss": 0.1273,
+      "step": 21440
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5888715835916012e-05,
+      "loss": 0.1064,
+      "step": 21441
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5856981068990147e-05,
+      "loss": 0.0842,
+      "step": 21442
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5825265509704877e-05,
+      "loss": 0.0641,
+      "step": 21443
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.579356915868558e-05,
+      "loss": 0.0611,
+      "step": 21444
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.576189201655732e-05,
+      "loss": 0.1199,
+      "step": 21445
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5730234083944593e-05,
+      "loss": 0.1027,
+      "step": 21446
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5698595361471677e-05,
+      "loss": 0.1066,
+      "step": 21447
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5666975849762408e-05,
+      "loss": 0.0594,
+      "step": 21448
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5635375549440176e-05,
+      "loss": 0.0892,
+      "step": 21449
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5603794461128037e-05,
+      "loss": 0.1038,
+      "step": 21450
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.557223258544872e-05,
+      "loss": 0.0515,
+      "step": 21451
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.554068992302472e-05,
+      "loss": 0.0775,
+      "step": 21452
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5509166474477764e-05,
+      "loss": 0.0807,
+      "step": 21453
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5477662240429466e-05,
+      "loss": 0.1041,
+      "step": 21454
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.544617722150111e-05,
+      "loss": 0.0723,
+      "step": 21455
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5414711418313197e-05,
+      "loss": 0.1541,
+      "step": 21456
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5383264831486453e-05,
+      "loss": 0.0473,
+      "step": 21457
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5351837461640937e-05,
+      "loss": 0.0997,
+      "step": 21458
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5320429309396153e-05,
+      "loss": 0.0427,
+      "step": 21459
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5289040375371498e-05,
+      "loss": 0.0995,
+      "step": 21460
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5257670660185917e-05,
+      "loss": 0.0694,
+      "step": 21461
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5226320164457807e-05,
+      "loss": 0.061,
+      "step": 21462
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5194988888805338e-05,
+      "loss": 0.0402,
+      "step": 21463
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.516367683384635e-05,
+      "loss": 0.1146,
+      "step": 21464
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.513238400019824e-05,
+      "loss": 0.0698,
+      "step": 21465
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5101110388477956e-05,
+      "loss": 0.1056,
+      "step": 21466
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5069855999302114e-05,
+      "loss": 0.061,
+      "step": 21467
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5038620833287228e-05,
+      "loss": 0.0994,
+      "step": 21468
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5007404891048914e-05,
+      "loss": 0.0898,
+      "step": 21469
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4976208173202786e-05,
+      "loss": 0.0554,
+      "step": 21470
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.494503068036369e-05,
+      "loss": 0.0615,
+      "step": 21471
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4913872413146797e-05,
+      "loss": 0.0405,
+      "step": 21472
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.488273337216629e-05,
+      "loss": 0.1274,
+      "step": 21473
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4851613558036114e-05,
+      "loss": 0.0854,
+      "step": 21474
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.482051297136989e-05,
+      "loss": 0.1037,
+      "step": 21475
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.47894316127808e-05,
+      "loss": 0.0404,
+      "step": 21476
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4758369482881793e-05,
+      "loss": 0.1076,
+      "step": 21477
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.472732658228516e-05,
+      "loss": 0.0829,
+      "step": 21478
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4696302911603185e-05,
+      "loss": 0.0731,
+      "step": 21479
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4665298471447605e-05,
+      "loss": 0.077,
+      "step": 21480
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4634313262429487e-05,
+      "loss": 0.0769,
+      "step": 21481
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4603347285160005e-05,
+      "loss": 0.0812,
+      "step": 21482
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4572400540249783e-05,
+      "loss": 0.0797,
+      "step": 21483
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.454147302830867e-05,
+      "loss": 0.1304,
+      "step": 21484
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.451056474994684e-05,
+      "loss": 0.0768,
+      "step": 21485
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4479675705773473e-05,
+      "loss": 0.0825,
+      "step": 21486
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4448805896397864e-05,
+      "loss": 0.0875,
+      "step": 21487
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.441795532242841e-05,
+      "loss": 0.0469,
+      "step": 21488
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4387123984473735e-05,
+      "loss": 0.1245,
+      "step": 21489
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.435631188314147e-05,
+      "loss": 0.0876,
+      "step": 21490
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4325519019039232e-05,
+      "loss": 0.0858,
+      "step": 21491
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.429474539277432e-05,
+      "loss": 0.0757,
+      "step": 21492
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4263991004953356e-05,
+      "loss": 0.0758,
+      "step": 21493
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4233255856182747e-05,
+      "loss": 0.0998,
+      "step": 21494
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.420253994706856e-05,
+      "loss": 0.0875,
+      "step": 21495
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.417184327821642e-05,
+      "loss": 0.0383,
+      "step": 21496
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4141165850231516e-05,
+      "loss": 0.0923,
+      "step": 21497
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.411050766371892e-05,
+      "loss": 0.0591,
+      "step": 21498
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4079868719282915e-05,
+      "loss": 0.0722,
+      "step": 21499
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.404924901752781e-05,
+      "loss": 0.074,
+      "step": 21500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.401864855905722e-05,
+      "loss": 0.0983,
+      "step": 21501
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3988067344474562e-05,
+      "loss": 0.0523,
+      "step": 21502
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3957505374382904e-05,
+      "loss": 0.1093,
+      "step": 21503
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3926962649384544e-05,
+      "loss": 0.1183,
+      "step": 21504
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3896439170082105e-05,
+      "loss": 0.0926,
+      "step": 21505
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3865934937077226e-05,
+      "loss": 0.0707,
+      "step": 21506
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.383544995097131e-05,
+      "loss": 0.0912,
+      "step": 21507
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3804984212365654e-05,
+      "loss": 0.0732,
+      "step": 21508
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3774537721860667e-05,
+      "loss": 0.0764,
+      "step": 21509
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.374411048005698e-05,
+      "loss": 0.0653,
+      "step": 21510
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3713702487554335e-05,
+      "loss": 0.1228,
+      "step": 21511
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3683313744952473e-05,
+      "loss": 0.1083,
+      "step": 21512
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3652944252850474e-05,
+      "loss": 0.0754,
+      "step": 21513
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3622594011847077e-05,
+      "loss": 0.0903,
+      "step": 21514
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3592263022540915e-05,
+      "loss": 0.1163,
+      "step": 21515
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3561951285529848e-05,
+      "loss": 0.1044,
+      "step": 21516
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3531658801411616e-05,
+      "loss": 0.0576,
+      "step": 21517
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3501385570783516e-05,
+      "loss": 0.0721,
+      "step": 21518
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.347113159424252e-05,
+      "loss": 0.0702,
+      "step": 21519
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.344089687238493e-05,
+      "loss": 0.0651,
+      "step": 21520
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3410681405807265e-05,
+      "loss": 0.0711,
+      "step": 21521
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.338048519510494e-05,
+      "loss": 0.0959,
+      "step": 21522
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3350308240873698e-05,
+      "loss": 0.1038,
+      "step": 21523
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3320150543708173e-05,
+      "loss": 0.0902,
+      "step": 21524
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3290012104203228e-05,
+      "loss": 0.0909,
+      "step": 21525
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3259892922953163e-05,
+      "loss": 0.0897,
+      "step": 21526
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3229793000551723e-05,
+      "loss": 0.0866,
+      "step": 21527
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.319971233759244e-05,
+      "loss": 0.0686,
+      "step": 21528
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3169650934668497e-05,
+      "loss": 0.1077,
+      "step": 21529
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3139608792372423e-05,
+      "loss": 0.0956,
+      "step": 21530
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.310958591129675e-05,
+      "loss": 0.0583,
+      "step": 21531
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3079582292033442e-05,
+      "loss": 0.0992,
+      "step": 21532
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3049597935174137e-05,
+      "loss": 0.0775,
+      "step": 21533
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.301963284130992e-05,
+      "loss": 0.1282,
+      "step": 21534
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.298968701103177e-05,
+      "loss": 0.0939,
+      "step": 21535
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.295976044492998e-05,
+      "loss": 0.0923,
+      "step": 21536
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2929853143594748e-05,
+      "loss": 0.1097,
+      "step": 21537
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2899965107615717e-05,
+      "loss": 0.1121,
+      "step": 21538
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.287009633758219e-05,
+      "loss": 0.0834,
+      "step": 21539
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2840246834083255e-05,
+      "loss": 0.1115,
+      "step": 21540
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.281041659770722e-05,
+      "loss": 0.1031,
+      "step": 21541
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2780605629042385e-05,
+      "loss": 0.0803,
+      "step": 21542
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2750813928676618e-05,
+      "loss": 0.0493,
+      "step": 21543
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2721041497197227e-05,
+      "loss": 0.1119,
+      "step": 21544
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2691288335191294e-05,
+      "loss": 0.1102,
+      "step": 21545
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.266155444324558e-05,
+      "loss": 0.1103,
+      "step": 21546
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2631839821946165e-05,
+      "loss": 0.0836,
+      "step": 21547
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2602144471879136e-05,
+      "loss": 0.0699,
+      "step": 21548
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2572468393629807e-05,
+      "loss": 0.0667,
+      "step": 21549
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2542811587783484e-05,
+      "loss": 0.0978,
+      "step": 21550
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2513174054924813e-05,
+      "loss": 0.0789,
+      "step": 21551
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.24835557956381e-05,
+      "loss": 0.1331,
+      "step": 21552
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.245395681050766e-05,
+      "loss": 0.0799,
+      "step": 21553
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.24243771001168e-05,
+      "loss": 0.0784,
+      "step": 21554
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2394816665048943e-05,
+      "loss": 0.0981,
+      "step": 21555
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.236527550588696e-05,
+      "loss": 0.0697,
+      "step": 21556
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.233575362321305e-05,
+      "loss": 0.0902,
+      "step": 21557
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2306251017609523e-05,
+      "loss": 0.0718,
+      "step": 21558
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.227676768965814e-05,
+      "loss": 0.0775,
+      "step": 21559
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.224730363994021e-05,
+      "loss": 0.0777,
+      "step": 21560
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.22178588690366e-05,
+      "loss": 0.1252,
+      "step": 21561
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.218843337752796e-05,
+      "loss": 0.0682,
+      "step": 21562
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2159027165994385e-05,
+      "loss": 0.0635,
+      "step": 21563
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2129640235015846e-05,
+      "loss": 0.0643,
+      "step": 21564
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2100272585171667e-05,
+      "loss": 0.1064,
+      "step": 21565
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2070924217040932e-05,
+      "loss": 0.0463,
+      "step": 21566
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2041595131202407e-05,
+      "loss": 0.0802,
+      "step": 21567
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.2012285328234295e-05,
+      "loss": 0.1045,
+      "step": 21568
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1982994808714464e-05,
+      "loss": 0.1293,
+      "step": 21569
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1953723573220564e-05,
+      "loss": 0.068,
+      "step": 21570
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1924471622329578e-05,
+      "loss": 0.0773,
+      "step": 21571
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.189523895661849e-05,
+      "loss": 0.1272,
+      "step": 21572
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1866025576663505e-05,
+      "loss": 0.1122,
+      "step": 21573
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1836831483040832e-05,
+      "loss": 0.0685,
+      "step": 21574
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1807656676326004e-05,
+      "loss": 0.068,
+      "step": 21575
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.177850115709423e-05,
+      "loss": 0.0807,
+      "step": 21576
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.174936492592039e-05,
+      "loss": 0.0906,
+      "step": 21577
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1720247983379127e-05,
+      "loss": 0.0963,
+      "step": 21578
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.169115033004443e-05,
+      "loss": 0.0593,
+      "step": 21579
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.166207196649006e-05,
+      "loss": 0.0836,
+      "step": 21580
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1633012893289227e-05,
+      "loss": 0.0756,
+      "step": 21581
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.160397311101514e-05,
+      "loss": 0.0841,
+      "step": 21582
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.157495262024023e-05,
+      "loss": 0.0911,
+      "step": 21583
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.154595142153659e-05,
+      "loss": 0.0813,
+      "step": 21584
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1516969515476438e-05,
+      "loss": 0.0878,
+      "step": 21585
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1488006902630975e-05,
+      "loss": 0.0665,
+      "step": 21586
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1459063583571302e-05,
+      "loss": 0.0745,
+      "step": 21587
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.143013955886808e-05,
+      "loss": 0.0912,
+      "step": 21588
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1401234829091733e-05,
+      "loss": 0.0754,
+      "step": 21589
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.137234939481192e-05,
+      "loss": 0.093,
+      "step": 21590
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1343483256598406e-05,
+      "loss": 0.0602,
+      "step": 21591
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1314636415020404e-05,
+      "loss": 0.1288,
+      "step": 21592
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1285808870646573e-05,
+      "loss": 0.1528,
+      "step": 21593
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.125700062404534e-05,
+      "loss": 0.0464,
+      "step": 21594
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1228211675784815e-05,
+      "loss": 0.0729,
+      "step": 21595
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1199442026432534e-05,
+      "loss": 0.0739,
+      "step": 21596
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.117069167655572e-05,
+      "loss": 0.1324,
+      "step": 21597
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.114196062672147e-05,
+      "loss": 0.0744,
+      "step": 21598
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1113248877496217e-05,
+      "loss": 0.0844,
+      "step": 21599
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1084556429445956e-05,
+      "loss": 0.097,
+      "step": 21600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1055883283136456e-05,
+      "loss": 0.1072,
+      "step": 21601
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1027229439133268e-05,
+      "loss": 0.1166,
+      "step": 21602
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0998594898001155e-05,
+      "loss": 0.0734,
+      "step": 21603
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0969979660304673e-05,
+      "loss": 0.0894,
+      "step": 21604
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0941383726608253e-05,
+      "loss": 0.0826,
+      "step": 21605
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0912807097475674e-05,
+      "loss": 0.1044,
+      "step": 21606
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0884249773470365e-05,
+      "loss": 0.0747,
+      "step": 21607
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0855711755155437e-05,
+      "loss": 0.0705,
+      "step": 21608
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.082719304309355e-05,
+      "loss": 0.1041,
+      "step": 21609
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0798693637847033e-05,
+      "loss": 0.0674,
+      "step": 21610
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0770213539977655e-05,
+      "loss": 0.0777,
+      "step": 21611
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0741752750047306e-05,
+      "loss": 0.1053,
+      "step": 21612
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0713311268616974e-05,
+      "loss": 0.0764,
+      "step": 21613
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.068488909624755e-05,
+      "loss": 0.1171,
+      "step": 21614
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.065648623349914e-05,
+      "loss": 0.1456,
+      "step": 21615
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.062810268093218e-05,
+      "loss": 0.0896,
+      "step": 21616
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.059973843910601e-05,
+      "loss": 0.0642,
+      "step": 21617
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0571393508580173e-05,
+      "loss": 0.1063,
+      "step": 21618
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0543067889913448e-05,
+      "loss": 0.1583,
+      "step": 21619
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0514761583664277e-05,
+      "loss": 0.0643,
+      "step": 21620
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.048647459039077e-05,
+      "loss": 0.0961,
+      "step": 21621
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.045820691065081e-05,
+      "loss": 0.0964,
+      "step": 21622
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0429958545001626e-05,
+      "loss": 0.183,
+      "step": 21623
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0401729494000322e-05,
+      "loss": 0.0776,
+      "step": 21624
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.037351975820334e-05,
+      "loss": 0.1031,
+      "step": 21625
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.034532933816713e-05,
+      "loss": 0.1399,
+      "step": 21626
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0317158234447354e-05,
+      "loss": 0.1597,
+      "step": 21627
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0289006447599567e-05,
+      "loss": 0.1354,
+      "step": 21628
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0260873978178883e-05,
+      "loss": 0.0906,
+      "step": 21629
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.023276082673997e-05,
+      "loss": 0.1144,
+      "step": 21630
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.020466699383694e-05,
+      "loss": 0.0918,
+      "step": 21631
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0176592480024126e-05,
+      "loss": 0.1149,
+      "step": 21632
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.014853728585475e-05,
+      "loss": 0.1036,
+      "step": 21633
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0120501411882264e-05,
+      "loss": 0.1451,
+      "step": 21634
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.009248485865922e-05,
+      "loss": 0.05,
+      "step": 21635
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.006448762673807e-05,
+      "loss": 0.0493,
+      "step": 21636
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0036509716671038e-05,
+      "loss": 0.0784,
+      "step": 21637
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0008551129009457e-05,
+      "loss": 0.0809,
+      "step": 21638
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9980611864305e-05,
+      "loss": 0.0526,
+      "step": 21639
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9952691923108224e-05,
+      "loss": 0.0699,
+      "step": 21640
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9924791305969915e-05,
+      "loss": 0.0878,
+      "step": 21641
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9896910013439963e-05,
+      "loss": 0.0684,
+      "step": 21642
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.986904804606815e-05,
+      "loss": 0.0432,
+      "step": 21643
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9841205404403927e-05,
+      "loss": 0.1432,
+      "step": 21644
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.98133820889963e-05,
+      "loss": 0.1101,
+      "step": 21645
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.978557810039383e-05,
+      "loss": 0.1102,
+      "step": 21646
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9757793439144743e-05,
+      "loss": 0.088,
+      "step": 21647
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9730028105796827e-05,
+      "loss": 0.0864,
+      "step": 21648
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.970228210089775e-05,
+      "loss": 0.0977,
+      "step": 21649
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9674555424994302e-05,
+      "loss": 0.0869,
+      "step": 21650
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9646848078633373e-05,
+      "loss": 0.0781,
+      "step": 21651
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9619160062361195e-05,
+      "loss": 0.053,
+      "step": 21652
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.959149137672378e-05,
+      "loss": 0.0605,
+      "step": 21653
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9563842022266686e-05,
+      "loss": 0.1057,
+      "step": 21654
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9536211999535037e-05,
+      "loss": 0.1012,
+      "step": 21655
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9508601309073616e-05,
+      "loss": 0.087,
+      "step": 21656
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9481009951426766e-05,
+      "loss": 0.078,
+      "step": 21657
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9453437927138717e-05,
+      "loss": 0.1074,
+      "step": 21658
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9425885236752928e-05,
+      "loss": 0.1145,
+      "step": 21659
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9398351880812848e-05,
+      "loss": 0.1475,
+      "step": 21660
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9370837859861267e-05,
+      "loss": 0.0909,
+      "step": 21661
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9343343174440753e-05,
+      "loss": 0.0972,
+      "step": 21662
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9315867825093316e-05,
+      "loss": 0.108,
+      "step": 21663
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.928841181236074e-05,
+      "loss": 0.101,
+      "step": 21664
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9260975136784375e-05,
+      "loss": 0.0938,
+      "step": 21665
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9233557798905342e-05,
+      "loss": 0.0875,
+      "step": 21666
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.92061597992641e-05,
+      "loss": 0.1387,
+      "step": 21667
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.917878113840088e-05,
+      "loss": 0.1041,
+      "step": 21668
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9151421816855584e-05,
+      "loss": 0.0645,
+      "step": 21669
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9124081835167673e-05,
+      "loss": 0.1414,
+      "step": 21670
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9096761193876043e-05,
+      "loss": 0.106,
+      "step": 21671
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.90694598935196e-05,
+      "loss": 0.0946,
+      "step": 21672
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9042177934636584e-05,
+      "loss": 0.0756,
+      "step": 21673
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9014915317765004e-05,
+      "loss": 0.1531,
+      "step": 21674
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8987672043442207e-05,
+      "loss": 0.0652,
+      "step": 21675
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8960448112205652e-05,
+      "loss": 0.1002,
+      "step": 21676
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8933243524591693e-05,
+      "loss": 0.108,
+      "step": 21677
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.890605828113723e-05,
+      "loss": 0.06,
+      "step": 21678
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8878892382377945e-05,
+      "loss": 0.0812,
+      "step": 21679
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.885174582884963e-05,
+      "loss": 0.0663,
+      "step": 21680
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.882461862108753e-05,
+      "loss": 0.0991,
+      "step": 21681
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8797510759626545e-05,
+      "loss": 0.0746,
+      "step": 21682
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.877042224500103e-05,
+      "loss": 0.0608,
+      "step": 21683
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8743353077745216e-05,
+      "loss": 0.0496,
+      "step": 21684
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8716303258392907e-05,
+      "loss": 0.1256,
+      "step": 21685
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8689272787477452e-05,
+      "loss": 0.0891,
+      "step": 21686
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8662261665531645e-05,
+      "loss": 0.0699,
+      "step": 21687
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8635269893088168e-05,
+      "loss": 0.0721,
+      "step": 21688
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.860829747067927e-05,
+      "loss": 0.0742,
+      "step": 21689
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.858134439883674e-05,
+      "loss": 0.095,
+      "step": 21690
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.855441067809205e-05,
+      "loss": 0.1022,
+      "step": 21691
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.852749630897621e-05,
+      "loss": 0.0632,
+      "step": 21692
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8500601292020026e-05,
+      "loss": 0.1567,
+      "step": 21693
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8473725627753736e-05,
+      "loss": 0.0775,
+      "step": 21694
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.844686931670725e-05,
+      "loss": 0.0697,
+      "step": 21695
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8420032359410144e-05,
+      "loss": 0.1168,
+      "step": 21696
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.839321475639133e-05,
+      "loss": 0.1002,
+      "step": 21697
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.836641650817994e-05,
+      "loss": 0.1217,
+      "step": 21698
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8339637615304216e-05,
+      "loss": 0.0829,
+      "step": 21699
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8312878078292074e-05,
+      "loss": 0.1127,
+      "step": 21700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.828613789767142e-05,
+      "loss": 0.1182,
+      "step": 21701
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8259417073969165e-05,
+      "loss": 0.093,
+      "step": 21702
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8232715607712448e-05,
+      "loss": 0.0609,
+      "step": 21703
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.820603349942751e-05,
+      "loss": 0.1167,
+      "step": 21704
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.817937074964071e-05,
+      "loss": 0.0836,
+      "step": 21705
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.815272735887763e-05,
+      "loss": 0.0865,
+      "step": 21706
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8126103327663624e-05,
+      "loss": 0.0841,
+      "step": 21707
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8099498656523606e-05,
+      "loss": 0.0833,
+      "step": 21708
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8072913345982266e-05,
+      "loss": 0.0819,
+      "step": 21709
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8046347396563633e-05,
+      "loss": 0.1102,
+      "step": 21710
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.801980080879162e-05,
+      "loss": 0.0385,
+      "step": 21711
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7993273583189694e-05,
+      "loss": 0.0692,
+      "step": 21712
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7966765720280887e-05,
+      "loss": 0.0887,
+      "step": 21713
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7940277220587887e-05,
+      "loss": 0.1154,
+      "step": 21714
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7913808084632943e-05,
+      "loss": 0.0718,
+      "step": 21715
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7887358312937975e-05,
+      "loss": 0.063,
+      "step": 21716
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7860927906024338e-05,
+      "loss": 0.0696,
+      "step": 21717
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7834516864413508e-05,
+      "loss": 0.075,
+      "step": 21718
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7808125188625955e-05,
+      "loss": 0.0645,
+      "step": 21719
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7781752879182268e-05,
+      "loss": 0.108,
+      "step": 21720
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7755399936602135e-05,
+      "loss": 0.1381,
+      "step": 21721
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7729066361405587e-05,
+      "loss": 0.0803,
+      "step": 21722
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7702752154111544e-05,
+      "loss": 0.0534,
+      "step": 21723
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7676457315238925e-05,
+      "loss": 0.0768,
+      "step": 21724
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7650181845306312e-05,
+      "loss": 0.0835,
+      "step": 21725
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7623925744831626e-05,
+      "loss": 0.0769,
+      "step": 21726
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7597689014332674e-05,
+      "loss": 0.119,
+      "step": 21727
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.757147165432682e-05,
+      "loss": 0.1571,
+      "step": 21728
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7545273665330874e-05,
+      "loss": 0.0797,
+      "step": 21729
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7519095047861423e-05,
+      "loss": 0.1305,
+      "step": 21730
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7492935802434717e-05,
+      "loss": 0.1136,
+      "step": 21731
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.746679592956657e-05,
+      "loss": 0.1144,
+      "step": 21732
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7440675429772234e-05,
+      "loss": 0.1125,
+      "step": 21733
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7414574303566854e-05,
+      "loss": 0.0793,
+      "step": 21734
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.738849255146513e-05,
+      "loss": 0.0642,
+      "step": 21735
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7362430173981313e-05,
+      "loss": 0.2068,
+      "step": 21736
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.733638717162911e-05,
+      "loss": 0.0709,
+      "step": 21737
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7310363544922212e-05,
+      "loss": 0.1055,
+      "step": 21738
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7284359294373664e-05,
+      "loss": 0.0861,
+      "step": 21739
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.725837442049627e-05,
+      "loss": 0.11,
+      "step": 21740
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.723240892380229e-05,
+      "loss": 0.089,
+      "step": 21741
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7206462804803647e-05,
+      "loss": 0.1373,
+      "step": 21742
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7180536064012155e-05,
+      "loss": 0.1,
+      "step": 21743
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7154628701938734e-05,
+      "loss": 0.0496,
+      "step": 21744
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7128740719094537e-05,
+      "loss": 0.0748,
+      "step": 21745
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.710287211598982e-05,
+      "loss": 0.0702,
+      "step": 21746
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.707702289313473e-05,
+      "loss": 0.066,
+      "step": 21747
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.705119305103886e-05,
+      "loss": 0.0524,
+      "step": 21748
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.702538259021147e-05,
+      "loss": 0.0796,
+      "step": 21749
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6999591511161593e-05,
+      "loss": 0.0862,
+      "step": 21750
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.697381981439772e-05,
+      "loss": 0.0649,
+      "step": 21751
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6948067500427988e-05,
+      "loss": 0.0701,
+      "step": 21752
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6922334569760224e-05,
+      "loss": 0.1364,
+      "step": 21753
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.689662102290168e-05,
+      "loss": 0.0649,
+      "step": 21754
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6870926860359516e-05,
+      "loss": 0.1415,
+      "step": 21755
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6845252082640317e-05,
+      "loss": 0.1213,
+      "step": 21756
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.681959669025024e-05,
+      "loss": 0.0769,
+      "step": 21757
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.679396068369532e-05,
+      "loss": 0.0933,
+      "step": 21758
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6768344063480932e-05,
+      "loss": 0.076,
+      "step": 21759
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.674274683011212e-05,
+      "loss": 0.0588,
+      "step": 21760
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.671716898409359e-05,
+      "loss": 0.0763,
+      "step": 21761
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6691610525929823e-05,
+      "loss": 0.0919,
+      "step": 21762
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6666071456124532e-05,
+      "loss": 0.0916,
+      "step": 21763
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6640551775181535e-05,
+      "loss": 0.0527,
+      "step": 21764
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.661505148360376e-05,
+      "loss": 0.0984,
+      "step": 21765
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6589570581894253e-05,
+      "loss": 0.0749,
+      "step": 21766
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.656410907055539e-05,
+      "loss": 0.0822,
+      "step": 21767
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6538666950088988e-05,
+      "loss": 0.0859,
+      "step": 21768
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.651324422099698e-05,
+      "loss": 0.0974,
+      "step": 21769
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6487840883780416e-05,
+      "loss": 0.0914,
+      "step": 21770
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.646245693894044e-05,
+      "loss": 0.1434,
+      "step": 21771
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6437092386977326e-05,
+      "loss": 0.087,
+      "step": 21772
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6411747228391226e-05,
+      "loss": 0.1157,
+      "step": 21773
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6386421463681967e-05,
+      "loss": 0.0812,
+      "step": 21774
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6361115093348922e-05,
+      "loss": 0.0997,
+      "step": 21775
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6335828117890917e-05,
+      "loss": 0.0897,
+      "step": 21776
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6310560537806663e-05,
+      "loss": 0.0644,
+      "step": 21777
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.628531235359454e-05,
+      "loss": 0.0598,
+      "step": 21778
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.626008356575204e-05,
+      "loss": 0.1333,
+      "step": 21779
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.623487417477687e-05,
+      "loss": 0.0665,
+      "step": 21780
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6209684181165973e-05,
+      "loss": 0.1076,
+      "step": 21781
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.618451358541606e-05,
+      "loss": 0.0593,
+      "step": 21782
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.61593623880234e-05,
+      "loss": 0.0837,
+      "step": 21783
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6134230589483934e-05,
+      "loss": 0.062,
+      "step": 21784
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.610911819029315e-05,
+      "loss": 0.0776,
+      "step": 21785
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6084025190946318e-05,
+      "loss": 0.056,
+      "step": 21786
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6058951591938153e-05,
+      "loss": 0.1018,
+      "step": 21787
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6033897393763043e-05,
+      "loss": 0.0529,
+      "step": 21788
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.600886259691503e-05,
+      "loss": 0.0861,
+      "step": 21789
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5983847201887504e-05,
+      "loss": 0.0781,
+      "step": 21790
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5958851209174065e-05,
+      "loss": 0.1304,
+      "step": 21791
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.593387461926743e-05,
+      "loss": 0.1052,
+      "step": 21792
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5908917432659987e-05,
+      "loss": 0.1616,
+      "step": 21793
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5883979649843892e-05,
+      "loss": 0.1028,
+      "step": 21794
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.585906127131076e-05,
+      "loss": 0.0807,
+      "step": 21795
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.583416229755219e-05,
+      "loss": 0.1171,
+      "step": 21796
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5809282729058793e-05,
+      "loss": 0.077,
+      "step": 21797
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.578442256632129e-05,
+      "loss": 0.1017,
+      "step": 21798
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5759581809829948e-05,
+      "loss": 0.0717,
+      "step": 21799
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5734760460074495e-05,
+      "loss": 0.1753,
+      "step": 21800
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5709958517544197e-05,
+      "loss": 0.1732,
+      "step": 21801
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.568517598272834e-05,
+      "loss": 0.0655,
+      "step": 21802
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.56604128561153e-05,
+      "loss": 0.0737,
+      "step": 21803
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5635669138193475e-05,
+      "loss": 0.0736,
+      "step": 21804
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5610944829450912e-05,
+      "loss": 0.092,
+      "step": 21805
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.558623993037489e-05,
+      "loss": 0.0583,
+      "step": 21806
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5561554441452574e-05,
+      "loss": 0.0876,
+      "step": 21807
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5536888363170686e-05,
+      "loss": 0.0829,
+      "step": 21808
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5512241696015727e-05,
+      "loss": 0.0874,
+      "step": 21809
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5487614440473418e-05,
+      "loss": 0.0834,
+      "step": 21810
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5463006597029595e-05,
+      "loss": 0.0878,
+      "step": 21811
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.543841816616931e-05,
+      "loss": 0.0854,
+      "step": 21812
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.541384914837751e-05,
+      "loss": 0.0856,
+      "step": 21813
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.538929954413848e-05,
+      "loss": 0.0803,
+      "step": 21814
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.536476935393627e-05,
+      "loss": 0.0855,
+      "step": 21815
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5340258578254717e-05,
+      "loss": 0.1401,
+      "step": 21816
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5315767217576994e-05,
+      "loss": 0.0685,
+      "step": 21817
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.529129527238593e-05,
+      "loss": 0.069,
+      "step": 21818
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5266842743164366e-05,
+      "loss": 0.0922,
+      "step": 21819
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5242409630394138e-05,
+      "loss": 0.0526,
+      "step": 21820
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5217995934557082e-05,
+      "loss": 0.1321,
+      "step": 21821
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5193601656134703e-05,
+      "loss": 0.0817,
+      "step": 21822
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5169226795607837e-05,
+      "loss": 0.0927,
+      "step": 21823
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.514487135345699e-05,
+      "loss": 0.063,
+      "step": 21824
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5120535330162776e-05,
+      "loss": 0.0778,
+      "step": 21825
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5096218726204702e-05,
+      "loss": 0.0872,
+      "step": 21826
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5071921542062383e-05,
+      "loss": 0.0838,
+      "step": 21827
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.504764377821477e-05,
+      "loss": 0.1194,
+      "step": 21828
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5023385435140701e-05,
+      "loss": 0.0849,
+      "step": 21829
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4999146513318352e-05,
+      "loss": 0.0788,
+      "step": 21830
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4974927013225781e-05,
+      "loss": 0.0873,
+      "step": 21831
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4950726935340498e-05,
+      "loss": 0.0928,
+      "step": 21832
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4926546280139675e-05,
+      "loss": 0.0936,
+      "step": 21833
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4902385048100042e-05,
+      "loss": 0.0432,
+      "step": 21834
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4878243239698108e-05,
+      "loss": 0.0516,
+      "step": 21835
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4854120855409714e-05,
+      "loss": 0.1428,
+      "step": 21836
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4830017895710591e-05,
+      "loss": 0.0663,
+      "step": 21837
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4805934361076023e-05,
+      "loss": 0.0642,
+      "step": 21838
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4781870251980744e-05,
+      "loss": 0.1449,
+      "step": 21839
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4757825568899485e-05,
+      "loss": 0.1074,
+      "step": 21840
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4733800312306089e-05,
+      "loss": 0.0573,
+      "step": 21841
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.470979448267451e-05,
+      "loss": 0.0851,
+      "step": 21842
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4685808080477814e-05,
+      "loss": 0.0966,
+      "step": 21843
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4661841106189066e-05,
+      "loss": 0.1054,
+      "step": 21844
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4637893560281002e-05,
+      "loss": 0.0498,
+      "step": 21845
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4613965443225574e-05,
+      "loss": 0.2316,
+      "step": 21846
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.459005675549463e-05,
+      "loss": 0.0885,
+      "step": 21847
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.456616749755979e-05,
+      "loss": 0.092,
+      "step": 21848
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.454229766989168e-05,
+      "loss": 0.0869,
+      "step": 21849
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4518447272961366e-05,
+      "loss": 0.1072,
+      "step": 21850
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4494616307238917e-05,
+      "loss": 0.0934,
+      "step": 21851
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.447080477319429e-05,
+      "loss": 0.0885,
+      "step": 21852
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4447012671297e-05,
+      "loss": 0.077,
+      "step": 21853
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4423240002016113e-05,
+      "loss": 0.0687,
+      "step": 21854
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4399486765820258e-05,
+      "loss": 0.0823,
+      "step": 21855
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4375752963178057e-05,
+      "loss": 0.0569,
+      "step": 21856
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4352038594557137e-05,
+      "loss": 0.0656,
+      "step": 21857
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4328343660425459e-05,
+      "loss": 0.1136,
+      "step": 21858
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4304668161249979e-05,
+      "loss": 0.0794,
+      "step": 21859
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4281012097497548e-05,
+      "loss": 0.0673,
+      "step": 21860
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4257375469634682e-05,
+      "loss": 0.104,
+      "step": 21861
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4233758278127452e-05,
+      "loss": 0.1125,
+      "step": 21862
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4210160523441485e-05,
+      "loss": 0.0778,
+      "step": 21863
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4186582206041965e-05,
+      "loss": 0.0626,
+      "step": 21864
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4163023326393854e-05,
+      "loss": 0.0904,
+      "step": 21865
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4139483884961891e-05,
+      "loss": 0.1097,
+      "step": 21866
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4115963882209926e-05,
+      "loss": 0.1283,
+      "step": 21867
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4092463318601812e-05,
+      "loss": 0.0935,
+      "step": 21868
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4068982194600844e-05,
+      "loss": 0.042,
+      "step": 21869
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4045520510670207e-05,
+      "loss": 0.0814,
+      "step": 21870
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.402207826727231e-05,
+      "loss": 0.0913,
+      "step": 21871
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3998655464869558e-05,
+      "loss": 0.0557,
+      "step": 21872
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3975252103923697e-05,
+      "loss": 0.0576,
+      "step": 21873
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3951868184896132e-05,
+      "loss": 0.0887,
+      "step": 21874
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3928503708248053e-05,
+      "loss": 0.1028,
+      "step": 21875
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3905158674439978e-05,
+      "loss": 0.0757,
+      "step": 21876
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3881833083932315e-05,
+      "loss": 0.0852,
+      "step": 21877
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3858526937185035e-05,
+      "loss": 0.0964,
+      "step": 21878
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3835240234657543e-05,
+      "loss": 0.0837,
+      "step": 21879
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.381197297680914e-05,
+      "loss": 0.0789,
+      "step": 21880
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3788725164098458e-05,
+      "loss": 0.0733,
+      "step": 21881
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3765496796984023e-05,
+      "loss": 0.0775,
+      "step": 21882
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3742287875923798e-05,
+      "loss": 0.0784,
+      "step": 21883
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3719098401375307e-05,
+      "loss": 0.0522,
+      "step": 21884
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3695928373795852e-05,
+      "loss": 0.0682,
+      "step": 21885
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.36727777936424e-05,
+      "loss": 0.078,
+      "step": 21886
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3649646661371251e-05,
+      "loss": 0.0861,
+      "step": 21887
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3626534977438598e-05,
+      "loss": 0.1016,
+      "step": 21888
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3603442742300077e-05,
+      "loss": 0.1488,
+      "step": 21889
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3580369956410987e-05,
+      "loss": 0.0762,
+      "step": 21890
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3557316620226411e-05,
+      "loss": 0.1367,
+      "step": 21891
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3534282734200765e-05,
+      "loss": 0.0725,
+      "step": 21892
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3511268298788238e-05,
+      "loss": 0.0712,
+      "step": 21893
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.348827331444269e-05,
+      "loss": 0.0797,
+      "step": 21894
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3465297781617424e-05,
+      "loss": 0.1243,
+      "step": 21895
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3442341700765637e-05,
+      "loss": 0.045,
+      "step": 21896
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.341940507233963e-05,
+      "loss": 0.0538,
+      "step": 21897
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3396487896792042e-05,
+      "loss": 0.1042,
+      "step": 21898
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3373590174574401e-05,
+      "loss": 0.0757,
+      "step": 21899
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3350711906138568e-05,
+      "loss": 0.0667,
+      "step": 21900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3327853091935294e-05,
+      "loss": 0.0609,
+      "step": 21901
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3305013732415438e-05,
+      "loss": 0.0765,
+      "step": 21902
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.328219382802931e-05,
+      "loss": 0.0804,
+      "step": 21903
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.325939337922688e-05,
+      "loss": 0.0743,
+      "step": 21904
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3236612386457792e-05,
+      "loss": 0.0788,
+      "step": 21905
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3213850850171127e-05,
+      "loss": 0.0696,
+      "step": 21906
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.319110877081564e-05,
+      "loss": 0.0858,
+      "step": 21907
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3168386148839973e-05,
+      "loss": 0.0782,
+      "step": 21908
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3145682984691875e-05,
+      "loss": 0.1149,
+      "step": 21909
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3122999278819103e-05,
+      "loss": 0.0671,
+      "step": 21910
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3100335031669076e-05,
+      "loss": 0.0958,
+      "step": 21911
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3077690243688544e-05,
+      "loss": 0.1093,
+      "step": 21912
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3055064915323823e-05,
+      "loss": 0.0635,
+      "step": 21913
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3032459047021327e-05,
+      "loss": 0.1158,
+      "step": 21914
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3009872639226706e-05,
+      "loss": 0.069,
+      "step": 21915
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2987305692385265e-05,
+      "loss": 0.0685,
+      "step": 21916
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2964758206941874e-05,
+      "loss": 0.1306,
+      "step": 21917
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2942230183341286e-05,
+      "loss": 0.085,
+      "step": 21918
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2919721622027703e-05,
+      "loss": 0.0743,
+      "step": 21919
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2897232523444769e-05,
+      "loss": 0.0957,
+      "step": 21920
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2874762888036018e-05,
+      "loss": 0.0837,
+      "step": 21921
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2852312716244542e-05,
+      "loss": 0.0849,
+      "step": 21922
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.282988200851276e-05,
+      "loss": 0.0699,
+      "step": 21923
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2807470765283213e-05,
+      "loss": 0.066,
+      "step": 21924
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2785078986997656e-05,
+      "loss": 0.0828,
+      "step": 21925
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2762706674097734e-05,
+      "loss": 0.0785,
+      "step": 21926
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2740353827024542e-05,
+      "loss": 0.0753,
+      "step": 21927
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2718020446218726e-05,
+      "loss": 0.1415,
+      "step": 21928
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.26957065321206e-05,
+      "loss": 0.0999,
+      "step": 21929
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2673412085170255e-05,
+      "loss": 0.0743,
+      "step": 21930
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2651137105807232e-05,
+      "loss": 0.1106,
+      "step": 21931
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2628881594470843e-05,
+      "loss": 0.1088,
+      "step": 21932
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2606645551599738e-05,
+      "loss": 0.0806,
+      "step": 21933
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2584428977632456e-05,
+      "loss": 0.1063,
+      "step": 21934
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.256223187300709e-05,
+      "loss": 0.1194,
+      "step": 21935
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.254005423816118e-05,
+      "loss": 0.0747,
+      "step": 21936
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2517896073532043e-05,
+      "loss": 0.0865,
+      "step": 21937
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2495757379556772e-05,
+      "loss": 0.111,
+      "step": 21938
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2473638156671685e-05,
+      "loss": 0.0637,
+      "step": 21939
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.245153840531299e-05,
+      "loss": 0.0847,
+      "step": 21940
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2429458125916449e-05,
+      "loss": 0.1066,
+      "step": 21941
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2407397318917269e-05,
+      "loss": 0.0722,
+      "step": 21942
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2385355984750657e-05,
+      "loss": 0.1039,
+      "step": 21943
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2363334123851156e-05,
+      "loss": 0.0944,
+      "step": 21944
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2341331736652861e-05,
+      "loss": 0.0648,
+      "step": 21945
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.231934882358976e-05,
+      "loss": 0.108,
+      "step": 21946
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2297385385095172e-05,
+      "loss": 0.0949,
+      "step": 21947
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2275441421602306e-05,
+      "loss": 0.0378,
+      "step": 21948
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2253516933543706e-05,
+      "loss": 0.0824,
+      "step": 21949
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2231611921351693e-05,
+      "loss": 0.1193,
+      "step": 21950
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2209726385458254e-05,
+      "loss": 0.084,
+      "step": 21951
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.218786032629482e-05,
+      "loss": 0.0593,
+      "step": 21952
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2166013744292603e-05,
+      "loss": 0.0782,
+      "step": 21953
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2144186639882371e-05,
+      "loss": 0.0629,
+      "step": 21954
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2122379013494333e-05,
+      "loss": 0.0745,
+      "step": 21955
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2100590865558703e-05,
+      "loss": 0.1375,
+      "step": 21956
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2078822196505024e-05,
+      "loss": 0.0963,
+      "step": 21957
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2057073006762509e-05,
+      "loss": 0.0947,
+      "step": 21958
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2035343296759927e-05,
+      "loss": 0.0544,
+      "step": 21959
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2013633066925821e-05,
+      "loss": 0.0928,
+      "step": 21960
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1991942317688299e-05,
+      "loss": 0.0834,
+      "step": 21961
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1970271049474902e-05,
+      "loss": 0.0719,
+      "step": 21962
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1948619262712957e-05,
+      "loss": 0.0792,
+      "step": 21963
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1926986957829456e-05,
+      "loss": 0.1229,
+      "step": 21964
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1905374135251057e-05,
+      "loss": 0.0886,
+      "step": 21965
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1883780795403531e-05,
+      "loss": 0.0643,
+      "step": 21966
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1862206938713094e-05,
+      "loss": 0.0723,
+      "step": 21967
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1840652565604736e-05,
+      "loss": 0.0784,
+      "step": 21968
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1819117676503788e-05,
+      "loss": 0.1255,
+      "step": 21969
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1797602271834573e-05,
+      "loss": 0.1323,
+      "step": 21970
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1776106352021532e-05,
+      "loss": 0.0748,
+      "step": 21971
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1754629917488435e-05,
+      "loss": 0.0634,
+      "step": 21972
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1733172968658723e-05,
+      "loss": 0.1503,
+      "step": 21973
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1711735505955611e-05,
+      "loss": 0.0842,
+      "step": 21974
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1690317529801541e-05,
+      "loss": 0.0768,
+      "step": 21975
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1668919040618953e-05,
+      "loss": 0.1099,
+      "step": 21976
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.164754003882973e-05,
+      "loss": 0.1475,
+      "step": 21977
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1626180524855534e-05,
+      "loss": 0.0773,
+      "step": 21978
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1604840499117365e-05,
+      "loss": 0.0882,
+      "step": 21979
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1583519962036103e-05,
+      "loss": 0.1161,
+      "step": 21980
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1562218914032196e-05,
+      "loss": 0.0756,
+      "step": 21981
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1540937355525416e-05,
+      "loss": 0.0483,
+      "step": 21982
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1519675286935539e-05,
+      "loss": 0.0894,
+      "step": 21983
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1498432708681893e-05,
+      "loss": 0.1694,
+      "step": 21984
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1477209621183149e-05,
+      "loss": 0.099,
+      "step": 21985
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1456006024857745e-05,
+      "loss": 0.0984,
+      "step": 21986
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1434821920123905e-05,
+      "loss": 0.0546,
+      "step": 21987
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1413657307399295e-05,
+      "loss": 0.1019,
+      "step": 21988
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1392512187101244e-05,
+      "loss": 0.0654,
+      "step": 21989
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1371386559646534e-05,
+      "loss": 0.1125,
+      "step": 21990
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1350280425451941e-05,
+      "loss": 0.1205,
+      "step": 21991
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1329193784933468e-05,
+      "loss": 0.1293,
+      "step": 21992
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1308126638506889e-05,
+      "loss": 0.0627,
+      "step": 21993
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1287078986587762e-05,
+      "loss": 0.0578,
+      "step": 21994
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1266050829590869e-05,
+      "loss": 0.0809,
+      "step": 21995
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1245042167930875e-05,
+      "loss": 0.0942,
+      "step": 21996
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1224053002022006e-05,
+      "loss": 0.0729,
+      "step": 21997
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1203083332278263e-05,
+      "loss": 0.0982,
+      "step": 21998
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1182133159113096e-05,
+      "loss": 0.1069,
+      "step": 21999
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1161202482939392e-05,
+      "loss": 0.072,
+      "step": 22000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1140291304170158e-05,
+      "loss": 0.0933,
+      "step": 22001
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1119399623217396e-05,
+      "loss": 0.0632,
+      "step": 22002
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1098527440493222e-05,
+      "loss": 0.1202,
+      "step": 22003
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1077674756409085e-05,
+      "loss": 0.1023,
+      "step": 22004
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1056841571376208e-05,
+      "loss": 0.125,
+      "step": 22005
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1036027885805489e-05,
+      "loss": 0.0765,
+      "step": 22006
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1015233700107042e-05,
+      "loss": 0.1302,
+      "step": 22007
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0994459014691094e-05,
+      "loss": 0.1108,
+      "step": 22008
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0973703829967096e-05,
+      "loss": 0.1337,
+      "step": 22009
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.09529681463445e-05,
+      "loss": 0.0855,
+      "step": 22010
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0932251964231976e-05,
+      "loss": 0.0759,
+      "step": 22011
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0911555284038198e-05,
+      "loss": 0.0861,
+      "step": 22012
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0890878106171064e-05,
+      "loss": 0.115,
+      "step": 22013
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0870220431038358e-05,
+      "loss": 0.0776,
+      "step": 22014
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0849582259047308e-05,
+      "loss": 0.1028,
+      "step": 22015
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0828963590605035e-05,
+      "loss": 0.0956,
+      "step": 22016
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0808364426117767e-05,
+      "loss": 0.0632,
+      "step": 22017
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.078778476599207e-05,
+      "loss": 0.093,
+      "step": 22018
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0767224610633287e-05,
+      "loss": 0.0832,
+      "step": 22019
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0746683960447312e-05,
+      "loss": 0.1249,
+      "step": 22020
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0726162815838714e-05,
+      "loss": 0.067,
+      "step": 22021
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.070566117721239e-05,
+      "loss": 0.1301,
+      "step": 22022
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.068517904497246e-05,
+      "loss": 0.095,
+      "step": 22023
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0664716419522713e-05,
+      "loss": 0.08,
+      "step": 22024
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0644273301266828e-05,
+      "loss": 0.0755,
+      "step": 22025
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0623849690607813e-05,
+      "loss": 0.0903,
+      "step": 22026
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0603445587948234e-05,
+      "loss": 0.1836,
+      "step": 22027
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0583060993690552e-05,
+      "loss": 0.077,
+      "step": 22028
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0562695908236552e-05,
+      "loss": 0.0765,
+      "step": 22029
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0542350331988027e-05,
+      "loss": 0.1328,
+      "step": 22030
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0522024265345875e-05,
+      "loss": 0.0942,
+      "step": 22031
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0501717708711111e-05,
+      "loss": 0.0611,
+      "step": 22032
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.048143066248397e-05,
+      "loss": 0.0778,
+      "step": 22033
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0461163127064466e-05,
+      "loss": 0.0989,
+      "step": 22034
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0440915102852388e-05,
+      "loss": 0.0926,
+      "step": 22035
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0420686590246752e-05,
+      "loss": 0.0483,
+      "step": 22036
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0400477589646462e-05,
+      "loss": 0.0651,
+      "step": 22037
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0380288101450087e-05,
+      "loss": 0.1027,
+      "step": 22038
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.036011812605575e-05,
+      "loss": 0.1328,
+      "step": 22039
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0339967663860916e-05,
+      "loss": 0.0972,
+      "step": 22040
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0319836715263153e-05,
+      "loss": 0.0986,
+      "step": 22041
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0299725280659255e-05,
+      "loss": 0.085,
+      "step": 22042
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0279633360445795e-05,
+      "loss": 0.0652,
+      "step": 22043
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0259560955019009e-05,
+      "loss": 0.1077,
+      "step": 22044
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0239508064774582e-05,
+      "loss": 0.0701,
+      "step": 22045
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0219474690107978e-05,
+      "loss": 0.0881,
+      "step": 22046
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0199460831414099e-05,
+      "loss": 0.1151,
+      "step": 22047
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.017946648908774e-05,
+      "loss": 0.0681,
+      "step": 22048
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0159491663522924e-05,
+      "loss": 0.0736,
+      "step": 22049
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0139536355113554e-05,
+      "loss": 0.0855,
+      "step": 22050
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0119600564253206e-05,
+      "loss": 0.0906,
+      "step": 22051
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0099684291334899e-05,
+      "loss": 0.0746,
+      "step": 22052
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0079787536751428e-05,
+      "loss": 0.0695,
+      "step": 22053
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0059910300895037e-05,
+      "loss": 0.0563,
+      "step": 22054
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0040052584157522e-05,
+      "loss": 0.059,
+      "step": 22055
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0020214386930682e-05,
+      "loss": 0.0936,
+      "step": 22056
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0000395709605314e-05,
+      "loss": 0.1378,
+      "step": 22057
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.980596552572663e-06,
+      "loss": 0.0815,
+      "step": 22058
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.960816916222747e-06,
+      "loss": 0.0772,
+      "step": 22059
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.941056800945702e-06,
+      "loss": 0.1118,
+      "step": 22060
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.921316207131214e-06,
+      "loss": 0.082,
+      "step": 22061
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.901595135168418e-06,
+      "loss": 0.1106,
+      "step": 22062
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.88189358544611e-06,
+      "loss": 0.105,
+      "step": 22063
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.862211558352874e-06,
+      "loss": 0.1237,
+      "step": 22064
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.84254905427684e-06,
+      "loss": 0.0735,
+      "step": 22065
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.822906073605586e-06,
+      "loss": 0.0549,
+      "step": 22066
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.803282616726362e-06,
+      "loss": 0.0646,
+      "step": 22067
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.783678684026299e-06,
+      "loss": 0.11,
+      "step": 22068
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.764094275891755e-06,
+      "loss": 0.0699,
+      "step": 22069
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.744529392708868e-06,
+      "loss": 0.0837,
+      "step": 22070
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.72498403486366e-06,
+      "loss": 0.0662,
+      "step": 22071
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.705458202741268e-06,
+      "loss": 0.1515,
+      "step": 22072
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.68595189672672e-06,
+      "loss": 0.1057,
+      "step": 22073
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.666465117204815e-06,
+      "loss": 0.0786,
+      "step": 22074
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.646997864559581e-06,
+      "loss": 0.1511,
+      "step": 22075
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.627550139174934e-06,
+      "loss": 0.1055,
+      "step": 22076
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.608121941434233e-06,
+      "loss": 0.0693,
+      "step": 22077
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.588713271720728e-06,
+      "loss": 0.1086,
+      "step": 22078
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.569324130417111e-06,
+      "loss": 0.086,
+      "step": 22079
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.549954517905523e-06,
+      "loss": 0.0388,
+      "step": 22080
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.530604434567991e-06,
+      "loss": 0.0914,
+      "step": 22081
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.511273880785986e-06,
+      "loss": 0.0609,
+      "step": 22082
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.49196285694076e-06,
+      "loss": 0.0638,
+      "step": 22083
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.47267136341301e-06,
+      "loss": 0.1278,
+      "step": 22084
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.453399400583096e-06,
+      "loss": 0.0835,
+      "step": 22085
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.434146968831158e-06,
+      "loss": 0.0781,
+      "step": 22086
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.414914068536562e-06,
+      "loss": 0.0554,
+      "step": 22087
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.395700700078669e-06,
+      "loss": 0.0852,
+      "step": 22088
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.376506863836398e-06,
+      "loss": 0.0626,
+      "step": 22089
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.357332560188003e-06,
+      "loss": 0.0925,
+      "step": 22090
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.338177789511736e-06,
+      "loss": 0.1226,
+      "step": 22091
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.319042552185186e-06,
+      "loss": 0.0751,
+      "step": 22092
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.299926848585604e-06,
+      "loss": 0.0938,
+      "step": 22093
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.280830679090024e-06,
+      "loss": 0.0731,
+      "step": 22094
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.261754044074921e-06,
+      "loss": 0.1008,
+      "step": 22095
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.24269694391644e-06,
+      "loss": 0.1609,
+      "step": 22096
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.223659378990278e-06,
+      "loss": 0.0818,
+      "step": 22097
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.204641349672027e-06,
+      "loss": 0.0608,
+      "step": 22098
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.185642856336385e-06,
+      "loss": 0.0576,
+      "step": 22099
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.166663899358162e-06,
+      "loss": 0.0596,
+      "step": 22100
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.147704479111506e-06,
+      "loss": 0.0779,
+      "step": 22101
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.128764595970118e-06,
+      "loss": 0.1233,
+      "step": 22102
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.109844250307586e-06,
+      "loss": 0.1042,
+      "step": 22103
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.090943442497058e-06,
+      "loss": 0.0423,
+      "step": 22104
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.07206217291101e-06,
+      "loss": 0.0564,
+      "step": 22105
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.053200441921927e-06,
+      "loss": 0.0734,
+      "step": 22106
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.034358249901509e-06,
+      "loss": 0.0985,
+      "step": 22107
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.015535597221458e-06,
+      "loss": 0.0679,
+      "step": 22108
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.996732484252813e-06,
+      "loss": 0.0663,
+      "step": 22109
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.977948911366274e-06,
+      "loss": 0.0807,
+      "step": 22110
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.959184878932435e-06,
+      "loss": 0.1161,
+      "step": 22111
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.940440387321004e-06,
+      "loss": 0.0603,
+      "step": 22112
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.921715436901679e-06,
+      "loss": 0.0938,
+      "step": 22113
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.903010028043723e-06,
+      "loss": 0.0673,
+      "step": 22114
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.884324161115953e-06,
+      "loss": 0.0786,
+      "step": 22115
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.86565783648663e-06,
+      "loss": 0.0824,
+      "step": 22116
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.847011054524013e-06,
+      "loss": 0.0807,
+      "step": 22117
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.82838381559581e-06,
+      "loss": 0.1143,
+      "step": 22118
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.809776120069057e-06,
+      "loss": 0.0578,
+      "step": 22119
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.79118796831091e-06,
+      "loss": 0.0767,
+      "step": 22120
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.772619360687629e-06,
+      "loss": 0.0901,
+      "step": 22121
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.754070297565586e-06,
+      "loss": 0.1383,
+      "step": 22122
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.735540779310268e-06,
+      "loss": 0.0829,
+      "step": 22123
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.717030806287274e-06,
+      "loss": 0.0903,
+      "step": 22124
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.69854037886142e-06,
+      "loss": 0.0775,
+      "step": 22125
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.680069497397303e-06,
+      "loss": 0.1251,
+      "step": 22126
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.6616181622593e-06,
+      "loss": 0.0751,
+      "step": 22127
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.643186373811007e-06,
+      "loss": 0.0938,
+      "step": 22128
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.624774132415913e-06,
+      "loss": 0.0793,
+      "step": 22129
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.606381438437061e-06,
+      "loss": 0.123,
+      "step": 22130
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.58800829223716e-06,
+      "loss": 0.0911,
+      "step": 22131
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.569654694178364e-06,
+      "loss": 0.0701,
+      "step": 22132
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.551320644622717e-06,
+      "loss": 0.1056,
+      "step": 22133
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.53300614393171e-06,
+      "loss": 0.0986,
+      "step": 22134
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.514711192466274e-06,
+      "loss": 0.129,
+      "step": 22135
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.496435790587232e-06,
+      "loss": 0.063,
+      "step": 22136
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.478179938654962e-06,
+      "loss": 0.0667,
+      "step": 22137
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.45994363702951e-06,
+      "loss": 0.0635,
+      "step": 22138
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.441726886070367e-06,
+      "loss": 0.0543,
+      "step": 22139
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.423529686136688e-06,
+      "loss": 0.0725,
+      "step": 22140
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.4053520375873e-06,
+      "loss": 0.0843,
+      "step": 22141
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.38719394078058e-06,
+      "loss": 0.0817,
+      "step": 22142
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.369055396074576e-06,
+      "loss": 0.0696,
+      "step": 22143
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.350936403827003e-06,
+      "loss": 0.0749,
+      "step": 22144
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.332836964395018e-06,
+      "loss": 0.0743,
+      "step": 22145
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.31475707813567e-06,
+      "loss": 0.0852,
+      "step": 22146
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.29669674540512e-06,
+      "loss": 0.0547,
+      "step": 22147
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.278655966559856e-06,
+      "loss": 0.0506,
+      "step": 22148
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.260634741955376e-06,
+      "loss": 0.0917,
+      "step": 22149
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.242633071947058e-06,
+      "loss": 0.149,
+      "step": 22150
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.224650956889957e-06,
+      "loss": 0.0432,
+      "step": 22151
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.20668839713845e-06,
+      "loss": 0.08,
+      "step": 22152
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.188745393046815e-06,
+      "loss": 0.0622,
+      "step": 22153
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.170821944968764e-06,
+      "loss": 0.092,
+      "step": 22154
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.152918053257797e-06,
+      "loss": 0.0969,
+      "step": 22155
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.135033718266849e-06,
+      "loss": 0.1199,
+      "step": 22156
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.11716894034864e-06,
+      "loss": 0.072,
+      "step": 22157
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.099323719855335e-06,
+      "loss": 0.1132,
+      "step": 22158
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.08149805713898e-06,
+      "loss": 0.0867,
+      "step": 22159
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.063691952550856e-06,
+      "loss": 0.0635,
+      "step": 22160
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.045905406442011e-06,
+      "loss": 0.1431,
+      "step": 22161
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.02813841916339e-06,
+      "loss": 0.1217,
+      "step": 22162
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.010390991065042e-06,
+      "loss": 0.0648,
+      "step": 22163
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.992663122497135e-06,
+      "loss": 0.1116,
+      "step": 22164
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.974954813809166e-06,
+      "loss": 0.1162,
+      "step": 22165
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.957266065350189e-06,
+      "loss": 0.0985,
+      "step": 22166
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.939596877469035e-06,
+      "loss": 0.0906,
+      "step": 22167
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.921947250513983e-06,
+      "loss": 0.187,
+      "step": 22168
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.904317184833309e-06,
+      "loss": 0.115,
+      "step": 22169
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.886706680774403e-06,
+      "loss": 0.0842,
+      "step": 22170
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.869115738684652e-06,
+      "loss": 0.0883,
+      "step": 22171
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.851544358910778e-06,
+      "loss": 0.1356,
+      "step": 22172
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.833992541799395e-06,
+      "loss": 0.1025,
+      "step": 22173
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.816460287696336e-06,
+      "loss": 0.0922,
+      "step": 22174
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.798947596947548e-06,
+      "loss": 0.1144,
+      "step": 22175
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.781454469898198e-06,
+      "loss": 0.0708,
+      "step": 22176
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.76398090689323e-06,
+      "loss": 0.1154,
+      "step": 22177
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.746526908277262e-06,
+      "loss": 0.0923,
+      "step": 22178
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.72909247439424e-06,
+      "loss": 0.111,
+      "step": 22179
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.711677605588108e-06,
+      "loss": 0.0931,
+      "step": 22180
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.69428230220226e-06,
+      "loss": 0.1109,
+      "step": 22181
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.676906564579534e-06,
+      "loss": 0.1309,
+      "step": 22182
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.659550393062653e-06,
+      "loss": 0.1294,
+      "step": 22183
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.64221378799379e-06,
+      "loss": 0.0527,
+      "step": 22184
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.624896749714783e-06,
+      "loss": 0.0975,
+      "step": 22185
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.607599278567134e-06,
+      "loss": 0.0635,
+      "step": 22186
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.590321374891795e-06,
+      "loss": 0.1293,
+      "step": 22187
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.573063039029604e-06,
+      "loss": 0.0548,
+      "step": 22188
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.555824271320622e-06,
+      "loss": 0.0923,
+      "step": 22189
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.53860507210502e-06,
+      "loss": 0.0884,
+      "step": 22190
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.521405441722085e-06,
+      "loss": 0.1199,
+      "step": 22191
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.504225380511098e-06,
+      "loss": 0.0988,
+      "step": 22192
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.4870648888107905e-06,
+      "loss": 0.0836,
+      "step": 22193
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.469923966959446e-06,
+      "loss": 0.1156,
+      "step": 22194
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.452802615295129e-06,
+      "loss": 0.0916,
+      "step": 22195
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.435700834155457e-06,
+      "loss": 0.0818,
+      "step": 22196
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.418618623877382e-06,
+      "loss": 0.1396,
+      "step": 22197
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.401555984797969e-06,
+      "loss": 0.0541,
+      "step": 22198
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.384512917253727e-06,
+      "loss": 0.0917,
+      "step": 22199
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.367489421580387e-06,
+      "loss": 0.0906,
+      "step": 22200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.35048549811379e-06,
+      "loss": 0.0957,
+      "step": 22201
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.333501147189337e-06,
+      "loss": 0.0785,
+      "step": 22202
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.316536369141646e-06,
+      "loss": 0.087,
+      "step": 22203
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.299591164305452e-06,
+      "loss": 0.0723,
+      "step": 22204
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.2826655330147095e-06,
+      "loss": 0.1326,
+      "step": 22205
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.2657594756032615e-06,
+      "loss": 0.1118,
+      "step": 22206
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.2488729924043985e-06,
+      "loss": 0.0799,
+      "step": 22207
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.232006083751075e-06,
+      "loss": 0.0932,
+      "step": 22208
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.215158749975914e-06,
+      "loss": 0.1127,
+      "step": 22209
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.198330991410984e-06,
+      "loss": 0.0884,
+      "step": 22210
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.181522808388241e-06,
+      "loss": 0.0507,
+      "step": 22211
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.164734201238976e-06,
+      "loss": 0.0716,
+      "step": 22212
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.147965170294368e-06,
+      "loss": 0.1132,
+      "step": 22213
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.13121571588482e-06,
+      "loss": 0.0536,
+      "step": 22214
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.114485838340845e-06,
+      "loss": 0.0903,
+      "step": 22215
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.097775537992179e-06,
+      "loss": 0.1399,
+      "step": 22216
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.081084815168226e-06,
+      "loss": 0.0933,
+      "step": 22217
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.064413670198277e-06,
+      "loss": 0.047,
+      "step": 22218
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.047762103410959e-06,
+      "loss": 0.0831,
+      "step": 22219
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.031130115134565e-06,
+      "loss": 0.0991,
+      "step": 22220
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.014517705697054e-06,
+      "loss": 0.1934,
+      "step": 22221
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.997924875426053e-06,
+      "loss": 0.0609,
+      "step": 22222
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.9813516246485245e-06,
+      "loss": 0.0841,
+      "step": 22223
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.9647979536915375e-06,
+      "loss": 0.0956,
+      "step": 22224
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.948263862881276e-06,
+      "loss": 0.0889,
+      "step": 22225
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.931749352543926e-06,
+      "loss": 0.0872,
+      "step": 22226
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.915254423004891e-06,
+      "loss": 0.0652,
+      "step": 22227
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.898779074589578e-06,
+      "loss": 0.0917,
+      "step": 22228
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.8823233076227286e-06,
+      "loss": 0.0739,
+      "step": 22229
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.86588712242886e-06,
+      "loss": 0.103,
+      "step": 22230
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.849470519332157e-06,
+      "loss": 0.0967,
+      "step": 22231
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.83307349865614e-06,
+      "loss": 0.1415,
+      "step": 22232
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.816696060724215e-06,
+      "loss": 0.1093,
+      "step": 22233
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.800338205859236e-06,
+      "loss": 0.1165,
+      "step": 22234
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.7839999343838336e-06,
+      "loss": 0.063,
+      "step": 22235
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.7676812466201945e-06,
+      "loss": 0.0859,
+      "step": 22236
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.7513821428898395e-06,
+      "loss": 0.0797,
+      "step": 22237
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.7351026235144e-06,
+      "loss": 0.1102,
+      "step": 22238
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.71884268881473e-06,
+      "loss": 0.072,
+      "step": 22239
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.702602339111463e-06,
+      "loss": 0.0621,
+      "step": 22240
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.686381574724787e-06,
+      "loss": 0.0797,
+      "step": 22241
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.6701803959746675e-06,
+      "loss": 0.0747,
+      "step": 22242
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.6539988031802945e-06,
+      "loss": 0.0465,
+      "step": 22243
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.637836796661079e-06,
+      "loss": 0.0404,
+      "step": 22244
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.621694376735432e-06,
+      "loss": 0.0649,
+      "step": 22245
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.605571543721767e-06,
+      "loss": 0.0788,
+      "step": 22246
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.589468297937829e-06,
+      "loss": 0.0721,
+      "step": 22247
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.573384639701363e-06,
+      "loss": 0.0708,
+      "step": 22248
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.557320569329339e-06,
+      "loss": 0.0503,
+      "step": 22249
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.541276087138503e-06,
+      "loss": 0.088,
+      "step": 22250
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.5252511934451585e-06,
+      "loss": 0.0599,
+      "step": 22251
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.509245888565496e-06,
+      "loss": 0.0908,
+      "step": 22252
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.49326017281493e-06,
+      "loss": 0.1312,
+      "step": 22253
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.477294046508653e-06,
+      "loss": 0.0846,
+      "step": 22254
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.4613475099616345e-06,
+      "loss": 0.0585,
+      "step": 22255
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.445420563488069e-06,
+      "loss": 0.0616,
+      "step": 22256
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.4295132074021486e-06,
+      "loss": 0.0802,
+      "step": 22257
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.413625442017401e-06,
+      "loss": 0.1158,
+      "step": 22258
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.397757267647353e-06,
+      "loss": 0.0733,
+      "step": 22259
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.381908684604531e-06,
+      "loss": 0.0878,
+      "step": 22260
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.366079693201799e-06,
+      "loss": 0.0432,
+      "step": 22261
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.350270293750904e-06,
+      "loss": 0.1097,
+      "step": 22262
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.3344804865638205e-06,
+      "loss": 0.067,
+      "step": 22263
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.318710271951744e-06,
+      "loss": 0.0755,
+      "step": 22264
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.302959650225759e-06,
+      "loss": 0.0727,
+      "step": 22265
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.287228621696284e-06,
+      "loss": 0.0929,
+      "step": 22266
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.271517186673625e-06,
+      "loss": 0.0939,
+      "step": 22267
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.255825345467425e-06,
+      "loss": 0.0924,
+      "step": 22268
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.240153098387213e-06,
+      "loss": 0.0944,
+      "step": 22269
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.224500445741965e-06,
+      "loss": 0.0919,
+      "step": 22270
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.208867387840322e-06,
+      "loss": 0.0682,
+      "step": 22271
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.193253924990483e-06,
+      "loss": 0.0569,
+      "step": 22272
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.177660057500422e-06,
+      "loss": 0.0553,
+      "step": 22273
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.162085785677451e-06,
+      "loss": 0.0728,
+      "step": 22274
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.146531109828768e-06,
+      "loss": 0.1027,
+      "step": 22275
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.130996030261015e-06,
+      "loss": 0.0668,
+      "step": 22276
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.115480547280505e-06,
+      "loss": 0.08,
+      "step": 22277
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.099984661193214e-06,
+      "loss": 0.0938,
+      "step": 22278
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.0845083723046754e-06,
+      "loss": 0.1144,
+      "step": 22279
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.0690516809199795e-06,
+      "loss": 0.1163,
+      "step": 22280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.0536145873439915e-06,
+      "loss": 0.0694,
+      "step": 22281
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.038197091881026e-06,
+      "loss": 0.1399,
+      "step": 22282
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.02279919483495e-06,
+      "loss": 0.1231,
+      "step": 22283
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.00742089650963e-06,
+      "loss": 0.0847,
+      "step": 22284
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.992062197208159e-06,
+      "loss": 0.0857,
+      "step": 22285
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.976723097233294e-06,
+      "loss": 0.1176,
+      "step": 22286
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.961403596887682e-06,
+      "loss": 0.0871,
+      "step": 22287
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.946103696473193e-06,
+      "loss": 0.0831,
+      "step": 22288
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.930823396291585e-06,
+      "loss": 0.0724,
+      "step": 22289
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.915562696644172e-06,
+      "loss": 0.0959,
+      "step": 22290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.900321597831826e-06,
+      "loss": 0.0962,
+      "step": 22291
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.885100100155083e-06,
+      "loss": 0.0876,
+      "step": 22292
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.869898203914037e-06,
+      "loss": 0.1049,
+      "step": 22293
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.854715909408337e-06,
+      "loss": 0.0974,
+      "step": 22294
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.839553216937521e-06,
+      "loss": 0.1062,
+      "step": 22295
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8244101268004616e-06,
+      "loss": 0.0883,
+      "step": 22296
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.8092866392958076e-06,
+      "loss": 0.1337,
+      "step": 22297
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.794182754721766e-06,
+      "loss": 0.0836,
+      "step": 22298
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.779098473375988e-06,
+      "loss": 0.0724,
+      "step": 22299
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.764033795556123e-06,
+      "loss": 0.066,
+      "step": 22300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7489887215590455e-06,
+      "loss": 0.0639,
+      "step": 22301
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.733963251681407e-06,
+      "loss": 0.0878,
+      "step": 22302
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.7189573862195255e-06,
+      "loss": 0.0821,
+      "step": 22303
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.703971125469276e-06,
+      "loss": 0.0483,
+      "step": 22304
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6890044697262e-06,
+      "loss": 0.0785,
+      "step": 22305
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.6740574192851725e-06,
+      "loss": 0.0799,
+      "step": 22306
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.659129974441291e-06,
+      "loss": 0.1176,
+      "step": 22307
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.644222135488541e-06,
+      "loss": 0.0699,
+      "step": 22308
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.629333902721134e-06,
+      "loss": 0.0647,
+      "step": 22309
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.61446527643239e-06,
+      "loss": 0.0605,
+      "step": 22310
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.599616256915629e-06,
+      "loss": 0.0522,
+      "step": 22311
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.584786844463619e-06,
+      "loss": 0.0727,
+      "step": 22312
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.569977039368679e-06,
+      "loss": 0.0935,
+      "step": 22313
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.555186841923021e-06,
+      "loss": 0.0919,
+      "step": 22314
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.540416252417968e-06,
+      "loss": 0.1053,
+      "step": 22315
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.525665271144953e-06,
+      "loss": 0.1377,
+      "step": 22316
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.5109338983948545e-06,
+      "loss": 0.0675,
+      "step": 22317
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.496222134457996e-06,
+      "loss": 0.0573,
+      "step": 22318
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4815299796247e-06,
+      "loss": 0.0895,
+      "step": 22319
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.466857434184402e-06,
+      "loss": 0.0528,
+      "step": 22320
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.452204498426428e-06,
+      "loss": 0.0676,
+      "step": 22321
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.437571172639877e-06,
+      "loss": 0.071,
+      "step": 22322
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.422957457113076e-06,
+      "loss": 0.0905,
+      "step": 22323
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.40836335213446e-06,
+      "loss": 0.0475,
+      "step": 22324
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.393788857991466e-06,
+      "loss": 0.0955,
+      "step": 22325
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.379233974971642e-06,
+      "loss": 0.0701,
+      "step": 22326
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.36469870336187e-06,
+      "loss": 0.0653,
+      "step": 22327
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3501830434488085e-06,
+      "loss": 0.0806,
+      "step": 22328
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.335686995518674e-06,
+      "loss": 0.0966,
+      "step": 22329
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.321210559857348e-06,
+      "loss": 0.057,
+      "step": 22330
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.306753736750158e-06,
+      "loss": 0.0668,
+      "step": 22331
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.292316526482321e-06,
+      "loss": 0.1235,
+      "step": 22332
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.277898929338276e-06,
+      "loss": 0.1146,
+      "step": 22333
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2635009456024615e-06,
+      "loss": 0.0988,
+      "step": 22334
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.249122575558763e-06,
+      "loss": 0.0677,
+      "step": 22335
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.234763819490507e-06,
+      "loss": 0.0822,
+      "step": 22336
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.2204246776810235e-06,
+      "loss": 0.16,
+      "step": 22337
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.206105150412976e-06,
+      "loss": 0.0706,
+      "step": 22338
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.191805237968694e-06,
+      "loss": 0.0784,
+      "step": 22339
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.177524940630174e-06,
+      "loss": 0.1201,
+      "step": 22340
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.163264258678968e-06,
+      "loss": 0.0701,
+      "step": 22341
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.149023192396296e-06,
+      "loss": 0.0916,
+      "step": 22342
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.134801742062822e-06,
+      "loss": 0.1118,
+      "step": 22343
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.1205999079591e-06,
+      "loss": 0.1454,
+      "step": 22344
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.106417690365128e-06,
+      "loss": 0.0527,
+      "step": 22345
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.09225508956046e-06,
+      "loss": 0.0788,
+      "step": 22346
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.078112105824428e-06,
+      "loss": 0.1099,
+      "step": 22347
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.0639887394358095e-06,
+      "loss": 0.0793,
+      "step": 22348
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.049884990673271e-06,
+      "loss": 0.0481,
+      "step": 22349
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.035800859814587e-06,
+      "loss": 0.101,
+      "step": 22350
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.021736347137762e-06,
+      "loss": 0.1201,
+      "step": 22351
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.007691452920016e-06,
+      "loss": 0.0441,
+      "step": 22352
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.9936661774383495e-06,
+      "loss": 0.0577,
+      "step": 22353
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.979660520969098e-06,
+      "loss": 0.1045,
+      "step": 22354
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.965674483788596e-06,
+      "loss": 0.0896,
+      "step": 22355
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.951708066172511e-06,
+      "loss": 0.073,
+      "step": 22356
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.93776126839629e-06,
+      "loss": 0.0688,
+      "step": 22357
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.923834090734936e-06,
+      "loss": 0.1005,
+      "step": 22358
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.909926533462894e-06,
+      "loss": 0.0813,
+      "step": 22359
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.896038596854724e-06,
+      "loss": 0.0927,
+      "step": 22360
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.882170281183984e-06,
+      "loss": 0.0461,
+      "step": 22361
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.868321586724122e-06,
+      "loss": 0.0637,
+      "step": 22362
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.854492513748366e-06,
+      "loss": 0.0718,
+      "step": 22363
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.840683062529272e-06,
+      "loss": 0.0881,
+      "step": 22364
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.826893233339181e-06,
+      "loss": 0.0842,
+      "step": 22365
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.813123026449873e-06,
+      "loss": 0.1013,
+      "step": 22366
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7993724421330205e-06,
+      "loss": 0.0526,
+      "step": 22367
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.785641480659741e-06,
+      "loss": 0.0834,
+      "step": 22368
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.771930142300706e-06,
+      "loss": 0.0885,
+      "step": 22369
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.758238427326256e-06,
+      "loss": 0.0914,
+      "step": 22370
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.744566336006284e-06,
+      "loss": 0.1107,
+      "step": 22371
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.730913868610576e-06,
+      "loss": 0.1388,
+      "step": 22372
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.7172810254082485e-06,
+      "loss": 0.1125,
+      "step": 22373
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.703667806667977e-06,
+      "loss": 0.0879,
+      "step": 22374
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.690074212658435e-06,
+      "loss": 0.1018,
+      "step": 22375
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.676500243647297e-06,
+      "loss": 0.0862,
+      "step": 22376
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.662945899902571e-06,
+      "loss": 0.0958,
+      "step": 22377
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.649411181691266e-06,
+      "loss": 0.0867,
+      "step": 22378
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.6358960892803896e-06,
+      "loss": 0.0697,
+      "step": 22379
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.622400622936285e-06,
+      "loss": 0.058,
+      "step": 22380
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.608924782925183e-06,
+      "loss": 0.1072,
+      "step": 22381
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.595468569512762e-06,
+      "loss": 0.085,
+      "step": 22382
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.582031982964252e-06,
+      "loss": 0.0709,
+      "step": 22383
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.568615023544776e-06,
+      "loss": 0.0667,
+      "step": 22384
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.555217691518787e-06,
+      "loss": 0.1339,
+      "step": 22385
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.541839987150298e-06,
+      "loss": 0.11,
+      "step": 22386
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.528481910703319e-06,
+      "loss": 0.0818,
+      "step": 22387
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.515143462441195e-06,
+      "loss": 0.0778,
+      "step": 22388
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.501824642626717e-06,
+      "loss": 0.0795,
+      "step": 22389
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.488525451522674e-06,
+      "loss": 0.0774,
+      "step": 22390
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4752458893912994e-06,
+      "loss": 0.0621,
+      "step": 22391
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.461985956494496e-06,
+      "loss": 0.0903,
+      "step": 22392
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4487456530935e-06,
+      "loss": 0.0673,
+      "step": 22393
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.435524979449545e-06,
+      "loss": 0.0818,
+      "step": 22394
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.422323935823314e-06,
+      "loss": 0.1021,
+      "step": 22395
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.409142522474929e-06,
+      "loss": 0.0758,
+      "step": 22396
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.395980739664518e-06,
+      "loss": 0.1056,
+      "step": 22397
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.382838587651428e-06,
+      "loss": 0.088,
+      "step": 22398
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.369716066694895e-06,
+      "loss": 0.0923,
+      "step": 22399
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.356613177053603e-06,
+      "loss": 0.082,
+      "step": 22400
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.343529918985789e-06,
+      "loss": 0.0681,
+      "step": 22401
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.330466292749691e-06,
+      "loss": 0.107,
+      "step": 22402
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.317422298602769e-06,
+      "loss": 0.1119,
+      "step": 22403
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.304397936802151e-06,
+      "loss": 0.0912,
+      "step": 22404
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.291393207604743e-06,
+      "loss": 0.0859,
+      "step": 22405
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.278408111267007e-06,
+      "loss": 0.0854,
+      "step": 22406
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.265442648044737e-06,
+      "loss": 0.0875,
+      "step": 22407
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2524968181938406e-06,
+      "loss": 0.2061,
+      "step": 22408
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.239570621969446e-06,
+      "loss": 0.0999,
+      "step": 22409
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.22666405962635e-06,
+      "loss": 0.0819,
+      "step": 22410
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.2137771314192385e-06,
+      "loss": 0.0712,
+      "step": 22411
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.200909837602018e-06,
+      "loss": 0.0857,
+      "step": 22412
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.188062178428597e-06,
+      "loss": 0.0871,
+      "step": 22413
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.175234154152108e-06,
+      "loss": 0.1039,
+      "step": 22414
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.162425765025679e-06,
+      "loss": 0.1406,
+      "step": 22415
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.1496370113015545e-06,
+      "loss": 0.0743,
+      "step": 22416
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.13686789323231e-06,
+      "loss": 0.0889,
+      "step": 22417
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.124118411069411e-06,
+      "loss": 0.0734,
+      "step": 22418
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.111388565064322e-06,
+      "loss": 0.0892,
+      "step": 22419
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.098678355468066e-06,
+      "loss": 0.0519,
+      "step": 22420
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0859877825312195e-06,
+      "loss": 0.0928,
+      "step": 22421
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.073316846504027e-06,
+      "loss": 0.1086,
+      "step": 22422
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0606655476364e-06,
+      "loss": 0.0784,
+      "step": 22423
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.0480338861775825e-06,
+      "loss": 0.0746,
+      "step": 22424
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.035421862376932e-06,
+      "loss": 0.0679,
+      "step": 22425
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.022829476483025e-06,
+      "loss": 0.1096,
+      "step": 22426
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.010256728743999e-06,
+      "loss": 0.1212,
+      "step": 22427
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.997703619407877e-06,
+      "loss": 0.0769,
+      "step": 22428
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.985170148722128e-06,
+      "loss": 0.0703,
+      "step": 22429
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.972656316933998e-06,
+      "loss": 0.1481,
+      "step": 22430
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.960162124290067e-06,
+      "loss": 0.1034,
+      "step": 22431
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.947687571036806e-06,
+      "loss": 0.0619,
+      "step": 22432
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.9352326574200185e-06,
+      "loss": 0.0659,
+      "step": 22433
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.922797383685506e-06,
+      "loss": 0.0719,
+      "step": 22434
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.910381750078406e-06,
+      "loss": 0.1011,
+      "step": 22435
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.897985756843414e-06,
+      "loss": 0.1299,
+      "step": 22436
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.8856094042249986e-06,
+      "loss": 0.1002,
+      "step": 22437
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.873252692467299e-06,
+      "loss": 0.0778,
+      "step": 22438
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.860915621813787e-06,
+      "loss": 0.0548,
+      "step": 22439
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.848598192507824e-06,
+      "loss": 0.0447,
+      "step": 22440
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.836300404792215e-06,
+      "loss": 0.0719,
+      "step": 22441
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.824022258909432e-06,
+      "loss": 0.0998,
+      "step": 22442
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.811763755101616e-06,
+      "loss": 0.0706,
+      "step": 22443
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.799524893610573e-06,
+      "loss": 0.07,
+      "step": 22444
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7873056746774436e-06,
+      "loss": 0.033,
+      "step": 22445
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7751060985431463e-06,
+      "loss": 0.0748,
+      "step": 22446
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7629261654484882e-06,
+      "loss": 0.0957,
+      "step": 22447
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7507658756333885e-06,
+      "loss": 0.16,
+      "step": 22448
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7386252293376553e-06,
+      "loss": 0.0813,
+      "step": 22449
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7265042268006533e-06,
+      "loss": 0.0896,
+      "step": 22450
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.7144028682615237e-06,
+      "loss": 0.0743,
+      "step": 22451
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.702321153958743e-06,
+      "loss": 0.0942,
+      "step": 22452
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.690259084130565e-06,
+      "loss": 0.0822,
+      "step": 22453
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.678216659014799e-06,
+      "loss": 0.0979,
+      "step": 22454
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6661938788489224e-06,
+      "loss": 0.0922,
+      "step": 22455
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6541907438699673e-06,
+      "loss": 0.0615,
+      "step": 22456
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.6422072543146332e-06,
+      "loss": 0.1588,
+      "step": 22457
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.630243410419176e-06,
+      "loss": 0.0906,
+      "step": 22458
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.618299212419518e-06,
+      "loss": 0.0448,
+      "step": 22459
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.606374660551248e-06,
+      "loss": 0.0554,
+      "step": 22460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.59446975504929e-06,
+      "loss": 0.1723,
+      "step": 22461
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5825844961485665e-06,
+      "loss": 0.0796,
+      "step": 22462
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.570718884083224e-06,
+      "loss": 0.0977,
+      "step": 22463
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.558872919087519e-06,
+      "loss": 0.0604,
+      "step": 22464
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.54704660139471e-06,
+      "loss": 0.0868,
+      "step": 22465
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5352399312381655e-06,
+      "loss": 0.0701,
+      "step": 22466
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5234529088506994e-06,
+      "loss": 0.1342,
+      "step": 22467
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.5116855344645705e-06,
+      "loss": 0.0746,
+      "step": 22468
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.499937808311926e-06,
+      "loss": 0.0978,
+      "step": 22469
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.488209730624359e-06,
+      "loss": 0.0858,
+      "step": 22470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4765013016332393e-06,
+      "loss": 0.0648,
+      "step": 22471
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4648125215691607e-06,
+      "loss": 0.1102,
+      "step": 22472
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4531433906628275e-06,
+      "loss": 0.0944,
+      "step": 22473
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4414939091442775e-06,
+      "loss": 0.0612,
+      "step": 22474
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4298640772431057e-06,
+      "loss": 0.0595,
+      "step": 22475
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.4182538951887944e-06,
+      "loss": 0.0521,
+      "step": 22476
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.406663363210161e-06,
+      "loss": 0.071,
+      "step": 22477
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3950924815356887e-06,
+      "loss": 0.0674,
+      "step": 22478
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.38354125039364e-06,
+      "loss": 0.0842,
+      "step": 22479
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.372009670011722e-06,
+      "loss": 0.0895,
+      "step": 22480
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3604977406173076e-06,
+      "loss": 0.1157,
+      "step": 22481
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3490054624373267e-06,
+      "loss": 0.0684,
+      "step": 22482
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3375328356984867e-06,
+      "loss": 0.0806,
+      "step": 22483
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3260798606269403e-06,
+      "loss": 0.0726,
+      "step": 22484
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.314646537448618e-06,
+      "loss": 0.0852,
+      "step": 22485
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3032328663886725e-06,
+      "loss": 0.1187,
+      "step": 22486
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.2918388476724793e-06,
+      "loss": 0.0703,
+      "step": 22487
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.2804644815245256e-06,
+      "loss": 0.1521,
+      "step": 22488
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2691097681691872e-06,
+      "loss": 0.1013,
+      "step": 22489
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2577747078301747e-06,
+      "loss": 0.0815,
+      "step": 22490
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.246459300731086e-06,
+      "loss": 0.0984,
+      "step": 22491
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.235163547095188e-06,
+      "loss": 0.1124,
+      "step": 22492
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2238874471449685e-06,
+      "loss": 0.0723,
+      "step": 22493
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2126310011028057e-06,
+      "loss": 0.0559,
+      "step": 22494
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.2013942091907442e-06,
+      "loss": 0.1018,
+      "step": 22495
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1901770716302737e-06,
+      "loss": 0.1034,
+      "step": 22496
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.178979588642661e-06,
+      "loss": 0.0555,
+      "step": 22497
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1678017604485078e-06,
+      "loss": 0.0955,
+      "step": 22498
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1566435872684152e-06,
+      "loss": 0.0706,
+      "step": 22499
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1455050693223188e-06,
+      "loss": 0.1191,
+      "step": 22500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.13438620682982e-06,
+      "loss": 0.0765,
+      "step": 22501
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1232870000101886e-06,
+      "loss": 0.0757,
+      "step": 22502
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.1122074490821383e-06,
+      "loss": 0.0767,
+      "step": 22503
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.101147554264272e-06,
+      "loss": 0.0513,
+      "step": 22504
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0901073157747482e-06,
+      "loss": 0.0594,
+      "step": 22505
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0790867338310604e-06,
+      "loss": 0.0652,
+      "step": 22506
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.068085808650478e-06,
+      "loss": 0.0859,
+      "step": 22507
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.057104540450162e-06,
+      "loss": 0.0716,
+      "step": 22508
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0461429294464936e-06,
+      "loss": 0.0399,
+      "step": 22509
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.035200975855523e-06,
+      "loss": 0.0872,
+      "step": 22510
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0242786798930775e-06,
+      "loss": 0.0805,
+      "step": 22511
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0133760417745405e-06,
+      "loss": 0.0479,
+      "step": 22512
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.0024930617148507e-06,
+      "loss": 0.0582,
+      "step": 22513
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9916297399287253e-06,
+      "loss": 0.1348,
+      "step": 22514
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9807860766301042e-06,
+      "loss": 0.0924,
+      "step": 22515
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.969962072032928e-06,
+      "loss": 0.0463,
+      "step": 22516
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.959157726350692e-06,
+      "loss": 0.0764,
+      "step": 22517
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.948373039796337e-06,
+      "loss": 0.0931,
+      "step": 22518
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.937608012582471e-06,
+      "loss": 0.1053,
+      "step": 22519
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.92686264492148e-06,
+      "loss": 0.1078,
+      "step": 22520
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9161369370251933e-06,
+      "loss": 0.081,
+      "step": 22521
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.9054308891049985e-06,
+      "loss": 0.115,
+      "step": 22522
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8947445013721706e-06,
+      "loss": 0.1137,
+      "step": 22523
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.884077774037208e-06,
+      "loss": 0.1703,
+      "step": 22524
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.873430707310609e-06,
+      "loss": 0.0775,
+      "step": 22525
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.862803301402317e-06,
+      "loss": 0.1125,
+      "step": 22526
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8521955565216084e-06,
+      "loss": 0.1217,
+      "step": 22527
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8416074728779827e-06,
+      "loss": 0.0963,
+      "step": 22528
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.83103905067994e-06,
+      "loss": 0.0668,
+      "step": 22529
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8204902901360907e-06,
+      "loss": 0.0533,
+      "step": 22530
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.809961191454269e-06,
+      "loss": 0.0704,
+      "step": 22531
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.799451754842086e-06,
+      "loss": 0.0576,
+      "step": 22532
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7889619805069323e-06,
+      "loss": 0.0789,
+      "step": 22533
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7784918686554196e-06,
+      "loss": 0.0558,
+      "step": 22534
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7680414194940493e-06,
+      "loss": 0.0927,
+      "step": 22535
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7576106332289907e-06,
+      "loss": 0.0662,
+      "step": 22536
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7471995100658562e-06,
+      "loss": 0.0836,
+      "step": 22537
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.736808050209816e-06,
+      "loss": 0.0569,
+      "step": 22538
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7264362538659272e-06,
+      "loss": 0.1147,
+      "step": 22539
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7160841212386935e-06,
+      "loss": 0.0736,
+      "step": 22540
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.7057516525320624e-06,
+      "loss": 0.1093,
+      "step": 22541
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6954388479498717e-06,
+      "loss": 0.1215,
+      "step": 22542
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6851457076954024e-06,
+      "loss": 0.1151,
+      "step": 22543
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6748722319718256e-06,
+      "loss": 0.098,
+      "step": 22544
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.664618420981424e-06,
+      "loss": 0.0652,
+      "step": 22545
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6543842749265912e-06,
+      "loss": 0.0873,
+      "step": 22546
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6441697940089438e-06,
+      "loss": 0.0435,
+      "step": 22547
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.633974978429987e-06,
+      "loss": 0.1174,
+      "step": 22548
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6237998283906715e-06,
+      "loss": 0.0495,
+      "step": 22549
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6136443440918368e-06,
+      "loss": 0.0654,
+      "step": 22550
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6035085257334336e-06,
+      "loss": 0.0768,
+      "step": 22551
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.593392373515413e-06,
+      "loss": 0.0984,
+      "step": 22552
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5832958876372825e-06,
+      "loss": 0.1313,
+      "step": 22553
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.573219068298216e-06,
+      "loss": 0.0779,
+      "step": 22554
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5631619156967213e-06,
+      "loss": 0.0768,
+      "step": 22555
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.553124430031084e-06,
+      "loss": 0.0681,
+      "step": 22556
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5431066114994794e-06,
+      "loss": 0.1219,
+      "step": 22557
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.533108460299083e-06,
+      "loss": 0.069,
+      "step": 22558
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.523129976627403e-06,
+      "loss": 0.0634,
+      "step": 22559
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5131711606808382e-06,
+      "loss": 0.08,
+      "step": 22560
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5032320126558984e-06,
+      "loss": 0.0776,
+      "step": 22561
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4933125327486484e-06,
+      "loss": 0.0639,
+      "step": 22562
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.483412721154599e-06,
+      "loss": 0.0769,
+      "step": 22563
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.473532578068927e-06,
+      "loss": 0.0552,
+      "step": 22564
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.463672103686476e-06,
+      "loss": 0.098,
+      "step": 22565
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4538312982016476e-06,
+      "loss": 0.0884,
+      "step": 22566
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4440101618085075e-06,
+      "loss": 0.084,
+      "step": 22567
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.434208694700568e-06,
+      "loss": 0.0797,
+      "step": 22568
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4244268970713414e-06,
+      "loss": 0.1226,
+      "step": 22569
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4146647691134503e-06,
+      "loss": 0.0958,
+      "step": 22570
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.4049223110196306e-06,
+      "loss": 0.0757,
+      "step": 22571
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3951995229817283e-06,
+      "loss": 0.0709,
+      "step": 22572
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3854964051915894e-06,
+      "loss": 0.0762,
+      "step": 22573
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.375812957840506e-06,
+      "loss": 0.082,
+      "step": 22574
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.366149181119437e-06,
+      "loss": 0.0565,
+      "step": 22575
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3565050752188954e-06,
+      "loss": 0.075,
+      "step": 22576
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3468806403290634e-06,
+      "loss": 0.1201,
+      "step": 22577
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3372758766396774e-06,
+      "loss": 0.0812,
+      "step": 22578
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.327690784340142e-06,
+      "loss": 0.0468,
+      "step": 22579
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.3181253636195276e-06,
+      "loss": 0.0677,
+      "step": 22580
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.30857961466624e-06,
+      "loss": 0.1003,
+      "step": 22581
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2990535376686826e-06,
+      "loss": 0.0668,
+      "step": 22582
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2895471328144846e-06,
+      "loss": 0.0941,
+      "step": 22583
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.280060400291384e-06,
+      "loss": 0.0542,
+      "step": 22584
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.270593340286231e-06,
+      "loss": 0.0835,
+      "step": 22585
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.261145952985766e-06,
+      "loss": 0.0551,
+      "step": 22586
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2517182385761724e-06,
+      "loss": 0.1266,
+      "step": 22587
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2423101972434133e-06,
+      "loss": 0.125,
+      "step": 22588
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.232921829173007e-06,
+      "loss": 0.1097,
+      "step": 22589
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.223553134550027e-06,
+      "loss": 0.0685,
+      "step": 22590
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.2142041135593262e-06,
+      "loss": 0.0693,
+      "step": 22591
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.204874766384979e-06,
+      "loss": 0.0922,
+      "step": 22592
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1955650932111713e-06,
+      "loss": 0.074,
+      "step": 22593
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.186275094221424e-06,
+      "loss": 0.0981,
+      "step": 22594
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1770047695988115e-06,
+      "loss": 0.1151,
+      "step": 22595
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1677541195261887e-06,
+      "loss": 0.0917,
+      "step": 22596
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1585231441859642e-06,
+      "loss": 0.0876,
+      "step": 22597
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1493118437601043e-06,
+      "loss": 0.1057,
+      "step": 22598
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1401202184303524e-06,
+      "loss": 0.0861,
+      "step": 22599
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.130948268377786e-06,
+      "loss": 0.0879,
+      "step": 22600
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1217959937833707e-06,
+      "loss": 0.0885,
+      "step": 22601
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1126633948275187e-06,
+      "loss": 0.0761,
+      "step": 22602
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1035504716903074e-06,
+      "loss": 0.0986,
+      "step": 22603
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0944572245513717e-06,
+      "loss": 0.062,
+      "step": 22604
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0853836535901226e-06,
+      "loss": 0.1119,
+      "step": 22605
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.076329758985307e-06,
+      "loss": 0.0871,
+      "step": 22606
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.067295540915559e-06,
+      "loss": 0.0891,
+      "step": 22607
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.05828099955907e-06,
+      "loss": 0.0977,
+      "step": 22608
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0492861350934754e-06,
+      "loss": 0.1327,
+      "step": 22609
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.040310947696078e-06,
+      "loss": 0.0655,
+      "step": 22610
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.031355437543958e-06,
+      "loss": 0.0699,
+      "step": 22611
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.022419604813752e-06,
+      "loss": 0.0667,
+      "step": 22612
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.013503449681431e-06,
+      "loss": 0.0944,
+      "step": 22613
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0046069723229645e-06,
+      "loss": 0.0721,
+      "step": 22614
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9957301729137677e-06,
+      "loss": 0.1049,
+      "step": 22615
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9868730516288124e-06,
+      "loss": 0.1219,
+      "step": 22616
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9780356086428476e-06,
+      "loss": 0.0616,
+      "step": 22617
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9692178441299557e-06,
+      "loss": 0.1066,
+      "step": 22618
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9604197582641093e-06,
+      "loss": 0.0983,
+      "step": 22619
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.951641351218836e-06,
+      "loss": 0.1749,
+      "step": 22620
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.942882623166997e-06,
+      "loss": 0.1125,
+      "step": 22621
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.934143574281566e-06,
+      "loss": 0.0677,
+      "step": 22622
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.925424204734627e-06,
+      "loss": 0.136,
+      "step": 22623
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9167245146982647e-06,
+      "loss": 0.0619,
+      "step": 22624
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9080445043440087e-06,
+      "loss": 0.0776,
+      "step": 22625
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.899384173842833e-06,
+      "loss": 0.0978,
+      "step": 22626
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8907435233657122e-06,
+      "loss": 0.0551,
+      "step": 22627
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8821225530829545e-06,
+      "loss": 0.0593,
+      "step": 22628
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8735212631644238e-06,
+      "loss": 0.1079,
+      "step": 22629
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8649396537798735e-06,
+      "loss": 0.0687,
+      "step": 22630
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8563777250985014e-06,
+      "loss": 0.0472,
+      "step": 22631
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8478354772889504e-06,
+      "loss": 0.1117,
+      "step": 22632
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8393129105198637e-06,
+      "loss": 0.0679,
+      "step": 22633
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8308100249592174e-06,
+      "loss": 0.0726,
+      "step": 22634
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8223268207745447e-06,
+      "loss": 0.087,
+      "step": 22635
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.813863298133267e-06,
+      "loss": 0.0804,
+      "step": 22636
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.8054194572021398e-06,
+      "loss": 0.0549,
+      "step": 22637
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7969952981478077e-06,
+      "loss": 0.0706,
+      "step": 22638
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.78859082113636e-06,
+      "loss": 0.0967,
+      "step": 22639
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.780206026333331e-06,
+      "loss": 0.0726,
+      "step": 22640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7718409139041435e-06,
+      "loss": 0.0609,
+      "step": 22641
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7634954840137773e-06,
+      "loss": 0.0837,
+      "step": 22642
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7551697368267671e-06,
+      "loss": 0.0524,
+      "step": 22643
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.746863672507315e-06,
+      "loss": 0.0834,
+      "step": 22644
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7385772912190678e-06,
+      "loss": 0.0637,
+      "step": 22645
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7303105931255614e-06,
+      "loss": 0.1078,
+      "step": 22646
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7220635783896654e-06,
+      "loss": 0.1211,
+      "step": 22647
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7138362471741386e-06,
+      "loss": 0.0989,
+      "step": 22648
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7056285996409626e-06,
+      "loss": 0.1144,
+      "step": 22649
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.69744063595223e-06,
+      "loss": 0.0512,
+      "step": 22650
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6892723562692558e-06,
+      "loss": 0.144,
+      "step": 22651
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6811237607531337e-06,
+      "loss": 0.0977,
+      "step": 22652
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6729948495645131e-06,
+      "loss": 0.1112,
+      "step": 22653
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6648856228637098e-06,
+      "loss": 0.1099,
+      "step": 22654
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6567960808105965e-06,
+      "loss": 0.0629,
+      "step": 22655
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6487262235646005e-06,
+      "loss": 0.1161,
+      "step": 22656
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6406760512849284e-06,
+      "loss": 0.0573,
+      "step": 22657
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6326455641303418e-06,
+      "loss": 0.0891,
+      "step": 22658
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6246347622590474e-06,
+      "loss": 0.0485,
+      "step": 22659
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6166436458292522e-06,
+      "loss": 0.1091,
+      "step": 22660
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6086722149981637e-06,
+      "loss": 0.0445,
+      "step": 22661
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6007204699232115e-06,
+      "loss": 0.093,
+      "step": 22662
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.592788410761159e-06,
+      "loss": 0.0984,
+      "step": 22663
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5848760376683258e-06,
+      "loss": 0.1702,
+      "step": 22664
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5769833508008091e-06,
+      "loss": 0.0637,
+      "step": 22665
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5691103503141512e-06,
+      "loss": 0.131,
+      "step": 22666
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5612570363635614e-06,
+      "loss": 0.0847,
+      "step": 22667
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5534234091040268e-06,
+      "loss": 0.1589,
+      "step": 22668
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5456094686899792e-06,
+      "loss": 0.0709,
+      "step": 22669
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5378152152752956e-06,
+      "loss": 0.0815,
+      "step": 22670
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.530040649013964e-06,
+      "loss": 0.0787,
+      "step": 22671
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5222857700590842e-06,
+      "loss": 0.0518,
+      "step": 22672
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.514550578563534e-06,
+      "loss": 0.064,
+      "step": 22673
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5068350746798575e-06,
+      "loss": 0.1869,
+      "step": 22674
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.499139258560378e-06,
+      "loss": 0.0861,
+      "step": 22675
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4914631303565296e-06,
+      "loss": 0.0906,
+      "step": 22676
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4838066902198578e-06,
+      "loss": 0.0839,
+      "step": 22677
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4761699383013527e-06,
+      "loss": 0.0825,
+      "step": 22678
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4685528747514499e-06,
+      "loss": 0.0753,
+      "step": 22679
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4609554997204732e-06,
+      "loss": 0.065,
+      "step": 22680
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.453377813358192e-06,
+      "loss": 0.0892,
+      "step": 22681
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4458198158139313e-06,
+      "loss": 0.0485,
+      "step": 22682
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4382815072367938e-06,
+      "loss": 0.157,
+      "step": 22683
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4307628877754386e-06,
+      "loss": 0.0575,
+      "step": 22684
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4232639575779693e-06,
+      "loss": 0.1077,
+      "step": 22685
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4157847167924898e-06,
+      "loss": 0.0843,
+      "step": 22686
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4083251655663265e-06,
+      "loss": 0.0764,
+      "step": 22687
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.4008853040464732e-06,
+      "loss": 0.1067,
+      "step": 22688
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.393465132379701e-06,
+      "loss": 0.076,
+      "step": 22689
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3860646507124486e-06,
+      "loss": 0.0738,
+      "step": 22690
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3786838591904882e-06,
+      "loss": 0.1022,
+      "step": 22691
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.37132275795937e-06,
+      "loss": 0.0977,
+      "step": 22692
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3639813471643115e-06,
+      "loss": 0.0662,
+      "step": 22693
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3566596269499743e-06,
+      "loss": 0.0742,
+      "step": 22694
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3493575974607986e-06,
+      "loss": 0.0991,
+      "step": 22695
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3420752588406693e-06,
+      "loss": 0.0923,
+      "step": 22696
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3348126112332492e-06,
+      "loss": 0.0996,
+      "step": 22697
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3275696547817572e-06,
+      "loss": 0.0613,
+      "step": 22698
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.320346389628857e-06,
+      "loss": 0.1001,
+      "step": 22699
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.3131428159172121e-06,
+      "loss": 0.0574,
+      "step": 22700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.305958933788598e-06,
+      "loss": 0.1182,
+      "step": 22701
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2987947433849011e-06,
+      "loss": 0.0439,
+      "step": 22702
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2916502448472311e-06,
+      "loss": 0.0939,
+      "step": 22703
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.284525438316586e-06,
+      "loss": 0.0945,
+      "step": 22704
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2774203239332981e-06,
+      "loss": 0.075,
+      "step": 22705
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2703349018374778e-06,
+      "loss": 0.05,
+      "step": 22706
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2632691721690125e-06,
+      "loss": 0.1057,
+      "step": 22707
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.256223135067014e-06,
+      "loss": 0.0826,
+      "step": 22708
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2491967906704815e-06,
+      "loss": 0.0949,
+      "step": 22709
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2421901391179712e-06,
+      "loss": 0.0527,
+      "step": 22710
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.235203180547706e-06,
+      "loss": 0.082,
+      "step": 22711
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2282359150973532e-06,
+      "loss": 0.0826,
+      "step": 22712
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2212883429042476e-06,
+      "loss": 0.1151,
+      "step": 22713
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2143604641055017e-06,
+      "loss": 0.1002,
+      "step": 22714
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.207452278837673e-06,
+      "loss": 0.0735,
+      "step": 22715
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2005637872368747e-06,
+      "loss": 0.0724,
+      "step": 22716
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1936949894391091e-06,
+      "loss": 0.1089,
+      "step": 22717
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1868458855796017e-06,
+      "loss": 0.0756,
+      "step": 22718
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1800164757936882e-06,
+      "loss": 0.0776,
+      "step": 22719
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1732067602157058e-06,
+      "loss": 0.0842,
+      "step": 22720
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1664167389801028e-06,
+      "loss": 0.0625,
+      "step": 22721
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1596464122207716e-06,
+      "loss": 0.092,
+      "step": 22722
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1528957800711615e-06,
+      "loss": 0.0696,
+      "step": 22723
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1461648426643877e-06,
+      "loss": 0.0424,
+      "step": 22724
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1394536001331223e-06,
+      "loss": 0.0706,
+      "step": 22725
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1327620526097038e-06,
+      "loss": 0.0438,
+      "step": 22726
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1260902002262486e-06,
+      "loss": 0.1471,
+      "step": 22727
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.119438043114096e-06,
+      "loss": 0.0618,
+      "step": 22728
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1128055814043636e-06,
+      "loss": 0.0986,
+      "step": 22729
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.1061928152280577e-06,
+      "loss": 0.1052,
+      "step": 22730
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0995997447154071e-06,
+      "loss": 0.1025,
+      "step": 22731
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0930263699964193e-06,
+      "loss": 0.0582,
+      "step": 22732
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0864726912008794e-06,
+      "loss": 0.1311,
+      "step": 22733
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.079938708457684e-06,
+      "loss": 0.0487,
+      "step": 22734
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0734244218959522e-06,
+      "loss": 0.0583,
+      "step": 22735
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0669298316440256e-06,
+      "loss": 0.0585,
+      "step": 22736
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0604549378299133e-06,
+      "loss": 0.0842,
+      "step": 22737
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0539997405814016e-06,
+      "loss": 0.0604,
+      "step": 22738
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0475642400256113e-06,
+      "loss": 0.0863,
+      "step": 22739
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0411484362896628e-06,
+      "loss": 0.0627,
+      "step": 22740
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0347523294997884e-06,
+      "loss": 0.0866,
+      "step": 22741
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0283759197822207e-06,
+      "loss": 0.0947,
+      "step": 22742
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0220192072627477e-06,
+      "loss": 0.1019,
+      "step": 22743
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0156821920667136e-06,
+      "loss": 0.0583,
+      "step": 22744
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0093648743190187e-06,
+      "loss": 0.0793,
+      "step": 22745
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0030672541441188e-06,
+      "loss": 0.0776,
+      "step": 22746
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.96789331666359e-07,
+      "loss": 0.1178,
+      "step": 22747
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.90531107009529e-07,
+      "loss": 0.0626,
+      "step": 22748
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.84292580296864e-07,
+      "loss": 0.0718,
+      "step": 22749
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.78073751651487e-07,
+      "loss": 0.0561,
+      "step": 22750
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.718746211959673e-07,
+      "loss": 0.0677,
+      "step": 22751
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.656951890525401e-07,
+      "loss": 0.058,
+      "step": 22752
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.59535455343108e-07,
+      "loss": 0.0895,
+      "step": 22753
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.533954201890182e-07,
+      "loss": 0.1508,
+      "step": 22754
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.472750837115074e-07,
+      "loss": 0.0955,
+      "step": 22755
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.411744460310345e-07,
+      "loss": 0.1119,
+      "step": 22756
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.350935072680588e-07,
+      "loss": 0.0988,
+      "step": 22757
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.290322675424844e-07,
+      "loss": 0.1282,
+      "step": 22758
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.229907269736604e-07,
+      "loss": 0.0629,
+      "step": 22759
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.169688856809355e-07,
+      "loss": 0.0839,
+      "step": 22760
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.109667437827707e-07,
+      "loss": 0.0815,
+      "step": 22761
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.049843013977377e-07,
+      "loss": 0.1464,
+      "step": 22762
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.990215586437422e-07,
+      "loss": 0.0734,
+      "step": 22763
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.930785156383569e-07,
+      "loss": 0.0941,
+      "step": 22764
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.871551724987103e-07,
+      "loss": 0.0973,
+      "step": 22765
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.812515293415979e-07,
+      "loss": 0.0913,
+      "step": 22766
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.75367586283593e-07,
+      "loss": 0.1167,
+      "step": 22767
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.69503343440492e-07,
+      "loss": 0.0508,
+      "step": 22768
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.636588009278689e-07,
+      "loss": 0.1261,
+      "step": 22769
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.578339588612983e-07,
+      "loss": 0.1046,
+      "step": 22770
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.520288173553548e-07,
+      "loss": 0.0754,
+      "step": 22771
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.462433765246136e-07,
+      "loss": 0.0643,
+      "step": 22772
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.404776364829836e-07,
+      "loss": 0.1119,
+      "step": 22773
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.347315973443736e-07,
+      "loss": 0.0827,
+      "step": 22774
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.290052592220265e-07,
+      "loss": 0.0889,
+      "step": 22775
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.232986222288519e-07,
+      "loss": 0.0971,
+      "step": 22776
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.176116864772043e-07,
+      "loss": 0.0844,
+      "step": 22777
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.119444520794384e-07,
+      "loss": 0.0916,
+      "step": 22778
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.062969191471315e-07,
+      "loss": 0.1034,
+      "step": 22779
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.0066908779175e-07,
+      "loss": 0.0563,
+      "step": 22780
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.950609581242052e-07,
+      "loss": 0.1028,
+      "step": 22781
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.894725302551864e-07,
+      "loss": 0.088,
+      "step": 22782
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.839038042946056e-07,
+      "loss": 0.1033,
+      "step": 22783
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.78354780352597e-07,
+      "loss": 0.1165,
+      "step": 22784
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.728254585384065e-07,
+      "loss": 0.1615,
+      "step": 22785
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.673158389609469e-07,
+      "loss": 0.0934,
+      "step": 22786
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.618259217291312e-07,
+      "loss": 0.079,
+      "step": 22787
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.563557069509841e-07,
+      "loss": 0.0726,
+      "step": 22788
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.509051947344193e-07,
+      "loss": 0.0737,
+      "step": 22789
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.454743851870172e-07,
+      "loss": 0.1244,
+      "step": 22790
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.400632784156924e-07,
+      "loss": 0.1096,
+      "step": 22791
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.346718745272484e-07,
+      "loss": 0.0764,
+      "step": 22792
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.293001736279336e-07,
+      "loss": 0.0717,
+      "step": 22793
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.239481758236632e-07,
+      "loss": 0.0894,
+      "step": 22794
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.186158812200195e-07,
+      "loss": 0.11,
+      "step": 22795
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.133032899221403e-07,
+      "loss": 0.1094,
+      "step": 22796
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.0801040203472e-07,
+      "loss": 0.07,
+      "step": 22797
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.027372176621194e-07,
+      "loss": 0.084,
+      "step": 22798
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.974837369083665e-07,
+      "loss": 0.0774,
+      "step": 22799
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.922499598770449e-07,
+      "loss": 0.0746,
+      "step": 22800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.870358866712945e-07,
+      "loss": 0.0934,
+      "step": 22801
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.818415173939219e-07,
+      "loss": 0.0697,
+      "step": 22802
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.766668521474006e-07,
+      "loss": 0.0908,
+      "step": 22803
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.715118910337603e-07,
+      "loss": 0.0773,
+      "step": 22804
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.663766341545863e-07,
+      "loss": 0.085,
+      "step": 22805
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.612610816112418e-07,
+      "loss": 0.0734,
+      "step": 22806
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.561652335045354e-07,
+      "loss": 0.0739,
+      "step": 22807
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.51089089934831e-07,
+      "loss": 0.0702,
+      "step": 22808
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.460326510023818e-07,
+      "loss": 0.1027,
+      "step": 22809
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.409959168068857e-07,
+      "loss": 0.0687,
+      "step": 22810
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.359788874474859e-07,
+      "loss": 0.0781,
+      "step": 22811
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.309815630233251e-07,
+      "loss": 0.0601,
+      "step": 22812
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.260039436327691e-07,
+      "loss": 0.0578,
+      "step": 22813
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.210460293739617e-07,
+      "loss": 0.0861,
+      "step": 22814
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.161078203447135e-07,
+      "loss": 0.0759,
+      "step": 22815
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.11189316642502e-07,
+      "loss": 0.1028,
+      "step": 22816
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.062905183641388e-07,
+      "loss": 0.0813,
+      "step": 22817
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.014114256063241e-07,
+      "loss": 0.0761,
+      "step": 22818
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.965520384652034e-07,
+      "loss": 0.092,
+      "step": 22819
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.91712357036589e-07,
+      "loss": 0.0892,
+      "step": 22820
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.868923814159599e-07,
+      "loss": 0.0817,
+      "step": 22821
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.820921116982403e-07,
+      "loss": 0.0532,
+      "step": 22822
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.773115479782431e-07,
+      "loss": 0.0637,
+      "step": 22823
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.725506903501154e-07,
+      "loss": 0.066,
+      "step": 22824
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.67809538907782e-07,
+      "loss": 0.1006,
+      "step": 22825
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.630880937447236e-07,
+      "loss": 0.1077,
+      "step": 22826
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.58386354953977e-07,
+      "loss": 0.0748,
+      "step": 22827
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.537043226282457e-07,
+      "loss": 0.1091,
+      "step": 22828
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.490419968599003e-07,
+      "loss": 0.1152,
+      "step": 22829
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.443993777409784e-07,
+      "loss": 0.0835,
+      "step": 22830
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.397764653627401e-07,
+      "loss": 0.1083,
+      "step": 22831
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.351732598165571e-07,
+      "loss": 0.1144,
+      "step": 22832
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.305897611932453e-07,
+      "loss": 0.1155,
+      "step": 22833
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.26025969582955e-07,
+      "loss": 0.0597,
+      "step": 22834
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.214818850757252e-07,
+      "loss": 0.0619,
+      "step": 22835
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.16957507761373e-07,
+      "loss": 0.0573,
+      "step": 22836
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.124528377288274e-07,
+      "loss": 0.1429,
+      "step": 22837
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.079678750670169e-07,
+      "loss": 0.1353,
+      "step": 22838
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.035026198644266e-07,
+      "loss": 0.1128,
+      "step": 22839
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.99057072209097e-07,
+      "loss": 0.0626,
+      "step": 22840
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.946312321886248e-07,
+      "loss": 0.0814,
+      "step": 22841
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.902250998902735e-07,
+      "loss": 0.0784,
+      "step": 22842
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.858386754009736e-07,
+      "loss": 0.1138,
+      "step": 22843
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.814719588072114e-07,
+      "loss": 0.0854,
+      "step": 22844
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.771249501950293e-07,
+      "loss": 0.1378,
+      "step": 22845
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.727976496502473e-07,
+      "loss": 0.1294,
+      "step": 22846
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.6849005725801977e-07,
+      "loss": 0.1273,
+      "step": 22847
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.642021731033896e-07,
+      "loss": 0.0897,
+      "step": 22848
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.5993399727084494e-07,
+      "loss": 0.0987,
+      "step": 22849
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.5568552984465163e-07,
+      "loss": 0.0831,
+      "step": 22850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.514567709085204e-07,
+      "loss": 0.0749,
+      "step": 22851
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.4724772054582917e-07,
+      "loss": 0.053,
+      "step": 22852
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.430583788396225e-07,
+      "loss": 0.1243,
+      "step": 22853
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.3888874587239e-07,
+      "loss": 0.1204,
+      "step": 22854
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.347388217263992e-07,
+      "loss": 0.1183,
+      "step": 22855
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.306086064833625e-07,
+      "loss": 0.0515,
+      "step": 22856
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.2649810022499234e-07,
+      "loss": 0.0906,
+      "step": 22857
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.2240730303211297e-07,
+      "loss": 0.2285,
+      "step": 22858
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.183362149855485e-07,
+      "loss": 0.093,
+      "step": 22859
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1428483616534617e-07,
+      "loss": 0.0441,
+      "step": 22860
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.10253166651664e-07,
+      "loss": 0.1333,
+      "step": 22861
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.062412065236609e-07,
+      "loss": 0.0736,
+      "step": 22862
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.022489558608289e-07,
+      "loss": 0.0953,
+      "step": 22863
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.982764147415496e-07,
+      "loss": 0.0894,
+      "step": 22864
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.9432358324431593e-07,
+      "loss": 0.0499,
+      "step": 22865
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.903904614470655e-07,
+      "loss": 0.1107,
+      "step": 22866
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.864770494274028e-07,
+      "loss": 0.108,
+      "step": 22867
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.825833472622664e-07,
+      "loss": 0.0707,
+      "step": 22868
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7870935502870576e-07,
+      "loss": 0.0609,
+      "step": 22869
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.748550728028821e-07,
+      "loss": 0.0862,
+      "step": 22870
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.710205006609568e-07,
+      "loss": 0.0593,
+      "step": 22871
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6720563867842503e-07,
+      "loss": 0.0999,
+      "step": 22872
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.634104869305599e-07,
+      "loss": 0.1481,
+      "step": 22873
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.596350454921904e-07,
+      "loss": 0.0963,
+      "step": 22874
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.558793144377015e-07,
+      "loss": 0.1123,
+      "step": 22875
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.521432938412561e-07,
+      "loss": 0.0927,
+      "step": 22876
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.48426983776462e-07,
+      "loss": 0.0533,
+      "step": 22877
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4473038431659387e-07,
+      "loss": 0.1574,
+      "step": 22878
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4105349553448237e-07,
+      "loss": 0.0765,
+      "step": 22879
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3739631750273615e-07,
+      "loss": 0.1699,
+      "step": 22880
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3375885029340856e-07,
+      "loss": 0.0529,
+      "step": 22881
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.301410939782201e-07,
+      "loss": 0.1086,
+      "step": 22882
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.265430486284471e-07,
+      "loss": 0.0745,
+      "step": 22883
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.229647143151437e-07,
+      "loss": 0.0769,
+      "step": 22884
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1940609110880925e-07,
+      "loss": 0.0923,
+      "step": 22885
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1586717907960973e-07,
+      "loss": 0.1313,
+      "step": 22886
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1234797829737817e-07,
+      "loss": 0.0725,
+      "step": 22887
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.0884848883150354e-07,
+      "loss": 0.0709,
+      "step": 22888
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.053687107508196e-07,
+      "loss": 0.1064,
+      "step": 22889
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.019086441241603e-07,
+      "loss": 0.0709,
+      "step": 22890
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.984682890195822e-07,
+      "loss": 0.0908,
+      "step": 22891
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9504764550492e-07,
+      "loss": 0.058,
+      "step": 22892
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.9164671364778627e-07,
+      "loss": 0.0814,
+      "step": 22893
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.882654935151274e-07,
+      "loss": 0.0981,
+      "step": 22894
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8490398517366787e-07,
+      "loss": 0.0785,
+      "step": 22895
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.8156218868957697e-07,
+      "loss": 0.1025,
+      "step": 22896
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.782401041289129e-07,
+      "loss": 0.0765,
+      "step": 22897
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.749377315569568e-07,
+      "loss": 0.0797,
+      "step": 22898
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.7165507103898977e-07,
+      "loss": 0.1152,
+      "step": 22899
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.683921226397379e-07,
+      "loss": 0.0709,
+      "step": 22900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.65148886423483e-07,
+      "loss": 0.111,
+      "step": 22901
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.6192536245417396e-07,
+      "loss": 0.0896,
+      "step": 22902
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5872155079531554e-07,
+      "loss": 0.0696,
+      "step": 22903
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.555374515103015e-07,
+      "loss": 0.0629,
+      "step": 22904
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5237306466163737e-07,
+      "loss": 0.1071,
+      "step": 22905
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4922839031182867e-07,
+      "loss": 0.0975,
+      "step": 22906
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4610342852293687e-07,
+      "loss": 0.0923,
+      "step": 22907
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4299817935657943e-07,
+      "loss": 0.1088,
+      "step": 22908
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3991264287381855e-07,
+      "loss": 0.0821,
+      "step": 22909
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.368468191357165e-07,
+      "loss": 0.0649,
+      "step": 22910
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.338007082025584e-07,
+      "loss": 0.0628,
+      "step": 22911
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3077431013451832e-07,
+      "loss": 0.0958,
+      "step": 22912
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2776762499121528e-07,
+      "loss": 0.1093,
+      "step": 22913
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2478065283193516e-07,
+      "loss": 0.0675,
+      "step": 22914
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2181339371551978e-07,
+      "loss": 0.0774,
+      "step": 22915
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.188658477004779e-07,
+      "loss": 0.1006,
+      "step": 22916
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1593801484509624e-07,
+      "loss": 0.0876,
+      "step": 22917
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1302989520699533e-07,
+      "loss": 0.0576,
+      "step": 22918
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1014148884346273e-07,
+      "loss": 0.0939,
+      "step": 22919
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0727279581145286e-07,
+      "loss": 0.1227,
+      "step": 22920
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.044238161675871e-07,
+      "loss": 0.0851,
+      "step": 22921
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.0159454996804273e-07,
+      "loss": 0.0512,
+      "step": 22922
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9878499726866395e-07,
+      "loss": 0.0722,
+      "step": 22923
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9599515812462887e-07,
+      "loss": 0.0906,
+      "step": 22924
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9322503259122658e-07,
+      "loss": 0.0778,
+      "step": 22925
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9047462072285804e-07,
+      "loss": 0.0829,
+      "step": 22926
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8774392257381312e-07,
+      "loss": 0.0729,
+      "step": 22927
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8503293819804866e-07,
+      "loss": 0.1013,
+      "step": 22928
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.823416676488554e-07,
+      "loss": 0.0668,
+      "step": 22929
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.79670110979413e-07,
+      "loss": 0.097,
+      "step": 22930
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7701826824234602e-07,
+      "loss": 0.1064,
+      "step": 22931
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7438613948994597e-07,
+      "loss": 0.0698,
+      "step": 22932
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7177372477406028e-07,
+      "loss": 0.0584,
+      "step": 22933
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6918102414631432e-07,
+      "loss": 0.1019,
+      "step": 22934
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6660803765777832e-07,
+      "loss": 0.0717,
+      "step": 22935
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.640547653591895e-07,
+      "loss": 0.2148,
+      "step": 22936
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6152120730084097e-07,
+      "loss": 0.0662,
+      "step": 22937
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5900736353269273e-07,
+      "loss": 0.0742,
+      "step": 22938
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.565132341043718e-07,
+      "loss": 0.08,
+      "step": 22939
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.54038819065061e-07,
+      "loss": 0.127,
+      "step": 22940
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5158411846349918e-07,
+      "loss": 0.0934,
+      "step": 22941
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.49149132348092e-07,
+      "loss": 0.0748,
+      "step": 22942
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.467338607669122e-07,
+      "loss": 0.1115,
+      "step": 22943
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4433830376747725e-07,
+      "loss": 0.11,
+      "step": 22944
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4196246139719372e-07,
+      "loss": 0.0637,
+      "step": 22945
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3960633370269094e-07,
+      "loss": 0.0618,
+      "step": 22946
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.372699207305983e-07,
+      "loss": 0.054,
+      "step": 22947
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3495322252687902e-07,
+      "loss": 0.1492,
+      "step": 22948
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.326562391371633e-07,
+      "loss": 0.0626,
+      "step": 22949
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.3037897060697023e-07,
+      "loss": 0.0587,
+      "step": 22950
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2812141698093082e-07,
+      "loss": 0.1091,
+      "step": 22951
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2588357830378704e-07,
+      "loss": 0.0745,
+      "step": 22952
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2366545461950375e-07,
+      "loss": 0.0909,
+      "step": 22953
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.214670459719347e-07,
+      "loss": 0.1407,
+      "step": 22954
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.192883524042676e-07,
+      "loss": 0.0835,
+      "step": 22955
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1712937395957912e-07,
+      "loss": 0.087,
+      "step": 22956
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.149901106805018e-07,
+      "loss": 0.0815,
+      "step": 22957
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.128705626091131e-07,
+      "loss": 0.0911,
+      "step": 22958
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1077072978715741e-07,
+      "loss": 0.1161,
+      "step": 22959
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0869061225615706e-07,
+      "loss": 0.0984,
+      "step": 22960
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0663021005707929e-07,
+      "loss": 0.0538,
+      "step": 22961
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0458952323055825e-07,
+      "loss": 0.1349,
+      "step": 22962
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0256855181678403e-07,
+      "loss": 0.1115,
+      "step": 22963
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0056729585572466e-07,
+      "loss": 0.061,
+      "step": 22964
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.858575538668202e-08,
+      "loss": 0.0848,
+      "step": 22965
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.662393044884698e-08,
+      "loss": 0.0732,
+      "step": 22966
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.46818210807443e-08,
+      "loss": 0.0654,
+      "step": 22967
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.27594273208987e-08,
+      "loss": 0.0957,
+      "step": 22968
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.085674920705778e-08,
+      "loss": 0.1285,
+      "step": 22969
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.897378677674706e-08,
+      "loss": 0.0654,
+      "step": 22970
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.711054006715902e-08,
+      "loss": 0.0787,
+      "step": 22971
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.526700911504204e-08,
+      "loss": 0.0834,
+      "step": 22972
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.344319395670041e-08,
+      "loss": 0.0764,
+      "step": 22973
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.163909462799435e-08,
+      "loss": 0.0688,
+      "step": 22974
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.985471116467302e-08,
+      "loss": 0.1066,
+      "step": 22975
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.809004360181948e-08,
+      "loss": 0.1249,
+      "step": 22976
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.634509197418371e-08,
+      "loss": 0.0985,
+      "step": 22977
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.461985631640467e-08,
+      "loss": 0.07,
+      "step": 22978
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.291433666223312e-08,
+      "loss": 0.0554,
+      "step": 22979
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.122853304541987e-08,
+      "loss": 0.113,
+      "step": 22980
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.956244549916058e-08,
+      "loss": 0.0507,
+      "step": 22981
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.791607405631783e-08,
+      "loss": 0.1117,
+      "step": 22982
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.628941874942119e-08,
+      "loss": 0.1401,
+      "step": 22983
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.468247961044504e-08,
+      "loss": 0.0803,
+      "step": 22984
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.30952566711418e-08,
+      "loss": 0.0919,
+      "step": 22985
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.152774996270871e-08,
+      "loss": 0.1139,
+      "step": 22986
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.997995951623203e-08,
+      "loss": 0.0654,
+      "step": 22987
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.845188536213186e-08,
+      "loss": 0.0889,
+      "step": 22988
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.6943527530495254e-08,
+      "loss": 0.0482,
+      "step": 22989
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.545488605107618e-08,
+      "loss": 0.1005,
+      "step": 22990
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.398596095329555e-08,
+      "loss": 0.0929,
+      "step": 22991
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.253675226601917e-08,
+      "loss": 0.0997,
+      "step": 22992
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.110726001789079e-08,
+      "loss": 0.163,
+      "step": 22993
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.9697484237110066e-08,
+      "loss": 0.0558,
+      "step": 22994
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.830742495143259e-08,
+      "loss": 0.0719,
+      "step": 22995
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.693708218828086e-08,
+      "loss": 0.1169,
+      "step": 22996
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.558645597474431e-08,
+      "loss": 0.0896,
+      "step": 22997
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.4255546337246264e-08,
+      "loss": 0.0473,
+      "step": 22998
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.294435330221003e-08,
+      "loss": 0.0928,
+      "step": 22999
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.165287689550379e-08,
+      "loss": 0.0825,
+      "step": 23000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.038111714255166e-08,
+      "loss": 0.0756,
+      "step": 23001
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.912907406833366e-08,
+      "loss": 0.0632,
+      "step": 23002
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.789674769760776e-08,
+      "loss": 0.0787,
+      "step": 23003
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.6684138054687846e-08,
+      "loss": 0.0748,
+      "step": 23004
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.549124516355473e-08,
+      "loss": 0.0909,
+      "step": 23005
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.43180690475231e-08,
+      "loss": 0.0792,
+      "step": 23006
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.316460973001867e-08,
+      "loss": 0.0875,
+      "step": 23007
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.203086723346793e-08,
+      "loss": 0.0333,
+      "step": 23008
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.091684158040842e-08,
+      "loss": 0.123,
+      "step": 23009
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.9822532792822546e-08,
+      "loss": 0.053,
+      "step": 23010
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.8747940892137615e-08,
+      "loss": 0.1028,
+      "step": 23011
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.7693065899780934e-08,
+      "loss": 0.1049,
+      "step": 23012
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.665790783629163e-08,
+      "loss": 0.0864,
+      "step": 23013
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5642466722208824e-08,
+      "loss": 0.0642,
+      "step": 23014
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.4646742577627557e-08,
+      "loss": 0.0836,
+      "step": 23015
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.3670735422087753e-08,
+      "loss": 0.088,
+      "step": 23016
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.271444527479627e-08,
+      "loss": 0.0532,
+      "step": 23017
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.1777872154626897e-08,
+      "loss": 0.0922,
+      "step": 23018
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0861016080120366e-08,
+      "loss": 0.0877,
+      "step": 23019
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.9963877069373305e-08,
+      "loss": 0.1249,
+      "step": 23020
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.9086455139927238e-08,
+      "loss": 0.0717,
+      "step": 23021
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.8228750309212673e-08,
+      "loss": 0.1113,
+      "step": 23022
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7390762594104992e-08,
+      "loss": 0.0845,
+      "step": 23023
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6572492011146523e-08,
+      "loss": 0.0806,
+      "step": 23024
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5773938576324476e-08,
+      "loss": 0.1244,
+      "step": 23025
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4995102305626063e-08,
+      "loss": 0.1029,
+      "step": 23026
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4235983214261339e-08,
+      "loss": 0.0865,
+      "step": 23027
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3496581317218314e-08,
+      "loss": 0.0648,
+      "step": 23028
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2776896629040913e-08,
+      "loss": 0.0544,
+      "step": 23029
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2076929163939988e-08,
+      "loss": 0.0723,
+      "step": 23030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1396678935793326e-08,
+      "loss": 0.0636,
+      "step": 23031
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0736145958034627e-08,
+      "loss": 0.0945,
+      "step": 23032
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0095330243431455e-08,
+      "loss": 0.1454,
+      "step": 23033
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.47423180497342e-09,
+      "loss": 0.1151,
+      "step": 23034
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.872850654650933e-09,
+      "loss": 0.0855,
+      "step": 23035
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.291186804343376e-09,
+      "loss": 0.0964,
+      "step": 23036
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.729240265708093e-09,
+      "loss": 0.1631,
+      "step": 23037
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.187011049625269e-09,
+      "loss": 0.0959,
+      "step": 23038
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.664499166864068e-09,
+      "loss": 0.0732,
+      "step": 23039
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.161704627749565e-09,
+      "loss": 0.049,
+      "step": 23040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.67862744205172e-09,
+      "loss": 0.1016,
+      "step": 23041
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.215267619540498e-09,
+      "loss": 0.1057,
+      "step": 23042
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.771625169208704e-09,
+      "loss": 0.0925,
+      "step": 23043
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.347700099827101e-09,
+      "loss": 0.0572,
+      "step": 23044
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.943492419611339e-09,
+      "loss": 0.0977,
+      "step": 23045
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.5590021367770676e-09,
+      "loss": 0.072,
+      "step": 23046
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.194229258762782e-09,
+      "loss": 0.0515,
+      "step": 23047
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.8491737928959537e-09,
+      "loss": 0.0641,
+      "step": 23048
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.523835745726899e-09,
+      "loss": 0.1049,
+      "step": 23049
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.2182151239169558e-09,
+      "loss": 0.0782,
+      "step": 23050
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.9323119333503057e-09,
+      "loss": 0.1201,
+      "step": 23051
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.666126179689087e-09,
+      "loss": 0.0748,
+      "step": 23052
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4196578682623695e-09,
+      "loss": 0.1256,
+      "step": 23053
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.1929070038441125e-09,
+      "loss": 0.0787,
+      "step": 23054
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.85873590875208e-10,
+      "loss": 0.0637,
+      "step": 23055
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.985576335745037e-10,
+      "loss": 0.0462,
+      "step": 23056
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.309591354947131e-10,
+      "loss": 0.0854,
+      "step": 23057
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.830780999665052e-10,
+      "loss": 0.1108,
+      "step": 23058
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.549145299874823e-10,
+      "loss": 0.0828,
+      "step": 23059
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.4646842811115733e-10,
+      "loss": 0.0446,
+      "step": 23060
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5773979633593172e-10,
+      "loss": 0.1028,
+      "step": 23061
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.872863643816232e-11,
+      "loss": 0.2047,
+      "step": 23062
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.9434949861139046e-11,
+      "loss": 0.0753,
+      "step": 23063
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.858737493040337e-12,
+      "loss": 0.0328,
+      "step": 23064
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.0894,
+      "step": 23065
+    },
+    {
+      "epoch": 5.0,
+      "step": 23065,
+      "total_flos": 2.3572991246751236e+19,
+      "train_loss": 0.1152699925133031,
+      "train_runtime": 62306.6717,
+      "train_samples_per_second": 23.691,
+      "train_steps_per_second": 0.37
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 23065,
+  "num_train_epochs": 5,
+  "save_steps": 8000,
+  "total_flos": 2.3572991246751236e+19,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/README.md b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..dcce45792f17c6f4a217b759836daa4a81605ce7
--- /dev/null
+++ b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/README.md
@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.5.0
diff --git a/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/adapter_config.json b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..1c8bdb2518eaadf68e9b8214b69b4aaed406673f
--- /dev/null
+++ b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/adapter_config.json
@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "./vicuna-v1-3-7b",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 256,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 128,
+  "revision": null,
+  "target_modules": [
+    "up_proj",
+    "gate_proj",
+    "q_proj",
+    "v_proj",
+    "k_proj",
+    "o_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}
\ No newline at end of file
diff --git a/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/adapter_model.bin b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/adapter_model.bin
new file mode 100644
index 0000000000000000000000000000000000000000..200d9860f2cda34dcf9abdf441ec005f43ce4fbd
--- /dev/null
+++ b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/adapter_model.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:84efdc09bf8be031148e59d5fbdb4ad49463495be923e0333731d6f26a3cd0bf
+size 639786637
diff --git a/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/config.json b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..b4edd8c579787fbd3ad760a5c62f6041c9d8b928
--- /dev/null
+++ b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/config.json
@@ -0,0 +1,39 @@
+{
+  "_name_or_path": "./vicuna-v1-3-7b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "freeze_mm_mlp_adapter": false,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "mlp_bias": false,
+  "mm_graph_tower": "hvqvae2",
+  "mm_hidden_size": 308,
+  "mm_projector_type": "hlinear",
+  "mm_use_im_patch_token": false,
+  "mm_use_im_start_end": false,
+  "model_type": "llava_graph",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.41.0",
+  "tune_mm_mlp_adapter": false,
+  "use_cache": true,
+  "use_lap_pe": true,
+  "use_mm_proj": true,
+  "vocab_size": 32000
+}
diff --git a/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/non_lora_trainables.bin b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/non_lora_trainables.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b1140a36b86e1344d047ffb18f7d6442044882d1
--- /dev/null
+++ b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/non_lora_trainables.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1b1709a04ed78d9e9be71a9a58720eae5356875436ec2b7fd8db2ee13d59235
+size 11335231
diff --git a/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/trainer_state.json b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..46090fc5a90e102c67575ffe2dd3f1bcd2bb1968
--- /dev/null
+++ b/vicuna/MoleculeNet-llava-hvqvae2-vicuna-v1-3-7b-finetune_lora-large/trainer_state.json
@@ -0,0 +1,39242 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "eval_steps": 500,
+  "global_step": 5600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0035714285714285713,
+      "grad_norm": 37.74194890615875,
+      "learning_rate": 2.3809523809523811e-07,
+      "loss": 7.0055,
+      "step": 1
+    },
+    {
+      "epoch": 0.007142857142857143,
+      "grad_norm": 48.38335960660328,
+      "learning_rate": 4.7619047619047623e-07,
+      "loss": 7.0857,
+      "step": 2
+    },
+    {
+      "epoch": 0.010714285714285714,
+      "grad_norm": 38.960595475357955,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 6.7904,
+      "step": 3
+    },
+    {
+      "epoch": 0.014285714285714285,
+      "grad_norm": 39.838203222476544,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 7.0175,
+      "step": 4
+    },
+    {
+      "epoch": 0.017857142857142856,
+      "grad_norm": 43.08774032417851,
+      "learning_rate": 1.1904761904761906e-06,
+      "loss": 6.9778,
+      "step": 5
+    },
+    {
+      "epoch": 0.02142857142857143,
+      "grad_norm": 45.92431304666535,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 6.9022,
+      "step": 6
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 47.256583389061774,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 6.9437,
+      "step": 7
+    },
+    {
+      "epoch": 0.02857142857142857,
+      "grad_norm": 34.96669336855623,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 6.8819,
+      "step": 8
+    },
+    {
+      "epoch": 0.03214285714285714,
+      "grad_norm": 37.02770083557375,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 6.9417,
+      "step": 9
+    },
+    {
+      "epoch": 0.03571428571428571,
+      "grad_norm": 33.785072644117356,
+      "learning_rate": 2.380952380952381e-06,
+      "loss": 6.6221,
+      "step": 10
+    },
+    {
+      "epoch": 0.039285714285714285,
+      "grad_norm": 52.955065998670726,
+      "learning_rate": 2.6190476190476192e-06,
+      "loss": 6.4243,
+      "step": 11
+    },
+    {
+      "epoch": 0.04285714285714286,
+      "grad_norm": 44.23018209594778,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 6.0617,
+      "step": 12
+    },
+    {
+      "epoch": 0.04642857142857143,
+      "grad_norm": 35.333566119256425,
+      "learning_rate": 3.0952380952380957e-06,
+      "loss": 5.9115,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 36.55277519580499,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 5.6001,
+      "step": 14
+    },
+    {
+      "epoch": 0.05357142857142857,
+      "grad_norm": 35.47460129656955,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 5.1356,
+      "step": 15
+    },
+    {
+      "epoch": 0.05714285714285714,
+      "grad_norm": 44.01044794179343,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 4.7047,
+      "step": 16
+    },
+    {
+      "epoch": 0.060714285714285714,
+      "grad_norm": 44.44891621152886,
+      "learning_rate": 4.047619047619048e-06,
+      "loss": 4.2867,
+      "step": 17
+    },
+    {
+      "epoch": 0.06428571428571428,
+      "grad_norm": 53.15130780751647,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 4.0392,
+      "step": 18
+    },
+    {
+      "epoch": 0.06785714285714285,
+      "grad_norm": 42.515558811936735,
+      "learning_rate": 4.523809523809524e-06,
+      "loss": 3.5619,
+      "step": 19
+    },
+    {
+      "epoch": 0.07142857142857142,
+      "grad_norm": 34.44475973130888,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 2.8813,
+      "step": 20
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 29.61866117769178,
+      "learning_rate": 5e-06,
+      "loss": 2.4187,
+      "step": 21
+    },
+    {
+      "epoch": 0.07857142857142857,
+      "grad_norm": 29.109420145156427,
+      "learning_rate": 5.2380952380952384e-06,
+      "loss": 1.8695,
+      "step": 22
+    },
+    {
+      "epoch": 0.08214285714285714,
+      "grad_norm": 17.20076401851678,
+      "learning_rate": 5.476190476190477e-06,
+      "loss": 1.4794,
+      "step": 23
+    },
+    {
+      "epoch": 0.08571428571428572,
+      "grad_norm": 14.687159834128565,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.3102,
+      "step": 24
+    },
+    {
+      "epoch": 0.08928571428571429,
+      "grad_norm": 13.479977096638445,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 1.1169,
+      "step": 25
+    },
+    {
+      "epoch": 0.09285714285714286,
+      "grad_norm": 10.941035031507756,
+      "learning_rate": 6.1904761904761914e-06,
+      "loss": 0.861,
+      "step": 26
+    },
+    {
+      "epoch": 0.09642857142857143,
+      "grad_norm": 9.288212935047026,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.7398,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 8.911876360350698,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7292,
+      "step": 28
+    },
+    {
+      "epoch": 0.10357142857142858,
+      "grad_norm": 7.285406784116684,
+      "learning_rate": 6.9047619047619055e-06,
+      "loss": 0.4857,
+      "step": 29
+    },
+    {
+      "epoch": 0.10714285714285714,
+      "grad_norm": 7.461945373804952,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.5016,
+      "step": 30
+    },
+    {
+      "epoch": 0.11071428571428571,
+      "grad_norm": 7.079998516950622,
+      "learning_rate": 7.380952380952382e-06,
+      "loss": 0.4283,
+      "step": 31
+    },
+    {
+      "epoch": 0.11428571428571428,
+      "grad_norm": 5.297644151811423,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.4342,
+      "step": 32
+    },
+    {
+      "epoch": 0.11785714285714285,
+      "grad_norm": 4.79843580171955,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.3465,
+      "step": 33
+    },
+    {
+      "epoch": 0.12142857142857143,
+      "grad_norm": 6.121677091994911,
+      "learning_rate": 8.095238095238097e-06,
+      "loss": 0.4642,
+      "step": 34
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 3.128683351979678,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.2193,
+      "step": 35
+    },
+    {
+      "epoch": 0.12857142857142856,
+      "grad_norm": 4.1703660197435335,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.2849,
+      "step": 36
+    },
+    {
+      "epoch": 0.13214285714285715,
+      "grad_norm": 3.878628315979179,
+      "learning_rate": 8.80952380952381e-06,
+      "loss": 0.1966,
+      "step": 37
+    },
+    {
+      "epoch": 0.1357142857142857,
+      "grad_norm": 1.6912361635518387,
+      "learning_rate": 9.047619047619049e-06,
+      "loss": 0.1563,
+      "step": 38
+    },
+    {
+      "epoch": 0.1392857142857143,
+      "grad_norm": 5.940085516999055,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.3126,
+      "step": 39
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 3.218547773388503,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.2187,
+      "step": 40
+    },
+    {
+      "epoch": 0.14642857142857144,
+      "grad_norm": 2.6090151523867133,
+      "learning_rate": 9.761904761904762e-06,
+      "loss": 0.1919,
+      "step": 41
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 3.2958764686205515,
+      "learning_rate": 1e-05,
+      "loss": 0.1873,
+      "step": 42
+    },
+    {
+      "epoch": 0.15357142857142858,
+      "grad_norm": 1.962640059756362,
+      "learning_rate": 1.0238095238095238e-05,
+      "loss": 0.1386,
+      "step": 43
+    },
+    {
+      "epoch": 0.15714285714285714,
+      "grad_norm": 1.6713044448326093,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.1214,
+      "step": 44
+    },
+    {
+      "epoch": 0.16071428571428573,
+      "grad_norm": 1.7946393250584707,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.1846,
+      "step": 45
+    },
+    {
+      "epoch": 0.16428571428571428,
+      "grad_norm": 2.8756893081697648,
+      "learning_rate": 1.0952380952380955e-05,
+      "loss": 0.1393,
+      "step": 46
+    },
+    {
+      "epoch": 0.16785714285714284,
+      "grad_norm": 3.5367297096353334,
+      "learning_rate": 1.1190476190476192e-05,
+      "loss": 0.1535,
+      "step": 47
+    },
+    {
+      "epoch": 0.17142857142857143,
+      "grad_norm": 3.7327692225118803,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.1534,
+      "step": 48
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 3.2578823189463613,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.1355,
+      "step": 49
+    },
+    {
+      "epoch": 0.17857142857142858,
+      "grad_norm": 4.249319493552774,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.1581,
+      "step": 50
+    },
+    {
+      "epoch": 0.18214285714285713,
+      "grad_norm": 2.516946328902159,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.1353,
+      "step": 51
+    },
+    {
+      "epoch": 0.18571428571428572,
+      "grad_norm": 1.7028365953906799,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.1118,
+      "step": 52
+    },
+    {
+      "epoch": 0.18928571428571428,
+      "grad_norm": 2.003726440208407,
+      "learning_rate": 1.261904761904762e-05,
+      "loss": 0.1353,
+      "step": 53
+    },
+    {
+      "epoch": 0.19285714285714287,
+      "grad_norm": 1.3239388754521,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.1248,
+      "step": 54
+    },
+    {
+      "epoch": 0.19642857142857142,
+      "grad_norm": 2.4839055934902765,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.1466,
+      "step": 55
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.1292947795927484,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0973,
+      "step": 56
+    },
+    {
+      "epoch": 0.20357142857142857,
+      "grad_norm": 1.462908519792513,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.1497,
+      "step": 57
+    },
+    {
+      "epoch": 0.20714285714285716,
+      "grad_norm": 0.8954850906871515,
+      "learning_rate": 1.3809523809523811e-05,
+      "loss": 0.0894,
+      "step": 58
+    },
+    {
+      "epoch": 0.21071428571428572,
+      "grad_norm": 1.848675455353086,
+      "learning_rate": 1.4047619047619048e-05,
+      "loss": 0.0932,
+      "step": 59
+    },
+    {
+      "epoch": 0.21428571428571427,
+      "grad_norm": 1.8271275854399032,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.0926,
+      "step": 60
+    },
+    {
+      "epoch": 0.21785714285714286,
+      "grad_norm": 2.004331406425219,
+      "learning_rate": 1.4523809523809524e-05,
+      "loss": 0.0543,
+      "step": 61
+    },
+    {
+      "epoch": 0.22142857142857142,
+      "grad_norm": 1.9111756785102334,
+      "learning_rate": 1.4761904761904763e-05,
+      "loss": 0.1064,
+      "step": 62
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 1.0249475365343874,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0818,
+      "step": 63
+    },
+    {
+      "epoch": 0.22857142857142856,
+      "grad_norm": 1.0739526383280837,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.0839,
+      "step": 64
+    },
+    {
+      "epoch": 0.23214285714285715,
+      "grad_norm": 1.707812577710304,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.1078,
+      "step": 65
+    },
+    {
+      "epoch": 0.2357142857142857,
+      "grad_norm": 1.755306760009629,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.1201,
+      "step": 66
+    },
+    {
+      "epoch": 0.2392857142857143,
+      "grad_norm": 1.8304574201497188,
+      "learning_rate": 1.5952380952380954e-05,
+      "loss": 0.0819,
+      "step": 67
+    },
+    {
+      "epoch": 0.24285714285714285,
+      "grad_norm": 1.7113094274636416,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.168,
+      "step": 68
+    },
+    {
+      "epoch": 0.24642857142857144,
+      "grad_norm": 3.115930471295973,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.1025,
+      "step": 69
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.9542415937029245,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0775,
+      "step": 70
+    },
+    {
+      "epoch": 0.25357142857142856,
+      "grad_norm": 1.2879949337863554,
+      "learning_rate": 1.6904761904761906e-05,
+      "loss": 0.0861,
+      "step": 71
+    },
+    {
+      "epoch": 0.2571428571428571,
+      "grad_norm": 1.1797993713802457,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.0645,
+      "step": 72
+    },
+    {
+      "epoch": 0.26071428571428573,
+      "grad_norm": 3.3841193443761655,
+      "learning_rate": 1.7380952380952384e-05,
+      "loss": 0.1469,
+      "step": 73
+    },
+    {
+      "epoch": 0.2642857142857143,
+      "grad_norm": 0.8460269535054071,
+      "learning_rate": 1.761904761904762e-05,
+      "loss": 0.0552,
+      "step": 74
+    },
+    {
+      "epoch": 0.26785714285714285,
+      "grad_norm": 0.9455751745946019,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.0634,
+      "step": 75
+    },
+    {
+      "epoch": 0.2714285714285714,
+      "grad_norm": 2.342847803810098,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.1396,
+      "step": 76
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 1.2884062839618575,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.0816,
+      "step": 77
+    },
+    {
+      "epoch": 0.2785714285714286,
+      "grad_norm": 1.6628732713300585,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.0795,
+      "step": 78
+    },
+    {
+      "epoch": 0.28214285714285714,
+      "grad_norm": 1.2182947834703313,
+      "learning_rate": 1.880952380952381e-05,
+      "loss": 0.0785,
+      "step": 79
+    },
+    {
+      "epoch": 0.2857142857142857,
+      "grad_norm": 3.4068523237690864,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.0702,
+      "step": 80
+    },
+    {
+      "epoch": 0.2892857142857143,
+      "grad_norm": 1.3321499794988305,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.0658,
+      "step": 81
+    },
+    {
+      "epoch": 0.29285714285714287,
+      "grad_norm": 1.4666939419844447,
+      "learning_rate": 1.9523809523809524e-05,
+      "loss": 0.0589,
+      "step": 82
+    },
+    {
+      "epoch": 0.29642857142857143,
+      "grad_norm": 1.4568567885650205,
+      "learning_rate": 1.9761904761904763e-05,
+      "loss": 0.0745,
+      "step": 83
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3011194104287231,
+      "learning_rate": 2e-05,
+      "loss": 0.1013,
+      "step": 84
+    },
+    {
+      "epoch": 0.30357142857142855,
+      "grad_norm": 1.8328773107884164,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.0572,
+      "step": 85
+    },
+    {
+      "epoch": 0.30714285714285716,
+      "grad_norm": 1.856313770045762,
+      "learning_rate": 2.0476190476190476e-05,
+      "loss": 0.114,
+      "step": 86
+    },
+    {
+      "epoch": 0.3107142857142857,
+      "grad_norm": 1.7378946487369082,
+      "learning_rate": 2.0714285714285718e-05,
+      "loss": 0.057,
+      "step": 87
+    },
+    {
+      "epoch": 0.3142857142857143,
+      "grad_norm": 1.8635162072625109,
+      "learning_rate": 2.0952380952380954e-05,
+      "loss": 0.1166,
+      "step": 88
+    },
+    {
+      "epoch": 0.31785714285714284,
+      "grad_norm": 3.2998481652298906,
+      "learning_rate": 2.1190476190476193e-05,
+      "loss": 0.1213,
+      "step": 89
+    },
+    {
+      "epoch": 0.32142857142857145,
+      "grad_norm": 1.5590643729053062,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.0922,
+      "step": 90
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 1.5876673068731728,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.1009,
+      "step": 91
+    },
+    {
+      "epoch": 0.32857142857142857,
+      "grad_norm": 2.5662191125350633,
+      "learning_rate": 2.190476190476191e-05,
+      "loss": 0.1247,
+      "step": 92
+    },
+    {
+      "epoch": 0.33214285714285713,
+      "grad_norm": 2.3290946844637177,
+      "learning_rate": 2.2142857142857145e-05,
+      "loss": 0.0706,
+      "step": 93
+    },
+    {
+      "epoch": 0.3357142857142857,
+      "grad_norm": 1.7454939009619057,
+      "learning_rate": 2.2380952380952384e-05,
+      "loss": 0.12,
+      "step": 94
+    },
+    {
+      "epoch": 0.3392857142857143,
+      "grad_norm": 1.418856857352545,
+      "learning_rate": 2.261904761904762e-05,
+      "loss": 0.0487,
+      "step": 95
+    },
+    {
+      "epoch": 0.34285714285714286,
+      "grad_norm": 1.4773283102249588,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.0697,
+      "step": 96
+    },
+    {
+      "epoch": 0.3464285714285714,
+      "grad_norm": 1.7029230294423938,
+      "learning_rate": 2.3095238095238094e-05,
+      "loss": 0.0902,
+      "step": 97
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0272515685172343,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.1178,
+      "step": 98
+    },
+    {
+      "epoch": 0.3535714285714286,
+      "grad_norm": 1.1905802230811404,
+      "learning_rate": 2.3571428571428575e-05,
+      "loss": 0.0877,
+      "step": 99
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 1.8320823160442867,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.1276,
+      "step": 100
+    },
+    {
+      "epoch": 0.3607142857142857,
+      "grad_norm": 1.720811162672069,
+      "learning_rate": 2.404761904761905e-05,
+      "loss": 0.0547,
+      "step": 101
+    },
+    {
+      "epoch": 0.36428571428571427,
+      "grad_norm": 1.6443372353118653,
+      "learning_rate": 2.4285714285714285e-05,
+      "loss": 0.0781,
+      "step": 102
+    },
+    {
+      "epoch": 0.3678571428571429,
+      "grad_norm": 2.496633119718356,
+      "learning_rate": 2.4523809523809527e-05,
+      "loss": 0.1408,
+      "step": 103
+    },
+    {
+      "epoch": 0.37142857142857144,
+      "grad_norm": 1.172630108215848,
+      "learning_rate": 2.4761904761904766e-05,
+      "loss": 0.038,
+      "step": 104
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 3.7620742696759137,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1563,
+      "step": 105
+    },
+    {
+      "epoch": 0.37857142857142856,
+      "grad_norm": 2.038149201879098,
+      "learning_rate": 2.523809523809524e-05,
+      "loss": 0.1177,
+      "step": 106
+    },
+    {
+      "epoch": 0.3821428571428571,
+      "grad_norm": 1.3775610258454989,
+      "learning_rate": 2.5476190476190476e-05,
+      "loss": 0.0596,
+      "step": 107
+    },
+    {
+      "epoch": 0.38571428571428573,
+      "grad_norm": 2.1541611947495283,
+      "learning_rate": 2.5714285714285718e-05,
+      "loss": 0.1176,
+      "step": 108
+    },
+    {
+      "epoch": 0.3892857142857143,
+      "grad_norm": 1.7681460206416166,
+      "learning_rate": 2.5952380952380957e-05,
+      "loss": 0.0961,
+      "step": 109
+    },
+    {
+      "epoch": 0.39285714285714285,
+      "grad_norm": 1.7885355926029933,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.155,
+      "step": 110
+    },
+    {
+      "epoch": 0.3964285714285714,
+      "grad_norm": 0.9594960062750463,
+      "learning_rate": 2.642857142857143e-05,
+      "loss": 0.138,
+      "step": 111
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6467799531440577,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.1296,
+      "step": 112
+    },
+    {
+      "epoch": 0.4035714285714286,
+      "grad_norm": 0.8584540721575626,
+      "learning_rate": 2.690476190476191e-05,
+      "loss": 0.1206,
+      "step": 113
+    },
+    {
+      "epoch": 0.40714285714285714,
+      "grad_norm": 1.322202136913476,
+      "learning_rate": 2.7142857142857148e-05,
+      "loss": 0.1482,
+      "step": 114
+    },
+    {
+      "epoch": 0.4107142857142857,
+      "grad_norm": 2.3213418758522457,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.1045,
+      "step": 115
+    },
+    {
+      "epoch": 0.4142857142857143,
+      "grad_norm": 0.7522057778397764,
+      "learning_rate": 2.7619047619047622e-05,
+      "loss": 0.0857,
+      "step": 116
+    },
+    {
+      "epoch": 0.41785714285714287,
+      "grad_norm": 1.5220287668982417,
+      "learning_rate": 2.7857142857142858e-05,
+      "loss": 0.0681,
+      "step": 117
+    },
+    {
+      "epoch": 0.42142857142857143,
+      "grad_norm": 0.9333957271541317,
+      "learning_rate": 2.8095238095238096e-05,
+      "loss": 0.0981,
+      "step": 118
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 1.5939499407990598,
+      "learning_rate": 2.833333333333334e-05,
+      "loss": 0.1058,
+      "step": 119
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 1.0565001935977638,
+      "learning_rate": 2.8571428571428574e-05,
+      "loss": 0.0851,
+      "step": 120
+    },
+    {
+      "epoch": 0.43214285714285716,
+      "grad_norm": 0.8724859338301517,
+      "learning_rate": 2.8809523809523813e-05,
+      "loss": 0.0911,
+      "step": 121
+    },
+    {
+      "epoch": 0.4357142857142857,
+      "grad_norm": 2.003092649603173,
+      "learning_rate": 2.904761904761905e-05,
+      "loss": 0.1154,
+      "step": 122
+    },
+    {
+      "epoch": 0.4392857142857143,
+      "grad_norm": 1.6509833790093358,
+      "learning_rate": 2.9285714285714288e-05,
+      "loss": 0.1157,
+      "step": 123
+    },
+    {
+      "epoch": 0.44285714285714284,
+      "grad_norm": 1.6226668220589093,
+      "learning_rate": 2.9523809523809526e-05,
+      "loss": 0.1071,
+      "step": 124
+    },
+    {
+      "epoch": 0.44642857142857145,
+      "grad_norm": 1.6059902199838492,
+      "learning_rate": 2.9761904761904765e-05,
+      "loss": 0.0779,
+      "step": 125
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.2843523780126018,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.102,
+      "step": 126
+    },
+    {
+      "epoch": 0.45357142857142857,
+      "grad_norm": 0.7757988418090723,
+      "learning_rate": 3.023809523809524e-05,
+      "loss": 0.0787,
+      "step": 127
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 1.864776705392358,
+      "learning_rate": 3.047619047619048e-05,
+      "loss": 0.0935,
+      "step": 128
+    },
+    {
+      "epoch": 0.4607142857142857,
+      "grad_norm": 1.635246990121031,
+      "learning_rate": 3.071428571428572e-05,
+      "loss": 0.0951,
+      "step": 129
+    },
+    {
+      "epoch": 0.4642857142857143,
+      "grad_norm": 0.6283859975011667,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.0633,
+      "step": 130
+    },
+    {
+      "epoch": 0.46785714285714286,
+      "grad_norm": 3.265041337054097,
+      "learning_rate": 3.1190476190476195e-05,
+      "loss": 0.1215,
+      "step": 131
+    },
+    {
+      "epoch": 0.4714285714285714,
+      "grad_norm": 1.5873533755239178,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.0671,
+      "step": 132
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 1.1832350812981725,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.074,
+      "step": 133
+    },
+    {
+      "epoch": 0.4785714285714286,
+      "grad_norm": 0.9992361553670249,
+      "learning_rate": 3.190476190476191e-05,
+      "loss": 0.0831,
+      "step": 134
+    },
+    {
+      "epoch": 0.48214285714285715,
+      "grad_norm": 1.9822939155533887,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.0931,
+      "step": 135
+    },
+    {
+      "epoch": 0.4857142857142857,
+      "grad_norm": 1.7702080923140466,
+      "learning_rate": 3.2380952380952386e-05,
+      "loss": 0.1084,
+      "step": 136
+    },
+    {
+      "epoch": 0.48928571428571427,
+      "grad_norm": 1.3589337110275708,
+      "learning_rate": 3.261904761904762e-05,
+      "loss": 0.0481,
+      "step": 137
+    },
+    {
+      "epoch": 0.4928571428571429,
+      "grad_norm": 1.5230605663225705,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 0.1209,
+      "step": 138
+    },
+    {
+      "epoch": 0.49642857142857144,
+      "grad_norm": 1.116744273635823,
+      "learning_rate": 3.309523809523809e-05,
+      "loss": 0.1082,
+      "step": 139
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0458246316775517,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1702,
+      "step": 140
+    },
+    {
+      "epoch": 0.5035714285714286,
+      "grad_norm": 1.890979626046139,
+      "learning_rate": 3.357142857142858e-05,
+      "loss": 0.0474,
+      "step": 141
+    },
+    {
+      "epoch": 0.5071428571428571,
+      "grad_norm": 1.2021419616872644,
+      "learning_rate": 3.380952380952381e-05,
+      "loss": 0.1201,
+      "step": 142
+    },
+    {
+      "epoch": 0.5107142857142857,
+      "grad_norm": 0.9874717423563248,
+      "learning_rate": 3.404761904761905e-05,
+      "loss": 0.0431,
+      "step": 143
+    },
+    {
+      "epoch": 0.5142857142857142,
+      "grad_norm": 2.796537651330358,
+      "learning_rate": 3.4285714285714284e-05,
+      "loss": 0.1196,
+      "step": 144
+    },
+    {
+      "epoch": 0.5178571428571429,
+      "grad_norm": 1.3272731681515693,
+      "learning_rate": 3.4523809523809526e-05,
+      "loss": 0.0842,
+      "step": 145
+    },
+    {
+      "epoch": 0.5214285714285715,
+      "grad_norm": 1.283551706990249,
+      "learning_rate": 3.476190476190477e-05,
+      "loss": 0.0671,
+      "step": 146
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 1.4978091320236977,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 0.0478,
+      "step": 147
+    },
+    {
+      "epoch": 0.5285714285714286,
+      "grad_norm": 1.3975082760941162,
+      "learning_rate": 3.523809523809524e-05,
+      "loss": 0.0624,
+      "step": 148
+    },
+    {
+      "epoch": 0.5321428571428571,
+      "grad_norm": 1.438623420848124,
+      "learning_rate": 3.5476190476190475e-05,
+      "loss": 0.1068,
+      "step": 149
+    },
+    {
+      "epoch": 0.5357142857142857,
+      "grad_norm": 0.8474186162676811,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.0741,
+      "step": 150
+    },
+    {
+      "epoch": 0.5392857142857143,
+      "grad_norm": 2.1717193972779993,
+      "learning_rate": 3.595238095238096e-05,
+      "loss": 0.0744,
+      "step": 151
+    },
+    {
+      "epoch": 0.5428571428571428,
+      "grad_norm": 1.9917104971618487,
+      "learning_rate": 3.6190476190476195e-05,
+      "loss": 0.1232,
+      "step": 152
+    },
+    {
+      "epoch": 0.5464285714285714,
+      "grad_norm": 2.179978730923194,
+      "learning_rate": 3.642857142857143e-05,
+      "loss": 0.0807,
+      "step": 153
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.3913716519692871,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.114,
+      "step": 154
+    },
+    {
+      "epoch": 0.5535714285714286,
+      "grad_norm": 1.6246457505049139,
+      "learning_rate": 3.690476190476191e-05,
+      "loss": 0.1371,
+      "step": 155
+    },
+    {
+      "epoch": 0.5571428571428572,
+      "grad_norm": 0.944105851104798,
+      "learning_rate": 3.714285714285715e-05,
+      "loss": 0.1277,
+      "step": 156
+    },
+    {
+      "epoch": 0.5607142857142857,
+      "grad_norm": 2.454869359604237,
+      "learning_rate": 3.7380952380952386e-05,
+      "loss": 0.1535,
+      "step": 157
+    },
+    {
+      "epoch": 0.5642857142857143,
+      "grad_norm": 2.7484341753379895,
+      "learning_rate": 3.761904761904762e-05,
+      "loss": 0.1257,
+      "step": 158
+    },
+    {
+      "epoch": 0.5678571428571428,
+      "grad_norm": 4.0111620891092095,
+      "learning_rate": 3.785714285714286e-05,
+      "loss": 0.107,
+      "step": 159
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 1.216324178099285,
+      "learning_rate": 3.80952380952381e-05,
+      "loss": 0.1339,
+      "step": 160
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 1.3469917951952834,
+      "learning_rate": 3.833333333333334e-05,
+      "loss": 0.086,
+      "step": 161
+    },
+    {
+      "epoch": 0.5785714285714286,
+      "grad_norm": 1.2527275437826204,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 0.0918,
+      "step": 162
+    },
+    {
+      "epoch": 0.5821428571428572,
+      "grad_norm": 2.480709801857708,
+      "learning_rate": 3.880952380952381e-05,
+      "loss": 0.1362,
+      "step": 163
+    },
+    {
+      "epoch": 0.5857142857142857,
+      "grad_norm": 1.3683898413044697,
+      "learning_rate": 3.904761904761905e-05,
+      "loss": 0.1015,
+      "step": 164
+    },
+    {
+      "epoch": 0.5892857142857143,
+      "grad_norm": 1.0339533402354204,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.0777,
+      "step": 165
+    },
+    {
+      "epoch": 0.5928571428571429,
+      "grad_norm": 0.715306267644628,
+      "learning_rate": 3.9523809523809526e-05,
+      "loss": 0.0615,
+      "step": 166
+    },
+    {
+      "epoch": 0.5964285714285714,
+      "grad_norm": 1.24483946049628,
+      "learning_rate": 3.976190476190477e-05,
+      "loss": 0.053,
+      "step": 167
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.233685677876475,
+      "learning_rate": 4e-05,
+      "loss": 0.0858,
+      "step": 168
+    },
+    {
+      "epoch": 0.6035714285714285,
+      "grad_norm": 0.42115219457676595,
+      "learning_rate": 3.9999996655122555e-05,
+      "loss": 0.0941,
+      "step": 169
+    },
+    {
+      "epoch": 0.6071428571428571,
+      "grad_norm": 0.9462723900912754,
+      "learning_rate": 3.999998662049132e-05,
+      "loss": 0.0846,
+      "step": 170
+    },
+    {
+      "epoch": 0.6107142857142858,
+      "grad_norm": 0.8158818562574234,
+      "learning_rate": 3.999996989610966e-05,
+      "loss": 0.1022,
+      "step": 171
+    },
+    {
+      "epoch": 0.6142857142857143,
+      "grad_norm": 0.74403456703164,
+      "learning_rate": 3.999994648198317e-05,
+      "loss": 0.1153,
+      "step": 172
+    },
+    {
+      "epoch": 0.6178571428571429,
+      "grad_norm": 2.1161847952144255,
+      "learning_rate": 3.999991637811968e-05,
+      "loss": 0.0637,
+      "step": 173
+    },
+    {
+      "epoch": 0.6214285714285714,
+      "grad_norm": 1.868082503618805,
+      "learning_rate": 3.9999879584529264e-05,
+      "loss": 0.1243,
+      "step": 174
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 2.2510781811089786,
+      "learning_rate": 3.999983610122422e-05,
+      "loss": 0.1243,
+      "step": 175
+    },
+    {
+      "epoch": 0.6285714285714286,
+      "grad_norm": 1.105498419722501,
+      "learning_rate": 3.9999785928219085e-05,
+      "loss": 0.0657,
+      "step": 176
+    },
+    {
+      "epoch": 0.6321428571428571,
+      "grad_norm": 0.8180425056849024,
+      "learning_rate": 3.999972906553067e-05,
+      "loss": 0.054,
+      "step": 177
+    },
+    {
+      "epoch": 0.6357142857142857,
+      "grad_norm": 0.861050775218869,
+      "learning_rate": 3.9999665513177974e-05,
+      "loss": 0.0645,
+      "step": 178
+    },
+    {
+      "epoch": 0.6392857142857142,
+      "grad_norm": 0.7499184401519348,
+      "learning_rate": 3.9999595271182255e-05,
+      "loss": 0.0853,
+      "step": 179
+    },
+    {
+      "epoch": 0.6428571428571429,
+      "grad_norm": 0.9901883330374264,
+      "learning_rate": 3.9999518339567016e-05,
+      "loss": 0.11,
+      "step": 180
+    },
+    {
+      "epoch": 0.6464285714285715,
+      "grad_norm": 0.8981211342569786,
+      "learning_rate": 3.9999434718357985e-05,
+      "loss": 0.0805,
+      "step": 181
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8795249380570482,
+      "learning_rate": 3.999934440758313e-05,
+      "loss": 0.0758,
+      "step": 182
+    },
+    {
+      "epoch": 0.6535714285714286,
+      "grad_norm": 0.6910724514366515,
+      "learning_rate": 3.999924740727266e-05,
+      "loss": 0.062,
+      "step": 183
+    },
+    {
+      "epoch": 0.6571428571428571,
+      "grad_norm": 1.3331820533917738,
+      "learning_rate": 3.999914371745902e-05,
+      "loss": 0.0822,
+      "step": 184
+    },
+    {
+      "epoch": 0.6607142857142857,
+      "grad_norm": 0.8301882970323585,
+      "learning_rate": 3.9999033338176906e-05,
+      "loss": 0.0539,
+      "step": 185
+    },
+    {
+      "epoch": 0.6642857142857143,
+      "grad_norm": 0.8905457933586546,
+      "learning_rate": 3.9998916269463215e-05,
+      "loss": 0.07,
+      "step": 186
+    },
+    {
+      "epoch": 0.6678571428571428,
+      "grad_norm": 0.9810587140851165,
+      "learning_rate": 3.999879251135712e-05,
+      "loss": 0.1109,
+      "step": 187
+    },
+    {
+      "epoch": 0.6714285714285714,
+      "grad_norm": 0.845088660560638,
+      "learning_rate": 3.999866206390002e-05,
+      "loss": 0.102,
+      "step": 188
+    },
+    {
+      "epoch": 0.675,
+      "grad_norm": 0.9203643906501286,
+      "learning_rate": 3.9998524927135544e-05,
+      "loss": 0.105,
+      "step": 189
+    },
+    {
+      "epoch": 0.6785714285714286,
+      "grad_norm": 0.6044756082021804,
+      "learning_rate": 3.999838110110956e-05,
+      "loss": 0.0654,
+      "step": 190
+    },
+    {
+      "epoch": 0.6821428571428572,
+      "grad_norm": 1.3818918621638805,
+      "learning_rate": 3.9998230585870174e-05,
+      "loss": 0.0382,
+      "step": 191
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 1.0986564741500602,
+      "learning_rate": 3.999807338146773e-05,
+      "loss": 0.112,
+      "step": 192
+    },
+    {
+      "epoch": 0.6892857142857143,
+      "grad_norm": 0.5857845662750193,
+      "learning_rate": 3.999790948795481e-05,
+      "loss": 0.0567,
+      "step": 193
+    },
+    {
+      "epoch": 0.6928571428571428,
+      "grad_norm": 0.599515437334773,
+      "learning_rate": 3.9997738905386255e-05,
+      "loss": 0.0612,
+      "step": 194
+    },
+    {
+      "epoch": 0.6964285714285714,
+      "grad_norm": 0.4517165893340363,
+      "learning_rate": 3.99975616338191e-05,
+      "loss": 0.0569,
+      "step": 195
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 3.4562329720993623,
+      "learning_rate": 3.9997377673312644e-05,
+      "loss": 0.1851,
+      "step": 196
+    },
+    {
+      "epoch": 0.7035714285714286,
+      "grad_norm": 1.192296278365767,
+      "learning_rate": 3.999718702392843e-05,
+      "loss": 0.0885,
+      "step": 197
+    },
+    {
+      "epoch": 0.7071428571428572,
+      "grad_norm": 0.9330501986516581,
+      "learning_rate": 3.999698968573021e-05,
+      "loss": 0.0611,
+      "step": 198
+    },
+    {
+      "epoch": 0.7107142857142857,
+      "grad_norm": 1.2741065311524322,
+      "learning_rate": 3.999678565878402e-05,
+      "loss": 0.0908,
+      "step": 199
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 0.6465801556976934,
+      "learning_rate": 3.999657494315807e-05,
+      "loss": 0.0586,
+      "step": 200
+    },
+    {
+      "epoch": 0.7178571428571429,
+      "grad_norm": 0.8115361201898217,
+      "learning_rate": 3.999635753892286e-05,
+      "loss": 0.1035,
+      "step": 201
+    },
+    {
+      "epoch": 0.7214285714285714,
+      "grad_norm": 0.6991771190578112,
+      "learning_rate": 3.999613344615112e-05,
+      "loss": 0.0738,
+      "step": 202
+    },
+    {
+      "epoch": 0.725,
+      "grad_norm": 1.3938421924448294,
+      "learning_rate": 3.999590266491778e-05,
+      "loss": 0.1313,
+      "step": 203
+    },
+    {
+      "epoch": 0.7285714285714285,
+      "grad_norm": 1.2477569002844668,
+      "learning_rate": 3.999566519530006e-05,
+      "loss": 0.071,
+      "step": 204
+    },
+    {
+      "epoch": 0.7321428571428571,
+      "grad_norm": 0.9353442067127694,
+      "learning_rate": 3.9995421037377366e-05,
+      "loss": 0.1282,
+      "step": 205
+    },
+    {
+      "epoch": 0.7357142857142858,
+      "grad_norm": 0.6271313655274752,
+      "learning_rate": 3.999517019123139e-05,
+      "loss": 0.1061,
+      "step": 206
+    },
+    {
+      "epoch": 0.7392857142857143,
+      "grad_norm": 1.3404978181561522,
+      "learning_rate": 3.999491265694601e-05,
+      "loss": 0.0775,
+      "step": 207
+    },
+    {
+      "epoch": 0.7428571428571429,
+      "grad_norm": 1.047189435704196,
+      "learning_rate": 3.99946484346074e-05,
+      "loss": 0.0741,
+      "step": 208
+    },
+    {
+      "epoch": 0.7464285714285714,
+      "grad_norm": 1.0881643470468925,
+      "learning_rate": 3.999437752430391e-05,
+      "loss": 0.065,
+      "step": 209
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5936806173650533,
+      "learning_rate": 3.999409992612616e-05,
+      "loss": 0.0747,
+      "step": 210
+    },
+    {
+      "epoch": 0.7535714285714286,
+      "grad_norm": 1.0940241519367941,
+      "learning_rate": 3.999381564016703e-05,
+      "loss": 0.0641,
+      "step": 211
+    },
+    {
+      "epoch": 0.7571428571428571,
+      "grad_norm": 1.651938436208599,
+      "learning_rate": 3.999352466652158e-05,
+      "loss": 0.1111,
+      "step": 212
+    },
+    {
+      "epoch": 0.7607142857142857,
+      "grad_norm": 1.1638607123995697,
+      "learning_rate": 3.999322700528715e-05,
+      "loss": 0.1078,
+      "step": 213
+    },
+    {
+      "epoch": 0.7642857142857142,
+      "grad_norm": 1.3236842000392444,
+      "learning_rate": 3.9992922656563306e-05,
+      "loss": 0.0411,
+      "step": 214
+    },
+    {
+      "epoch": 0.7678571428571429,
+      "grad_norm": 1.3758017417584147,
+      "learning_rate": 3.999261162045184e-05,
+      "loss": 0.1033,
+      "step": 215
+    },
+    {
+      "epoch": 0.7714285714285715,
+      "grad_norm": 0.6746285140055772,
+      "learning_rate": 3.99922938970568e-05,
+      "loss": 0.0573,
+      "step": 216
+    },
+    {
+      "epoch": 0.775,
+      "grad_norm": 0.679440635160988,
+      "learning_rate": 3.999196948648446e-05,
+      "loss": 0.0628,
+      "step": 217
+    },
+    {
+      "epoch": 0.7785714285714286,
+      "grad_norm": 1.67846704940794,
+      "learning_rate": 3.999163838884332e-05,
+      "loss": 0.0773,
+      "step": 218
+    },
+    {
+      "epoch": 0.7821428571428571,
+      "grad_norm": 1.4147416588406725,
+      "learning_rate": 3.999130060424414e-05,
+      "loss": 0.0682,
+      "step": 219
+    },
+    {
+      "epoch": 0.7857142857142857,
+      "grad_norm": 0.5166188161509333,
+      "learning_rate": 3.9990956132799896e-05,
+      "loss": 0.0562,
+      "step": 220
+    },
+    {
+      "epoch": 0.7892857142857143,
+      "grad_norm": 1.7052681172521749,
+      "learning_rate": 3.999060497462581e-05,
+      "loss": 0.1044,
+      "step": 221
+    },
+    {
+      "epoch": 0.7928571428571428,
+      "grad_norm": 1.2617405660340528,
+      "learning_rate": 3.999024712983935e-05,
+      "loss": 0.1224,
+      "step": 222
+    },
+    {
+      "epoch": 0.7964285714285714,
+      "grad_norm": 1.119412388592049,
+      "learning_rate": 3.99898825985602e-05,
+      "loss": 0.074,
+      "step": 223
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8002041534828582,
+      "learning_rate": 3.998951138091031e-05,
+      "loss": 0.0814,
+      "step": 224
+    },
+    {
+      "epoch": 0.8035714285714286,
+      "grad_norm": 1.4029809358324539,
+      "learning_rate": 3.998913347701381e-05,
+      "loss": 0.1389,
+      "step": 225
+    },
+    {
+      "epoch": 0.8071428571428572,
+      "grad_norm": 0.9338101326176634,
+      "learning_rate": 3.9988748886997144e-05,
+      "loss": 0.0797,
+      "step": 226
+    },
+    {
+      "epoch": 0.8107142857142857,
+      "grad_norm": 0.6630314577468278,
+      "learning_rate": 3.9988357610988935e-05,
+      "loss": 0.0855,
+      "step": 227
+    },
+    {
+      "epoch": 0.8142857142857143,
+      "grad_norm": 1.3654243901672047,
+      "learning_rate": 3.998795964912006e-05,
+      "loss": 0.0666,
+      "step": 228
+    },
+    {
+      "epoch": 0.8178571428571428,
+      "grad_norm": 1.1370639401078195,
+      "learning_rate": 3.998755500152362e-05,
+      "loss": 0.092,
+      "step": 229
+    },
+    {
+      "epoch": 0.8214285714285714,
+      "grad_norm": 1.7075818383578598,
+      "learning_rate": 3.9987143668334996e-05,
+      "loss": 0.1161,
+      "step": 230
+    },
+    {
+      "epoch": 0.825,
+      "grad_norm": 0.8724447318266455,
+      "learning_rate": 3.9986725649691756e-05,
+      "loss": 0.0357,
+      "step": 231
+    },
+    {
+      "epoch": 0.8285714285714286,
+      "grad_norm": 0.6858426503795594,
+      "learning_rate": 3.998630094573371e-05,
+      "loss": 0.0851,
+      "step": 232
+    },
+    {
+      "epoch": 0.8321428571428572,
+      "grad_norm": 0.7460750388024511,
+      "learning_rate": 3.9985869556602935e-05,
+      "loss": 0.0672,
+      "step": 233
+    },
+    {
+      "epoch": 0.8357142857142857,
+      "grad_norm": 1.4144947548914417,
+      "learning_rate": 3.9985431482443726e-05,
+      "loss": 0.0516,
+      "step": 234
+    },
+    {
+      "epoch": 0.8392857142857143,
+      "grad_norm": 0.8301077736790788,
+      "learning_rate": 3.99849867234026e-05,
+      "loss": 0.086,
+      "step": 235
+    },
+    {
+      "epoch": 0.8428571428571429,
+      "grad_norm": 0.8425570239951746,
+      "learning_rate": 3.998453527962833e-05,
+      "loss": 0.0586,
+      "step": 236
+    },
+    {
+      "epoch": 0.8464285714285714,
+      "grad_norm": 1.1993398561971251,
+      "learning_rate": 3.998407715127192e-05,
+      "loss": 0.0868,
+      "step": 237
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.0228378080456106,
+      "learning_rate": 3.9983612338486604e-05,
+      "loss": 0.1009,
+      "step": 238
+    },
+    {
+      "epoch": 0.8535714285714285,
+      "grad_norm": 1.179333632964974,
+      "learning_rate": 3.998314084142786e-05,
+      "loss": 0.12,
+      "step": 239
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 1.0510376705223217,
+      "learning_rate": 3.9982662660253396e-05,
+      "loss": 0.1188,
+      "step": 240
+    },
+    {
+      "epoch": 0.8607142857142858,
+      "grad_norm": 2.0030764485783896,
+      "learning_rate": 3.998217779512316e-05,
+      "loss": 0.0577,
+      "step": 241
+    },
+    {
+      "epoch": 0.8642857142857143,
+      "grad_norm": 1.3223071169735103,
+      "learning_rate": 3.9981686246199334e-05,
+      "loss": 0.1063,
+      "step": 242
+    },
+    {
+      "epoch": 0.8678571428571429,
+      "grad_norm": 0.8825542338185316,
+      "learning_rate": 3.998118801364633e-05,
+      "loss": 0.041,
+      "step": 243
+    },
+    {
+      "epoch": 0.8714285714285714,
+      "grad_norm": 1.0083204099797398,
+      "learning_rate": 3.99806830976308e-05,
+      "loss": 0.0587,
+      "step": 244
+    },
+    {
+      "epoch": 0.875,
+      "grad_norm": 0.6439777459230878,
+      "learning_rate": 3.998017149832164e-05,
+      "loss": 0.0856,
+      "step": 245
+    },
+    {
+      "epoch": 0.8785714285714286,
+      "grad_norm": 0.8485629458126653,
+      "learning_rate": 3.997965321588997e-05,
+      "loss": 0.1202,
+      "step": 246
+    },
+    {
+      "epoch": 0.8821428571428571,
+      "grad_norm": 2.39217856917238,
+      "learning_rate": 3.9979128250509155e-05,
+      "loss": 0.1062,
+      "step": 247
+    },
+    {
+      "epoch": 0.8857142857142857,
+      "grad_norm": 1.8878683564636158,
+      "learning_rate": 3.9978596602354774e-05,
+      "loss": 0.1035,
+      "step": 248
+    },
+    {
+      "epoch": 0.8892857142857142,
+      "grad_norm": 0.8930817641720431,
+      "learning_rate": 3.9978058271604666e-05,
+      "loss": 0.0893,
+      "step": 249
+    },
+    {
+      "epoch": 0.8928571428571429,
+      "grad_norm": 1.369439975224411,
+      "learning_rate": 3.99775132584389e-05,
+      "loss": 0.115,
+      "step": 250
+    },
+    {
+      "epoch": 0.8964285714285715,
+      "grad_norm": 1.2660740617287565,
+      "learning_rate": 3.9976961563039775e-05,
+      "loss": 0.0687,
+      "step": 251
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0650701867468773,
+      "learning_rate": 3.997640318559182e-05,
+      "loss": 0.0724,
+      "step": 252
+    },
+    {
+      "epoch": 0.9035714285714286,
+      "grad_norm": 0.8409442479884971,
+      "learning_rate": 3.9975838126281806e-05,
+      "loss": 0.0787,
+      "step": 253
+    },
+    {
+      "epoch": 0.9071428571428571,
+      "grad_norm": 1.6789967504678176,
+      "learning_rate": 3.9975266385298744e-05,
+      "loss": 0.0716,
+      "step": 254
+    },
+    {
+      "epoch": 0.9107142857142857,
+      "grad_norm": 1.037539170692534,
+      "learning_rate": 3.997468796283388e-05,
+      "loss": 0.0874,
+      "step": 255
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 0.8312605754433313,
+      "learning_rate": 3.997410285908067e-05,
+      "loss": 0.1128,
+      "step": 256
+    },
+    {
+      "epoch": 0.9178571428571428,
+      "grad_norm": 0.7001440504835422,
+      "learning_rate": 3.997351107423484e-05,
+      "loss": 0.1049,
+      "step": 257
+    },
+    {
+      "epoch": 0.9214285714285714,
+      "grad_norm": 1.3415056653178543,
+      "learning_rate": 3.9972912608494325e-05,
+      "loss": 0.1316,
+      "step": 258
+    },
+    {
+      "epoch": 0.925,
+      "grad_norm": 1.7162636811218717,
+      "learning_rate": 3.9972307462059315e-05,
+      "loss": 0.147,
+      "step": 259
+    },
+    {
+      "epoch": 0.9285714285714286,
+      "grad_norm": 0.9977311840901677,
+      "learning_rate": 3.9971695635132214e-05,
+      "loss": 0.0824,
+      "step": 260
+    },
+    {
+      "epoch": 0.9321428571428572,
+      "grad_norm": 1.4509870587192308,
+      "learning_rate": 3.997107712791768e-05,
+      "loss": 0.1079,
+      "step": 261
+    },
+    {
+      "epoch": 0.9357142857142857,
+      "grad_norm": 1.3389255451315258,
+      "learning_rate": 3.997045194062259e-05,
+      "loss": 0.0678,
+      "step": 262
+    },
+    {
+      "epoch": 0.9392857142857143,
+      "grad_norm": 0.7180200986309652,
+      "learning_rate": 3.9969820073456065e-05,
+      "loss": 0.0875,
+      "step": 263
+    },
+    {
+      "epoch": 0.9428571428571428,
+      "grad_norm": 0.9526644757948708,
+      "learning_rate": 3.9969181526629445e-05,
+      "loss": 0.0946,
+      "step": 264
+    },
+    {
+      "epoch": 0.9464285714285714,
+      "grad_norm": 1.7933759575911479,
+      "learning_rate": 3.996853630035634e-05,
+      "loss": 0.119,
+      "step": 265
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6521610211883466,
+      "learning_rate": 3.9967884394852543e-05,
+      "loss": 0.079,
+      "step": 266
+    },
+    {
+      "epoch": 0.9535714285714286,
+      "grad_norm": 1.3839780504654737,
+      "learning_rate": 3.9967225810336127e-05,
+      "loss": 0.0747,
+      "step": 267
+    },
+    {
+      "epoch": 0.9571428571428572,
+      "grad_norm": 1.1442373013425053,
+      "learning_rate": 3.996656054702737e-05,
+      "loss": 0.0864,
+      "step": 268
+    },
+    {
+      "epoch": 0.9607142857142857,
+      "grad_norm": 0.8943920225939885,
+      "learning_rate": 3.99658886051488e-05,
+      "loss": 0.0685,
+      "step": 269
+    },
+    {
+      "epoch": 0.9642857142857143,
+      "grad_norm": 0.8284751440849266,
+      "learning_rate": 3.996520998492518e-05,
+      "loss": 0.0478,
+      "step": 270
+    },
+    {
+      "epoch": 0.9678571428571429,
+      "grad_norm": 2.870308775480328,
+      "learning_rate": 3.9964524686583486e-05,
+      "loss": 0.0972,
+      "step": 271
+    },
+    {
+      "epoch": 0.9714285714285714,
+      "grad_norm": 0.9534650927111473,
+      "learning_rate": 3.9963832710352945e-05,
+      "loss": 0.0854,
+      "step": 272
+    },
+    {
+      "epoch": 0.975,
+      "grad_norm": 1.537372099137238,
+      "learning_rate": 3.996313405646503e-05,
+      "loss": 0.093,
+      "step": 273
+    },
+    {
+      "epoch": 0.9785714285714285,
+      "grad_norm": 1.3986871637975078,
+      "learning_rate": 3.996242872515341e-05,
+      "loss": 0.1116,
+      "step": 274
+    },
+    {
+      "epoch": 0.9821428571428571,
+      "grad_norm": 0.5532425332265273,
+      "learning_rate": 3.996171671665402e-05,
+      "loss": 0.0511,
+      "step": 275
+    },
+    {
+      "epoch": 0.9857142857142858,
+      "grad_norm": 3.129212386700657,
+      "learning_rate": 3.9960998031205026e-05,
+      "loss": 0.1323,
+      "step": 276
+    },
+    {
+      "epoch": 0.9892857142857143,
+      "grad_norm": 1.8404179557206803,
+      "learning_rate": 3.996027266904681e-05,
+      "loss": 0.1246,
+      "step": 277
+    },
+    {
+      "epoch": 0.9928571428571429,
+      "grad_norm": 1.0818463932394426,
+      "learning_rate": 3.995954063042199e-05,
+      "loss": 0.0974,
+      "step": 278
+    },
+    {
+      "epoch": 0.9964285714285714,
+      "grad_norm": 3.424553258570884,
+      "learning_rate": 3.995880191557543e-05,
+      "loss": 0.0998,
+      "step": 279
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3350022639686183,
+      "learning_rate": 3.995805652475424e-05,
+      "loss": 0.0662,
+      "step": 280
+    },
+    {
+      "epoch": 1.0035714285714286,
+      "grad_norm": 1.8655597883487487,
+      "learning_rate": 3.995730445820771e-05,
+      "loss": 0.0721,
+      "step": 281
+    },
+    {
+      "epoch": 1.0071428571428571,
+      "grad_norm": 0.7639774907802587,
+      "learning_rate": 3.9956545716187423e-05,
+      "loss": 0.0552,
+      "step": 282
+    },
+    {
+      "epoch": 1.0107142857142857,
+      "grad_norm": 1.4259023041603565,
+      "learning_rate": 3.9955780298947166e-05,
+      "loss": 0.075,
+      "step": 283
+    },
+    {
+      "epoch": 1.0142857142857142,
+      "grad_norm": 1.340971179126857,
+      "learning_rate": 3.995500820674294e-05,
+      "loss": 0.0844,
+      "step": 284
+    },
+    {
+      "epoch": 1.0178571428571428,
+      "grad_norm": 0.7575543165019466,
+      "learning_rate": 3.995422943983303e-05,
+      "loss": 0.0776,
+      "step": 285
+    },
+    {
+      "epoch": 1.0214285714285714,
+      "grad_norm": 1.834924531928753,
+      "learning_rate": 3.9953443998477905e-05,
+      "loss": 0.1556,
+      "step": 286
+    },
+    {
+      "epoch": 1.025,
+      "grad_norm": 0.8129262650308676,
+      "learning_rate": 3.9952651882940294e-05,
+      "loss": 0.0726,
+      "step": 287
+    },
+    {
+      "epoch": 1.0285714285714285,
+      "grad_norm": 1.0905034914695726,
+      "learning_rate": 3.995185309348515e-05,
+      "loss": 0.0562,
+      "step": 288
+    },
+    {
+      "epoch": 1.032142857142857,
+      "grad_norm": 0.8545385359813322,
+      "learning_rate": 3.9951047630379645e-05,
+      "loss": 0.0541,
+      "step": 289
+    },
+    {
+      "epoch": 1.0357142857142858,
+      "grad_norm": 0.4894072472252566,
+      "learning_rate": 3.995023549389321e-05,
+      "loss": 0.0673,
+      "step": 290
+    },
+    {
+      "epoch": 1.0392857142857144,
+      "grad_norm": 0.79764700054824,
+      "learning_rate": 3.994941668429749e-05,
+      "loss": 0.0728,
+      "step": 291
+    },
+    {
+      "epoch": 1.042857142857143,
+      "grad_norm": 0.5962045998130294,
+      "learning_rate": 3.994859120186637e-05,
+      "loss": 0.0437,
+      "step": 292
+    },
+    {
+      "epoch": 1.0464285714285715,
+      "grad_norm": 1.0979253918048237,
+      "learning_rate": 3.994775904687596e-05,
+      "loss": 0.0816,
+      "step": 293
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6598431026847716,
+      "learning_rate": 3.994692021960461e-05,
+      "loss": 0.0743,
+      "step": 294
+    },
+    {
+      "epoch": 1.0535714285714286,
+      "grad_norm": 0.9783659700721974,
+      "learning_rate": 3.9946074720332894e-05,
+      "loss": 0.1065,
+      "step": 295
+    },
+    {
+      "epoch": 1.0571428571428572,
+      "grad_norm": 1.0078781892528923,
+      "learning_rate": 3.994522254934363e-05,
+      "loss": 0.078,
+      "step": 296
+    },
+    {
+      "epoch": 1.0607142857142857,
+      "grad_norm": 1.1439496111184553,
+      "learning_rate": 3.994436370692184e-05,
+      "loss": 0.0613,
+      "step": 297
+    },
+    {
+      "epoch": 1.0642857142857143,
+      "grad_norm": 0.5872017484902617,
+      "learning_rate": 3.9943498193354816e-05,
+      "loss": 0.0789,
+      "step": 298
+    },
+    {
+      "epoch": 1.0678571428571428,
+      "grad_norm": 1.4037463194054076,
+      "learning_rate": 3.994262600893204e-05,
+      "loss": 0.139,
+      "step": 299
+    },
+    {
+      "epoch": 1.0714285714285714,
+      "grad_norm": 1.0235993945393385,
+      "learning_rate": 3.994174715394527e-05,
+      "loss": 0.0947,
+      "step": 300
+    },
+    {
+      "epoch": 1.075,
+      "grad_norm": 1.4875965321669373,
+      "learning_rate": 3.994086162868846e-05,
+      "loss": 0.0599,
+      "step": 301
+    },
+    {
+      "epoch": 1.0785714285714285,
+      "grad_norm": 0.9123264833936583,
+      "learning_rate": 3.9939969433457804e-05,
+      "loss": 0.0749,
+      "step": 302
+    },
+    {
+      "epoch": 1.082142857142857,
+      "grad_norm": 0.9922143099067368,
+      "learning_rate": 3.9939070568551746e-05,
+      "loss": 0.1287,
+      "step": 303
+    },
+    {
+      "epoch": 1.0857142857142856,
+      "grad_norm": 0.8402849550851401,
+      "learning_rate": 3.993816503427093e-05,
+      "loss": 0.1123,
+      "step": 304
+    },
+    {
+      "epoch": 1.0892857142857142,
+      "grad_norm": 0.7470041355573542,
+      "learning_rate": 3.9937252830918244e-05,
+      "loss": 0.0725,
+      "step": 305
+    },
+    {
+      "epoch": 1.092857142857143,
+      "grad_norm": 0.9481632746032271,
+      "learning_rate": 3.9936333958798825e-05,
+      "loss": 0.0929,
+      "step": 306
+    },
+    {
+      "epoch": 1.0964285714285715,
+      "grad_norm": 1.9193964824264547,
+      "learning_rate": 3.993540841822001e-05,
+      "loss": 0.1464,
+      "step": 307
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8233081682317402,
+      "learning_rate": 3.993447620949139e-05,
+      "loss": 0.0986,
+      "step": 308
+    },
+    {
+      "epoch": 1.1035714285714286,
+      "grad_norm": 0.7551248974653422,
+      "learning_rate": 3.9933537332924776e-05,
+      "loss": 0.0728,
+      "step": 309
+    },
+    {
+      "epoch": 1.1071428571428572,
+      "grad_norm": 0.7285774714620464,
+      "learning_rate": 3.993259178883421e-05,
+      "loss": 0.0816,
+      "step": 310
+    },
+    {
+      "epoch": 1.1107142857142858,
+      "grad_norm": 1.421972830540872,
+      "learning_rate": 3.993163957753595e-05,
+      "loss": 0.0989,
+      "step": 311
+    },
+    {
+      "epoch": 1.1142857142857143,
+      "grad_norm": 0.4686999062938734,
+      "learning_rate": 3.9930680699348524e-05,
+      "loss": 0.0518,
+      "step": 312
+    },
+    {
+      "epoch": 1.1178571428571429,
+      "grad_norm": 1.3000883640850383,
+      "learning_rate": 3.9929715154592647e-05,
+      "loss": 0.1386,
+      "step": 313
+    },
+    {
+      "epoch": 1.1214285714285714,
+      "grad_norm": 1.0096646757843055,
+      "learning_rate": 3.99287429435913e-05,
+      "loss": 0.0766,
+      "step": 314
+    },
+    {
+      "epoch": 1.125,
+      "grad_norm": 0.9939012106123866,
+      "learning_rate": 3.9927764066669654e-05,
+      "loss": 0.1353,
+      "step": 315
+    },
+    {
+      "epoch": 1.1285714285714286,
+      "grad_norm": 0.7452176515429224,
+      "learning_rate": 3.992677852415514e-05,
+      "loss": 0.0594,
+      "step": 316
+    },
+    {
+      "epoch": 1.1321428571428571,
+      "grad_norm": 0.7964215206519661,
+      "learning_rate": 3.9925786316377416e-05,
+      "loss": 0.0744,
+      "step": 317
+    },
+    {
+      "epoch": 1.1357142857142857,
+      "grad_norm": 0.7499370318306604,
+      "learning_rate": 3.9924787443668355e-05,
+      "loss": 0.0808,
+      "step": 318
+    },
+    {
+      "epoch": 1.1392857142857142,
+      "grad_norm": 0.6403887968670355,
+      "learning_rate": 3.9923781906362076e-05,
+      "loss": 0.0629,
+      "step": 319
+    },
+    {
+      "epoch": 1.1428571428571428,
+      "grad_norm": 0.6482135831961022,
+      "learning_rate": 3.992276970479491e-05,
+      "loss": 0.0654,
+      "step": 320
+    },
+    {
+      "epoch": 1.1464285714285714,
+      "grad_norm": 0.9242171324214844,
+      "learning_rate": 3.9921750839305436e-05,
+      "loss": 0.0879,
+      "step": 321
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6056116631009741,
+      "learning_rate": 3.992072531023444e-05,
+      "loss": 0.0891,
+      "step": 322
+    },
+    {
+      "epoch": 1.1535714285714285,
+      "grad_norm": 0.6130430046281948,
+      "learning_rate": 3.9919693117924954e-05,
+      "loss": 0.0892,
+      "step": 323
+    },
+    {
+      "epoch": 1.157142857142857,
+      "grad_norm": 0.7838669290040703,
+      "learning_rate": 3.991865426272224e-05,
+      "loss": 0.0553,
+      "step": 324
+    },
+    {
+      "epoch": 1.1607142857142858,
+      "grad_norm": 0.6550527039205734,
+      "learning_rate": 3.991760874497377e-05,
+      "loss": 0.0874,
+      "step": 325
+    },
+    {
+      "epoch": 1.1642857142857144,
+      "grad_norm": 1.5322553277772553,
+      "learning_rate": 3.9916556565029275e-05,
+      "loss": 0.0706,
+      "step": 326
+    },
+    {
+      "epoch": 1.167857142857143,
+      "grad_norm": 1.1422878397919607,
+      "learning_rate": 3.9915497723240674e-05,
+      "loss": 0.0892,
+      "step": 327
+    },
+    {
+      "epoch": 1.1714285714285715,
+      "grad_norm": 1.0217508750622994,
+      "learning_rate": 3.991443221996216e-05,
+      "loss": 0.0322,
+      "step": 328
+    },
+    {
+      "epoch": 1.175,
+      "grad_norm": 0.911275554115598,
+      "learning_rate": 3.991336005555011e-05,
+      "loss": 0.0901,
+      "step": 329
+    },
+    {
+      "epoch": 1.1785714285714286,
+      "grad_norm": 0.4781164713715452,
+      "learning_rate": 3.991228123036316e-05,
+      "loss": 0.0515,
+      "step": 330
+    },
+    {
+      "epoch": 1.1821428571428572,
+      "grad_norm": 1.537768625157379,
+      "learning_rate": 3.991119574476216e-05,
+      "loss": 0.1068,
+      "step": 331
+    },
+    {
+      "epoch": 1.1857142857142857,
+      "grad_norm": 0.8121418776224743,
+      "learning_rate": 3.9910103599110196e-05,
+      "loss": 0.0603,
+      "step": 332
+    },
+    {
+      "epoch": 1.1892857142857143,
+      "grad_norm": 0.7291960365355258,
+      "learning_rate": 3.990900479377258e-05,
+      "loss": 0.0938,
+      "step": 333
+    },
+    {
+      "epoch": 1.1928571428571428,
+      "grad_norm": 0.7983413564852868,
+      "learning_rate": 3.9907899329116835e-05,
+      "loss": 0.0987,
+      "step": 334
+    },
+    {
+      "epoch": 1.1964285714285714,
+      "grad_norm": 0.8194134879623393,
+      "learning_rate": 3.990678720551274e-05,
+      "loss": 0.0713,
+      "step": 335
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7519108269804139,
+      "learning_rate": 3.990566842333228e-05,
+      "loss": 0.0641,
+      "step": 336
+    },
+    {
+      "epoch": 1.2035714285714285,
+      "grad_norm": 1.1176548596415925,
+      "learning_rate": 3.9904542982949676e-05,
+      "loss": 0.0994,
+      "step": 337
+    },
+    {
+      "epoch": 1.207142857142857,
+      "grad_norm": 2.1805361801212655,
+      "learning_rate": 3.990341088474138e-05,
+      "loss": 0.1011,
+      "step": 338
+    },
+    {
+      "epoch": 1.2107142857142856,
+      "grad_norm": 2.7429445313154823,
+      "learning_rate": 3.990227212908605e-05,
+      "loss": 0.1346,
+      "step": 339
+    },
+    {
+      "epoch": 1.2142857142857142,
+      "grad_norm": 1.037915748454411,
+      "learning_rate": 3.99011267163646e-05,
+      "loss": 0.0392,
+      "step": 340
+    },
+    {
+      "epoch": 1.217857142857143,
+      "grad_norm": 1.2639561432710893,
+      "learning_rate": 3.989997464696014e-05,
+      "loss": 0.1047,
+      "step": 341
+    },
+    {
+      "epoch": 1.2214285714285715,
+      "grad_norm": 1.580873998119708,
+      "learning_rate": 3.9898815921258044e-05,
+      "loss": 0.0727,
+      "step": 342
+    },
+    {
+      "epoch": 1.225,
+      "grad_norm": 0.818805708353473,
+      "learning_rate": 3.989765053964587e-05,
+      "loss": 0.0511,
+      "step": 343
+    },
+    {
+      "epoch": 1.2285714285714286,
+      "grad_norm": 0.8626835302198238,
+      "learning_rate": 3.989647850251344e-05,
+      "loss": 0.0287,
+      "step": 344
+    },
+    {
+      "epoch": 1.2321428571428572,
+      "grad_norm": 1.8052434382994489,
+      "learning_rate": 3.989529981025279e-05,
+      "loss": 0.1137,
+      "step": 345
+    },
+    {
+      "epoch": 1.2357142857142858,
+      "grad_norm": 0.9235738327485407,
+      "learning_rate": 3.989411446325816e-05,
+      "loss": 0.0793,
+      "step": 346
+    },
+    {
+      "epoch": 1.2392857142857143,
+      "grad_norm": 1.4260320970552771,
+      "learning_rate": 3.989292246192605e-05,
+      "loss": 0.0767,
+      "step": 347
+    },
+    {
+      "epoch": 1.2428571428571429,
+      "grad_norm": 0.7046180354291478,
+      "learning_rate": 3.989172380665516e-05,
+      "loss": 0.1042,
+      "step": 348
+    },
+    {
+      "epoch": 1.2464285714285714,
+      "grad_norm": 2.0230951607599192,
+      "learning_rate": 3.989051849784642e-05,
+      "loss": 0.1331,
+      "step": 349
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.0409713268751815,
+      "learning_rate": 3.9889306535903015e-05,
+      "loss": 0.0777,
+      "step": 350
+    },
+    {
+      "epoch": 1.2535714285714286,
+      "grad_norm": 1.9948414800462204,
+      "learning_rate": 3.98880879212303e-05,
+      "loss": 0.0794,
+      "step": 351
+    },
+    {
+      "epoch": 1.2571428571428571,
+      "grad_norm": 2.088723381579366,
+      "learning_rate": 3.988686265423591e-05,
+      "loss": 0.0724,
+      "step": 352
+    },
+    {
+      "epoch": 1.2607142857142857,
+      "grad_norm": 0.8947070525342539,
+      "learning_rate": 3.988563073532968e-05,
+      "loss": 0.088,
+      "step": 353
+    },
+    {
+      "epoch": 1.2642857142857142,
+      "grad_norm": 0.9025244875323668,
+      "learning_rate": 3.988439216492366e-05,
+      "loss": 0.0913,
+      "step": 354
+    },
+    {
+      "epoch": 1.2678571428571428,
+      "grad_norm": 0.8711289116873793,
+      "learning_rate": 3.988314694343215e-05,
+      "loss": 0.1126,
+      "step": 355
+    },
+    {
+      "epoch": 1.2714285714285714,
+      "grad_norm": 2.6212800327897936,
+      "learning_rate": 3.988189507127165e-05,
+      "loss": 0.1507,
+      "step": 356
+    },
+    {
+      "epoch": 1.275,
+      "grad_norm": 1.0004012459705633,
+      "learning_rate": 3.98806365488609e-05,
+      "loss": 0.0762,
+      "step": 357
+    },
+    {
+      "epoch": 1.2785714285714285,
+      "grad_norm": 1.184451250323154,
+      "learning_rate": 3.987937137662086e-05,
+      "loss": 0.095,
+      "step": 358
+    },
+    {
+      "epoch": 1.282142857142857,
+      "grad_norm": 0.8255835984359969,
+      "learning_rate": 3.987809955497471e-05,
+      "loss": 0.1261,
+      "step": 359
+    },
+    {
+      "epoch": 1.2857142857142856,
+      "grad_norm": 0.5978265519740309,
+      "learning_rate": 3.9876821084347875e-05,
+      "loss": 0.0468,
+      "step": 360
+    },
+    {
+      "epoch": 1.2892857142857144,
+      "grad_norm": 0.7028099453795392,
+      "learning_rate": 3.9875535965167965e-05,
+      "loss": 0.0607,
+      "step": 361
+    },
+    {
+      "epoch": 1.292857142857143,
+      "grad_norm": 0.4613216512397272,
+      "learning_rate": 3.9874244197864856e-05,
+      "loss": 0.0477,
+      "step": 362
+    },
+    {
+      "epoch": 1.2964285714285715,
+      "grad_norm": 0.829260676627165,
+      "learning_rate": 3.9872945782870614e-05,
+      "loss": 0.0661,
+      "step": 363
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8159383712254581,
+      "learning_rate": 3.987164072061956e-05,
+      "loss": 0.0563,
+      "step": 364
+    },
+    {
+      "epoch": 1.3035714285714286,
+      "grad_norm": 0.6381112110911379,
+      "learning_rate": 3.98703290115482e-05,
+      "loss": 0.0774,
+      "step": 365
+    },
+    {
+      "epoch": 1.3071428571428572,
+      "grad_norm": 0.8506433650301102,
+      "learning_rate": 3.98690106560953e-05,
+      "loss": 0.075,
+      "step": 366
+    },
+    {
+      "epoch": 1.3107142857142857,
+      "grad_norm": 1.399445837945221,
+      "learning_rate": 3.986768565470183e-05,
+      "loss": 0.1006,
+      "step": 367
+    },
+    {
+      "epoch": 1.3142857142857143,
+      "grad_norm": 1.0237573527681194,
+      "learning_rate": 3.986635400781099e-05,
+      "loss": 0.1049,
+      "step": 368
+    },
+    {
+      "epoch": 1.3178571428571428,
+      "grad_norm": 0.7789767441049664,
+      "learning_rate": 3.986501571586819e-05,
+      "loss": 0.0693,
+      "step": 369
+    },
+    {
+      "epoch": 1.3214285714285714,
+      "grad_norm": 1.1095790175804825,
+      "learning_rate": 3.986367077932107e-05,
+      "loss": 0.1108,
+      "step": 370
+    },
+    {
+      "epoch": 1.325,
+      "grad_norm": 0.7184760117139838,
+      "learning_rate": 3.9862319198619514e-05,
+      "loss": 0.0699,
+      "step": 371
+    },
+    {
+      "epoch": 1.3285714285714285,
+      "grad_norm": 0.9718106215016105,
+      "learning_rate": 3.9860960974215595e-05,
+      "loss": 0.0555,
+      "step": 372
+    },
+    {
+      "epoch": 1.332142857142857,
+      "grad_norm": 1.386998777046129,
+      "learning_rate": 3.9859596106563625e-05,
+      "loss": 0.1607,
+      "step": 373
+    },
+    {
+      "epoch": 1.3357142857142856,
+      "grad_norm": 1.0384468350835614,
+      "learning_rate": 3.9858224596120134e-05,
+      "loss": 0.1572,
+      "step": 374
+    },
+    {
+      "epoch": 1.3392857142857144,
+      "grad_norm": 1.6510647866826962,
+      "learning_rate": 3.985684644334388e-05,
+      "loss": 0.1099,
+      "step": 375
+    },
+    {
+      "epoch": 1.342857142857143,
+      "grad_norm": 1.4331103274443373,
+      "learning_rate": 3.985546164869583e-05,
+      "loss": 0.106,
+      "step": 376
+    },
+    {
+      "epoch": 1.3464285714285715,
+      "grad_norm": 0.9566248294275949,
+      "learning_rate": 3.985407021263919e-05,
+      "loss": 0.0641,
+      "step": 377
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.1462744865388128,
+      "learning_rate": 3.985267213563937e-05,
+      "loss": 0.0843,
+      "step": 378
+    },
+    {
+      "epoch": 1.3535714285714286,
+      "grad_norm": 0.841100081094384,
+      "learning_rate": 3.985126741816402e-05,
+      "loss": 0.0846,
+      "step": 379
+    },
+    {
+      "epoch": 1.3571428571428572,
+      "grad_norm": 0.7512069932536123,
+      "learning_rate": 3.984985606068299e-05,
+      "loss": 0.0761,
+      "step": 380
+    },
+    {
+      "epoch": 1.3607142857142858,
+      "grad_norm": 0.8089418107732764,
+      "learning_rate": 3.984843806366837e-05,
+      "loss": 0.0914,
+      "step": 381
+    },
+    {
+      "epoch": 1.3642857142857143,
+      "grad_norm": 0.817025239497121,
+      "learning_rate": 3.984701342759446e-05,
+      "loss": 0.0871,
+      "step": 382
+    },
+    {
+      "epoch": 1.3678571428571429,
+      "grad_norm": 1.9590096473103655,
+      "learning_rate": 3.984558215293778e-05,
+      "loss": 0.1299,
+      "step": 383
+    },
+    {
+      "epoch": 1.3714285714285714,
+      "grad_norm": 1.0695532498073057,
+      "learning_rate": 3.9844144240177074e-05,
+      "loss": 0.0988,
+      "step": 384
+    },
+    {
+      "epoch": 1.375,
+      "grad_norm": 0.8101555392329796,
+      "learning_rate": 3.984269968979331e-05,
+      "loss": 0.1102,
+      "step": 385
+    },
+    {
+      "epoch": 1.3785714285714286,
+      "grad_norm": 1.0612227661570137,
+      "learning_rate": 3.984124850226968e-05,
+      "loss": 0.112,
+      "step": 386
+    },
+    {
+      "epoch": 1.3821428571428571,
+      "grad_norm": 0.9700463769335379,
+      "learning_rate": 3.9839790678091566e-05,
+      "loss": 0.1113,
+      "step": 387
+    },
+    {
+      "epoch": 1.3857142857142857,
+      "grad_norm": 0.7067835865361205,
+      "learning_rate": 3.9838326217746616e-05,
+      "loss": 0.1002,
+      "step": 388
+    },
+    {
+      "epoch": 1.3892857142857142,
+      "grad_norm": 0.8582945358744356,
+      "learning_rate": 3.983685512172466e-05,
+      "loss": 0.1087,
+      "step": 389
+    },
+    {
+      "epoch": 1.3928571428571428,
+      "grad_norm": 0.8812011195548888,
+      "learning_rate": 3.983537739051776e-05,
+      "loss": 0.0976,
+      "step": 390
+    },
+    {
+      "epoch": 1.3964285714285714,
+      "grad_norm": 1.444613959054698,
+      "learning_rate": 3.983389302462021e-05,
+      "loss": 0.1092,
+      "step": 391
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.342766866635371,
+      "learning_rate": 3.983240202452851e-05,
+      "loss": 0.1131,
+      "step": 392
+    },
+    {
+      "epoch": 1.4035714285714285,
+      "grad_norm": 1.0532177325963952,
+      "learning_rate": 3.983090439074137e-05,
+      "loss": 0.0838,
+      "step": 393
+    },
+    {
+      "epoch": 1.407142857142857,
+      "grad_norm": 0.582675834340328,
+      "learning_rate": 3.982940012375974e-05,
+      "loss": 0.0693,
+      "step": 394
+    },
+    {
+      "epoch": 1.4107142857142856,
+      "grad_norm": 0.5053832596458228,
+      "learning_rate": 3.982788922408678e-05,
+      "loss": 0.0748,
+      "step": 395
+    },
+    {
+      "epoch": 1.4142857142857144,
+      "grad_norm": 1.4830417485566878,
+      "learning_rate": 3.982637169222786e-05,
+      "loss": 0.0541,
+      "step": 396
+    },
+    {
+      "epoch": 1.417857142857143,
+      "grad_norm": 1.1475644391891535,
+      "learning_rate": 3.982484752869058e-05,
+      "loss": 0.0711,
+      "step": 397
+    },
+    {
+      "epoch": 1.4214285714285715,
+      "grad_norm": 0.9654287570695997,
+      "learning_rate": 3.982331673398476e-05,
+      "loss": 0.0806,
+      "step": 398
+    },
+    {
+      "epoch": 1.425,
+      "grad_norm": 0.6481885480387872,
+      "learning_rate": 3.982177930862242e-05,
+      "loss": 0.0706,
+      "step": 399
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 1.4320219768090863,
+      "learning_rate": 3.9820235253117817e-05,
+      "loss": 0.1333,
+      "step": 400
+    },
+    {
+      "epoch": 1.4321428571428572,
+      "grad_norm": 0.9210504805059367,
+      "learning_rate": 3.981868456798742e-05,
+      "loss": 0.1386,
+      "step": 401
+    },
+    {
+      "epoch": 1.4357142857142857,
+      "grad_norm": 0.9486631225748333,
+      "learning_rate": 3.981712725374991e-05,
+      "loss": 0.0876,
+      "step": 402
+    },
+    {
+      "epoch": 1.4392857142857143,
+      "grad_norm": 1.1810223433255438,
+      "learning_rate": 3.9815563310926195e-05,
+      "loss": 0.0936,
+      "step": 403
+    },
+    {
+      "epoch": 1.4428571428571428,
+      "grad_norm": 0.8843009653542324,
+      "learning_rate": 3.9813992740039384e-05,
+      "loss": 0.1084,
+      "step": 404
+    },
+    {
+      "epoch": 1.4464285714285714,
+      "grad_norm": 1.125607855197495,
+      "learning_rate": 3.981241554161483e-05,
+      "loss": 0.0991,
+      "step": 405
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.2628347148484842,
+      "learning_rate": 3.981083171618007e-05,
+      "loss": 0.0829,
+      "step": 406
+    },
+    {
+      "epoch": 1.4535714285714285,
+      "grad_norm": 1.2137035317357656,
+      "learning_rate": 3.9809241264264885e-05,
+      "loss": 0.0931,
+      "step": 407
+    },
+    {
+      "epoch": 1.457142857142857,
+      "grad_norm": 1.2561306303135797,
+      "learning_rate": 3.980764418640125e-05,
+      "loss": 0.0732,
+      "step": 408
+    },
+    {
+      "epoch": 1.4607142857142856,
+      "grad_norm": 0.9754601432438966,
+      "learning_rate": 3.980604048312339e-05,
+      "loss": 0.0576,
+      "step": 409
+    },
+    {
+      "epoch": 1.4642857142857144,
+      "grad_norm": 0.7391540764275938,
+      "learning_rate": 3.98044301549677e-05,
+      "loss": 0.1365,
+      "step": 410
+    },
+    {
+      "epoch": 1.467857142857143,
+      "grad_norm": 0.6163859315725281,
+      "learning_rate": 3.980281320247283e-05,
+      "loss": 0.0438,
+      "step": 411
+    },
+    {
+      "epoch": 1.4714285714285715,
+      "grad_norm": 0.4050818790354135,
+      "learning_rate": 3.980118962617963e-05,
+      "loss": 0.0514,
+      "step": 412
+    },
+    {
+      "epoch": 1.475,
+      "grad_norm": 1.108178221637504,
+      "learning_rate": 3.9799559426631155e-05,
+      "loss": 0.1016,
+      "step": 413
+    },
+    {
+      "epoch": 1.4785714285714286,
+      "grad_norm": 1.206506439016465,
+      "learning_rate": 3.97979226043727e-05,
+      "loss": 0.0689,
+      "step": 414
+    },
+    {
+      "epoch": 1.4821428571428572,
+      "grad_norm": 0.47463567098843223,
+      "learning_rate": 3.979627915995175e-05,
+      "loss": 0.0786,
+      "step": 415
+    },
+    {
+      "epoch": 1.4857142857142858,
+      "grad_norm": 1.3906848356649029,
+      "learning_rate": 3.979462909391803e-05,
+      "loss": 0.118,
+      "step": 416
+    },
+    {
+      "epoch": 1.4892857142857143,
+      "grad_norm": 0.6975554975158951,
+      "learning_rate": 3.979297240682345e-05,
+      "loss": 0.0802,
+      "step": 417
+    },
+    {
+      "epoch": 1.4928571428571429,
+      "grad_norm": 1.238414986981063,
+      "learning_rate": 3.9791309099222176e-05,
+      "loss": 0.129,
+      "step": 418
+    },
+    {
+      "epoch": 1.4964285714285714,
+      "grad_norm": 1.3007883867864174,
+      "learning_rate": 3.978963917167054e-05,
+      "loss": 0.1267,
+      "step": 419
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.7409909880013532,
+      "learning_rate": 3.9787962624727126e-05,
+      "loss": 0.1531,
+      "step": 420
+    },
+    {
+      "epoch": 1.5035714285714286,
+      "grad_norm": 0.30615883661854043,
+      "learning_rate": 3.978627945895271e-05,
+      "loss": 0.0633,
+      "step": 421
+    },
+    {
+      "epoch": 1.5071428571428571,
+      "grad_norm": 0.827295950768099,
+      "learning_rate": 3.97845896749103e-05,
+      "loss": 0.072,
+      "step": 422
+    },
+    {
+      "epoch": 1.5107142857142857,
+      "grad_norm": 1.4059892994015082,
+      "learning_rate": 3.978289327316509e-05,
+      "loss": 0.1441,
+      "step": 423
+    },
+    {
+      "epoch": 1.5142857142857142,
+      "grad_norm": 1.4884933362666437,
+      "learning_rate": 3.978119025428454e-05,
+      "loss": 0.0896,
+      "step": 424
+    },
+    {
+      "epoch": 1.5178571428571428,
+      "grad_norm": 1.45999497065778,
+      "learning_rate": 3.977948061883825e-05,
+      "loss": 0.1124,
+      "step": 425
+    },
+    {
+      "epoch": 1.5214285714285714,
+      "grad_norm": 1.6964310877657691,
+      "learning_rate": 3.977776436739809e-05,
+      "loss": 0.087,
+      "step": 426
+    },
+    {
+      "epoch": 1.525,
+      "grad_norm": 1.1068095975490757,
+      "learning_rate": 3.9776041500538134e-05,
+      "loss": 0.0791,
+      "step": 427
+    },
+    {
+      "epoch": 1.5285714285714285,
+      "grad_norm": 0.4247884918899206,
+      "learning_rate": 3.9774312018834654e-05,
+      "loss": 0.061,
+      "step": 428
+    },
+    {
+      "epoch": 1.532142857142857,
+      "grad_norm": 0.3051147766651246,
+      "learning_rate": 3.9772575922866124e-05,
+      "loss": 0.0592,
+      "step": 429
+    },
+    {
+      "epoch": 1.5357142857142856,
+      "grad_norm": 0.8811723952245779,
+      "learning_rate": 3.977083321321326e-05,
+      "loss": 0.1029,
+      "step": 430
+    },
+    {
+      "epoch": 1.5392857142857141,
+      "grad_norm": 0.3656994953569929,
+      "learning_rate": 3.976908389045898e-05,
+      "loss": 0.0165,
+      "step": 431
+    },
+    {
+      "epoch": 1.5428571428571427,
+      "grad_norm": 0.506266135967698,
+      "learning_rate": 3.976732795518842e-05,
+      "loss": 0.0568,
+      "step": 432
+    },
+    {
+      "epoch": 1.5464285714285713,
+      "grad_norm": 0.966255966550785,
+      "learning_rate": 3.976556540798889e-05,
+      "loss": 0.0757,
+      "step": 433
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.042983892930037,
+      "learning_rate": 3.976379624944996e-05,
+      "loss": 0.1276,
+      "step": 434
+    },
+    {
+      "epoch": 1.5535714285714286,
+      "grad_norm": 1.1177977260913634,
+      "learning_rate": 3.976202048016339e-05,
+      "loss": 0.1064,
+      "step": 435
+    },
+    {
+      "epoch": 1.5571428571428572,
+      "grad_norm": 0.9429731104857307,
+      "learning_rate": 3.976023810072315e-05,
+      "loss": 0.1042,
+      "step": 436
+    },
+    {
+      "epoch": 1.5607142857142857,
+      "grad_norm": 0.29549097879316927,
+      "learning_rate": 3.975844911172544e-05,
+      "loss": 0.0252,
+      "step": 437
+    },
+    {
+      "epoch": 1.5642857142857143,
+      "grad_norm": 1.5248006048036924,
+      "learning_rate": 3.975665351376862e-05,
+      "loss": 0.1204,
+      "step": 438
+    },
+    {
+      "epoch": 1.5678571428571428,
+      "grad_norm": 0.8645818165039436,
+      "learning_rate": 3.975485130745332e-05,
+      "loss": 0.104,
+      "step": 439
+    },
+    {
+      "epoch": 1.5714285714285714,
+      "grad_norm": 0.4134110971716588,
+      "learning_rate": 3.9753042493382365e-05,
+      "loss": 0.0953,
+      "step": 440
+    },
+    {
+      "epoch": 1.575,
+      "grad_norm": 0.8630002140025468,
+      "learning_rate": 3.975122707216076e-05,
+      "loss": 0.0925,
+      "step": 441
+    },
+    {
+      "epoch": 1.5785714285714287,
+      "grad_norm": 0.7446336142594386,
+      "learning_rate": 3.974940504439575e-05,
+      "loss": 0.0965,
+      "step": 442
+    },
+    {
+      "epoch": 1.5821428571428573,
+      "grad_norm": 1.0680284032132228,
+      "learning_rate": 3.9747576410696786e-05,
+      "loss": 0.103,
+      "step": 443
+    },
+    {
+      "epoch": 1.5857142857142859,
+      "grad_norm": 1.0529909953512822,
+      "learning_rate": 3.974574117167551e-05,
+      "loss": 0.0966,
+      "step": 444
+    },
+    {
+      "epoch": 1.5892857142857144,
+      "grad_norm": 0.610898874664773,
+      "learning_rate": 3.97438993279458e-05,
+      "loss": 0.0823,
+      "step": 445
+    },
+    {
+      "epoch": 1.592857142857143,
+      "grad_norm": 0.4905877051586956,
+      "learning_rate": 3.974205088012372e-05,
+      "loss": 0.0796,
+      "step": 446
+    },
+    {
+      "epoch": 1.5964285714285715,
+      "grad_norm": 0.6992079224171457,
+      "learning_rate": 3.974019582882757e-05,
+      "loss": 0.068,
+      "step": 447
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.6269136580845739,
+      "learning_rate": 3.9738334174677816e-05,
+      "loss": 0.0896,
+      "step": 448
+    },
+    {
+      "epoch": 1.6035714285714286,
+      "grad_norm": 0.8760536456809249,
+      "learning_rate": 3.973646591829718e-05,
+      "loss": 0.1063,
+      "step": 449
+    },
+    {
+      "epoch": 1.6071428571428572,
+      "grad_norm": 0.3948628963679801,
+      "learning_rate": 3.973459106031056e-05,
+      "loss": 0.0751,
+      "step": 450
+    },
+    {
+      "epoch": 1.6107142857142858,
+      "grad_norm": 0.9894037801549839,
+      "learning_rate": 3.973270960134508e-05,
+      "loss": 0.1035,
+      "step": 451
+    },
+    {
+      "epoch": 1.6142857142857143,
+      "grad_norm": 0.6398652662095126,
+      "learning_rate": 3.973082154203006e-05,
+      "loss": 0.1015,
+      "step": 452
+    },
+    {
+      "epoch": 1.6178571428571429,
+      "grad_norm": 1.1342888357456913,
+      "learning_rate": 3.9728926882997034e-05,
+      "loss": 0.1529,
+      "step": 453
+    },
+    {
+      "epoch": 1.6214285714285714,
+      "grad_norm": 1.1179973178512337,
+      "learning_rate": 3.972702562487975e-05,
+      "loss": 0.1191,
+      "step": 454
+    },
+    {
+      "epoch": 1.625,
+      "grad_norm": 0.9247583052767986,
+      "learning_rate": 3.9725117768314133e-05,
+      "loss": 0.132,
+      "step": 455
+    },
+    {
+      "epoch": 1.6285714285714286,
+      "grad_norm": 0.3570117265615051,
+      "learning_rate": 3.972320331393837e-05,
+      "loss": 0.0737,
+      "step": 456
+    },
+    {
+      "epoch": 1.6321428571428571,
+      "grad_norm": 0.7194249654158054,
+      "learning_rate": 3.9721282262392795e-05,
+      "loss": 0.0678,
+      "step": 457
+    },
+    {
+      "epoch": 1.6357142857142857,
+      "grad_norm": 0.5195033476736866,
+      "learning_rate": 3.971935461431999e-05,
+      "loss": 0.0743,
+      "step": 458
+    },
+    {
+      "epoch": 1.6392857142857142,
+      "grad_norm": 0.6515903796253151,
+      "learning_rate": 3.971742037036472e-05,
+      "loss": 0.1099,
+      "step": 459
+    },
+    {
+      "epoch": 1.6428571428571428,
+      "grad_norm": 1.001627771127134,
+      "learning_rate": 3.971547953117398e-05,
+      "loss": 0.0591,
+      "step": 460
+    },
+    {
+      "epoch": 1.6464285714285714,
+      "grad_norm": 0.4674606696720902,
+      "learning_rate": 3.971353209739694e-05,
+      "loss": 0.0919,
+      "step": 461
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.47648458078650646,
+      "learning_rate": 3.971157806968501e-05,
+      "loss": 0.0884,
+      "step": 462
+    },
+    {
+      "epoch": 1.6535714285714285,
+      "grad_norm": 0.7387907972154028,
+      "learning_rate": 3.970961744869177e-05,
+      "loss": 0.0918,
+      "step": 463
+    },
+    {
+      "epoch": 1.657142857142857,
+      "grad_norm": 0.5337698131273113,
+      "learning_rate": 3.9707650235073045e-05,
+      "loss": 0.0949,
+      "step": 464
+    },
+    {
+      "epoch": 1.6607142857142856,
+      "grad_norm": 0.6551818606783406,
+      "learning_rate": 3.970567642948683e-05,
+      "loss": 0.0574,
+      "step": 465
+    },
+    {
+      "epoch": 1.6642857142857141,
+      "grad_norm": 0.29387636867272143,
+      "learning_rate": 3.970369603259334e-05,
+      "loss": 0.0964,
+      "step": 466
+    },
+    {
+      "epoch": 1.6678571428571427,
+      "grad_norm": 0.521464082858247,
+      "learning_rate": 3.970170904505499e-05,
+      "loss": 0.1113,
+      "step": 467
+    },
+    {
+      "epoch": 1.6714285714285713,
+      "grad_norm": 0.5970874838299207,
+      "learning_rate": 3.969971546753641e-05,
+      "loss": 0.0824,
+      "step": 468
+    },
+    {
+      "epoch": 1.675,
+      "grad_norm": 0.7911440664816791,
+      "learning_rate": 3.969771530070443e-05,
+      "loss": 0.1442,
+      "step": 469
+    },
+    {
+      "epoch": 1.6785714285714286,
+      "grad_norm": 0.6333580768169138,
+      "learning_rate": 3.9695708545228076e-05,
+      "loss": 0.0892,
+      "step": 470
+    },
+    {
+      "epoch": 1.6821428571428572,
+      "grad_norm": 0.36649732759947323,
+      "learning_rate": 3.969369520177858e-05,
+      "loss": 0.0691,
+      "step": 471
+    },
+    {
+      "epoch": 1.6857142857142857,
+      "grad_norm": 0.5835696376307341,
+      "learning_rate": 3.9691675271029384e-05,
+      "loss": 0.0928,
+      "step": 472
+    },
+    {
+      "epoch": 1.6892857142857143,
+      "grad_norm": 0.47999349088457244,
+      "learning_rate": 3.968964875365613e-05,
+      "loss": 0.0538,
+      "step": 473
+    },
+    {
+      "epoch": 1.6928571428571428,
+      "grad_norm": 0.6030471125733031,
+      "learning_rate": 3.968761565033666e-05,
+      "loss": 0.0811,
+      "step": 474
+    },
+    {
+      "epoch": 1.6964285714285714,
+      "grad_norm": 0.615781898024002,
+      "learning_rate": 3.968557596175103e-05,
+      "loss": 0.067,
+      "step": 475
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6123951297281729,
+      "learning_rate": 3.968352968858149e-05,
+      "loss": 0.1035,
+      "step": 476
+    },
+    {
+      "epoch": 1.7035714285714287,
+      "grad_norm": 0.8935309845192123,
+      "learning_rate": 3.968147683151248e-05,
+      "loss": 0.0996,
+      "step": 477
+    },
+    {
+      "epoch": 1.7071428571428573,
+      "grad_norm": 0.44742215555320913,
+      "learning_rate": 3.967941739123068e-05,
+      "loss": 0.0885,
+      "step": 478
+    },
+    {
+      "epoch": 1.7107142857142859,
+      "grad_norm": 0.7264672607972537,
+      "learning_rate": 3.9677351368424915e-05,
+      "loss": 0.0883,
+      "step": 479
+    },
+    {
+      "epoch": 1.7142857142857144,
+      "grad_norm": 0.4715617599312736,
+      "learning_rate": 3.9675278763786274e-05,
+      "loss": 0.0759,
+      "step": 480
+    },
+    {
+      "epoch": 1.717857142857143,
+      "grad_norm": 0.6356710667686848,
+      "learning_rate": 3.967319957800801e-05,
+      "loss": 0.0567,
+      "step": 481
+    },
+    {
+      "epoch": 1.7214285714285715,
+      "grad_norm": 0.6680959587728996,
+      "learning_rate": 3.9671113811785564e-05,
+      "loss": 0.0899,
+      "step": 482
+    },
+    {
+      "epoch": 1.725,
+      "grad_norm": 0.7135311753381732,
+      "learning_rate": 3.966902146581663e-05,
+      "loss": 0.1104,
+      "step": 483
+    },
+    {
+      "epoch": 1.7285714285714286,
+      "grad_norm": 1.8329574531235893,
+      "learning_rate": 3.966692254080106e-05,
+      "loss": 0.0999,
+      "step": 484
+    },
+    {
+      "epoch": 1.7321428571428572,
+      "grad_norm": 0.3662287747489552,
+      "learning_rate": 3.966481703744091e-05,
+      "loss": 0.0655,
+      "step": 485
+    },
+    {
+      "epoch": 1.7357142857142858,
+      "grad_norm": 0.6697647366308543,
+      "learning_rate": 3.966270495644045e-05,
+      "loss": 0.0709,
+      "step": 486
+    },
+    {
+      "epoch": 1.7392857142857143,
+      "grad_norm": 0.5180756619512662,
+      "learning_rate": 3.966058629850615e-05,
+      "loss": 0.09,
+      "step": 487
+    },
+    {
+      "epoch": 1.7428571428571429,
+      "grad_norm": 0.9099970762030544,
+      "learning_rate": 3.9658461064346674e-05,
+      "loss": 0.102,
+      "step": 488
+    },
+    {
+      "epoch": 1.7464285714285714,
+      "grad_norm": 0.6995175653162985,
+      "learning_rate": 3.965632925467289e-05,
+      "loss": 0.0635,
+      "step": 489
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.0196060537683398,
+      "learning_rate": 3.965419087019785e-05,
+      "loss": 0.1328,
+      "step": 490
+    },
+    {
+      "epoch": 1.7535714285714286,
+      "grad_norm": 0.7875412232074495,
+      "learning_rate": 3.965204591163683e-05,
+      "loss": 0.1016,
+      "step": 491
+    },
+    {
+      "epoch": 1.7571428571428571,
+      "grad_norm": 0.590694576262596,
+      "learning_rate": 3.9649894379707284e-05,
+      "loss": 0.123,
+      "step": 492
+    },
+    {
+      "epoch": 1.7607142857142857,
+      "grad_norm": 0.5065091983260875,
+      "learning_rate": 3.964773627512888e-05,
+      "loss": 0.0947,
+      "step": 493
+    },
+    {
+      "epoch": 1.7642857142857142,
+      "grad_norm": 0.5830411896963708,
+      "learning_rate": 3.964557159862347e-05,
+      "loss": 0.0503,
+      "step": 494
+    },
+    {
+      "epoch": 1.7678571428571428,
+      "grad_norm": 0.9030098210081953,
+      "learning_rate": 3.964340035091511e-05,
+      "loss": 0.0973,
+      "step": 495
+    },
+    {
+      "epoch": 1.7714285714285714,
+      "grad_norm": 0.8776821608769888,
+      "learning_rate": 3.964122253273007e-05,
+      "loss": 0.0779,
+      "step": 496
+    },
+    {
+      "epoch": 1.775,
+      "grad_norm": 0.5417586231714271,
+      "learning_rate": 3.9639038144796794e-05,
+      "loss": 0.0704,
+      "step": 497
+    },
+    {
+      "epoch": 1.7785714285714285,
+      "grad_norm": 0.516444432421725,
+      "learning_rate": 3.9636847187845936e-05,
+      "loss": 0.0783,
+      "step": 498
+    },
+    {
+      "epoch": 1.782142857142857,
+      "grad_norm": 0.5568575508798261,
+      "learning_rate": 3.963464966261034e-05,
+      "loss": 0.0563,
+      "step": 499
+    },
+    {
+      "epoch": 1.7857142857142856,
+      "grad_norm": 1.0831346146505347,
+      "learning_rate": 3.963244556982505e-05,
+      "loss": 0.098,
+      "step": 500
+    },
+    {
+      "epoch": 1.7892857142857141,
+      "grad_norm": 0.6337505827337594,
+      "learning_rate": 3.9630234910227316e-05,
+      "loss": 0.0972,
+      "step": 501
+    },
+    {
+      "epoch": 1.7928571428571427,
+      "grad_norm": 0.9639209794928328,
+      "learning_rate": 3.962801768455657e-05,
+      "loss": 0.0552,
+      "step": 502
+    },
+    {
+      "epoch": 1.7964285714285713,
+      "grad_norm": 0.719333526549779,
+      "learning_rate": 3.9625793893554447e-05,
+      "loss": 0.0576,
+      "step": 503
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5005248932269417,
+      "learning_rate": 3.9623563537964784e-05,
+      "loss": 0.0406,
+      "step": 504
+    },
+    {
+      "epoch": 1.8035714285714286,
+      "grad_norm": 0.7968105437352502,
+      "learning_rate": 3.96213266185336e-05,
+      "loss": 0.0843,
+      "step": 505
+    },
+    {
+      "epoch": 1.8071428571428572,
+      "grad_norm": 1.0948001788859887,
+      "learning_rate": 3.961908313600912e-05,
+      "loss": 0.1022,
+      "step": 506
+    },
+    {
+      "epoch": 1.8107142857142857,
+      "grad_norm": 0.6664500019524523,
+      "learning_rate": 3.961683309114176e-05,
+      "loss": 0.0829,
+      "step": 507
+    },
+    {
+      "epoch": 1.8142857142857143,
+      "grad_norm": 0.4171606812293665,
+      "learning_rate": 3.9614576484684144e-05,
+      "loss": 0.0655,
+      "step": 508
+    },
+    {
+      "epoch": 1.8178571428571428,
+      "grad_norm": 0.5066741030255496,
+      "learning_rate": 3.961231331739106e-05,
+      "loss": 0.0836,
+      "step": 509
+    },
+    {
+      "epoch": 1.8214285714285714,
+      "grad_norm": 0.948123776416021,
+      "learning_rate": 3.961004359001952e-05,
+      "loss": 0.0938,
+      "step": 510
+    },
+    {
+      "epoch": 1.825,
+      "grad_norm": 0.7444628370471363,
+      "learning_rate": 3.960776730332872e-05,
+      "loss": 0.0649,
+      "step": 511
+    },
+    {
+      "epoch": 1.8285714285714287,
+      "grad_norm": 0.46320985249074526,
+      "learning_rate": 3.9605484458080055e-05,
+      "loss": 0.0826,
+      "step": 512
+    },
+    {
+      "epoch": 1.8321428571428573,
+      "grad_norm": 0.9685903813386301,
+      "learning_rate": 3.96031950550371e-05,
+      "loss": 0.0493,
+      "step": 513
+    },
+    {
+      "epoch": 1.8357142857142859,
+      "grad_norm": 0.9053887303605164,
+      "learning_rate": 3.960089909496563e-05,
+      "loss": 0.0751,
+      "step": 514
+    },
+    {
+      "epoch": 1.8392857142857144,
+      "grad_norm": 0.5091118961422333,
+      "learning_rate": 3.9598596578633624e-05,
+      "loss": 0.0942,
+      "step": 515
+    },
+    {
+      "epoch": 1.842857142857143,
+      "grad_norm": 0.6997959239033796,
+      "learning_rate": 3.959628750681124e-05,
+      "loss": 0.0791,
+      "step": 516
+    },
+    {
+      "epoch": 1.8464285714285715,
+      "grad_norm": 0.48629943340402765,
+      "learning_rate": 3.959397188027083e-05,
+      "loss": 0.035,
+      "step": 517
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5344325568326405,
+      "learning_rate": 3.9591649699786965e-05,
+      "loss": 0.1079,
+      "step": 518
+    },
+    {
+      "epoch": 1.8535714285714286,
+      "grad_norm": 1.1054255371437887,
+      "learning_rate": 3.958932096613636e-05,
+      "loss": 0.0955,
+      "step": 519
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.519575768616812,
+      "learning_rate": 3.958698568009796e-05,
+      "loss": 0.0796,
+      "step": 520
+    },
+    {
+      "epoch": 1.8607142857142858,
+      "grad_norm": 1.2980512245744011,
+      "learning_rate": 3.958464384245288e-05,
+      "loss": 0.051,
+      "step": 521
+    },
+    {
+      "epoch": 1.8642857142857143,
+      "grad_norm": 0.6213683679397631,
+      "learning_rate": 3.9582295453984454e-05,
+      "loss": 0.0453,
+      "step": 522
+    },
+    {
+      "epoch": 1.8678571428571429,
+      "grad_norm": 0.7023553253528808,
+      "learning_rate": 3.9579940515478176e-05,
+      "loss": 0.0847,
+      "step": 523
+    },
+    {
+      "epoch": 1.8714285714285714,
+      "grad_norm": 1.4564355717871746,
+      "learning_rate": 3.957757902772175e-05,
+      "loss": 0.1615,
+      "step": 524
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.8163730166772489,
+      "learning_rate": 3.9575210991505056e-05,
+      "loss": 0.0807,
+      "step": 525
+    },
+    {
+      "epoch": 1.8785714285714286,
+      "grad_norm": 0.5177151496643684,
+      "learning_rate": 3.957283640762018e-05,
+      "loss": 0.1103,
+      "step": 526
+    },
+    {
+      "epoch": 1.8821428571428571,
+      "grad_norm": 1.0024778437826498,
+      "learning_rate": 3.9570455276861386e-05,
+      "loss": 0.08,
+      "step": 527
+    },
+    {
+      "epoch": 1.8857142857142857,
+      "grad_norm": 0.6591600451368589,
+      "learning_rate": 3.956806760002514e-05,
+      "loss": 0.1093,
+      "step": 528
+    },
+    {
+      "epoch": 1.8892857142857142,
+      "grad_norm": 0.9189492378786976,
+      "learning_rate": 3.956567337791009e-05,
+      "loss": 0.1142,
+      "step": 529
+    },
+    {
+      "epoch": 1.8928571428571428,
+      "grad_norm": 1.0609827947736592,
+      "learning_rate": 3.9563272611317066e-05,
+      "loss": 0.0677,
+      "step": 530
+    },
+    {
+      "epoch": 1.8964285714285714,
+      "grad_norm": 0.8011196782068092,
+      "learning_rate": 3.95608653010491e-05,
+      "loss": 0.1074,
+      "step": 531
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.7318658258475825,
+      "learning_rate": 3.955845144791142e-05,
+      "loss": 0.098,
+      "step": 532
+    },
+    {
+      "epoch": 1.9035714285714285,
+      "grad_norm": 0.35419126214226976,
+      "learning_rate": 3.9556031052711406e-05,
+      "loss": 0.0477,
+      "step": 533
+    },
+    {
+      "epoch": 1.907142857142857,
+      "grad_norm": 0.47734110467317475,
+      "learning_rate": 3.955360411625867e-05,
+      "loss": 0.0485,
+      "step": 534
+    },
+    {
+      "epoch": 1.9107142857142856,
+      "grad_norm": 0.8332851174201977,
+      "learning_rate": 3.955117063936498e-05,
+      "loss": 0.0802,
+      "step": 535
+    },
+    {
+      "epoch": 1.9142857142857141,
+      "grad_norm": 0.5625694207662685,
+      "learning_rate": 3.954873062284431e-05,
+      "loss": 0.0909,
+      "step": 536
+    },
+    {
+      "epoch": 1.9178571428571427,
+      "grad_norm": 1.1831879729570653,
+      "learning_rate": 3.9546284067512816e-05,
+      "loss": 0.1416,
+      "step": 537
+    },
+    {
+      "epoch": 1.9214285714285713,
+      "grad_norm": 0.977803783283507,
+      "learning_rate": 3.9543830974188837e-05,
+      "loss": 0.1132,
+      "step": 538
+    },
+    {
+      "epoch": 1.925,
+      "grad_norm": 0.5857679902693327,
+      "learning_rate": 3.954137134369292e-05,
+      "loss": 0.0932,
+      "step": 539
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.6312733124343726,
+      "learning_rate": 3.953890517684775e-05,
+      "loss": 0.0764,
+      "step": 540
+    },
+    {
+      "epoch": 1.9321428571428572,
+      "grad_norm": 0.5567332135205346,
+      "learning_rate": 3.9536432474478256e-05,
+      "loss": 0.0517,
+      "step": 541
+    },
+    {
+      "epoch": 1.9357142857142857,
+      "grad_norm": 0.8878740815327976,
+      "learning_rate": 3.953395323741151e-05,
+      "loss": 0.0901,
+      "step": 542
+    },
+    {
+      "epoch": 1.9392857142857143,
+      "grad_norm": 0.5502224457021146,
+      "learning_rate": 3.95314674664768e-05,
+      "loss": 0.0681,
+      "step": 543
+    },
+    {
+      "epoch": 1.9428571428571428,
+      "grad_norm": 0.989940660163851,
+      "learning_rate": 3.9528975162505574e-05,
+      "loss": 0.0854,
+      "step": 544
+    },
+    {
+      "epoch": 1.9464285714285714,
+      "grad_norm": 0.7108268511988862,
+      "learning_rate": 3.952647632633149e-05,
+      "loss": 0.0888,
+      "step": 545
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.932163992430506,
+      "learning_rate": 3.952397095879036e-05,
+      "loss": 0.0722,
+      "step": 546
+    },
+    {
+      "epoch": 1.9535714285714287,
+      "grad_norm": 0.9633480632074612,
+      "learning_rate": 3.952145906072022e-05,
+      "loss": 0.1058,
+      "step": 547
+    },
+    {
+      "epoch": 1.9571428571428573,
+      "grad_norm": 1.048124174855216,
+      "learning_rate": 3.9518940632961255e-05,
+      "loss": 0.1311,
+      "step": 548
+    },
+    {
+      "epoch": 1.9607142857142859,
+      "grad_norm": 0.5967223824086821,
+      "learning_rate": 3.951641567635585e-05,
+      "loss": 0.06,
+      "step": 549
+    },
+    {
+      "epoch": 1.9642857142857144,
+      "grad_norm": 0.587501963351525,
+      "learning_rate": 3.951388419174857e-05,
+      "loss": 0.0774,
+      "step": 550
+    },
+    {
+      "epoch": 1.967857142857143,
+      "grad_norm": 0.8089546085180309,
+      "learning_rate": 3.9511346179986175e-05,
+      "loss": 0.0981,
+      "step": 551
+    },
+    {
+      "epoch": 1.9714285714285715,
+      "grad_norm": 0.6094358356701864,
+      "learning_rate": 3.9508801641917596e-05,
+      "loss": 0.12,
+      "step": 552
+    },
+    {
+      "epoch": 1.975,
+      "grad_norm": 0.7976995203477562,
+      "learning_rate": 3.950625057839394e-05,
+      "loss": 0.0847,
+      "step": 553
+    },
+    {
+      "epoch": 1.9785714285714286,
+      "grad_norm": 0.3890469906984489,
+      "learning_rate": 3.950369299026852e-05,
+      "loss": 0.0479,
+      "step": 554
+    },
+    {
+      "epoch": 1.9821428571428572,
+      "grad_norm": 0.570908966493566,
+      "learning_rate": 3.950112887839681e-05,
+      "loss": 0.0695,
+      "step": 555
+    },
+    {
+      "epoch": 1.9857142857142858,
+      "grad_norm": 0.8625987660927652,
+      "learning_rate": 3.949855824363647e-05,
+      "loss": 0.0512,
+      "step": 556
+    },
+    {
+      "epoch": 1.9892857142857143,
+      "grad_norm": 0.6901195801560999,
+      "learning_rate": 3.949598108684736e-05,
+      "loss": 0.0834,
+      "step": 557
+    },
+    {
+      "epoch": 1.9928571428571429,
+      "grad_norm": 0.506672362878381,
+      "learning_rate": 3.9493397408891497e-05,
+      "loss": 0.0875,
+      "step": 558
+    },
+    {
+      "epoch": 1.9964285714285714,
+      "grad_norm": 0.8258034827675103,
+      "learning_rate": 3.949080721063308e-05,
+      "loss": 0.0981,
+      "step": 559
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8895943830405975,
+      "learning_rate": 3.948821049293853e-05,
+      "loss": 0.072,
+      "step": 560
+    },
+    {
+      "epoch": 2.0035714285714286,
+      "grad_norm": 0.5812890708804186,
+      "learning_rate": 3.948560725667638e-05,
+      "loss": 0.0601,
+      "step": 561
+    },
+    {
+      "epoch": 2.007142857142857,
+      "grad_norm": 0.5262661724008607,
+      "learning_rate": 3.94829975027174e-05,
+      "loss": 0.0726,
+      "step": 562
+    },
+    {
+      "epoch": 2.0107142857142857,
+      "grad_norm": 0.6185420666869119,
+      "learning_rate": 3.9480381231934525e-05,
+      "loss": 0.0434,
+      "step": 563
+    },
+    {
+      "epoch": 2.0142857142857142,
+      "grad_norm": 0.3993443980046951,
+      "learning_rate": 3.947775844520285e-05,
+      "loss": 0.0583,
+      "step": 564
+    },
+    {
+      "epoch": 2.017857142857143,
+      "grad_norm": 0.6638803209589612,
+      "learning_rate": 3.947512914339968e-05,
+      "loss": 0.0683,
+      "step": 565
+    },
+    {
+      "epoch": 2.0214285714285714,
+      "grad_norm": 0.9928475153170945,
+      "learning_rate": 3.9472493327404485e-05,
+      "loss": 0.1333,
+      "step": 566
+    },
+    {
+      "epoch": 2.025,
+      "grad_norm": 0.5669061833762513,
+      "learning_rate": 3.946985099809889e-05,
+      "loss": 0.0743,
+      "step": 567
+    },
+    {
+      "epoch": 2.0285714285714285,
+      "grad_norm": 0.9234711866077135,
+      "learning_rate": 3.9467202156366754e-05,
+      "loss": 0.0904,
+      "step": 568
+    },
+    {
+      "epoch": 2.032142857142857,
+      "grad_norm": 0.765826870673619,
+      "learning_rate": 3.9464546803094056e-05,
+      "loss": 0.0462,
+      "step": 569
+    },
+    {
+      "epoch": 2.0357142857142856,
+      "grad_norm": 0.50390083661875,
+      "learning_rate": 3.946188493916899e-05,
+      "loss": 0.0807,
+      "step": 570
+    },
+    {
+      "epoch": 2.039285714285714,
+      "grad_norm": 0.6762953900263295,
+      "learning_rate": 3.945921656548191e-05,
+      "loss": 0.0793,
+      "step": 571
+    },
+    {
+      "epoch": 2.0428571428571427,
+      "grad_norm": 0.6163694313837468,
+      "learning_rate": 3.945654168292538e-05,
+      "loss": 0.0594,
+      "step": 572
+    },
+    {
+      "epoch": 2.0464285714285713,
+      "grad_norm": 1.2421021027036525,
+      "learning_rate": 3.9453860292394075e-05,
+      "loss": 0.1097,
+      "step": 573
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.4424567902837839,
+      "learning_rate": 3.945117239478492e-05,
+      "loss": 0.0779,
+      "step": 574
+    },
+    {
+      "epoch": 2.0535714285714284,
+      "grad_norm": 0.7603636911125126,
+      "learning_rate": 3.9448477990996975e-05,
+      "loss": 0.0995,
+      "step": 575
+    },
+    {
+      "epoch": 2.057142857142857,
+      "grad_norm": 0.6845509680748996,
+      "learning_rate": 3.9445777081931476e-05,
+      "loss": 0.0576,
+      "step": 576
+    },
+    {
+      "epoch": 2.0607142857142855,
+      "grad_norm": 0.8857248618966495,
+      "learning_rate": 3.944306966849185e-05,
+      "loss": 0.0575,
+      "step": 577
+    },
+    {
+      "epoch": 2.064285714285714,
+      "grad_norm": 0.8306122409659006,
+      "learning_rate": 3.944035575158369e-05,
+      "loss": 0.0719,
+      "step": 578
+    },
+    {
+      "epoch": 2.067857142857143,
+      "grad_norm": 0.9076074805539973,
+      "learning_rate": 3.9437635332114774e-05,
+      "loss": 0.0931,
+      "step": 579
+    },
+    {
+      "epoch": 2.0714285714285716,
+      "grad_norm": 0.9173961487919557,
+      "learning_rate": 3.943490841099505e-05,
+      "loss": 0.0849,
+      "step": 580
+    },
+    {
+      "epoch": 2.075,
+      "grad_norm": 0.3809440922813365,
+      "learning_rate": 3.9432174989136636e-05,
+      "loss": 0.0824,
+      "step": 581
+    },
+    {
+      "epoch": 2.0785714285714287,
+      "grad_norm": 1.2143386361330617,
+      "learning_rate": 3.942943506745382e-05,
+      "loss": 0.0824,
+      "step": 582
+    },
+    {
+      "epoch": 2.0821428571428573,
+      "grad_norm": 0.4732493829709655,
+      "learning_rate": 3.942668864686309e-05,
+      "loss": 0.0638,
+      "step": 583
+    },
+    {
+      "epoch": 2.085714285714286,
+      "grad_norm": 0.5909491573137307,
+      "learning_rate": 3.9423935728283085e-05,
+      "loss": 0.0658,
+      "step": 584
+    },
+    {
+      "epoch": 2.0892857142857144,
+      "grad_norm": 1.1046272116062879,
+      "learning_rate": 3.942117631263461e-05,
+      "loss": 0.0382,
+      "step": 585
+    },
+    {
+      "epoch": 2.092857142857143,
+      "grad_norm": 0.9837255124282164,
+      "learning_rate": 3.9418410400840666e-05,
+      "loss": 0.1082,
+      "step": 586
+    },
+    {
+      "epoch": 2.0964285714285715,
+      "grad_norm": 0.723012751161953,
+      "learning_rate": 3.941563799382641e-05,
+      "loss": 0.0827,
+      "step": 587
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.3249812896817181,
+      "learning_rate": 3.9412859092519184e-05,
+      "loss": 0.11,
+      "step": 588
+    },
+    {
+      "epoch": 2.1035714285714286,
+      "grad_norm": 0.8481273402658095,
+      "learning_rate": 3.94100736978485e-05,
+      "loss": 0.0806,
+      "step": 589
+    },
+    {
+      "epoch": 2.107142857142857,
+      "grad_norm": 0.5944870914828062,
+      "learning_rate": 3.940728181074603e-05,
+      "loss": 0.0711,
+      "step": 590
+    },
+    {
+      "epoch": 2.1107142857142858,
+      "grad_norm": 0.7578132903948145,
+      "learning_rate": 3.940448343214563e-05,
+      "loss": 0.0942,
+      "step": 591
+    },
+    {
+      "epoch": 2.1142857142857143,
+      "grad_norm": 0.9480975877224131,
+      "learning_rate": 3.940167856298332e-05,
+      "loss": 0.0607,
+      "step": 592
+    },
+    {
+      "epoch": 2.117857142857143,
+      "grad_norm": 0.7196148687201784,
+      "learning_rate": 3.93988672041973e-05,
+      "loss": 0.0567,
+      "step": 593
+    },
+    {
+      "epoch": 2.1214285714285714,
+      "grad_norm": 0.3866538454432723,
+      "learning_rate": 3.9396049356727933e-05,
+      "loss": 0.0617,
+      "step": 594
+    },
+    {
+      "epoch": 2.125,
+      "grad_norm": 0.44730033579423306,
+      "learning_rate": 3.9393225021517757e-05,
+      "loss": 0.0722,
+      "step": 595
+    },
+    {
+      "epoch": 2.1285714285714286,
+      "grad_norm": 0.4708679731232688,
+      "learning_rate": 3.939039419951147e-05,
+      "loss": 0.0547,
+      "step": 596
+    },
+    {
+      "epoch": 2.132142857142857,
+      "grad_norm": 0.7356061980373692,
+      "learning_rate": 3.9387556891655946e-05,
+      "loss": 0.0692,
+      "step": 597
+    },
+    {
+      "epoch": 2.1357142857142857,
+      "grad_norm": 0.7953591831046649,
+      "learning_rate": 3.938471309890024e-05,
+      "loss": 0.0607,
+      "step": 598
+    },
+    {
+      "epoch": 2.1392857142857142,
+      "grad_norm": 1.452159454797582,
+      "learning_rate": 3.938186282219556e-05,
+      "loss": 0.0768,
+      "step": 599
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.8059391961453279,
+      "learning_rate": 3.93790060624953e-05,
+      "loss": 0.0974,
+      "step": 600
+    },
+    {
+      "epoch": 2.1464285714285714,
+      "grad_norm": 0.7431215517566792,
+      "learning_rate": 3.937614282075499e-05,
+      "loss": 0.0804,
+      "step": 601
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.808219353235476,
+      "learning_rate": 3.9373273097932354e-05,
+      "loss": 0.0807,
+      "step": 602
+    },
+    {
+      "epoch": 2.1535714285714285,
+      "grad_norm": 0.6665354511252982,
+      "learning_rate": 3.93703968949873e-05,
+      "loss": 0.0498,
+      "step": 603
+    },
+    {
+      "epoch": 2.157142857142857,
+      "grad_norm": 0.5300244637975766,
+      "learning_rate": 3.936751421288186e-05,
+      "loss": 0.0855,
+      "step": 604
+    },
+    {
+      "epoch": 2.1607142857142856,
+      "grad_norm": 0.540136835641528,
+      "learning_rate": 3.936462505258026e-05,
+      "loss": 0.0518,
+      "step": 605
+    },
+    {
+      "epoch": 2.164285714285714,
+      "grad_norm": 0.5884020180535653,
+      "learning_rate": 3.93617294150489e-05,
+      "loss": 0.0911,
+      "step": 606
+    },
+    {
+      "epoch": 2.1678571428571427,
+      "grad_norm": 0.8205809326074744,
+      "learning_rate": 3.935882730125633e-05,
+      "loss": 0.0995,
+      "step": 607
+    },
+    {
+      "epoch": 2.1714285714285713,
+      "grad_norm": 2.0418783443294535,
+      "learning_rate": 3.935591871217326e-05,
+      "loss": 0.1146,
+      "step": 608
+    },
+    {
+      "epoch": 2.175,
+      "grad_norm": 0.7180352550723419,
+      "learning_rate": 3.935300364877259e-05,
+      "loss": 0.0843,
+      "step": 609
+    },
+    {
+      "epoch": 2.1785714285714284,
+      "grad_norm": 1.3974099372827111,
+      "learning_rate": 3.935008211202937e-05,
+      "loss": 0.0839,
+      "step": 610
+    },
+    {
+      "epoch": 2.182142857142857,
+      "grad_norm": 0.8381616542153555,
+      "learning_rate": 3.9347154102920814e-05,
+      "loss": 0.0482,
+      "step": 611
+    },
+    {
+      "epoch": 2.185714285714286,
+      "grad_norm": 1.292541950368239,
+      "learning_rate": 3.934421962242631e-05,
+      "loss": 0.1032,
+      "step": 612
+    },
+    {
+      "epoch": 2.189285714285714,
+      "grad_norm": 0.574845983314466,
+      "learning_rate": 3.9341278671527404e-05,
+      "loss": 0.0638,
+      "step": 613
+    },
+    {
+      "epoch": 2.192857142857143,
+      "grad_norm": 0.9785710964950036,
+      "learning_rate": 3.9338331251207814e-05,
+      "loss": 0.0612,
+      "step": 614
+    },
+    {
+      "epoch": 2.1964285714285716,
+      "grad_norm": 1.3302581374995974,
+      "learning_rate": 3.933537736245341e-05,
+      "loss": 0.106,
+      "step": 615
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.6045494888545162,
+      "learning_rate": 3.933241700625223e-05,
+      "loss": 0.0446,
+      "step": 616
+    },
+    {
+      "epoch": 2.2035714285714287,
+      "grad_norm": 0.8548929265926066,
+      "learning_rate": 3.932945018359448e-05,
+      "loss": 0.0717,
+      "step": 617
+    },
+    {
+      "epoch": 2.2071428571428573,
+      "grad_norm": 0.7817602971403889,
+      "learning_rate": 3.9326476895472517e-05,
+      "loss": 0.0749,
+      "step": 618
+    },
+    {
+      "epoch": 2.210714285714286,
+      "grad_norm": 1.0180957806032138,
+      "learning_rate": 3.932349714288089e-05,
+      "loss": 0.0799,
+      "step": 619
+    },
+    {
+      "epoch": 2.2142857142857144,
+      "grad_norm": 0.6447097923426459,
+      "learning_rate": 3.9320510926816266e-05,
+      "loss": 0.0757,
+      "step": 620
+    },
+    {
+      "epoch": 2.217857142857143,
+      "grad_norm": 0.7064319499400328,
+      "learning_rate": 3.9317518248277514e-05,
+      "loss": 0.0966,
+      "step": 621
+    },
+    {
+      "epoch": 2.2214285714285715,
+      "grad_norm": 1.3841115587354116,
+      "learning_rate": 3.931451910826565e-05,
+      "loss": 0.0763,
+      "step": 622
+    },
+    {
+      "epoch": 2.225,
+      "grad_norm": 0.5427076845089599,
+      "learning_rate": 3.931151350778383e-05,
+      "loss": 0.0367,
+      "step": 623
+    },
+    {
+      "epoch": 2.2285714285714286,
+      "grad_norm": 0.806753866327309,
+      "learning_rate": 3.930850144783741e-05,
+      "loss": 0.0973,
+      "step": 624
+    },
+    {
+      "epoch": 2.232142857142857,
+      "grad_norm": 0.6828256334929033,
+      "learning_rate": 3.930548292943388e-05,
+      "loss": 0.0491,
+      "step": 625
+    },
+    {
+      "epoch": 2.2357142857142858,
+      "grad_norm": 1.2055747780783428,
+      "learning_rate": 3.9302457953582896e-05,
+      "loss": 0.0691,
+      "step": 626
+    },
+    {
+      "epoch": 2.2392857142857143,
+      "grad_norm": 0.8727525196602515,
+      "learning_rate": 3.929942652129627e-05,
+      "loss": 0.1158,
+      "step": 627
+    },
+    {
+      "epoch": 2.242857142857143,
+      "grad_norm": 1.0446780625974874,
+      "learning_rate": 3.9296388633588e-05,
+      "loss": 0.0648,
+      "step": 628
+    },
+    {
+      "epoch": 2.2464285714285714,
+      "grad_norm": 0.8537445137390867,
+      "learning_rate": 3.929334429147421e-05,
+      "loss": 0.0932,
+      "step": 629
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.7968028247467981,
+      "learning_rate": 3.929029349597318e-05,
+      "loss": 0.0857,
+      "step": 630
+    },
+    {
+      "epoch": 2.2535714285714286,
+      "grad_norm": 0.962437655603322,
+      "learning_rate": 3.928723624810539e-05,
+      "loss": 0.0755,
+      "step": 631
+    },
+    {
+      "epoch": 2.257142857142857,
+      "grad_norm": 0.5203018666173721,
+      "learning_rate": 3.9284172548893426e-05,
+      "loss": 0.08,
+      "step": 632
+    },
+    {
+      "epoch": 2.2607142857142857,
+      "grad_norm": 1.1477934407457648,
+      "learning_rate": 3.928110239936207e-05,
+      "loss": 0.0788,
+      "step": 633
+    },
+    {
+      "epoch": 2.2642857142857142,
+      "grad_norm": 0.9892151595734521,
+      "learning_rate": 3.9278025800538266e-05,
+      "loss": 0.0993,
+      "step": 634
+    },
+    {
+      "epoch": 2.267857142857143,
+      "grad_norm": 0.5977979337651129,
+      "learning_rate": 3.927494275345107e-05,
+      "loss": 0.0377,
+      "step": 635
+    },
+    {
+      "epoch": 2.2714285714285714,
+      "grad_norm": 1.221512314232381,
+      "learning_rate": 3.9271853259131736e-05,
+      "loss": 0.0827,
+      "step": 636
+    },
+    {
+      "epoch": 2.275,
+      "grad_norm": 0.7564907946306709,
+      "learning_rate": 3.926875731861367e-05,
+      "loss": 0.0574,
+      "step": 637
+    },
+    {
+      "epoch": 2.2785714285714285,
+      "grad_norm": 1.1968369807237562,
+      "learning_rate": 3.9265654932932415e-05,
+      "loss": 0.0689,
+      "step": 638
+    },
+    {
+      "epoch": 2.282142857142857,
+      "grad_norm": 0.3514697039654983,
+      "learning_rate": 3.9262546103125683e-05,
+      "loss": 0.0235,
+      "step": 639
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.48732741094239324,
+      "learning_rate": 3.925943083023334e-05,
+      "loss": 0.0749,
+      "step": 640
+    },
+    {
+      "epoch": 2.289285714285714,
+      "grad_norm": 0.8423610853441953,
+      "learning_rate": 3.925630911529741e-05,
+      "loss": 0.1047,
+      "step": 641
+    },
+    {
+      "epoch": 2.2928571428571427,
+      "grad_norm": 0.9794515091445838,
+      "learning_rate": 3.925318095936207e-05,
+      "loss": 0.1282,
+      "step": 642
+    },
+    {
+      "epoch": 2.2964285714285713,
+      "grad_norm": 0.9155560456863151,
+      "learning_rate": 3.925004636347364e-05,
+      "loss": 0.0796,
+      "step": 643
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.7222059673435268,
+      "learning_rate": 3.924690532868061e-05,
+      "loss": 0.0511,
+      "step": 644
+    },
+    {
+      "epoch": 2.3035714285714284,
+      "grad_norm": 0.5306881946456448,
+      "learning_rate": 3.9243757856033615e-05,
+      "loss": 0.1114,
+      "step": 645
+    },
+    {
+      "epoch": 2.307142857142857,
+      "grad_norm": 0.8936123619185099,
+      "learning_rate": 3.9240603946585454e-05,
+      "loss": 0.0646,
+      "step": 646
+    },
+    {
+      "epoch": 2.310714285714286,
+      "grad_norm": 0.6814178528176416,
+      "learning_rate": 3.9237443601391064e-05,
+      "loss": 0.0814,
+      "step": 647
+    },
+    {
+      "epoch": 2.314285714285714,
+      "grad_norm": 0.6513030898274469,
+      "learning_rate": 3.923427682150754e-05,
+      "loss": 0.0695,
+      "step": 648
+    },
+    {
+      "epoch": 2.317857142857143,
+      "grad_norm": 0.7437129855008155,
+      "learning_rate": 3.923110360799413e-05,
+      "loss": 0.108,
+      "step": 649
+    },
+    {
+      "epoch": 2.3214285714285716,
+      "grad_norm": 0.5787313651785617,
+      "learning_rate": 3.922792396191225e-05,
+      "loss": 0.0563,
+      "step": 650
+    },
+    {
+      "epoch": 2.325,
+      "grad_norm": 0.5174046360891601,
+      "learning_rate": 3.9224737884325435e-05,
+      "loss": 0.0443,
+      "step": 651
+    },
+    {
+      "epoch": 2.3285714285714287,
+      "grad_norm": 1.2750837338728969,
+      "learning_rate": 3.92215453762994e-05,
+      "loss": 0.1171,
+      "step": 652
+    },
+    {
+      "epoch": 2.3321428571428573,
+      "grad_norm": 0.6358478538854582,
+      "learning_rate": 3.9218346438901996e-05,
+      "loss": 0.0727,
+      "step": 653
+    },
+    {
+      "epoch": 2.335714285714286,
+      "grad_norm": 0.8543317768466425,
+      "learning_rate": 3.9215141073203225e-05,
+      "loss": 0.1132,
+      "step": 654
+    },
+    {
+      "epoch": 2.3392857142857144,
+      "grad_norm": 0.4743806966507369,
+      "learning_rate": 3.921192928027524e-05,
+      "loss": 0.0821,
+      "step": 655
+    },
+    {
+      "epoch": 2.342857142857143,
+      "grad_norm": 0.993934184571449,
+      "learning_rate": 3.9208711061192363e-05,
+      "loss": 0.0593,
+      "step": 656
+    },
+    {
+      "epoch": 2.3464285714285715,
+      "grad_norm": 1.1964066878096,
+      "learning_rate": 3.920548641703103e-05,
+      "loss": 0.1477,
+      "step": 657
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.1940904622225497,
+      "learning_rate": 3.920225534886986e-05,
+      "loss": 0.0966,
+      "step": 658
+    },
+    {
+      "epoch": 2.3535714285714286,
+      "grad_norm": 0.7670652918586472,
+      "learning_rate": 3.919901785778959e-05,
+      "loss": 0.1283,
+      "step": 659
+    },
+    {
+      "epoch": 2.357142857142857,
+      "grad_norm": 1.6967750239184871,
+      "learning_rate": 3.9195773944873135e-05,
+      "loss": 0.1035,
+      "step": 660
+    },
+    {
+      "epoch": 2.3607142857142858,
+      "grad_norm": 1.6749135313240933,
+      "learning_rate": 3.9192523611205535e-05,
+      "loss": 0.1003,
+      "step": 661
+    },
+    {
+      "epoch": 2.3642857142857143,
+      "grad_norm": 0.8727380878009077,
+      "learning_rate": 3.9189266857873995e-05,
+      "loss": 0.1046,
+      "step": 662
+    },
+    {
+      "epoch": 2.367857142857143,
+      "grad_norm": 1.1049402605371905,
+      "learning_rate": 3.918600368596785e-05,
+      "loss": 0.0884,
+      "step": 663
+    },
+    {
+      "epoch": 2.3714285714285714,
+      "grad_norm": 1.5520018137193214,
+      "learning_rate": 3.9182734096578595e-05,
+      "loss": 0.1206,
+      "step": 664
+    },
+    {
+      "epoch": 2.375,
+      "grad_norm": 0.5672030570935855,
+      "learning_rate": 3.917945809079987e-05,
+      "loss": 0.0683,
+      "step": 665
+    },
+    {
+      "epoch": 2.3785714285714286,
+      "grad_norm": 1.0226161855234819,
+      "learning_rate": 3.9176175669727455e-05,
+      "loss": 0.0832,
+      "step": 666
+    },
+    {
+      "epoch": 2.382142857142857,
+      "grad_norm": 0.7802718741260426,
+      "learning_rate": 3.9172886834459286e-05,
+      "loss": 0.0537,
+      "step": 667
+    },
+    {
+      "epoch": 2.3857142857142857,
+      "grad_norm": 0.4508064186021185,
+      "learning_rate": 3.916959158609543e-05,
+      "loss": 0.0537,
+      "step": 668
+    },
+    {
+      "epoch": 2.3892857142857142,
+      "grad_norm": 1.139204851051718,
+      "learning_rate": 3.916628992573811e-05,
+      "loss": 0.1369,
+      "step": 669
+    },
+    {
+      "epoch": 2.392857142857143,
+      "grad_norm": 1.2554042800693943,
+      "learning_rate": 3.9162981854491695e-05,
+      "loss": 0.1255,
+      "step": 670
+    },
+    {
+      "epoch": 2.3964285714285714,
+      "grad_norm": 0.9569750666396515,
+      "learning_rate": 3.915966737346269e-05,
+      "loss": 0.1219,
+      "step": 671
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.7993636314682899,
+      "learning_rate": 3.915634648375974e-05,
+      "loss": 0.0867,
+      "step": 672
+    },
+    {
+      "epoch": 2.4035714285714285,
+      "grad_norm": 1.4093936719613727,
+      "learning_rate": 3.9153019186493664e-05,
+      "loss": 0.1114,
+      "step": 673
+    },
+    {
+      "epoch": 2.407142857142857,
+      "grad_norm": 2.4552668063289893,
+      "learning_rate": 3.914968548277738e-05,
+      "loss": 0.1053,
+      "step": 674
+    },
+    {
+      "epoch": 2.4107142857142856,
+      "grad_norm": 1.2885191965028355,
+      "learning_rate": 3.9146345373725984e-05,
+      "loss": 0.1214,
+      "step": 675
+    },
+    {
+      "epoch": 2.414285714285714,
+      "grad_norm": 1.0319431601817644,
+      "learning_rate": 3.914299886045671e-05,
+      "loss": 0.0746,
+      "step": 676
+    },
+    {
+      "epoch": 2.4178571428571427,
+      "grad_norm": 0.810997860271014,
+      "learning_rate": 3.913964594408889e-05,
+      "loss": 0.0535,
+      "step": 677
+    },
+    {
+      "epoch": 2.4214285714285713,
+      "grad_norm": 0.432493335100842,
+      "learning_rate": 3.9136286625744074e-05,
+      "loss": 0.086,
+      "step": 678
+    },
+    {
+      "epoch": 2.425,
+      "grad_norm": 0.689822516356383,
+      "learning_rate": 3.913292090654589e-05,
+      "loss": 0.1039,
+      "step": 679
+    },
+    {
+      "epoch": 2.4285714285714284,
+      "grad_norm": 0.9044527925471296,
+      "learning_rate": 3.9129548787620135e-05,
+      "loss": 0.07,
+      "step": 680
+    },
+    {
+      "epoch": 2.432142857142857,
+      "grad_norm": 0.8588932257803473,
+      "learning_rate": 3.912617027009475e-05,
+      "loss": 0.0909,
+      "step": 681
+    },
+    {
+      "epoch": 2.435714285714286,
+      "grad_norm": 1.180960016625176,
+      "learning_rate": 3.9122785355099796e-05,
+      "loss": 0.1582,
+      "step": 682
+    },
+    {
+      "epoch": 2.439285714285714,
+      "grad_norm": 0.871312322253799,
+      "learning_rate": 3.911939404376748e-05,
+      "loss": 0.1184,
+      "step": 683
+    },
+    {
+      "epoch": 2.442857142857143,
+      "grad_norm": 0.5704061055336533,
+      "learning_rate": 3.9115996337232174e-05,
+      "loss": 0.0588,
+      "step": 684
+    },
+    {
+      "epoch": 2.4464285714285716,
+      "grad_norm": 0.8805453623675099,
+      "learning_rate": 3.911259223663036e-05,
+      "loss": 0.0793,
+      "step": 685
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.6905293051459056,
+      "learning_rate": 3.910918174310066e-05,
+      "loss": 0.1084,
+      "step": 686
+    },
+    {
+      "epoch": 2.4535714285714287,
+      "grad_norm": 1.0902860256235027,
+      "learning_rate": 3.910576485778385e-05,
+      "loss": 0.1091,
+      "step": 687
+    },
+    {
+      "epoch": 2.4571428571428573,
+      "grad_norm": 0.7694650342196058,
+      "learning_rate": 3.910234158182283e-05,
+      "loss": 0.0643,
+      "step": 688
+    },
+    {
+      "epoch": 2.460714285714286,
+      "grad_norm": 0.44799356164363086,
+      "learning_rate": 3.909891191636265e-05,
+      "loss": 0.0684,
+      "step": 689
+    },
+    {
+      "epoch": 2.4642857142857144,
+      "grad_norm": 0.9310624593893404,
+      "learning_rate": 3.90954758625505e-05,
+      "loss": 0.0732,
+      "step": 690
+    },
+    {
+      "epoch": 2.467857142857143,
+      "grad_norm": 0.8723962411098155,
+      "learning_rate": 3.9092033421535685e-05,
+      "loss": 0.0893,
+      "step": 691
+    },
+    {
+      "epoch": 2.4714285714285715,
+      "grad_norm": 1.243547390525058,
+      "learning_rate": 3.908858459446966e-05,
+      "loss": 0.1187,
+      "step": 692
+    },
+    {
+      "epoch": 2.475,
+      "grad_norm": 0.6746811359290301,
+      "learning_rate": 3.908512938250602e-05,
+      "loss": 0.1003,
+      "step": 693
+    },
+    {
+      "epoch": 2.4785714285714286,
+      "grad_norm": 0.6425235387768927,
+      "learning_rate": 3.908166778680048e-05,
+      "loss": 0.0927,
+      "step": 694
+    },
+    {
+      "epoch": 2.482142857142857,
+      "grad_norm": 0.6482930064947436,
+      "learning_rate": 3.9078199808510924e-05,
+      "loss": 0.0753,
+      "step": 695
+    },
+    {
+      "epoch": 2.4857142857142858,
+      "grad_norm": 0.8136372909211631,
+      "learning_rate": 3.907472544879732e-05,
+      "loss": 0.0915,
+      "step": 696
+    },
+    {
+      "epoch": 2.4892857142857143,
+      "grad_norm": 0.8242595735449603,
+      "learning_rate": 3.907124470882182e-05,
+      "loss": 0.0797,
+      "step": 697
+    },
+    {
+      "epoch": 2.492857142857143,
+      "grad_norm": 0.6627937608598636,
+      "learning_rate": 3.906775758974869e-05,
+      "loss": 0.0989,
+      "step": 698
+    },
+    {
+      "epoch": 2.4964285714285714,
+      "grad_norm": 0.7487713282921086,
+      "learning_rate": 3.906426409274431e-05,
+      "loss": 0.1132,
+      "step": 699
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.4794040675847486,
+      "learning_rate": 3.906076421897722e-05,
+      "loss": 0.0558,
+      "step": 700
+    },
+    {
+      "epoch": 2.5035714285714286,
+      "grad_norm": 0.4180183571822546,
+      "learning_rate": 3.9057257969618095e-05,
+      "loss": 0.0874,
+      "step": 701
+    },
+    {
+      "epoch": 2.507142857142857,
+      "grad_norm": 0.8493370704992458,
+      "learning_rate": 3.905374534583972e-05,
+      "loss": 0.1345,
+      "step": 702
+    },
+    {
+      "epoch": 2.5107142857142857,
+      "grad_norm": 0.9439320161325945,
+      "learning_rate": 3.905022634881703e-05,
+      "loss": 0.0706,
+      "step": 703
+    },
+    {
+      "epoch": 2.5142857142857142,
+      "grad_norm": 0.5215570333300609,
+      "learning_rate": 3.904670097972709e-05,
+      "loss": 0.0852,
+      "step": 704
+    },
+    {
+      "epoch": 2.517857142857143,
+      "grad_norm": 0.6501045495770245,
+      "learning_rate": 3.904316923974908e-05,
+      "loss": 0.0845,
+      "step": 705
+    },
+    {
+      "epoch": 2.5214285714285714,
+      "grad_norm": 0.6874369933523368,
+      "learning_rate": 3.903963113006434e-05,
+      "loss": 0.0606,
+      "step": 706
+    },
+    {
+      "epoch": 2.525,
+      "grad_norm": 0.34685724317919314,
+      "learning_rate": 3.9036086651856314e-05,
+      "loss": 0.0808,
+      "step": 707
+    },
+    {
+      "epoch": 2.5285714285714285,
+      "grad_norm": 0.9429659489528421,
+      "learning_rate": 3.90325358063106e-05,
+      "loss": 0.0796,
+      "step": 708
+    },
+    {
+      "epoch": 2.532142857142857,
+      "grad_norm": 1.0995647635121353,
+      "learning_rate": 3.902897859461488e-05,
+      "loss": 0.1081,
+      "step": 709
+    },
+    {
+      "epoch": 2.5357142857142856,
+      "grad_norm": 0.5787323770932422,
+      "learning_rate": 3.902541501795903e-05,
+      "loss": 0.0934,
+      "step": 710
+    },
+    {
+      "epoch": 2.539285714285714,
+      "grad_norm": 0.7836156282691568,
+      "learning_rate": 3.9021845077535014e-05,
+      "loss": 0.0779,
+      "step": 711
+    },
+    {
+      "epoch": 2.5428571428571427,
+      "grad_norm": 0.7448100116610351,
+      "learning_rate": 3.901826877453693e-05,
+      "loss": 0.0465,
+      "step": 712
+    },
+    {
+      "epoch": 2.5464285714285713,
+      "grad_norm": 0.6694094073698567,
+      "learning_rate": 3.9014686110161005e-05,
+      "loss": 0.0807,
+      "step": 713
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.4359074005086277,
+      "learning_rate": 3.901109708560561e-05,
+      "loss": 0.0421,
+      "step": 714
+    },
+    {
+      "epoch": 2.553571428571429,
+      "grad_norm": 0.6543363176296414,
+      "learning_rate": 3.900750170207121e-05,
+      "loss": 0.1181,
+      "step": 715
+    },
+    {
+      "epoch": 2.557142857142857,
+      "grad_norm": 0.5037036507630038,
+      "learning_rate": 3.900389996076043e-05,
+      "loss": 0.0626,
+      "step": 716
+    },
+    {
+      "epoch": 2.560714285714286,
+      "grad_norm": 0.26173227959884904,
+      "learning_rate": 3.900029186287801e-05,
+      "loss": 0.0182,
+      "step": 717
+    },
+    {
+      "epoch": 2.564285714285714,
+      "grad_norm": 0.7565060960442598,
+      "learning_rate": 3.8996677409630794e-05,
+      "loss": 0.0802,
+      "step": 718
+    },
+    {
+      "epoch": 2.567857142857143,
+      "grad_norm": 0.725466588656459,
+      "learning_rate": 3.89930566022278e-05,
+      "loss": 0.1019,
+      "step": 719
+    },
+    {
+      "epoch": 2.571428571428571,
+      "grad_norm": 0.9597572520040345,
+      "learning_rate": 3.898942944188013e-05,
+      "loss": 0.1247,
+      "step": 720
+    },
+    {
+      "epoch": 2.575,
+      "grad_norm": 0.8728921079560074,
+      "learning_rate": 3.8985795929801024e-05,
+      "loss": 0.081,
+      "step": 721
+    },
+    {
+      "epoch": 2.5785714285714287,
+      "grad_norm": 0.6854351954339186,
+      "learning_rate": 3.8982156067205844e-05,
+      "loss": 0.0666,
+      "step": 722
+    },
+    {
+      "epoch": 2.5821428571428573,
+      "grad_norm": 0.653550076135066,
+      "learning_rate": 3.8978509855312096e-05,
+      "loss": 0.0907,
+      "step": 723
+    },
+    {
+      "epoch": 2.585714285714286,
+      "grad_norm": 0.6294019657873029,
+      "learning_rate": 3.8974857295339376e-05,
+      "loss": 0.0751,
+      "step": 724
+    },
+    {
+      "epoch": 2.5892857142857144,
+      "grad_norm": 0.683625030685315,
+      "learning_rate": 3.897119838850943e-05,
+      "loss": 0.0739,
+      "step": 725
+    },
+    {
+      "epoch": 2.592857142857143,
+      "grad_norm": 0.9890321715064776,
+      "learning_rate": 3.896753313604611e-05,
+      "loss": 0.0987,
+      "step": 726
+    },
+    {
+      "epoch": 2.5964285714285715,
+      "grad_norm": 2.0830359154114158,
+      "learning_rate": 3.89638615391754e-05,
+      "loss": 0.0931,
+      "step": 727
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.40889668037168,
+      "learning_rate": 3.896018359912541e-05,
+      "loss": 0.1018,
+      "step": 728
+    },
+    {
+      "epoch": 2.6035714285714286,
+      "grad_norm": 0.8104625556431391,
+      "learning_rate": 3.895649931712637e-05,
+      "loss": 0.1093,
+      "step": 729
+    },
+    {
+      "epoch": 2.607142857142857,
+      "grad_norm": 0.3766645799093778,
+      "learning_rate": 3.8952808694410606e-05,
+      "loss": 0.0562,
+      "step": 730
+    },
+    {
+      "epoch": 2.6107142857142858,
+      "grad_norm": 0.44483492835224553,
+      "learning_rate": 3.89491117322126e-05,
+      "loss": 0.0614,
+      "step": 731
+    },
+    {
+      "epoch": 2.6142857142857143,
+      "grad_norm": 0.5554206630549883,
+      "learning_rate": 3.894540843176895e-05,
+      "loss": 0.1079,
+      "step": 732
+    },
+    {
+      "epoch": 2.617857142857143,
+      "grad_norm": 1.021387030435115,
+      "learning_rate": 3.894169879431835e-05,
+      "loss": 0.0849,
+      "step": 733
+    },
+    {
+      "epoch": 2.6214285714285714,
+      "grad_norm": 0.4535999736762529,
+      "learning_rate": 3.8937982821101626e-05,
+      "loss": 0.0432,
+      "step": 734
+    },
+    {
+      "epoch": 2.625,
+      "grad_norm": 0.6146729386145838,
+      "learning_rate": 3.8934260513361734e-05,
+      "loss": 0.0791,
+      "step": 735
+    },
+    {
+      "epoch": 2.6285714285714286,
+      "grad_norm": 0.8029510481243191,
+      "learning_rate": 3.893053187234375e-05,
+      "loss": 0.0898,
+      "step": 736
+    },
+    {
+      "epoch": 2.632142857142857,
+      "grad_norm": 1.9611583113781619,
+      "learning_rate": 3.8926796899294837e-05,
+      "loss": 0.1878,
+      "step": 737
+    },
+    {
+      "epoch": 2.6357142857142857,
+      "grad_norm": 0.7271082698286748,
+      "learning_rate": 3.8923055595464315e-05,
+      "loss": 0.088,
+      "step": 738
+    },
+    {
+      "epoch": 2.6392857142857142,
+      "grad_norm": 0.6454356855426903,
+      "learning_rate": 3.891930796210359e-05,
+      "loss": 0.0854,
+      "step": 739
+    },
+    {
+      "epoch": 2.642857142857143,
+      "grad_norm": 0.5046977798852078,
+      "learning_rate": 3.8915554000466205e-05,
+      "loss": 0.0788,
+      "step": 740
+    },
+    {
+      "epoch": 2.6464285714285714,
+      "grad_norm": 0.35549103140919924,
+      "learning_rate": 3.891179371180783e-05,
+      "loss": 0.0651,
+      "step": 741
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.46979843959821643,
+      "learning_rate": 3.8908027097386205e-05,
+      "loss": 0.0836,
+      "step": 742
+    },
+    {
+      "epoch": 2.6535714285714285,
+      "grad_norm": 0.5203634193148317,
+      "learning_rate": 3.890425415846124e-05,
+      "loss": 0.0989,
+      "step": 743
+    },
+    {
+      "epoch": 2.657142857142857,
+      "grad_norm": 0.6019402414670744,
+      "learning_rate": 3.890047489629493e-05,
+      "loss": 0.0751,
+      "step": 744
+    },
+    {
+      "epoch": 2.6607142857142856,
+      "grad_norm": 0.4891591373510291,
+      "learning_rate": 3.889668931215139e-05,
+      "loss": 0.0994,
+      "step": 745
+    },
+    {
+      "epoch": 2.664285714285714,
+      "grad_norm": 0.6752351528079952,
+      "learning_rate": 3.889289740729685e-05,
+      "loss": 0.0854,
+      "step": 746
+    },
+    {
+      "epoch": 2.6678571428571427,
+      "grad_norm": 0.6313098404649151,
+      "learning_rate": 3.888909918299966e-05,
+      "loss": 0.0745,
+      "step": 747
+    },
+    {
+      "epoch": 2.6714285714285713,
+      "grad_norm": 0.4291138108328092,
+      "learning_rate": 3.888529464053029e-05,
+      "loss": 0.0985,
+      "step": 748
+    },
+    {
+      "epoch": 2.675,
+      "grad_norm": 0.49388446496538807,
+      "learning_rate": 3.8881483781161286e-05,
+      "loss": 0.0872,
+      "step": 749
+    },
+    {
+      "epoch": 2.678571428571429,
+      "grad_norm": 0.4279318257596009,
+      "learning_rate": 3.8877666606167354e-05,
+      "loss": 0.0969,
+      "step": 750
+    },
+    {
+      "epoch": 2.682142857142857,
+      "grad_norm": 0.5861161328812649,
+      "learning_rate": 3.887384311682529e-05,
+      "loss": 0.0446,
+      "step": 751
+    },
+    {
+      "epoch": 2.685714285714286,
+      "grad_norm": 0.4450674276141422,
+      "learning_rate": 3.8870013314413996e-05,
+      "loss": 0.0526,
+      "step": 752
+    },
+    {
+      "epoch": 2.689285714285714,
+      "grad_norm": 0.8687494978749438,
+      "learning_rate": 3.88661772002145e-05,
+      "loss": 0.1145,
+      "step": 753
+    },
+    {
+      "epoch": 2.692857142857143,
+      "grad_norm": 0.7507301854278623,
+      "learning_rate": 3.8862334775509934e-05,
+      "loss": 0.062,
+      "step": 754
+    },
+    {
+      "epoch": 2.696428571428571,
+      "grad_norm": 0.8444780244916804,
+      "learning_rate": 3.8858486041585546e-05,
+      "loss": 0.0654,
+      "step": 755
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.3335207526565628,
+      "learning_rate": 3.885463099972869e-05,
+      "loss": 0.05,
+      "step": 756
+    },
+    {
+      "epoch": 2.7035714285714287,
+      "grad_norm": 0.5845498244687236,
+      "learning_rate": 3.885076965122882e-05,
+      "loss": 0.0807,
+      "step": 757
+    },
+    {
+      "epoch": 2.7071428571428573,
+      "grad_norm": 0.8053931239650283,
+      "learning_rate": 3.884690199737752e-05,
+      "loss": 0.0461,
+      "step": 758
+    },
+    {
+      "epoch": 2.710714285714286,
+      "grad_norm": 0.8276356026478784,
+      "learning_rate": 3.8843028039468466e-05,
+      "loss": 0.108,
+      "step": 759
+    },
+    {
+      "epoch": 2.7142857142857144,
+      "grad_norm": 0.49437904807893635,
+      "learning_rate": 3.883914777879746e-05,
+      "loss": 0.087,
+      "step": 760
+    },
+    {
+      "epoch": 2.717857142857143,
+      "grad_norm": 0.6628871042089105,
+      "learning_rate": 3.883526121666239e-05,
+      "loss": 0.0963,
+      "step": 761
+    },
+    {
+      "epoch": 2.7214285714285715,
+      "grad_norm": 0.8525793209025375,
+      "learning_rate": 3.883136835436327e-05,
+      "loss": 0.1157,
+      "step": 762
+    },
+    {
+      "epoch": 2.725,
+      "grad_norm": 0.6841115844962428,
+      "learning_rate": 3.882746919320221e-05,
+      "loss": 0.1119,
+      "step": 763
+    },
+    {
+      "epoch": 2.7285714285714286,
+      "grad_norm": 0.9457168676810153,
+      "learning_rate": 3.8823563734483435e-05,
+      "loss": 0.0567,
+      "step": 764
+    },
+    {
+      "epoch": 2.732142857142857,
+      "grad_norm": 0.6813392871746714,
+      "learning_rate": 3.881965197951327e-05,
+      "loss": 0.0363,
+      "step": 765
+    },
+    {
+      "epoch": 2.7357142857142858,
+      "grad_norm": 0.40808777132506147,
+      "learning_rate": 3.881573392960015e-05,
+      "loss": 0.0786,
+      "step": 766
+    },
+    {
+      "epoch": 2.7392857142857143,
+      "grad_norm": 0.7865186492083779,
+      "learning_rate": 3.881180958605463e-05,
+      "loss": 0.0764,
+      "step": 767
+    },
+    {
+      "epoch": 2.742857142857143,
+      "grad_norm": 0.47417252109870606,
+      "learning_rate": 3.880787895018933e-05,
+      "loss": 0.0546,
+      "step": 768
+    },
+    {
+      "epoch": 2.7464285714285714,
+      "grad_norm": 0.7550734083722881,
+      "learning_rate": 3.880394202331901e-05,
+      "loss": 0.0861,
+      "step": 769
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.5904550814832419,
+      "learning_rate": 3.879999880676053e-05,
+      "loss": 0.077,
+      "step": 770
+    },
+    {
+      "epoch": 2.7535714285714286,
+      "grad_norm": 0.5109907260061688,
+      "learning_rate": 3.879604930183284e-05,
+      "loss": 0.1012,
+      "step": 771
+    },
+    {
+      "epoch": 2.757142857142857,
+      "grad_norm": 0.8003694946548795,
+      "learning_rate": 3.8792093509857e-05,
+      "loss": 0.1071,
+      "step": 772
+    },
+    {
+      "epoch": 2.7607142857142857,
+      "grad_norm": 0.9986006997819653,
+      "learning_rate": 3.878813143215618e-05,
+      "loss": 0.1043,
+      "step": 773
+    },
+    {
+      "epoch": 2.7642857142857142,
+      "grad_norm": 1.0782752630961052,
+      "learning_rate": 3.878416307005565e-05,
+      "loss": 0.1076,
+      "step": 774
+    },
+    {
+      "epoch": 2.767857142857143,
+      "grad_norm": 0.5817680731959481,
+      "learning_rate": 3.878018842488276e-05,
+      "loss": 0.1282,
+      "step": 775
+    },
+    {
+      "epoch": 2.7714285714285714,
+      "grad_norm": 0.5278845831271766,
+      "learning_rate": 3.877620749796699e-05,
+      "loss": 0.1226,
+      "step": 776
+    },
+    {
+      "epoch": 2.775,
+      "grad_norm": 0.3266007338793975,
+      "learning_rate": 3.8772220290639924e-05,
+      "loss": 0.0606,
+      "step": 777
+    },
+    {
+      "epoch": 2.7785714285714285,
+      "grad_norm": 0.4340292930882877,
+      "learning_rate": 3.876822680423522e-05,
+      "loss": 0.0742,
+      "step": 778
+    },
+    {
+      "epoch": 2.782142857142857,
+      "grad_norm": 0.8957163339709835,
+      "learning_rate": 3.876422704008866e-05,
+      "loss": 0.0985,
+      "step": 779
+    },
+    {
+      "epoch": 2.7857142857142856,
+      "grad_norm": 1.0496708928835439,
+      "learning_rate": 3.8760220999538095e-05,
+      "loss": 0.1076,
+      "step": 780
+    },
+    {
+      "epoch": 2.789285714285714,
+      "grad_norm": 0.47051407235112325,
+      "learning_rate": 3.875620868392352e-05,
+      "loss": 0.0566,
+      "step": 781
+    },
+    {
+      "epoch": 2.7928571428571427,
+      "grad_norm": 0.9610156902457999,
+      "learning_rate": 3.8752190094586994e-05,
+      "loss": 0.1049,
+      "step": 782
+    },
+    {
+      "epoch": 2.7964285714285713,
+      "grad_norm": 0.7986509283544702,
+      "learning_rate": 3.874816523287269e-05,
+      "loss": 0.0976,
+      "step": 783
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.7626379719259644,
+      "learning_rate": 3.874413410012688e-05,
+      "loss": 0.054,
+      "step": 784
+    },
+    {
+      "epoch": 2.803571428571429,
+      "grad_norm": 0.7838279863022131,
+      "learning_rate": 3.874009669769791e-05,
+      "loss": 0.0998,
+      "step": 785
+    },
+    {
+      "epoch": 2.807142857142857,
+      "grad_norm": 0.46153098719134056,
+      "learning_rate": 3.873605302693626e-05,
+      "loss": 0.0609,
+      "step": 786
+    },
+    {
+      "epoch": 2.810714285714286,
+      "grad_norm": 0.4640861819811759,
+      "learning_rate": 3.8732003089194476e-05,
+      "loss": 0.0361,
+      "step": 787
+    },
+    {
+      "epoch": 2.814285714285714,
+      "grad_norm": 0.33024449753067653,
+      "learning_rate": 3.8727946885827224e-05,
+      "loss": 0.0523,
+      "step": 788
+    },
+    {
+      "epoch": 2.817857142857143,
+      "grad_norm": 0.4229024364605341,
+      "learning_rate": 3.8723884418191245e-05,
+      "loss": 0.0505,
+      "step": 789
+    },
+    {
+      "epoch": 2.821428571428571,
+      "grad_norm": 0.6672921072136103,
+      "learning_rate": 3.87198156876454e-05,
+      "loss": 0.061,
+      "step": 790
+    },
+    {
+      "epoch": 2.825,
+      "grad_norm": 0.6308969573309644,
+      "learning_rate": 3.87157406955506e-05,
+      "loss": 0.0872,
+      "step": 791
+    },
+    {
+      "epoch": 2.8285714285714287,
+      "grad_norm": 0.6288383541919587,
+      "learning_rate": 3.8711659443269904e-05,
+      "loss": 0.0892,
+      "step": 792
+    },
+    {
+      "epoch": 2.8321428571428573,
+      "grad_norm": 0.6861966328758808,
+      "learning_rate": 3.870757193216844e-05,
+      "loss": 0.0484,
+      "step": 793
+    },
+    {
+      "epoch": 2.835714285714286,
+      "grad_norm": 0.5177690985031326,
+      "learning_rate": 3.870347816361342e-05,
+      "loss": 0.0581,
+      "step": 794
+    },
+    {
+      "epoch": 2.8392857142857144,
+      "grad_norm": 0.5660687823995036,
+      "learning_rate": 3.869937813897416e-05,
+      "loss": 0.0743,
+      "step": 795
+    },
+    {
+      "epoch": 2.842857142857143,
+      "grad_norm": 0.9019450605529574,
+      "learning_rate": 3.869527185962208e-05,
+      "loss": 0.0821,
+      "step": 796
+    },
+    {
+      "epoch": 2.8464285714285715,
+      "grad_norm": 0.7936530714400899,
+      "learning_rate": 3.8691159326930676e-05,
+      "loss": 0.0842,
+      "step": 797
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.9140017588906301,
+      "learning_rate": 3.868704054227553e-05,
+      "loss": 0.0836,
+      "step": 798
+    },
+    {
+      "epoch": 2.8535714285714286,
+      "grad_norm": 0.7958519388142878,
+      "learning_rate": 3.868291550703433e-05,
+      "loss": 0.0733,
+      "step": 799
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.9594410621741134,
+      "learning_rate": 3.867878422258685e-05,
+      "loss": 0.0632,
+      "step": 800
+    },
+    {
+      "epoch": 2.8607142857142858,
+      "grad_norm": 2.0448545679517083,
+      "learning_rate": 3.867464669031496e-05,
+      "loss": 0.1143,
+      "step": 801
+    },
+    {
+      "epoch": 2.8642857142857143,
+      "grad_norm": 0.5342561739887258,
+      "learning_rate": 3.867050291160261e-05,
+      "loss": 0.0757,
+      "step": 802
+    },
+    {
+      "epoch": 2.867857142857143,
+      "grad_norm": 0.6475296535251237,
+      "learning_rate": 3.8666352887835834e-05,
+      "loss": 0.0441,
+      "step": 803
+    },
+    {
+      "epoch": 2.8714285714285714,
+      "grad_norm": 0.6049863092553899,
+      "learning_rate": 3.8662196620402783e-05,
+      "loss": 0.0844,
+      "step": 804
+    },
+    {
+      "epoch": 2.875,
+      "grad_norm": 0.6368511646791394,
+      "learning_rate": 3.865803411069366e-05,
+      "loss": 0.0761,
+      "step": 805
+    },
+    {
+      "epoch": 2.8785714285714286,
+      "grad_norm": 0.5071136495522707,
+      "learning_rate": 3.8653865360100774e-05,
+      "loss": 0.0525,
+      "step": 806
+    },
+    {
+      "epoch": 2.882142857142857,
+      "grad_norm": 0.9651490692424916,
+      "learning_rate": 3.864969037001854e-05,
+      "loss": 0.0987,
+      "step": 807
+    },
+    {
+      "epoch": 2.8857142857142857,
+      "grad_norm": 0.5955395730768902,
+      "learning_rate": 3.8645509141843425e-05,
+      "loss": 0.0548,
+      "step": 808
+    },
+    {
+      "epoch": 2.8892857142857142,
+      "grad_norm": 0.60678853647265,
+      "learning_rate": 3.864132167697399e-05,
+      "loss": 0.0544,
+      "step": 809
+    },
+    {
+      "epoch": 2.892857142857143,
+      "grad_norm": 0.876302227126944,
+      "learning_rate": 3.8637127976810914e-05,
+      "loss": 0.0862,
+      "step": 810
+    },
+    {
+      "epoch": 2.8964285714285714,
+      "grad_norm": 1.0921013731408866,
+      "learning_rate": 3.8632928042756926e-05,
+      "loss": 0.1129,
+      "step": 811
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.6083058767636192,
+      "learning_rate": 3.862872187621685e-05,
+      "loss": 0.0716,
+      "step": 812
+    },
+    {
+      "epoch": 2.9035714285714285,
+      "grad_norm": 0.9358748171851029,
+      "learning_rate": 3.86245094785976e-05,
+      "loss": 0.0715,
+      "step": 813
+    },
+    {
+      "epoch": 2.907142857142857,
+      "grad_norm": 0.550141004474312,
+      "learning_rate": 3.862029085130817e-05,
+      "loss": 0.0829,
+      "step": 814
+    },
+    {
+      "epoch": 2.9107142857142856,
+      "grad_norm": 0.3874085368469643,
+      "learning_rate": 3.861606599575964e-05,
+      "loss": 0.0883,
+      "step": 815
+    },
+    {
+      "epoch": 2.914285714285714,
+      "grad_norm": 1.0032968905182122,
+      "learning_rate": 3.861183491336518e-05,
+      "loss": 0.0981,
+      "step": 816
+    },
+    {
+      "epoch": 2.9178571428571427,
+      "grad_norm": 0.7336321876328805,
+      "learning_rate": 3.860759760554003e-05,
+      "loss": 0.0512,
+      "step": 817
+    },
+    {
+      "epoch": 2.9214285714285713,
+      "grad_norm": 0.6100794729054968,
+      "learning_rate": 3.8603354073701514e-05,
+      "loss": 0.0715,
+      "step": 818
+    },
+    {
+      "epoch": 2.925,
+      "grad_norm": 0.6473152629131884,
+      "learning_rate": 3.859910431926904e-05,
+      "loss": 0.068,
+      "step": 819
+    },
+    {
+      "epoch": 2.928571428571429,
+      "grad_norm": 0.724793158701529,
+      "learning_rate": 3.85948483436641e-05,
+      "loss": 0.0793,
+      "step": 820
+    },
+    {
+      "epoch": 2.932142857142857,
+      "grad_norm": 0.6582324659262825,
+      "learning_rate": 3.8590586148310276e-05,
+      "loss": 0.0825,
+      "step": 821
+    },
+    {
+      "epoch": 2.935714285714286,
+      "grad_norm": 0.7874248944038483,
+      "learning_rate": 3.85863177346332e-05,
+      "loss": 0.0593,
+      "step": 822
+    },
+    {
+      "epoch": 2.939285714285714,
+      "grad_norm": 0.5455224106601851,
+      "learning_rate": 3.858204310406063e-05,
+      "loss": 0.1122,
+      "step": 823
+    },
+    {
+      "epoch": 2.942857142857143,
+      "grad_norm": 0.5182386179446102,
+      "learning_rate": 3.8577762258022354e-05,
+      "loss": 0.0625,
+      "step": 824
+    },
+    {
+      "epoch": 2.946428571428571,
+      "grad_norm": 0.4601989363233952,
+      "learning_rate": 3.8573475197950276e-05,
+      "loss": 0.072,
+      "step": 825
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.5792145599723899,
+      "learning_rate": 3.856918192527836e-05,
+      "loss": 0.0831,
+      "step": 826
+    },
+    {
+      "epoch": 2.9535714285714287,
+      "grad_norm": 0.9896287050709341,
+      "learning_rate": 3.856488244144265e-05,
+      "loss": 0.1074,
+      "step": 827
+    },
+    {
+      "epoch": 2.9571428571428573,
+      "grad_norm": 0.6999395687378457,
+      "learning_rate": 3.8560576747881275e-05,
+      "loss": 0.0873,
+      "step": 828
+    },
+    {
+      "epoch": 2.960714285714286,
+      "grad_norm": 0.49087294697186973,
+      "learning_rate": 3.855626484603445e-05,
+      "loss": 0.0569,
+      "step": 829
+    },
+    {
+      "epoch": 2.9642857142857144,
+      "grad_norm": 0.4004153013621006,
+      "learning_rate": 3.855194673734443e-05,
+      "loss": 0.0843,
+      "step": 830
+    },
+    {
+      "epoch": 2.967857142857143,
+      "grad_norm": 1.112702252157592,
+      "learning_rate": 3.854762242325558e-05,
+      "loss": 0.1078,
+      "step": 831
+    },
+    {
+      "epoch": 2.9714285714285715,
+      "grad_norm": 1.025002597522379,
+      "learning_rate": 3.854329190521433e-05,
+      "loss": 0.1147,
+      "step": 832
+    },
+    {
+      "epoch": 2.975,
+      "grad_norm": 0.505058244232013,
+      "learning_rate": 3.853895518466918e-05,
+      "loss": 0.0621,
+      "step": 833
+    },
+    {
+      "epoch": 2.9785714285714286,
+      "grad_norm": 0.32605130693977924,
+      "learning_rate": 3.853461226307072e-05,
+      "loss": 0.0393,
+      "step": 834
+    },
+    {
+      "epoch": 2.982142857142857,
+      "grad_norm": 0.3898411625936498,
+      "learning_rate": 3.85302631418716e-05,
+      "loss": 0.0438,
+      "step": 835
+    },
+    {
+      "epoch": 2.9857142857142858,
+      "grad_norm": 1.0959073988862615,
+      "learning_rate": 3.8525907822526545e-05,
+      "loss": 0.1131,
+      "step": 836
+    },
+    {
+      "epoch": 2.9892857142857143,
+      "grad_norm": 1.3462970145766286,
+      "learning_rate": 3.852154630649235e-05,
+      "loss": 0.0735,
+      "step": 837
+    },
+    {
+      "epoch": 2.992857142857143,
+      "grad_norm": 0.9361883017432029,
+      "learning_rate": 3.8517178595227905e-05,
+      "loss": 0.128,
+      "step": 838
+    },
+    {
+      "epoch": 2.9964285714285714,
+      "grad_norm": 0.8563932574239526,
+      "learning_rate": 3.8512804690194144e-05,
+      "loss": 0.0795,
+      "step": 839
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.5479696390344655,
+      "learning_rate": 3.8508424592854085e-05,
+      "loss": 0.0769,
+      "step": 840
+    },
+    {
+      "epoch": 3.0035714285714286,
+      "grad_norm": 0.9641407708304004,
+      "learning_rate": 3.8504038304672825e-05,
+      "loss": 0.0918,
+      "step": 841
+    },
+    {
+      "epoch": 3.007142857142857,
+      "grad_norm": 0.7037590451007083,
+      "learning_rate": 3.849964582711751e-05,
+      "loss": 0.0605,
+      "step": 842
+    },
+    {
+      "epoch": 3.0107142857142857,
+      "grad_norm": 0.49099164456606365,
+      "learning_rate": 3.849524716165739e-05,
+      "loss": 0.0423,
+      "step": 843
+    },
+    {
+      "epoch": 3.0142857142857142,
+      "grad_norm": 0.6243660034040295,
+      "learning_rate": 3.8490842309763746e-05,
+      "loss": 0.089,
+      "step": 844
+    },
+    {
+      "epoch": 3.017857142857143,
+      "grad_norm": 1.03960474291041,
+      "learning_rate": 3.848643127290995e-05,
+      "loss": 0.0998,
+      "step": 845
+    },
+    {
+      "epoch": 3.0214285714285714,
+      "grad_norm": 0.546118574164898,
+      "learning_rate": 3.8482014052571446e-05,
+      "loss": 0.0652,
+      "step": 846
+    },
+    {
+      "epoch": 3.025,
+      "grad_norm": 0.46111118597676215,
+      "learning_rate": 3.8477590650225735e-05,
+      "loss": 0.0556,
+      "step": 847
+    },
+    {
+      "epoch": 3.0285714285714285,
+      "grad_norm": 0.6374644597827944,
+      "learning_rate": 3.84731610673524e-05,
+      "loss": 0.0421,
+      "step": 848
+    },
+    {
+      "epoch": 3.032142857142857,
+      "grad_norm": 0.7168607775069339,
+      "learning_rate": 3.846872530543307e-05,
+      "loss": 0.026,
+      "step": 849
+    },
+    {
+      "epoch": 3.0357142857142856,
+      "grad_norm": 0.6828883421972055,
+      "learning_rate": 3.846428336595146e-05,
+      "loss": 0.1061,
+      "step": 850
+    },
+    {
+      "epoch": 3.039285714285714,
+      "grad_norm": 0.5825059880854055,
+      "learning_rate": 3.8459835250393344e-05,
+      "loss": 0.0503,
+      "step": 851
+    },
+    {
+      "epoch": 3.0428571428571427,
+      "grad_norm": 0.4782438433294523,
+      "learning_rate": 3.845538096024657e-05,
+      "loss": 0.0683,
+      "step": 852
+    },
+    {
+      "epoch": 3.0464285714285713,
+      "grad_norm": 1.1489603872919751,
+      "learning_rate": 3.8450920497001016e-05,
+      "loss": 0.1112,
+      "step": 853
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 1.3874181051146737,
+      "learning_rate": 3.844645386214868e-05,
+      "loss": 0.1208,
+      "step": 854
+    },
+    {
+      "epoch": 3.0535714285714284,
+      "grad_norm": 0.5767383273388068,
+      "learning_rate": 3.8441981057183584e-05,
+      "loss": 0.0716,
+      "step": 855
+    },
+    {
+      "epoch": 3.057142857142857,
+      "grad_norm": 0.4809220478755356,
+      "learning_rate": 3.843750208360184e-05,
+      "loss": 0.0505,
+      "step": 856
+    },
+    {
+      "epoch": 3.0607142857142855,
+      "grad_norm": 0.7790156406398195,
+      "learning_rate": 3.8433016942901586e-05,
+      "loss": 0.095,
+      "step": 857
+    },
+    {
+      "epoch": 3.064285714285714,
+      "grad_norm": 1.1703866759619797,
+      "learning_rate": 3.8428525636583066e-05,
+      "loss": 0.0886,
+      "step": 858
+    },
+    {
+      "epoch": 3.067857142857143,
+      "grad_norm": 0.7055850238142221,
+      "learning_rate": 3.842402816614856e-05,
+      "loss": 0.0643,
+      "step": 859
+    },
+    {
+      "epoch": 3.0714285714285716,
+      "grad_norm": 1.2165605925456058,
+      "learning_rate": 3.8419524533102424e-05,
+      "loss": 0.1101,
+      "step": 860
+    },
+    {
+      "epoch": 3.075,
+      "grad_norm": 0.5951127662845987,
+      "learning_rate": 3.841501473895106e-05,
+      "loss": 0.0906,
+      "step": 861
+    },
+    {
+      "epoch": 3.0785714285714287,
+      "grad_norm": 0.4277443042462981,
+      "learning_rate": 3.841049878520293e-05,
+      "loss": 0.0457,
+      "step": 862
+    },
+    {
+      "epoch": 3.0821428571428573,
+      "grad_norm": 0.6296878854075085,
+      "learning_rate": 3.8405976673368586e-05,
+      "loss": 0.0988,
+      "step": 863
+    },
+    {
+      "epoch": 3.085714285714286,
+      "grad_norm": 0.7712867302002314,
+      "learning_rate": 3.840144840496061e-05,
+      "loss": 0.0781,
+      "step": 864
+    },
+    {
+      "epoch": 3.0892857142857144,
+      "grad_norm": 0.5895665679286034,
+      "learning_rate": 3.839691398149365e-05,
+      "loss": 0.0717,
+      "step": 865
+    },
+    {
+      "epoch": 3.092857142857143,
+      "grad_norm": 0.5081220654134758,
+      "learning_rate": 3.839237340448441e-05,
+      "loss": 0.082,
+      "step": 866
+    },
+    {
+      "epoch": 3.0964285714285715,
+      "grad_norm": 0.555357930187513,
+      "learning_rate": 3.838782667545167e-05,
+      "loss": 0.0487,
+      "step": 867
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 1.0645316178687283,
+      "learning_rate": 3.8383273795916245e-05,
+      "loss": 0.1249,
+      "step": 868
+    },
+    {
+      "epoch": 3.1035714285714286,
+      "grad_norm": 0.34424406881235553,
+      "learning_rate": 3.8378714767401026e-05,
+      "loss": 0.067,
+      "step": 869
+    },
+    {
+      "epoch": 3.107142857142857,
+      "grad_norm": 0.5793396142009867,
+      "learning_rate": 3.8374149591430947e-05,
+      "loss": 0.0837,
+      "step": 870
+    },
+    {
+      "epoch": 3.1107142857142858,
+      "grad_norm": 0.7133634747220637,
+      "learning_rate": 3.8369578269533e-05,
+      "loss": 0.0999,
+      "step": 871
+    },
+    {
+      "epoch": 3.1142857142857143,
+      "grad_norm": 0.6892551900443395,
+      "learning_rate": 3.8365000803236244e-05,
+      "loss": 0.0896,
+      "step": 872
+    },
+    {
+      "epoch": 3.117857142857143,
+      "grad_norm": 0.7129000199365135,
+      "learning_rate": 3.8360417194071775e-05,
+      "loss": 0.0854,
+      "step": 873
+    },
+    {
+      "epoch": 3.1214285714285714,
+      "grad_norm": 0.8272387287883026,
+      "learning_rate": 3.835582744357277e-05,
+      "loss": 0.0574,
+      "step": 874
+    },
+    {
+      "epoch": 3.125,
+      "grad_norm": 0.6225185682248656,
+      "learning_rate": 3.8351231553274424e-05,
+      "loss": 0.0966,
+      "step": 875
+    },
+    {
+      "epoch": 3.1285714285714286,
+      "grad_norm": 0.4387242818435944,
+      "learning_rate": 3.834662952471402e-05,
+      "loss": 0.0881,
+      "step": 876
+    },
+    {
+      "epoch": 3.132142857142857,
+      "grad_norm": 0.7910067515846866,
+      "learning_rate": 3.834202135943088e-05,
+      "loss": 0.05,
+      "step": 877
+    },
+    {
+      "epoch": 3.1357142857142857,
+      "grad_norm": 0.726256868921072,
+      "learning_rate": 3.833740705896638e-05,
+      "loss": 0.0475,
+      "step": 878
+    },
+    {
+      "epoch": 3.1392857142857142,
+      "grad_norm": 0.91657935552738,
+      "learning_rate": 3.833278662486393e-05,
+      "loss": 0.0653,
+      "step": 879
+    },
+    {
+      "epoch": 3.142857142857143,
+      "grad_norm": 0.8042346160729629,
+      "learning_rate": 3.832816005866903e-05,
+      "loss": 0.1066,
+      "step": 880
+    },
+    {
+      "epoch": 3.1464285714285714,
+      "grad_norm": 0.5935681902275555,
+      "learning_rate": 3.83235273619292e-05,
+      "loss": 0.0458,
+      "step": 881
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.5327431227536524,
+      "learning_rate": 3.8318888536194025e-05,
+      "loss": 0.025,
+      "step": 882
+    },
+    {
+      "epoch": 3.1535714285714285,
+      "grad_norm": 0.8436400798744544,
+      "learning_rate": 3.8314243583015126e-05,
+      "loss": 0.1196,
+      "step": 883
+    },
+    {
+      "epoch": 3.157142857142857,
+      "grad_norm": 0.9871205983753772,
+      "learning_rate": 3.8309592503946184e-05,
+      "loss": 0.0783,
+      "step": 884
+    },
+    {
+      "epoch": 3.1607142857142856,
+      "grad_norm": 0.9198388059443876,
+      "learning_rate": 3.8304935300542945e-05,
+      "loss": 0.0879,
+      "step": 885
+    },
+    {
+      "epoch": 3.164285714285714,
+      "grad_norm": 0.7830779536695768,
+      "learning_rate": 3.8300271974363165e-05,
+      "loss": 0.0852,
+      "step": 886
+    },
+    {
+      "epoch": 3.1678571428571427,
+      "grad_norm": 0.6279850633872576,
+      "learning_rate": 3.829560252696668e-05,
+      "loss": 0.0549,
+      "step": 887
+    },
+    {
+      "epoch": 3.1714285714285713,
+      "grad_norm": 1.1891933235093868,
+      "learning_rate": 3.829092695991536e-05,
+      "loss": 0.0755,
+      "step": 888
+    },
+    {
+      "epoch": 3.175,
+      "grad_norm": 1.927338974998825,
+      "learning_rate": 3.8286245274773134e-05,
+      "loss": 0.1556,
+      "step": 889
+    },
+    {
+      "epoch": 3.1785714285714284,
+      "grad_norm": 0.505201653770062,
+      "learning_rate": 3.828155747310595e-05,
+      "loss": 0.0543,
+      "step": 890
+    },
+    {
+      "epoch": 3.182142857142857,
+      "grad_norm": 0.7532131038907506,
+      "learning_rate": 3.8276863556481834e-05,
+      "loss": 0.0478,
+      "step": 891
+    },
+    {
+      "epoch": 3.185714285714286,
+      "grad_norm": 0.413271753875231,
+      "learning_rate": 3.827216352647084e-05,
+      "loss": 0.0481,
+      "step": 892
+    },
+    {
+      "epoch": 3.189285714285714,
+      "grad_norm": 0.5738067497742363,
+      "learning_rate": 3.8267457384645076e-05,
+      "loss": 0.0562,
+      "step": 893
+    },
+    {
+      "epoch": 3.192857142857143,
+      "grad_norm": 0.9673542287671832,
+      "learning_rate": 3.826274513257868e-05,
+      "loss": 0.0654,
+      "step": 894
+    },
+    {
+      "epoch": 3.1964285714285716,
+      "grad_norm": 0.643337762142059,
+      "learning_rate": 3.8258026771847846e-05,
+      "loss": 0.0569,
+      "step": 895
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.8398035332721172,
+      "learning_rate": 3.825330230403081e-05,
+      "loss": 0.0603,
+      "step": 896
+    },
+    {
+      "epoch": 3.2035714285714287,
+      "grad_norm": 0.6400234288019518,
+      "learning_rate": 3.824857173070784e-05,
+      "loss": 0.0716,
+      "step": 897
+    },
+    {
+      "epoch": 3.2071428571428573,
+      "grad_norm": 0.499324951862545,
+      "learning_rate": 3.824383505346127e-05,
+      "loss": 0.0698,
+      "step": 898
+    },
+    {
+      "epoch": 3.210714285714286,
+      "grad_norm": 0.6586882954678486,
+      "learning_rate": 3.823909227387544e-05,
+      "loss": 0.0718,
+      "step": 899
+    },
+    {
+      "epoch": 3.2142857142857144,
+      "grad_norm": 1.1017991649383374,
+      "learning_rate": 3.823434339353678e-05,
+      "loss": 0.0726,
+      "step": 900
+    },
+    {
+      "epoch": 3.217857142857143,
+      "grad_norm": 0.6244140941368843,
+      "learning_rate": 3.82295884140337e-05,
+      "loss": 0.0856,
+      "step": 901
+    },
+    {
+      "epoch": 3.2214285714285715,
+      "grad_norm": 0.7982877310752905,
+      "learning_rate": 3.822482733695671e-05,
+      "loss": 0.0817,
+      "step": 902
+    },
+    {
+      "epoch": 3.225,
+      "grad_norm": 0.7473659040070937,
+      "learning_rate": 3.8220060163898315e-05,
+      "loss": 0.0837,
+      "step": 903
+    },
+    {
+      "epoch": 3.2285714285714286,
+      "grad_norm": 0.5873527758877337,
+      "learning_rate": 3.821528689645307e-05,
+      "loss": 0.0353,
+      "step": 904
+    },
+    {
+      "epoch": 3.232142857142857,
+      "grad_norm": 1.4383626939615313,
+      "learning_rate": 3.82105075362176e-05,
+      "loss": 0.1105,
+      "step": 905
+    },
+    {
+      "epoch": 3.2357142857142858,
+      "grad_norm": 0.574853515126818,
+      "learning_rate": 3.8205722084790525e-05,
+      "loss": 0.0715,
+      "step": 906
+    },
+    {
+      "epoch": 3.2392857142857143,
+      "grad_norm": 0.8574694039402991,
+      "learning_rate": 3.8200930543772524e-05,
+      "loss": 0.0751,
+      "step": 907
+    },
+    {
+      "epoch": 3.242857142857143,
+      "grad_norm": 0.7964896123959989,
+      "learning_rate": 3.81961329147663e-05,
+      "loss": 0.0517,
+      "step": 908
+    },
+    {
+      "epoch": 3.2464285714285714,
+      "grad_norm": 0.7669387192135553,
+      "learning_rate": 3.8191329199376615e-05,
+      "loss": 0.102,
+      "step": 909
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.8090364752561764,
+      "learning_rate": 3.818651939921025e-05,
+      "loss": 0.089,
+      "step": 910
+    },
+    {
+      "epoch": 3.2535714285714286,
+      "grad_norm": 1.0445790900481973,
+      "learning_rate": 3.818170351587601e-05,
+      "loss": 0.0762,
+      "step": 911
+    },
+    {
+      "epoch": 3.257142857142857,
+      "grad_norm": 0.4572568776140298,
+      "learning_rate": 3.817688155098477e-05,
+      "loss": 0.08,
+      "step": 912
+    },
+    {
+      "epoch": 3.2607142857142857,
+      "grad_norm": 0.7765726620640928,
+      "learning_rate": 3.81720535061494e-05,
+      "loss": 0.0642,
+      "step": 913
+    },
+    {
+      "epoch": 3.2642857142857142,
+      "grad_norm": 0.5998259684766428,
+      "learning_rate": 3.816721938298484e-05,
+      "loss": 0.0711,
+      "step": 914
+    },
+    {
+      "epoch": 3.267857142857143,
+      "grad_norm": 0.7054531489742752,
+      "learning_rate": 3.816237918310802e-05,
+      "loss": 0.1251,
+      "step": 915
+    },
+    {
+      "epoch": 3.2714285714285714,
+      "grad_norm": 0.7673887215230498,
+      "learning_rate": 3.8157532908137954e-05,
+      "loss": 0.0513,
+      "step": 916
+    },
+    {
+      "epoch": 3.275,
+      "grad_norm": 1.15543176308973,
+      "learning_rate": 3.815268055969564e-05,
+      "loss": 0.1407,
+      "step": 917
+    },
+    {
+      "epoch": 3.2785714285714285,
+      "grad_norm": 1.0314567135899655,
+      "learning_rate": 3.814782213940415e-05,
+      "loss": 0.1207,
+      "step": 918
+    },
+    {
+      "epoch": 3.282142857142857,
+      "grad_norm": 0.9038889519145629,
+      "learning_rate": 3.814295764888855e-05,
+      "loss": 0.0664,
+      "step": 919
+    },
+    {
+      "epoch": 3.2857142857142856,
+      "grad_norm": 0.6762006010334416,
+      "learning_rate": 3.8138087089775956e-05,
+      "loss": 0.083,
+      "step": 920
+    },
+    {
+      "epoch": 3.289285714285714,
+      "grad_norm": 0.6104005244405359,
+      "learning_rate": 3.813321046369551e-05,
+      "loss": 0.0575,
+      "step": 921
+    },
+    {
+      "epoch": 3.2928571428571427,
+      "grad_norm": 0.5807919358470252,
+      "learning_rate": 3.812832777227839e-05,
+      "loss": 0.0648,
+      "step": 922
+    },
+    {
+      "epoch": 3.2964285714285713,
+      "grad_norm": 0.6088738699657089,
+      "learning_rate": 3.812343901715779e-05,
+      "loss": 0.052,
+      "step": 923
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.7274636820610884,
+      "learning_rate": 3.811854419996894e-05,
+      "loss": 0.1021,
+      "step": 924
+    },
+    {
+      "epoch": 3.3035714285714284,
+      "grad_norm": 0.4689931443172558,
+      "learning_rate": 3.811364332234909e-05,
+      "loss": 0.1046,
+      "step": 925
+    },
+    {
+      "epoch": 3.307142857142857,
+      "grad_norm": 0.5128806143771191,
+      "learning_rate": 3.810873638593754e-05,
+      "loss": 0.0553,
+      "step": 926
+    },
+    {
+      "epoch": 3.310714285714286,
+      "grad_norm": 0.7693709249993729,
+      "learning_rate": 3.810382339237559e-05,
+      "loss": 0.1024,
+      "step": 927
+    },
+    {
+      "epoch": 3.314285714285714,
+      "grad_norm": 1.3855391289590189,
+      "learning_rate": 3.809890434330658e-05,
+      "loss": 0.0567,
+      "step": 928
+    },
+    {
+      "epoch": 3.317857142857143,
+      "grad_norm": 0.6480828137970878,
+      "learning_rate": 3.809397924037586e-05,
+      "loss": 0.0757,
+      "step": 929
+    },
+    {
+      "epoch": 3.3214285714285716,
+      "grad_norm": 0.5785247063191642,
+      "learning_rate": 3.8089048085230826e-05,
+      "loss": 0.0497,
+      "step": 930
+    },
+    {
+      "epoch": 3.325,
+      "grad_norm": 0.6956557690567091,
+      "learning_rate": 3.8084110879520893e-05,
+      "loss": 0.0695,
+      "step": 931
+    },
+    {
+      "epoch": 3.3285714285714287,
+      "grad_norm": 0.9359090278529301,
+      "learning_rate": 3.807916762489749e-05,
+      "loss": 0.123,
+      "step": 932
+    },
+    {
+      "epoch": 3.3321428571428573,
+      "grad_norm": 0.6700332677323655,
+      "learning_rate": 3.807421832301407e-05,
+      "loss": 0.1139,
+      "step": 933
+    },
+    {
+      "epoch": 3.335714285714286,
+      "grad_norm": 0.6105682350257381,
+      "learning_rate": 3.806926297552612e-05,
+      "loss": 0.0554,
+      "step": 934
+    },
+    {
+      "epoch": 3.3392857142857144,
+      "grad_norm": 0.6757805635987638,
+      "learning_rate": 3.806430158409114e-05,
+      "loss": 0.0457,
+      "step": 935
+    },
+    {
+      "epoch": 3.342857142857143,
+      "grad_norm": 0.5870310041878234,
+      "learning_rate": 3.8059334150368666e-05,
+      "loss": 0.1277,
+      "step": 936
+    },
+    {
+      "epoch": 3.3464285714285715,
+      "grad_norm": 0.8666582448602532,
+      "learning_rate": 3.805436067602023e-05,
+      "loss": 0.0661,
+      "step": 937
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.7254697107058538,
+      "learning_rate": 3.80493811627094e-05,
+      "loss": 0.0854,
+      "step": 938
+    },
+    {
+      "epoch": 3.3535714285714286,
+      "grad_norm": 0.6169749192625623,
+      "learning_rate": 3.804439561210177e-05,
+      "loss": 0.0685,
+      "step": 939
+    },
+    {
+      "epoch": 3.357142857142857,
+      "grad_norm": 0.5096611117005678,
+      "learning_rate": 3.803940402586494e-05,
+      "loss": 0.0671,
+      "step": 940
+    },
+    {
+      "epoch": 3.3607142857142858,
+      "grad_norm": 1.1552479685525818,
+      "learning_rate": 3.803440640566853e-05,
+      "loss": 0.0519,
+      "step": 941
+    },
+    {
+      "epoch": 3.3642857142857143,
+      "grad_norm": 0.43715321082368913,
+      "learning_rate": 3.80294027531842e-05,
+      "loss": 0.0263,
+      "step": 942
+    },
+    {
+      "epoch": 3.367857142857143,
+      "grad_norm": 0.40976638170981333,
+      "learning_rate": 3.802439307008559e-05,
+      "loss": 0.0627,
+      "step": 943
+    },
+    {
+      "epoch": 3.3714285714285714,
+      "grad_norm": 0.6469701798859473,
+      "learning_rate": 3.801937735804838e-05,
+      "loss": 0.0337,
+      "step": 944
+    },
+    {
+      "epoch": 3.375,
+      "grad_norm": 0.3756692006890433,
+      "learning_rate": 3.801435561875028e-05,
+      "loss": 0.0366,
+      "step": 945
+    },
+    {
+      "epoch": 3.3785714285714286,
+      "grad_norm": 1.0830187232391923,
+      "learning_rate": 3.800932785387099e-05,
+      "loss": 0.0852,
+      "step": 946
+    },
+    {
+      "epoch": 3.382142857142857,
+      "grad_norm": 1.3007776186535454,
+      "learning_rate": 3.8004294065092234e-05,
+      "loss": 0.148,
+      "step": 947
+    },
+    {
+      "epoch": 3.3857142857142857,
+      "grad_norm": 0.7647783410880568,
+      "learning_rate": 3.7999254254097756e-05,
+      "loss": 0.0601,
+      "step": 948
+    },
+    {
+      "epoch": 3.3892857142857142,
+      "grad_norm": 0.8046704081504243,
+      "learning_rate": 3.799420842257331e-05,
+      "loss": 0.0777,
+      "step": 949
+    },
+    {
+      "epoch": 3.392857142857143,
+      "grad_norm": 0.7361136561256391,
+      "learning_rate": 3.7989156572206655e-05,
+      "loss": 0.0785,
+      "step": 950
+    },
+    {
+      "epoch": 3.3964285714285714,
+      "grad_norm": 0.8218693615115771,
+      "learning_rate": 3.79840987046876e-05,
+      "loss": 0.0846,
+      "step": 951
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.6809402179526319,
+      "learning_rate": 3.797903482170791e-05,
+      "loss": 0.0622,
+      "step": 952
+    },
+    {
+      "epoch": 3.4035714285714285,
+      "grad_norm": 0.3757566870999362,
+      "learning_rate": 3.797396492496141e-05,
+      "loss": 0.026,
+      "step": 953
+    },
+    {
+      "epoch": 3.407142857142857,
+      "grad_norm": 0.8013299748541074,
+      "learning_rate": 3.7968889016143904e-05,
+      "loss": 0.0853,
+      "step": 954
+    },
+    {
+      "epoch": 3.4107142857142856,
+      "grad_norm": 0.7625483710640657,
+      "learning_rate": 3.796380709695324e-05,
+      "loss": 0.1434,
+      "step": 955
+    },
+    {
+      "epoch": 3.414285714285714,
+      "grad_norm": 0.46796425187630014,
+      "learning_rate": 3.795871916908924e-05,
+      "loss": 0.0799,
+      "step": 956
+    },
+    {
+      "epoch": 3.4178571428571427,
+      "grad_norm": 0.5608915724639482,
+      "learning_rate": 3.795362523425377e-05,
+      "loss": 0.0611,
+      "step": 957
+    },
+    {
+      "epoch": 3.4214285714285713,
+      "grad_norm": 0.5144069241120354,
+      "learning_rate": 3.7948525294150676e-05,
+      "loss": 0.0858,
+      "step": 958
+    },
+    {
+      "epoch": 3.425,
+      "grad_norm": 0.8991145590755437,
+      "learning_rate": 3.794341935048582e-05,
+      "loss": 0.0767,
+      "step": 959
+    },
+    {
+      "epoch": 3.4285714285714284,
+      "grad_norm": 0.5776173037096167,
+      "learning_rate": 3.79383074049671e-05,
+      "loss": 0.0634,
+      "step": 960
+    },
+    {
+      "epoch": 3.432142857142857,
+      "grad_norm": 0.4230850223055977,
+      "learning_rate": 3.7933189459304376e-05,
+      "loss": 0.066,
+      "step": 961
+    },
+    {
+      "epoch": 3.435714285714286,
+      "grad_norm": 0.5355409894143948,
+      "learning_rate": 3.792806551520955e-05,
+      "loss": 0.0866,
+      "step": 962
+    },
+    {
+      "epoch": 3.439285714285714,
+      "grad_norm": 0.45686455402832626,
+      "learning_rate": 3.792293557439652e-05,
+      "loss": 0.1031,
+      "step": 963
+    },
+    {
+      "epoch": 3.442857142857143,
+      "grad_norm": 0.6050344614378894,
+      "learning_rate": 3.7917799638581184e-05,
+      "loss": 0.0945,
+      "step": 964
+    },
+    {
+      "epoch": 3.4464285714285716,
+      "grad_norm": 0.9707129244278566,
+      "learning_rate": 3.7912657709481454e-05,
+      "loss": 0.0942,
+      "step": 965
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.7349410394546875,
+      "learning_rate": 3.7907509788817234e-05,
+      "loss": 0.1046,
+      "step": 966
+    },
+    {
+      "epoch": 3.4535714285714287,
+      "grad_norm": 0.6558204330542616,
+      "learning_rate": 3.790235587831045e-05,
+      "loss": 0.0845,
+      "step": 967
+    },
+    {
+      "epoch": 3.4571428571428573,
+      "grad_norm": 0.659545618451158,
+      "learning_rate": 3.789719597968501e-05,
+      "loss": 0.0577,
+      "step": 968
+    },
+    {
+      "epoch": 3.460714285714286,
+      "grad_norm": 0.720987613308211,
+      "learning_rate": 3.7892030094666846e-05,
+      "loss": 0.05,
+      "step": 969
+    },
+    {
+      "epoch": 3.4642857142857144,
+      "grad_norm": 0.9196434247508526,
+      "learning_rate": 3.788685822498388e-05,
+      "loss": 0.1075,
+      "step": 970
+    },
+    {
+      "epoch": 3.467857142857143,
+      "grad_norm": 0.48255744403020556,
+      "learning_rate": 3.788168037236604e-05,
+      "loss": 0.0562,
+      "step": 971
+    },
+    {
+      "epoch": 3.4714285714285715,
+      "grad_norm": 0.4405090383194801,
+      "learning_rate": 3.787649653854527e-05,
+      "loss": 0.0333,
+      "step": 972
+    },
+    {
+      "epoch": 3.475,
+      "grad_norm": 1.015010156925469,
+      "learning_rate": 3.787130672525546e-05,
+      "loss": 0.1148,
+      "step": 973
+    },
+    {
+      "epoch": 3.4785714285714286,
+      "grad_norm": 0.548383713640926,
+      "learning_rate": 3.786611093423258e-05,
+      "loss": 0.0384,
+      "step": 974
+    },
+    {
+      "epoch": 3.482142857142857,
+      "grad_norm": 1.1567928725327659,
+      "learning_rate": 3.786090916721453e-05,
+      "loss": 0.1471,
+      "step": 975
+    },
+    {
+      "epoch": 3.4857142857142858,
+      "grad_norm": 0.9690633019509476,
+      "learning_rate": 3.785570142594125e-05,
+      "loss": 0.082,
+      "step": 976
+    },
+    {
+      "epoch": 3.4892857142857143,
+      "grad_norm": 1.4058675058300791,
+      "learning_rate": 3.7850487712154664e-05,
+      "loss": 0.0847,
+      "step": 977
+    },
+    {
+      "epoch": 3.492857142857143,
+      "grad_norm": 1.1600398457966927,
+      "learning_rate": 3.78452680275987e-05,
+      "loss": 0.0646,
+      "step": 978
+    },
+    {
+      "epoch": 3.4964285714285714,
+      "grad_norm": 1.1724925666003267,
+      "learning_rate": 3.784004237401927e-05,
+      "loss": 0.0794,
+      "step": 979
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.9649097124408625,
+      "learning_rate": 3.783481075316429e-05,
+      "loss": 0.064,
+      "step": 980
+    },
+    {
+      "epoch": 3.5035714285714286,
+      "grad_norm": 0.9281904577155056,
+      "learning_rate": 3.782957316678368e-05,
+      "loss": 0.0678,
+      "step": 981
+    },
+    {
+      "epoch": 3.507142857142857,
+      "grad_norm": 0.7354481452664775,
+      "learning_rate": 3.7824329616629356e-05,
+      "loss": 0.054,
+      "step": 982
+    },
+    {
+      "epoch": 3.5107142857142857,
+      "grad_norm": 0.7806456754271344,
+      "learning_rate": 3.78190801044552e-05,
+      "loss": 0.1092,
+      "step": 983
+    },
+    {
+      "epoch": 3.5142857142857142,
+      "grad_norm": 0.5383500053324658,
+      "learning_rate": 3.781382463201712e-05,
+      "loss": 0.0971,
+      "step": 984
+    },
+    {
+      "epoch": 3.517857142857143,
+      "grad_norm": 0.6013474544860657,
+      "learning_rate": 3.780856320107301e-05,
+      "loss": 0.04,
+      "step": 985
+    },
+    {
+      "epoch": 3.5214285714285714,
+      "grad_norm": 1.2269652698099127,
+      "learning_rate": 3.780329581338276e-05,
+      "loss": 0.1086,
+      "step": 986
+    },
+    {
+      "epoch": 3.525,
+      "grad_norm": 0.6506047062076193,
+      "learning_rate": 3.7798022470708236e-05,
+      "loss": 0.0776,
+      "step": 987
+    },
+    {
+      "epoch": 3.5285714285714285,
+      "grad_norm": 0.9218846772273696,
+      "learning_rate": 3.77927431748133e-05,
+      "loss": 0.083,
+      "step": 988
+    },
+    {
+      "epoch": 3.532142857142857,
+      "grad_norm": 0.4631906490231701,
+      "learning_rate": 3.7787457927463834e-05,
+      "loss": 0.0706,
+      "step": 989
+    },
+    {
+      "epoch": 3.5357142857142856,
+      "grad_norm": 1.2933436386470765,
+      "learning_rate": 3.778216673042767e-05,
+      "loss": 0.1376,
+      "step": 990
+    },
+    {
+      "epoch": 3.539285714285714,
+      "grad_norm": 0.6337232568416719,
+      "learning_rate": 3.777686958547465e-05,
+      "loss": 0.098,
+      "step": 991
+    },
+    {
+      "epoch": 3.5428571428571427,
+      "grad_norm": 0.649568289523289,
+      "learning_rate": 3.777156649437661e-05,
+      "loss": 0.0638,
+      "step": 992
+    },
+    {
+      "epoch": 3.5464285714285713,
+      "grad_norm": 1.0491367613056086,
+      "learning_rate": 3.7766257458907374e-05,
+      "loss": 0.1057,
+      "step": 993
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.8516767998885332,
+      "learning_rate": 3.776094248084273e-05,
+      "loss": 0.0403,
+      "step": 994
+    },
+    {
+      "epoch": 3.553571428571429,
+      "grad_norm": 1.0084331146133463,
+      "learning_rate": 3.77556215619605e-05,
+      "loss": 0.0875,
+      "step": 995
+    },
+    {
+      "epoch": 3.557142857142857,
+      "grad_norm": 0.8977548857853122,
+      "learning_rate": 3.775029470404044e-05,
+      "loss": 0.093,
+      "step": 996
+    },
+    {
+      "epoch": 3.560714285714286,
+      "grad_norm": 1.2912073766436472,
+      "learning_rate": 3.774496190886433e-05,
+      "loss": 0.1,
+      "step": 997
+    },
+    {
+      "epoch": 3.564285714285714,
+      "grad_norm": 0.5480345113065584,
+      "learning_rate": 3.773962317821593e-05,
+      "loss": 0.1328,
+      "step": 998
+    },
+    {
+      "epoch": 3.567857142857143,
+      "grad_norm": 0.7350106576691643,
+      "learning_rate": 3.773427851388098e-05,
+      "loss": 0.0953,
+      "step": 999
+    },
+    {
+      "epoch": 3.571428571428571,
+      "grad_norm": 0.6559415734032069,
+      "learning_rate": 3.7728927917647195e-05,
+      "loss": 0.1032,
+      "step": 1000
+    },
+    {
+      "epoch": 3.575,
+      "grad_norm": 0.5504602711278983,
+      "learning_rate": 3.7723571391304285e-05,
+      "loss": 0.0563,
+      "step": 1001
+    },
+    {
+      "epoch": 3.5785714285714287,
+      "grad_norm": 0.7194248277649935,
+      "learning_rate": 3.7718208936643954e-05,
+      "loss": 0.0764,
+      "step": 1002
+    },
+    {
+      "epoch": 3.5821428571428573,
+      "grad_norm": 0.4648929885924148,
+      "learning_rate": 3.771284055545986e-05,
+      "loss": 0.0785,
+      "step": 1003
+    },
+    {
+      "epoch": 3.585714285714286,
+      "grad_norm": 1.0742214874752882,
+      "learning_rate": 3.7707466249547675e-05,
+      "loss": 0.1353,
+      "step": 1004
+    },
+    {
+      "epoch": 3.5892857142857144,
+      "grad_norm": 0.5590926860123052,
+      "learning_rate": 3.770208602070503e-05,
+      "loss": 0.0537,
+      "step": 1005
+    },
+    {
+      "epoch": 3.592857142857143,
+      "grad_norm": 0.915359126833675,
+      "learning_rate": 3.769669987073155e-05,
+      "loss": 0.0775,
+      "step": 1006
+    },
+    {
+      "epoch": 3.5964285714285715,
+      "grad_norm": 0.6096919760143035,
+      "learning_rate": 3.7691307801428844e-05,
+      "loss": 0.0658,
+      "step": 1007
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.45751185936179906,
+      "learning_rate": 3.768590981460047e-05,
+      "loss": 0.1127,
+      "step": 1008
+    },
+    {
+      "epoch": 3.6035714285714286,
+      "grad_norm": 0.5978334503383611,
+      "learning_rate": 3.768050591205201e-05,
+      "loss": 0.0696,
+      "step": 1009
+    },
+    {
+      "epoch": 3.607142857142857,
+      "grad_norm": 1.2406396448451613,
+      "learning_rate": 3.7675096095591e-05,
+      "loss": 0.0946,
+      "step": 1010
+    },
+    {
+      "epoch": 3.6107142857142858,
+      "grad_norm": 0.6421929448449027,
+      "learning_rate": 3.7669680367026953e-05,
+      "loss": 0.0514,
+      "step": 1011
+    },
+    {
+      "epoch": 3.6142857142857143,
+      "grad_norm": 0.9416341577774511,
+      "learning_rate": 3.766425872817136e-05,
+      "loss": 0.077,
+      "step": 1012
+    },
+    {
+      "epoch": 3.617857142857143,
+      "grad_norm": 0.6473645103219646,
+      "learning_rate": 3.7658831180837696e-05,
+      "loss": 0.0657,
+      "step": 1013
+    },
+    {
+      "epoch": 3.6214285714285714,
+      "grad_norm": 2.286769175504995,
+      "learning_rate": 3.765339772684141e-05,
+      "loss": 0.1624,
+      "step": 1014
+    },
+    {
+      "epoch": 3.625,
+      "grad_norm": 0.8516235688865064,
+      "learning_rate": 3.764795836799993e-05,
+      "loss": 0.0894,
+      "step": 1015
+    },
+    {
+      "epoch": 3.6285714285714286,
+      "grad_norm": 0.6498856293582111,
+      "learning_rate": 3.764251310613265e-05,
+      "loss": 0.1129,
+      "step": 1016
+    },
+    {
+      "epoch": 3.632142857142857,
+      "grad_norm": 0.6260668759446473,
+      "learning_rate": 3.763706194306094e-05,
+      "loss": 0.0943,
+      "step": 1017
+    },
+    {
+      "epoch": 3.6357142857142857,
+      "grad_norm": 0.7646002574449627,
+      "learning_rate": 3.763160488060816e-05,
+      "loss": 0.0773,
+      "step": 1018
+    },
+    {
+      "epoch": 3.6392857142857142,
+      "grad_norm": 1.3899017739406356,
+      "learning_rate": 3.762614192059961e-05,
+      "loss": 0.0971,
+      "step": 1019
+    },
+    {
+      "epoch": 3.642857142857143,
+      "grad_norm": 1.9734017847795429,
+      "learning_rate": 3.76206730648626e-05,
+      "loss": 0.1011,
+      "step": 1020
+    },
+    {
+      "epoch": 3.6464285714285714,
+      "grad_norm": 0.6441427452745135,
+      "learning_rate": 3.7615198315226394e-05,
+      "loss": 0.1071,
+      "step": 1021
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.5025739606162641,
+      "learning_rate": 3.760971767352222e-05,
+      "loss": 0.0914,
+      "step": 1022
+    },
+    {
+      "epoch": 3.6535714285714285,
+      "grad_norm": 0.3621903330440995,
+      "learning_rate": 3.760423114158329e-05,
+      "loss": 0.0366,
+      "step": 1023
+    },
+    {
+      "epoch": 3.657142857142857,
+      "grad_norm": 0.5747647412967415,
+      "learning_rate": 3.759873872124479e-05,
+      "loss": 0.0505,
+      "step": 1024
+    },
+    {
+      "epoch": 3.6607142857142856,
+      "grad_norm": 0.6120674055313959,
+      "learning_rate": 3.7593240414343846e-05,
+      "loss": 0.0759,
+      "step": 1025
+    },
+    {
+      "epoch": 3.664285714285714,
+      "grad_norm": 0.39680731044084616,
+      "learning_rate": 3.75877362227196e-05,
+      "loss": 0.0847,
+      "step": 1026
+    },
+    {
+      "epoch": 3.6678571428571427,
+      "grad_norm": 0.5114178562884194,
+      "learning_rate": 3.758222614821311e-05,
+      "loss": 0.0919,
+      "step": 1027
+    },
+    {
+      "epoch": 3.6714285714285713,
+      "grad_norm": 0.977594858581119,
+      "learning_rate": 3.757671019266746e-05,
+      "loss": 0.0868,
+      "step": 1028
+    },
+    {
+      "epoch": 3.675,
+      "grad_norm": 0.6223621743422217,
+      "learning_rate": 3.757118835792764e-05,
+      "loss": 0.0696,
+      "step": 1029
+    },
+    {
+      "epoch": 3.678571428571429,
+      "grad_norm": 0.5415015528288507,
+      "learning_rate": 3.756566064584065e-05,
+      "loss": 0.0911,
+      "step": 1030
+    },
+    {
+      "epoch": 3.682142857142857,
+      "grad_norm": 0.3549209961340171,
+      "learning_rate": 3.7560127058255443e-05,
+      "loss": 0.0447,
+      "step": 1031
+    },
+    {
+      "epoch": 3.685714285714286,
+      "grad_norm": 0.7548979099433814,
+      "learning_rate": 3.755458759702293e-05,
+      "loss": 0.1189,
+      "step": 1032
+    },
+    {
+      "epoch": 3.689285714285714,
+      "grad_norm": 0.5154569296736006,
+      "learning_rate": 3.7549042263996e-05,
+      "loss": 0.0383,
+      "step": 1033
+    },
+    {
+      "epoch": 3.692857142857143,
+      "grad_norm": 1.0234030923702622,
+      "learning_rate": 3.7543491061029496e-05,
+      "loss": 0.0694,
+      "step": 1034
+    },
+    {
+      "epoch": 3.696428571428571,
+      "grad_norm": 0.7377495101614261,
+      "learning_rate": 3.753793398998022e-05,
+      "loss": 0.0837,
+      "step": 1035
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.7847209082584407,
+      "learning_rate": 3.753237105270696e-05,
+      "loss": 0.0618,
+      "step": 1036
+    },
+    {
+      "epoch": 3.7035714285714287,
+      "grad_norm": 0.6810026262249758,
+      "learning_rate": 3.752680225107043e-05,
+      "loss": 0.0802,
+      "step": 1037
+    },
+    {
+      "epoch": 3.7071428571428573,
+      "grad_norm": 0.43513144626748734,
+      "learning_rate": 3.7521227586933344e-05,
+      "loss": 0.055,
+      "step": 1038
+    },
+    {
+      "epoch": 3.710714285714286,
+      "grad_norm": 0.586896127054456,
+      "learning_rate": 3.751564706216035e-05,
+      "loss": 0.0733,
+      "step": 1039
+    },
+    {
+      "epoch": 3.7142857142857144,
+      "grad_norm": 0.7479158119492524,
+      "learning_rate": 3.751006067861807e-05,
+      "loss": 0.0796,
+      "step": 1040
+    },
+    {
+      "epoch": 3.717857142857143,
+      "grad_norm": 0.5541371849606521,
+      "learning_rate": 3.7504468438175076e-05,
+      "loss": 0.0859,
+      "step": 1041
+    },
+    {
+      "epoch": 3.7214285714285715,
+      "grad_norm": 0.5969036396050937,
+      "learning_rate": 3.7498870342701906e-05,
+      "loss": 0.0371,
+      "step": 1042
+    },
+    {
+      "epoch": 3.725,
+      "grad_norm": 0.39899038138087306,
+      "learning_rate": 3.749326639407105e-05,
+      "loss": 0.0792,
+      "step": 1043
+    },
+    {
+      "epoch": 3.7285714285714286,
+      "grad_norm": 0.5627675578330008,
+      "learning_rate": 3.748765659415697e-05,
+      "loss": 0.0735,
+      "step": 1044
+    },
+    {
+      "epoch": 3.732142857142857,
+      "grad_norm": 0.5289235923705053,
+      "learning_rate": 3.748204094483607e-05,
+      "loss": 0.0667,
+      "step": 1045
+    },
+    {
+      "epoch": 3.7357142857142858,
+      "grad_norm": 0.5169462530598051,
+      "learning_rate": 3.747641944798671e-05,
+      "loss": 0.0591,
+      "step": 1046
+    },
+    {
+      "epoch": 3.7392857142857143,
+      "grad_norm": 0.591796628220982,
+      "learning_rate": 3.747079210548922e-05,
+      "loss": 0.0796,
+      "step": 1047
+    },
+    {
+      "epoch": 3.742857142857143,
+      "grad_norm": 0.3756083138897675,
+      "learning_rate": 3.746515891922587e-05,
+      "loss": 0.0484,
+      "step": 1048
+    },
+    {
+      "epoch": 3.7464285714285714,
+      "grad_norm": 1.3980853978421632,
+      "learning_rate": 3.7459519891080905e-05,
+      "loss": 0.0964,
+      "step": 1049
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.4766986574827897,
+      "learning_rate": 3.7453875022940494e-05,
+      "loss": 0.061,
+      "step": 1050
+    },
+    {
+      "epoch": 3.7535714285714286,
+      "grad_norm": 0.8293875488463219,
+      "learning_rate": 3.744822431669279e-05,
+      "loss": 0.0694,
+      "step": 1051
+    },
+    {
+      "epoch": 3.757142857142857,
+      "grad_norm": 0.5669160184455709,
+      "learning_rate": 3.744256777422788e-05,
+      "loss": 0.0681,
+      "step": 1052
+    },
+    {
+      "epoch": 3.7607142857142857,
+      "grad_norm": 0.5121643010342525,
+      "learning_rate": 3.7436905397437794e-05,
+      "loss": 0.1005,
+      "step": 1053
+    },
+    {
+      "epoch": 3.7642857142857142,
+      "grad_norm": 0.7689177392050482,
+      "learning_rate": 3.743123718821656e-05,
+      "loss": 0.052,
+      "step": 1054
+    },
+    {
+      "epoch": 3.767857142857143,
+      "grad_norm": 0.7583667818478721,
+      "learning_rate": 3.742556314846009e-05,
+      "loss": 0.0836,
+      "step": 1055
+    },
+    {
+      "epoch": 3.7714285714285714,
+      "grad_norm": 1.0078384293185376,
+      "learning_rate": 3.7419883280066305e-05,
+      "loss": 0.0491,
+      "step": 1056
+    },
+    {
+      "epoch": 3.775,
+      "grad_norm": 0.8141921123325013,
+      "learning_rate": 3.741419758493504e-05,
+      "loss": 0.08,
+      "step": 1057
+    },
+    {
+      "epoch": 3.7785714285714285,
+      "grad_norm": 0.9037596002699976,
+      "learning_rate": 3.740850606496809e-05,
+      "loss": 0.1206,
+      "step": 1058
+    },
+    {
+      "epoch": 3.782142857142857,
+      "grad_norm": 1.1710259314311737,
+      "learning_rate": 3.74028087220692e-05,
+      "loss": 0.0736,
+      "step": 1059
+    },
+    {
+      "epoch": 3.7857142857142856,
+      "grad_norm": 0.9249690463923623,
+      "learning_rate": 3.739710555814407e-05,
+      "loss": 0.0967,
+      "step": 1060
+    },
+    {
+      "epoch": 3.789285714285714,
+      "grad_norm": 0.9844914388879862,
+      "learning_rate": 3.739139657510033e-05,
+      "loss": 0.1041,
+      "step": 1061
+    },
+    {
+      "epoch": 3.7928571428571427,
+      "grad_norm": 0.3570135235779621,
+      "learning_rate": 3.738568177484756e-05,
+      "loss": 0.0385,
+      "step": 1062
+    },
+    {
+      "epoch": 3.7964285714285713,
+      "grad_norm": 0.8536721272829356,
+      "learning_rate": 3.73799611592973e-05,
+      "loss": 0.0665,
+      "step": 1063
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 1.0506230981680398,
+      "learning_rate": 3.737423473036303e-05,
+      "loss": 0.0747,
+      "step": 1064
+    },
+    {
+      "epoch": 3.803571428571429,
+      "grad_norm": 0.8430839198082948,
+      "learning_rate": 3.736850248996015e-05,
+      "loss": 0.0751,
+      "step": 1065
+    },
+    {
+      "epoch": 3.807142857142857,
+      "grad_norm": 1.9034408933723848,
+      "learning_rate": 3.7362764440006046e-05,
+      "loss": 0.1198,
+      "step": 1066
+    },
+    {
+      "epoch": 3.810714285714286,
+      "grad_norm": 1.1723104838009704,
+      "learning_rate": 3.7357020582420013e-05,
+      "loss": 0.0942,
+      "step": 1067
+    },
+    {
+      "epoch": 3.814285714285714,
+      "grad_norm": 0.6398403800880634,
+      "learning_rate": 3.7351270919123305e-05,
+      "loss": 0.0554,
+      "step": 1068
+    },
+    {
+      "epoch": 3.817857142857143,
+      "grad_norm": 1.2075427188189034,
+      "learning_rate": 3.734551545203911e-05,
+      "loss": 0.0759,
+      "step": 1069
+    },
+    {
+      "epoch": 3.821428571428571,
+      "grad_norm": 0.7632409068981203,
+      "learning_rate": 3.733975418309257e-05,
+      "loss": 0.0708,
+      "step": 1070
+    },
+    {
+      "epoch": 3.825,
+      "grad_norm": 0.8455931026734321,
+      "learning_rate": 3.7333987114210754e-05,
+      "loss": 0.1195,
+      "step": 1071
+    },
+    {
+      "epoch": 3.8285714285714287,
+      "grad_norm": 1.022788045819547,
+      "learning_rate": 3.7328214247322675e-05,
+      "loss": 0.0857,
+      "step": 1072
+    },
+    {
+      "epoch": 3.8321428571428573,
+      "grad_norm": 0.7139053441752193,
+      "learning_rate": 3.7322435584359285e-05,
+      "loss": 0.0657,
+      "step": 1073
+    },
+    {
+      "epoch": 3.835714285714286,
+      "grad_norm": 0.4006751178624328,
+      "learning_rate": 3.7316651127253474e-05,
+      "loss": 0.046,
+      "step": 1074
+    },
+    {
+      "epoch": 3.8392857142857144,
+      "grad_norm": 0.4411573196028674,
+      "learning_rate": 3.731086087794008e-05,
+      "loss": 0.062,
+      "step": 1075
+    },
+    {
+      "epoch": 3.842857142857143,
+      "grad_norm": 0.7552034987024792,
+      "learning_rate": 3.7305064838355865e-05,
+      "loss": 0.0377,
+      "step": 1076
+    },
+    {
+      "epoch": 3.8464285714285715,
+      "grad_norm": 1.329358666464491,
+      "learning_rate": 3.7299263010439535e-05,
+      "loss": 0.0649,
+      "step": 1077
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.5562272735152852,
+      "learning_rate": 3.729345539613173e-05,
+      "loss": 0.1,
+      "step": 1078
+    },
+    {
+      "epoch": 3.8535714285714286,
+      "grad_norm": 0.43387948240942326,
+      "learning_rate": 3.7287641997375025e-05,
+      "loss": 0.0547,
+      "step": 1079
+    },
+    {
+      "epoch": 3.857142857142857,
+      "grad_norm": 0.759041638333593,
+      "learning_rate": 3.728182281611393e-05,
+      "loss": 0.0613,
+      "step": 1080
+    },
+    {
+      "epoch": 3.8607142857142858,
+      "grad_norm": 0.5261618859162001,
+      "learning_rate": 3.7275997854294885e-05,
+      "loss": 0.063,
+      "step": 1081
+    },
+    {
+      "epoch": 3.8642857142857143,
+      "grad_norm": 0.8172180197021022,
+      "learning_rate": 3.727016711386628e-05,
+      "loss": 0.0784,
+      "step": 1082
+    },
+    {
+      "epoch": 3.867857142857143,
+      "grad_norm": 0.5935162663380993,
+      "learning_rate": 3.726433059677843e-05,
+      "loss": 0.0801,
+      "step": 1083
+    },
+    {
+      "epoch": 3.8714285714285714,
+      "grad_norm": 0.9975209838813925,
+      "learning_rate": 3.7258488304983547e-05,
+      "loss": 0.0814,
+      "step": 1084
+    },
+    {
+      "epoch": 3.875,
+      "grad_norm": 0.3870010092459377,
+      "learning_rate": 3.7252640240435844e-05,
+      "loss": 0.0416,
+      "step": 1085
+    },
+    {
+      "epoch": 3.8785714285714286,
+      "grad_norm": 0.768590467365692,
+      "learning_rate": 3.7246786405091405e-05,
+      "loss": 0.0974,
+      "step": 1086
+    },
+    {
+      "epoch": 3.882142857142857,
+      "grad_norm": 0.9232652199980819,
+      "learning_rate": 3.724092680090827e-05,
+      "loss": 0.1099,
+      "step": 1087
+    },
+    {
+      "epoch": 3.8857142857142857,
+      "grad_norm": 1.0744338405725986,
+      "learning_rate": 3.723506142984642e-05,
+      "loss": 0.0913,
+      "step": 1088
+    },
+    {
+      "epoch": 3.8892857142857142,
+      "grad_norm": 1.2548690511268066,
+      "learning_rate": 3.7229190293867725e-05,
+      "loss": 0.1043,
+      "step": 1089
+    },
+    {
+      "epoch": 3.892857142857143,
+      "grad_norm": 0.9118463433529003,
+      "learning_rate": 3.722331339493602e-05,
+      "loss": 0.0815,
+      "step": 1090
+    },
+    {
+      "epoch": 3.8964285714285714,
+      "grad_norm": 0.9440857222681995,
+      "learning_rate": 3.721743073501706e-05,
+      "loss": 0.0997,
+      "step": 1091
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.6581261311934389,
+      "learning_rate": 3.7211542316078506e-05,
+      "loss": 0.1022,
+      "step": 1092
+    },
+    {
+      "epoch": 3.9035714285714285,
+      "grad_norm": 0.9326735530666729,
+      "learning_rate": 3.7205648140089986e-05,
+      "loss": 0.1051,
+      "step": 1093
+    },
+    {
+      "epoch": 3.907142857142857,
+      "grad_norm": 0.558896050652002,
+      "learning_rate": 3.7199748209023005e-05,
+      "loss": 0.0583,
+      "step": 1094
+    },
+    {
+      "epoch": 3.9107142857142856,
+      "grad_norm": 0.9607272571275116,
+      "learning_rate": 3.719384252485104e-05,
+      "loss": 0.0895,
+      "step": 1095
+    },
+    {
+      "epoch": 3.914285714285714,
+      "grad_norm": 1.302688506173105,
+      "learning_rate": 3.718793108954946e-05,
+      "loss": 0.045,
+      "step": 1096
+    },
+    {
+      "epoch": 3.9178571428571427,
+      "grad_norm": 0.7725086995762989,
+      "learning_rate": 3.7182013905095565e-05,
+      "loss": 0.0874,
+      "step": 1097
+    },
+    {
+      "epoch": 3.9214285714285713,
+      "grad_norm": 0.536682495156433,
+      "learning_rate": 3.7176090973468584e-05,
+      "loss": 0.0548,
+      "step": 1098
+    },
+    {
+      "epoch": 3.925,
+      "grad_norm": 0.7342314832168169,
+      "learning_rate": 3.717016229664966e-05,
+      "loss": 0.06,
+      "step": 1099
+    },
+    {
+      "epoch": 3.928571428571429,
+      "grad_norm": 1.1336781486177103,
+      "learning_rate": 3.716422787662188e-05,
+      "loss": 0.0827,
+      "step": 1100
+    },
+    {
+      "epoch": 3.932142857142857,
+      "grad_norm": 0.8178520273007425,
+      "learning_rate": 3.715828771537021e-05,
+      "loss": 0.1027,
+      "step": 1101
+    },
+    {
+      "epoch": 3.935714285714286,
+      "grad_norm": 1.0219129039794173,
+      "learning_rate": 3.715234181488158e-05,
+      "loss": 0.0573,
+      "step": 1102
+    },
+    {
+      "epoch": 3.939285714285714,
+      "grad_norm": 0.6138021614836072,
+      "learning_rate": 3.714639017714481e-05,
+      "loss": 0.0593,
+      "step": 1103
+    },
+    {
+      "epoch": 3.942857142857143,
+      "grad_norm": 0.48961825390125524,
+      "learning_rate": 3.714043280415065e-05,
+      "loss": 0.048,
+      "step": 1104
+    },
+    {
+      "epoch": 3.946428571428571,
+      "grad_norm": 1.073797993268229,
+      "learning_rate": 3.7134469697891774e-05,
+      "loss": 0.1224,
+      "step": 1105
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.377873615314867,
+      "learning_rate": 3.7128500860362775e-05,
+      "loss": 0.0467,
+      "step": 1106
+    },
+    {
+      "epoch": 3.9535714285714287,
+      "grad_norm": 0.9440215495171782,
+      "learning_rate": 3.712252629356014e-05,
+      "loss": 0.0724,
+      "step": 1107
+    },
+    {
+      "epoch": 3.9571428571428573,
+      "grad_norm": 0.7537245433398065,
+      "learning_rate": 3.7116545999482295e-05,
+      "loss": 0.0643,
+      "step": 1108
+    },
+    {
+      "epoch": 3.960714285714286,
+      "grad_norm": 0.9286678023059525,
+      "learning_rate": 3.711055998012958e-05,
+      "loss": 0.0534,
+      "step": 1109
+    },
+    {
+      "epoch": 3.9642857142857144,
+      "grad_norm": 1.9584008010209852,
+      "learning_rate": 3.710456823750423e-05,
+      "loss": 0.0951,
+      "step": 1110
+    },
+    {
+      "epoch": 3.967857142857143,
+      "grad_norm": 0.7639072901523708,
+      "learning_rate": 3.7098570773610426e-05,
+      "loss": 0.0526,
+      "step": 1111
+    },
+    {
+      "epoch": 3.9714285714285715,
+      "grad_norm": 0.9727850353819623,
+      "learning_rate": 3.709256759045424e-05,
+      "loss": 0.1319,
+      "step": 1112
+    },
+    {
+      "epoch": 3.975,
+      "grad_norm": 1.0742750772967211,
+      "learning_rate": 3.708655869004367e-05,
+      "loss": 0.0516,
+      "step": 1113
+    },
+    {
+      "epoch": 3.9785714285714286,
+      "grad_norm": 1.2089963261536665,
+      "learning_rate": 3.708054407438861e-05,
+      "loss": 0.0856,
+      "step": 1114
+    },
+    {
+      "epoch": 3.982142857142857,
+      "grad_norm": 0.8803186844837751,
+      "learning_rate": 3.707452374550087e-05,
+      "loss": 0.1396,
+      "step": 1115
+    },
+    {
+      "epoch": 3.9857142857142858,
+      "grad_norm": 0.9287285261564457,
+      "learning_rate": 3.7068497705394196e-05,
+      "loss": 0.1104,
+      "step": 1116
+    },
+    {
+      "epoch": 3.9892857142857143,
+      "grad_norm": 1.09484530268284,
+      "learning_rate": 3.70624659560842e-05,
+      "loss": 0.0838,
+      "step": 1117
+    },
+    {
+      "epoch": 3.992857142857143,
+      "grad_norm": 1.1342581721375824,
+      "learning_rate": 3.705642849958845e-05,
+      "loss": 0.0848,
+      "step": 1118
+    },
+    {
+      "epoch": 3.9964285714285714,
+      "grad_norm": 1.8980020230997392,
+      "learning_rate": 3.7050385337926396e-05,
+      "loss": 0.0913,
+      "step": 1119
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.6078892525317021,
+      "learning_rate": 3.7044336473119386e-05,
+      "loss": 0.1328,
+      "step": 1120
+    },
+    {
+      "epoch": 4.003571428571429,
+      "grad_norm": 0.7413061159624806,
+      "learning_rate": 3.703828190719071e-05,
+      "loss": 0.0563,
+      "step": 1121
+    },
+    {
+      "epoch": 4.007142857142857,
+      "grad_norm": 0.46420285498145214,
+      "learning_rate": 3.7032221642165545e-05,
+      "loss": 0.0274,
+      "step": 1122
+    },
+    {
+      "epoch": 4.010714285714286,
+      "grad_norm": 0.5527200964492022,
+      "learning_rate": 3.7026155680070954e-05,
+      "loss": 0.0774,
+      "step": 1123
+    },
+    {
+      "epoch": 4.014285714285714,
+      "grad_norm": 0.785937684547749,
+      "learning_rate": 3.7020084022935956e-05,
+      "loss": 0.11,
+      "step": 1124
+    },
+    {
+      "epoch": 4.017857142857143,
+      "grad_norm": 1.243891027010887,
+      "learning_rate": 3.701400667279143e-05,
+      "loss": 0.0929,
+      "step": 1125
+    },
+    {
+      "epoch": 4.021428571428571,
+      "grad_norm": 0.6216137573653318,
+      "learning_rate": 3.700792363167018e-05,
+      "loss": 0.042,
+      "step": 1126
+    },
+    {
+      "epoch": 4.025,
+      "grad_norm": 0.55461669366272,
+      "learning_rate": 3.70018349016069e-05,
+      "loss": 0.0499,
+      "step": 1127
+    },
+    {
+      "epoch": 4.0285714285714285,
+      "grad_norm": 1.0853242743783047,
+      "learning_rate": 3.699574048463821e-05,
+      "loss": 0.0651,
+      "step": 1128
+    },
+    {
+      "epoch": 4.0321428571428575,
+      "grad_norm": 0.5071409478889147,
+      "learning_rate": 3.69896403828026e-05,
+      "loss": 0.0978,
+      "step": 1129
+    },
+    {
+      "epoch": 4.035714285714286,
+      "grad_norm": 0.635933945337284,
+      "learning_rate": 3.69835345981405e-05,
+      "loss": 0.0372,
+      "step": 1130
+    },
+    {
+      "epoch": 4.039285714285715,
+      "grad_norm": 0.7806353946209003,
+      "learning_rate": 3.6977423132694195e-05,
+      "loss": 0.1111,
+      "step": 1131
+    },
+    {
+      "epoch": 4.042857142857143,
+      "grad_norm": 0.8419598471573492,
+      "learning_rate": 3.697130598850792e-05,
+      "loss": 0.0929,
+      "step": 1132
+    },
+    {
+      "epoch": 4.046428571428572,
+      "grad_norm": 0.6462893250756092,
+      "learning_rate": 3.6965183167627776e-05,
+      "loss": 0.1193,
+      "step": 1133
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.43632509311943457,
+      "learning_rate": 3.695905467210176e-05,
+      "loss": 0.0483,
+      "step": 1134
+    },
+    {
+      "epoch": 4.053571428571429,
+      "grad_norm": 0.6316464002029112,
+      "learning_rate": 3.69529205039798e-05,
+      "loss": 0.0974,
+      "step": 1135
+    },
+    {
+      "epoch": 4.057142857142857,
+      "grad_norm": 0.7323202620988976,
+      "learning_rate": 3.694678066531368e-05,
+      "loss": 0.0871,
+      "step": 1136
+    },
+    {
+      "epoch": 4.060714285714286,
+      "grad_norm": 0.5112124567288203,
+      "learning_rate": 3.694063515815711e-05,
+      "loss": 0.0311,
+      "step": 1137
+    },
+    {
+      "epoch": 4.064285714285714,
+      "grad_norm": 0.6510625807760643,
+      "learning_rate": 3.6934483984565684e-05,
+      "loss": 0.1111,
+      "step": 1138
+    },
+    {
+      "epoch": 4.067857142857143,
+      "grad_norm": 0.9359172093850497,
+      "learning_rate": 3.69283271465969e-05,
+      "loss": 0.0879,
+      "step": 1139
+    },
+    {
+      "epoch": 4.071428571428571,
+      "grad_norm": 0.577599783422523,
+      "learning_rate": 3.692216464631014e-05,
+      "loss": 0.0844,
+      "step": 1140
+    },
+    {
+      "epoch": 4.075,
+      "grad_norm": 0.7983920083589532,
+      "learning_rate": 3.691599648576669e-05,
+      "loss": 0.082,
+      "step": 1141
+    },
+    {
+      "epoch": 4.078571428571428,
+      "grad_norm": 0.555220186271298,
+      "learning_rate": 3.690982266702972e-05,
+      "loss": 0.0797,
+      "step": 1142
+    },
+    {
+      "epoch": 4.082142857142857,
+      "grad_norm": 0.6447237545273758,
+      "learning_rate": 3.690364319216429e-05,
+      "loss": 0.0767,
+      "step": 1143
+    },
+    {
+      "epoch": 4.085714285714285,
+      "grad_norm": 0.46497991283399304,
+      "learning_rate": 3.6897458063237375e-05,
+      "loss": 0.0545,
+      "step": 1144
+    },
+    {
+      "epoch": 4.089285714285714,
+      "grad_norm": 0.5658566113904032,
+      "learning_rate": 3.689126728231781e-05,
+      "loss": 0.0714,
+      "step": 1145
+    },
+    {
+      "epoch": 4.0928571428571425,
+      "grad_norm": 0.8182222032939,
+      "learning_rate": 3.6885070851476335e-05,
+      "loss": 0.0942,
+      "step": 1146
+    },
+    {
+      "epoch": 4.0964285714285715,
+      "grad_norm": 0.6736463352497092,
+      "learning_rate": 3.687886877278559e-05,
+      "loss": 0.0725,
+      "step": 1147
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.46870405091766454,
+      "learning_rate": 3.6872661048320096e-05,
+      "loss": 0.0657,
+      "step": 1148
+    },
+    {
+      "epoch": 4.103571428571429,
+      "grad_norm": 0.55934259519813,
+      "learning_rate": 3.686644768015625e-05,
+      "loss": 0.0952,
+      "step": 1149
+    },
+    {
+      "epoch": 4.107142857142857,
+      "grad_norm": 0.672563008609227,
+      "learning_rate": 3.6860228670372354e-05,
+      "loss": 0.0826,
+      "step": 1150
+    },
+    {
+      "epoch": 4.110714285714286,
+      "grad_norm": 0.46261808256622605,
+      "learning_rate": 3.6854004021048587e-05,
+      "loss": 0.0484,
+      "step": 1151
+    },
+    {
+      "epoch": 4.114285714285714,
+      "grad_norm": 0.6588242882726189,
+      "learning_rate": 3.684777373426702e-05,
+      "loss": 0.084,
+      "step": 1152
+    },
+    {
+      "epoch": 4.117857142857143,
+      "grad_norm": 0.3340170863921649,
+      "learning_rate": 3.6841537812111606e-05,
+      "loss": 0.0737,
+      "step": 1153
+    },
+    {
+      "epoch": 4.121428571428571,
+      "grad_norm": 0.38143061570897374,
+      "learning_rate": 3.6835296256668186e-05,
+      "loss": 0.0827,
+      "step": 1154
+    },
+    {
+      "epoch": 4.125,
+      "grad_norm": 0.719657447636293,
+      "learning_rate": 3.682904907002449e-05,
+      "loss": 0.1263,
+      "step": 1155
+    },
+    {
+      "epoch": 4.128571428571428,
+      "grad_norm": 0.49577893858631755,
+      "learning_rate": 3.6822796254270114e-05,
+      "loss": 0.0737,
+      "step": 1156
+    },
+    {
+      "epoch": 4.132142857142857,
+      "grad_norm": 0.7555641491758125,
+      "learning_rate": 3.681653781149655e-05,
+      "loss": 0.0714,
+      "step": 1157
+    },
+    {
+      "epoch": 4.135714285714286,
+      "grad_norm": 0.8444970172531743,
+      "learning_rate": 3.681027374379718e-05,
+      "loss": 0.0919,
+      "step": 1158
+    },
+    {
+      "epoch": 4.139285714285714,
+      "grad_norm": 0.9324104521206126,
+      "learning_rate": 3.6804004053267245e-05,
+      "loss": 0.1031,
+      "step": 1159
+    },
+    {
+      "epoch": 4.142857142857143,
+      "grad_norm": 0.5266424971987246,
+      "learning_rate": 3.679772874200389e-05,
+      "loss": 0.0479,
+      "step": 1160
+    },
+    {
+      "epoch": 4.146428571428571,
+      "grad_norm": 0.5730646543179072,
+      "learning_rate": 3.679144781210613e-05,
+      "loss": 0.0916,
+      "step": 1161
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 1.1163274449939313,
+      "learning_rate": 3.6785161265674847e-05,
+      "loss": 0.0553,
+      "step": 1162
+    },
+    {
+      "epoch": 4.1535714285714285,
+      "grad_norm": 0.5631566708005266,
+      "learning_rate": 3.677886910481283e-05,
+      "loss": 0.0552,
+      "step": 1163
+    },
+    {
+      "epoch": 4.1571428571428575,
+      "grad_norm": 0.6433978914110037,
+      "learning_rate": 3.677257133162472e-05,
+      "loss": 0.0251,
+      "step": 1164
+    },
+    {
+      "epoch": 4.160714285714286,
+      "grad_norm": 0.9043430103607214,
+      "learning_rate": 3.676626794821705e-05,
+      "loss": 0.0624,
+      "step": 1165
+    },
+    {
+      "epoch": 4.164285714285715,
+      "grad_norm": 1.832418156112379,
+      "learning_rate": 3.675995895669821e-05,
+      "loss": 0.0915,
+      "step": 1166
+    },
+    {
+      "epoch": 4.167857142857143,
+      "grad_norm": 1.01698719400646,
+      "learning_rate": 3.67536443591785e-05,
+      "loss": 0.0639,
+      "step": 1167
+    },
+    {
+      "epoch": 4.171428571428572,
+      "grad_norm": 0.6120369680168735,
+      "learning_rate": 3.6747324157770056e-05,
+      "loss": 0.0658,
+      "step": 1168
+    },
+    {
+      "epoch": 4.175,
+      "grad_norm": 0.8202122324024148,
+      "learning_rate": 3.6740998354586925e-05,
+      "loss": 0.0659,
+      "step": 1169
+    },
+    {
+      "epoch": 4.178571428571429,
+      "grad_norm": 0.5602191672099119,
+      "learning_rate": 3.673466695174501e-05,
+      "loss": 0.088,
+      "step": 1170
+    },
+    {
+      "epoch": 4.182142857142857,
+      "grad_norm": 0.5584343377902388,
+      "learning_rate": 3.6728329951362075e-05,
+      "loss": 0.0148,
+      "step": 1171
+    },
+    {
+      "epoch": 4.185714285714286,
+      "grad_norm": 1.3596546401913663,
+      "learning_rate": 3.672198735555778e-05,
+      "loss": 0.0977,
+      "step": 1172
+    },
+    {
+      "epoch": 4.189285714285714,
+      "grad_norm": 0.9852511439959148,
+      "learning_rate": 3.671563916645364e-05,
+      "loss": 0.0856,
+      "step": 1173
+    },
+    {
+      "epoch": 4.192857142857143,
+      "grad_norm": 1.6463899377469762,
+      "learning_rate": 3.670928538617305e-05,
+      "loss": 0.0909,
+      "step": 1174
+    },
+    {
+      "epoch": 4.196428571428571,
+      "grad_norm": 1.0460477819692116,
+      "learning_rate": 3.6702926016841266e-05,
+      "loss": 0.1027,
+      "step": 1175
+    },
+    {
+      "epoch": 4.2,
+      "grad_norm": 1.0691046526171428,
+      "learning_rate": 3.6696561060585424e-05,
+      "loss": 0.1158,
+      "step": 1176
+    },
+    {
+      "epoch": 4.203571428571428,
+      "grad_norm": 0.8211063901650147,
+      "learning_rate": 3.6690190519534525e-05,
+      "loss": 0.1667,
+      "step": 1177
+    },
+    {
+      "epoch": 4.207142857142857,
+      "grad_norm": 0.9972729216406091,
+      "learning_rate": 3.668381439581944e-05,
+      "loss": 0.0551,
+      "step": 1178
+    },
+    {
+      "epoch": 4.210714285714285,
+      "grad_norm": 0.8755757595109852,
+      "learning_rate": 3.667743269157289e-05,
+      "loss": 0.0728,
+      "step": 1179
+    },
+    {
+      "epoch": 4.214285714285714,
+      "grad_norm": 0.5180899168557346,
+      "learning_rate": 3.667104540892949e-05,
+      "loss": 0.067,
+      "step": 1180
+    },
+    {
+      "epoch": 4.2178571428571425,
+      "grad_norm": 0.4717913851023377,
+      "learning_rate": 3.666465255002569e-05,
+      "loss": 0.0747,
+      "step": 1181
+    },
+    {
+      "epoch": 4.2214285714285715,
+      "grad_norm": 0.4009214501291442,
+      "learning_rate": 3.6658254116999857e-05,
+      "loss": 0.0543,
+      "step": 1182
+    },
+    {
+      "epoch": 4.225,
+      "grad_norm": 0.5741022845498963,
+      "learning_rate": 3.665185011199215e-05,
+      "loss": 0.0626,
+      "step": 1183
+    },
+    {
+      "epoch": 4.228571428571429,
+      "grad_norm": 0.6348530897416798,
+      "learning_rate": 3.6645440537144665e-05,
+      "loss": 0.0712,
+      "step": 1184
+    },
+    {
+      "epoch": 4.232142857142857,
+      "grad_norm": 0.7098716145872329,
+      "learning_rate": 3.66390253946013e-05,
+      "loss": 0.1417,
+      "step": 1185
+    },
+    {
+      "epoch": 4.235714285714286,
+      "grad_norm": 0.39727935134240705,
+      "learning_rate": 3.663260468650785e-05,
+      "loss": 0.0271,
+      "step": 1186
+    },
+    {
+      "epoch": 4.239285714285714,
+      "grad_norm": 0.6299019589767009,
+      "learning_rate": 3.662617841501196e-05,
+      "loss": 0.0954,
+      "step": 1187
+    },
+    {
+      "epoch": 4.242857142857143,
+      "grad_norm": 0.5710949819060528,
+      "learning_rate": 3.661974658226315e-05,
+      "loss": 0.0584,
+      "step": 1188
+    },
+    {
+      "epoch": 4.246428571428572,
+      "grad_norm": 0.4466062023299109,
+      "learning_rate": 3.661330919041278e-05,
+      "loss": 0.0371,
+      "step": 1189
+    },
+    {
+      "epoch": 4.25,
+      "grad_norm": 0.8946462537062475,
+      "learning_rate": 3.6606866241614085e-05,
+      "loss": 0.1066,
+      "step": 1190
+    },
+    {
+      "epoch": 4.253571428571428,
+      "grad_norm": 0.6802974252281927,
+      "learning_rate": 3.660041773802214e-05,
+      "loss": 0.0741,
+      "step": 1191
+    },
+    {
+      "epoch": 4.257142857142857,
+      "grad_norm": 0.7513045843182,
+      "learning_rate": 3.65939636817939e-05,
+      "loss": 0.0909,
+      "step": 1192
+    },
+    {
+      "epoch": 4.260714285714286,
+      "grad_norm": 0.6908755330362365,
+      "learning_rate": 3.6587504075088176e-05,
+      "loss": 0.0815,
+      "step": 1193
+    },
+    {
+      "epoch": 4.264285714285714,
+      "grad_norm": 0.7361776111395507,
+      "learning_rate": 3.658103892006561e-05,
+      "loss": 0.0761,
+      "step": 1194
+    },
+    {
+      "epoch": 4.267857142857143,
+      "grad_norm": 0.39074575390604416,
+      "learning_rate": 3.657456821888873e-05,
+      "loss": 0.0412,
+      "step": 1195
+    },
+    {
+      "epoch": 4.271428571428571,
+      "grad_norm": 0.634120341055703,
+      "learning_rate": 3.65680919737219e-05,
+      "loss": 0.0916,
+      "step": 1196
+    },
+    {
+      "epoch": 4.275,
+      "grad_norm": 0.4887302627161152,
+      "learning_rate": 3.656161018673135e-05,
+      "loss": 0.0978,
+      "step": 1197
+    },
+    {
+      "epoch": 4.2785714285714285,
+      "grad_norm": 0.5877574322688968,
+      "learning_rate": 3.655512286008514e-05,
+      "loss": 0.0977,
+      "step": 1198
+    },
+    {
+      "epoch": 4.2821428571428575,
+      "grad_norm": 0.7811789688214734,
+      "learning_rate": 3.654862999595322e-05,
+      "loss": 0.0521,
+      "step": 1199
+    },
+    {
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.6707279881220468,
+      "learning_rate": 3.6542131596507376e-05,
+      "loss": 0.096,
+      "step": 1200
+    },
+    {
+      "epoch": 4.289285714285715,
+      "grad_norm": 0.8459291765830396,
+      "learning_rate": 3.653562766392123e-05,
+      "loss": 0.0963,
+      "step": 1201
+    },
+    {
+      "epoch": 4.292857142857143,
+      "grad_norm": 0.9578215025574297,
+      "learning_rate": 3.6529118200370284e-05,
+      "loss": 0.0849,
+      "step": 1202
+    },
+    {
+      "epoch": 4.296428571428572,
+      "grad_norm": 0.6807650045030499,
+      "learning_rate": 3.652260320803185e-05,
+      "loss": 0.0582,
+      "step": 1203
+    },
+    {
+      "epoch": 4.3,
+      "grad_norm": 0.5652465483318732,
+      "learning_rate": 3.651608268908513e-05,
+      "loss": 0.0671,
+      "step": 1204
+    },
+    {
+      "epoch": 4.303571428571429,
+      "grad_norm": 0.406611748133779,
+      "learning_rate": 3.650955664571116e-05,
+      "loss": 0.0633,
+      "step": 1205
+    },
+    {
+      "epoch": 4.307142857142857,
+      "grad_norm": 0.8385360302941074,
+      "learning_rate": 3.6503025080092814e-05,
+      "loss": 0.0779,
+      "step": 1206
+    },
+    {
+      "epoch": 4.310714285714286,
+      "grad_norm": 0.7615971083172868,
+      "learning_rate": 3.649648799441482e-05,
+      "loss": 0.1033,
+      "step": 1207
+    },
+    {
+      "epoch": 4.314285714285714,
+      "grad_norm": 0.4383733027176645,
+      "learning_rate": 3.648994539086375e-05,
+      "loss": 0.0649,
+      "step": 1208
+    },
+    {
+      "epoch": 4.317857142857143,
+      "grad_norm": 0.5129791439706098,
+      "learning_rate": 3.648339727162804e-05,
+      "loss": 0.0698,
+      "step": 1209
+    },
+    {
+      "epoch": 4.321428571428571,
+      "grad_norm": 0.978083650187757,
+      "learning_rate": 3.647684363889794e-05,
+      "loss": 0.0938,
+      "step": 1210
+    },
+    {
+      "epoch": 4.325,
+      "grad_norm": 1.097164105874134,
+      "learning_rate": 3.647028449486557e-05,
+      "loss": 0.085,
+      "step": 1211
+    },
+    {
+      "epoch": 4.328571428571428,
+      "grad_norm": 0.9187676536909503,
+      "learning_rate": 3.646371984172487e-05,
+      "loss": 0.1168,
+      "step": 1212
+    },
+    {
+      "epoch": 4.332142857142857,
+      "grad_norm": 1.2245245681529344,
+      "learning_rate": 3.645714968167165e-05,
+      "loss": 0.0826,
+      "step": 1213
+    },
+    {
+      "epoch": 4.335714285714285,
+      "grad_norm": 1.189750978991794,
+      "learning_rate": 3.645057401690355e-05,
+      "loss": 0.0774,
+      "step": 1214
+    },
+    {
+      "epoch": 4.339285714285714,
+      "grad_norm": 0.9420585453989379,
+      "learning_rate": 3.644399284962003e-05,
+      "loss": 0.0538,
+      "step": 1215
+    },
+    {
+      "epoch": 4.3428571428571425,
+      "grad_norm": 0.7815883452926162,
+      "learning_rate": 3.643740618202243e-05,
+      "loss": 0.1007,
+      "step": 1216
+    },
+    {
+      "epoch": 4.3464285714285715,
+      "grad_norm": 0.8915631223225716,
+      "learning_rate": 3.6430814016313904e-05,
+      "loss": 0.0787,
+      "step": 1217
+    },
+    {
+      "epoch": 4.35,
+      "grad_norm": 0.8138930120849838,
+      "learning_rate": 3.6424216354699444e-05,
+      "loss": 0.1083,
+      "step": 1218
+    },
+    {
+      "epoch": 4.353571428571429,
+      "grad_norm": 0.6912485062554811,
+      "learning_rate": 3.641761319938589e-05,
+      "loss": 0.0271,
+      "step": 1219
+    },
+    {
+      "epoch": 4.357142857142857,
+      "grad_norm": 0.896917791124702,
+      "learning_rate": 3.641100455258192e-05,
+      "loss": 0.1254,
+      "step": 1220
+    },
+    {
+      "epoch": 4.360714285714286,
+      "grad_norm": 0.7652087514333035,
+      "learning_rate": 3.640439041649804e-05,
+      "loss": 0.0941,
+      "step": 1221
+    },
+    {
+      "epoch": 4.364285714285714,
+      "grad_norm": 0.6464239884202996,
+      "learning_rate": 3.639777079334659e-05,
+      "loss": 0.0556,
+      "step": 1222
+    },
+    {
+      "epoch": 4.367857142857143,
+      "grad_norm": 0.9580681496167621,
+      "learning_rate": 3.6391145685341785e-05,
+      "loss": 0.0549,
+      "step": 1223
+    },
+    {
+      "epoch": 4.371428571428572,
+      "grad_norm": 0.9304331576717936,
+      "learning_rate": 3.638451509469961e-05,
+      "loss": 0.1234,
+      "step": 1224
+    },
+    {
+      "epoch": 4.375,
+      "grad_norm": 0.7685010526382748,
+      "learning_rate": 3.637787902363793e-05,
+      "loss": 0.1046,
+      "step": 1225
+    },
+    {
+      "epoch": 4.378571428571428,
+      "grad_norm": 0.8534091144357915,
+      "learning_rate": 3.637123747437643e-05,
+      "loss": 0.0889,
+      "step": 1226
+    },
+    {
+      "epoch": 4.382142857142857,
+      "grad_norm": 0.7579328273090233,
+      "learning_rate": 3.636459044913661e-05,
+      "loss": 0.085,
+      "step": 1227
+    },
+    {
+      "epoch": 4.385714285714286,
+      "grad_norm": 0.4288051806856814,
+      "learning_rate": 3.635793795014184e-05,
+      "loss": 0.0614,
+      "step": 1228
+    },
+    {
+      "epoch": 4.389285714285714,
+      "grad_norm": 0.5335975193128619,
+      "learning_rate": 3.6351279979617295e-05,
+      "loss": 0.0902,
+      "step": 1229
+    },
+    {
+      "epoch": 4.392857142857143,
+      "grad_norm": 0.9843632908107669,
+      "learning_rate": 3.634461653978998e-05,
+      "loss": 0.1007,
+      "step": 1230
+    },
+    {
+      "epoch": 4.396428571428571,
+      "grad_norm": 0.8402440335799808,
+      "learning_rate": 3.633794763288873e-05,
+      "loss": 0.0786,
+      "step": 1231
+    },
+    {
+      "epoch": 4.4,
+      "grad_norm": 0.7390771797756899,
+      "learning_rate": 3.633127326114422e-05,
+      "loss": 0.1078,
+      "step": 1232
+    },
+    {
+      "epoch": 4.4035714285714285,
+      "grad_norm": 0.8033213620352317,
+      "learning_rate": 3.6324593426788943e-05,
+      "loss": 0.062,
+      "step": 1233
+    },
+    {
+      "epoch": 4.4071428571428575,
+      "grad_norm": 0.7930831939638257,
+      "learning_rate": 3.631790813205722e-05,
+      "loss": 0.0983,
+      "step": 1234
+    },
+    {
+      "epoch": 4.410714285714286,
+      "grad_norm": 0.6222156751513389,
+      "learning_rate": 3.631121737918521e-05,
+      "loss": 0.0547,
+      "step": 1235
+    },
+    {
+      "epoch": 4.414285714285715,
+      "grad_norm": 1.2437309872871445,
+      "learning_rate": 3.630452117041087e-05,
+      "loss": 0.0782,
+      "step": 1236
+    },
+    {
+      "epoch": 4.417857142857143,
+      "grad_norm": 0.4771007484342013,
+      "learning_rate": 3.6297819507974016e-05,
+      "loss": 0.0811,
+      "step": 1237
+    },
+    {
+      "epoch": 4.421428571428572,
+      "grad_norm": 0.8284548032657117,
+      "learning_rate": 3.629111239411626e-05,
+      "loss": 0.0878,
+      "step": 1238
+    },
+    {
+      "epoch": 4.425,
+      "grad_norm": 0.8025578251359954,
+      "learning_rate": 3.628439983108106e-05,
+      "loss": 0.1078,
+      "step": 1239
+    },
+    {
+      "epoch": 4.428571428571429,
+      "grad_norm": 0.4723413035154494,
+      "learning_rate": 3.627768182111367e-05,
+      "loss": 0.0376,
+      "step": 1240
+    },
+    {
+      "epoch": 4.432142857142857,
+      "grad_norm": 0.5039925805825226,
+      "learning_rate": 3.6270958366461206e-05,
+      "loss": 0.0583,
+      "step": 1241
+    },
+    {
+      "epoch": 4.435714285714286,
+      "grad_norm": 0.666061794298106,
+      "learning_rate": 3.626422946937256e-05,
+      "loss": 0.0634,
+      "step": 1242
+    },
+    {
+      "epoch": 4.439285714285714,
+      "grad_norm": 0.445683369936083,
+      "learning_rate": 3.625749513209848e-05,
+      "loss": 0.0434,
+      "step": 1243
+    },
+    {
+      "epoch": 4.442857142857143,
+      "grad_norm": 0.44543768187970495,
+      "learning_rate": 3.625075535689151e-05,
+      "loss": 0.0435,
+      "step": 1244
+    },
+    {
+      "epoch": 4.446428571428571,
+      "grad_norm": 0.5324756171377111,
+      "learning_rate": 3.6244010146006025e-05,
+      "loss": 0.0652,
+      "step": 1245
+    },
+    {
+      "epoch": 4.45,
+      "grad_norm": 0.9931396801354075,
+      "learning_rate": 3.623725950169821e-05,
+      "loss": 0.0622,
+      "step": 1246
+    },
+    {
+      "epoch": 4.453571428571428,
+      "grad_norm": 0.6366205348595203,
+      "learning_rate": 3.623050342622609e-05,
+      "loss": 0.0396,
+      "step": 1247
+    },
+    {
+      "epoch": 4.457142857142857,
+      "grad_norm": 0.5115830219491899,
+      "learning_rate": 3.6223741921849466e-05,
+      "loss": 0.0734,
+      "step": 1248
+    },
+    {
+      "epoch": 4.460714285714285,
+      "grad_norm": 0.8275445314282458,
+      "learning_rate": 3.621697499083e-05,
+      "loss": 0.0831,
+      "step": 1249
+    },
+    {
+      "epoch": 4.464285714285714,
+      "grad_norm": 1.526349822505108,
+      "learning_rate": 3.6210202635431134e-05,
+      "loss": 0.1132,
+      "step": 1250
+    },
+    {
+      "epoch": 4.4678571428571425,
+      "grad_norm": 0.7593975631640416,
+      "learning_rate": 3.620342485791814e-05,
+      "loss": 0.0963,
+      "step": 1251
+    },
+    {
+      "epoch": 4.4714285714285715,
+      "grad_norm": 0.5484668545529099,
+      "learning_rate": 3.61966416605581e-05,
+      "loss": 0.0382,
+      "step": 1252
+    },
+    {
+      "epoch": 4.475,
+      "grad_norm": 0.6451931834915436,
+      "learning_rate": 3.6189853045619916e-05,
+      "loss": 0.083,
+      "step": 1253
+    },
+    {
+      "epoch": 4.478571428571429,
+      "grad_norm": 0.9593598063274972,
+      "learning_rate": 3.6183059015374296e-05,
+      "loss": 0.1022,
+      "step": 1254
+    },
+    {
+      "epoch": 4.482142857142857,
+      "grad_norm": 0.8230351101040276,
+      "learning_rate": 3.6176259572093766e-05,
+      "loss": 0.0806,
+      "step": 1255
+    },
+    {
+      "epoch": 4.485714285714286,
+      "grad_norm": 0.5694166135973292,
+      "learning_rate": 3.616945471805263e-05,
+      "loss": 0.0474,
+      "step": 1256
+    },
+    {
+      "epoch": 4.489285714285714,
+      "grad_norm": 0.6095675041063908,
+      "learning_rate": 3.616264445552706e-05,
+      "loss": 0.0576,
+      "step": 1257
+    },
+    {
+      "epoch": 4.492857142857143,
+      "grad_norm": 0.8201383260370262,
+      "learning_rate": 3.615582878679499e-05,
+      "loss": 0.0555,
+      "step": 1258
+    },
+    {
+      "epoch": 4.496428571428572,
+      "grad_norm": 0.4420441332279392,
+      "learning_rate": 3.614900771413617e-05,
+      "loss": 0.0476,
+      "step": 1259
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 0.4754094476492968,
+      "learning_rate": 3.614218123983219e-05,
+      "loss": 0.055,
+      "step": 1260
+    },
+    {
+      "epoch": 4.503571428571428,
+      "grad_norm": 0.885585337297035,
+      "learning_rate": 3.61353493661664e-05,
+      "loss": 0.0905,
+      "step": 1261
+    },
+    {
+      "epoch": 4.507142857142857,
+      "grad_norm": 0.3770433213994182,
+      "learning_rate": 3.612851209542398e-05,
+      "loss": 0.0455,
+      "step": 1262
+    },
+    {
+      "epoch": 4.510714285714286,
+      "grad_norm": 1.173590762297546,
+      "learning_rate": 3.612166942989192e-05,
+      "loss": 0.0868,
+      "step": 1263
+    },
+    {
+      "epoch": 4.514285714285714,
+      "grad_norm": 0.34009433509686476,
+      "learning_rate": 3.611482137185901e-05,
+      "loss": 0.0465,
+      "step": 1264
+    },
+    {
+      "epoch": 4.517857142857143,
+      "grad_norm": 0.7577291639382515,
+      "learning_rate": 3.6107967923615836e-05,
+      "loss": 0.076,
+      "step": 1265
+    },
+    {
+      "epoch": 4.521428571428571,
+      "grad_norm": 0.4810175562080691,
+      "learning_rate": 3.61011090874548e-05,
+      "loss": 0.0407,
+      "step": 1266
+    },
+    {
+      "epoch": 4.525,
+      "grad_norm": 0.5505887438450808,
+      "learning_rate": 3.609424486567008e-05,
+      "loss": 0.0806,
+      "step": 1267
+    },
+    {
+      "epoch": 4.5285714285714285,
+      "grad_norm": 0.965229449431817,
+      "learning_rate": 3.6087375260557697e-05,
+      "loss": 0.1042,
+      "step": 1268
+    },
+    {
+      "epoch": 4.5321428571428575,
+      "grad_norm": 0.5913129507793394,
+      "learning_rate": 3.6080500274415434e-05,
+      "loss": 0.0416,
+      "step": 1269
+    },
+    {
+      "epoch": 4.535714285714286,
+      "grad_norm": 0.8358915803579491,
+      "learning_rate": 3.607361990954289e-05,
+      "loss": 0.0829,
+      "step": 1270
+    },
+    {
+      "epoch": 4.539285714285715,
+      "grad_norm": 0.5397665472694985,
+      "learning_rate": 3.6066734168241476e-05,
+      "loss": 0.0625,
+      "step": 1271
+    },
+    {
+      "epoch": 4.542857142857143,
+      "grad_norm": 0.9070137816869608,
+      "learning_rate": 3.605984305281437e-05,
+      "loss": 0.0656,
+      "step": 1272
+    },
+    {
+      "epoch": 4.546428571428572,
+      "grad_norm": 0.5961528243087447,
+      "learning_rate": 3.605294656556659e-05,
+      "loss": 0.0898,
+      "step": 1273
+    },
+    {
+      "epoch": 4.55,
+      "grad_norm": 0.5167871720758072,
+      "learning_rate": 3.60460447088049e-05,
+      "loss": 0.0468,
+      "step": 1274
+    },
+    {
+      "epoch": 4.553571428571429,
+      "grad_norm": 0.7401178713550168,
+      "learning_rate": 3.6039137484837905e-05,
+      "loss": 0.0558,
+      "step": 1275
+    },
+    {
+      "epoch": 4.557142857142857,
+      "grad_norm": 0.729645374662742,
+      "learning_rate": 3.603222489597597e-05,
+      "loss": 0.0915,
+      "step": 1276
+    },
+    {
+      "epoch": 4.560714285714286,
+      "grad_norm": 0.48337595669413597,
+      "learning_rate": 3.602530694453129e-05,
+      "loss": 0.0798,
+      "step": 1277
+    },
+    {
+      "epoch": 4.564285714285714,
+      "grad_norm": 0.48944773112096945,
+      "learning_rate": 3.601838363281783e-05,
+      "loss": 0.0459,
+      "step": 1278
+    },
+    {
+      "epoch": 4.567857142857143,
+      "grad_norm": 0.7190984422000163,
+      "learning_rate": 3.601145496315134e-05,
+      "loss": 0.0853,
+      "step": 1279
+    },
+    {
+      "epoch": 4.571428571428571,
+      "grad_norm": 0.5925627336952264,
+      "learning_rate": 3.600452093784939e-05,
+      "loss": 0.0438,
+      "step": 1280
+    },
+    {
+      "epoch": 4.575,
+      "grad_norm": 1.0776338613962266,
+      "learning_rate": 3.599758155923132e-05,
+      "loss": 0.1152,
+      "step": 1281
+    },
+    {
+      "epoch": 4.578571428571428,
+      "grad_norm": 0.9919505079163456,
+      "learning_rate": 3.599063682961826e-05,
+      "loss": 0.0949,
+      "step": 1282
+    },
+    {
+      "epoch": 4.582142857142857,
+      "grad_norm": 0.6302399530761069,
+      "learning_rate": 3.598368675133315e-05,
+      "loss": 0.0665,
+      "step": 1283
+    },
+    {
+      "epoch": 4.585714285714285,
+      "grad_norm": 0.6150098645758008,
+      "learning_rate": 3.5976731326700705e-05,
+      "loss": 0.0504,
+      "step": 1284
+    },
+    {
+      "epoch": 4.589285714285714,
+      "grad_norm": 1.194589714353097,
+      "learning_rate": 3.5969770558047424e-05,
+      "loss": 0.1178,
+      "step": 1285
+    },
+    {
+      "epoch": 4.5928571428571425,
+      "grad_norm": 0.6454241320062177,
+      "learning_rate": 3.59628044477016e-05,
+      "loss": 0.075,
+      "step": 1286
+    },
+    {
+      "epoch": 4.5964285714285715,
+      "grad_norm": 0.6071777618885226,
+      "learning_rate": 3.595583299799331e-05,
+      "loss": 0.101,
+      "step": 1287
+    },
+    {
+      "epoch": 4.6,
+      "grad_norm": 0.627812059415677,
+      "learning_rate": 3.594885621125442e-05,
+      "loss": 0.0841,
+      "step": 1288
+    },
+    {
+      "epoch": 4.603571428571429,
+      "grad_norm": 1.1104170972359388,
+      "learning_rate": 3.5941874089818585e-05,
+      "loss": 0.0865,
+      "step": 1289
+    },
+    {
+      "epoch": 4.607142857142857,
+      "grad_norm": 0.7033900869286741,
+      "learning_rate": 3.593488663602122e-05,
+      "loss": 0.0333,
+      "step": 1290
+    },
+    {
+      "epoch": 4.610714285714286,
+      "grad_norm": 0.5383739614116897,
+      "learning_rate": 3.592789385219956e-05,
+      "loss": 0.1005,
+      "step": 1291
+    },
+    {
+      "epoch": 4.614285714285714,
+      "grad_norm": 0.3958277175511265,
+      "learning_rate": 3.592089574069261e-05,
+      "loss": 0.037,
+      "step": 1292
+    },
+    {
+      "epoch": 4.617857142857143,
+      "grad_norm": 0.9162393813720697,
+      "learning_rate": 3.591389230384114e-05,
+      "loss": 0.0692,
+      "step": 1293
+    },
+    {
+      "epoch": 4.621428571428572,
+      "grad_norm": 0.5803572022734732,
+      "learning_rate": 3.590688354398772e-05,
+      "loss": 0.0706,
+      "step": 1294
+    },
+    {
+      "epoch": 4.625,
+      "grad_norm": 0.8304687058628434,
+      "learning_rate": 3.5899869463476696e-05,
+      "loss": 0.084,
+      "step": 1295
+    },
+    {
+      "epoch": 4.628571428571428,
+      "grad_norm": 0.5786053492509414,
+      "learning_rate": 3.589285006465418e-05,
+      "loss": 0.0632,
+      "step": 1296
+    },
+    {
+      "epoch": 4.632142857142857,
+      "grad_norm": 1.1587688964788547,
+      "learning_rate": 3.5885825349868093e-05,
+      "loss": 0.067,
+      "step": 1297
+    },
+    {
+      "epoch": 4.635714285714286,
+      "grad_norm": 0.8229084330698092,
+      "learning_rate": 3.58787953214681e-05,
+      "loss": 0.0829,
+      "step": 1298
+    },
+    {
+      "epoch": 4.639285714285714,
+      "grad_norm": 0.8064726803058346,
+      "learning_rate": 3.5871759981805666e-05,
+      "loss": 0.0821,
+      "step": 1299
+    },
+    {
+      "epoch": 4.642857142857143,
+      "grad_norm": 0.6367085639463929,
+      "learning_rate": 3.5864719333234025e-05,
+      "loss": 0.0605,
+      "step": 1300
+    },
+    {
+      "epoch": 4.646428571428571,
+      "grad_norm": 0.6589421862115458,
+      "learning_rate": 3.5857673378108196e-05,
+      "loss": 0.0731,
+      "step": 1301
+    },
+    {
+      "epoch": 4.65,
+      "grad_norm": 1.0670677082699798,
+      "learning_rate": 3.5850622118784945e-05,
+      "loss": 0.0547,
+      "step": 1302
+    },
+    {
+      "epoch": 4.6535714285714285,
+      "grad_norm": 0.8201416881577017,
+      "learning_rate": 3.584356555762285e-05,
+      "loss": 0.0744,
+      "step": 1303
+    },
+    {
+      "epoch": 4.6571428571428575,
+      "grad_norm": 0.8128960489100597,
+      "learning_rate": 3.5836503696982236e-05,
+      "loss": 0.066,
+      "step": 1304
+    },
+    {
+      "epoch": 4.660714285714286,
+      "grad_norm": 1.0156152289750424,
+      "learning_rate": 3.582943653922521e-05,
+      "loss": 0.0833,
+      "step": 1305
+    },
+    {
+      "epoch": 4.664285714285715,
+      "grad_norm": 0.8493750638419065,
+      "learning_rate": 3.5822364086715655e-05,
+      "loss": 0.0572,
+      "step": 1306
+    },
+    {
+      "epoch": 4.667857142857143,
+      "grad_norm": 0.8360580538868084,
+      "learning_rate": 3.581528634181921e-05,
+      "loss": 0.1083,
+      "step": 1307
+    },
+    {
+      "epoch": 4.671428571428572,
+      "grad_norm": 0.8003238671431894,
+      "learning_rate": 3.580820330690331e-05,
+      "loss": 0.0796,
+      "step": 1308
+    },
+    {
+      "epoch": 4.675,
+      "grad_norm": 0.6148553480428056,
+      "learning_rate": 3.580111498433712e-05,
+      "loss": 0.0383,
+      "step": 1309
+    },
+    {
+      "epoch": 4.678571428571429,
+      "grad_norm": 0.6116119939355008,
+      "learning_rate": 3.579402137649162e-05,
+      "loss": 0.0377,
+      "step": 1310
+    },
+    {
+      "epoch": 4.682142857142857,
+      "grad_norm": 0.8129224144469598,
+      "learning_rate": 3.5786922485739515e-05,
+      "loss": 0.0876,
+      "step": 1311
+    },
+    {
+      "epoch": 4.685714285714286,
+      "grad_norm": 0.8667127971861558,
+      "learning_rate": 3.5779818314455305e-05,
+      "loss": 0.0658,
+      "step": 1312
+    },
+    {
+      "epoch": 4.689285714285714,
+      "grad_norm": 0.6410085448598549,
+      "learning_rate": 3.577270886501525e-05,
+      "loss": 0.1062,
+      "step": 1313
+    },
+    {
+      "epoch": 4.692857142857143,
+      "grad_norm": 0.662953385291473,
+      "learning_rate": 3.5765594139797386e-05,
+      "loss": 0.0811,
+      "step": 1314
+    },
+    {
+      "epoch": 4.696428571428571,
+      "grad_norm": 0.3911112551705129,
+      "learning_rate": 3.5758474141181475e-05,
+      "loss": 0.066,
+      "step": 1315
+    },
+    {
+      "epoch": 4.7,
+      "grad_norm": 0.7338175111034914,
+      "learning_rate": 3.575134887154909e-05,
+      "loss": 0.0599,
+      "step": 1316
+    },
+    {
+      "epoch": 4.703571428571428,
+      "grad_norm": 0.617065401011239,
+      "learning_rate": 3.5744218333283536e-05,
+      "loss": 0.1058,
+      "step": 1317
+    },
+    {
+      "epoch": 4.707142857142857,
+      "grad_norm": 0.54922797953636,
+      "learning_rate": 3.57370825287699e-05,
+      "loss": 0.0619,
+      "step": 1318
+    },
+    {
+      "epoch": 4.710714285714285,
+      "grad_norm": 0.9810694656311799,
+      "learning_rate": 3.572994146039501e-05,
+      "loss": 0.1394,
+      "step": 1319
+    },
+    {
+      "epoch": 4.714285714285714,
+      "grad_norm": 0.9332910752170943,
+      "learning_rate": 3.572279513054747e-05,
+      "loss": 0.0719,
+      "step": 1320
+    },
+    {
+      "epoch": 4.7178571428571425,
+      "grad_norm": 0.4254903296051315,
+      "learning_rate": 3.5715643541617646e-05,
+      "loss": 0.0427,
+      "step": 1321
+    },
+    {
+      "epoch": 4.7214285714285715,
+      "grad_norm": 0.676348373012402,
+      "learning_rate": 3.570848669599765e-05,
+      "loss": 0.0958,
+      "step": 1322
+    },
+    {
+      "epoch": 4.725,
+      "grad_norm": 0.4274312743711668,
+      "learning_rate": 3.570132459608135e-05,
+      "loss": 0.054,
+      "step": 1323
+    },
+    {
+      "epoch": 4.728571428571429,
+      "grad_norm": 0.6559946483930749,
+      "learning_rate": 3.5694157244264404e-05,
+      "loss": 0.0973,
+      "step": 1324
+    },
+    {
+      "epoch": 4.732142857142857,
+      "grad_norm": 0.4885817292849657,
+      "learning_rate": 3.568698464294419e-05,
+      "loss": 0.0749,
+      "step": 1325
+    },
+    {
+      "epoch": 4.735714285714286,
+      "grad_norm": 0.6203298988740008,
+      "learning_rate": 3.567980679451985e-05,
+      "loss": 0.0482,
+      "step": 1326
+    },
+    {
+      "epoch": 4.739285714285714,
+      "grad_norm": 0.7223121485719798,
+      "learning_rate": 3.5672623701392304e-05,
+      "loss": 0.0719,
+      "step": 1327
+    },
+    {
+      "epoch": 4.742857142857143,
+      "grad_norm": 0.8035824945809367,
+      "learning_rate": 3.5665435365964185e-05,
+      "loss": 0.1136,
+      "step": 1328
+    },
+    {
+      "epoch": 4.746428571428572,
+      "grad_norm": 0.7639516638950494,
+      "learning_rate": 3.5658241790639915e-05,
+      "loss": 0.0884,
+      "step": 1329
+    },
+    {
+      "epoch": 4.75,
+      "grad_norm": 0.6229330778460267,
+      "learning_rate": 3.5651042977825666e-05,
+      "loss": 0.0569,
+      "step": 1330
+    },
+    {
+      "epoch": 4.753571428571428,
+      "grad_norm": 1.3139821919316998,
+      "learning_rate": 3.5643838929929334e-05,
+      "loss": 0.114,
+      "step": 1331
+    },
+    {
+      "epoch": 4.757142857142857,
+      "grad_norm": 0.5151458061182829,
+      "learning_rate": 3.56366296493606e-05,
+      "loss": 0.0683,
+      "step": 1332
+    },
+    {
+      "epoch": 4.760714285714286,
+      "grad_norm": 0.45734290196738475,
+      "learning_rate": 3.562941513853087e-05,
+      "loss": 0.0932,
+      "step": 1333
+    },
+    {
+      "epoch": 4.764285714285714,
+      "grad_norm": 0.48064331975048613,
+      "learning_rate": 3.562219539985332e-05,
+      "loss": 0.058,
+      "step": 1334
+    },
+    {
+      "epoch": 4.767857142857143,
+      "grad_norm": 0.4676091230585495,
+      "learning_rate": 3.5614970435742854e-05,
+      "loss": 0.054,
+      "step": 1335
+    },
+    {
+      "epoch": 4.771428571428571,
+      "grad_norm": 0.4171431944887826,
+      "learning_rate": 3.560774024861614e-05,
+      "loss": 0.0703,
+      "step": 1336
+    },
+    {
+      "epoch": 4.775,
+      "grad_norm": 0.5078358982385583,
+      "learning_rate": 3.5600504840891576e-05,
+      "loss": 0.0965,
+      "step": 1337
+    },
+    {
+      "epoch": 4.7785714285714285,
+      "grad_norm": 0.5661125970320657,
+      "learning_rate": 3.559326421498934e-05,
+      "loss": 0.0736,
+      "step": 1338
+    },
+    {
+      "epoch": 4.7821428571428575,
+      "grad_norm": 0.47997331310751573,
+      "learning_rate": 3.5586018373331313e-05,
+      "loss": 0.0603,
+      "step": 1339
+    },
+    {
+      "epoch": 4.785714285714286,
+      "grad_norm": 1.0593181516062835,
+      "learning_rate": 3.557876731834114e-05,
+      "loss": 0.1108,
+      "step": 1340
+    },
+    {
+      "epoch": 4.789285714285715,
+      "grad_norm": 0.7972760434193639,
+      "learning_rate": 3.557151105244423e-05,
+      "loss": 0.0657,
+      "step": 1341
+    },
+    {
+      "epoch": 4.792857142857143,
+      "grad_norm": 1.1118987844085253,
+      "learning_rate": 3.5564249578067686e-05,
+      "loss": 0.0859,
+      "step": 1342
+    },
+    {
+      "epoch": 4.796428571428572,
+      "grad_norm": 0.4456004082725553,
+      "learning_rate": 3.5556982897640404e-05,
+      "loss": 0.1008,
+      "step": 1343
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.3339090747345181,
+      "learning_rate": 3.5549711013592995e-05,
+      "loss": 0.0556,
+      "step": 1344
+    },
+    {
+      "epoch": 4.803571428571429,
+      "grad_norm": 0.4142406588245405,
+      "learning_rate": 3.5542433928357804e-05,
+      "loss": 0.0602,
+      "step": 1345
+    },
+    {
+      "epoch": 4.807142857142857,
+      "grad_norm": 0.5795361109610488,
+      "learning_rate": 3.553515164436894e-05,
+      "loss": 0.0788,
+      "step": 1346
+    },
+    {
+      "epoch": 4.810714285714286,
+      "grad_norm": 0.6498956672341994,
+      "learning_rate": 3.5527864164062236e-05,
+      "loss": 0.0736,
+      "step": 1347
+    },
+    {
+      "epoch": 4.814285714285714,
+      "grad_norm": 0.5854013309965186,
+      "learning_rate": 3.5520571489875265e-05,
+      "loss": 0.0402,
+      "step": 1348
+    },
+    {
+      "epoch": 4.817857142857143,
+      "grad_norm": 0.30477377150342877,
+      "learning_rate": 3.551327362424732e-05,
+      "loss": 0.0454,
+      "step": 1349
+    },
+    {
+      "epoch": 4.821428571428571,
+      "grad_norm": 0.7842968965446533,
+      "learning_rate": 3.550597056961948e-05,
+      "loss": 0.0583,
+      "step": 1350
+    },
+    {
+      "epoch": 4.825,
+      "grad_norm": 0.6663836375042876,
+      "learning_rate": 3.54986623284345e-05,
+      "loss": 0.0478,
+      "step": 1351
+    },
+    {
+      "epoch": 4.828571428571428,
+      "grad_norm": 0.6518351399675405,
+      "learning_rate": 3.549134890313691e-05,
+      "loss": 0.0619,
+      "step": 1352
+    },
+    {
+      "epoch": 4.832142857142857,
+      "grad_norm": 0.7317713606362746,
+      "learning_rate": 3.5484030296172954e-05,
+      "loss": 0.1013,
+      "step": 1353
+    },
+    {
+      "epoch": 4.835714285714285,
+      "grad_norm": 0.645122796536155,
+      "learning_rate": 3.547670650999062e-05,
+      "loss": 0.0638,
+      "step": 1354
+    },
+    {
+      "epoch": 4.839285714285714,
+      "grad_norm": 0.4750096160753898,
+      "learning_rate": 3.546937754703963e-05,
+      "loss": 0.0664,
+      "step": 1355
+    },
+    {
+      "epoch": 4.8428571428571425,
+      "grad_norm": 0.3419895659272165,
+      "learning_rate": 3.546204340977142e-05,
+      "loss": 0.072,
+      "step": 1356
+    },
+    {
+      "epoch": 4.8464285714285715,
+      "grad_norm": 1.0626229828486067,
+      "learning_rate": 3.545470410063917e-05,
+      "loss": 0.073,
+      "step": 1357
+    },
+    {
+      "epoch": 4.85,
+      "grad_norm": 0.7219182005538108,
+      "learning_rate": 3.544735962209781e-05,
+      "loss": 0.0873,
+      "step": 1358
+    },
+    {
+      "epoch": 4.853571428571429,
+      "grad_norm": 0.6398975361388535,
+      "learning_rate": 3.5440009976603954e-05,
+      "loss": 0.0576,
+      "step": 1359
+    },
+    {
+      "epoch": 4.857142857142857,
+      "grad_norm": 1.2098672328924547,
+      "learning_rate": 3.5432655166615976e-05,
+      "loss": 0.11,
+      "step": 1360
+    },
+    {
+      "epoch": 4.860714285714286,
+      "grad_norm": 0.8446535764654766,
+      "learning_rate": 3.5425295194593976e-05,
+      "loss": 0.0737,
+      "step": 1361
+    },
+    {
+      "epoch": 4.864285714285714,
+      "grad_norm": 0.6476827168082533,
+      "learning_rate": 3.541793006299977e-05,
+      "loss": 0.0679,
+      "step": 1362
+    },
+    {
+      "epoch": 4.867857142857143,
+      "grad_norm": 0.5298150351334668,
+      "learning_rate": 3.54105597742969e-05,
+      "loss": 0.0892,
+      "step": 1363
+    },
+    {
+      "epoch": 4.871428571428572,
+      "grad_norm": 0.4723824825656757,
+      "learning_rate": 3.540318433095064e-05,
+      "loss": 0.0934,
+      "step": 1364
+    },
+    {
+      "epoch": 4.875,
+      "grad_norm": 1.2831595681892687,
+      "learning_rate": 3.5395803735427994e-05,
+      "loss": 0.0999,
+      "step": 1365
+    },
+    {
+      "epoch": 4.878571428571428,
+      "grad_norm": 0.6447156163571397,
+      "learning_rate": 3.5388417990197665e-05,
+      "loss": 0.0537,
+      "step": 1366
+    },
+    {
+      "epoch": 4.882142857142857,
+      "grad_norm": 0.6682222676742932,
+      "learning_rate": 3.5381027097730105e-05,
+      "loss": 0.0525,
+      "step": 1367
+    },
+    {
+      "epoch": 4.885714285714286,
+      "grad_norm": 0.8333884198211053,
+      "learning_rate": 3.537363106049748e-05,
+      "loss": 0.1158,
+      "step": 1368
+    },
+    {
+      "epoch": 4.889285714285714,
+      "grad_norm": 0.5408685275830517,
+      "learning_rate": 3.5366229880973665e-05,
+      "loss": 0.067,
+      "step": 1369
+    },
+    {
+      "epoch": 4.892857142857143,
+      "grad_norm": 1.004639503978672,
+      "learning_rate": 3.535882356163426e-05,
+      "loss": 0.0603,
+      "step": 1370
+    },
+    {
+      "epoch": 4.896428571428571,
+      "grad_norm": 0.7665751687482658,
+      "learning_rate": 3.5351412104956606e-05,
+      "loss": 0.1015,
+      "step": 1371
+    },
+    {
+      "epoch": 4.9,
+      "grad_norm": 0.6678436738460631,
+      "learning_rate": 3.534399551341972e-05,
+      "loss": 0.1103,
+      "step": 1372
+    },
+    {
+      "epoch": 4.9035714285714285,
+      "grad_norm": 0.625935358352219,
+      "learning_rate": 3.533657378950439e-05,
+      "loss": 0.041,
+      "step": 1373
+    },
+    {
+      "epoch": 4.9071428571428575,
+      "grad_norm": 0.8141287376987785,
+      "learning_rate": 3.5329146935693064e-05,
+      "loss": 0.0626,
+      "step": 1374
+    },
+    {
+      "epoch": 4.910714285714286,
+      "grad_norm": 0.6437032811294984,
+      "learning_rate": 3.532171495446996e-05,
+      "loss": 0.0836,
+      "step": 1375
+    },
+    {
+      "epoch": 4.914285714285715,
+      "grad_norm": 0.5502535538596447,
+      "learning_rate": 3.531427784832096e-05,
+      "loss": 0.042,
+      "step": 1376
+    },
+    {
+      "epoch": 4.917857142857143,
+      "grad_norm": 0.5107343048103736,
+      "learning_rate": 3.530683561973369e-05,
+      "loss": 0.0621,
+      "step": 1377
+    },
+    {
+      "epoch": 4.921428571428572,
+      "grad_norm": 0.5783410001447952,
+      "learning_rate": 3.5299388271197497e-05,
+      "loss": 0.0532,
+      "step": 1378
+    },
+    {
+      "epoch": 4.925,
+      "grad_norm": 0.5200326544041551,
+      "learning_rate": 3.529193580520342e-05,
+      "loss": 0.093,
+      "step": 1379
+    },
+    {
+      "epoch": 4.928571428571429,
+      "grad_norm": 0.2927241175616995,
+      "learning_rate": 3.528447822424422e-05,
+      "loss": 0.0437,
+      "step": 1380
+    },
+    {
+      "epoch": 4.932142857142857,
+      "grad_norm": 0.49894592152946377,
+      "learning_rate": 3.527701553081437e-05,
+      "loss": 0.0363,
+      "step": 1381
+    },
+    {
+      "epoch": 4.935714285714286,
+      "grad_norm": 1.295836669108994,
+      "learning_rate": 3.526954772741003e-05,
+      "loss": 0.0954,
+      "step": 1382
+    },
+    {
+      "epoch": 4.939285714285714,
+      "grad_norm": 1.4669988761130903,
+      "learning_rate": 3.526207481652911e-05,
+      "loss": 0.1739,
+      "step": 1383
+    },
+    {
+      "epoch": 4.942857142857143,
+      "grad_norm": 1.1784279983172254,
+      "learning_rate": 3.52545968006712e-05,
+      "loss": 0.0888,
+      "step": 1384
+    },
+    {
+      "epoch": 4.946428571428571,
+      "grad_norm": 0.6936213456396214,
+      "learning_rate": 3.5247113682337595e-05,
+      "loss": 0.0995,
+      "step": 1385
+    },
+    {
+      "epoch": 4.95,
+      "grad_norm": 0.4537718436441563,
+      "learning_rate": 3.523962546403133e-05,
+      "loss": 0.0419,
+      "step": 1386
+    },
+    {
+      "epoch": 4.953571428571428,
+      "grad_norm": 0.9990481878036627,
+      "learning_rate": 3.52321321482571e-05,
+      "loss": 0.1056,
+      "step": 1387
+    },
+    {
+      "epoch": 4.957142857142857,
+      "grad_norm": 0.9421962146567046,
+      "learning_rate": 3.522463373752134e-05,
+      "loss": 0.0775,
+      "step": 1388
+    },
+    {
+      "epoch": 4.960714285714285,
+      "grad_norm": 0.8387054915416472,
+      "learning_rate": 3.521713023433217e-05,
+      "loss": 0.0794,
+      "step": 1389
+    },
+    {
+      "epoch": 4.964285714285714,
+      "grad_norm": 0.4285434789963143,
+      "learning_rate": 3.520962164119942e-05,
+      "loss": 0.0822,
+      "step": 1390
+    },
+    {
+      "epoch": 4.9678571428571425,
+      "grad_norm": 0.5769077559519581,
+      "learning_rate": 3.520210796063462e-05,
+      "loss": 0.0803,
+      "step": 1391
+    },
+    {
+      "epoch": 4.9714285714285715,
+      "grad_norm": 0.774940364383544,
+      "learning_rate": 3.519458919515102e-05,
+      "loss": 0.0808,
+      "step": 1392
+    },
+    {
+      "epoch": 4.975,
+      "grad_norm": 0.886949456952134,
+      "learning_rate": 3.518706534726353e-05,
+      "loss": 0.1012,
+      "step": 1393
+    },
+    {
+      "epoch": 4.978571428571429,
+      "grad_norm": 0.5864621683845239,
+      "learning_rate": 3.517953641948881e-05,
+      "loss": 0.0606,
+      "step": 1394
+    },
+    {
+      "epoch": 4.982142857142857,
+      "grad_norm": 0.730736125352962,
+      "learning_rate": 3.517200241434517e-05,
+      "loss": 0.0828,
+      "step": 1395
+    },
+    {
+      "epoch": 4.985714285714286,
+      "grad_norm": 0.715565802096532,
+      "learning_rate": 3.516446333435266e-05,
+      "loss": 0.0767,
+      "step": 1396
+    },
+    {
+      "epoch": 4.989285714285714,
+      "grad_norm": 0.768730179588837,
+      "learning_rate": 3.5156919182033005e-05,
+      "loss": 0.0737,
+      "step": 1397
+    },
+    {
+      "epoch": 4.992857142857143,
+      "grad_norm": 0.378407289831163,
+      "learning_rate": 3.5149369959909636e-05,
+      "loss": 0.0591,
+      "step": 1398
+    },
+    {
+      "epoch": 4.996428571428572,
+      "grad_norm": 0.3055245249435555,
+      "learning_rate": 3.514181567050766e-05,
+      "loss": 0.0198,
+      "step": 1399
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.3926888694699638,
+      "learning_rate": 3.513425631635391e-05,
+      "loss": 0.0296,
+      "step": 1400
+    },
+    {
+      "epoch": 5.003571428571429,
+      "grad_norm": 0.6029248163240349,
+      "learning_rate": 3.512669189997689e-05,
+      "loss": 0.0633,
+      "step": 1401
+    },
+    {
+      "epoch": 5.007142857142857,
+      "grad_norm": 0.7688697342997175,
+      "learning_rate": 3.51191224239068e-05,
+      "loss": 0.0797,
+      "step": 1402
+    },
+    {
+      "epoch": 5.010714285714286,
+      "grad_norm": 0.741538753339733,
+      "learning_rate": 3.5111547890675554e-05,
+      "loss": 0.0634,
+      "step": 1403
+    },
+    {
+      "epoch": 5.014285714285714,
+      "grad_norm": 0.5053481445624609,
+      "learning_rate": 3.5103968302816725e-05,
+      "loss": 0.042,
+      "step": 1404
+    },
+    {
+      "epoch": 5.017857142857143,
+      "grad_norm": 0.625788604579296,
+      "learning_rate": 3.50963836628656e-05,
+      "loss": 0.033,
+      "step": 1405
+    },
+    {
+      "epoch": 5.021428571428571,
+      "grad_norm": 0.9092921934291821,
+      "learning_rate": 3.508879397335914e-05,
+      "loss": 0.0748,
+      "step": 1406
+    },
+    {
+      "epoch": 5.025,
+      "grad_norm": 0.8356712567819203,
+      "learning_rate": 3.508119923683602e-05,
+      "loss": 0.0838,
+      "step": 1407
+    },
+    {
+      "epoch": 5.0285714285714285,
+      "grad_norm": 1.337458579539149,
+      "learning_rate": 3.507359945583656e-05,
+      "loss": 0.0617,
+      "step": 1408
+    },
+    {
+      "epoch": 5.0321428571428575,
+      "grad_norm": 1.3460432975746928,
+      "learning_rate": 3.506599463290281e-05,
+      "loss": 0.1464,
+      "step": 1409
+    },
+    {
+      "epoch": 5.035714285714286,
+      "grad_norm": 1.673356323526979,
+      "learning_rate": 3.505838477057849e-05,
+      "loss": 0.0603,
+      "step": 1410
+    },
+    {
+      "epoch": 5.039285714285715,
+      "grad_norm": 1.4446754182946289,
+      "learning_rate": 3.505076987140901e-05,
+      "loss": 0.1419,
+      "step": 1411
+    },
+    {
+      "epoch": 5.042857142857143,
+      "grad_norm": 1.2161677104795163,
+      "learning_rate": 3.504314993794145e-05,
+      "loss": 0.1005,
+      "step": 1412
+    },
+    {
+      "epoch": 5.046428571428572,
+      "grad_norm": 0.5496859081005876,
+      "learning_rate": 3.503552497272459e-05,
+      "loss": 0.1016,
+      "step": 1413
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 1.0389646982640102,
+      "learning_rate": 3.5027894978308886e-05,
+      "loss": 0.0878,
+      "step": 1414
+    },
+    {
+      "epoch": 5.053571428571429,
+      "grad_norm": 0.45238097028513247,
+      "learning_rate": 3.502025995724648e-05,
+      "loss": 0.0681,
+      "step": 1415
+    },
+    {
+      "epoch": 5.057142857142857,
+      "grad_norm": 1.4654183361546018,
+      "learning_rate": 3.501261991209118e-05,
+      "loss": 0.1093,
+      "step": 1416
+    },
+    {
+      "epoch": 5.060714285714286,
+      "grad_norm": 0.8192278407915476,
+      "learning_rate": 3.500497484539851e-05,
+      "loss": 0.0775,
+      "step": 1417
+    },
+    {
+      "epoch": 5.064285714285714,
+      "grad_norm": 0.4064824789611165,
+      "learning_rate": 3.499732475972563e-05,
+      "loss": 0.0329,
+      "step": 1418
+    },
+    {
+      "epoch": 5.067857142857143,
+      "grad_norm": 0.9959882127639348,
+      "learning_rate": 3.4989669657631417e-05,
+      "loss": 0.0595,
+      "step": 1419
+    },
+    {
+      "epoch": 5.071428571428571,
+      "grad_norm": 0.8665466847313087,
+      "learning_rate": 3.49820095416764e-05,
+      "loss": 0.0651,
+      "step": 1420
+    },
+    {
+      "epoch": 5.075,
+      "grad_norm": 0.7715643422942644,
+      "learning_rate": 3.497434441442279e-05,
+      "loss": 0.1053,
+      "step": 1421
+    },
+    {
+      "epoch": 5.078571428571428,
+      "grad_norm": 0.9275354619560882,
+      "learning_rate": 3.496667427843449e-05,
+      "loss": 0.0656,
+      "step": 1422
+    },
+    {
+      "epoch": 5.082142857142857,
+      "grad_norm": 0.469668914021755,
+      "learning_rate": 3.4958999136277056e-05,
+      "loss": 0.0486,
+      "step": 1423
+    },
+    {
+      "epoch": 5.085714285714285,
+      "grad_norm": 0.41834363154071175,
+      "learning_rate": 3.495131899051773e-05,
+      "loss": 0.0219,
+      "step": 1424
+    },
+    {
+      "epoch": 5.089285714285714,
+      "grad_norm": 0.5292551807748199,
+      "learning_rate": 3.494363384372543e-05,
+      "loss": 0.0584,
+      "step": 1425
+    },
+    {
+      "epoch": 5.0928571428571425,
+      "grad_norm": 0.7557659536546928,
+      "learning_rate": 3.493594369847075e-05,
+      "loss": 0.0897,
+      "step": 1426
+    },
+    {
+      "epoch": 5.0964285714285715,
+      "grad_norm": 0.4881069692332945,
+      "learning_rate": 3.4928248557325934e-05,
+      "loss": 0.0432,
+      "step": 1427
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.7313991497526544,
+      "learning_rate": 3.4920548422864926e-05,
+      "loss": 0.0933,
+      "step": 1428
+    },
+    {
+      "epoch": 5.103571428571429,
+      "grad_norm": 0.7064927386205304,
+      "learning_rate": 3.4912843297663315e-05,
+      "loss": 0.0533,
+      "step": 1429
+    },
+    {
+      "epoch": 5.107142857142857,
+      "grad_norm": 1.1200218552975993,
+      "learning_rate": 3.490513318429838e-05,
+      "loss": 0.0858,
+      "step": 1430
+    },
+    {
+      "epoch": 5.110714285714286,
+      "grad_norm": 1.28399215634829,
+      "learning_rate": 3.489741808534906e-05,
+      "loss": 0.0787,
+      "step": 1431
+    },
+    {
+      "epoch": 5.114285714285714,
+      "grad_norm": 1.241806009555398,
+      "learning_rate": 3.488969800339595e-05,
+      "loss": 0.0572,
+      "step": 1432
+    },
+    {
+      "epoch": 5.117857142857143,
+      "grad_norm": 0.8880862118543567,
+      "learning_rate": 3.4881972941021335e-05,
+      "loss": 0.0861,
+      "step": 1433
+    },
+    {
+      "epoch": 5.121428571428571,
+      "grad_norm": 0.5628925772121165,
+      "learning_rate": 3.487424290080915e-05,
+      "loss": 0.0987,
+      "step": 1434
+    },
+    {
+      "epoch": 5.125,
+      "grad_norm": 0.612406317474032,
+      "learning_rate": 3.4866507885344994e-05,
+      "loss": 0.0327,
+      "step": 1435
+    },
+    {
+      "epoch": 5.128571428571428,
+      "grad_norm": 0.5709966722848054,
+      "learning_rate": 3.485876789721614e-05,
+      "loss": 0.0824,
+      "step": 1436
+    },
+    {
+      "epoch": 5.132142857142857,
+      "grad_norm": 0.4723296258718329,
+      "learning_rate": 3.485102293901152e-05,
+      "loss": 0.0538,
+      "step": 1437
+    },
+    {
+      "epoch": 5.135714285714286,
+      "grad_norm": 0.5819054797364808,
+      "learning_rate": 3.4843273013321726e-05,
+      "loss": 0.0623,
+      "step": 1438
+    },
+    {
+      "epoch": 5.139285714285714,
+      "grad_norm": 0.7982750059551007,
+      "learning_rate": 3.4835518122739e-05,
+      "loss": 0.0771,
+      "step": 1439
+    },
+    {
+      "epoch": 5.142857142857143,
+      "grad_norm": 1.1761607467532476,
+      "learning_rate": 3.482775826985729e-05,
+      "loss": 0.102,
+      "step": 1440
+    },
+    {
+      "epoch": 5.146428571428571,
+      "grad_norm": 1.0817551689656228,
+      "learning_rate": 3.4819993457272135e-05,
+      "loss": 0.1226,
+      "step": 1441
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 1.1275471633688352,
+      "learning_rate": 3.48122236875808e-05,
+      "loss": 0.0931,
+      "step": 1442
+    },
+    {
+      "epoch": 5.1535714285714285,
+      "grad_norm": 0.8137810707161074,
+      "learning_rate": 3.480444896338216e-05,
+      "loss": 0.0507,
+      "step": 1443
+    },
+    {
+      "epoch": 5.1571428571428575,
+      "grad_norm": 1.0857294495148968,
+      "learning_rate": 3.479666928727676e-05,
+      "loss": 0.0558,
+      "step": 1444
+    },
+    {
+      "epoch": 5.160714285714286,
+      "grad_norm": 0.8611046155988049,
+      "learning_rate": 3.4788884661866826e-05,
+      "loss": 0.08,
+      "step": 1445
+    },
+    {
+      "epoch": 5.164285714285715,
+      "grad_norm": 1.4102386852721784,
+      "learning_rate": 3.478109508975621e-05,
+      "loss": 0.1192,
+      "step": 1446
+    },
+    {
+      "epoch": 5.167857142857143,
+      "grad_norm": 1.0599008717230938,
+      "learning_rate": 3.4773300573550425e-05,
+      "loss": 0.0925,
+      "step": 1447
+    },
+    {
+      "epoch": 5.171428571428572,
+      "grad_norm": 0.572326682912422,
+      "learning_rate": 3.4765501115856645e-05,
+      "loss": 0.0772,
+      "step": 1448
+    },
+    {
+      "epoch": 5.175,
+      "grad_norm": 0.5809694205559898,
+      "learning_rate": 3.475769671928369e-05,
+      "loss": 0.0606,
+      "step": 1449
+    },
+    {
+      "epoch": 5.178571428571429,
+      "grad_norm": 0.3039131978062911,
+      "learning_rate": 3.474988738644204e-05,
+      "loss": 0.0394,
+      "step": 1450
+    },
+    {
+      "epoch": 5.182142857142857,
+      "grad_norm": 1.2001953961113947,
+      "learning_rate": 3.4742073119943816e-05,
+      "loss": 0.1284,
+      "step": 1451
+    },
+    {
+      "epoch": 5.185714285714286,
+      "grad_norm": 2.295089105463374,
+      "learning_rate": 3.47342539224028e-05,
+      "loss": 0.2099,
+      "step": 1452
+    },
+    {
+      "epoch": 5.189285714285714,
+      "grad_norm": 0.4270870982715354,
+      "learning_rate": 3.472642979643441e-05,
+      "loss": 0.0481,
+      "step": 1453
+    },
+    {
+      "epoch": 5.192857142857143,
+      "grad_norm": 0.30108263085175874,
+      "learning_rate": 3.471860074465573e-05,
+      "loss": 0.0589,
+      "step": 1454
+    },
+    {
+      "epoch": 5.196428571428571,
+      "grad_norm": 0.7884929850238561,
+      "learning_rate": 3.4710766769685476e-05,
+      "loss": 0.0914,
+      "step": 1455
+    },
+    {
+      "epoch": 5.2,
+      "grad_norm": 0.44401339714927174,
+      "learning_rate": 3.4702927874144015e-05,
+      "loss": 0.0722,
+      "step": 1456
+    },
+    {
+      "epoch": 5.203571428571428,
+      "grad_norm": 0.8139622863688308,
+      "learning_rate": 3.4695084060653355e-05,
+      "loss": 0.0719,
+      "step": 1457
+    },
+    {
+      "epoch": 5.207142857142857,
+      "grad_norm": 0.5534957142318963,
+      "learning_rate": 3.4687235331837176e-05,
+      "loss": 0.061,
+      "step": 1458
+    },
+    {
+      "epoch": 5.210714285714285,
+      "grad_norm": 0.5797446235763496,
+      "learning_rate": 3.4679381690320766e-05,
+      "loss": 0.0897,
+      "step": 1459
+    },
+    {
+      "epoch": 5.214285714285714,
+      "grad_norm": 0.6213551691974762,
+      "learning_rate": 3.467152313873108e-05,
+      "loss": 0.0859,
+      "step": 1460
+    },
+    {
+      "epoch": 5.2178571428571425,
+      "grad_norm": 0.6837160262523164,
+      "learning_rate": 3.4663659679696695e-05,
+      "loss": 0.0784,
+      "step": 1461
+    },
+    {
+      "epoch": 5.2214285714285715,
+      "grad_norm": 0.6965176824290548,
+      "learning_rate": 3.465579131584785e-05,
+      "loss": 0.0715,
+      "step": 1462
+    },
+    {
+      "epoch": 5.225,
+      "grad_norm": 0.7606472931867871,
+      "learning_rate": 3.464791804981642e-05,
+      "loss": 0.0705,
+      "step": 1463
+    },
+    {
+      "epoch": 5.228571428571429,
+      "grad_norm": 0.6499927726284372,
+      "learning_rate": 3.464003988423591e-05,
+      "loss": 0.0727,
+      "step": 1464
+    },
+    {
+      "epoch": 5.232142857142857,
+      "grad_norm": 0.5226378928302332,
+      "learning_rate": 3.4632156821741475e-05,
+      "loss": 0.051,
+      "step": 1465
+    },
+    {
+      "epoch": 5.235714285714286,
+      "grad_norm": 0.6206554043190272,
+      "learning_rate": 3.46242688649699e-05,
+      "loss": 0.0762,
+      "step": 1466
+    },
+    {
+      "epoch": 5.239285714285714,
+      "grad_norm": 0.8741407162384907,
+      "learning_rate": 3.46163760165596e-05,
+      "loss": 0.0622,
+      "step": 1467
+    },
+    {
+      "epoch": 5.242857142857143,
+      "grad_norm": 0.88372212750573,
+      "learning_rate": 3.460847827915065e-05,
+      "loss": 0.1526,
+      "step": 1468
+    },
+    {
+      "epoch": 5.246428571428572,
+      "grad_norm": 0.4839112178942813,
+      "learning_rate": 3.460057565538474e-05,
+      "loss": 0.0712,
+      "step": 1469
+    },
+    {
+      "epoch": 5.25,
+      "grad_norm": 0.6896261948705177,
+      "learning_rate": 3.459266814790521e-05,
+      "loss": 0.0725,
+      "step": 1470
+    },
+    {
+      "epoch": 5.253571428571428,
+      "grad_norm": 1.1047993978126844,
+      "learning_rate": 3.458475575935701e-05,
+      "loss": 0.1127,
+      "step": 1471
+    },
+    {
+      "epoch": 5.257142857142857,
+      "grad_norm": 0.594572220052609,
+      "learning_rate": 3.4576838492386745e-05,
+      "loss": 0.06,
+      "step": 1472
+    },
+    {
+      "epoch": 5.260714285714286,
+      "grad_norm": 0.4811171543488252,
+      "learning_rate": 3.456891634964264e-05,
+      "loss": 0.0338,
+      "step": 1473
+    },
+    {
+      "epoch": 5.264285714285714,
+      "grad_norm": 0.45163404041256094,
+      "learning_rate": 3.4560989333774566e-05,
+      "loss": 0.0693,
+      "step": 1474
+    },
+    {
+      "epoch": 5.267857142857143,
+      "grad_norm": 0.6125594142484266,
+      "learning_rate": 3.4553057447433996e-05,
+      "loss": 0.0485,
+      "step": 1475
+    },
+    {
+      "epoch": 5.271428571428571,
+      "grad_norm": 0.35663544416810133,
+      "learning_rate": 3.454512069327407e-05,
+      "loss": 0.029,
+      "step": 1476
+    },
+    {
+      "epoch": 5.275,
+      "grad_norm": 0.6891419905233552,
+      "learning_rate": 3.45371790739495e-05,
+      "loss": 0.0662,
+      "step": 1477
+    },
+    {
+      "epoch": 5.2785714285714285,
+      "grad_norm": 0.8430909774023881,
+      "learning_rate": 3.45292325921167e-05,
+      "loss": 0.0457,
+      "step": 1478
+    },
+    {
+      "epoch": 5.2821428571428575,
+      "grad_norm": 0.6662060212100281,
+      "learning_rate": 3.4521281250433645e-05,
+      "loss": 0.0437,
+      "step": 1479
+    },
+    {
+      "epoch": 5.285714285714286,
+      "grad_norm": 0.526826962789405,
+      "learning_rate": 3.451332505155997e-05,
+      "loss": 0.0978,
+      "step": 1480
+    },
+    {
+      "epoch": 5.289285714285715,
+      "grad_norm": 1.0159231789990555,
+      "learning_rate": 3.450536399815692e-05,
+      "loss": 0.0694,
+      "step": 1481
+    },
+    {
+      "epoch": 5.292857142857143,
+      "grad_norm": 0.6225196934682656,
+      "learning_rate": 3.449739809288739e-05,
+      "loss": 0.078,
+      "step": 1482
+    },
+    {
+      "epoch": 5.296428571428572,
+      "grad_norm": 0.6001634675299203,
+      "learning_rate": 3.448942733841585e-05,
+      "loss": 0.0766,
+      "step": 1483
+    },
+    {
+      "epoch": 5.3,
+      "grad_norm": 0.9256831071649431,
+      "learning_rate": 3.4481451737408437e-05,
+      "loss": 0.0547,
+      "step": 1484
+    },
+    {
+      "epoch": 5.303571428571429,
+      "grad_norm": 0.8883498998132696,
+      "learning_rate": 3.447347129253288e-05,
+      "loss": 0.0882,
+      "step": 1485
+    },
+    {
+      "epoch": 5.307142857142857,
+      "grad_norm": 0.8158314568143361,
+      "learning_rate": 3.446548600645856e-05,
+      "loss": 0.0559,
+      "step": 1486
+    },
+    {
+      "epoch": 5.310714285714286,
+      "grad_norm": 0.7975743354112304,
+      "learning_rate": 3.4457495881856436e-05,
+      "loss": 0.0806,
+      "step": 1487
+    },
+    {
+      "epoch": 5.314285714285714,
+      "grad_norm": 0.6397821192059817,
+      "learning_rate": 3.4449500921399116e-05,
+      "loss": 0.0819,
+      "step": 1488
+    },
+    {
+      "epoch": 5.317857142857143,
+      "grad_norm": 0.7254933573576322,
+      "learning_rate": 3.4441501127760814e-05,
+      "loss": 0.0564,
+      "step": 1489
+    },
+    {
+      "epoch": 5.321428571428571,
+      "grad_norm": 0.8990873767713223,
+      "learning_rate": 3.443349650361737e-05,
+      "loss": 0.0524,
+      "step": 1490
+    },
+    {
+      "epoch": 5.325,
+      "grad_norm": 0.6136089044967238,
+      "learning_rate": 3.442548705164622e-05,
+      "loss": 0.0799,
+      "step": 1491
+    },
+    {
+      "epoch": 5.328571428571428,
+      "grad_norm": 0.866650435223203,
+      "learning_rate": 3.441747277452644e-05,
+      "loss": 0.0622,
+      "step": 1492
+    },
+    {
+      "epoch": 5.332142857142857,
+      "grad_norm": 0.43357185296261846,
+      "learning_rate": 3.44094536749387e-05,
+      "loss": 0.0463,
+      "step": 1493
+    },
+    {
+      "epoch": 5.335714285714285,
+      "grad_norm": 0.35361125422207257,
+      "learning_rate": 3.44014297555653e-05,
+      "loss": 0.0633,
+      "step": 1494
+    },
+    {
+      "epoch": 5.339285714285714,
+      "grad_norm": 0.5600082647459157,
+      "learning_rate": 3.439340101909013e-05,
+      "loss": 0.0887,
+      "step": 1495
+    },
+    {
+      "epoch": 5.3428571428571425,
+      "grad_norm": 0.9467722842965755,
+      "learning_rate": 3.438536746819871e-05,
+      "loss": 0.0626,
+      "step": 1496
+    },
+    {
+      "epoch": 5.3464285714285715,
+      "grad_norm": 1.0598555197511093,
+      "learning_rate": 3.437732910557816e-05,
+      "loss": 0.1114,
+      "step": 1497
+    },
+    {
+      "epoch": 5.35,
+      "grad_norm": 0.4940918245267479,
+      "learning_rate": 3.436928593391722e-05,
+      "loss": 0.0272,
+      "step": 1498
+    },
+    {
+      "epoch": 5.353571428571429,
+      "grad_norm": 0.4586303405967186,
+      "learning_rate": 3.436123795590623e-05,
+      "loss": 0.0532,
+      "step": 1499
+    },
+    {
+      "epoch": 5.357142857142857,
+      "grad_norm": 0.5609563048262174,
+      "learning_rate": 3.4353185174237144e-05,
+      "loss": 0.0674,
+      "step": 1500
+    },
+    {
+      "epoch": 5.360714285714286,
+      "grad_norm": 0.722587619120987,
+      "learning_rate": 3.4345127591603506e-05,
+      "loss": 0.0754,
+      "step": 1501
+    },
+    {
+      "epoch": 5.364285714285714,
+      "grad_norm": 0.49965447958630843,
+      "learning_rate": 3.4337065210700494e-05,
+      "loss": 0.0653,
+      "step": 1502
+    },
+    {
+      "epoch": 5.367857142857143,
+      "grad_norm": 0.5815648189218486,
+      "learning_rate": 3.432899803422487e-05,
+      "loss": 0.044,
+      "step": 1503
+    },
+    {
+      "epoch": 5.371428571428572,
+      "grad_norm": 1.4789405783929528,
+      "learning_rate": 3.4320926064875e-05,
+      "loss": 0.0854,
+      "step": 1504
+    },
+    {
+      "epoch": 5.375,
+      "grad_norm": 0.49060916065587407,
+      "learning_rate": 3.4312849305350866e-05,
+      "loss": 0.0887,
+      "step": 1505
+    },
+    {
+      "epoch": 5.378571428571428,
+      "grad_norm": 0.5026031688798049,
+      "learning_rate": 3.430476775835404e-05,
+      "loss": 0.0549,
+      "step": 1506
+    },
+    {
+      "epoch": 5.382142857142857,
+      "grad_norm": 0.7280732760948375,
+      "learning_rate": 3.42966814265877e-05,
+      "loss": 0.1059,
+      "step": 1507
+    },
+    {
+      "epoch": 5.385714285714286,
+      "grad_norm": 0.5537746746130177,
+      "learning_rate": 3.428859031275664e-05,
+      "loss": 0.0579,
+      "step": 1508
+    },
+    {
+      "epoch": 5.389285714285714,
+      "grad_norm": 0.6819102461238966,
+      "learning_rate": 3.428049441956722e-05,
+      "loss": 0.0894,
+      "step": 1509
+    },
+    {
+      "epoch": 5.392857142857143,
+      "grad_norm": 0.42844297242291146,
+      "learning_rate": 3.4272393749727417e-05,
+      "loss": 0.0614,
+      "step": 1510
+    },
+    {
+      "epoch": 5.396428571428571,
+      "grad_norm": 0.4681045941230553,
+      "learning_rate": 3.426428830594682e-05,
+      "loss": 0.1064,
+      "step": 1511
+    },
+    {
+      "epoch": 5.4,
+      "grad_norm": 0.49801631844272226,
+      "learning_rate": 3.425617809093659e-05,
+      "loss": 0.0821,
+      "step": 1512
+    },
+    {
+      "epoch": 5.4035714285714285,
+      "grad_norm": 0.5587594545739437,
+      "learning_rate": 3.42480631074095e-05,
+      "loss": 0.0663,
+      "step": 1513
+    },
+    {
+      "epoch": 5.4071428571428575,
+      "grad_norm": 0.5853922621754243,
+      "learning_rate": 3.42399433580799e-05,
+      "loss": 0.0614,
+      "step": 1514
+    },
+    {
+      "epoch": 5.410714285714286,
+      "grad_norm": 0.4796949693940109,
+      "learning_rate": 3.4231818845663765e-05,
+      "loss": 0.055,
+      "step": 1515
+    },
+    {
+      "epoch": 5.414285714285715,
+      "grad_norm": 0.5119931119313258,
+      "learning_rate": 3.422368957287864e-05,
+      "loss": 0.0755,
+      "step": 1516
+    },
+    {
+      "epoch": 5.417857142857143,
+      "grad_norm": 0.7777582804195667,
+      "learning_rate": 3.4215555542443655e-05,
+      "loss": 0.0904,
+      "step": 1517
+    },
+    {
+      "epoch": 5.421428571428572,
+      "grad_norm": 0.8982999108252855,
+      "learning_rate": 3.420741675707955e-05,
+      "loss": 0.1038,
+      "step": 1518
+    },
+    {
+      "epoch": 5.425,
+      "grad_norm": 0.4223897096049981,
+      "learning_rate": 3.419927321950866e-05,
+      "loss": 0.0478,
+      "step": 1519
+    },
+    {
+      "epoch": 5.428571428571429,
+      "grad_norm": 1.2467352530713804,
+      "learning_rate": 3.419112493245488e-05,
+      "loss": 0.1312,
+      "step": 1520
+    },
+    {
+      "epoch": 5.432142857142857,
+      "grad_norm": 0.5192755753418115,
+      "learning_rate": 3.418297189864373e-05,
+      "loss": 0.0667,
+      "step": 1521
+    },
+    {
+      "epoch": 5.435714285714286,
+      "grad_norm": 0.6157976939253785,
+      "learning_rate": 3.417481412080229e-05,
+      "loss": 0.0545,
+      "step": 1522
+    },
+    {
+      "epoch": 5.439285714285714,
+      "grad_norm": 0.9620183949692173,
+      "learning_rate": 3.416665160165924e-05,
+      "loss": 0.0667,
+      "step": 1523
+    },
+    {
+      "epoch": 5.442857142857143,
+      "grad_norm": 0.7732906468551863,
+      "learning_rate": 3.4158484343944834e-05,
+      "loss": 0.0907,
+      "step": 1524
+    },
+    {
+      "epoch": 5.446428571428571,
+      "grad_norm": 0.6502059093366062,
+      "learning_rate": 3.415031235039093e-05,
+      "loss": 0.0517,
+      "step": 1525
+    },
+    {
+      "epoch": 5.45,
+      "grad_norm": 0.6778953131244417,
+      "learning_rate": 3.4142135623730954e-05,
+      "loss": 0.0458,
+      "step": 1526
+    },
+    {
+      "epoch": 5.453571428571428,
+      "grad_norm": 0.7446582740323636,
+      "learning_rate": 3.413395416669992e-05,
+      "loss": 0.1013,
+      "step": 1527
+    },
+    {
+      "epoch": 5.457142857142857,
+      "grad_norm": 0.5165737121691433,
+      "learning_rate": 3.4125767982034435e-05,
+      "loss": 0.0442,
+      "step": 1528
+    },
+    {
+      "epoch": 5.460714285714285,
+      "grad_norm": 0.7604097344524313,
+      "learning_rate": 3.411757707247267e-05,
+      "loss": 0.0711,
+      "step": 1529
+    },
+    {
+      "epoch": 5.464285714285714,
+      "grad_norm": 0.8132711485653437,
+      "learning_rate": 3.4109381440754384e-05,
+      "loss": 0.0392,
+      "step": 1530
+    },
+    {
+      "epoch": 5.4678571428571425,
+      "grad_norm": 0.7233446102888489,
+      "learning_rate": 3.410118108962091e-05,
+      "loss": 0.0776,
+      "step": 1531
+    },
+    {
+      "epoch": 5.4714285714285715,
+      "grad_norm": 1.6474527603018836,
+      "learning_rate": 3.409297602181518e-05,
+      "loss": 0.1008,
+      "step": 1532
+    },
+    {
+      "epoch": 5.475,
+      "grad_norm": 0.4426964727193784,
+      "learning_rate": 3.408476624008167e-05,
+      "loss": 0.0252,
+      "step": 1533
+    },
+    {
+      "epoch": 5.478571428571429,
+      "grad_norm": 0.6053752213099841,
+      "learning_rate": 3.4076551747166466e-05,
+      "loss": 0.0487,
+      "step": 1534
+    },
+    {
+      "epoch": 5.482142857142857,
+      "grad_norm": 0.8176093162247311,
+      "learning_rate": 3.406833254581721e-05,
+      "loss": 0.1051,
+      "step": 1535
+    },
+    {
+      "epoch": 5.485714285714286,
+      "grad_norm": 0.7215917200195855,
+      "learning_rate": 3.4060108638783126e-05,
+      "loss": 0.0616,
+      "step": 1536
+    },
+    {
+      "epoch": 5.489285714285714,
+      "grad_norm": 0.6637153306590116,
+      "learning_rate": 3.4051880028815e-05,
+      "loss": 0.0756,
+      "step": 1537
+    },
+    {
+      "epoch": 5.492857142857143,
+      "grad_norm": 0.7390830283027705,
+      "learning_rate": 3.404364671866521e-05,
+      "loss": 0.0807,
+      "step": 1538
+    },
+    {
+      "epoch": 5.496428571428572,
+      "grad_norm": 1.080415725652449,
+      "learning_rate": 3.403540871108771e-05,
+      "loss": 0.0915,
+      "step": 1539
+    },
+    {
+      "epoch": 5.5,
+      "grad_norm": 0.580709020944053,
+      "learning_rate": 3.402716600883799e-05,
+      "loss": 0.0478,
+      "step": 1540
+    },
+    {
+      "epoch": 5.503571428571428,
+      "grad_norm": 0.467676310218209,
+      "learning_rate": 3.401891861467314e-05,
+      "loss": 0.0572,
+      "step": 1541
+    },
+    {
+      "epoch": 5.507142857142857,
+      "grad_norm": 0.8307019074956753,
+      "learning_rate": 3.401066653135182e-05,
+      "loss": 0.099,
+      "step": 1542
+    },
+    {
+      "epoch": 5.510714285714286,
+      "grad_norm": 0.5835255498339291,
+      "learning_rate": 3.400240976163424e-05,
+      "loss": 0.1049,
+      "step": 1543
+    },
+    {
+      "epoch": 5.514285714285714,
+      "grad_norm": 0.6927274163695396,
+      "learning_rate": 3.399414830828219e-05,
+      "loss": 0.0848,
+      "step": 1544
+    },
+    {
+      "epoch": 5.517857142857143,
+      "grad_norm": 0.5538346150729959,
+      "learning_rate": 3.3985882174059037e-05,
+      "loss": 0.0622,
+      "step": 1545
+    },
+    {
+      "epoch": 5.521428571428571,
+      "grad_norm": 0.7482821766415303,
+      "learning_rate": 3.397761136172969e-05,
+      "loss": 0.0648,
+      "step": 1546
+    },
+    {
+      "epoch": 5.525,
+      "grad_norm": 0.583141605008031,
+      "learning_rate": 3.396933587406064e-05,
+      "loss": 0.0952,
+      "step": 1547
+    },
+    {
+      "epoch": 5.5285714285714285,
+      "grad_norm": 1.4208118843161324,
+      "learning_rate": 3.396105571381993e-05,
+      "loss": 0.0614,
+      "step": 1548
+    },
+    {
+      "epoch": 5.5321428571428575,
+      "grad_norm": 0.765104979150866,
+      "learning_rate": 3.395277088377718e-05,
+      "loss": 0.0778,
+      "step": 1549
+    },
+    {
+      "epoch": 5.535714285714286,
+      "grad_norm": 0.7711519006385451,
+      "learning_rate": 3.394448138670355e-05,
+      "loss": 0.0572,
+      "step": 1550
+    },
+    {
+      "epoch": 5.539285714285715,
+      "grad_norm": 0.42361730065268716,
+      "learning_rate": 3.393618722537179e-05,
+      "loss": 0.0421,
+      "step": 1551
+    },
+    {
+      "epoch": 5.542857142857143,
+      "grad_norm": 0.8292549555015144,
+      "learning_rate": 3.3927888402556194e-05,
+      "loss": 0.1389,
+      "step": 1552
+    },
+    {
+      "epoch": 5.546428571428572,
+      "grad_norm": 0.5712416334592197,
+      "learning_rate": 3.3919584921032614e-05,
+      "loss": 0.0523,
+      "step": 1553
+    },
+    {
+      "epoch": 5.55,
+      "grad_norm": 0.4012536457109647,
+      "learning_rate": 3.391127678357846e-05,
+      "loss": 0.0636,
+      "step": 1554
+    },
+    {
+      "epoch": 5.553571428571429,
+      "grad_norm": 0.6398130576428169,
+      "learning_rate": 3.3902963992972695e-05,
+      "loss": 0.0867,
+      "step": 1555
+    },
+    {
+      "epoch": 5.557142857142857,
+      "grad_norm": 0.4062731786949905,
+      "learning_rate": 3.3894646551995866e-05,
+      "loss": 0.0353,
+      "step": 1556
+    },
+    {
+      "epoch": 5.560714285714286,
+      "grad_norm": 0.48801776035170247,
+      "learning_rate": 3.3886324463430035e-05,
+      "loss": 0.0725,
+      "step": 1557
+    },
+    {
+      "epoch": 5.564285714285714,
+      "grad_norm": 0.5473727077417677,
+      "learning_rate": 3.387799773005885e-05,
+      "loss": 0.1305,
+      "step": 1558
+    },
+    {
+      "epoch": 5.567857142857143,
+      "grad_norm": 0.4593919638263103,
+      "learning_rate": 3.38696663546675e-05,
+      "loss": 0.0664,
+      "step": 1559
+    },
+    {
+      "epoch": 5.571428571428571,
+      "grad_norm": 0.9642617789159486,
+      "learning_rate": 3.386133034004273e-05,
+      "loss": 0.1109,
+      "step": 1560
+    },
+    {
+      "epoch": 5.575,
+      "grad_norm": 0.4996473741059193,
+      "learning_rate": 3.3852989688972824e-05,
+      "loss": 0.0564,
+      "step": 1561
+    },
+    {
+      "epoch": 5.578571428571428,
+      "grad_norm": 0.6490229507820638,
+      "learning_rate": 3.384464440424764e-05,
+      "loss": 0.059,
+      "step": 1562
+    },
+    {
+      "epoch": 5.582142857142857,
+      "grad_norm": 0.47557781810311467,
+      "learning_rate": 3.383629448865856e-05,
+      "loss": 0.096,
+      "step": 1563
+    },
+    {
+      "epoch": 5.585714285714285,
+      "grad_norm": 0.724654769664419,
+      "learning_rate": 3.3827939944998544e-05,
+      "loss": 0.1143,
+      "step": 1564
+    },
+    {
+      "epoch": 5.589285714285714,
+      "grad_norm": 1.1851199623385826,
+      "learning_rate": 3.381958077606208e-05,
+      "loss": 0.0979,
+      "step": 1565
+    },
+    {
+      "epoch": 5.5928571428571425,
+      "grad_norm": 0.4719995710704958,
+      "learning_rate": 3.3811216984645196e-05,
+      "loss": 0.0753,
+      "step": 1566
+    },
+    {
+      "epoch": 5.5964285714285715,
+      "grad_norm": 0.42734389071379403,
+      "learning_rate": 3.380284857354549e-05,
+      "loss": 0.0462,
+      "step": 1567
+    },
+    {
+      "epoch": 5.6,
+      "grad_norm": 0.5433157291471181,
+      "learning_rate": 3.379447554556209e-05,
+      "loss": 0.0588,
+      "step": 1568
+    },
+    {
+      "epoch": 5.603571428571429,
+      "grad_norm": 0.6663646838605957,
+      "learning_rate": 3.3786097903495666e-05,
+      "loss": 0.0745,
+      "step": 1569
+    },
+    {
+      "epoch": 5.607142857142857,
+      "grad_norm": 0.8069999134382498,
+      "learning_rate": 3.377771565014845e-05,
+      "loss": 0.0692,
+      "step": 1570
+    },
+    {
+      "epoch": 5.610714285714286,
+      "grad_norm": 0.459209826067309,
+      "learning_rate": 3.3769328788324184e-05,
+      "loss": 0.0635,
+      "step": 1571
+    },
+    {
+      "epoch": 5.614285714285714,
+      "grad_norm": 0.8614682610147112,
+      "learning_rate": 3.376093732082819e-05,
+      "loss": 0.0737,
+      "step": 1572
+    },
+    {
+      "epoch": 5.617857142857143,
+      "grad_norm": 0.47109145978480443,
+      "learning_rate": 3.3752541250467295e-05,
+      "loss": 0.0683,
+      "step": 1573
+    },
+    {
+      "epoch": 5.621428571428572,
+      "grad_norm": 0.8694524837731129,
+      "learning_rate": 3.374414058004989e-05,
+      "loss": 0.0722,
+      "step": 1574
+    },
+    {
+      "epoch": 5.625,
+      "grad_norm": 0.8419587913839013,
+      "learning_rate": 3.373573531238589e-05,
+      "loss": 0.12,
+      "step": 1575
+    },
+    {
+      "epoch": 5.628571428571428,
+      "grad_norm": 1.1519153374446875,
+      "learning_rate": 3.372732545028677e-05,
+      "loss": 0.1541,
+      "step": 1576
+    },
+    {
+      "epoch": 5.632142857142857,
+      "grad_norm": 0.608728210024841,
+      "learning_rate": 3.371891099656551e-05,
+      "loss": 0.0478,
+      "step": 1577
+    },
+    {
+      "epoch": 5.635714285714286,
+      "grad_norm": 0.8540536106129224,
+      "learning_rate": 3.3710491954036645e-05,
+      "loss": 0.0917,
+      "step": 1578
+    },
+    {
+      "epoch": 5.639285714285714,
+      "grad_norm": 0.6375877980621725,
+      "learning_rate": 3.370206832551624e-05,
+      "loss": 0.0632,
+      "step": 1579
+    },
+    {
+      "epoch": 5.642857142857143,
+      "grad_norm": 0.6770380519639472,
+      "learning_rate": 3.36936401138219e-05,
+      "loss": 0.0613,
+      "step": 1580
+    },
+    {
+      "epoch": 5.646428571428571,
+      "grad_norm": 1.0088386670873493,
+      "learning_rate": 3.368520732177276e-05,
+      "loss": 0.0721,
+      "step": 1581
+    },
+    {
+      "epoch": 5.65,
+      "grad_norm": 0.4279275180078502,
+      "learning_rate": 3.3676769952189476e-05,
+      "loss": 0.0402,
+      "step": 1582
+    },
+    {
+      "epoch": 5.6535714285714285,
+      "grad_norm": 0.6368505244734077,
+      "learning_rate": 3.366832800789426e-05,
+      "loss": 0.0407,
+      "step": 1583
+    },
+    {
+      "epoch": 5.6571428571428575,
+      "grad_norm": 0.7270407182030377,
+      "learning_rate": 3.365988149171082e-05,
+      "loss": 0.0563,
+      "step": 1584
+    },
+    {
+      "epoch": 5.660714285714286,
+      "grad_norm": 1.1270064124619124,
+      "learning_rate": 3.365143040646443e-05,
+      "loss": 0.1272,
+      "step": 1585
+    },
+    {
+      "epoch": 5.664285714285715,
+      "grad_norm": 1.1515803193166738,
+      "learning_rate": 3.364297475498186e-05,
+      "loss": 0.0667,
+      "step": 1586
+    },
+    {
+      "epoch": 5.667857142857143,
+      "grad_norm": 0.9171948507311306,
+      "learning_rate": 3.363451454009143e-05,
+      "loss": 0.076,
+      "step": 1587
+    },
+    {
+      "epoch": 5.671428571428572,
+      "grad_norm": 0.6419350973826998,
+      "learning_rate": 3.3626049764622984e-05,
+      "loss": 0.0687,
+      "step": 1588
+    },
+    {
+      "epoch": 5.675,
+      "grad_norm": 0.6989083592431253,
+      "learning_rate": 3.3617580431407866e-05,
+      "loss": 0.0654,
+      "step": 1589
+    },
+    {
+      "epoch": 5.678571428571429,
+      "grad_norm": 0.6021261473070199,
+      "learning_rate": 3.360910654327898e-05,
+      "loss": 0.0664,
+      "step": 1590
+    },
+    {
+      "epoch": 5.682142857142857,
+      "grad_norm": 0.792885855095265,
+      "learning_rate": 3.3600628103070745e-05,
+      "loss": 0.08,
+      "step": 1591
+    },
+    {
+      "epoch": 5.685714285714286,
+      "grad_norm": 0.6377322845179819,
+      "learning_rate": 3.359214511361907e-05,
+      "loss": 0.0582,
+      "step": 1592
+    },
+    {
+      "epoch": 5.689285714285714,
+      "grad_norm": 1.1688686161845556,
+      "learning_rate": 3.3583657577761435e-05,
+      "loss": 0.1224,
+      "step": 1593
+    },
+    {
+      "epoch": 5.692857142857143,
+      "grad_norm": 0.9190316657866606,
+      "learning_rate": 3.3575165498336806e-05,
+      "loss": 0.0847,
+      "step": 1594
+    },
+    {
+      "epoch": 5.696428571428571,
+      "grad_norm": 0.5134298431790334,
+      "learning_rate": 3.356666887818568e-05,
+      "loss": 0.0473,
+      "step": 1595
+    },
+    {
+      "epoch": 5.7,
+      "grad_norm": 0.5268079474528679,
+      "learning_rate": 3.3558167720150064e-05,
+      "loss": 0.0589,
+      "step": 1596
+    },
+    {
+      "epoch": 5.703571428571428,
+      "grad_norm": 0.6963198774612741,
+      "learning_rate": 3.354966202707351e-05,
+      "loss": 0.1252,
+      "step": 1597
+    },
+    {
+      "epoch": 5.707142857142857,
+      "grad_norm": 0.7008492981837539,
+      "learning_rate": 3.354115180180106e-05,
+      "loss": 0.092,
+      "step": 1598
+    },
+    {
+      "epoch": 5.710714285714285,
+      "grad_norm": 0.5371556250209782,
+      "learning_rate": 3.3532637047179274e-05,
+      "loss": 0.0628,
+      "step": 1599
+    },
+    {
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.6415273646703847,
+      "learning_rate": 3.352411776605624e-05,
+      "loss": 0.0764,
+      "step": 1600
+    },
+    {
+      "epoch": 5.7178571428571425,
+      "grad_norm": 0.4758356391997447,
+      "learning_rate": 3.3515593961281546e-05,
+      "loss": 0.0587,
+      "step": 1601
+    },
+    {
+      "epoch": 5.7214285714285715,
+      "grad_norm": 0.601597267419943,
+      "learning_rate": 3.3507065635706305e-05,
+      "loss": 0.0846,
+      "step": 1602
+    },
+    {
+      "epoch": 5.725,
+      "grad_norm": 0.5618480490452086,
+      "learning_rate": 3.349853279218314e-05,
+      "loss": 0.0679,
+      "step": 1603
+    },
+    {
+      "epoch": 5.728571428571429,
+      "grad_norm": 0.940588501838182,
+      "learning_rate": 3.348999543356618e-05,
+      "loss": 0.0622,
+      "step": 1604
+    },
+    {
+      "epoch": 5.732142857142857,
+      "grad_norm": 0.5746554914435508,
+      "learning_rate": 3.3481453562711064e-05,
+      "loss": 0.0656,
+      "step": 1605
+    },
+    {
+      "epoch": 5.735714285714286,
+      "grad_norm": 1.0043359486942023,
+      "learning_rate": 3.347290718247495e-05,
+      "loss": 0.0889,
+      "step": 1606
+    },
+    {
+      "epoch": 5.739285714285714,
+      "grad_norm": 1.1303089720307113,
+      "learning_rate": 3.346435629571649e-05,
+      "loss": 0.0822,
+      "step": 1607
+    },
+    {
+      "epoch": 5.742857142857143,
+      "grad_norm": 1.064500155252076,
+      "learning_rate": 3.345580090529585e-05,
+      "loss": 0.0738,
+      "step": 1608
+    },
+    {
+      "epoch": 5.746428571428572,
+      "grad_norm": 0.9179420498854473,
+      "learning_rate": 3.344724101407471e-05,
+      "loss": 0.0614,
+      "step": 1609
+    },
+    {
+      "epoch": 5.75,
+      "grad_norm": 1.1167239192462641,
+      "learning_rate": 3.3438676624916246e-05,
+      "loss": 0.086,
+      "step": 1610
+    },
+    {
+      "epoch": 5.753571428571428,
+      "grad_norm": 0.9995000933016261,
+      "learning_rate": 3.343010774068515e-05,
+      "loss": 0.0774,
+      "step": 1611
+    },
+    {
+      "epoch": 5.757142857142857,
+      "grad_norm": 0.6929487349718625,
+      "learning_rate": 3.342153436424759e-05,
+      "loss": 0.0628,
+      "step": 1612
+    },
+    {
+      "epoch": 5.760714285714286,
+      "grad_norm": 0.4896679752613396,
+      "learning_rate": 3.341295649847127e-05,
+      "loss": 0.0294,
+      "step": 1613
+    },
+    {
+      "epoch": 5.764285714285714,
+      "grad_norm": 1.293627985469536,
+      "learning_rate": 3.3404374146225375e-05,
+      "loss": 0.0755,
+      "step": 1614
+    },
+    {
+      "epoch": 5.767857142857143,
+      "grad_norm": 0.5072901520888614,
+      "learning_rate": 3.33957873103806e-05,
+      "loss": 0.0408,
+      "step": 1615
+    },
+    {
+      "epoch": 5.771428571428571,
+      "grad_norm": 0.7581145505066798,
+      "learning_rate": 3.338719599380913e-05,
+      "loss": 0.0766,
+      "step": 1616
+    },
+    {
+      "epoch": 5.775,
+      "grad_norm": 0.5489372106508994,
+      "learning_rate": 3.3378600199384664e-05,
+      "loss": 0.0681,
+      "step": 1617
+    },
+    {
+      "epoch": 5.7785714285714285,
+      "grad_norm": 0.8122643304640509,
+      "learning_rate": 3.336999992998238e-05,
+      "loss": 0.0608,
+      "step": 1618
+    },
+    {
+      "epoch": 5.7821428571428575,
+      "grad_norm": 0.7220733346849562,
+      "learning_rate": 3.336139518847897e-05,
+      "loss": 0.0725,
+      "step": 1619
+    },
+    {
+      "epoch": 5.785714285714286,
+      "grad_norm": 0.5989833571836286,
+      "learning_rate": 3.335278597775261e-05,
+      "loss": 0.088,
+      "step": 1620
+    },
+    {
+      "epoch": 5.789285714285715,
+      "grad_norm": 0.4949535122752444,
+      "learning_rate": 3.334417230068298e-05,
+      "loss": 0.028,
+      "step": 1621
+    },
+    {
+      "epoch": 5.792857142857143,
+      "grad_norm": 0.5146716057111027,
+      "learning_rate": 3.333555416015125e-05,
+      "loss": 0.0671,
+      "step": 1622
+    },
+    {
+      "epoch": 5.796428571428572,
+      "grad_norm": 0.9337637587961043,
+      "learning_rate": 3.3326931559040084e-05,
+      "loss": 0.1002,
+      "step": 1623
+    },
+    {
+      "epoch": 5.8,
+      "grad_norm": 0.5543247159317684,
+      "learning_rate": 3.331830450023362e-05,
+      "loss": 0.0417,
+      "step": 1624
+    },
+    {
+      "epoch": 5.803571428571429,
+      "grad_norm": 0.8011590848759809,
+      "learning_rate": 3.330967298661753e-05,
+      "loss": 0.0993,
+      "step": 1625
+    },
+    {
+      "epoch": 5.807142857142857,
+      "grad_norm": 0.511374976666901,
+      "learning_rate": 3.330103702107892e-05,
+      "loss": 0.057,
+      "step": 1626
+    },
+    {
+      "epoch": 5.810714285714286,
+      "grad_norm": 1.1516564285109545,
+      "learning_rate": 3.3292396606506445e-05,
+      "loss": 0.0546,
+      "step": 1627
+    },
+    {
+      "epoch": 5.814285714285714,
+      "grad_norm": 1.325753549228182,
+      "learning_rate": 3.3283751745790197e-05,
+      "loss": 0.0598,
+      "step": 1628
+    },
+    {
+      "epoch": 5.817857142857143,
+      "grad_norm": 0.5573744389276002,
+      "learning_rate": 3.327510244182178e-05,
+      "loss": 0.1115,
+      "step": 1629
+    },
+    {
+      "epoch": 5.821428571428571,
+      "grad_norm": 1.3674838469334591,
+      "learning_rate": 3.326644869749428e-05,
+      "loss": 0.1258,
+      "step": 1630
+    },
+    {
+      "epoch": 5.825,
+      "grad_norm": 0.7076019911984487,
+      "learning_rate": 3.325779051570227e-05,
+      "loss": 0.0724,
+      "step": 1631
+    },
+    {
+      "epoch": 5.828571428571428,
+      "grad_norm": 1.089760844797165,
+      "learning_rate": 3.3249127899341806e-05,
+      "loss": 0.1285,
+      "step": 1632
+    },
+    {
+      "epoch": 5.832142857142857,
+      "grad_norm": 0.8359719529332756,
+      "learning_rate": 3.3240460851310435e-05,
+      "loss": 0.0883,
+      "step": 1633
+    },
+    {
+      "epoch": 5.835714285714285,
+      "grad_norm": 0.8207703577805007,
+      "learning_rate": 3.323178937450717e-05,
+      "loss": 0.1009,
+      "step": 1634
+    },
+    {
+      "epoch": 5.839285714285714,
+      "grad_norm": 0.7985685315345563,
+      "learning_rate": 3.3223113471832506e-05,
+      "loss": 0.0665,
+      "step": 1635
+    },
+    {
+      "epoch": 5.8428571428571425,
+      "grad_norm": 0.7613449868480089,
+      "learning_rate": 3.3214433146188436e-05,
+      "loss": 0.0918,
+      "step": 1636
+    },
+    {
+      "epoch": 5.8464285714285715,
+      "grad_norm": 0.7056898264450547,
+      "learning_rate": 3.320574840047842e-05,
+      "loss": 0.0372,
+      "step": 1637
+    },
+    {
+      "epoch": 5.85,
+      "grad_norm": 0.5874371313201198,
+      "learning_rate": 3.31970592376074e-05,
+      "loss": 0.0615,
+      "step": 1638
+    },
+    {
+      "epoch": 5.853571428571429,
+      "grad_norm": 1.0047361601231377,
+      "learning_rate": 3.3188365660481794e-05,
+      "loss": 0.0699,
+      "step": 1639
+    },
+    {
+      "epoch": 5.857142857142857,
+      "grad_norm": 0.5260123641770439,
+      "learning_rate": 3.3179667672009506e-05,
+      "loss": 0.0402,
+      "step": 1640
+    },
+    {
+      "epoch": 5.860714285714286,
+      "grad_norm": 0.9979680543970817,
+      "learning_rate": 3.3170965275099885e-05,
+      "loss": 0.0976,
+      "step": 1641
+    },
+    {
+      "epoch": 5.864285714285714,
+      "grad_norm": 0.841027332989944,
+      "learning_rate": 3.3162258472663796e-05,
+      "loss": 0.1161,
+      "step": 1642
+    },
+    {
+      "epoch": 5.867857142857143,
+      "grad_norm": 1.022492618955773,
+      "learning_rate": 3.315354726761354e-05,
+      "loss": 0.0987,
+      "step": 1643
+    },
+    {
+      "epoch": 5.871428571428572,
+      "grad_norm": 0.9327204679707837,
+      "learning_rate": 3.314483166286293e-05,
+      "loss": 0.0882,
+      "step": 1644
+    },
+    {
+      "epoch": 5.875,
+      "grad_norm": 0.44216333548872533,
+      "learning_rate": 3.313611166132722e-05,
+      "loss": 0.0468,
+      "step": 1645
+    },
+    {
+      "epoch": 5.878571428571428,
+      "grad_norm": 0.9794158374072535,
+      "learning_rate": 3.312738726592314e-05,
+      "loss": 0.0854,
+      "step": 1646
+    },
+    {
+      "epoch": 5.882142857142857,
+      "grad_norm": 0.590974852378728,
+      "learning_rate": 3.311865847956889e-05,
+      "loss": 0.0357,
+      "step": 1647
+    },
+    {
+      "epoch": 5.885714285714286,
+      "grad_norm": 0.7717462111945098,
+      "learning_rate": 3.3109925305184156e-05,
+      "loss": 0.0522,
+      "step": 1648
+    },
+    {
+      "epoch": 5.889285714285714,
+      "grad_norm": 0.5642039267009152,
+      "learning_rate": 3.3101187745690055e-05,
+      "loss": 0.0418,
+      "step": 1649
+    },
+    {
+      "epoch": 5.892857142857143,
+      "grad_norm": 1.0907336338440539,
+      "learning_rate": 3.309244580400922e-05,
+      "loss": 0.0893,
+      "step": 1650
+    },
+    {
+      "epoch": 5.896428571428571,
+      "grad_norm": 0.7082153382020161,
+      "learning_rate": 3.30836994830657e-05,
+      "loss": 0.0503,
+      "step": 1651
+    },
+    {
+      "epoch": 5.9,
+      "grad_norm": 1.2255226050499322,
+      "learning_rate": 3.3074948785785054e-05,
+      "loss": 0.0741,
+      "step": 1652
+    },
+    {
+      "epoch": 5.9035714285714285,
+      "grad_norm": 0.5390543074574875,
+      "learning_rate": 3.3066193715094263e-05,
+      "loss": 0.0767,
+      "step": 1653
+    },
+    {
+      "epoch": 5.9071428571428575,
+      "grad_norm": 0.49747941693742304,
+      "learning_rate": 3.30574342739218e-05,
+      "loss": 0.033,
+      "step": 1654
+    },
+    {
+      "epoch": 5.910714285714286,
+      "grad_norm": 0.7577213230275777,
+      "learning_rate": 3.3048670465197595e-05,
+      "loss": 0.1197,
+      "step": 1655
+    },
+    {
+      "epoch": 5.914285714285715,
+      "grad_norm": 1.3393038137016424,
+      "learning_rate": 3.303990229185303e-05,
+      "loss": 0.1027,
+      "step": 1656
+    },
+    {
+      "epoch": 5.917857142857143,
+      "grad_norm": 0.6967586009071692,
+      "learning_rate": 3.303112975682095e-05,
+      "loss": 0.0804,
+      "step": 1657
+    },
+    {
+      "epoch": 5.921428571428572,
+      "grad_norm": 0.7516058030278857,
+      "learning_rate": 3.302235286303566e-05,
+      "loss": 0.1129,
+      "step": 1658
+    },
+    {
+      "epoch": 5.925,
+      "grad_norm": 0.804652344029168,
+      "learning_rate": 3.301357161343293e-05,
+      "loss": 0.0822,
+      "step": 1659
+    },
+    {
+      "epoch": 5.928571428571429,
+      "grad_norm": 0.5785596125244727,
+      "learning_rate": 3.300478601094998e-05,
+      "loss": 0.0719,
+      "step": 1660
+    },
+    {
+      "epoch": 5.932142857142857,
+      "grad_norm": 1.1114912812491016,
+      "learning_rate": 3.2995996058525473e-05,
+      "loss": 0.0796,
+      "step": 1661
+    },
+    {
+      "epoch": 5.935714285714286,
+      "grad_norm": 0.8150678686193886,
+      "learning_rate": 3.2987201759099554e-05,
+      "loss": 0.0464,
+      "step": 1662
+    },
+    {
+      "epoch": 5.939285714285714,
+      "grad_norm": 0.3543939423937253,
+      "learning_rate": 3.29784031156138e-05,
+      "loss": 0.0463,
+      "step": 1663
+    },
+    {
+      "epoch": 5.942857142857143,
+      "grad_norm": 0.2947125727613372,
+      "learning_rate": 3.2969600131011254e-05,
+      "loss": 0.0377,
+      "step": 1664
+    },
+    {
+      "epoch": 5.946428571428571,
+      "grad_norm": 0.49279972367926805,
+      "learning_rate": 3.2960792808236415e-05,
+      "loss": 0.0859,
+      "step": 1665
+    },
+    {
+      "epoch": 5.95,
+      "grad_norm": 0.5641878231170565,
+      "learning_rate": 3.2951981150235205e-05,
+      "loss": 0.0471,
+      "step": 1666
+    },
+    {
+      "epoch": 5.953571428571428,
+      "grad_norm": 0.7438652299534987,
+      "learning_rate": 3.294316515995503e-05,
+      "loss": 0.0477,
+      "step": 1667
+    },
+    {
+      "epoch": 5.957142857142857,
+      "grad_norm": 0.545540533378132,
+      "learning_rate": 3.293434484034473e-05,
+      "loss": 0.0857,
+      "step": 1668
+    },
+    {
+      "epoch": 5.960714285714285,
+      "grad_norm": 0.3714322799243328,
+      "learning_rate": 3.2925520194354586e-05,
+      "loss": 0.0212,
+      "step": 1669
+    },
+    {
+      "epoch": 5.964285714285714,
+      "grad_norm": 0.29902337891147046,
+      "learning_rate": 3.291669122493634e-05,
+      "loss": 0.0273,
+      "step": 1670
+    },
+    {
+      "epoch": 5.9678571428571425,
+      "grad_norm": 1.2839799718883587,
+      "learning_rate": 3.2907857935043167e-05,
+      "loss": 0.1243,
+      "step": 1671
+    },
+    {
+      "epoch": 5.9714285714285715,
+      "grad_norm": 1.326948161154676,
+      "learning_rate": 3.289902032762971e-05,
+      "loss": 0.1327,
+      "step": 1672
+    },
+    {
+      "epoch": 5.975,
+      "grad_norm": 0.4355472253943165,
+      "learning_rate": 3.289017840565202e-05,
+      "loss": 0.0335,
+      "step": 1673
+    },
+    {
+      "epoch": 5.978571428571429,
+      "grad_norm": 0.45489348335549273,
+      "learning_rate": 3.2881332172067636e-05,
+      "loss": 0.0306,
+      "step": 1674
+    },
+    {
+      "epoch": 5.982142857142857,
+      "grad_norm": 0.6130060409187914,
+      "learning_rate": 3.287248162983549e-05,
+      "loss": 0.0348,
+      "step": 1675
+    },
+    {
+      "epoch": 5.985714285714286,
+      "grad_norm": 0.7785561825581222,
+      "learning_rate": 3.286362678191599e-05,
+      "loss": 0.0697,
+      "step": 1676
+    },
+    {
+      "epoch": 5.989285714285714,
+      "grad_norm": 0.6679363304735413,
+      "learning_rate": 3.285476763127097e-05,
+      "loss": 0.0575,
+      "step": 1677
+    },
+    {
+      "epoch": 5.992857142857143,
+      "grad_norm": 0.7774266381946087,
+      "learning_rate": 3.284590418086372e-05,
+      "loss": 0.1464,
+      "step": 1678
+    },
+    {
+      "epoch": 5.996428571428572,
+      "grad_norm": 0.5539218120865095,
+      "learning_rate": 3.2837036433658944e-05,
+      "loss": 0.0529,
+      "step": 1679
+    },
+    {
+      "epoch": 6.0,
+      "grad_norm": 0.614452317217278,
+      "learning_rate": 3.2828164392622804e-05,
+      "loss": 0.0687,
+      "step": 1680
+    },
+    {
+      "epoch": 6.003571428571429,
+      "grad_norm": 0.7083452942583913,
+      "learning_rate": 3.281928806072288e-05,
+      "loss": 0.0493,
+      "step": 1681
+    },
+    {
+      "epoch": 6.007142857142857,
+      "grad_norm": 0.32376501269825364,
+      "learning_rate": 3.28104074409282e-05,
+      "loss": 0.0327,
+      "step": 1682
+    },
+    {
+      "epoch": 6.010714285714286,
+      "grad_norm": 0.9416101141773034,
+      "learning_rate": 3.2801522536209214e-05,
+      "loss": 0.0962,
+      "step": 1683
+    },
+    {
+      "epoch": 6.014285714285714,
+      "grad_norm": 0.8975712146565344,
+      "learning_rate": 3.279263334953783e-05,
+      "loss": 0.0586,
+      "step": 1684
+    },
+    {
+      "epoch": 6.017857142857143,
+      "grad_norm": 0.6367000270327544,
+      "learning_rate": 3.278373988388736e-05,
+      "loss": 0.0723,
+      "step": 1685
+    },
+    {
+      "epoch": 6.021428571428571,
+      "grad_norm": 0.6461113114401705,
+      "learning_rate": 3.2774842142232565e-05,
+      "loss": 0.0922,
+      "step": 1686
+    },
+    {
+      "epoch": 6.025,
+      "grad_norm": 0.2423069751994496,
+      "learning_rate": 3.276594012754962e-05,
+      "loss": 0.0259,
+      "step": 1687
+    },
+    {
+      "epoch": 6.0285714285714285,
+      "grad_norm": 0.7971304967897935,
+      "learning_rate": 3.2757033842816155e-05,
+      "loss": 0.0754,
+      "step": 1688
+    },
+    {
+      "epoch": 6.0321428571428575,
+      "grad_norm": 0.8712880757840296,
+      "learning_rate": 3.274812329101121e-05,
+      "loss": 0.0976,
+      "step": 1689
+    },
+    {
+      "epoch": 6.035714285714286,
+      "grad_norm": 0.4761017922522581,
+      "learning_rate": 3.273920847511525e-05,
+      "loss": 0.0664,
+      "step": 1690
+    },
+    {
+      "epoch": 6.039285714285715,
+      "grad_norm": 0.48734335910746124,
+      "learning_rate": 3.273028939811017e-05,
+      "loss": 0.0605,
+      "step": 1691
+    },
+    {
+      "epoch": 6.042857142857143,
+      "grad_norm": 0.9219260946578877,
+      "learning_rate": 3.2721366062979293e-05,
+      "loss": 0.0845,
+      "step": 1692
+    },
+    {
+      "epoch": 6.046428571428572,
+      "grad_norm": 0.5494864308493974,
+      "learning_rate": 3.2712438472707376e-05,
+      "loss": 0.0601,
+      "step": 1693
+    },
+    {
+      "epoch": 6.05,
+      "grad_norm": 0.5459145014884595,
+      "learning_rate": 3.270350663028057e-05,
+      "loss": 0.0622,
+      "step": 1694
+    },
+    {
+      "epoch": 6.053571428571429,
+      "grad_norm": 0.9872556383504931,
+      "learning_rate": 3.269457053868649e-05,
+      "loss": 0.0555,
+      "step": 1695
+    },
+    {
+      "epoch": 6.057142857142857,
+      "grad_norm": 0.6805425654716813,
+      "learning_rate": 3.268563020091413e-05,
+      "loss": 0.078,
+      "step": 1696
+    },
+    {
+      "epoch": 6.060714285714286,
+      "grad_norm": 0.6391286084495992,
+      "learning_rate": 3.267668561995392e-05,
+      "loss": 0.074,
+      "step": 1697
+    },
+    {
+      "epoch": 6.064285714285714,
+      "grad_norm": 0.5795901134835514,
+      "learning_rate": 3.266773679879773e-05,
+      "loss": 0.0371,
+      "step": 1698
+    },
+    {
+      "epoch": 6.067857142857143,
+      "grad_norm": 0.507290258024719,
+      "learning_rate": 3.2658783740438825e-05,
+      "loss": 0.0518,
+      "step": 1699
+    },
+    {
+      "epoch": 6.071428571428571,
+      "grad_norm": 0.6855927508714245,
+      "learning_rate": 3.264982644787189e-05,
+      "loss": 0.0801,
+      "step": 1700
+    },
+    {
+      "epoch": 6.075,
+      "grad_norm": 0.7723373740458989,
+      "learning_rate": 3.264086492409303e-05,
+      "loss": 0.0483,
+      "step": 1701
+    },
+    {
+      "epoch": 6.078571428571428,
+      "grad_norm": 0.3482610524574822,
+      "learning_rate": 3.2631899172099774e-05,
+      "loss": 0.0556,
+      "step": 1702
+    },
+    {
+      "epoch": 6.082142857142857,
+      "grad_norm": 0.4560718762558169,
+      "learning_rate": 3.2622929194891034e-05,
+      "loss": 0.0653,
+      "step": 1703
+    },
+    {
+      "epoch": 6.085714285714285,
+      "grad_norm": 0.8051553823591943,
+      "learning_rate": 3.261395499546718e-05,
+      "loss": 0.094,
+      "step": 1704
+    },
+    {
+      "epoch": 6.089285714285714,
+      "grad_norm": 0.7633641909299375,
+      "learning_rate": 3.2604976576829964e-05,
+      "loss": 0.1138,
+      "step": 1705
+    },
+    {
+      "epoch": 6.0928571428571425,
+      "grad_norm": 0.7786972291392573,
+      "learning_rate": 3.2595993941982554e-05,
+      "loss": 0.0355,
+      "step": 1706
+    },
+    {
+      "epoch": 6.0964285714285715,
+      "grad_norm": 0.7346310589613672,
+      "learning_rate": 3.2587007093929534e-05,
+      "loss": 0.0652,
+      "step": 1707
+    },
+    {
+      "epoch": 6.1,
+      "grad_norm": 0.4849189227748302,
+      "learning_rate": 3.2578016035676895e-05,
+      "loss": 0.06,
+      "step": 1708
+    },
+    {
+      "epoch": 6.103571428571429,
+      "grad_norm": 0.4166430391909311,
+      "learning_rate": 3.256902077023204e-05,
+      "loss": 0.0412,
+      "step": 1709
+    },
+    {
+      "epoch": 6.107142857142857,
+      "grad_norm": 0.43073287428184065,
+      "learning_rate": 3.256002130060376e-05,
+      "loss": 0.0586,
+      "step": 1710
+    },
+    {
+      "epoch": 6.110714285714286,
+      "grad_norm": 0.8169961517927412,
+      "learning_rate": 3.2551017629802275e-05,
+      "loss": 0.0388,
+      "step": 1711
+    },
+    {
+      "epoch": 6.114285714285714,
+      "grad_norm": 0.8669116475125964,
+      "learning_rate": 3.254200976083921e-05,
+      "loss": 0.0891,
+      "step": 1712
+    },
+    {
+      "epoch": 6.117857142857143,
+      "grad_norm": 1.0383724185346204,
+      "learning_rate": 3.253299769672759e-05,
+      "loss": 0.0629,
+      "step": 1713
+    },
+    {
+      "epoch": 6.121428571428571,
+      "grad_norm": 0.6557242074933367,
+      "learning_rate": 3.252398144048182e-05,
+      "loss": 0.0904,
+      "step": 1714
+    },
+    {
+      "epoch": 6.125,
+      "grad_norm": 0.6638543654350725,
+      "learning_rate": 3.2514960995117746e-05,
+      "loss": 0.0879,
+      "step": 1715
+    },
+    {
+      "epoch": 6.128571428571428,
+      "grad_norm": 1.0137910270132173,
+      "learning_rate": 3.250593636365259e-05,
+      "loss": 0.0626,
+      "step": 1716
+    },
+    {
+      "epoch": 6.132142857142857,
+      "grad_norm": 0.5091053640551411,
+      "learning_rate": 3.2496907549104974e-05,
+      "loss": 0.0688,
+      "step": 1717
+    },
+    {
+      "epoch": 6.135714285714286,
+      "grad_norm": 0.8938630578403035,
+      "learning_rate": 3.248787455449493e-05,
+      "loss": 0.1036,
+      "step": 1718
+    },
+    {
+      "epoch": 6.139285714285714,
+      "grad_norm": 0.579000811005397,
+      "learning_rate": 3.2478837382843896e-05,
+      "loss": 0.0348,
+      "step": 1719
+    },
+    {
+      "epoch": 6.142857142857143,
+      "grad_norm": 0.7977296058695565,
+      "learning_rate": 3.246979603717467e-05,
+      "loss": 0.0979,
+      "step": 1720
+    },
+    {
+      "epoch": 6.146428571428571,
+      "grad_norm": 0.9339507585925242,
+      "learning_rate": 3.24607505205115e-05,
+      "loss": 0.1085,
+      "step": 1721
+    },
+    {
+      "epoch": 6.15,
+      "grad_norm": 0.7387737236546918,
+      "learning_rate": 3.245170083587998e-05,
+      "loss": 0.0688,
+      "step": 1722
+    },
+    {
+      "epoch": 6.1535714285714285,
+      "grad_norm": 0.563892391172653,
+      "learning_rate": 3.244264698630713e-05,
+      "loss": 0.0595,
+      "step": 1723
+    },
+    {
+      "epoch": 6.1571428571428575,
+      "grad_norm": 0.7352843692433322,
+      "learning_rate": 3.243358897482134e-05,
+      "loss": 0.0644,
+      "step": 1724
+    },
+    {
+      "epoch": 6.160714285714286,
+      "grad_norm": 0.671443339301165,
+      "learning_rate": 3.242452680445242e-05,
+      "loss": 0.106,
+      "step": 1725
+    },
+    {
+      "epoch": 6.164285714285715,
+      "grad_norm": 0.702378329176265,
+      "learning_rate": 3.2415460478231535e-05,
+      "loss": 0.0749,
+      "step": 1726
+    },
+    {
+      "epoch": 6.167857142857143,
+      "grad_norm": 0.7791644147526529,
+      "learning_rate": 3.240638999919128e-05,
+      "loss": 0.0982,
+      "step": 1727
+    },
+    {
+      "epoch": 6.171428571428572,
+      "grad_norm": 0.7287959875557102,
+      "learning_rate": 3.23973153703656e-05,
+      "loss": 0.0571,
+      "step": 1728
+    },
+    {
+      "epoch": 6.175,
+      "grad_norm": 0.7832558768271445,
+      "learning_rate": 3.238823659478987e-05,
+      "loss": 0.0902,
+      "step": 1729
+    },
+    {
+      "epoch": 6.178571428571429,
+      "grad_norm": 0.4907204393080552,
+      "learning_rate": 3.237915367550081e-05,
+      "loss": 0.0582,
+      "step": 1730
+    },
+    {
+      "epoch": 6.182142857142857,
+      "grad_norm": 0.6524980354652751,
+      "learning_rate": 3.237006661553655e-05,
+      "loss": 0.1006,
+      "step": 1731
+    },
+    {
+      "epoch": 6.185714285714286,
+      "grad_norm": 0.6344065964904931,
+      "learning_rate": 3.23609754179366e-05,
+      "loss": 0.101,
+      "step": 1732
+    },
+    {
+      "epoch": 6.189285714285714,
+      "grad_norm": 0.4521689691603983,
+      "learning_rate": 3.235188008574186e-05,
+      "loss": 0.0702,
+      "step": 1733
+    },
+    {
+      "epoch": 6.192857142857143,
+      "grad_norm": 0.6340595980461224,
+      "learning_rate": 3.2342780621994606e-05,
+      "loss": 0.0745,
+      "step": 1734
+    },
+    {
+      "epoch": 6.196428571428571,
+      "grad_norm": 0.652054213111625,
+      "learning_rate": 3.233367702973849e-05,
+      "loss": 0.0952,
+      "step": 1735
+    },
+    {
+      "epoch": 6.2,
+      "grad_norm": 0.7740161625805466,
+      "learning_rate": 3.232456931201855e-05,
+      "loss": 0.0745,
+      "step": 1736
+    },
+    {
+      "epoch": 6.203571428571428,
+      "grad_norm": 0.550959659299997,
+      "learning_rate": 3.231545747188122e-05,
+      "loss": 0.055,
+      "step": 1737
+    },
+    {
+      "epoch": 6.207142857142857,
+      "grad_norm": 0.9271382230307901,
+      "learning_rate": 3.230634151237429e-05,
+      "loss": 0.0766,
+      "step": 1738
+    },
+    {
+      "epoch": 6.210714285714285,
+      "grad_norm": 0.5867126125764389,
+      "learning_rate": 3.229722143654693e-05,
+      "loss": 0.111,
+      "step": 1739
+    },
+    {
+      "epoch": 6.214285714285714,
+      "grad_norm": 1.027944881763432,
+      "learning_rate": 3.22880972474497e-05,
+      "loss": 0.0654,
+      "step": 1740
+    },
+    {
+      "epoch": 6.2178571428571425,
+      "grad_norm": 0.7764785130287666,
+      "learning_rate": 3.227896894813454e-05,
+      "loss": 0.0465,
+      "step": 1741
+    },
+    {
+      "epoch": 6.2214285714285715,
+      "grad_norm": 0.7391220191808566,
+      "learning_rate": 3.2269836541654746e-05,
+      "loss": 0.0435,
+      "step": 1742
+    },
+    {
+      "epoch": 6.225,
+      "grad_norm": 0.9403117132859078,
+      "learning_rate": 3.2260700031064995e-05,
+      "loss": 0.1575,
+      "step": 1743
+    },
+    {
+      "epoch": 6.228571428571429,
+      "grad_norm": 0.8881009299590196,
+      "learning_rate": 3.2251559419421336e-05,
+      "loss": 0.0814,
+      "step": 1744
+    },
+    {
+      "epoch": 6.232142857142857,
+      "grad_norm": 0.41922649211300117,
+      "learning_rate": 3.2242414709781185e-05,
+      "loss": 0.0483,
+      "step": 1745
+    },
+    {
+      "epoch": 6.235714285714286,
+      "grad_norm": 0.6564597898510307,
+      "learning_rate": 3.2233265905203354e-05,
+      "loss": 0.0658,
+      "step": 1746
+    },
+    {
+      "epoch": 6.239285714285714,
+      "grad_norm": 0.7500291310879718,
+      "learning_rate": 3.222411300874799e-05,
+      "loss": 0.0928,
+      "step": 1747
+    },
+    {
+      "epoch": 6.242857142857143,
+      "grad_norm": 0.7852920920983244,
+      "learning_rate": 3.221495602347663e-05,
+      "loss": 0.1143,
+      "step": 1748
+    },
+    {
+      "epoch": 6.246428571428572,
+      "grad_norm": 0.9026777381800908,
+      "learning_rate": 3.220579495245218e-05,
+      "loss": 0.0673,
+      "step": 1749
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 0.6707872093095024,
+      "learning_rate": 3.219662979873889e-05,
+      "loss": 0.0669,
+      "step": 1750
+    },
+    {
+      "epoch": 6.253571428571428,
+      "grad_norm": 0.8784631387649141,
+      "learning_rate": 3.2187460565402414e-05,
+      "loss": 0.0648,
+      "step": 1751
+    },
+    {
+      "epoch": 6.257142857142857,
+      "grad_norm": 0.554525260133819,
+      "learning_rate": 3.2178287255509726e-05,
+      "loss": 0.0848,
+      "step": 1752
+    },
+    {
+      "epoch": 6.260714285714286,
+      "grad_norm": 1.2231144065963857,
+      "learning_rate": 3.21691098721292e-05,
+      "loss": 0.078,
+      "step": 1753
+    },
+    {
+      "epoch": 6.264285714285714,
+      "grad_norm": 0.7599892518917266,
+      "learning_rate": 3.215992841833055e-05,
+      "loss": 0.1006,
+      "step": 1754
+    },
+    {
+      "epoch": 6.267857142857143,
+      "grad_norm": 0.9804169862443344,
+      "learning_rate": 3.2150742897184865e-05,
+      "loss": 0.064,
+      "step": 1755
+    },
+    {
+      "epoch": 6.271428571428571,
+      "grad_norm": 0.8027932583862647,
+      "learning_rate": 3.2141553311764587e-05,
+      "loss": 0.0802,
+      "step": 1756
+    },
+    {
+      "epoch": 6.275,
+      "grad_norm": 0.42545242552345974,
+      "learning_rate": 3.2132359665143524e-05,
+      "loss": 0.0601,
+      "step": 1757
+    },
+    {
+      "epoch": 6.2785714285714285,
+      "grad_norm": 0.3487689551914776,
+      "learning_rate": 3.2123161960396826e-05,
+      "loss": 0.0242,
+      "step": 1758
+    },
+    {
+      "epoch": 6.2821428571428575,
+      "grad_norm": 1.2627449010234035,
+      "learning_rate": 3.2113960200601026e-05,
+      "loss": 0.1164,
+      "step": 1759
+    },
+    {
+      "epoch": 6.285714285714286,
+      "grad_norm": 0.5633055226801511,
+      "learning_rate": 3.2104754388834e-05,
+      "loss": 0.0623,
+      "step": 1760
+    },
+    {
+      "epoch": 6.289285714285715,
+      "grad_norm": 0.8278283160110987,
+      "learning_rate": 3.209554452817497e-05,
+      "loss": 0.1048,
+      "step": 1761
+    },
+    {
+      "epoch": 6.292857142857143,
+      "grad_norm": 0.37364166327285836,
+      "learning_rate": 3.208633062170452e-05,
+      "loss": 0.0381,
+      "step": 1762
+    },
+    {
+      "epoch": 6.296428571428572,
+      "grad_norm": 0.7705676468188305,
+      "learning_rate": 3.20771126725046e-05,
+      "loss": 0.1262,
+      "step": 1763
+    },
+    {
+      "epoch": 6.3,
+      "grad_norm": 0.639339095969274,
+      "learning_rate": 3.20678906836585e-05,
+      "loss": 0.0623,
+      "step": 1764
+    },
+    {
+      "epoch": 6.303571428571429,
+      "grad_norm": 0.6426748183802145,
+      "learning_rate": 3.205866465825085e-05,
+      "loss": 0.1093,
+      "step": 1765
+    },
+    {
+      "epoch": 6.307142857142857,
+      "grad_norm": 1.2149930337410575,
+      "learning_rate": 3.204943459936765e-05,
+      "loss": 0.1153,
+      "step": 1766
+    },
+    {
+      "epoch": 6.310714285714286,
+      "grad_norm": 0.5171934869379107,
+      "learning_rate": 3.204020051009624e-05,
+      "loss": 0.0527,
+      "step": 1767
+    },
+    {
+      "epoch": 6.314285714285714,
+      "grad_norm": 0.8381955144636298,
+      "learning_rate": 3.2030962393525315e-05,
+      "loss": 0.0944,
+      "step": 1768
+    },
+    {
+      "epoch": 6.317857142857143,
+      "grad_norm": 0.5621520428018479,
+      "learning_rate": 3.2021720252744914e-05,
+      "loss": 0.0704,
+      "step": 1769
+    },
+    {
+      "epoch": 6.321428571428571,
+      "grad_norm": 0.44231571567351335,
+      "learning_rate": 3.2012474090846406e-05,
+      "loss": 0.0542,
+      "step": 1770
+    },
+    {
+      "epoch": 6.325,
+      "grad_norm": 0.6009157226602944,
+      "learning_rate": 3.200322391092253e-05,
+      "loss": 0.0485,
+      "step": 1771
+    },
+    {
+      "epoch": 6.328571428571428,
+      "grad_norm": 0.36701298015625733,
+      "learning_rate": 3.199396971606735e-05,
+      "loss": 0.0362,
+      "step": 1772
+    },
+    {
+      "epoch": 6.332142857142857,
+      "grad_norm": 0.6809716049186376,
+      "learning_rate": 3.198471150937629e-05,
+      "loss": 0.0835,
+      "step": 1773
+    },
+    {
+      "epoch": 6.335714285714285,
+      "grad_norm": 0.4696546938906532,
+      "learning_rate": 3.19754492939461e-05,
+      "loss": 0.059,
+      "step": 1774
+    },
+    {
+      "epoch": 6.339285714285714,
+      "grad_norm": 0.7866206044811809,
+      "learning_rate": 3.196618307287488e-05,
+      "loss": 0.0959,
+      "step": 1775
+    },
+    {
+      "epoch": 6.3428571428571425,
+      "grad_norm": 0.7799511461218107,
+      "learning_rate": 3.1956912849262065e-05,
+      "loss": 0.0475,
+      "step": 1776
+    },
+    {
+      "epoch": 6.3464285714285715,
+      "grad_norm": 1.020835441006009,
+      "learning_rate": 3.194763862620844e-05,
+      "loss": 0.1319,
+      "step": 1777
+    },
+    {
+      "epoch": 6.35,
+      "grad_norm": 0.4070208377047479,
+      "learning_rate": 3.1938360406816104e-05,
+      "loss": 0.0491,
+      "step": 1778
+    },
+    {
+      "epoch": 6.353571428571429,
+      "grad_norm": 0.5621775207583817,
+      "learning_rate": 3.192907819418852e-05,
+      "loss": 0.0541,
+      "step": 1779
+    },
+    {
+      "epoch": 6.357142857142857,
+      "grad_norm": 0.5107769038496517,
+      "learning_rate": 3.191979199143047e-05,
+      "loss": 0.0367,
+      "step": 1780
+    },
+    {
+      "epoch": 6.360714285714286,
+      "grad_norm": 0.5371986219103853,
+      "learning_rate": 3.1910501801648077e-05,
+      "loss": 0.0528,
+      "step": 1781
+    },
+    {
+      "epoch": 6.364285714285714,
+      "grad_norm": 0.5924796021015644,
+      "learning_rate": 3.1901207627948795e-05,
+      "loss": 0.0776,
+      "step": 1782
+    },
+    {
+      "epoch": 6.367857142857143,
+      "grad_norm": 0.69187362623501,
+      "learning_rate": 3.1891909473441405e-05,
+      "loss": 0.0883,
+      "step": 1783
+    },
+    {
+      "epoch": 6.371428571428572,
+      "grad_norm": 0.6821549027876379,
+      "learning_rate": 3.1882607341236036e-05,
+      "loss": 0.0654,
+      "step": 1784
+    },
+    {
+      "epoch": 6.375,
+      "grad_norm": 0.7907509147083367,
+      "learning_rate": 3.1873301234444125e-05,
+      "loss": 0.0681,
+      "step": 1785
+    },
+    {
+      "epoch": 6.378571428571428,
+      "grad_norm": 0.9201265526452135,
+      "learning_rate": 3.186399115617846e-05,
+      "loss": 0.0574,
+      "step": 1786
+    },
+    {
+      "epoch": 6.382142857142857,
+      "grad_norm": 0.8943350993374393,
+      "learning_rate": 3.1854677109553157e-05,
+      "loss": 0.031,
+      "step": 1787
+    },
+    {
+      "epoch": 6.385714285714286,
+      "grad_norm": 0.8412921069367935,
+      "learning_rate": 3.184535909768363e-05,
+      "loss": 0.0402,
+      "step": 1788
+    },
+    {
+      "epoch": 6.389285714285714,
+      "grad_norm": 0.8718338961086827,
+      "learning_rate": 3.1836037123686644e-05,
+      "loss": 0.0973,
+      "step": 1789
+    },
+    {
+      "epoch": 6.392857142857143,
+      "grad_norm": 0.7354739578531267,
+      "learning_rate": 3.18267111906803e-05,
+      "loss": 0.0435,
+      "step": 1790
+    },
+    {
+      "epoch": 6.396428571428571,
+      "grad_norm": 0.508807578100025,
+      "learning_rate": 3.181738130178399e-05,
+      "loss": 0.0391,
+      "step": 1791
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.4847624731936382,
+      "learning_rate": 3.1808047460118454e-05,
+      "loss": 0.0594,
+      "step": 1792
+    },
+    {
+      "epoch": 6.4035714285714285,
+      "grad_norm": 0.8981011572254152,
+      "learning_rate": 3.179870966880576e-05,
+      "loss": 0.0939,
+      "step": 1793
+    },
+    {
+      "epoch": 6.4071428571428575,
+      "grad_norm": 0.5651847556903731,
+      "learning_rate": 3.1789367930969264e-05,
+      "loss": 0.0742,
+      "step": 1794
+    },
+    {
+      "epoch": 6.410714285714286,
+      "grad_norm": 0.787170779772436,
+      "learning_rate": 3.178002224973368e-05,
+      "loss": 0.0911,
+      "step": 1795
+    },
+    {
+      "epoch": 6.414285714285715,
+      "grad_norm": 1.1246559746331422,
+      "learning_rate": 3.177067262822502e-05,
+      "loss": 0.1198,
+      "step": 1796
+    },
+    {
+      "epoch": 6.417857142857143,
+      "grad_norm": 0.6708557381326734,
+      "learning_rate": 3.1761319069570606e-05,
+      "loss": 0.0606,
+      "step": 1797
+    },
+    {
+      "epoch": 6.421428571428572,
+      "grad_norm": 1.302563051026848,
+      "learning_rate": 3.17519615768991e-05,
+      "loss": 0.0757,
+      "step": 1798
+    },
+    {
+      "epoch": 6.425,
+      "grad_norm": 0.5933381679172223,
+      "learning_rate": 3.174260015334047e-05,
+      "loss": 0.0925,
+      "step": 1799
+    },
+    {
+      "epoch": 6.428571428571429,
+      "grad_norm": 0.7610612656974565,
+      "learning_rate": 3.173323480202599e-05,
+      "loss": 0.0718,
+      "step": 1800
+    },
+    {
+      "epoch": 6.432142857142857,
+      "grad_norm": 0.548465114741662,
+      "learning_rate": 3.172386552608826e-05,
+      "loss": 0.0831,
+      "step": 1801
+    },
+    {
+      "epoch": 6.435714285714286,
+      "grad_norm": 0.970781259931781,
+      "learning_rate": 3.1714492328661185e-05,
+      "loss": 0.0712,
+      "step": 1802
+    },
+    {
+      "epoch": 6.439285714285714,
+      "grad_norm": 0.8091768751341839,
+      "learning_rate": 3.170511521287999e-05,
+      "loss": 0.0762,
+      "step": 1803
+    },
+    {
+      "epoch": 6.442857142857143,
+      "grad_norm": 0.48869794919598675,
+      "learning_rate": 3.1695734181881205e-05,
+      "loss": 0.0743,
+      "step": 1804
+    },
+    {
+      "epoch": 6.446428571428571,
+      "grad_norm": 0.5893466830240713,
+      "learning_rate": 3.168634923880266e-05,
+      "loss": 0.1005,
+      "step": 1805
+    },
+    {
+      "epoch": 6.45,
+      "grad_norm": 0.5321421371303581,
+      "learning_rate": 3.1676960386783507e-05,
+      "loss": 0.0502,
+      "step": 1806
+    },
+    {
+      "epoch": 6.453571428571428,
+      "grad_norm": 0.42199613029258315,
+      "learning_rate": 3.166756762896421e-05,
+      "loss": 0.0181,
+      "step": 1807
+    },
+    {
+      "epoch": 6.457142857142857,
+      "grad_norm": 0.4588604224041588,
+      "learning_rate": 3.165817096848653e-05,
+      "loss": 0.0437,
+      "step": 1808
+    },
+    {
+      "epoch": 6.460714285714285,
+      "grad_norm": 0.6131488225602503,
+      "learning_rate": 3.164877040849353e-05,
+      "loss": 0.0906,
+      "step": 1809
+    },
+    {
+      "epoch": 6.464285714285714,
+      "grad_norm": 0.6071336096604741,
+      "learning_rate": 3.163936595212958e-05,
+      "loss": 0.0392,
+      "step": 1810
+    },
+    {
+      "epoch": 6.4678571428571425,
+      "grad_norm": 0.5367199889455316,
+      "learning_rate": 3.1629957602540366e-05,
+      "loss": 0.0579,
+      "step": 1811
+    },
+    {
+      "epoch": 6.4714285714285715,
+      "grad_norm": 0.6079354579191076,
+      "learning_rate": 3.162054536287285e-05,
+      "loss": 0.0513,
+      "step": 1812
+    },
+    {
+      "epoch": 6.475,
+      "grad_norm": 0.9562425732592041,
+      "learning_rate": 3.161112923627532e-05,
+      "loss": 0.1278,
+      "step": 1813
+    },
+    {
+      "epoch": 6.478571428571429,
+      "grad_norm": 0.5887384564821594,
+      "learning_rate": 3.1601709225897355e-05,
+      "loss": 0.0365,
+      "step": 1814
+    },
+    {
+      "epoch": 6.482142857142857,
+      "grad_norm": 0.834734920290735,
+      "learning_rate": 3.1592285334889835e-05,
+      "loss": 0.0974,
+      "step": 1815
+    },
+    {
+      "epoch": 6.485714285714286,
+      "grad_norm": 0.5813684481389928,
+      "learning_rate": 3.158285756640493e-05,
+      "loss": 0.0841,
+      "step": 1816
+    },
+    {
+      "epoch": 6.489285714285714,
+      "grad_norm": 0.8168254263786457,
+      "learning_rate": 3.157342592359612e-05,
+      "loss": 0.0763,
+      "step": 1817
+    },
+    {
+      "epoch": 6.492857142857143,
+      "grad_norm": 0.6204735301141291,
+      "learning_rate": 3.156399040961817e-05,
+      "loss": 0.0727,
+      "step": 1818
+    },
+    {
+      "epoch": 6.496428571428572,
+      "grad_norm": 0.4761372253710653,
+      "learning_rate": 3.155455102762714e-05,
+      "loss": 0.0533,
+      "step": 1819
+    },
+    {
+      "epoch": 6.5,
+      "grad_norm": 1.0287083504310948,
+      "learning_rate": 3.154510778078039e-05,
+      "loss": 0.1174,
+      "step": 1820
+    },
+    {
+      "epoch": 6.503571428571428,
+      "grad_norm": 1.8602571758377155,
+      "learning_rate": 3.153566067223658e-05,
+      "loss": 0.1508,
+      "step": 1821
+    },
+    {
+      "epoch": 6.507142857142857,
+      "grad_norm": 0.5223910727313513,
+      "learning_rate": 3.152620970515564e-05,
+      "loss": 0.0571,
+      "step": 1822
+    },
+    {
+      "epoch": 6.510714285714286,
+      "grad_norm": 0.8449308106080102,
+      "learning_rate": 3.1516754882698805e-05,
+      "loss": 0.0579,
+      "step": 1823
+    },
+    {
+      "epoch": 6.514285714285714,
+      "grad_norm": 0.667857419854795,
+      "learning_rate": 3.15072962080286e-05,
+      "loss": 0.0969,
+      "step": 1824
+    },
+    {
+      "epoch": 6.517857142857143,
+      "grad_norm": 0.939940401609305,
+      "learning_rate": 3.149783368430884e-05,
+      "loss": 0.0404,
+      "step": 1825
+    },
+    {
+      "epoch": 6.521428571428571,
+      "grad_norm": 0.31991431986377455,
+      "learning_rate": 3.148836731470461e-05,
+      "loss": 0.0362,
+      "step": 1826
+    },
+    {
+      "epoch": 6.525,
+      "grad_norm": 0.5998739840736109,
+      "learning_rate": 3.14788971023823e-05,
+      "loss": 0.0415,
+      "step": 1827
+    },
+    {
+      "epoch": 6.5285714285714285,
+      "grad_norm": 1.1869917660194111,
+      "learning_rate": 3.146942305050959e-05,
+      "loss": 0.0938,
+      "step": 1828
+    },
+    {
+      "epoch": 6.5321428571428575,
+      "grad_norm": 1.1797235194546996,
+      "learning_rate": 3.1459945162255426e-05,
+      "loss": 0.1156,
+      "step": 1829
+    },
+    {
+      "epoch": 6.535714285714286,
+      "grad_norm": 0.7193573510849244,
+      "learning_rate": 3.145046344079004e-05,
+      "loss": 0.0942,
+      "step": 1830
+    },
+    {
+      "epoch": 6.539285714285715,
+      "grad_norm": 0.5748418626246747,
+      "learning_rate": 3.1440977889284966e-05,
+      "loss": 0.076,
+      "step": 1831
+    },
+    {
+      "epoch": 6.542857142857143,
+      "grad_norm": 0.6742591616946236,
+      "learning_rate": 3.1431488510912994e-05,
+      "loss": 0.0765,
+      "step": 1832
+    },
+    {
+      "epoch": 6.546428571428572,
+      "grad_norm": 0.7310264228132335,
+      "learning_rate": 3.142199530884821e-05,
+      "loss": 0.069,
+      "step": 1833
+    },
+    {
+      "epoch": 6.55,
+      "grad_norm": 0.5714556303466054,
+      "learning_rate": 3.1412498286265964e-05,
+      "loss": 0.0935,
+      "step": 1834
+    },
+    {
+      "epoch": 6.553571428571429,
+      "grad_norm": 0.300078550825974,
+      "learning_rate": 3.140299744634291e-05,
+      "loss": 0.0264,
+      "step": 1835
+    },
+    {
+      "epoch": 6.557142857142857,
+      "grad_norm": 0.9274500330185059,
+      "learning_rate": 3.139349279225695e-05,
+      "loss": 0.0714,
+      "step": 1836
+    },
+    {
+      "epoch": 6.560714285714286,
+      "grad_norm": 0.8370431590553782,
+      "learning_rate": 3.138398432718727e-05,
+      "loss": 0.0821,
+      "step": 1837
+    },
+    {
+      "epoch": 6.564285714285714,
+      "grad_norm": 0.9512462884881889,
+      "learning_rate": 3.137447205431435e-05,
+      "loss": 0.1012,
+      "step": 1838
+    },
+    {
+      "epoch": 6.567857142857143,
+      "grad_norm": 1.0106597584592714,
+      "learning_rate": 3.1364955976819915e-05,
+      "loss": 0.1039,
+      "step": 1839
+    },
+    {
+      "epoch": 6.571428571428571,
+      "grad_norm": 0.5070647095934988,
+      "learning_rate": 3.135543609788699e-05,
+      "loss": 0.0751,
+      "step": 1840
+    },
+    {
+      "epoch": 6.575,
+      "grad_norm": 0.8536267575983327,
+      "learning_rate": 3.1345912420699845e-05,
+      "loss": 0.0853,
+      "step": 1841
+    },
+    {
+      "epoch": 6.578571428571428,
+      "grad_norm": 0.8771556078736245,
+      "learning_rate": 3.1336384948444045e-05,
+      "loss": 0.0833,
+      "step": 1842
+    },
+    {
+      "epoch": 6.582142857142857,
+      "grad_norm": 0.6564845934830128,
+      "learning_rate": 3.13268536843064e-05,
+      "loss": 0.0653,
+      "step": 1843
+    },
+    {
+      "epoch": 6.585714285714285,
+      "grad_norm": 0.7274702529342185,
+      "learning_rate": 3.131731863147501e-05,
+      "loss": 0.0489,
+      "step": 1844
+    },
+    {
+      "epoch": 6.589285714285714,
+      "grad_norm": 0.38206331906415836,
+      "learning_rate": 3.130777979313924e-05,
+      "loss": 0.0647,
+      "step": 1845
+    },
+    {
+      "epoch": 6.5928571428571425,
+      "grad_norm": 0.2799145680792345,
+      "learning_rate": 3.129823717248969e-05,
+      "loss": 0.0379,
+      "step": 1846
+    },
+    {
+      "epoch": 6.5964285714285715,
+      "grad_norm": 0.40582788874802334,
+      "learning_rate": 3.128869077271827e-05,
+      "loss": 0.0516,
+      "step": 1847
+    },
+    {
+      "epoch": 6.6,
+      "grad_norm": 0.5615162533969114,
+      "learning_rate": 3.1279140597018135e-05,
+      "loss": 0.0397,
+      "step": 1848
+    },
+    {
+      "epoch": 6.603571428571429,
+      "grad_norm": 0.6207548409762152,
+      "learning_rate": 3.126958664858369e-05,
+      "loss": 0.0594,
+      "step": 1849
+    },
+    {
+      "epoch": 6.607142857142857,
+      "grad_norm": 0.8270207130759678,
+      "learning_rate": 3.1260028930610615e-05,
+      "loss": 0.0762,
+      "step": 1850
+    },
+    {
+      "epoch": 6.610714285714286,
+      "grad_norm": 0.5020272305998789,
+      "learning_rate": 3.1250467446295864e-05,
+      "loss": 0.0363,
+      "step": 1851
+    },
+    {
+      "epoch": 6.614285714285714,
+      "grad_norm": 1.0393298410924838,
+      "learning_rate": 3.1240902198837616e-05,
+      "loss": 0.0806,
+      "step": 1852
+    },
+    {
+      "epoch": 6.617857142857143,
+      "grad_norm": 0.5113627338807503,
+      "learning_rate": 3.123133319143535e-05,
+      "loss": 0.0622,
+      "step": 1853
+    },
+    {
+      "epoch": 6.621428571428572,
+      "grad_norm": 0.25001902209287924,
+      "learning_rate": 3.1221760427289765e-05,
+      "loss": 0.0164,
+      "step": 1854
+    },
+    {
+      "epoch": 6.625,
+      "grad_norm": 0.7989893994355101,
+      "learning_rate": 3.121218390960284e-05,
+      "loss": 0.112,
+      "step": 1855
+    },
+    {
+      "epoch": 6.628571428571428,
+      "grad_norm": 0.4379250646995418,
+      "learning_rate": 3.12026036415778e-05,
+      "loss": 0.0486,
+      "step": 1856
+    },
+    {
+      "epoch": 6.632142857142857,
+      "grad_norm": 0.6208732069398152,
+      "learning_rate": 3.119301962641913e-05,
+      "loss": 0.0309,
+      "step": 1857
+    },
+    {
+      "epoch": 6.635714285714286,
+      "grad_norm": 0.5472727443628951,
+      "learning_rate": 3.118343186733257e-05,
+      "loss": 0.0502,
+      "step": 1858
+    },
+    {
+      "epoch": 6.639285714285714,
+      "grad_norm": 0.923480779287465,
+      "learning_rate": 3.1173840367525094e-05,
+      "loss": 0.0867,
+      "step": 1859
+    },
+    {
+      "epoch": 6.642857142857143,
+      "grad_norm": 0.442186832997458,
+      "learning_rate": 3.116424513020495e-05,
+      "loss": 0.0702,
+      "step": 1860
+    },
+    {
+      "epoch": 6.646428571428571,
+      "grad_norm": 0.3694346373016995,
+      "learning_rate": 3.115464615858163e-05,
+      "loss": 0.0587,
+      "step": 1861
+    },
+    {
+      "epoch": 6.65,
+      "grad_norm": 0.839864055705499,
+      "learning_rate": 3.114504345586587e-05,
+      "loss": 0.0728,
+      "step": 1862
+    },
+    {
+      "epoch": 6.6535714285714285,
+      "grad_norm": 0.4143208378955332,
+      "learning_rate": 3.1135437025269664e-05,
+      "loss": 0.0692,
+      "step": 1863
+    },
+    {
+      "epoch": 6.6571428571428575,
+      "grad_norm": 0.38521575984398393,
+      "learning_rate": 3.1125826870006226e-05,
+      "loss": 0.0647,
+      "step": 1864
+    },
+    {
+      "epoch": 6.660714285714286,
+      "grad_norm": 0.5474502265548524,
+      "learning_rate": 3.111621299329005e-05,
+      "loss": 0.0507,
+      "step": 1865
+    },
+    {
+      "epoch": 6.664285714285715,
+      "grad_norm": 0.4305664989190576,
+      "learning_rate": 3.110659539833686e-05,
+      "loss": 0.0752,
+      "step": 1866
+    },
+    {
+      "epoch": 6.667857142857143,
+      "grad_norm": 0.6015544280642342,
+      "learning_rate": 3.1096974088363615e-05,
+      "loss": 0.0639,
+      "step": 1867
+    },
+    {
+      "epoch": 6.671428571428572,
+      "grad_norm": 0.9411658523537092,
+      "learning_rate": 3.108734906658852e-05,
+      "loss": 0.1165,
+      "step": 1868
+    },
+    {
+      "epoch": 6.675,
+      "grad_norm": 0.5514036851930807,
+      "learning_rate": 3.1077720336231054e-05,
+      "loss": 0.051,
+      "step": 1869
+    },
+    {
+      "epoch": 6.678571428571429,
+      "grad_norm": 1.470182360542381,
+      "learning_rate": 3.106808790051189e-05,
+      "loss": 0.146,
+      "step": 1870
+    },
+    {
+      "epoch": 6.682142857142857,
+      "grad_norm": 0.6370322457403715,
+      "learning_rate": 3.1058451762652955e-05,
+      "loss": 0.035,
+      "step": 1871
+    },
+    {
+      "epoch": 6.685714285714286,
+      "grad_norm": 0.6709179645477086,
+      "learning_rate": 3.1048811925877434e-05,
+      "loss": 0.0736,
+      "step": 1872
+    },
+    {
+      "epoch": 6.689285714285714,
+      "grad_norm": 0.4277570134833592,
+      "learning_rate": 3.103916839340972e-05,
+      "loss": 0.0551,
+      "step": 1873
+    },
+    {
+      "epoch": 6.692857142857143,
+      "grad_norm": 0.8207095010040879,
+      "learning_rate": 3.1029521168475466e-05,
+      "loss": 0.098,
+      "step": 1874
+    },
+    {
+      "epoch": 6.696428571428571,
+      "grad_norm": 0.5804348397305463,
+      "learning_rate": 3.1019870254301546e-05,
+      "loss": 0.0577,
+      "step": 1875
+    },
+    {
+      "epoch": 6.7,
+      "grad_norm": 1.1580120780152576,
+      "learning_rate": 3.1010215654116075e-05,
+      "loss": 0.0955,
+      "step": 1876
+    },
+    {
+      "epoch": 6.703571428571428,
+      "grad_norm": 1.0186412296533645,
+      "learning_rate": 3.1000557371148396e-05,
+      "loss": 0.0924,
+      "step": 1877
+    },
+    {
+      "epoch": 6.707142857142857,
+      "grad_norm": 1.1159326751130847,
+      "learning_rate": 3.0990895408629085e-05,
+      "loss": 0.1016,
+      "step": 1878
+    },
+    {
+      "epoch": 6.710714285714285,
+      "grad_norm": 0.6445657878665647,
+      "learning_rate": 3.098122976978996e-05,
+      "loss": 0.0451,
+      "step": 1879
+    },
+    {
+      "epoch": 6.714285714285714,
+      "grad_norm": 0.4837637496593846,
+      "learning_rate": 3.097156045786404e-05,
+      "loss": 0.0242,
+      "step": 1880
+    },
+    {
+      "epoch": 6.7178571428571425,
+      "grad_norm": 0.887840848817431,
+      "learning_rate": 3.096188747608562e-05,
+      "loss": 0.0644,
+      "step": 1881
+    },
+    {
+      "epoch": 6.7214285714285715,
+      "grad_norm": 0.8765363187090507,
+      "learning_rate": 3.095221082769016e-05,
+      "loss": 0.0748,
+      "step": 1882
+    },
+    {
+      "epoch": 6.725,
+      "grad_norm": 1.459434573549939,
+      "learning_rate": 3.0942530515914405e-05,
+      "loss": 0.1311,
+      "step": 1883
+    },
+    {
+      "epoch": 6.728571428571429,
+      "grad_norm": 0.8355477446261595,
+      "learning_rate": 3.093284654399629e-05,
+      "loss": 0.0504,
+      "step": 1884
+    },
+    {
+      "epoch": 6.732142857142857,
+      "grad_norm": 0.46805514965001166,
+      "learning_rate": 3.092315891517499e-05,
+      "loss": 0.0309,
+      "step": 1885
+    },
+    {
+      "epoch": 6.735714285714286,
+      "grad_norm": 1.2388563004509545,
+      "learning_rate": 3.0913467632690885e-05,
+      "loss": 0.0629,
+      "step": 1886
+    },
+    {
+      "epoch": 6.739285714285714,
+      "grad_norm": 0.4750961133436672,
+      "learning_rate": 3.090377269978562e-05,
+      "loss": 0.0378,
+      "step": 1887
+    },
+    {
+      "epoch": 6.742857142857143,
+      "grad_norm": 1.4144056251145112,
+      "learning_rate": 3.0894074119702e-05,
+      "loss": 0.1071,
+      "step": 1888
+    },
+    {
+      "epoch": 6.746428571428572,
+      "grad_norm": 0.65534342240714,
+      "learning_rate": 3.088437189568409e-05,
+      "loss": 0.0569,
+      "step": 1889
+    },
+    {
+      "epoch": 6.75,
+      "grad_norm": 1.0282993402814964,
+      "learning_rate": 3.087466603097717e-05,
+      "loss": 0.1051,
+      "step": 1890
+    },
+    {
+      "epoch": 6.753571428571428,
+      "grad_norm": 0.4693991766548451,
+      "learning_rate": 3.086495652882774e-05,
+      "loss": 0.0264,
+      "step": 1891
+    },
+    {
+      "epoch": 6.757142857142857,
+      "grad_norm": 0.990731507529586,
+      "learning_rate": 3.085524339248349e-05,
+      "loss": 0.0524,
+      "step": 1892
+    },
+    {
+      "epoch": 6.760714285714286,
+      "grad_norm": 0.9066260120485604,
+      "learning_rate": 3.084552662519336e-05,
+      "loss": 0.0457,
+      "step": 1893
+    },
+    {
+      "epoch": 6.764285714285714,
+      "grad_norm": 0.6427645388887454,
+      "learning_rate": 3.083580623020749e-05,
+      "loss": 0.0719,
+      "step": 1894
+    },
+    {
+      "epoch": 6.767857142857143,
+      "grad_norm": 0.6577654329613617,
+      "learning_rate": 3.082608221077722e-05,
+      "loss": 0.0654,
+      "step": 1895
+    },
+    {
+      "epoch": 6.771428571428571,
+      "grad_norm": 0.7839108987881716,
+      "learning_rate": 3.081635457015513e-05,
+      "loss": 0.0892,
+      "step": 1896
+    },
+    {
+      "epoch": 6.775,
+      "grad_norm": 0.7777616538386505,
+      "learning_rate": 3.080662331159499e-05,
+      "loss": 0.055,
+      "step": 1897
+    },
+    {
+      "epoch": 6.7785714285714285,
+      "grad_norm": 0.6446199836483303,
+      "learning_rate": 3.079688843835178e-05,
+      "loss": 0.08,
+      "step": 1898
+    },
+    {
+      "epoch": 6.7821428571428575,
+      "grad_norm": 0.6205806028447809,
+      "learning_rate": 3.07871499536817e-05,
+      "loss": 0.1264,
+      "step": 1899
+    },
+    {
+      "epoch": 6.785714285714286,
+      "grad_norm": 0.37418371677544005,
+      "learning_rate": 3.077740786084217e-05,
+      "loss": 0.0601,
+      "step": 1900
+    },
+    {
+      "epoch": 6.789285714285715,
+      "grad_norm": 1.0133322070755664,
+      "learning_rate": 3.076766216309178e-05,
+      "loss": 0.095,
+      "step": 1901
+    },
+    {
+      "epoch": 6.792857142857143,
+      "grad_norm": 0.4693300171041947,
+      "learning_rate": 3.075791286369035e-05,
+      "loss": 0.061,
+      "step": 1902
+    },
+    {
+      "epoch": 6.796428571428572,
+      "grad_norm": 0.4168395163138502,
+      "learning_rate": 3.0748159965898916e-05,
+      "loss": 0.0687,
+      "step": 1903
+    },
+    {
+      "epoch": 6.8,
+      "grad_norm": 0.5524099164009653,
+      "learning_rate": 3.073840347297968e-05,
+      "loss": 0.0863,
+      "step": 1904
+    },
+    {
+      "epoch": 6.803571428571429,
+      "grad_norm": 0.25473401623491043,
+      "learning_rate": 3.072864338819609e-05,
+      "loss": 0.0308,
+      "step": 1905
+    },
+    {
+      "epoch": 6.807142857142857,
+      "grad_norm": 1.0498786148597374,
+      "learning_rate": 3.0718879714812755e-05,
+      "loss": 0.0859,
+      "step": 1906
+    },
+    {
+      "epoch": 6.810714285714286,
+      "grad_norm": 0.6880083269327826,
+      "learning_rate": 3.070911245609552e-05,
+      "loss": 0.0845,
+      "step": 1907
+    },
+    {
+      "epoch": 6.814285714285714,
+      "grad_norm": 0.48900313310674987,
+      "learning_rate": 3.069934161531141e-05,
+      "loss": 0.0722,
+      "step": 1908
+    },
+    {
+      "epoch": 6.817857142857143,
+      "grad_norm": 0.24949252008140105,
+      "learning_rate": 3.068956719572864e-05,
+      "loss": 0.0389,
+      "step": 1909
+    },
+    {
+      "epoch": 6.821428571428571,
+      "grad_norm": 0.2569355378752272,
+      "learning_rate": 3.067978920061666e-05,
+      "loss": 0.0312,
+      "step": 1910
+    },
+    {
+      "epoch": 6.825,
+      "grad_norm": 0.4653297081953886,
+      "learning_rate": 3.0670007633246055e-05,
+      "loss": 0.0674,
+      "step": 1911
+    },
+    {
+      "epoch": 6.828571428571428,
+      "grad_norm": 0.6154588922147478,
+      "learning_rate": 3.0660222496888665e-05,
+      "loss": 0.0465,
+      "step": 1912
+    },
+    {
+      "epoch": 6.832142857142857,
+      "grad_norm": 0.4000328344713736,
+      "learning_rate": 3.06504337948175e-05,
+      "loss": 0.0349,
+      "step": 1913
+    },
+    {
+      "epoch": 6.835714285714285,
+      "grad_norm": 0.611707798873997,
+      "learning_rate": 3.064064153030673e-05,
+      "loss": 0.063,
+      "step": 1914
+    },
+    {
+      "epoch": 6.839285714285714,
+      "grad_norm": 0.7545289563730623,
+      "learning_rate": 3.063084570663178e-05,
+      "loss": 0.0886,
+      "step": 1915
+    },
+    {
+      "epoch": 6.8428571428571425,
+      "grad_norm": 0.7914498385045253,
+      "learning_rate": 3.0621046327069226e-05,
+      "loss": 0.065,
+      "step": 1916
+    },
+    {
+      "epoch": 6.8464285714285715,
+      "grad_norm": 0.43218162575018637,
+      "learning_rate": 3.061124339489683e-05,
+      "loss": 0.0756,
+      "step": 1917
+    },
+    {
+      "epoch": 6.85,
+      "grad_norm": 1.0195392558029697,
+      "learning_rate": 3.060143691339356e-05,
+      "loss": 0.0989,
+      "step": 1918
+    },
+    {
+      "epoch": 6.853571428571429,
+      "grad_norm": 0.7060030331569246,
+      "learning_rate": 3.059162688583956e-05,
+      "loss": 0.0722,
+      "step": 1919
+    },
+    {
+      "epoch": 6.857142857142857,
+      "grad_norm": 0.4153401958316504,
+      "learning_rate": 3.058181331551617e-05,
+      "loss": 0.0619,
+      "step": 1920
+    },
+    {
+      "epoch": 6.860714285714286,
+      "grad_norm": 0.9935705304148125,
+      "learning_rate": 3.05719962057059e-05,
+      "loss": 0.102,
+      "step": 1921
+    },
+    {
+      "epoch": 6.864285714285714,
+      "grad_norm": 0.7588145855352455,
+      "learning_rate": 3.056217555969247e-05,
+      "loss": 0.1079,
+      "step": 1922
+    },
+    {
+      "epoch": 6.867857142857143,
+      "grad_norm": 0.4866783357367315,
+      "learning_rate": 3.0552351380760746e-05,
+      "loss": 0.064,
+      "step": 1923
+    },
+    {
+      "epoch": 6.871428571428572,
+      "grad_norm": 0.34889517732182096,
+      "learning_rate": 3.0542523672196806e-05,
+      "loss": 0.0606,
+      "step": 1924
+    },
+    {
+      "epoch": 6.875,
+      "grad_norm": 0.44881245735580566,
+      "learning_rate": 3.05326924372879e-05,
+      "loss": 0.0313,
+      "step": 1925
+    },
+    {
+      "epoch": 6.878571428571428,
+      "grad_norm": 0.4555939311373143,
+      "learning_rate": 3.0522857679322444e-05,
+      "loss": 0.0384,
+      "step": 1926
+    },
+    {
+      "epoch": 6.882142857142857,
+      "grad_norm": 0.48171528806999336,
+      "learning_rate": 3.0513019401590046e-05,
+      "loss": 0.0481,
+      "step": 1927
+    },
+    {
+      "epoch": 6.885714285714286,
+      "grad_norm": 0.4370197702310799,
+      "learning_rate": 3.0503177607381506e-05,
+      "loss": 0.0541,
+      "step": 1928
+    },
+    {
+      "epoch": 6.889285714285714,
+      "grad_norm": 0.44815592325952275,
+      "learning_rate": 3.049333229998877e-05,
+      "loss": 0.046,
+      "step": 1929
+    },
+    {
+      "epoch": 6.892857142857143,
+      "grad_norm": 0.5682266855374412,
+      "learning_rate": 3.0483483482704968e-05,
+      "loss": 0.094,
+      "step": 1930
+    },
+    {
+      "epoch": 6.896428571428571,
+      "grad_norm": 0.7824260526346556,
+      "learning_rate": 3.0473631158824418e-05,
+      "loss": 0.1208,
+      "step": 1931
+    },
+    {
+      "epoch": 6.9,
+      "grad_norm": 0.7435677470714179,
+      "learning_rate": 3.04637753316426e-05,
+      "loss": 0.1086,
+      "step": 1932
+    },
+    {
+      "epoch": 6.9035714285714285,
+      "grad_norm": 0.3629857278578948,
+      "learning_rate": 3.045391600445617e-05,
+      "loss": 0.0343,
+      "step": 1933
+    },
+    {
+      "epoch": 6.9071428571428575,
+      "grad_norm": 1.0127715154358148,
+      "learning_rate": 3.0444053180562945e-05,
+      "loss": 0.0886,
+      "step": 1934
+    },
+    {
+      "epoch": 6.910714285714286,
+      "grad_norm": 0.7880946613706957,
+      "learning_rate": 3.0434186863261925e-05,
+      "loss": 0.0612,
+      "step": 1935
+    },
+    {
+      "epoch": 6.914285714285715,
+      "grad_norm": 0.5313207268584157,
+      "learning_rate": 3.042431705585327e-05,
+      "loss": 0.0859,
+      "step": 1936
+    },
+    {
+      "epoch": 6.917857142857143,
+      "grad_norm": 0.40335284377838876,
+      "learning_rate": 3.0414443761638313e-05,
+      "loss": 0.064,
+      "step": 1937
+    },
+    {
+      "epoch": 6.921428571428572,
+      "grad_norm": 1.0043669628396623,
+      "learning_rate": 3.0404566983919537e-05,
+      "loss": 0.0635,
+      "step": 1938
+    },
+    {
+      "epoch": 6.925,
+      "grad_norm": 0.789195657159531,
+      "learning_rate": 3.0394686726000617e-05,
+      "loss": 0.0693,
+      "step": 1939
+    },
+    {
+      "epoch": 6.928571428571429,
+      "grad_norm": 0.5767407372099836,
+      "learning_rate": 3.0384802991186374e-05,
+      "loss": 0.0337,
+      "step": 1940
+    },
+    {
+      "epoch": 6.932142857142857,
+      "grad_norm": 0.943576361902715,
+      "learning_rate": 3.037491578278279e-05,
+      "loss": 0.0614,
+      "step": 1941
+    },
+    {
+      "epoch": 6.935714285714286,
+      "grad_norm": 0.5489811381181934,
+      "learning_rate": 3.036502510409703e-05,
+      "loss": 0.0934,
+      "step": 1942
+    },
+    {
+      "epoch": 6.939285714285714,
+      "grad_norm": 0.33833920686222735,
+      "learning_rate": 3.0355130958437386e-05,
+      "loss": 0.0366,
+      "step": 1943
+    },
+    {
+      "epoch": 6.942857142857143,
+      "grad_norm": 0.9869133005664424,
+      "learning_rate": 3.0345233349113346e-05,
+      "loss": 0.1446,
+      "step": 1944
+    },
+    {
+      "epoch": 6.946428571428571,
+      "grad_norm": 0.6010613878780248,
+      "learning_rate": 3.0335332279435524e-05,
+      "loss": 0.0748,
+      "step": 1945
+    },
+    {
+      "epoch": 6.95,
+      "grad_norm": 0.603025556862844,
+      "learning_rate": 3.032542775271572e-05,
+      "loss": 0.0462,
+      "step": 1946
+    },
+    {
+      "epoch": 6.953571428571428,
+      "grad_norm": 0.8310533165920394,
+      "learning_rate": 3.0315519772266863e-05,
+      "loss": 0.064,
+      "step": 1947
+    },
+    {
+      "epoch": 6.957142857142857,
+      "grad_norm": 0.5210775027537463,
+      "learning_rate": 3.030560834140306e-05,
+      "loss": 0.0612,
+      "step": 1948
+    },
+    {
+      "epoch": 6.960714285714285,
+      "grad_norm": 0.6000032443533403,
+      "learning_rate": 3.029569346343957e-05,
+      "loss": 0.0895,
+      "step": 1949
+    },
+    {
+      "epoch": 6.964285714285714,
+      "grad_norm": 0.5913230568296698,
+      "learning_rate": 3.0285775141692785e-05,
+      "loss": 0.1006,
+      "step": 1950
+    },
+    {
+      "epoch": 6.9678571428571425,
+      "grad_norm": 0.9927558539729031,
+      "learning_rate": 3.0275853379480265e-05,
+      "loss": 0.1198,
+      "step": 1951
+    },
+    {
+      "epoch": 6.9714285714285715,
+      "grad_norm": 0.8106957189788327,
+      "learning_rate": 3.0265928180120717e-05,
+      "loss": 0.1049,
+      "step": 1952
+    },
+    {
+      "epoch": 6.975,
+      "grad_norm": 1.2632032049313413,
+      "learning_rate": 3.0255999546934002e-05,
+      "loss": 0.0713,
+      "step": 1953
+    },
+    {
+      "epoch": 6.978571428571429,
+      "grad_norm": 0.5824324893127132,
+      "learning_rate": 3.0246067483241127e-05,
+      "loss": 0.0352,
+      "step": 1954
+    },
+    {
+      "epoch": 6.982142857142857,
+      "grad_norm": 1.4439533336552834,
+      "learning_rate": 3.0236131992364243e-05,
+      "loss": 0.0934,
+      "step": 1955
+    },
+    {
+      "epoch": 6.985714285714286,
+      "grad_norm": 0.4998041516396315,
+      "learning_rate": 3.0226193077626652e-05,
+      "loss": 0.0485,
+      "step": 1956
+    },
+    {
+      "epoch": 6.989285714285714,
+      "grad_norm": 0.3543942577452194,
+      "learning_rate": 3.0216250742352798e-05,
+      "loss": 0.0252,
+      "step": 1957
+    },
+    {
+      "epoch": 6.992857142857143,
+      "grad_norm": 0.974707058892318,
+      "learning_rate": 3.020630498986827e-05,
+      "loss": 0.0929,
+      "step": 1958
+    },
+    {
+      "epoch": 6.996428571428572,
+      "grad_norm": 0.7778242041816037,
+      "learning_rate": 3.0196355823499802e-05,
+      "loss": 0.0491,
+      "step": 1959
+    },
+    {
+      "epoch": 7.0,
+      "grad_norm": 0.8449257267188708,
+      "learning_rate": 3.0186403246575263e-05,
+      "loss": 0.0543,
+      "step": 1960
+    },
+    {
+      "epoch": 7.003571428571429,
+      "grad_norm": 0.6712453688599759,
+      "learning_rate": 3.017644726242367e-05,
+      "loss": 0.0573,
+      "step": 1961
+    },
+    {
+      "epoch": 7.007142857142857,
+      "grad_norm": 0.7132509110704268,
+      "learning_rate": 3.016648787437519e-05,
+      "loss": 0.0823,
+      "step": 1962
+    },
+    {
+      "epoch": 7.010714285714286,
+      "grad_norm": 0.7002747031837383,
+      "learning_rate": 3.0156525085761097e-05,
+      "loss": 0.0403,
+      "step": 1963
+    },
+    {
+      "epoch": 7.014285714285714,
+      "grad_norm": 0.7392236891753348,
+      "learning_rate": 3.0146558899913833e-05,
+      "loss": 0.0656,
+      "step": 1964
+    },
+    {
+      "epoch": 7.017857142857143,
+      "grad_norm": 0.9035640975423979,
+      "learning_rate": 3.0136589320166964e-05,
+      "loss": 0.0579,
+      "step": 1965
+    },
+    {
+      "epoch": 7.021428571428571,
+      "grad_norm": 0.3650430330550526,
+      "learning_rate": 3.0126616349855186e-05,
+      "loss": 0.036,
+      "step": 1966
+    },
+    {
+      "epoch": 7.025,
+      "grad_norm": 1.0693302656811483,
+      "learning_rate": 3.0116639992314343e-05,
+      "loss": 0.098,
+      "step": 1967
+    },
+    {
+      "epoch": 7.0285714285714285,
+      "grad_norm": 0.6902530676017324,
+      "learning_rate": 3.0106660250881403e-05,
+      "loss": 0.1054,
+      "step": 1968
+    },
+    {
+      "epoch": 7.0321428571428575,
+      "grad_norm": 0.7558512393457515,
+      "learning_rate": 3.0096677128894464e-05,
+      "loss": 0.0774,
+      "step": 1969
+    },
+    {
+      "epoch": 7.035714285714286,
+      "grad_norm": 0.9024737463651683,
+      "learning_rate": 3.0086690629692757e-05,
+      "loss": 0.0698,
+      "step": 1970
+    },
+    {
+      "epoch": 7.039285714285715,
+      "grad_norm": 0.7587215596536265,
+      "learning_rate": 3.007670075661665e-05,
+      "loss": 0.0819,
+      "step": 1971
+    },
+    {
+      "epoch": 7.042857142857143,
+      "grad_norm": 1.24746699752787,
+      "learning_rate": 3.006670751300763e-05,
+      "loss": 0.0835,
+      "step": 1972
+    },
+    {
+      "epoch": 7.046428571428572,
+      "grad_norm": 0.6301074545066738,
+      "learning_rate": 3.005671090220831e-05,
+      "loss": 0.0574,
+      "step": 1973
+    },
+    {
+      "epoch": 7.05,
+      "grad_norm": 0.7968782669998499,
+      "learning_rate": 3.0046710927562442e-05,
+      "loss": 0.0672,
+      "step": 1974
+    },
+    {
+      "epoch": 7.053571428571429,
+      "grad_norm": 0.9254008030585812,
+      "learning_rate": 3.003670759241489e-05,
+      "loss": 0.0923,
+      "step": 1975
+    },
+    {
+      "epoch": 7.057142857142857,
+      "grad_norm": 0.49953374771562803,
+      "learning_rate": 3.0026700900111642e-05,
+      "loss": 0.0912,
+      "step": 1976
+    },
+    {
+      "epoch": 7.060714285714286,
+      "grad_norm": 1.0247967650118595,
+      "learning_rate": 3.0016690853999827e-05,
+      "loss": 0.1071,
+      "step": 1977
+    },
+    {
+      "epoch": 7.064285714285714,
+      "grad_norm": 0.42724468643092933,
+      "learning_rate": 3.0006677457427674e-05,
+      "loss": 0.0356,
+      "step": 1978
+    },
+    {
+      "epoch": 7.067857142857143,
+      "grad_norm": 0.45975588963978625,
+      "learning_rate": 2.9996660713744538e-05,
+      "loss": 0.0358,
+      "step": 1979
+    },
+    {
+      "epoch": 7.071428571428571,
+      "grad_norm": 0.7270981210481786,
+      "learning_rate": 2.9986640626300906e-05,
+      "loss": 0.0959,
+      "step": 1980
+    },
+    {
+      "epoch": 7.075,
+      "grad_norm": 0.7270647910887004,
+      "learning_rate": 2.9976617198448364e-05,
+      "loss": 0.0365,
+      "step": 1981
+    },
+    {
+      "epoch": 7.078571428571428,
+      "grad_norm": 0.4564633700075659,
+      "learning_rate": 2.9966590433539636e-05,
+      "loss": 0.0526,
+      "step": 1982
+    },
+    {
+      "epoch": 7.082142857142857,
+      "grad_norm": 0.3727560368124852,
+      "learning_rate": 2.9956560334928548e-05,
+      "loss": 0.0448,
+      "step": 1983
+    },
+    {
+      "epoch": 7.085714285714285,
+      "grad_norm": 1.2040017903427158,
+      "learning_rate": 2.9946526905970045e-05,
+      "loss": 0.1756,
+      "step": 1984
+    },
+    {
+      "epoch": 7.089285714285714,
+      "grad_norm": 1.26918593602553,
+      "learning_rate": 2.993649015002019e-05,
+      "loss": 0.1157,
+      "step": 1985
+    },
+    {
+      "epoch": 7.0928571428571425,
+      "grad_norm": 1.1902843149725113,
+      "learning_rate": 2.992645007043615e-05,
+      "loss": 0.1046,
+      "step": 1986
+    },
+    {
+      "epoch": 7.0964285714285715,
+      "grad_norm": 0.5006908653964292,
+      "learning_rate": 2.9916406670576205e-05,
+      "loss": 0.0534,
+      "step": 1987
+    },
+    {
+      "epoch": 7.1,
+      "grad_norm": 0.4269327610101025,
+      "learning_rate": 2.9906359953799756e-05,
+      "loss": 0.0512,
+      "step": 1988
+    },
+    {
+      "epoch": 7.103571428571429,
+      "grad_norm": 0.5951286388740064,
+      "learning_rate": 2.9896309923467305e-05,
+      "loss": 0.0882,
+      "step": 1989
+    },
+    {
+      "epoch": 7.107142857142857,
+      "grad_norm": 0.7120366042132479,
+      "learning_rate": 2.988625658294046e-05,
+      "loss": 0.0585,
+      "step": 1990
+    },
+    {
+      "epoch": 7.110714285714286,
+      "grad_norm": 0.5402443871213362,
+      "learning_rate": 2.9876199935581945e-05,
+      "loss": 0.1049,
+      "step": 1991
+    },
+    {
+      "epoch": 7.114285714285714,
+      "grad_norm": 0.726658118304981,
+      "learning_rate": 2.9866139984755587e-05,
+      "loss": 0.0722,
+      "step": 1992
+    },
+    {
+      "epoch": 7.117857142857143,
+      "grad_norm": 0.45068586655074155,
+      "learning_rate": 2.9856076733826307e-05,
+      "loss": 0.0289,
+      "step": 1993
+    },
+    {
+      "epoch": 7.121428571428571,
+      "grad_norm": 0.4294331123366745,
+      "learning_rate": 2.9846010186160148e-05,
+      "loss": 0.0589,
+      "step": 1994
+    },
+    {
+      "epoch": 7.125,
+      "grad_norm": 0.7597867843159476,
+      "learning_rate": 2.9835940345124245e-05,
+      "loss": 0.0564,
+      "step": 1995
+    },
+    {
+      "epoch": 7.128571428571428,
+      "grad_norm": 0.4947865635350962,
+      "learning_rate": 2.982586721408684e-05,
+      "loss": 0.0561,
+      "step": 1996
+    },
+    {
+      "epoch": 7.132142857142857,
+      "grad_norm": 0.30739375881490705,
+      "learning_rate": 2.9815790796417263e-05,
+      "loss": 0.0369,
+      "step": 1997
+    },
+    {
+      "epoch": 7.135714285714286,
+      "grad_norm": 1.206327387977939,
+      "learning_rate": 2.9805711095485955e-05,
+      "loss": 0.0966,
+      "step": 1998
+    },
+    {
+      "epoch": 7.139285714285714,
+      "grad_norm": 1.2383058785658734,
+      "learning_rate": 2.9795628114664456e-05,
+      "loss": 0.1157,
+      "step": 1999
+    },
+    {
+      "epoch": 7.142857142857143,
+      "grad_norm": 0.3840093547290718,
+      "learning_rate": 2.9785541857325397e-05,
+      "loss": 0.0407,
+      "step": 2000
+    },
+    {
+      "epoch": 7.146428571428571,
+      "grad_norm": 0.649588595988064,
+      "learning_rate": 2.9775452326842508e-05,
+      "loss": 0.0631,
+      "step": 2001
+    },
+    {
+      "epoch": 7.15,
+      "grad_norm": 0.5779028075691239,
+      "learning_rate": 2.9765359526590614e-05,
+      "loss": 0.078,
+      "step": 2002
+    },
+    {
+      "epoch": 7.1535714285714285,
+      "grad_norm": 0.6192854015752687,
+      "learning_rate": 2.9755263459945628e-05,
+      "loss": 0.0543,
+      "step": 2003
+    },
+    {
+      "epoch": 7.1571428571428575,
+      "grad_norm": 0.6268981119092637,
+      "learning_rate": 2.974516413028457e-05,
+      "loss": 0.0731,
+      "step": 2004
+    },
+    {
+      "epoch": 7.160714285714286,
+      "grad_norm": 0.46804841400899744,
+      "learning_rate": 2.9735061540985533e-05,
+      "loss": 0.0601,
+      "step": 2005
+    },
+    {
+      "epoch": 7.164285714285715,
+      "grad_norm": 0.7357031703029466,
+      "learning_rate": 2.972495569542772e-05,
+      "loss": 0.0637,
+      "step": 2006
+    },
+    {
+      "epoch": 7.167857142857143,
+      "grad_norm": 0.42505468738210533,
+      "learning_rate": 2.9714846596991398e-05,
+      "loss": 0.0412,
+      "step": 2007
+    },
+    {
+      "epoch": 7.171428571428572,
+      "grad_norm": 0.4636700970658616,
+      "learning_rate": 2.970473424905795e-05,
+      "loss": 0.0666,
+      "step": 2008
+    },
+    {
+      "epoch": 7.175,
+      "grad_norm": 0.5713715693405613,
+      "learning_rate": 2.969461865500982e-05,
+      "loss": 0.0477,
+      "step": 2009
+    },
+    {
+      "epoch": 7.178571428571429,
+      "grad_norm": 0.669065508781555,
+      "learning_rate": 2.9684499818230565e-05,
+      "loss": 0.06,
+      "step": 2010
+    },
+    {
+      "epoch": 7.182142857142857,
+      "grad_norm": 0.8536671750408703,
+      "learning_rate": 2.9674377742104798e-05,
+      "loss": 0.0575,
+      "step": 2011
+    },
+    {
+      "epoch": 7.185714285714286,
+      "grad_norm": 0.5614276934972854,
+      "learning_rate": 2.966425243001824e-05,
+      "loss": 0.0724,
+      "step": 2012
+    },
+    {
+      "epoch": 7.189285714285714,
+      "grad_norm": 0.6438816293611394,
+      "learning_rate": 2.9654123885357672e-05,
+      "loss": 0.0853,
+      "step": 2013
+    },
+    {
+      "epoch": 7.192857142857143,
+      "grad_norm": 1.0593993943167186,
+      "learning_rate": 2.964399211151098e-05,
+      "loss": 0.0864,
+      "step": 2014
+    },
+    {
+      "epoch": 7.196428571428571,
+      "grad_norm": 0.3328717183311614,
+      "learning_rate": 2.9633857111867113e-05,
+      "loss": 0.0164,
+      "step": 2015
+    },
+    {
+      "epoch": 7.2,
+      "grad_norm": 0.6808367779690029,
+      "learning_rate": 2.9623718889816105e-05,
+      "loss": 0.0538,
+      "step": 2016
+    },
+    {
+      "epoch": 7.203571428571428,
+      "grad_norm": 0.6555524308868562,
+      "learning_rate": 2.9613577448749064e-05,
+      "loss": 0.0463,
+      "step": 2017
+    },
+    {
+      "epoch": 7.207142857142857,
+      "grad_norm": 0.9737395658599877,
+      "learning_rate": 2.9603432792058177e-05,
+      "loss": 0.1164,
+      "step": 2018
+    },
+    {
+      "epoch": 7.210714285714285,
+      "grad_norm": 1.044664224607262,
+      "learning_rate": 2.9593284923136716e-05,
+      "loss": 0.0576,
+      "step": 2019
+    },
+    {
+      "epoch": 7.214285714285714,
+      "grad_norm": 0.7239141701527848,
+      "learning_rate": 2.958313384537902e-05,
+      "loss": 0.0731,
+      "step": 2020
+    },
+    {
+      "epoch": 7.2178571428571425,
+      "grad_norm": 0.46311470976738117,
+      "learning_rate": 2.9572979562180482e-05,
+      "loss": 0.0556,
+      "step": 2021
+    },
+    {
+      "epoch": 7.2214285714285715,
+      "grad_norm": 0.6085987562550119,
+      "learning_rate": 2.95628220769376e-05,
+      "loss": 0.091,
+      "step": 2022
+    },
+    {
+      "epoch": 7.225,
+      "grad_norm": 0.6516077235972756,
+      "learning_rate": 2.955266139304793e-05,
+      "loss": 0.0455,
+      "step": 2023
+    },
+    {
+      "epoch": 7.228571428571429,
+      "grad_norm": 0.8791760094734675,
+      "learning_rate": 2.9542497513910084e-05,
+      "loss": 0.0465,
+      "step": 2024
+    },
+    {
+      "epoch": 7.232142857142857,
+      "grad_norm": 0.7115709404668651,
+      "learning_rate": 2.9532330442923768e-05,
+      "loss": 0.0333,
+      "step": 2025
+    },
+    {
+      "epoch": 7.235714285714286,
+      "grad_norm": 0.5846805569243129,
+      "learning_rate": 2.9522160183489733e-05,
+      "loss": 0.0323,
+      "step": 2026
+    },
+    {
+      "epoch": 7.239285714285714,
+      "grad_norm": 0.6995511152057403,
+      "learning_rate": 2.9511986739009812e-05,
+      "loss": 0.0706,
+      "step": 2027
+    },
+    {
+      "epoch": 7.242857142857143,
+      "grad_norm": 1.0244609599406405,
+      "learning_rate": 2.950181011288689e-05,
+      "loss": 0.0482,
+      "step": 2028
+    },
+    {
+      "epoch": 7.246428571428572,
+      "grad_norm": 0.7535564827743133,
+      "learning_rate": 2.9491630308524937e-05,
+      "loss": 0.0439,
+      "step": 2029
+    },
+    {
+      "epoch": 7.25,
+      "grad_norm": 0.9887705146525638,
+      "learning_rate": 2.9481447329328965e-05,
+      "loss": 0.0848,
+      "step": 2030
+    },
+    {
+      "epoch": 7.253571428571428,
+      "grad_norm": 0.8151798920008275,
+      "learning_rate": 2.947126117870505e-05,
+      "loss": 0.0627,
+      "step": 2031
+    },
+    {
+      "epoch": 7.257142857142857,
+      "grad_norm": 0.6452423771722877,
+      "learning_rate": 2.9461071860060343e-05,
+      "loss": 0.0663,
+      "step": 2032
+    },
+    {
+      "epoch": 7.260714285714286,
+      "grad_norm": 0.8687761128727773,
+      "learning_rate": 2.9450879376803053e-05,
+      "loss": 0.0316,
+      "step": 2033
+    },
+    {
+      "epoch": 7.264285714285714,
+      "grad_norm": 0.48452816957334444,
+      "learning_rate": 2.944068373234242e-05,
+      "loss": 0.042,
+      "step": 2034
+    },
+    {
+      "epoch": 7.267857142857143,
+      "grad_norm": 1.0794203923467016,
+      "learning_rate": 2.9430484930088775e-05,
+      "loss": 0.0666,
+      "step": 2035
+    },
+    {
+      "epoch": 7.271428571428571,
+      "grad_norm": 1.0380293663284659,
+      "learning_rate": 2.9420282973453497e-05,
+      "loss": 0.0981,
+      "step": 2036
+    },
+    {
+      "epoch": 7.275,
+      "grad_norm": 1.304766780557781,
+      "learning_rate": 2.9410077865849003e-05,
+      "loss": 0.1361,
+      "step": 2037
+    },
+    {
+      "epoch": 7.2785714285714285,
+      "grad_norm": 0.8661619591567666,
+      "learning_rate": 2.9399869610688788e-05,
+      "loss": 0.0504,
+      "step": 2038
+    },
+    {
+      "epoch": 7.2821428571428575,
+      "grad_norm": 0.6131619861663105,
+      "learning_rate": 2.938965821138738e-05,
+      "loss": 0.0754,
+      "step": 2039
+    },
+    {
+      "epoch": 7.285714285714286,
+      "grad_norm": 0.7477187139210263,
+      "learning_rate": 2.9379443671360367e-05,
+      "loss": 0.0901,
+      "step": 2040
+    },
+    {
+      "epoch": 7.289285714285715,
+      "grad_norm": 0.6225860334407884,
+      "learning_rate": 2.9369225994024394e-05,
+      "loss": 0.0602,
+      "step": 2041
+    },
+    {
+      "epoch": 7.292857142857143,
+      "grad_norm": 0.7247457935555545,
+      "learning_rate": 2.9359005182797145e-05,
+      "loss": 0.0784,
+      "step": 2042
+    },
+    {
+      "epoch": 7.296428571428572,
+      "grad_norm": 0.6790761658548092,
+      "learning_rate": 2.9348781241097358e-05,
+      "loss": 0.0451,
+      "step": 2043
+    },
+    {
+      "epoch": 7.3,
+      "grad_norm": 0.898773054020782,
+      "learning_rate": 2.9338554172344813e-05,
+      "loss": 0.0598,
+      "step": 2044
+    },
+    {
+      "epoch": 7.303571428571429,
+      "grad_norm": 0.7208778364952277,
+      "learning_rate": 2.9328323979960346e-05,
+      "loss": 0.0609,
+      "step": 2045
+    },
+    {
+      "epoch": 7.307142857142857,
+      "grad_norm": 0.6513036309691149,
+      "learning_rate": 2.9318090667365823e-05,
+      "loss": 0.0474,
+      "step": 2046
+    },
+    {
+      "epoch": 7.310714285714286,
+      "grad_norm": 0.5203436537059007,
+      "learning_rate": 2.9307854237984158e-05,
+      "loss": 0.0662,
+      "step": 2047
+    },
+    {
+      "epoch": 7.314285714285714,
+      "grad_norm": 0.5128937307217091,
+      "learning_rate": 2.9297614695239324e-05,
+      "loss": 0.0178,
+      "step": 2048
+    },
+    {
+      "epoch": 7.317857142857143,
+      "grad_norm": 1.1289580270478656,
+      "learning_rate": 2.928737204255631e-05,
+      "loss": 0.0445,
+      "step": 2049
+    },
+    {
+      "epoch": 7.321428571428571,
+      "grad_norm": 1.363672687869839,
+      "learning_rate": 2.9277126283361173e-05,
+      "loss": 0.1002,
+      "step": 2050
+    },
+    {
+      "epoch": 7.325,
+      "grad_norm": 0.835123466824242,
+      "learning_rate": 2.9266877421080976e-05,
+      "loss": 0.0773,
+      "step": 2051
+    },
+    {
+      "epoch": 7.328571428571428,
+      "grad_norm": 0.6398214973778992,
+      "learning_rate": 2.925662545914385e-05,
+      "loss": 0.0541,
+      "step": 2052
+    },
+    {
+      "epoch": 7.332142857142857,
+      "grad_norm": 0.8274511089766811,
+      "learning_rate": 2.9246370400978953e-05,
+      "loss": 0.0642,
+      "step": 2053
+    },
+    {
+      "epoch": 7.335714285714285,
+      "grad_norm": 0.9732112839495851,
+      "learning_rate": 2.923611225001646e-05,
+      "loss": 0.0849,
+      "step": 2054
+    },
+    {
+      "epoch": 7.339285714285714,
+      "grad_norm": 0.6647899231213188,
+      "learning_rate": 2.9225851009687614e-05,
+      "loss": 0.0452,
+      "step": 2055
+    },
+    {
+      "epoch": 7.3428571428571425,
+      "grad_norm": 1.0699087064532573,
+      "learning_rate": 2.921558668342467e-05,
+      "loss": 0.0596,
+      "step": 2056
+    },
+    {
+      "epoch": 7.3464285714285715,
+      "grad_norm": 0.9896139666042898,
+      "learning_rate": 2.920531927466091e-05,
+      "loss": 0.0799,
+      "step": 2057
+    },
+    {
+      "epoch": 7.35,
+      "grad_norm": 0.6735656838801739,
+      "learning_rate": 2.9195048786830672e-05,
+      "loss": 0.0549,
+      "step": 2058
+    },
+    {
+      "epoch": 7.353571428571429,
+      "grad_norm": 0.3794916070874405,
+      "learning_rate": 2.9184775223369294e-05,
+      "loss": 0.0412,
+      "step": 2059
+    },
+    {
+      "epoch": 7.357142857142857,
+      "grad_norm": 0.6527431059189714,
+      "learning_rate": 2.917449858771316e-05,
+      "loss": 0.0601,
+      "step": 2060
+    },
+    {
+      "epoch": 7.360714285714286,
+      "grad_norm": 0.8671718715615856,
+      "learning_rate": 2.9164218883299686e-05,
+      "loss": 0.1325,
+      "step": 2061
+    },
+    {
+      "epoch": 7.364285714285714,
+      "grad_norm": 0.5866846664848318,
+      "learning_rate": 2.9153936113567302e-05,
+      "loss": 0.0854,
+      "step": 2062
+    },
+    {
+      "epoch": 7.367857142857143,
+      "grad_norm": 0.44214139218838994,
+      "learning_rate": 2.9143650281955466e-05,
+      "loss": 0.0415,
+      "step": 2063
+    },
+    {
+      "epoch": 7.371428571428572,
+      "grad_norm": 0.6098814701306855,
+      "learning_rate": 2.9133361391904667e-05,
+      "loss": 0.0617,
+      "step": 2064
+    },
+    {
+      "epoch": 7.375,
+      "grad_norm": 0.7353176973768965,
+      "learning_rate": 2.912306944685641e-05,
+      "loss": 0.0956,
+      "step": 2065
+    },
+    {
+      "epoch": 7.378571428571428,
+      "grad_norm": 0.6908630891848049,
+      "learning_rate": 2.911277445025323e-05,
+      "loss": 0.0944,
+      "step": 2066
+    },
+    {
+      "epoch": 7.382142857142857,
+      "grad_norm": 0.43217578053827405,
+      "learning_rate": 2.9102476405538673e-05,
+      "loss": 0.0701,
+      "step": 2067
+    },
+    {
+      "epoch": 7.385714285714286,
+      "grad_norm": 0.5387522453685842,
+      "learning_rate": 2.90921753161573e-05,
+      "loss": 0.0442,
+      "step": 2068
+    },
+    {
+      "epoch": 7.389285714285714,
+      "grad_norm": 0.5875282218618699,
+      "learning_rate": 2.9081871185554718e-05,
+      "loss": 0.0567,
+      "step": 2069
+    },
+    {
+      "epoch": 7.392857142857143,
+      "grad_norm": 0.6879295231978868,
+      "learning_rate": 2.9071564017177523e-05,
+      "loss": 0.0736,
+      "step": 2070
+    },
+    {
+      "epoch": 7.396428571428571,
+      "grad_norm": 0.993709650249994,
+      "learning_rate": 2.906125381447333e-05,
+      "loss": 0.1129,
+      "step": 2071
+    },
+    {
+      "epoch": 7.4,
+      "grad_norm": 0.5650485333941964,
+      "learning_rate": 2.9050940580890783e-05,
+      "loss": 0.0477,
+      "step": 2072
+    },
+    {
+      "epoch": 7.4035714285714285,
+      "grad_norm": 0.5384665086818657,
+      "learning_rate": 2.904062431987953e-05,
+      "loss": 0.0519,
+      "step": 2073
+    },
+    {
+      "epoch": 7.4071428571428575,
+      "grad_norm": 0.5483549975571231,
+      "learning_rate": 2.9030305034890235e-05,
+      "loss": 0.0805,
+      "step": 2074
+    },
+    {
+      "epoch": 7.410714285714286,
+      "grad_norm": 0.743993040254456,
+      "learning_rate": 2.9019982729374566e-05,
+      "loss": 0.0711,
+      "step": 2075
+    },
+    {
+      "epoch": 7.414285714285715,
+      "grad_norm": 1.0180249055915893,
+      "learning_rate": 2.900965740678522e-05,
+      "loss": 0.0877,
+      "step": 2076
+    },
+    {
+      "epoch": 7.417857142857143,
+      "grad_norm": 0.7114672633802002,
+      "learning_rate": 2.899932907057588e-05,
+      "loss": 0.0973,
+      "step": 2077
+    },
+    {
+      "epoch": 7.421428571428572,
+      "grad_norm": 0.5235676224503608,
+      "learning_rate": 2.8988997724201255e-05,
+      "loss": 0.047,
+      "step": 2078
+    },
+    {
+      "epoch": 7.425,
+      "grad_norm": 0.7491112852430354,
+      "learning_rate": 2.897866337111705e-05,
+      "loss": 0.0671,
+      "step": 2079
+    },
+    {
+      "epoch": 7.428571428571429,
+      "grad_norm": 0.3955391332982998,
+      "learning_rate": 2.8968326014779982e-05,
+      "loss": 0.0652,
+      "step": 2080
+    },
+    {
+      "epoch": 7.432142857142857,
+      "grad_norm": 0.5428604020661156,
+      "learning_rate": 2.8957985658647767e-05,
+      "loss": 0.0808,
+      "step": 2081
+    },
+    {
+      "epoch": 7.435714285714286,
+      "grad_norm": 1.008651101953109,
+      "learning_rate": 2.894764230617913e-05,
+      "loss": 0.1027,
+      "step": 2082
+    },
+    {
+      "epoch": 7.439285714285714,
+      "grad_norm": 0.46579509311710177,
+      "learning_rate": 2.8937295960833793e-05,
+      "loss": 0.0879,
+      "step": 2083
+    },
+    {
+      "epoch": 7.442857142857143,
+      "grad_norm": 0.9103171411720223,
+      "learning_rate": 2.8926946626072482e-05,
+      "loss": 0.0651,
+      "step": 2084
+    },
+    {
+      "epoch": 7.446428571428571,
+      "grad_norm": 0.6992158988047553,
+      "learning_rate": 2.8916594305356923e-05,
+      "loss": 0.0655,
+      "step": 2085
+    },
+    {
+      "epoch": 7.45,
+      "grad_norm": 0.6213166392735381,
+      "learning_rate": 2.890623900214985e-05,
+      "loss": 0.0568,
+      "step": 2086
+    },
+    {
+      "epoch": 7.453571428571428,
+      "grad_norm": 0.4953391930093722,
+      "learning_rate": 2.889588071991497e-05,
+      "loss": 0.0938,
+      "step": 2087
+    },
+    {
+      "epoch": 7.457142857142857,
+      "grad_norm": 0.41665642408016795,
+      "learning_rate": 2.8885519462117004e-05,
+      "loss": 0.048,
+      "step": 2088
+    },
+    {
+      "epoch": 7.460714285714285,
+      "grad_norm": 0.6636447696184257,
+      "learning_rate": 2.8875155232221673e-05,
+      "loss": 0.0745,
+      "step": 2089
+    },
+    {
+      "epoch": 7.464285714285714,
+      "grad_norm": 0.7110785117716738,
+      "learning_rate": 2.8864788033695678e-05,
+      "loss": 0.0377,
+      "step": 2090
+    },
+    {
+      "epoch": 7.4678571428571425,
+      "grad_norm": 0.5847390313241465,
+      "learning_rate": 2.8854417870006733e-05,
+      "loss": 0.0958,
+      "step": 2091
+    },
+    {
+      "epoch": 7.4714285714285715,
+      "grad_norm": 0.6364853304386928,
+      "learning_rate": 2.8844044744623514e-05,
+      "loss": 0.0489,
+      "step": 2092
+    },
+    {
+      "epoch": 7.475,
+      "grad_norm": 0.5077237174248657,
+      "learning_rate": 2.8833668661015714e-05,
+      "loss": 0.0616,
+      "step": 2093
+    },
+    {
+      "epoch": 7.478571428571429,
+      "grad_norm": 0.45442828253187423,
+      "learning_rate": 2.8823289622654e-05,
+      "loss": 0.0513,
+      "step": 2094
+    },
+    {
+      "epoch": 7.482142857142857,
+      "grad_norm": 0.7564115628893106,
+      "learning_rate": 2.8812907633010035e-05,
+      "loss": 0.0951,
+      "step": 2095
+    },
+    {
+      "epoch": 7.485714285714286,
+      "grad_norm": 0.7687577195820005,
+      "learning_rate": 2.880252269555647e-05,
+      "loss": 0.0545,
+      "step": 2096
+    },
+    {
+      "epoch": 7.489285714285714,
+      "grad_norm": 0.439864128292456,
+      "learning_rate": 2.879213481376694e-05,
+      "loss": 0.064,
+      "step": 2097
+    },
+    {
+      "epoch": 7.492857142857143,
+      "grad_norm": 0.6778046192555219,
+      "learning_rate": 2.8781743991116057e-05,
+      "loss": 0.0587,
+      "step": 2098
+    },
+    {
+      "epoch": 7.496428571428572,
+      "grad_norm": 0.5952096987565308,
+      "learning_rate": 2.8771350231079435e-05,
+      "loss": 0.0745,
+      "step": 2099
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 0.5360457128876778,
+      "learning_rate": 2.876095353713365e-05,
+      "loss": 0.064,
+      "step": 2100
+    },
+    {
+      "epoch": 7.503571428571428,
+      "grad_norm": 0.5581839850197868,
+      "learning_rate": 2.875055391275627e-05,
+      "loss": 0.0919,
+      "step": 2101
+    },
+    {
+      "epoch": 7.507142857142857,
+      "grad_norm": 0.6123532985767934,
+      "learning_rate": 2.8740151361425844e-05,
+      "loss": 0.0794,
+      "step": 2102
+    },
+    {
+      "epoch": 7.510714285714286,
+      "grad_norm": 1.11871098938474,
+      "learning_rate": 2.8729745886621902e-05,
+      "loss": 0.0776,
+      "step": 2103
+    },
+    {
+      "epoch": 7.514285714285714,
+      "grad_norm": 0.5610442103766712,
+      "learning_rate": 2.8719337491824943e-05,
+      "loss": 0.0654,
+      "step": 2104
+    },
+    {
+      "epoch": 7.517857142857143,
+      "grad_norm": 0.6053554305091994,
+      "learning_rate": 2.8708926180516443e-05,
+      "loss": 0.0719,
+      "step": 2105
+    },
+    {
+      "epoch": 7.521428571428571,
+      "grad_norm": 1.3113506838288063,
+      "learning_rate": 2.8698511956178868e-05,
+      "loss": 0.0797,
+      "step": 2106
+    },
+    {
+      "epoch": 7.525,
+      "grad_norm": 0.7639951612738582,
+      "learning_rate": 2.868809482229564e-05,
+      "loss": 0.0481,
+      "step": 2107
+    },
+    {
+      "epoch": 7.5285714285714285,
+      "grad_norm": 0.6209066148621736,
+      "learning_rate": 2.8677674782351164e-05,
+      "loss": 0.0423,
+      "step": 2108
+    },
+    {
+      "epoch": 7.5321428571428575,
+      "grad_norm": 0.6451106082974598,
+      "learning_rate": 2.866725183983082e-05,
+      "loss": 0.082,
+      "step": 2109
+    },
+    {
+      "epoch": 7.535714285714286,
+      "grad_norm": 0.5619509623572307,
+      "learning_rate": 2.8656825998220954e-05,
+      "loss": 0.0638,
+      "step": 2110
+    },
+    {
+      "epoch": 7.539285714285715,
+      "grad_norm": 0.8059982056451351,
+      "learning_rate": 2.8646397261008882e-05,
+      "loss": 0.0475,
+      "step": 2111
+    },
+    {
+      "epoch": 7.542857142857143,
+      "grad_norm": 0.8191259261477638,
+      "learning_rate": 2.8635965631682883e-05,
+      "loss": 0.0553,
+      "step": 2112
+    },
+    {
+      "epoch": 7.546428571428572,
+      "grad_norm": 1.6191913091494126,
+      "learning_rate": 2.8625531113732212e-05,
+      "loss": 0.0617,
+      "step": 2113
+    },
+    {
+      "epoch": 7.55,
+      "grad_norm": 0.7300494991072307,
+      "learning_rate": 2.8615093710647098e-05,
+      "loss": 0.0512,
+      "step": 2114
+    },
+    {
+      "epoch": 7.553571428571429,
+      "grad_norm": 0.7016209407108143,
+      "learning_rate": 2.8604653425918705e-05,
+      "loss": 0.0773,
+      "step": 2115
+    },
+    {
+      "epoch": 7.557142857142857,
+      "grad_norm": 1.0697671839091385,
+      "learning_rate": 2.85942102630392e-05,
+      "loss": 0.1053,
+      "step": 2116
+    },
+    {
+      "epoch": 7.560714285714286,
+      "grad_norm": 0.9004042283967888,
+      "learning_rate": 2.858376422550168e-05,
+      "loss": 0.0818,
+      "step": 2117
+    },
+    {
+      "epoch": 7.564285714285714,
+      "grad_norm": 0.6848363448705084,
+      "learning_rate": 2.8573315316800213e-05,
+      "loss": 0.0821,
+      "step": 2118
+    },
+    {
+      "epoch": 7.567857142857143,
+      "grad_norm": 0.6192725713697306,
+      "learning_rate": 2.856286354042985e-05,
+      "loss": 0.0818,
+      "step": 2119
+    },
+    {
+      "epoch": 7.571428571428571,
+      "grad_norm": 1.196125071486183,
+      "learning_rate": 2.855240889988656e-05,
+      "loss": 0.0908,
+      "step": 2120
+    },
+    {
+      "epoch": 7.575,
+      "grad_norm": 0.6805434299676187,
+      "learning_rate": 2.8541951398667307e-05,
+      "loss": 0.0755,
+      "step": 2121
+    },
+    {
+      "epoch": 7.578571428571428,
+      "grad_norm": 0.881927025112184,
+      "learning_rate": 2.853149104026999e-05,
+      "loss": 0.0758,
+      "step": 2122
+    },
+    {
+      "epoch": 7.582142857142857,
+      "grad_norm": 0.6832544549334831,
+      "learning_rate": 2.852102782819348e-05,
+      "loss": 0.0535,
+      "step": 2123
+    },
+    {
+      "epoch": 7.585714285714285,
+      "grad_norm": 0.389765133687767,
+      "learning_rate": 2.8510561765937574e-05,
+      "loss": 0.0349,
+      "step": 2124
+    },
+    {
+      "epoch": 7.589285714285714,
+      "grad_norm": 0.5429587546057193,
+      "learning_rate": 2.8500092857003064e-05,
+      "loss": 0.0833,
+      "step": 2125
+    },
+    {
+      "epoch": 7.5928571428571425,
+      "grad_norm": 0.4198838609023052,
+      "learning_rate": 2.8489621104891654e-05,
+      "loss": 0.0272,
+      "step": 2126
+    },
+    {
+      "epoch": 7.5964285714285715,
+      "grad_norm": 0.7940608637144116,
+      "learning_rate": 2.847914651310603e-05,
+      "loss": 0.0923,
+      "step": 2127
+    },
+    {
+      "epoch": 7.6,
+      "grad_norm": 0.680712761600877,
+      "learning_rate": 2.8468669085149812e-05,
+      "loss": 0.0778,
+      "step": 2128
+    },
+    {
+      "epoch": 7.603571428571429,
+      "grad_norm": 1.5467884039180861,
+      "learning_rate": 2.8458188824527558e-05,
+      "loss": 0.1308,
+      "step": 2129
+    },
+    {
+      "epoch": 7.607142857142857,
+      "grad_norm": 0.6779848218829921,
+      "learning_rate": 2.84477057347448e-05,
+      "loss": 0.0753,
+      "step": 2130
+    },
+    {
+      "epoch": 7.610714285714286,
+      "grad_norm": 0.8144294399567079,
+      "learning_rate": 2.8437219819308005e-05,
+      "loss": 0.0941,
+      "step": 2131
+    },
+    {
+      "epoch": 7.614285714285714,
+      "grad_norm": 0.823599477180809,
+      "learning_rate": 2.8426731081724576e-05,
+      "loss": 0.0364,
+      "step": 2132
+    },
+    {
+      "epoch": 7.617857142857143,
+      "grad_norm": 0.7735415683151773,
+      "learning_rate": 2.8416239525502867e-05,
+      "loss": 0.1527,
+      "step": 2133
+    },
+    {
+      "epoch": 7.621428571428572,
+      "grad_norm": 0.8705949727868263,
+      "learning_rate": 2.8405745154152174e-05,
+      "loss": 0.0807,
+      "step": 2134
+    },
+    {
+      "epoch": 7.625,
+      "grad_norm": 0.6952396747974863,
+      "learning_rate": 2.8395247971182747e-05,
+      "loss": 0.0786,
+      "step": 2135
+    },
+    {
+      "epoch": 7.628571428571428,
+      "grad_norm": 1.0499793917257698,
+      "learning_rate": 2.8384747980105755e-05,
+      "loss": 0.0673,
+      "step": 2136
+    },
+    {
+      "epoch": 7.632142857142857,
+      "grad_norm": 0.7310262028105273,
+      "learning_rate": 2.8374245184433316e-05,
+      "loss": 0.0789,
+      "step": 2137
+    },
+    {
+      "epoch": 7.635714285714286,
+      "grad_norm": 0.4725716968199061,
+      "learning_rate": 2.8363739587678483e-05,
+      "loss": 0.0611,
+      "step": 2138
+    },
+    {
+      "epoch": 7.639285714285714,
+      "grad_norm": 1.0196163774365377,
+      "learning_rate": 2.8353231193355263e-05,
+      "loss": 0.1197,
+      "step": 2139
+    },
+    {
+      "epoch": 7.642857142857143,
+      "grad_norm": 0.682520595517436,
+      "learning_rate": 2.8342720004978575e-05,
+      "loss": 0.0778,
+      "step": 2140
+    },
+    {
+      "epoch": 7.646428571428571,
+      "grad_norm": 0.5113779078148857,
+      "learning_rate": 2.8332206026064287e-05,
+      "loss": 0.0683,
+      "step": 2141
+    },
+    {
+      "epoch": 7.65,
+      "grad_norm": 0.6620552628922637,
+      "learning_rate": 2.8321689260129192e-05,
+      "loss": 0.0697,
+      "step": 2142
+    },
+    {
+      "epoch": 7.6535714285714285,
+      "grad_norm": 0.6929316855437333,
+      "learning_rate": 2.8311169710691024e-05,
+      "loss": 0.1087,
+      "step": 2143
+    },
+    {
+      "epoch": 7.6571428571428575,
+      "grad_norm": 0.4216580773785832,
+      "learning_rate": 2.830064738126844e-05,
+      "loss": 0.0347,
+      "step": 2144
+    },
+    {
+      "epoch": 7.660714285714286,
+      "grad_norm": 0.3368653073939218,
+      "learning_rate": 2.829012227538103e-05,
+      "loss": 0.0205,
+      "step": 2145
+    },
+    {
+      "epoch": 7.664285714285715,
+      "grad_norm": 0.3094684454099374,
+      "learning_rate": 2.8279594396549313e-05,
+      "loss": 0.0305,
+      "step": 2146
+    },
+    {
+      "epoch": 7.667857142857143,
+      "grad_norm": 0.9282876528708638,
+      "learning_rate": 2.8269063748294736e-05,
+      "loss": 0.0727,
+      "step": 2147
+    },
+    {
+      "epoch": 7.671428571428572,
+      "grad_norm": 0.5332949287866922,
+      "learning_rate": 2.825853033413967e-05,
+      "loss": 0.0472,
+      "step": 2148
+    },
+    {
+      "epoch": 7.675,
+      "grad_norm": 0.5844532849173837,
+      "learning_rate": 2.8247994157607415e-05,
+      "loss": 0.1028,
+      "step": 2149
+    },
+    {
+      "epoch": 7.678571428571429,
+      "grad_norm": 0.4896938129021951,
+      "learning_rate": 2.8237455222222196e-05,
+      "loss": 0.0783,
+      "step": 2150
+    },
+    {
+      "epoch": 7.682142857142857,
+      "grad_norm": 0.511959746683513,
+      "learning_rate": 2.8226913531509157e-05,
+      "loss": 0.0316,
+      "step": 2151
+    },
+    {
+      "epoch": 7.685714285714286,
+      "grad_norm": 0.4925118975901381,
+      "learning_rate": 2.821636908899436e-05,
+      "loss": 0.0431,
+      "step": 2152
+    },
+    {
+      "epoch": 7.689285714285714,
+      "grad_norm": 0.6646905098689618,
+      "learning_rate": 2.8205821898204794e-05,
+      "loss": 0.0884,
+      "step": 2153
+    },
+    {
+      "epoch": 7.692857142857143,
+      "grad_norm": 0.5082733986768463,
+      "learning_rate": 2.8195271962668366e-05,
+      "loss": 0.067,
+      "step": 2154
+    },
+    {
+      "epoch": 7.696428571428571,
+      "grad_norm": 0.7175869006773854,
+      "learning_rate": 2.8184719285913894e-05,
+      "loss": 0.1301,
+      "step": 2155
+    },
+    {
+      "epoch": 7.7,
+      "grad_norm": 0.5946031859498466,
+      "learning_rate": 2.817416387147113e-05,
+      "loss": 0.0489,
+      "step": 2156
+    },
+    {
+      "epoch": 7.703571428571428,
+      "grad_norm": 0.8040954755122298,
+      "learning_rate": 2.816360572287072e-05,
+      "loss": 0.0495,
+      "step": 2157
+    },
+    {
+      "epoch": 7.707142857142857,
+      "grad_norm": 0.48181155922746777,
+      "learning_rate": 2.8153044843644243e-05,
+      "loss": 0.0731,
+      "step": 2158
+    },
+    {
+      "epoch": 7.710714285714285,
+      "grad_norm": 0.6098658217197004,
+      "learning_rate": 2.814248123732418e-05,
+      "loss": 0.0422,
+      "step": 2159
+    },
+    {
+      "epoch": 7.714285714285714,
+      "grad_norm": 0.7813732135363861,
+      "learning_rate": 2.8131914907443926e-05,
+      "loss": 0.0386,
+      "step": 2160
+    },
+    {
+      "epoch": 7.7178571428571425,
+      "grad_norm": 0.40609129025815893,
+      "learning_rate": 2.81213458575378e-05,
+      "loss": 0.0643,
+      "step": 2161
+    },
+    {
+      "epoch": 7.7214285714285715,
+      "grad_norm": 0.7894848515949036,
+      "learning_rate": 2.8110774091141003e-05,
+      "loss": 0.071,
+      "step": 2162
+    },
+    {
+      "epoch": 7.725,
+      "grad_norm": 0.32585393828316866,
+      "learning_rate": 2.8100199611789675e-05,
+      "loss": 0.0204,
+      "step": 2163
+    },
+    {
+      "epoch": 7.728571428571429,
+      "grad_norm": 0.32364294197099996,
+      "learning_rate": 2.8089622423020837e-05,
+      "loss": 0.037,
+      "step": 2164
+    },
+    {
+      "epoch": 7.732142857142857,
+      "grad_norm": 0.9878022371256581,
+      "learning_rate": 2.8079042528372446e-05,
+      "loss": 0.082,
+      "step": 2165
+    },
+    {
+      "epoch": 7.735714285714286,
+      "grad_norm": 0.4020608331820184,
+      "learning_rate": 2.8068459931383333e-05,
+      "loss": 0.044,
+      "step": 2166
+    },
+    {
+      "epoch": 7.739285714285714,
+      "grad_norm": 0.7419820392118738,
+      "learning_rate": 2.805787463559325e-05,
+      "loss": 0.0985,
+      "step": 2167
+    },
+    {
+      "epoch": 7.742857142857143,
+      "grad_norm": 1.0154728878143366,
+      "learning_rate": 2.804728664454285e-05,
+      "loss": 0.0906,
+      "step": 2168
+    },
+    {
+      "epoch": 7.746428571428572,
+      "grad_norm": 0.9253782326628939,
+      "learning_rate": 2.803669596177369e-05,
+      "loss": 0.0814,
+      "step": 2169
+    },
+    {
+      "epoch": 7.75,
+      "grad_norm": 0.5644240073284394,
+      "learning_rate": 2.8026102590828216e-05,
+      "loss": 0.0572,
+      "step": 2170
+    },
+    {
+      "epoch": 7.753571428571428,
+      "grad_norm": 0.5801432151893517,
+      "learning_rate": 2.8015506535249785e-05,
+      "loss": 0.0819,
+      "step": 2171
+    },
+    {
+      "epoch": 7.757142857142857,
+      "grad_norm": 0.449204363799899,
+      "learning_rate": 2.8004907798582652e-05,
+      "loss": 0.0418,
+      "step": 2172
+    },
+    {
+      "epoch": 7.760714285714286,
+      "grad_norm": 0.6633797243914661,
+      "learning_rate": 2.7994306384371954e-05,
+      "loss": 0.0722,
+      "step": 2173
+    },
+    {
+      "epoch": 7.764285714285714,
+      "grad_norm": 0.9680392869003973,
+      "learning_rate": 2.7983702296163745e-05,
+      "loss": 0.082,
+      "step": 2174
+    },
+    {
+      "epoch": 7.767857142857143,
+      "grad_norm": 0.7183743445457019,
+      "learning_rate": 2.7973095537504958e-05,
+      "loss": 0.0548,
+      "step": 2175
+    },
+    {
+      "epoch": 7.771428571428571,
+      "grad_norm": 0.5187566731099855,
+      "learning_rate": 2.7962486111943424e-05,
+      "loss": 0.0556,
+      "step": 2176
+    },
+    {
+      "epoch": 7.775,
+      "grad_norm": 0.8327727632549154,
+      "learning_rate": 2.7951874023027863e-05,
+      "loss": 0.0682,
+      "step": 2177
+    },
+    {
+      "epoch": 7.7785714285714285,
+      "grad_norm": 0.7397598674486464,
+      "learning_rate": 2.794125927430789e-05,
+      "loss": 0.0721,
+      "step": 2178
+    },
+    {
+      "epoch": 7.7821428571428575,
+      "grad_norm": 0.5231215832277089,
+      "learning_rate": 2.7930641869334012e-05,
+      "loss": 0.0507,
+      "step": 2179
+    },
+    {
+      "epoch": 7.785714285714286,
+      "grad_norm": 0.481725730656144,
+      "learning_rate": 2.7920021811657617e-05,
+      "loss": 0.0531,
+      "step": 2180
+    },
+    {
+      "epoch": 7.789285714285715,
+      "grad_norm": 0.8785426462255891,
+      "learning_rate": 2.7909399104830984e-05,
+      "loss": 0.0489,
+      "step": 2181
+    },
+    {
+      "epoch": 7.792857142857143,
+      "grad_norm": 0.7936510404754399,
+      "learning_rate": 2.7898773752407282e-05,
+      "loss": 0.0879,
+      "step": 2182
+    },
+    {
+      "epoch": 7.796428571428572,
+      "grad_norm": 0.5278268182163611,
+      "learning_rate": 2.7888145757940556e-05,
+      "loss": 0.0467,
+      "step": 2183
+    },
+    {
+      "epoch": 7.8,
+      "grad_norm": 1.1630910044342737,
+      "learning_rate": 2.7877515124985745e-05,
+      "loss": 0.0926,
+      "step": 2184
+    },
+    {
+      "epoch": 7.803571428571429,
+      "grad_norm": 1.187732325002114,
+      "learning_rate": 2.7866881857098664e-05,
+      "loss": 0.1112,
+      "step": 2185
+    },
+    {
+      "epoch": 7.807142857142857,
+      "grad_norm": 0.49435141980777286,
+      "learning_rate": 2.7856245957836006e-05,
+      "loss": 0.0679,
+      "step": 2186
+    },
+    {
+      "epoch": 7.810714285714286,
+      "grad_norm": 0.829765220317081,
+      "learning_rate": 2.7845607430755365e-05,
+      "loss": 0.0728,
+      "step": 2187
+    },
+    {
+      "epoch": 7.814285714285714,
+      "grad_norm": 0.8447095074578798,
+      "learning_rate": 2.7834966279415174e-05,
+      "loss": 0.0769,
+      "step": 2188
+    },
+    {
+      "epoch": 7.817857142857143,
+      "grad_norm": 0.8264899117973687,
+      "learning_rate": 2.7824322507374783e-05,
+      "loss": 0.0757,
+      "step": 2189
+    },
+    {
+      "epoch": 7.821428571428571,
+      "grad_norm": 0.49453595645875525,
+      "learning_rate": 2.7813676118194397e-05,
+      "loss": 0.0676,
+      "step": 2190
+    },
+    {
+      "epoch": 7.825,
+      "grad_norm": 0.6655422916224029,
+      "learning_rate": 2.780302711543511e-05,
+      "loss": 0.0527,
+      "step": 2191
+    },
+    {
+      "epoch": 7.828571428571428,
+      "grad_norm": 1.018456306454061,
+      "learning_rate": 2.7792375502658875e-05,
+      "loss": 0.0866,
+      "step": 2192
+    },
+    {
+      "epoch": 7.832142857142857,
+      "grad_norm": 0.9644922686375345,
+      "learning_rate": 2.778172128342853e-05,
+      "loss": 0.0822,
+      "step": 2193
+    },
+    {
+      "epoch": 7.835714285714285,
+      "grad_norm": 0.6386580658261525,
+      "learning_rate": 2.777106446130778e-05,
+      "loss": 0.0359,
+      "step": 2194
+    },
+    {
+      "epoch": 7.839285714285714,
+      "grad_norm": 0.6472584417681567,
+      "learning_rate": 2.77604050398612e-05,
+      "loss": 0.0549,
+      "step": 2195
+    },
+    {
+      "epoch": 7.8428571428571425,
+      "grad_norm": 0.9174492831586673,
+      "learning_rate": 2.7749743022654242e-05,
+      "loss": 0.1024,
+      "step": 2196
+    },
+    {
+      "epoch": 7.8464285714285715,
+      "grad_norm": 0.5082320882672253,
+      "learning_rate": 2.7739078413253216e-05,
+      "loss": 0.0546,
+      "step": 2197
+    },
+    {
+      "epoch": 7.85,
+      "grad_norm": 0.7263390230396983,
+      "learning_rate": 2.7728411215225296e-05,
+      "loss": 0.0731,
+      "step": 2198
+    },
+    {
+      "epoch": 7.853571428571429,
+      "grad_norm": 0.9131163089081853,
+      "learning_rate": 2.771774143213855e-05,
+      "loss": 0.0598,
+      "step": 2199
+    },
+    {
+      "epoch": 7.857142857142857,
+      "grad_norm": 0.8985386124239599,
+      "learning_rate": 2.7707069067561865e-05,
+      "loss": 0.0839,
+      "step": 2200
+    },
+    {
+      "epoch": 7.860714285714286,
+      "grad_norm": 0.953174004445544,
+      "learning_rate": 2.7696394125065026e-05,
+      "loss": 0.0524,
+      "step": 2201
+    },
+    {
+      "epoch": 7.864285714285714,
+      "grad_norm": 0.43689446794356884,
+      "learning_rate": 2.7685716608218674e-05,
+      "loss": 0.0453,
+      "step": 2202
+    },
+    {
+      "epoch": 7.867857142857143,
+      "grad_norm": 0.6545595018968131,
+      "learning_rate": 2.7675036520594303e-05,
+      "loss": 0.0381,
+      "step": 2203
+    },
+    {
+      "epoch": 7.871428571428572,
+      "grad_norm": 0.6455093402166366,
+      "learning_rate": 2.766435386576427e-05,
+      "loss": 0.0279,
+      "step": 2204
+    },
+    {
+      "epoch": 7.875,
+      "grad_norm": 0.5613953167585573,
+      "learning_rate": 2.7653668647301797e-05,
+      "loss": 0.0462,
+      "step": 2205
+    },
+    {
+      "epoch": 7.878571428571428,
+      "grad_norm": 0.9157682216385818,
+      "learning_rate": 2.7642980868780963e-05,
+      "loss": 0.0857,
+      "step": 2206
+    },
+    {
+      "epoch": 7.882142857142857,
+      "grad_norm": 0.628268088636355,
+      "learning_rate": 2.7632290533776683e-05,
+      "loss": 0.1148,
+      "step": 2207
+    },
+    {
+      "epoch": 7.885714285714286,
+      "grad_norm": 0.6038489769920918,
+      "learning_rate": 2.762159764586476e-05,
+      "loss": 0.0655,
+      "step": 2208
+    },
+    {
+      "epoch": 7.889285714285714,
+      "grad_norm": 0.5834240994170908,
+      "learning_rate": 2.7610902208621825e-05,
+      "loss": 0.0727,
+      "step": 2209
+    },
+    {
+      "epoch": 7.892857142857143,
+      "grad_norm": 0.42902563058486165,
+      "learning_rate": 2.7600204225625367e-05,
+      "loss": 0.0637,
+      "step": 2210
+    },
+    {
+      "epoch": 7.896428571428571,
+      "grad_norm": 0.6098473902728715,
+      "learning_rate": 2.7589503700453742e-05,
+      "loss": 0.0622,
+      "step": 2211
+    },
+    {
+      "epoch": 7.9,
+      "grad_norm": 0.7881904500104906,
+      "learning_rate": 2.757880063668614e-05,
+      "loss": 0.07,
+      "step": 2212
+    },
+    {
+      "epoch": 7.9035714285714285,
+      "grad_norm": 0.5310553178280611,
+      "learning_rate": 2.75680950379026e-05,
+      "loss": 0.0881,
+      "step": 2213
+    },
+    {
+      "epoch": 7.9071428571428575,
+      "grad_norm": 1.0091031742155987,
+      "learning_rate": 2.7557386907684013e-05,
+      "loss": 0.0505,
+      "step": 2214
+    },
+    {
+      "epoch": 7.910714285714286,
+      "grad_norm": 0.6036068879048744,
+      "learning_rate": 2.754667624961212e-05,
+      "loss": 0.0641,
+      "step": 2215
+    },
+    {
+      "epoch": 7.914285714285715,
+      "grad_norm": 0.892785738525611,
+      "learning_rate": 2.753596306726951e-05,
+      "loss": 0.1075,
+      "step": 2216
+    },
+    {
+      "epoch": 7.917857142857143,
+      "grad_norm": 0.7927230885831497,
+      "learning_rate": 2.7525247364239604e-05,
+      "loss": 0.0692,
+      "step": 2217
+    },
+    {
+      "epoch": 7.921428571428572,
+      "grad_norm": 0.6335727028185648,
+      "learning_rate": 2.751452914410668e-05,
+      "loss": 0.0619,
+      "step": 2218
+    },
+    {
+      "epoch": 7.925,
+      "grad_norm": 0.6617938802291603,
+      "learning_rate": 2.7503808410455843e-05,
+      "loss": 0.0683,
+      "step": 2219
+    },
+    {
+      "epoch": 7.928571428571429,
+      "grad_norm": 0.5849995809298449,
+      "learning_rate": 2.7493085166873058e-05,
+      "loss": 0.0841,
+      "step": 2220
+    },
+    {
+      "epoch": 7.932142857142857,
+      "grad_norm": 0.5757447501375226,
+      "learning_rate": 2.7482359416945105e-05,
+      "loss": 0.0414,
+      "step": 2221
+    },
+    {
+      "epoch": 7.935714285714286,
+      "grad_norm": 0.4794695027830706,
+      "learning_rate": 2.7471631164259625e-05,
+      "loss": 0.0489,
+      "step": 2222
+    },
+    {
+      "epoch": 7.939285714285714,
+      "grad_norm": 0.3742321655071042,
+      "learning_rate": 2.746090041240509e-05,
+      "loss": 0.0397,
+      "step": 2223
+    },
+    {
+      "epoch": 7.942857142857143,
+      "grad_norm": 0.37173066009105604,
+      "learning_rate": 2.7450167164970793e-05,
+      "loss": 0.0506,
+      "step": 2224
+    },
+    {
+      "epoch": 7.946428571428571,
+      "grad_norm": 0.46901298243517364,
+      "learning_rate": 2.7439431425546884e-05,
+      "loss": 0.0444,
+      "step": 2225
+    },
+    {
+      "epoch": 7.95,
+      "grad_norm": 0.829151823769199,
+      "learning_rate": 2.7428693197724335e-05,
+      "loss": 0.0801,
+      "step": 2226
+    },
+    {
+      "epoch": 7.953571428571428,
+      "grad_norm": 0.506491803446384,
+      "learning_rate": 2.741795248509494e-05,
+      "loss": 0.0579,
+      "step": 2227
+    },
+    {
+      "epoch": 7.957142857142857,
+      "grad_norm": 0.44805818187923036,
+      "learning_rate": 2.7407209291251356e-05,
+      "loss": 0.0499,
+      "step": 2228
+    },
+    {
+      "epoch": 7.960714285714285,
+      "grad_norm": 0.6296061138622759,
+      "learning_rate": 2.7396463619787038e-05,
+      "loss": 0.0856,
+      "step": 2229
+    },
+    {
+      "epoch": 7.964285714285714,
+      "grad_norm": 0.9894980011218004,
+      "learning_rate": 2.7385715474296274e-05,
+      "loss": 0.092,
+      "step": 2230
+    },
+    {
+      "epoch": 7.9678571428571425,
+      "grad_norm": 0.5244511763441166,
+      "learning_rate": 2.7374964858374203e-05,
+      "loss": 0.0367,
+      "step": 2231
+    },
+    {
+      "epoch": 7.9714285714285715,
+      "grad_norm": 0.49390000919785115,
+      "learning_rate": 2.736421177561676e-05,
+      "loss": 0.0446,
+      "step": 2232
+    },
+    {
+      "epoch": 7.975,
+      "grad_norm": 0.4751716569125071,
+      "learning_rate": 2.7353456229620733e-05,
+      "loss": 0.0709,
+      "step": 2233
+    },
+    {
+      "epoch": 7.978571428571429,
+      "grad_norm": 0.4335746069550736,
+      "learning_rate": 2.7342698223983714e-05,
+      "loss": 0.0653,
+      "step": 2234
+    },
+    {
+      "epoch": 7.982142857142857,
+      "grad_norm": 0.9569457219063116,
+      "learning_rate": 2.7331937762304116e-05,
+      "loss": 0.0594,
+      "step": 2235
+    },
+    {
+      "epoch": 7.985714285714286,
+      "grad_norm": 1.2910938594067267,
+      "learning_rate": 2.73211748481812e-05,
+      "loss": 0.0736,
+      "step": 2236
+    },
+    {
+      "epoch": 7.989285714285714,
+      "grad_norm": 0.8086996874369712,
+      "learning_rate": 2.7310409485215007e-05,
+      "loss": 0.0461,
+      "step": 2237
+    },
+    {
+      "epoch": 7.992857142857143,
+      "grad_norm": 1.6879434359954626,
+      "learning_rate": 2.7299641677006436e-05,
+      "loss": 0.1095,
+      "step": 2238
+    },
+    {
+      "epoch": 7.996428571428572,
+      "grad_norm": 0.6975308517897917,
+      "learning_rate": 2.728887142715718e-05,
+      "loss": 0.0572,
+      "step": 2239
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 1.0723512884895798,
+      "learning_rate": 2.7278098739269757e-05,
+      "loss": 0.065,
+      "step": 2240
+    },
+    {
+      "epoch": 8.003571428571428,
+      "grad_norm": 0.6244073364217272,
+      "learning_rate": 2.726732361694749e-05,
+      "loss": 0.0546,
+      "step": 2241
+    },
+    {
+      "epoch": 8.007142857142858,
+      "grad_norm": 0.8928770498867531,
+      "learning_rate": 2.7256546063794545e-05,
+      "loss": 0.0754,
+      "step": 2242
+    },
+    {
+      "epoch": 8.010714285714286,
+      "grad_norm": 0.6505041614909743,
+      "learning_rate": 2.7245766083415865e-05,
+      "loss": 0.0515,
+      "step": 2243
+    },
+    {
+      "epoch": 8.014285714285714,
+      "grad_norm": 1.8611450428625402,
+      "learning_rate": 2.723498367941723e-05,
+      "loss": 0.081,
+      "step": 2244
+    },
+    {
+      "epoch": 8.017857142857142,
+      "grad_norm": 0.633041700882408,
+      "learning_rate": 2.722419885540521e-05,
+      "loss": 0.0847,
+      "step": 2245
+    },
+    {
+      "epoch": 8.021428571428572,
+      "grad_norm": 0.7702067225475056,
+      "learning_rate": 2.721341161498721e-05,
+      "loss": 0.0725,
+      "step": 2246
+    },
+    {
+      "epoch": 8.025,
+      "grad_norm": 0.6238909617694331,
+      "learning_rate": 2.7202621961771427e-05,
+      "loss": 0.0507,
+      "step": 2247
+    },
+    {
+      "epoch": 8.028571428571428,
+      "grad_norm": 0.7358439049627676,
+      "learning_rate": 2.7191829899366866e-05,
+      "loss": 0.0405,
+      "step": 2248
+    },
+    {
+      "epoch": 8.032142857142857,
+      "grad_norm": 0.7001686209565093,
+      "learning_rate": 2.7181035431383335e-05,
+      "loss": 0.0509,
+      "step": 2249
+    },
+    {
+      "epoch": 8.035714285714286,
+      "grad_norm": 0.40687952551182116,
+      "learning_rate": 2.7170238561431456e-05,
+      "loss": 0.0422,
+      "step": 2250
+    },
+    {
+      "epoch": 8.039285714285715,
+      "grad_norm": 1.4485425052923817,
+      "learning_rate": 2.715943929312265e-05,
+      "loss": 0.1133,
+      "step": 2251
+    },
+    {
+      "epoch": 8.042857142857143,
+      "grad_norm": 0.8599643381853159,
+      "learning_rate": 2.714863763006914e-05,
+      "loss": 0.0766,
+      "step": 2252
+    },
+    {
+      "epoch": 8.04642857142857,
+      "grad_norm": 0.34411032468004416,
+      "learning_rate": 2.7137833575883948e-05,
+      "loss": 0.0257,
+      "step": 2253
+    },
+    {
+      "epoch": 8.05,
+      "grad_norm": 0.5496028642522537,
+      "learning_rate": 2.71270271341809e-05,
+      "loss": 0.023,
+      "step": 2254
+    },
+    {
+      "epoch": 8.053571428571429,
+      "grad_norm": 0.5838356956854382,
+      "learning_rate": 2.7116218308574613e-05,
+      "loss": 0.0679,
+      "step": 2255
+    },
+    {
+      "epoch": 8.057142857142857,
+      "grad_norm": 0.5328969046106565,
+      "learning_rate": 2.7105407102680518e-05,
+      "loss": 0.0787,
+      "step": 2256
+    },
+    {
+      "epoch": 8.060714285714285,
+      "grad_norm": 0.5945503984200184,
+      "learning_rate": 2.709459352011482e-05,
+      "loss": 0.0455,
+      "step": 2257
+    },
+    {
+      "epoch": 8.064285714285715,
+      "grad_norm": 0.4865619877377172,
+      "learning_rate": 2.708377756449453e-05,
+      "loss": 0.0337,
+      "step": 2258
+    },
+    {
+      "epoch": 8.067857142857143,
+      "grad_norm": 1.0637049741405094,
+      "learning_rate": 2.707295923943746e-05,
+      "loss": 0.0761,
+      "step": 2259
+    },
+    {
+      "epoch": 8.071428571428571,
+      "grad_norm": 0.6273573760945412,
+      "learning_rate": 2.7062138548562203e-05,
+      "loss": 0.0531,
+      "step": 2260
+    },
+    {
+      "epoch": 8.075,
+      "grad_norm": 0.9559285317602675,
+      "learning_rate": 2.7051315495488146e-05,
+      "loss": 0.0576,
+      "step": 2261
+    },
+    {
+      "epoch": 8.07857142857143,
+      "grad_norm": 0.5518882593702803,
+      "learning_rate": 2.7040490083835468e-05,
+      "loss": 0.0735,
+      "step": 2262
+    },
+    {
+      "epoch": 8.082142857142857,
+      "grad_norm": 0.75656617048279,
+      "learning_rate": 2.7029662317225137e-05,
+      "loss": 0.0537,
+      "step": 2263
+    },
+    {
+      "epoch": 8.085714285714285,
+      "grad_norm": 0.4365387777592226,
+      "learning_rate": 2.701883219927891e-05,
+      "loss": 0.0542,
+      "step": 2264
+    },
+    {
+      "epoch": 8.089285714285714,
+      "grad_norm": 0.40995686842206164,
+      "learning_rate": 2.7007999733619335e-05,
+      "loss": 0.0335,
+      "step": 2265
+    },
+    {
+      "epoch": 8.092857142857143,
+      "grad_norm": 0.4965747822290575,
+      "learning_rate": 2.6997164923869723e-05,
+      "loss": 0.0333,
+      "step": 2266
+    },
+    {
+      "epoch": 8.096428571428572,
+      "grad_norm": 0.8590423069031194,
+      "learning_rate": 2.69863277736542e-05,
+      "loss": 0.0676,
+      "step": 2267
+    },
+    {
+      "epoch": 8.1,
+      "grad_norm": 0.4622582269054704,
+      "learning_rate": 2.6975488286597643e-05,
+      "loss": 0.0176,
+      "step": 2268
+    },
+    {
+      "epoch": 8.103571428571428,
+      "grad_norm": 0.45643547423449393,
+      "learning_rate": 2.696464646632575e-05,
+      "loss": 0.023,
+      "step": 2269
+    },
+    {
+      "epoch": 8.107142857142858,
+      "grad_norm": 0.6850416664165243,
+      "learning_rate": 2.695380231646496e-05,
+      "loss": 0.031,
+      "step": 2270
+    },
+    {
+      "epoch": 8.110714285714286,
+      "grad_norm": 1.5893607445671776,
+      "learning_rate": 2.6942955840642517e-05,
+      "loss": 0.0948,
+      "step": 2271
+    },
+    {
+      "epoch": 8.114285714285714,
+      "grad_norm": 1.4137943953668666,
+      "learning_rate": 2.693210704248643e-05,
+      "loss": 0.0484,
+      "step": 2272
+    },
+    {
+      "epoch": 8.117857142857142,
+      "grad_norm": 0.7293768204496713,
+      "learning_rate": 2.6921255925625485e-05,
+      "loss": 0.051,
+      "step": 2273
+    },
+    {
+      "epoch": 8.121428571428572,
+      "grad_norm": 0.7318423251970745,
+      "learning_rate": 2.6910402493689258e-05,
+      "loss": 0.0437,
+      "step": 2274
+    },
+    {
+      "epoch": 8.125,
+      "grad_norm": 0.3829885090402272,
+      "learning_rate": 2.689954675030808e-05,
+      "loss": 0.0523,
+      "step": 2275
+    },
+    {
+      "epoch": 8.128571428571428,
+      "grad_norm": 0.527239855561519,
+      "learning_rate": 2.688868869911307e-05,
+      "loss": 0.0614,
+      "step": 2276
+    },
+    {
+      "epoch": 8.132142857142858,
+      "grad_norm": 0.5521300241153527,
+      "learning_rate": 2.6877828343736106e-05,
+      "loss": 0.0561,
+      "step": 2277
+    },
+    {
+      "epoch": 8.135714285714286,
+      "grad_norm": 1.0547532956530423,
+      "learning_rate": 2.6866965687809855e-05,
+      "loss": 0.0335,
+      "step": 2278
+    },
+    {
+      "epoch": 8.139285714285714,
+      "grad_norm": 0.8564970492642026,
+      "learning_rate": 2.685610073496773e-05,
+      "loss": 0.0677,
+      "step": 2279
+    },
+    {
+      "epoch": 8.142857142857142,
+      "grad_norm": 0.550403737514541,
+      "learning_rate": 2.6845233488843932e-05,
+      "loss": 0.0477,
+      "step": 2280
+    },
+    {
+      "epoch": 8.146428571428572,
+      "grad_norm": 0.9790243625881927,
+      "learning_rate": 2.6834363953073423e-05,
+      "loss": 0.05,
+      "step": 2281
+    },
+    {
+      "epoch": 8.15,
+      "grad_norm": 0.7497077889685201,
+      "learning_rate": 2.6823492131291923e-05,
+      "loss": 0.0767,
+      "step": 2282
+    },
+    {
+      "epoch": 8.153571428571428,
+      "grad_norm": 1.3659578764967655,
+      "learning_rate": 2.681261802713593e-05,
+      "loss": 0.0833,
+      "step": 2283
+    },
+    {
+      "epoch": 8.157142857142857,
+      "grad_norm": 0.6295797295057135,
+      "learning_rate": 2.6801741644242696e-05,
+      "loss": 0.0625,
+      "step": 2284
+    },
+    {
+      "epoch": 8.160714285714286,
+      "grad_norm": 0.9248127145177161,
+      "learning_rate": 2.6790862986250236e-05,
+      "loss": 0.0739,
+      "step": 2285
+    },
+    {
+      "epoch": 8.164285714285715,
+      "grad_norm": 0.5309315114753757,
+      "learning_rate": 2.6779982056797322e-05,
+      "loss": 0.0784,
+      "step": 2286
+    },
+    {
+      "epoch": 8.167857142857143,
+      "grad_norm": 0.556860109765514,
+      "learning_rate": 2.67690988595235e-05,
+      "loss": 0.0704,
+      "step": 2287
+    },
+    {
+      "epoch": 8.17142857142857,
+      "grad_norm": 0.7729098991675768,
+      "learning_rate": 2.6758213398069073e-05,
+      "loss": 0.0653,
+      "step": 2288
+    },
+    {
+      "epoch": 8.175,
+      "grad_norm": 0.6087325992030439,
+      "learning_rate": 2.6747325676075077e-05,
+      "loss": 0.0618,
+      "step": 2289
+    },
+    {
+      "epoch": 8.178571428571429,
+      "grad_norm": 0.7982667097298745,
+      "learning_rate": 2.6736435697183335e-05,
+      "loss": 0.069,
+      "step": 2290
+    },
+    {
+      "epoch": 8.182142857142857,
+      "grad_norm": 1.127387507158282,
+      "learning_rate": 2.6725543465036407e-05,
+      "loss": 0.0873,
+      "step": 2291
+    },
+    {
+      "epoch": 8.185714285714285,
+      "grad_norm": 0.5005854687300633,
+      "learning_rate": 2.6714648983277605e-05,
+      "loss": 0.0344,
+      "step": 2292
+    },
+    {
+      "epoch": 8.189285714285715,
+      "grad_norm": 0.4788147335446108,
+      "learning_rate": 2.670375225555101e-05,
+      "loss": 0.0655,
+      "step": 2293
+    },
+    {
+      "epoch": 8.192857142857143,
+      "grad_norm": 0.8638374124440461,
+      "learning_rate": 2.669285328550144e-05,
+      "loss": 0.042,
+      "step": 2294
+    },
+    {
+      "epoch": 8.196428571428571,
+      "grad_norm": 0.9998483061404423,
+      "learning_rate": 2.6681952076774472e-05,
+      "loss": 0.0756,
+      "step": 2295
+    },
+    {
+      "epoch": 8.2,
+      "grad_norm": 1.2889909818797332,
+      "learning_rate": 2.6671048633016416e-05,
+      "loss": 0.1018,
+      "step": 2296
+    },
+    {
+      "epoch": 8.20357142857143,
+      "grad_norm": 0.7614205173947978,
+      "learning_rate": 2.6660142957874344e-05,
+      "loss": 0.0995,
+      "step": 2297
+    },
+    {
+      "epoch": 8.207142857142857,
+      "grad_norm": 1.0447098985046752,
+      "learning_rate": 2.6649235054996075e-05,
+      "loss": 0.0729,
+      "step": 2298
+    },
+    {
+      "epoch": 8.210714285714285,
+      "grad_norm": 0.4841373304183714,
+      "learning_rate": 2.6638324928030166e-05,
+      "loss": 0.0502,
+      "step": 2299
+    },
+    {
+      "epoch": 8.214285714285714,
+      "grad_norm": 0.861034139726897,
+      "learning_rate": 2.662741258062592e-05,
+      "loss": 0.0442,
+      "step": 2300
+    },
+    {
+      "epoch": 8.217857142857143,
+      "grad_norm": 0.7233607697649818,
+      "learning_rate": 2.6616498016433384e-05,
+      "loss": 0.0389,
+      "step": 2301
+    },
+    {
+      "epoch": 8.221428571428572,
+      "grad_norm": 0.8912100961100915,
+      "learning_rate": 2.6605581239103347e-05,
+      "loss": 0.0583,
+      "step": 2302
+    },
+    {
+      "epoch": 8.225,
+      "grad_norm": 1.0885229188718775,
+      "learning_rate": 2.6594662252287334e-05,
+      "loss": 0.0849,
+      "step": 2303
+    },
+    {
+      "epoch": 8.228571428571428,
+      "grad_norm": 0.8367141256128918,
+      "learning_rate": 2.6583741059637618e-05,
+      "loss": 0.0717,
+      "step": 2304
+    },
+    {
+      "epoch": 8.232142857142858,
+      "grad_norm": 0.8369634196196195,
+      "learning_rate": 2.65728176648072e-05,
+      "loss": 0.0606,
+      "step": 2305
+    },
+    {
+      "epoch": 8.235714285714286,
+      "grad_norm": 0.5112354122254636,
+      "learning_rate": 2.6561892071449816e-05,
+      "loss": 0.0639,
+      "step": 2306
+    },
+    {
+      "epoch": 8.239285714285714,
+      "grad_norm": 1.0022950269362025,
+      "learning_rate": 2.6550964283219962e-05,
+      "loss": 0.1083,
+      "step": 2307
+    },
+    {
+      "epoch": 8.242857142857142,
+      "grad_norm": 0.7908206695772291,
+      "learning_rate": 2.654003430377283e-05,
+      "loss": 0.0379,
+      "step": 2308
+    },
+    {
+      "epoch": 8.246428571428572,
+      "grad_norm": 0.6364178055113779,
+      "learning_rate": 2.6529102136764374e-05,
+      "loss": 0.0281,
+      "step": 2309
+    },
+    {
+      "epoch": 8.25,
+      "grad_norm": 1.1942361332232005,
+      "learning_rate": 2.6518167785851266e-05,
+      "loss": 0.106,
+      "step": 2310
+    },
+    {
+      "epoch": 8.253571428571428,
+      "grad_norm": 1.0310371203587116,
+      "learning_rate": 2.650723125469091e-05,
+      "loss": 0.0836,
+      "step": 2311
+    },
+    {
+      "epoch": 8.257142857142856,
+      "grad_norm": 1.2938526849611551,
+      "learning_rate": 2.649629254694145e-05,
+      "loss": 0.1086,
+      "step": 2312
+    },
+    {
+      "epoch": 8.260714285714286,
+      "grad_norm": 0.733465366176365,
+      "learning_rate": 2.6485351666261742e-05,
+      "loss": 0.0641,
+      "step": 2313
+    },
+    {
+      "epoch": 8.264285714285714,
+      "grad_norm": 0.6792440641298911,
+      "learning_rate": 2.6474408616311378e-05,
+      "loss": 0.0587,
+      "step": 2314
+    },
+    {
+      "epoch": 8.267857142857142,
+      "grad_norm": 0.8334190334985002,
+      "learning_rate": 2.6463463400750678e-05,
+      "loss": 0.1037,
+      "step": 2315
+    },
+    {
+      "epoch": 8.271428571428572,
+      "grad_norm": 0.36451803732106325,
+      "learning_rate": 2.6452516023240678e-05,
+      "loss": 0.0401,
+      "step": 2316
+    },
+    {
+      "epoch": 8.275,
+      "grad_norm": 0.4152114563773884,
+      "learning_rate": 2.6441566487443143e-05,
+      "loss": 0.0251,
+      "step": 2317
+    },
+    {
+      "epoch": 8.278571428571428,
+      "grad_norm": 0.3902322389044926,
+      "learning_rate": 2.643061479702056e-05,
+      "loss": 0.0453,
+      "step": 2318
+    },
+    {
+      "epoch": 8.282142857142857,
+      "grad_norm": 0.4344725449960206,
+      "learning_rate": 2.641966095563613e-05,
+      "loss": 0.0433,
+      "step": 2319
+    },
+    {
+      "epoch": 8.285714285714286,
+      "grad_norm": 0.44456797399493236,
+      "learning_rate": 2.6408704966953796e-05,
+      "loss": 0.0615,
+      "step": 2320
+    },
+    {
+      "epoch": 8.289285714285715,
+      "grad_norm": 0.5171640079411556,
+      "learning_rate": 2.6397746834638184e-05,
+      "loss": 0.0562,
+      "step": 2321
+    },
+    {
+      "epoch": 8.292857142857143,
+      "grad_norm": 0.4688691978750208,
+      "learning_rate": 2.6386786562354655e-05,
+      "loss": 0.0352,
+      "step": 2322
+    },
+    {
+      "epoch": 8.29642857142857,
+      "grad_norm": 0.6164675073031614,
+      "learning_rate": 2.637582415376929e-05,
+      "loss": 0.106,
+      "step": 2323
+    },
+    {
+      "epoch": 8.3,
+      "grad_norm": 0.7245728215296077,
+      "learning_rate": 2.6364859612548884e-05,
+      "loss": 0.1091,
+      "step": 2324
+    },
+    {
+      "epoch": 8.303571428571429,
+      "grad_norm": 0.5189380265961911,
+      "learning_rate": 2.6353892942360935e-05,
+      "loss": 0.0486,
+      "step": 2325
+    },
+    {
+      "epoch": 8.307142857142857,
+      "grad_norm": 0.5290909691392081,
+      "learning_rate": 2.6342924146873667e-05,
+      "loss": 0.0557,
+      "step": 2326
+    },
+    {
+      "epoch": 8.310714285714285,
+      "grad_norm": 0.6628734227182588,
+      "learning_rate": 2.6331953229756e-05,
+      "loss": 0.092,
+      "step": 2327
+    },
+    {
+      "epoch": 8.314285714285715,
+      "grad_norm": 1.0577735582745567,
+      "learning_rate": 2.632098019467758e-05,
+      "loss": 0.0755,
+      "step": 2328
+    },
+    {
+      "epoch": 8.317857142857143,
+      "grad_norm": 0.5314270287175397,
+      "learning_rate": 2.6310005045308745e-05,
+      "loss": 0.0572,
+      "step": 2329
+    },
+    {
+      "epoch": 8.321428571428571,
+      "grad_norm": 0.7616868730364257,
+      "learning_rate": 2.629902778532055e-05,
+      "loss": 0.0558,
+      "step": 2330
+    },
+    {
+      "epoch": 8.325,
+      "grad_norm": 0.6389136885798141,
+      "learning_rate": 2.6288048418384756e-05,
+      "loss": 0.091,
+      "step": 2331
+    },
+    {
+      "epoch": 8.32857142857143,
+      "grad_norm": 0.48957654391826694,
+      "learning_rate": 2.6277066948173834e-05,
+      "loss": 0.0695,
+      "step": 2332
+    },
+    {
+      "epoch": 8.332142857142857,
+      "grad_norm": 0.5709270505585573,
+      "learning_rate": 2.6266083378360933e-05,
+      "loss": 0.1057,
+      "step": 2333
+    },
+    {
+      "epoch": 8.335714285714285,
+      "grad_norm": 0.533948809534112,
+      "learning_rate": 2.6255097712619932e-05,
+      "loss": 0.0599,
+      "step": 2334
+    },
+    {
+      "epoch": 8.339285714285714,
+      "grad_norm": 0.9732285073412356,
+      "learning_rate": 2.6244109954625406e-05,
+      "loss": 0.0847,
+      "step": 2335
+    },
+    {
+      "epoch": 8.342857142857143,
+      "grad_norm": 0.8029033434184046,
+      "learning_rate": 2.623312010805262e-05,
+      "loss": 0.0534,
+      "step": 2336
+    },
+    {
+      "epoch": 8.346428571428572,
+      "grad_norm": 0.6813811395406109,
+      "learning_rate": 2.6222128176577544e-05,
+      "loss": 0.0837,
+      "step": 2337
+    },
+    {
+      "epoch": 8.35,
+      "grad_norm": 0.6763490111162483,
+      "learning_rate": 2.6211134163876847e-05,
+      "loss": 0.1172,
+      "step": 2338
+    },
+    {
+      "epoch": 8.353571428571428,
+      "grad_norm": 0.8111412211806007,
+      "learning_rate": 2.620013807362789e-05,
+      "loss": 0.1265,
+      "step": 2339
+    },
+    {
+      "epoch": 8.357142857142858,
+      "grad_norm": 0.6075784404285682,
+      "learning_rate": 2.618913990950873e-05,
+      "loss": 0.0484,
+      "step": 2340
+    },
+    {
+      "epoch": 8.360714285714286,
+      "grad_norm": 1.0537298019802444,
+      "learning_rate": 2.6178139675198117e-05,
+      "loss": 0.0948,
+      "step": 2341
+    },
+    {
+      "epoch": 8.364285714285714,
+      "grad_norm": 0.611642234049137,
+      "learning_rate": 2.616713737437549e-05,
+      "loss": 0.0616,
+      "step": 2342
+    },
+    {
+      "epoch": 8.367857142857144,
+      "grad_norm": 0.6255027819939633,
+      "learning_rate": 2.6156133010720996e-05,
+      "loss": 0.0933,
+      "step": 2343
+    },
+    {
+      "epoch": 8.371428571428572,
+      "grad_norm": 0.9305794053561403,
+      "learning_rate": 2.614512658791545e-05,
+      "loss": 0.0602,
+      "step": 2344
+    },
+    {
+      "epoch": 8.375,
+      "grad_norm": 0.7668586825954715,
+      "learning_rate": 2.613411810964037e-05,
+      "loss": 0.0717,
+      "step": 2345
+    },
+    {
+      "epoch": 8.378571428571428,
+      "grad_norm": 0.7597416603371199,
+      "learning_rate": 2.6123107579577954e-05,
+      "loss": 0.0582,
+      "step": 2346
+    },
+    {
+      "epoch": 8.382142857142856,
+      "grad_norm": 1.3005109307783933,
+      "learning_rate": 2.611209500141109e-05,
+      "loss": 0.1142,
+      "step": 2347
+    },
+    {
+      "epoch": 8.385714285714286,
+      "grad_norm": 0.5158872191614323,
+      "learning_rate": 2.610108037882335e-05,
+      "loss": 0.0722,
+      "step": 2348
+    },
+    {
+      "epoch": 8.389285714285714,
+      "grad_norm": 1.0578379140389276,
+      "learning_rate": 2.6090063715498998e-05,
+      "loss": 0.0846,
+      "step": 2349
+    },
+    {
+      "epoch": 8.392857142857142,
+      "grad_norm": 0.4924409228601598,
+      "learning_rate": 2.6079045015122962e-05,
+      "loss": 0.0468,
+      "step": 2350
+    },
+    {
+      "epoch": 8.396428571428572,
+      "grad_norm": 0.8501966682807108,
+      "learning_rate": 2.606802428138087e-05,
+      "loss": 0.0475,
+      "step": 2351
+    },
+    {
+      "epoch": 8.4,
+      "grad_norm": 0.5801320331151759,
+      "learning_rate": 2.6057001517959015e-05,
+      "loss": 0.0755,
+      "step": 2352
+    },
+    {
+      "epoch": 8.403571428571428,
+      "grad_norm": 0.8083734231590455,
+      "learning_rate": 2.6045976728544383e-05,
+      "loss": 0.0573,
+      "step": 2353
+    },
+    {
+      "epoch": 8.407142857142857,
+      "grad_norm": 0.9310925068905402,
+      "learning_rate": 2.603494991682463e-05,
+      "loss": 0.0803,
+      "step": 2354
+    },
+    {
+      "epoch": 8.410714285714286,
+      "grad_norm": 0.9154050704685205,
+      "learning_rate": 2.602392108648809e-05,
+      "loss": 0.0587,
+      "step": 2355
+    },
+    {
+      "epoch": 8.414285714285715,
+      "grad_norm": 1.1966780798798364,
+      "learning_rate": 2.6012890241223764e-05,
+      "loss": 0.1161,
+      "step": 2356
+    },
+    {
+      "epoch": 8.417857142857143,
+      "grad_norm": 0.7247974312864167,
+      "learning_rate": 2.600185738472134e-05,
+      "loss": 0.0537,
+      "step": 2357
+    },
+    {
+      "epoch": 8.42142857142857,
+      "grad_norm": 1.3547442561962724,
+      "learning_rate": 2.5990822520671172e-05,
+      "loss": 0.1123,
+      "step": 2358
+    },
+    {
+      "epoch": 8.425,
+      "grad_norm": 1.06229197657291,
+      "learning_rate": 2.5979785652764288e-05,
+      "loss": 0.0705,
+      "step": 2359
+    },
+    {
+      "epoch": 8.428571428571429,
+      "grad_norm": 1.2140247821165226,
+      "learning_rate": 2.596874678469239e-05,
+      "loss": 0.0866,
+      "step": 2360
+    },
+    {
+      "epoch": 8.432142857142857,
+      "grad_norm": 1.3353968779703984,
+      "learning_rate": 2.5957705920147836e-05,
+      "loss": 0.1095,
+      "step": 2361
+    },
+    {
+      "epoch": 8.435714285714285,
+      "grad_norm": 0.6954107787068609,
+      "learning_rate": 2.5946663062823663e-05,
+      "loss": 0.0588,
+      "step": 2362
+    },
+    {
+      "epoch": 8.439285714285715,
+      "grad_norm": 0.608647916052578,
+      "learning_rate": 2.5935618216413573e-05,
+      "loss": 0.0396,
+      "step": 2363
+    },
+    {
+      "epoch": 8.442857142857143,
+      "grad_norm": 0.6982542874715005,
+      "learning_rate": 2.5924571384611928e-05,
+      "loss": 0.1049,
+      "step": 2364
+    },
+    {
+      "epoch": 8.446428571428571,
+      "grad_norm": 0.5682973692685659,
+      "learning_rate": 2.5913522571113766e-05,
+      "loss": 0.058,
+      "step": 2365
+    },
+    {
+      "epoch": 8.45,
+      "grad_norm": 0.6160380347643004,
+      "learning_rate": 2.590247177961477e-05,
+      "loss": 0.0612,
+      "step": 2366
+    },
+    {
+      "epoch": 8.45357142857143,
+      "grad_norm": 0.7165632375096864,
+      "learning_rate": 2.5891419013811295e-05,
+      "loss": 0.044,
+      "step": 2367
+    },
+    {
+      "epoch": 8.457142857142857,
+      "grad_norm": 0.7185451153029216,
+      "learning_rate": 2.588036427740036e-05,
+      "loss": 0.0876,
+      "step": 2368
+    },
+    {
+      "epoch": 8.460714285714285,
+      "grad_norm": 0.6881230929656619,
+      "learning_rate": 2.5869307574079642e-05,
+      "loss": 0.0769,
+      "step": 2369
+    },
+    {
+      "epoch": 8.464285714285714,
+      "grad_norm": 0.7456095646942129,
+      "learning_rate": 2.585824890754747e-05,
+      "loss": 0.0592,
+      "step": 2370
+    },
+    {
+      "epoch": 8.467857142857143,
+      "grad_norm": 0.5407314294313816,
+      "learning_rate": 2.5847188281502824e-05,
+      "loss": 0.0374,
+      "step": 2371
+    },
+    {
+      "epoch": 8.471428571428572,
+      "grad_norm": 0.6918704630689249,
+      "learning_rate": 2.583612569964536e-05,
+      "loss": 0.0653,
+      "step": 2372
+    },
+    {
+      "epoch": 8.475,
+      "grad_norm": 0.8130904260251933,
+      "learning_rate": 2.5825061165675366e-05,
+      "loss": 0.0958,
+      "step": 2373
+    },
+    {
+      "epoch": 8.478571428571428,
+      "grad_norm": 0.4720903923253732,
+      "learning_rate": 2.58139946832938e-05,
+      "loss": 0.0366,
+      "step": 2374
+    },
+    {
+      "epoch": 8.482142857142858,
+      "grad_norm": 1.0491369727177693,
+      "learning_rate": 2.580292625620226e-05,
+      "loss": 0.0986,
+      "step": 2375
+    },
+    {
+      "epoch": 8.485714285714286,
+      "grad_norm": 0.7536922109151479,
+      "learning_rate": 2.5791855888103006e-05,
+      "loss": 0.0656,
+      "step": 2376
+    },
+    {
+      "epoch": 8.489285714285714,
+      "grad_norm": 0.5874959412009565,
+      "learning_rate": 2.5780783582698937e-05,
+      "loss": 0.0536,
+      "step": 2377
+    },
+    {
+      "epoch": 8.492857142857144,
+      "grad_norm": 0.5226561460170586,
+      "learning_rate": 2.57697093436936e-05,
+      "loss": 0.0522,
+      "step": 2378
+    },
+    {
+      "epoch": 8.496428571428572,
+      "grad_norm": 0.36018070251199863,
+      "learning_rate": 2.5758633174791196e-05,
+      "loss": 0.074,
+      "step": 2379
+    },
+    {
+      "epoch": 8.5,
+      "grad_norm": 0.5020876082864432,
+      "learning_rate": 2.574755507969657e-05,
+      "loss": 0.0485,
+      "step": 2380
+    },
+    {
+      "epoch": 8.503571428571428,
+      "grad_norm": 0.4968538649149364,
+      "learning_rate": 2.5736475062115198e-05,
+      "loss": 0.0587,
+      "step": 2381
+    },
+    {
+      "epoch": 8.507142857142856,
+      "grad_norm": 0.6165316958073126,
+      "learning_rate": 2.5725393125753222e-05,
+      "loss": 0.061,
+      "step": 2382
+    },
+    {
+      "epoch": 8.510714285714286,
+      "grad_norm": 0.5213572212838676,
+      "learning_rate": 2.5714309274317414e-05,
+      "loss": 0.0384,
+      "step": 2383
+    },
+    {
+      "epoch": 8.514285714285714,
+      "grad_norm": 0.6181301434191901,
+      "learning_rate": 2.5703223511515175e-05,
+      "loss": 0.0405,
+      "step": 2384
+    },
+    {
+      "epoch": 8.517857142857142,
+      "grad_norm": 0.6063203455412464,
+      "learning_rate": 2.5692135841054565e-05,
+      "loss": 0.0352,
+      "step": 2385
+    },
+    {
+      "epoch": 8.521428571428572,
+      "grad_norm": 0.9592316300195228,
+      "learning_rate": 2.568104626664428e-05,
+      "loss": 0.1074,
+      "step": 2386
+    },
+    {
+      "epoch": 8.525,
+      "grad_norm": 0.9935329515784529,
+      "learning_rate": 2.5669954791993635e-05,
+      "loss": 0.1321,
+      "step": 2387
+    },
+    {
+      "epoch": 8.528571428571428,
+      "grad_norm": 0.5503228449126264,
+      "learning_rate": 2.5658861420812595e-05,
+      "loss": 0.0579,
+      "step": 2388
+    },
+    {
+      "epoch": 8.532142857142857,
+      "grad_norm": 0.5968485790147477,
+      "learning_rate": 2.5647766156811757e-05,
+      "loss": 0.0474,
+      "step": 2389
+    },
+    {
+      "epoch": 8.535714285714286,
+      "grad_norm": 1.0549403044369796,
+      "learning_rate": 2.563666900370236e-05,
+      "loss": 0.1094,
+      "step": 2390
+    },
+    {
+      "epoch": 8.539285714285715,
+      "grad_norm": 0.8743939696512325,
+      "learning_rate": 2.562556996519626e-05,
+      "loss": 0.1045,
+      "step": 2391
+    },
+    {
+      "epoch": 8.542857142857143,
+      "grad_norm": 0.4960796422130419,
+      "learning_rate": 2.561446904500594e-05,
+      "loss": 0.0682,
+      "step": 2392
+    },
+    {
+      "epoch": 8.54642857142857,
+      "grad_norm": 1.1238094198274147,
+      "learning_rate": 2.5603366246844533e-05,
+      "loss": 0.0941,
+      "step": 2393
+    },
+    {
+      "epoch": 8.55,
+      "grad_norm": 0.8846884266764115,
+      "learning_rate": 2.559226157442578e-05,
+      "loss": 0.0738,
+      "step": 2394
+    },
+    {
+      "epoch": 8.553571428571429,
+      "grad_norm": 0.8529912179302127,
+      "learning_rate": 2.5581155031464067e-05,
+      "loss": 0.0566,
+      "step": 2395
+    },
+    {
+      "epoch": 8.557142857142857,
+      "grad_norm": 0.7590744636016832,
+      "learning_rate": 2.5570046621674397e-05,
+      "loss": 0.0807,
+      "step": 2396
+    },
+    {
+      "epoch": 8.560714285714285,
+      "grad_norm": 1.1757993402665674,
+      "learning_rate": 2.5558936348772388e-05,
+      "loss": 0.0818,
+      "step": 2397
+    },
+    {
+      "epoch": 8.564285714285715,
+      "grad_norm": 1.2515671009799254,
+      "learning_rate": 2.5547824216474295e-05,
+      "loss": 0.0755,
+      "step": 2398
+    },
+    {
+      "epoch": 8.567857142857143,
+      "grad_norm": 0.7587264197859337,
+      "learning_rate": 2.5536710228496986e-05,
+      "loss": 0.0725,
+      "step": 2399
+    },
+    {
+      "epoch": 8.571428571428571,
+      "grad_norm": 0.7914449802872751,
+      "learning_rate": 2.552559438855797e-05,
+      "loss": 0.0666,
+      "step": 2400
+    },
+    {
+      "epoch": 8.575,
+      "grad_norm": 0.5556372326650835,
+      "learning_rate": 2.5514476700375337e-05,
+      "loss": 0.061,
+      "step": 2401
+    },
+    {
+      "epoch": 8.57857142857143,
+      "grad_norm": 0.3531775540804425,
+      "learning_rate": 2.5503357167667827e-05,
+      "loss": 0.0229,
+      "step": 2402
+    },
+    {
+      "epoch": 8.582142857142857,
+      "grad_norm": 0.39291670470427537,
+      "learning_rate": 2.5492235794154793e-05,
+      "loss": 0.0434,
+      "step": 2403
+    },
+    {
+      "epoch": 8.585714285714285,
+      "grad_norm": 0.7007395049127855,
+      "learning_rate": 2.5481112583556192e-05,
+      "loss": 0.0677,
+      "step": 2404
+    },
+    {
+      "epoch": 8.589285714285714,
+      "grad_norm": 0.6386837514909929,
+      "learning_rate": 2.54699875395926e-05,
+      "loss": 0.0408,
+      "step": 2405
+    },
+    {
+      "epoch": 8.592857142857143,
+      "grad_norm": 0.8135684770789139,
+      "learning_rate": 2.545886066598521e-05,
+      "loss": 0.0677,
+      "step": 2406
+    },
+    {
+      "epoch": 8.596428571428572,
+      "grad_norm": 0.5841576561631285,
+      "learning_rate": 2.544773196645583e-05,
+      "loss": 0.0291,
+      "step": 2407
+    },
+    {
+      "epoch": 8.6,
+      "grad_norm": 0.5960647005505993,
+      "learning_rate": 2.5436601444726862e-05,
+      "loss": 0.0263,
+      "step": 2408
+    },
+    {
+      "epoch": 8.603571428571428,
+      "grad_norm": 0.5030499036287768,
+      "learning_rate": 2.5425469104521343e-05,
+      "loss": 0.0611,
+      "step": 2409
+    },
+    {
+      "epoch": 8.607142857142858,
+      "grad_norm": 1.0901337120175,
+      "learning_rate": 2.5414334949562892e-05,
+      "loss": 0.0655,
+      "step": 2410
+    },
+    {
+      "epoch": 8.610714285714286,
+      "grad_norm": 0.7716130666142421,
+      "learning_rate": 2.5403198983575748e-05,
+      "loss": 0.0956,
+      "step": 2411
+    },
+    {
+      "epoch": 8.614285714285714,
+      "grad_norm": 0.8426466097166225,
+      "learning_rate": 2.539206121028477e-05,
+      "loss": 0.088,
+      "step": 2412
+    },
+    {
+      "epoch": 8.617857142857144,
+      "grad_norm": 0.4921658780584418,
+      "learning_rate": 2.538092163341539e-05,
+      "loss": 0.0651,
+      "step": 2413
+    },
+    {
+      "epoch": 8.621428571428572,
+      "grad_norm": 0.4972993917366206,
+      "learning_rate": 2.5369780256693668e-05,
+      "loss": 0.0697,
+      "step": 2414
+    },
+    {
+      "epoch": 8.625,
+      "grad_norm": 0.6487957756547671,
+      "learning_rate": 2.5358637083846263e-05,
+      "loss": 0.088,
+      "step": 2415
+    },
+    {
+      "epoch": 8.628571428571428,
+      "grad_norm": 0.49310123602873107,
+      "learning_rate": 2.5347492118600417e-05,
+      "loss": 0.0407,
+      "step": 2416
+    },
+    {
+      "epoch": 8.632142857142856,
+      "grad_norm": 0.4756276510889417,
+      "learning_rate": 2.533634536468399e-05,
+      "loss": 0.0819,
+      "step": 2417
+    },
+    {
+      "epoch": 8.635714285714286,
+      "grad_norm": 0.6590125319683003,
+      "learning_rate": 2.5325196825825433e-05,
+      "loss": 0.0737,
+      "step": 2418
+    },
+    {
+      "epoch": 8.639285714285714,
+      "grad_norm": 0.7840365965756044,
+      "learning_rate": 2.5314046505753802e-05,
+      "loss": 0.0756,
+      "step": 2419
+    },
+    {
+      "epoch": 8.642857142857142,
+      "grad_norm": 0.4288628155293837,
+      "learning_rate": 2.5302894408198738e-05,
+      "loss": 0.0192,
+      "step": 2420
+    },
+    {
+      "epoch": 8.646428571428572,
+      "grad_norm": 0.768547118430207,
+      "learning_rate": 2.5291740536890478e-05,
+      "loss": 0.077,
+      "step": 2421
+    },
+    {
+      "epoch": 8.65,
+      "grad_norm": 1.1383029603798185,
+      "learning_rate": 2.5280584895559864e-05,
+      "loss": 0.1078,
+      "step": 2422
+    },
+    {
+      "epoch": 8.653571428571428,
+      "grad_norm": 0.6550926278473135,
+      "learning_rate": 2.5269427487938313e-05,
+      "loss": 0.0407,
+      "step": 2423
+    },
+    {
+      "epoch": 8.657142857142857,
+      "grad_norm": 0.504774915353213,
+      "learning_rate": 2.5258268317757838e-05,
+      "loss": 0.0498,
+      "step": 2424
+    },
+    {
+      "epoch": 8.660714285714286,
+      "grad_norm": 1.2989941740464388,
+      "learning_rate": 2.5247107388751063e-05,
+      "loss": 0.0939,
+      "step": 2425
+    },
+    {
+      "epoch": 8.664285714285715,
+      "grad_norm": 1.3332501514066737,
+      "learning_rate": 2.523594470465116e-05,
+      "loss": 0.0672,
+      "step": 2426
+    },
+    {
+      "epoch": 8.667857142857143,
+      "grad_norm": 0.9108839142425146,
+      "learning_rate": 2.5224780269191924e-05,
+      "loss": 0.0895,
+      "step": 2427
+    },
+    {
+      "epoch": 8.67142857142857,
+      "grad_norm": 0.7489868085010597,
+      "learning_rate": 2.5213614086107713e-05,
+      "loss": 0.0494,
+      "step": 2428
+    },
+    {
+      "epoch": 8.675,
+      "grad_norm": 0.9300874525469229,
+      "learning_rate": 2.520244615913349e-05,
+      "loss": 0.1148,
+      "step": 2429
+    },
+    {
+      "epoch": 8.678571428571429,
+      "grad_norm": 0.977484989409436,
+      "learning_rate": 2.519127649200477e-05,
+      "loss": 0.1362,
+      "step": 2430
+    },
+    {
+      "epoch": 8.682142857142857,
+      "grad_norm": 0.672526816767021,
+      "learning_rate": 2.5180105088457697e-05,
+      "loss": 0.0435,
+      "step": 2431
+    },
+    {
+      "epoch": 8.685714285714285,
+      "grad_norm": 0.4695592862487934,
+      "learning_rate": 2.516893195222894e-05,
+      "loss": 0.0564,
+      "step": 2432
+    },
+    {
+      "epoch": 8.689285714285715,
+      "grad_norm": 0.5148089798797766,
+      "learning_rate": 2.5157757087055797e-05,
+      "loss": 0.0543,
+      "step": 2433
+    },
+    {
+      "epoch": 8.692857142857143,
+      "grad_norm": 0.921836302892691,
+      "learning_rate": 2.5146580496676118e-05,
+      "loss": 0.0779,
+      "step": 2434
+    },
+    {
+      "epoch": 8.696428571428571,
+      "grad_norm": 0.4481718694755972,
+      "learning_rate": 2.5135402184828334e-05,
+      "loss": 0.0422,
+      "step": 2435
+    },
+    {
+      "epoch": 8.7,
+      "grad_norm": 1.1059593463814525,
+      "learning_rate": 2.5124222155251445e-05,
+      "loss": 0.0788,
+      "step": 2436
+    },
+    {
+      "epoch": 8.70357142857143,
+      "grad_norm": 0.7485260348952466,
+      "learning_rate": 2.511304041168505e-05,
+      "loss": 0.0705,
+      "step": 2437
+    },
+    {
+      "epoch": 8.707142857142857,
+      "grad_norm": 0.5185190036317168,
+      "learning_rate": 2.51018569578693e-05,
+      "loss": 0.0573,
+      "step": 2438
+    },
+    {
+      "epoch": 8.710714285714285,
+      "grad_norm": 0.45093257920296353,
+      "learning_rate": 2.5090671797544913e-05,
+      "loss": 0.0724,
+      "step": 2439
+    },
+    {
+      "epoch": 8.714285714285714,
+      "grad_norm": 0.5717613439292226,
+      "learning_rate": 2.50794849344532e-05,
+      "loss": 0.0332,
+      "step": 2440
+    },
+    {
+      "epoch": 8.717857142857143,
+      "grad_norm": 0.7338363847679655,
+      "learning_rate": 2.5068296372336027e-05,
+      "loss": 0.0602,
+      "step": 2441
+    },
+    {
+      "epoch": 8.721428571428572,
+      "grad_norm": 0.5185095506198661,
+      "learning_rate": 2.5057106114935822e-05,
+      "loss": 0.0303,
+      "step": 2442
+    },
+    {
+      "epoch": 8.725,
+      "grad_norm": 0.6621629890314719,
+      "learning_rate": 2.50459141659956e-05,
+      "loss": 0.0599,
+      "step": 2443
+    },
+    {
+      "epoch": 8.728571428571428,
+      "grad_norm": 0.2648607603259433,
+      "learning_rate": 2.5034720529258926e-05,
+      "loss": 0.0264,
+      "step": 2444
+    },
+    {
+      "epoch": 8.732142857142858,
+      "grad_norm": 0.7087477142564769,
+      "learning_rate": 2.502352520846994e-05,
+      "loss": 0.0555,
+      "step": 2445
+    },
+    {
+      "epoch": 8.735714285714286,
+      "grad_norm": 0.5151561737906561,
+      "learning_rate": 2.5012328207373328e-05,
+      "loss": 0.0594,
+      "step": 2446
+    },
+    {
+      "epoch": 8.739285714285714,
+      "grad_norm": 0.5480862726067176,
+      "learning_rate": 2.5001129529714357e-05,
+      "loss": 0.0541,
+      "step": 2447
+    },
+    {
+      "epoch": 8.742857142857144,
+      "grad_norm": 1.2195702601826954,
+      "learning_rate": 2.498992917923885e-05,
+      "loss": 0.0965,
+      "step": 2448
+    },
+    {
+      "epoch": 8.746428571428572,
+      "grad_norm": 0.7642105881133852,
+      "learning_rate": 2.4978727159693187e-05,
+      "loss": 0.0767,
+      "step": 2449
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 0.5483057408712694,
+      "learning_rate": 2.49675234748243e-05,
+      "loss": 0.0638,
+      "step": 2450
+    },
+    {
+      "epoch": 8.753571428571428,
+      "grad_norm": 0.8602826190502453,
+      "learning_rate": 2.4956318128379688e-05,
+      "loss": 0.0486,
+      "step": 2451
+    },
+    {
+      "epoch": 8.757142857142856,
+      "grad_norm": 0.9044720351857859,
+      "learning_rate": 2.4945111124107396e-05,
+      "loss": 0.0471,
+      "step": 2452
+    },
+    {
+      "epoch": 8.760714285714286,
+      "grad_norm": 0.7049029336459119,
+      "learning_rate": 2.4933902465756042e-05,
+      "loss": 0.0587,
+      "step": 2453
+    },
+    {
+      "epoch": 8.764285714285714,
+      "grad_norm": 0.7120499138697716,
+      "learning_rate": 2.492269215707477e-05,
+      "loss": 0.0874,
+      "step": 2454
+    },
+    {
+      "epoch": 8.767857142857142,
+      "grad_norm": 0.8626105038548666,
+      "learning_rate": 2.491148020181331e-05,
+      "loss": 0.0672,
+      "step": 2455
+    },
+    {
+      "epoch": 8.771428571428572,
+      "grad_norm": 0.9323629838897878,
+      "learning_rate": 2.4900266603721903e-05,
+      "loss": 0.0636,
+      "step": 2456
+    },
+    {
+      "epoch": 8.775,
+      "grad_norm": 0.7355796141658046,
+      "learning_rate": 2.4889051366551375e-05,
+      "loss": 0.0478,
+      "step": 2457
+    },
+    {
+      "epoch": 8.778571428571428,
+      "grad_norm": 1.0335614836099312,
+      "learning_rate": 2.487783449405308e-05,
+      "loss": 0.1102,
+      "step": 2458
+    },
+    {
+      "epoch": 8.782142857142857,
+      "grad_norm": 0.9615992916358482,
+      "learning_rate": 2.4866615989978916e-05,
+      "loss": 0.0663,
+      "step": 2459
+    },
+    {
+      "epoch": 8.785714285714286,
+      "grad_norm": 1.4535967943956325,
+      "learning_rate": 2.4855395858081353e-05,
+      "loss": 0.1047,
+      "step": 2460
+    },
+    {
+      "epoch": 8.789285714285715,
+      "grad_norm": 0.6127344850400973,
+      "learning_rate": 2.4844174102113373e-05,
+      "loss": 0.0865,
+      "step": 2461
+    },
+    {
+      "epoch": 8.792857142857143,
+      "grad_norm": 0.7494682874118054,
+      "learning_rate": 2.4832950725828527e-05,
+      "loss": 0.0916,
+      "step": 2462
+    },
+    {
+      "epoch": 8.79642857142857,
+      "grad_norm": 1.0066393173814903,
+      "learning_rate": 2.4821725732980887e-05,
+      "loss": 0.0729,
+      "step": 2463
+    },
+    {
+      "epoch": 8.8,
+      "grad_norm": 0.9560873228908882,
+      "learning_rate": 2.4810499127325077e-05,
+      "loss": 0.0811,
+      "step": 2464
+    },
+    {
+      "epoch": 8.803571428571429,
+      "grad_norm": 0.7805846479194406,
+      "learning_rate": 2.479927091261626e-05,
+      "loss": 0.0605,
+      "step": 2465
+    },
+    {
+      "epoch": 8.807142857142857,
+      "grad_norm": 0.6268345126991758,
+      "learning_rate": 2.478804109261014e-05,
+      "loss": 0.0663,
+      "step": 2466
+    },
+    {
+      "epoch": 8.810714285714285,
+      "grad_norm": 0.36167194878211845,
+      "learning_rate": 2.4776809671062955e-05,
+      "loss": 0.0304,
+      "step": 2467
+    },
+    {
+      "epoch": 8.814285714285715,
+      "grad_norm": 0.48923847445562557,
+      "learning_rate": 2.476557665173147e-05,
+      "loss": 0.0566,
+      "step": 2468
+    },
+    {
+      "epoch": 8.817857142857143,
+      "grad_norm": 0.3329231017774557,
+      "learning_rate": 2.4754342038373e-05,
+      "loss": 0.0386,
+      "step": 2469
+    },
+    {
+      "epoch": 8.821428571428571,
+      "grad_norm": 0.9199821591366284,
+      "learning_rate": 2.4743105834745376e-05,
+      "loss": 0.0899,
+      "step": 2470
+    },
+    {
+      "epoch": 8.825,
+      "grad_norm": 0.9422780357925594,
+      "learning_rate": 2.4731868044606983e-05,
+      "loss": 0.0742,
+      "step": 2471
+    },
+    {
+      "epoch": 8.82857142857143,
+      "grad_norm": 1.0334638330629597,
+      "learning_rate": 2.472062867171672e-05,
+      "loss": 0.0695,
+      "step": 2472
+    },
+    {
+      "epoch": 8.832142857142857,
+      "grad_norm": 0.33352554765975495,
+      "learning_rate": 2.470938771983401e-05,
+      "loss": 0.0368,
+      "step": 2473
+    },
+    {
+      "epoch": 8.835714285714285,
+      "grad_norm": 0.6156564846374802,
+      "learning_rate": 2.469814519271883e-05,
+      "loss": 0.0632,
+      "step": 2474
+    },
+    {
+      "epoch": 8.839285714285714,
+      "grad_norm": 0.48505703734506656,
+      "learning_rate": 2.4686901094131654e-05,
+      "loss": 0.0646,
+      "step": 2475
+    },
+    {
+      "epoch": 8.842857142857143,
+      "grad_norm": 1.0668024200115853,
+      "learning_rate": 2.4675655427833497e-05,
+      "loss": 0.1458,
+      "step": 2476
+    },
+    {
+      "epoch": 8.846428571428572,
+      "grad_norm": 0.964153220148512,
+      "learning_rate": 2.46644081975859e-05,
+      "loss": 0.042,
+      "step": 2477
+    },
+    {
+      "epoch": 8.85,
+      "grad_norm": 0.8643468052970582,
+      "learning_rate": 2.4653159407150927e-05,
+      "loss": 0.0417,
+      "step": 2478
+    },
+    {
+      "epoch": 8.853571428571428,
+      "grad_norm": 0.5027721530977775,
+      "learning_rate": 2.4641909060291155e-05,
+      "loss": 0.0546,
+      "step": 2479
+    },
+    {
+      "epoch": 8.857142857142858,
+      "grad_norm": 0.7797697318560458,
+      "learning_rate": 2.4630657160769683e-05,
+      "loss": 0.0657,
+      "step": 2480
+    },
+    {
+      "epoch": 8.860714285714286,
+      "grad_norm": 1.2516202866145443,
+      "learning_rate": 2.4619403712350147e-05,
+      "loss": 0.0852,
+      "step": 2481
+    },
+    {
+      "epoch": 8.864285714285714,
+      "grad_norm": 0.8100636320693259,
+      "learning_rate": 2.4608148718796672e-05,
+      "loss": 0.0686,
+      "step": 2482
+    },
+    {
+      "epoch": 8.867857142857144,
+      "grad_norm": 0.44680637460073963,
+      "learning_rate": 2.459689218387393e-05,
+      "loss": 0.0236,
+      "step": 2483
+    },
+    {
+      "epoch": 8.871428571428572,
+      "grad_norm": 0.7679235491071481,
+      "learning_rate": 2.4585634111347086e-05,
+      "loss": 0.0601,
+      "step": 2484
+    },
+    {
+      "epoch": 8.875,
+      "grad_norm": 0.8529925814900318,
+      "learning_rate": 2.4574374504981828e-05,
+      "loss": 0.0811,
+      "step": 2485
+    },
+    {
+      "epoch": 8.878571428571428,
+      "grad_norm": 0.26523157500545985,
+      "learning_rate": 2.4563113368544355e-05,
+      "loss": 0.0254,
+      "step": 2486
+    },
+    {
+      "epoch": 8.882142857142856,
+      "grad_norm": 1.0268339600625456,
+      "learning_rate": 2.4551850705801385e-05,
+      "loss": 0.082,
+      "step": 2487
+    },
+    {
+      "epoch": 8.885714285714286,
+      "grad_norm": 0.7609744705519187,
+      "learning_rate": 2.4540586520520132e-05,
+      "loss": 0.0486,
+      "step": 2488
+    },
+    {
+      "epoch": 8.889285714285714,
+      "grad_norm": 1.2952912228759939,
+      "learning_rate": 2.4529320816468333e-05,
+      "loss": 0.1283,
+      "step": 2489
+    },
+    {
+      "epoch": 8.892857142857142,
+      "grad_norm": 0.38312035113450843,
+      "learning_rate": 2.4518053597414228e-05,
+      "loss": 0.0442,
+      "step": 2490
+    },
+    {
+      "epoch": 8.896428571428572,
+      "grad_norm": 0.6680263744657537,
+      "learning_rate": 2.4506784867126564e-05,
+      "loss": 0.0525,
+      "step": 2491
+    },
+    {
+      "epoch": 8.9,
+      "grad_norm": 0.5100245178608173,
+      "learning_rate": 2.4495514629374592e-05,
+      "loss": 0.0492,
+      "step": 2492
+    },
+    {
+      "epoch": 8.903571428571428,
+      "grad_norm": 0.7915144124825184,
+      "learning_rate": 2.448424288792807e-05,
+      "loss": 0.096,
+      "step": 2493
+    },
+    {
+      "epoch": 8.907142857142857,
+      "grad_norm": 0.35315781976260596,
+      "learning_rate": 2.4472969646557253e-05,
+      "loss": 0.0436,
+      "step": 2494
+    },
+    {
+      "epoch": 8.910714285714286,
+      "grad_norm": 0.9764778859317311,
+      "learning_rate": 2.4461694909032906e-05,
+      "loss": 0.113,
+      "step": 2495
+    },
+    {
+      "epoch": 8.914285714285715,
+      "grad_norm": 0.9909978131107476,
+      "learning_rate": 2.445041867912629e-05,
+      "loss": 0.0801,
+      "step": 2496
+    },
+    {
+      "epoch": 8.917857142857143,
+      "grad_norm": 0.7757017203830077,
+      "learning_rate": 2.4439140960609166e-05,
+      "loss": 0.0458,
+      "step": 2497
+    },
+    {
+      "epoch": 8.92142857142857,
+      "grad_norm": 1.0884215063988452,
+      "learning_rate": 2.442786175725379e-05,
+      "loss": 0.1092,
+      "step": 2498
+    },
+    {
+      "epoch": 8.925,
+      "grad_norm": 0.5274062758151958,
+      "learning_rate": 2.441658107283293e-05,
+      "loss": 0.0787,
+      "step": 2499
+    },
+    {
+      "epoch": 8.928571428571429,
+      "grad_norm": 0.8114858086672402,
+      "learning_rate": 2.4405298911119815e-05,
+      "loss": 0.0697,
+      "step": 2500
+    },
+    {
+      "epoch": 8.932142857142857,
+      "grad_norm": 0.6875218888646812,
+      "learning_rate": 2.4394015275888204e-05,
+      "loss": 0.0646,
+      "step": 2501
+    },
+    {
+      "epoch": 8.935714285714285,
+      "grad_norm": 0.6470997614431683,
+      "learning_rate": 2.438273017091233e-05,
+      "loss": 0.0642,
+      "step": 2502
+    },
+    {
+      "epoch": 8.939285714285715,
+      "grad_norm": 0.6380604947747622,
+      "learning_rate": 2.437144359996692e-05,
+      "loss": 0.0471,
+      "step": 2503
+    },
+    {
+      "epoch": 8.942857142857143,
+      "grad_norm": 0.6311522667213902,
+      "learning_rate": 2.4360155566827202e-05,
+      "loss": 0.0755,
+      "step": 2504
+    },
+    {
+      "epoch": 8.946428571428571,
+      "grad_norm": 1.0220335314685904,
+      "learning_rate": 2.4348866075268886e-05,
+      "loss": 0.0559,
+      "step": 2505
+    },
+    {
+      "epoch": 8.95,
+      "grad_norm": 0.5955483173140359,
+      "learning_rate": 2.4337575129068157e-05,
+      "loss": 0.0816,
+      "step": 2506
+    },
+    {
+      "epoch": 8.95357142857143,
+      "grad_norm": 0.7834158769005107,
+      "learning_rate": 2.4326282732001704e-05,
+      "loss": 0.0707,
+      "step": 2507
+    },
+    {
+      "epoch": 8.957142857142857,
+      "grad_norm": 0.8048657549097726,
+      "learning_rate": 2.43149888878467e-05,
+      "loss": 0.0565,
+      "step": 2508
+    },
+    {
+      "epoch": 8.960714285714285,
+      "grad_norm": 0.5368714330901325,
+      "learning_rate": 2.4303693600380792e-05,
+      "loss": 0.0583,
+      "step": 2509
+    },
+    {
+      "epoch": 8.964285714285714,
+      "grad_norm": 0.6206759072663475,
+      "learning_rate": 2.4292396873382113e-05,
+      "loss": 0.0693,
+      "step": 2510
+    },
+    {
+      "epoch": 8.967857142857143,
+      "grad_norm": 0.606149953646086,
+      "learning_rate": 2.4281098710629284e-05,
+      "loss": 0.0464,
+      "step": 2511
+    },
+    {
+      "epoch": 8.971428571428572,
+      "grad_norm": 0.8030743216757655,
+      "learning_rate": 2.42697991159014e-05,
+      "loss": 0.085,
+      "step": 2512
+    },
+    {
+      "epoch": 8.975,
+      "grad_norm": 0.6584251114191445,
+      "learning_rate": 2.425849809297804e-05,
+      "loss": 0.0672,
+      "step": 2513
+    },
+    {
+      "epoch": 8.978571428571428,
+      "grad_norm": 0.6139381707339112,
+      "learning_rate": 2.4247195645639253e-05,
+      "loss": 0.0661,
+      "step": 2514
+    },
+    {
+      "epoch": 8.982142857142858,
+      "grad_norm": 0.6661948651932524,
+      "learning_rate": 2.423589177766557e-05,
+      "loss": 0.0627,
+      "step": 2515
+    },
+    {
+      "epoch": 8.985714285714286,
+      "grad_norm": 0.44049205594166935,
+      "learning_rate": 2.4224586492838e-05,
+      "loss": 0.0554,
+      "step": 2516
+    },
+    {
+      "epoch": 8.989285714285714,
+      "grad_norm": 0.5733695569416843,
+      "learning_rate": 2.4213279794938026e-05,
+      "loss": 0.0394,
+      "step": 2517
+    },
+    {
+      "epoch": 8.992857142857144,
+      "grad_norm": 0.5826438993775913,
+      "learning_rate": 2.420197168774759e-05,
+      "loss": 0.0605,
+      "step": 2518
+    },
+    {
+      "epoch": 8.996428571428572,
+      "grad_norm": 0.3917848164649797,
+      "learning_rate": 2.4190662175049114e-05,
+      "loss": 0.0449,
+      "step": 2519
+    },
+    {
+      "epoch": 9.0,
+      "grad_norm": 0.48786661549845245,
+      "learning_rate": 2.417935126062551e-05,
+      "loss": 0.0597,
+      "step": 2520
+    },
+    {
+      "epoch": 9.003571428571428,
+      "grad_norm": 0.7918251268711789,
+      "learning_rate": 2.4168038948260114e-05,
+      "loss": 0.0999,
+      "step": 2521
+    },
+    {
+      "epoch": 9.007142857142858,
+      "grad_norm": 0.9426549047417694,
+      "learning_rate": 2.4156725241736773e-05,
+      "loss": 0.0987,
+      "step": 2522
+    },
+    {
+      "epoch": 9.010714285714286,
+      "grad_norm": 0.5079060228819897,
+      "learning_rate": 2.414541014483978e-05,
+      "loss": 0.0764,
+      "step": 2523
+    },
+    {
+      "epoch": 9.014285714285714,
+      "grad_norm": 0.742088897054447,
+      "learning_rate": 2.4134093661353893e-05,
+      "loss": 0.0703,
+      "step": 2524
+    },
+    {
+      "epoch": 9.017857142857142,
+      "grad_norm": 0.5768056711851393,
+      "learning_rate": 2.4122775795064342e-05,
+      "loss": 0.0379,
+      "step": 2525
+    },
+    {
+      "epoch": 9.021428571428572,
+      "grad_norm": 0.4029863041904266,
+      "learning_rate": 2.4111456549756808e-05,
+      "loss": 0.0583,
+      "step": 2526
+    },
+    {
+      "epoch": 9.025,
+      "grad_norm": 0.5022978972753577,
+      "learning_rate": 2.410013592921745e-05,
+      "loss": 0.0308,
+      "step": 2527
+    },
+    {
+      "epoch": 9.028571428571428,
+      "grad_norm": 0.38841883733391586,
+      "learning_rate": 2.4088813937232867e-05,
+      "loss": 0.0525,
+      "step": 2528
+    },
+    {
+      "epoch": 9.032142857142857,
+      "grad_norm": 0.4935891722265799,
+      "learning_rate": 2.4077490577590127e-05,
+      "loss": 0.0584,
+      "step": 2529
+    },
+    {
+      "epoch": 9.035714285714286,
+      "grad_norm": 0.42624849431955847,
+      "learning_rate": 2.4066165854076762e-05,
+      "loss": 0.0592,
+      "step": 2530
+    },
+    {
+      "epoch": 9.039285714285715,
+      "grad_norm": 0.28152329385111535,
+      "learning_rate": 2.4054839770480748e-05,
+      "loss": 0.0179,
+      "step": 2531
+    },
+    {
+      "epoch": 9.042857142857143,
+      "grad_norm": 0.6537891685569786,
+      "learning_rate": 2.4043512330590522e-05,
+      "loss": 0.0632,
+      "step": 2532
+    },
+    {
+      "epoch": 9.04642857142857,
+      "grad_norm": 0.5131845931618266,
+      "learning_rate": 2.403218353819498e-05,
+      "loss": 0.0657,
+      "step": 2533
+    },
+    {
+      "epoch": 9.05,
+      "grad_norm": 0.6036292674943938,
+      "learning_rate": 2.4020853397083456e-05,
+      "loss": 0.0357,
+      "step": 2534
+    },
+    {
+      "epoch": 9.053571428571429,
+      "grad_norm": 0.5406536992278551,
+      "learning_rate": 2.4009521911045746e-05,
+      "loss": 0.0486,
+      "step": 2535
+    },
+    {
+      "epoch": 9.057142857142857,
+      "grad_norm": 0.7050376097044551,
+      "learning_rate": 2.3998189083872094e-05,
+      "loss": 0.0799,
+      "step": 2536
+    },
+    {
+      "epoch": 9.060714285714285,
+      "grad_norm": 0.37614020457089464,
+      "learning_rate": 2.398685491935319e-05,
+      "loss": 0.0346,
+      "step": 2537
+    },
+    {
+      "epoch": 9.064285714285715,
+      "grad_norm": 0.84109675756461,
+      "learning_rate": 2.397551942128017e-05,
+      "loss": 0.0881,
+      "step": 2538
+    },
+    {
+      "epoch": 9.067857142857143,
+      "grad_norm": 0.8166515917690778,
+      "learning_rate": 2.3964182593444626e-05,
+      "loss": 0.0717,
+      "step": 2539
+    },
+    {
+      "epoch": 9.071428571428571,
+      "grad_norm": 0.4629870178686782,
+      "learning_rate": 2.395284443963859e-05,
+      "loss": 0.0484,
+      "step": 2540
+    },
+    {
+      "epoch": 9.075,
+      "grad_norm": 0.9348759461282414,
+      "learning_rate": 2.3941504963654522e-05,
+      "loss": 0.0893,
+      "step": 2541
+    },
+    {
+      "epoch": 9.07857142857143,
+      "grad_norm": 0.6063129060515882,
+      "learning_rate": 2.3930164169285354e-05,
+      "loss": 0.0563,
+      "step": 2542
+    },
+    {
+      "epoch": 9.082142857142857,
+      "grad_norm": 0.9271533696161666,
+      "learning_rate": 2.3918822060324432e-05,
+      "loss": 0.0529,
+      "step": 2543
+    },
+    {
+      "epoch": 9.085714285714285,
+      "grad_norm": 0.6426076421636133,
+      "learning_rate": 2.3907478640565557e-05,
+      "loss": 0.0485,
+      "step": 2544
+    },
+    {
+      "epoch": 9.089285714285714,
+      "grad_norm": 0.5612966488437116,
+      "learning_rate": 2.3896133913802966e-05,
+      "loss": 0.0812,
+      "step": 2545
+    },
+    {
+      "epoch": 9.092857142857143,
+      "grad_norm": 0.6863932511980912,
+      "learning_rate": 2.3884787883831324e-05,
+      "loss": 0.0463,
+      "step": 2546
+    },
+    {
+      "epoch": 9.096428571428572,
+      "grad_norm": 0.7057967517389377,
+      "learning_rate": 2.387344055444574e-05,
+      "loss": 0.0573,
+      "step": 2547
+    },
+    {
+      "epoch": 9.1,
+      "grad_norm": 0.9551044861006828,
+      "learning_rate": 2.3862091929441764e-05,
+      "loss": 0.0542,
+      "step": 2548
+    },
+    {
+      "epoch": 9.103571428571428,
+      "grad_norm": 0.7352023690898936,
+      "learning_rate": 2.3850742012615363e-05,
+      "loss": 0.045,
+      "step": 2549
+    },
+    {
+      "epoch": 9.107142857142858,
+      "grad_norm": 0.6145725232455961,
+      "learning_rate": 2.383939080776295e-05,
+      "loss": 0.052,
+      "step": 2550
+    },
+    {
+      "epoch": 9.110714285714286,
+      "grad_norm": 0.46743374637458995,
+      "learning_rate": 2.3828038318681368e-05,
+      "loss": 0.0473,
+      "step": 2551
+    },
+    {
+      "epoch": 9.114285714285714,
+      "grad_norm": 1.0589916169549682,
+      "learning_rate": 2.3816684549167873e-05,
+      "loss": 0.1234,
+      "step": 2552
+    },
+    {
+      "epoch": 9.117857142857142,
+      "grad_norm": 0.8591553976419788,
+      "learning_rate": 2.380532950302018e-05,
+      "loss": 0.0867,
+      "step": 2553
+    },
+    {
+      "epoch": 9.121428571428572,
+      "grad_norm": 0.6757013323711496,
+      "learning_rate": 2.3793973184036392e-05,
+      "loss": 0.0471,
+      "step": 2554
+    },
+    {
+      "epoch": 9.125,
+      "grad_norm": 0.5531067411121916,
+      "learning_rate": 2.378261559601507e-05,
+      "loss": 0.0546,
+      "step": 2555
+    },
+    {
+      "epoch": 9.128571428571428,
+      "grad_norm": 0.850658741218356,
+      "learning_rate": 2.3771256742755196e-05,
+      "loss": 0.0414,
+      "step": 2556
+    },
+    {
+      "epoch": 9.132142857142858,
+      "grad_norm": 0.7199335468458806,
+      "learning_rate": 2.3759896628056155e-05,
+      "loss": 0.0593,
+      "step": 2557
+    },
+    {
+      "epoch": 9.135714285714286,
+      "grad_norm": 0.8790502550447409,
+      "learning_rate": 2.374853525571777e-05,
+      "loss": 0.058,
+      "step": 2558
+    },
+    {
+      "epoch": 9.139285714285714,
+      "grad_norm": 0.8227417525078601,
+      "learning_rate": 2.3737172629540274e-05,
+      "loss": 0.0703,
+      "step": 2559
+    },
+    {
+      "epoch": 9.142857142857142,
+      "grad_norm": 0.6749958102012888,
+      "learning_rate": 2.3725808753324335e-05,
+      "loss": 0.0549,
+      "step": 2560
+    },
+    {
+      "epoch": 9.146428571428572,
+      "grad_norm": 0.5930756304529378,
+      "learning_rate": 2.3714443630871026e-05,
+      "loss": 0.0737,
+      "step": 2561
+    },
+    {
+      "epoch": 9.15,
+      "grad_norm": 0.6381916787701427,
+      "learning_rate": 2.3703077265981844e-05,
+      "loss": 0.0778,
+      "step": 2562
+    },
+    {
+      "epoch": 9.153571428571428,
+      "grad_norm": 0.49763000376402017,
+      "learning_rate": 2.3691709662458698e-05,
+      "loss": 0.0268,
+      "step": 2563
+    },
+    {
+      "epoch": 9.157142857142857,
+      "grad_norm": 0.9913465947746835,
+      "learning_rate": 2.3680340824103915e-05,
+      "loss": 0.0597,
+      "step": 2564
+    },
+    {
+      "epoch": 9.160714285714286,
+      "grad_norm": 0.8036292476090104,
+      "learning_rate": 2.3668970754720218e-05,
+      "loss": 0.0768,
+      "step": 2565
+    },
+    {
+      "epoch": 9.164285714285715,
+      "grad_norm": 1.0066316551271262,
+      "learning_rate": 2.365759945811078e-05,
+      "loss": 0.0762,
+      "step": 2566
+    },
+    {
+      "epoch": 9.167857142857143,
+      "grad_norm": 0.5785426566525185,
+      "learning_rate": 2.3646226938079138e-05,
+      "loss": 0.0656,
+      "step": 2567
+    },
+    {
+      "epoch": 9.17142857142857,
+      "grad_norm": 1.0909378713911684,
+      "learning_rate": 2.3634853198429273e-05,
+      "loss": 0.0756,
+      "step": 2568
+    },
+    {
+      "epoch": 9.175,
+      "grad_norm": 0.6131949844326061,
+      "learning_rate": 2.3623478242965555e-05,
+      "loss": 0.0451,
+      "step": 2569
+    },
+    {
+      "epoch": 9.178571428571429,
+      "grad_norm": 0.6523506151162185,
+      "learning_rate": 2.3612102075492776e-05,
+      "loss": 0.0363,
+      "step": 2570
+    },
+    {
+      "epoch": 9.182142857142857,
+      "grad_norm": 0.7900141399976883,
+      "learning_rate": 2.360072469981612e-05,
+      "loss": 0.1068,
+      "step": 2571
+    },
+    {
+      "epoch": 9.185714285714285,
+      "grad_norm": 0.8381521737160151,
+      "learning_rate": 2.358934611974117e-05,
+      "loss": 0.0735,
+      "step": 2572
+    },
+    {
+      "epoch": 9.189285714285715,
+      "grad_norm": 0.6545123461189739,
+      "learning_rate": 2.3577966339073933e-05,
+      "loss": 0.0578,
+      "step": 2573
+    },
+    {
+      "epoch": 9.192857142857143,
+      "grad_norm": 0.8892488387600991,
+      "learning_rate": 2.3566585361620806e-05,
+      "loss": 0.0358,
+      "step": 2574
+    },
+    {
+      "epoch": 9.196428571428571,
+      "grad_norm": 0.32872923407261284,
+      "learning_rate": 2.355520319118858e-05,
+      "loss": 0.0528,
+      "step": 2575
+    },
+    {
+      "epoch": 9.2,
+      "grad_norm": 0.700851147533614,
+      "learning_rate": 2.354381983158446e-05,
+      "loss": 0.0365,
+      "step": 2576
+    },
+    {
+      "epoch": 9.20357142857143,
+      "grad_norm": 0.6181932437933318,
+      "learning_rate": 2.353243528661603e-05,
+      "loss": 0.0656,
+      "step": 2577
+    },
+    {
+      "epoch": 9.207142857142857,
+      "grad_norm": 0.956668633009719,
+      "learning_rate": 2.352104956009129e-05,
+      "loss": 0.0579,
+      "step": 2578
+    },
+    {
+      "epoch": 9.210714285714285,
+      "grad_norm": 0.3436512264160102,
+      "learning_rate": 2.3509662655818623e-05,
+      "loss": 0.031,
+      "step": 2579
+    },
+    {
+      "epoch": 9.214285714285714,
+      "grad_norm": 0.5941227680735618,
+      "learning_rate": 2.3498274577606807e-05,
+      "loss": 0.0784,
+      "step": 2580
+    },
+    {
+      "epoch": 9.217857142857143,
+      "grad_norm": 1.0517476497206082,
+      "learning_rate": 2.3486885329265013e-05,
+      "loss": 0.0509,
+      "step": 2581
+    },
+    {
+      "epoch": 9.221428571428572,
+      "grad_norm": 0.2883635412639624,
+      "learning_rate": 2.3475494914602818e-05,
+      "loss": 0.0189,
+      "step": 2582
+    },
+    {
+      "epoch": 9.225,
+      "grad_norm": 0.8467093931913651,
+      "learning_rate": 2.3464103337430163e-05,
+      "loss": 0.0958,
+      "step": 2583
+    },
+    {
+      "epoch": 9.228571428571428,
+      "grad_norm": 1.7092941017819043,
+      "learning_rate": 2.3452710601557386e-05,
+      "loss": 0.1177,
+      "step": 2584
+    },
+    {
+      "epoch": 9.232142857142858,
+      "grad_norm": 0.7411869977107941,
+      "learning_rate": 2.3441316710795232e-05,
+      "loss": 0.036,
+      "step": 2585
+    },
+    {
+      "epoch": 9.235714285714286,
+      "grad_norm": 0.7314400835066779,
+      "learning_rate": 2.342992166895481e-05,
+      "loss": 0.0716,
+      "step": 2586
+    },
+    {
+      "epoch": 9.239285714285714,
+      "grad_norm": 0.832213792286631,
+      "learning_rate": 2.3418525479847615e-05,
+      "loss": 0.0875,
+      "step": 2587
+    },
+    {
+      "epoch": 9.242857142857142,
+      "grad_norm": 0.8723672573051194,
+      "learning_rate": 2.340712814728555e-05,
+      "loss": 0.1319,
+      "step": 2588
+    },
+    {
+      "epoch": 9.246428571428572,
+      "grad_norm": 0.8302930268675202,
+      "learning_rate": 2.339572967508087e-05,
+      "loss": 0.0699,
+      "step": 2589
+    },
+    {
+      "epoch": 9.25,
+      "grad_norm": 0.7010147724302662,
+      "learning_rate": 2.3384330067046233e-05,
+      "loss": 0.0748,
+      "step": 2590
+    },
+    {
+      "epoch": 9.253571428571428,
+      "grad_norm": 0.8044135980534868,
+      "learning_rate": 2.3372929326994664e-05,
+      "loss": 0.0543,
+      "step": 2591
+    },
+    {
+      "epoch": 9.257142857142856,
+      "grad_norm": 1.1069263142294652,
+      "learning_rate": 2.3361527458739567e-05,
+      "loss": 0.0549,
+      "step": 2592
+    },
+    {
+      "epoch": 9.260714285714286,
+      "grad_norm": 0.6687406345998721,
+      "learning_rate": 2.335012446609473e-05,
+      "loss": 0.0448,
+      "step": 2593
+    },
+    {
+      "epoch": 9.264285714285714,
+      "grad_norm": 0.46875213502109603,
+      "learning_rate": 2.3338720352874318e-05,
+      "loss": 0.0414,
+      "step": 2594
+    },
+    {
+      "epoch": 9.267857142857142,
+      "grad_norm": 0.5051013889707137,
+      "learning_rate": 2.3327315122892856e-05,
+      "loss": 0.0807,
+      "step": 2595
+    },
+    {
+      "epoch": 9.271428571428572,
+      "grad_norm": 0.3810339964078406,
+      "learning_rate": 2.331590877996527e-05,
+      "loss": 0.0377,
+      "step": 2596
+    },
+    {
+      "epoch": 9.275,
+      "grad_norm": 0.7440901541896721,
+      "learning_rate": 2.330450132790683e-05,
+      "loss": 0.0907,
+      "step": 2597
+    },
+    {
+      "epoch": 9.278571428571428,
+      "grad_norm": 0.39046302711832886,
+      "learning_rate": 2.329309277053319e-05,
+      "loss": 0.0449,
+      "step": 2598
+    },
+    {
+      "epoch": 9.282142857142857,
+      "grad_norm": 0.9102529536865773,
+      "learning_rate": 2.3281683111660377e-05,
+      "loss": 0.0544,
+      "step": 2599
+    },
+    {
+      "epoch": 9.285714285714286,
+      "grad_norm": 0.7029072408811584,
+      "learning_rate": 2.327027235510478e-05,
+      "loss": 0.0452,
+      "step": 2600
+    },
+    {
+      "epoch": 9.289285714285715,
+      "grad_norm": 0.7136829738111642,
+      "learning_rate": 2.3258860504683158e-05,
+      "loss": 0.0999,
+      "step": 2601
+    },
+    {
+      "epoch": 9.292857142857143,
+      "grad_norm": 0.861828633490484,
+      "learning_rate": 2.324744756421263e-05,
+      "loss": 0.0764,
+      "step": 2602
+    },
+    {
+      "epoch": 9.29642857142857,
+      "grad_norm": 0.5362198950336028,
+      "learning_rate": 2.3236033537510692e-05,
+      "loss": 0.0357,
+      "step": 2603
+    },
+    {
+      "epoch": 9.3,
+      "grad_norm": 0.4978639821700624,
+      "learning_rate": 2.3224618428395198e-05,
+      "loss": 0.035,
+      "step": 2604
+    },
+    {
+      "epoch": 9.303571428571429,
+      "grad_norm": 0.8527180643810932,
+      "learning_rate": 2.3213202240684356e-05,
+      "loss": 0.06,
+      "step": 2605
+    },
+    {
+      "epoch": 9.307142857142857,
+      "grad_norm": 1.2562164738711703,
+      "learning_rate": 2.320178497819674e-05,
+      "loss": 0.1059,
+      "step": 2606
+    },
+    {
+      "epoch": 9.310714285714285,
+      "grad_norm": 0.767841518574634,
+      "learning_rate": 2.3190366644751282e-05,
+      "loss": 0.0775,
+      "step": 2607
+    },
+    {
+      "epoch": 9.314285714285715,
+      "grad_norm": 0.607435566473406,
+      "learning_rate": 2.3178947244167284e-05,
+      "loss": 0.0586,
+      "step": 2608
+    },
+    {
+      "epoch": 9.317857142857143,
+      "grad_norm": 0.5909607382955244,
+      "learning_rate": 2.316752678026439e-05,
+      "loss": 0.0755,
+      "step": 2609
+    },
+    {
+      "epoch": 9.321428571428571,
+      "grad_norm": 0.7157222686676222,
+      "learning_rate": 2.3156105256862603e-05,
+      "loss": 0.1107,
+      "step": 2610
+    },
+    {
+      "epoch": 9.325,
+      "grad_norm": 0.6001041092789494,
+      "learning_rate": 2.314468267778229e-05,
+      "loss": 0.0396,
+      "step": 2611
+    },
+    {
+      "epoch": 9.32857142857143,
+      "grad_norm": 0.44431648829162396,
+      "learning_rate": 2.3133259046844153e-05,
+      "loss": 0.0236,
+      "step": 2612
+    },
+    {
+      "epoch": 9.332142857142857,
+      "grad_norm": 0.5985122428173444,
+      "learning_rate": 2.3121834367869268e-05,
+      "loss": 0.0465,
+      "step": 2613
+    },
+    {
+      "epoch": 9.335714285714285,
+      "grad_norm": 0.7850283449169978,
+      "learning_rate": 2.3110408644679043e-05,
+      "loss": 0.0866,
+      "step": 2614
+    },
+    {
+      "epoch": 9.339285714285714,
+      "grad_norm": 1.198277789840684,
+      "learning_rate": 2.3098981881095247e-05,
+      "loss": 0.0721,
+      "step": 2615
+    },
+    {
+      "epoch": 9.342857142857143,
+      "grad_norm": 0.9134420472463015,
+      "learning_rate": 2.3087554080939988e-05,
+      "loss": 0.0532,
+      "step": 2616
+    },
+    {
+      "epoch": 9.346428571428572,
+      "grad_norm": 0.44663855525164736,
+      "learning_rate": 2.3076125248035725e-05,
+      "loss": 0.0429,
+      "step": 2617
+    },
+    {
+      "epoch": 9.35,
+      "grad_norm": 1.0994012046006345,
+      "learning_rate": 2.3064695386205264e-05,
+      "loss": 0.0664,
+      "step": 2618
+    },
+    {
+      "epoch": 9.353571428571428,
+      "grad_norm": 0.4565328471635104,
+      "learning_rate": 2.3053264499271756e-05,
+      "loss": 0.0428,
+      "step": 2619
+    },
+    {
+      "epoch": 9.357142857142858,
+      "grad_norm": 0.6756107083778264,
+      "learning_rate": 2.304183259105869e-05,
+      "loss": 0.0819,
+      "step": 2620
+    },
+    {
+      "epoch": 9.360714285714286,
+      "grad_norm": 1.2253994251951905,
+      "learning_rate": 2.30303996653899e-05,
+      "loss": 0.0535,
+      "step": 2621
+    },
+    {
+      "epoch": 9.364285714285714,
+      "grad_norm": 0.992440444067353,
+      "learning_rate": 2.3018965726089557e-05,
+      "loss": 0.0786,
+      "step": 2622
+    },
+    {
+      "epoch": 9.367857142857144,
+      "grad_norm": 1.0564501160085586,
+      "learning_rate": 2.3007530776982182e-05,
+      "loss": 0.0734,
+      "step": 2623
+    },
+    {
+      "epoch": 9.371428571428572,
+      "grad_norm": 0.8455878668161415,
+      "learning_rate": 2.2996094821892615e-05,
+      "loss": 0.0495,
+      "step": 2624
+    },
+    {
+      "epoch": 9.375,
+      "grad_norm": 0.605514781157727,
+      "learning_rate": 2.2984657864646045e-05,
+      "loss": 0.0254,
+      "step": 2625
+    },
+    {
+      "epoch": 9.378571428571428,
+      "grad_norm": 0.770572988253507,
+      "learning_rate": 2.2973219909067998e-05,
+      "loss": 0.0737,
+      "step": 2626
+    },
+    {
+      "epoch": 9.382142857142856,
+      "grad_norm": 0.8547243922668767,
+      "learning_rate": 2.296178095898433e-05,
+      "loss": 0.0578,
+      "step": 2627
+    },
+    {
+      "epoch": 9.385714285714286,
+      "grad_norm": 0.5968003717743072,
+      "learning_rate": 2.295034101822123e-05,
+      "loss": 0.0631,
+      "step": 2628
+    },
+    {
+      "epoch": 9.389285714285714,
+      "grad_norm": 0.9139409851481666,
+      "learning_rate": 2.293890009060521e-05,
+      "loss": 0.0648,
+      "step": 2629
+    },
+    {
+      "epoch": 9.392857142857142,
+      "grad_norm": 1.019536631460233,
+      "learning_rate": 2.2927458179963126e-05,
+      "loss": 0.0456,
+      "step": 2630
+    },
+    {
+      "epoch": 9.396428571428572,
+      "grad_norm": 0.4111661947718684,
+      "learning_rate": 2.2916015290122154e-05,
+      "loss": 0.0447,
+      "step": 2631
+    },
+    {
+      "epoch": 9.4,
+      "grad_norm": 0.6470074767513081,
+      "learning_rate": 2.290457142490981e-05,
+      "loss": 0.1032,
+      "step": 2632
+    },
+    {
+      "epoch": 9.403571428571428,
+      "grad_norm": 1.4786722273987922,
+      "learning_rate": 2.2893126588153916e-05,
+      "loss": 0.1304,
+      "step": 2633
+    },
+    {
+      "epoch": 9.407142857142857,
+      "grad_norm": 1.3771406034873481,
+      "learning_rate": 2.2881680783682635e-05,
+      "loss": 0.0819,
+      "step": 2634
+    },
+    {
+      "epoch": 9.410714285714286,
+      "grad_norm": 0.7275890830985224,
+      "learning_rate": 2.2870234015324445e-05,
+      "loss": 0.0583,
+      "step": 2635
+    },
+    {
+      "epoch": 9.414285714285715,
+      "grad_norm": 1.0092509384604404,
+      "learning_rate": 2.285878628690815e-05,
+      "loss": 0.0999,
+      "step": 2636
+    },
+    {
+      "epoch": 9.417857142857143,
+      "grad_norm": 0.5292088207280802,
+      "learning_rate": 2.284733760226288e-05,
+      "loss": 0.0468,
+      "step": 2637
+    },
+    {
+      "epoch": 9.42142857142857,
+      "grad_norm": 0.8384144812755503,
+      "learning_rate": 2.2835887965218072e-05,
+      "loss": 0.0796,
+      "step": 2638
+    },
+    {
+      "epoch": 9.425,
+      "grad_norm": 0.7986089948284776,
+      "learning_rate": 2.2824437379603496e-05,
+      "loss": 0.0898,
+      "step": 2639
+    },
+    {
+      "epoch": 9.428571428571429,
+      "grad_norm": 1.1596040049154073,
+      "learning_rate": 2.2812985849249228e-05,
+      "loss": 0.0549,
+      "step": 2640
+    },
+    {
+      "epoch": 9.432142857142857,
+      "grad_norm": 0.9039821472613709,
+      "learning_rate": 2.2801533377985673e-05,
+      "loss": 0.0682,
+      "step": 2641
+    },
+    {
+      "epoch": 9.435714285714285,
+      "grad_norm": 0.9356390644787252,
+      "learning_rate": 2.279007996964353e-05,
+      "loss": 0.04,
+      "step": 2642
+    },
+    {
+      "epoch": 9.439285714285715,
+      "grad_norm": 0.9039401698465891,
+      "learning_rate": 2.2778625628053833e-05,
+      "loss": 0.079,
+      "step": 2643
+    },
+    {
+      "epoch": 9.442857142857143,
+      "grad_norm": 0.6673800174453586,
+      "learning_rate": 2.276717035704791e-05,
+      "loss": 0.0649,
+      "step": 2644
+    },
+    {
+      "epoch": 9.446428571428571,
+      "grad_norm": 0.6516305872869529,
+      "learning_rate": 2.2755714160457417e-05,
+      "loss": 0.0501,
+      "step": 2645
+    },
+    {
+      "epoch": 9.45,
+      "grad_norm": 0.5554420147159396,
+      "learning_rate": 2.274425704211431e-05,
+      "loss": 0.063,
+      "step": 2646
+    },
+    {
+      "epoch": 9.45357142857143,
+      "grad_norm": 0.5854173635505191,
+      "learning_rate": 2.273279900585085e-05,
+      "loss": 0.0387,
+      "step": 2647
+    },
+    {
+      "epoch": 9.457142857142857,
+      "grad_norm": 0.9047125587723067,
+      "learning_rate": 2.2721340055499613e-05,
+      "loss": 0.0813,
+      "step": 2648
+    },
+    {
+      "epoch": 9.460714285714285,
+      "grad_norm": 0.6346395141698932,
+      "learning_rate": 2.2709880194893478e-05,
+      "loss": 0.0731,
+      "step": 2649
+    },
+    {
+      "epoch": 9.464285714285714,
+      "grad_norm": 0.8681622702019024,
+      "learning_rate": 2.2698419427865627e-05,
+      "loss": 0.0667,
+      "step": 2650
+    },
+    {
+      "epoch": 9.467857142857143,
+      "grad_norm": 0.593943598733648,
+      "learning_rate": 2.268695775824955e-05,
+      "loss": 0.048,
+      "step": 2651
+    },
+    {
+      "epoch": 9.471428571428572,
+      "grad_norm": 0.9154938778800696,
+      "learning_rate": 2.2675495189879023e-05,
+      "loss": 0.0971,
+      "step": 2652
+    },
+    {
+      "epoch": 9.475,
+      "grad_norm": 0.5956698054137287,
+      "learning_rate": 2.2664031726588155e-05,
+      "loss": 0.0317,
+      "step": 2653
+    },
+    {
+      "epoch": 9.478571428571428,
+      "grad_norm": 0.8315538350642412,
+      "learning_rate": 2.2652567372211312e-05,
+      "loss": 0.0735,
+      "step": 2654
+    },
+    {
+      "epoch": 9.482142857142858,
+      "grad_norm": 1.0437892214682813,
+      "learning_rate": 2.2641102130583198e-05,
+      "loss": 0.0778,
+      "step": 2655
+    },
+    {
+      "epoch": 9.485714285714286,
+      "grad_norm": 0.599584940092499,
+      "learning_rate": 2.262963600553878e-05,
+      "loss": 0.0844,
+      "step": 2656
+    },
+    {
+      "epoch": 9.489285714285714,
+      "grad_norm": 0.9015514742948305,
+      "learning_rate": 2.2618169000913347e-05,
+      "loss": 0.0881,
+      "step": 2657
+    },
+    {
+      "epoch": 9.492857142857144,
+      "grad_norm": 0.4270950204918469,
+      "learning_rate": 2.2606701120542473e-05,
+      "loss": 0.0535,
+      "step": 2658
+    },
+    {
+      "epoch": 9.496428571428572,
+      "grad_norm": 0.714056768044811,
+      "learning_rate": 2.259523236826201e-05,
+      "loss": 0.0477,
+      "step": 2659
+    },
+    {
+      "epoch": 9.5,
+      "grad_norm": 0.6788250506476103,
+      "learning_rate": 2.2583762747908132e-05,
+      "loss": 0.0976,
+      "step": 2660
+    },
+    {
+      "epoch": 9.503571428571428,
+      "grad_norm": 0.8700982185665851,
+      "learning_rate": 2.2572292263317273e-05,
+      "loss": 0.0688,
+      "step": 2661
+    },
+    {
+      "epoch": 9.507142857142856,
+      "grad_norm": 0.5443304273124907,
+      "learning_rate": 2.2560820918326183e-05,
+      "loss": 0.0647,
+      "step": 2662
+    },
+    {
+      "epoch": 9.510714285714286,
+      "grad_norm": 0.5413180552374376,
+      "learning_rate": 2.254934871677187e-05,
+      "loss": 0.0498,
+      "step": 2663
+    },
+    {
+      "epoch": 9.514285714285714,
+      "grad_norm": 0.4986964516171025,
+      "learning_rate": 2.253787566249166e-05,
+      "loss": 0.049,
+      "step": 2664
+    },
+    {
+      "epoch": 9.517857142857142,
+      "grad_norm": 1.0606437616566748,
+      "learning_rate": 2.252640175932314e-05,
+      "loss": 0.1099,
+      "step": 2665
+    },
+    {
+      "epoch": 9.521428571428572,
+      "grad_norm": 0.6113245620886507,
+      "learning_rate": 2.2514927011104196e-05,
+      "loss": 0.054,
+      "step": 2666
+    },
+    {
+      "epoch": 9.525,
+      "grad_norm": 0.8361543371796707,
+      "learning_rate": 2.250345142167298e-05,
+      "loss": 0.0674,
+      "step": 2667
+    },
+    {
+      "epoch": 9.528571428571428,
+      "grad_norm": 1.261045823697038,
+      "learning_rate": 2.249197499486795e-05,
+      "loss": 0.1231,
+      "step": 2668
+    },
+    {
+      "epoch": 9.532142857142857,
+      "grad_norm": 0.46688978450071095,
+      "learning_rate": 2.2480497734527817e-05,
+      "loss": 0.0368,
+      "step": 2669
+    },
+    {
+      "epoch": 9.535714285714286,
+      "grad_norm": 0.5992430807689797,
+      "learning_rate": 2.246901964449159e-05,
+      "loss": 0.033,
+      "step": 2670
+    },
+    {
+      "epoch": 9.539285714285715,
+      "grad_norm": 0.3575036605187835,
+      "learning_rate": 2.2457540728598553e-05,
+      "loss": 0.0495,
+      "step": 2671
+    },
+    {
+      "epoch": 9.542857142857143,
+      "grad_norm": 0.5338289440504088,
+      "learning_rate": 2.2446060990688254e-05,
+      "loss": 0.054,
+      "step": 2672
+    },
+    {
+      "epoch": 9.54642857142857,
+      "grad_norm": 0.7067063073763276,
+      "learning_rate": 2.2434580434600536e-05,
+      "loss": 0.1021,
+      "step": 2673
+    },
+    {
+      "epoch": 9.55,
+      "grad_norm": 0.23820994787897423,
+      "learning_rate": 2.2423099064175498e-05,
+      "loss": 0.0311,
+      "step": 2674
+    },
+    {
+      "epoch": 9.553571428571429,
+      "grad_norm": 0.5281662852242939,
+      "learning_rate": 2.241161688325352e-05,
+      "loss": 0.0355,
+      "step": 2675
+    },
+    {
+      "epoch": 9.557142857142857,
+      "grad_norm": 0.5709625151195952,
+      "learning_rate": 2.2400133895675244e-05,
+      "loss": 0.0504,
+      "step": 2676
+    },
+    {
+      "epoch": 9.560714285714285,
+      "grad_norm": 0.5804127419334045,
+      "learning_rate": 2.2388650105281595e-05,
+      "loss": 0.0607,
+      "step": 2677
+    },
+    {
+      "epoch": 9.564285714285715,
+      "grad_norm": 0.4631058018263968,
+      "learning_rate": 2.2377165515913756e-05,
+      "loss": 0.0625,
+      "step": 2678
+    },
+    {
+      "epoch": 9.567857142857143,
+      "grad_norm": 0.8325127170653037,
+      "learning_rate": 2.236568013141319e-05,
+      "loss": 0.0844,
+      "step": 2679
+    },
+    {
+      "epoch": 9.571428571428571,
+      "grad_norm": 0.56517935745875,
+      "learning_rate": 2.2354193955621607e-05,
+      "loss": 0.0453,
+      "step": 2680
+    },
+    {
+      "epoch": 9.575,
+      "grad_norm": 0.5222466739143934,
+      "learning_rate": 2.2342706992380997e-05,
+      "loss": 0.0471,
+      "step": 2681
+    },
+    {
+      "epoch": 9.57857142857143,
+      "grad_norm": 0.6233443146421394,
+      "learning_rate": 2.2331219245533607e-05,
+      "loss": 0.0881,
+      "step": 2682
+    },
+    {
+      "epoch": 9.582142857142857,
+      "grad_norm": 0.5139458055877835,
+      "learning_rate": 2.231973071892195e-05,
+      "loss": 0.0432,
+      "step": 2683
+    },
+    {
+      "epoch": 9.585714285714285,
+      "grad_norm": 1.2712104496468966,
+      "learning_rate": 2.2308241416388796e-05,
+      "loss": 0.1046,
+      "step": 2684
+    },
+    {
+      "epoch": 9.589285714285714,
+      "grad_norm": 0.5591446159036514,
+      "learning_rate": 2.2296751341777173e-05,
+      "loss": 0.0893,
+      "step": 2685
+    },
+    {
+      "epoch": 9.592857142857143,
+      "grad_norm": 0.41863375522773544,
+      "learning_rate": 2.2285260498930374e-05,
+      "loss": 0.0344,
+      "step": 2686
+    },
+    {
+      "epoch": 9.596428571428572,
+      "grad_norm": 0.48641941159819835,
+      "learning_rate": 2.227376889169195e-05,
+      "loss": 0.0385,
+      "step": 2687
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 0.9431317696046368,
+      "learning_rate": 2.226227652390569e-05,
+      "loss": 0.0752,
+      "step": 2688
+    },
+    {
+      "epoch": 9.603571428571428,
+      "grad_norm": 0.5152466069209086,
+      "learning_rate": 2.2250783399415656e-05,
+      "loss": 0.0669,
+      "step": 2689
+    },
+    {
+      "epoch": 9.607142857142858,
+      "grad_norm": 0.6490482681110574,
+      "learning_rate": 2.2239289522066157e-05,
+      "loss": 0.1213,
+      "step": 2690
+    },
+    {
+      "epoch": 9.610714285714286,
+      "grad_norm": 0.4732970281636899,
+      "learning_rate": 2.222779489570176e-05,
+      "loss": 0.053,
+      "step": 2691
+    },
+    {
+      "epoch": 9.614285714285714,
+      "grad_norm": 0.5040954771800618,
+      "learning_rate": 2.221629952416727e-05,
+      "loss": 0.0442,
+      "step": 2692
+    },
+    {
+      "epoch": 9.617857142857144,
+      "grad_norm": 0.4692348790901352,
+      "learning_rate": 2.220480341130775e-05,
+      "loss": 0.049,
+      "step": 2693
+    },
+    {
+      "epoch": 9.621428571428572,
+      "grad_norm": 0.5967719050749721,
+      "learning_rate": 2.219330656096851e-05,
+      "loss": 0.0604,
+      "step": 2694
+    },
+    {
+      "epoch": 9.625,
+      "grad_norm": 0.37714682109789865,
+      "learning_rate": 2.21818089769951e-05,
+      "loss": 0.0307,
+      "step": 2695
+    },
+    {
+      "epoch": 9.628571428571428,
+      "grad_norm": 0.49370285736956787,
+      "learning_rate": 2.2170310663233327e-05,
+      "loss": 0.0627,
+      "step": 2696
+    },
+    {
+      "epoch": 9.632142857142856,
+      "grad_norm": 0.4187822483246256,
+      "learning_rate": 2.215881162352924e-05,
+      "loss": 0.0286,
+      "step": 2697
+    },
+    {
+      "epoch": 9.635714285714286,
+      "grad_norm": 0.40846905116515225,
+      "learning_rate": 2.2147311861729112e-05,
+      "loss": 0.0518,
+      "step": 2698
+    },
+    {
+      "epoch": 9.639285714285714,
+      "grad_norm": 0.8823915506113049,
+      "learning_rate": 2.213581138167949e-05,
+      "loss": 0.0813,
+      "step": 2699
+    },
+    {
+      "epoch": 9.642857142857142,
+      "grad_norm": 0.6819014616355303,
+      "learning_rate": 2.212431018722713e-05,
+      "loss": 0.0959,
+      "step": 2700
+    },
+    {
+      "epoch": 9.646428571428572,
+      "grad_norm": 0.9363657351310862,
+      "learning_rate": 2.2112808282219043e-05,
+      "loss": 0.079,
+      "step": 2701
+    },
+    {
+      "epoch": 9.65,
+      "grad_norm": 0.9877057950992237,
+      "learning_rate": 2.210130567050248e-05,
+      "loss": 0.0982,
+      "step": 2702
+    },
+    {
+      "epoch": 9.653571428571428,
+      "grad_norm": 0.8578923409083974,
+      "learning_rate": 2.2089802355924927e-05,
+      "loss": 0.0866,
+      "step": 2703
+    },
+    {
+      "epoch": 9.657142857142857,
+      "grad_norm": 0.7328276764118163,
+      "learning_rate": 2.207829834233409e-05,
+      "loss": 0.0406,
+      "step": 2704
+    },
+    {
+      "epoch": 9.660714285714286,
+      "grad_norm": 0.5609772649624416,
+      "learning_rate": 2.206679363357793e-05,
+      "loss": 0.0636,
+      "step": 2705
+    },
+    {
+      "epoch": 9.664285714285715,
+      "grad_norm": 0.6094277481381001,
+      "learning_rate": 2.2055288233504625e-05,
+      "loss": 0.0574,
+      "step": 2706
+    },
+    {
+      "epoch": 9.667857142857143,
+      "grad_norm": 0.5872050731573315,
+      "learning_rate": 2.2043782145962592e-05,
+      "loss": 0.0396,
+      "step": 2707
+    },
+    {
+      "epoch": 9.67142857142857,
+      "grad_norm": 0.46639296469924574,
+      "learning_rate": 2.2032275374800482e-05,
+      "loss": 0.046,
+      "step": 2708
+    },
+    {
+      "epoch": 9.675,
+      "grad_norm": 0.43162046696663875,
+      "learning_rate": 2.2020767923867164e-05,
+      "loss": 0.0464,
+      "step": 2709
+    },
+    {
+      "epoch": 9.678571428571429,
+      "grad_norm": 0.6345593019020721,
+      "learning_rate": 2.200925979701174e-05,
+      "loss": 0.062,
+      "step": 2710
+    },
+    {
+      "epoch": 9.682142857142857,
+      "grad_norm": 0.6060070820513224,
+      "learning_rate": 2.1997750998083534e-05,
+      "loss": 0.0532,
+      "step": 2711
+    },
+    {
+      "epoch": 9.685714285714285,
+      "grad_norm": 0.8848618227854355,
+      "learning_rate": 2.198624153093211e-05,
+      "loss": 0.0739,
+      "step": 2712
+    },
+    {
+      "epoch": 9.689285714285715,
+      "grad_norm": 0.8641442157557853,
+      "learning_rate": 2.1974731399407232e-05,
+      "loss": 0.0515,
+      "step": 2713
+    },
+    {
+      "epoch": 9.692857142857143,
+      "grad_norm": 0.6997342946237541,
+      "learning_rate": 2.1963220607358898e-05,
+      "loss": 0.0513,
+      "step": 2714
+    },
+    {
+      "epoch": 9.696428571428571,
+      "grad_norm": 0.5400872897466906,
+      "learning_rate": 2.1951709158637328e-05,
+      "loss": 0.0522,
+      "step": 2715
+    },
+    {
+      "epoch": 9.7,
+      "grad_norm": 0.46390870106174,
+      "learning_rate": 2.1940197057092964e-05,
+      "loss": 0.0344,
+      "step": 2716
+    },
+    {
+      "epoch": 9.70357142857143,
+      "grad_norm": 0.6399806219321906,
+      "learning_rate": 2.1928684306576465e-05,
+      "loss": 0.1071,
+      "step": 2717
+    },
+    {
+      "epoch": 9.707142857142857,
+      "grad_norm": 0.7116422813046882,
+      "learning_rate": 2.1917170910938695e-05,
+      "loss": 0.0651,
+      "step": 2718
+    },
+    {
+      "epoch": 9.710714285714285,
+      "grad_norm": 0.9015897670934058,
+      "learning_rate": 2.1905656874030753e-05,
+      "loss": 0.0811,
+      "step": 2719
+    },
+    {
+      "epoch": 9.714285714285714,
+      "grad_norm": 0.31539874978814936,
+      "learning_rate": 2.1894142199703944e-05,
+      "loss": 0.021,
+      "step": 2720
+    },
+    {
+      "epoch": 9.717857142857143,
+      "grad_norm": 0.7765060112877015,
+      "learning_rate": 2.1882626891809776e-05,
+      "loss": 0.0306,
+      "step": 2721
+    },
+    {
+      "epoch": 9.721428571428572,
+      "grad_norm": 0.7091751455877214,
+      "learning_rate": 2.187111095419999e-05,
+      "loss": 0.0435,
+      "step": 2722
+    },
+    {
+      "epoch": 9.725,
+      "grad_norm": 0.626578506697214,
+      "learning_rate": 2.1859594390726517e-05,
+      "loss": 0.0626,
+      "step": 2723
+    },
+    {
+      "epoch": 9.728571428571428,
+      "grad_norm": 1.4546008354532347,
+      "learning_rate": 2.1848077205241513e-05,
+      "loss": 0.0672,
+      "step": 2724
+    },
+    {
+      "epoch": 9.732142857142858,
+      "grad_norm": 1.7543993349995608,
+      "learning_rate": 2.183655940159733e-05,
+      "loss": 0.0846,
+      "step": 2725
+    },
+    {
+      "epoch": 9.735714285714286,
+      "grad_norm": 1.0904159646788818,
+      "learning_rate": 2.1825040983646532e-05,
+      "loss": 0.062,
+      "step": 2726
+    },
+    {
+      "epoch": 9.739285714285714,
+      "grad_norm": 1.218089761619436,
+      "learning_rate": 2.181352195524189e-05,
+      "loss": 0.091,
+      "step": 2727
+    },
+    {
+      "epoch": 9.742857142857144,
+      "grad_norm": 0.5210946725173388,
+      "learning_rate": 2.180200232023638e-05,
+      "loss": 0.053,
+      "step": 2728
+    },
+    {
+      "epoch": 9.746428571428572,
+      "grad_norm": 0.5906302910041407,
+      "learning_rate": 2.179048208248318e-05,
+      "loss": 0.0726,
+      "step": 2729
+    },
+    {
+      "epoch": 9.75,
+      "grad_norm": 0.5772550184132398,
+      "learning_rate": 2.177896124583566e-05,
+      "loss": 0.0449,
+      "step": 2730
+    },
+    {
+      "epoch": 9.753571428571428,
+      "grad_norm": 0.5529851977784586,
+      "learning_rate": 2.176743981414741e-05,
+      "loss": 0.0856,
+      "step": 2731
+    },
+    {
+      "epoch": 9.757142857142856,
+      "grad_norm": 0.6217926575290101,
+      "learning_rate": 2.1755917791272195e-05,
+      "loss": 0.0572,
+      "step": 2732
+    },
+    {
+      "epoch": 9.760714285714286,
+      "grad_norm": 0.6202037601133039,
+      "learning_rate": 2.1744395181064003e-05,
+      "loss": 0.0438,
+      "step": 2733
+    },
+    {
+      "epoch": 9.764285714285714,
+      "grad_norm": 0.4884313509126108,
+      "learning_rate": 2.1732871987377e-05,
+      "loss": 0.0309,
+      "step": 2734
+    },
+    {
+      "epoch": 9.767857142857142,
+      "grad_norm": 0.8977959672841186,
+      "learning_rate": 2.172134821406555e-05,
+      "loss": 0.0628,
+      "step": 2735
+    },
+    {
+      "epoch": 9.771428571428572,
+      "grad_norm": 0.7370112830459992,
+      "learning_rate": 2.1709823864984228e-05,
+      "loss": 0.0567,
+      "step": 2736
+    },
+    {
+      "epoch": 9.775,
+      "grad_norm": 0.6585264418671256,
+      "learning_rate": 2.1698298943987767e-05,
+      "loss": 0.0462,
+      "step": 2737
+    },
+    {
+      "epoch": 9.778571428571428,
+      "grad_norm": 0.579605997689196,
+      "learning_rate": 2.1686773454931123e-05,
+      "loss": 0.0432,
+      "step": 2738
+    },
+    {
+      "epoch": 9.782142857142857,
+      "grad_norm": 0.702947907696597,
+      "learning_rate": 2.167524740166943e-05,
+      "loss": 0.0729,
+      "step": 2739
+    },
+    {
+      "epoch": 9.785714285714286,
+      "grad_norm": 1.1214203744760838,
+      "learning_rate": 2.1663720788058013e-05,
+      "loss": 0.043,
+      "step": 2740
+    },
+    {
+      "epoch": 9.789285714285715,
+      "grad_norm": 0.7513997930571733,
+      "learning_rate": 2.165219361795238e-05,
+      "loss": 0.0579,
+      "step": 2741
+    },
+    {
+      "epoch": 9.792857142857143,
+      "grad_norm": 0.4123744335986051,
+      "learning_rate": 2.164066589520823e-05,
+      "loss": 0.0361,
+      "step": 2742
+    },
+    {
+      "epoch": 9.79642857142857,
+      "grad_norm": 0.5326287834257172,
+      "learning_rate": 2.1629137623681446e-05,
+      "loss": 0.0437,
+      "step": 2743
+    },
+    {
+      "epoch": 9.8,
+      "grad_norm": 0.6482606996517498,
+      "learning_rate": 2.1617608807228087e-05,
+      "loss": 0.0693,
+      "step": 2744
+    },
+    {
+      "epoch": 9.803571428571429,
+      "grad_norm": 0.8733659803535443,
+      "learning_rate": 2.160607944970441e-05,
+      "loss": 0.1021,
+      "step": 2745
+    },
+    {
+      "epoch": 9.807142857142857,
+      "grad_norm": 0.35771500260944206,
+      "learning_rate": 2.1594549554966834e-05,
+      "loss": 0.0269,
+      "step": 2746
+    },
+    {
+      "epoch": 9.810714285714285,
+      "grad_norm": 1.1619514275851164,
+      "learning_rate": 2.1583019126871978e-05,
+      "loss": 0.0529,
+      "step": 2747
+    },
+    {
+      "epoch": 9.814285714285715,
+      "grad_norm": 0.8442457299361693,
+      "learning_rate": 2.157148816927662e-05,
+      "loss": 0.0891,
+      "step": 2748
+    },
+    {
+      "epoch": 9.817857142857143,
+      "grad_norm": 0.8848990378992801,
+      "learning_rate": 2.1559956686037737e-05,
+      "loss": 0.0906,
+      "step": 2749
+    },
+    {
+      "epoch": 9.821428571428571,
+      "grad_norm": 0.6230535216298332,
+      "learning_rate": 2.1548424681012454e-05,
+      "loss": 0.0616,
+      "step": 2750
+    },
+    {
+      "epoch": 9.825,
+      "grad_norm": 0.678200075562014,
+      "learning_rate": 2.1536892158058088e-05,
+      "loss": 0.0236,
+      "step": 2751
+    },
+    {
+      "epoch": 9.82857142857143,
+      "grad_norm": 0.4354715385040191,
+      "learning_rate": 2.1525359121032135e-05,
+      "loss": 0.0567,
+      "step": 2752
+    },
+    {
+      "epoch": 9.832142857142857,
+      "grad_norm": 0.5874186682345245,
+      "learning_rate": 2.1513825573792244e-05,
+      "loss": 0.0406,
+      "step": 2753
+    },
+    {
+      "epoch": 9.835714285714285,
+      "grad_norm": 0.6457729061275236,
+      "learning_rate": 2.150229152019625e-05,
+      "loss": 0.0386,
+      "step": 2754
+    },
+    {
+      "epoch": 9.839285714285714,
+      "grad_norm": 0.7661511192739392,
+      "learning_rate": 2.1490756964102158e-05,
+      "loss": 0.0986,
+      "step": 2755
+    },
+    {
+      "epoch": 9.842857142857143,
+      "grad_norm": 0.4710439396120435,
+      "learning_rate": 2.1479221909368128e-05,
+      "loss": 0.0462,
+      "step": 2756
+    },
+    {
+      "epoch": 9.846428571428572,
+      "grad_norm": 0.6192763330998342,
+      "learning_rate": 2.1467686359852494e-05,
+      "loss": 0.078,
+      "step": 2757
+    },
+    {
+      "epoch": 9.85,
+      "grad_norm": 0.549204124260435,
+      "learning_rate": 2.1456150319413762e-05,
+      "loss": 0.0683,
+      "step": 2758
+    },
+    {
+      "epoch": 9.853571428571428,
+      "grad_norm": 0.23562074346590553,
+      "learning_rate": 2.1444613791910595e-05,
+      "loss": 0.0325,
+      "step": 2759
+    },
+    {
+      "epoch": 9.857142857142858,
+      "grad_norm": 0.756766616276899,
+      "learning_rate": 2.1433076781201815e-05,
+      "loss": 0.0715,
+      "step": 2760
+    },
+    {
+      "epoch": 9.860714285714286,
+      "grad_norm": 0.992973269236979,
+      "learning_rate": 2.1421539291146418e-05,
+      "loss": 0.0708,
+      "step": 2761
+    },
+    {
+      "epoch": 9.864285714285714,
+      "grad_norm": 0.7085876652547423,
+      "learning_rate": 2.1410001325603544e-05,
+      "loss": 0.052,
+      "step": 2762
+    },
+    {
+      "epoch": 9.867857142857144,
+      "grad_norm": 0.6884944822130163,
+      "learning_rate": 2.139846288843251e-05,
+      "loss": 0.0693,
+      "step": 2763
+    },
+    {
+      "epoch": 9.871428571428572,
+      "grad_norm": 1.0409298849962667,
+      "learning_rate": 2.1386923983492777e-05,
+      "loss": 0.0628,
+      "step": 2764
+    },
+    {
+      "epoch": 9.875,
+      "grad_norm": 0.3152832465301387,
+      "learning_rate": 2.1375384614643968e-05,
+      "loss": 0.0269,
+      "step": 2765
+    },
+    {
+      "epoch": 9.878571428571428,
+      "grad_norm": 0.42055165262573857,
+      "learning_rate": 2.136384478574586e-05,
+      "loss": 0.0444,
+      "step": 2766
+    },
+    {
+      "epoch": 9.882142857142856,
+      "grad_norm": 0.8783150042562635,
+      "learning_rate": 2.1352304500658386e-05,
+      "loss": 0.0674,
+      "step": 2767
+    },
+    {
+      "epoch": 9.885714285714286,
+      "grad_norm": 0.5094887047357832,
+      "learning_rate": 2.134076376324163e-05,
+      "loss": 0.0776,
+      "step": 2768
+    },
+    {
+      "epoch": 9.889285714285714,
+      "grad_norm": 0.8342525564117368,
+      "learning_rate": 2.132922257735582e-05,
+      "loss": 0.0565,
+      "step": 2769
+    },
+    {
+      "epoch": 9.892857142857142,
+      "grad_norm": 0.7696026104512671,
+      "learning_rate": 2.1317680946861356e-05,
+      "loss": 0.0354,
+      "step": 2770
+    },
+    {
+      "epoch": 9.896428571428572,
+      "grad_norm": 0.5784004907114461,
+      "learning_rate": 2.130613887561876e-05,
+      "loss": 0.0515,
+      "step": 2771
+    },
+    {
+      "epoch": 9.9,
+      "grad_norm": 0.7310480949051474,
+      "learning_rate": 2.1294596367488717e-05,
+      "loss": 0.0964,
+      "step": 2772
+    },
+    {
+      "epoch": 9.903571428571428,
+      "grad_norm": 0.7127679676034544,
+      "learning_rate": 2.1283053426332056e-05,
+      "loss": 0.0718,
+      "step": 2773
+    },
+    {
+      "epoch": 9.907142857142857,
+      "grad_norm": 0.8976729410984619,
+      "learning_rate": 2.1271510056009744e-05,
+      "loss": 0.0825,
+      "step": 2774
+    },
+    {
+      "epoch": 9.910714285714286,
+      "grad_norm": 1.002100070289599,
+      "learning_rate": 2.1259966260382902e-05,
+      "loss": 0.0653,
+      "step": 2775
+    },
+    {
+      "epoch": 9.914285714285715,
+      "grad_norm": 0.7803623261998682,
+      "learning_rate": 2.1248422043312784e-05,
+      "loss": 0.0819,
+      "step": 2776
+    },
+    {
+      "epoch": 9.917857142857143,
+      "grad_norm": 2.0122947409938545,
+      "learning_rate": 2.1236877408660797e-05,
+      "loss": 0.1295,
+      "step": 2777
+    },
+    {
+      "epoch": 9.92142857142857,
+      "grad_norm": 0.5876915009221436,
+      "learning_rate": 2.1225332360288472e-05,
+      "loss": 0.052,
+      "step": 2778
+    },
+    {
+      "epoch": 9.925,
+      "grad_norm": 0.7813157947968362,
+      "learning_rate": 2.1213786902057487e-05,
+      "loss": 0.0558,
+      "step": 2779
+    },
+    {
+      "epoch": 9.928571428571429,
+      "grad_norm": 1.0107791780991457,
+      "learning_rate": 2.1202241037829657e-05,
+      "loss": 0.0976,
+      "step": 2780
+    },
+    {
+      "epoch": 9.932142857142857,
+      "grad_norm": 0.585352360696538,
+      "learning_rate": 2.119069477146694e-05,
+      "loss": 0.0664,
+      "step": 2781
+    },
+    {
+      "epoch": 9.935714285714285,
+      "grad_norm": 1.4553746399768739,
+      "learning_rate": 2.1179148106831407e-05,
+      "loss": 0.0759,
+      "step": 2782
+    },
+    {
+      "epoch": 9.939285714285715,
+      "grad_norm": 0.7976481043736428,
+      "learning_rate": 2.1167601047785292e-05,
+      "loss": 0.0667,
+      "step": 2783
+    },
+    {
+      "epoch": 9.942857142857143,
+      "grad_norm": 0.9057662852254439,
+      "learning_rate": 2.1156053598190927e-05,
+      "loss": 0.0593,
+      "step": 2784
+    },
+    {
+      "epoch": 9.946428571428571,
+      "grad_norm": 0.6712159627600629,
+      "learning_rate": 2.1144505761910806e-05,
+      "loss": 0.0386,
+      "step": 2785
+    },
+    {
+      "epoch": 9.95,
+      "grad_norm": 0.42730117490478364,
+      "learning_rate": 2.1132957542807527e-05,
+      "loss": 0.0398,
+      "step": 2786
+    },
+    {
+      "epoch": 9.95357142857143,
+      "grad_norm": 0.3323599982374471,
+      "learning_rate": 2.1121408944743838e-05,
+      "loss": 0.0178,
+      "step": 2787
+    },
+    {
+      "epoch": 9.957142857142857,
+      "grad_norm": 0.31583860482372794,
+      "learning_rate": 2.11098599715826e-05,
+      "loss": 0.0306,
+      "step": 2788
+    },
+    {
+      "epoch": 9.960714285714285,
+      "grad_norm": 1.2443860179650836,
+      "learning_rate": 2.10983106271868e-05,
+      "loss": 0.0787,
+      "step": 2789
+    },
+    {
+      "epoch": 9.964285714285714,
+      "grad_norm": 0.7310962165040014,
+      "learning_rate": 2.108676091541956e-05,
+      "loss": 0.0471,
+      "step": 2790
+    },
+    {
+      "epoch": 9.967857142857143,
+      "grad_norm": 0.6908773935711429,
+      "learning_rate": 2.1075210840144108e-05,
+      "loss": 0.0554,
+      "step": 2791
+    },
+    {
+      "epoch": 9.971428571428572,
+      "grad_norm": 0.5668372913187126,
+      "learning_rate": 2.1063660405223808e-05,
+      "loss": 0.0455,
+      "step": 2792
+    },
+    {
+      "epoch": 9.975,
+      "grad_norm": 1.408079172636642,
+      "learning_rate": 2.1052109614522135e-05,
+      "loss": 0.1295,
+      "step": 2793
+    },
+    {
+      "epoch": 9.978571428571428,
+      "grad_norm": 0.7225307693351014,
+      "learning_rate": 2.1040558471902694e-05,
+      "loss": 0.0961,
+      "step": 2794
+    },
+    {
+      "epoch": 9.982142857142858,
+      "grad_norm": 0.8428822024315218,
+      "learning_rate": 2.1029006981229195e-05,
+      "loss": 0.0624,
+      "step": 2795
+    },
+    {
+      "epoch": 9.985714285714286,
+      "grad_norm": 0.32947461069728884,
+      "learning_rate": 2.101745514636547e-05,
+      "loss": 0.0343,
+      "step": 2796
+    },
+    {
+      "epoch": 9.989285714285714,
+      "grad_norm": 0.568263436606735,
+      "learning_rate": 2.1005902971175472e-05,
+      "loss": 0.0643,
+      "step": 2797
+    },
+    {
+      "epoch": 9.992857142857144,
+      "grad_norm": 0.4928645462561523,
+      "learning_rate": 2.0994350459523248e-05,
+      "loss": 0.0592,
+      "step": 2798
+    },
+    {
+      "epoch": 9.996428571428572,
+      "grad_norm": 1.1054609937867534,
+      "learning_rate": 2.098279761527299e-05,
+      "loss": 0.1191,
+      "step": 2799
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.6638464043134148,
+      "learning_rate": 2.097124444228897e-05,
+      "loss": 0.0877,
+      "step": 2800
+    },
+    {
+      "epoch": 10.003571428571428,
+      "grad_norm": 0.7161777889620585,
+      "learning_rate": 2.095969094443558e-05,
+      "loss": 0.0506,
+      "step": 2801
+    },
+    {
+      "epoch": 10.007142857142858,
+      "grad_norm": 0.5754943772334498,
+      "learning_rate": 2.0948137125577342e-05,
+      "loss": 0.0437,
+      "step": 2802
+    },
+    {
+      "epoch": 10.010714285714286,
+      "grad_norm": 0.5473545165853269,
+      "learning_rate": 2.0936582989578848e-05,
+      "loss": 0.0639,
+      "step": 2803
+    },
+    {
+      "epoch": 10.014285714285714,
+      "grad_norm": 0.44330861979008035,
+      "learning_rate": 2.092502854030482e-05,
+      "loss": 0.0571,
+      "step": 2804
+    },
+    {
+      "epoch": 10.017857142857142,
+      "grad_norm": 0.4732870460829038,
+      "learning_rate": 2.0913473781620088e-05,
+      "loss": 0.0341,
+      "step": 2805
+    },
+    {
+      "epoch": 10.021428571428572,
+      "grad_norm": 0.4673094807893729,
+      "learning_rate": 2.0901918717389567e-05,
+      "loss": 0.0582,
+      "step": 2806
+    },
+    {
+      "epoch": 10.025,
+      "grad_norm": 1.4669088165055413,
+      "learning_rate": 2.0890363351478288e-05,
+      "loss": 0.0879,
+      "step": 2807
+    },
+    {
+      "epoch": 10.028571428571428,
+      "grad_norm": 0.577552823075164,
+      "learning_rate": 2.087880768775138e-05,
+      "loss": 0.049,
+      "step": 2808
+    },
+    {
+      "epoch": 10.032142857142857,
+      "grad_norm": 0.7163131099354108,
+      "learning_rate": 2.086725173007407e-05,
+      "loss": 0.0601,
+      "step": 2809
+    },
+    {
+      "epoch": 10.035714285714286,
+      "grad_norm": 0.4133653715807572,
+      "learning_rate": 2.0855695482311687e-05,
+      "loss": 0.0259,
+      "step": 2810
+    },
+    {
+      "epoch": 10.039285714285715,
+      "grad_norm": 0.7074917771746546,
+      "learning_rate": 2.0844138948329648e-05,
+      "loss": 0.0852,
+      "step": 2811
+    },
+    {
+      "epoch": 10.042857142857143,
+      "grad_norm": 0.5020745978599952,
+      "learning_rate": 2.0832582131993475e-05,
+      "loss": 0.0564,
+      "step": 2812
+    },
+    {
+      "epoch": 10.04642857142857,
+      "grad_norm": 0.6673226180104362,
+      "learning_rate": 2.082102503716878e-05,
+      "loss": 0.0271,
+      "step": 2813
+    },
+    {
+      "epoch": 10.05,
+      "grad_norm": 0.45101020194554914,
+      "learning_rate": 2.0809467667721277e-05,
+      "loss": 0.0276,
+      "step": 2814
+    },
+    {
+      "epoch": 10.053571428571429,
+      "grad_norm": 0.7995425718958755,
+      "learning_rate": 2.0797910027516757e-05,
+      "loss": 0.0339,
+      "step": 2815
+    },
+    {
+      "epoch": 10.057142857142857,
+      "grad_norm": 0.7230796226216754,
+      "learning_rate": 2.0786352120421112e-05,
+      "loss": 0.0883,
+      "step": 2816
+    },
+    {
+      "epoch": 10.060714285714285,
+      "grad_norm": 1.0686061764373056,
+      "learning_rate": 2.0774793950300318e-05,
+      "loss": 0.0686,
+      "step": 2817
+    },
+    {
+      "epoch": 10.064285714285715,
+      "grad_norm": 0.5674035517545992,
+      "learning_rate": 2.0763235521020442e-05,
+      "loss": 0.0457,
+      "step": 2818
+    },
+    {
+      "epoch": 10.067857142857143,
+      "grad_norm": 0.5471792674761848,
+      "learning_rate": 2.0751676836447642e-05,
+      "loss": 0.0487,
+      "step": 2819
+    },
+    {
+      "epoch": 10.071428571428571,
+      "grad_norm": 0.6299467673472893,
+      "learning_rate": 2.074011790044815e-05,
+      "loss": 0.04,
+      "step": 2820
+    },
+    {
+      "epoch": 10.075,
+      "grad_norm": 0.5363203650116851,
+      "learning_rate": 2.072855871688829e-05,
+      "loss": 0.0573,
+      "step": 2821
+    },
+    {
+      "epoch": 10.07857142857143,
+      "grad_norm": 0.5485306860553332,
+      "learning_rate": 2.071699928963447e-05,
+      "loss": 0.0369,
+      "step": 2822
+    },
+    {
+      "epoch": 10.082142857142857,
+      "grad_norm": 0.39144282655132484,
+      "learning_rate": 2.0705439622553167e-05,
+      "loss": 0.0268,
+      "step": 2823
+    },
+    {
+      "epoch": 10.085714285714285,
+      "grad_norm": 0.858263860005623,
+      "learning_rate": 2.0693879719510962e-05,
+      "loss": 0.0504,
+      "step": 2824
+    },
+    {
+      "epoch": 10.089285714285714,
+      "grad_norm": 0.945045312868014,
+      "learning_rate": 2.068231958437449e-05,
+      "loss": 0.0547,
+      "step": 2825
+    },
+    {
+      "epoch": 10.092857142857143,
+      "grad_norm": 0.8306001797065758,
+      "learning_rate": 2.0670759221010483e-05,
+      "loss": 0.0496,
+      "step": 2826
+    },
+    {
+      "epoch": 10.096428571428572,
+      "grad_norm": 0.2759995146499259,
+      "learning_rate": 2.0659198633285733e-05,
+      "loss": 0.0271,
+      "step": 2827
+    },
+    {
+      "epoch": 10.1,
+      "grad_norm": 0.6518543604151986,
+      "learning_rate": 2.0647637825067123e-05,
+      "loss": 0.073,
+      "step": 2828
+    },
+    {
+      "epoch": 10.103571428571428,
+      "grad_norm": 0.600741127944514,
+      "learning_rate": 2.0636076800221595e-05,
+      "loss": 0.0402,
+      "step": 2829
+    },
+    {
+      "epoch": 10.107142857142858,
+      "grad_norm": 0.9656202138653676,
+      "learning_rate": 2.0624515562616166e-05,
+      "loss": 0.0732,
+      "step": 2830
+    },
+    {
+      "epoch": 10.110714285714286,
+      "grad_norm": 0.8456522911720228,
+      "learning_rate": 2.0612954116117946e-05,
+      "loss": 0.1158,
+      "step": 2831
+    },
+    {
+      "epoch": 10.114285714285714,
+      "grad_norm": 0.5252592045844757,
+      "learning_rate": 2.0601392464594082e-05,
+      "loss": 0.0716,
+      "step": 2832
+    },
+    {
+      "epoch": 10.117857142857142,
+      "grad_norm": 0.6200572826894512,
+      "learning_rate": 2.0589830611911806e-05,
+      "loss": 0.0449,
+      "step": 2833
+    },
+    {
+      "epoch": 10.121428571428572,
+      "grad_norm": 0.48861170561269623,
+      "learning_rate": 2.057826856193842e-05,
+      "loss": 0.0304,
+      "step": 2834
+    },
+    {
+      "epoch": 10.125,
+      "grad_norm": 0.5555077796395699,
+      "learning_rate": 2.056670631854129e-05,
+      "loss": 0.0547,
+      "step": 2835
+    },
+    {
+      "epoch": 10.128571428571428,
+      "grad_norm": 0.9390306372548302,
+      "learning_rate": 2.0555143885587837e-05,
+      "loss": 0.0516,
+      "step": 2836
+    },
+    {
+      "epoch": 10.132142857142858,
+      "grad_norm": 0.9675874991277102,
+      "learning_rate": 2.0543581266945556e-05,
+      "loss": 0.0938,
+      "step": 2837
+    },
+    {
+      "epoch": 10.135714285714286,
+      "grad_norm": 0.8619649375459468,
+      "learning_rate": 2.0532018466482005e-05,
+      "loss": 0.0632,
+      "step": 2838
+    },
+    {
+      "epoch": 10.139285714285714,
+      "grad_norm": 0.8540611121666298,
+      "learning_rate": 2.0520455488064797e-05,
+      "loss": 0.0459,
+      "step": 2839
+    },
+    {
+      "epoch": 10.142857142857142,
+      "grad_norm": 1.2232241954624987,
+      "learning_rate": 2.0508892335561607e-05,
+      "loss": 0.0622,
+      "step": 2840
+    },
+    {
+      "epoch": 10.146428571428572,
+      "grad_norm": 0.9654991095941894,
+      "learning_rate": 2.0497329012840166e-05,
+      "loss": 0.0959,
+      "step": 2841
+    },
+    {
+      "epoch": 10.15,
+      "grad_norm": 1.0015676030465661,
+      "learning_rate": 2.0485765523768265e-05,
+      "loss": 0.0573,
+      "step": 2842
+    },
+    {
+      "epoch": 10.153571428571428,
+      "grad_norm": 0.5760242430619013,
+      "learning_rate": 2.0474201872213747e-05,
+      "loss": 0.0493,
+      "step": 2843
+    },
+    {
+      "epoch": 10.157142857142857,
+      "grad_norm": 0.5684982762272885,
+      "learning_rate": 2.046263806204452e-05,
+      "loss": 0.043,
+      "step": 2844
+    },
+    {
+      "epoch": 10.160714285714286,
+      "grad_norm": 0.49890555544990645,
+      "learning_rate": 2.045107409712853e-05,
+      "loss": 0.0539,
+      "step": 2845
+    },
+    {
+      "epoch": 10.164285714285715,
+      "grad_norm": 0.7300631887316745,
+      "learning_rate": 2.0439509981333776e-05,
+      "loss": 0.0508,
+      "step": 2846
+    },
+    {
+      "epoch": 10.167857142857143,
+      "grad_norm": 0.6959320614803701,
+      "learning_rate": 2.0427945718528326e-05,
+      "loss": 0.048,
+      "step": 2847
+    },
+    {
+      "epoch": 10.17142857142857,
+      "grad_norm": 0.8959381630162413,
+      "learning_rate": 2.0416381312580275e-05,
+      "loss": 0.048,
+      "step": 2848
+    },
+    {
+      "epoch": 10.175,
+      "grad_norm": 0.5708267996395814,
+      "learning_rate": 2.0404816767357774e-05,
+      "loss": 0.0545,
+      "step": 2849
+    },
+    {
+      "epoch": 10.178571428571429,
+      "grad_norm": 1.297558709869371,
+      "learning_rate": 2.039325208672903e-05,
+      "loss": 0.0667,
+      "step": 2850
+    },
+    {
+      "epoch": 10.182142857142857,
+      "grad_norm": 0.5711675927431521,
+      "learning_rate": 2.038168727456228e-05,
+      "loss": 0.0319,
+      "step": 2851
+    },
+    {
+      "epoch": 10.185714285714285,
+      "grad_norm": 0.5598567086499713,
+      "learning_rate": 2.0370122334725818e-05,
+      "loss": 0.026,
+      "step": 2852
+    },
+    {
+      "epoch": 10.189285714285715,
+      "grad_norm": 0.908888446362121,
+      "learning_rate": 2.0358557271087966e-05,
+      "loss": 0.0323,
+      "step": 2853
+    },
+    {
+      "epoch": 10.192857142857143,
+      "grad_norm": 0.6860634726232858,
+      "learning_rate": 2.034699208751711e-05,
+      "loss": 0.0479,
+      "step": 2854
+    },
+    {
+      "epoch": 10.196428571428571,
+      "grad_norm": 0.9749511919370509,
+      "learning_rate": 2.0335426787881648e-05,
+      "loss": 0.0635,
+      "step": 2855
+    },
+    {
+      "epoch": 10.2,
+      "grad_norm": 1.0091507128174904,
+      "learning_rate": 2.0323861376050035e-05,
+      "loss": 0.0851,
+      "step": 2856
+    },
+    {
+      "epoch": 10.20357142857143,
+      "grad_norm": 0.7259416211806436,
+      "learning_rate": 2.031229585589076e-05,
+      "loss": 0.0393,
+      "step": 2857
+    },
+    {
+      "epoch": 10.207142857142857,
+      "grad_norm": 0.8890563241988404,
+      "learning_rate": 2.0300730231272347e-05,
+      "loss": 0.0794,
+      "step": 2858
+    },
+    {
+      "epoch": 10.210714285714285,
+      "grad_norm": 1.0986484376008152,
+      "learning_rate": 2.028916450606336e-05,
+      "loss": 0.0442,
+      "step": 2859
+    },
+    {
+      "epoch": 10.214285714285714,
+      "grad_norm": 1.6871962762401427,
+      "learning_rate": 2.0277598684132386e-05,
+      "loss": 0.0827,
+      "step": 2860
+    },
+    {
+      "epoch": 10.217857142857143,
+      "grad_norm": 0.7328766263415006,
+      "learning_rate": 2.0266032769348055e-05,
+      "loss": 0.068,
+      "step": 2861
+    },
+    {
+      "epoch": 10.221428571428572,
+      "grad_norm": 1.546064167746562,
+      "learning_rate": 2.0254466765579024e-05,
+      "loss": 0.1111,
+      "step": 2862
+    },
+    {
+      "epoch": 10.225,
+      "grad_norm": 0.7116807037828246,
+      "learning_rate": 2.024290067669398e-05,
+      "loss": 0.0506,
+      "step": 2863
+    },
+    {
+      "epoch": 10.228571428571428,
+      "grad_norm": 0.8661688987418102,
+      "learning_rate": 2.0231334506561634e-05,
+      "loss": 0.0763,
+      "step": 2864
+    },
+    {
+      "epoch": 10.232142857142858,
+      "grad_norm": 0.9149009741360072,
+      "learning_rate": 2.021976825905073e-05,
+      "loss": 0.0597,
+      "step": 2865
+    },
+    {
+      "epoch": 10.235714285714286,
+      "grad_norm": 1.2311149191953288,
+      "learning_rate": 2.0208201938030046e-05,
+      "loss": 0.0657,
+      "step": 2866
+    },
+    {
+      "epoch": 10.239285714285714,
+      "grad_norm": 1.0759987332692398,
+      "learning_rate": 2.0196635547368352e-05,
+      "loss": 0.0602,
+      "step": 2867
+    },
+    {
+      "epoch": 10.242857142857142,
+      "grad_norm": 0.44457874960389243,
+      "learning_rate": 2.0185069090934485e-05,
+      "loss": 0.0356,
+      "step": 2868
+    },
+    {
+      "epoch": 10.246428571428572,
+      "grad_norm": 0.296290112058532,
+      "learning_rate": 2.017350257259727e-05,
+      "loss": 0.0371,
+      "step": 2869
+    },
+    {
+      "epoch": 10.25,
+      "grad_norm": 1.0200664548617373,
+      "learning_rate": 2.0161935996225573e-05,
+      "loss": 0.0795,
+      "step": 2870
+    },
+    {
+      "epoch": 10.253571428571428,
+      "grad_norm": 0.5348972362451084,
+      "learning_rate": 2.0150369365688275e-05,
+      "loss": 0.0365,
+      "step": 2871
+    },
+    {
+      "epoch": 10.257142857142856,
+      "grad_norm": 1.5757928135829815,
+      "learning_rate": 2.013880268485426e-05,
+      "loss": 0.0318,
+      "step": 2872
+    },
+    {
+      "epoch": 10.260714285714286,
+      "grad_norm": 0.7108285455363333,
+      "learning_rate": 2.012723595759245e-05,
+      "loss": 0.0619,
+      "step": 2873
+    },
+    {
+      "epoch": 10.264285714285714,
+      "grad_norm": 0.4640698039661836,
+      "learning_rate": 2.0115669187771768e-05,
+      "loss": 0.0448,
+      "step": 2874
+    },
+    {
+      "epoch": 10.267857142857142,
+      "grad_norm": 0.9292674122301323,
+      "learning_rate": 2.0104102379261165e-05,
+      "loss": 0.0945,
+      "step": 2875
+    },
+    {
+      "epoch": 10.271428571428572,
+      "grad_norm": 0.5725046535570756,
+      "learning_rate": 2.009253553592959e-05,
+      "loss": 0.0445,
+      "step": 2876
+    },
+    {
+      "epoch": 10.275,
+      "grad_norm": 1.4963510897134096,
+      "learning_rate": 2.0080968661646013e-05,
+      "loss": 0.0386,
+      "step": 2877
+    },
+    {
+      "epoch": 10.278571428571428,
+      "grad_norm": 0.9439040572715294,
+      "learning_rate": 2.0069401760279413e-05,
+      "loss": 0.0627,
+      "step": 2878
+    },
+    {
+      "epoch": 10.282142857142857,
+      "grad_norm": 0.4327275306877417,
+      "learning_rate": 2.005783483569877e-05,
+      "loss": 0.0306,
+      "step": 2879
+    },
+    {
+      "epoch": 10.285714285714286,
+      "grad_norm": 1.6321017549197971,
+      "learning_rate": 2.004626789177309e-05,
+      "loss": 0.0904,
+      "step": 2880
+    },
+    {
+      "epoch": 10.289285714285715,
+      "grad_norm": 0.6131208568446622,
+      "learning_rate": 2.003470093237136e-05,
+      "loss": 0.0392,
+      "step": 2881
+    },
+    {
+      "epoch": 10.292857142857143,
+      "grad_norm": 0.8300803511253126,
+      "learning_rate": 2.00231339613626e-05,
+      "loss": 0.1227,
+      "step": 2882
+    },
+    {
+      "epoch": 10.29642857142857,
+      "grad_norm": 0.5705632283667403,
+      "learning_rate": 2.0011566982615807e-05,
+      "loss": 0.0685,
+      "step": 2883
+    },
+    {
+      "epoch": 10.3,
+      "grad_norm": 0.940129489380279,
+      "learning_rate": 2e-05,
+      "loss": 0.0976,
+      "step": 2884
+    },
+    {
+      "epoch": 10.303571428571429,
+      "grad_norm": 0.6018888160610654,
+      "learning_rate": 1.99884330173842e-05,
+      "loss": 0.0806,
+      "step": 2885
+    },
+    {
+      "epoch": 10.307142857142857,
+      "grad_norm": 0.6496784869566228,
+      "learning_rate": 1.9976866038637407e-05,
+      "loss": 0.0465,
+      "step": 2886
+    },
+    {
+      "epoch": 10.310714285714285,
+      "grad_norm": 0.6000998378051325,
+      "learning_rate": 1.9965299067628643e-05,
+      "loss": 0.0326,
+      "step": 2887
+    },
+    {
+      "epoch": 10.314285714285715,
+      "grad_norm": 0.6188129125973005,
+      "learning_rate": 1.9953732108226917e-05,
+      "loss": 0.0458,
+      "step": 2888
+    },
+    {
+      "epoch": 10.317857142857143,
+      "grad_norm": 0.6698113357569176,
+      "learning_rate": 1.9942165164301232e-05,
+      "loss": 0.034,
+      "step": 2889
+    },
+    {
+      "epoch": 10.321428571428571,
+      "grad_norm": 1.1825736157512208,
+      "learning_rate": 1.993059823972059e-05,
+      "loss": 0.0503,
+      "step": 2890
+    },
+    {
+      "epoch": 10.325,
+      "grad_norm": 0.5659014590639861,
+      "learning_rate": 1.9919031338353994e-05,
+      "loss": 0.0345,
+      "step": 2891
+    },
+    {
+      "epoch": 10.32857142857143,
+      "grad_norm": 0.6415822652967791,
+      "learning_rate": 1.9907464464070416e-05,
+      "loss": 0.0356,
+      "step": 2892
+    },
+    {
+      "epoch": 10.332142857142857,
+      "grad_norm": 0.6213629069285482,
+      "learning_rate": 1.989589762073884e-05,
+      "loss": 0.0563,
+      "step": 2893
+    },
+    {
+      "epoch": 10.335714285714285,
+      "grad_norm": 0.7539118321241752,
+      "learning_rate": 1.9884330812228235e-05,
+      "loss": 0.0842,
+      "step": 2894
+    },
+    {
+      "epoch": 10.339285714285714,
+      "grad_norm": 1.3445536634160065,
+      "learning_rate": 1.9872764042407557e-05,
+      "loss": 0.0663,
+      "step": 2895
+    },
+    {
+      "epoch": 10.342857142857143,
+      "grad_norm": 0.44771439834368904,
+      "learning_rate": 1.9861197315145747e-05,
+      "loss": 0.0376,
+      "step": 2896
+    },
+    {
+      "epoch": 10.346428571428572,
+      "grad_norm": 1.65637368289212,
+      "learning_rate": 1.9849630634311735e-05,
+      "loss": 0.0803,
+      "step": 2897
+    },
+    {
+      "epoch": 10.35,
+      "grad_norm": 0.5631853890820345,
+      "learning_rate": 1.983806400377443e-05,
+      "loss": 0.0294,
+      "step": 2898
+    },
+    {
+      "epoch": 10.353571428571428,
+      "grad_norm": 0.7175663298933782,
+      "learning_rate": 1.9826497427402734e-05,
+      "loss": 0.0519,
+      "step": 2899
+    },
+    {
+      "epoch": 10.357142857142858,
+      "grad_norm": 0.6501560617838033,
+      "learning_rate": 1.981493090906552e-05,
+      "loss": 0.0796,
+      "step": 2900
+    },
+    {
+      "epoch": 10.360714285714286,
+      "grad_norm": 0.8899575173135584,
+      "learning_rate": 1.980336445263165e-05,
+      "loss": 0.0482,
+      "step": 2901
+    },
+    {
+      "epoch": 10.364285714285714,
+      "grad_norm": 1.076929443228369,
+      "learning_rate": 1.979179806196996e-05,
+      "loss": 0.092,
+      "step": 2902
+    },
+    {
+      "epoch": 10.367857142857144,
+      "grad_norm": 0.6589555096897288,
+      "learning_rate": 1.9780231740949273e-05,
+      "loss": 0.0521,
+      "step": 2903
+    },
+    {
+      "epoch": 10.371428571428572,
+      "grad_norm": 0.4696711691083142,
+      "learning_rate": 1.9768665493438373e-05,
+      "loss": 0.0471,
+      "step": 2904
+    },
+    {
+      "epoch": 10.375,
+      "grad_norm": 0.6123356987166726,
+      "learning_rate": 1.9757099323306027e-05,
+      "loss": 0.0582,
+      "step": 2905
+    },
+    {
+      "epoch": 10.378571428571428,
+      "grad_norm": 0.842956146058949,
+      "learning_rate": 1.9745533234420982e-05,
+      "loss": 0.0494,
+      "step": 2906
+    },
+    {
+      "epoch": 10.382142857142856,
+      "grad_norm": 0.6855224761431693,
+      "learning_rate": 1.9733967230651948e-05,
+      "loss": 0.05,
+      "step": 2907
+    },
+    {
+      "epoch": 10.385714285714286,
+      "grad_norm": 0.5302559221992539,
+      "learning_rate": 1.9722401315867618e-05,
+      "loss": 0.0251,
+      "step": 2908
+    },
+    {
+      "epoch": 10.389285714285714,
+      "grad_norm": 0.37086672767717604,
+      "learning_rate": 1.971083549393665e-05,
+      "loss": 0.0278,
+      "step": 2909
+    },
+    {
+      "epoch": 10.392857142857142,
+      "grad_norm": 0.6062612225366985,
+      "learning_rate": 1.969926976872766e-05,
+      "loss": 0.0537,
+      "step": 2910
+    },
+    {
+      "epoch": 10.396428571428572,
+      "grad_norm": 0.9648355982461191,
+      "learning_rate": 1.9687704144109248e-05,
+      "loss": 0.0471,
+      "step": 2911
+    },
+    {
+      "epoch": 10.4,
+      "grad_norm": 1.0416495108060913,
+      "learning_rate": 1.967613862394997e-05,
+      "loss": 0.0553,
+      "step": 2912
+    },
+    {
+      "epoch": 10.403571428571428,
+      "grad_norm": 0.5435234499273289,
+      "learning_rate": 1.966457321211836e-05,
+      "loss": 0.0206,
+      "step": 2913
+    },
+    {
+      "epoch": 10.407142857142857,
+      "grad_norm": 0.5079038679932493,
+      "learning_rate": 1.9653007912482894e-05,
+      "loss": 0.0472,
+      "step": 2914
+    },
+    {
+      "epoch": 10.410714285714286,
+      "grad_norm": 0.5897674855154287,
+      "learning_rate": 1.9641442728912037e-05,
+      "loss": 0.0279,
+      "step": 2915
+    },
+    {
+      "epoch": 10.414285714285715,
+      "grad_norm": 0.5261491360247622,
+      "learning_rate": 1.962987766527419e-05,
+      "loss": 0.035,
+      "step": 2916
+    },
+    {
+      "epoch": 10.417857142857143,
+      "grad_norm": 0.6565773576211303,
+      "learning_rate": 1.9618312725437723e-05,
+      "loss": 0.0399,
+      "step": 2917
+    },
+    {
+      "epoch": 10.42142857142857,
+      "grad_norm": 1.131116427772526,
+      "learning_rate": 1.9606747913270973e-05,
+      "loss": 0.0871,
+      "step": 2918
+    },
+    {
+      "epoch": 10.425,
+      "grad_norm": 0.8590485532576627,
+      "learning_rate": 1.959518323264223e-05,
+      "loss": 0.1027,
+      "step": 2919
+    },
+    {
+      "epoch": 10.428571428571429,
+      "grad_norm": 1.0605710648795297,
+      "learning_rate": 1.958361868741973e-05,
+      "loss": 0.0842,
+      "step": 2920
+    },
+    {
+      "epoch": 10.432142857142857,
+      "grad_norm": 0.822383806571356,
+      "learning_rate": 1.957205428147168e-05,
+      "loss": 0.0693,
+      "step": 2921
+    },
+    {
+      "epoch": 10.435714285714285,
+      "grad_norm": 0.6710946337232778,
+      "learning_rate": 1.956049001866623e-05,
+      "loss": 0.0647,
+      "step": 2922
+    },
+    {
+      "epoch": 10.439285714285715,
+      "grad_norm": 0.6373971221644786,
+      "learning_rate": 1.9548925902871477e-05,
+      "loss": 0.0675,
+      "step": 2923
+    },
+    {
+      "epoch": 10.442857142857143,
+      "grad_norm": 0.9300142912716366,
+      "learning_rate": 1.9537361937955487e-05,
+      "loss": 0.0824,
+      "step": 2924
+    },
+    {
+      "epoch": 10.446428571428571,
+      "grad_norm": 0.5249150989335596,
+      "learning_rate": 1.9525798127786257e-05,
+      "loss": 0.0485,
+      "step": 2925
+    },
+    {
+      "epoch": 10.45,
+      "grad_norm": 0.5843986937086613,
+      "learning_rate": 1.9514234476231742e-05,
+      "loss": 0.023,
+      "step": 2926
+    },
+    {
+      "epoch": 10.45357142857143,
+      "grad_norm": 0.9569993458692233,
+      "learning_rate": 1.9502670987159844e-05,
+      "loss": 0.0438,
+      "step": 2927
+    },
+    {
+      "epoch": 10.457142857142857,
+      "grad_norm": 0.49471514902806124,
+      "learning_rate": 1.9491107664438403e-05,
+      "loss": 0.0327,
+      "step": 2928
+    },
+    {
+      "epoch": 10.460714285714285,
+      "grad_norm": 0.9726428937252292,
+      "learning_rate": 1.9479544511935213e-05,
+      "loss": 0.0797,
+      "step": 2929
+    },
+    {
+      "epoch": 10.464285714285714,
+      "grad_norm": 0.3325858594841096,
+      "learning_rate": 1.9467981533518005e-05,
+      "loss": 0.0251,
+      "step": 2930
+    },
+    {
+      "epoch": 10.467857142857143,
+      "grad_norm": 1.351125671851442,
+      "learning_rate": 1.9456418733054454e-05,
+      "loss": 0.0844,
+      "step": 2931
+    },
+    {
+      "epoch": 10.471428571428572,
+      "grad_norm": 0.6960078272916675,
+      "learning_rate": 1.9444856114412176e-05,
+      "loss": 0.076,
+      "step": 2932
+    },
+    {
+      "epoch": 10.475,
+      "grad_norm": 0.8865719524177519,
+      "learning_rate": 1.943329368145872e-05,
+      "loss": 0.0524,
+      "step": 2933
+    },
+    {
+      "epoch": 10.478571428571428,
+      "grad_norm": 0.7492104795162168,
+      "learning_rate": 1.942173143806159e-05,
+      "loss": 0.0424,
+      "step": 2934
+    },
+    {
+      "epoch": 10.482142857142858,
+      "grad_norm": 1.036094751566807,
+      "learning_rate": 1.9410169388088204e-05,
+      "loss": 0.0899,
+      "step": 2935
+    },
+    {
+      "epoch": 10.485714285714286,
+      "grad_norm": 0.37031401372932266,
+      "learning_rate": 1.9398607535405928e-05,
+      "loss": 0.0313,
+      "step": 2936
+    },
+    {
+      "epoch": 10.489285714285714,
+      "grad_norm": 0.718590586260465,
+      "learning_rate": 1.9387045883882064e-05,
+      "loss": 0.0655,
+      "step": 2937
+    },
+    {
+      "epoch": 10.492857142857144,
+      "grad_norm": 0.9707674413180071,
+      "learning_rate": 1.9375484437383837e-05,
+      "loss": 0.0935,
+      "step": 2938
+    },
+    {
+      "epoch": 10.496428571428572,
+      "grad_norm": 0.26792405210286646,
+      "learning_rate": 1.9363923199778415e-05,
+      "loss": 0.0234,
+      "step": 2939
+    },
+    {
+      "epoch": 10.5,
+      "grad_norm": 0.6428595942759616,
+      "learning_rate": 1.9352362174932887e-05,
+      "loss": 0.0509,
+      "step": 2940
+    },
+    {
+      "epoch": 10.503571428571428,
+      "grad_norm": 0.7856447132126981,
+      "learning_rate": 1.9340801366714277e-05,
+      "loss": 0.0903,
+      "step": 2941
+    },
+    {
+      "epoch": 10.507142857142856,
+      "grad_norm": 0.5505593973076193,
+      "learning_rate": 1.9329240778989527e-05,
+      "loss": 0.059,
+      "step": 2942
+    },
+    {
+      "epoch": 10.510714285714286,
+      "grad_norm": 0.7587776430232979,
+      "learning_rate": 1.931768041562552e-05,
+      "loss": 0.0379,
+      "step": 2943
+    },
+    {
+      "epoch": 10.514285714285714,
+      "grad_norm": 0.8492077159597945,
+      "learning_rate": 1.9306120280489048e-05,
+      "loss": 0.0357,
+      "step": 2944
+    },
+    {
+      "epoch": 10.517857142857142,
+      "grad_norm": 0.8513414555795104,
+      "learning_rate": 1.929456037744684e-05,
+      "loss": 0.0578,
+      "step": 2945
+    },
+    {
+      "epoch": 10.521428571428572,
+      "grad_norm": 0.586250582040497,
+      "learning_rate": 1.928300071036554e-05,
+      "loss": 0.0544,
+      "step": 2946
+    },
+    {
+      "epoch": 10.525,
+      "grad_norm": 0.9663692862990074,
+      "learning_rate": 1.927144128311172e-05,
+      "loss": 0.0586,
+      "step": 2947
+    },
+    {
+      "epoch": 10.528571428571428,
+      "grad_norm": 0.6663656651179961,
+      "learning_rate": 1.925988209955186e-05,
+      "loss": 0.0697,
+      "step": 2948
+    },
+    {
+      "epoch": 10.532142857142857,
+      "grad_norm": 0.7343031379327044,
+      "learning_rate": 1.9248323163552365e-05,
+      "loss": 0.0192,
+      "step": 2949
+    },
+    {
+      "epoch": 10.535714285714286,
+      "grad_norm": 0.8982656928796485,
+      "learning_rate": 1.923676447897956e-05,
+      "loss": 0.0473,
+      "step": 2950
+    },
+    {
+      "epoch": 10.539285714285715,
+      "grad_norm": 0.6459093708296627,
+      "learning_rate": 1.922520604969969e-05,
+      "loss": 0.0683,
+      "step": 2951
+    },
+    {
+      "epoch": 10.542857142857143,
+      "grad_norm": 0.6053610394407971,
+      "learning_rate": 1.9213647879578898e-05,
+      "loss": 0.058,
+      "step": 2952
+    },
+    {
+      "epoch": 10.54642857142857,
+      "grad_norm": 0.7502276860712761,
+      "learning_rate": 1.9202089972483253e-05,
+      "loss": 0.1016,
+      "step": 2953
+    },
+    {
+      "epoch": 10.55,
+      "grad_norm": 0.5357323746646512,
+      "learning_rate": 1.9190532332278733e-05,
+      "loss": 0.0551,
+      "step": 2954
+    },
+    {
+      "epoch": 10.553571428571429,
+      "grad_norm": 0.5720850778766904,
+      "learning_rate": 1.9178974962831227e-05,
+      "loss": 0.0582,
+      "step": 2955
+    },
+    {
+      "epoch": 10.557142857142857,
+      "grad_norm": 1.3171961125522371,
+      "learning_rate": 1.9167417868006535e-05,
+      "loss": 0.0901,
+      "step": 2956
+    },
+    {
+      "epoch": 10.560714285714285,
+      "grad_norm": 0.6926061932115317,
+      "learning_rate": 1.9155861051670362e-05,
+      "loss": 0.0711,
+      "step": 2957
+    },
+    {
+      "epoch": 10.564285714285715,
+      "grad_norm": 0.6134916210671876,
+      "learning_rate": 1.9144304517688326e-05,
+      "loss": 0.0448,
+      "step": 2958
+    },
+    {
+      "epoch": 10.567857142857143,
+      "grad_norm": 0.680003811128146,
+      "learning_rate": 1.913274826992594e-05,
+      "loss": 0.0453,
+      "step": 2959
+    },
+    {
+      "epoch": 10.571428571428571,
+      "grad_norm": 0.862516007796391,
+      "learning_rate": 1.9121192312248627e-05,
+      "loss": 0.0929,
+      "step": 2960
+    },
+    {
+      "epoch": 10.575,
+      "grad_norm": 0.9286178039571926,
+      "learning_rate": 1.910963664852172e-05,
+      "loss": 0.0629,
+      "step": 2961
+    },
+    {
+      "epoch": 10.57857142857143,
+      "grad_norm": 0.6901417261393803,
+      "learning_rate": 1.909808128261044e-05,
+      "loss": 0.0687,
+      "step": 2962
+    },
+    {
+      "epoch": 10.582142857142857,
+      "grad_norm": 0.8091078043986958,
+      "learning_rate": 1.908652621837992e-05,
+      "loss": 0.0579,
+      "step": 2963
+    },
+    {
+      "epoch": 10.585714285714285,
+      "grad_norm": 0.7045141937529862,
+      "learning_rate": 1.907497145969519e-05,
+      "loss": 0.0571,
+      "step": 2964
+    },
+    {
+      "epoch": 10.589285714285714,
+      "grad_norm": 0.7216655586613726,
+      "learning_rate": 1.9063417010421163e-05,
+      "loss": 0.0436,
+      "step": 2965
+    },
+    {
+      "epoch": 10.592857142857143,
+      "grad_norm": 0.9452824206039484,
+      "learning_rate": 1.905186287442267e-05,
+      "loss": 0.0647,
+      "step": 2966
+    },
+    {
+      "epoch": 10.596428571428572,
+      "grad_norm": 0.5807840827117725,
+      "learning_rate": 1.9040309055564425e-05,
+      "loss": 0.0785,
+      "step": 2967
+    },
+    {
+      "epoch": 10.6,
+      "grad_norm": 1.1288586487838284,
+      "learning_rate": 1.9028755557711043e-05,
+      "loss": 0.0842,
+      "step": 2968
+    },
+    {
+      "epoch": 10.603571428571428,
+      "grad_norm": 0.4576468723379745,
+      "learning_rate": 1.9017202384727013e-05,
+      "loss": 0.0585,
+      "step": 2969
+    },
+    {
+      "epoch": 10.607142857142858,
+      "grad_norm": 1.2086863530311414,
+      "learning_rate": 1.9005649540476752e-05,
+      "loss": 0.0648,
+      "step": 2970
+    },
+    {
+      "epoch": 10.610714285714286,
+      "grad_norm": 0.8999165862366966,
+      "learning_rate": 1.8994097028824535e-05,
+      "loss": 0.0895,
+      "step": 2971
+    },
+    {
+      "epoch": 10.614285714285714,
+      "grad_norm": 0.8190105303403288,
+      "learning_rate": 1.8982544853634533e-05,
+      "loss": 0.0916,
+      "step": 2972
+    },
+    {
+      "epoch": 10.617857142857144,
+      "grad_norm": 0.3461674346849705,
+      "learning_rate": 1.897099301877081e-05,
+      "loss": 0.0252,
+      "step": 2973
+    },
+    {
+      "epoch": 10.621428571428572,
+      "grad_norm": 0.5557490022117234,
+      "learning_rate": 1.895944152809731e-05,
+      "loss": 0.0485,
+      "step": 2974
+    },
+    {
+      "epoch": 10.625,
+      "grad_norm": 0.8245300486455402,
+      "learning_rate": 1.8947890385477865e-05,
+      "loss": 0.0477,
+      "step": 2975
+    },
+    {
+      "epoch": 10.628571428571428,
+      "grad_norm": 1.3737898594511382,
+      "learning_rate": 1.8936339594776195e-05,
+      "loss": 0.0894,
+      "step": 2976
+    },
+    {
+      "epoch": 10.632142857142856,
+      "grad_norm": 0.5564396604195343,
+      "learning_rate": 1.8924789159855895e-05,
+      "loss": 0.0365,
+      "step": 2977
+    },
+    {
+      "epoch": 10.635714285714286,
+      "grad_norm": 0.5782116658060663,
+      "learning_rate": 1.891323908458044e-05,
+      "loss": 0.0445,
+      "step": 2978
+    },
+    {
+      "epoch": 10.639285714285714,
+      "grad_norm": 0.5484143809002796,
+      "learning_rate": 1.89016893728132e-05,
+      "loss": 0.0511,
+      "step": 2979
+    },
+    {
+      "epoch": 10.642857142857142,
+      "grad_norm": 0.6967157201810433,
+      "learning_rate": 1.88901400284174e-05,
+      "loss": 0.0778,
+      "step": 2980
+    },
+    {
+      "epoch": 10.646428571428572,
+      "grad_norm": 0.8760214859735271,
+      "learning_rate": 1.8878591055256165e-05,
+      "loss": 0.0467,
+      "step": 2981
+    },
+    {
+      "epoch": 10.65,
+      "grad_norm": 0.5211028281439638,
+      "learning_rate": 1.8867042457192473e-05,
+      "loss": 0.0502,
+      "step": 2982
+    },
+    {
+      "epoch": 10.653571428571428,
+      "grad_norm": 0.7123212096198962,
+      "learning_rate": 1.8855494238089197e-05,
+      "loss": 0.0439,
+      "step": 2983
+    },
+    {
+      "epoch": 10.657142857142857,
+      "grad_norm": 0.3732640493576592,
+      "learning_rate": 1.8843946401809076e-05,
+      "loss": 0.0159,
+      "step": 2984
+    },
+    {
+      "epoch": 10.660714285714286,
+      "grad_norm": 1.567659209033469,
+      "learning_rate": 1.8832398952214714e-05,
+      "loss": 0.1027,
+      "step": 2985
+    },
+    {
+      "epoch": 10.664285714285715,
+      "grad_norm": 0.4960926560540979,
+      "learning_rate": 1.8820851893168593e-05,
+      "loss": 0.0202,
+      "step": 2986
+    },
+    {
+      "epoch": 10.667857142857143,
+      "grad_norm": 0.6822855296593388,
+      "learning_rate": 1.8809305228533064e-05,
+      "loss": 0.0361,
+      "step": 2987
+    },
+    {
+      "epoch": 10.67142857142857,
+      "grad_norm": 0.4683212318685272,
+      "learning_rate": 1.8797758962170342e-05,
+      "loss": 0.05,
+      "step": 2988
+    },
+    {
+      "epoch": 10.675,
+      "grad_norm": 1.0846092599894335,
+      "learning_rate": 1.8786213097942513e-05,
+      "loss": 0.0394,
+      "step": 2989
+    },
+    {
+      "epoch": 10.678571428571429,
+      "grad_norm": 1.4885161608215403,
+      "learning_rate": 1.877466763971153e-05,
+      "loss": 0.0737,
+      "step": 2990
+    },
+    {
+      "epoch": 10.682142857142857,
+      "grad_norm": 1.3014875376983968,
+      "learning_rate": 1.8763122591339206e-05,
+      "loss": 0.0787,
+      "step": 2991
+    },
+    {
+      "epoch": 10.685714285714285,
+      "grad_norm": 0.8348579122361156,
+      "learning_rate": 1.8751577956687216e-05,
+      "loss": 0.0383,
+      "step": 2992
+    },
+    {
+      "epoch": 10.689285714285715,
+      "grad_norm": 0.8119555413465663,
+      "learning_rate": 1.87400337396171e-05,
+      "loss": 0.0568,
+      "step": 2993
+    },
+    {
+      "epoch": 10.692857142857143,
+      "grad_norm": 1.1183809761308943,
+      "learning_rate": 1.872848994399026e-05,
+      "loss": 0.1071,
+      "step": 2994
+    },
+    {
+      "epoch": 10.696428571428571,
+      "grad_norm": 0.9206072439145958,
+      "learning_rate": 1.8716946573667947e-05,
+      "loss": 0.0719,
+      "step": 2995
+    },
+    {
+      "epoch": 10.7,
+      "grad_norm": 1.0115395366175135,
+      "learning_rate": 1.8705403632511286e-05,
+      "loss": 0.0441,
+      "step": 2996
+    },
+    {
+      "epoch": 10.70357142857143,
+      "grad_norm": 1.2706269380911162,
+      "learning_rate": 1.8693861124381243e-05,
+      "loss": 0.072,
+      "step": 2997
+    },
+    {
+      "epoch": 10.707142857142857,
+      "grad_norm": 0.7312970421787034,
+      "learning_rate": 1.8682319053138647e-05,
+      "loss": 0.0544,
+      "step": 2998
+    },
+    {
+      "epoch": 10.710714285714285,
+      "grad_norm": 0.6978499738984506,
+      "learning_rate": 1.867077742264418e-05,
+      "loss": 0.0315,
+      "step": 2999
+    },
+    {
+      "epoch": 10.714285714285714,
+      "grad_norm": 0.9017068116144364,
+      "learning_rate": 1.8659236236758375e-05,
+      "loss": 0.0546,
+      "step": 3000
+    },
+    {
+      "epoch": 10.717857142857143,
+      "grad_norm": 1.408368486187249,
+      "learning_rate": 1.8647695499341617e-05,
+      "loss": 0.0916,
+      "step": 3001
+    },
+    {
+      "epoch": 10.721428571428572,
+      "grad_norm": 0.94310654760992,
+      "learning_rate": 1.863615521425414e-05,
+      "loss": 0.0401,
+      "step": 3002
+    },
+    {
+      "epoch": 10.725,
+      "grad_norm": 1.0311512280451678,
+      "learning_rate": 1.8624615385356032e-05,
+      "loss": 0.0742,
+      "step": 3003
+    },
+    {
+      "epoch": 10.728571428571428,
+      "grad_norm": 0.761631630116814,
+      "learning_rate": 1.8613076016507223e-05,
+      "loss": 0.0533,
+      "step": 3004
+    },
+    {
+      "epoch": 10.732142857142858,
+      "grad_norm": 1.1018440331420347,
+      "learning_rate": 1.8601537111567493e-05,
+      "loss": 0.0993,
+      "step": 3005
+    },
+    {
+      "epoch": 10.735714285714286,
+      "grad_norm": 0.7001518855534044,
+      "learning_rate": 1.8589998674396453e-05,
+      "loss": 0.0419,
+      "step": 3006
+    },
+    {
+      "epoch": 10.739285714285714,
+      "grad_norm": 0.36673039315415573,
+      "learning_rate": 1.857846070885358e-05,
+      "loss": 0.0202,
+      "step": 3007
+    },
+    {
+      "epoch": 10.742857142857144,
+      "grad_norm": 0.6712988475603748,
+      "learning_rate": 1.8566923218798188e-05,
+      "loss": 0.0431,
+      "step": 3008
+    },
+    {
+      "epoch": 10.746428571428572,
+      "grad_norm": 0.7426662115470223,
+      "learning_rate": 1.855538620808941e-05,
+      "loss": 0.0662,
+      "step": 3009
+    },
+    {
+      "epoch": 10.75,
+      "grad_norm": 0.6808944428829229,
+      "learning_rate": 1.854384968058624e-05,
+      "loss": 0.0678,
+      "step": 3010
+    },
+    {
+      "epoch": 10.753571428571428,
+      "grad_norm": 1.2415673952485633,
+      "learning_rate": 1.853231364014751e-05,
+      "loss": 0.0651,
+      "step": 3011
+    },
+    {
+      "epoch": 10.757142857142856,
+      "grad_norm": 1.0859392995013277,
+      "learning_rate": 1.852077809063188e-05,
+      "loss": 0.091,
+      "step": 3012
+    },
+    {
+      "epoch": 10.760714285714286,
+      "grad_norm": 0.6509951994128335,
+      "learning_rate": 1.850924303589785e-05,
+      "loss": 0.0466,
+      "step": 3013
+    },
+    {
+      "epoch": 10.764285714285714,
+      "grad_norm": 0.658247265092681,
+      "learning_rate": 1.8497708479803754e-05,
+      "loss": 0.0674,
+      "step": 3014
+    },
+    {
+      "epoch": 10.767857142857142,
+      "grad_norm": 0.45364738774732466,
+      "learning_rate": 1.848617442620776e-05,
+      "loss": 0.039,
+      "step": 3015
+    },
+    {
+      "epoch": 10.771428571428572,
+      "grad_norm": 0.747532869172951,
+      "learning_rate": 1.8474640878967872e-05,
+      "loss": 0.0684,
+      "step": 3016
+    },
+    {
+      "epoch": 10.775,
+      "grad_norm": 0.8407281081277329,
+      "learning_rate": 1.8463107841941915e-05,
+      "loss": 0.0508,
+      "step": 3017
+    },
+    {
+      "epoch": 10.778571428571428,
+      "grad_norm": 0.9113998236495843,
+      "learning_rate": 1.8451575318987552e-05,
+      "loss": 0.091,
+      "step": 3018
+    },
+    {
+      "epoch": 10.782142857142857,
+      "grad_norm": 0.9571819792500953,
+      "learning_rate": 1.844004331396227e-05,
+      "loss": 0.0888,
+      "step": 3019
+    },
+    {
+      "epoch": 10.785714285714286,
+      "grad_norm": 1.1735352217779071,
+      "learning_rate": 1.8428511830723384e-05,
+      "loss": 0.1303,
+      "step": 3020
+    },
+    {
+      "epoch": 10.789285714285715,
+      "grad_norm": 0.858518634429606,
+      "learning_rate": 1.841698087312803e-05,
+      "loss": 0.0843,
+      "step": 3021
+    },
+    {
+      "epoch": 10.792857142857143,
+      "grad_norm": 1.0239036246819038,
+      "learning_rate": 1.840545044503317e-05,
+      "loss": 0.093,
+      "step": 3022
+    },
+    {
+      "epoch": 10.79642857142857,
+      "grad_norm": 1.0946535857979391,
+      "learning_rate": 1.83939205502956e-05,
+      "loss": 0.0715,
+      "step": 3023
+    },
+    {
+      "epoch": 10.8,
+      "grad_norm": 0.6419495710837825,
+      "learning_rate": 1.838239119277192e-05,
+      "loss": 0.0633,
+      "step": 3024
+    },
+    {
+      "epoch": 10.803571428571429,
+      "grad_norm": 0.6908363824870192,
+      "learning_rate": 1.8370862376318564e-05,
+      "loss": 0.078,
+      "step": 3025
+    },
+    {
+      "epoch": 10.807142857142857,
+      "grad_norm": 0.7367143461548613,
+      "learning_rate": 1.8359334104791777e-05,
+      "loss": 0.0782,
+      "step": 3026
+    },
+    {
+      "epoch": 10.810714285714285,
+      "grad_norm": 0.6671558804497557,
+      "learning_rate": 1.8347806382047625e-05,
+      "loss": 0.0824,
+      "step": 3027
+    },
+    {
+      "epoch": 10.814285714285715,
+      "grad_norm": 0.42455763059034574,
+      "learning_rate": 1.833627921194199e-05,
+      "loss": 0.0294,
+      "step": 3028
+    },
+    {
+      "epoch": 10.817857142857143,
+      "grad_norm": 0.6036512990485147,
+      "learning_rate": 1.8324752598330572e-05,
+      "loss": 0.0656,
+      "step": 3029
+    },
+    {
+      "epoch": 10.821428571428571,
+      "grad_norm": 0.7890740106942237,
+      "learning_rate": 1.831322654506888e-05,
+      "loss": 0.0578,
+      "step": 3030
+    },
+    {
+      "epoch": 10.825,
+      "grad_norm": 0.3790336458013265,
+      "learning_rate": 1.8301701056012236e-05,
+      "loss": 0.0423,
+      "step": 3031
+    },
+    {
+      "epoch": 10.82857142857143,
+      "grad_norm": 0.4134330231224775,
+      "learning_rate": 1.829017613501578e-05,
+      "loss": 0.0356,
+      "step": 3032
+    },
+    {
+      "epoch": 10.832142857142857,
+      "grad_norm": 0.675892689610788,
+      "learning_rate": 1.8278651785934453e-05,
+      "loss": 0.0505,
+      "step": 3033
+    },
+    {
+      "epoch": 10.835714285714285,
+      "grad_norm": 0.8742700170684282,
+      "learning_rate": 1.8267128012623007e-05,
+      "loss": 0.0562,
+      "step": 3034
+    },
+    {
+      "epoch": 10.839285714285714,
+      "grad_norm": 0.5254736131774003,
+      "learning_rate": 1.8255604818936e-05,
+      "loss": 0.0362,
+      "step": 3035
+    },
+    {
+      "epoch": 10.842857142857143,
+      "grad_norm": 1.0083575400155642,
+      "learning_rate": 1.8244082208727808e-05,
+      "loss": 0.0739,
+      "step": 3036
+    },
+    {
+      "epoch": 10.846428571428572,
+      "grad_norm": 0.858062637163032,
+      "learning_rate": 1.8232560185852598e-05,
+      "loss": 0.061,
+      "step": 3037
+    },
+    {
+      "epoch": 10.85,
+      "grad_norm": 0.45544917444784044,
+      "learning_rate": 1.8221038754164348e-05,
+      "loss": 0.034,
+      "step": 3038
+    },
+    {
+      "epoch": 10.853571428571428,
+      "grad_norm": 0.7952919200576439,
+      "learning_rate": 1.8209517917516828e-05,
+      "loss": 0.0754,
+      "step": 3039
+    },
+    {
+      "epoch": 10.857142857142858,
+      "grad_norm": 0.6108667754030724,
+      "learning_rate": 1.8197997679763625e-05,
+      "loss": 0.0598,
+      "step": 3040
+    },
+    {
+      "epoch": 10.860714285714286,
+      "grad_norm": 0.727485399739345,
+      "learning_rate": 1.8186478044758115e-05,
+      "loss": 0.0462,
+      "step": 3041
+    },
+    {
+      "epoch": 10.864285714285714,
+      "grad_norm": 1.128278513583264,
+      "learning_rate": 1.8174959016353475e-05,
+      "loss": 0.0755,
+      "step": 3042
+    },
+    {
+      "epoch": 10.867857142857144,
+      "grad_norm": 0.5794542207874849,
+      "learning_rate": 1.8163440598402677e-05,
+      "loss": 0.0691,
+      "step": 3043
+    },
+    {
+      "epoch": 10.871428571428572,
+      "grad_norm": 0.5971577516564467,
+      "learning_rate": 1.8151922794758494e-05,
+      "loss": 0.0452,
+      "step": 3044
+    },
+    {
+      "epoch": 10.875,
+      "grad_norm": 1.0974645160280485,
+      "learning_rate": 1.814040560927349e-05,
+      "loss": 0.0552,
+      "step": 3045
+    },
+    {
+      "epoch": 10.878571428571428,
+      "grad_norm": 0.8531094874827071,
+      "learning_rate": 1.812888904580002e-05,
+      "loss": 0.044,
+      "step": 3046
+    },
+    {
+      "epoch": 10.882142857142856,
+      "grad_norm": 0.9910109536082851,
+      "learning_rate": 1.8117373108190227e-05,
+      "loss": 0.08,
+      "step": 3047
+    },
+    {
+      "epoch": 10.885714285714286,
+      "grad_norm": 0.8411655973539393,
+      "learning_rate": 1.8105857800296063e-05,
+      "loss": 0.0532,
+      "step": 3048
+    },
+    {
+      "epoch": 10.889285714285714,
+      "grad_norm": 0.6470474801790685,
+      "learning_rate": 1.809434312596925e-05,
+      "loss": 0.08,
+      "step": 3049
+    },
+    {
+      "epoch": 10.892857142857142,
+      "grad_norm": 1.1078158726373102,
+      "learning_rate": 1.8082829089061308e-05,
+      "loss": 0.0805,
+      "step": 3050
+    },
+    {
+      "epoch": 10.896428571428572,
+      "grad_norm": 2.203115524452754,
+      "learning_rate": 1.8071315693423545e-05,
+      "loss": 0.0379,
+      "step": 3051
+    },
+    {
+      "epoch": 10.9,
+      "grad_norm": 0.9330760163857095,
+      "learning_rate": 1.805980294290704e-05,
+      "loss": 0.0386,
+      "step": 3052
+    },
+    {
+      "epoch": 10.903571428571428,
+      "grad_norm": 0.8848565472796136,
+      "learning_rate": 1.8048290841362676e-05,
+      "loss": 0.0642,
+      "step": 3053
+    },
+    {
+      "epoch": 10.907142857142857,
+      "grad_norm": 0.8594847858267846,
+      "learning_rate": 1.8036779392641112e-05,
+      "loss": 0.0727,
+      "step": 3054
+    },
+    {
+      "epoch": 10.910714285714286,
+      "grad_norm": 0.7775651600942342,
+      "learning_rate": 1.8025268600592774e-05,
+      "loss": 0.0771,
+      "step": 3055
+    },
+    {
+      "epoch": 10.914285714285715,
+      "grad_norm": 0.7013154214138669,
+      "learning_rate": 1.8013758469067895e-05,
+      "loss": 0.0511,
+      "step": 3056
+    },
+    {
+      "epoch": 10.917857142857143,
+      "grad_norm": 1.1251457174000796,
+      "learning_rate": 1.800224900191647e-05,
+      "loss": 0.0647,
+      "step": 3057
+    },
+    {
+      "epoch": 10.92142857142857,
+      "grad_norm": 0.5735324066018822,
+      "learning_rate": 1.7990740202988264e-05,
+      "loss": 0.0276,
+      "step": 3058
+    },
+    {
+      "epoch": 10.925,
+      "grad_norm": 1.2611848594110933,
+      "learning_rate": 1.797923207613284e-05,
+      "loss": 0.0937,
+      "step": 3059
+    },
+    {
+      "epoch": 10.928571428571429,
+      "grad_norm": 0.9211568260051466,
+      "learning_rate": 1.796772462519952e-05,
+      "loss": 0.053,
+      "step": 3060
+    },
+    {
+      "epoch": 10.932142857142857,
+      "grad_norm": 2.0776512014375106,
+      "learning_rate": 1.795621785403741e-05,
+      "loss": 0.1077,
+      "step": 3061
+    },
+    {
+      "epoch": 10.935714285714285,
+      "grad_norm": 0.7794617856947188,
+      "learning_rate": 1.794471176649538e-05,
+      "loss": 0.1225,
+      "step": 3062
+    },
+    {
+      "epoch": 10.939285714285715,
+      "grad_norm": 1.1598542404653516,
+      "learning_rate": 1.7933206366422075e-05,
+      "loss": 0.0536,
+      "step": 3063
+    },
+    {
+      "epoch": 10.942857142857143,
+      "grad_norm": 1.5662293589924212,
+      "learning_rate": 1.7921701657665916e-05,
+      "loss": 0.0998,
+      "step": 3064
+    },
+    {
+      "epoch": 10.946428571428571,
+      "grad_norm": 0.6536027936845302,
+      "learning_rate": 1.791019764407508e-05,
+      "loss": 0.0371,
+      "step": 3065
+    },
+    {
+      "epoch": 10.95,
+      "grad_norm": 0.6517593665664602,
+      "learning_rate": 1.7898694329497523e-05,
+      "loss": 0.0322,
+      "step": 3066
+    },
+    {
+      "epoch": 10.95357142857143,
+      "grad_norm": 0.9368907174840537,
+      "learning_rate": 1.788719171778096e-05,
+      "loss": 0.0467,
+      "step": 3067
+    },
+    {
+      "epoch": 10.957142857142857,
+      "grad_norm": 1.4093914625861421,
+      "learning_rate": 1.7875689812772876e-05,
+      "loss": 0.0661,
+      "step": 3068
+    },
+    {
+      "epoch": 10.960714285714285,
+      "grad_norm": 0.8557046056583082,
+      "learning_rate": 1.7864188618320517e-05,
+      "loss": 0.038,
+      "step": 3069
+    },
+    {
+      "epoch": 10.964285714285714,
+      "grad_norm": 1.0250565639785971,
+      "learning_rate": 1.7852688138270895e-05,
+      "loss": 0.1096,
+      "step": 3070
+    },
+    {
+      "epoch": 10.967857142857143,
+      "grad_norm": 0.7032164025419321,
+      "learning_rate": 1.784118837647077e-05,
+      "loss": 0.0734,
+      "step": 3071
+    },
+    {
+      "epoch": 10.971428571428572,
+      "grad_norm": 1.3603428353764289,
+      "learning_rate": 1.782968933676668e-05,
+      "loss": 0.0413,
+      "step": 3072
+    },
+    {
+      "epoch": 10.975,
+      "grad_norm": 0.6609780340696908,
+      "learning_rate": 1.7818191023004904e-05,
+      "loss": 0.0285,
+      "step": 3073
+    },
+    {
+      "epoch": 10.978571428571428,
+      "grad_norm": 1.0444466167403115,
+      "learning_rate": 1.7806693439031496e-05,
+      "loss": 0.0855,
+      "step": 3074
+    },
+    {
+      "epoch": 10.982142857142858,
+      "grad_norm": 0.7710682493740484,
+      "learning_rate": 1.7795196588692257e-05,
+      "loss": 0.0293,
+      "step": 3075
+    },
+    {
+      "epoch": 10.985714285714286,
+      "grad_norm": 1.3156795603340048,
+      "learning_rate": 1.7783700475832736e-05,
+      "loss": 0.1318,
+      "step": 3076
+    },
+    {
+      "epoch": 10.989285714285714,
+      "grad_norm": 0.725814708905628,
+      "learning_rate": 1.7772205104298246e-05,
+      "loss": 0.0738,
+      "step": 3077
+    },
+    {
+      "epoch": 10.992857142857144,
+      "grad_norm": 0.8809515205801798,
+      "learning_rate": 1.7760710477933846e-05,
+      "loss": 0.1057,
+      "step": 3078
+    },
+    {
+      "epoch": 10.996428571428572,
+      "grad_norm": 1.1592457043674238,
+      "learning_rate": 1.774921660058435e-05,
+      "loss": 0.0705,
+      "step": 3079
+    },
+    {
+      "epoch": 11.0,
+      "grad_norm": 0.7112564315798022,
+      "learning_rate": 1.7737723476094317e-05,
+      "loss": 0.0536,
+      "step": 3080
+    },
+    {
+      "epoch": 11.003571428571428,
+      "grad_norm": 1.094558850150742,
+      "learning_rate": 1.7726231108308058e-05,
+      "loss": 0.0769,
+      "step": 3081
+    },
+    {
+      "epoch": 11.007142857142858,
+      "grad_norm": 0.634207265567949,
+      "learning_rate": 1.771473950106963e-05,
+      "loss": 0.0533,
+      "step": 3082
+    },
+    {
+      "epoch": 11.010714285714286,
+      "grad_norm": 0.5237541723706625,
+      "learning_rate": 1.7703248658222833e-05,
+      "loss": 0.054,
+      "step": 3083
+    },
+    {
+      "epoch": 11.014285714285714,
+      "grad_norm": 0.4961435647794548,
+      "learning_rate": 1.769175858361121e-05,
+      "loss": 0.0245,
+      "step": 3084
+    },
+    {
+      "epoch": 11.017857142857142,
+      "grad_norm": 0.3953670117157097,
+      "learning_rate": 1.7680269281078054e-05,
+      "loss": 0.0557,
+      "step": 3085
+    },
+    {
+      "epoch": 11.021428571428572,
+      "grad_norm": 0.624856560771699,
+      "learning_rate": 1.76687807544664e-05,
+      "loss": 0.0384,
+      "step": 3086
+    },
+    {
+      "epoch": 11.025,
+      "grad_norm": 0.37214713076248346,
+      "learning_rate": 1.765729300761901e-05,
+      "loss": 0.0229,
+      "step": 3087
+    },
+    {
+      "epoch": 11.028571428571428,
+      "grad_norm": 0.5400445188303606,
+      "learning_rate": 1.76458060443784e-05,
+      "loss": 0.0388,
+      "step": 3088
+    },
+    {
+      "epoch": 11.032142857142857,
+      "grad_norm": 1.278550492837449,
+      "learning_rate": 1.7634319868586815e-05,
+      "loss": 0.0907,
+      "step": 3089
+    },
+    {
+      "epoch": 11.035714285714286,
+      "grad_norm": 0.7867755719709602,
+      "learning_rate": 1.7622834484086248e-05,
+      "loss": 0.0693,
+      "step": 3090
+    },
+    {
+      "epoch": 11.039285714285715,
+      "grad_norm": 1.0136925668742953,
+      "learning_rate": 1.761134989471841e-05,
+      "loss": 0.0505,
+      "step": 3091
+    },
+    {
+      "epoch": 11.042857142857143,
+      "grad_norm": 0.9503542082602852,
+      "learning_rate": 1.759986610432476e-05,
+      "loss": 0.0516,
+      "step": 3092
+    },
+    {
+      "epoch": 11.04642857142857,
+      "grad_norm": 0.6329916016975558,
+      "learning_rate": 1.7588383116746486e-05,
+      "loss": 0.085,
+      "step": 3093
+    },
+    {
+      "epoch": 11.05,
+      "grad_norm": 0.7385996678729405,
+      "learning_rate": 1.757690093582451e-05,
+      "loss": 0.0625,
+      "step": 3094
+    },
+    {
+      "epoch": 11.053571428571429,
+      "grad_norm": 1.1546441477500824,
+      "learning_rate": 1.756541956539947e-05,
+      "loss": 0.0569,
+      "step": 3095
+    },
+    {
+      "epoch": 11.057142857142857,
+      "grad_norm": 0.5189854449775055,
+      "learning_rate": 1.755393900931175e-05,
+      "loss": 0.0216,
+      "step": 3096
+    },
+    {
+      "epoch": 11.060714285714285,
+      "grad_norm": 0.5998036514833788,
+      "learning_rate": 1.7542459271401457e-05,
+      "loss": 0.0386,
+      "step": 3097
+    },
+    {
+      "epoch": 11.064285714285715,
+      "grad_norm": 0.8524506009990548,
+      "learning_rate": 1.753098035550842e-05,
+      "loss": 0.06,
+      "step": 3098
+    },
+    {
+      "epoch": 11.067857142857143,
+      "grad_norm": 0.7468857488518992,
+      "learning_rate": 1.7519502265472194e-05,
+      "loss": 0.0405,
+      "step": 3099
+    },
+    {
+      "epoch": 11.071428571428571,
+      "grad_norm": 0.5210754170804901,
+      "learning_rate": 1.750802500513206e-05,
+      "loss": 0.0303,
+      "step": 3100
+    },
+    {
+      "epoch": 11.075,
+      "grad_norm": 0.8019857638683237,
+      "learning_rate": 1.7496548578327027e-05,
+      "loss": 0.0708,
+      "step": 3101
+    },
+    {
+      "epoch": 11.07857142857143,
+      "grad_norm": 0.6800858296252895,
+      "learning_rate": 1.7485072988895817e-05,
+      "loss": 0.0426,
+      "step": 3102
+    },
+    {
+      "epoch": 11.082142857142857,
+      "grad_norm": 0.6965283919501518,
+      "learning_rate": 1.747359824067687e-05,
+      "loss": 0.0809,
+      "step": 3103
+    },
+    {
+      "epoch": 11.085714285714285,
+      "grad_norm": 0.7391753965713747,
+      "learning_rate": 1.7462124337508347e-05,
+      "loss": 0.0625,
+      "step": 3104
+    },
+    {
+      "epoch": 11.089285714285714,
+      "grad_norm": 0.7015440965324952,
+      "learning_rate": 1.7450651283228134e-05,
+      "loss": 0.0445,
+      "step": 3105
+    },
+    {
+      "epoch": 11.092857142857143,
+      "grad_norm": 0.7577634608809038,
+      "learning_rate": 1.743917908167383e-05,
+      "loss": 0.0559,
+      "step": 3106
+    },
+    {
+      "epoch": 11.096428571428572,
+      "grad_norm": 0.7421951032785536,
+      "learning_rate": 1.7427707736682737e-05,
+      "loss": 0.0509,
+      "step": 3107
+    },
+    {
+      "epoch": 11.1,
+      "grad_norm": 0.6761667247465336,
+      "learning_rate": 1.7416237252091878e-05,
+      "loss": 0.0417,
+      "step": 3108
+    },
+    {
+      "epoch": 11.103571428571428,
+      "grad_norm": 0.5277957775888356,
+      "learning_rate": 1.7404767631738e-05,
+      "loss": 0.0308,
+      "step": 3109
+    },
+    {
+      "epoch": 11.107142857142858,
+      "grad_norm": 0.7717505439503484,
+      "learning_rate": 1.739329887945754e-05,
+      "loss": 0.0556,
+      "step": 3110
+    },
+    {
+      "epoch": 11.110714285714286,
+      "grad_norm": 0.7165593408543237,
+      "learning_rate": 1.7381830999086663e-05,
+      "loss": 0.0582,
+      "step": 3111
+    },
+    {
+      "epoch": 11.114285714285714,
+      "grad_norm": 0.8486990304196762,
+      "learning_rate": 1.737036399446123e-05,
+      "loss": 0.0487,
+      "step": 3112
+    },
+    {
+      "epoch": 11.117857142857142,
+      "grad_norm": 0.8470734995682413,
+      "learning_rate": 1.7358897869416815e-05,
+      "loss": 0.0712,
+      "step": 3113
+    },
+    {
+      "epoch": 11.121428571428572,
+      "grad_norm": 1.0974303576195499,
+      "learning_rate": 1.7347432627788698e-05,
+      "loss": 0.0422,
+      "step": 3114
+    },
+    {
+      "epoch": 11.125,
+      "grad_norm": 1.005263272921582,
+      "learning_rate": 1.7335968273411855e-05,
+      "loss": 0.0372,
+      "step": 3115
+    },
+    {
+      "epoch": 11.128571428571428,
+      "grad_norm": 1.0530787310704273,
+      "learning_rate": 1.732450481012098e-05,
+      "loss": 0.0552,
+      "step": 3116
+    },
+    {
+      "epoch": 11.132142857142858,
+      "grad_norm": 0.6849679589132616,
+      "learning_rate": 1.731304224175046e-05,
+      "loss": 0.0421,
+      "step": 3117
+    },
+    {
+      "epoch": 11.135714285714286,
+      "grad_norm": 0.7931803182651854,
+      "learning_rate": 1.7301580572134377e-05,
+      "loss": 0.1019,
+      "step": 3118
+    },
+    {
+      "epoch": 11.139285714285714,
+      "grad_norm": 1.1831294286398035,
+      "learning_rate": 1.7290119805106532e-05,
+      "loss": 0.054,
+      "step": 3119
+    },
+    {
+      "epoch": 11.142857142857142,
+      "grad_norm": 0.8388185634456794,
+      "learning_rate": 1.7278659944500397e-05,
+      "loss": 0.043,
+      "step": 3120
+    },
+    {
+      "epoch": 11.146428571428572,
+      "grad_norm": 1.2451743225471132,
+      "learning_rate": 1.726720099414916e-05,
+      "loss": 0.0541,
+      "step": 3121
+    },
+    {
+      "epoch": 11.15,
+      "grad_norm": 0.6694610224543569,
+      "learning_rate": 1.72557429578857e-05,
+      "loss": 0.0566,
+      "step": 3122
+    },
+    {
+      "epoch": 11.153571428571428,
+      "grad_norm": 0.7767370028849051,
+      "learning_rate": 1.724428583954259e-05,
+      "loss": 0.0539,
+      "step": 3123
+    },
+    {
+      "epoch": 11.157142857142857,
+      "grad_norm": 0.8861641057706432,
+      "learning_rate": 1.7232829642952097e-05,
+      "loss": 0.0691,
+      "step": 3124
+    },
+    {
+      "epoch": 11.160714285714286,
+      "grad_norm": 0.8561729284952132,
+      "learning_rate": 1.7221374371946177e-05,
+      "loss": 0.0336,
+      "step": 3125
+    },
+    {
+      "epoch": 11.164285714285715,
+      "grad_norm": 0.7396649035979828,
+      "learning_rate": 1.7209920030356476e-05,
+      "loss": 0.0429,
+      "step": 3126
+    },
+    {
+      "epoch": 11.167857142857143,
+      "grad_norm": 0.7655577735769139,
+      "learning_rate": 1.7198466622014337e-05,
+      "loss": 0.0394,
+      "step": 3127
+    },
+    {
+      "epoch": 11.17142857142857,
+      "grad_norm": 0.5865297913127783,
+      "learning_rate": 1.7187014150750775e-05,
+      "loss": 0.0689,
+      "step": 3128
+    },
+    {
+      "epoch": 11.175,
+      "grad_norm": 0.5760564107995063,
+      "learning_rate": 1.7175562620396507e-05,
+      "loss": 0.0446,
+      "step": 3129
+    },
+    {
+      "epoch": 11.178571428571429,
+      "grad_norm": 0.7285740765547117,
+      "learning_rate": 1.716411203478193e-05,
+      "loss": 0.0417,
+      "step": 3130
+    },
+    {
+      "epoch": 11.182142857142857,
+      "grad_norm": 1.6628703669184295,
+      "learning_rate": 1.715266239773713e-05,
+      "loss": 0.0728,
+      "step": 3131
+    },
+    {
+      "epoch": 11.185714285714285,
+      "grad_norm": 0.6080495262863216,
+      "learning_rate": 1.7141213713091858e-05,
+      "loss": 0.0601,
+      "step": 3132
+    },
+    {
+      "epoch": 11.189285714285715,
+      "grad_norm": 0.5246743673977147,
+      "learning_rate": 1.7129765984675565e-05,
+      "loss": 0.0343,
+      "step": 3133
+    },
+    {
+      "epoch": 11.192857142857143,
+      "grad_norm": 1.1844686142324259,
+      "learning_rate": 1.7118319216317375e-05,
+      "loss": 0.0688,
+      "step": 3134
+    },
+    {
+      "epoch": 11.196428571428571,
+      "grad_norm": 0.6113601080538386,
+      "learning_rate": 1.7106873411846094e-05,
+      "loss": 0.0496,
+      "step": 3135
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 0.7930866980860216,
+      "learning_rate": 1.70954285750902e-05,
+      "loss": 0.0203,
+      "step": 3136
+    },
+    {
+      "epoch": 11.20357142857143,
+      "grad_norm": 0.7275284444350958,
+      "learning_rate": 1.7083984709877853e-05,
+      "loss": 0.0459,
+      "step": 3137
+    },
+    {
+      "epoch": 11.207142857142857,
+      "grad_norm": 0.7839606132809325,
+      "learning_rate": 1.7072541820036884e-05,
+      "loss": 0.0621,
+      "step": 3138
+    },
+    {
+      "epoch": 11.210714285714285,
+      "grad_norm": 0.5059417543385538,
+      "learning_rate": 1.7061099909394798e-05,
+      "loss": 0.0498,
+      "step": 3139
+    },
+    {
+      "epoch": 11.214285714285714,
+      "grad_norm": 0.7781488775415953,
+      "learning_rate": 1.7049658981778778e-05,
+      "loss": 0.0636,
+      "step": 3140
+    },
+    {
+      "epoch": 11.217857142857143,
+      "grad_norm": 0.4926881217970539,
+      "learning_rate": 1.7038219041015674e-05,
+      "loss": 0.0552,
+      "step": 3141
+    },
+    {
+      "epoch": 11.221428571428572,
+      "grad_norm": 0.5549642002476148,
+      "learning_rate": 1.7026780090932005e-05,
+      "loss": 0.0298,
+      "step": 3142
+    },
+    {
+      "epoch": 11.225,
+      "grad_norm": 1.0451394681782455,
+      "learning_rate": 1.701534213535396e-05,
+      "loss": 0.0795,
+      "step": 3143
+    },
+    {
+      "epoch": 11.228571428571428,
+      "grad_norm": 0.7632934577000375,
+      "learning_rate": 1.700390517810739e-05,
+      "loss": 0.0586,
+      "step": 3144
+    },
+    {
+      "epoch": 11.232142857142858,
+      "grad_norm": 0.6721652367093097,
+      "learning_rate": 1.699246922301782e-05,
+      "loss": 0.0353,
+      "step": 3145
+    },
+    {
+      "epoch": 11.235714285714286,
+      "grad_norm": 0.7308936134570315,
+      "learning_rate": 1.6981034273910443e-05,
+      "loss": 0.0779,
+      "step": 3146
+    },
+    {
+      "epoch": 11.239285714285714,
+      "grad_norm": 0.6887921619429379,
+      "learning_rate": 1.69696003346101e-05,
+      "loss": 0.0282,
+      "step": 3147
+    },
+    {
+      "epoch": 11.242857142857142,
+      "grad_norm": 0.39980160590760644,
+      "learning_rate": 1.695816740894131e-05,
+      "loss": 0.0248,
+      "step": 3148
+    },
+    {
+      "epoch": 11.246428571428572,
+      "grad_norm": 0.38531520644997136,
+      "learning_rate": 1.6946735500728244e-05,
+      "loss": 0.02,
+      "step": 3149
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 0.4635060644667977,
+      "learning_rate": 1.693530461379474e-05,
+      "loss": 0.0347,
+      "step": 3150
+    },
+    {
+      "epoch": 11.253571428571428,
+      "grad_norm": 0.7827888977537165,
+      "learning_rate": 1.692387475196428e-05,
+      "loss": 0.0369,
+      "step": 3151
+    },
+    {
+      "epoch": 11.257142857142856,
+      "grad_norm": 0.5042124340013584,
+      "learning_rate": 1.6912445919060015e-05,
+      "loss": 0.0375,
+      "step": 3152
+    },
+    {
+      "epoch": 11.260714285714286,
+      "grad_norm": 0.930577900654215,
+      "learning_rate": 1.6901018118904756e-05,
+      "loss": 0.0546,
+      "step": 3153
+    },
+    {
+      "epoch": 11.264285714285714,
+      "grad_norm": 0.5871250263167571,
+      "learning_rate": 1.6889591355320957e-05,
+      "loss": 0.0152,
+      "step": 3154
+    },
+    {
+      "epoch": 11.267857142857142,
+      "grad_norm": 0.7078591889765632,
+      "learning_rate": 1.6878165632130736e-05,
+      "loss": 0.0448,
+      "step": 3155
+    },
+    {
+      "epoch": 11.271428571428572,
+      "grad_norm": 0.5310354510233967,
+      "learning_rate": 1.6866740953155847e-05,
+      "loss": 0.0392,
+      "step": 3156
+    },
+    {
+      "epoch": 11.275,
+      "grad_norm": 0.661358842438068,
+      "learning_rate": 1.6855317322217714e-05,
+      "loss": 0.0525,
+      "step": 3157
+    },
+    {
+      "epoch": 11.278571428571428,
+      "grad_norm": 0.5607399200244877,
+      "learning_rate": 1.68438947431374e-05,
+      "loss": 0.0118,
+      "step": 3158
+    },
+    {
+      "epoch": 11.282142857142857,
+      "grad_norm": 0.9615588477507081,
+      "learning_rate": 1.683247321973561e-05,
+      "loss": 0.0541,
+      "step": 3159
+    },
+    {
+      "epoch": 11.285714285714286,
+      "grad_norm": 0.8350026485554047,
+      "learning_rate": 1.682105275583272e-05,
+      "loss": 0.0395,
+      "step": 3160
+    },
+    {
+      "epoch": 11.289285714285715,
+      "grad_norm": 0.7400002823991955,
+      "learning_rate": 1.6809633355248718e-05,
+      "loss": 0.0241,
+      "step": 3161
+    },
+    {
+      "epoch": 11.292857142857143,
+      "grad_norm": 1.06876733503379,
+      "learning_rate": 1.6798215021803264e-05,
+      "loss": 0.0655,
+      "step": 3162
+    },
+    {
+      "epoch": 11.29642857142857,
+      "grad_norm": 0.6608562332932203,
+      "learning_rate": 1.678679775931565e-05,
+      "loss": 0.0406,
+      "step": 3163
+    },
+    {
+      "epoch": 11.3,
+      "grad_norm": 1.4548484180897607,
+      "learning_rate": 1.6775381571604806e-05,
+      "loss": 0.097,
+      "step": 3164
+    },
+    {
+      "epoch": 11.303571428571429,
+      "grad_norm": 0.9230808206032975,
+      "learning_rate": 1.6763966462489308e-05,
+      "loss": 0.0399,
+      "step": 3165
+    },
+    {
+      "epoch": 11.307142857142857,
+      "grad_norm": 0.8644539700711701,
+      "learning_rate": 1.6752552435787367e-05,
+      "loss": 0.0468,
+      "step": 3166
+    },
+    {
+      "epoch": 11.310714285714285,
+      "grad_norm": 1.1547598010693572,
+      "learning_rate": 1.6741139495316846e-05,
+      "loss": 0.0382,
+      "step": 3167
+    },
+    {
+      "epoch": 11.314285714285715,
+      "grad_norm": 1.4420482191563937,
+      "learning_rate": 1.6729727644895223e-05,
+      "loss": 0.0847,
+      "step": 3168
+    },
+    {
+      "epoch": 11.317857142857143,
+      "grad_norm": 0.6623129831824299,
+      "learning_rate": 1.6718316888339626e-05,
+      "loss": 0.0203,
+      "step": 3169
+    },
+    {
+      "epoch": 11.321428571428571,
+      "grad_norm": 0.7511141756986175,
+      "learning_rate": 1.6706907229466813e-05,
+      "loss": 0.055,
+      "step": 3170
+    },
+    {
+      "epoch": 11.325,
+      "grad_norm": 0.7180055084984591,
+      "learning_rate": 1.6695498672093174e-05,
+      "loss": 0.0667,
+      "step": 3171
+    },
+    {
+      "epoch": 11.32857142857143,
+      "grad_norm": 1.2625042749809956,
+      "learning_rate": 1.6684091220034732e-05,
+      "loss": 0.0883,
+      "step": 3172
+    },
+    {
+      "epoch": 11.332142857142857,
+      "grad_norm": 0.8794811895522788,
+      "learning_rate": 1.667268487710714e-05,
+      "loss": 0.067,
+      "step": 3173
+    },
+    {
+      "epoch": 11.335714285714285,
+      "grad_norm": 1.212647031772798,
+      "learning_rate": 1.6661279647125686e-05,
+      "loss": 0.0839,
+      "step": 3174
+    },
+    {
+      "epoch": 11.339285714285714,
+      "grad_norm": 1.1674728411254218,
+      "learning_rate": 1.6649875533905276e-05,
+      "loss": 0.0697,
+      "step": 3175
+    },
+    {
+      "epoch": 11.342857142857143,
+      "grad_norm": 0.7413954420352095,
+      "learning_rate": 1.6638472541260437e-05,
+      "loss": 0.0495,
+      "step": 3176
+    },
+    {
+      "epoch": 11.346428571428572,
+      "grad_norm": 0.9700841387283672,
+      "learning_rate": 1.6627070673005342e-05,
+      "loss": 0.0629,
+      "step": 3177
+    },
+    {
+      "epoch": 11.35,
+      "grad_norm": 0.606505778973568,
+      "learning_rate": 1.661566993295377e-05,
+      "loss": 0.0321,
+      "step": 3178
+    },
+    {
+      "epoch": 11.353571428571428,
+      "grad_norm": 0.6979926122194983,
+      "learning_rate": 1.660427032491913e-05,
+      "loss": 0.0664,
+      "step": 3179
+    },
+    {
+      "epoch": 11.357142857142858,
+      "grad_norm": 0.7826768491242607,
+      "learning_rate": 1.659287185271445e-05,
+      "loss": 0.0427,
+      "step": 3180
+    },
+    {
+      "epoch": 11.360714285714286,
+      "grad_norm": 0.4880888727420604,
+      "learning_rate": 1.658147452015238e-05,
+      "loss": 0.0178,
+      "step": 3181
+    },
+    {
+      "epoch": 11.364285714285714,
+      "grad_norm": 0.755515237615738,
+      "learning_rate": 1.6570078331045198e-05,
+      "loss": 0.0363,
+      "step": 3182
+    },
+    {
+      "epoch": 11.367857142857144,
+      "grad_norm": 1.3201753301546204,
+      "learning_rate": 1.6558683289204775e-05,
+      "loss": 0.0821,
+      "step": 3183
+    },
+    {
+      "epoch": 11.371428571428572,
+      "grad_norm": 1.0274737583713505,
+      "learning_rate": 1.6547289398442618e-05,
+      "loss": 0.047,
+      "step": 3184
+    },
+    {
+      "epoch": 11.375,
+      "grad_norm": 0.6518275871993316,
+      "learning_rate": 1.6535896662569847e-05,
+      "loss": 0.0314,
+      "step": 3185
+    },
+    {
+      "epoch": 11.378571428571428,
+      "grad_norm": 0.6290818534852334,
+      "learning_rate": 1.6524505085397185e-05,
+      "loss": 0.0214,
+      "step": 3186
+    },
+    {
+      "epoch": 11.382142857142856,
+      "grad_norm": 0.8279652996811565,
+      "learning_rate": 1.651311467073499e-05,
+      "loss": 0.0926,
+      "step": 3187
+    },
+    {
+      "epoch": 11.385714285714286,
+      "grad_norm": 0.6486187835506956,
+      "learning_rate": 1.6501725422393196e-05,
+      "loss": 0.0673,
+      "step": 3188
+    },
+    {
+      "epoch": 11.389285714285714,
+      "grad_norm": 1.3010431696552465,
+      "learning_rate": 1.649033734418138e-05,
+      "loss": 0.0895,
+      "step": 3189
+    },
+    {
+      "epoch": 11.392857142857142,
+      "grad_norm": 0.8639475928415028,
+      "learning_rate": 1.6478950439908715e-05,
+      "loss": 0.0697,
+      "step": 3190
+    },
+    {
+      "epoch": 11.396428571428572,
+      "grad_norm": 0.7962416479311397,
+      "learning_rate": 1.6467564713383975e-05,
+      "loss": 0.0465,
+      "step": 3191
+    },
+    {
+      "epoch": 11.4,
+      "grad_norm": 0.5845980549540531,
+      "learning_rate": 1.6456180168415546e-05,
+      "loss": 0.0444,
+      "step": 3192
+    },
+    {
+      "epoch": 11.403571428571428,
+      "grad_norm": 0.8609526509116989,
+      "learning_rate": 1.6444796808811424e-05,
+      "loss": 0.0471,
+      "step": 3193
+    },
+    {
+      "epoch": 11.407142857142857,
+      "grad_norm": 0.6859623907739063,
+      "learning_rate": 1.64334146383792e-05,
+      "loss": 0.0623,
+      "step": 3194
+    },
+    {
+      "epoch": 11.410714285714286,
+      "grad_norm": 1.1073058418047952,
+      "learning_rate": 1.642203366092607e-05,
+      "loss": 0.1074,
+      "step": 3195
+    },
+    {
+      "epoch": 11.414285714285715,
+      "grad_norm": 0.8200907392562626,
+      "learning_rate": 1.6410653880258834e-05,
+      "loss": 0.0737,
+      "step": 3196
+    },
+    {
+      "epoch": 11.417857142857143,
+      "grad_norm": 1.0318652405279471,
+      "learning_rate": 1.639927530018389e-05,
+      "loss": 0.0543,
+      "step": 3197
+    },
+    {
+      "epoch": 11.42142857142857,
+      "grad_norm": 0.7732537655606138,
+      "learning_rate": 1.6387897924507227e-05,
+      "loss": 0.0403,
+      "step": 3198
+    },
+    {
+      "epoch": 11.425,
+      "grad_norm": 0.45872975862011345,
+      "learning_rate": 1.637652175703445e-05,
+      "loss": 0.044,
+      "step": 3199
+    },
+    {
+      "epoch": 11.428571428571429,
+      "grad_norm": 1.0971023484666111,
+      "learning_rate": 1.6365146801570733e-05,
+      "loss": 0.0561,
+      "step": 3200
+    },
+    {
+      "epoch": 11.432142857142857,
+      "grad_norm": 0.6281326034772039,
+      "learning_rate": 1.635377306192087e-05,
+      "loss": 0.041,
+      "step": 3201
+    },
+    {
+      "epoch": 11.435714285714285,
+      "grad_norm": 0.7850823025077914,
+      "learning_rate": 1.634240054188923e-05,
+      "loss": 0.0772,
+      "step": 3202
+    },
+    {
+      "epoch": 11.439285714285715,
+      "grad_norm": 0.5914043087987272,
+      "learning_rate": 1.6331029245279785e-05,
+      "loss": 0.0524,
+      "step": 3203
+    },
+    {
+      "epoch": 11.442857142857143,
+      "grad_norm": 0.8774905336963342,
+      "learning_rate": 1.6319659175896095e-05,
+      "loss": 0.0594,
+      "step": 3204
+    },
+    {
+      "epoch": 11.446428571428571,
+      "grad_norm": 0.7004222248647135,
+      "learning_rate": 1.630829033754131e-05,
+      "loss": 0.0273,
+      "step": 3205
+    },
+    {
+      "epoch": 11.45,
+      "grad_norm": 0.6762427550008836,
+      "learning_rate": 1.6296922734018162e-05,
+      "loss": 0.0229,
+      "step": 3206
+    },
+    {
+      "epoch": 11.45357142857143,
+      "grad_norm": 0.3002810030972318,
+      "learning_rate": 1.6285556369128978e-05,
+      "loss": 0.0165,
+      "step": 3207
+    },
+    {
+      "epoch": 11.457142857142857,
+      "grad_norm": 0.6147432492904666,
+      "learning_rate": 1.6274191246675672e-05,
+      "loss": 0.0744,
+      "step": 3208
+    },
+    {
+      "epoch": 11.460714285714285,
+      "grad_norm": 0.9667707487872569,
+      "learning_rate": 1.6262827370459733e-05,
+      "loss": 0.051,
+      "step": 3209
+    },
+    {
+      "epoch": 11.464285714285714,
+      "grad_norm": 0.7937679941009674,
+      "learning_rate": 1.625146474428224e-05,
+      "loss": 0.0818,
+      "step": 3210
+    },
+    {
+      "epoch": 11.467857142857143,
+      "grad_norm": 1.185752360848995,
+      "learning_rate": 1.624010337194385e-05,
+      "loss": 0.0633,
+      "step": 3211
+    },
+    {
+      "epoch": 11.471428571428572,
+      "grad_norm": 1.4200189465775443,
+      "learning_rate": 1.622874325724481e-05,
+      "loss": 0.0664,
+      "step": 3212
+    },
+    {
+      "epoch": 11.475,
+      "grad_norm": 1.424477054695648,
+      "learning_rate": 1.6217384403984932e-05,
+      "loss": 0.1175,
+      "step": 3213
+    },
+    {
+      "epoch": 11.478571428571428,
+      "grad_norm": 0.9158940344479644,
+      "learning_rate": 1.620602681596361e-05,
+      "loss": 0.0981,
+      "step": 3214
+    },
+    {
+      "epoch": 11.482142857142858,
+      "grad_norm": 0.7812208100170569,
+      "learning_rate": 1.6194670496979828e-05,
+      "loss": 0.0539,
+      "step": 3215
+    },
+    {
+      "epoch": 11.485714285714286,
+      "grad_norm": 1.1242008678699587,
+      "learning_rate": 1.618331545083213e-05,
+      "loss": 0.0828,
+      "step": 3216
+    },
+    {
+      "epoch": 11.489285714285714,
+      "grad_norm": 0.5579098144094847,
+      "learning_rate": 1.617196168131864e-05,
+      "loss": 0.0483,
+      "step": 3217
+    },
+    {
+      "epoch": 11.492857142857144,
+      "grad_norm": 0.7171683543803092,
+      "learning_rate": 1.6160609192237052e-05,
+      "loss": 0.045,
+      "step": 3218
+    },
+    {
+      "epoch": 11.496428571428572,
+      "grad_norm": 0.7794972086949973,
+      "learning_rate": 1.6149257987384643e-05,
+      "loss": 0.08,
+      "step": 3219
+    },
+    {
+      "epoch": 11.5,
+      "grad_norm": 0.5995749357286215,
+      "learning_rate": 1.6137908070558243e-05,
+      "loss": 0.0294,
+      "step": 3220
+    },
+    {
+      "epoch": 11.503571428571428,
+      "grad_norm": 0.9169863819638305,
+      "learning_rate": 1.6126559445554265e-05,
+      "loss": 0.0796,
+      "step": 3221
+    },
+    {
+      "epoch": 11.507142857142856,
+      "grad_norm": 0.9509456684860064,
+      "learning_rate": 1.6115212116168682e-05,
+      "loss": 0.0586,
+      "step": 3222
+    },
+    {
+      "epoch": 11.510714285714286,
+      "grad_norm": 0.8868184944697867,
+      "learning_rate": 1.610386608619704e-05,
+      "loss": 0.0653,
+      "step": 3223
+    },
+    {
+      "epoch": 11.514285714285714,
+      "grad_norm": 0.4358465933955549,
+      "learning_rate": 1.609252135943445e-05,
+      "loss": 0.0331,
+      "step": 3224
+    },
+    {
+      "epoch": 11.517857142857142,
+      "grad_norm": 1.4620766801300384,
+      "learning_rate": 1.6081177939675575e-05,
+      "loss": 0.0935,
+      "step": 3225
+    },
+    {
+      "epoch": 11.521428571428572,
+      "grad_norm": 1.0876518353738391,
+      "learning_rate": 1.6069835830714653e-05,
+      "loss": 0.0863,
+      "step": 3226
+    },
+    {
+      "epoch": 11.525,
+      "grad_norm": 0.5211127003724888,
+      "learning_rate": 1.605849503634548e-05,
+      "loss": 0.0377,
+      "step": 3227
+    },
+    {
+      "epoch": 11.528571428571428,
+      "grad_norm": 0.9238763595871686,
+      "learning_rate": 1.6047155560361417e-05,
+      "loss": 0.0663,
+      "step": 3228
+    },
+    {
+      "epoch": 11.532142857142857,
+      "grad_norm": 0.6103945466045333,
+      "learning_rate": 1.6035817406555377e-05,
+      "loss": 0.022,
+      "step": 3229
+    },
+    {
+      "epoch": 11.535714285714286,
+      "grad_norm": 1.4590425556332867,
+      "learning_rate": 1.6024480578719836e-05,
+      "loss": 0.0782,
+      "step": 3230
+    },
+    {
+      "epoch": 11.539285714285715,
+      "grad_norm": 0.5366054588822258,
+      "learning_rate": 1.6013145080646817e-05,
+      "loss": 0.0454,
+      "step": 3231
+    },
+    {
+      "epoch": 11.542857142857143,
+      "grad_norm": 0.7825586882472252,
+      "learning_rate": 1.6001810916127912e-05,
+      "loss": 0.0785,
+      "step": 3232
+    },
+    {
+      "epoch": 11.54642857142857,
+      "grad_norm": 1.4142754494677983,
+      "learning_rate": 1.5990478088954257e-05,
+      "loss": 0.0858,
+      "step": 3233
+    },
+    {
+      "epoch": 11.55,
+      "grad_norm": 0.9327709391440182,
+      "learning_rate": 1.5979146602916548e-05,
+      "loss": 0.1009,
+      "step": 3234
+    },
+    {
+      "epoch": 11.553571428571429,
+      "grad_norm": 1.784300362564964,
+      "learning_rate": 1.5967816461805023e-05,
+      "loss": 0.0965,
+      "step": 3235
+    },
+    {
+      "epoch": 11.557142857142857,
+      "grad_norm": 0.8457196210095927,
+      "learning_rate": 1.595648766940948e-05,
+      "loss": 0.0432,
+      "step": 3236
+    },
+    {
+      "epoch": 11.560714285714285,
+      "grad_norm": 0.6457083635094626,
+      "learning_rate": 1.5945160229519256e-05,
+      "loss": 0.0677,
+      "step": 3237
+    },
+    {
+      "epoch": 11.564285714285715,
+      "grad_norm": 0.8565507387099256,
+      "learning_rate": 1.593383414592324e-05,
+      "loss": 0.0486,
+      "step": 3238
+    },
+    {
+      "epoch": 11.567857142857143,
+      "grad_norm": 0.8101599166027517,
+      "learning_rate": 1.5922509422409876e-05,
+      "loss": 0.0839,
+      "step": 3239
+    },
+    {
+      "epoch": 11.571428571428571,
+      "grad_norm": 0.6224033434213084,
+      "learning_rate": 1.591118606276714e-05,
+      "loss": 0.0553,
+      "step": 3240
+    },
+    {
+      "epoch": 11.575,
+      "grad_norm": 0.8920207316567703,
+      "learning_rate": 1.5899864070782557e-05,
+      "loss": 0.104,
+      "step": 3241
+    },
+    {
+      "epoch": 11.57857142857143,
+      "grad_norm": 0.719298931709627,
+      "learning_rate": 1.5888543450243195e-05,
+      "loss": 0.0433,
+      "step": 3242
+    },
+    {
+      "epoch": 11.582142857142857,
+      "grad_norm": 1.109189531675062,
+      "learning_rate": 1.5877224204935664e-05,
+      "loss": 0.0974,
+      "step": 3243
+    },
+    {
+      "epoch": 11.585714285714285,
+      "grad_norm": 0.4955778142126495,
+      "learning_rate": 1.5865906338646113e-05,
+      "loss": 0.0202,
+      "step": 3244
+    },
+    {
+      "epoch": 11.589285714285714,
+      "grad_norm": 0.6078327358020424,
+      "learning_rate": 1.5854589855160227e-05,
+      "loss": 0.0487,
+      "step": 3245
+    },
+    {
+      "epoch": 11.592857142857143,
+      "grad_norm": 1.029216162085094,
+      "learning_rate": 1.584327475826323e-05,
+      "loss": 0.0433,
+      "step": 3246
+    },
+    {
+      "epoch": 11.596428571428572,
+      "grad_norm": 0.8235798414781398,
+      "learning_rate": 1.5831961051739893e-05,
+      "loss": 0.0565,
+      "step": 3247
+    },
+    {
+      "epoch": 11.6,
+      "grad_norm": 0.7061462687551673,
+      "learning_rate": 1.58206487393745e-05,
+      "loss": 0.0651,
+      "step": 3248
+    },
+    {
+      "epoch": 11.603571428571428,
+      "grad_norm": 1.007836890826492,
+      "learning_rate": 1.5809337824950892e-05,
+      "loss": 0.066,
+      "step": 3249
+    },
+    {
+      "epoch": 11.607142857142858,
+      "grad_norm": 0.7330291799597911,
+      "learning_rate": 1.579802831225242e-05,
+      "loss": 0.0458,
+      "step": 3250
+    },
+    {
+      "epoch": 11.610714285714286,
+      "grad_norm": 1.3154070579395494,
+      "learning_rate": 1.578672020506198e-05,
+      "loss": 0.0587,
+      "step": 3251
+    },
+    {
+      "epoch": 11.614285714285714,
+      "grad_norm": 1.0183117483415927,
+      "learning_rate": 1.5775413507162004e-05,
+      "loss": 0.0504,
+      "step": 3252
+    },
+    {
+      "epoch": 11.617857142857144,
+      "grad_norm": 0.45352595845637034,
+      "learning_rate": 1.5764108222334435e-05,
+      "loss": 0.0273,
+      "step": 3253
+    },
+    {
+      "epoch": 11.621428571428572,
+      "grad_norm": 0.6758252005362834,
+      "learning_rate": 1.5752804354360754e-05,
+      "loss": 0.0517,
+      "step": 3254
+    },
+    {
+      "epoch": 11.625,
+      "grad_norm": 0.8912476242384538,
+      "learning_rate": 1.5741501907021966e-05,
+      "loss": 0.0812,
+      "step": 3255
+    },
+    {
+      "epoch": 11.628571428571428,
+      "grad_norm": 1.185801217608816,
+      "learning_rate": 1.5730200884098605e-05,
+      "loss": 0.0536,
+      "step": 3256
+    },
+    {
+      "epoch": 11.632142857142856,
+      "grad_norm": 0.5353974930989358,
+      "learning_rate": 1.5718901289370723e-05,
+      "loss": 0.0293,
+      "step": 3257
+    },
+    {
+      "epoch": 11.635714285714286,
+      "grad_norm": 0.5857308348138736,
+      "learning_rate": 1.5707603126617893e-05,
+      "loss": 0.0655,
+      "step": 3258
+    },
+    {
+      "epoch": 11.639285714285714,
+      "grad_norm": 0.3804283919166935,
+      "learning_rate": 1.5696306399619215e-05,
+      "loss": 0.0173,
+      "step": 3259
+    },
+    {
+      "epoch": 11.642857142857142,
+      "grad_norm": 0.9449099345963733,
+      "learning_rate": 1.5685011112153305e-05,
+      "loss": 0.0841,
+      "step": 3260
+    },
+    {
+      "epoch": 11.646428571428572,
+      "grad_norm": 1.1803927046627283,
+      "learning_rate": 1.56737172679983e-05,
+      "loss": 0.0576,
+      "step": 3261
+    },
+    {
+      "epoch": 11.65,
+      "grad_norm": 1.0159026090916055,
+      "learning_rate": 1.566242487093185e-05,
+      "loss": 0.0755,
+      "step": 3262
+    },
+    {
+      "epoch": 11.653571428571428,
+      "grad_norm": 0.7425878791435593,
+      "learning_rate": 1.565113392473112e-05,
+      "loss": 0.0583,
+      "step": 3263
+    },
+    {
+      "epoch": 11.657142857142857,
+      "grad_norm": 1.141192943175944,
+      "learning_rate": 1.56398444331728e-05,
+      "loss": 0.065,
+      "step": 3264
+    },
+    {
+      "epoch": 11.660714285714286,
+      "grad_norm": 0.35095644706187434,
+      "learning_rate": 1.5628556400033083e-05,
+      "loss": 0.0134,
+      "step": 3265
+    },
+    {
+      "epoch": 11.664285714285715,
+      "grad_norm": 0.6248367106391123,
+      "learning_rate": 1.561726982908768e-05,
+      "loss": 0.0344,
+      "step": 3266
+    },
+    {
+      "epoch": 11.667857142857143,
+      "grad_norm": 0.9076875101040113,
+      "learning_rate": 1.5605984724111806e-05,
+      "loss": 0.0489,
+      "step": 3267
+    },
+    {
+      "epoch": 11.67142857142857,
+      "grad_norm": 0.7320350907586559,
+      "learning_rate": 1.559470108888019e-05,
+      "loss": 0.0596,
+      "step": 3268
+    },
+    {
+      "epoch": 11.675,
+      "grad_norm": 0.7862834041310277,
+      "learning_rate": 1.5583418927167082e-05,
+      "loss": 0.0662,
+      "step": 3269
+    },
+    {
+      "epoch": 11.678571428571429,
+      "grad_norm": 0.8864278820416818,
+      "learning_rate": 1.5572138242746212e-05,
+      "loss": 0.0372,
+      "step": 3270
+    },
+    {
+      "epoch": 11.682142857142857,
+      "grad_norm": 0.750080016907396,
+      "learning_rate": 1.556085903939084e-05,
+      "loss": 0.0723,
+      "step": 3271
+    },
+    {
+      "epoch": 11.685714285714285,
+      "grad_norm": 1.2563871666744106,
+      "learning_rate": 1.5549581320873715e-05,
+      "loss": 0.0716,
+      "step": 3272
+    },
+    {
+      "epoch": 11.689285714285715,
+      "grad_norm": 1.03324683444874,
+      "learning_rate": 1.5538305090967104e-05,
+      "loss": 0.0519,
+      "step": 3273
+    },
+    {
+      "epoch": 11.692857142857143,
+      "grad_norm": 1.5220248029383316,
+      "learning_rate": 1.5527030353442757e-05,
+      "loss": 0.0643,
+      "step": 3274
+    },
+    {
+      "epoch": 11.696428571428571,
+      "grad_norm": 0.3505740447603997,
+      "learning_rate": 1.5515757112071943e-05,
+      "loss": 0.0241,
+      "step": 3275
+    },
+    {
+      "epoch": 11.7,
+      "grad_norm": 1.193215635010233,
+      "learning_rate": 1.5504485370625418e-05,
+      "loss": 0.0513,
+      "step": 3276
+    },
+    {
+      "epoch": 11.70357142857143,
+      "grad_norm": 0.7852268171899398,
+      "learning_rate": 1.5493215132873446e-05,
+      "loss": 0.0416,
+      "step": 3277
+    },
+    {
+      "epoch": 11.707142857142857,
+      "grad_norm": 0.5126228439539189,
+      "learning_rate": 1.5481946402585782e-05,
+      "loss": 0.0233,
+      "step": 3278
+    },
+    {
+      "epoch": 11.710714285714285,
+      "grad_norm": 0.8080961153405309,
+      "learning_rate": 1.5470679183531677e-05,
+      "loss": 0.04,
+      "step": 3279
+    },
+    {
+      "epoch": 11.714285714285714,
+      "grad_norm": 1.2777888635935435,
+      "learning_rate": 1.5459413479479878e-05,
+      "loss": 0.0359,
+      "step": 3280
+    },
+    {
+      "epoch": 11.717857142857143,
+      "grad_norm": 1.6420368930578073,
+      "learning_rate": 1.5448149294198628e-05,
+      "loss": 0.08,
+      "step": 3281
+    },
+    {
+      "epoch": 11.721428571428572,
+      "grad_norm": 0.5337913447205482,
+      "learning_rate": 1.5436886631455652e-05,
+      "loss": 0.0456,
+      "step": 3282
+    },
+    {
+      "epoch": 11.725,
+      "grad_norm": 0.7102063597159102,
+      "learning_rate": 1.5425625495018182e-05,
+      "loss": 0.0666,
+      "step": 3283
+    },
+    {
+      "epoch": 11.728571428571428,
+      "grad_norm": 1.1789551562318916,
+      "learning_rate": 1.541436588865292e-05,
+      "loss": 0.0408,
+      "step": 3284
+    },
+    {
+      "epoch": 11.732142857142858,
+      "grad_norm": 0.9900746885699613,
+      "learning_rate": 1.5403107816126078e-05,
+      "loss": 0.0623,
+      "step": 3285
+    },
+    {
+      "epoch": 11.735714285714286,
+      "grad_norm": 0.938029788063016,
+      "learning_rate": 1.5391851281203338e-05,
+      "loss": 0.0301,
+      "step": 3286
+    },
+    {
+      "epoch": 11.739285714285714,
+      "grad_norm": 1.3986625090782316,
+      "learning_rate": 1.5380596287649864e-05,
+      "loss": 0.0534,
+      "step": 3287
+    },
+    {
+      "epoch": 11.742857142857144,
+      "grad_norm": 1.003724147730019,
+      "learning_rate": 1.5369342839230327e-05,
+      "loss": 0.0675,
+      "step": 3288
+    },
+    {
+      "epoch": 11.746428571428572,
+      "grad_norm": 0.6834720973350638,
+      "learning_rate": 1.535809093970886e-05,
+      "loss": 0.0675,
+      "step": 3289
+    },
+    {
+      "epoch": 11.75,
+      "grad_norm": 0.6611600170365026,
+      "learning_rate": 1.5346840592849083e-05,
+      "loss": 0.0245,
+      "step": 3290
+    },
+    {
+      "epoch": 11.753571428571428,
+      "grad_norm": 0.958742381596911,
+      "learning_rate": 1.5335591802414106e-05,
+      "loss": 0.0502,
+      "step": 3291
+    },
+    {
+      "epoch": 11.757142857142856,
+      "grad_norm": 0.8254260576789345,
+      "learning_rate": 1.5324344572166513e-05,
+      "loss": 0.0357,
+      "step": 3292
+    },
+    {
+      "epoch": 11.760714285714286,
+      "grad_norm": 0.8587154143860137,
+      "learning_rate": 1.531309890586836e-05,
+      "loss": 0.0762,
+      "step": 3293
+    },
+    {
+      "epoch": 11.764285714285714,
+      "grad_norm": 0.9163183620257555,
+      "learning_rate": 1.530185480728118e-05,
+      "loss": 0.0349,
+      "step": 3294
+    },
+    {
+      "epoch": 11.767857142857142,
+      "grad_norm": 1.0553156637759138,
+      "learning_rate": 1.5290612280165995e-05,
+      "loss": 0.1292,
+      "step": 3295
+    },
+    {
+      "epoch": 11.771428571428572,
+      "grad_norm": 0.8096515298564535,
+      "learning_rate": 1.5279371328283288e-05,
+      "loss": 0.0712,
+      "step": 3296
+    },
+    {
+      "epoch": 11.775,
+      "grad_norm": 0.8859772188985551,
+      "learning_rate": 1.526813195539302e-05,
+      "loss": 0.0581,
+      "step": 3297
+    },
+    {
+      "epoch": 11.778571428571428,
+      "grad_norm": 0.684829772686069,
+      "learning_rate": 1.525689416525463e-05,
+      "loss": 0.0502,
+      "step": 3298
+    },
+    {
+      "epoch": 11.782142857142857,
+      "grad_norm": 0.823482451145626,
+      "learning_rate": 1.5245657961627011e-05,
+      "loss": 0.0368,
+      "step": 3299
+    },
+    {
+      "epoch": 11.785714285714286,
+      "grad_norm": 1.2088812713229133,
+      "learning_rate": 1.523442334826854e-05,
+      "loss": 0.0855,
+      "step": 3300
+    },
+    {
+      "epoch": 11.789285714285715,
+      "grad_norm": 0.4987499785366555,
+      "learning_rate": 1.5223190328937057e-05,
+      "loss": 0.0289,
+      "step": 3301
+    },
+    {
+      "epoch": 11.792857142857143,
+      "grad_norm": 1.0812531091443465,
+      "learning_rate": 1.5211958907389868e-05,
+      "loss": 0.0682,
+      "step": 3302
+    },
+    {
+      "epoch": 11.79642857142857,
+      "grad_norm": 0.890149362953474,
+      "learning_rate": 1.5200729087383748e-05,
+      "loss": 0.0561,
+      "step": 3303
+    },
+    {
+      "epoch": 11.8,
+      "grad_norm": 0.7310251720694966,
+      "learning_rate": 1.5189500872674934e-05,
+      "loss": 0.0333,
+      "step": 3304
+    },
+    {
+      "epoch": 11.803571428571429,
+      "grad_norm": 0.3995717123056673,
+      "learning_rate": 1.5178274267019124e-05,
+      "loss": 0.0346,
+      "step": 3305
+    },
+    {
+      "epoch": 11.807142857142857,
+      "grad_norm": 0.5552356756021857,
+      "learning_rate": 1.5167049274171483e-05,
+      "loss": 0.0483,
+      "step": 3306
+    },
+    {
+      "epoch": 11.810714285714285,
+      "grad_norm": 0.4705195587142091,
+      "learning_rate": 1.5155825897886633e-05,
+      "loss": 0.0296,
+      "step": 3307
+    },
+    {
+      "epoch": 11.814285714285715,
+      "grad_norm": 1.123488565200705,
+      "learning_rate": 1.5144604141918656e-05,
+      "loss": 0.0743,
+      "step": 3308
+    },
+    {
+      "epoch": 11.817857142857143,
+      "grad_norm": 0.6709922350325744,
+      "learning_rate": 1.5133384010021084e-05,
+      "loss": 0.0443,
+      "step": 3309
+    },
+    {
+      "epoch": 11.821428571428571,
+      "grad_norm": 0.9003377406157557,
+      "learning_rate": 1.5122165505946925e-05,
+      "loss": 0.0901,
+      "step": 3310
+    },
+    {
+      "epoch": 11.825,
+      "grad_norm": 0.744807798440444,
+      "learning_rate": 1.5110948633448627e-05,
+      "loss": 0.0815,
+      "step": 3311
+    },
+    {
+      "epoch": 11.82857142857143,
+      "grad_norm": 1.165994994222213,
+      "learning_rate": 1.5099733396278095e-05,
+      "loss": 0.0721,
+      "step": 3312
+    },
+    {
+      "epoch": 11.832142857142857,
+      "grad_norm": 0.9465310202119072,
+      "learning_rate": 1.5088519798186693e-05,
+      "loss": 0.0648,
+      "step": 3313
+    },
+    {
+      "epoch": 11.835714285714285,
+      "grad_norm": 0.624874707605096,
+      "learning_rate": 1.5077307842925227e-05,
+      "loss": 0.1087,
+      "step": 3314
+    },
+    {
+      "epoch": 11.839285714285714,
+      "grad_norm": 0.799977657923022,
+      "learning_rate": 1.506609753424396e-05,
+      "loss": 0.0689,
+      "step": 3315
+    },
+    {
+      "epoch": 11.842857142857143,
+      "grad_norm": 0.8730417119823992,
+      "learning_rate": 1.50548888758926e-05,
+      "loss": 0.0546,
+      "step": 3316
+    },
+    {
+      "epoch": 11.846428571428572,
+      "grad_norm": 0.616973404302292,
+      "learning_rate": 1.5043681871620317e-05,
+      "loss": 0.0291,
+      "step": 3317
+    },
+    {
+      "epoch": 11.85,
+      "grad_norm": 0.6289720314379856,
+      "learning_rate": 1.5032476525175703e-05,
+      "loss": 0.0267,
+      "step": 3318
+    },
+    {
+      "epoch": 11.853571428571428,
+      "grad_norm": 1.2327348435747678,
+      "learning_rate": 1.5021272840306816e-05,
+      "loss": 0.0553,
+      "step": 3319
+    },
+    {
+      "epoch": 11.857142857142858,
+      "grad_norm": 1.043274709732153,
+      "learning_rate": 1.5010070820761148e-05,
+      "loss": 0.0649,
+      "step": 3320
+    },
+    {
+      "epoch": 11.860714285714286,
+      "grad_norm": 0.608166386089052,
+      "learning_rate": 1.4998870470285641e-05,
+      "loss": 0.0367,
+      "step": 3321
+    },
+    {
+      "epoch": 11.864285714285714,
+      "grad_norm": 0.7203001960190467,
+      "learning_rate": 1.4987671792626672e-05,
+      "loss": 0.032,
+      "step": 3322
+    },
+    {
+      "epoch": 11.867857142857144,
+      "grad_norm": 0.7073514906755034,
+      "learning_rate": 1.4976474791530065e-05,
+      "loss": 0.0449,
+      "step": 3323
+    },
+    {
+      "epoch": 11.871428571428572,
+      "grad_norm": 0.603473949883832,
+      "learning_rate": 1.4965279470741072e-05,
+      "loss": 0.0583,
+      "step": 3324
+    },
+    {
+      "epoch": 11.875,
+      "grad_norm": 1.0769460479099426,
+      "learning_rate": 1.49540858340044e-05,
+      "loss": 0.0567,
+      "step": 3325
+    },
+    {
+      "epoch": 11.878571428571428,
+      "grad_norm": 0.40166937430686495,
+      "learning_rate": 1.4942893885064177e-05,
+      "loss": 0.0384,
+      "step": 3326
+    },
+    {
+      "epoch": 11.882142857142856,
+      "grad_norm": 0.7859726983037544,
+      "learning_rate": 1.4931703627663977e-05,
+      "loss": 0.0368,
+      "step": 3327
+    },
+    {
+      "epoch": 11.885714285714286,
+      "grad_norm": 1.063730516436604,
+      "learning_rate": 1.4920515065546801e-05,
+      "loss": 0.0754,
+      "step": 3328
+    },
+    {
+      "epoch": 11.889285714285714,
+      "grad_norm": 0.8717181116214139,
+      "learning_rate": 1.4909328202455089e-05,
+      "loss": 0.1185,
+      "step": 3329
+    },
+    {
+      "epoch": 11.892857142857142,
+      "grad_norm": 1.1816817509473587,
+      "learning_rate": 1.4898143042130705e-05,
+      "loss": 0.0391,
+      "step": 3330
+    },
+    {
+      "epoch": 11.896428571428572,
+      "grad_norm": 0.6527636596263849,
+      "learning_rate": 1.4886959588314951e-05,
+      "loss": 0.0581,
+      "step": 3331
+    },
+    {
+      "epoch": 11.9,
+      "grad_norm": 1.07353271642682,
+      "learning_rate": 1.4875777844748553e-05,
+      "loss": 0.0464,
+      "step": 3332
+    },
+    {
+      "epoch": 11.903571428571428,
+      "grad_norm": 0.9948080609165498,
+      "learning_rate": 1.4864597815171671e-05,
+      "loss": 0.0626,
+      "step": 3333
+    },
+    {
+      "epoch": 11.907142857142857,
+      "grad_norm": 1.5175279226178433,
+      "learning_rate": 1.4853419503323885e-05,
+      "loss": 0.0429,
+      "step": 3334
+    },
+    {
+      "epoch": 11.910714285714286,
+      "grad_norm": 0.8599169984866643,
+      "learning_rate": 1.4842242912944203e-05,
+      "loss": 0.0402,
+      "step": 3335
+    },
+    {
+      "epoch": 11.914285714285715,
+      "grad_norm": 2.0118868500650735,
+      "learning_rate": 1.4831068047771057e-05,
+      "loss": 0.059,
+      "step": 3336
+    },
+    {
+      "epoch": 11.917857142857143,
+      "grad_norm": 1.262746043930487,
+      "learning_rate": 1.4819894911542308e-05,
+      "loss": 0.0614,
+      "step": 3337
+    },
+    {
+      "epoch": 11.92142857142857,
+      "grad_norm": 1.910960947799482,
+      "learning_rate": 1.4808723507995226e-05,
+      "loss": 0.0708,
+      "step": 3338
+    },
+    {
+      "epoch": 11.925,
+      "grad_norm": 0.7498341703151995,
+      "learning_rate": 1.4797553840866514e-05,
+      "loss": 0.0615,
+      "step": 3339
+    },
+    {
+      "epoch": 11.928571428571429,
+      "grad_norm": 0.3904480245837106,
+      "learning_rate": 1.4786385913892287e-05,
+      "loss": 0.0169,
+      "step": 3340
+    },
+    {
+      "epoch": 11.932142857142857,
+      "grad_norm": 0.5964249253937536,
+      "learning_rate": 1.477521973080808e-05,
+      "loss": 0.0322,
+      "step": 3341
+    },
+    {
+      "epoch": 11.935714285714285,
+      "grad_norm": 0.5190844267506831,
+      "learning_rate": 1.4764055295348842e-05,
+      "loss": 0.0502,
+      "step": 3342
+    },
+    {
+      "epoch": 11.939285714285715,
+      "grad_norm": 0.46997230046483857,
+      "learning_rate": 1.4752892611248943e-05,
+      "loss": 0.0228,
+      "step": 3343
+    },
+    {
+      "epoch": 11.942857142857143,
+      "grad_norm": 1.366345341152276,
+      "learning_rate": 1.474173168224216e-05,
+      "loss": 0.067,
+      "step": 3344
+    },
+    {
+      "epoch": 11.946428571428571,
+      "grad_norm": 1.0278137479501532,
+      "learning_rate": 1.4730572512061692e-05,
+      "loss": 0.0582,
+      "step": 3345
+    },
+    {
+      "epoch": 11.95,
+      "grad_norm": 0.6729632217152701,
+      "learning_rate": 1.471941510444014e-05,
+      "loss": 0.0512,
+      "step": 3346
+    },
+    {
+      "epoch": 11.95357142857143,
+      "grad_norm": 1.3400193501112212,
+      "learning_rate": 1.4708259463109522e-05,
+      "loss": 0.0916,
+      "step": 3347
+    },
+    {
+      "epoch": 11.957142857142857,
+      "grad_norm": 1.3476080614394463,
+      "learning_rate": 1.4697105591801263e-05,
+      "loss": 0.0755,
+      "step": 3348
+    },
+    {
+      "epoch": 11.960714285714285,
+      "grad_norm": 1.169347059287544,
+      "learning_rate": 1.4685953494246201e-05,
+      "loss": 0.0746,
+      "step": 3349
+    },
+    {
+      "epoch": 11.964285714285714,
+      "grad_norm": 0.63668635569703,
+      "learning_rate": 1.4674803174174567e-05,
+      "loss": 0.0553,
+      "step": 3350
+    },
+    {
+      "epoch": 11.967857142857143,
+      "grad_norm": 0.7448423401796431,
+      "learning_rate": 1.4663654635316014e-05,
+      "loss": 0.0371,
+      "step": 3351
+    },
+    {
+      "epoch": 11.971428571428572,
+      "grad_norm": 1.0914486684765858,
+      "learning_rate": 1.465250788139959e-05,
+      "loss": 0.0551,
+      "step": 3352
+    },
+    {
+      "epoch": 11.975,
+      "grad_norm": 0.7502380064227216,
+      "learning_rate": 1.4641362916153742e-05,
+      "loss": 0.0593,
+      "step": 3353
+    },
+    {
+      "epoch": 11.978571428571428,
+      "grad_norm": 0.863473285459803,
+      "learning_rate": 1.4630219743306336e-05,
+      "loss": 0.1145,
+      "step": 3354
+    },
+    {
+      "epoch": 11.982142857142858,
+      "grad_norm": 1.0521520877708515,
+      "learning_rate": 1.4619078366584613e-05,
+      "loss": 0.0371,
+      "step": 3355
+    },
+    {
+      "epoch": 11.985714285714286,
+      "grad_norm": 0.5738283115515526,
+      "learning_rate": 1.460793878971524e-05,
+      "loss": 0.0529,
+      "step": 3356
+    },
+    {
+      "epoch": 11.989285714285714,
+      "grad_norm": 0.5572000610848324,
+      "learning_rate": 1.4596801016424255e-05,
+      "loss": 0.0497,
+      "step": 3357
+    },
+    {
+      "epoch": 11.992857142857144,
+      "grad_norm": 1.2171991335946573,
+      "learning_rate": 1.4585665050437116e-05,
+      "loss": 0.0298,
+      "step": 3358
+    },
+    {
+      "epoch": 11.996428571428572,
+      "grad_norm": 0.27742775645161827,
+      "learning_rate": 1.4574530895478668e-05,
+      "loss": 0.0238,
+      "step": 3359
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 0.8169519608895074,
+      "learning_rate": 1.4563398555273143e-05,
+      "loss": 0.0791,
+      "step": 3360
+    },
+    {
+      "epoch": 12.003571428571428,
+      "grad_norm": 0.8938718837946952,
+      "learning_rate": 1.4552268033544179e-05,
+      "loss": 0.0429,
+      "step": 3361
+    },
+    {
+      "epoch": 12.007142857142858,
+      "grad_norm": 0.8778225965498936,
+      "learning_rate": 1.4541139334014795e-05,
+      "loss": 0.0608,
+      "step": 3362
+    },
+    {
+      "epoch": 12.010714285714286,
+      "grad_norm": 0.5169018456380303,
+      "learning_rate": 1.4530012460407406e-05,
+      "loss": 0.0173,
+      "step": 3363
+    },
+    {
+      "epoch": 12.014285714285714,
+      "grad_norm": 0.38518910922989624,
+      "learning_rate": 1.4518887416443815e-05,
+      "loss": 0.0363,
+      "step": 3364
+    },
+    {
+      "epoch": 12.017857142857142,
+      "grad_norm": 0.5555364400087046,
+      "learning_rate": 1.450776420584521e-05,
+      "loss": 0.0349,
+      "step": 3365
+    },
+    {
+      "epoch": 12.021428571428572,
+      "grad_norm": 0.7045123116442956,
+      "learning_rate": 1.4496642832332176e-05,
+      "loss": 0.0407,
+      "step": 3366
+    },
+    {
+      "epoch": 12.025,
+      "grad_norm": 0.6034379161801056,
+      "learning_rate": 1.4485523299624671e-05,
+      "loss": 0.0521,
+      "step": 3367
+    },
+    {
+      "epoch": 12.028571428571428,
+      "grad_norm": 1.2222941418886257,
+      "learning_rate": 1.447440561144204e-05,
+      "loss": 0.0408,
+      "step": 3368
+    },
+    {
+      "epoch": 12.032142857142857,
+      "grad_norm": 0.7655071185775568,
+      "learning_rate": 1.4463289771503015e-05,
+      "loss": 0.0561,
+      "step": 3369
+    },
+    {
+      "epoch": 12.035714285714286,
+      "grad_norm": 0.6902492952626574,
+      "learning_rate": 1.445217578352571e-05,
+      "loss": 0.0452,
+      "step": 3370
+    },
+    {
+      "epoch": 12.039285714285715,
+      "grad_norm": 0.6899919582905154,
+      "learning_rate": 1.4441063651227616e-05,
+      "loss": 0.034,
+      "step": 3371
+    },
+    {
+      "epoch": 12.042857142857143,
+      "grad_norm": 0.6494064395141637,
+      "learning_rate": 1.442995337832561e-05,
+      "loss": 0.0633,
+      "step": 3372
+    },
+    {
+      "epoch": 12.04642857142857,
+      "grad_norm": 0.7770459016555268,
+      "learning_rate": 1.4418844968535934e-05,
+      "loss": 0.0507,
+      "step": 3373
+    },
+    {
+      "epoch": 12.05,
+      "grad_norm": 0.9007868587305013,
+      "learning_rate": 1.4407738425574223e-05,
+      "loss": 0.0594,
+      "step": 3374
+    },
+    {
+      "epoch": 12.053571428571429,
+      "grad_norm": 0.7172140719368991,
+      "learning_rate": 1.4396633753155473e-05,
+      "loss": 0.0311,
+      "step": 3375
+    },
+    {
+      "epoch": 12.057142857142857,
+      "grad_norm": 1.425510759856527,
+      "learning_rate": 1.4385530954994064e-05,
+      "loss": 0.0436,
+      "step": 3376
+    },
+    {
+      "epoch": 12.060714285714285,
+      "grad_norm": 0.9958035141469781,
+      "learning_rate": 1.4374430034803748e-05,
+      "loss": 0.0511,
+      "step": 3377
+    },
+    {
+      "epoch": 12.064285714285715,
+      "grad_norm": 0.8663509257559798,
+      "learning_rate": 1.4363330996297646e-05,
+      "loss": 0.0309,
+      "step": 3378
+    },
+    {
+      "epoch": 12.067857142857143,
+      "grad_norm": 1.0977649607312614,
+      "learning_rate": 1.4352233843188246e-05,
+      "loss": 0.0893,
+      "step": 3379
+    },
+    {
+      "epoch": 12.071428571428571,
+      "grad_norm": 1.4010857079258792,
+      "learning_rate": 1.434113857918741e-05,
+      "loss": 0.0556,
+      "step": 3380
+    },
+    {
+      "epoch": 12.075,
+      "grad_norm": 0.6631221084348473,
+      "learning_rate": 1.4330045208006374e-05,
+      "loss": 0.0413,
+      "step": 3381
+    },
+    {
+      "epoch": 12.07857142857143,
+      "grad_norm": 0.6149515963846406,
+      "learning_rate": 1.4318953733355728e-05,
+      "loss": 0.0189,
+      "step": 3382
+    },
+    {
+      "epoch": 12.082142857142857,
+      "grad_norm": 0.7853189712402519,
+      "learning_rate": 1.4307864158945437e-05,
+      "loss": 0.0319,
+      "step": 3383
+    },
+    {
+      "epoch": 12.085714285714285,
+      "grad_norm": 1.1854005188880408,
+      "learning_rate": 1.4296776488484828e-05,
+      "loss": 0.0826,
+      "step": 3384
+    },
+    {
+      "epoch": 12.089285714285714,
+      "grad_norm": 1.019457121164725,
+      "learning_rate": 1.4285690725682595e-05,
+      "loss": 0.0523,
+      "step": 3385
+    },
+    {
+      "epoch": 12.092857142857143,
+      "grad_norm": 0.9370994772150845,
+      "learning_rate": 1.4274606874246781e-05,
+      "loss": 0.0483,
+      "step": 3386
+    },
+    {
+      "epoch": 12.096428571428572,
+      "grad_norm": 1.101971263513939,
+      "learning_rate": 1.4263524937884806e-05,
+      "loss": 0.064,
+      "step": 3387
+    },
+    {
+      "epoch": 12.1,
+      "grad_norm": 1.6632620359272272,
+      "learning_rate": 1.4252444920303438e-05,
+      "loss": 0.0469,
+      "step": 3388
+    },
+    {
+      "epoch": 12.103571428571428,
+      "grad_norm": 0.3835989527339793,
+      "learning_rate": 1.4241366825208807e-05,
+      "loss": 0.0146,
+      "step": 3389
+    },
+    {
+      "epoch": 12.107142857142858,
+      "grad_norm": 1.5298704372564405,
+      "learning_rate": 1.4230290656306402e-05,
+      "loss": 0.0621,
+      "step": 3390
+    },
+    {
+      "epoch": 12.110714285714286,
+      "grad_norm": 1.336693225640036,
+      "learning_rate": 1.4219216417301071e-05,
+      "loss": 0.0478,
+      "step": 3391
+    },
+    {
+      "epoch": 12.114285714285714,
+      "grad_norm": 0.8127941732542067,
+      "learning_rate": 1.4208144111897e-05,
+      "loss": 0.0544,
+      "step": 3392
+    },
+    {
+      "epoch": 12.117857142857142,
+      "grad_norm": 1.796418307475875,
+      "learning_rate": 1.4197073743797745e-05,
+      "loss": 0.0748,
+      "step": 3393
+    },
+    {
+      "epoch": 12.121428571428572,
+      "grad_norm": 1.0039723340109932,
+      "learning_rate": 1.4186005316706207e-05,
+      "loss": 0.0269,
+      "step": 3394
+    },
+    {
+      "epoch": 12.125,
+      "grad_norm": 0.8761233534507042,
+      "learning_rate": 1.417493883432464e-05,
+      "loss": 0.0529,
+      "step": 3395
+    },
+    {
+      "epoch": 12.128571428571428,
+      "grad_norm": 1.0277233318169754,
+      "learning_rate": 1.416387430035465e-05,
+      "loss": 0.0498,
+      "step": 3396
+    },
+    {
+      "epoch": 12.132142857142858,
+      "grad_norm": 1.1051506928251336,
+      "learning_rate": 1.4152811718497184e-05,
+      "loss": 0.0712,
+      "step": 3397
+    },
+    {
+      "epoch": 12.135714285714286,
+      "grad_norm": 0.9517893263153414,
+      "learning_rate": 1.414175109245254e-05,
+      "loss": 0.0376,
+      "step": 3398
+    },
+    {
+      "epoch": 12.139285714285714,
+      "grad_norm": 0.8362407838830498,
+      "learning_rate": 1.4130692425920363e-05,
+      "loss": 0.1171,
+      "step": 3399
+    },
+    {
+      "epoch": 12.142857142857142,
+      "grad_norm": 1.2680633495631446,
+      "learning_rate": 1.411963572259964e-05,
+      "loss": 0.0599,
+      "step": 3400
+    },
+    {
+      "epoch": 12.146428571428572,
+      "grad_norm": 0.9635254005493409,
+      "learning_rate": 1.4108580986188706e-05,
+      "loss": 0.0438,
+      "step": 3401
+    },
+    {
+      "epoch": 12.15,
+      "grad_norm": 1.363013084103542,
+      "learning_rate": 1.4097528220385235e-05,
+      "loss": 0.052,
+      "step": 3402
+    },
+    {
+      "epoch": 12.153571428571428,
+      "grad_norm": 0.513863770305192,
+      "learning_rate": 1.4086477428886243e-05,
+      "loss": 0.0345,
+      "step": 3403
+    },
+    {
+      "epoch": 12.157142857142857,
+      "grad_norm": 1.0316310374140512,
+      "learning_rate": 1.4075428615388077e-05,
+      "loss": 0.0719,
+      "step": 3404
+    },
+    {
+      "epoch": 12.160714285714286,
+      "grad_norm": 0.8425777513807606,
+      "learning_rate": 1.4064381783586432e-05,
+      "loss": 0.0472,
+      "step": 3405
+    },
+    {
+      "epoch": 12.164285714285715,
+      "grad_norm": 0.8231425395961113,
+      "learning_rate": 1.405333693717634e-05,
+      "loss": 0.0475,
+      "step": 3406
+    },
+    {
+      "epoch": 12.167857142857143,
+      "grad_norm": 0.6269103205799467,
+      "learning_rate": 1.4042294079852167e-05,
+      "loss": 0.0399,
+      "step": 3407
+    },
+    {
+      "epoch": 12.17142857142857,
+      "grad_norm": 0.9779795106712009,
+      "learning_rate": 1.4031253215307614e-05,
+      "loss": 0.0529,
+      "step": 3408
+    },
+    {
+      "epoch": 12.175,
+      "grad_norm": 1.5900627674119416,
+      "learning_rate": 1.4020214347235714e-05,
+      "loss": 0.0951,
+      "step": 3409
+    },
+    {
+      "epoch": 12.178571428571429,
+      "grad_norm": 1.5327528456357569,
+      "learning_rate": 1.4009177479328833e-05,
+      "loss": 0.0442,
+      "step": 3410
+    },
+    {
+      "epoch": 12.182142857142857,
+      "grad_norm": 1.4014017674037562,
+      "learning_rate": 1.3998142615278667e-05,
+      "loss": 0.0627,
+      "step": 3411
+    },
+    {
+      "epoch": 12.185714285714285,
+      "grad_norm": 0.5215234863398119,
+      "learning_rate": 1.3987109758776243e-05,
+      "loss": 0.0243,
+      "step": 3412
+    },
+    {
+      "epoch": 12.189285714285715,
+      "grad_norm": 0.5274045668518653,
+      "learning_rate": 1.397607891351192e-05,
+      "loss": 0.0333,
+      "step": 3413
+    },
+    {
+      "epoch": 12.192857142857143,
+      "grad_norm": 0.6157452763880351,
+      "learning_rate": 1.3965050083175374e-05,
+      "loss": 0.0657,
+      "step": 3414
+    },
+    {
+      "epoch": 12.196428571428571,
+      "grad_norm": 0.6680020485596639,
+      "learning_rate": 1.3954023271455622e-05,
+      "loss": 0.0635,
+      "step": 3415
+    },
+    {
+      "epoch": 12.2,
+      "grad_norm": 0.5652565866412272,
+      "learning_rate": 1.394299848204099e-05,
+      "loss": 0.0358,
+      "step": 3416
+    },
+    {
+      "epoch": 12.20357142857143,
+      "grad_norm": 0.6749171760687549,
+      "learning_rate": 1.3931975718619137e-05,
+      "loss": 0.0435,
+      "step": 3417
+    },
+    {
+      "epoch": 12.207142857142857,
+      "grad_norm": 0.6434460218622178,
+      "learning_rate": 1.3920954984877041e-05,
+      "loss": 0.0398,
+      "step": 3418
+    },
+    {
+      "epoch": 12.210714285714285,
+      "grad_norm": 0.8153251041942121,
+      "learning_rate": 1.3909936284501008e-05,
+      "loss": 0.0469,
+      "step": 3419
+    },
+    {
+      "epoch": 12.214285714285714,
+      "grad_norm": 0.9274964105190079,
+      "learning_rate": 1.3898919621176652e-05,
+      "loss": 0.0797,
+      "step": 3420
+    },
+    {
+      "epoch": 12.217857142857143,
+      "grad_norm": 0.4415099822009994,
+      "learning_rate": 1.3887904998588914e-05,
+      "loss": 0.0347,
+      "step": 3421
+    },
+    {
+      "epoch": 12.221428571428572,
+      "grad_norm": 1.7442510853285635,
+      "learning_rate": 1.387689242042205e-05,
+      "loss": 0.0837,
+      "step": 3422
+    },
+    {
+      "epoch": 12.225,
+      "grad_norm": 0.7555660390477041,
+      "learning_rate": 1.3865881890359636e-05,
+      "loss": 0.0482,
+      "step": 3423
+    },
+    {
+      "epoch": 12.228571428571428,
+      "grad_norm": 0.6382836787916353,
+      "learning_rate": 1.3854873412084552e-05,
+      "loss": 0.0447,
+      "step": 3424
+    },
+    {
+      "epoch": 12.232142857142858,
+      "grad_norm": 0.5669322500322365,
+      "learning_rate": 1.3843866989279009e-05,
+      "loss": 0.0393,
+      "step": 3425
+    },
+    {
+      "epoch": 12.235714285714286,
+      "grad_norm": 0.9119413688772505,
+      "learning_rate": 1.3832862625624512e-05,
+      "loss": 0.0744,
+      "step": 3426
+    },
+    {
+      "epoch": 12.239285714285714,
+      "grad_norm": 0.6136237714815085,
+      "learning_rate": 1.3821860324801888e-05,
+      "loss": 0.0228,
+      "step": 3427
+    },
+    {
+      "epoch": 12.242857142857142,
+      "grad_norm": 0.5663336826568265,
+      "learning_rate": 1.381086009049128e-05,
+      "loss": 0.0414,
+      "step": 3428
+    },
+    {
+      "epoch": 12.246428571428572,
+      "grad_norm": 0.9771940032014801,
+      "learning_rate": 1.3799861926372118e-05,
+      "loss": 0.0609,
+      "step": 3429
+    },
+    {
+      "epoch": 12.25,
+      "grad_norm": 0.7417045209153936,
+      "learning_rate": 1.3788865836123158e-05,
+      "loss": 0.0674,
+      "step": 3430
+    },
+    {
+      "epoch": 12.253571428571428,
+      "grad_norm": 0.6331839389798467,
+      "learning_rate": 1.377787182342246e-05,
+      "loss": 0.0395,
+      "step": 3431
+    },
+    {
+      "epoch": 12.257142857142856,
+      "grad_norm": 1.1265526409931357,
+      "learning_rate": 1.3766879891947384e-05,
+      "loss": 0.0666,
+      "step": 3432
+    },
+    {
+      "epoch": 12.260714285714286,
+      "grad_norm": 1.1349625105558516,
+      "learning_rate": 1.37558900453746e-05,
+      "loss": 0.0624,
+      "step": 3433
+    },
+    {
+      "epoch": 12.264285714285714,
+      "grad_norm": 0.6215686932449002,
+      "learning_rate": 1.3744902287380071e-05,
+      "loss": 0.0494,
+      "step": 3434
+    },
+    {
+      "epoch": 12.267857142857142,
+      "grad_norm": 0.9785634605437323,
+      "learning_rate": 1.3733916621639074e-05,
+      "loss": 0.0459,
+      "step": 3435
+    },
+    {
+      "epoch": 12.271428571428572,
+      "grad_norm": 0.6406634416391339,
+      "learning_rate": 1.3722933051826176e-05,
+      "loss": 0.0422,
+      "step": 3436
+    },
+    {
+      "epoch": 12.275,
+      "grad_norm": 0.98139734127228,
+      "learning_rate": 1.3711951581615246e-05,
+      "loss": 0.043,
+      "step": 3437
+    },
+    {
+      "epoch": 12.278571428571428,
+      "grad_norm": 0.648382029939972,
+      "learning_rate": 1.3700972214679453e-05,
+      "loss": 0.0278,
+      "step": 3438
+    },
+    {
+      "epoch": 12.282142857142857,
+      "grad_norm": 0.5429428405506126,
+      "learning_rate": 1.3689994954691261e-05,
+      "loss": 0.0374,
+      "step": 3439
+    },
+    {
+      "epoch": 12.285714285714286,
+      "grad_norm": 1.0948444425494408,
+      "learning_rate": 1.367901980532243e-05,
+      "loss": 0.0884,
+      "step": 3440
+    },
+    {
+      "epoch": 12.289285714285715,
+      "grad_norm": 0.923743042981013,
+      "learning_rate": 1.3668046770244008e-05,
+      "loss": 0.0252,
+      "step": 3441
+    },
+    {
+      "epoch": 12.292857142857143,
+      "grad_norm": 1.240987860037262,
+      "learning_rate": 1.3657075853126345e-05,
+      "loss": 0.0504,
+      "step": 3442
+    },
+    {
+      "epoch": 12.29642857142857,
+      "grad_norm": 1.4129665051759024,
+      "learning_rate": 1.3646107057639075e-05,
+      "loss": 0.062,
+      "step": 3443
+    },
+    {
+      "epoch": 12.3,
+      "grad_norm": 0.42852260750499477,
+      "learning_rate": 1.3635140387451129e-05,
+      "loss": 0.0153,
+      "step": 3444
+    },
+    {
+      "epoch": 12.303571428571429,
+      "grad_norm": 1.2626943909095178,
+      "learning_rate": 1.3624175846230721e-05,
+      "loss": 0.0462,
+      "step": 3445
+    },
+    {
+      "epoch": 12.307142857142857,
+      "grad_norm": 1.3492084102836337,
+      "learning_rate": 1.3613213437645359e-05,
+      "loss": 0.0804,
+      "step": 3446
+    },
+    {
+      "epoch": 12.310714285714285,
+      "grad_norm": 1.8574807385113659,
+      "learning_rate": 1.360225316536183e-05,
+      "loss": 0.0567,
+      "step": 3447
+    },
+    {
+      "epoch": 12.314285714285715,
+      "grad_norm": 0.6996940059540947,
+      "learning_rate": 1.3591295033046214e-05,
+      "loss": 0.0339,
+      "step": 3448
+    },
+    {
+      "epoch": 12.317857142857143,
+      "grad_norm": 2.0317774325153155,
+      "learning_rate": 1.358033904436387e-05,
+      "loss": 0.0823,
+      "step": 3449
+    },
+    {
+      "epoch": 12.321428571428571,
+      "grad_norm": 0.8521864128645733,
+      "learning_rate": 1.3569385202979443e-05,
+      "loss": 0.0351,
+      "step": 3450
+    },
+    {
+      "epoch": 12.325,
+      "grad_norm": 1.825070069946692,
+      "learning_rate": 1.355843351255686e-05,
+      "loss": 0.065,
+      "step": 3451
+    },
+    {
+      "epoch": 12.32857142857143,
+      "grad_norm": 1.0170774407575967,
+      "learning_rate": 1.354748397675933e-05,
+      "loss": 0.043,
+      "step": 3452
+    },
+    {
+      "epoch": 12.332142857142857,
+      "grad_norm": 0.8249629182986851,
+      "learning_rate": 1.3536536599249332e-05,
+      "loss": 0.0569,
+      "step": 3453
+    },
+    {
+      "epoch": 12.335714285714285,
+      "grad_norm": 0.6285398063276767,
+      "learning_rate": 1.352559138368863e-05,
+      "loss": 0.0379,
+      "step": 3454
+    },
+    {
+      "epoch": 12.339285714285714,
+      "grad_norm": 1.39249356492616,
+      "learning_rate": 1.351464833373827e-05,
+      "loss": 0.0451,
+      "step": 3455
+    },
+    {
+      "epoch": 12.342857142857143,
+      "grad_norm": 0.8301845755234408,
+      "learning_rate": 1.350370745305856e-05,
+      "loss": 0.032,
+      "step": 3456
+    },
+    {
+      "epoch": 12.346428571428572,
+      "grad_norm": 1.5331996587771821,
+      "learning_rate": 1.3492768745309098e-05,
+      "loss": 0.0342,
+      "step": 3457
+    },
+    {
+      "epoch": 12.35,
+      "grad_norm": 1.6211409536061543,
+      "learning_rate": 1.3481832214148744e-05,
+      "loss": 0.0799,
+      "step": 3458
+    },
+    {
+      "epoch": 12.353571428571428,
+      "grad_norm": 1.6176466867972215,
+      "learning_rate": 1.3470897863235637e-05,
+      "loss": 0.0621,
+      "step": 3459
+    },
+    {
+      "epoch": 12.357142857142858,
+      "grad_norm": 1.5008010653076784,
+      "learning_rate": 1.3459965696227177e-05,
+      "loss": 0.0425,
+      "step": 3460
+    },
+    {
+      "epoch": 12.360714285714286,
+      "grad_norm": 0.7122311251125087,
+      "learning_rate": 1.3449035716780046e-05,
+      "loss": 0.0258,
+      "step": 3461
+    },
+    {
+      "epoch": 12.364285714285714,
+      "grad_norm": 1.2078787559126045,
+      "learning_rate": 1.3438107928550185e-05,
+      "loss": 0.0423,
+      "step": 3462
+    },
+    {
+      "epoch": 12.367857142857144,
+      "grad_norm": 1.635705064665459,
+      "learning_rate": 1.3427182335192808e-05,
+      "loss": 0.0814,
+      "step": 3463
+    },
+    {
+      "epoch": 12.371428571428572,
+      "grad_norm": 1.1024481011095408,
+      "learning_rate": 1.3416258940362394e-05,
+      "loss": 0.0786,
+      "step": 3464
+    },
+    {
+      "epoch": 12.375,
+      "grad_norm": 1.0257117044043882,
+      "learning_rate": 1.3405337747712677e-05,
+      "loss": 0.05,
+      "step": 3465
+    },
+    {
+      "epoch": 12.378571428571428,
+      "grad_norm": 1.0108212007055128,
+      "learning_rate": 1.3394418760896665e-05,
+      "loss": 0.0823,
+      "step": 3466
+    },
+    {
+      "epoch": 12.382142857142856,
+      "grad_norm": 1.5586157824248525,
+      "learning_rate": 1.3383501983566628e-05,
+      "loss": 0.0413,
+      "step": 3467
+    },
+    {
+      "epoch": 12.385714285714286,
+      "grad_norm": 1.322214395532231,
+      "learning_rate": 1.3372587419374092e-05,
+      "loss": 0.0742,
+      "step": 3468
+    },
+    {
+      "epoch": 12.389285714285714,
+      "grad_norm": 0.9433100065005364,
+      "learning_rate": 1.3361675071969844e-05,
+      "loss": 0.0688,
+      "step": 3469
+    },
+    {
+      "epoch": 12.392857142857142,
+      "grad_norm": 0.7275259020754357,
+      "learning_rate": 1.3350764945003935e-05,
+      "loss": 0.032,
+      "step": 3470
+    },
+    {
+      "epoch": 12.396428571428572,
+      "grad_norm": 1.0442970003983194,
+      "learning_rate": 1.3339857042125664e-05,
+      "loss": 0.0395,
+      "step": 3471
+    },
+    {
+      "epoch": 12.4,
+      "grad_norm": 0.5450968482816492,
+      "learning_rate": 1.3328951366983594e-05,
+      "loss": 0.0262,
+      "step": 3472
+    },
+    {
+      "epoch": 12.403571428571428,
+      "grad_norm": 0.8483336725177391,
+      "learning_rate": 1.3318047923225537e-05,
+      "loss": 0.0744,
+      "step": 3473
+    },
+    {
+      "epoch": 12.407142857142857,
+      "grad_norm": 0.7769396404094209,
+      "learning_rate": 1.3307146714498562e-05,
+      "loss": 0.0488,
+      "step": 3474
+    },
+    {
+      "epoch": 12.410714285714286,
+      "grad_norm": 0.8295974775999287,
+      "learning_rate": 1.3296247744448994e-05,
+      "loss": 0.0791,
+      "step": 3475
+    },
+    {
+      "epoch": 12.414285714285715,
+      "grad_norm": 0.7184935107581222,
+      "learning_rate": 1.3285351016722398e-05,
+      "loss": 0.035,
+      "step": 3476
+    },
+    {
+      "epoch": 12.417857142857143,
+      "grad_norm": 0.7390134308571701,
+      "learning_rate": 1.3274456534963605e-05,
+      "loss": 0.0639,
+      "step": 3477
+    },
+    {
+      "epoch": 12.42142857142857,
+      "grad_norm": 0.7998834907467782,
+      "learning_rate": 1.3263564302816673e-05,
+      "loss": 0.0379,
+      "step": 3478
+    },
+    {
+      "epoch": 12.425,
+      "grad_norm": 1.0606910426348921,
+      "learning_rate": 1.3252674323924923e-05,
+      "loss": 0.031,
+      "step": 3479
+    },
+    {
+      "epoch": 12.428571428571429,
+      "grad_norm": 2.0302188263535093,
+      "learning_rate": 1.324178660193093e-05,
+      "loss": 0.0907,
+      "step": 3480
+    },
+    {
+      "epoch": 12.432142857142857,
+      "grad_norm": 0.5312731530512131,
+      "learning_rate": 1.3230901140476498e-05,
+      "loss": 0.0614,
+      "step": 3481
+    },
+    {
+      "epoch": 12.435714285714285,
+      "grad_norm": 0.5417877233665039,
+      "learning_rate": 1.3220017943202678e-05,
+      "loss": 0.0302,
+      "step": 3482
+    },
+    {
+      "epoch": 12.439285714285715,
+      "grad_norm": 0.6175320348726432,
+      "learning_rate": 1.3209137013749769e-05,
+      "loss": 0.0712,
+      "step": 3483
+    },
+    {
+      "epoch": 12.442857142857143,
+      "grad_norm": 1.855758960013007,
+      "learning_rate": 1.3198258355757311e-05,
+      "loss": 0.099,
+      "step": 3484
+    },
+    {
+      "epoch": 12.446428571428571,
+      "grad_norm": 0.8897650658729824,
+      "learning_rate": 1.3187381972864076e-05,
+      "loss": 0.0517,
+      "step": 3485
+    },
+    {
+      "epoch": 12.45,
+      "grad_norm": 0.8002027011344701,
+      "learning_rate": 1.3176507868708077e-05,
+      "loss": 0.0521,
+      "step": 3486
+    },
+    {
+      "epoch": 12.45357142857143,
+      "grad_norm": 0.6916192155570705,
+      "learning_rate": 1.3165636046926577e-05,
+      "loss": 0.0361,
+      "step": 3487
+    },
+    {
+      "epoch": 12.457142857142857,
+      "grad_norm": 0.5834593652593872,
+      "learning_rate": 1.3154766511156068e-05,
+      "loss": 0.0378,
+      "step": 3488
+    },
+    {
+      "epoch": 12.460714285714285,
+      "grad_norm": 1.0176849093266673,
+      "learning_rate": 1.314389926503227e-05,
+      "loss": 0.0328,
+      "step": 3489
+    },
+    {
+      "epoch": 12.464285714285714,
+      "grad_norm": 2.2471319648751567,
+      "learning_rate": 1.3133034312190146e-05,
+      "loss": 0.0672,
+      "step": 3490
+    },
+    {
+      "epoch": 12.467857142857143,
+      "grad_norm": 0.9461364017680091,
+      "learning_rate": 1.3122171656263894e-05,
+      "loss": 0.0623,
+      "step": 3491
+    },
+    {
+      "epoch": 12.471428571428572,
+      "grad_norm": 1.0991560144736332,
+      "learning_rate": 1.3111311300886932e-05,
+      "loss": 0.0609,
+      "step": 3492
+    },
+    {
+      "epoch": 12.475,
+      "grad_norm": 1.1160368847812319,
+      "learning_rate": 1.3100453249691922e-05,
+      "loss": 0.0654,
+      "step": 3493
+    },
+    {
+      "epoch": 12.478571428571428,
+      "grad_norm": 1.0072591297618456,
+      "learning_rate": 1.3089597506310745e-05,
+      "loss": 0.062,
+      "step": 3494
+    },
+    {
+      "epoch": 12.482142857142858,
+      "grad_norm": 1.1767663112012958,
+      "learning_rate": 1.3078744074374515e-05,
+      "loss": 0.0522,
+      "step": 3495
+    },
+    {
+      "epoch": 12.485714285714286,
+      "grad_norm": 1.5139308336458894,
+      "learning_rate": 1.3067892957513578e-05,
+      "loss": 0.0844,
+      "step": 3496
+    },
+    {
+      "epoch": 12.489285714285714,
+      "grad_norm": 0.8980972511969009,
+      "learning_rate": 1.3057044159357488e-05,
+      "loss": 0.0777,
+      "step": 3497
+    },
+    {
+      "epoch": 12.492857142857144,
+      "grad_norm": 0.6175206880681237,
+      "learning_rate": 1.3046197683535042e-05,
+      "loss": 0.0378,
+      "step": 3498
+    },
+    {
+      "epoch": 12.496428571428572,
+      "grad_norm": 0.7467617760517559,
+      "learning_rate": 1.3035353533674254e-05,
+      "loss": 0.047,
+      "step": 3499
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 1.014036272037455,
+      "learning_rate": 1.3024511713402355e-05,
+      "loss": 0.0636,
+      "step": 3500
+    },
+    {
+      "epoch": 12.503571428571428,
+      "grad_norm": 0.6562133414099965,
+      "learning_rate": 1.3013672226345807e-05,
+      "loss": 0.053,
+      "step": 3501
+    },
+    {
+      "epoch": 12.507142857142856,
+      "grad_norm": 1.1808947668283185,
+      "learning_rate": 1.300283507613028e-05,
+      "loss": 0.0489,
+      "step": 3502
+    },
+    {
+      "epoch": 12.510714285714286,
+      "grad_norm": 0.7431398533384982,
+      "learning_rate": 1.299200026638067e-05,
+      "loss": 0.046,
+      "step": 3503
+    },
+    {
+      "epoch": 12.514285714285714,
+      "grad_norm": 0.47572709419200243,
+      "learning_rate": 1.2981167800721085e-05,
+      "loss": 0.0218,
+      "step": 3504
+    },
+    {
+      "epoch": 12.517857142857142,
+      "grad_norm": 0.5443792161626382,
+      "learning_rate": 1.2970337682774861e-05,
+      "loss": 0.0434,
+      "step": 3505
+    },
+    {
+      "epoch": 12.521428571428572,
+      "grad_norm": 0.9825000394124331,
+      "learning_rate": 1.2959509916164532e-05,
+      "loss": 0.0666,
+      "step": 3506
+    },
+    {
+      "epoch": 12.525,
+      "grad_norm": 0.5862870898046149,
+      "learning_rate": 1.2948684504511854e-05,
+      "loss": 0.0621,
+      "step": 3507
+    },
+    {
+      "epoch": 12.528571428571428,
+      "grad_norm": 0.5794527290650222,
+      "learning_rate": 1.2937861451437804e-05,
+      "loss": 0.0164,
+      "step": 3508
+    },
+    {
+      "epoch": 12.532142857142857,
+      "grad_norm": 1.1494446835859022,
+      "learning_rate": 1.2927040760562543e-05,
+      "loss": 0.0809,
+      "step": 3509
+    },
+    {
+      "epoch": 12.535714285714286,
+      "grad_norm": 0.8238063638368176,
+      "learning_rate": 1.2916222435505471e-05,
+      "loss": 0.0374,
+      "step": 3510
+    },
+    {
+      "epoch": 12.539285714285715,
+      "grad_norm": 0.7273937595483569,
+      "learning_rate": 1.2905406479885185e-05,
+      "loss": 0.073,
+      "step": 3511
+    },
+    {
+      "epoch": 12.542857142857143,
+      "grad_norm": 0.402798715283014,
+      "learning_rate": 1.2894592897319487e-05,
+      "loss": 0.0177,
+      "step": 3512
+    },
+    {
+      "epoch": 12.54642857142857,
+      "grad_norm": 1.117505639105896,
+      "learning_rate": 1.2883781691425385e-05,
+      "loss": 0.0603,
+      "step": 3513
+    },
+    {
+      "epoch": 12.55,
+      "grad_norm": 0.6437702942044804,
+      "learning_rate": 1.28729728658191e-05,
+      "loss": 0.0593,
+      "step": 3514
+    },
+    {
+      "epoch": 12.553571428571429,
+      "grad_norm": 1.1785824718726563,
+      "learning_rate": 1.2862166424116053e-05,
+      "loss": 0.0624,
+      "step": 3515
+    },
+    {
+      "epoch": 12.557142857142857,
+      "grad_norm": 0.9091748490112618,
+      "learning_rate": 1.2851362369930862e-05,
+      "loss": 0.0685,
+      "step": 3516
+    },
+    {
+      "epoch": 12.560714285714285,
+      "grad_norm": 0.6284559202479295,
+      "learning_rate": 1.2840560706877349e-05,
+      "loss": 0.0375,
+      "step": 3517
+    },
+    {
+      "epoch": 12.564285714285715,
+      "grad_norm": 1.1883516189156444,
+      "learning_rate": 1.2829761438568544e-05,
+      "loss": 0.0557,
+      "step": 3518
+    },
+    {
+      "epoch": 12.567857142857143,
+      "grad_norm": 1.13971796424839,
+      "learning_rate": 1.2818964568616665e-05,
+      "loss": 0.0546,
+      "step": 3519
+    },
+    {
+      "epoch": 12.571428571428571,
+      "grad_norm": 1.1349981234665825,
+      "learning_rate": 1.2808170100633137e-05,
+      "loss": 0.1096,
+      "step": 3520
+    },
+    {
+      "epoch": 12.575,
+      "grad_norm": 0.5038740825065792,
+      "learning_rate": 1.2797378038228578e-05,
+      "loss": 0.0247,
+      "step": 3521
+    },
+    {
+      "epoch": 12.57857142857143,
+      "grad_norm": 1.0622367726208861,
+      "learning_rate": 1.2786588385012793e-05,
+      "loss": 0.0672,
+      "step": 3522
+    },
+    {
+      "epoch": 12.582142857142857,
+      "grad_norm": 0.7267240918532057,
+      "learning_rate": 1.2775801144594792e-05,
+      "loss": 0.0463,
+      "step": 3523
+    },
+    {
+      "epoch": 12.585714285714285,
+      "grad_norm": 1.2391845814506315,
+      "learning_rate": 1.276501632058278e-05,
+      "loss": 0.082,
+      "step": 3524
+    },
+    {
+      "epoch": 12.589285714285714,
+      "grad_norm": 0.6725686588298596,
+      "learning_rate": 1.2754233916584141e-05,
+      "loss": 0.0387,
+      "step": 3525
+    },
+    {
+      "epoch": 12.592857142857143,
+      "grad_norm": 0.4739056187449256,
+      "learning_rate": 1.2743453936205461e-05,
+      "loss": 0.0218,
+      "step": 3526
+    },
+    {
+      "epoch": 12.596428571428572,
+      "grad_norm": 1.137455729606227,
+      "learning_rate": 1.273267638305251e-05,
+      "loss": 0.0519,
+      "step": 3527
+    },
+    {
+      "epoch": 12.6,
+      "grad_norm": 1.0696914237648276,
+      "learning_rate": 1.2721901260730252e-05,
+      "loss": 0.0486,
+      "step": 3528
+    },
+    {
+      "epoch": 12.603571428571428,
+      "grad_norm": 0.7136159679369731,
+      "learning_rate": 1.2711128572842825e-05,
+      "loss": 0.0754,
+      "step": 3529
+    },
+    {
+      "epoch": 12.607142857142858,
+      "grad_norm": 0.6159416406025134,
+      "learning_rate": 1.2700358322993566e-05,
+      "loss": 0.0511,
+      "step": 3530
+    },
+    {
+      "epoch": 12.610714285714286,
+      "grad_norm": 0.5366951916342539,
+      "learning_rate": 1.2689590514784996e-05,
+      "loss": 0.0252,
+      "step": 3531
+    },
+    {
+      "epoch": 12.614285714285714,
+      "grad_norm": 1.3858147031589514,
+      "learning_rate": 1.2678825151818805e-05,
+      "loss": 0.0982,
+      "step": 3532
+    },
+    {
+      "epoch": 12.617857142857144,
+      "grad_norm": 1.3073301068835572,
+      "learning_rate": 1.2668062237695886e-05,
+      "loss": 0.0681,
+      "step": 3533
+    },
+    {
+      "epoch": 12.621428571428572,
+      "grad_norm": 0.700259967918617,
+      "learning_rate": 1.2657301776016293e-05,
+      "loss": 0.0542,
+      "step": 3534
+    },
+    {
+      "epoch": 12.625,
+      "grad_norm": 1.1310775646511422,
+      "learning_rate": 1.2646543770379272e-05,
+      "loss": 0.0776,
+      "step": 3535
+    },
+    {
+      "epoch": 12.628571428571428,
+      "grad_norm": 1.133327635020009,
+      "learning_rate": 1.2635788224383242e-05,
+      "loss": 0.049,
+      "step": 3536
+    },
+    {
+      "epoch": 12.632142857142856,
+      "grad_norm": 0.5050927065749166,
+      "learning_rate": 1.2625035141625804e-05,
+      "loss": 0.0215,
+      "step": 3537
+    },
+    {
+      "epoch": 12.635714285714286,
+      "grad_norm": 1.3733313400282234,
+      "learning_rate": 1.2614284525703728e-05,
+      "loss": 0.0367,
+      "step": 3538
+    },
+    {
+      "epoch": 12.639285714285714,
+      "grad_norm": 0.4609067904461785,
+      "learning_rate": 1.260353638021297e-05,
+      "loss": 0.0188,
+      "step": 3539
+    },
+    {
+      "epoch": 12.642857142857142,
+      "grad_norm": 0.6185160287985434,
+      "learning_rate": 1.259279070874865e-05,
+      "loss": 0.0272,
+      "step": 3540
+    },
+    {
+      "epoch": 12.646428571428572,
+      "grad_norm": 0.5157363032735087,
+      "learning_rate": 1.258204751490506e-05,
+      "loss": 0.0441,
+      "step": 3541
+    },
+    {
+      "epoch": 12.65,
+      "grad_norm": 0.8283591810207279,
+      "learning_rate": 1.2571306802275673e-05,
+      "loss": 0.0524,
+      "step": 3542
+    },
+    {
+      "epoch": 12.653571428571428,
+      "grad_norm": 1.1507291740928143,
+      "learning_rate": 1.256056857445312e-05,
+      "loss": 0.0591,
+      "step": 3543
+    },
+    {
+      "epoch": 12.657142857142857,
+      "grad_norm": 0.520999811457747,
+      "learning_rate": 1.254983283502921e-05,
+      "loss": 0.0373,
+      "step": 3544
+    },
+    {
+      "epoch": 12.660714285714286,
+      "grad_norm": 1.2238730508158833,
+      "learning_rate": 1.2539099587594921e-05,
+      "loss": 0.0842,
+      "step": 3545
+    },
+    {
+      "epoch": 12.664285714285715,
+      "grad_norm": 0.8165846788942648,
+      "learning_rate": 1.252836883574038e-05,
+      "loss": 0.0572,
+      "step": 3546
+    },
+    {
+      "epoch": 12.667857142857143,
+      "grad_norm": 0.720206484503646,
+      "learning_rate": 1.25176405830549e-05,
+      "loss": 0.0601,
+      "step": 3547
+    },
+    {
+      "epoch": 12.67142857142857,
+      "grad_norm": 1.5451542870646302,
+      "learning_rate": 1.250691483312695e-05,
+      "loss": 0.0527,
+      "step": 3548
+    },
+    {
+      "epoch": 12.675,
+      "grad_norm": 0.6813151741753039,
+      "learning_rate": 1.249619158954416e-05,
+      "loss": 0.0561,
+      "step": 3549
+    },
+    {
+      "epoch": 12.678571428571429,
+      "grad_norm": 0.8015111239221692,
+      "learning_rate": 1.2485470855893324e-05,
+      "loss": 0.0396,
+      "step": 3550
+    },
+    {
+      "epoch": 12.682142857142857,
+      "grad_norm": 0.9145711927489796,
+      "learning_rate": 1.2474752635760399e-05,
+      "loss": 0.0376,
+      "step": 3551
+    },
+    {
+      "epoch": 12.685714285714285,
+      "grad_norm": 1.3638667423420872,
+      "learning_rate": 1.2464036932730494e-05,
+      "loss": 0.0652,
+      "step": 3552
+    },
+    {
+      "epoch": 12.689285714285715,
+      "grad_norm": 0.7482284641102696,
+      "learning_rate": 1.2453323750387882e-05,
+      "loss": 0.0498,
+      "step": 3553
+    },
+    {
+      "epoch": 12.692857142857143,
+      "grad_norm": 2.05236154791854,
+      "learning_rate": 1.2442613092315994e-05,
+      "loss": 0.0607,
+      "step": 3554
+    },
+    {
+      "epoch": 12.696428571428571,
+      "grad_norm": 1.4726407521179259,
+      "learning_rate": 1.2431904962097407e-05,
+      "loss": 0.0537,
+      "step": 3555
+    },
+    {
+      "epoch": 12.7,
+      "grad_norm": 1.0883838282080658,
+      "learning_rate": 1.2421199363313866e-05,
+      "loss": 0.0336,
+      "step": 3556
+    },
+    {
+      "epoch": 12.70357142857143,
+      "grad_norm": 0.6724538370726215,
+      "learning_rate": 1.2410496299546266e-05,
+      "loss": 0.0295,
+      "step": 3557
+    },
+    {
+      "epoch": 12.707142857142857,
+      "grad_norm": 0.6344014984898164,
+      "learning_rate": 1.2399795774374636e-05,
+      "loss": 0.04,
+      "step": 3558
+    },
+    {
+      "epoch": 12.710714285714285,
+      "grad_norm": 0.8265461414074474,
+      "learning_rate": 1.2389097791378185e-05,
+      "loss": 0.0564,
+      "step": 3559
+    },
+    {
+      "epoch": 12.714285714285714,
+      "grad_norm": 0.5688267100247411,
+      "learning_rate": 1.2378402354135246e-05,
+      "loss": 0.0348,
+      "step": 3560
+    },
+    {
+      "epoch": 12.717857142857143,
+      "grad_norm": 0.31299671628617043,
+      "learning_rate": 1.236770946622332e-05,
+      "loss": 0.0112,
+      "step": 3561
+    },
+    {
+      "epoch": 12.721428571428572,
+      "grad_norm": 0.390485534507098,
+      "learning_rate": 1.2357019131219045e-05,
+      "loss": 0.0201,
+      "step": 3562
+    },
+    {
+      "epoch": 12.725,
+      "grad_norm": 0.49044221686295114,
+      "learning_rate": 1.2346331352698206e-05,
+      "loss": 0.0304,
+      "step": 3563
+    },
+    {
+      "epoch": 12.728571428571428,
+      "grad_norm": 2.4733274036600625,
+      "learning_rate": 1.2335646134235735e-05,
+      "loss": 0.0651,
+      "step": 3564
+    },
+    {
+      "epoch": 12.732142857142858,
+      "grad_norm": 0.9174506709239981,
+      "learning_rate": 1.2324963479405705e-05,
+      "loss": 0.0341,
+      "step": 3565
+    },
+    {
+      "epoch": 12.735714285714286,
+      "grad_norm": 1.034901214392877,
+      "learning_rate": 1.2314283391781334e-05,
+      "loss": 0.0515,
+      "step": 3566
+    },
+    {
+      "epoch": 12.739285714285714,
+      "grad_norm": 0.7874422711241122,
+      "learning_rate": 1.230360587493498e-05,
+      "loss": 0.0435,
+      "step": 3567
+    },
+    {
+      "epoch": 12.742857142857144,
+      "grad_norm": 0.6676610134208859,
+      "learning_rate": 1.2292930932438144e-05,
+      "loss": 0.0145,
+      "step": 3568
+    },
+    {
+      "epoch": 12.746428571428572,
+      "grad_norm": 0.683937531777004,
+      "learning_rate": 1.2282258567861458e-05,
+      "loss": 0.0644,
+      "step": 3569
+    },
+    {
+      "epoch": 12.75,
+      "grad_norm": 1.107518825752781,
+      "learning_rate": 1.2271588784774706e-05,
+      "loss": 0.086,
+      "step": 3570
+    },
+    {
+      "epoch": 12.753571428571428,
+      "grad_norm": 0.9555050607764228,
+      "learning_rate": 1.226092158674679e-05,
+      "loss": 0.0801,
+      "step": 3571
+    },
+    {
+      "epoch": 12.757142857142856,
+      "grad_norm": 1.2001420438738575,
+      "learning_rate": 1.2250256977345763e-05,
+      "loss": 0.072,
+      "step": 3572
+    },
+    {
+      "epoch": 12.760714285714286,
+      "grad_norm": 0.7220330194467786,
+      "learning_rate": 1.22395949601388e-05,
+      "loss": 0.054,
+      "step": 3573
+    },
+    {
+      "epoch": 12.764285714285714,
+      "grad_norm": 0.5555183877490165,
+      "learning_rate": 1.2228935538692226e-05,
+      "loss": 0.0172,
+      "step": 3574
+    },
+    {
+      "epoch": 12.767857142857142,
+      "grad_norm": 1.0345146447921254,
+      "learning_rate": 1.2218278716571477e-05,
+      "loss": 0.0802,
+      "step": 3575
+    },
+    {
+      "epoch": 12.771428571428572,
+      "grad_norm": 0.6764001017919147,
+      "learning_rate": 1.2207624497341133e-05,
+      "loss": 0.0365,
+      "step": 3576
+    },
+    {
+      "epoch": 12.775,
+      "grad_norm": 0.8945904868566492,
+      "learning_rate": 1.21969728845649e-05,
+      "loss": 0.0443,
+      "step": 3577
+    },
+    {
+      "epoch": 12.778571428571428,
+      "grad_norm": 0.6186200916922168,
+      "learning_rate": 1.2186323881805608e-05,
+      "loss": 0.0372,
+      "step": 3578
+    },
+    {
+      "epoch": 12.782142857142857,
+      "grad_norm": 1.8601182124827134,
+      "learning_rate": 1.2175677492625224e-05,
+      "loss": 0.0871,
+      "step": 3579
+    },
+    {
+      "epoch": 12.785714285714286,
+      "grad_norm": 0.8505006587553249,
+      "learning_rate": 1.2165033720584831e-05,
+      "loss": 0.0336,
+      "step": 3580
+    },
+    {
+      "epoch": 12.789285714285715,
+      "grad_norm": 0.7852137862793603,
+      "learning_rate": 1.2154392569244645e-05,
+      "loss": 0.0649,
+      "step": 3581
+    },
+    {
+      "epoch": 12.792857142857143,
+      "grad_norm": 0.37715940599882963,
+      "learning_rate": 1.2143754042163997e-05,
+      "loss": 0.0195,
+      "step": 3582
+    },
+    {
+      "epoch": 12.79642857142857,
+      "grad_norm": 0.4962003606063101,
+      "learning_rate": 1.2133118142901341e-05,
+      "loss": 0.0214,
+      "step": 3583
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 1.3075435642985698,
+      "learning_rate": 1.2122484875014261e-05,
+      "loss": 0.0374,
+      "step": 3584
+    },
+    {
+      "epoch": 12.803571428571429,
+      "grad_norm": 0.8015005440033395,
+      "learning_rate": 1.2111854242059449e-05,
+      "loss": 0.051,
+      "step": 3585
+    },
+    {
+      "epoch": 12.807142857142857,
+      "grad_norm": 1.5268875165369022,
+      "learning_rate": 1.2101226247592726e-05,
+      "loss": 0.1036,
+      "step": 3586
+    },
+    {
+      "epoch": 12.810714285714285,
+      "grad_norm": 0.8381380327796094,
+      "learning_rate": 1.209060089516902e-05,
+      "loss": 0.0416,
+      "step": 3587
+    },
+    {
+      "epoch": 12.814285714285715,
+      "grad_norm": 1.081460870205811,
+      "learning_rate": 1.207997818834239e-05,
+      "loss": 0.0407,
+      "step": 3588
+    },
+    {
+      "epoch": 12.817857142857143,
+      "grad_norm": 0.5329398992977065,
+      "learning_rate": 1.2069358130665996e-05,
+      "loss": 0.0245,
+      "step": 3589
+    },
+    {
+      "epoch": 12.821428571428571,
+      "grad_norm": 1.6150760013703096,
+      "learning_rate": 1.2058740725692115e-05,
+      "loss": 0.0519,
+      "step": 3590
+    },
+    {
+      "epoch": 12.825,
+      "grad_norm": 1.4622024433738137,
+      "learning_rate": 1.2048125976972144e-05,
+      "loss": 0.0558,
+      "step": 3591
+    },
+    {
+      "epoch": 12.82857142857143,
+      "grad_norm": 1.0662805994996711,
+      "learning_rate": 1.2037513888056583e-05,
+      "loss": 0.1067,
+      "step": 3592
+    },
+    {
+      "epoch": 12.832142857142857,
+      "grad_norm": 1.124001603552458,
+      "learning_rate": 1.2026904462495045e-05,
+      "loss": 0.0374,
+      "step": 3593
+    },
+    {
+      "epoch": 12.835714285714285,
+      "grad_norm": 1.0083142291227525,
+      "learning_rate": 1.2016297703836262e-05,
+      "loss": 0.0555,
+      "step": 3594
+    },
+    {
+      "epoch": 12.839285714285714,
+      "grad_norm": 1.2277740622108302,
+      "learning_rate": 1.200569361562805e-05,
+      "loss": 0.0535,
+      "step": 3595
+    },
+    {
+      "epoch": 12.842857142857143,
+      "grad_norm": 0.7748319420956415,
+      "learning_rate": 1.1995092201417355e-05,
+      "loss": 0.0648,
+      "step": 3596
+    },
+    {
+      "epoch": 12.846428571428572,
+      "grad_norm": 0.8272284338669492,
+      "learning_rate": 1.198449346475022e-05,
+      "loss": 0.0216,
+      "step": 3597
+    },
+    {
+      "epoch": 12.85,
+      "grad_norm": 1.1000695371547424,
+      "learning_rate": 1.197389740917179e-05,
+      "loss": 0.0509,
+      "step": 3598
+    },
+    {
+      "epoch": 12.853571428571428,
+      "grad_norm": 1.3393848750077606,
+      "learning_rate": 1.196330403822632e-05,
+      "loss": 0.0858,
+      "step": 3599
+    },
+    {
+      "epoch": 12.857142857142858,
+      "grad_norm": 1.2732087483669274,
+      "learning_rate": 1.1952713355457157e-05,
+      "loss": 0.0618,
+      "step": 3600
+    },
+    {
+      "epoch": 12.860714285714286,
+      "grad_norm": 0.6791576133479447,
+      "learning_rate": 1.1942125364406758e-05,
+      "loss": 0.0344,
+      "step": 3601
+    },
+    {
+      "epoch": 12.864285714285714,
+      "grad_norm": 0.8735417869000636,
+      "learning_rate": 1.1931540068616676e-05,
+      "loss": 0.0322,
+      "step": 3602
+    },
+    {
+      "epoch": 12.867857142857144,
+      "grad_norm": 1.7840380761208179,
+      "learning_rate": 1.1920957471627563e-05,
+      "loss": 0.1084,
+      "step": 3603
+    },
+    {
+      "epoch": 12.871428571428572,
+      "grad_norm": 0.997786589333906,
+      "learning_rate": 1.1910377576979166e-05,
+      "loss": 0.0384,
+      "step": 3604
+    },
+    {
+      "epoch": 12.875,
+      "grad_norm": 1.102841922732878,
+      "learning_rate": 1.1899800388210334e-05,
+      "loss": 0.0368,
+      "step": 3605
+    },
+    {
+      "epoch": 12.878571428571428,
+      "grad_norm": 1.0435166151156365,
+      "learning_rate": 1.1889225908859005e-05,
+      "loss": 0.0909,
+      "step": 3606
+    },
+    {
+      "epoch": 12.882142857142856,
+      "grad_norm": 0.9913351239549503,
+      "learning_rate": 1.1878654142462213e-05,
+      "loss": 0.086,
+      "step": 3607
+    },
+    {
+      "epoch": 12.885714285714286,
+      "grad_norm": 0.8623289903270395,
+      "learning_rate": 1.1868085092556081e-05,
+      "loss": 0.0408,
+      "step": 3608
+    },
+    {
+      "epoch": 12.889285714285714,
+      "grad_norm": 1.2857417341261488,
+      "learning_rate": 1.1857518762675829e-05,
+      "loss": 0.036,
+      "step": 3609
+    },
+    {
+      "epoch": 12.892857142857142,
+      "grad_norm": 0.6236726987283671,
+      "learning_rate": 1.1846955156355765e-05,
+      "loss": 0.0272,
+      "step": 3610
+    },
+    {
+      "epoch": 12.896428571428572,
+      "grad_norm": 0.6491552280317491,
+      "learning_rate": 1.1836394277129291e-05,
+      "loss": 0.0295,
+      "step": 3611
+    },
+    {
+      "epoch": 12.9,
+      "grad_norm": 1.6072447842631115,
+      "learning_rate": 1.1825836128528882e-05,
+      "loss": 0.084,
+      "step": 3612
+    },
+    {
+      "epoch": 12.903571428571428,
+      "grad_norm": 1.7346369158889752,
+      "learning_rate": 1.1815280714086112e-05,
+      "loss": 0.0595,
+      "step": 3613
+    },
+    {
+      "epoch": 12.907142857142857,
+      "grad_norm": 0.718019613840628,
+      "learning_rate": 1.1804728037331645e-05,
+      "loss": 0.0413,
+      "step": 3614
+    },
+    {
+      "epoch": 12.910714285714286,
+      "grad_norm": 0.9872409819930928,
+      "learning_rate": 1.1794178101795215e-05,
+      "loss": 0.0388,
+      "step": 3615
+    },
+    {
+      "epoch": 12.914285714285715,
+      "grad_norm": 1.6041956308315688,
+      "learning_rate": 1.1783630911005648e-05,
+      "loss": 0.0522,
+      "step": 3616
+    },
+    {
+      "epoch": 12.917857142857143,
+      "grad_norm": 1.1245134569370818,
+      "learning_rate": 1.1773086468490851e-05,
+      "loss": 0.0834,
+      "step": 3617
+    },
+    {
+      "epoch": 12.92142857142857,
+      "grad_norm": 1.278238807994833,
+      "learning_rate": 1.1762544777777807e-05,
+      "loss": 0.0462,
+      "step": 3618
+    },
+    {
+      "epoch": 12.925,
+      "grad_norm": 0.8480686515804075,
+      "learning_rate": 1.1752005842392591e-05,
+      "loss": 0.0331,
+      "step": 3619
+    },
+    {
+      "epoch": 12.928571428571429,
+      "grad_norm": 2.3637505112709993,
+      "learning_rate": 1.174146966586034e-05,
+      "loss": 0.0788,
+      "step": 3620
+    },
+    {
+      "epoch": 12.932142857142857,
+      "grad_norm": 1.0204620603179828,
+      "learning_rate": 1.1730936251705276e-05,
+      "loss": 0.0387,
+      "step": 3621
+    },
+    {
+      "epoch": 12.935714285714285,
+      "grad_norm": 1.175558860524099,
+      "learning_rate": 1.17204056034507e-05,
+      "loss": 0.0447,
+      "step": 3622
+    },
+    {
+      "epoch": 12.939285714285715,
+      "grad_norm": 0.9449282300146445,
+      "learning_rate": 1.1709877724618983e-05,
+      "loss": 0.0278,
+      "step": 3623
+    },
+    {
+      "epoch": 12.942857142857143,
+      "grad_norm": 0.6642065829804515,
+      "learning_rate": 1.1699352618731571e-05,
+      "loss": 0.025,
+      "step": 3624
+    },
+    {
+      "epoch": 12.946428571428571,
+      "grad_norm": 0.9705890426815594,
+      "learning_rate": 1.1688830289308983e-05,
+      "loss": 0.0481,
+      "step": 3625
+    },
+    {
+      "epoch": 12.95,
+      "grad_norm": 0.6027801910188397,
+      "learning_rate": 1.1678310739870815e-05,
+      "loss": 0.0259,
+      "step": 3626
+    },
+    {
+      "epoch": 12.95357142857143,
+      "grad_norm": 0.7485292566888742,
+      "learning_rate": 1.1667793973935724e-05,
+      "loss": 0.0366,
+      "step": 3627
+    },
+    {
+      "epoch": 12.957142857142857,
+      "grad_norm": 0.5740879901947831,
+      "learning_rate": 1.165727999502143e-05,
+      "loss": 0.046,
+      "step": 3628
+    },
+    {
+      "epoch": 12.960714285714285,
+      "grad_norm": 0.9600345336210674,
+      "learning_rate": 1.1646768806644745e-05,
+      "loss": 0.0591,
+      "step": 3629
+    },
+    {
+      "epoch": 12.964285714285714,
+      "grad_norm": 0.5358550418520266,
+      "learning_rate": 1.1636260412321518e-05,
+      "loss": 0.039,
+      "step": 3630
+    },
+    {
+      "epoch": 12.967857142857143,
+      "grad_norm": 1.0090491162426785,
+      "learning_rate": 1.1625754815566697e-05,
+      "loss": 0.0959,
+      "step": 3631
+    },
+    {
+      "epoch": 12.971428571428572,
+      "grad_norm": 0.911283084563295,
+      "learning_rate": 1.1615252019894253e-05,
+      "loss": 0.0292,
+      "step": 3632
+    },
+    {
+      "epoch": 12.975,
+      "grad_norm": 0.608472965147132,
+      "learning_rate": 1.1604752028817263e-05,
+      "loss": 0.041,
+      "step": 3633
+    },
+    {
+      "epoch": 12.978571428571428,
+      "grad_norm": 0.8088224609528538,
+      "learning_rate": 1.1594254845847827e-05,
+      "loss": 0.0372,
+      "step": 3634
+    },
+    {
+      "epoch": 12.982142857142858,
+      "grad_norm": 0.657909317212124,
+      "learning_rate": 1.1583760474497143e-05,
+      "loss": 0.0313,
+      "step": 3635
+    },
+    {
+      "epoch": 12.985714285714286,
+      "grad_norm": 0.8180909129705112,
+      "learning_rate": 1.157326891827543e-05,
+      "loss": 0.0279,
+      "step": 3636
+    },
+    {
+      "epoch": 12.989285714285714,
+      "grad_norm": 0.48975042460552887,
+      "learning_rate": 1.1562780180692003e-05,
+      "loss": 0.0226,
+      "step": 3637
+    },
+    {
+      "epoch": 12.992857142857144,
+      "grad_norm": 0.8358726527839857,
+      "learning_rate": 1.1552294265255211e-05,
+      "loss": 0.0541,
+      "step": 3638
+    },
+    {
+      "epoch": 12.996428571428572,
+      "grad_norm": 0.687210731650881,
+      "learning_rate": 1.1541811175472448e-05,
+      "loss": 0.0559,
+      "step": 3639
+    },
+    {
+      "epoch": 13.0,
+      "grad_norm": 0.5874940643223119,
+      "learning_rate": 1.1531330914850204e-05,
+      "loss": 0.0475,
+      "step": 3640
+    },
+    {
+      "epoch": 13.003571428571428,
+      "grad_norm": 0.7503546273781969,
+      "learning_rate": 1.1520853486893977e-05,
+      "loss": 0.0652,
+      "step": 3641
+    },
+    {
+      "epoch": 13.007142857142858,
+      "grad_norm": 0.4195234504432884,
+      "learning_rate": 1.1510378895108354e-05,
+      "loss": 0.0263,
+      "step": 3642
+    },
+    {
+      "epoch": 13.010714285714286,
+      "grad_norm": 0.4619690486074972,
+      "learning_rate": 1.1499907142996945e-05,
+      "loss": 0.0179,
+      "step": 3643
+    },
+    {
+      "epoch": 13.014285714285714,
+      "grad_norm": 0.33884527212164817,
+      "learning_rate": 1.1489438234062433e-05,
+      "loss": 0.018,
+      "step": 3644
+    },
+    {
+      "epoch": 13.017857142857142,
+      "grad_norm": 0.99558924235675,
+      "learning_rate": 1.147897217180653e-05,
+      "loss": 0.0596,
+      "step": 3645
+    },
+    {
+      "epoch": 13.021428571428572,
+      "grad_norm": 0.8366484505698112,
+      "learning_rate": 1.1468508959730017e-05,
+      "loss": 0.0322,
+      "step": 3646
+    },
+    {
+      "epoch": 13.025,
+      "grad_norm": 0.47733333980601694,
+      "learning_rate": 1.1458048601332698e-05,
+      "loss": 0.0238,
+      "step": 3647
+    },
+    {
+      "epoch": 13.028571428571428,
+      "grad_norm": 0.8141557872878797,
+      "learning_rate": 1.1447591100113442e-05,
+      "loss": 0.0146,
+      "step": 3648
+    },
+    {
+      "epoch": 13.032142857142857,
+      "grad_norm": 0.6017859481238899,
+      "learning_rate": 1.143713645957016e-05,
+      "loss": 0.049,
+      "step": 3649
+    },
+    {
+      "epoch": 13.035714285714286,
+      "grad_norm": 1.20694036251531,
+      "learning_rate": 1.1426684683199785e-05,
+      "loss": 0.057,
+      "step": 3650
+    },
+    {
+      "epoch": 13.039285714285715,
+      "grad_norm": 0.9251036968756815,
+      "learning_rate": 1.1416235774498328e-05,
+      "loss": 0.0671,
+      "step": 3651
+    },
+    {
+      "epoch": 13.042857142857143,
+      "grad_norm": 0.9593699936627412,
+      "learning_rate": 1.1405789736960802e-05,
+      "loss": 0.079,
+      "step": 3652
+    },
+    {
+      "epoch": 13.04642857142857,
+      "grad_norm": 1.3443173476361934,
+      "learning_rate": 1.1395346574081296e-05,
+      "loss": 0.0727,
+      "step": 3653
+    },
+    {
+      "epoch": 13.05,
+      "grad_norm": 0.7442332454405785,
+      "learning_rate": 1.1384906289352902e-05,
+      "loss": 0.0542,
+      "step": 3654
+    },
+    {
+      "epoch": 13.053571428571429,
+      "grad_norm": 1.5400145297773937,
+      "learning_rate": 1.1374468886267786e-05,
+      "loss": 0.0368,
+      "step": 3655
+    },
+    {
+      "epoch": 13.057142857142857,
+      "grad_norm": 1.1457839050873289,
+      "learning_rate": 1.1364034368317123e-05,
+      "loss": 0.0645,
+      "step": 3656
+    },
+    {
+      "epoch": 13.060714285714285,
+      "grad_norm": 1.0632255001039892,
+      "learning_rate": 1.1353602738991123e-05,
+      "loss": 0.0403,
+      "step": 3657
+    },
+    {
+      "epoch": 13.064285714285715,
+      "grad_norm": 0.7044673189919591,
+      "learning_rate": 1.134317400177905e-05,
+      "loss": 0.0187,
+      "step": 3658
+    },
+    {
+      "epoch": 13.067857142857143,
+      "grad_norm": 0.5545817871098612,
+      "learning_rate": 1.1332748160169177e-05,
+      "loss": 0.0226,
+      "step": 3659
+    },
+    {
+      "epoch": 13.071428571428571,
+      "grad_norm": 0.8737882998445812,
+      "learning_rate": 1.132232521764884e-05,
+      "loss": 0.0589,
+      "step": 3660
+    },
+    {
+      "epoch": 13.075,
+      "grad_norm": 1.1164738077727734,
+      "learning_rate": 1.1311905177704363e-05,
+      "loss": 0.0661,
+      "step": 3661
+    },
+    {
+      "epoch": 13.07857142857143,
+      "grad_norm": 0.6733206685892643,
+      "learning_rate": 1.130148804382114e-05,
+      "loss": 0.0458,
+      "step": 3662
+    },
+    {
+      "epoch": 13.082142857142857,
+      "grad_norm": 1.6836995758963396,
+      "learning_rate": 1.1291073819483559e-05,
+      "loss": 0.0455,
+      "step": 3663
+    },
+    {
+      "epoch": 13.085714285714285,
+      "grad_norm": 1.4042195125062884,
+      "learning_rate": 1.1280662508175065e-05,
+      "loss": 0.0414,
+      "step": 3664
+    },
+    {
+      "epoch": 13.089285714285714,
+      "grad_norm": 0.7114766371261345,
+      "learning_rate": 1.1270254113378098e-05,
+      "loss": 0.0507,
+      "step": 3665
+    },
+    {
+      "epoch": 13.092857142857143,
+      "grad_norm": 0.6926748101566316,
+      "learning_rate": 1.1259848638574158e-05,
+      "loss": 0.053,
+      "step": 3666
+    },
+    {
+      "epoch": 13.096428571428572,
+      "grad_norm": 0.9848509517584919,
+      "learning_rate": 1.124944608724373e-05,
+      "loss": 0.0253,
+      "step": 3667
+    },
+    {
+      "epoch": 13.1,
+      "grad_norm": 0.983808579430212,
+      "learning_rate": 1.1239046462866354e-05,
+      "loss": 0.0292,
+      "step": 3668
+    },
+    {
+      "epoch": 13.103571428571428,
+      "grad_norm": 1.025007589414967,
+      "learning_rate": 1.1228649768920572e-05,
+      "loss": 0.0461,
+      "step": 3669
+    },
+    {
+      "epoch": 13.107142857142858,
+      "grad_norm": 0.5637750994578558,
+      "learning_rate": 1.1218256008883943e-05,
+      "loss": 0.0321,
+      "step": 3670
+    },
+    {
+      "epoch": 13.110714285714286,
+      "grad_norm": 1.4132011889052405,
+      "learning_rate": 1.1207865186233067e-05,
+      "loss": 0.0659,
+      "step": 3671
+    },
+    {
+      "epoch": 13.114285714285714,
+      "grad_norm": 0.7329293157359233,
+      "learning_rate": 1.1197477304443529e-05,
+      "loss": 0.0423,
+      "step": 3672
+    },
+    {
+      "epoch": 13.117857142857142,
+      "grad_norm": 1.241939811305907,
+      "learning_rate": 1.1187092366989969e-05,
+      "loss": 0.0327,
+      "step": 3673
+    },
+    {
+      "epoch": 13.121428571428572,
+      "grad_norm": 2.2082853338988184,
+      "learning_rate": 1.1176710377346004e-05,
+      "loss": 0.067,
+      "step": 3674
+    },
+    {
+      "epoch": 13.125,
+      "grad_norm": 0.9507271265531219,
+      "learning_rate": 1.1166331338984293e-05,
+      "loss": 0.0381,
+      "step": 3675
+    },
+    {
+      "epoch": 13.128571428571428,
+      "grad_norm": 1.6457387654582956,
+      "learning_rate": 1.1155955255376485e-05,
+      "loss": 0.0874,
+      "step": 3676
+    },
+    {
+      "epoch": 13.132142857142858,
+      "grad_norm": 0.8182666656239269,
+      "learning_rate": 1.1145582129993274e-05,
+      "loss": 0.0176,
+      "step": 3677
+    },
+    {
+      "epoch": 13.135714285714286,
+      "grad_norm": 1.0973422337458187,
+      "learning_rate": 1.1135211966304315e-05,
+      "loss": 0.0438,
+      "step": 3678
+    },
+    {
+      "epoch": 13.139285714285714,
+      "grad_norm": 0.8729669361778867,
+      "learning_rate": 1.1124844767778328e-05,
+      "loss": 0.0341,
+      "step": 3679
+    },
+    {
+      "epoch": 13.142857142857142,
+      "grad_norm": 0.807073876918893,
+      "learning_rate": 1.1114480537882995e-05,
+      "loss": 0.0282,
+      "step": 3680
+    },
+    {
+      "epoch": 13.146428571428572,
+      "grad_norm": 0.5749622086763535,
+      "learning_rate": 1.1104119280085035e-05,
+      "loss": 0.042,
+      "step": 3681
+    },
+    {
+      "epoch": 13.15,
+      "grad_norm": 1.3156258669358225,
+      "learning_rate": 1.1093760997850158e-05,
+      "loss": 0.0624,
+      "step": 3682
+    },
+    {
+      "epoch": 13.153571428571428,
+      "grad_norm": 0.705168776103595,
+      "learning_rate": 1.1083405694643075e-05,
+      "loss": 0.0403,
+      "step": 3683
+    },
+    {
+      "epoch": 13.157142857142857,
+      "grad_norm": 1.1535017890410022,
+      "learning_rate": 1.1073053373927523e-05,
+      "loss": 0.0454,
+      "step": 3684
+    },
+    {
+      "epoch": 13.160714285714286,
+      "grad_norm": 0.489740932661799,
+      "learning_rate": 1.106270403916621e-05,
+      "loss": 0.0262,
+      "step": 3685
+    },
+    {
+      "epoch": 13.164285714285715,
+      "grad_norm": 1.3715053927710572,
+      "learning_rate": 1.1052357693820874e-05,
+      "loss": 0.053,
+      "step": 3686
+    },
+    {
+      "epoch": 13.167857142857143,
+      "grad_norm": 0.8481772456423593,
+      "learning_rate": 1.1042014341352233e-05,
+      "loss": 0.0409,
+      "step": 3687
+    },
+    {
+      "epoch": 13.17142857142857,
+      "grad_norm": 0.4965480986960627,
+      "learning_rate": 1.1031673985220022e-05,
+      "loss": 0.0168,
+      "step": 3688
+    },
+    {
+      "epoch": 13.175,
+      "grad_norm": 1.249698328518641,
+      "learning_rate": 1.1021336628882948e-05,
+      "loss": 0.0593,
+      "step": 3689
+    },
+    {
+      "epoch": 13.178571428571429,
+      "grad_norm": 1.0833094222459125,
+      "learning_rate": 1.1011002275798748e-05,
+      "loss": 0.0576,
+      "step": 3690
+    },
+    {
+      "epoch": 13.182142857142857,
+      "grad_norm": 1.0499255943120032,
+      "learning_rate": 1.1000670929424117e-05,
+      "loss": 0.0304,
+      "step": 3691
+    },
+    {
+      "epoch": 13.185714285714285,
+      "grad_norm": 1.0312421696358396,
+      "learning_rate": 1.0990342593214787e-05,
+      "loss": 0.0792,
+      "step": 3692
+    },
+    {
+      "epoch": 13.189285714285715,
+      "grad_norm": 1.1650773390697875,
+      "learning_rate": 1.098001727062544e-05,
+      "loss": 0.0455,
+      "step": 3693
+    },
+    {
+      "epoch": 13.192857142857143,
+      "grad_norm": 0.9263648670052056,
+      "learning_rate": 1.0969694965109773e-05,
+      "loss": 0.0509,
+      "step": 3694
+    },
+    {
+      "epoch": 13.196428571428571,
+      "grad_norm": 0.8673349743209109,
+      "learning_rate": 1.095937568012048e-05,
+      "loss": 0.0718,
+      "step": 3695
+    },
+    {
+      "epoch": 13.2,
+      "grad_norm": 1.161815276743186,
+      "learning_rate": 1.0949059419109225e-05,
+      "loss": 0.0444,
+      "step": 3696
+    },
+    {
+      "epoch": 13.20357142857143,
+      "grad_norm": 0.9619351613446598,
+      "learning_rate": 1.0938746185526678e-05,
+      "loss": 0.0398,
+      "step": 3697
+    },
+    {
+      "epoch": 13.207142857142857,
+      "grad_norm": 0.3890941112046188,
+      "learning_rate": 1.0928435982822482e-05,
+      "loss": 0.0169,
+      "step": 3698
+    },
+    {
+      "epoch": 13.210714285714285,
+      "grad_norm": 0.6329864006193454,
+      "learning_rate": 1.0918128814445285e-05,
+      "loss": 0.0521,
+      "step": 3699
+    },
+    {
+      "epoch": 13.214285714285714,
+      "grad_norm": 1.819823905005964,
+      "learning_rate": 1.0907824683842697e-05,
+      "loss": 0.0322,
+      "step": 3700
+    },
+    {
+      "epoch": 13.217857142857143,
+      "grad_norm": 1.4536755663088818,
+      "learning_rate": 1.0897523594461334e-05,
+      "loss": 0.053,
+      "step": 3701
+    },
+    {
+      "epoch": 13.221428571428572,
+      "grad_norm": 0.5482966280161263,
+      "learning_rate": 1.0887225549746771e-05,
+      "loss": 0.0229,
+      "step": 3702
+    },
+    {
+      "epoch": 13.225,
+      "grad_norm": 0.7304609984863781,
+      "learning_rate": 1.0876930553143592e-05,
+      "loss": 0.0279,
+      "step": 3703
+    },
+    {
+      "epoch": 13.228571428571428,
+      "grad_norm": 0.940413374807461,
+      "learning_rate": 1.0866638608095332e-05,
+      "loss": 0.0306,
+      "step": 3704
+    },
+    {
+      "epoch": 13.232142857142858,
+      "grad_norm": 1.373930503726066,
+      "learning_rate": 1.0856349718044536e-05,
+      "loss": 0.0305,
+      "step": 3705
+    },
+    {
+      "epoch": 13.235714285714286,
+      "grad_norm": 2.24684649728429,
+      "learning_rate": 1.0846063886432707e-05,
+      "loss": 0.0826,
+      "step": 3706
+    },
+    {
+      "epoch": 13.239285714285714,
+      "grad_norm": 1.1011405008733492,
+      "learning_rate": 1.0835781116700317e-05,
+      "loss": 0.0531,
+      "step": 3707
+    },
+    {
+      "epoch": 13.242857142857142,
+      "grad_norm": 1.2882344074123355,
+      "learning_rate": 1.0825501412286847e-05,
+      "loss": 0.0362,
+      "step": 3708
+    },
+    {
+      "epoch": 13.246428571428572,
+      "grad_norm": 1.40489918688851,
+      "learning_rate": 1.0815224776630711e-05,
+      "loss": 0.0456,
+      "step": 3709
+    },
+    {
+      "epoch": 13.25,
+      "grad_norm": 1.5580142086204296,
+      "learning_rate": 1.080495121316934e-05,
+      "loss": 0.0968,
+      "step": 3710
+    },
+    {
+      "epoch": 13.253571428571428,
+      "grad_norm": 0.8332217683530528,
+      "learning_rate": 1.079468072533909e-05,
+      "loss": 0.0482,
+      "step": 3711
+    },
+    {
+      "epoch": 13.257142857142856,
+      "grad_norm": 1.085224822436533,
+      "learning_rate": 1.0784413316575337e-05,
+      "loss": 0.0417,
+      "step": 3712
+    },
+    {
+      "epoch": 13.260714285714286,
+      "grad_norm": 0.7066931480909956,
+      "learning_rate": 1.0774148990312386e-05,
+      "loss": 0.0234,
+      "step": 3713
+    },
+    {
+      "epoch": 13.264285714285714,
+      "grad_norm": 1.177097641012233,
+      "learning_rate": 1.076388774998354e-05,
+      "loss": 0.0483,
+      "step": 3714
+    },
+    {
+      "epoch": 13.267857142857142,
+      "grad_norm": 0.5954018118560492,
+      "learning_rate": 1.0753629599021052e-05,
+      "loss": 0.0239,
+      "step": 3715
+    },
+    {
+      "epoch": 13.271428571428572,
+      "grad_norm": 0.624005559674872,
+      "learning_rate": 1.0743374540856153e-05,
+      "loss": 0.0422,
+      "step": 3716
+    },
+    {
+      "epoch": 13.275,
+      "grad_norm": 0.7961484974053569,
+      "learning_rate": 1.0733122578919024e-05,
+      "loss": 0.0583,
+      "step": 3717
+    },
+    {
+      "epoch": 13.278571428571428,
+      "grad_norm": 0.8426788923407763,
+      "learning_rate": 1.0722873716638835e-05,
+      "loss": 0.0523,
+      "step": 3718
+    },
+    {
+      "epoch": 13.282142857142857,
+      "grad_norm": 0.6398574849678915,
+      "learning_rate": 1.0712627957443696e-05,
+      "loss": 0.0268,
+      "step": 3719
+    },
+    {
+      "epoch": 13.285714285714286,
+      "grad_norm": 0.7323269367676272,
+      "learning_rate": 1.0702385304760682e-05,
+      "loss": 0.0318,
+      "step": 3720
+    },
+    {
+      "epoch": 13.289285714285715,
+      "grad_norm": 1.0293144607082207,
+      "learning_rate": 1.0692145762015849e-05,
+      "loss": 0.0192,
+      "step": 3721
+    },
+    {
+      "epoch": 13.292857142857143,
+      "grad_norm": 0.7238436947985429,
+      "learning_rate": 1.0681909332634187e-05,
+      "loss": 0.0159,
+      "step": 3722
+    },
+    {
+      "epoch": 13.29642857142857,
+      "grad_norm": 0.7603954967039263,
+      "learning_rate": 1.0671676020039664e-05,
+      "loss": 0.0465,
+      "step": 3723
+    },
+    {
+      "epoch": 13.3,
+      "grad_norm": 1.1823595033618641,
+      "learning_rate": 1.0661445827655187e-05,
+      "loss": 0.0657,
+      "step": 3724
+    },
+    {
+      "epoch": 13.303571428571429,
+      "grad_norm": 1.643290888116586,
+      "learning_rate": 1.0651218758902649e-05,
+      "loss": 0.07,
+      "step": 3725
+    },
+    {
+      "epoch": 13.307142857142857,
+      "grad_norm": 1.9718305304284307,
+      "learning_rate": 1.0640994817202855e-05,
+      "loss": 0.0717,
+      "step": 3726
+    },
+    {
+      "epoch": 13.310714285714285,
+      "grad_norm": 1.0014307517540522,
+      "learning_rate": 1.0630774005975608e-05,
+      "loss": 0.0571,
+      "step": 3727
+    },
+    {
+      "epoch": 13.314285714285715,
+      "grad_norm": 0.4559838672479564,
+      "learning_rate": 1.0620556328639631e-05,
+      "loss": 0.0112,
+      "step": 3728
+    },
+    {
+      "epoch": 13.317857142857143,
+      "grad_norm": 0.9657547654164231,
+      "learning_rate": 1.0610341788612625e-05,
+      "loss": 0.0376,
+      "step": 3729
+    },
+    {
+      "epoch": 13.321428571428571,
+      "grad_norm": 0.9121986931008419,
+      "learning_rate": 1.060013038931122e-05,
+      "loss": 0.0279,
+      "step": 3730
+    },
+    {
+      "epoch": 13.325,
+      "grad_norm": 1.639789952841768,
+      "learning_rate": 1.0589922134151e-05,
+      "loss": 0.0435,
+      "step": 3731
+    },
+    {
+      "epoch": 13.32857142857143,
+      "grad_norm": 1.1622558122822464,
+      "learning_rate": 1.0579717026546512e-05,
+      "loss": 0.0586,
+      "step": 3732
+    },
+    {
+      "epoch": 13.332142857142857,
+      "grad_norm": 1.2313752417250612,
+      "learning_rate": 1.0569515069911227e-05,
+      "loss": 0.0544,
+      "step": 3733
+    },
+    {
+      "epoch": 13.335714285714285,
+      "grad_norm": 1.1895554824221357,
+      "learning_rate": 1.0559316267657586e-05,
+      "loss": 0.0432,
+      "step": 3734
+    },
+    {
+      "epoch": 13.339285714285714,
+      "grad_norm": 1.3891933242780279,
+      "learning_rate": 1.0549120623196954e-05,
+      "loss": 0.0547,
+      "step": 3735
+    },
+    {
+      "epoch": 13.342857142857143,
+      "grad_norm": 0.8433626871841273,
+      "learning_rate": 1.053892813993966e-05,
+      "loss": 0.0348,
+      "step": 3736
+    },
+    {
+      "epoch": 13.346428571428572,
+      "grad_norm": 1.0135859012722999,
+      "learning_rate": 1.0528738821294949e-05,
+      "loss": 0.0208,
+      "step": 3737
+    },
+    {
+      "epoch": 13.35,
+      "grad_norm": 1.265547143269222,
+      "learning_rate": 1.0518552670671043e-05,
+      "loss": 0.0776,
+      "step": 3738
+    },
+    {
+      "epoch": 13.353571428571428,
+      "grad_norm": 1.1257697515765783,
+      "learning_rate": 1.0508369691475065e-05,
+      "loss": 0.0487,
+      "step": 3739
+    },
+    {
+      "epoch": 13.357142857142858,
+      "grad_norm": 1.2253838039979923,
+      "learning_rate": 1.0498189887113112e-05,
+      "loss": 0.0465,
+      "step": 3740
+    },
+    {
+      "epoch": 13.360714285714286,
+      "grad_norm": 1.3677429569928243,
+      "learning_rate": 1.0488013260990191e-05,
+      "loss": 0.0419,
+      "step": 3741
+    },
+    {
+      "epoch": 13.364285714285714,
+      "grad_norm": 1.4377287810261934,
+      "learning_rate": 1.0477839816510274e-05,
+      "loss": 0.1078,
+      "step": 3742
+    },
+    {
+      "epoch": 13.367857142857144,
+      "grad_norm": 1.2721516031190792,
+      "learning_rate": 1.0467669557076246e-05,
+      "loss": 0.0802,
+      "step": 3743
+    },
+    {
+      "epoch": 13.371428571428572,
+      "grad_norm": 0.7563748734959916,
+      "learning_rate": 1.0457502486089921e-05,
+      "loss": 0.0375,
+      "step": 3744
+    },
+    {
+      "epoch": 13.375,
+      "grad_norm": 1.2025790898175548,
+      "learning_rate": 1.0447338606952082e-05,
+      "loss": 0.0478,
+      "step": 3745
+    },
+    {
+      "epoch": 13.378571428571428,
+      "grad_norm": 1.33695553998437,
+      "learning_rate": 1.0437177923062402e-05,
+      "loss": 0.0497,
+      "step": 3746
+    },
+    {
+      "epoch": 13.382142857142856,
+      "grad_norm": 1.039478488744914,
+      "learning_rate": 1.0427020437819526e-05,
+      "loss": 0.0471,
+      "step": 3747
+    },
+    {
+      "epoch": 13.385714285714286,
+      "grad_norm": 1.6222969302514858,
+      "learning_rate": 1.0416866154620988e-05,
+      "loss": 0.0717,
+      "step": 3748
+    },
+    {
+      "epoch": 13.389285714285714,
+      "grad_norm": 0.7532153357331571,
+      "learning_rate": 1.0406715076863287e-05,
+      "loss": 0.0358,
+      "step": 3749
+    },
+    {
+      "epoch": 13.392857142857142,
+      "grad_norm": 1.1937449118511376,
+      "learning_rate": 1.0396567207941821e-05,
+      "loss": 0.0515,
+      "step": 3750
+    },
+    {
+      "epoch": 13.396428571428572,
+      "grad_norm": 1.3346955521582962,
+      "learning_rate": 1.0386422551250942e-05,
+      "loss": 0.0369,
+      "step": 3751
+    },
+    {
+      "epoch": 13.4,
+      "grad_norm": 0.8840889888103255,
+      "learning_rate": 1.03762811101839e-05,
+      "loss": 0.0456,
+      "step": 3752
+    },
+    {
+      "epoch": 13.403571428571428,
+      "grad_norm": 1.397734302939845,
+      "learning_rate": 1.0366142888132892e-05,
+      "loss": 0.07,
+      "step": 3753
+    },
+    {
+      "epoch": 13.407142857142857,
+      "grad_norm": 0.978252689631685,
+      "learning_rate": 1.0356007888489021e-05,
+      "loss": 0.0411,
+      "step": 3754
+    },
+    {
+      "epoch": 13.410714285714286,
+      "grad_norm": 0.9227731950223286,
+      "learning_rate": 1.0345876114642333e-05,
+      "loss": 0.0686,
+      "step": 3755
+    },
+    {
+      "epoch": 13.414285714285715,
+      "grad_norm": 0.7822542919346913,
+      "learning_rate": 1.0335747569981774e-05,
+      "loss": 0.0159,
+      "step": 3756
+    },
+    {
+      "epoch": 13.417857142857143,
+      "grad_norm": 1.0129006897402006,
+      "learning_rate": 1.0325622257895205e-05,
+      "loss": 0.052,
+      "step": 3757
+    },
+    {
+      "epoch": 13.42142857142857,
+      "grad_norm": 0.5495394469023159,
+      "learning_rate": 1.0315500181769445e-05,
+      "loss": 0.0279,
+      "step": 3758
+    },
+    {
+      "epoch": 13.425,
+      "grad_norm": 1.0607865513443153,
+      "learning_rate": 1.0305381344990181e-05,
+      "loss": 0.0333,
+      "step": 3759
+    },
+    {
+      "epoch": 13.428571428571429,
+      "grad_norm": 1.0474685520451585,
+      "learning_rate": 1.0295265750942059e-05,
+      "loss": 0.0582,
+      "step": 3760
+    },
+    {
+      "epoch": 13.432142857142857,
+      "grad_norm": 0.5706176792495276,
+      "learning_rate": 1.0285153403008606e-05,
+      "loss": 0.0309,
+      "step": 3761
+    },
+    {
+      "epoch": 13.435714285714285,
+      "grad_norm": 0.7589103606841239,
+      "learning_rate": 1.027504430457229e-05,
+      "loss": 0.0387,
+      "step": 3762
+    },
+    {
+      "epoch": 13.439285714285715,
+      "grad_norm": 1.2743466637134275,
+      "learning_rate": 1.0264938459014467e-05,
+      "loss": 0.0528,
+      "step": 3763
+    },
+    {
+      "epoch": 13.442857142857143,
+      "grad_norm": 0.7894928189318487,
+      "learning_rate": 1.0254835869715435e-05,
+      "loss": 0.0505,
+      "step": 3764
+    },
+    {
+      "epoch": 13.446428571428571,
+      "grad_norm": 0.8566576914988726,
+      "learning_rate": 1.0244736540054372e-05,
+      "loss": 0.0345,
+      "step": 3765
+    },
+    {
+      "epoch": 13.45,
+      "grad_norm": 0.6938299558356255,
+      "learning_rate": 1.0234640473409394e-05,
+      "loss": 0.027,
+      "step": 3766
+    },
+    {
+      "epoch": 13.45357142857143,
+      "grad_norm": 0.8328489365762632,
+      "learning_rate": 1.0224547673157504e-05,
+      "loss": 0.0289,
+      "step": 3767
+    },
+    {
+      "epoch": 13.457142857142857,
+      "grad_norm": 1.0444732980763003,
+      "learning_rate": 1.0214458142674611e-05,
+      "loss": 0.0412,
+      "step": 3768
+    },
+    {
+      "epoch": 13.460714285714285,
+      "grad_norm": 0.6445872743507707,
+      "learning_rate": 1.0204371885335553e-05,
+      "loss": 0.0361,
+      "step": 3769
+    },
+    {
+      "epoch": 13.464285714285714,
+      "grad_norm": 1.7352795115335213,
+      "learning_rate": 1.0194288904514052e-05,
+      "loss": 0.0558,
+      "step": 3770
+    },
+    {
+      "epoch": 13.467857142857143,
+      "grad_norm": 1.3556745520490803,
+      "learning_rate": 1.0184209203582747e-05,
+      "loss": 0.0596,
+      "step": 3771
+    },
+    {
+      "epoch": 13.471428571428572,
+      "grad_norm": 0.5771143567060644,
+      "learning_rate": 1.0174132785913166e-05,
+      "loss": 0.0269,
+      "step": 3772
+    },
+    {
+      "epoch": 13.475,
+      "grad_norm": 0.8116652398834817,
+      "learning_rate": 1.016405965487576e-05,
+      "loss": 0.0475,
+      "step": 3773
+    },
+    {
+      "epoch": 13.478571428571428,
+      "grad_norm": 0.9573363724277762,
+      "learning_rate": 1.0153989813839853e-05,
+      "loss": 0.0319,
+      "step": 3774
+    },
+    {
+      "epoch": 13.482142857142858,
+      "grad_norm": 0.5657934690318372,
+      "learning_rate": 1.01439232661737e-05,
+      "loss": 0.0301,
+      "step": 3775
+    },
+    {
+      "epoch": 13.485714285714286,
+      "grad_norm": 0.6067566903463788,
+      "learning_rate": 1.0133860015244418e-05,
+      "loss": 0.0234,
+      "step": 3776
+    },
+    {
+      "epoch": 13.489285714285714,
+      "grad_norm": 1.2548114353690585,
+      "learning_rate": 1.0123800064418061e-05,
+      "loss": 0.031,
+      "step": 3777
+    },
+    {
+      "epoch": 13.492857142857144,
+      "grad_norm": 0.8968542926621386,
+      "learning_rate": 1.0113743417059544e-05,
+      "loss": 0.0327,
+      "step": 3778
+    },
+    {
+      "epoch": 13.496428571428572,
+      "grad_norm": 0.9733563831431039,
+      "learning_rate": 1.0103690076532704e-05,
+      "loss": 0.039,
+      "step": 3779
+    },
+    {
+      "epoch": 13.5,
+      "grad_norm": 0.4601473456187191,
+      "learning_rate": 1.0093640046200257e-05,
+      "loss": 0.0187,
+      "step": 3780
+    },
+    {
+      "epoch": 13.503571428571428,
+      "grad_norm": 0.7888524797042894,
+      "learning_rate": 1.0083593329423804e-05,
+      "loss": 0.0243,
+      "step": 3781
+    },
+    {
+      "epoch": 13.507142857142856,
+      "grad_norm": 0.8331038532452698,
+      "learning_rate": 1.0073549929563863e-05,
+      "loss": 0.0297,
+      "step": 3782
+    },
+    {
+      "epoch": 13.510714285714286,
+      "grad_norm": 1.0886347864805113,
+      "learning_rate": 1.0063509849979818e-05,
+      "loss": 0.0519,
+      "step": 3783
+    },
+    {
+      "epoch": 13.514285714285714,
+      "grad_norm": 1.1608106370073565,
+      "learning_rate": 1.0053473094029962e-05,
+      "loss": 0.052,
+      "step": 3784
+    },
+    {
+      "epoch": 13.517857142857142,
+      "grad_norm": 0.947084045458095,
+      "learning_rate": 1.0043439665071455e-05,
+      "loss": 0.0599,
+      "step": 3785
+    },
+    {
+      "epoch": 13.521428571428572,
+      "grad_norm": 0.8338375959056858,
+      "learning_rate": 1.0033409566460372e-05,
+      "loss": 0.0433,
+      "step": 3786
+    },
+    {
+      "epoch": 13.525,
+      "grad_norm": 1.7783886230206816,
+      "learning_rate": 1.002338280155164e-05,
+      "loss": 0.1015,
+      "step": 3787
+    },
+    {
+      "epoch": 13.528571428571428,
+      "grad_norm": 1.3946029055475366,
+      "learning_rate": 1.0013359373699108e-05,
+      "loss": 0.0488,
+      "step": 3788
+    },
+    {
+      "epoch": 13.532142857142857,
+      "grad_norm": 1.378412300068479,
+      "learning_rate": 1.0003339286255469e-05,
+      "loss": 0.0689,
+      "step": 3789
+    },
+    {
+      "epoch": 13.535714285714286,
+      "grad_norm": 0.757798483100772,
+      "learning_rate": 9.99332254257234e-06,
+      "loss": 0.0421,
+      "step": 3790
+    },
+    {
+      "epoch": 13.539285714285715,
+      "grad_norm": 1.0135541486457504,
+      "learning_rate": 9.98330914600018e-06,
+      "loss": 0.0327,
+      "step": 3791
+    },
+    {
+      "epoch": 13.542857142857143,
+      "grad_norm": 0.9107881588717984,
+      "learning_rate": 9.973299099888365e-06,
+      "loss": 0.0569,
+      "step": 3792
+    },
+    {
+      "epoch": 13.54642857142857,
+      "grad_norm": 0.8857380586694045,
+      "learning_rate": 9.963292407585124e-06,
+      "loss": 0.0435,
+      "step": 3793
+    },
+    {
+      "epoch": 13.55,
+      "grad_norm": 1.5455244581007563,
+      "learning_rate": 9.953289072437566e-06,
+      "loss": 0.0661,
+      "step": 3794
+    },
+    {
+      "epoch": 13.553571428571429,
+      "grad_norm": 1.208599295833445,
+      "learning_rate": 9.9432890977917e-06,
+      "loss": 0.0269,
+      "step": 3795
+    },
+    {
+      "epoch": 13.557142857142857,
+      "grad_norm": 1.2601513691572324,
+      "learning_rate": 9.933292486992376e-06,
+      "loss": 0.0276,
+      "step": 3796
+    },
+    {
+      "epoch": 13.560714285714285,
+      "grad_norm": 1.006069809696858,
+      "learning_rate": 9.92329924338336e-06,
+      "loss": 0.0482,
+      "step": 3797
+    },
+    {
+      "epoch": 13.564285714285715,
+      "grad_norm": 1.392773741897402,
+      "learning_rate": 9.913309370307246e-06,
+      "loss": 0.0422,
+      "step": 3798
+    },
+    {
+      "epoch": 13.567857142857143,
+      "grad_norm": 1.0977107121279015,
+      "learning_rate": 9.903322871105545e-06,
+      "loss": 0.0428,
+      "step": 3799
+    },
+    {
+      "epoch": 13.571428571428571,
+      "grad_norm": 1.3499500321566154,
+      "learning_rate": 9.893339749118602e-06,
+      "loss": 0.0604,
+      "step": 3800
+    },
+    {
+      "epoch": 13.575,
+      "grad_norm": 0.8233775016194604,
+      "learning_rate": 9.883360007685665e-06,
+      "loss": 0.0442,
+      "step": 3801
+    },
+    {
+      "epoch": 13.57857142857143,
+      "grad_norm": 1.1119286637978294,
+      "learning_rate": 9.873383650144819e-06,
+      "loss": 0.0619,
+      "step": 3802
+    },
+    {
+      "epoch": 13.582142857142857,
+      "grad_norm": 1.262478725587829,
+      "learning_rate": 9.863410679833045e-06,
+      "loss": 0.0878,
+      "step": 3803
+    },
+    {
+      "epoch": 13.585714285714285,
+      "grad_norm": 0.7698576021559611,
+      "learning_rate": 9.853441100086179e-06,
+      "loss": 0.0433,
+      "step": 3804
+    },
+    {
+      "epoch": 13.589285714285714,
+      "grad_norm": 1.0015986399455257,
+      "learning_rate": 9.84347491423891e-06,
+      "loss": 0.0277,
+      "step": 3805
+    },
+    {
+      "epoch": 13.592857142857143,
+      "grad_norm": 1.1556171811149036,
+      "learning_rate": 9.833512125624825e-06,
+      "loss": 0.0502,
+      "step": 3806
+    },
+    {
+      "epoch": 13.596428571428572,
+      "grad_norm": 1.4374554414360814,
+      "learning_rate": 9.823552737576333e-06,
+      "loss": 0.0481,
+      "step": 3807
+    },
+    {
+      "epoch": 13.6,
+      "grad_norm": 1.6273239060706093,
+      "learning_rate": 9.813596753424747e-06,
+      "loss": 0.0656,
+      "step": 3808
+    },
+    {
+      "epoch": 13.603571428571428,
+      "grad_norm": 1.2456068686989963,
+      "learning_rate": 9.803644176500207e-06,
+      "loss": 0.0547,
+      "step": 3809
+    },
+    {
+      "epoch": 13.607142857142858,
+      "grad_norm": 1.3122455123670869,
+      "learning_rate": 9.793695010131741e-06,
+      "loss": 0.0606,
+      "step": 3810
+    },
+    {
+      "epoch": 13.610714285714286,
+      "grad_norm": 0.9374239918076646,
+      "learning_rate": 9.783749257647209e-06,
+      "loss": 0.0448,
+      "step": 3811
+    },
+    {
+      "epoch": 13.614285714285714,
+      "grad_norm": 0.9493077858426636,
+      "learning_rate": 9.773806922373358e-06,
+      "loss": 0.0613,
+      "step": 3812
+    },
+    {
+      "epoch": 13.617857142857144,
+      "grad_norm": 0.874487007266945,
+      "learning_rate": 9.763868007635762e-06,
+      "loss": 0.0446,
+      "step": 3813
+    },
+    {
+      "epoch": 13.621428571428572,
+      "grad_norm": 0.8578388098141476,
+      "learning_rate": 9.753932516758881e-06,
+      "loss": 0.0486,
+      "step": 3814
+    },
+    {
+      "epoch": 13.625,
+      "grad_norm": 0.7720607290208743,
+      "learning_rate": 9.744000453066003e-06,
+      "loss": 0.058,
+      "step": 3815
+    },
+    {
+      "epoch": 13.628571428571428,
+      "grad_norm": 0.9449129876124508,
+      "learning_rate": 9.734071819879293e-06,
+      "loss": 0.0462,
+      "step": 3816
+    },
+    {
+      "epoch": 13.632142857142856,
+      "grad_norm": 0.7023281974966408,
+      "learning_rate": 9.724146620519752e-06,
+      "loss": 0.0468,
+      "step": 3817
+    },
+    {
+      "epoch": 13.635714285714286,
+      "grad_norm": 0.8134360604575944,
+      "learning_rate": 9.71422485830722e-06,
+      "loss": 0.06,
+      "step": 3818
+    },
+    {
+      "epoch": 13.639285714285714,
+      "grad_norm": 1.0209146712997346,
+      "learning_rate": 9.704306536560436e-06,
+      "loss": 0.0642,
+      "step": 3819
+    },
+    {
+      "epoch": 13.642857142857142,
+      "grad_norm": 1.122200140194372,
+      "learning_rate": 9.694391658596936e-06,
+      "loss": 0.0728,
+      "step": 3820
+    },
+    {
+      "epoch": 13.646428571428572,
+      "grad_norm": 1.8691979246141126,
+      "learning_rate": 9.68448022773314e-06,
+      "loss": 0.0534,
+      "step": 3821
+    },
+    {
+      "epoch": 13.65,
+      "grad_norm": 0.585796353207801,
+      "learning_rate": 9.674572247284282e-06,
+      "loss": 0.0293,
+      "step": 3822
+    },
+    {
+      "epoch": 13.653571428571428,
+      "grad_norm": 0.8547034587576453,
+      "learning_rate": 9.664667720564478e-06,
+      "loss": 0.0338,
+      "step": 3823
+    },
+    {
+      "epoch": 13.657142857142857,
+      "grad_norm": 1.441616523241506,
+      "learning_rate": 9.654766650886662e-06,
+      "loss": 0.0574,
+      "step": 3824
+    },
+    {
+      "epoch": 13.660714285714286,
+      "grad_norm": 0.99816668644188,
+      "learning_rate": 9.644869041562614e-06,
+      "loss": 0.0395,
+      "step": 3825
+    },
+    {
+      "epoch": 13.664285714285715,
+      "grad_norm": 0.8396406154333108,
+      "learning_rate": 9.63497489590298e-06,
+      "loss": 0.0594,
+      "step": 3826
+    },
+    {
+      "epoch": 13.667857142857143,
+      "grad_norm": 1.0908485669768766,
+      "learning_rate": 9.625084217217208e-06,
+      "loss": 0.0473,
+      "step": 3827
+    },
+    {
+      "epoch": 13.67142857142857,
+      "grad_norm": 0.5581279384665438,
+      "learning_rate": 9.615197008813633e-06,
+      "loss": 0.0263,
+      "step": 3828
+    },
+    {
+      "epoch": 13.675,
+      "grad_norm": 0.9272850255943798,
+      "learning_rate": 9.605313273999387e-06,
+      "loss": 0.0534,
+      "step": 3829
+    },
+    {
+      "epoch": 13.678571428571429,
+      "grad_norm": 1.0798411147811289,
+      "learning_rate": 9.595433016080466e-06,
+      "loss": 0.0422,
+      "step": 3830
+    },
+    {
+      "epoch": 13.682142857142857,
+      "grad_norm": 1.1748761615510381,
+      "learning_rate": 9.58555623836169e-06,
+      "loss": 0.0579,
+      "step": 3831
+    },
+    {
+      "epoch": 13.685714285714285,
+      "grad_norm": 0.8945696198623708,
+      "learning_rate": 9.575682944146733e-06,
+      "loss": 0.041,
+      "step": 3832
+    },
+    {
+      "epoch": 13.689285714285715,
+      "grad_norm": 1.604213620328639,
+      "learning_rate": 9.565813136738072e-06,
+      "loss": 0.0555,
+      "step": 3833
+    },
+    {
+      "epoch": 13.692857142857143,
+      "grad_norm": 1.1915283457640067,
+      "learning_rate": 9.555946819437055e-06,
+      "loss": 0.0443,
+      "step": 3834
+    },
+    {
+      "epoch": 13.696428571428571,
+      "grad_norm": 1.850337985634166,
+      "learning_rate": 9.546083995543834e-06,
+      "loss": 0.0737,
+      "step": 3835
+    },
+    {
+      "epoch": 13.7,
+      "grad_norm": 0.4954159075649003,
+      "learning_rate": 9.5362246683574e-06,
+      "loss": 0.0245,
+      "step": 3836
+    },
+    {
+      "epoch": 13.70357142857143,
+      "grad_norm": 1.307494211681837,
+      "learning_rate": 9.526368841175584e-06,
+      "loss": 0.04,
+      "step": 3837
+    },
+    {
+      "epoch": 13.707142857142857,
+      "grad_norm": 1.1786062862584707,
+      "learning_rate": 9.516516517295033e-06,
+      "loss": 0.0546,
+      "step": 3838
+    },
+    {
+      "epoch": 13.710714285714285,
+      "grad_norm": 2.1013146428728753,
+      "learning_rate": 9.506667700011238e-06,
+      "loss": 0.0781,
+      "step": 3839
+    },
+    {
+      "epoch": 13.714285714285714,
+      "grad_norm": 0.9103691486556399,
+      "learning_rate": 9.496822392618496e-06,
+      "loss": 0.024,
+      "step": 3840
+    },
+    {
+      "epoch": 13.717857142857143,
+      "grad_norm": 0.6908026045436505,
+      "learning_rate": 9.486980598409956e-06,
+      "loss": 0.0344,
+      "step": 3841
+    },
+    {
+      "epoch": 13.721428571428572,
+      "grad_norm": 1.0779466032507785,
+      "learning_rate": 9.47714232067756e-06,
+      "loss": 0.0332,
+      "step": 3842
+    },
+    {
+      "epoch": 13.725,
+      "grad_norm": 0.16820397900793266,
+      "learning_rate": 9.46730756271211e-06,
+      "loss": 0.005,
+      "step": 3843
+    },
+    {
+      "epoch": 13.728571428571428,
+      "grad_norm": 0.9547414287959984,
+      "learning_rate": 9.457476327803194e-06,
+      "loss": 0.0499,
+      "step": 3844
+    },
+    {
+      "epoch": 13.732142857142858,
+      "grad_norm": 0.6785393085385902,
+      "learning_rate": 9.447648619239256e-06,
+      "loss": 0.0283,
+      "step": 3845
+    },
+    {
+      "epoch": 13.735714285714286,
+      "grad_norm": 0.8917406818136087,
+      "learning_rate": 9.43782444030753e-06,
+      "loss": 0.0606,
+      "step": 3846
+    },
+    {
+      "epoch": 13.739285714285714,
+      "grad_norm": 1.6722847186381695,
+      "learning_rate": 9.428003794294096e-06,
+      "loss": 0.0738,
+      "step": 3847
+    },
+    {
+      "epoch": 13.742857142857144,
+      "grad_norm": 0.6614238789957837,
+      "learning_rate": 9.418186684483834e-06,
+      "loss": 0.0314,
+      "step": 3848
+    },
+    {
+      "epoch": 13.746428571428572,
+      "grad_norm": 1.0867442852510931,
+      "learning_rate": 9.408373114160438e-06,
+      "loss": 0.0553,
+      "step": 3849
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 0.5284775548537107,
+      "learning_rate": 9.398563086606444e-06,
+      "loss": 0.0223,
+      "step": 3850
+    },
+    {
+      "epoch": 13.753571428571428,
+      "grad_norm": 0.6847643132753649,
+      "learning_rate": 9.38875660510317e-06,
+      "loss": 0.016,
+      "step": 3851
+    },
+    {
+      "epoch": 13.757142857142856,
+      "grad_norm": 0.8518744994074231,
+      "learning_rate": 9.378953672930779e-06,
+      "loss": 0.0601,
+      "step": 3852
+    },
+    {
+      "epoch": 13.760714285714286,
+      "grad_norm": 0.7299987776319737,
+      "learning_rate": 9.369154293368217e-06,
+      "loss": 0.0515,
+      "step": 3853
+    },
+    {
+      "epoch": 13.764285714285714,
+      "grad_norm": 1.4486240000538395,
+      "learning_rate": 9.359358469693272e-06,
+      "loss": 0.0281,
+      "step": 3854
+    },
+    {
+      "epoch": 13.767857142857142,
+      "grad_norm": 0.682401623536968,
+      "learning_rate": 9.349566205182508e-06,
+      "loss": 0.0359,
+      "step": 3855
+    },
+    {
+      "epoch": 13.771428571428572,
+      "grad_norm": 0.7034524648449169,
+      "learning_rate": 9.339777503111338e-06,
+      "loss": 0.0266,
+      "step": 3856
+    },
+    {
+      "epoch": 13.775,
+      "grad_norm": 1.4196624378649654,
+      "learning_rate": 9.329992366753943e-06,
+      "loss": 0.0495,
+      "step": 3857
+    },
+    {
+      "epoch": 13.778571428571428,
+      "grad_norm": 0.6087547887669549,
+      "learning_rate": 9.32021079938335e-06,
+      "loss": 0.0303,
+      "step": 3858
+    },
+    {
+      "epoch": 13.782142857142857,
+      "grad_norm": 0.860028833950986,
+      "learning_rate": 9.310432804271357e-06,
+      "loss": 0.0509,
+      "step": 3859
+    },
+    {
+      "epoch": 13.785714285714286,
+      "grad_norm": 0.7407089600997344,
+      "learning_rate": 9.300658384688598e-06,
+      "loss": 0.0458,
+      "step": 3860
+    },
+    {
+      "epoch": 13.789285714285715,
+      "grad_norm": 0.6284626650865216,
+      "learning_rate": 9.290887543904488e-06,
+      "loss": 0.0366,
+      "step": 3861
+    },
+    {
+      "epoch": 13.792857142857143,
+      "grad_norm": 1.1970809722954188,
+      "learning_rate": 9.281120285187248e-06,
+      "loss": 0.0483,
+      "step": 3862
+    },
+    {
+      "epoch": 13.79642857142857,
+      "grad_norm": 0.8995199712589678,
+      "learning_rate": 9.271356611803922e-06,
+      "loss": 0.0201,
+      "step": 3863
+    },
+    {
+      "epoch": 13.8,
+      "grad_norm": 1.4217084933542563,
+      "learning_rate": 9.261596527020324e-06,
+      "loss": 0.0373,
+      "step": 3864
+    },
+    {
+      "epoch": 13.803571428571429,
+      "grad_norm": 0.6807892166859242,
+      "learning_rate": 9.251840034101095e-06,
+      "loss": 0.0402,
+      "step": 3865
+    },
+    {
+      "epoch": 13.807142857142857,
+      "grad_norm": 1.726662861460695,
+      "learning_rate": 9.24208713630965e-06,
+      "loss": 0.0549,
+      "step": 3866
+    },
+    {
+      "epoch": 13.810714285714285,
+      "grad_norm": 1.066051659017834,
+      "learning_rate": 9.232337836908228e-06,
+      "loss": 0.0402,
+      "step": 3867
+    },
+    {
+      "epoch": 13.814285714285715,
+      "grad_norm": 1.342012972479746,
+      "learning_rate": 9.222592139157834e-06,
+      "loss": 0.0313,
+      "step": 3868
+    },
+    {
+      "epoch": 13.817857142857143,
+      "grad_norm": 0.7305107410621678,
+      "learning_rate": 9.212850046318299e-06,
+      "loss": 0.0275,
+      "step": 3869
+    },
+    {
+      "epoch": 13.821428571428571,
+      "grad_norm": 1.7630936721969321,
+      "learning_rate": 9.203111561648222e-06,
+      "loss": 0.0803,
+      "step": 3870
+    },
+    {
+      "epoch": 13.825,
+      "grad_norm": 1.2595485486366875,
+      "learning_rate": 9.193376688405018e-06,
+      "loss": 0.0259,
+      "step": 3871
+    },
+    {
+      "epoch": 13.82857142857143,
+      "grad_norm": 2.558502004339864,
+      "learning_rate": 9.183645429844881e-06,
+      "loss": 0.038,
+      "step": 3872
+    },
+    {
+      "epoch": 13.832142857142857,
+      "grad_norm": 0.863471288471859,
+      "learning_rate": 9.173917789222784e-06,
+      "loss": 0.0244,
+      "step": 3873
+    },
+    {
+      "epoch": 13.835714285714285,
+      "grad_norm": 0.9864931582250712,
+      "learning_rate": 9.16419376979252e-06,
+      "loss": 0.0391,
+      "step": 3874
+    },
+    {
+      "epoch": 13.839285714285714,
+      "grad_norm": 1.1783659721511404,
+      "learning_rate": 9.154473374806641e-06,
+      "loss": 0.0362,
+      "step": 3875
+    },
+    {
+      "epoch": 13.842857142857143,
+      "grad_norm": 0.9967696643110084,
+      "learning_rate": 9.144756607516515e-06,
+      "loss": 0.067,
+      "step": 3876
+    },
+    {
+      "epoch": 13.846428571428572,
+      "grad_norm": 1.0079899247818156,
+      "learning_rate": 9.135043471172264e-06,
+      "loss": 0.0339,
+      "step": 3877
+    },
+    {
+      "epoch": 13.85,
+      "grad_norm": 1.6270662883177325,
+      "learning_rate": 9.125333969022831e-06,
+      "loss": 0.0553,
+      "step": 3878
+    },
+    {
+      "epoch": 13.853571428571428,
+      "grad_norm": 0.6153375597503611,
+      "learning_rate": 9.115628104315912e-06,
+      "loss": 0.0208,
+      "step": 3879
+    },
+    {
+      "epoch": 13.857142857142858,
+      "grad_norm": 0.9759351917938506,
+      "learning_rate": 9.10592588029801e-06,
+      "loss": 0.0539,
+      "step": 3880
+    },
+    {
+      "epoch": 13.860714285714286,
+      "grad_norm": 0.8588346545648441,
+      "learning_rate": 9.096227300214387e-06,
+      "loss": 0.0513,
+      "step": 3881
+    },
+    {
+      "epoch": 13.864285714285714,
+      "grad_norm": 1.051913930437755,
+      "learning_rate": 9.086532367309115e-06,
+      "loss": 0.0289,
+      "step": 3882
+    },
+    {
+      "epoch": 13.867857142857144,
+      "grad_norm": 0.7378549608932323,
+      "learning_rate": 9.076841084825015e-06,
+      "loss": 0.0263,
+      "step": 3883
+    },
+    {
+      "epoch": 13.871428571428572,
+      "grad_norm": 1.1381667819964219,
+      "learning_rate": 9.067153456003716e-06,
+      "loss": 0.0539,
+      "step": 3884
+    },
+    {
+      "epoch": 13.875,
+      "grad_norm": 0.9482138217716097,
+      "learning_rate": 9.057469484085605e-06,
+      "loss": 0.0447,
+      "step": 3885
+    },
+    {
+      "epoch": 13.878571428571428,
+      "grad_norm": 0.9653744451884618,
+      "learning_rate": 9.047789172309844e-06,
+      "loss": 0.0501,
+      "step": 3886
+    },
+    {
+      "epoch": 13.882142857142856,
+      "grad_norm": 1.141834650481675,
+      "learning_rate": 9.038112523914393e-06,
+      "loss": 0.0379,
+      "step": 3887
+    },
+    {
+      "epoch": 13.885714285714286,
+      "grad_norm": 1.7052652375073034,
+      "learning_rate": 9.02843954213596e-06,
+      "loss": 0.0819,
+      "step": 3888
+    },
+    {
+      "epoch": 13.889285714285714,
+      "grad_norm": 0.6259715959069986,
+      "learning_rate": 9.01877023021005e-06,
+      "loss": 0.0325,
+      "step": 3889
+    },
+    {
+      "epoch": 13.892857142857142,
+      "grad_norm": 1.0926838225299693,
+      "learning_rate": 9.009104591370917e-06,
+      "loss": 0.0528,
+      "step": 3890
+    },
+    {
+      "epoch": 13.896428571428572,
+      "grad_norm": 1.2630426461765047,
+      "learning_rate": 8.999442628851611e-06,
+      "loss": 0.0841,
+      "step": 3891
+    },
+    {
+      "epoch": 13.9,
+      "grad_norm": 1.0512005886459779,
+      "learning_rate": 8.98978434588393e-06,
+      "loss": 0.0294,
+      "step": 3892
+    },
+    {
+      "epoch": 13.903571428571428,
+      "grad_norm": 1.068955332811711,
+      "learning_rate": 8.98012974569846e-06,
+      "loss": 0.0238,
+      "step": 3893
+    },
+    {
+      "epoch": 13.907142857142857,
+      "grad_norm": 0.6785012306275714,
+      "learning_rate": 8.970478831524536e-06,
+      "loss": 0.0228,
+      "step": 3894
+    },
+    {
+      "epoch": 13.910714285714286,
+      "grad_norm": 0.8598542221086015,
+      "learning_rate": 8.960831606590286e-06,
+      "loss": 0.0588,
+      "step": 3895
+    },
+    {
+      "epoch": 13.914285714285715,
+      "grad_norm": 0.8274086013305003,
+      "learning_rate": 8.95118807412257e-06,
+      "loss": 0.0431,
+      "step": 3896
+    },
+    {
+      "epoch": 13.917857142857143,
+      "grad_norm": 0.812963235117756,
+      "learning_rate": 8.941548237347048e-06,
+      "loss": 0.0451,
+      "step": 3897
+    },
+    {
+      "epoch": 13.92142857142857,
+      "grad_norm": 1.0348332935381774,
+      "learning_rate": 8.93191209948812e-06,
+      "loss": 0.0364,
+      "step": 3898
+    },
+    {
+      "epoch": 13.925,
+      "grad_norm": 0.7681432468514433,
+      "learning_rate": 8.92227966376895e-06,
+      "loss": 0.027,
+      "step": 3899
+    },
+    {
+      "epoch": 13.928571428571429,
+      "grad_norm": 0.8523095443004485,
+      "learning_rate": 8.91265093341148e-06,
+      "loss": 0.028,
+      "step": 3900
+    },
+    {
+      "epoch": 13.932142857142857,
+      "grad_norm": 1.0230127492400496,
+      "learning_rate": 8.903025911636393e-06,
+      "loss": 0.0563,
+      "step": 3901
+    },
+    {
+      "epoch": 13.935714285714285,
+      "grad_norm": 1.6337672749153658,
+      "learning_rate": 8.893404601663152e-06,
+      "loss": 0.0617,
+      "step": 3902
+    },
+    {
+      "epoch": 13.939285714285715,
+      "grad_norm": 0.8376657716930895,
+      "learning_rate": 8.883787006709953e-06,
+      "loss": 0.0474,
+      "step": 3903
+    },
+    {
+      "epoch": 13.942857142857143,
+      "grad_norm": 1.4444313951134753,
+      "learning_rate": 8.874173129993782e-06,
+      "loss": 0.0795,
+      "step": 3904
+    },
+    {
+      "epoch": 13.946428571428571,
+      "grad_norm": 1.5623325356506268,
+      "learning_rate": 8.864562974730344e-06,
+      "loss": 0.0638,
+      "step": 3905
+    },
+    {
+      "epoch": 13.95,
+      "grad_norm": 0.9027541405762148,
+      "learning_rate": 8.854956544134132e-06,
+      "loss": 0.0302,
+      "step": 3906
+    },
+    {
+      "epoch": 13.95357142857143,
+      "grad_norm": 1.1310834059480732,
+      "learning_rate": 8.845353841418369e-06,
+      "loss": 0.0475,
+      "step": 3907
+    },
+    {
+      "epoch": 13.957142857142857,
+      "grad_norm": 0.9405400689789333,
+      "learning_rate": 8.835754869795052e-06,
+      "loss": 0.0514,
+      "step": 3908
+    },
+    {
+      "epoch": 13.960714285714285,
+      "grad_norm": 0.7595849999233651,
+      "learning_rate": 8.826159632474916e-06,
+      "loss": 0.0455,
+      "step": 3909
+    },
+    {
+      "epoch": 13.964285714285714,
+      "grad_norm": 0.9022942731759905,
+      "learning_rate": 8.81656813266744e-06,
+      "loss": 0.046,
+      "step": 3910
+    },
+    {
+      "epoch": 13.967857142857143,
+      "grad_norm": 1.2447984618477985,
+      "learning_rate": 8.806980373580879e-06,
+      "loss": 0.0335,
+      "step": 3911
+    },
+    {
+      "epoch": 13.971428571428572,
+      "grad_norm": 0.9005209590299482,
+      "learning_rate": 8.797396358422206e-06,
+      "loss": 0.044,
+      "step": 3912
+    },
+    {
+      "epoch": 13.975,
+      "grad_norm": 0.8428970992593292,
+      "learning_rate": 8.78781609039717e-06,
+      "loss": 0.0508,
+      "step": 3913
+    },
+    {
+      "epoch": 13.978571428571428,
+      "grad_norm": 1.3721296034590116,
+      "learning_rate": 8.778239572710238e-06,
+      "loss": 0.0606,
+      "step": 3914
+    },
+    {
+      "epoch": 13.982142857142858,
+      "grad_norm": 1.9283300975341975,
+      "learning_rate": 8.768666808564658e-06,
+      "loss": 0.0384,
+      "step": 3915
+    },
+    {
+      "epoch": 13.985714285714286,
+      "grad_norm": 0.7261547011629556,
+      "learning_rate": 8.759097801162382e-06,
+      "loss": 0.0234,
+      "step": 3916
+    },
+    {
+      "epoch": 13.989285714285714,
+      "grad_norm": 0.8318882436139999,
+      "learning_rate": 8.749532553704142e-06,
+      "loss": 0.0222,
+      "step": 3917
+    },
+    {
+      "epoch": 13.992857142857144,
+      "grad_norm": 0.6150836107988323,
+      "learning_rate": 8.739971069389383e-06,
+      "loss": 0.0373,
+      "step": 3918
+    },
+    {
+      "epoch": 13.996428571428572,
+      "grad_norm": 1.8048133050868973,
+      "learning_rate": 8.730413351416315e-06,
+      "loss": 0.0317,
+      "step": 3919
+    },
+    {
+      "epoch": 14.0,
+      "grad_norm": 1.1382000303984054,
+      "learning_rate": 8.72085940298187e-06,
+      "loss": 0.0697,
+      "step": 3920
+    },
+    {
+      "epoch": 14.003571428571428,
+      "grad_norm": 0.7466249703863652,
+      "learning_rate": 8.711309227281732e-06,
+      "loss": 0.0312,
+      "step": 3921
+    },
+    {
+      "epoch": 14.007142857142858,
+      "grad_norm": 0.8429514780044993,
+      "learning_rate": 8.701762827510318e-06,
+      "loss": 0.0474,
+      "step": 3922
+    },
+    {
+      "epoch": 14.010714285714286,
+      "grad_norm": 0.8356440361990425,
+      "learning_rate": 8.692220206860771e-06,
+      "loss": 0.044,
+      "step": 3923
+    },
+    {
+      "epoch": 14.014285714285714,
+      "grad_norm": 0.6264230843568228,
+      "learning_rate": 8.682681368524996e-06,
+      "loss": 0.0382,
+      "step": 3924
+    },
+    {
+      "epoch": 14.017857142857142,
+      "grad_norm": 0.7930713967218075,
+      "learning_rate": 8.673146315693604e-06,
+      "loss": 0.0301,
+      "step": 3925
+    },
+    {
+      "epoch": 14.021428571428572,
+      "grad_norm": 0.7635847879532822,
+      "learning_rate": 8.663615051555965e-06,
+      "loss": 0.0351,
+      "step": 3926
+    },
+    {
+      "epoch": 14.025,
+      "grad_norm": 0.8340809431717688,
+      "learning_rate": 8.654087579300156e-06,
+      "loss": 0.0318,
+      "step": 3927
+    },
+    {
+      "epoch": 14.028571428571428,
+      "grad_norm": 1.9509103522077142,
+      "learning_rate": 8.644563902113017e-06,
+      "loss": 0.0479,
+      "step": 3928
+    },
+    {
+      "epoch": 14.032142857142857,
+      "grad_norm": 0.468583138483572,
+      "learning_rate": 8.635044023180086e-06,
+      "loss": 0.0171,
+      "step": 3929
+    },
+    {
+      "epoch": 14.035714285714286,
+      "grad_norm": 0.8640564271707438,
+      "learning_rate": 8.62552794568566e-06,
+      "loss": 0.0355,
+      "step": 3930
+    },
+    {
+      "epoch": 14.039285714285715,
+      "grad_norm": 1.2959256153160224,
+      "learning_rate": 8.616015672812732e-06,
+      "loss": 0.0835,
+      "step": 3931
+    },
+    {
+      "epoch": 14.042857142857143,
+      "grad_norm": 0.809520731277387,
+      "learning_rate": 8.606507207743062e-06,
+      "loss": 0.0479,
+      "step": 3932
+    },
+    {
+      "epoch": 14.04642857142857,
+      "grad_norm": 1.0136260159549064,
+      "learning_rate": 8.597002553657097e-06,
+      "loss": 0.0306,
+      "step": 3933
+    },
+    {
+      "epoch": 14.05,
+      "grad_norm": 0.9762782297045478,
+      "learning_rate": 8.587501713734039e-06,
+      "loss": 0.0332,
+      "step": 3934
+    },
+    {
+      "epoch": 14.053571428571429,
+      "grad_norm": 0.822174878858265,
+      "learning_rate": 8.5780046911518e-06,
+      "loss": 0.053,
+      "step": 3935
+    },
+    {
+      "epoch": 14.057142857142857,
+      "grad_norm": 0.8994458680374542,
+      "learning_rate": 8.56851148908701e-06,
+      "loss": 0.0431,
+      "step": 3936
+    },
+    {
+      "epoch": 14.060714285714285,
+      "grad_norm": 1.404447541490762,
+      "learning_rate": 8.55902211071504e-06,
+      "loss": 0.0529,
+      "step": 3937
+    },
+    {
+      "epoch": 14.064285714285715,
+      "grad_norm": 0.8709069423769285,
+      "learning_rate": 8.54953655920996e-06,
+      "loss": 0.0458,
+      "step": 3938
+    },
+    {
+      "epoch": 14.067857142857143,
+      "grad_norm": 1.0305856600264358,
+      "learning_rate": 8.540054837744583e-06,
+      "loss": 0.0543,
+      "step": 3939
+    },
+    {
+      "epoch": 14.071428571428571,
+      "grad_norm": 1.3051232521112832,
+      "learning_rate": 8.530576949490412e-06,
+      "loss": 0.0394,
+      "step": 3940
+    },
+    {
+      "epoch": 14.075,
+      "grad_norm": 0.9082270865201982,
+      "learning_rate": 8.521102897617703e-06,
+      "loss": 0.0493,
+      "step": 3941
+    },
+    {
+      "epoch": 14.07857142857143,
+      "grad_norm": 1.4003557324884104,
+      "learning_rate": 8.511632685295397e-06,
+      "loss": 0.0371,
+      "step": 3942
+    },
+    {
+      "epoch": 14.082142857142857,
+      "grad_norm": 1.943797669274648,
+      "learning_rate": 8.502166315691172e-06,
+      "loss": 0.0507,
+      "step": 3943
+    },
+    {
+      "epoch": 14.085714285714285,
+      "grad_norm": 1.085115706150346,
+      "learning_rate": 8.492703791971404e-06,
+      "loss": 0.0463,
+      "step": 3944
+    },
+    {
+      "epoch": 14.089285714285714,
+      "grad_norm": 0.8766177224406929,
+      "learning_rate": 8.483245117301202e-06,
+      "loss": 0.0243,
+      "step": 3945
+    },
+    {
+      "epoch": 14.092857142857143,
+      "grad_norm": 0.48257329819191985,
+      "learning_rate": 8.473790294844372e-06,
+      "loss": 0.023,
+      "step": 3946
+    },
+    {
+      "epoch": 14.096428571428572,
+      "grad_norm": 0.7531980720641436,
+      "learning_rate": 8.464339327763429e-06,
+      "loss": 0.0225,
+      "step": 3947
+    },
+    {
+      "epoch": 14.1,
+      "grad_norm": 1.123127774844,
+      "learning_rate": 8.454892219219617e-06,
+      "loss": 0.0223,
+      "step": 3948
+    },
+    {
+      "epoch": 14.103571428571428,
+      "grad_norm": 0.9973373914716441,
+      "learning_rate": 8.445448972372868e-06,
+      "loss": 0.0601,
+      "step": 3949
+    },
+    {
+      "epoch": 14.107142857142858,
+      "grad_norm": 0.979804922188906,
+      "learning_rate": 8.436009590381844e-06,
+      "loss": 0.0234,
+      "step": 3950
+    },
+    {
+      "epoch": 14.110714285714286,
+      "grad_norm": 0.6196184828646573,
+      "learning_rate": 8.426574076403887e-06,
+      "loss": 0.0201,
+      "step": 3951
+    },
+    {
+      "epoch": 14.114285714285714,
+      "grad_norm": 0.6987200663083484,
+      "learning_rate": 8.417142433595077e-06,
+      "loss": 0.0177,
+      "step": 3952
+    },
+    {
+      "epoch": 14.117857142857142,
+      "grad_norm": 0.8171548035071896,
+      "learning_rate": 8.407714665110171e-06,
+      "loss": 0.0205,
+      "step": 3953
+    },
+    {
+      "epoch": 14.121428571428572,
+      "grad_norm": 1.1222562608865951,
+      "learning_rate": 8.398290774102653e-06,
+      "loss": 0.0549,
+      "step": 3954
+    },
+    {
+      "epoch": 14.125,
+      "grad_norm": 1.3420221261653293,
+      "learning_rate": 8.388870763724685e-06,
+      "loss": 0.0432,
+      "step": 3955
+    },
+    {
+      "epoch": 14.128571428571428,
+      "grad_norm": 0.9189967727339683,
+      "learning_rate": 8.37945463712716e-06,
+      "loss": 0.0432,
+      "step": 3956
+    },
+    {
+      "epoch": 14.132142857142858,
+      "grad_norm": 1.0774247294684915,
+      "learning_rate": 8.370042397459644e-06,
+      "loss": 0.0401,
+      "step": 3957
+    },
+    {
+      "epoch": 14.135714285714286,
+      "grad_norm": 0.9850347859076346,
+      "learning_rate": 8.360634047870427e-06,
+      "loss": 0.0151,
+      "step": 3958
+    },
+    {
+      "epoch": 14.139285714285714,
+      "grad_norm": 1.7119178188465556,
+      "learning_rate": 8.351229591506482e-06,
+      "loss": 0.075,
+      "step": 3959
+    },
+    {
+      "epoch": 14.142857142857142,
+      "grad_norm": 0.9748324602844927,
+      "learning_rate": 8.341829031513476e-06,
+      "loss": 0.0525,
+      "step": 3960
+    },
+    {
+      "epoch": 14.146428571428572,
+      "grad_norm": 1.2031360759627807,
+      "learning_rate": 8.332432371035797e-06,
+      "loss": 0.0392,
+      "step": 3961
+    },
+    {
+      "epoch": 14.15,
+      "grad_norm": 1.6301297821516278,
+      "learning_rate": 8.323039613216495e-06,
+      "loss": 0.0248,
+      "step": 3962
+    },
+    {
+      "epoch": 14.153571428571428,
+      "grad_norm": 0.5692600327238201,
+      "learning_rate": 8.313650761197352e-06,
+      "loss": 0.0124,
+      "step": 3963
+    },
+    {
+      "epoch": 14.157142857142857,
+      "grad_norm": 1.0705452299651732,
+      "learning_rate": 8.304265818118804e-06,
+      "loss": 0.0339,
+      "step": 3964
+    },
+    {
+      "epoch": 14.160714285714286,
+      "grad_norm": 0.9090274800128738,
+      "learning_rate": 8.294884787120019e-06,
+      "loss": 0.0244,
+      "step": 3965
+    },
+    {
+      "epoch": 14.164285714285715,
+      "grad_norm": 1.6161129701374235,
+      "learning_rate": 8.285507671338817e-06,
+      "loss": 0.0704,
+      "step": 3966
+    },
+    {
+      "epoch": 14.167857142857143,
+      "grad_norm": 1.1627423041947573,
+      "learning_rate": 8.276134473911747e-06,
+      "loss": 0.0243,
+      "step": 3967
+    },
+    {
+      "epoch": 14.17142857142857,
+      "grad_norm": 0.7856678241331625,
+      "learning_rate": 8.266765197974013e-06,
+      "loss": 0.0321,
+      "step": 3968
+    },
+    {
+      "epoch": 14.175,
+      "grad_norm": 0.6775883014257397,
+      "learning_rate": 8.25739984665954e-06,
+      "loss": 0.0392,
+      "step": 3969
+    },
+    {
+      "epoch": 14.178571428571429,
+      "grad_norm": 1.1151381850306052,
+      "learning_rate": 8.248038423100904e-06,
+      "loss": 0.0477,
+      "step": 3970
+    },
+    {
+      "epoch": 14.182142857142857,
+      "grad_norm": 0.970335806640711,
+      "learning_rate": 8.238680930429402e-06,
+      "loss": 0.0377,
+      "step": 3971
+    },
+    {
+      "epoch": 14.185714285714285,
+      "grad_norm": 0.8253197370206999,
+      "learning_rate": 8.229327371774995e-06,
+      "loss": 0.0355,
+      "step": 3972
+    },
+    {
+      "epoch": 14.189285714285715,
+      "grad_norm": 0.5714413943118246,
+      "learning_rate": 8.219977750266326e-06,
+      "loss": 0.0263,
+      "step": 3973
+    },
+    {
+      "epoch": 14.192857142857143,
+      "grad_norm": 0.4953404569529705,
+      "learning_rate": 8.210632069030744e-06,
+      "loss": 0.0234,
+      "step": 3974
+    },
+    {
+      "epoch": 14.196428571428571,
+      "grad_norm": 1.360027821549407,
+      "learning_rate": 8.20129033119425e-06,
+      "loss": 0.0618,
+      "step": 3975
+    },
+    {
+      "epoch": 14.2,
+      "grad_norm": 1.1689125875731987,
+      "learning_rate": 8.191952539881554e-06,
+      "loss": 0.022,
+      "step": 3976
+    },
+    {
+      "epoch": 14.20357142857143,
+      "grad_norm": 1.5036354946736095,
+      "learning_rate": 8.182618698216019e-06,
+      "loss": 0.0713,
+      "step": 3977
+    },
+    {
+      "epoch": 14.207142857142857,
+      "grad_norm": 0.4856382064188858,
+      "learning_rate": 8.173288809319716e-06,
+      "loss": 0.0154,
+      "step": 3978
+    },
+    {
+      "epoch": 14.210714285714285,
+      "grad_norm": 1.4801800787366834,
+      "learning_rate": 8.163962876313363e-06,
+      "loss": 0.0528,
+      "step": 3979
+    },
+    {
+      "epoch": 14.214285714285714,
+      "grad_norm": 2.034200611570026,
+      "learning_rate": 8.154640902316385e-06,
+      "loss": 0.0922,
+      "step": 3980
+    },
+    {
+      "epoch": 14.217857142857143,
+      "grad_norm": 0.8233752198334379,
+      "learning_rate": 8.145322890446853e-06,
+      "loss": 0.0392,
+      "step": 3981
+    },
+    {
+      "epoch": 14.221428571428572,
+      "grad_norm": 0.801382031586967,
+      "learning_rate": 8.136008843821545e-06,
+      "loss": 0.0195,
+      "step": 3982
+    },
+    {
+      "epoch": 14.225,
+      "grad_norm": 1.216770993871357,
+      "learning_rate": 8.126698765555885e-06,
+      "loss": 0.0346,
+      "step": 3983
+    },
+    {
+      "epoch": 14.228571428571428,
+      "grad_norm": 1.1291642445925798,
+      "learning_rate": 8.117392658763974e-06,
+      "loss": 0.0607,
+      "step": 3984
+    },
+    {
+      "epoch": 14.232142857142858,
+      "grad_norm": 0.8075411999673058,
+      "learning_rate": 8.108090526558605e-06,
+      "loss": 0.0185,
+      "step": 3985
+    },
+    {
+      "epoch": 14.235714285714286,
+      "grad_norm": 0.7776185464954325,
+      "learning_rate": 8.098792372051213e-06,
+      "loss": 0.0208,
+      "step": 3986
+    },
+    {
+      "epoch": 14.239285714285714,
+      "grad_norm": 0.9225667831987769,
+      "learning_rate": 8.089498198351934e-06,
+      "loss": 0.0293,
+      "step": 3987
+    },
+    {
+      "epoch": 14.242857142857142,
+      "grad_norm": 0.5050846393722702,
+      "learning_rate": 8.080208008569525e-06,
+      "loss": 0.0281,
+      "step": 3988
+    },
+    {
+      "epoch": 14.246428571428572,
+      "grad_norm": 1.3303122204159623,
+      "learning_rate": 8.07092180581148e-06,
+      "loss": 0.0332,
+      "step": 3989
+    },
+    {
+      "epoch": 14.25,
+      "grad_norm": 0.3461533291880898,
+      "learning_rate": 8.0616395931839e-06,
+      "loss": 0.0099,
+      "step": 3990
+    },
+    {
+      "epoch": 14.253571428571428,
+      "grad_norm": 1.0014525190261794,
+      "learning_rate": 8.052361373791565e-06,
+      "loss": 0.0415,
+      "step": 3991
+    },
+    {
+      "epoch": 14.257142857142856,
+      "grad_norm": 0.7127686162622905,
+      "learning_rate": 8.043087150737936e-06,
+      "loss": 0.0369,
+      "step": 3992
+    },
+    {
+      "epoch": 14.260714285714286,
+      "grad_norm": 1.272122770281648,
+      "learning_rate": 8.03381692712512e-06,
+      "loss": 0.0369,
+      "step": 3993
+    },
+    {
+      "epoch": 14.264285714285714,
+      "grad_norm": 1.7624674112132188,
+      "learning_rate": 8.024550706053904e-06,
+      "loss": 0.0819,
+      "step": 3994
+    },
+    {
+      "epoch": 14.267857142857142,
+      "grad_norm": 1.0526613964801803,
+      "learning_rate": 8.015288490623712e-06,
+      "loss": 0.0524,
+      "step": 3995
+    },
+    {
+      "epoch": 14.271428571428572,
+      "grad_norm": 0.6966699888928789,
+      "learning_rate": 8.006030283932654e-06,
+      "loss": 0.0207,
+      "step": 3996
+    },
+    {
+      "epoch": 14.275,
+      "grad_norm": 1.1910837501088964,
+      "learning_rate": 7.996776089077473e-06,
+      "loss": 0.0411,
+      "step": 3997
+    },
+    {
+      "epoch": 14.278571428571428,
+      "grad_norm": 0.8777840279200068,
+      "learning_rate": 7.9875259091536e-06,
+      "loss": 0.0432,
+      "step": 3998
+    },
+    {
+      "epoch": 14.282142857142857,
+      "grad_norm": 0.7847763624911668,
+      "learning_rate": 7.978279747255089e-06,
+      "loss": 0.0432,
+      "step": 3999
+    },
+    {
+      "epoch": 14.285714285714286,
+      "grad_norm": 0.8759093420950511,
+      "learning_rate": 7.969037606474686e-06,
+      "loss": 0.0538,
+      "step": 4000
+    },
+    {
+      "epoch": 14.289285714285715,
+      "grad_norm": 1.3797878332160691,
+      "learning_rate": 7.959799489903757e-06,
+      "loss": 0.0472,
+      "step": 4001
+    },
+    {
+      "epoch": 14.292857142857143,
+      "grad_norm": 0.815474910467155,
+      "learning_rate": 7.950565400632353e-06,
+      "loss": 0.0298,
+      "step": 4002
+    },
+    {
+      "epoch": 14.29642857142857,
+      "grad_norm": 0.7109624544220678,
+      "learning_rate": 7.941335341749158e-06,
+      "loss": 0.0418,
+      "step": 4003
+    },
+    {
+      "epoch": 14.3,
+      "grad_norm": 1.177899028191375,
+      "learning_rate": 7.932109316341508e-06,
+      "loss": 0.0443,
+      "step": 4004
+    },
+    {
+      "epoch": 14.303571428571429,
+      "grad_norm": 1.2077341997596327,
+      "learning_rate": 7.922887327495404e-06,
+      "loss": 0.0316,
+      "step": 4005
+    },
+    {
+      "epoch": 14.307142857142857,
+      "grad_norm": 0.741135312860146,
+      "learning_rate": 7.913669378295479e-06,
+      "loss": 0.0524,
+      "step": 4006
+    },
+    {
+      "epoch": 14.310714285714285,
+      "grad_norm": 1.1648929301253894,
+      "learning_rate": 7.904455471825037e-06,
+      "loss": 0.0537,
+      "step": 4007
+    },
+    {
+      "epoch": 14.314285714285715,
+      "grad_norm": 0.7212583966609362,
+      "learning_rate": 7.895245611166003e-06,
+      "loss": 0.0427,
+      "step": 4008
+    },
+    {
+      "epoch": 14.317857142857143,
+      "grad_norm": 0.5674795171037618,
+      "learning_rate": 7.886039799398974e-06,
+      "loss": 0.029,
+      "step": 4009
+    },
+    {
+      "epoch": 14.321428571428571,
+      "grad_norm": 1.3738923588969636,
+      "learning_rate": 7.876838039603169e-06,
+      "loss": 0.0375,
+      "step": 4010
+    },
+    {
+      "epoch": 14.325,
+      "grad_norm": 1.1950889149223816,
+      "learning_rate": 7.867640334856481e-06,
+      "loss": 0.048,
+      "step": 4011
+    },
+    {
+      "epoch": 14.32857142857143,
+      "grad_norm": 1.0713634829700185,
+      "learning_rate": 7.85844668823541e-06,
+      "loss": 0.0317,
+      "step": 4012
+    },
+    {
+      "epoch": 14.332142857142857,
+      "grad_norm": 1.0127797411153323,
+      "learning_rate": 7.849257102815138e-06,
+      "loss": 0.0333,
+      "step": 4013
+    },
+    {
+      "epoch": 14.335714285714285,
+      "grad_norm": 0.9026744674348927,
+      "learning_rate": 7.840071581669455e-06,
+      "loss": 0.0366,
+      "step": 4014
+    },
+    {
+      "epoch": 14.339285714285714,
+      "grad_norm": 1.486232855161559,
+      "learning_rate": 7.830890127870803e-06,
+      "loss": 0.0576,
+      "step": 4015
+    },
+    {
+      "epoch": 14.342857142857143,
+      "grad_norm": 0.9106674451604208,
+      "learning_rate": 7.821712744490278e-06,
+      "loss": 0.0342,
+      "step": 4016
+    },
+    {
+      "epoch": 14.346428571428572,
+      "grad_norm": 1.049059588748774,
+      "learning_rate": 7.812539434597591e-06,
+      "loss": 0.0268,
+      "step": 4017
+    },
+    {
+      "epoch": 14.35,
+      "grad_norm": 0.9140514394506649,
+      "learning_rate": 7.803370201261108e-06,
+      "loss": 0.0447,
+      "step": 4018
+    },
+    {
+      "epoch": 14.353571428571428,
+      "grad_norm": 0.7224137374521817,
+      "learning_rate": 7.794205047547823e-06,
+      "loss": 0.0381,
+      "step": 4019
+    },
+    {
+      "epoch": 14.357142857142858,
+      "grad_norm": 1.4023521628009594,
+      "learning_rate": 7.785043976523372e-06,
+      "loss": 0.0431,
+      "step": 4020
+    },
+    {
+      "epoch": 14.360714285714286,
+      "grad_norm": 1.1215090607507585,
+      "learning_rate": 7.77588699125201e-06,
+      "loss": 0.048,
+      "step": 4021
+    },
+    {
+      "epoch": 14.364285714285714,
+      "grad_norm": 0.9157579846100271,
+      "learning_rate": 7.76673409479665e-06,
+      "loss": 0.0369,
+      "step": 4022
+    },
+    {
+      "epoch": 14.367857142857144,
+      "grad_norm": 0.8233768953494535,
+      "learning_rate": 7.757585290218815e-06,
+      "loss": 0.0179,
+      "step": 4023
+    },
+    {
+      "epoch": 14.371428571428572,
+      "grad_norm": 5.791998493227808,
+      "learning_rate": 7.748440580578674e-06,
+      "loss": 0.0453,
+      "step": 4024
+    },
+    {
+      "epoch": 14.375,
+      "grad_norm": 1.2370491255983356,
+      "learning_rate": 7.739299968935008e-06,
+      "loss": 0.0555,
+      "step": 4025
+    },
+    {
+      "epoch": 14.378571428571428,
+      "grad_norm": 0.8089564164186059,
+      "learning_rate": 7.730163458345257e-06,
+      "loss": 0.0222,
+      "step": 4026
+    },
+    {
+      "epoch": 14.382142857142856,
+      "grad_norm": 0.9824917935002203,
+      "learning_rate": 7.721031051865461e-06,
+      "loss": 0.0614,
+      "step": 4027
+    },
+    {
+      "epoch": 14.385714285714286,
+      "grad_norm": 0.5811396798114765,
+      "learning_rate": 7.711902752550298e-06,
+      "loss": 0.0312,
+      "step": 4028
+    },
+    {
+      "epoch": 14.389285714285714,
+      "grad_norm": 0.9971548647295526,
+      "learning_rate": 7.702778563453078e-06,
+      "loss": 0.0492,
+      "step": 4029
+    },
+    {
+      "epoch": 14.392857142857142,
+      "grad_norm": 0.716383667208903,
+      "learning_rate": 7.69365848762572e-06,
+      "loss": 0.0424,
+      "step": 4030
+    },
+    {
+      "epoch": 14.396428571428572,
+      "grad_norm": 0.9981842721534688,
+      "learning_rate": 7.68454252811879e-06,
+      "loss": 0.0461,
+      "step": 4031
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 0.9429663788284476,
+      "learning_rate": 7.675430687981454e-06,
+      "loss": 0.0441,
+      "step": 4032
+    },
+    {
+      "epoch": 14.403571428571428,
+      "grad_norm": 1.0124310020487153,
+      "learning_rate": 7.666322970261521e-06,
+      "loss": 0.0522,
+      "step": 4033
+    },
+    {
+      "epoch": 14.407142857142857,
+      "grad_norm": 0.8490485286608483,
+      "learning_rate": 7.657219378005402e-06,
+      "loss": 0.036,
+      "step": 4034
+    },
+    {
+      "epoch": 14.410714285714286,
+      "grad_norm": 1.1567052391231418,
+      "learning_rate": 7.648119914258145e-06,
+      "loss": 0.0488,
+      "step": 4035
+    },
+    {
+      "epoch": 14.414285714285715,
+      "grad_norm": 1.4901795436551093,
+      "learning_rate": 7.639024582063401e-06,
+      "loss": 0.0504,
+      "step": 4036
+    },
+    {
+      "epoch": 14.417857142857143,
+      "grad_norm": 1.0202406611133157,
+      "learning_rate": 7.629933384463457e-06,
+      "loss": 0.0376,
+      "step": 4037
+    },
+    {
+      "epoch": 14.42142857142857,
+      "grad_norm": 1.0791682179608144,
+      "learning_rate": 7.620846324499194e-06,
+      "loss": 0.0532,
+      "step": 4038
+    },
+    {
+      "epoch": 14.425,
+      "grad_norm": 0.6477694529299298,
+      "learning_rate": 7.611763405210135e-06,
+      "loss": 0.0244,
+      "step": 4039
+    },
+    {
+      "epoch": 14.428571428571429,
+      "grad_norm": 1.3946771108822718,
+      "learning_rate": 7.602684629634402e-06,
+      "loss": 0.0359,
+      "step": 4040
+    },
+    {
+      "epoch": 14.432142857142857,
+      "grad_norm": 1.1539342756708177,
+      "learning_rate": 7.5936100008087245e-06,
+      "loss": 0.027,
+      "step": 4041
+    },
+    {
+      "epoch": 14.435714285714285,
+      "grad_norm": 0.5773821803854401,
+      "learning_rate": 7.58453952176847e-06,
+      "loss": 0.0332,
+      "step": 4042
+    },
+    {
+      "epoch": 14.439285714285715,
+      "grad_norm": 0.6244411995377929,
+      "learning_rate": 7.5754731955475865e-06,
+      "loss": 0.0168,
+      "step": 4043
+    },
+    {
+      "epoch": 14.442857142857143,
+      "grad_norm": 1.4566552466452816,
+      "learning_rate": 7.5664110251786636e-06,
+      "loss": 0.0362,
+      "step": 4044
+    },
+    {
+      "epoch": 14.446428571428571,
+      "grad_norm": 1.4223847004567267,
+      "learning_rate": 7.557353013692874e-06,
+      "loss": 0.0452,
+      "step": 4045
+    },
+    {
+      "epoch": 14.45,
+      "grad_norm": 1.6792349178610364,
+      "learning_rate": 7.5482991641200256e-06,
+      "loss": 0.0495,
+      "step": 4046
+    },
+    {
+      "epoch": 14.45357142857143,
+      "grad_norm": 1.5150558426736578,
+      "learning_rate": 7.539249479488504e-06,
+      "loss": 0.0423,
+      "step": 4047
+    },
+    {
+      "epoch": 14.457142857142857,
+      "grad_norm": 1.1800525437291358,
+      "learning_rate": 7.530203962825331e-06,
+      "loss": 0.0157,
+      "step": 4048
+    },
+    {
+      "epoch": 14.460714285714285,
+      "grad_norm": 1.1840875216976174,
+      "learning_rate": 7.52116261715611e-06,
+      "loss": 0.0266,
+      "step": 4049
+    },
+    {
+      "epoch": 14.464285714285714,
+      "grad_norm": 1.25310256072758,
+      "learning_rate": 7.512125445505074e-06,
+      "loss": 0.0391,
+      "step": 4050
+    },
+    {
+      "epoch": 14.467857142857143,
+      "grad_norm": 1.9183481271202747,
+      "learning_rate": 7.503092450895037e-06,
+      "loss": 0.0509,
+      "step": 4051
+    },
+    {
+      "epoch": 14.471428571428572,
+      "grad_norm": 1.358181448378258,
+      "learning_rate": 7.49406363634742e-06,
+      "loss": 0.0552,
+      "step": 4052
+    },
+    {
+      "epoch": 14.475,
+      "grad_norm": 1.0897326237584175,
+      "learning_rate": 7.485039004882264e-06,
+      "loss": 0.0666,
+      "step": 4053
+    },
+    {
+      "epoch": 14.478571428571428,
+      "grad_norm": 1.3631187868141685,
+      "learning_rate": 7.476018559518184e-06,
+      "loss": 0.0331,
+      "step": 4054
+    },
+    {
+      "epoch": 14.482142857142858,
+      "grad_norm": 1.1119007945337047,
+      "learning_rate": 7.467002303272421e-06,
+      "loss": 0.041,
+      "step": 4055
+    },
+    {
+      "epoch": 14.485714285714286,
+      "grad_norm": 1.1565601788237005,
+      "learning_rate": 7.457990239160791e-06,
+      "loss": 0.0363,
+      "step": 4056
+    },
+    {
+      "epoch": 14.489285714285714,
+      "grad_norm": 0.8047954207343399,
+      "learning_rate": 7.448982370197728e-06,
+      "loss": 0.024,
+      "step": 4057
+    },
+    {
+      "epoch": 14.492857142857144,
+      "grad_norm": 0.7591433912894009,
+      "learning_rate": 7.439978699396246e-06,
+      "loss": 0.0476,
+      "step": 4058
+    },
+    {
+      "epoch": 14.496428571428572,
+      "grad_norm": 0.7620177265292419,
+      "learning_rate": 7.430979229767974e-06,
+      "loss": 0.0488,
+      "step": 4059
+    },
+    {
+      "epoch": 14.5,
+      "grad_norm": 1.3314543394857423,
+      "learning_rate": 7.421983964323109e-06,
+      "loss": 0.0575,
+      "step": 4060
+    },
+    {
+      "epoch": 14.503571428571428,
+      "grad_norm": 0.7948087636975563,
+      "learning_rate": 7.4129929060704735e-06,
+      "loss": 0.0569,
+      "step": 4061
+    },
+    {
+      "epoch": 14.507142857142856,
+      "grad_norm": 1.3339294054691575,
+      "learning_rate": 7.4040060580174475e-06,
+      "loss": 0.0463,
+      "step": 4062
+    },
+    {
+      "epoch": 14.510714285714286,
+      "grad_norm": 1.903783342737482,
+      "learning_rate": 7.395023423170042e-06,
+      "loss": 0.084,
+      "step": 4063
+    },
+    {
+      "epoch": 14.514285714285714,
+      "grad_norm": 1.373621333429697,
+      "learning_rate": 7.386045004532827e-06,
+      "loss": 0.0305,
+      "step": 4064
+    },
+    {
+      "epoch": 14.517857142857142,
+      "grad_norm": 1.3347800843212283,
+      "learning_rate": 7.377070805108968e-06,
+      "loss": 0.0497,
+      "step": 4065
+    },
+    {
+      "epoch": 14.521428571428572,
+      "grad_norm": 0.9525226393221096,
+      "learning_rate": 7.36810082790024e-06,
+      "loss": 0.036,
+      "step": 4066
+    },
+    {
+      "epoch": 14.525,
+      "grad_norm": 0.7434284212643028,
+      "learning_rate": 7.359135075906971e-06,
+      "loss": 0.022,
+      "step": 4067
+    },
+    {
+      "epoch": 14.528571428571428,
+      "grad_norm": 0.7446622458795659,
+      "learning_rate": 7.350173552128117e-06,
+      "loss": 0.0324,
+      "step": 4068
+    },
+    {
+      "epoch": 14.532142857142857,
+      "grad_norm": 0.9071343165107048,
+      "learning_rate": 7.341216259561177e-06,
+      "loss": 0.0347,
+      "step": 4069
+    },
+    {
+      "epoch": 14.535714285714286,
+      "grad_norm": 1.546402029709079,
+      "learning_rate": 7.332263201202274e-06,
+      "loss": 0.0687,
+      "step": 4070
+    },
+    {
+      "epoch": 14.539285714285715,
+      "grad_norm": 0.9683869480960069,
+      "learning_rate": 7.32331438004608e-06,
+      "loss": 0.0512,
+      "step": 4071
+    },
+    {
+      "epoch": 14.542857142857143,
+      "grad_norm": 0.9716710257982811,
+      "learning_rate": 7.314369799085881e-06,
+      "loss": 0.0475,
+      "step": 4072
+    },
+    {
+      "epoch": 14.54642857142857,
+      "grad_norm": 1.8405345404020685,
+      "learning_rate": 7.305429461313516e-06,
+      "loss": 0.091,
+      "step": 4073
+    },
+    {
+      "epoch": 14.55,
+      "grad_norm": 0.4251877110300742,
+      "learning_rate": 7.296493369719433e-06,
+      "loss": 0.0102,
+      "step": 4074
+    },
+    {
+      "epoch": 14.553571428571429,
+      "grad_norm": 0.7188239945478129,
+      "learning_rate": 7.287561527292628e-06,
+      "loss": 0.0332,
+      "step": 4075
+    },
+    {
+      "epoch": 14.557142857142857,
+      "grad_norm": 0.827414254374259,
+      "learning_rate": 7.278633937020709e-06,
+      "loss": 0.0296,
+      "step": 4076
+    },
+    {
+      "epoch": 14.560714285714285,
+      "grad_norm": 0.7363647941055106,
+      "learning_rate": 7.269710601889839e-06,
+      "loss": 0.0341,
+      "step": 4077
+    },
+    {
+      "epoch": 14.564285714285715,
+      "grad_norm": 0.4313844602774652,
+      "learning_rate": 7.260791524884756e-06,
+      "loss": 0.0184,
+      "step": 4078
+    },
+    {
+      "epoch": 14.567857142857143,
+      "grad_norm": 0.7786662438026264,
+      "learning_rate": 7.251876708988799e-06,
+      "loss": 0.0384,
+      "step": 4079
+    },
+    {
+      "epoch": 14.571428571428571,
+      "grad_norm": 1.5398907258904135,
+      "learning_rate": 7.242966157183846e-06,
+      "loss": 0.0488,
+      "step": 4080
+    },
+    {
+      "epoch": 14.575,
+      "grad_norm": 1.4963756402628845,
+      "learning_rate": 7.234059872450385e-06,
+      "loss": 0.0277,
+      "step": 4081
+    },
+    {
+      "epoch": 14.57857142857143,
+      "grad_norm": 2.0018064158170343,
+      "learning_rate": 7.225157857767442e-06,
+      "loss": 0.0303,
+      "step": 4082
+    },
+    {
+      "epoch": 14.582142857142857,
+      "grad_norm": 1.4692129827529152,
+      "learning_rate": 7.21626011611265e-06,
+      "loss": 0.063,
+      "step": 4083
+    },
+    {
+      "epoch": 14.585714285714285,
+      "grad_norm": 0.9493183502681226,
+      "learning_rate": 7.207366650462177e-06,
+      "loss": 0.0258,
+      "step": 4084
+    },
+    {
+      "epoch": 14.589285714285714,
+      "grad_norm": 1.038644799880571,
+      "learning_rate": 7.198477463790794e-06,
+      "loss": 0.0192,
+      "step": 4085
+    },
+    {
+      "epoch": 14.592857142857143,
+      "grad_norm": 1.409034448899324,
+      "learning_rate": 7.189592559071809e-06,
+      "loss": 0.0331,
+      "step": 4086
+    },
+    {
+      "epoch": 14.596428571428572,
+      "grad_norm": 1.0370436398817986,
+      "learning_rate": 7.18071193927713e-06,
+      "loss": 0.0347,
+      "step": 4087
+    },
+    {
+      "epoch": 14.6,
+      "grad_norm": 0.9189478124712223,
+      "learning_rate": 7.171835607377206e-06,
+      "loss": 0.0304,
+      "step": 4088
+    },
+    {
+      "epoch": 14.603571428571428,
+      "grad_norm": 1.0973107107585236,
+      "learning_rate": 7.162963566341059e-06,
+      "loss": 0.0447,
+      "step": 4089
+    },
+    {
+      "epoch": 14.607142857142858,
+      "grad_norm": 0.9635697442605489,
+      "learning_rate": 7.154095819136287e-06,
+      "loss": 0.0635,
+      "step": 4090
+    },
+    {
+      "epoch": 14.610714285714286,
+      "grad_norm": 1.1805798581749964,
+      "learning_rate": 7.145232368729031e-06,
+      "loss": 0.0276,
+      "step": 4091
+    },
+    {
+      "epoch": 14.614285714285714,
+      "grad_norm": 3.5489368557215957,
+      "learning_rate": 7.136373218084021e-06,
+      "loss": 0.0657,
+      "step": 4092
+    },
+    {
+      "epoch": 14.617857142857144,
+      "grad_norm": 0.8667643934916195,
+      "learning_rate": 7.127518370164519e-06,
+      "loss": 0.0281,
+      "step": 4093
+    },
+    {
+      "epoch": 14.621428571428572,
+      "grad_norm": 0.5782990714099793,
+      "learning_rate": 7.118667827932377e-06,
+      "loss": 0.0186,
+      "step": 4094
+    },
+    {
+      "epoch": 14.625,
+      "grad_norm": 0.7706890762505277,
+      "learning_rate": 7.109821594347981e-06,
+      "loss": 0.0244,
+      "step": 4095
+    },
+    {
+      "epoch": 14.628571428571428,
+      "grad_norm": 1.1932635648330543,
+      "learning_rate": 7.1009796723702985e-06,
+      "loss": 0.0529,
+      "step": 4096
+    },
+    {
+      "epoch": 14.632142857142856,
+      "grad_norm": 0.5728804818075194,
+      "learning_rate": 7.0921420649568325e-06,
+      "loss": 0.0182,
+      "step": 4097
+    },
+    {
+      "epoch": 14.635714285714286,
+      "grad_norm": 0.7769628729975598,
+      "learning_rate": 7.083308775063669e-06,
+      "loss": 0.0325,
+      "step": 4098
+    },
+    {
+      "epoch": 14.639285714285714,
+      "grad_norm": 1.1136237008862304,
+      "learning_rate": 7.074479805645418e-06,
+      "loss": 0.0599,
+      "step": 4099
+    },
+    {
+      "epoch": 14.642857142857142,
+      "grad_norm": 1.437711724482509,
+      "learning_rate": 7.065655159655278e-06,
+      "loss": 0.0416,
+      "step": 4100
+    },
+    {
+      "epoch": 14.646428571428572,
+      "grad_norm": 0.8080262275845915,
+      "learning_rate": 7.056834840044977e-06,
+      "loss": 0.0383,
+      "step": 4101
+    },
+    {
+      "epoch": 14.65,
+      "grad_norm": 1.542197713908992,
+      "learning_rate": 7.0480188497648e-06,
+      "loss": 0.0453,
+      "step": 4102
+    },
+    {
+      "epoch": 14.653571428571428,
+      "grad_norm": 1.3090616214911999,
+      "learning_rate": 7.039207191763595e-06,
+      "loss": 0.044,
+      "step": 4103
+    },
+    {
+      "epoch": 14.657142857142857,
+      "grad_norm": 0.8231154419354503,
+      "learning_rate": 7.030399868988746e-06,
+      "loss": 0.0375,
+      "step": 4104
+    },
+    {
+      "epoch": 14.660714285714286,
+      "grad_norm": 1.2548246416569209,
+      "learning_rate": 7.021596884386206e-06,
+      "loss": 0.0398,
+      "step": 4105
+    },
+    {
+      "epoch": 14.664285714285715,
+      "grad_norm": 0.6022820066397878,
+      "learning_rate": 7.0127982409004515e-06,
+      "loss": 0.0274,
+      "step": 4106
+    },
+    {
+      "epoch": 14.667857142857143,
+      "grad_norm": 1.2791778547307233,
+      "learning_rate": 7.004003941474533e-06,
+      "loss": 0.0608,
+      "step": 4107
+    },
+    {
+      "epoch": 14.67142857142857,
+      "grad_norm": 0.9660748707014634,
+      "learning_rate": 6.995213989050027e-06,
+      "loss": 0.0244,
+      "step": 4108
+    },
+    {
+      "epoch": 14.675,
+      "grad_norm": 0.9565346503516746,
+      "learning_rate": 6.9864283865670725e-06,
+      "loss": 0.0556,
+      "step": 4109
+    },
+    {
+      "epoch": 14.678571428571429,
+      "grad_norm": 1.1513673221791896,
+      "learning_rate": 6.977647136964338e-06,
+      "loss": 0.0311,
+      "step": 4110
+    },
+    {
+      "epoch": 14.682142857142857,
+      "grad_norm": 1.346490405375332,
+      "learning_rate": 6.9688702431790535e-06,
+      "loss": 0.0712,
+      "step": 4111
+    },
+    {
+      "epoch": 14.685714285714285,
+      "grad_norm": 0.9694910438113863,
+      "learning_rate": 6.9600977081469715e-06,
+      "loss": 0.0699,
+      "step": 4112
+    },
+    {
+      "epoch": 14.689285714285715,
+      "grad_norm": 0.9379895003053705,
+      "learning_rate": 6.951329534802407e-06,
+      "loss": 0.0329,
+      "step": 4113
+    },
+    {
+      "epoch": 14.692857142857143,
+      "grad_norm": 1.6106161291467394,
+      "learning_rate": 6.942565726078203e-06,
+      "loss": 0.0467,
+      "step": 4114
+    },
+    {
+      "epoch": 14.696428571428571,
+      "grad_norm": 0.837271079828966,
+      "learning_rate": 6.93380628490574e-06,
+      "loss": 0.0402,
+      "step": 4115
+    },
+    {
+      "epoch": 14.7,
+      "grad_norm": 1.788111275593519,
+      "learning_rate": 6.925051214214955e-06,
+      "loss": 0.0699,
+      "step": 4116
+    },
+    {
+      "epoch": 14.70357142857143,
+      "grad_norm": 1.2122980259486749,
+      "learning_rate": 6.9163005169343e-06,
+      "loss": 0.021,
+      "step": 4117
+    },
+    {
+      "epoch": 14.707142857142857,
+      "grad_norm": 0.9677184719158805,
+      "learning_rate": 6.9075541959907885e-06,
+      "loss": 0.0484,
+      "step": 4118
+    },
+    {
+      "epoch": 14.710714285714285,
+      "grad_norm": 0.731298638217899,
+      "learning_rate": 6.898812254309946e-06,
+      "loss": 0.0176,
+      "step": 4119
+    },
+    {
+      "epoch": 14.714285714285714,
+      "grad_norm": 0.6429577795810101,
+      "learning_rate": 6.8900746948158556e-06,
+      "loss": 0.0227,
+      "step": 4120
+    },
+    {
+      "epoch": 14.717857142857143,
+      "grad_norm": 1.1057961378473875,
+      "learning_rate": 6.8813415204311105e-06,
+      "loss": 0.0294,
+      "step": 4121
+    },
+    {
+      "epoch": 14.721428571428572,
+      "grad_norm": 1.3240239083018734,
+      "learning_rate": 6.8726127340768664e-06,
+      "loss": 0.046,
+      "step": 4122
+    },
+    {
+      "epoch": 14.725,
+      "grad_norm": 1.7408022311294613,
+      "learning_rate": 6.863888338672782e-06,
+      "loss": 0.04,
+      "step": 4123
+    },
+    {
+      "epoch": 14.728571428571428,
+      "grad_norm": 0.997633328581503,
+      "learning_rate": 6.85516833713707e-06,
+      "loss": 0.0459,
+      "step": 4124
+    },
+    {
+      "epoch": 14.732142857142858,
+      "grad_norm": 1.6601445578766467,
+      "learning_rate": 6.846452732386462e-06,
+      "loss": 0.0211,
+      "step": 4125
+    },
+    {
+      "epoch": 14.735714285714286,
+      "grad_norm": 1.9176657042996457,
+      "learning_rate": 6.837741527336212e-06,
+      "loss": 0.0538,
+      "step": 4126
+    },
+    {
+      "epoch": 14.739285714285714,
+      "grad_norm": 0.4414513742118312,
+      "learning_rate": 6.829034724900123e-06,
+      "loss": 0.0112,
+      "step": 4127
+    },
+    {
+      "epoch": 14.742857142857144,
+      "grad_norm": 0.8975833600740171,
+      "learning_rate": 6.8203323279905066e-06,
+      "loss": 0.0162,
+      "step": 4128
+    },
+    {
+      "epoch": 14.746428571428572,
+      "grad_norm": 1.8939348951775883,
+      "learning_rate": 6.811634339518214e-06,
+      "loss": 0.0543,
+      "step": 4129
+    },
+    {
+      "epoch": 14.75,
+      "grad_norm": 0.8226311988954756,
+      "learning_rate": 6.802940762392605e-06,
+      "loss": 0.0296,
+      "step": 4130
+    },
+    {
+      "epoch": 14.753571428571428,
+      "grad_norm": 1.5422062275328223,
+      "learning_rate": 6.794251599521591e-06,
+      "loss": 0.0694,
+      "step": 4131
+    },
+    {
+      "epoch": 14.757142857142856,
+      "grad_norm": 0.6083348479800497,
+      "learning_rate": 6.785566853811572e-06,
+      "loss": 0.0226,
+      "step": 4132
+    },
+    {
+      "epoch": 14.760714285714286,
+      "grad_norm": 1.896071547768543,
+      "learning_rate": 6.7768865281675055e-06,
+      "loss": 0.0473,
+      "step": 4133
+    },
+    {
+      "epoch": 14.764285714285714,
+      "grad_norm": 0.47271607062058774,
+      "learning_rate": 6.768210625492842e-06,
+      "loss": 0.0149,
+      "step": 4134
+    },
+    {
+      "epoch": 14.767857142857142,
+      "grad_norm": 0.6563509397987902,
+      "learning_rate": 6.759539148689573e-06,
+      "loss": 0.0337,
+      "step": 4135
+    },
+    {
+      "epoch": 14.771428571428572,
+      "grad_norm": 1.1014494817878215,
+      "learning_rate": 6.750872100658195e-06,
+      "loss": 0.0553,
+      "step": 4136
+    },
+    {
+      "epoch": 14.775,
+      "grad_norm": 2.718625075043433,
+      "learning_rate": 6.742209484297735e-06,
+      "loss": 0.038,
+      "step": 4137
+    },
+    {
+      "epoch": 14.778571428571428,
+      "grad_norm": 1.2085589269236783,
+      "learning_rate": 6.7335513025057294e-06,
+      "loss": 0.0585,
+      "step": 4138
+    },
+    {
+      "epoch": 14.782142857142857,
+      "grad_norm": 1.6358057580523073,
+      "learning_rate": 6.72489755817823e-06,
+      "loss": 0.0512,
+      "step": 4139
+    },
+    {
+      "epoch": 14.785714285714286,
+      "grad_norm": 1.4622729260904674,
+      "learning_rate": 6.716248254209816e-06,
+      "loss": 0.0588,
+      "step": 4140
+    },
+    {
+      "epoch": 14.789285714285715,
+      "grad_norm": 1.2461246293077473,
+      "learning_rate": 6.707603393493563e-06,
+      "loss": 0.0457,
+      "step": 4141
+    },
+    {
+      "epoch": 14.792857142857143,
+      "grad_norm": 1.2031483437004367,
+      "learning_rate": 6.698962978921084e-06,
+      "loss": 0.0391,
+      "step": 4142
+    },
+    {
+      "epoch": 14.79642857142857,
+      "grad_norm": 1.069666819079905,
+      "learning_rate": 6.690327013382478e-06,
+      "loss": 0.0334,
+      "step": 4143
+    },
+    {
+      "epoch": 14.8,
+      "grad_norm": 0.7253595647397953,
+      "learning_rate": 6.681695499766383e-06,
+      "loss": 0.0295,
+      "step": 4144
+    },
+    {
+      "epoch": 14.803571428571429,
+      "grad_norm": 0.6527234335215744,
+      "learning_rate": 6.6730684409599225e-06,
+      "loss": 0.0216,
+      "step": 4145
+    },
+    {
+      "epoch": 14.807142857142857,
+      "grad_norm": 0.8850745376035565,
+      "learning_rate": 6.664445839848754e-06,
+      "loss": 0.0231,
+      "step": 4146
+    },
+    {
+      "epoch": 14.810714285714285,
+      "grad_norm": 0.797827711001986,
+      "learning_rate": 6.655827699317019e-06,
+      "loss": 0.02,
+      "step": 4147
+    },
+    {
+      "epoch": 14.814285714285715,
+      "grad_norm": 1.7356433909144748,
+      "learning_rate": 6.6472140222473945e-06,
+      "loss": 0.0457,
+      "step": 4148
+    },
+    {
+      "epoch": 14.817857142857143,
+      "grad_norm": 1.5576080101887833,
+      "learning_rate": 6.6386048115210346e-06,
+      "loss": 0.0389,
+      "step": 4149
+    },
+    {
+      "epoch": 14.821428571428571,
+      "grad_norm": 0.8247931637569824,
+      "learning_rate": 6.630000070017628e-06,
+      "loss": 0.052,
+      "step": 4150
+    },
+    {
+      "epoch": 14.825,
+      "grad_norm": 0.9007472586261556,
+      "learning_rate": 6.621399800615349e-06,
+      "loss": 0.0254,
+      "step": 4151
+    },
+    {
+      "epoch": 14.82857142857143,
+      "grad_norm": 1.2470142944512836,
+      "learning_rate": 6.612804006190876e-06,
+      "loss": 0.0568,
+      "step": 4152
+    },
+    {
+      "epoch": 14.832142857142857,
+      "grad_norm": 1.1357646991773418,
+      "learning_rate": 6.6042126896194115e-06,
+      "loss": 0.0296,
+      "step": 4153
+    },
+    {
+      "epoch": 14.835714285714285,
+      "grad_norm": 1.081855347452007,
+      "learning_rate": 6.5956258537746325e-06,
+      "loss": 0.046,
+      "step": 4154
+    },
+    {
+      "epoch": 14.839285714285714,
+      "grad_norm": 1.8071376553725393,
+      "learning_rate": 6.587043501528738e-06,
+      "loss": 0.0533,
+      "step": 4155
+    },
+    {
+      "epoch": 14.842857142857143,
+      "grad_norm": 0.8839092011622246,
+      "learning_rate": 6.578465635752413e-06,
+      "loss": 0.0297,
+      "step": 4156
+    },
+    {
+      "epoch": 14.846428571428572,
+      "grad_norm": 1.9292680404155005,
+      "learning_rate": 6.56989225931486e-06,
+      "loss": 0.0599,
+      "step": 4157
+    },
+    {
+      "epoch": 14.85,
+      "grad_norm": 1.2172018945360665,
+      "learning_rate": 6.561323375083752e-06,
+      "loss": 0.0681,
+      "step": 4158
+    },
+    {
+      "epoch": 14.853571428571428,
+      "grad_norm": 1.1497268114540726,
+      "learning_rate": 6.552758985925287e-06,
+      "loss": 0.0487,
+      "step": 4159
+    },
+    {
+      "epoch": 14.857142857142858,
+      "grad_norm": 0.6919105647609916,
+      "learning_rate": 6.544199094704153e-06,
+      "loss": 0.0312,
+      "step": 4160
+    },
+    {
+      "epoch": 14.860714285714286,
+      "grad_norm": 0.9367373844139334,
+      "learning_rate": 6.5356437042835145e-06,
+      "loss": 0.0183,
+      "step": 4161
+    },
+    {
+      "epoch": 14.864285714285714,
+      "grad_norm": 1.742287139244388,
+      "learning_rate": 6.527092817525058e-06,
+      "loss": 0.0435,
+      "step": 4162
+    },
+    {
+      "epoch": 14.867857142857144,
+      "grad_norm": 0.5367591714190931,
+      "learning_rate": 6.5185464372889375e-06,
+      "loss": 0.0288,
+      "step": 4163
+    },
+    {
+      "epoch": 14.871428571428572,
+      "grad_norm": 0.9318846154656314,
+      "learning_rate": 6.510004566433825e-06,
+      "loss": 0.0257,
+      "step": 4164
+    },
+    {
+      "epoch": 14.875,
+      "grad_norm": 1.3576241788435788,
+      "learning_rate": 6.501467207816863e-06,
+      "loss": 0.0301,
+      "step": 4165
+    },
+    {
+      "epoch": 14.878571428571428,
+      "grad_norm": 0.8624662726354021,
+      "learning_rate": 6.492934364293699e-06,
+      "loss": 0.0394,
+      "step": 4166
+    },
+    {
+      "epoch": 14.882142857142856,
+      "grad_norm": 0.5207881989207346,
+      "learning_rate": 6.484406038718456e-06,
+      "loss": 0.0144,
+      "step": 4167
+    },
+    {
+      "epoch": 14.885714285714286,
+      "grad_norm": 0.4727764911096665,
+      "learning_rate": 6.475882233943769e-06,
+      "loss": 0.0202,
+      "step": 4168
+    },
+    {
+      "epoch": 14.889285714285714,
+      "grad_norm": 0.47475158791394356,
+      "learning_rate": 6.467362952820735e-06,
+      "loss": 0.0148,
+      "step": 4169
+    },
+    {
+      "epoch": 14.892857142857142,
+      "grad_norm": 1.7371705605471361,
+      "learning_rate": 6.458848198198946e-06,
+      "loss": 0.0428,
+      "step": 4170
+    },
+    {
+      "epoch": 14.896428571428572,
+      "grad_norm": 0.9225558440404791,
+      "learning_rate": 6.450337972926495e-06,
+      "loss": 0.0261,
+      "step": 4171
+    },
+    {
+      "epoch": 14.9,
+      "grad_norm": 0.8242904484795991,
+      "learning_rate": 6.4418322798499355e-06,
+      "loss": 0.0183,
+      "step": 4172
+    },
+    {
+      "epoch": 14.903571428571428,
+      "grad_norm": 0.9253583395570607,
+      "learning_rate": 6.43333112181433e-06,
+      "loss": 0.033,
+      "step": 4173
+    },
+    {
+      "epoch": 14.907142857142857,
+      "grad_norm": 0.9084640123094445,
+      "learning_rate": 6.424834501663199e-06,
+      "loss": 0.04,
+      "step": 4174
+    },
+    {
+      "epoch": 14.910714285714286,
+      "grad_norm": 1.5447739407933814,
+      "learning_rate": 6.416342422238571e-06,
+      "loss": 0.0301,
+      "step": 4175
+    },
+    {
+      "epoch": 14.914285714285715,
+      "grad_norm": 0.8397670935249396,
+      "learning_rate": 6.407854886380929e-06,
+      "loss": 0.0299,
+      "step": 4176
+    },
+    {
+      "epoch": 14.917857142857143,
+      "grad_norm": 1.3664064432771632,
+      "learning_rate": 6.399371896929263e-06,
+      "loss": 0.034,
+      "step": 4177
+    },
+    {
+      "epoch": 14.92142857142857,
+      "grad_norm": 0.5844081065816753,
+      "learning_rate": 6.3908934567210165e-06,
+      "loss": 0.0111,
+      "step": 4178
+    },
+    {
+      "epoch": 14.925,
+      "grad_norm": 0.7941701494140625,
+      "learning_rate": 6.382419568592135e-06,
+      "loss": 0.0346,
+      "step": 4179
+    },
+    {
+      "epoch": 14.928571428571429,
+      "grad_norm": 0.9644365353855104,
+      "learning_rate": 6.37395023537702e-06,
+      "loss": 0.0564,
+      "step": 4180
+    },
+    {
+      "epoch": 14.932142857142857,
+      "grad_norm": 0.6392124843222944,
+      "learning_rate": 6.365485459908572e-06,
+      "loss": 0.0099,
+      "step": 4181
+    },
+    {
+      "epoch": 14.935714285714285,
+      "grad_norm": 0.7349756212798939,
+      "learning_rate": 6.357025245018147e-06,
+      "loss": 0.0176,
+      "step": 4182
+    },
+    {
+      "epoch": 14.939285714285715,
+      "grad_norm": 1.2493396104688945,
+      "learning_rate": 6.348569593535576e-06,
+      "loss": 0.0379,
+      "step": 4183
+    },
+    {
+      "epoch": 14.942857142857143,
+      "grad_norm": 1.5241560067667912,
+      "learning_rate": 6.340118508289186e-06,
+      "loss": 0.0473,
+      "step": 4184
+    },
+    {
+      "epoch": 14.946428571428571,
+      "grad_norm": 1.3059520935139821,
+      "learning_rate": 6.331671992105748e-06,
+      "loss": 0.0428,
+      "step": 4185
+    },
+    {
+      "epoch": 14.95,
+      "grad_norm": 0.8707959710623371,
+      "learning_rate": 6.323230047810529e-06,
+      "loss": 0.0371,
+      "step": 4186
+    },
+    {
+      "epoch": 14.95357142857143,
+      "grad_norm": 1.0590934088609378,
+      "learning_rate": 6.314792678227244e-06,
+      "loss": 0.0411,
+      "step": 4187
+    },
+    {
+      "epoch": 14.957142857142857,
+      "grad_norm": 0.7051287762727501,
+      "learning_rate": 6.306359886178104e-06,
+      "loss": 0.03,
+      "step": 4188
+    },
+    {
+      "epoch": 14.960714285714285,
+      "grad_norm": 1.6575160765551484,
+      "learning_rate": 6.2979316744837615e-06,
+      "loss": 0.065,
+      "step": 4189
+    },
+    {
+      "epoch": 14.964285714285714,
+      "grad_norm": 1.3033348408962144,
+      "learning_rate": 6.289508045963362e-06,
+      "loss": 0.0244,
+      "step": 4190
+    },
+    {
+      "epoch": 14.967857142857143,
+      "grad_norm": 0.6608817262569924,
+      "learning_rate": 6.281089003434493e-06,
+      "loss": 0.0398,
+      "step": 4191
+    },
+    {
+      "epoch": 14.971428571428572,
+      "grad_norm": 1.2197637739376985,
+      "learning_rate": 6.272674549713236e-06,
+      "loss": 0.0327,
+      "step": 4192
+    },
+    {
+      "epoch": 14.975,
+      "grad_norm": 0.5367943566836677,
+      "learning_rate": 6.264264687614112e-06,
+      "loss": 0.0196,
+      "step": 4193
+    },
+    {
+      "epoch": 14.978571428571428,
+      "grad_norm": 0.9890294065520959,
+      "learning_rate": 6.255859419950114e-06,
+      "loss": 0.0445,
+      "step": 4194
+    },
+    {
+      "epoch": 14.982142857142858,
+      "grad_norm": 0.6886581714008342,
+      "learning_rate": 6.2474587495327135e-06,
+      "loss": 0.0342,
+      "step": 4195
+    },
+    {
+      "epoch": 14.985714285714286,
+      "grad_norm": 0.7589822668025668,
+      "learning_rate": 6.239062679171816e-06,
+      "loss": 0.0431,
+      "step": 4196
+    },
+    {
+      "epoch": 14.989285714285714,
+      "grad_norm": 0.7928219241476149,
+      "learning_rate": 6.230671211675821e-06,
+      "loss": 0.0279,
+      "step": 4197
+    },
+    {
+      "epoch": 14.992857142857144,
+      "grad_norm": 0.5467952514726666,
+      "learning_rate": 6.222284349851557e-06,
+      "loss": 0.0111,
+      "step": 4198
+    },
+    {
+      "epoch": 14.996428571428572,
+      "grad_norm": 0.7893726450545717,
+      "learning_rate": 6.2139020965043386e-06,
+      "loss": 0.0489,
+      "step": 4199
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 0.935342288867424,
+      "learning_rate": 6.2055244544379145e-06,
+      "loss": 0.0372,
+      "step": 4200
+    },
+    {
+      "epoch": 15.003571428571428,
+      "grad_norm": 0.5444191644236142,
+      "learning_rate": 6.197151426454517e-06,
+      "loss": 0.0176,
+      "step": 4201
+    },
+    {
+      "epoch": 15.007142857142858,
+      "grad_norm": 1.0283236804682396,
+      "learning_rate": 6.188783015354807e-06,
+      "loss": 0.0503,
+      "step": 4202
+    },
+    {
+      "epoch": 15.010714285714286,
+      "grad_norm": 0.8296667767636738,
+      "learning_rate": 6.18041922393793e-06,
+      "loss": 0.0286,
+      "step": 4203
+    },
+    {
+      "epoch": 15.014285714285714,
+      "grad_norm": 0.812005231602574,
+      "learning_rate": 6.172060055001456e-06,
+      "loss": 0.0183,
+      "step": 4204
+    },
+    {
+      "epoch": 15.017857142857142,
+      "grad_norm": 0.4450270624669421,
+      "learning_rate": 6.1637055113414425e-06,
+      "loss": 0.0134,
+      "step": 4205
+    },
+    {
+      "epoch": 15.021428571428572,
+      "grad_norm": 0.42624312510507784,
+      "learning_rate": 6.155355595752371e-06,
+      "loss": 0.0118,
+      "step": 4206
+    },
+    {
+      "epoch": 15.025,
+      "grad_norm": 0.6853793265863115,
+      "learning_rate": 6.147010311027182e-06,
+      "loss": 0.0283,
+      "step": 4207
+    },
+    {
+      "epoch": 15.028571428571428,
+      "grad_norm": 0.4888403415297945,
+      "learning_rate": 6.138669659957281e-06,
+      "loss": 0.0314,
+      "step": 4208
+    },
+    {
+      "epoch": 15.032142857142857,
+      "grad_norm": 1.1799549639768767,
+      "learning_rate": 6.130333645332502e-06,
+      "loss": 0.0587,
+      "step": 4209
+    },
+    {
+      "epoch": 15.035714285714286,
+      "grad_norm": 1.058587097132987,
+      "learning_rate": 6.122002269941154e-06,
+      "loss": 0.0247,
+      "step": 4210
+    },
+    {
+      "epoch": 15.039285714285715,
+      "grad_norm": 1.2387952863912608,
+      "learning_rate": 6.113675536569967e-06,
+      "loss": 0.053,
+      "step": 4211
+    },
+    {
+      "epoch": 15.042857142857143,
+      "grad_norm": 1.0465939893250278,
+      "learning_rate": 6.105353448004143e-06,
+      "loss": 0.0644,
+      "step": 4212
+    },
+    {
+      "epoch": 15.04642857142857,
+      "grad_norm": 0.9658246154154129,
+      "learning_rate": 6.097036007027306e-06,
+      "loss": 0.0193,
+      "step": 4213
+    },
+    {
+      "epoch": 15.05,
+      "grad_norm": 0.8272127583508203,
+      "learning_rate": 6.088723216421551e-06,
+      "loss": 0.0184,
+      "step": 4214
+    },
+    {
+      "epoch": 15.053571428571429,
+      "grad_norm": 0.9260432754286021,
+      "learning_rate": 6.080415078967392e-06,
+      "loss": 0.0372,
+      "step": 4215
+    },
+    {
+      "epoch": 15.057142857142857,
+      "grad_norm": 1.49695767955174,
+      "learning_rate": 6.072111597443811e-06,
+      "loss": 0.0259,
+      "step": 4216
+    },
+    {
+      "epoch": 15.060714285714285,
+      "grad_norm": 1.1557268039520703,
+      "learning_rate": 6.063812774628208e-06,
+      "loss": 0.0427,
+      "step": 4217
+    },
+    {
+      "epoch": 15.064285714285715,
+      "grad_norm": 2.395526248512596,
+      "learning_rate": 6.0555186132964534e-06,
+      "loss": 0.0611,
+      "step": 4218
+    },
+    {
+      "epoch": 15.067857142857143,
+      "grad_norm": 1.3127546916578094,
+      "learning_rate": 6.047229116222833e-06,
+      "loss": 0.0524,
+      "step": 4219
+    },
+    {
+      "epoch": 15.071428571428571,
+      "grad_norm": 0.6159789827504456,
+      "learning_rate": 6.038944286180075e-06,
+      "loss": 0.02,
+      "step": 4220
+    },
+    {
+      "epoch": 15.075,
+      "grad_norm": 0.6934164967311914,
+      "learning_rate": 6.03066412593937e-06,
+      "loss": 0.031,
+      "step": 4221
+    },
+    {
+      "epoch": 15.07857142857143,
+      "grad_norm": 1.2478934493701919,
+      "learning_rate": 6.022388638270313e-06,
+      "loss": 0.06,
+      "step": 4222
+    },
+    {
+      "epoch": 15.082142857142857,
+      "grad_norm": 1.068899481735934,
+      "learning_rate": 6.014117825940968e-06,
+      "loss": 0.0297,
+      "step": 4223
+    },
+    {
+      "epoch": 15.085714285714285,
+      "grad_norm": 0.7387202458226113,
+      "learning_rate": 6.00585169171781e-06,
+      "loss": 0.0256,
+      "step": 4224
+    },
+    {
+      "epoch": 15.089285714285714,
+      "grad_norm": 1.0308042465349634,
+      "learning_rate": 5.9975902383657695e-06,
+      "loss": 0.0359,
+      "step": 4225
+    },
+    {
+      "epoch": 15.092857142857143,
+      "grad_norm": 0.6912967851918572,
+      "learning_rate": 5.989333468648186e-06,
+      "loss": 0.0257,
+      "step": 4226
+    },
+    {
+      "epoch": 15.096428571428572,
+      "grad_norm": 0.7764527227633871,
+      "learning_rate": 5.981081385326868e-06,
+      "loss": 0.0381,
+      "step": 4227
+    },
+    {
+      "epoch": 15.1,
+      "grad_norm": 1.1744859551226867,
+      "learning_rate": 5.972833991162017e-06,
+      "loss": 0.0305,
+      "step": 4228
+    },
+    {
+      "epoch": 15.103571428571428,
+      "grad_norm": 0.6385381812771718,
+      "learning_rate": 5.9645912889123e-06,
+      "loss": 0.0215,
+      "step": 4229
+    },
+    {
+      "epoch": 15.107142857142858,
+      "grad_norm": 0.5292446307375084,
+      "learning_rate": 5.956353281334792e-06,
+      "loss": 0.0175,
+      "step": 4230
+    },
+    {
+      "epoch": 15.110714285714286,
+      "grad_norm": 1.0965000536338407,
+      "learning_rate": 5.948119971185004e-06,
+      "loss": 0.0142,
+      "step": 4231
+    },
+    {
+      "epoch": 15.114285714285714,
+      "grad_norm": 1.2919343763689584,
+      "learning_rate": 5.939891361216885e-06,
+      "loss": 0.0279,
+      "step": 4232
+    },
+    {
+      "epoch": 15.117857142857142,
+      "grad_norm": 1.4196328924416621,
+      "learning_rate": 5.931667454182794e-06,
+      "loss": 0.0421,
+      "step": 4233
+    },
+    {
+      "epoch": 15.121428571428572,
+      "grad_norm": 1.118509009097837,
+      "learning_rate": 5.9234482528335414e-06,
+      "loss": 0.0427,
+      "step": 4234
+    },
+    {
+      "epoch": 15.125,
+      "grad_norm": 0.9486684826800291,
+      "learning_rate": 5.915233759918333e-06,
+      "loss": 0.033,
+      "step": 4235
+    },
+    {
+      "epoch": 15.128571428571428,
+      "grad_norm": 1.387381923121419,
+      "learning_rate": 5.907023978184829e-06,
+      "loss": 0.0229,
+      "step": 4236
+    },
+    {
+      "epoch": 15.132142857142858,
+      "grad_norm": 0.6862779082613656,
+      "learning_rate": 5.898818910379092e-06,
+      "loss": 0.0183,
+      "step": 4237
+    },
+    {
+      "epoch": 15.135714285714286,
+      "grad_norm": 1.0281360285959713,
+      "learning_rate": 5.890618559245624e-06,
+      "loss": 0.0373,
+      "step": 4238
+    },
+    {
+      "epoch": 15.139285714285714,
+      "grad_norm": 0.6008892866216341,
+      "learning_rate": 5.8824229275273335e-06,
+      "loss": 0.0306,
+      "step": 4239
+    },
+    {
+      "epoch": 15.142857142857142,
+      "grad_norm": 0.6363833979931013,
+      "learning_rate": 5.8742320179655684e-06,
+      "loss": 0.023,
+      "step": 4240
+    },
+    {
+      "epoch": 15.146428571428572,
+      "grad_norm": 1.172668980775189,
+      "learning_rate": 5.866045833300078e-06,
+      "loss": 0.0262,
+      "step": 4241
+    },
+    {
+      "epoch": 15.15,
+      "grad_norm": 0.6746708289314586,
+      "learning_rate": 5.857864376269051e-06,
+      "loss": 0.0299,
+      "step": 4242
+    },
+    {
+      "epoch": 15.153571428571428,
+      "grad_norm": 1.3100684072636664,
+      "learning_rate": 5.849687649609079e-06,
+      "loss": 0.0555,
+      "step": 4243
+    },
+    {
+      "epoch": 15.157142857142857,
+      "grad_norm": 0.7417415788243072,
+      "learning_rate": 5.841515656055172e-06,
+      "loss": 0.0066,
+      "step": 4244
+    },
+    {
+      "epoch": 15.160714285714286,
+      "grad_norm": 1.0012867847491018,
+      "learning_rate": 5.8333483983407725e-06,
+      "loss": 0.0348,
+      "step": 4245
+    },
+    {
+      "epoch": 15.164285714285715,
+      "grad_norm": 1.0076109611650121,
+      "learning_rate": 5.825185879197714e-06,
+      "loss": 0.022,
+      "step": 4246
+    },
+    {
+      "epoch": 15.167857142857143,
+      "grad_norm": 0.4005458389922747,
+      "learning_rate": 5.8170281013562745e-06,
+      "loss": 0.0115,
+      "step": 4247
+    },
+    {
+      "epoch": 15.17142857142857,
+      "grad_norm": 1.4902749610701211,
+      "learning_rate": 5.808875067545119e-06,
+      "loss": 0.0242,
+      "step": 4248
+    },
+    {
+      "epoch": 15.175,
+      "grad_norm": 0.9061106063870389,
+      "learning_rate": 5.800726780491348e-06,
+      "loss": 0.046,
+      "step": 4249
+    },
+    {
+      "epoch": 15.178571428571429,
+      "grad_norm": 1.4904302540806655,
+      "learning_rate": 5.792583242920449e-06,
+      "loss": 0.0607,
+      "step": 4250
+    },
+    {
+      "epoch": 15.182142857142857,
+      "grad_norm": 0.7889641403411009,
+      "learning_rate": 5.784444457556353e-06,
+      "loss": 0.015,
+      "step": 4251
+    },
+    {
+      "epoch": 15.185714285714285,
+      "grad_norm": 1.5507644371951812,
+      "learning_rate": 5.776310427121368e-06,
+      "loss": 0.023,
+      "step": 4252
+    },
+    {
+      "epoch": 15.189285714285715,
+      "grad_norm": 0.532755964778624,
+      "learning_rate": 5.768181154336238e-06,
+      "loss": 0.0172,
+      "step": 4253
+    },
+    {
+      "epoch": 15.192857142857143,
+      "grad_norm": 0.9957332342016993,
+      "learning_rate": 5.7600566419200996e-06,
+      "loss": 0.0441,
+      "step": 4254
+    },
+    {
+      "epoch": 15.196428571428571,
+      "grad_norm": 0.8057489664321906,
+      "learning_rate": 5.75193689259051e-06,
+      "loss": 0.0267,
+      "step": 4255
+    },
+    {
+      "epoch": 15.2,
+      "grad_norm": 0.569758264359997,
+      "learning_rate": 5.7438219090634205e-06,
+      "loss": 0.0121,
+      "step": 4256
+    },
+    {
+      "epoch": 15.20357142857143,
+      "grad_norm": 1.3007550461895387,
+      "learning_rate": 5.7357116940531855e-06,
+      "loss": 0.041,
+      "step": 4257
+    },
+    {
+      "epoch": 15.207142857142857,
+      "grad_norm": 1.367837536562525,
+      "learning_rate": 5.72760625027259e-06,
+      "loss": 0.0523,
+      "step": 4258
+    },
+    {
+      "epoch": 15.210714285714285,
+      "grad_norm": 1.0656390877596356,
+      "learning_rate": 5.719505580432789e-06,
+      "loss": 0.0365,
+      "step": 4259
+    },
+    {
+      "epoch": 15.214285714285714,
+      "grad_norm": 0.6531112646318133,
+      "learning_rate": 5.71140968724337e-06,
+      "loss": 0.045,
+      "step": 4260
+    },
+    {
+      "epoch": 15.217857142857143,
+      "grad_norm": 0.7919464531745223,
+      "learning_rate": 5.7033185734123e-06,
+      "loss": 0.0169,
+      "step": 4261
+    },
+    {
+      "epoch": 15.221428571428572,
+      "grad_norm": 0.9503202230059469,
+      "learning_rate": 5.695232241645967e-06,
+      "loss": 0.0306,
+      "step": 4262
+    },
+    {
+      "epoch": 15.225,
+      "grad_norm": 0.875445710065734,
+      "learning_rate": 5.687150694649139e-06,
+      "loss": 0.0326,
+      "step": 4263
+    },
+    {
+      "epoch": 15.228571428571428,
+      "grad_norm": 0.8219838696731596,
+      "learning_rate": 5.679073935125008e-06,
+      "loss": 0.0226,
+      "step": 4264
+    },
+    {
+      "epoch": 15.232142857142858,
+      "grad_norm": 0.9751873102887382,
+      "learning_rate": 5.671001965775136e-06,
+      "loss": 0.0284,
+      "step": 4265
+    },
+    {
+      "epoch": 15.235714285714286,
+      "grad_norm": 1.0025595099754925,
+      "learning_rate": 5.662934789299512e-06,
+      "loss": 0.052,
+      "step": 4266
+    },
+    {
+      "epoch": 15.239285714285714,
+      "grad_norm": 1.0611690967134635,
+      "learning_rate": 5.654872408396501e-06,
+      "loss": 0.0348,
+      "step": 4267
+    },
+    {
+      "epoch": 15.242857142857142,
+      "grad_norm": 1.4928947370187475,
+      "learning_rate": 5.646814825762866e-06,
+      "loss": 0.0451,
+      "step": 4268
+    },
+    {
+      "epoch": 15.246428571428572,
+      "grad_norm": 0.543444928801803,
+      "learning_rate": 5.638762044093778e-06,
+      "loss": 0.0192,
+      "step": 4269
+    },
+    {
+      "epoch": 15.25,
+      "grad_norm": 0.6474655673506455,
+      "learning_rate": 5.630714066082785e-06,
+      "loss": 0.0116,
+      "step": 4270
+    },
+    {
+      "epoch": 15.253571428571428,
+      "grad_norm": 0.8204486068423901,
+      "learning_rate": 5.622670894421849e-06,
+      "loss": 0.0191,
+      "step": 4271
+    },
+    {
+      "epoch": 15.257142857142856,
+      "grad_norm": 2.2437699868218877,
+      "learning_rate": 5.614632531801298e-06,
+      "loss": 0.0496,
+      "step": 4272
+    },
+    {
+      "epoch": 15.260714285714286,
+      "grad_norm": 1.053536000194414,
+      "learning_rate": 5.606598980909881e-06,
+      "loss": 0.0275,
+      "step": 4273
+    },
+    {
+      "epoch": 15.264285714285714,
+      "grad_norm": 0.5085275664689646,
+      "learning_rate": 5.598570244434707e-06,
+      "loss": 0.0172,
+      "step": 4274
+    },
+    {
+      "epoch": 15.267857142857142,
+      "grad_norm": 0.9679012507394454,
+      "learning_rate": 5.5905463250613035e-06,
+      "loss": 0.0349,
+      "step": 4275
+    },
+    {
+      "epoch": 15.271428571428572,
+      "grad_norm": 0.6567112835578323,
+      "learning_rate": 5.582527225473562e-06,
+      "loss": 0.0163,
+      "step": 4276
+    },
+    {
+      "epoch": 15.275,
+      "grad_norm": 0.689612356883945,
+      "learning_rate": 5.574512948353783e-06,
+      "loss": 0.0102,
+      "step": 4277
+    },
+    {
+      "epoch": 15.278571428571428,
+      "grad_norm": 1.4549002927561445,
+      "learning_rate": 5.5665034963826335e-06,
+      "loss": 0.0609,
+      "step": 4278
+    },
+    {
+      "epoch": 15.282142857142857,
+      "grad_norm": 1.3437588733956067,
+      "learning_rate": 5.558498872239191e-06,
+      "loss": 0.0283,
+      "step": 4279
+    },
+    {
+      "epoch": 15.285714285714286,
+      "grad_norm": 0.8480542163626382,
+      "learning_rate": 5.550499078600895e-06,
+      "loss": 0.0217,
+      "step": 4280
+    },
+    {
+      "epoch": 15.289285714285715,
+      "grad_norm": 1.253496427090327,
+      "learning_rate": 5.542504118143573e-06,
+      "loss": 0.0964,
+      "step": 4281
+    },
+    {
+      "epoch": 15.292857142857143,
+      "grad_norm": 0.7887336890063723,
+      "learning_rate": 5.534513993541453e-06,
+      "loss": 0.0259,
+      "step": 4282
+    },
+    {
+      "epoch": 15.29642857142857,
+      "grad_norm": 0.5532033467437123,
+      "learning_rate": 5.526528707467122e-06,
+      "loss": 0.0076,
+      "step": 4283
+    },
+    {
+      "epoch": 15.3,
+      "grad_norm": 1.6364850614869264,
+      "learning_rate": 5.518548262591574e-06,
+      "loss": 0.0313,
+      "step": 4284
+    },
+    {
+      "epoch": 15.303571428571429,
+      "grad_norm": 2.012006377055822,
+      "learning_rate": 5.510572661584157e-06,
+      "loss": 0.0501,
+      "step": 4285
+    },
+    {
+      "epoch": 15.307142857142857,
+      "grad_norm": 0.7249035393906462,
+      "learning_rate": 5.502601907112624e-06,
+      "loss": 0.0165,
+      "step": 4286
+    },
+    {
+      "epoch": 15.310714285714285,
+      "grad_norm": 1.301110653544984,
+      "learning_rate": 5.494636001843081e-06,
+      "loss": 0.0328,
+      "step": 4287
+    },
+    {
+      "epoch": 15.314285714285715,
+      "grad_norm": 1.3011896566063812,
+      "learning_rate": 5.486674948440038e-06,
+      "loss": 0.0313,
+      "step": 4288
+    },
+    {
+      "epoch": 15.317857142857143,
+      "grad_norm": 0.5192677110215158,
+      "learning_rate": 5.478718749566361e-06,
+      "loss": 0.0103,
+      "step": 4289
+    },
+    {
+      "epoch": 15.321428571428571,
+      "grad_norm": 1.3261397257610374,
+      "learning_rate": 5.470767407883311e-06,
+      "loss": 0.0253,
+      "step": 4290
+    },
+    {
+      "epoch": 15.325,
+      "grad_norm": 1.0078052369861448,
+      "learning_rate": 5.4628209260505004e-06,
+      "loss": 0.026,
+      "step": 4291
+    },
+    {
+      "epoch": 15.32857142857143,
+      "grad_norm": 0.6491416067423474,
+      "learning_rate": 5.454879306725944e-06,
+      "loss": 0.0226,
+      "step": 4292
+    },
+    {
+      "epoch": 15.332142857142857,
+      "grad_norm": 1.5187785255799466,
+      "learning_rate": 5.44694255256601e-06,
+      "loss": 0.0293,
+      "step": 4293
+    },
+    {
+      "epoch": 15.335714285714285,
+      "grad_norm": 1.0986052128186057,
+      "learning_rate": 5.439010666225439e-06,
+      "loss": 0.0339,
+      "step": 4294
+    },
+    {
+      "epoch": 15.339285714285714,
+      "grad_norm": 0.31164791867447666,
+      "learning_rate": 5.4310836503573625e-06,
+      "loss": 0.0093,
+      "step": 4295
+    },
+    {
+      "epoch": 15.342857142857143,
+      "grad_norm": 0.6130922669926,
+      "learning_rate": 5.42316150761326e-06,
+      "loss": 0.0261,
+      "step": 4296
+    },
+    {
+      "epoch": 15.346428571428572,
+      "grad_norm": 1.3324582528152582,
+      "learning_rate": 5.415244240642996e-06,
+      "loss": 0.0365,
+      "step": 4297
+    },
+    {
+      "epoch": 15.35,
+      "grad_norm": 0.5183997796441783,
+      "learning_rate": 5.407331852094795e-06,
+      "loss": 0.0146,
+      "step": 4298
+    },
+    {
+      "epoch": 15.353571428571428,
+      "grad_norm": 1.7839354964308878,
+      "learning_rate": 5.3994243446152624e-06,
+      "loss": 0.0673,
+      "step": 4299
+    },
+    {
+      "epoch": 15.357142857142858,
+      "grad_norm": 0.7754846631908334,
+      "learning_rate": 5.391521720849353e-06,
+      "loss": 0.0233,
+      "step": 4300
+    },
+    {
+      "epoch": 15.360714285714286,
+      "grad_norm": 1.155935936316608,
+      "learning_rate": 5.383623983440407e-06,
+      "loss": 0.0448,
+      "step": 4301
+    },
+    {
+      "epoch": 15.364285714285714,
+      "grad_norm": 0.9045127439520891,
+      "learning_rate": 5.375731135030109e-06,
+      "loss": 0.0386,
+      "step": 4302
+    },
+    {
+      "epoch": 15.367857142857144,
+      "grad_norm": 1.9234474857881267,
+      "learning_rate": 5.367843178258534e-06,
+      "loss": 0.0608,
+      "step": 4303
+    },
+    {
+      "epoch": 15.371428571428572,
+      "grad_norm": 1.0682540579907451,
+      "learning_rate": 5.359960115764098e-06,
+      "loss": 0.0392,
+      "step": 4304
+    },
+    {
+      "epoch": 15.375,
+      "grad_norm": 0.40835973621328125,
+      "learning_rate": 5.352081950183585e-06,
+      "loss": 0.0116,
+      "step": 4305
+    },
+    {
+      "epoch": 15.378571428571428,
+      "grad_norm": 1.9326392164560302,
+      "learning_rate": 5.344208684152157e-06,
+      "loss": 0.025,
+      "step": 4306
+    },
+    {
+      "epoch": 15.382142857142856,
+      "grad_norm": 1.8092715717488594,
+      "learning_rate": 5.33634032030331e-06,
+      "loss": 0.0448,
+      "step": 4307
+    },
+    {
+      "epoch": 15.385714285714286,
+      "grad_norm": 1.7115950467036063,
+      "learning_rate": 5.328476861268932e-06,
+      "loss": 0.0295,
+      "step": 4308
+    },
+    {
+      "epoch": 15.389285714285714,
+      "grad_norm": 2.1936963113475834,
+      "learning_rate": 5.32061830967924e-06,
+      "loss": 0.0333,
+      "step": 4309
+    },
+    {
+      "epoch": 15.392857142857142,
+      "grad_norm": 0.8185306157739954,
+      "learning_rate": 5.312764668162831e-06,
+      "loss": 0.0309,
+      "step": 4310
+    },
+    {
+      "epoch": 15.396428571428572,
+      "grad_norm": 0.5126573442977819,
+      "learning_rate": 5.304915939346647e-06,
+      "loss": 0.0181,
+      "step": 4311
+    },
+    {
+      "epoch": 15.4,
+      "grad_norm": 0.6879451919609796,
+      "learning_rate": 5.297072125855998e-06,
+      "loss": 0.0257,
+      "step": 4312
+    },
+    {
+      "epoch": 15.403571428571428,
+      "grad_norm": 0.9390974677675614,
+      "learning_rate": 5.289233230314532e-06,
+      "loss": 0.0227,
+      "step": 4313
+    },
+    {
+      "epoch": 15.407142857142857,
+      "grad_norm": 1.0111456153446672,
+      "learning_rate": 5.281399255344277e-06,
+      "loss": 0.0175,
+      "step": 4314
+    },
+    {
+      "epoch": 15.410714285714286,
+      "grad_norm": 0.7421721208227556,
+      "learning_rate": 5.273570203565592e-06,
+      "loss": 0.0204,
+      "step": 4315
+    },
+    {
+      "epoch": 15.414285714285715,
+      "grad_norm": 1.3440606973021114,
+      "learning_rate": 5.265746077597207e-06,
+      "loss": 0.0667,
+      "step": 4316
+    },
+    {
+      "epoch": 15.417857142857143,
+      "grad_norm": 1.0363225029569285,
+      "learning_rate": 5.257926880056192e-06,
+      "loss": 0.0208,
+      "step": 4317
+    },
+    {
+      "epoch": 15.42142857142857,
+      "grad_norm": 1.1560196546398096,
+      "learning_rate": 5.250112613557965e-06,
+      "loss": 0.0559,
+      "step": 4318
+    },
+    {
+      "epoch": 15.425,
+      "grad_norm": 0.9463126407334292,
+      "learning_rate": 5.242303280716317e-06,
+      "loss": 0.0277,
+      "step": 4319
+    },
+    {
+      "epoch": 15.428571428571429,
+      "grad_norm": 0.8492498896536613,
+      "learning_rate": 5.23449888414336e-06,
+      "loss": 0.036,
+      "step": 4320
+    },
+    {
+      "epoch": 15.432142857142857,
+      "grad_norm": 0.5729425227248747,
+      "learning_rate": 5.226699426449582e-06,
+      "loss": 0.0224,
+      "step": 4321
+    },
+    {
+      "epoch": 15.435714285714285,
+      "grad_norm": 1.0711028226014987,
+      "learning_rate": 5.218904910243794e-06,
+      "loss": 0.0433,
+      "step": 4322
+    },
+    {
+      "epoch": 15.439285714285715,
+      "grad_norm": 0.3653415906089986,
+      "learning_rate": 5.211115338133179e-06,
+      "loss": 0.0114,
+      "step": 4323
+    },
+    {
+      "epoch": 15.442857142857143,
+      "grad_norm": 0.833465802447322,
+      "learning_rate": 5.20333071272324e-06,
+      "loss": 0.0307,
+      "step": 4324
+    },
+    {
+      "epoch": 15.446428571428571,
+      "grad_norm": 0.9420028263858004,
+      "learning_rate": 5.195551036617849e-06,
+      "loss": 0.038,
+      "step": 4325
+    },
+    {
+      "epoch": 15.45,
+      "grad_norm": 1.3777123001119875,
+      "learning_rate": 5.187776312419206e-06,
+      "loss": 0.0357,
+      "step": 4326
+    },
+    {
+      "epoch": 15.45357142857143,
+      "grad_norm": 1.080085450310553,
+      "learning_rate": 5.180006542727862e-06,
+      "loss": 0.0334,
+      "step": 4327
+    },
+    {
+      "epoch": 15.457142857142857,
+      "grad_norm": 0.6657115580926481,
+      "learning_rate": 5.172241730142717e-06,
+      "loss": 0.0233,
+      "step": 4328
+    },
+    {
+      "epoch": 15.460714285714285,
+      "grad_norm": 1.914795682576943,
+      "learning_rate": 5.164481877260994e-06,
+      "loss": 0.0808,
+      "step": 4329
+    },
+    {
+      "epoch": 15.464285714285714,
+      "grad_norm": 1.2013546465639862,
+      "learning_rate": 5.156726986678282e-06,
+      "loss": 0.0329,
+      "step": 4330
+    },
+    {
+      "epoch": 15.467857142857143,
+      "grad_norm": 0.8603038452568224,
+      "learning_rate": 5.148977060988483e-06,
+      "loss": 0.051,
+      "step": 4331
+    },
+    {
+      "epoch": 15.471428571428572,
+      "grad_norm": 0.9733125517191629,
+      "learning_rate": 5.141232102783862e-06,
+      "loss": 0.0406,
+      "step": 4332
+    },
+    {
+      "epoch": 15.475,
+      "grad_norm": 0.8404474072546432,
+      "learning_rate": 5.1334921146550055e-06,
+      "loss": 0.0356,
+      "step": 4333
+    },
+    {
+      "epoch": 15.478571428571428,
+      "grad_norm": 1.2926761288958337,
+      "learning_rate": 5.125757099190856e-06,
+      "loss": 0.027,
+      "step": 4334
+    },
+    {
+      "epoch": 15.482142857142858,
+      "grad_norm": 1.0391750839375773,
+      "learning_rate": 5.11802705897867e-06,
+      "loss": 0.0408,
+      "step": 4335
+    },
+    {
+      "epoch": 15.485714285714286,
+      "grad_norm": 0.7655811106408599,
+      "learning_rate": 5.110301996604052e-06,
+      "loss": 0.0207,
+      "step": 4336
+    },
+    {
+      "epoch": 15.489285714285714,
+      "grad_norm": 1.0320169362085172,
+      "learning_rate": 5.102581914650948e-06,
+      "loss": 0.0455,
+      "step": 4337
+    },
+    {
+      "epoch": 15.492857142857144,
+      "grad_norm": 0.44412101919598296,
+      "learning_rate": 5.0948668157016225e-06,
+      "loss": 0.0128,
+      "step": 4338
+    },
+    {
+      "epoch": 15.496428571428572,
+      "grad_norm": 0.5411915308224027,
+      "learning_rate": 5.087156702336689e-06,
+      "loss": 0.022,
+      "step": 4339
+    },
+    {
+      "epoch": 15.5,
+      "grad_norm": 1.4380420944332373,
+      "learning_rate": 5.079451577135079e-06,
+      "loss": 0.0411,
+      "step": 4340
+    },
+    {
+      "epoch": 15.503571428571428,
+      "grad_norm": 0.6763369101911993,
+      "learning_rate": 5.071751442674071e-06,
+      "loss": 0.026,
+      "step": 4341
+    },
+    {
+      "epoch": 15.507142857142856,
+      "grad_norm": 0.49499681286691893,
+      "learning_rate": 5.064056301529254e-06,
+      "loss": 0.0102,
+      "step": 4342
+    },
+    {
+      "epoch": 15.510714285714286,
+      "grad_norm": 1.5809167609746193,
+      "learning_rate": 5.05636615627457e-06,
+      "loss": 0.0596,
+      "step": 4343
+    },
+    {
+      "epoch": 15.514285714285714,
+      "grad_norm": 0.8629041623309527,
+      "learning_rate": 5.0486810094822706e-06,
+      "loss": 0.032,
+      "step": 4344
+    },
+    {
+      "epoch": 15.517857142857142,
+      "grad_norm": 1.9261433581813752,
+      "learning_rate": 5.041000863722949e-06,
+      "loss": 0.0282,
+      "step": 4345
+    },
+    {
+      "epoch": 15.521428571428572,
+      "grad_norm": 1.1833687656879424,
+      "learning_rate": 5.0333257215655116e-06,
+      "loss": 0.0268,
+      "step": 4346
+    },
+    {
+      "epoch": 15.525,
+      "grad_norm": 0.49974018668919096,
+      "learning_rate": 5.025655585577212e-06,
+      "loss": 0.0137,
+      "step": 4347
+    },
+    {
+      "epoch": 15.528571428571428,
+      "grad_norm": 1.1015598954113308,
+      "learning_rate": 5.017990458323607e-06,
+      "loss": 0.0253,
+      "step": 4348
+    },
+    {
+      "epoch": 15.532142857142857,
+      "grad_norm": 0.9332591411361861,
+      "learning_rate": 5.010330342368586e-06,
+      "loss": 0.0417,
+      "step": 4349
+    },
+    {
+      "epoch": 15.535714285714286,
+      "grad_norm": 0.5773828344349604,
+      "learning_rate": 5.002675240274373e-06,
+      "loss": 0.0254,
+      "step": 4350
+    },
+    {
+      "epoch": 15.539285714285715,
+      "grad_norm": 1.3266452279337326,
+      "learning_rate": 4.995025154601494e-06,
+      "loss": 0.035,
+      "step": 4351
+    },
+    {
+      "epoch": 15.542857142857143,
+      "grad_norm": 1.1881218840775112,
+      "learning_rate": 4.987380087908822e-06,
+      "loss": 0.0306,
+      "step": 4352
+    },
+    {
+      "epoch": 15.54642857142857,
+      "grad_norm": 0.8426670888171601,
+      "learning_rate": 4.979740042753527e-06,
+      "loss": 0.0362,
+      "step": 4353
+    },
+    {
+      "epoch": 15.55,
+      "grad_norm": 1.074285369888006,
+      "learning_rate": 4.97210502169112e-06,
+      "loss": 0.0495,
+      "step": 4354
+    },
+    {
+      "epoch": 15.553571428571429,
+      "grad_norm": 2.1745937419185193,
+      "learning_rate": 4.9644750272754126e-06,
+      "loss": 0.1091,
+      "step": 4355
+    },
+    {
+      "epoch": 15.557142857142857,
+      "grad_norm": 0.6054456711704427,
+      "learning_rate": 4.956850062058554e-06,
+      "loss": 0.0252,
+      "step": 4356
+    },
+    {
+      "epoch": 15.560714285714285,
+      "grad_norm": 2.3651635458779063,
+      "learning_rate": 4.949230128590992e-06,
+      "loss": 0.053,
+      "step": 4357
+    },
+    {
+      "epoch": 15.564285714285715,
+      "grad_norm": 1.4861964501182194,
+      "learning_rate": 4.94161522942151e-06,
+      "loss": 0.0476,
+      "step": 4358
+    },
+    {
+      "epoch": 15.567857142857143,
+      "grad_norm": 1.7599964957571077,
+      "learning_rate": 4.934005367097188e-06,
+      "loss": 0.029,
+      "step": 4359
+    },
+    {
+      "epoch": 15.571428571428571,
+      "grad_norm": 1.6125492885381025,
+      "learning_rate": 4.926400544163445e-06,
+      "loss": 0.0387,
+      "step": 4360
+    },
+    {
+      "epoch": 15.575,
+      "grad_norm": 2.531108373562281,
+      "learning_rate": 4.918800763163994e-06,
+      "loss": 0.0923,
+      "step": 4361
+    },
+    {
+      "epoch": 15.57857142857143,
+      "grad_norm": 0.9773375501038677,
+      "learning_rate": 4.91120602664086e-06,
+      "loss": 0.0248,
+      "step": 4362
+    },
+    {
+      "epoch": 15.582142857142857,
+      "grad_norm": 1.0170089120160024,
+      "learning_rate": 4.9036163371344066e-06,
+      "loss": 0.0244,
+      "step": 4363
+    },
+    {
+      "epoch": 15.585714285714285,
+      "grad_norm": 0.79023127905267,
+      "learning_rate": 4.896031697183276e-06,
+      "loss": 0.0173,
+      "step": 4364
+    },
+    {
+      "epoch": 15.589285714285714,
+      "grad_norm": 1.1012869794278093,
+      "learning_rate": 4.8884521093244506e-06,
+      "loss": 0.0345,
+      "step": 4365
+    },
+    {
+      "epoch": 15.592857142857143,
+      "grad_norm": 1.0755848871762448,
+      "learning_rate": 4.880877576093197e-06,
+      "loss": 0.0381,
+      "step": 4366
+    },
+    {
+      "epoch": 15.596428571428572,
+      "grad_norm": 1.066842195724301,
+      "learning_rate": 4.873308100023118e-06,
+      "loss": 0.0557,
+      "step": 4367
+    },
+    {
+      "epoch": 15.6,
+      "grad_norm": 0.7719261751788339,
+      "learning_rate": 4.865743683646094e-06,
+      "loss": 0.0195,
+      "step": 4368
+    },
+    {
+      "epoch": 15.603571428571428,
+      "grad_norm": 1.9893171818197541,
+      "learning_rate": 4.858184329492346e-06,
+      "loss": 0.0285,
+      "step": 4369
+    },
+    {
+      "epoch": 15.607142857142858,
+      "grad_norm": 1.1378706408260832,
+      "learning_rate": 4.850630040090372e-06,
+      "loss": 0.0328,
+      "step": 4370
+    },
+    {
+      "epoch": 15.610714285714286,
+      "grad_norm": 1.2208155899930615,
+      "learning_rate": 4.843080817966999e-06,
+      "loss": 0.0315,
+      "step": 4371
+    },
+    {
+      "epoch": 15.614285714285714,
+      "grad_norm": 1.1907251895263449,
+      "learning_rate": 4.835536665647347e-06,
+      "loss": 0.0411,
+      "step": 4372
+    },
+    {
+      "epoch": 15.617857142857144,
+      "grad_norm": 0.8908850437994043,
+      "learning_rate": 4.827997585654833e-06,
+      "loss": 0.0275,
+      "step": 4373
+    },
+    {
+      "epoch": 15.621428571428572,
+      "grad_norm": 1.8423697894892876,
+      "learning_rate": 4.820463580511203e-06,
+      "loss": 0.0316,
+      "step": 4374
+    },
+    {
+      "epoch": 15.625,
+      "grad_norm": 0.9315643869776691,
+      "learning_rate": 4.812934652736474e-06,
+      "loss": 0.0288,
+      "step": 4375
+    },
+    {
+      "epoch": 15.628571428571428,
+      "grad_norm": 0.9786976477935484,
+      "learning_rate": 4.805410804848991e-06,
+      "loss": 0.0277,
+      "step": 4376
+    },
+    {
+      "epoch": 15.632142857142856,
+      "grad_norm": 0.8383961782275929,
+      "learning_rate": 4.797892039365382e-06,
+      "loss": 0.0157,
+      "step": 4377
+    },
+    {
+      "epoch": 15.635714285714286,
+      "grad_norm": 0.9117786278251998,
+      "learning_rate": 4.790378358800589e-06,
+      "loss": 0.024,
+      "step": 4378
+    },
+    {
+      "epoch": 15.639285714285714,
+      "grad_norm": 0.7826526374171506,
+      "learning_rate": 4.782869765667837e-06,
+      "loss": 0.0206,
+      "step": 4379
+    },
+    {
+      "epoch": 15.642857142857142,
+      "grad_norm": 1.2344219660576117,
+      "learning_rate": 4.775366262478669e-06,
+      "loss": 0.0213,
+      "step": 4380
+    },
+    {
+      "epoch": 15.646428571428572,
+      "grad_norm": 1.404267580130676,
+      "learning_rate": 4.767867851742902e-06,
+      "loss": 0.0338,
+      "step": 4381
+    },
+    {
+      "epoch": 15.65,
+      "grad_norm": 1.4008315420071638,
+      "learning_rate": 4.760374535968677e-06,
+      "loss": 0.0275,
+      "step": 4382
+    },
+    {
+      "epoch": 15.653571428571428,
+      "grad_norm": 0.6250611944666815,
+      "learning_rate": 4.752886317662402e-06,
+      "loss": 0.0228,
+      "step": 4383
+    },
+    {
+      "epoch": 15.657142857142857,
+      "grad_norm": 2.006260113582937,
+      "learning_rate": 4.745403199328809e-06,
+      "loss": 0.0806,
+      "step": 4384
+    },
+    {
+      "epoch": 15.660714285714286,
+      "grad_norm": 0.9892054807384,
+      "learning_rate": 4.7379251834709e-06,
+      "loss": 0.0216,
+      "step": 4385
+    },
+    {
+      "epoch": 15.664285714285715,
+      "grad_norm": 1.4386770132381153,
+      "learning_rate": 4.730452272589976e-06,
+      "loss": 0.0515,
+      "step": 4386
+    },
+    {
+      "epoch": 15.667857142857143,
+      "grad_norm": 0.49031137649142703,
+      "learning_rate": 4.722984469185643e-06,
+      "loss": 0.0113,
+      "step": 4387
+    },
+    {
+      "epoch": 15.67142857142857,
+      "grad_norm": 1.2843997039806196,
+      "learning_rate": 4.715521775755783e-06,
+      "loss": 0.0461,
+      "step": 4388
+    },
+    {
+      "epoch": 15.675,
+      "grad_norm": 1.322544390221808,
+      "learning_rate": 4.708064194796585e-06,
+      "loss": 0.0405,
+      "step": 4389
+    },
+    {
+      "epoch": 15.678571428571429,
+      "grad_norm": 1.3809572604831812,
+      "learning_rate": 4.700611728802504e-06,
+      "loss": 0.059,
+      "step": 4390
+    },
+    {
+      "epoch": 15.682142857142857,
+      "grad_norm": 1.1229324378919339,
+      "learning_rate": 4.693164380266313e-06,
+      "loss": 0.023,
+      "step": 4391
+    },
+    {
+      "epoch": 15.685714285714285,
+      "grad_norm": 1.384147982128602,
+      "learning_rate": 4.685722151679046e-06,
+      "loss": 0.0377,
+      "step": 4392
+    },
+    {
+      "epoch": 15.689285714285715,
+      "grad_norm": 1.3524666788343032,
+      "learning_rate": 4.678285045530051e-06,
+      "loss": 0.0246,
+      "step": 4393
+    },
+    {
+      "epoch": 15.692857142857143,
+      "grad_norm": 1.095328536937415,
+      "learning_rate": 4.670853064306935e-06,
+      "loss": 0.0208,
+      "step": 4394
+    },
+    {
+      "epoch": 15.696428571428571,
+      "grad_norm": 1.266226323618783,
+      "learning_rate": 4.663426210495616e-06,
+      "loss": 0.0214,
+      "step": 4395
+    },
+    {
+      "epoch": 15.7,
+      "grad_norm": 1.242207649299209,
+      "learning_rate": 4.656004486580276e-06,
+      "loss": 0.0196,
+      "step": 4396
+    },
+    {
+      "epoch": 15.70357142857143,
+      "grad_norm": 0.6593718210326333,
+      "learning_rate": 4.648587895043402e-06,
+      "loss": 0.014,
+      "step": 4397
+    },
+    {
+      "epoch": 15.707142857142857,
+      "grad_norm": 1.4054564737220614,
+      "learning_rate": 4.641176438365745e-06,
+      "loss": 0.0239,
+      "step": 4398
+    },
+    {
+      "epoch": 15.710714285714285,
+      "grad_norm": 1.0734962710609148,
+      "learning_rate": 4.633770119026344e-06,
+      "loss": 0.0243,
+      "step": 4399
+    },
+    {
+      "epoch": 15.714285714285714,
+      "grad_norm": 0.983860761191185,
+      "learning_rate": 4.62636893950253e-06,
+      "loss": 0.0301,
+      "step": 4400
+    },
+    {
+      "epoch": 15.717857142857143,
+      "grad_norm": 0.9385297546879534,
+      "learning_rate": 4.6189729022698955e-06,
+      "loss": 0.0243,
+      "step": 4401
+    },
+    {
+      "epoch": 15.721428571428572,
+      "grad_norm": 0.7116242853781233,
+      "learning_rate": 4.61158200980234e-06,
+      "loss": 0.0205,
+      "step": 4402
+    },
+    {
+      "epoch": 15.725,
+      "grad_norm": 1.9975491616824357,
+      "learning_rate": 4.604196264572011e-06,
+      "loss": 0.0269,
+      "step": 4403
+    },
+    {
+      "epoch": 15.728571428571428,
+      "grad_norm": 0.6915444570007092,
+      "learning_rate": 4.596815669049364e-06,
+      "loss": 0.0223,
+      "step": 4404
+    },
+    {
+      "epoch": 15.732142857142858,
+      "grad_norm": 1.064506868604621,
+      "learning_rate": 4.589440225703104e-06,
+      "loss": 0.0242,
+      "step": 4405
+    },
+    {
+      "epoch": 15.735714285714286,
+      "grad_norm": 1.3430600366779066,
+      "learning_rate": 4.582069937000237e-06,
+      "loss": 0.0336,
+      "step": 4406
+    },
+    {
+      "epoch": 15.739285714285714,
+      "grad_norm": 1.8810914695225323,
+      "learning_rate": 4.574704805406027e-06,
+      "loss": 0.0267,
+      "step": 4407
+    },
+    {
+      "epoch": 15.742857142857144,
+      "grad_norm": 0.9470855481506882,
+      "learning_rate": 4.567344833384027e-06,
+      "loss": 0.0444,
+      "step": 4408
+    },
+    {
+      "epoch": 15.746428571428572,
+      "grad_norm": 0.9839228663803006,
+      "learning_rate": 4.559990023396055e-06,
+      "loss": 0.0232,
+      "step": 4409
+    },
+    {
+      "epoch": 15.75,
+      "grad_norm": 0.9882734100144852,
+      "learning_rate": 4.552640377902197e-06,
+      "loss": 0.0342,
+      "step": 4410
+    },
+    {
+      "epoch": 15.753571428571428,
+      "grad_norm": 1.2600305145689912,
+      "learning_rate": 4.5452958993608335e-06,
+      "loss": 0.0407,
+      "step": 4411
+    },
+    {
+      "epoch": 15.757142857142856,
+      "grad_norm": 0.8938952057722452,
+      "learning_rate": 4.5379565902285895e-06,
+      "loss": 0.0272,
+      "step": 4412
+    },
+    {
+      "epoch": 15.760714285714286,
+      "grad_norm": 1.0420109676538492,
+      "learning_rate": 4.530622452960385e-06,
+      "loss": 0.0356,
+      "step": 4413
+    },
+    {
+      "epoch": 15.764285714285714,
+      "grad_norm": 1.4733431636827008,
+      "learning_rate": 4.523293490009386e-06,
+      "loss": 0.0498,
+      "step": 4414
+    },
+    {
+      "epoch": 15.767857142857142,
+      "grad_norm": 1.3317487682746576,
+      "learning_rate": 4.515969703827057e-06,
+      "loss": 0.0412,
+      "step": 4415
+    },
+    {
+      "epoch": 15.771428571428572,
+      "grad_norm": 1.3333424249509205,
+      "learning_rate": 4.508651096863097e-06,
+      "loss": 0.0291,
+      "step": 4416
+    },
+    {
+      "epoch": 15.775,
+      "grad_norm": 2.0947302293634222,
+      "learning_rate": 4.501337671565509e-06,
+      "loss": 0.0346,
+      "step": 4417
+    },
+    {
+      "epoch": 15.778571428571428,
+      "grad_norm": 0.9058754080050702,
+      "learning_rate": 4.494029430380527e-06,
+      "loss": 0.0416,
+      "step": 4418
+    },
+    {
+      "epoch": 15.782142857142857,
+      "grad_norm": 0.7216888805908538,
+      "learning_rate": 4.486726375752681e-06,
+      "loss": 0.0176,
+      "step": 4419
+    },
+    {
+      "epoch": 15.785714285714286,
+      "grad_norm": 1.7482969359601421,
+      "learning_rate": 4.4794285101247435e-06,
+      "loss": 0.0524,
+      "step": 4420
+    },
+    {
+      "epoch": 15.789285714285715,
+      "grad_norm": 1.4891601927609228,
+      "learning_rate": 4.47213583593777e-06,
+      "loss": 0.0328,
+      "step": 4421
+    },
+    {
+      "epoch": 15.792857142857143,
+      "grad_norm": 1.5241845284659843,
+      "learning_rate": 4.464848355631066e-06,
+      "loss": 0.0505,
+      "step": 4422
+    },
+    {
+      "epoch": 15.79642857142857,
+      "grad_norm": 0.5786124513364627,
+      "learning_rate": 4.457566071642201e-06,
+      "loss": 0.0162,
+      "step": 4423
+    },
+    {
+      "epoch": 15.8,
+      "grad_norm": 2.260754632040175,
+      "learning_rate": 4.450288986407019e-06,
+      "loss": 0.04,
+      "step": 4424
+    },
+    {
+      "epoch": 15.803571428571429,
+      "grad_norm": 1.2872908884633403,
+      "learning_rate": 4.443017102359601e-06,
+      "loss": 0.0335,
+      "step": 4425
+    },
+    {
+      "epoch": 15.807142857142857,
+      "grad_norm": 1.122348643809646,
+      "learning_rate": 4.435750421932321e-06,
+      "loss": 0.0354,
+      "step": 4426
+    },
+    {
+      "epoch": 15.810714285714285,
+      "grad_norm": 1.5691805378636021,
+      "learning_rate": 4.4284889475557805e-06,
+      "loss": 0.0527,
+      "step": 4427
+    },
+    {
+      "epoch": 15.814285714285715,
+      "grad_norm": 0.8812073192983372,
+      "learning_rate": 4.421232681658865e-06,
+      "loss": 0.0224,
+      "step": 4428
+    },
+    {
+      "epoch": 15.817857142857143,
+      "grad_norm": 0.767948174283224,
+      "learning_rate": 4.413981626668693e-06,
+      "loss": 0.0104,
+      "step": 4429
+    },
+    {
+      "epoch": 15.821428571428571,
+      "grad_norm": 0.8536462459834202,
+      "learning_rate": 4.406735785010668e-06,
+      "loss": 0.0264,
+      "step": 4430
+    },
+    {
+      "epoch": 15.825,
+      "grad_norm": 1.1858896414166598,
+      "learning_rate": 4.399495159108423e-06,
+      "loss": 0.0418,
+      "step": 4431
+    },
+    {
+      "epoch": 15.82857142857143,
+      "grad_norm": 2.106916586176277,
+      "learning_rate": 4.392259751383867e-06,
+      "loss": 0.0139,
+      "step": 4432
+    },
+    {
+      "epoch": 15.832142857142857,
+      "grad_norm": 1.4737613899092359,
+      "learning_rate": 4.385029564257149e-06,
+      "loss": 0.0462,
+      "step": 4433
+    },
+    {
+      "epoch": 15.835714285714285,
+      "grad_norm": 2.2595077637884926,
+      "learning_rate": 4.377804600146687e-06,
+      "loss": 0.0557,
+      "step": 4434
+    },
+    {
+      "epoch": 15.839285714285714,
+      "grad_norm": 1.5811255541040952,
+      "learning_rate": 4.370584861469136e-06,
+      "loss": 0.0395,
+      "step": 4435
+    },
+    {
+      "epoch": 15.842857142857143,
+      "grad_norm": 0.9512787977647834,
+      "learning_rate": 4.363370350639405e-06,
+      "loss": 0.0155,
+      "step": 4436
+    },
+    {
+      "epoch": 15.846428571428572,
+      "grad_norm": 1.0362885654263065,
+      "learning_rate": 4.3561610700706725e-06,
+      "loss": 0.028,
+      "step": 4437
+    },
+    {
+      "epoch": 15.85,
+      "grad_norm": 1.1841839870328899,
+      "learning_rate": 4.348957022174343e-06,
+      "loss": 0.0317,
+      "step": 4438
+    },
+    {
+      "epoch": 15.853571428571428,
+      "grad_norm": 0.9976241823771064,
+      "learning_rate": 4.34175820936009e-06,
+      "loss": 0.0381,
+      "step": 4439
+    },
+    {
+      "epoch": 15.857142857142858,
+      "grad_norm": 0.8445163714744034,
+      "learning_rate": 4.33456463403582e-06,
+      "loss": 0.0235,
+      "step": 4440
+    },
+    {
+      "epoch": 15.860714285714286,
+      "grad_norm": 0.7897573285243012,
+      "learning_rate": 4.3273762986077085e-06,
+      "loss": 0.0236,
+      "step": 4441
+    },
+    {
+      "epoch": 15.864285714285714,
+      "grad_norm": 1.185543592591235,
+      "learning_rate": 4.320193205480152e-06,
+      "loss": 0.0224,
+      "step": 4442
+    },
+    {
+      "epoch": 15.867857142857144,
+      "grad_norm": 2.5133339470238143,
+      "learning_rate": 4.313015357055817e-06,
+      "loss": 0.0442,
+      "step": 4443
+    },
+    {
+      "epoch": 15.871428571428572,
+      "grad_norm": 0.8077869428231158,
+      "learning_rate": 4.305842755735599e-06,
+      "loss": 0.0174,
+      "step": 4444
+    },
+    {
+      "epoch": 15.875,
+      "grad_norm": 0.8435146654279381,
+      "learning_rate": 4.29867540391865e-06,
+      "loss": 0.0261,
+      "step": 4445
+    },
+    {
+      "epoch": 15.878571428571428,
+      "grad_norm": 0.5769857656019688,
+      "learning_rate": 4.291513304002362e-06,
+      "loss": 0.0117,
+      "step": 4446
+    },
+    {
+      "epoch": 15.882142857142856,
+      "grad_norm": 1.271289119362958,
+      "learning_rate": 4.284356458382362e-06,
+      "loss": 0.0472,
+      "step": 4447
+    },
+    {
+      "epoch": 15.885714285714286,
+      "grad_norm": 1.2098497417778356,
+      "learning_rate": 4.2772048694525384e-06,
+      "loss": 0.0495,
+      "step": 4448
+    },
+    {
+      "epoch": 15.889285714285714,
+      "grad_norm": 0.9228735046957088,
+      "learning_rate": 4.2700585396049954e-06,
+      "loss": 0.0257,
+      "step": 4449
+    },
+    {
+      "epoch": 15.892857142857142,
+      "grad_norm": 0.9248607066795157,
+      "learning_rate": 4.26291747123011e-06,
+      "loss": 0.0174,
+      "step": 4450
+    },
+    {
+      "epoch": 15.896428571428572,
+      "grad_norm": 1.2292896222998233,
+      "learning_rate": 4.2557816667164655e-06,
+      "loss": 0.0383,
+      "step": 4451
+    },
+    {
+      "epoch": 15.9,
+      "grad_norm": 1.1102301769034282,
+      "learning_rate": 4.248651128450916e-06,
+      "loss": 0.0549,
+      "step": 4452
+    },
+    {
+      "epoch": 15.903571428571428,
+      "grad_norm": 1.0500376416275472,
+      "learning_rate": 4.241525858818527e-06,
+      "loss": 0.0216,
+      "step": 4453
+    },
+    {
+      "epoch": 15.907142857142857,
+      "grad_norm": 1.06758920065016,
+      "learning_rate": 4.234405860202624e-06,
+      "loss": 0.028,
+      "step": 4454
+    },
+    {
+      "epoch": 15.910714285714286,
+      "grad_norm": 0.9231692697289245,
+      "learning_rate": 4.227291134984748e-06,
+      "loss": 0.0272,
+      "step": 4455
+    },
+    {
+      "epoch": 15.914285714285715,
+      "grad_norm": 0.660029094521949,
+      "learning_rate": 4.2201816855447e-06,
+      "loss": 0.0144,
+      "step": 4456
+    },
+    {
+      "epoch": 15.917857142857143,
+      "grad_norm": 0.9012575056972798,
+      "learning_rate": 4.2130775142604905e-06,
+      "loss": 0.0247,
+      "step": 4457
+    },
+    {
+      "epoch": 15.92142857142857,
+      "grad_norm": 0.768316151926101,
+      "learning_rate": 4.205978623508391e-06,
+      "loss": 0.0192,
+      "step": 4458
+    },
+    {
+      "epoch": 15.925,
+      "grad_norm": 1.448319716638753,
+      "learning_rate": 4.198885015662888e-06,
+      "loss": 0.055,
+      "step": 4459
+    },
+    {
+      "epoch": 15.928571428571429,
+      "grad_norm": 0.3563336973299812,
+      "learning_rate": 4.191796693096699e-06,
+      "loss": 0.0096,
+      "step": 4460
+    },
+    {
+      "epoch": 15.932142857142857,
+      "grad_norm": 1.1022809839898846,
+      "learning_rate": 4.184713658180794e-06,
+      "loss": 0.0196,
+      "step": 4461
+    },
+    {
+      "epoch": 15.935714285714285,
+      "grad_norm": 1.5151048150423978,
+      "learning_rate": 4.177635913284348e-06,
+      "loss": 0.0729,
+      "step": 4462
+    },
+    {
+      "epoch": 15.939285714285715,
+      "grad_norm": 0.7531199252038745,
+      "learning_rate": 4.170563460774795e-06,
+      "loss": 0.0294,
+      "step": 4463
+    },
+    {
+      "epoch": 15.942857142857143,
+      "grad_norm": 0.4278062492201082,
+      "learning_rate": 4.163496303017766e-06,
+      "loss": 0.0113,
+      "step": 4464
+    },
+    {
+      "epoch": 15.946428571428571,
+      "grad_norm": 1.8059792509152413,
+      "learning_rate": 4.1564344423771574e-06,
+      "loss": 0.0663,
+      "step": 4465
+    },
+    {
+      "epoch": 15.95,
+      "grad_norm": 1.785058396919337,
+      "learning_rate": 4.149377881215058e-06,
+      "loss": 0.0423,
+      "step": 4466
+    },
+    {
+      "epoch": 15.95357142857143,
+      "grad_norm": 1.4213994244132444,
+      "learning_rate": 4.142326621891815e-06,
+      "loss": 0.0238,
+      "step": 4467
+    },
+    {
+      "epoch": 15.957142857142857,
+      "grad_norm": 0.8221797220789192,
+      "learning_rate": 4.135280666765977e-06,
+      "loss": 0.0245,
+      "step": 4468
+    },
+    {
+      "epoch": 15.960714285714285,
+      "grad_norm": 0.8792303658603298,
+      "learning_rate": 4.1282400181943405e-06,
+      "loss": 0.0187,
+      "step": 4469
+    },
+    {
+      "epoch": 15.964285714285714,
+      "grad_norm": 0.9928800933903147,
+      "learning_rate": 4.121204678531903e-06,
+      "loss": 0.0178,
+      "step": 4470
+    },
+    {
+      "epoch": 15.967857142857143,
+      "grad_norm": 1.8830689811957493,
+      "learning_rate": 4.114174650131915e-06,
+      "loss": 0.0333,
+      "step": 4471
+    },
+    {
+      "epoch": 15.971428571428572,
+      "grad_norm": 0.816250501182946,
+      "learning_rate": 4.1071499353458245e-06,
+      "loss": 0.0249,
+      "step": 4472
+    },
+    {
+      "epoch": 15.975,
+      "grad_norm": 0.6581922047901969,
+      "learning_rate": 4.10013053652331e-06,
+      "loss": 0.0098,
+      "step": 4473
+    },
+    {
+      "epoch": 15.978571428571428,
+      "grad_norm": 1.160537145693653,
+      "learning_rate": 4.093116456012285e-06,
+      "loss": 0.0321,
+      "step": 4474
+    },
+    {
+      "epoch": 15.982142857142858,
+      "grad_norm": 0.5593818779539091,
+      "learning_rate": 4.086107696158861e-06,
+      "loss": 0.0086,
+      "step": 4475
+    },
+    {
+      "epoch": 15.985714285714286,
+      "grad_norm": 1.1563583652024865,
+      "learning_rate": 4.079104259307394e-06,
+      "loss": 0.0314,
+      "step": 4476
+    },
+    {
+      "epoch": 15.989285714285714,
+      "grad_norm": 1.2810313004340577,
+      "learning_rate": 4.072106147800438e-06,
+      "loss": 0.032,
+      "step": 4477
+    },
+    {
+      "epoch": 15.992857142857144,
+      "grad_norm": 0.38435401806310315,
+      "learning_rate": 4.065113363978785e-06,
+      "loss": 0.0086,
+      "step": 4478
+    },
+    {
+      "epoch": 15.996428571428572,
+      "grad_norm": 0.7171175736940356,
+      "learning_rate": 4.0581259101814255e-06,
+      "loss": 0.0254,
+      "step": 4479
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 0.9776864954692295,
+      "learning_rate": 4.051143788745588e-06,
+      "loss": 0.0223,
+      "step": 4480
+    },
+    {
+      "epoch": 16.00357142857143,
+      "grad_norm": 1.7945207151685825,
+      "learning_rate": 4.044167002006694e-06,
+      "loss": 0.0434,
+      "step": 4481
+    },
+    {
+      "epoch": 16.007142857142856,
+      "grad_norm": 1.2030155119293586,
+      "learning_rate": 4.037195552298407e-06,
+      "loss": 0.0422,
+      "step": 4482
+    },
+    {
+      "epoch": 16.010714285714286,
+      "grad_norm": 1.1219485663391846,
+      "learning_rate": 4.030229441952585e-06,
+      "loss": 0.0397,
+      "step": 4483
+    },
+    {
+      "epoch": 16.014285714285716,
+      "grad_norm": 0.7868542990659215,
+      "learning_rate": 4.0232686732992985e-06,
+      "loss": 0.0202,
+      "step": 4484
+    },
+    {
+      "epoch": 16.017857142857142,
+      "grad_norm": 0.7335929338094682,
+      "learning_rate": 4.016313248666854e-06,
+      "loss": 0.0194,
+      "step": 4485
+    },
+    {
+      "epoch": 16.021428571428572,
+      "grad_norm": 1.0466971457050085,
+      "learning_rate": 4.009363170381742e-06,
+      "loss": 0.0328,
+      "step": 4486
+    },
+    {
+      "epoch": 16.025,
+      "grad_norm": 1.4401441932880545,
+      "learning_rate": 4.002418440768692e-06,
+      "loss": 0.0298,
+      "step": 4487
+    },
+    {
+      "epoch": 16.02857142857143,
+      "grad_norm": 0.4761779689834046,
+      "learning_rate": 3.9954790621506175e-06,
+      "loss": 0.0113,
+      "step": 4488
+    },
+    {
+      "epoch": 16.03214285714286,
+      "grad_norm": 1.510186806810632,
+      "learning_rate": 3.988545036848668e-06,
+      "loss": 0.0376,
+      "step": 4489
+    },
+    {
+      "epoch": 16.035714285714285,
+      "grad_norm": 0.8332715363999682,
+      "learning_rate": 3.98161636718218e-06,
+      "loss": 0.022,
+      "step": 4490
+    },
+    {
+      "epoch": 16.039285714285715,
+      "grad_norm": 1.0714882791579632,
+      "learning_rate": 3.974693055468714e-06,
+      "loss": 0.0321,
+      "step": 4491
+    },
+    {
+      "epoch": 16.042857142857144,
+      "grad_norm": 1.5411593623385047,
+      "learning_rate": 3.96777510402403e-06,
+      "loss": 0.0309,
+      "step": 4492
+    },
+    {
+      "epoch": 16.04642857142857,
+      "grad_norm": 1.1137709648440013,
+      "learning_rate": 3.960862515162105e-06,
+      "loss": 0.0262,
+      "step": 4493
+    },
+    {
+      "epoch": 16.05,
+      "grad_norm": 0.736613514124241,
+      "learning_rate": 3.953955291195104e-06,
+      "loss": 0.0144,
+      "step": 4494
+    },
+    {
+      "epoch": 16.053571428571427,
+      "grad_norm": 0.6000138375117703,
+      "learning_rate": 3.947053434433421e-06,
+      "loss": 0.0092,
+      "step": 4495
+    },
+    {
+      "epoch": 16.057142857142857,
+      "grad_norm": 0.7509143475555277,
+      "learning_rate": 3.940156947185634e-06,
+      "loss": 0.0173,
+      "step": 4496
+    },
+    {
+      "epoch": 16.060714285714287,
+      "grad_norm": 1.1950411211909493,
+      "learning_rate": 3.933265831758526e-06,
+      "loss": 0.0532,
+      "step": 4497
+    },
+    {
+      "epoch": 16.064285714285713,
+      "grad_norm": 0.9843711748881424,
+      "learning_rate": 3.926380090457111e-06,
+      "loss": 0.0522,
+      "step": 4498
+    },
+    {
+      "epoch": 16.067857142857143,
+      "grad_norm": 0.6169781005219693,
+      "learning_rate": 3.9194997255845706e-06,
+      "loss": 0.0212,
+      "step": 4499
+    },
+    {
+      "epoch": 16.071428571428573,
+      "grad_norm": 0.9302934822447426,
+      "learning_rate": 3.91262473944231e-06,
+      "loss": 0.041,
+      "step": 4500
+    },
+    {
+      "epoch": 16.075,
+      "grad_norm": 0.5879129479265613,
+      "learning_rate": 3.9057551343299185e-06,
+      "loss": 0.0125,
+      "step": 4501
+    },
+    {
+      "epoch": 16.07857142857143,
+      "grad_norm": 0.5910288645353989,
+      "learning_rate": 3.898890912545208e-06,
+      "loss": 0.0131,
+      "step": 4502
+    },
+    {
+      "epoch": 16.082142857142856,
+      "grad_norm": 1.5524522537070358,
+      "learning_rate": 3.892032076384167e-06,
+      "loss": 0.0571,
+      "step": 4503
+    },
+    {
+      "epoch": 16.085714285714285,
+      "grad_norm": 1.1894955112706733,
+      "learning_rate": 3.885178628140988e-06,
+      "loss": 0.0601,
+      "step": 4504
+    },
+    {
+      "epoch": 16.089285714285715,
+      "grad_norm": 2.0635309447453443,
+      "learning_rate": 3.87833057010808e-06,
+      "loss": 0.0371,
+      "step": 4505
+    },
+    {
+      "epoch": 16.09285714285714,
+      "grad_norm": 1.002178510938788,
+      "learning_rate": 3.871487904576019e-06,
+      "loss": 0.0205,
+      "step": 4506
+    },
+    {
+      "epoch": 16.09642857142857,
+      "grad_norm": 0.909591981331953,
+      "learning_rate": 3.864650633833606e-06,
+      "loss": 0.0196,
+      "step": 4507
+    },
+    {
+      "epoch": 16.1,
+      "grad_norm": 0.7711346483212361,
+      "learning_rate": 3.857818760167813e-06,
+      "loss": 0.027,
+      "step": 4508
+    },
+    {
+      "epoch": 16.103571428571428,
+      "grad_norm": 1.699818116945288,
+      "learning_rate": 3.850992285863828e-06,
+      "loss": 0.0269,
+      "step": 4509
+    },
+    {
+      "epoch": 16.107142857142858,
+      "grad_norm": 0.6891541858708813,
+      "learning_rate": 3.8441712132050125e-06,
+      "loss": 0.0151,
+      "step": 4510
+    },
+    {
+      "epoch": 16.110714285714284,
+      "grad_norm": 0.9388728162397588,
+      "learning_rate": 3.837355544472943e-06,
+      "loss": 0.0206,
+      "step": 4511
+    },
+    {
+      "epoch": 16.114285714285714,
+      "grad_norm": 1.1931231565097318,
+      "learning_rate": 3.830545281947367e-06,
+      "loss": 0.0138,
+      "step": 4512
+    },
+    {
+      "epoch": 16.117857142857144,
+      "grad_norm": 1.6568994700862174,
+      "learning_rate": 3.823740427906244e-06,
+      "loss": 0.0326,
+      "step": 4513
+    },
+    {
+      "epoch": 16.12142857142857,
+      "grad_norm": 1.0087693613280595,
+      "learning_rate": 3.816940984625708e-06,
+      "loss": 0.0264,
+      "step": 4514
+    },
+    {
+      "epoch": 16.125,
+      "grad_norm": 1.5274566979213215,
+      "learning_rate": 3.8101469543800828e-06,
+      "loss": 0.0338,
+      "step": 4515
+    },
+    {
+      "epoch": 16.12857142857143,
+      "grad_norm": 0.691761919785751,
+      "learning_rate": 3.8033583394419026e-06,
+      "loss": 0.0263,
+      "step": 4516
+    },
+    {
+      "epoch": 16.132142857142856,
+      "grad_norm": 0.5402536450922536,
+      "learning_rate": 3.796575142081864e-06,
+      "loss": 0.0104,
+      "step": 4517
+    },
+    {
+      "epoch": 16.135714285714286,
+      "grad_norm": 1.12668986319769,
+      "learning_rate": 3.7897973645688744e-06,
+      "loss": 0.0327,
+      "step": 4518
+    },
+    {
+      "epoch": 16.139285714285716,
+      "grad_norm": 0.8931465668260825,
+      "learning_rate": 3.783025009170005e-06,
+      "loss": 0.0226,
+      "step": 4519
+    },
+    {
+      "epoch": 16.142857142857142,
+      "grad_norm": 0.657047598790759,
+      "learning_rate": 3.7762580781505363e-06,
+      "loss": 0.0201,
+      "step": 4520
+    },
+    {
+      "epoch": 16.146428571428572,
+      "grad_norm": 2.074216768116552,
+      "learning_rate": 3.7694965737739165e-06,
+      "loss": 0.0553,
+      "step": 4521
+    },
+    {
+      "epoch": 16.15,
+      "grad_norm": 1.6041803528571632,
+      "learning_rate": 3.762740498301791e-06,
+      "loss": 0.0127,
+      "step": 4522
+    },
+    {
+      "epoch": 16.15357142857143,
+      "grad_norm": 1.3265629661868708,
+      "learning_rate": 3.7559898539939797e-06,
+      "loss": 0.0434,
+      "step": 4523
+    },
+    {
+      "epoch": 16.15714285714286,
+      "grad_norm": 1.7560036971081536,
+      "learning_rate": 3.7492446431084963e-06,
+      "loss": 0.0179,
+      "step": 4524
+    },
+    {
+      "epoch": 16.160714285714285,
+      "grad_norm": 1.4133663680227428,
+      "learning_rate": 3.742504867901524e-06,
+      "loss": 0.0419,
+      "step": 4525
+    },
+    {
+      "epoch": 16.164285714285715,
+      "grad_norm": 1.5136820275981422,
+      "learning_rate": 3.735770530627443e-06,
+      "loss": 0.0561,
+      "step": 4526
+    },
+    {
+      "epoch": 16.167857142857144,
+      "grad_norm": 0.9058105388758774,
+      "learning_rate": 3.7290416335388014e-06,
+      "loss": 0.0221,
+      "step": 4527
+    },
+    {
+      "epoch": 16.17142857142857,
+      "grad_norm": 1.2657096919172155,
+      "learning_rate": 3.7223181788863306e-06,
+      "loss": 0.0359,
+      "step": 4528
+    },
+    {
+      "epoch": 16.175,
+      "grad_norm": 0.6210804385282244,
+      "learning_rate": 3.7156001689189492e-06,
+      "loss": 0.0207,
+      "step": 4529
+    },
+    {
+      "epoch": 16.178571428571427,
+      "grad_norm": 1.3272684215331734,
+      "learning_rate": 3.708887605883744e-06,
+      "loss": 0.0506,
+      "step": 4530
+    },
+    {
+      "epoch": 16.182142857142857,
+      "grad_norm": 0.8738497247768638,
+      "learning_rate": 3.7021804920259928e-06,
+      "loss": 0.0209,
+      "step": 4531
+    },
+    {
+      "epoch": 16.185714285714287,
+      "grad_norm": 1.5517424251977032,
+      "learning_rate": 3.695478829589132e-06,
+      "loss": 0.0305,
+      "step": 4532
+    },
+    {
+      "epoch": 16.189285714285713,
+      "grad_norm": 1.6368017356102837,
+      "learning_rate": 3.6887826208147968e-06,
+      "loss": 0.0251,
+      "step": 4533
+    },
+    {
+      "epoch": 16.192857142857143,
+      "grad_norm": 1.7871488469095378,
+      "learning_rate": 3.6820918679427787e-06,
+      "loss": 0.0334,
+      "step": 4534
+    },
+    {
+      "epoch": 16.196428571428573,
+      "grad_norm": 0.7696967111242426,
+      "learning_rate": 3.67540657321106e-06,
+      "loss": 0.0171,
+      "step": 4535
+    },
+    {
+      "epoch": 16.2,
+      "grad_norm": 0.6702437952380141,
+      "learning_rate": 3.668726738855779e-06,
+      "loss": 0.0156,
+      "step": 4536
+    },
+    {
+      "epoch": 16.20357142857143,
+      "grad_norm": 0.6999002161778081,
+      "learning_rate": 3.6620523671112728e-06,
+      "loss": 0.0161,
+      "step": 4537
+    },
+    {
+      "epoch": 16.207142857142856,
+      "grad_norm": 0.7836892162126025,
+      "learning_rate": 3.6553834602100226e-06,
+      "loss": 0.0119,
+      "step": 4538
+    },
+    {
+      "epoch": 16.210714285714285,
+      "grad_norm": 1.030804285469787,
+      "learning_rate": 3.648720020382708e-06,
+      "loss": 0.0347,
+      "step": 4539
+    },
+    {
+      "epoch": 16.214285714285715,
+      "grad_norm": 1.1957323250231902,
+      "learning_rate": 3.6420620498581614e-06,
+      "loss": 0.0424,
+      "step": 4540
+    },
+    {
+      "epoch": 16.21785714285714,
+      "grad_norm": 0.6258461313449996,
+      "learning_rate": 3.6354095508633892e-06,
+      "loss": 0.021,
+      "step": 4541
+    },
+    {
+      "epoch": 16.22142857142857,
+      "grad_norm": 0.5688196889736339,
+      "learning_rate": 3.628762525623579e-06,
+      "loss": 0.0082,
+      "step": 4542
+    },
+    {
+      "epoch": 16.225,
+      "grad_norm": 0.7763437350797503,
+      "learning_rate": 3.6221209763620713e-06,
+      "loss": 0.0308,
+      "step": 4543
+    },
+    {
+      "epoch": 16.228571428571428,
+      "grad_norm": 1.4291912789901324,
+      "learning_rate": 3.6154849053003927e-06,
+      "loss": 0.0384,
+      "step": 4544
+    },
+    {
+      "epoch": 16.232142857142858,
+      "grad_norm": 0.9035808711163116,
+      "learning_rate": 3.6088543146582165e-06,
+      "loss": 0.0125,
+      "step": 4545
+    },
+    {
+      "epoch": 16.235714285714284,
+      "grad_norm": 1.580634608106845,
+      "learning_rate": 3.602229206653405e-06,
+      "loss": 0.0287,
+      "step": 4546
+    },
+    {
+      "epoch": 16.239285714285714,
+      "grad_norm": 0.6215016954787931,
+      "learning_rate": 3.5956095835019645e-06,
+      "loss": 0.0128,
+      "step": 4547
+    },
+    {
+      "epoch": 16.242857142857144,
+      "grad_norm": 1.7027135242432367,
+      "learning_rate": 3.5889954474180887e-06,
+      "loss": 0.0559,
+      "step": 4548
+    },
+    {
+      "epoch": 16.24642857142857,
+      "grad_norm": 0.9864969144524797,
+      "learning_rate": 3.5823868006141147e-06,
+      "loss": 0.0261,
+      "step": 4549
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 0.8807162898299915,
+      "learning_rate": 3.5757836453005633e-06,
+      "loss": 0.0139,
+      "step": 4550
+    },
+    {
+      "epoch": 16.25357142857143,
+      "grad_norm": 0.5064829811892271,
+      "learning_rate": 3.5691859836861053e-06,
+      "loss": 0.0108,
+      "step": 4551
+    },
+    {
+      "epoch": 16.257142857142856,
+      "grad_norm": 1.246633769576147,
+      "learning_rate": 3.5625938179775733e-06,
+      "loss": 0.0326,
+      "step": 4552
+    },
+    {
+      "epoch": 16.260714285714286,
+      "grad_norm": 1.2799709669385348,
+      "learning_rate": 3.556007150379974e-06,
+      "loss": 0.0219,
+      "step": 4553
+    },
+    {
+      "epoch": 16.264285714285716,
+      "grad_norm": 1.1820140104855135,
+      "learning_rate": 3.5494259830964573e-06,
+      "loss": 0.052,
+      "step": 4554
+    },
+    {
+      "epoch": 16.267857142857142,
+      "grad_norm": 1.1484502285621752,
+      "learning_rate": 3.5428503183283523e-06,
+      "loss": 0.0346,
+      "step": 4555
+    },
+    {
+      "epoch": 16.271428571428572,
+      "grad_norm": 0.7315533710018199,
+      "learning_rate": 3.5362801582751296e-06,
+      "loss": 0.0259,
+      "step": 4556
+    },
+    {
+      "epoch": 16.275,
+      "grad_norm": 1.3023499006556334,
+      "learning_rate": 3.529715505134439e-06,
+      "loss": 0.0602,
+      "step": 4557
+    },
+    {
+      "epoch": 16.27857142857143,
+      "grad_norm": 0.2510895303778787,
+      "learning_rate": 3.5231563611020624e-06,
+      "loss": 0.0082,
+      "step": 4558
+    },
+    {
+      "epoch": 16.28214285714286,
+      "grad_norm": 0.6347129415923156,
+      "learning_rate": 3.5166027283719674e-06,
+      "loss": 0.0128,
+      "step": 4559
+    },
+    {
+      "epoch": 16.285714285714285,
+      "grad_norm": 1.275106614304084,
+      "learning_rate": 3.5100546091362487e-06,
+      "loss": 0.0422,
+      "step": 4560
+    },
+    {
+      "epoch": 16.289285714285715,
+      "grad_norm": 0.2985730647798565,
+      "learning_rate": 3.5035120055851878e-06,
+      "loss": 0.0066,
+      "step": 4561
+    },
+    {
+      "epoch": 16.292857142857144,
+      "grad_norm": 0.953994199770007,
+      "learning_rate": 3.49697491990719e-06,
+      "loss": 0.0244,
+      "step": 4562
+    },
+    {
+      "epoch": 16.29642857142857,
+      "grad_norm": 0.8363960397780554,
+      "learning_rate": 3.4904433542888437e-06,
+      "loss": 0.022,
+      "step": 4563
+    },
+    {
+      "epoch": 16.3,
+      "grad_norm": 0.47014966752676113,
+      "learning_rate": 3.4839173109148728e-06,
+      "loss": 0.0159,
+      "step": 4564
+    },
+    {
+      "epoch": 16.303571428571427,
+      "grad_norm": 1.0743294478946783,
+      "learning_rate": 3.477396791968153e-06,
+      "loss": 0.0402,
+      "step": 4565
+    },
+    {
+      "epoch": 16.307142857142857,
+      "grad_norm": 0.7204948432608362,
+      "learning_rate": 3.470881799629726e-06,
+      "loss": 0.0144,
+      "step": 4566
+    },
+    {
+      "epoch": 16.310714285714287,
+      "grad_norm": 1.345676804577197,
+      "learning_rate": 3.4643723360787697e-06,
+      "loss": 0.044,
+      "step": 4567
+    },
+    {
+      "epoch": 16.314285714285713,
+      "grad_norm": 2.5637476425346635,
+      "learning_rate": 3.457868403492628e-06,
+      "loss": 0.0441,
+      "step": 4568
+    },
+    {
+      "epoch": 16.317857142857143,
+      "grad_norm": 1.1521732850039903,
+      "learning_rate": 3.4513700040467792e-06,
+      "loss": 0.0218,
+      "step": 4569
+    },
+    {
+      "epoch": 16.321428571428573,
+      "grad_norm": 1.11495218322723,
+      "learning_rate": 3.4448771399148638e-06,
+      "loss": 0.0273,
+      "step": 4570
+    },
+    {
+      "epoch": 16.325,
+      "grad_norm": 0.9919829476878738,
+      "learning_rate": 3.4383898132686613e-06,
+      "loss": 0.0269,
+      "step": 4571
+    },
+    {
+      "epoch": 16.32857142857143,
+      "grad_norm": 1.1921634972257122,
+      "learning_rate": 3.4319080262781057e-06,
+      "loss": 0.0333,
+      "step": 4572
+    },
+    {
+      "epoch": 16.332142857142856,
+      "grad_norm": 1.0006699378249466,
+      "learning_rate": 3.4254317811112724e-06,
+      "loss": 0.0218,
+      "step": 4573
+    },
+    {
+      "epoch": 16.335714285714285,
+      "grad_norm": 0.77416977165098,
+      "learning_rate": 3.418961079934391e-06,
+      "loss": 0.0158,
+      "step": 4574
+    },
+    {
+      "epoch": 16.339285714285715,
+      "grad_norm": 0.5098917934471304,
+      "learning_rate": 3.412495924911825e-06,
+      "loss": 0.0113,
+      "step": 4575
+    },
+    {
+      "epoch": 16.34285714285714,
+      "grad_norm": 0.6439064345159072,
+      "learning_rate": 3.4060363182060987e-06,
+      "loss": 0.0191,
+      "step": 4576
+    },
+    {
+      "epoch": 16.34642857142857,
+      "grad_norm": 0.9427260140195994,
+      "learning_rate": 3.399582261977863e-06,
+      "loss": 0.0205,
+      "step": 4577
+    },
+    {
+      "epoch": 16.35,
+      "grad_norm": 1.6084936438195527,
+      "learning_rate": 3.39313375838592e-06,
+      "loss": 0.0348,
+      "step": 4578
+    },
+    {
+      "epoch": 16.353571428571428,
+      "grad_norm": 2.2935003023164926,
+      "learning_rate": 3.386690809587225e-06,
+      "loss": 0.0481,
+      "step": 4579
+    },
+    {
+      "epoch": 16.357142857142858,
+      "grad_norm": 0.4707589544696899,
+      "learning_rate": 3.380253417736854e-06,
+      "loss": 0.0115,
+      "step": 4580
+    },
+    {
+      "epoch": 16.360714285714284,
+      "grad_norm": 0.7579453355894322,
+      "learning_rate": 3.373821584988044e-06,
+      "loss": 0.0224,
+      "step": 4581
+    },
+    {
+      "epoch": 16.364285714285714,
+      "grad_norm": 1.9383914173317665,
+      "learning_rate": 3.3673953134921566e-06,
+      "loss": 0.0271,
+      "step": 4582
+    },
+    {
+      "epoch": 16.367857142857144,
+      "grad_norm": 1.670679714095103,
+      "learning_rate": 3.36097460539871e-06,
+      "loss": 0.0398,
+      "step": 4583
+    },
+    {
+      "epoch": 16.37142857142857,
+      "grad_norm": 0.8981325610332586,
+      "learning_rate": 3.3545594628553425e-06,
+      "loss": 0.0271,
+      "step": 4584
+    },
+    {
+      "epoch": 16.375,
+      "grad_norm": 1.1863917725119442,
+      "learning_rate": 3.3481498880078523e-06,
+      "loss": 0.0195,
+      "step": 4585
+    },
+    {
+      "epoch": 16.37857142857143,
+      "grad_norm": 1.5960903293257407,
+      "learning_rate": 3.3417458830001514e-06,
+      "loss": 0.027,
+      "step": 4586
+    },
+    {
+      "epoch": 16.382142857142856,
+      "grad_norm": 0.8495890499840826,
+      "learning_rate": 3.3353474499743112e-06,
+      "loss": 0.0285,
+      "step": 4587
+    },
+    {
+      "epoch": 16.385714285714286,
+      "grad_norm": 1.756950024043528,
+      "learning_rate": 3.3289545910705235e-06,
+      "loss": 0.0302,
+      "step": 4588
+    },
+    {
+      "epoch": 16.389285714285716,
+      "grad_norm": 1.2953625999022262,
+      "learning_rate": 3.3225673084271182e-06,
+      "loss": 0.0298,
+      "step": 4589
+    },
+    {
+      "epoch": 16.392857142857142,
+      "grad_norm": 0.760745031301915,
+      "learning_rate": 3.316185604180573e-06,
+      "loss": 0.0238,
+      "step": 4590
+    },
+    {
+      "epoch": 16.396428571428572,
+      "grad_norm": 0.4072412592375771,
+      "learning_rate": 3.3098094804654778e-06,
+      "loss": 0.0064,
+      "step": 4591
+    },
+    {
+      "epoch": 16.4,
+      "grad_norm": 2.850083491855655,
+      "learning_rate": 3.3034389394145806e-06,
+      "loss": 0.0415,
+      "step": 4592
+    },
+    {
+      "epoch": 16.40357142857143,
+      "grad_norm": 2.4672241931920276,
+      "learning_rate": 3.297073983158736e-06,
+      "loss": 0.0505,
+      "step": 4593
+    },
+    {
+      "epoch": 16.40714285714286,
+      "grad_norm": 0.666340740810865,
+      "learning_rate": 3.2907146138269575e-06,
+      "loss": 0.0167,
+      "step": 4594
+    },
+    {
+      "epoch": 16.410714285714285,
+      "grad_norm": 1.957775140759181,
+      "learning_rate": 3.284360833546363e-06,
+      "loss": 0.029,
+      "step": 4595
+    },
+    {
+      "epoch": 16.414285714285715,
+      "grad_norm": 0.5145935826105209,
+      "learning_rate": 3.2780126444422256e-06,
+      "loss": 0.0098,
+      "step": 4596
+    },
+    {
+      "epoch": 16.417857142857144,
+      "grad_norm": 1.079804618008352,
+      "learning_rate": 3.2716700486379273e-06,
+      "loss": 0.0386,
+      "step": 4597
+    },
+    {
+      "epoch": 16.42142857142857,
+      "grad_norm": 1.020812640339745,
+      "learning_rate": 3.265333048254997e-06,
+      "loss": 0.0214,
+      "step": 4598
+    },
+    {
+      "epoch": 16.425,
+      "grad_norm": 1.1113863173033696,
+      "learning_rate": 3.2590016454130734e-06,
+      "loss": 0.0356,
+      "step": 4599
+    },
+    {
+      "epoch": 16.428571428571427,
+      "grad_norm": 1.1249342627450603,
+      "learning_rate": 3.2526758422299466e-06,
+      "loss": 0.0122,
+      "step": 4600
+    },
+    {
+      "epoch": 16.432142857142857,
+      "grad_norm": 1.1758496995855994,
+      "learning_rate": 3.2463556408215122e-06,
+      "loss": 0.0254,
+      "step": 4601
+    },
+    {
+      "epoch": 16.435714285714287,
+      "grad_norm": 0.7456019047584492,
+      "learning_rate": 3.240041043301796e-06,
+      "loss": 0.0169,
+      "step": 4602
+    },
+    {
+      "epoch": 16.439285714285713,
+      "grad_norm": 1.0751454188440637,
+      "learning_rate": 3.233732051782965e-06,
+      "loss": 0.0493,
+      "step": 4603
+    },
+    {
+      "epoch": 16.442857142857143,
+      "grad_norm": 2.2190637638960293,
+      "learning_rate": 3.2274286683752877e-06,
+      "loss": 0.0442,
+      "step": 4604
+    },
+    {
+      "epoch": 16.446428571428573,
+      "grad_norm": 1.3890579988899614,
+      "learning_rate": 3.221130895187177e-06,
+      "loss": 0.0668,
+      "step": 4605
+    },
+    {
+      "epoch": 16.45,
+      "grad_norm": 1.51770686434472,
+      "learning_rate": 3.214838734325154e-06,
+      "loss": 0.0292,
+      "step": 4606
+    },
+    {
+      "epoch": 16.45357142857143,
+      "grad_norm": 0.8211429665301386,
+      "learning_rate": 3.2085521878938785e-06,
+      "loss": 0.021,
+      "step": 4607
+    },
+    {
+      "epoch": 16.457142857142856,
+      "grad_norm": 1.027405457679702,
+      "learning_rate": 3.2022712579961125e-06,
+      "loss": 0.0314,
+      "step": 4608
+    },
+    {
+      "epoch": 16.460714285714285,
+      "grad_norm": 0.45578969604207464,
+      "learning_rate": 3.1959959467327596e-06,
+      "loss": 0.0105,
+      "step": 4609
+    },
+    {
+      "epoch": 16.464285714285715,
+      "grad_norm": 1.0550099846763752,
+      "learning_rate": 3.189726256202825e-06,
+      "loss": 0.0397,
+      "step": 4610
+    },
+    {
+      "epoch": 16.46785714285714,
+      "grad_norm": 1.2834988526180626,
+      "learning_rate": 3.1834621885034566e-06,
+      "loss": 0.0639,
+      "step": 4611
+    },
+    {
+      "epoch": 16.47142857142857,
+      "grad_norm": 0.7042343310262464,
+      "learning_rate": 3.177203745729893e-06,
+      "loss": 0.0233,
+      "step": 4612
+    },
+    {
+      "epoch": 16.475,
+      "grad_norm": 1.3311804831276477,
+      "learning_rate": 3.17095092997552e-06,
+      "loss": 0.0358,
+      "step": 4613
+    },
+    {
+      "epoch": 16.478571428571428,
+      "grad_norm": 2.0384675410335045,
+      "learning_rate": 3.1647037433318206e-06,
+      "loss": 0.0573,
+      "step": 4614
+    },
+    {
+      "epoch": 16.482142857142858,
+      "grad_norm": 0.8536529742847849,
+      "learning_rate": 3.1584621878884004e-06,
+      "loss": 0.0209,
+      "step": 4615
+    },
+    {
+      "epoch": 16.485714285714284,
+      "grad_norm": 0.9921091105768822,
+      "learning_rate": 3.1522262657329892e-06,
+      "loss": 0.0336,
+      "step": 4616
+    },
+    {
+      "epoch": 16.489285714285714,
+      "grad_norm": 0.5099259219613381,
+      "learning_rate": 3.1459959789514192e-06,
+      "loss": 0.0088,
+      "step": 4617
+    },
+    {
+      "epoch": 16.492857142857144,
+      "grad_norm": 0.8001334106660839,
+      "learning_rate": 3.1397713296276566e-06,
+      "loss": 0.0181,
+      "step": 4618
+    },
+    {
+      "epoch": 16.49642857142857,
+      "grad_norm": 0.9029038538920481,
+      "learning_rate": 3.1335523198437557e-06,
+      "loss": 0.0165,
+      "step": 4619
+    },
+    {
+      "epoch": 16.5,
+      "grad_norm": 1.3492724939097387,
+      "learning_rate": 3.1273389516799126e-06,
+      "loss": 0.0329,
+      "step": 4620
+    },
+    {
+      "epoch": 16.50357142857143,
+      "grad_norm": 0.8978176411424224,
+      "learning_rate": 3.121131227214411e-06,
+      "loss": 0.0238,
+      "step": 4621
+    },
+    {
+      "epoch": 16.507142857142856,
+      "grad_norm": 0.910567016169253,
+      "learning_rate": 3.1149291485236687e-06,
+      "loss": 0.015,
+      "step": 4622
+    },
+    {
+      "epoch": 16.510714285714286,
+      "grad_norm": 1.1283398321187421,
+      "learning_rate": 3.108732717682197e-06,
+      "loss": 0.029,
+      "step": 4623
+    },
+    {
+      "epoch": 16.514285714285712,
+      "grad_norm": 0.6520969866760938,
+      "learning_rate": 3.102541936762633e-06,
+      "loss": 0.0178,
+      "step": 4624
+    },
+    {
+      "epoch": 16.517857142857142,
+      "grad_norm": 1.2719722983932724,
+      "learning_rate": 3.0963568078357143e-06,
+      "loss": 0.0272,
+      "step": 4625
+    },
+    {
+      "epoch": 16.521428571428572,
+      "grad_norm": 0.8310513989661229,
+      "learning_rate": 3.0901773329702877e-06,
+      "loss": 0.0139,
+      "step": 4626
+    },
+    {
+      "epoch": 16.525,
+      "grad_norm": 1.0228801584489482,
+      "learning_rate": 3.084003514233316e-06,
+      "loss": 0.0354,
+      "step": 4627
+    },
+    {
+      "epoch": 16.52857142857143,
+      "grad_norm": 1.5333876372734292,
+      "learning_rate": 3.077835353689862e-06,
+      "loss": 0.0492,
+      "step": 4628
+    },
+    {
+      "epoch": 16.53214285714286,
+      "grad_norm": 0.8152822439012507,
+      "learning_rate": 3.071672853403105e-06,
+      "loss": 0.012,
+      "step": 4629
+    },
+    {
+      "epoch": 16.535714285714285,
+      "grad_norm": 1.1510654751609084,
+      "learning_rate": 3.0655160154343177e-06,
+      "loss": 0.0313,
+      "step": 4630
+    },
+    {
+      "epoch": 16.539285714285715,
+      "grad_norm": 0.7735091779619782,
+      "learning_rate": 3.059364841842898e-06,
+      "loss": 0.03,
+      "step": 4631
+    },
+    {
+      "epoch": 16.542857142857144,
+      "grad_norm": 0.8101195661087374,
+      "learning_rate": 3.053219334686326e-06,
+      "loss": 0.0156,
+      "step": 4632
+    },
+    {
+      "epoch": 16.54642857142857,
+      "grad_norm": 0.5473245181953326,
+      "learning_rate": 3.0470794960202112e-06,
+      "loss": 0.0106,
+      "step": 4633
+    },
+    {
+      "epoch": 16.55,
+      "grad_norm": 1.2938160136237127,
+      "learning_rate": 3.040945327898244e-06,
+      "loss": 0.0285,
+      "step": 4634
+    },
+    {
+      "epoch": 16.553571428571427,
+      "grad_norm": 1.6775429133317619,
+      "learning_rate": 3.034816832372234e-06,
+      "loss": 0.0398,
+      "step": 4635
+    },
+    {
+      "epoch": 16.557142857142857,
+      "grad_norm": 1.4183719271343707,
+      "learning_rate": 3.0286940114920837e-06,
+      "loss": 0.029,
+      "step": 4636
+    },
+    {
+      "epoch": 16.560714285714287,
+      "grad_norm": 0.7738894995223476,
+      "learning_rate": 3.0225768673058087e-06,
+      "loss": 0.009,
+      "step": 4637
+    },
+    {
+      "epoch": 16.564285714285713,
+      "grad_norm": 1.6966063341920719,
+      "learning_rate": 3.016465401859512e-06,
+      "loss": 0.0274,
+      "step": 4638
+    },
+    {
+      "epoch": 16.567857142857143,
+      "grad_norm": 0.6856336049702227,
+      "learning_rate": 3.010359617197405e-06,
+      "loss": 0.0312,
+      "step": 4639
+    },
+    {
+      "epoch": 16.571428571428573,
+      "grad_norm": 2.2373383717018056,
+      "learning_rate": 3.0042595153618004e-06,
+      "loss": 0.0606,
+      "step": 4640
+    },
+    {
+      "epoch": 16.575,
+      "grad_norm": 0.4464441415473557,
+      "learning_rate": 2.9981650983931044e-06,
+      "loss": 0.0085,
+      "step": 4641
+    },
+    {
+      "epoch": 16.57857142857143,
+      "grad_norm": 0.9308708007325788,
+      "learning_rate": 2.9920763683298282e-06,
+      "loss": 0.0259,
+      "step": 4642
+    },
+    {
+      "epoch": 16.582142857142856,
+      "grad_norm": 0.8035634102860653,
+      "learning_rate": 2.9859933272085716e-06,
+      "loss": 0.0135,
+      "step": 4643
+    },
+    {
+      "epoch": 16.585714285714285,
+      "grad_norm": 1.2174553231720144,
+      "learning_rate": 2.9799159770640474e-06,
+      "loss": 0.0186,
+      "step": 4644
+    },
+    {
+      "epoch": 16.589285714285715,
+      "grad_norm": 0.8591849175118619,
+      "learning_rate": 2.9738443199290445e-06,
+      "loss": 0.017,
+      "step": 4645
+    },
+    {
+      "epoch": 16.59285714285714,
+      "grad_norm": 0.6738793049601218,
+      "learning_rate": 2.9677783578344655e-06,
+      "loss": 0.0198,
+      "step": 4646
+    },
+    {
+      "epoch": 16.59642857142857,
+      "grad_norm": 1.2092954712200414,
+      "learning_rate": 2.9617180928092916e-06,
+      "loss": 0.0415,
+      "step": 4647
+    },
+    {
+      "epoch": 16.6,
+      "grad_norm": 0.9822077762270148,
+      "learning_rate": 2.9556635268806165e-06,
+      "loss": 0.0256,
+      "step": 4648
+    },
+    {
+      "epoch": 16.603571428571428,
+      "grad_norm": 0.5687513791059137,
+      "learning_rate": 2.949614662073612e-06,
+      "loss": 0.011,
+      "step": 4649
+    },
+    {
+      "epoch": 16.607142857142858,
+      "grad_norm": 1.5794800627719963,
+      "learning_rate": 2.943571500411555e-06,
+      "loss": 0.0307,
+      "step": 4650
+    },
+    {
+      "epoch": 16.610714285714288,
+      "grad_norm": 0.9101207755915429,
+      "learning_rate": 2.9375340439158063e-06,
+      "loss": 0.0295,
+      "step": 4651
+    },
+    {
+      "epoch": 16.614285714285714,
+      "grad_norm": 1.6273120139256816,
+      "learning_rate": 2.9315022946058148e-06,
+      "loss": 0.028,
+      "step": 4652
+    },
+    {
+      "epoch": 16.617857142857144,
+      "grad_norm": 1.1283197410752743,
+      "learning_rate": 2.9254762544991377e-06,
+      "loss": 0.0262,
+      "step": 4653
+    },
+    {
+      "epoch": 16.62142857142857,
+      "grad_norm": 1.3990894019929292,
+      "learning_rate": 2.919455925611401e-06,
+      "loss": 0.0429,
+      "step": 4654
+    },
+    {
+      "epoch": 16.625,
+      "grad_norm": 1.3201288171412688,
+      "learning_rate": 2.9134413099563397e-06,
+      "loss": 0.0388,
+      "step": 4655
+    },
+    {
+      "epoch": 16.62857142857143,
+      "grad_norm": 2.8304658218458667,
+      "learning_rate": 2.9074324095457607e-06,
+      "loss": 0.0643,
+      "step": 4656
+    },
+    {
+      "epoch": 16.632142857142856,
+      "grad_norm": 1.5194097728702038,
+      "learning_rate": 2.901429226389578e-06,
+      "loss": 0.0304,
+      "step": 4657
+    },
+    {
+      "epoch": 16.635714285714286,
+      "grad_norm": 1.0086727038587289,
+      "learning_rate": 2.8954317624957727e-06,
+      "loss": 0.0187,
+      "step": 4658
+    },
+    {
+      "epoch": 16.639285714285712,
+      "grad_norm": 1.1557375000236103,
+      "learning_rate": 2.889440019870433e-06,
+      "loss": 0.0217,
+      "step": 4659
+    },
+    {
+      "epoch": 16.642857142857142,
+      "grad_norm": 1.208567187067679,
+      "learning_rate": 2.883454000517711e-06,
+      "loss": 0.0282,
+      "step": 4660
+    },
+    {
+      "epoch": 16.646428571428572,
+      "grad_norm": 1.104669388908561,
+      "learning_rate": 2.8774737064398685e-06,
+      "loss": 0.0229,
+      "step": 4661
+    },
+    {
+      "epoch": 16.65,
+      "grad_norm": 1.139610615707823,
+      "learning_rate": 2.871499139637237e-06,
+      "loss": 0.0317,
+      "step": 4662
+    },
+    {
+      "epoch": 16.65357142857143,
+      "grad_norm": 2.2192143793571,
+      "learning_rate": 2.8655303021082283e-06,
+      "loss": 0.0344,
+      "step": 4663
+    },
+    {
+      "epoch": 16.65714285714286,
+      "grad_norm": 1.1060890287402803,
+      "learning_rate": 2.8595671958493553e-06,
+      "loss": 0.0266,
+      "step": 4664
+    },
+    {
+      "epoch": 16.660714285714285,
+      "grad_norm": 1.0708978366963178,
+      "learning_rate": 2.853609822855199e-06,
+      "loss": 0.023,
+      "step": 4665
+    },
+    {
+      "epoch": 16.664285714285715,
+      "grad_norm": 1.7417301480557787,
+      "learning_rate": 2.8476581851184314e-06,
+      "loss": 0.025,
+      "step": 4666
+    },
+    {
+      "epoch": 16.667857142857144,
+      "grad_norm": 0.9129400458672893,
+      "learning_rate": 2.8417122846297897e-06,
+      "loss": 0.0285,
+      "step": 4667
+    },
+    {
+      "epoch": 16.67142857142857,
+      "grad_norm": 0.6768817733036536,
+      "learning_rate": 2.8357721233781267e-06,
+      "loss": 0.0103,
+      "step": 4668
+    },
+    {
+      "epoch": 16.675,
+      "grad_norm": 2.0773883984965615,
+      "learning_rate": 2.82983770335034e-06,
+      "loss": 0.0806,
+      "step": 4669
+    },
+    {
+      "epoch": 16.678571428571427,
+      "grad_norm": 0.9349857646826576,
+      "learning_rate": 2.823909026531417e-06,
+      "loss": 0.0234,
+      "step": 4670
+    },
+    {
+      "epoch": 16.682142857142857,
+      "grad_norm": 1.1692825472688528,
+      "learning_rate": 2.817986094904439e-06,
+      "loss": 0.0228,
+      "step": 4671
+    },
+    {
+      "epoch": 16.685714285714287,
+      "grad_norm": 1.0313049954458027,
+      "learning_rate": 2.8120689104505404e-06,
+      "loss": 0.0139,
+      "step": 4672
+    },
+    {
+      "epoch": 16.689285714285713,
+      "grad_norm": 1.4696449041652362,
+      "learning_rate": 2.806157475148961e-06,
+      "loss": 0.0247,
+      "step": 4673
+    },
+    {
+      "epoch": 16.692857142857143,
+      "grad_norm": 1.0778764271023993,
+      "learning_rate": 2.800251790976991e-06,
+      "loss": 0.0405,
+      "step": 4674
+    },
+    {
+      "epoch": 16.696428571428573,
+      "grad_norm": 0.5075417304519372,
+      "learning_rate": 2.7943518599100207e-06,
+      "loss": 0.0132,
+      "step": 4675
+    },
+    {
+      "epoch": 16.7,
+      "grad_norm": 0.7434635073662726,
+      "learning_rate": 2.788457683921495e-06,
+      "loss": 0.0177,
+      "step": 4676
+    },
+    {
+      "epoch": 16.70357142857143,
+      "grad_norm": 1.745202470657059,
+      "learning_rate": 2.7825692649829505e-06,
+      "loss": 0.0243,
+      "step": 4677
+    },
+    {
+      "epoch": 16.707142857142856,
+      "grad_norm": 1.928514996848797,
+      "learning_rate": 2.7766866050639828e-06,
+      "loss": 0.0327,
+      "step": 4678
+    },
+    {
+      "epoch": 16.710714285714285,
+      "grad_norm": 1.4407431656578653,
+      "learning_rate": 2.7708097061322804e-06,
+      "loss": 0.0256,
+      "step": 4679
+    },
+    {
+      "epoch": 16.714285714285715,
+      "grad_norm": 1.540701317598601,
+      "learning_rate": 2.7649385701535856e-06,
+      "loss": 0.0372,
+      "step": 4680
+    },
+    {
+      "epoch": 16.71785714285714,
+      "grad_norm": 0.8047009824721665,
+      "learning_rate": 2.7590731990917263e-06,
+      "loss": 0.0144,
+      "step": 4681
+    },
+    {
+      "epoch": 16.72142857142857,
+      "grad_norm": 0.5627743317527769,
+      "learning_rate": 2.753213594908597e-06,
+      "loss": 0.0127,
+      "step": 4682
+    },
+    {
+      "epoch": 16.725,
+      "grad_norm": 0.6383604312344275,
+      "learning_rate": 2.7473597595641565e-06,
+      "loss": 0.0137,
+      "step": 4683
+    },
+    {
+      "epoch": 16.728571428571428,
+      "grad_norm": 0.5917042586588401,
+      "learning_rate": 2.741511695016452e-06,
+      "loss": 0.0116,
+      "step": 4684
+    },
+    {
+      "epoch": 16.732142857142858,
+      "grad_norm": 2.2970930912013743,
+      "learning_rate": 2.735669403221579e-06,
+      "loss": 0.0406,
+      "step": 4685
+    },
+    {
+      "epoch": 16.735714285714288,
+      "grad_norm": 0.5270313610064824,
+      "learning_rate": 2.729832886133721e-06,
+      "loss": 0.0122,
+      "step": 4686
+    },
+    {
+      "epoch": 16.739285714285714,
+      "grad_norm": 1.1975724048350043,
+      "learning_rate": 2.7240021457051136e-06,
+      "loss": 0.0158,
+      "step": 4687
+    },
+    {
+      "epoch": 16.742857142857144,
+      "grad_norm": 0.9743472685241252,
+      "learning_rate": 2.718177183886077e-06,
+      "loss": 0.0244,
+      "step": 4688
+    },
+    {
+      "epoch": 16.74642857142857,
+      "grad_norm": 1.2117029483789847,
+      "learning_rate": 2.7123580026249794e-06,
+      "loss": 0.0328,
+      "step": 4689
+    },
+    {
+      "epoch": 16.75,
+      "grad_norm": 1.963350167973712,
+      "learning_rate": 2.7065446038682752e-06,
+      "loss": 0.0306,
+      "step": 4690
+    },
+    {
+      "epoch": 16.75357142857143,
+      "grad_norm": 1.0688511285076316,
+      "learning_rate": 2.700736989560466e-06,
+      "loss": 0.0354,
+      "step": 4691
+    },
+    {
+      "epoch": 16.757142857142856,
+      "grad_norm": 0.3657029188906679,
+      "learning_rate": 2.6949351616441366e-06,
+      "loss": 0.0067,
+      "step": 4692
+    },
+    {
+      "epoch": 16.760714285714286,
+      "grad_norm": 1.3082144468648509,
+      "learning_rate": 2.689139122059925e-06,
+      "loss": 0.0404,
+      "step": 4693
+    },
+    {
+      "epoch": 16.764285714285712,
+      "grad_norm": 2.1052416807728056,
+      "learning_rate": 2.6833488727465274e-06,
+      "loss": 0.0408,
+      "step": 4694
+    },
+    {
+      "epoch": 16.767857142857142,
+      "grad_norm": 1.6824003296830545,
+      "learning_rate": 2.677564415640721e-06,
+      "loss": 0.0577,
+      "step": 4695
+    },
+    {
+      "epoch": 16.771428571428572,
+      "grad_norm": 1.3395929833904072,
+      "learning_rate": 2.6717857526773294e-06,
+      "loss": 0.0102,
+      "step": 4696
+    },
+    {
+      "epoch": 16.775,
+      "grad_norm": 1.1614528364178311,
+      "learning_rate": 2.666012885789251e-06,
+      "loss": 0.0441,
+      "step": 4697
+    },
+    {
+      "epoch": 16.77857142857143,
+      "grad_norm": 0.8803926505984481,
+      "learning_rate": 2.6602458169074295e-06,
+      "loss": 0.0224,
+      "step": 4698
+    },
+    {
+      "epoch": 16.78214285714286,
+      "grad_norm": 2.158496172134839,
+      "learning_rate": 2.6544845479608917e-06,
+      "loss": 0.0257,
+      "step": 4699
+    },
+    {
+      "epoch": 16.785714285714285,
+      "grad_norm": 0.905195490815638,
+      "learning_rate": 2.6487290808766975e-06,
+      "loss": 0.0316,
+      "step": 4700
+    },
+    {
+      "epoch": 16.789285714285715,
+      "grad_norm": 0.7087167910267272,
+      "learning_rate": 2.6429794175799915e-06,
+      "loss": 0.0232,
+      "step": 4701
+    },
+    {
+      "epoch": 16.792857142857144,
+      "grad_norm": 1.8421751771686414,
+      "learning_rate": 2.6372355599939583e-06,
+      "loss": 0.017,
+      "step": 4702
+    },
+    {
+      "epoch": 16.79642857142857,
+      "grad_norm": 0.33478243641221156,
+      "learning_rate": 2.631497510039853e-06,
+      "loss": 0.0072,
+      "step": 4703
+    },
+    {
+      "epoch": 16.8,
+      "grad_norm": 0.5735073483492211,
+      "learning_rate": 2.6257652696369773e-06,
+      "loss": 0.0125,
+      "step": 4704
+    },
+    {
+      "epoch": 16.803571428571427,
+      "grad_norm": 1.5610944607942059,
+      "learning_rate": 2.6200388407027034e-06,
+      "loss": 0.0341,
+      "step": 4705
+    },
+    {
+      "epoch": 16.807142857142857,
+      "grad_norm": 0.4480414345192395,
+      "learning_rate": 2.6143182251524456e-06,
+      "loss": 0.012,
+      "step": 4706
+    },
+    {
+      "epoch": 16.810714285714287,
+      "grad_norm": 0.48117338421739686,
+      "learning_rate": 2.608603424899678e-06,
+      "loss": 0.0056,
+      "step": 4707
+    },
+    {
+      "epoch": 16.814285714285713,
+      "grad_norm": 0.8173641686381258,
+      "learning_rate": 2.6028944418559367e-06,
+      "loss": 0.0179,
+      "step": 4708
+    },
+    {
+      "epoch": 16.817857142857143,
+      "grad_norm": 1.7219501452041928,
+      "learning_rate": 2.597191277930802e-06,
+      "loss": 0.0327,
+      "step": 4709
+    },
+    {
+      "epoch": 16.821428571428573,
+      "grad_norm": 1.0722105999240321,
+      "learning_rate": 2.591493935031917e-06,
+      "loss": 0.0287,
+      "step": 4710
+    },
+    {
+      "epoch": 16.825,
+      "grad_norm": 1.0231103890760953,
+      "learning_rate": 2.5858024150649664e-06,
+      "loss": 0.0308,
+      "step": 4711
+    },
+    {
+      "epoch": 16.82857142857143,
+      "grad_norm": 1.5106225871056298,
+      "learning_rate": 2.580116719933703e-06,
+      "loss": 0.0395,
+      "step": 4712
+    },
+    {
+      "epoch": 16.832142857142856,
+      "grad_norm": 1.092888392993338,
+      "learning_rate": 2.574436851539912e-06,
+      "loss": 0.0188,
+      "step": 4713
+    },
+    {
+      "epoch": 16.835714285714285,
+      "grad_norm": 0.6198844871520761,
+      "learning_rate": 2.5687628117834497e-06,
+      "loss": 0.0172,
+      "step": 4714
+    },
+    {
+      "epoch": 16.839285714285715,
+      "grad_norm": 0.8293012658208252,
+      "learning_rate": 2.563094602562204e-06,
+      "loss": 0.0135,
+      "step": 4715
+    },
+    {
+      "epoch": 16.84285714285714,
+      "grad_norm": 1.2594119325008997,
+      "learning_rate": 2.5574322257721275e-06,
+      "loss": 0.0217,
+      "step": 4716
+    },
+    {
+      "epoch": 16.84642857142857,
+      "grad_norm": 1.3055446590843978,
+      "learning_rate": 2.5517756833072115e-06,
+      "loss": 0.0277,
+      "step": 4717
+    },
+    {
+      "epoch": 16.85,
+      "grad_norm": 0.9199064407476283,
+      "learning_rate": 2.5461249770595074e-06,
+      "loss": 0.0203,
+      "step": 4718
+    },
+    {
+      "epoch": 16.853571428571428,
+      "grad_norm": 1.2106924200756277,
+      "learning_rate": 2.540480108919101e-06,
+      "loss": 0.021,
+      "step": 4719
+    },
+    {
+      "epoch": 16.857142857142858,
+      "grad_norm": 0.8564637055516642,
+      "learning_rate": 2.5348410807741286e-06,
+      "loss": 0.0308,
+      "step": 4720
+    },
+    {
+      "epoch": 16.860714285714288,
+      "grad_norm": 1.3090381083150573,
+      "learning_rate": 2.5292078945107832e-06,
+      "loss": 0.0294,
+      "step": 4721
+    },
+    {
+      "epoch": 16.864285714285714,
+      "grad_norm": 1.1494135458587744,
+      "learning_rate": 2.5235805520132917e-06,
+      "loss": 0.0316,
+      "step": 4722
+    },
+    {
+      "epoch": 16.867857142857144,
+      "grad_norm": 1.2057465706521218,
+      "learning_rate": 2.5179590551639365e-06,
+      "loss": 0.0375,
+      "step": 4723
+    },
+    {
+      "epoch": 16.87142857142857,
+      "grad_norm": 0.8753596092680567,
+      "learning_rate": 2.512343405843032e-06,
+      "loss": 0.0254,
+      "step": 4724
+    },
+    {
+      "epoch": 16.875,
+      "grad_norm": 1.4211977287893132,
+      "learning_rate": 2.5067336059289502e-06,
+      "loss": 0.0405,
+      "step": 4725
+    },
+    {
+      "epoch": 16.87857142857143,
+      "grad_norm": 0.8759592688755455,
+      "learning_rate": 2.501129657298096e-06,
+      "loss": 0.0381,
+      "step": 4726
+    },
+    {
+      "epoch": 16.882142857142856,
+      "grad_norm": 1.4832326336176453,
+      "learning_rate": 2.4955315618249263e-06,
+      "loss": 0.0401,
+      "step": 4727
+    },
+    {
+      "epoch": 16.885714285714286,
+      "grad_norm": 1.7942765947027033,
+      "learning_rate": 2.4899393213819313e-06,
+      "loss": 0.0256,
+      "step": 4728
+    },
+    {
+      "epoch": 16.889285714285712,
+      "grad_norm": 1.2904341837627267,
+      "learning_rate": 2.484352937839651e-06,
+      "loss": 0.0259,
+      "step": 4729
+    },
+    {
+      "epoch": 16.892857142857142,
+      "grad_norm": 2.537433876439498,
+      "learning_rate": 2.478772413066659e-06,
+      "loss": 0.0393,
+      "step": 4730
+    },
+    {
+      "epoch": 16.896428571428572,
+      "grad_norm": 1.516834845811811,
+      "learning_rate": 2.47319774892957e-06,
+      "loss": 0.0416,
+      "step": 4731
+    },
+    {
+      "epoch": 16.9,
+      "grad_norm": 1.1991168015321905,
+      "learning_rate": 2.467628947293048e-06,
+      "loss": 0.0227,
+      "step": 4732
+    },
+    {
+      "epoch": 16.90357142857143,
+      "grad_norm": 1.062462796517959,
+      "learning_rate": 2.4620660100197815e-06,
+      "loss": 0.0146,
+      "step": 4733
+    },
+    {
+      "epoch": 16.90714285714286,
+      "grad_norm": 1.2304719726202409,
+      "learning_rate": 2.4565089389705123e-06,
+      "loss": 0.0258,
+      "step": 4734
+    },
+    {
+      "epoch": 16.910714285714285,
+      "grad_norm": 0.6075050813620225,
+      "learning_rate": 2.4509577360040026e-06,
+      "loss": 0.0177,
+      "step": 4735
+    },
+    {
+      "epoch": 16.914285714285715,
+      "grad_norm": 1.3222963362667255,
+      "learning_rate": 2.445412402977072e-06,
+      "loss": 0.0385,
+      "step": 4736
+    },
+    {
+      "epoch": 16.917857142857144,
+      "grad_norm": 0.9041987875982604,
+      "learning_rate": 2.4398729417445586e-06,
+      "loss": 0.0323,
+      "step": 4737
+    },
+    {
+      "epoch": 16.92142857142857,
+      "grad_norm": 1.327601778876104,
+      "learning_rate": 2.434339354159354e-06,
+      "loss": 0.0249,
+      "step": 4738
+    },
+    {
+      "epoch": 16.925,
+      "grad_norm": 1.1562343755235034,
+      "learning_rate": 2.428811642072364e-06,
+      "loss": 0.046,
+      "step": 4739
+    },
+    {
+      "epoch": 16.928571428571427,
+      "grad_norm": 1.5429974463516534,
+      "learning_rate": 2.4232898073325495e-06,
+      "loss": 0.0335,
+      "step": 4740
+    },
+    {
+      "epoch": 16.932142857142857,
+      "grad_norm": 1.6771472302080757,
+      "learning_rate": 2.4177738517868865e-06,
+      "loss": 0.0269,
+      "step": 4741
+    },
+    {
+      "epoch": 16.935714285714287,
+      "grad_norm": 1.4692384927682292,
+      "learning_rate": 2.412263777280408e-06,
+      "loss": 0.0169,
+      "step": 4742
+    },
+    {
+      "epoch": 16.939285714285713,
+      "grad_norm": 1.299768290450774,
+      "learning_rate": 2.4067595856561577e-06,
+      "loss": 0.027,
+      "step": 4743
+    },
+    {
+      "epoch": 16.942857142857143,
+      "grad_norm": 1.2025483445627843,
+      "learning_rate": 2.401261278755218e-06,
+      "loss": 0.0143,
+      "step": 4744
+    },
+    {
+      "epoch": 16.946428571428573,
+      "grad_norm": 1.4927904291207839,
+      "learning_rate": 2.395768858416714e-06,
+      "loss": 0.0155,
+      "step": 4745
+    },
+    {
+      "epoch": 16.95,
+      "grad_norm": 1.4036092012856078,
+      "learning_rate": 2.390282326477784e-06,
+      "loss": 0.0218,
+      "step": 4746
+    },
+    {
+      "epoch": 16.95357142857143,
+      "grad_norm": 0.4209862924920148,
+      "learning_rate": 2.3848016847736123e-06,
+      "loss": 0.0141,
+      "step": 4747
+    },
+    {
+      "epoch": 16.957142857142856,
+      "grad_norm": 1.2218553882147556,
+      "learning_rate": 2.3793269351374004e-06,
+      "loss": 0.0081,
+      "step": 4748
+    },
+    {
+      "epoch": 16.960714285714285,
+      "grad_norm": 1.1291857572635815,
+      "learning_rate": 2.373858079400393e-06,
+      "loss": 0.0274,
+      "step": 4749
+    },
+    {
+      "epoch": 16.964285714285715,
+      "grad_norm": 0.5111478728401788,
+      "learning_rate": 2.3683951193918466e-06,
+      "loss": 0.0149,
+      "step": 4750
+    },
+    {
+      "epoch": 16.96785714285714,
+      "grad_norm": 1.241113541471217,
+      "learning_rate": 2.3629380569390612e-06,
+      "loss": 0.0295,
+      "step": 4751
+    },
+    {
+      "epoch": 16.97142857142857,
+      "grad_norm": 1.3286049843276384,
+      "learning_rate": 2.3574868938673533e-06,
+      "loss": 0.0399,
+      "step": 4752
+    },
+    {
+      "epoch": 16.975,
+      "grad_norm": 0.6266643603390298,
+      "learning_rate": 2.3520416320000728e-06,
+      "loss": 0.0174,
+      "step": 4753
+    },
+    {
+      "epoch": 16.978571428571428,
+      "grad_norm": 1.7425883642114743,
+      "learning_rate": 2.3466022731585893e-06,
+      "loss": 0.0346,
+      "step": 4754
+    },
+    {
+      "epoch": 16.982142857142858,
+      "grad_norm": 1.9266124384469576,
+      "learning_rate": 2.3411688191623074e-06,
+      "loss": 0.0262,
+      "step": 4755
+    },
+    {
+      "epoch": 16.985714285714288,
+      "grad_norm": 1.6145537010609015,
+      "learning_rate": 2.3357412718286466e-06,
+      "loss": 0.0412,
+      "step": 4756
+    },
+    {
+      "epoch": 16.989285714285714,
+      "grad_norm": 2.018172944295882,
+      "learning_rate": 2.330319632973053e-06,
+      "loss": 0.0395,
+      "step": 4757
+    },
+    {
+      "epoch": 16.992857142857144,
+      "grad_norm": 0.6119102212396907,
+      "learning_rate": 2.324903904409004e-06,
+      "loss": 0.0184,
+      "step": 4758
+    },
+    {
+      "epoch": 16.99642857142857,
+      "grad_norm": 2.0481372939982174,
+      "learning_rate": 2.319494087947989e-06,
+      "loss": 0.0283,
+      "step": 4759
+    },
+    {
+      "epoch": 17.0,
+      "grad_norm": 1.2396901357686805,
+      "learning_rate": 2.3140901853995313e-06,
+      "loss": 0.0216,
+      "step": 4760
+    },
+    {
+      "epoch": 17.00357142857143,
+      "grad_norm": 1.6788099232988416,
+      "learning_rate": 2.308692198571163e-06,
+      "loss": 0.0182,
+      "step": 4761
+    },
+    {
+      "epoch": 17.007142857142856,
+      "grad_norm": 1.5179266986504834,
+      "learning_rate": 2.303300129268451e-06,
+      "loss": 0.0494,
+      "step": 4762
+    },
+    {
+      "epoch": 17.010714285714286,
+      "grad_norm": 0.6160606059600789,
+      "learning_rate": 2.29791397929497e-06,
+      "loss": 0.0123,
+      "step": 4763
+    },
+    {
+      "epoch": 17.014285714285716,
+      "grad_norm": 0.6787467540580258,
+      "learning_rate": 2.2925337504523303e-06,
+      "loss": 0.015,
+      "step": 4764
+    },
+    {
+      "epoch": 17.017857142857142,
+      "grad_norm": 1.5094448192290526,
+      "learning_rate": 2.2871594445401433e-06,
+      "loss": 0.0457,
+      "step": 4765
+    },
+    {
+      "epoch": 17.021428571428572,
+      "grad_norm": 0.9585093664252607,
+      "learning_rate": 2.281791063356056e-06,
+      "loss": 0.01,
+      "step": 4766
+    },
+    {
+      "epoch": 17.025,
+      "grad_norm": 1.1619624345007233,
+      "learning_rate": 2.2764286086957198e-06,
+      "loss": 0.0266,
+      "step": 4767
+    },
+    {
+      "epoch": 17.02857142857143,
+      "grad_norm": 1.5235966272266321,
+      "learning_rate": 2.271072082352812e-06,
+      "loss": 0.0334,
+      "step": 4768
+    },
+    {
+      "epoch": 17.03214285714286,
+      "grad_norm": 1.4506604265977956,
+      "learning_rate": 2.265721486119028e-06,
+      "loss": 0.0378,
+      "step": 4769
+    },
+    {
+      "epoch": 17.035714285714285,
+      "grad_norm": 0.32792775750905784,
+      "learning_rate": 2.2603768217840693e-06,
+      "loss": 0.007,
+      "step": 4770
+    },
+    {
+      "epoch": 17.039285714285715,
+      "grad_norm": 1.0097524349939406,
+      "learning_rate": 2.2550380911356708e-06,
+      "loss": 0.0244,
+      "step": 4771
+    },
+    {
+      "epoch": 17.042857142857144,
+      "grad_norm": 1.8096398010463035,
+      "learning_rate": 2.2497052959595635e-06,
+      "loss": 0.0377,
+      "step": 4772
+    },
+    {
+      "epoch": 17.04642857142857,
+      "grad_norm": 1.1852591006147177,
+      "learning_rate": 2.2443784380395094e-06,
+      "loss": 0.0301,
+      "step": 4773
+    },
+    {
+      "epoch": 17.05,
+      "grad_norm": 1.4555562987387736,
+      "learning_rate": 2.2390575191572703e-06,
+      "loss": 0.0276,
+      "step": 4774
+    },
+    {
+      "epoch": 17.053571428571427,
+      "grad_norm": 0.7689845025548527,
+      "learning_rate": 2.233742541092636e-06,
+      "loss": 0.0128,
+      "step": 4775
+    },
+    {
+      "epoch": 17.057142857142857,
+      "grad_norm": 0.47290937337814803,
+      "learning_rate": 2.228433505623393e-06,
+      "loss": 0.0118,
+      "step": 4776
+    },
+    {
+      "epoch": 17.060714285714287,
+      "grad_norm": 1.2432513852008742,
+      "learning_rate": 2.2231304145253565e-06,
+      "loss": 0.03,
+      "step": 4777
+    },
+    {
+      "epoch": 17.064285714285713,
+      "grad_norm": 0.7641456048085765,
+      "learning_rate": 2.217833269572336e-06,
+      "loss": 0.0275,
+      "step": 4778
+    },
+    {
+      "epoch": 17.067857142857143,
+      "grad_norm": 1.6101946094237527,
+      "learning_rate": 2.2125420725361745e-06,
+      "loss": 0.0304,
+      "step": 4779
+    },
+    {
+      "epoch": 17.071428571428573,
+      "grad_norm": 0.6761338421882047,
+      "learning_rate": 2.2072568251867034e-06,
+      "loss": 0.0129,
+      "step": 4780
+    },
+    {
+      "epoch": 17.075,
+      "grad_norm": 2.3118162781507445,
+      "learning_rate": 2.2019775292917723e-06,
+      "loss": 0.0378,
+      "step": 4781
+    },
+    {
+      "epoch": 17.07857142857143,
+      "grad_norm": 1.16155432240221,
+      "learning_rate": 2.1967041866172467e-06,
+      "loss": 0.0376,
+      "step": 4782
+    },
+    {
+      "epoch": 17.082142857142856,
+      "grad_norm": 0.8136953997461924,
+      "learning_rate": 2.1914367989269892e-06,
+      "loss": 0.026,
+      "step": 4783
+    },
+    {
+      "epoch": 17.085714285714285,
+      "grad_norm": 0.5845830609620729,
+      "learning_rate": 2.1861753679828835e-06,
+      "loss": 0.0092,
+      "step": 4784
+    },
+    {
+      "epoch": 17.089285714285715,
+      "grad_norm": 0.7017870893059434,
+      "learning_rate": 2.1809198955448065e-06,
+      "loss": 0.0204,
+      "step": 4785
+    },
+    {
+      "epoch": 17.09285714285714,
+      "grad_norm": 0.712190073221824,
+      "learning_rate": 2.175670383370654e-06,
+      "loss": 0.0176,
+      "step": 4786
+    },
+    {
+      "epoch": 17.09642857142857,
+      "grad_norm": 0.3801393964341104,
+      "learning_rate": 2.1704268332163215e-06,
+      "loss": 0.0059,
+      "step": 4787
+    },
+    {
+      "epoch": 17.1,
+      "grad_norm": 1.2590229862672193,
+      "learning_rate": 2.165189246835715e-06,
+      "loss": 0.0289,
+      "step": 4788
+    },
+    {
+      "epoch": 17.103571428571428,
+      "grad_norm": 0.5943457132943293,
+      "learning_rate": 2.1599576259807354e-06,
+      "loss": 0.0143,
+      "step": 4789
+    },
+    {
+      "epoch": 17.107142857142858,
+      "grad_norm": 0.32127405726908126,
+      "learning_rate": 2.154731972401307e-06,
+      "loss": 0.0071,
+      "step": 4790
+    },
+    {
+      "epoch": 17.110714285714284,
+      "grad_norm": 1.0281966754113816,
+      "learning_rate": 2.149512287845337e-06,
+      "loss": 0.0325,
+      "step": 4791
+    },
+    {
+      "epoch": 17.114285714285714,
+      "grad_norm": 0.8362351399382772,
+      "learning_rate": 2.144298574058754e-06,
+      "loss": 0.0146,
+      "step": 4792
+    },
+    {
+      "epoch": 17.117857142857144,
+      "grad_norm": 1.7343889530690284,
+      "learning_rate": 2.1390908327854776e-06,
+      "loss": 0.0516,
+      "step": 4793
+    },
+    {
+      "epoch": 17.12142857142857,
+      "grad_norm": 1.8208981955868324,
+      "learning_rate": 2.1338890657674293e-06,
+      "loss": 0.024,
+      "step": 4794
+    },
+    {
+      "epoch": 17.125,
+      "grad_norm": 1.1905631252862028,
+      "learning_rate": 2.128693274744542e-06,
+      "loss": 0.0162,
+      "step": 4795
+    },
+    {
+      "epoch": 17.12857142857143,
+      "grad_norm": 1.2704318190240869,
+      "learning_rate": 2.1235034614547413e-06,
+      "loss": 0.0188,
+      "step": 4796
+    },
+    {
+      "epoch": 17.132142857142856,
+      "grad_norm": 2.2518724030454704,
+      "learning_rate": 2.11831962763396e-06,
+      "loss": 0.0354,
+      "step": 4797
+    },
+    {
+      "epoch": 17.135714285714286,
+      "grad_norm": 1.7691545138042362,
+      "learning_rate": 2.113141775016121e-06,
+      "loss": 0.0361,
+      "step": 4798
+    },
+    {
+      "epoch": 17.139285714285716,
+      "grad_norm": 1.0600551615492098,
+      "learning_rate": 2.1079699053331605e-06,
+      "loss": 0.0182,
+      "step": 4799
+    },
+    {
+      "epoch": 17.142857142857142,
+      "grad_norm": 1.6808621267612134,
+      "learning_rate": 2.102804020314995e-06,
+      "loss": 0.0282,
+      "step": 4800
+    },
+    {
+      "epoch": 17.146428571428572,
+      "grad_norm": 1.7695298366259409,
+      "learning_rate": 2.0976441216895593e-06,
+      "loss": 0.0304,
+      "step": 4801
+    },
+    {
+      "epoch": 17.15,
+      "grad_norm": 1.0996050264990758,
+      "learning_rate": 2.0924902111827694e-06,
+      "loss": 0.0276,
+      "step": 4802
+    },
+    {
+      "epoch": 17.15357142857143,
+      "grad_norm": 1.898019033810932,
+      "learning_rate": 2.0873422905185525e-06,
+      "loss": 0.0232,
+      "step": 4803
+    },
+    {
+      "epoch": 17.15714285714286,
+      "grad_norm": 1.6942891666871822,
+      "learning_rate": 2.08220036141882e-06,
+      "loss": 0.0463,
+      "step": 4804
+    },
+    {
+      "epoch": 17.160714285714285,
+      "grad_norm": 0.76225454549598,
+      "learning_rate": 2.0770644256034834e-06,
+      "loss": 0.0127,
+      "step": 4805
+    },
+    {
+      "epoch": 17.164285714285715,
+      "grad_norm": 1.3889220742788182,
+      "learning_rate": 2.0719344847904543e-06,
+      "loss": 0.02,
+      "step": 4806
+    },
+    {
+      "epoch": 17.167857142857144,
+      "grad_norm": 1.1754160344330287,
+      "learning_rate": 2.0668105406956275e-06,
+      "loss": 0.0232,
+      "step": 4807
+    },
+    {
+      "epoch": 17.17142857142857,
+      "grad_norm": 1.069502062717993,
+      "learning_rate": 2.0616925950329094e-06,
+      "loss": 0.0172,
+      "step": 4808
+    },
+    {
+      "epoch": 17.175,
+      "grad_norm": 1.5430359883083948,
+      "learning_rate": 2.0565806495141815e-06,
+      "loss": 0.015,
+      "step": 4809
+    },
+    {
+      "epoch": 17.178571428571427,
+      "grad_norm": 1.3866873814101623,
+      "learning_rate": 2.0514747058493346e-06,
+      "loss": 0.0265,
+      "step": 4810
+    },
+    {
+      "epoch": 17.182142857142857,
+      "grad_norm": 1.6994110284324888,
+      "learning_rate": 2.0463747657462376e-06,
+      "loss": 0.0389,
+      "step": 4811
+    },
+    {
+      "epoch": 17.185714285714287,
+      "grad_norm": 1.0704973187534292,
+      "learning_rate": 2.0412808309107632e-06,
+      "loss": 0.0199,
+      "step": 4812
+    },
+    {
+      "epoch": 17.189285714285713,
+      "grad_norm": 2.901254688824483,
+      "learning_rate": 2.0361929030467654e-06,
+      "loss": 0.0308,
+      "step": 4813
+    },
+    {
+      "epoch": 17.192857142857143,
+      "grad_norm": 0.7742793944276846,
+      "learning_rate": 2.0311109838560992e-06,
+      "loss": 0.0228,
+      "step": 4814
+    },
+    {
+      "epoch": 17.196428571428573,
+      "grad_norm": 1.3402247303431116,
+      "learning_rate": 2.026035075038597e-06,
+      "loss": 0.0269,
+      "step": 4815
+    },
+    {
+      "epoch": 17.2,
+      "grad_norm": 1.5824739325432544,
+      "learning_rate": 2.020965178292096e-06,
+      "loss": 0.0524,
+      "step": 4816
+    },
+    {
+      "epoch": 17.20357142857143,
+      "grad_norm": 0.7013571756342006,
+      "learning_rate": 2.01590129531241e-06,
+      "loss": 0.0143,
+      "step": 4817
+    },
+    {
+      "epoch": 17.207142857142856,
+      "grad_norm": 1.4138070895543393,
+      "learning_rate": 2.0108434277933454e-06,
+      "loss": 0.013,
+      "step": 4818
+    },
+    {
+      "epoch": 17.210714285714285,
+      "grad_norm": 1.8629080253155588,
+      "learning_rate": 2.005791577426699e-06,
+      "loss": 0.0331,
+      "step": 4819
+    },
+    {
+      "epoch": 17.214285714285715,
+      "grad_norm": 0.898174955786597,
+      "learning_rate": 2.0007457459022506e-06,
+      "loss": 0.016,
+      "step": 4820
+    },
+    {
+      "epoch": 17.21785714285714,
+      "grad_norm": 1.5825654211220357,
+      "learning_rate": 1.9957059349077723e-06,
+      "loss": 0.0195,
+      "step": 4821
+    },
+    {
+      "epoch": 17.22142857142857,
+      "grad_norm": 1.642315158733262,
+      "learning_rate": 1.9906721461290135e-06,
+      "loss": 0.024,
+      "step": 4822
+    },
+    {
+      "epoch": 17.225,
+      "grad_norm": 1.438594997273554,
+      "learning_rate": 1.9856443812497227e-06,
+      "loss": 0.025,
+      "step": 4823
+    },
+    {
+      "epoch": 17.228571428571428,
+      "grad_norm": 0.9536188138145985,
+      "learning_rate": 1.9806226419516195e-06,
+      "loss": 0.0167,
+      "step": 4824
+    },
+    {
+      "epoch": 17.232142857142858,
+      "grad_norm": 1.7468588534749432,
+      "learning_rate": 1.9756069299144176e-06,
+      "loss": 0.0316,
+      "step": 4825
+    },
+    {
+      "epoch": 17.235714285714284,
+      "grad_norm": 2.0090492185662865,
+      "learning_rate": 1.9705972468158064e-06,
+      "loss": 0.0277,
+      "step": 4826
+    },
+    {
+      "epoch": 17.239285714285714,
+      "grad_norm": 1.7070238246363412,
+      "learning_rate": 1.9655935943314695e-06,
+      "loss": 0.0232,
+      "step": 4827
+    },
+    {
+      "epoch": 17.242857142857144,
+      "grad_norm": 0.5051807718693688,
+      "learning_rate": 1.9605959741350644e-06,
+      "loss": 0.0103,
+      "step": 4828
+    },
+    {
+      "epoch": 17.24642857142857,
+      "grad_norm": 2.0174872257392193,
+      "learning_rate": 1.9556043878982356e-06,
+      "loss": 0.0243,
+      "step": 4829
+    },
+    {
+      "epoch": 17.25,
+      "grad_norm": 0.6256066447822731,
+      "learning_rate": 1.9506188372906056e-06,
+      "loss": 0.0162,
+      "step": 4830
+    },
+    {
+      "epoch": 17.25357142857143,
+      "grad_norm": 1.0000141150253241,
+      "learning_rate": 1.9456393239797754e-06,
+      "loss": 0.0223,
+      "step": 4831
+    },
+    {
+      "epoch": 17.257142857142856,
+      "grad_norm": 1.605396658956464,
+      "learning_rate": 1.9406658496313424e-06,
+      "loss": 0.0355,
+      "step": 4832
+    },
+    {
+      "epoch": 17.260714285714286,
+      "grad_norm": 0.503654290617971,
+      "learning_rate": 1.935698415908862e-06,
+      "loss": 0.0104,
+      "step": 4833
+    },
+    {
+      "epoch": 17.264285714285716,
+      "grad_norm": 1.5749906340719837,
+      "learning_rate": 1.930737024473888e-06,
+      "loss": 0.0186,
+      "step": 4834
+    },
+    {
+      "epoch": 17.267857142857142,
+      "grad_norm": 0.6424169735996684,
+      "learning_rate": 1.925781676985938e-06,
+      "loss": 0.0137,
+      "step": 4835
+    },
+    {
+      "epoch": 17.271428571428572,
+      "grad_norm": 0.6134665051817221,
+      "learning_rate": 1.9208323751025216e-06,
+      "loss": 0.0124,
+      "step": 4836
+    },
+    {
+      "epoch": 17.275,
+      "grad_norm": 2.154447094596528,
+      "learning_rate": 1.9158891204791154e-06,
+      "loss": 0.02,
+      "step": 4837
+    },
+    {
+      "epoch": 17.27857142857143,
+      "grad_norm": 0.4433446226049816,
+      "learning_rate": 1.910951914769175e-06,
+      "loss": 0.0062,
+      "step": 4838
+    },
+    {
+      "epoch": 17.28214285714286,
+      "grad_norm": 1.7321325206165077,
+      "learning_rate": 1.9060207596241454e-06,
+      "loss": 0.0351,
+      "step": 4839
+    },
+    {
+      "epoch": 17.285714285714285,
+      "grad_norm": 0.522562000379934,
+      "learning_rate": 1.9010956566934259e-06,
+      "loss": 0.0134,
+      "step": 4840
+    },
+    {
+      "epoch": 17.289285714285715,
+      "grad_norm": 1.1003273566127758,
+      "learning_rate": 1.8961766076244114e-06,
+      "loss": 0.0154,
+      "step": 4841
+    },
+    {
+      "epoch": 17.292857142857144,
+      "grad_norm": 1.1150506754748368,
+      "learning_rate": 1.8912636140624574e-06,
+      "loss": 0.0293,
+      "step": 4842
+    },
+    {
+      "epoch": 17.29642857142857,
+      "grad_norm": 1.8715827153426308,
+      "learning_rate": 1.8863566776509068e-06,
+      "loss": 0.0175,
+      "step": 4843
+    },
+    {
+      "epoch": 17.3,
+      "grad_norm": 0.6963014057768735,
+      "learning_rate": 1.8814558000310623e-06,
+      "loss": 0.0104,
+      "step": 4844
+    },
+    {
+      "epoch": 17.303571428571427,
+      "grad_norm": 1.6491728873688571,
+      "learning_rate": 1.876560982842215e-06,
+      "loss": 0.0308,
+      "step": 4845
+    },
+    {
+      "epoch": 17.307142857142857,
+      "grad_norm": 0.6537379363933794,
+      "learning_rate": 1.8716722277216125e-06,
+      "loss": 0.0128,
+      "step": 4846
+    },
+    {
+      "epoch": 17.310714285714287,
+      "grad_norm": 1.2128114778153885,
+      "learning_rate": 1.8667895363044918e-06,
+      "loss": 0.0364,
+      "step": 4847
+    },
+    {
+      "epoch": 17.314285714285713,
+      "grad_norm": 0.3910164817830802,
+      "learning_rate": 1.8619129102240507e-06,
+      "loss": 0.0058,
+      "step": 4848
+    },
+    {
+      "epoch": 17.317857142857143,
+      "grad_norm": 1.550128287932739,
+      "learning_rate": 1.8570423511114555e-06,
+      "loss": 0.0221,
+      "step": 4849
+    },
+    {
+      "epoch": 17.321428571428573,
+      "grad_norm": 1.20765160936562,
+      "learning_rate": 1.8521778605958563e-06,
+      "loss": 0.0139,
+      "step": 4850
+    },
+    {
+      "epoch": 17.325,
+      "grad_norm": 1.269472685801916,
+      "learning_rate": 1.847319440304358e-06,
+      "loss": 0.0143,
+      "step": 4851
+    },
+    {
+      "epoch": 17.32857142857143,
+      "grad_norm": 1.0747804841933728,
+      "learning_rate": 1.8424670918620524e-06,
+      "loss": 0.0171,
+      "step": 4852
+    },
+    {
+      "epoch": 17.332142857142856,
+      "grad_norm": 1.3681474990304994,
+      "learning_rate": 1.8376208168919807e-06,
+      "loss": 0.0288,
+      "step": 4853
+    },
+    {
+      "epoch": 17.335714285714285,
+      "grad_norm": 1.056722979489449,
+      "learning_rate": 1.8327806170151686e-06,
+      "loss": 0.015,
+      "step": 4854
+    },
+    {
+      "epoch": 17.339285714285715,
+      "grad_norm": 0.6470472273945911,
+      "learning_rate": 1.827946493850601e-06,
+      "loss": 0.0134,
+      "step": 4855
+    },
+    {
+      "epoch": 17.34285714285714,
+      "grad_norm": 0.49132524898871327,
+      "learning_rate": 1.8231184490152354e-06,
+      "loss": 0.0173,
+      "step": 4856
+    },
+    {
+      "epoch": 17.34642857142857,
+      "grad_norm": 1.199723294935427,
+      "learning_rate": 1.818296484123989e-06,
+      "loss": 0.0332,
+      "step": 4857
+    },
+    {
+      "epoch": 17.35,
+      "grad_norm": 0.9251511529279307,
+      "learning_rate": 1.8134806007897566e-06,
+      "loss": 0.0261,
+      "step": 4858
+    },
+    {
+      "epoch": 17.353571428571428,
+      "grad_norm": 0.9159723034087244,
+      "learning_rate": 1.8086708006233844e-06,
+      "loss": 0.011,
+      "step": 4859
+    },
+    {
+      "epoch": 17.357142857142858,
+      "grad_norm": 0.6376970482433116,
+      "learning_rate": 1.8038670852337014e-06,
+      "loss": 0.0146,
+      "step": 4860
+    },
+    {
+      "epoch": 17.360714285714284,
+      "grad_norm": 0.7196953294332574,
+      "learning_rate": 1.7990694562274847e-06,
+      "loss": 0.0096,
+      "step": 4861
+    },
+    {
+      "epoch": 17.364285714285714,
+      "grad_norm": 0.6278374977926299,
+      "learning_rate": 1.794277915209479e-06,
+      "loss": 0.0169,
+      "step": 4862
+    },
+    {
+      "epoch": 17.367857142857144,
+      "grad_norm": 1.4486633101760413,
+      "learning_rate": 1.789492463782403e-06,
+      "loss": 0.0287,
+      "step": 4863
+    },
+    {
+      "epoch": 17.37142857142857,
+      "grad_norm": 2.623080774578703,
+      "learning_rate": 1.7847131035469267e-06,
+      "loss": 0.023,
+      "step": 4864
+    },
+    {
+      "epoch": 17.375,
+      "grad_norm": 1.3949278706902721,
+      "learning_rate": 1.779939836101694e-06,
+      "loss": 0.0336,
+      "step": 4865
+    },
+    {
+      "epoch": 17.37857142857143,
+      "grad_norm": 1.565462870017105,
+      "learning_rate": 1.7751726630432986e-06,
+      "loss": 0.0388,
+      "step": 4866
+    },
+    {
+      "epoch": 17.382142857142856,
+      "grad_norm": 1.3271605329334295,
+      "learning_rate": 1.7704115859663028e-06,
+      "loss": 0.0265,
+      "step": 4867
+    },
+    {
+      "epoch": 17.385714285714286,
+      "grad_norm": 1.5640480991925048,
+      "learning_rate": 1.7656566064632286e-06,
+      "loss": 0.0428,
+      "step": 4868
+    },
+    {
+      "epoch": 17.389285714285716,
+      "grad_norm": 0.8094145175706486,
+      "learning_rate": 1.760907726124561e-06,
+      "loss": 0.0165,
+      "step": 4869
+    },
+    {
+      "epoch": 17.392857142857142,
+      "grad_norm": 0.8027171622702524,
+      "learning_rate": 1.7561649465387366e-06,
+      "loss": 0.0148,
+      "step": 4870
+    },
+    {
+      "epoch": 17.396428571428572,
+      "grad_norm": 0.9556208234671005,
+      "learning_rate": 1.7514282692921635e-06,
+      "loss": 0.0165,
+      "step": 4871
+    },
+    {
+      "epoch": 17.4,
+      "grad_norm": 1.4772974492437416,
+      "learning_rate": 1.7466976959691994e-06,
+      "loss": 0.0372,
+      "step": 4872
+    },
+    {
+      "epoch": 17.40357142857143,
+      "grad_norm": 0.7311572512664736,
+      "learning_rate": 1.7419732281521583e-06,
+      "loss": 0.012,
+      "step": 4873
+    },
+    {
+      "epoch": 17.40714285714286,
+      "grad_norm": 1.4024010914844682,
+      "learning_rate": 1.7372548674213251e-06,
+      "loss": 0.0196,
+      "step": 4874
+    },
+    {
+      "epoch": 17.410714285714285,
+      "grad_norm": 0.7507540806693126,
+      "learning_rate": 1.7325426153549284e-06,
+      "loss": 0.0149,
+      "step": 4875
+    },
+    {
+      "epoch": 17.414285714285715,
+      "grad_norm": 0.8472850176590015,
+      "learning_rate": 1.7278364735291608e-06,
+      "loss": 0.0218,
+      "step": 4876
+    },
+    {
+      "epoch": 17.417857142857144,
+      "grad_norm": 0.7752187684665649,
+      "learning_rate": 1.7231364435181673e-06,
+      "loss": 0.0211,
+      "step": 4877
+    },
+    {
+      "epoch": 17.42142857142857,
+      "grad_norm": 1.8161211384796463,
+      "learning_rate": 1.7184425268940552e-06,
+      "loss": 0.032,
+      "step": 4878
+    },
+    {
+      "epoch": 17.425,
+      "grad_norm": 1.5641126039604776,
+      "learning_rate": 1.7137547252268727e-06,
+      "loss": 0.0346,
+      "step": 4879
+    },
+    {
+      "epoch": 17.428571428571427,
+      "grad_norm": 1.704380355091173,
+      "learning_rate": 1.7090730400846433e-06,
+      "loss": 0.0155,
+      "step": 4880
+    },
+    {
+      "epoch": 17.432142857142857,
+      "grad_norm": 1.2724938232859464,
+      "learning_rate": 1.7043974730333235e-06,
+      "loss": 0.0411,
+      "step": 4881
+    },
+    {
+      "epoch": 17.435714285714287,
+      "grad_norm": 1.1092579429994263,
+      "learning_rate": 1.6997280256368397e-06,
+      "loss": 0.0297,
+      "step": 4882
+    },
+    {
+      "epoch": 17.439285714285713,
+      "grad_norm": 1.8897346831755844,
+      "learning_rate": 1.6950646994570607e-06,
+      "loss": 0.0174,
+      "step": 4883
+    },
+    {
+      "epoch": 17.442857142857143,
+      "grad_norm": 0.4557343238158434,
+      "learning_rate": 1.6904074960538165e-06,
+      "loss": 0.0076,
+      "step": 4884
+    },
+    {
+      "epoch": 17.446428571428573,
+      "grad_norm": 1.3929535401889073,
+      "learning_rate": 1.6857564169848807e-06,
+      "loss": 0.0497,
+      "step": 4885
+    },
+    {
+      "epoch": 17.45,
+      "grad_norm": 1.0329060489661783,
+      "learning_rate": 1.6811114638059822e-06,
+      "loss": 0.0218,
+      "step": 4886
+    },
+    {
+      "epoch": 17.45357142857143,
+      "grad_norm": 0.6896300816200639,
+      "learning_rate": 1.6764726380708029e-06,
+      "loss": 0.0112,
+      "step": 4887
+    },
+    {
+      "epoch": 17.457142857142856,
+      "grad_norm": 0.655823007745991,
+      "learning_rate": 1.6718399413309705e-06,
+      "loss": 0.0154,
+      "step": 4888
+    },
+    {
+      "epoch": 17.460714285714285,
+      "grad_norm": 0.8929215757829039,
+      "learning_rate": 1.667213375136072e-06,
+      "loss": 0.0205,
+      "step": 4889
+    },
+    {
+      "epoch": 17.464285714285715,
+      "grad_norm": 1.335341633328343,
+      "learning_rate": 1.6625929410336273e-06,
+      "loss": 0.0236,
+      "step": 4890
+    },
+    {
+      "epoch": 17.46785714285714,
+      "grad_norm": 1.0029637954023187,
+      "learning_rate": 1.6579786405691224e-06,
+      "loss": 0.0249,
+      "step": 4891
+    },
+    {
+      "epoch": 17.47142857142857,
+      "grad_norm": 1.823327829732967,
+      "learning_rate": 1.6533704752859802e-06,
+      "loss": 0.0377,
+      "step": 4892
+    },
+    {
+      "epoch": 17.475,
+      "grad_norm": 1.140594461600517,
+      "learning_rate": 1.6487684467255793e-06,
+      "loss": 0.0488,
+      "step": 4893
+    },
+    {
+      "epoch": 17.478571428571428,
+      "grad_norm": 1.7537950175681898,
+      "learning_rate": 1.6441725564272371e-06,
+      "loss": 0.0181,
+      "step": 4894
+    },
+    {
+      "epoch": 17.482142857142858,
+      "grad_norm": 1.4520244045172146,
+      "learning_rate": 1.6395828059282282e-06,
+      "loss": 0.0225,
+      "step": 4895
+    },
+    {
+      "epoch": 17.485714285714284,
+      "grad_norm": 0.291516808947317,
+      "learning_rate": 1.6349991967637601e-06,
+      "loss": 0.0043,
+      "step": 4896
+    },
+    {
+      "epoch": 17.489285714285714,
+      "grad_norm": 2.4677593317032525,
+      "learning_rate": 1.6304217304670046e-06,
+      "loss": 0.0457,
+      "step": 4897
+    },
+    {
+      "epoch": 17.492857142857144,
+      "grad_norm": 1.257656234495535,
+      "learning_rate": 1.6258504085690607e-06,
+      "loss": 0.0215,
+      "step": 4898
+    },
+    {
+      "epoch": 17.49642857142857,
+      "grad_norm": 1.046983953511427,
+      "learning_rate": 1.621285232598977e-06,
+      "loss": 0.0234,
+      "step": 4899
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 0.7705206879730521,
+      "learning_rate": 1.6167262040837583e-06,
+      "loss": 0.0149,
+      "step": 4900
+    },
+    {
+      "epoch": 17.50357142857143,
+      "grad_norm": 0.39244099399188326,
+      "learning_rate": 1.612173324548334e-06,
+      "loss": 0.0054,
+      "step": 4901
+    },
+    {
+      "epoch": 17.507142857142856,
+      "grad_norm": 0.7299925833832304,
+      "learning_rate": 1.6076265955155922e-06,
+      "loss": 0.0186,
+      "step": 4902
+    },
+    {
+      "epoch": 17.510714285714286,
+      "grad_norm": 0.7739950408540693,
+      "learning_rate": 1.6030860185063547e-06,
+      "loss": 0.0193,
+      "step": 4903
+    },
+    {
+      "epoch": 17.514285714285712,
+      "grad_norm": 0.6880396584508319,
+      "learning_rate": 1.5985515950393948e-06,
+      "loss": 0.0137,
+      "step": 4904
+    },
+    {
+      "epoch": 17.517857142857142,
+      "grad_norm": 1.4590892794727568,
+      "learning_rate": 1.5940233266314154e-06,
+      "loss": 0.0187,
+      "step": 4905
+    },
+    {
+      "epoch": 17.521428571428572,
+      "grad_norm": 2.8570967259152726,
+      "learning_rate": 1.5895012147970713e-06,
+      "loss": 0.0336,
+      "step": 4906
+    },
+    {
+      "epoch": 17.525,
+      "grad_norm": 1.2631669410194768,
+      "learning_rate": 1.5849852610489459e-06,
+      "loss": 0.0186,
+      "step": 4907
+    },
+    {
+      "epoch": 17.52857142857143,
+      "grad_norm": 8.72166685554689,
+      "learning_rate": 1.5804754668975819e-06,
+      "loss": 0.0334,
+      "step": 4908
+    },
+    {
+      "epoch": 17.53214285714286,
+      "grad_norm": 0.9999055398863553,
+      "learning_rate": 1.5759718338514418e-06,
+      "loss": 0.017,
+      "step": 4909
+    },
+    {
+      "epoch": 17.535714285714285,
+      "grad_norm": 0.8092753618783898,
+      "learning_rate": 1.571474363416936e-06,
+      "loss": 0.0202,
+      "step": 4910
+    },
+    {
+      "epoch": 17.539285714285715,
+      "grad_norm": 2.898161853808311,
+      "learning_rate": 1.566983057098419e-06,
+      "loss": 0.0436,
+      "step": 4911
+    },
+    {
+      "epoch": 17.542857142857144,
+      "grad_norm": 0.37106020691130215,
+      "learning_rate": 1.5624979163981692e-06,
+      "loss": 0.0062,
+      "step": 4912
+    },
+    {
+      "epoch": 17.54642857142857,
+      "grad_norm": 1.0970070762976827,
+      "learning_rate": 1.5580189428164193e-06,
+      "loss": 0.0322,
+      "step": 4913
+    },
+    {
+      "epoch": 17.55,
+      "grad_norm": 1.004365526521415,
+      "learning_rate": 1.5535461378513227e-06,
+      "loss": 0.0185,
+      "step": 4914
+    },
+    {
+      "epoch": 17.553571428571427,
+      "grad_norm": 0.6920638135176657,
+      "learning_rate": 1.549079502998989e-06,
+      "loss": 0.0136,
+      "step": 4915
+    },
+    {
+      "epoch": 17.557142857142857,
+      "grad_norm": 1.5915970512331585,
+      "learning_rate": 1.544619039753441e-06,
+      "loss": 0.0237,
+      "step": 4916
+    },
+    {
+      "epoch": 17.560714285714287,
+      "grad_norm": 1.4510418251545385,
+      "learning_rate": 1.5401647496066586e-06,
+      "loss": 0.0249,
+      "step": 4917
+    },
+    {
+      "epoch": 17.564285714285713,
+      "grad_norm": 0.8941329085790762,
+      "learning_rate": 1.5357166340485407e-06,
+      "loss": 0.0262,
+      "step": 4918
+    },
+    {
+      "epoch": 17.567857142857143,
+      "grad_norm": 0.409097721908107,
+      "learning_rate": 1.5312746945669333e-06,
+      "loss": 0.0064,
+      "step": 4919
+    },
+    {
+      "epoch": 17.571428571428573,
+      "grad_norm": 0.8437051096096372,
+      "learning_rate": 1.526838932647603e-06,
+      "loss": 0.0095,
+      "step": 4920
+    },
+    {
+      "epoch": 17.575,
+      "grad_norm": 2.7787384015288814,
+      "learning_rate": 1.5224093497742654e-06,
+      "loss": 0.0665,
+      "step": 4921
+    },
+    {
+      "epoch": 17.57857142857143,
+      "grad_norm": 0.8840934636962283,
+      "learning_rate": 1.5179859474285596e-06,
+      "loss": 0.0149,
+      "step": 4922
+    },
+    {
+      "epoch": 17.582142857142856,
+      "grad_norm": 1.175225461071797,
+      "learning_rate": 1.5135687270900533e-06,
+      "loss": 0.0121,
+      "step": 4923
+    },
+    {
+      "epoch": 17.585714285714285,
+      "grad_norm": 0.5839503576484311,
+      "learning_rate": 1.5091576902362626e-06,
+      "loss": 0.0181,
+      "step": 4924
+    },
+    {
+      "epoch": 17.589285714285715,
+      "grad_norm": 0.8613098408446219,
+      "learning_rate": 1.5047528383426157e-06,
+      "loss": 0.0157,
+      "step": 4925
+    },
+    {
+      "epoch": 17.59285714285714,
+      "grad_norm": 1.9562937373327516,
+      "learning_rate": 1.5003541728824921e-06,
+      "loss": 0.0305,
+      "step": 4926
+    },
+    {
+      "epoch": 17.59642857142857,
+      "grad_norm": 1.7191623734908124,
+      "learning_rate": 1.4959616953271794e-06,
+      "loss": 0.0511,
+      "step": 4927
+    },
+    {
+      "epoch": 17.6,
+      "grad_norm": 1.4317580833518517,
+      "learning_rate": 1.4915754071459176e-06,
+      "loss": 0.0132,
+      "step": 4928
+    },
+    {
+      "epoch": 17.603571428571428,
+      "grad_norm": 0.5919183005150265,
+      "learning_rate": 1.4871953098058601e-06,
+      "loss": 0.0108,
+      "step": 4929
+    },
+    {
+      "epoch": 17.607142857142858,
+      "grad_norm": 2.5422513934419624,
+      "learning_rate": 1.4828214047721012e-06,
+      "loss": 0.0371,
+      "step": 4930
+    },
+    {
+      "epoch": 17.610714285714288,
+      "grad_norm": 0.7455024369349319,
+      "learning_rate": 1.478453693507651e-06,
+      "loss": 0.0127,
+      "step": 4931
+    },
+    {
+      "epoch": 17.614285714285714,
+      "grad_norm": 0.5119307051516336,
+      "learning_rate": 1.474092177473463e-06,
+      "loss": 0.0113,
+      "step": 4932
+    },
+    {
+      "epoch": 17.617857142857144,
+      "grad_norm": 1.7820572401636632,
+      "learning_rate": 1.4697368581284055e-06,
+      "loss": 0.0235,
+      "step": 4933
+    },
+    {
+      "epoch": 17.62142857142857,
+      "grad_norm": 1.3012265779802863,
+      "learning_rate": 1.4653877369292847e-06,
+      "loss": 0.0227,
+      "step": 4934
+    },
+    {
+      "epoch": 17.625,
+      "grad_norm": 1.053725847670408,
+      "learning_rate": 1.4610448153308253e-06,
+      "loss": 0.0201,
+      "step": 4935
+    },
+    {
+      "epoch": 17.62857142857143,
+      "grad_norm": 0.6935991935299918,
+      "learning_rate": 1.4567080947856772e-06,
+      "loss": 0.0179,
+      "step": 4936
+    },
+    {
+      "epoch": 17.632142857142856,
+      "grad_norm": 0.4601417455494584,
+      "learning_rate": 1.452377576744428e-06,
+      "loss": 0.0074,
+      "step": 4937
+    },
+    {
+      "epoch": 17.635714285714286,
+      "grad_norm": 1.0326458632735473,
+      "learning_rate": 1.4480532626555777e-06,
+      "loss": 0.0098,
+      "step": 4938
+    },
+    {
+      "epoch": 17.639285714285712,
+      "grad_norm": 2.1541172329986105,
+      "learning_rate": 1.4437351539655597e-06,
+      "loss": 0.0204,
+      "step": 4939
+    },
+    {
+      "epoch": 17.642857142857142,
+      "grad_norm": 0.5264203838096475,
+      "learning_rate": 1.4394232521187235e-06,
+      "loss": 0.0089,
+      "step": 4940
+    },
+    {
+      "epoch": 17.646428571428572,
+      "grad_norm": 1.3309824174098146,
+      "learning_rate": 1.435117558557355e-06,
+      "loss": 0.0463,
+      "step": 4941
+    },
+    {
+      "epoch": 17.65,
+      "grad_norm": 1.4317343347869207,
+      "learning_rate": 1.4308180747216471e-06,
+      "loss": 0.0223,
+      "step": 4942
+    },
+    {
+      "epoch": 17.65357142857143,
+      "grad_norm": 0.7899284359661259,
+      "learning_rate": 1.4265248020497314e-06,
+      "loss": 0.0177,
+      "step": 4943
+    },
+    {
+      "epoch": 17.65714285714286,
+      "grad_norm": 1.03345127091419,
+      "learning_rate": 1.42223774197765e-06,
+      "loss": 0.0183,
+      "step": 4944
+    },
+    {
+      "epoch": 17.660714285714285,
+      "grad_norm": 1.7330726735161217,
+      "learning_rate": 1.4179568959393786e-06,
+      "loss": 0.0583,
+      "step": 4945
+    },
+    {
+      "epoch": 17.664285714285715,
+      "grad_norm": 1.7176707236306987,
+      "learning_rate": 1.4136822653668026e-06,
+      "loss": 0.0139,
+      "step": 4946
+    },
+    {
+      "epoch": 17.667857142857144,
+      "grad_norm": 1.0009986981991457,
+      "learning_rate": 1.4094138516897315e-06,
+      "loss": 0.0298,
+      "step": 4947
+    },
+    {
+      "epoch": 17.67142857142857,
+      "grad_norm": 0.6094533611235902,
+      "learning_rate": 1.4051516563359058e-06,
+      "loss": 0.0142,
+      "step": 4948
+    },
+    {
+      "epoch": 17.675,
+      "grad_norm": 0.21403996167256276,
+      "learning_rate": 1.400895680730967e-06,
+      "loss": 0.0038,
+      "step": 4949
+    },
+    {
+      "epoch": 17.678571428571427,
+      "grad_norm": 1.254074665040999,
+      "learning_rate": 1.3966459262984945e-06,
+      "loss": 0.0205,
+      "step": 4950
+    },
+    {
+      "epoch": 17.682142857142857,
+      "grad_norm": 0.8255757679345641,
+      "learning_rate": 1.392402394459975e-06,
+      "loss": 0.016,
+      "step": 4951
+    },
+    {
+      "epoch": 17.685714285714287,
+      "grad_norm": 1.062894648167904,
+      "learning_rate": 1.3881650866348206e-06,
+      "loss": 0.0354,
+      "step": 4952
+    },
+    {
+      "epoch": 17.689285714285713,
+      "grad_norm": 0.8885555037871087,
+      "learning_rate": 1.3839340042403571e-06,
+      "loss": 0.0177,
+      "step": 4953
+    },
+    {
+      "epoch": 17.692857142857143,
+      "grad_norm": 0.597534804458311,
+      "learning_rate": 1.3797091486918324e-06,
+      "loss": 0.0078,
+      "step": 4954
+    },
+    {
+      "epoch": 17.696428571428573,
+      "grad_norm": 1.0747750840593515,
+      "learning_rate": 1.3754905214024027e-06,
+      "loss": 0.0186,
+      "step": 4955
+    },
+    {
+      "epoch": 17.7,
+      "grad_norm": 0.8754530774672366,
+      "learning_rate": 1.371278123783155e-06,
+      "loss": 0.0142,
+      "step": 4956
+    },
+    {
+      "epoch": 17.70357142857143,
+      "grad_norm": 0.4315748855103528,
+      "learning_rate": 1.367071957243078e-06,
+      "loss": 0.0082,
+      "step": 4957
+    },
+    {
+      "epoch": 17.707142857142856,
+      "grad_norm": 2.322127855680281,
+      "learning_rate": 1.3628720231890901e-06,
+      "loss": 0.0376,
+      "step": 4958
+    },
+    {
+      "epoch": 17.710714285714285,
+      "grad_norm": 2.730430999150055,
+      "learning_rate": 1.3586783230260126e-06,
+      "loss": 0.0189,
+      "step": 4959
+    },
+    {
+      "epoch": 17.714285714285715,
+      "grad_norm": 1.507806395877517,
+      "learning_rate": 1.3544908581565853e-06,
+      "loss": 0.0156,
+      "step": 4960
+    },
+    {
+      "epoch": 17.71785714285714,
+      "grad_norm": 1.4563556840306746,
+      "learning_rate": 1.3503096299814678e-06,
+      "loss": 0.0365,
+      "step": 4961
+    },
+    {
+      "epoch": 17.72142857142857,
+      "grad_norm": 1.0391353592781056,
+      "learning_rate": 1.346134639899226e-06,
+      "loss": 0.0153,
+      "step": 4962
+    },
+    {
+      "epoch": 17.725,
+      "grad_norm": 0.8327856125257661,
+      "learning_rate": 1.3419658893063469e-06,
+      "loss": 0.0129,
+      "step": 4963
+    },
+    {
+      "epoch": 17.728571428571428,
+      "grad_norm": 1.0097192090354004,
+      "learning_rate": 1.3378033795972245e-06,
+      "loss": 0.0206,
+      "step": 4964
+    },
+    {
+      "epoch": 17.732142857142858,
+      "grad_norm": 0.8417921980103665,
+      "learning_rate": 1.3336471121641669e-06,
+      "loss": 0.0157,
+      "step": 4965
+    },
+    {
+      "epoch": 17.735714285714288,
+      "grad_norm": 0.8530027042942606,
+      "learning_rate": 1.3294970883973956e-06,
+      "loss": 0.0177,
+      "step": 4966
+    },
+    {
+      "epoch": 17.739285714285714,
+      "grad_norm": 1.7507332414947792,
+      "learning_rate": 1.3253533096850423e-06,
+      "loss": 0.0415,
+      "step": 4967
+    },
+    {
+      "epoch": 17.742857142857144,
+      "grad_norm": 0.6248503399641775,
+      "learning_rate": 1.3212157774131495e-06,
+      "loss": 0.0111,
+      "step": 4968
+    },
+    {
+      "epoch": 17.74642857142857,
+      "grad_norm": 2.6813425329400964,
+      "learning_rate": 1.3170844929656744e-06,
+      "loss": 0.0398,
+      "step": 4969
+    },
+    {
+      "epoch": 17.75,
+      "grad_norm": 2.870846918258749,
+      "learning_rate": 1.3129594577244742e-06,
+      "loss": 0.0546,
+      "step": 4970
+    },
+    {
+      "epoch": 17.75357142857143,
+      "grad_norm": 1.0284735175154358,
+      "learning_rate": 1.3088406730693293e-06,
+      "loss": 0.0273,
+      "step": 4971
+    },
+    {
+      "epoch": 17.757142857142856,
+      "grad_norm": 2.92297592322307,
+      "learning_rate": 1.3047281403779222e-06,
+      "loss": 0.0447,
+      "step": 4972
+    },
+    {
+      "epoch": 17.760714285714286,
+      "grad_norm": 0.9336738767764833,
+      "learning_rate": 1.300621861025837e-06,
+      "loss": 0.0187,
+      "step": 4973
+    },
+    {
+      "epoch": 17.764285714285712,
+      "grad_norm": 1.0395820847237742,
+      "learning_rate": 1.296521836386584e-06,
+      "loss": 0.0158,
+      "step": 4974
+    },
+    {
+      "epoch": 17.767857142857142,
+      "grad_norm": 0.9376220675809761,
+      "learning_rate": 1.2924280678315638e-06,
+      "loss": 0.0234,
+      "step": 4975
+    },
+    {
+      "epoch": 17.771428571428572,
+      "grad_norm": 0.9680082373631425,
+      "learning_rate": 1.2883405567300987e-06,
+      "loss": 0.0129,
+      "step": 4976
+    },
+    {
+      "epoch": 17.775,
+      "grad_norm": 1.9088665108394889,
+      "learning_rate": 1.2842593044494023e-06,
+      "loss": 0.0399,
+      "step": 4977
+    },
+    {
+      "epoch": 17.77857142857143,
+      "grad_norm": 0.7241345129904646,
+      "learning_rate": 1.2801843123546109e-06,
+      "loss": 0.0101,
+      "step": 4978
+    },
+    {
+      "epoch": 17.78214285714286,
+      "grad_norm": 0.902386492807282,
+      "learning_rate": 1.2761155818087545e-06,
+      "loss": 0.0114,
+      "step": 4979
+    },
+    {
+      "epoch": 17.785714285714285,
+      "grad_norm": 0.7675222860334839,
+      "learning_rate": 1.27205311417278e-06,
+      "loss": 0.0044,
+      "step": 4980
+    },
+    {
+      "epoch": 17.789285714285715,
+      "grad_norm": 1.4026634156134488,
+      "learning_rate": 1.2679969108055245e-06,
+      "loss": 0.0328,
+      "step": 4981
+    },
+    {
+      "epoch": 17.792857142857144,
+      "grad_norm": 2.158633025427657,
+      "learning_rate": 1.2639469730637455e-06,
+      "loss": 0.0178,
+      "step": 4982
+    },
+    {
+      "epoch": 17.79642857142857,
+      "grad_norm": 1.11337088260151,
+      "learning_rate": 1.2599033023020967e-06,
+      "loss": 0.0148,
+      "step": 4983
+    },
+    {
+      "epoch": 17.8,
+      "grad_norm": 0.6910269058265268,
+      "learning_rate": 1.2558658998731298e-06,
+      "loss": 0.0092,
+      "step": 4984
+    },
+    {
+      "epoch": 17.803571428571427,
+      "grad_norm": 1.4423187699431257,
+      "learning_rate": 1.2518347671273135e-06,
+      "loss": 0.0343,
+      "step": 4985
+    },
+    {
+      "epoch": 17.807142857142857,
+      "grad_norm": 1.288319802545973,
+      "learning_rate": 1.2478099054130067e-06,
+      "loss": 0.0202,
+      "step": 4986
+    },
+    {
+      "epoch": 17.810714285714287,
+      "grad_norm": 1.0785470785060693,
+      "learning_rate": 1.2437913160764836e-06,
+      "loss": 0.0205,
+      "step": 4987
+    },
+    {
+      "epoch": 17.814285714285713,
+      "grad_norm": 0.5399305276221995,
+      "learning_rate": 1.239779000461907e-06,
+      "loss": 0.0166,
+      "step": 4988
+    },
+    {
+      "epoch": 17.817857142857143,
+      "grad_norm": 0.8547076670889755,
+      "learning_rate": 1.23577295991135e-06,
+      "loss": 0.0176,
+      "step": 4989
+    },
+    {
+      "epoch": 17.821428571428573,
+      "grad_norm": 2.992101420939746,
+      "learning_rate": 1.2317731957647827e-06,
+      "loss": 0.0646,
+      "step": 4990
+    },
+    {
+      "epoch": 17.825,
+      "grad_norm": 0.9737435376444998,
+      "learning_rate": 1.2277797093600795e-06,
+      "loss": 0.0228,
+      "step": 4991
+    },
+    {
+      "epoch": 17.82857142857143,
+      "grad_norm": 0.6685119807296258,
+      "learning_rate": 1.2237925020330078e-06,
+      "loss": 0.0134,
+      "step": 4992
+    },
+    {
+      "epoch": 17.832142857142856,
+      "grad_norm": 1.0420042171784911,
+      "learning_rate": 1.2198115751172468e-06,
+      "loss": 0.0508,
+      "step": 4993
+    },
+    {
+      "epoch": 17.835714285714285,
+      "grad_norm": 1.2375424757986515,
+      "learning_rate": 1.2158369299443584e-06,
+      "loss": 0.0288,
+      "step": 4994
+    },
+    {
+      "epoch": 17.839285714285715,
+      "grad_norm": 1.8214633111719551,
+      "learning_rate": 1.2118685678438235e-06,
+      "loss": 0.0158,
+      "step": 4995
+    },
+    {
+      "epoch": 17.84285714285714,
+      "grad_norm": 0.7174848716043107,
+      "learning_rate": 1.207906490143005e-06,
+      "loss": 0.0139,
+      "step": 4996
+    },
+    {
+      "epoch": 17.84642857142857,
+      "grad_norm": 0.9518283041720248,
+      "learning_rate": 1.2039506981671667e-06,
+      "loss": 0.0144,
+      "step": 4997
+    },
+    {
+      "epoch": 17.85,
+      "grad_norm": 0.7953915477476293,
+      "learning_rate": 1.2000011932394773e-06,
+      "loss": 0.0119,
+      "step": 4998
+    },
+    {
+      "epoch": 17.853571428571428,
+      "grad_norm": 1.8508520087304101,
+      "learning_rate": 1.1960579766809932e-06,
+      "loss": 0.0237,
+      "step": 4999
+    },
+    {
+      "epoch": 17.857142857142858,
+      "grad_norm": 0.9938720172529588,
+      "learning_rate": 1.1921210498106772e-06,
+      "loss": 0.0155,
+      "step": 5000
+    },
+    {
+      "epoch": 17.860714285714288,
+      "grad_norm": 0.7622409635602423,
+      "learning_rate": 1.188190413945378e-06,
+      "loss": 0.0121,
+      "step": 5001
+    },
+    {
+      "epoch": 17.864285714285714,
+      "grad_norm": 0.8667972823529276,
+      "learning_rate": 1.184266070399851e-06,
+      "loss": 0.0196,
+      "step": 5002
+    },
+    {
+      "epoch": 17.867857142857144,
+      "grad_norm": 1.3156949675293919,
+      "learning_rate": 1.1803480204867323e-06,
+      "loss": 0.0331,
+      "step": 5003
+    },
+    {
+      "epoch": 17.87142857142857,
+      "grad_norm": 0.5415047790891804,
+      "learning_rate": 1.176436265516574e-06,
+      "loss": 0.0133,
+      "step": 5004
+    },
+    {
+      "epoch": 17.875,
+      "grad_norm": 2.23767078814403,
+      "learning_rate": 1.172530806797798e-06,
+      "loss": 0.0397,
+      "step": 5005
+    },
+    {
+      "epoch": 17.87857142857143,
+      "grad_norm": 1.1393706326143007,
+      "learning_rate": 1.1686316456367376e-06,
+      "loss": 0.0142,
+      "step": 5006
+    },
+    {
+      "epoch": 17.882142857142856,
+      "grad_norm": 1.2746158325029981,
+      "learning_rate": 1.1647387833376178e-06,
+      "loss": 0.0218,
+      "step": 5007
+    },
+    {
+      "epoch": 17.885714285714286,
+      "grad_norm": 1.107816754014842,
+      "learning_rate": 1.1608522212025463e-06,
+      "loss": 0.0354,
+      "step": 5008
+    },
+    {
+      "epoch": 17.889285714285712,
+      "grad_norm": 1.4828861452557227,
+      "learning_rate": 1.1569719605315365e-06,
+      "loss": 0.0206,
+      "step": 5009
+    },
+    {
+      "epoch": 17.892857142857142,
+      "grad_norm": 0.30869485010174597,
+      "learning_rate": 1.1530980026224836e-06,
+      "loss": 0.0058,
+      "step": 5010
+    },
+    {
+      "epoch": 17.896428571428572,
+      "grad_norm": 1.169386387113122,
+      "learning_rate": 1.1492303487711842e-06,
+      "loss": 0.0189,
+      "step": 5011
+    },
+    {
+      "epoch": 17.9,
+      "grad_norm": 1.1295667433320367,
+      "learning_rate": 1.1453690002713147e-06,
+      "loss": 0.0216,
+      "step": 5012
+    },
+    {
+      "epoch": 17.90357142857143,
+      "grad_norm": 0.8679660922993934,
+      "learning_rate": 1.1415139584144551e-06,
+      "loss": 0.0102,
+      "step": 5013
+    },
+    {
+      "epoch": 17.90714285714286,
+      "grad_norm": 2.6565409670147964,
+      "learning_rate": 1.1376652244900676e-06,
+      "loss": 0.0302,
+      "step": 5014
+    },
+    {
+      "epoch": 17.910714285714285,
+      "grad_norm": 0.8442236252748672,
+      "learning_rate": 1.1338227997855022e-06,
+      "loss": 0.0249,
+      "step": 5015
+    },
+    {
+      "epoch": 17.914285714285715,
+      "grad_norm": 1.8244643703318397,
+      "learning_rate": 1.1299866855860086e-06,
+      "loss": 0.0346,
+      "step": 5016
+    },
+    {
+      "epoch": 17.917857142857144,
+      "grad_norm": 1.6814865866320607,
+      "learning_rate": 1.126156883174716e-06,
+      "loss": 0.0281,
+      "step": 5017
+    },
+    {
+      "epoch": 17.92142857142857,
+      "grad_norm": 0.8647707500327112,
+      "learning_rate": 1.1223333938326486e-06,
+      "loss": 0.0112,
+      "step": 5018
+    },
+    {
+      "epoch": 17.925,
+      "grad_norm": 0.9096803044922727,
+      "learning_rate": 1.1185162188387167e-06,
+      "loss": 0.0247,
+      "step": 5019
+    },
+    {
+      "epoch": 17.928571428571427,
+      "grad_norm": 0.8191438486413735,
+      "learning_rate": 1.114705359469719e-06,
+      "loss": 0.0148,
+      "step": 5020
+    },
+    {
+      "epoch": 17.932142857142857,
+      "grad_norm": 0.5369024161399691,
+      "learning_rate": 1.1109008170003377e-06,
+      "loss": 0.0137,
+      "step": 5021
+    },
+    {
+      "epoch": 17.935714285714287,
+      "grad_norm": 1.191545088292136,
+      "learning_rate": 1.1071025927031509e-06,
+      "loss": 0.0206,
+      "step": 5022
+    },
+    {
+      "epoch": 17.939285714285713,
+      "grad_norm": 1.2749670951672591,
+      "learning_rate": 1.1033106878486133e-06,
+      "loss": 0.0263,
+      "step": 5023
+    },
+    {
+      "epoch": 17.942857142857143,
+      "grad_norm": 1.4871363732447622,
+      "learning_rate": 1.0995251037050748e-06,
+      "loss": 0.0413,
+      "step": 5024
+    },
+    {
+      "epoch": 17.946428571428573,
+      "grad_norm": 0.7864115428215891,
+      "learning_rate": 1.0957458415387622e-06,
+      "loss": 0.032,
+      "step": 5025
+    },
+    {
+      "epoch": 17.95,
+      "grad_norm": 1.0535358356915534,
+      "learning_rate": 1.0919729026137982e-06,
+      "loss": 0.0191,
+      "step": 5026
+    },
+    {
+      "epoch": 17.95357142857143,
+      "grad_norm": 1.1738593043235341,
+      "learning_rate": 1.0882062881921818e-06,
+      "loss": 0.0146,
+      "step": 5027
+    },
+    {
+      "epoch": 17.957142857142856,
+      "grad_norm": 0.7085022937561217,
+      "learning_rate": 1.084445999533794e-06,
+      "loss": 0.0126,
+      "step": 5028
+    },
+    {
+      "epoch": 17.960714285714285,
+      "grad_norm": 0.3883904818398044,
+      "learning_rate": 1.080692037896416e-06,
+      "loss": 0.0063,
+      "step": 5029
+    },
+    {
+      "epoch": 17.964285714285715,
+      "grad_norm": 1.1722694247293686,
+      "learning_rate": 1.0769444045356913e-06,
+      "loss": 0.014,
+      "step": 5030
+    },
+    {
+      "epoch": 17.96785714285714,
+      "grad_norm": 1.9883344656965023,
+      "learning_rate": 1.073203100705167e-06,
+      "loss": 0.0259,
+      "step": 5031
+    },
+    {
+      "epoch": 17.97142857142857,
+      "grad_norm": 1.444818775238297,
+      "learning_rate": 1.0694681276562546e-06,
+      "loss": 0.0276,
+      "step": 5032
+    },
+    {
+      "epoch": 17.975,
+      "grad_norm": 1.4831630255987822,
+      "learning_rate": 1.0657394866382642e-06,
+      "loss": 0.0287,
+      "step": 5033
+    },
+    {
+      "epoch": 17.978571428571428,
+      "grad_norm": 0.9705534087734358,
+      "learning_rate": 1.0620171788983756e-06,
+      "loss": 0.0186,
+      "step": 5034
+    },
+    {
+      "epoch": 17.982142857142858,
+      "grad_norm": 0.9732313330353358,
+      "learning_rate": 1.0583012056816577e-06,
+      "loss": 0.04,
+      "step": 5035
+    },
+    {
+      "epoch": 17.985714285714288,
+      "grad_norm": 2.4200414865873334,
+      "learning_rate": 1.0545915682310536e-06,
+      "loss": 0.0415,
+      "step": 5036
+    },
+    {
+      "epoch": 17.989285714285714,
+      "grad_norm": 0.9682908881564976,
+      "learning_rate": 1.0508882677874e-06,
+      "loss": 0.0126,
+      "step": 5037
+    },
+    {
+      "epoch": 17.992857142857144,
+      "grad_norm": 1.025293488840276,
+      "learning_rate": 1.0471913055893967e-06,
+      "loss": 0.0271,
+      "step": 5038
+    },
+    {
+      "epoch": 17.99642857142857,
+      "grad_norm": 1.1092486586099675,
+      "learning_rate": 1.0435006828736393e-06,
+      "loss": 0.0108,
+      "step": 5039
+    },
+    {
+      "epoch": 18.0,
+      "grad_norm": 0.9921277150920967,
+      "learning_rate": 1.0398164008745916e-06,
+      "loss": 0.0117,
+      "step": 5040
+    },
+    {
+      "epoch": 18.00357142857143,
+      "grad_norm": 0.4497857691145876,
+      "learning_rate": 1.0361384608246006e-06,
+      "loss": 0.0075,
+      "step": 5041
+    },
+    {
+      "epoch": 18.007142857142856,
+      "grad_norm": 1.8171999379144004,
+      "learning_rate": 1.0324668639538938e-06,
+      "loss": 0.0291,
+      "step": 5042
+    },
+    {
+      "epoch": 18.010714285714286,
+      "grad_norm": 0.7560974785568662,
+      "learning_rate": 1.0288016114905752e-06,
+      "loss": 0.0081,
+      "step": 5043
+    },
+    {
+      "epoch": 18.014285714285716,
+      "grad_norm": 0.6772817570589654,
+      "learning_rate": 1.0251427046606267e-06,
+      "loss": 0.0129,
+      "step": 5044
+    },
+    {
+      "epoch": 18.017857142857142,
+      "grad_norm": 0.636964963899172,
+      "learning_rate": 1.0214901446879067e-06,
+      "loss": 0.0069,
+      "step": 5045
+    },
+    {
+      "epoch": 18.021428571428572,
+      "grad_norm": 0.6705112447047986,
+      "learning_rate": 1.0178439327941558e-06,
+      "loss": 0.0147,
+      "step": 5046
+    },
+    {
+      "epoch": 18.025,
+      "grad_norm": 0.34447548200377726,
+      "learning_rate": 1.0142040701989785e-06,
+      "loss": 0.0048,
+      "step": 5047
+    },
+    {
+      "epoch": 18.02857142857143,
+      "grad_norm": 0.8078511950196574,
+      "learning_rate": 1.0105705581198766e-06,
+      "loss": 0.0139,
+      "step": 5048
+    },
+    {
+      "epoch": 18.03214285714286,
+      "grad_norm": 1.4464659620301843,
+      "learning_rate": 1.006943397772202e-06,
+      "loss": 0.0225,
+      "step": 5049
+    },
+    {
+      "epoch": 18.035714285714285,
+      "grad_norm": 0.5121907377287992,
+      "learning_rate": 1.0033225903692068e-06,
+      "loss": 0.0085,
+      "step": 5050
+    },
+    {
+      "epoch": 18.039285714285715,
+      "grad_norm": 0.715484376542434,
+      "learning_rate": 9.997081371220019e-07,
+      "loss": 0.0149,
+      "step": 5051
+    },
+    {
+      "epoch": 18.042857142857144,
+      "grad_norm": 0.5558257753865922,
+      "learning_rate": 9.96100039239576e-07,
+      "loss": 0.0105,
+      "step": 5052
+    },
+    {
+      "epoch": 18.04642857142857,
+      "grad_norm": 2.1582343404549262,
+      "learning_rate": 9.924982979287944e-07,
+      "loss": 0.035,
+      "step": 5053
+    },
+    {
+      "epoch": 18.05,
+      "grad_norm": 0.6531283658998401,
+      "learning_rate": 9.889029143943963e-07,
+      "loss": 0.0156,
+      "step": 5054
+    },
+    {
+      "epoch": 18.053571428571427,
+      "grad_norm": 1.3802956021395472,
+      "learning_rate": 9.85313889838997e-07,
+      "loss": 0.0284,
+      "step": 5055
+    },
+    {
+      "epoch": 18.057142857142857,
+      "grad_norm": 1.4345446387360588,
+      "learning_rate": 9.817312254630718e-07,
+      "loss": 0.023,
+      "step": 5056
+    },
+    {
+      "epoch": 18.060714285714287,
+      "grad_norm": 0.898228137598783,
+      "learning_rate": 9.781549224649889e-07,
+      "loss": 0.0095,
+      "step": 5057
+    },
+    {
+      "epoch": 18.064285714285713,
+      "grad_norm": 0.81193984663263,
+      "learning_rate": 9.745849820409692e-07,
+      "loss": 0.0269,
+      "step": 5058
+    },
+    {
+      "epoch": 18.067857142857143,
+      "grad_norm": 0.4152777774199575,
+      "learning_rate": 9.710214053851219e-07,
+      "loss": 0.0073,
+      "step": 5059
+    },
+    {
+      "epoch": 18.071428571428573,
+      "grad_norm": 1.1435084252157866,
+      "learning_rate": 9.674641936894114e-07,
+      "loss": 0.0254,
+      "step": 5060
+    },
+    {
+      "epoch": 18.075,
+      "grad_norm": 0.8919505585820409,
+      "learning_rate": 9.63913348143688e-07,
+      "loss": 0.02,
+      "step": 5061
+    },
+    {
+      "epoch": 18.07857142857143,
+      "grad_norm": 1.4255225524969215,
+      "learning_rate": 9.603688699356616e-07,
+      "loss": 0.0271,
+      "step": 5062
+    },
+    {
+      "epoch": 18.082142857142856,
+      "grad_norm": 0.8108425820357359,
+      "learning_rate": 9.568307602509196e-07,
+      "loss": 0.0259,
+      "step": 5063
+    },
+    {
+      "epoch": 18.085714285714285,
+      "grad_norm": 0.866327802699689,
+      "learning_rate": 9.532990202729175e-07,
+      "loss": 0.0154,
+      "step": 5064
+    },
+    {
+      "epoch": 18.089285714285715,
+      "grad_norm": 0.6709262976464745,
+      "learning_rate": 9.497736511829725e-07,
+      "loss": 0.0141,
+      "step": 5065
+    },
+    {
+      "epoch": 18.09285714285714,
+      "grad_norm": 0.5084230053038211,
+      "learning_rate": 9.462546541602857e-07,
+      "loss": 0.009,
+      "step": 5066
+    },
+    {
+      "epoch": 18.09642857142857,
+      "grad_norm": 1.7317023893266008,
+      "learning_rate": 9.427420303819113e-07,
+      "loss": 0.0248,
+      "step": 5067
+    },
+    {
+      "epoch": 18.1,
+      "grad_norm": 0.33740609157608903,
+      "learning_rate": 9.392357810227826e-07,
+      "loss": 0.006,
+      "step": 5068
+    },
+    {
+      "epoch": 18.103571428571428,
+      "grad_norm": 1.3514731467553105,
+      "learning_rate": 9.35735907255697e-07,
+      "loss": 0.0295,
+      "step": 5069
+    },
+    {
+      "epoch": 18.107142857142858,
+      "grad_norm": 1.3750281443225003,
+      "learning_rate": 9.322424102513205e-07,
+      "loss": 0.0107,
+      "step": 5070
+    },
+    {
+      "epoch": 18.110714285714284,
+      "grad_norm": 1.8761337895511851,
+      "learning_rate": 9.287552911781805e-07,
+      "loss": 0.0417,
+      "step": 5071
+    },
+    {
+      "epoch": 18.114285714285714,
+      "grad_norm": 0.6216219668817541,
+      "learning_rate": 9.252745512026817e-07,
+      "loss": 0.0139,
+      "step": 5072
+    },
+    {
+      "epoch": 18.117857142857144,
+      "grad_norm": 0.6090778591991869,
+      "learning_rate": 9.21800191489084e-07,
+      "loss": 0.017,
+      "step": 5073
+    },
+    {
+      "epoch": 18.12142857142857,
+      "grad_norm": 0.43654398248422593,
+      "learning_rate": 9.183322131995198e-07,
+      "loss": 0.0079,
+      "step": 5074
+    },
+    {
+      "epoch": 18.125,
+      "grad_norm": 0.34349635916558147,
+      "learning_rate": 9.148706174939859e-07,
+      "loss": 0.0068,
+      "step": 5075
+    },
+    {
+      "epoch": 18.12857142857143,
+      "grad_norm": 0.5471406051371471,
+      "learning_rate": 9.114154055303449e-07,
+      "loss": 0.0129,
+      "step": 5076
+    },
+    {
+      "epoch": 18.132142857142856,
+      "grad_norm": 0.7857574861380596,
+      "learning_rate": 9.07966578464321e-07,
+      "loss": 0.0138,
+      "step": 5077
+    },
+    {
+      "epoch": 18.135714285714286,
+      "grad_norm": 0.9173920919952739,
+      "learning_rate": 9.045241374495029e-07,
+      "loss": 0.0079,
+      "step": 5078
+    },
+    {
+      "epoch": 18.139285714285716,
+      "grad_norm": 1.4126516611052453,
+      "learning_rate": 9.01088083637347e-07,
+      "loss": 0.0295,
+      "step": 5079
+    },
+    {
+      "epoch": 18.142857142857142,
+      "grad_norm": 0.6006159306079014,
+      "learning_rate": 8.976584181771719e-07,
+      "loss": 0.0193,
+      "step": 5080
+    },
+    {
+      "epoch": 18.146428571428572,
+      "grad_norm": 0.7972892962258762,
+      "learning_rate": 8.942351422161577e-07,
+      "loss": 0.0156,
+      "step": 5081
+    },
+    {
+      "epoch": 18.15,
+      "grad_norm": 1.5315739652315477,
+      "learning_rate": 8.908182568993462e-07,
+      "loss": 0.0216,
+      "step": 5082
+    },
+    {
+      "epoch": 18.15357142857143,
+      "grad_norm": 0.8289073156584561,
+      "learning_rate": 8.874077633696498e-07,
+      "loss": 0.0169,
+      "step": 5083
+    },
+    {
+      "epoch": 18.15714285714286,
+      "grad_norm": 0.6713963131219065,
+      "learning_rate": 8.840036627678294e-07,
+      "loss": 0.0154,
+      "step": 5084
+    },
+    {
+      "epoch": 18.160714285714285,
+      "grad_norm": 1.891751342396216,
+      "learning_rate": 8.806059562325209e-07,
+      "loss": 0.0319,
+      "step": 5085
+    },
+    {
+      "epoch": 18.164285714285715,
+      "grad_norm": 0.7530623487172079,
+      "learning_rate": 8.772146449002105e-07,
+      "loss": 0.0057,
+      "step": 5086
+    },
+    {
+      "epoch": 18.167857142857144,
+      "grad_norm": 1.1242941546741874,
+      "learning_rate": 8.738297299052556e-07,
+      "loss": 0.0254,
+      "step": 5087
+    },
+    {
+      "epoch": 18.17142857142857,
+      "grad_norm": 0.5958128449457942,
+      "learning_rate": 8.704512123798658e-07,
+      "loss": 0.0102,
+      "step": 5088
+    },
+    {
+      "epoch": 18.175,
+      "grad_norm": 0.6329884039039108,
+      "learning_rate": 8.670790934541107e-07,
+      "loss": 0.0132,
+      "step": 5089
+    },
+    {
+      "epoch": 18.178571428571427,
+      "grad_norm": 1.4204116895371675,
+      "learning_rate": 8.637133742559301e-07,
+      "loss": 0.0441,
+      "step": 5090
+    },
+    {
+      "epoch": 18.182142857142857,
+      "grad_norm": 1.2654307363131616,
+      "learning_rate": 8.603540559111079e-07,
+      "loss": 0.024,
+      "step": 5091
+    },
+    {
+      "epoch": 18.185714285714287,
+      "grad_norm": 1.682208117072008,
+      "learning_rate": 8.570011395433009e-07,
+      "loss": 0.0278,
+      "step": 5092
+    },
+    {
+      "epoch": 18.189285714285713,
+      "grad_norm": 1.250711501830989,
+      "learning_rate": 8.536546262740142e-07,
+      "loss": 0.0234,
+      "step": 5093
+    },
+    {
+      "epoch": 18.192857142857143,
+      "grad_norm": 0.7793599144456264,
+      "learning_rate": 8.503145172226191e-07,
+      "loss": 0.011,
+      "step": 5094
+    },
+    {
+      "epoch": 18.196428571428573,
+      "grad_norm": 1.227021719360323,
+      "learning_rate": 8.469808135063395e-07,
+      "loss": 0.0292,
+      "step": 5095
+    },
+    {
+      "epoch": 18.2,
+      "grad_norm": 0.7350181361775819,
+      "learning_rate": 8.436535162402592e-07,
+      "loss": 0.0123,
+      "step": 5096
+    },
+    {
+      "epoch": 18.20357142857143,
+      "grad_norm": 0.6048906219913142,
+      "learning_rate": 8.403326265373168e-07,
+      "loss": 0.0107,
+      "step": 5097
+    },
+    {
+      "epoch": 18.207142857142856,
+      "grad_norm": 1.2880807607425986,
+      "learning_rate": 8.370181455083104e-07,
+      "loss": 0.0126,
+      "step": 5098
+    },
+    {
+      "epoch": 18.210714285714285,
+      "grad_norm": 1.1410556953339273,
+      "learning_rate": 8.337100742618931e-07,
+      "loss": 0.0187,
+      "step": 5099
+    },
+    {
+      "epoch": 18.214285714285715,
+      "grad_norm": 0.704544591136182,
+      "learning_rate": 8.304084139045753e-07,
+      "loss": 0.0105,
+      "step": 5100
+    },
+    {
+      "epoch": 18.21785714285714,
+      "grad_norm": 0.7801645825752533,
+      "learning_rate": 8.271131655407205e-07,
+      "loss": 0.0234,
+      "step": 5101
+    },
+    {
+      "epoch": 18.22142857142857,
+      "grad_norm": 1.6170446043645388,
+      "learning_rate": 8.238243302725468e-07,
+      "loss": 0.0343,
+      "step": 5102
+    },
+    {
+      "epoch": 18.225,
+      "grad_norm": 0.5462263775317919,
+      "learning_rate": 8.205419092001343e-07,
+      "loss": 0.0059,
+      "step": 5103
+    },
+    {
+      "epoch": 18.228571428571428,
+      "grad_norm": 1.0128330629583708,
+      "learning_rate": 8.172659034214092e-07,
+      "loss": 0.0077,
+      "step": 5104
+    },
+    {
+      "epoch": 18.232142857142858,
+      "grad_norm": 2.644190739317922,
+      "learning_rate": 8.139963140321572e-07,
+      "loss": 0.0374,
+      "step": 5105
+    },
+    {
+      "epoch": 18.235714285714284,
+      "grad_norm": 1.4005924934602951,
+      "learning_rate": 8.107331421260123e-07,
+      "loss": 0.0203,
+      "step": 5106
+    },
+    {
+      "epoch": 18.239285714285714,
+      "grad_norm": 0.785518127848494,
+      "learning_rate": 8.074763887944703e-07,
+      "loss": 0.0114,
+      "step": 5107
+    },
+    {
+      "epoch": 18.242857142857144,
+      "grad_norm": 0.43422765241644573,
+      "learning_rate": 8.042260551268688e-07,
+      "loss": 0.0053,
+      "step": 5108
+    },
+    {
+      "epoch": 18.24642857142857,
+      "grad_norm": 1.8902397696770257,
+      "learning_rate": 8.009821422104136e-07,
+      "loss": 0.0313,
+      "step": 5109
+    },
+    {
+      "epoch": 18.25,
+      "grad_norm": 0.6852498329667261,
+      "learning_rate": 7.977446511301456e-07,
+      "loss": 0.0065,
+      "step": 5110
+    },
+    {
+      "epoch": 18.25357142857143,
+      "grad_norm": 1.2310251043266895,
+      "learning_rate": 7.94513582968972e-07,
+      "loss": 0.0204,
+      "step": 5111
+    },
+    {
+      "epoch": 18.257142857142856,
+      "grad_norm": 0.8376509753323375,
+      "learning_rate": 7.912889388076417e-07,
+      "loss": 0.0125,
+      "step": 5112
+    },
+    {
+      "epoch": 18.260714285714286,
+      "grad_norm": 0.9133760905408493,
+      "learning_rate": 7.880707197247606e-07,
+      "loss": 0.0162,
+      "step": 5113
+    },
+    {
+      "epoch": 18.264285714285716,
+      "grad_norm": 0.8748728354257689,
+      "learning_rate": 7.84858926796781e-07,
+      "loss": 0.0224,
+      "step": 5114
+    },
+    {
+      "epoch": 18.267857142857142,
+      "grad_norm": 0.9859734234994728,
+      "learning_rate": 7.816535610980103e-07,
+      "loss": 0.0264,
+      "step": 5115
+    },
+    {
+      "epoch": 18.271428571428572,
+      "grad_norm": 0.7142396890581321,
+      "learning_rate": 7.784546237006041e-07,
+      "loss": 0.0129,
+      "step": 5116
+    },
+    {
+      "epoch": 18.275,
+      "grad_norm": 1.060307138659144,
+      "learning_rate": 7.752621156745666e-07,
+      "loss": 0.0232,
+      "step": 5117
+    },
+    {
+      "epoch": 18.27857142857143,
+      "grad_norm": 1.2407432896535888,
+      "learning_rate": 7.720760380877545e-07,
+      "loss": 0.0174,
+      "step": 5118
+    },
+    {
+      "epoch": 18.28214285714286,
+      "grad_norm": 0.5200833363185612,
+      "learning_rate": 7.68896392005869e-07,
+      "loss": 0.0105,
+      "step": 5119
+    },
+    {
+      "epoch": 18.285714285714285,
+      "grad_norm": 1.3359068494412165,
+      "learning_rate": 7.657231784924657e-07,
+      "loss": 0.0348,
+      "step": 5120
+    },
+    {
+      "epoch": 18.289285714285715,
+      "grad_norm": 1.3352842299624734,
+      "learning_rate": 7.625563986089424e-07,
+      "loss": 0.0142,
+      "step": 5121
+    },
+    {
+      "epoch": 18.292857142857144,
+      "grad_norm": 1.5260176693954088,
+      "learning_rate": 7.593960534145516e-07,
+      "loss": 0.0344,
+      "step": 5122
+    },
+    {
+      "epoch": 18.29642857142857,
+      "grad_norm": 1.5358292836944236,
+      "learning_rate": 7.562421439663881e-07,
+      "loss": 0.0364,
+      "step": 5123
+    },
+    {
+      "epoch": 18.3,
+      "grad_norm": 1.0025477939955092,
+      "learning_rate": 7.530946713193965e-07,
+      "loss": 0.0097,
+      "step": 5124
+    },
+    {
+      "epoch": 18.303571428571427,
+      "grad_norm": 1.2267934541325711,
+      "learning_rate": 7.499536365263682e-07,
+      "loss": 0.0234,
+      "step": 5125
+    },
+    {
+      "epoch": 18.307142857142857,
+      "grad_norm": 1.842562649427403,
+      "learning_rate": 7.468190406379383e-07,
+      "loss": 0.0197,
+      "step": 5126
+    },
+    {
+      "epoch": 18.310714285714287,
+      "grad_norm": 0.8805196999512824,
+      "learning_rate": 7.436908847025948e-07,
+      "loss": 0.0142,
+      "step": 5127
+    },
+    {
+      "epoch": 18.314285714285713,
+      "grad_norm": 1.3730070533478749,
+      "learning_rate": 7.405691697666628e-07,
+      "loss": 0.0338,
+      "step": 5128
+    },
+    {
+      "epoch": 18.317857142857143,
+      "grad_norm": 1.34513776804142,
+      "learning_rate": 7.374538968743227e-07,
+      "loss": 0.0313,
+      "step": 5129
+    },
+    {
+      "epoch": 18.321428571428573,
+      "grad_norm": 0.7022198462849433,
+      "learning_rate": 7.343450670675901e-07,
+      "loss": 0.015,
+      "step": 5130
+    },
+    {
+      "epoch": 18.325,
+      "grad_norm": 0.8895562348704814,
+      "learning_rate": 7.312426813863371e-07,
+      "loss": 0.0158,
+      "step": 5131
+    },
+    {
+      "epoch": 18.32857142857143,
+      "grad_norm": 8.30826842798115,
+      "learning_rate": 7.281467408682652e-07,
+      "loss": 0.0273,
+      "step": 5132
+    },
+    {
+      "epoch": 18.332142857142856,
+      "grad_norm": 1.5972842670084206,
+      "learning_rate": 7.250572465489369e-07,
+      "loss": 0.0154,
+      "step": 5133
+    },
+    {
+      "epoch": 18.335714285714285,
+      "grad_norm": 0.4311764212447403,
+      "learning_rate": 7.219741994617413e-07,
+      "loss": 0.0086,
+      "step": 5134
+    },
+    {
+      "epoch": 18.339285714285715,
+      "grad_norm": 2.009480118224382,
+      "learning_rate": 7.188976006379311e-07,
+      "loss": 0.0311,
+      "step": 5135
+    },
+    {
+      "epoch": 18.34285714285714,
+      "grad_norm": 2.1276036818529853,
+      "learning_rate": 7.158274511065788e-07,
+      "loss": 0.0309,
+      "step": 5136
+    },
+    {
+      "epoch": 18.34642857142857,
+      "grad_norm": 0.9395495084492859,
+      "learning_rate": 7.127637518946228e-07,
+      "loss": 0.0241,
+      "step": 5137
+    },
+    {
+      "epoch": 18.35,
+      "grad_norm": 2.160821549551104,
+      "learning_rate": 7.097065040268258e-07,
+      "loss": 0.0254,
+      "step": 5138
+    },
+    {
+      "epoch": 18.353571428571428,
+      "grad_norm": 0.8787744278862812,
+      "learning_rate": 7.066557085258008e-07,
+      "loss": 0.0131,
+      "step": 5139
+    },
+    {
+      "epoch": 18.357142857142858,
+      "grad_norm": 0.47416118664454165,
+      "learning_rate": 7.036113664120026e-07,
+      "loss": 0.0097,
+      "step": 5140
+    },
+    {
+      "epoch": 18.360714285714284,
+      "grad_norm": 0.9221603356673101,
+      "learning_rate": 7.005734787037255e-07,
+      "loss": 0.0228,
+      "step": 5141
+    },
+    {
+      "epoch": 18.364285714285714,
+      "grad_norm": 1.1298368121555284,
+      "learning_rate": 6.975420464171079e-07,
+      "loss": 0.0158,
+      "step": 5142
+    },
+    {
+      "epoch": 18.367857142857144,
+      "grad_norm": 1.1220187550096563,
+      "learning_rate": 6.945170705661252e-07,
+      "loss": 0.018,
+      "step": 5143
+    },
+    {
+      "epoch": 18.37142857142857,
+      "grad_norm": 0.9592553104892357,
+      "learning_rate": 6.91498552162595e-07,
+      "loss": 0.0117,
+      "step": 5144
+    },
+    {
+      "epoch": 18.375,
+      "grad_norm": 1.0120246255792957,
+      "learning_rate": 6.884864922161716e-07,
+      "loss": 0.0159,
+      "step": 5145
+    },
+    {
+      "epoch": 18.37857142857143,
+      "grad_norm": 1.1745845429073094,
+      "learning_rate": 6.854808917343581e-07,
+      "loss": 0.018,
+      "step": 5146
+    },
+    {
+      "epoch": 18.382142857142856,
+      "grad_norm": 0.9304883530605262,
+      "learning_rate": 6.824817517224858e-07,
+      "loss": 0.0149,
+      "step": 5147
+    },
+    {
+      "epoch": 18.385714285714286,
+      "grad_norm": 0.8561038682302726,
+      "learning_rate": 6.794890731837367e-07,
+      "loss": 0.0136,
+      "step": 5148
+    },
+    {
+      "epoch": 18.389285714285716,
+      "grad_norm": 1.4215773428752945,
+      "learning_rate": 6.765028571191168e-07,
+      "loss": 0.0435,
+      "step": 5149
+    },
+    {
+      "epoch": 18.392857142857142,
+      "grad_norm": 0.8766723421725889,
+      "learning_rate": 6.735231045274848e-07,
+      "loss": 0.0214,
+      "step": 5150
+    },
+    {
+      "epoch": 18.396428571428572,
+      "grad_norm": 1.8935071943906912,
+      "learning_rate": 6.705498164055302e-07,
+      "loss": 0.0195,
+      "step": 5151
+    },
+    {
+      "epoch": 18.4,
+      "grad_norm": 1.2172919708134657,
+      "learning_rate": 6.675829937477773e-07,
+      "loss": 0.0212,
+      "step": 5152
+    },
+    {
+      "epoch": 18.40357142857143,
+      "grad_norm": 0.5517319835676942,
+      "learning_rate": 6.646226375465969e-07,
+      "loss": 0.0138,
+      "step": 5153
+    },
+    {
+      "epoch": 18.40714285714286,
+      "grad_norm": 0.7119869948687717,
+      "learning_rate": 6.616687487921902e-07,
+      "loss": 0.0067,
+      "step": 5154
+    },
+    {
+      "epoch": 18.410714285714285,
+      "grad_norm": 1.1703759868353285,
+      "learning_rate": 6.587213284725979e-07,
+      "loss": 0.0193,
+      "step": 5155
+    },
+    {
+      "epoch": 18.414285714285715,
+      "grad_norm": 0.7934913272781643,
+      "learning_rate": 6.557803775736937e-07,
+      "loss": 0.0201,
+      "step": 5156
+    },
+    {
+      "epoch": 18.417857142857144,
+      "grad_norm": 0.727829047230472,
+      "learning_rate": 6.528458970791929e-07,
+      "loss": 0.0122,
+      "step": 5157
+    },
+    {
+      "epoch": 18.42142857142857,
+      "grad_norm": 0.23890335353069053,
+      "learning_rate": 6.49917887970637e-07,
+      "loss": 0.0047,
+      "step": 5158
+    },
+    {
+      "epoch": 18.425,
+      "grad_norm": 1.1974978405536594,
+      "learning_rate": 6.46996351227418e-07,
+      "loss": 0.0237,
+      "step": 5159
+    },
+    {
+      "epoch": 18.428571428571427,
+      "grad_norm": 0.6970319713732243,
+      "learning_rate": 6.440812878267455e-07,
+      "loss": 0.0168,
+      "step": 5160
+    },
+    {
+      "epoch": 18.432142857142857,
+      "grad_norm": 1.1937582564425344,
+      "learning_rate": 6.411726987436818e-07,
+      "loss": 0.0148,
+      "step": 5161
+    },
+    {
+      "epoch": 18.435714285714287,
+      "grad_norm": 0.6763500011033323,
+      "learning_rate": 6.382705849511062e-07,
+      "loss": 0.0123,
+      "step": 5162
+    },
+    {
+      "epoch": 18.439285714285713,
+      "grad_norm": 0.7878399009758615,
+      "learning_rate": 6.353749474197424e-07,
+      "loss": 0.0098,
+      "step": 5163
+    },
+    {
+      "epoch": 18.442857142857143,
+      "grad_norm": 1.9117566993574748,
+      "learning_rate": 6.324857871181467e-07,
+      "loss": 0.0255,
+      "step": 5164
+    },
+    {
+      "epoch": 18.446428571428573,
+      "grad_norm": 0.615227178122797,
+      "learning_rate": 6.29603105012706e-07,
+      "loss": 0.012,
+      "step": 5165
+    },
+    {
+      "epoch": 18.45,
+      "grad_norm": 2.019066065295221,
+      "learning_rate": 6.26726902067647e-07,
+      "loss": 0.019,
+      "step": 5166
+    },
+    {
+      "epoch": 18.45357142857143,
+      "grad_norm": 1.1280862513786474,
+      "learning_rate": 6.238571792450177e-07,
+      "loss": 0.0265,
+      "step": 5167
+    },
+    {
+      "epoch": 18.457142857142856,
+      "grad_norm": 2.07622375495234,
+      "learning_rate": 6.209939375047103e-07,
+      "loss": 0.0305,
+      "step": 5168
+    },
+    {
+      "epoch": 18.460714285714285,
+      "grad_norm": 1.1101528537632523,
+      "learning_rate": 6.181371778044388e-07,
+      "loss": 0.0227,
+      "step": 5169
+    },
+    {
+      "epoch": 18.464285714285715,
+      "grad_norm": 1.0050462718657458,
+      "learning_rate": 6.152869010997631e-07,
+      "loss": 0.014,
+      "step": 5170
+    },
+    {
+      "epoch": 18.46785714285714,
+      "grad_norm": 0.851435914454913,
+      "learning_rate": 6.124431083440563e-07,
+      "loss": 0.018,
+      "step": 5171
+    },
+    {
+      "epoch": 18.47142857142857,
+      "grad_norm": 1.3587308856682112,
+      "learning_rate": 6.096058004885375e-07,
+      "loss": 0.0229,
+      "step": 5172
+    },
+    {
+      "epoch": 18.475,
+      "grad_norm": 0.8400413136650526,
+      "learning_rate": 6.067749784822497e-07,
+      "loss": 0.0175,
+      "step": 5173
+    },
+    {
+      "epoch": 18.478571428571428,
+      "grad_norm": 0.8229615860599694,
+      "learning_rate": 6.03950643272071e-07,
+      "loss": 0.0139,
+      "step": 5174
+    },
+    {
+      "epoch": 18.482142857142858,
+      "grad_norm": 0.7735714179358696,
+      "learning_rate": 6.011327958027036e-07,
+      "loss": 0.0159,
+      "step": 5175
+    },
+    {
+      "epoch": 18.485714285714284,
+      "grad_norm": 0.6991588277820403,
+      "learning_rate": 5.983214370166801e-07,
+      "loss": 0.0193,
+      "step": 5176
+    },
+    {
+      "epoch": 18.489285714285714,
+      "grad_norm": 0.8555849703462091,
+      "learning_rate": 5.955165678543728e-07,
+      "loss": 0.021,
+      "step": 5177
+    },
+    {
+      "epoch": 18.492857142857144,
+      "grad_norm": 1.8830379253844964,
+      "learning_rate": 5.927181892539735e-07,
+      "loss": 0.0342,
+      "step": 5178
+    },
+    {
+      "epoch": 18.49642857142857,
+      "grad_norm": 1.6157668852091471,
+      "learning_rate": 5.899263021515045e-07,
+      "loss": 0.0242,
+      "step": 5179
+    },
+    {
+      "epoch": 18.5,
+      "grad_norm": 0.5000096841475942,
+      "learning_rate": 5.871409074808166e-07,
+      "loss": 0.0067,
+      "step": 5180
+    },
+    {
+      "epoch": 18.50357142857143,
+      "grad_norm": 0.70669898484788,
+      "learning_rate": 5.843620061735933e-07,
+      "loss": 0.0082,
+      "step": 5181
+    },
+    {
+      "epoch": 18.507142857142856,
+      "grad_norm": 1.6266434218448267,
+      "learning_rate": 5.815895991593401e-07,
+      "loss": 0.0214,
+      "step": 5182
+    },
+    {
+      "epoch": 18.510714285714286,
+      "grad_norm": 1.9778827732348236,
+      "learning_rate": 5.788236873653952e-07,
+      "loss": 0.0381,
+      "step": 5183
+    },
+    {
+      "epoch": 18.514285714285712,
+      "grad_norm": 0.6868995696633106,
+      "learning_rate": 5.760642717169207e-07,
+      "loss": 0.0127,
+      "step": 5184
+    },
+    {
+      "epoch": 18.517857142857142,
+      "grad_norm": 0.7467759959776289,
+      "learning_rate": 5.733113531369094e-07,
+      "loss": 0.0079,
+      "step": 5185
+    },
+    {
+      "epoch": 18.521428571428572,
+      "grad_norm": 1.4296095114650866,
+      "learning_rate": 5.70564932546176e-07,
+      "loss": 0.0215,
+      "step": 5186
+    },
+    {
+      "epoch": 18.525,
+      "grad_norm": 2.2741889327273848,
+      "learning_rate": 5.678250108633676e-07,
+      "loss": 0.0293,
+      "step": 5187
+    },
+    {
+      "epoch": 18.52857142857143,
+      "grad_norm": 1.6028059401065182,
+      "learning_rate": 5.650915890049513e-07,
+      "loss": 0.0154,
+      "step": 5188
+    },
+    {
+      "epoch": 18.53214285714286,
+      "grad_norm": 2.2681445343391005,
+      "learning_rate": 5.623646678852246e-07,
+      "loss": 0.0425,
+      "step": 5189
+    },
+    {
+      "epoch": 18.535714285714285,
+      "grad_norm": 0.6119132527730761,
+      "learning_rate": 5.596442484163112e-07,
+      "loss": 0.0078,
+      "step": 5190
+    },
+    {
+      "epoch": 18.539285714285715,
+      "grad_norm": 1.6299698183882938,
+      "learning_rate": 5.569303315081543e-07,
+      "loss": 0.0343,
+      "step": 5191
+    },
+    {
+      "epoch": 18.542857142857144,
+      "grad_norm": 0.9852077307795143,
+      "learning_rate": 5.542229180685299e-07,
+      "loss": 0.0112,
+      "step": 5192
+    },
+    {
+      "epoch": 18.54642857142857,
+      "grad_norm": 1.0752319027411408,
+      "learning_rate": 5.515220090030316e-07,
+      "loss": 0.0222,
+      "step": 5193
+    },
+    {
+      "epoch": 18.55,
+      "grad_norm": 0.896229259259583,
+      "learning_rate": 5.48827605215081e-07,
+      "loss": 0.012,
+      "step": 5194
+    },
+    {
+      "epoch": 18.553571428571427,
+      "grad_norm": 0.9327956886692054,
+      "learning_rate": 5.461397076059238e-07,
+      "loss": 0.0217,
+      "step": 5195
+    },
+    {
+      "epoch": 18.557142857142857,
+      "grad_norm": 1.4232234658614042,
+      "learning_rate": 5.434583170746277e-07,
+      "loss": 0.0685,
+      "step": 5196
+    },
+    {
+      "epoch": 18.560714285714287,
+      "grad_norm": 1.2242721362484201,
+      "learning_rate": 5.407834345180862e-07,
+      "loss": 0.031,
+      "step": 5197
+    },
+    {
+      "epoch": 18.564285714285713,
+      "grad_norm": 1.2805702282288471,
+      "learning_rate": 5.381150608310148e-07,
+      "loss": 0.0221,
+      "step": 5198
+    },
+    {
+      "epoch": 18.567857142857143,
+      "grad_norm": 0.3815960470857988,
+      "learning_rate": 5.354531969059507e-07,
+      "loss": 0.0052,
+      "step": 5199
+    },
+    {
+      "epoch": 18.571428571428573,
+      "grad_norm": 0.8575161004721429,
+      "learning_rate": 5.327978436332549e-07,
+      "loss": 0.0107,
+      "step": 5200
+    },
+    {
+      "epoch": 18.575,
+      "grad_norm": 0.7352621441199261,
+      "learning_rate": 5.301490019011124e-07,
+      "loss": 0.0176,
+      "step": 5201
+    },
+    {
+      "epoch": 18.57857142857143,
+      "grad_norm": 1.0449589272641209,
+      "learning_rate": 5.275066725955235e-07,
+      "loss": 0.0297,
+      "step": 5202
+    },
+    {
+      "epoch": 18.582142857142856,
+      "grad_norm": 1.0576790384715087,
+      "learning_rate": 5.248708566003214e-07,
+      "loss": 0.0251,
+      "step": 5203
+    },
+    {
+      "epoch": 18.585714285714285,
+      "grad_norm": 1.4888060802366085,
+      "learning_rate": 5.222415547971493e-07,
+      "loss": 0.0204,
+      "step": 5204
+    },
+    {
+      "epoch": 18.589285714285715,
+      "grad_norm": 1.5378295477357948,
+      "learning_rate": 5.196187680654796e-07,
+      "loss": 0.0279,
+      "step": 5205
+    },
+    {
+      "epoch": 18.59285714285714,
+      "grad_norm": 0.6317622458550363,
+      "learning_rate": 5.170024972826016e-07,
+      "loss": 0.0135,
+      "step": 5206
+    },
+    {
+      "epoch": 18.59642857142857,
+      "grad_norm": 0.6214112186068386,
+      "learning_rate": 5.14392743323624e-07,
+      "loss": 0.0165,
+      "step": 5207
+    },
+    {
+      "epoch": 18.6,
+      "grad_norm": 0.30648055300586047,
+      "learning_rate": 5.117895070614797e-07,
+      "loss": 0.0054,
+      "step": 5208
+    },
+    {
+      "epoch": 18.603571428571428,
+      "grad_norm": 1.6754258361843717,
+      "learning_rate": 5.091927893669168e-07,
+      "loss": 0.0275,
+      "step": 5209
+    },
+    {
+      "epoch": 18.607142857142858,
+      "grad_norm": 0.8330989326065937,
+      "learning_rate": 5.066025911085093e-07,
+      "loss": 0.0126,
+      "step": 5210
+    },
+    {
+      "epoch": 18.610714285714288,
+      "grad_norm": 1.6350608183733948,
+      "learning_rate": 5.040189131526419e-07,
+      "loss": 0.0413,
+      "step": 5211
+    },
+    {
+      "epoch": 18.614285714285714,
+      "grad_norm": 0.3840392700948037,
+      "learning_rate": 5.014417563635276e-07,
+      "loss": 0.0063,
+      "step": 5212
+    },
+    {
+      "epoch": 18.617857142857144,
+      "grad_norm": 0.8232789757536394,
+      "learning_rate": 4.988711216031905e-07,
+      "loss": 0.0157,
+      "step": 5213
+    },
+    {
+      "epoch": 18.62142857142857,
+      "grad_norm": 0.8597769964050865,
+      "learning_rate": 4.963070097314804e-07,
+      "loss": 0.0217,
+      "step": 5214
+    },
+    {
+      "epoch": 18.625,
+      "grad_norm": 0.7346168712078245,
+      "learning_rate": 4.937494216060578e-07,
+      "loss": 0.0113,
+      "step": 5215
+    },
+    {
+      "epoch": 18.62857142857143,
+      "grad_norm": 0.4415474616565294,
+      "learning_rate": 4.911983580824053e-07,
+      "loss": 0.0055,
+      "step": 5216
+    },
+    {
+      "epoch": 18.632142857142856,
+      "grad_norm": 1.4127827283127272,
+      "learning_rate": 4.886538200138225e-07,
+      "loss": 0.0424,
+      "step": 5217
+    },
+    {
+      "epoch": 18.635714285714286,
+      "grad_norm": 1.5974690606297481,
+      "learning_rate": 4.861158082514283e-07,
+      "loss": 0.0122,
+      "step": 5218
+    },
+    {
+      "epoch": 18.639285714285712,
+      "grad_norm": 2.0214810034474993,
+      "learning_rate": 4.835843236441551e-07,
+      "loss": 0.0192,
+      "step": 5219
+    },
+    {
+      "epoch": 18.642857142857142,
+      "grad_norm": 0.4925555684126724,
+      "learning_rate": 4.810593670387498e-07,
+      "loss": 0.0121,
+      "step": 5220
+    },
+    {
+      "epoch": 18.646428571428572,
+      "grad_norm": 0.7516451119645438,
+      "learning_rate": 4.785409392797857e-07,
+      "loss": 0.0156,
+      "step": 5221
+    },
+    {
+      "epoch": 18.65,
+      "grad_norm": 1.0194385222228526,
+      "learning_rate": 4.760290412096402e-07,
+      "loss": 0.0105,
+      "step": 5222
+    },
+    {
+      "epoch": 18.65357142857143,
+      "grad_norm": 1.6292602223033599,
+      "learning_rate": 4.735236736685167e-07,
+      "loss": 0.013,
+      "step": 5223
+    },
+    {
+      "epoch": 18.65714285714286,
+      "grad_norm": 1.0246801737551503,
+      "learning_rate": 4.7102483749442707e-07,
+      "loss": 0.0179,
+      "step": 5224
+    },
+    {
+      "epoch": 18.660714285714285,
+      "grad_norm": 0.3237824832215185,
+      "learning_rate": 4.6853253352320717e-07,
+      "loss": 0.0051,
+      "step": 5225
+    },
+    {
+      "epoch": 18.664285714285715,
+      "grad_norm": 1.785765573440142,
+      "learning_rate": 4.6604676258849237e-07,
+      "loss": 0.0296,
+      "step": 5226
+    },
+    {
+      "epoch": 18.667857142857144,
+      "grad_norm": 0.7523511938343806,
+      "learning_rate": 4.635675255217509e-07,
+      "loss": 0.0105,
+      "step": 5227
+    },
+    {
+      "epoch": 18.67142857142857,
+      "grad_norm": 0.38606291569981727,
+      "learning_rate": 4.610948231522527e-07,
+      "loss": 0.0054,
+      "step": 5228
+    },
+    {
+      "epoch": 18.675,
+      "grad_norm": 1.8791952649448451,
+      "learning_rate": 4.5862865630709187e-07,
+      "loss": 0.0215,
+      "step": 5229
+    },
+    {
+      "epoch": 18.678571428571427,
+      "grad_norm": 1.4587534357215763,
+      "learning_rate": 4.5616902581116396e-07,
+      "loss": 0.0325,
+      "step": 5230
+    },
+    {
+      "epoch": 18.682142857142857,
+      "grad_norm": 0.7952939487806203,
+      "learning_rate": 4.537159324871887e-07,
+      "loss": 0.0143,
+      "step": 5231
+    },
+    {
+      "epoch": 18.685714285714287,
+      "grad_norm": 1.068231419701942,
+      "learning_rate": 4.512693771556964e-07,
+      "loss": 0.0147,
+      "step": 5232
+    },
+    {
+      "epoch": 18.689285714285713,
+      "grad_norm": 1.5934811852170925,
+      "learning_rate": 4.488293606350258e-07,
+      "loss": 0.0206,
+      "step": 5233
+    },
+    {
+      "epoch": 18.692857142857143,
+      "grad_norm": 0.8300407627055895,
+      "learning_rate": 4.4639588374133956e-07,
+      "loss": 0.0098,
+      "step": 5234
+    },
+    {
+      "epoch": 18.696428571428573,
+      "grad_norm": 1.8754116823945604,
+      "learning_rate": 4.4396894728859775e-07,
+      "loss": 0.0327,
+      "step": 5235
+    },
+    {
+      "epoch": 18.7,
+      "grad_norm": 1.3740950832110623,
+      "learning_rate": 4.415485520885887e-07,
+      "loss": 0.0361,
+      "step": 5236
+    },
+    {
+      "epoch": 18.70357142857143,
+      "grad_norm": 0.734409140778637,
+      "learning_rate": 4.3913469895089823e-07,
+      "loss": 0.0152,
+      "step": 5237
+    },
+    {
+      "epoch": 18.707142857142856,
+      "grad_norm": 0.5846807226578723,
+      "learning_rate": 4.3672738868293595e-07,
+      "loss": 0.0102,
+      "step": 5238
+    },
+    {
+      "epoch": 18.710714285714285,
+      "grad_norm": 1.1247154255068756,
+      "learning_rate": 4.3432662208991337e-07,
+      "loss": 0.0201,
+      "step": 5239
+    },
+    {
+      "epoch": 18.714285714285715,
+      "grad_norm": 1.3258879108942203,
+      "learning_rate": 4.319323999748615e-07,
+      "loss": 0.0263,
+      "step": 5240
+    },
+    {
+      "epoch": 18.71785714285714,
+      "grad_norm": 0.9949004919268483,
+      "learning_rate": 4.295447231386152e-07,
+      "loss": 0.016,
+      "step": 5241
+    },
+    {
+      "epoch": 18.72142857142857,
+      "grad_norm": 1.8041463901621306,
+      "learning_rate": 4.2716359237982453e-07,
+      "loss": 0.0316,
+      "step": 5242
+    },
+    {
+      "epoch": 18.725,
+      "grad_norm": 1.3182979794617493,
+      "learning_rate": 4.247890084949502e-07,
+      "loss": 0.0205,
+      "step": 5243
+    },
+    {
+      "epoch": 18.728571428571428,
+      "grad_norm": 1.9746238340552127,
+      "learning_rate": 4.2242097227825686e-07,
+      "loss": 0.0358,
+      "step": 5244
+    },
+    {
+      "epoch": 18.732142857142858,
+      "grad_norm": 0.7596043841064669,
+      "learning_rate": 4.2005948452182645e-07,
+      "loss": 0.0115,
+      "step": 5245
+    },
+    {
+      "epoch": 18.735714285714288,
+      "grad_norm": 1.5698339327011823,
+      "learning_rate": 4.177045460155471e-07,
+      "loss": 0.0341,
+      "step": 5246
+    },
+    {
+      "epoch": 18.739285714285714,
+      "grad_norm": 0.9368801030098314,
+      "learning_rate": 4.1535615754711766e-07,
+      "loss": 0.019,
+      "step": 5247
+    },
+    {
+      "epoch": 18.742857142857144,
+      "grad_norm": 2.112354733645379,
+      "learning_rate": 4.130143199020453e-07,
+      "loss": 0.0266,
+      "step": 5248
+    },
+    {
+      "epoch": 18.74642857142857,
+      "grad_norm": 1.7599177726085296,
+      "learning_rate": 4.106790338636457e-07,
+      "loss": 0.0248,
+      "step": 5249
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 0.7032116194814677,
+      "learning_rate": 4.0835030021304065e-07,
+      "loss": 0.0118,
+      "step": 5250
+    },
+    {
+      "epoch": 18.75357142857143,
+      "grad_norm": 0.8188410726292813,
+      "learning_rate": 4.060281197291693e-07,
+      "loss": 0.0141,
+      "step": 5251
+    },
+    {
+      "epoch": 18.757142857142856,
+      "grad_norm": 1.3006237231057667,
+      "learning_rate": 4.0371249318876595e-07,
+      "loss": 0.0294,
+      "step": 5252
+    },
+    {
+      "epoch": 18.760714285714286,
+      "grad_norm": 0.6283764021897271,
+      "learning_rate": 4.014034213663842e-07,
+      "loss": 0.0073,
+      "step": 5253
+    },
+    {
+      "epoch": 18.764285714285712,
+      "grad_norm": 1.735408709164191,
+      "learning_rate": 3.9910090503437526e-07,
+      "loss": 0.0225,
+      "step": 5254
+    },
+    {
+      "epoch": 18.767857142857142,
+      "grad_norm": 1.3118957561273719,
+      "learning_rate": 3.9680494496290964e-07,
+      "loss": 0.0342,
+      "step": 5255
+    },
+    {
+      "epoch": 18.771428571428572,
+      "grad_norm": 0.8051826443214862,
+      "learning_rate": 3.945155419199531e-07,
+      "loss": 0.0161,
+      "step": 5256
+    },
+    {
+      "epoch": 18.775,
+      "grad_norm": 1.0950836574701022,
+      "learning_rate": 3.922326966712819e-07,
+      "loss": 0.0247,
+      "step": 5257
+    },
+    {
+      "epoch": 18.77857142857143,
+      "grad_norm": 0.9756634445637494,
+      "learning_rate": 3.899564099804831e-07,
+      "loss": 0.018,
+      "step": 5258
+    },
+    {
+      "epoch": 18.78214285714286,
+      "grad_norm": 1.5553390648892407,
+      "learning_rate": 3.8768668260894314e-07,
+      "loss": 0.0172,
+      "step": 5259
+    },
+    {
+      "epoch": 18.785714285714285,
+      "grad_norm": 0.8717446907379096,
+      "learning_rate": 3.8542351531586363e-07,
+      "loss": 0.0202,
+      "step": 5260
+    },
+    {
+      "epoch": 18.789285714285715,
+      "grad_norm": 1.6297777258735433,
+      "learning_rate": 3.8316690885823905e-07,
+      "loss": 0.0229,
+      "step": 5261
+    },
+    {
+      "epoch": 18.792857142857144,
+      "grad_norm": 0.9489727801684863,
+      "learning_rate": 3.809168639908811e-07,
+      "loss": 0.0179,
+      "step": 5262
+    },
+    {
+      "epoch": 18.79642857142857,
+      "grad_norm": 0.7714049226894092,
+      "learning_rate": 3.786733814664012e-07,
+      "loss": 0.0145,
+      "step": 5263
+    },
+    {
+      "epoch": 18.8,
+      "grad_norm": 0.6286974285907664,
+      "learning_rate": 3.7643646203521897e-07,
+      "loss": 0.0073,
+      "step": 5264
+    },
+    {
+      "epoch": 18.803571428571427,
+      "grad_norm": 1.0141856163256247,
+      "learning_rate": 3.7420610644555375e-07,
+      "loss": 0.0314,
+      "step": 5265
+    },
+    {
+      "epoch": 18.807142857142857,
+      "grad_norm": 0.7949330119107355,
+      "learning_rate": 3.7198231544343324e-07,
+      "loss": 0.0145,
+      "step": 5266
+    },
+    {
+      "epoch": 18.810714285714287,
+      "grad_norm": 0.344132573873548,
+      "learning_rate": 3.697650897726868e-07,
+      "loss": 0.0053,
+      "step": 5267
+    },
+    {
+      "epoch": 18.814285714285713,
+      "grad_norm": 0.6577458946917193,
+      "learning_rate": 3.6755443017495006e-07,
+      "loss": 0.0075,
+      "step": 5268
+    },
+    {
+      "epoch": 18.817857142857143,
+      "grad_norm": 1.1304341260543176,
+      "learning_rate": 3.65350337389665e-07,
+      "loss": 0.0129,
+      "step": 5269
+    },
+    {
+      "epoch": 18.821428571428573,
+      "grad_norm": 0.9017290872370353,
+      "learning_rate": 3.631528121540662e-07,
+      "loss": 0.0163,
+      "step": 5270
+    },
+    {
+      "epoch": 18.825,
+      "grad_norm": 0.5500342382065622,
+      "learning_rate": 3.609618552032057e-07,
+      "loss": 0.008,
+      "step": 5271
+    },
+    {
+      "epoch": 18.82857142857143,
+      "grad_norm": 1.1033932314151764,
+      "learning_rate": 3.5877746726992845e-07,
+      "loss": 0.0083,
+      "step": 5272
+    },
+    {
+      "epoch": 18.832142857142856,
+      "grad_norm": 0.7210000742539291,
+      "learning_rate": 3.5659964908488774e-07,
+      "loss": 0.0137,
+      "step": 5273
+    },
+    {
+      "epoch": 18.835714285714285,
+      "grad_norm": 1.7154815403983261,
+      "learning_rate": 3.544284013765342e-07,
+      "loss": 0.0229,
+      "step": 5274
+    },
+    {
+      "epoch": 18.839285714285715,
+      "grad_norm": 1.13816620921045,
+      "learning_rate": 3.522637248711269e-07,
+      "loss": 0.0474,
+      "step": 5275
+    },
+    {
+      "epoch": 18.84285714285714,
+      "grad_norm": 1.601641819024826,
+      "learning_rate": 3.5010562029271775e-07,
+      "loss": 0.0205,
+      "step": 5276
+    },
+    {
+      "epoch": 18.84642857142857,
+      "grad_norm": 1.8398703604024607,
+      "learning_rate": 3.479540883631738e-07,
+      "loss": 0.0221,
+      "step": 5277
+    },
+    {
+      "epoch": 18.85,
+      "grad_norm": 0.9794288806861601,
+      "learning_rate": 3.4580912980215043e-07,
+      "loss": 0.0301,
+      "step": 5278
+    },
+    {
+      "epoch": 18.853571428571428,
+      "grad_norm": 0.5002166040993359,
+      "learning_rate": 3.43670745327116e-07,
+      "loss": 0.0067,
+      "step": 5279
+    },
+    {
+      "epoch": 18.857142857142858,
+      "grad_norm": 0.525180654043675,
+      "learning_rate": 3.415389356533272e-07,
+      "loss": 0.0083,
+      "step": 5280
+    },
+    {
+      "epoch": 18.860714285714288,
+      "grad_norm": 1.2977279345155248,
+      "learning_rate": 3.3941370149385147e-07,
+      "loss": 0.0517,
+      "step": 5281
+    },
+    {
+      "epoch": 18.864285714285714,
+      "grad_norm": 0.9541579323088871,
+      "learning_rate": 3.3729504355955346e-07,
+      "loss": 0.0122,
+      "step": 5282
+    },
+    {
+      "epoch": 18.867857142857144,
+      "grad_norm": 1.7224125616774584,
+      "learning_rate": 3.351829625590952e-07,
+      "loss": 0.0247,
+      "step": 5283
+    },
+    {
+      "epoch": 18.87142857142857,
+      "grad_norm": 1.79497748857978,
+      "learning_rate": 3.330774591989494e-07,
+      "loss": 0.0282,
+      "step": 5284
+    },
+    {
+      "epoch": 18.875,
+      "grad_norm": 1.637756385400897,
+      "learning_rate": 3.3097853418337267e-07,
+      "loss": 0.0236,
+      "step": 5285
+    },
+    {
+      "epoch": 18.87857142857143,
+      "grad_norm": 1.485612440994496,
+      "learning_rate": 3.2888618821443673e-07,
+      "loss": 0.0239,
+      "step": 5286
+    },
+    {
+      "epoch": 18.882142857142856,
+      "grad_norm": 1.6782721378849774,
+      "learning_rate": 3.268004219919996e-07,
+      "loss": 0.0206,
+      "step": 5287
+    },
+    {
+      "epoch": 18.885714285714286,
+      "grad_norm": 1.5665314256031948,
+      "learning_rate": 3.2472123621372975e-07,
+      "loss": 0.0316,
+      "step": 5288
+    },
+    {
+      "epoch": 18.889285714285712,
+      "grad_norm": 1.1383605356772075,
+      "learning_rate": 3.226486315750843e-07,
+      "loss": 0.0149,
+      "step": 5289
+    },
+    {
+      "epoch": 18.892857142857142,
+      "grad_norm": 1.0401138433248005,
+      "learning_rate": 3.205826087693287e-07,
+      "loss": 0.0266,
+      "step": 5290
+    },
+    {
+      "epoch": 18.896428571428572,
+      "grad_norm": 1.2250786683293808,
+      "learning_rate": 3.1852316848751894e-07,
+      "loss": 0.0218,
+      "step": 5291
+    },
+    {
+      "epoch": 18.9,
+      "grad_norm": 0.7373600410057495,
+      "learning_rate": 3.1647031141851526e-07,
+      "loss": 0.0167,
+      "step": 5292
+    },
+    {
+      "epoch": 18.90357142857143,
+      "grad_norm": 1.1493483967141085,
+      "learning_rate": 3.144240382489727e-07,
+      "loss": 0.0305,
+      "step": 5293
+    },
+    {
+      "epoch": 18.90714285714286,
+      "grad_norm": 2.241916182763609,
+      "learning_rate": 3.123843496633416e-07,
+      "loss": 0.0353,
+      "step": 5294
+    },
+    {
+      "epoch": 18.910714285714285,
+      "grad_norm": 2.0871821172443923,
+      "learning_rate": 3.1035124634387626e-07,
+      "loss": 0.0202,
+      "step": 5295
+    },
+    {
+      "epoch": 18.914285714285715,
+      "grad_norm": 1.5791311323196409,
+      "learning_rate": 3.083247289706215e-07,
+      "loss": 0.0149,
+      "step": 5296
+    },
+    {
+      "epoch": 18.917857142857144,
+      "grad_norm": 0.5058889322294206,
+      "learning_rate": 3.0630479822142847e-07,
+      "loss": 0.0129,
+      "step": 5297
+    },
+    {
+      "epoch": 18.92142857142857,
+      "grad_norm": 1.0171424795846051,
+      "learning_rate": 3.0429145477193e-07,
+      "loss": 0.012,
+      "step": 5298
+    },
+    {
+      "epoch": 18.925,
+      "grad_norm": 0.7249829505412103,
+      "learning_rate": 3.0228469929557414e-07,
+      "loss": 0.0137,
+      "step": 5299
+    },
+    {
+      "epoch": 18.928571428571427,
+      "grad_norm": 2.086414753147997,
+      "learning_rate": 3.0028453246359057e-07,
+      "loss": 0.0341,
+      "step": 5300
+    },
+    {
+      "epoch": 18.932142857142857,
+      "grad_norm": 0.9425684684558844,
+      "learning_rate": 2.9829095494501303e-07,
+      "loss": 0.0215,
+      "step": 5301
+    },
+    {
+      "epoch": 18.935714285714287,
+      "grad_norm": 0.879592455968814,
+      "learning_rate": 2.963039674066659e-07,
+      "loss": 0.0245,
+      "step": 5302
+    },
+    {
+      "epoch": 18.939285714285713,
+      "grad_norm": 1.8606565923754868,
+      "learning_rate": 2.9432357051317527e-07,
+      "loss": 0.0304,
+      "step": 5303
+    },
+    {
+      "epoch": 18.942857142857143,
+      "grad_norm": 0.9052135018239197,
+      "learning_rate": 2.9234976492696023e-07,
+      "loss": 0.0147,
+      "step": 5304
+    },
+    {
+      "epoch": 18.946428571428573,
+      "grad_norm": 0.8823024648150856,
+      "learning_rate": 2.903825513082281e-07,
+      "loss": 0.0192,
+      "step": 5305
+    },
+    {
+      "epoch": 18.95,
+      "grad_norm": 0.9098271484949991,
+      "learning_rate": 2.884219303149971e-07,
+      "loss": 0.0156,
+      "step": 5306
+    },
+    {
+      "epoch": 18.95357142857143,
+      "grad_norm": 1.204303631104701,
+      "learning_rate": 2.864679026030626e-07,
+      "loss": 0.0072,
+      "step": 5307
+    },
+    {
+      "epoch": 18.957142857142856,
+      "grad_norm": 0.6575186822059415,
+      "learning_rate": 2.8452046882602834e-07,
+      "loss": 0.0126,
+      "step": 5308
+    },
+    {
+      "epoch": 18.960714285714285,
+      "grad_norm": 2.150656007481457,
+      "learning_rate": 2.825796296352823e-07,
+      "loss": 0.0288,
+      "step": 5309
+    },
+    {
+      "epoch": 18.964285714285715,
+      "grad_norm": 0.8980653452031634,
+      "learning_rate": 2.8064538568001843e-07,
+      "loss": 0.0242,
+      "step": 5310
+    },
+    {
+      "epoch": 18.96785714285714,
+      "grad_norm": 0.8496702637243106,
+      "learning_rate": 2.7871773760721033e-07,
+      "loss": 0.0115,
+      "step": 5311
+    },
+    {
+      "epoch": 18.97142857142857,
+      "grad_norm": 1.3977838524021389,
+      "learning_rate": 2.767966860616378e-07,
+      "loss": 0.0296,
+      "step": 5312
+    },
+    {
+      "epoch": 18.975,
+      "grad_norm": 0.8958659013805436,
+      "learning_rate": 2.748822316858646e-07,
+      "loss": 0.0145,
+      "step": 5313
+    },
+    {
+      "epoch": 18.978571428571428,
+      "grad_norm": 2.104587547681915,
+      "learning_rate": 2.7297437512025626e-07,
+      "loss": 0.0274,
+      "step": 5314
+    },
+    {
+      "epoch": 18.982142857142858,
+      "grad_norm": 1.1171207591441905,
+      "learning_rate": 2.7107311700296677e-07,
+      "loss": 0.0155,
+      "step": 5315
+    },
+    {
+      "epoch": 18.985714285714288,
+      "grad_norm": 2.1706844631720847,
+      "learning_rate": 2.691784579699408e-07,
+      "loss": 0.0412,
+      "step": 5316
+    },
+    {
+      "epoch": 18.989285714285714,
+      "grad_norm": 1.3543395648826413,
+      "learning_rate": 2.6729039865492245e-07,
+      "loss": 0.0238,
+      "step": 5317
+    },
+    {
+      "epoch": 18.992857142857144,
+      "grad_norm": 0.9413488228830238,
+      "learning_rate": 2.6540893968944216e-07,
+      "loss": 0.0094,
+      "step": 5318
+    },
+    {
+      "epoch": 18.99642857142857,
+      "grad_norm": 0.7915310116691506,
+      "learning_rate": 2.6353408170282534e-07,
+      "loss": 0.0137,
+      "step": 5319
+    },
+    {
+      "epoch": 19.0,
+      "grad_norm": 1.2261650980091146,
+      "learning_rate": 2.61665825322186e-07,
+      "loss": 0.0189,
+      "step": 5320
+    },
+    {
+      "epoch": 19.00357142857143,
+      "grad_norm": 0.6346011891718591,
+      "learning_rate": 2.5980417117243973e-07,
+      "loss": 0.0086,
+      "step": 5321
+    },
+    {
+      "epoch": 19.007142857142856,
+      "grad_norm": 0.8688151385322014,
+      "learning_rate": 2.5794911987628177e-07,
+      "loss": 0.0142,
+      "step": 5322
+    },
+    {
+      "epoch": 19.010714285714286,
+      "grad_norm": 0.5303981772159176,
+      "learning_rate": 2.5610067205420473e-07,
+      "loss": 0.012,
+      "step": 5323
+    },
+    {
+      "epoch": 19.014285714285716,
+      "grad_norm": 1.6970249736655623,
+      "learning_rate": 2.542588283244918e-07,
+      "loss": 0.0264,
+      "step": 5324
+    },
+    {
+      "epoch": 19.017857142857142,
+      "grad_norm": 0.4608871912500528,
+      "learning_rate": 2.524235893032212e-07,
+      "loss": 0.0056,
+      "step": 5325
+    },
+    {
+      "epoch": 19.021428571428572,
+      "grad_norm": 0.6528775120442805,
+      "learning_rate": 2.5059495560425086e-07,
+      "loss": 0.0109,
+      "step": 5326
+    },
+    {
+      "epoch": 19.025,
+      "grad_norm": 0.9888746433548655,
+      "learning_rate": 2.487729278392426e-07,
+      "loss": 0.0141,
+      "step": 5327
+    },
+    {
+      "epoch": 19.02857142857143,
+      "grad_norm": 1.0195784339774607,
+      "learning_rate": 2.469575066176377e-07,
+      "loss": 0.024,
+      "step": 5328
+    },
+    {
+      "epoch": 19.03214285714286,
+      "grad_norm": 0.9220830577415371,
+      "learning_rate": 2.451486925466773e-07,
+      "loss": 0.0164,
+      "step": 5329
+    },
+    {
+      "epoch": 19.035714285714285,
+      "grad_norm": 1.286653039995291,
+      "learning_rate": 2.433464862313839e-07,
+      "loss": 0.0109,
+      "step": 5330
+    },
+    {
+      "epoch": 19.039285714285715,
+      "grad_norm": 0.8713568814699321,
+      "learning_rate": 2.4155088827457095e-07,
+      "loss": 0.0096,
+      "step": 5331
+    },
+    {
+      "epoch": 19.042857142857144,
+      "grad_norm": 1.0492820527766844,
+      "learning_rate": 2.397618992768513e-07,
+      "loss": 0.0174,
+      "step": 5332
+    },
+    {
+      "epoch": 19.04642857142857,
+      "grad_norm": 2.138526789651721,
+      "learning_rate": 2.3797951983661306e-07,
+      "loss": 0.025,
+      "step": 5333
+    },
+    {
+      "epoch": 19.05,
+      "grad_norm": 0.8202323641335345,
+      "learning_rate": 2.3620375055004364e-07,
+      "loss": 0.0133,
+      "step": 5334
+    },
+    {
+      "epoch": 19.053571428571427,
+      "grad_norm": 2.0378919647593348,
+      "learning_rate": 2.3443459201111685e-07,
+      "loss": 0.0266,
+      "step": 5335
+    },
+    {
+      "epoch": 19.057142857142857,
+      "grad_norm": 0.9910270085712838,
+      "learning_rate": 2.3267204481159045e-07,
+      "loss": 0.0206,
+      "step": 5336
+    },
+    {
+      "epoch": 19.060714285714287,
+      "grad_norm": 0.7764261327143317,
+      "learning_rate": 2.309161095410195e-07,
+      "loss": 0.0093,
+      "step": 5337
+    },
+    {
+      "epoch": 19.064285714285713,
+      "grad_norm": 1.2560823638959067,
+      "learning_rate": 2.2916678678674087e-07,
+      "loss": 0.0324,
+      "step": 5338
+    },
+    {
+      "epoch": 19.067857142857143,
+      "grad_norm": 0.8988131771705637,
+      "learning_rate": 2.274240771338798e-07,
+      "loss": 0.0168,
+      "step": 5339
+    },
+    {
+      "epoch": 19.071428571428573,
+      "grad_norm": 1.4339794318521304,
+      "learning_rate": 2.2568798116535673e-07,
+      "loss": 0.0184,
+      "step": 5340
+    },
+    {
+      "epoch": 19.075,
+      "grad_norm": 1.05632313144604,
+      "learning_rate": 2.2395849946186931e-07,
+      "loss": 0.0237,
+      "step": 5341
+    },
+    {
+      "epoch": 19.07857142857143,
+      "grad_norm": 0.806866397906503,
+      "learning_rate": 2.2223563260190817e-07,
+      "loss": 0.0101,
+      "step": 5342
+    },
+    {
+      "epoch": 19.082142857142856,
+      "grad_norm": 0.9347796695172643,
+      "learning_rate": 2.2051938116175232e-07,
+      "loss": 0.0135,
+      "step": 5343
+    },
+    {
+      "epoch": 19.085714285714285,
+      "grad_norm": 0.9192351327364195,
+      "learning_rate": 2.188097457154692e-07,
+      "loss": 0.0218,
+      "step": 5344
+    },
+    {
+      "epoch": 19.089285714285715,
+      "grad_norm": 1.1758807283201564,
+      "learning_rate": 2.1710672683490807e-07,
+      "loss": 0.0106,
+      "step": 5345
+    },
+    {
+      "epoch": 19.09285714285714,
+      "grad_norm": 1.5405722259415817,
+      "learning_rate": 2.154103250897044e-07,
+      "loss": 0.0253,
+      "step": 5346
+    },
+    {
+      "epoch": 19.09642857142857,
+      "grad_norm": 0.9989358217053825,
+      "learning_rate": 2.1372054104729312e-07,
+      "loss": 0.0145,
+      "step": 5347
+    },
+    {
+      "epoch": 19.1,
+      "grad_norm": 0.9627224621519744,
+      "learning_rate": 2.1203737527287994e-07,
+      "loss": 0.0105,
+      "step": 5348
+    },
+    {
+      "epoch": 19.103571428571428,
+      "grad_norm": 0.7514045485411617,
+      "learning_rate": 2.1036082832946336e-07,
+      "loss": 0.0161,
+      "step": 5349
+    },
+    {
+      "epoch": 19.107142857142858,
+      "grad_norm": 1.347291231478894,
+      "learning_rate": 2.086909007778304e-07,
+      "loss": 0.0283,
+      "step": 5350
+    },
+    {
+      "epoch": 19.110714285714284,
+      "grad_norm": 0.43600628552086096,
+      "learning_rate": 2.0702759317654753e-07,
+      "loss": 0.0089,
+      "step": 5351
+    },
+    {
+      "epoch": 19.114285714285714,
+      "grad_norm": 1.1589166494996137,
+      "learning_rate": 2.053709060819764e-07,
+      "loss": 0.0158,
+      "step": 5352
+    },
+    {
+      "epoch": 19.117857142857144,
+      "grad_norm": 0.9548231371519718,
+      "learning_rate": 2.0372084004825155e-07,
+      "loss": 0.0212,
+      "step": 5353
+    },
+    {
+      "epoch": 19.12142857142857,
+      "grad_norm": 0.39591933454498784,
+      "learning_rate": 2.0207739562730479e-07,
+      "loss": 0.0056,
+      "step": 5354
+    },
+    {
+      "epoch": 19.125,
+      "grad_norm": 0.772035360805719,
+      "learning_rate": 2.0044057336884749e-07,
+      "loss": 0.0217,
+      "step": 5355
+    },
+    {
+      "epoch": 19.12857142857143,
+      "grad_norm": 1.0673137967608692,
+      "learning_rate": 1.9881037382037504e-07,
+      "loss": 0.0139,
+      "step": 5356
+    },
+    {
+      "epoch": 19.132142857142856,
+      "grad_norm": 0.5952057387408205,
+      "learning_rate": 1.9718679752716907e-07,
+      "loss": 0.0096,
+      "step": 5357
+    },
+    {
+      "epoch": 19.135714285714286,
+      "grad_norm": 0.6920267002757055,
+      "learning_rate": 1.9556984503229958e-07,
+      "loss": 0.0121,
+      "step": 5358
+    },
+    {
+      "epoch": 19.139285714285716,
+      "grad_norm": 1.0748443650088495,
+      "learning_rate": 1.9395951687661175e-07,
+      "loss": 0.0206,
+      "step": 5359
+    },
+    {
+      "epoch": 19.142857142857142,
+      "grad_norm": 2.2950920163149022,
+      "learning_rate": 1.9235581359874577e-07,
+      "loss": 0.0199,
+      "step": 5360
+    },
+    {
+      "epoch": 19.146428571428572,
+      "grad_norm": 0.7635295984169685,
+      "learning_rate": 1.9075873573511926e-07,
+      "loss": 0.0223,
+      "step": 5361
+    },
+    {
+      "epoch": 19.15,
+      "grad_norm": 0.7698733814941253,
+      "learning_rate": 1.891682838199338e-07,
+      "loss": 0.0234,
+      "step": 5362
+    },
+    {
+      "epoch": 19.15357142857143,
+      "grad_norm": 0.8876395164712324,
+      "learning_rate": 1.8758445838517713e-07,
+      "loss": 0.0129,
+      "step": 5363
+    },
+    {
+      "epoch": 19.15714285714286,
+      "grad_norm": 2.3828024836626747,
+      "learning_rate": 1.8600725996061664e-07,
+      "loss": 0.034,
+      "step": 5364
+    },
+    {
+      "epoch": 19.160714285714285,
+      "grad_norm": 1.6286272758014828,
+      "learning_rate": 1.844366890738103e-07,
+      "loss": 0.0285,
+      "step": 5365
+    },
+    {
+      "epoch": 19.164285714285715,
+      "grad_norm": 0.9546913664337794,
+      "learning_rate": 1.8287274625009122e-07,
+      "loss": 0.0254,
+      "step": 5366
+    },
+    {
+      "epoch": 19.167857142857144,
+      "grad_norm": 0.9566191352050827,
+      "learning_rate": 1.8131543201258316e-07,
+      "loss": 0.0128,
+      "step": 5367
+    },
+    {
+      "epoch": 19.17142857142857,
+      "grad_norm": 0.9280358269593864,
+      "learning_rate": 1.7976474688218503e-07,
+      "loss": 0.0204,
+      "step": 5368
+    },
+    {
+      "epoch": 19.175,
+      "grad_norm": 2.219722014920552,
+      "learning_rate": 1.7822069137758414e-07,
+      "loss": 0.0163,
+      "step": 5369
+    },
+    {
+      "epoch": 19.178571428571427,
+      "grad_norm": 1.5039760295514166,
+      "learning_rate": 1.766832660152451e-07,
+      "loss": 0.0171,
+      "step": 5370
+    },
+    {
+      "epoch": 19.182142857142857,
+      "grad_norm": 1.8425906471111408,
+      "learning_rate": 1.7515247130942102e-07,
+      "loss": 0.0454,
+      "step": 5371
+    },
+    {
+      "epoch": 19.185714285714287,
+      "grad_norm": 1.2517672590853706,
+      "learning_rate": 1.7362830777214457e-07,
+      "loss": 0.0215,
+      "step": 5372
+    },
+    {
+      "epoch": 19.189285714285713,
+      "grad_norm": 0.810327131846394,
+      "learning_rate": 1.7211077591322566e-07,
+      "loss": 0.0116,
+      "step": 5373
+    },
+    {
+      "epoch": 19.192857142857143,
+      "grad_norm": 0.885730561495021,
+      "learning_rate": 1.7059987624026498e-07,
+      "loss": 0.0079,
+      "step": 5374
+    },
+    {
+      "epoch": 19.196428571428573,
+      "grad_norm": 1.8163659574541462,
+      "learning_rate": 1.6909560925863379e-07,
+      "loss": 0.0218,
+      "step": 5375
+    },
+    {
+      "epoch": 19.2,
+      "grad_norm": 1.090062891362715,
+      "learning_rate": 1.6759797547149849e-07,
+      "loss": 0.0177,
+      "step": 5376
+    },
+    {
+      "epoch": 19.20357142857143,
+      "grad_norm": 0.8618814786923563,
+      "learning_rate": 1.6610697537979171e-07,
+      "loss": 0.0135,
+      "step": 5377
+    },
+    {
+      "epoch": 19.207142857142856,
+      "grad_norm": 1.308916848598732,
+      "learning_rate": 1.6462260948224118e-07,
+      "loss": 0.0158,
+      "step": 5378
+    },
+    {
+      "epoch": 19.210714285714285,
+      "grad_norm": 0.6431099074345615,
+      "learning_rate": 1.6314487827534532e-07,
+      "loss": 0.0112,
+      "step": 5379
+    },
+    {
+      "epoch": 19.214285714285715,
+      "grad_norm": 0.8169301295176795,
+      "learning_rate": 1.616737822533887e-07,
+      "loss": 0.0193,
+      "step": 5380
+    },
+    {
+      "epoch": 19.21785714285714,
+      "grad_norm": 1.2823335363451758,
+      "learning_rate": 1.6020932190843553e-07,
+      "loss": 0.0123,
+      "step": 5381
+    },
+    {
+      "epoch": 19.22142857142857,
+      "grad_norm": 1.3885727257565954,
+      "learning_rate": 1.5875149773032728e-07,
+      "loss": 0.0166,
+      "step": 5382
+    },
+    {
+      "epoch": 19.225,
+      "grad_norm": 1.0799289525122728,
+      "learning_rate": 1.5730031020668946e-07,
+      "loss": 0.0165,
+      "step": 5383
+    },
+    {
+      "epoch": 19.228571428571428,
+      "grad_norm": 0.9328685183423945,
+      "learning_rate": 1.5585575982292712e-07,
+      "loss": 0.0304,
+      "step": 5384
+    },
+    {
+      "epoch": 19.232142857142858,
+      "grad_norm": 2.344710531303069,
+      "learning_rate": 1.5441784706222484e-07,
+      "loss": 0.0376,
+      "step": 5385
+    },
+    {
+      "epoch": 19.235714285714284,
+      "grad_norm": 1.1864508567400192,
+      "learning_rate": 1.529865724055446e-07,
+      "loss": 0.0243,
+      "step": 5386
+    },
+    {
+      "epoch": 19.239285714285714,
+      "grad_norm": 0.5471607320748831,
+      "learning_rate": 1.515619363316323e-07,
+      "loss": 0.0074,
+      "step": 5387
+    },
+    {
+      "epoch": 19.242857142857144,
+      "grad_norm": 1.3192940470788639,
+      "learning_rate": 1.501439393170112e-07,
+      "loss": 0.017,
+      "step": 5388
+    },
+    {
+      "epoch": 19.24642857142857,
+      "grad_norm": 0.46030473034920544,
+      "learning_rate": 1.4873258183598414e-07,
+      "loss": 0.0096,
+      "step": 5389
+    },
+    {
+      "epoch": 19.25,
+      "grad_norm": 0.8188875923074287,
+      "learning_rate": 1.4732786436063128e-07,
+      "loss": 0.0185,
+      "step": 5390
+    },
+    {
+      "epoch": 19.25357142857143,
+      "grad_norm": 1.0747179445044523,
+      "learning_rate": 1.4592978736081455e-07,
+      "loss": 0.015,
+      "step": 5391
+    },
+    {
+      "epoch": 19.257142857142856,
+      "grad_norm": 1.0907583907480625,
+      "learning_rate": 1.4453835130417315e-07,
+      "loss": 0.0244,
+      "step": 5392
+    },
+    {
+      "epoch": 19.260714285714286,
+      "grad_norm": 2.869659614049801,
+      "learning_rate": 1.4315355665612595e-07,
+      "loss": 0.0149,
+      "step": 5393
+    },
+    {
+      "epoch": 19.264285714285716,
+      "grad_norm": 0.8814792334066761,
+      "learning_rate": 1.417754038798691e-07,
+      "loss": 0.0192,
+      "step": 5394
+    },
+    {
+      "epoch": 19.267857142857142,
+      "grad_norm": 0.8396053799305839,
+      "learning_rate": 1.404038934363805e-07,
+      "loss": 0.0194,
+      "step": 5395
+    },
+    {
+      "epoch": 19.271428571428572,
+      "grad_norm": 1.123622401939535,
+      "learning_rate": 1.3903902578440875e-07,
+      "loss": 0.0077,
+      "step": 5396
+    },
+    {
+      "epoch": 19.275,
+      "grad_norm": 0.9263867810272571,
+      "learning_rate": 1.3768080138048868e-07,
+      "loss": 0.0106,
+      "step": 5397
+    },
+    {
+      "epoch": 19.27857142857143,
+      "grad_norm": 0.9533806373987318,
+      "learning_rate": 1.3632922067893019e-07,
+      "loss": 0.0198,
+      "step": 5398
+    },
+    {
+      "epoch": 19.28214285714286,
+      "grad_norm": 1.5384369803781694,
+      "learning_rate": 1.349842841318183e-07,
+      "loss": 0.0382,
+      "step": 5399
+    },
+    {
+      "epoch": 19.285714285714285,
+      "grad_norm": 0.8731050932624441,
+      "learning_rate": 1.3364599218901985e-07,
+      "loss": 0.0263,
+      "step": 5400
+    },
+    {
+      "epoch": 19.289285714285715,
+      "grad_norm": 0.695368928072874,
+      "learning_rate": 1.3231434529817234e-07,
+      "loss": 0.0124,
+      "step": 5401
+    },
+    {
+      "epoch": 19.292857142857144,
+      "grad_norm": 0.6606032312090985,
+      "learning_rate": 1.3098934390470163e-07,
+      "loss": 0.0094,
+      "step": 5402
+    },
+    {
+      "epoch": 19.29642857142857,
+      "grad_norm": 2.118361282312603,
+      "learning_rate": 1.2967098845180214e-07,
+      "loss": 0.0324,
+      "step": 5403
+    },
+    {
+      "epoch": 19.3,
+      "grad_norm": 0.6482794705806196,
+      "learning_rate": 1.2835927938044778e-07,
+      "loss": 0.0124,
+      "step": 5404
+    },
+    {
+      "epoch": 19.303571428571427,
+      "grad_norm": 1.2843441827699527,
+      "learning_rate": 1.2705421712938538e-07,
+      "loss": 0.0097,
+      "step": 5405
+    },
+    {
+      "epoch": 19.307142857142857,
+      "grad_norm": 1.4036373418758739,
+      "learning_rate": 1.2575580213514792e-07,
+      "loss": 0.023,
+      "step": 5406
+    },
+    {
+      "epoch": 19.310714285714287,
+      "grad_norm": 1.633667487730367,
+      "learning_rate": 1.2446403483203696e-07,
+      "loss": 0.0297,
+      "step": 5407
+    },
+    {
+      "epoch": 19.314285714285713,
+      "grad_norm": 1.7490068882969163,
+      "learning_rate": 1.2317891565213348e-07,
+      "loss": 0.0127,
+      "step": 5408
+    },
+    {
+      "epoch": 19.317857142857143,
+      "grad_norm": 0.3700475700798246,
+      "learning_rate": 1.2190044502529142e-07,
+      "loss": 0.0056,
+      "step": 5409
+    },
+    {
+      "epoch": 19.321428571428573,
+      "grad_norm": 1.491062556943054,
+      "learning_rate": 1.206286233791465e-07,
+      "loss": 0.0388,
+      "step": 5410
+    },
+    {
+      "epoch": 19.325,
+      "grad_norm": 1.7118461078440435,
+      "learning_rate": 1.193634511391073e-07,
+      "loss": 0.0237,
+      "step": 5411
+    },
+    {
+      "epoch": 19.32857142857143,
+      "grad_norm": 1.255052974867749,
+      "learning_rate": 1.1810492872835533e-07,
+      "loss": 0.0334,
+      "step": 5412
+    },
+    {
+      "epoch": 19.332142857142856,
+      "grad_norm": 1.0385339262373612,
+      "learning_rate": 1.1685305656785606e-07,
+      "loss": 0.0269,
+      "step": 5413
+    },
+    {
+      "epoch": 19.335714285714285,
+      "grad_norm": 1.517979873828689,
+      "learning_rate": 1.15607835076339e-07,
+      "loss": 0.0211,
+      "step": 5414
+    },
+    {
+      "epoch": 19.339285714285715,
+      "grad_norm": 1.0055378879850785,
+      "learning_rate": 1.1436926467032205e-07,
+      "loss": 0.0143,
+      "step": 5415
+    },
+    {
+      "epoch": 19.34285714285714,
+      "grad_norm": 0.8366764569674315,
+      "learning_rate": 1.1313734576408719e-07,
+      "loss": 0.0147,
+      "step": 5416
+    },
+    {
+      "epoch": 19.34642857142857,
+      "grad_norm": 0.88612190394732,
+      "learning_rate": 1.1191207876969812e-07,
+      "loss": 0.0177,
+      "step": 5417
+    },
+    {
+      "epoch": 19.35,
+      "grad_norm": 2.0831662579190295,
+      "learning_rate": 1.1069346409699144e-07,
+      "loss": 0.0258,
+      "step": 5418
+    },
+    {
+      "epoch": 19.353571428571428,
+      "grad_norm": 0.5684815083344572,
+      "learning_rate": 1.0948150215357889e-07,
+      "loss": 0.0078,
+      "step": 5419
+    },
+    {
+      "epoch": 19.357142857142858,
+      "grad_norm": 0.8438527625494998,
+      "learning_rate": 1.0827619334484729e-07,
+      "loss": 0.0091,
+      "step": 5420
+    },
+    {
+      "epoch": 19.360714285714284,
+      "grad_norm": 1.6197652244760568,
+      "learning_rate": 1.0707753807395637e-07,
+      "loss": 0.0386,
+      "step": 5421
+    },
+    {
+      "epoch": 19.364285714285714,
+      "grad_norm": 1.5904864584535585,
+      "learning_rate": 1.058855367418432e-07,
+      "loss": 0.0153,
+      "step": 5422
+    },
+    {
+      "epoch": 19.367857142857144,
+      "grad_norm": 2.1944453997742377,
+      "learning_rate": 1.0470018974721329e-07,
+      "loss": 0.0272,
+      "step": 5423
+    },
+    {
+      "epoch": 19.37142857142857,
+      "grad_norm": 1.499224236331948,
+      "learning_rate": 1.0352149748655838e-07,
+      "loss": 0.0214,
+      "step": 5424
+    },
+    {
+      "epoch": 19.375,
+      "grad_norm": 1.4106571676256738,
+      "learning_rate": 1.0234946035412973e-07,
+      "loss": 0.0267,
+      "step": 5425
+    },
+    {
+      "epoch": 19.37857142857143,
+      "grad_norm": 2.4552252922810487,
+      "learning_rate": 1.0118407874196267e-07,
+      "loss": 0.0231,
+      "step": 5426
+    },
+    {
+      "epoch": 19.382142857142856,
+      "grad_norm": 0.5130226091339126,
+      "learning_rate": 1.0002535303986316e-07,
+      "loss": 0.0068,
+      "step": 5427
+    },
+    {
+      "epoch": 19.385714285714286,
+      "grad_norm": 1.0201333344496462,
+      "learning_rate": 9.887328363541005e-08,
+      "loss": 0.0214,
+      "step": 5428
+    },
+    {
+      "epoch": 19.389285714285716,
+      "grad_norm": 1.4837628851393245,
+      "learning_rate": 9.772787091395508e-08,
+      "loss": 0.0353,
+      "step": 5429
+    },
+    {
+      "epoch": 19.392857142857142,
+      "grad_norm": 0.41095590204328325,
+      "learning_rate": 9.658911525862735e-08,
+      "loss": 0.0079,
+      "step": 5430
+    },
+    {
+      "epoch": 19.396428571428572,
+      "grad_norm": 1.1142285411677608,
+      "learning_rate": 9.54570170503244e-08,
+      "loss": 0.016,
+      "step": 5431
+    },
+    {
+      "epoch": 19.4,
+      "grad_norm": 0.7702834285871435,
+      "learning_rate": 9.433157666772109e-08,
+      "loss": 0.009,
+      "step": 5432
+    },
+    {
+      "epoch": 19.40357142857143,
+      "grad_norm": 0.8035745918614586,
+      "learning_rate": 9.321279448726073e-08,
+      "loss": 0.0179,
+      "step": 5433
+    },
+    {
+      "epoch": 19.40714285714286,
+      "grad_norm": 0.8573953609777022,
+      "learning_rate": 9.210067088316621e-08,
+      "loss": 0.0096,
+      "step": 5434
+    },
+    {
+      "epoch": 19.410714285714285,
+      "grad_norm": 0.2825072345591127,
+      "learning_rate": 9.099520622742663e-08,
+      "loss": 0.0044,
+      "step": 5435
+    },
+    {
+      "epoch": 19.414285714285715,
+      "grad_norm": 1.6889830712655218,
+      "learning_rate": 8.989640088980623e-08,
+      "loss": 0.037,
+      "step": 5436
+    },
+    {
+      "epoch": 19.417857142857144,
+      "grad_norm": 1.87245683343811,
+      "learning_rate": 8.880425523784209e-08,
+      "loss": 0.0375,
+      "step": 5437
+    },
+    {
+      "epoch": 19.42142857142857,
+      "grad_norm": 1.0436541288493015,
+      "learning_rate": 8.771876963684422e-08,
+      "loss": 0.0203,
+      "step": 5438
+    },
+    {
+      "epoch": 19.425,
+      "grad_norm": 1.2685369320055244,
+      "learning_rate": 8.663994444989554e-08,
+      "loss": 0.0182,
+      "step": 5439
+    },
+    {
+      "epoch": 19.428571428571427,
+      "grad_norm": 0.9314824827046295,
+      "learning_rate": 8.556778003784738e-08,
+      "loss": 0.01,
+      "step": 5440
+    },
+    {
+      "epoch": 19.432142857142857,
+      "grad_norm": 2.065305696732633,
+      "learning_rate": 8.450227675932621e-08,
+      "loss": 0.0207,
+      "step": 5441
+    },
+    {
+      "epoch": 19.435714285714287,
+      "grad_norm": 2.0617433894372,
+      "learning_rate": 8.344343497072915e-08,
+      "loss": 0.023,
+      "step": 5442
+    },
+    {
+      "epoch": 19.439285714285713,
+      "grad_norm": 0.3297579782396536,
+      "learning_rate": 8.239125502622847e-08,
+      "loss": 0.0053,
+      "step": 5443
+    },
+    {
+      "epoch": 19.442857142857143,
+      "grad_norm": 0.7723368799866238,
+      "learning_rate": 8.134573727776262e-08,
+      "loss": 0.0138,
+      "step": 5444
+    },
+    {
+      "epoch": 19.446428571428573,
+      "grad_norm": 2.4870968590678046,
+      "learning_rate": 8.030688207504744e-08,
+      "loss": 0.0224,
+      "step": 5445
+    },
+    {
+      "epoch": 19.45,
+      "grad_norm": 1.1092556562670413,
+      "learning_rate": 7.927468976556496e-08,
+      "loss": 0.0272,
+      "step": 5446
+    },
+    {
+      "epoch": 19.45357142857143,
+      "grad_norm": 0.5015212068291374,
+      "learning_rate": 7.824916069457011e-08,
+      "loss": 0.0084,
+      "step": 5447
+    },
+    {
+      "epoch": 19.457142857142856,
+      "grad_norm": 1.5341972988072117,
+      "learning_rate": 7.723029520509296e-08,
+      "loss": 0.0125,
+      "step": 5448
+    },
+    {
+      "epoch": 19.460714285714285,
+      "grad_norm": 0.7072912292761192,
+      "learning_rate": 7.621809363792753e-08,
+      "loss": 0.0097,
+      "step": 5449
+    },
+    {
+      "epoch": 19.464285714285715,
+      "grad_norm": 1.329912998330568,
+      "learning_rate": 7.521255633164748e-08,
+      "loss": 0.0202,
+      "step": 5450
+    },
+    {
+      "epoch": 19.46785714285714,
+      "grad_norm": 0.4578042328984412,
+      "learning_rate": 7.421368362258819e-08,
+      "loss": 0.0103,
+      "step": 5451
+    },
+    {
+      "epoch": 19.47142857142857,
+      "grad_norm": 0.18111924165184884,
+      "learning_rate": 7.322147584486239e-08,
+      "loss": 0.0023,
+      "step": 5452
+    },
+    {
+      "epoch": 19.475,
+      "grad_norm": 1.637058339476018,
+      "learning_rate": 7.223593333035129e-08,
+      "loss": 0.0104,
+      "step": 5453
+    },
+    {
+      "epoch": 19.478571428571428,
+      "grad_norm": 0.5273065790334182,
+      "learning_rate": 7.125705640870894e-08,
+      "loss": 0.0071,
+      "step": 5454
+    },
+    {
+      "epoch": 19.482142857142858,
+      "grad_norm": 1.2148383412976698,
+      "learning_rate": 7.028484540735347e-08,
+      "loss": 0.0189,
+      "step": 5455
+    },
+    {
+      "epoch": 19.485714285714284,
+      "grad_norm": 1.332066398884702,
+      "learning_rate": 6.93193006514803e-08,
+      "loss": 0.0187,
+      "step": 5456
+    },
+    {
+      "epoch": 19.489285714285714,
+      "grad_norm": 1.318440692875042,
+      "learning_rate": 6.836042246405106e-08,
+      "loss": 0.0137,
+      "step": 5457
+    },
+    {
+      "epoch": 19.492857142857144,
+      "grad_norm": 1.8113020810510452,
+      "learning_rate": 6.740821116580032e-08,
+      "loss": 0.0357,
+      "step": 5458
+    },
+    {
+      "epoch": 19.49642857142857,
+      "grad_norm": 0.909031634137426,
+      "learning_rate": 6.646266707522886e-08,
+      "loss": 0.0093,
+      "step": 5459
+    },
+    {
+      "epoch": 19.5,
+      "grad_norm": 0.7857335533930743,
+      "learning_rate": 6.552379050861257e-08,
+      "loss": 0.0132,
+      "step": 5460
+    },
+    {
+      "epoch": 19.50357142857143,
+      "grad_norm": 1.0420565437814064,
+      "learning_rate": 6.459158177999136e-08,
+      "loss": 0.0188,
+      "step": 5461
+    },
+    {
+      "epoch": 19.507142857142856,
+      "grad_norm": 1.1640444574041893,
+      "learning_rate": 6.366604120117803e-08,
+      "loss": 0.0151,
+      "step": 5462
+    },
+    {
+      "epoch": 19.510714285714286,
+      "grad_norm": 0.5811066111437392,
+      "learning_rate": 6.274716908175826e-08,
+      "loss": 0.016,
+      "step": 5463
+    },
+    {
+      "epoch": 19.514285714285712,
+      "grad_norm": 0.9002906243156895,
+      "learning_rate": 6.183496572907733e-08,
+      "loss": 0.0074,
+      "step": 5464
+    },
+    {
+      "epoch": 19.517857142857142,
+      "grad_norm": 0.7371523219688426,
+      "learning_rate": 6.092943144826002e-08,
+      "loss": 0.0142,
+      "step": 5465
+    },
+    {
+      "epoch": 19.521428571428572,
+      "grad_norm": 1.7073663414298095,
+      "learning_rate": 6.003056654219741e-08,
+      "loss": 0.0168,
+      "step": 5466
+    },
+    {
+      "epoch": 19.525,
+      "grad_norm": 2.7609685242355715,
+      "learning_rate": 5.9138371311546764e-08,
+      "loss": 0.0341,
+      "step": 5467
+    },
+    {
+      "epoch": 19.52857142857143,
+      "grad_norm": 0.4298861143632552,
+      "learning_rate": 5.82528460547338e-08,
+      "loss": 0.0077,
+      "step": 5468
+    },
+    {
+      "epoch": 19.53214285714286,
+      "grad_norm": 1.1040362906841563,
+      "learning_rate": 5.73739910679616e-08,
+      "loss": 0.0183,
+      "step": 5469
+    },
+    {
+      "epoch": 19.535714285714285,
+      "grad_norm": 0.5543239733257797,
+      "learning_rate": 5.650180664519278e-08,
+      "loss": 0.012,
+      "step": 5470
+    },
+    {
+      "epoch": 19.539285714285715,
+      "grad_norm": 0.9842481556861422,
+      "learning_rate": 5.563629307816287e-08,
+      "loss": 0.0093,
+      "step": 5471
+    },
+    {
+      "epoch": 19.542857142857144,
+      "grad_norm": 2.311627917069961,
+      "learning_rate": 5.477745065637807e-08,
+      "loss": 0.0269,
+      "step": 5472
+    },
+    {
+      "epoch": 19.54642857142857,
+      "grad_norm": 2.2637918051624437,
+      "learning_rate": 5.392527966710637e-08,
+      "loss": 0.0328,
+      "step": 5473
+    },
+    {
+      "epoch": 19.55,
+      "grad_norm": 0.3135250157532399,
+      "learning_rate": 5.3079780395390856e-08,
+      "loss": 0.006,
+      "step": 5474
+    },
+    {
+      "epoch": 19.553571428571427,
+      "grad_norm": 1.3856853112783114,
+      "learning_rate": 5.2240953124040874e-08,
+      "loss": 0.0171,
+      "step": 5475
+    },
+    {
+      "epoch": 19.557142857142857,
+      "grad_norm": 0.6710198891470134,
+      "learning_rate": 5.1408798133631977e-08,
+      "loss": 0.0134,
+      "step": 5476
+    },
+    {
+      "epoch": 19.560714285714287,
+      "grad_norm": 0.5788246832174612,
+      "learning_rate": 5.0583315702512624e-08,
+      "loss": 0.009,
+      "step": 5477
+    },
+    {
+      "epoch": 19.564285714285713,
+      "grad_norm": 1.7743375246801623,
+      "learning_rate": 4.976450610679529e-08,
+      "loss": 0.0297,
+      "step": 5478
+    },
+    {
+      "epoch": 19.567857142857143,
+      "grad_norm": 0.7721909235407681,
+      "learning_rate": 4.895236962036087e-08,
+      "loss": 0.0157,
+      "step": 5479
+    },
+    {
+      "epoch": 19.571428571428573,
+      "grad_norm": 1.1575582417260628,
+      "learning_rate": 4.814690651486098e-08,
+      "loss": 0.0134,
+      "step": 5480
+    },
+    {
+      "epoch": 19.575,
+      "grad_norm": 1.0195104523342509,
+      "learning_rate": 4.7348117059711206e-08,
+      "loss": 0.0188,
+      "step": 5481
+    },
+    {
+      "epoch": 19.57857142857143,
+      "grad_norm": 1.3240251646672108,
+      "learning_rate": 4.655600152209783e-08,
+      "loss": 0.0315,
+      "step": 5482
+    },
+    {
+      "epoch": 19.582142857142856,
+      "grad_norm": 1.2089670224130762,
+      "learning_rate": 4.577056016697334e-08,
+      "loss": 0.0274,
+      "step": 5483
+    },
+    {
+      "epoch": 19.585714285714285,
+      "grad_norm": 1.917628688950856,
+      "learning_rate": 4.499179325705871e-08,
+      "loss": 0.029,
+      "step": 5484
+    },
+    {
+      "epoch": 19.589285714285715,
+      "grad_norm": 0.7972181205796349,
+      "learning_rate": 4.421970105284113e-08,
+      "loss": 0.022,
+      "step": 5485
+    },
+    {
+      "epoch": 19.59285714285714,
+      "grad_norm": 1.6885675128084174,
+      "learning_rate": 4.3454283812578434e-08,
+      "loss": 0.0169,
+      "step": 5486
+    },
+    {
+      "epoch": 19.59642857142857,
+      "grad_norm": 0.7191281988685542,
+      "learning_rate": 4.2695541792290296e-08,
+      "loss": 0.0134,
+      "step": 5487
+    },
+    {
+      "epoch": 19.6,
+      "grad_norm": 1.34228755487406,
+      "learning_rate": 4.194347524576703e-08,
+      "loss": 0.026,
+      "step": 5488
+    },
+    {
+      "epoch": 19.603571428571428,
+      "grad_norm": 2.1114307212339534,
+      "learning_rate": 4.1198084424567406e-08,
+      "loss": 0.0303,
+      "step": 5489
+    },
+    {
+      "epoch": 19.607142857142858,
+      "grad_norm": 1.5022144170954026,
+      "learning_rate": 4.045936957801422e-08,
+      "loss": 0.0251,
+      "step": 5490
+    },
+    {
+      "epoch": 19.610714285714288,
+      "grad_norm": 1.9511110409485841,
+      "learning_rate": 3.9727330953198695e-08,
+      "loss": 0.0161,
+      "step": 5491
+    },
+    {
+      "epoch": 19.614285714285714,
+      "grad_norm": 1.5611085843971197,
+      "learning_rate": 3.900196879497831e-08,
+      "loss": 0.0121,
+      "step": 5492
+    },
+    {
+      "epoch": 19.617857142857144,
+      "grad_norm": 0.6523829154668997,
+      "learning_rate": 3.828328334597897e-08,
+      "loss": 0.0072,
+      "step": 5493
+    },
+    {
+      "epoch": 19.62142857142857,
+      "grad_norm": 1.3416283909175466,
+      "learning_rate": 3.757127484659284e-08,
+      "loss": 0.0289,
+      "step": 5494
+    },
+    {
+      "epoch": 19.625,
+      "grad_norm": 0.8649521517707942,
+      "learning_rate": 3.6865943534976076e-08,
+      "loss": 0.014,
+      "step": 5495
+    },
+    {
+      "epoch": 19.62857142857143,
+      "grad_norm": 0.6407355782431585,
+      "learning_rate": 3.616728964705329e-08,
+      "loss": 0.0106,
+      "step": 5496
+    },
+    {
+      "epoch": 19.632142857142856,
+      "grad_norm": 0.9632865637466769,
+      "learning_rate": 3.5475313416517555e-08,
+      "loss": 0.0182,
+      "step": 5497
+    },
+    {
+      "epoch": 19.635714285714286,
+      "grad_norm": 1.9638165259216958,
+      "learning_rate": 3.4790015074825935e-08,
+      "loss": 0.0284,
+      "step": 5498
+    },
+    {
+      "epoch": 19.639285714285712,
+      "grad_norm": 1.5000600936271424,
+      "learning_rate": 3.411139485119952e-08,
+      "loss": 0.0452,
+      "step": 5499
+    },
+    {
+      "epoch": 19.642857142857142,
+      "grad_norm": 1.2759578240543556,
+      "learning_rate": 3.3439452972632294e-08,
+      "loss": 0.0144,
+      "step": 5500
+    },
+    {
+      "epoch": 19.646428571428572,
+      "grad_norm": 1.2611762430029088,
+      "learning_rate": 3.27741896638778e-08,
+      "loss": 0.0258,
+      "step": 5501
+    },
+    {
+      "epoch": 19.65,
+      "grad_norm": 8.766578778291159,
+      "learning_rate": 3.211560514746248e-08,
+      "loss": 0.0141,
+      "step": 5502
+    },
+    {
+      "epoch": 19.65357142857143,
+      "grad_norm": 2.360359899934651,
+      "learning_rate": 3.146369964366791e-08,
+      "loss": 0.0137,
+      "step": 5503
+    },
+    {
+      "epoch": 19.65714285714286,
+      "grad_norm": 1.725281546027725,
+      "learning_rate": 3.0818473370555216e-08,
+      "loss": 0.0191,
+      "step": 5504
+    },
+    {
+      "epoch": 19.660714285714285,
+      "grad_norm": 1.3531439602560862,
+      "learning_rate": 3.017992654394064e-08,
+      "loss": 0.0234,
+      "step": 5505
+    },
+    {
+      "epoch": 19.664285714285715,
+      "grad_norm": 1.2710463222305293,
+      "learning_rate": 2.954805937741334e-08,
+      "loss": 0.0288,
+      "step": 5506
+    },
+    {
+      "epoch": 19.667857142857144,
+      "grad_norm": 1.0972598559002824,
+      "learning_rate": 2.892287208232203e-08,
+      "loss": 0.0152,
+      "step": 5507
+    },
+    {
+      "epoch": 19.67142857142857,
+      "grad_norm": 1.1284069756501023,
+      "learning_rate": 2.8304364867786095e-08,
+      "loss": 0.045,
+      "step": 5508
+    },
+    {
+      "epoch": 19.675,
+      "grad_norm": 0.4299082382771255,
+      "learning_rate": 2.7692537940688933e-08,
+      "loss": 0.0068,
+      "step": 5509
+    },
+    {
+      "epoch": 19.678571428571427,
+      "grad_norm": 1.904443121209477,
+      "learning_rate": 2.7087391505675743e-08,
+      "loss": 0.0281,
+      "step": 5510
+    },
+    {
+      "epoch": 19.682142857142857,
+      "grad_norm": 0.6345344091631395,
+      "learning_rate": 2.6488925765164598e-08,
+      "loss": 0.0141,
+      "step": 5511
+    },
+    {
+      "epoch": 19.685714285714287,
+      "grad_norm": 0.7645672498367444,
+      "learning_rate": 2.5897140919335374e-08,
+      "loss": 0.0111,
+      "step": 5512
+    },
+    {
+      "epoch": 19.689285714285713,
+      "grad_norm": 2.1291040996824746,
+      "learning_rate": 2.5312037166129734e-08,
+      "loss": 0.0278,
+      "step": 5513
+    },
+    {
+      "epoch": 19.692857142857143,
+      "grad_norm": 1.3170129512715418,
+      "learning_rate": 2.473361470125779e-08,
+      "loss": 0.0297,
+      "step": 5514
+    },
+    {
+      "epoch": 19.696428571428573,
+      "grad_norm": 1.585867244111678,
+      "learning_rate": 2.4161873718195895e-08,
+      "loss": 0.0156,
+      "step": 5515
+    },
+    {
+      "epoch": 19.7,
+      "grad_norm": 1.1031186214656374,
+      "learning_rate": 2.359681440818662e-08,
+      "loss": 0.0172,
+      "step": 5516
+    },
+    {
+      "epoch": 19.70357142857143,
+      "grad_norm": 1.1230084968311904,
+      "learning_rate": 2.3038436960229892e-08,
+      "loss": 0.0247,
+      "step": 5517
+    },
+    {
+      "epoch": 19.707142857142856,
+      "grad_norm": 0.8924211472551153,
+      "learning_rate": 2.2486741561102975e-08,
+      "loss": 0.0145,
+      "step": 5518
+    },
+    {
+      "epoch": 19.710714285714285,
+      "grad_norm": 1.2250273696909677,
+      "learning_rate": 2.1941728395336036e-08,
+      "loss": 0.0345,
+      "step": 5519
+    },
+    {
+      "epoch": 19.714285714285715,
+      "grad_norm": 0.7974957339616608,
+      "learning_rate": 2.140339764522992e-08,
+      "loss": 0.012,
+      "step": 5520
+    },
+    {
+      "epoch": 19.71785714285714,
+      "grad_norm": 0.6451678729730693,
+      "learning_rate": 2.0871749490851688e-08,
+      "loss": 0.008,
+      "step": 5521
+    },
+    {
+      "epoch": 19.72142857142857,
+      "grad_norm": 0.25042847056453255,
+      "learning_rate": 2.0346784110032436e-08,
+      "loss": 0.004,
+      "step": 5522
+    },
+    {
+      "epoch": 19.725,
+      "grad_norm": 0.804122185448423,
+      "learning_rate": 1.982850167836059e-08,
+      "loss": 0.0106,
+      "step": 5523
+    },
+    {
+      "epoch": 19.728571428571428,
+      "grad_norm": 0.5200864843352597,
+      "learning_rate": 1.9316902369201917e-08,
+      "loss": 0.0124,
+      "step": 5524
+    },
+    {
+      "epoch": 19.732142857142858,
+      "grad_norm": 0.8937100499484406,
+      "learning_rate": 1.8811986353675092e-08,
+      "loss": 0.0122,
+      "step": 5525
+    },
+    {
+      "epoch": 19.735714285714288,
+      "grad_norm": 2.276332210427932,
+      "learning_rate": 1.8313753800671686e-08,
+      "loss": 0.0499,
+      "step": 5526
+    },
+    {
+      "epoch": 19.739285714285714,
+      "grad_norm": 1.2711261732346788,
+      "learning_rate": 1.782220487684283e-08,
+      "loss": 0.0317,
+      "step": 5527
+    },
+    {
+      "epoch": 19.742857142857144,
+      "grad_norm": 0.7451442600940361,
+      "learning_rate": 1.7337339746608116e-08,
+      "loss": 0.0095,
+      "step": 5528
+    },
+    {
+      "epoch": 19.74642857142857,
+      "grad_norm": 0.871033893520522,
+      "learning_rate": 1.6859158572144484e-08,
+      "loss": 0.0149,
+      "step": 5529
+    },
+    {
+      "epoch": 19.75,
+      "grad_norm": 0.357391040643228,
+      "learning_rate": 1.6387661513399546e-08,
+      "loss": 0.0063,
+      "step": 5530
+    },
+    {
+      "epoch": 19.75357142857143,
+      "grad_norm": 0.38831916157235624,
+      "learning_rate": 1.5922848728084915e-08,
+      "loss": 0.0057,
+      "step": 5531
+    },
+    {
+      "epoch": 19.757142857142856,
+      "grad_norm": 0.7520067705952856,
+      "learning_rate": 1.5464720371674014e-08,
+      "loss": 0.0085,
+      "step": 5532
+    },
+    {
+      "epoch": 19.760714285714286,
+      "grad_norm": 0.5798355685764556,
+      "learning_rate": 1.5013276597404257e-08,
+      "loss": 0.009,
+      "step": 5533
+    },
+    {
+      "epoch": 19.764285714285712,
+      "grad_norm": 0.7762533557260889,
+      "learning_rate": 1.4568517556279304e-08,
+      "loss": 0.0173,
+      "step": 5534
+    },
+    {
+      "epoch": 19.767857142857142,
+      "grad_norm": 0.7915228203685105,
+      "learning_rate": 1.4130443397062376e-08,
+      "loss": 0.0206,
+      "step": 5535
+    },
+    {
+      "epoch": 19.771428571428572,
+      "grad_norm": 1.2680643353713381,
+      "learning_rate": 1.369905426628959e-08,
+      "loss": 0.0315,
+      "step": 5536
+    },
+    {
+      "epoch": 19.775,
+      "grad_norm": 1.1417758231876487,
+      "learning_rate": 1.3274350308249973e-08,
+      "loss": 0.0191,
+      "step": 5537
+    },
+    {
+      "epoch": 19.77857142857143,
+      "grad_norm": 1.6169444133050115,
+      "learning_rate": 1.2856331665005439e-08,
+      "loss": 0.0366,
+      "step": 5538
+    },
+    {
+      "epoch": 19.78214285714286,
+      "grad_norm": 1.6825390273558845,
+      "learning_rate": 1.2444998476375258e-08,
+      "loss": 0.0242,
+      "step": 5539
+    },
+    {
+      "epoch": 19.785714285714285,
+      "grad_norm": 0.6596665013093672,
+      "learning_rate": 1.2040350879947148e-08,
+      "loss": 0.0099,
+      "step": 5540
+    },
+    {
+      "epoch": 19.789285714285715,
+      "grad_norm": 0.9931954011548895,
+      "learning_rate": 1.164238901107062e-08,
+      "loss": 0.0145,
+      "step": 5541
+    },
+    {
+      "epoch": 19.792857142857144,
+      "grad_norm": 1.7240050601455132,
+      "learning_rate": 1.1251113002859193e-08,
+      "loss": 0.0176,
+      "step": 5542
+    },
+    {
+      "epoch": 19.79642857142857,
+      "grad_norm": 0.7686386326117168,
+      "learning_rate": 1.0866522986190397e-08,
+      "loss": 0.0141,
+      "step": 5543
+    },
+    {
+      "epoch": 19.8,
+      "grad_norm": 1.0465117392414407,
+      "learning_rate": 1.0488619089701335e-08,
+      "loss": 0.0144,
+      "step": 5544
+    },
+    {
+      "epoch": 19.803571428571427,
+      "grad_norm": 1.3854678547506172,
+      "learning_rate": 1.0117401439799778e-08,
+      "loss": 0.0162,
+      "step": 5545
+    },
+    {
+      "epoch": 19.807142857142857,
+      "grad_norm": 1.1785749308168618,
+      "learning_rate": 9.752870160653071e-09,
+      "loss": 0.0381,
+      "step": 5546
+    },
+    {
+      "epoch": 19.810714285714287,
+      "grad_norm": 1.610020883908374,
+      "learning_rate": 9.395025374192568e-09,
+      "loss": 0.0277,
+      "step": 5547
+    },
+    {
+      "epoch": 19.814285714285713,
+      "grad_norm": 0.27812506342829224,
+      "learning_rate": 9.043867200111412e-09,
+      "loss": 0.0058,
+      "step": 5548
+    },
+    {
+      "epoch": 19.817857142857143,
+      "grad_norm": 0.29048074225322407,
+      "learning_rate": 8.699395755868977e-09,
+      "loss": 0.0047,
+      "step": 5549
+    },
+    {
+      "epoch": 19.821428571428573,
+      "grad_norm": 0.8213871865786521,
+      "learning_rate": 8.361611156686433e-09,
+      "loss": 0.009,
+      "step": 5550
+    },
+    {
+      "epoch": 19.825,
+      "grad_norm": 1.7830783547954028,
+      "learning_rate": 8.030513515548954e-09,
+      "loss": 0.0169,
+      "step": 5551
+    },
+    {
+      "epoch": 19.82857142857143,
+      "grad_norm": 0.46047357508880615,
+      "learning_rate": 7.706102943203509e-09,
+      "loss": 0.0076,
+      "step": 5552
+    },
+    {
+      "epoch": 19.832142857142856,
+      "grad_norm": 1.9706918348496056,
+      "learning_rate": 7.388379548161073e-09,
+      "loss": 0.0267,
+      "step": 5553
+    },
+    {
+      "epoch": 19.835714285714285,
+      "grad_norm": 0.3954877024485901,
+      "learning_rate": 7.077343436698858e-09,
+      "loss": 0.0056,
+      "step": 5554
+    },
+    {
+      "epoch": 19.839285714285715,
+      "grad_norm": 0.7654017135297871,
+      "learning_rate": 6.772994712853642e-09,
+      "loss": 0.0163,
+      "step": 5555
+    },
+    {
+      "epoch": 19.84285714285714,
+      "grad_norm": 0.43550312917481976,
+      "learning_rate": 6.475333478423995e-09,
+      "loss": 0.0076,
+      "step": 5556
+    },
+    {
+      "epoch": 19.84642857142857,
+      "grad_norm": 2.865945911420734,
+      "learning_rate": 6.184359832976938e-09,
+      "loss": 0.0293,
+      "step": 5557
+    },
+    {
+      "epoch": 19.85,
+      "grad_norm": 1.4341033353382338,
+      "learning_rate": 5.900073873839063e-09,
+      "loss": 0.0399,
+      "step": 5558
+    },
+    {
+      "epoch": 19.853571428571428,
+      "grad_norm": 1.654438190609327,
+      "learning_rate": 5.62247569609875e-09,
+      "loss": 0.0204,
+      "step": 5559
+    },
+    {
+      "epoch": 19.857142857142858,
+      "grad_norm": 1.7559942899359502,
+      "learning_rate": 5.351565392610614e-09,
+      "loss": 0.0185,
+      "step": 5560
+    },
+    {
+      "epoch": 19.860714285714288,
+      "grad_norm": 0.9943093069108828,
+      "learning_rate": 5.087343053991056e-09,
+      "loss": 0.0102,
+      "step": 5561
+    },
+    {
+      "epoch": 19.864285714285714,
+      "grad_norm": 2.016310719467545,
+      "learning_rate": 4.829808768618272e-09,
+      "loss": 0.0378,
+      "step": 5562
+    },
+    {
+      "epoch": 19.867857142857144,
+      "grad_norm": 0.46001795289574965,
+      "learning_rate": 4.578962622636685e-09,
+      "loss": 0.0093,
+      "step": 5563
+    },
+    {
+      "epoch": 19.87142857142857,
+      "grad_norm": 0.5143728646061182,
+      "learning_rate": 4.33480469994585e-09,
+      "loss": 0.0118,
+      "step": 5564
+    },
+    {
+      "epoch": 19.875,
+      "grad_norm": 1.8902153894543496,
+      "learning_rate": 4.097335082220433e-09,
+      "loss": 0.0107,
+      "step": 5565
+    },
+    {
+      "epoch": 19.87857142857143,
+      "grad_norm": 1.0359415166822477,
+      "learning_rate": 3.86655384888579e-09,
+      "loss": 0.0168,
+      "step": 5566
+    },
+    {
+      "epoch": 19.882142857142856,
+      "grad_norm": 0.6917355346273424,
+      "learning_rate": 3.6424610771379486e-09,
+      "loss": 0.0148,
+      "step": 5567
+    },
+    {
+      "epoch": 19.885714285714286,
+      "grad_norm": 1.1156372667072416,
+      "learning_rate": 3.425056841932506e-09,
+      "loss": 0.0133,
+      "step": 5568
+    },
+    {
+      "epoch": 19.889285714285712,
+      "grad_norm": 0.9052681773587418,
+      "learning_rate": 3.2143412159868492e-09,
+      "loss": 0.0214,
+      "step": 5569
+    },
+    {
+      "epoch": 19.892857142857142,
+      "grad_norm": 1.4895771454613311,
+      "learning_rate": 3.010314269786818e-09,
+      "loss": 0.0166,
+      "step": 5570
+    },
+    {
+      "epoch": 19.896428571428572,
+      "grad_norm": 0.9195086705692631,
+      "learning_rate": 2.812976071573381e-09,
+      "loss": 0.0057,
+      "step": 5571
+    },
+    {
+      "epoch": 19.9,
+      "grad_norm": 0.9484501015558131,
+      "learning_rate": 2.6223266873559584e-09,
+      "loss": 0.0255,
+      "step": 5572
+    },
+    {
+      "epoch": 19.90357142857143,
+      "grad_norm": 0.8980057806420724,
+      "learning_rate": 2.43836618090354e-09,
+      "loss": 0.0115,
+      "step": 5573
+    },
+    {
+      "epoch": 19.90714285714286,
+      "grad_norm": 0.44420752682305625,
+      "learning_rate": 2.2610946137491264e-09,
+      "loss": 0.0079,
+      "step": 5574
+    },
+    {
+      "epoch": 19.910714285714285,
+      "grad_norm": 0.6532699280160803,
+      "learning_rate": 2.0905120451875094e-09,
+      "loss": 0.0108,
+      "step": 5575
+    },
+    {
+      "epoch": 19.914285714285715,
+      "grad_norm": 0.8425235218888137,
+      "learning_rate": 1.9266185322752707e-09,
+      "loss": 0.0167,
+      "step": 5576
+    },
+    {
+      "epoch": 19.917857142857144,
+      "grad_norm": 1.4844286654835126,
+      "learning_rate": 1.7694141298330026e-09,
+      "loss": 0.0254,
+      "step": 5577
+    },
+    {
+      "epoch": 19.92142857142857,
+      "grad_norm": 0.969721166613953,
+      "learning_rate": 1.618898890447529e-09,
+      "loss": 0.01,
+      "step": 5578
+    },
+    {
+      "epoch": 19.925,
+      "grad_norm": 0.44328982198489714,
+      "learning_rate": 1.4750728644585822e-09,
+      "loss": 0.0062,
+      "step": 5579
+    },
+    {
+      "epoch": 19.928571428571427,
+      "grad_norm": 1.8301877566041922,
+      "learning_rate": 1.3379360999787872e-09,
+      "loss": 0.0271,
+      "step": 5580
+    },
+    {
+      "epoch": 19.932142857142857,
+      "grad_norm": 1.8046767800340942,
+      "learning_rate": 1.207488642875898e-09,
+      "loss": 0.0147,
+      "step": 5581
+    },
+    {
+      "epoch": 19.935714285714287,
+      "grad_norm": 0.8740675481528634,
+      "learning_rate": 1.0837305367839001e-09,
+      "loss": 0.0085,
+      "step": 5582
+    },
+    {
+      "epoch": 19.939285714285713,
+      "grad_norm": 1.3960111641162127,
+      "learning_rate": 9.666618230985692e-10,
+      "loss": 0.0145,
+      "step": 5583
+    },
+    {
+      "epoch": 19.942857142857143,
+      "grad_norm": 1.6552655948464376,
+      "learning_rate": 8.562825409774711e-10,
+      "loss": 0.0267,
+      "step": 5584
+    },
+    {
+      "epoch": 19.946428571428573,
+      "grad_norm": 1.4847799567539777,
+      "learning_rate": 7.525927273421829e-10,
+      "loss": 0.0262,
+      "step": 5585
+    },
+    {
+      "epoch": 19.95,
+      "grad_norm": 1.0081289438217396,
+      "learning_rate": 6.555924168738514e-10,
+      "loss": 0.0097,
+      "step": 5586
+    },
+    {
+      "epoch": 19.95357142857143,
+      "grad_norm": 0.785158373672071,
+      "learning_rate": 5.652816420198548e-10,
+      "loss": 0.0181,
+      "step": 5587
+    },
+    {
+      "epoch": 19.957142857142856,
+      "grad_norm": 0.4212443728709062,
+      "learning_rate": 4.81660432987141e-10,
+      "loss": 0.0048,
+      "step": 5588
+    },
+    {
+      "epoch": 19.960714285714285,
+      "grad_norm": 1.2111823597732683,
+      "learning_rate": 4.0472881774666905e-10,
+      "loss": 0.0243,
+      "step": 5589
+    },
+    {
+      "epoch": 19.964285714285715,
+      "grad_norm": 1.5475172668618025,
+      "learning_rate": 3.344868220311881e-10,
+      "loss": 0.0124,
+      "step": 5590
+    },
+    {
+      "epoch": 19.96785714285714,
+      "grad_norm": 0.7572652841801586,
+      "learning_rate": 2.7093446933301736e-10,
+      "loss": 0.0171,
+      "step": 5591
+    },
+    {
+      "epoch": 19.97142857142857,
+      "grad_norm": 0.3719398601580941,
+      "learning_rate": 2.140717809129278e-10,
+      "loss": 0.0054,
+      "step": 5592
+    },
+    {
+      "epoch": 19.975,
+      "grad_norm": 1.1884507532616424,
+      "learning_rate": 1.638987757890398e-10,
+      "loss": 0.0251,
+      "step": 5593
+    },
+    {
+      "epoch": 19.978571428571428,
+      "grad_norm": 0.9961350352950346,
+      "learning_rate": 1.2041547074348458e-10,
+      "loss": 0.0133,
+      "step": 5594
+    },
+    {
+      "epoch": 19.982142857142858,
+      "grad_norm": 0.377500419333739,
+      "learning_rate": 8.362188032240426e-11,
+      "loss": 0.0074,
+      "step": 5595
+    },
+    {
+      "epoch": 19.985714285714288,
+      "grad_norm": 1.3160981746263587,
+      "learning_rate": 5.351801683151081e-11,
+      "loss": 0.0083,
+      "step": 5596
+    },
+    {
+      "epoch": 19.989285714285714,
+      "grad_norm": 1.5937463359901674,
+      "learning_rate": 3.010389034274752e-11,
+      "loss": 0.0258,
+      "step": 5597
+    },
+    {
+      "epoch": 19.992857142857144,
+      "grad_norm": 2.0010383427059897,
+      "learning_rate": 1.3379508683186715e-11,
+      "loss": 0.0302,
+      "step": 5598
+    },
+    {
+      "epoch": 19.99642857142857,
+      "grad_norm": 1.0162147252965172,
+      "learning_rate": 3.344877450572881e-12,
+      "loss": 0.0236,
+      "step": 5599
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.634060036283673,
+      "learning_rate": 0.0,
+      "loss": 0.0099,
+      "step": 5600
+    },
+    {
+      "epoch": 20.0,
+      "step": 5600,
+      "total_flos": 1.8486186976739328e+18,
+      "train_loss": 0.0793532435270442,
+      "train_runtime": 74354.0281,
+      "train_samples_per_second": 9.614,
+      "train_steps_per_second": 0.075
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5600,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.8486186976739328e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}